ATMEL AT91SAM9G45

Features
• 400 MHz ARM926EJ-S™ ARM® Thumb® Processor
– 32 KBytes Data Cache, 32 KBytes Instruction Cache, MMU
• Memories
•
•
•
•
– DDR2 Controller 4-bank DDR2/LPDDR, SDRAM/LPSDR
– External Bus Interface supporting 4-bank DDR2/LPDDR, SDRAM/LPSDR, Static
Memories, CompactFlash, SLC NAND Flash with ECC
– One 64-KByte internal SRAM, single-cycle access at system speed or processor
speed through TCM interface
– One 64-KByte internal ROM, embedding bootstrap routine
Peripherals
– LCD Controller supporting STN and TFT displays up to 1280*860
– ITU-R BT. 601/656 Image Sensor Interface
– USB Device High Speed, USB Host High Speed and USB Host Full Speed with OnChip Transceiver
– 10/100 Mbps Ethernet MAC Controller
– Two High Speed Memory Card Hosts (SDIO, SDCard, MMC)
– AC'97 controller
– Two Master/Slave Serial Peripheral Interfaces
– Two Three-channel 32-bit Timer/Counters
– Two Synchronous Serial Controllers (I2S mode)
– Four-channel 16-bit PWM Controller
– Two Two-wire Interfaces
– Four USARTs with ISO7816, IrDA, Manchester and SPI modes
– 8-channel 10-bit ADC with 4-wire Touch Screen support
System
– 133 MHz twelve 32-bit layer AHB Bus Matrix
– 37 DMA Channels
– Boot from NAND Flash, SDCard, DataFlash® or serial DataFlash
– Reset Controller with on-chip Power-on Reset
– Selectable 32768 Hz Low-power and 12 MHz Crystal Oscillators
– Internal Low-power 32 kHz RC Oscillator
– One PLL for the system and one 480 MHz PLL optimized for USB High Speed
– Two Programmable External Clock Signals
– Advanced Interrupt Controller and Debug Unit
– Periodic Interval Timer, Watchdog Timer, Real Time Timer and Real Time Clock
I/O
– Five 32-bit Parallel Input/Output Controllers
– 160 Programmable I/O Lines Multiplexed with up to Two Peripheral I/Os with
Schmitt trigger input
Package
– 324-ball TFBGA, pitch 0.8 mm
AT91 ARM
Thumb-based
Microcontrollers
AT91SAM9G45
Preliminary
6438D–ATARM–13-Oct-09
1. Description
The ARM926EJ-S based AT91SAM9G45 features the frequently demanded combination of user
interface functionality and high data rate connectivity, including LCD Controller, resistive touchscreen, camera interface, audio, Ethernet 10/100 and high speed USB and SDIO. With the processor running at 400MHz and multiple 100+ Mbps data rate peripherals, the AT91SAM9G45
has the performance and bandwidth to the network or local storage media to provide an adequate user experience.
The AT91SAM9G45 supports the latest generation of DDR2 and NAND Flash memory interfaces for program and data storage. An internal 133 MHz multi-layer bus architecture associated
with 37 DMA channels, a dual external bus interface and distributed memory including a 64KByte SRAM which can be configured as a tightly coupled memory (TCM) sustains the high
bandwidth required by the processor and the high speed peripherals.
The I/Os support 1.8V or 3.3V operation, which are independently configurable for the memory
interface and peripheral I/Os. This feature completely eliminates the need for any external level
shifters. In addition it supports 0.8 ball pitch package for low cost PCB manufacturing.
The AT91SAM9G45 power management controller features efficient clock gating and a battery
backup section minimizing power consumption in active and standby modes.
2
AT91SAM9G45
6438D–ATARM–13-Oct-09
6438D–ATARM–13-Oct-09
PDC
DBGU
AIC
MCI0/MCI1
SD/SDIO
CE ATA
TWI0
TWI1
PIOE
FIFO
PIOD
RSTC
PIOA
POR
RTC
PIOB
PIOC
POR
VDDCORE
SHDC
RTT
4
GPBR
RC
OSC 32K
PIT
WDT
OSC12M
PLLUTMI PMC
PLLA
VDDBU
NRST
XIN32
XOUT32
SHDN
WKUP
XIN
XOUT
DRXD
DTXD
FIQ
IRQ
System Controller
PIO
JTA
GS
EL
NT
RS
T
TD
I
TD
O
TM
TC S
K
RT
CK
USART0
USART1
USART2
USART3
PDC
ROM
64KB
SRAM
64KB
4-CH
PWM
I
TC0
TC1
TC2
D
DCache
ICache
MMU 32 Kbytes
32 Kbytes
ITCM DTCM Bus Interface
ARM926EJ-S
In-Circuit Emulator
JTAG / Boundary Scan
S
PB
TC3
TC4
TC5
DMA
DMA
HS
USB
HS
Transceiver
DMA
LCD
PIO
RNG
SPI0
SPI1
PDC
PIO
SSC0
SSC1
PDC
Peripheral
DMA
Controller
DMA
ISI
SSC0_, SSC1_
Peripheral
Bridge
Multi-Layer AHB Matrix
HS EHCI
USB HOST
PA
HS
Transceiver
AC97
PDC
DMA
EMAC
PDC
8-CH
10Bit ADC
TouchScreen
8-CH
DMA
HF
S
HH DP
SD A,H
P F
A
,H SDM
HS A
VB
DM
G
A
DF
S
DH DP/
SD HFS
P/H DP
HS B,
LC
D
D
PB FSD
D
LC D0
,D M
HS /H
D -L
D F
L VS CD
S
M
C
D YN D2
/H DM
HS B
L DO C 3
D
DM
D T ,LC
LC EN CK DH
B
SY
DP ,LC
NC
IS WR DC
I_
, C
I DO LC
S
I_ -IS DM
IS PCK I_D OD
I_
11
IS HS
I
_ Y
IS VS NC
I_M YN
C
C
K
ET
X
ET CK
X
E
EC EN RX
R -E C
ER S-E TX K
E
ERXER COL R
X ET 0-E ERX
X R D
EM 0-ET X3 V
EMDC X3
DI
O
BM
APB
SPI0_, SPI1_
M
CI
M 0_D
CI
0_ A0C M
M DA CI0
C
,M _
M I0_ CI DA
CI
C
1_ 7
1
_D K,M CD
A0 CI A
-M 1_C
C
I
1_ K
DA
TW
7
TW D0
CK -TW
0D
CT TW 1
S C
RT 0- K1
S CT
SC 0-R S3
K T
RD 0-S S3
X C
TX 0-R K3
D
D
0- X3
TX
PW
D3
M
0PW
TC
M
LK
3
T 0
I
O -TC
A0 L
-T K2
TI
IO
O
TC B0 A2
-T
L
TI K3 IO
O - B
TI A3 TCL 2
O B3 TIOK5
-T A5
IO
B
NP 5
NPCS
C 3
NP S2
NPCS
1
C
SP S0
C
M K
O
M SI
T ISO
K
0
TF -TK
TD 0-T 1
F
R 0-T 1
D
0 D
R -RD1
F
0
RK -R 1
0- F1
AC RK1
AC97C
9 K
AC 7F
S
AC97R
X
TS 97T
AD X
TR
A IG
D
0X
AD P
1
AD XM
2Y
GP
AD AD3 P
Y
4
TS -GP M
A
DVAD7
VD RE
DA F
GN NA
DA
N
Static
Memory
Controller
CF
NAND Flash
Controller
ECC
DDR2/
LPDDR/
SDRAM
Controller
EBI
DDR2
LPDDR
D16-D31
NWAIT
DQM[2..3]
A19-A24
NCS4/CFCS0
NCS5/CFCS1
A25/CFRNW
CFCE1-CFCE2
NCS2
NCS3/NANDCS
D0-D15
A0/NBS0
A1/NBS2/NWR2
A2-A15, A18
A16/BA0
A17/BA1
NCS1/SDCS
SDCK, #SDCK, SDCKE
RAS, CAS
SDWE, SDA10
DQM[0..1]
DQS[0..1]
NRD
NWR0/NWE
NWR1/NBS1
NWR3/NBS3
NCS0
NANDOE, NANDWE
DDR_CS
DDR_CLK,#DDR_CLK
DDR_CKE
DDR_RAS, DDR_CAS
DDR_WE
DDR_BA0, DDR_BA1
DDR_A0-DDR_A13
DDR_D0-DDR_D15
DDR_VREF
DDR_DQM[0..1]
DDR_DQS[0..1]
Figure 2-1.
PCK0-PCK1
TST
AT91SAM9G45
2. Block Diagram
AT91SAM9G45 Block Diagram
3
3. Signal Description
Table 3-1 gives details on the signal names classified by peripheral.
Table 3-1.
Signal Name
Signal Description List
Function
Type
Active
Level
Reference
Voltage
Comments
Power Supplies
VDDIOM0
DDR2 I/O Lines Power Supply
Power
1.65V to 1.95V
VDDIOM1
EBI I/O Lines Power Supply
Power
1.65V to 1.95V or 3.0V to3.6V
VDDIOP0
Peripherals I/O Lines Power Supply
Power
1.65V to 3.6V
VDDIOP1
Peripherals I/O Lines Power Supply
Power
1.65V to 3.6V
VDDIOP2
ISI I/O Lines Power Supply
Power
1.65V to 3.6V
VDDBU
Backup I/O Lines Power Supply
Power
1.8V to 3.6V
VDDANA
Analog Power Supply
Power
3.0V to 3.6V
VDDPLLA
PLLA Power Supply
Power
0.9V to 1.1V
VDDPLLUTMI
PLLUTMI Power Supply
Power
0.9V to 1.1V
VDDOSC
Oscillator Power Supply
Power
1.65V to 3.6V
VDDCORE
Core Chip Power Supply
Power
0.9V to 1.1V
VDDUTMIC
UDPHS and UHPHS UTMI+ Core
Power Supply
Power
0.9V to 1.1V
VDDUTMII
UDPHS and UHPHS UTMI+ interface
Power Supply
Power
3.0V to 3.6V
GNDIOM
DDR2 and EBI I/O Lines Ground
Ground
GNDIOP
Peripherals and ISI I/O lines Ground
Ground
GNDCORE
Core Chip Ground
Ground
GNDOSC
PLLA, PLLUTMI and Oscillator
Ground
Ground
GNDBU
Backup Ground
Ground
GNDUTMI
UDPHS and UHPHS UTMI+ Core and
interface Ground
Ground
GNDANA
Analog Ground
Ground
Clocks, Oscillators and PLLs
XIN
Main Oscillator Input
Input
XOUT
Main Oscillator Output
Output
XIN32
Slow Clock Oscillator Input
Input
XOUT32
Slow Clock Oscillator Output
Output
VBG
Bias Voltage Reference for USB
Analog
PCK0 - PCK1
Programmable Clock Output
Output
4
(1)
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Table 3-1.
Signal Description List (Continued)
Signal Name
Function
Type
Active
Level
Reference
Voltage
Comments
Shutdown, Wakeup Logic
SHDN
Shut-Down Control
Output
VDDBU
Driven at 0V only.
0: The device is in backup
mode
1: The device is running (not in
backup mode).
WKUP
Wake-Up Input
Input
VDDBU
Accept between 0V and
VDDBU.
ICE and JTAG
TCK
Test Clock
Input
VDDIOP0
No pull-up resistor, Schmitt
trigger
TDI
Test Data In
Input
VDDIOP0
No pull-up resistor, Schmitt
trigger
TDO
Test Data Out
Output
VDDIOP0
TMS
Test Mode Select
Input
VDDIOP0
No pull-up resistor, Schmitt
trigger
JTAGSEL
JTAG Selection
Input
VDDBU
Pull-down resistor (15 kΩ).
RTCK
Return Test Clock
Output
VDDIOP0
Reset/Test
VDDIOP0
Pull-Up resistor (100 kΩ),
Schmitt trigger
Input
VDDBU
Pull-down resistor (15 kΩ),
Schmitt trigger
Test Reset Signal
Input
VDDIOP0
Pull-Up resistor (100 kΩ),
Schmitt trigger
Boot Mode Select
Input
VDDIOP0
must be connected to GND or
VDDIOP0.
NRST
Microcontroller Reset (2)
I/O
TST
Test Mode Select
NTRST
BMS
Low
Debug Unit - DBGU
DRXD
Debug Receive Data
Input
(1)
DTXD
Debug Transmit Data
Output
(1)
Advanced Interrupt Controller - AIC
IRQ
External Interrupt Input
Input
(1)
FIQ
Fast Interrupt Input
Input
(1)
PIO Controller - PIOA- PIOB - PIOC - PIOD - PIOE
PA0 - PA31
Parallel IO Controller A
I/O
(1)
PB0 - PB31
Parallel IO Controller B
I/O
(1)
PC0 - PC31
Parallel IO Controller C
I/O
(1)
Pulled-up input at reset
(100kΩ)(3), Schmitt trigger
Pulled-up input at reset
(100kΩ)(3), Schmitt trigger
Pulled-up input at reset
(100kΩ)(3), Schmitt trigger
5
6438D–ATARM–13-Oct-09
Table 3-1.
Signal Description List (Continued)
Signal Name
Function
Type
Active
Level
Reference
Voltage
PD0 - PD31
Parallel IO Controller D
I/O
(1)
PE0 - PE31
Parallel IO Controller E
I/O
(1)
Comments
Pulled-up input at reset
(100kΩ)(3), Schmitt trigger
Pulled-up input at reset
(100kΩ)(3), Schmitt trigger
DDR Memory Interface- DDR2/SDRAM/LPDDR Controller
DDR_D0 DDR_D15
Data Bus
I/O
VDDIOM0
Pulled-up input at reset
DDR_A0 DDR_A13
Address Bus
Output
VDDIOM0
0 at reset
DDR_CLK#DDR_CLK
DDR differential clock input
Output
VDDIOM0
DDR_CKE
DDR Clock Enable
Output
High
VDDIOM0
DDR_CS
DDR Chip Select
Output
Low
VDDIOM0
DDR_WE
DDR Write Enable
Output
Low
VDDIOM0
DDR_RASDDR_CAS
Row and Column Signal
Output
Low
VDDIOM0
DDR_DQM[0..1]
Write Data Mask
Output
VDDIOM0
DDR_DQS[0..1]
Data Strobe
Output
VDDIOM0
DDR_BA0 DDR_BA1
Bank Select
Output
VDDIOM0
DDR_VREF
Reference Voltage
Input
VDDIOM0
External Bus Interface - EBI
D0 -D31
Data Bus
I/O
VDDIOM1
Pulled-up input at reset
A0 - A25
Address Bus
Output
VDDIOM1
0 at reset
NWAIT
External Wait Signal
Input
Low
VDDIOM1
Static Memory Controller - SMC
NCS0 - NCS5
Chip Select Lines
Output
Low
VDDIOM1
NWR0 - NWR3
Write Signal
Output
Low
VDDIOM1
NRD
Read Signal
Output
Low
VDDIOM1
NWE
Write Enable
Output
Low
VDDIOM1
NBS0 - NBS3
Byte Mask Signal
Output
Low
VDDIOM1
CompactFlash Support
CFCE1 - CFCE2
CompactFlash Chip Enable
Output
Low
VDDIOM1
CFOE
CompactFlash Output Enable
Output
Low
VDDIOM1
CFWE
CompactFlash Write Enable
Output
Low
VDDIOM1
CFIOR
CompactFlash IO Read
Output
Low
VDDIOM1
CFIOW
CompactFlash IO Write
Output
Low
VDDIOM1
CFRNW
CompactFlash Read Not Write
Output
6
VDDIOM1
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Table 3-1.
Signal Description List (Continued)
Signal Name
CFCS0 -CFCS1
Function
Type
CompactFlash Chip Select Lines
Output
Active
Level
Low
Reference
Voltage
Comments
VDDIOM1
NAND Flash Support
NANDCS
NAND Flash Chip Select
Output
Low
VDDIOM1
NANDOE
NAND Flash Output Enable
Output
Low
VDDIOM1
NANDWE
NAND Flash Write Enable
Output
Low
VDDIOM1
DDR2/SDRAM/LPDDR Controller
SDCK,#SDCK
DDR2/SDRAM differential clock
Output
VDDIOM1
SDCKE
DDR2/SDRAM Clock Enable
Output
High
VDDIOM1
SDCS
DDR2/SDRAM Controller Chip Select
Output
Low
VDDIOM1
BA0 - BA1
Bank Select
Output
SDWE
DDR2/SDRAM Write Enable
Output
Low
VDDIOM1
RAS - CAS
Row and Column Signal
Output
Low
VDDIOM1
SDA10
SDRAM Address 10 Line
Output
VDDIOM1
DQS[0..1]
Data Strobe
Output
VDDIOM1
DQM[0..3]
Write Data Mask
Output
VDDIOM1
VDDIOM1
High Speed Multimedia Card Interface - HSMCIx
MCIx_CK
Multimedia Card Clock
I/O
(1)
MCIx_CDA
Multimedia Card Slot A Command
I/O
(1)
MCIx_DA0 MCIx_DA7
Multimedia Card Slot A Data
I/O
(1)
Universal Synchronous Asynchronous Receiver Transmitter - USARTx
SCKx
USARTx Serial Clock
I/O
(1)
TXDx
USARTx Transmit Data
Output
(1)
RXDx
USARTx Receive Data
Input
(1)
RTSx
USARTx Request To Send
Output
(1)
CTSx
USARTx Clear To Send
Input
(1)
Synchronous Serial Controller - SSCx
TDx
SSC Transmit Data
Output
(1)
RDx
SSC Receive Data
Input
(1)
TKx
SSC Transmit Clock
I/O
(1)
RKx
SSC Receive Clock
I/O
(1)
TFx
SSC Transmit Frame Sync
I/O
(1)
RFx
SSC Receive Frame Sync
I/O
(1)
7
6438D–ATARM–13-Oct-09
Table 3-1.
Signal Name
Signal Description List (Continued)
Function
Type
Active
Level
Reference
Voltage
Comments
AC97 Controller - AC97C
AC97RX
AC97 Receive Signal
Input
(1)
AC97TX
AC97 Transmit Signal
Output
(1)
AC97FS
AC97 Frame Synchronization Signal
Output
(1)
AC97CK
AC97 Clock signal
Input
(1)
Time Counter - TCx
TCLKx
TC Channel x External Clock Input
Input
(1)
TIOAx
TC Channel x I/O Line A
I/O
(1)
TIOBx
TC Channel x I/O Line B
I/O
(1)
Pulse Width Modulation Controller - PWM
PWMx
Pulse Width Modulation Output
(1)
Output
Serial Peripheral Interface - SPIx_
SPIx_MISO
Master In Slave Out
I/O
(1)
SPIx_MOSI
Master Out Slave In
I/O
(1)
SPIx_SPCK
SPI Serial Clock
I/O
(1)
SPIx_NPCS0
SPI Peripheral Chip Select 0
I/O
Low
(1)
SPIx_NPCS1SPIx_NPCS3
SPI Peripheral Chip Select
Output
Low
(1)
Two-Wire Interface
TWDx
Two-wire Serial Data
I/O
(1)
TWCKx
Two-wire Serial Clock
I/O
(1)
USB Host High Speed Port - UHPHS
HFSDPA
USB Host Port A Full Speed Data +
Analog
VDDUTMII
HFSDMA
USB Host Port A Full Speed Data -
Analog
VDDUTMII
HHSDPA
USB Host Port A High Speed Data +
Analog
VDDUTMII
HHSDMA
USB Host Port A High Speed Data -
Analog
VDDUTMII
HFSDPB
USB Host Port B Full Speed Data +
Analog
VDDUTMII
Multiplexed with DFSDP
HFSDMB
USB Host Port B Full Speed Data -
Analog
VDDUTMII
Multiplexed with DFSDM
HHSDPB
USB Host Port B High Speed Data +
Analog
VDDUTMII
Multiplexed with DHSDP
HHSDMB
USB Host Port B High Speed Data -
Analog
VDDUTMII
Multiplexed with DHSDM
USB Device High Speed Port - UDPHS
DFSDM
USB Device Full Speed Data -
Analog
VDDUTMII
DFSDP
USB Device Full Speed Data +
Analog
VDDUTMII
DHSDM
USB Device High Speed Data -
Analog
VDDUTMII
DHSDP
USB Device High Speed Data +
Analog
VDDUTMII
8
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Table 3-1.
Signal Description List (Continued)
Signal Name
Function
Type
Active
Level
Reference
Voltage
Comments
Ethernet 10/100
ETXCK
ERXCK
Transmit Clock or Reference Clock
Receive Clock
Input
(1)
MII only, REFCK in RMII
Input
(1)
MII only
ETXEN
Transmit Enable
Output
(1)
ETX0-ETX3
Transmit Data
Output
(1)
ETX0-ETX1 only in RMII
Output
(1)
MII only
Input
(1)
RXDV in MII, CRSDV in RMII
ERX0-ERX1 only in RMII
ETXER
ERXDV
Transmit Coding Error
Receive Data Valid
ERX0-ERX3
Receive Data
Input
(1)
ERXER
Receive Error
Input
(1)
ECRS
Carrier Sense and Data Valid
Input
(1)
MII only
Input
(1)
MII only
ECOL
Collision Detect
EMDC
Management Data Clock
Output
(1)
EMDIO
Management Data Input/Output
I/O
(1)
Image Sensor Interface
ISI_D0-ISI_D11
Image Sensor Data
Input
VDDIOP2
ISI_MCK
Image sensor Reference clock
output
VDDIOP2
ISI_HSYNC
Image Sensor Horizontal Synchro
input
VDDIOP2
ISI_VSYNC
Image Sensor Vertical Synchro
input
VDDIOP2
ISI_PCK
Image Sensor Data clock
input
VDDIOP2
LCD Controller - LCDC
LCDD0 LCDD23
LCD Data Bus
Output
VDDIOP1
LCDVSYNC
LCD Vertical Synchronization
Output
VDDIOP1
LCDHSYNC
LCD Horizontal Synchronization
Output
VDDIOP1
LCDDOTCK
LCD Dot Clock
Output
VDDIOP1
LCDDEN
LCD Data Enable
Output
VDDIOP1
LCDCC
LCD Contrast Control
Output
VDDIOP1
LCDPWR
LCD panel Power enable control
Output
VDDIOP1
LCDMOD
LCD Modulation signal
Output
VDDIOP1
Touch Screen Analog-to-Digital Converter
AD0XP
Analog input channel 0 or
Touch Screen Top channel
Analog
VDDANA
Multiplexed with AD0
AD1XM
Analog input channel 1 or
Touch Screen Bottom channel
Analog
VDDANA
Multiplexed with AD1
AD2YP
Analog input channel 2 or
Touch Screen Right channel
Analog
VDDANA
Multiplexed with AD2
9
6438D–ATARM–13-Oct-09
Table 3-1.
Signal Description List (Continued)
Signal Name
Function
Type
Active
Level
Reference
Voltage
Comments
Multiplexed with AD3
AD3YM
Analog input channel 3 or
Touch Screen Left channel
Analog
VDDANA
GPAD4-GPAD7
Analog Inputs
Analog
VDDANA
TSADTRG
ADC Trigger
Input
VDDANA
TSADVREF
ADC Reference
Analog
VDDANA
Notes:
1. Refer to peripheral multiplexing tables in Section 8.4 “Peripheral Signals Multiplexing on I/O Lines” for these signals.
2. When configured as an input, the NRST pin enables asynchronous reset of the device when asserted low. This allows connection of a simple push button on the NRST pin as a system-user reset.
3. Programming of this pull-up resistor is performed independently for each I/O line through the PIO Controllers. After reset, all
the I/O lines default as inputs with pull-up resistors enabled, except those which are multiplexed with the External Bus Interface signals that require to be enabled as Peripheral at reset. This is explicitly indicated in the column “Reset State” of the
peripheral multiplexing tables.
10
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
4. Package and Pinout
The AT91SAM9G45 is delivered in a 324-ball TFBGA package.
4.1
Mechanical Overview of the 324-ball TFBGA Package
Figure 4-1 shows the orientation of the 324-ball TFBGA Package
Figure 4-1.
Orientation of the 324-ball TFBGA Package
Bottom VIEW
V
U
T
R
P
N
M
L
K
J
H
G
F
E
D
C
B
A
1 2
3 4 5 6 7 8 9 10 11 12 13 14 15
16 17 18
11
6438D–ATARM–13-Oct-09
4.2
324-ball TFBGA Package Pinout
Table 4-1.
AT91SAM9G45 Pinout for 324-ball BGA Package
Pin
Signal Name
Pin
Signal Name
Pin
Signal Name
Pin
A1
PC27
E10
NANDWE
K1
PE21
P10
Signal Name
TMS
A2
PC28
E11
DQS1
K2
PE23
P11
VDDPLLA
A3
PC25
E12
D13
K3
PE26
P12
PB20
A4
PC20
E13
D11
K4
PE22
P13
PB31
A5
PC12
E14
A4
K5
PE24
P14
DDR_D7
A6
PC7
E15
A8
K6
PE25
P15
DDR_D3
A7
PC5
E16
A9
K7
PE27
P16
DDR_D4
A8
PC0
E17
A7
K8
PE28
P17
DDR_D5
A9
NWR3/NBS3
E18
VDDCORE
K9
VDDIOP0
P18
DDR_D10
A10
NCS0
F1
PD22
K10
VDDIOP0
R1
PA18
A11
DQS0
F2
PD24
K11
GNDIOM
R2
PA20
A12
RAS
F3
SHDN
K12
GNDIOM
R3
PA24
A13
SDCK
F4
PE1
K13
VDDIOM0
R4
PA30
PB4
A14
NSDCK
F5
PE3
K14
DDR_A7
R5
A15
D7
F6
VDDIOM1
K15
DDR_A8
R6
PB13
A16
DDR_VREF
F7
PC19
K16
DDR_A9
R7
PD0
A17
D0
F8
PC14
K17
DDR_A11
R8
PD9
A18
A14
F9
PC4
K18
DDR_A10
R9
PD18
B1
PC31
F10
NCS1/SDCS
L1
PA0
R10
TDI
B2
PC29
F11
NRD
L2
PE30
R11
RTCK
B3
PC30
F12
SDWE
L3
PE29
R12
PB22
B4
PC22
F13
A0/NBS0
L4
PE31
R13
PB29
B5
PC17
F14
A1/NBS2/NWR2
L5
PA2
R14
DDR_D6
B6
PC10
F15
A3
L6
PA4
R15
DDR_D1
B7
PC11
F16
A6
L7
PA8
R16
DDR_D0
B8
PC2
F17
A5
L8
PD2
R17
HHSDMA
B9
SDA10
F18
A2
L9
PD13
R18
HFSDMA
B10
A17/BA1
G1
PD25
L10
PD29
T1
PA22
B11
DQM0
G2
PD23
L11
PD31
T2
PA25
B12
SDCKE
G3
PE6
L12
VDDIOM0
T3
PA26
B13
D12
G4
PE0
L13
VDDIOM0
T4
PB0
B14
D8
G5
PE2
L14
DDR_A1
T5
PB6
B15
D4
G6
PE8
L15
DDR_A3
T6
PB16
B16
D3
G7
PE4
L16
DDR_A4
T7
PD1
B17
A15
G8
PE11
L17
DDR_A6
T8
PD11
B18
A13
G9
GNDCORE
L18
DDR_A5
T9
PD19
C1
XIN32
G10
VDDIOM1
M1
PA1
T10
PD30
C2
GNDANA
G11
VDDIOM1
M2
PA5
T11
BMS
C3
WKUP
G12
VDDCORE
M3
PA6
T12
PB8
C4
PC26
G13
VDDCORE
M4
PA7
T13
PB30
C5
PC21
G14
DDR_DQM0
M5
PA10
T14
DDR_D2
C6
PC15
G15
DDR_DQS1
M6
PA14
T15
PB21
C7
PC9
G16
DDR_BA1
M7
PB14
T16
PB23
C8
PC3
G17
DDR_BA0
M8
PD4
T17
HHSDPA
C9
NWR0/NWE
G18
DDR_DQS0
M9
PD15
T18
HFSDPA
C10
A16/BA0
H1
PD26
M10
NRST
U1
PA27
C11
CAS
H2
PD27
M11
PB11
U2
PA29
C12
D15
H3
VDDIOP1
M12
PB25
U3
PA28
12
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Table 4-1.
AT91SAM9G45 Pinout for 324-ball BGA Package (Continued)
Pin
Signal Name
Pin
Signal Name
Pin
Signal Name
Pin
C13
D10
H4
PE13
M13
PB27
U4
Signal Name
PB3
C14
D6
H5
PE5
M14
VDDIOM0
U5
PB7
C15
D2
H6
PE7
M15
DDR_D14
U6
PB17
C16
GNDIOM
H7
PE9
M16
DDR_D15
U7
PD7
C17
A18
H8
PE10
M17
DDR_A0
U8
PD10
C18
A12
H9
GNDCORE
M18
DDR_A2
U9
PD14
D1
XOUT32
H10
GNDIOP
N1
PA3
U10
TCK
D2
PD20
H11
VDDCORE
N2
PA9
U11
VDDOSC
D3
GNDBU
H12
GNDIOM
N3
PA12
U12
GNDOSC
D4
VDDBU
H13
GNDIOM
N4
PA15
U13
PB10
D5
PC24
H14
DDR_CS
N5
PA16
U14
PB26
D6
PC18
H15
DDR_WE
N6
PA17
U15
HHSDPB/DHSDP
D7
PC13
H16
DDR_DQM1
N7
PB18
U16
HHSDMB/DHSDM
D8
PC6
H17
DDR_CAS
N8
PD6
U17
GNDUTMI
D9
NWR1/NBS1
H18
DDR_NCLK
N9
PD16
U18
VDDUTMIC
D10
NANDOE
J1
PE19
N10
NTRST
V1
PA31
D11
DQM1
J2
PE16
N11
PB9
V2
PB1
D12
D14
J3
PE14
N12
PB24
V3
PB2
D13
D9
J4
PE15
N13
PB28
V4
PB5
D14
D5
J5
PE12
N14
DDR_D13
V5
PB15
D15
D1
J6
PE17
N15
DDR_D8
V6
PD3
D16
VDDIOM1
J7
PE18
N16
DDR_D9
V7
PD5
D17
A11
J8
PE20
N17
DDR_D11
V8
PD12
D18
A10
J9
GNDCORE
N18
DDR_D12
V9
PD17
E1
PD21
J10
GNDCORE
P1
PA11
V10
TDO
E2
TSADVREF
J11
GNDIOP
P2
PA13
V11
XOUT
E3
VDDANA
J12
GNDIOM
P3
PA19
V12
XIN
E4
JTAGSEL
J13
GNDIOM
P4
PA21
V13
VDDPLLUTMI
E5
TST
J14
DDR_A12
P5
PA23
V14
VDDIOP2
E6
PC23
J15
DDR_A13
P6
PB12
V15
HFSDPB/DFSDP
E7
PC16
J16
DDR_CKE
P7
PB19
V16
HFSDMB/DFSDM
E8
PC8
J17
DDR_RAS
P8
PD8
V17
VDDUTMII
E9
PC1
J18
DDR_CLK
P9
PD28
V18
VBG
13
6438D–ATARM–13-Oct-09
5. Power Considerations
5.1
Power Supplies
The AT91SAM9G45 has several types of power supply pins:
• VDDCORE pins: Power the core, including the processor, the embedded memories and the
peripherals; voltage ranges from 0.9V to 1.1V, 1.0V nominal.
• VDDIOM0 pins: Power the DDR2/LPDDR I/O lines; voltage ranges between 1.65V and 1.95V
(1.8V typical).
• VDDIOM1 pins: Power the External Bus Interface 1 I/O lines; voltage ranges between 1.65V
and 1.95V (1.8V typical) or between 3.0V and 3.6V (3.3V nominal).
• VDDIOP0, VDDIOP1, VDDIOP2 pins: Power the Peripherals I/O lines; voltage ranges from
1.65V to 3.6V.
• VDDBU pin: Powers the Slow Clock oscillator, the internal RC oscillator and a part of the
System Controller; voltage ranges from 1.8V to 3.6V.
• VDDPLLUTMI Powers the PLLUTMI cell; voltage range from 0.9V to 1.1V.
• VDDUTMIC pin: Powers the USB device and host UTMI+ core; voltage range from 0.9V to
1.1V, 1.0V nominal.
• VDDUTMII pin: Powers the USB device and host UTMI+ interface; voltage range from 3.0V to
3.6V, 3.3V nominal.
• VDDPLLA pin: Powers the PLLA cell; voltage ranges from 0.9V to 1.1V.
• VDDOSC pin: Powers the Main Oscillator cells; voltage ranges from 1.65V to 3.6V
• VDDANA pin: Powers the Analog to Digital Converter; voltage ranges from 3.0V to 3.6V, 3.3V
nominal.
Ground pins GND are common to VDDIOM0, VDDIOM1, VDDIOP0, VDDIOP1 and VDDIOP2
power supplies. Separated ground pins are provided for VDDUTMIC, VDDUTMII, VDDBU,
VDDOSC, VDDPLLA, VDDPLLUTMI and VDDANA. These ground pins are respectively
GNDUTMIC, GNDUTMII, GNDBU, GNDOSC, GNDPLLA, GNDPLLUTMI and GNDANA.
14
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
6. Memories
Figure 6-1.
AT91SAM9G45 Memory Mapping
0x00000000
Address Memory Space
Internal Memories
0x00000000
Internal Memories
0xFFFF0000
System Controller
Boot Memory
Reserved
0x00100000
0xFFFFE400
ITCM
0x10000000
DDRSDRC1
0x00200000
0xFFFFE600
DTCM
EBI Chip Select 0
DDRSDRC0
0x00300000
0xFFFFE800
SRAM
0x20000000
SMC
0x00400000
EBI Chip Select 1
DDR2-LPDDR-SDRAM
0x30000000
0xFFFFEA00
ROM
MATRIX
0x00500000
0xFFFFEC00
LCDC
23
0x00600000
DMAC
0xFFFFEE00
UDPHS (DMA)
EBI Chip Select 2
0x00700000
0xFFFFF000
UHP OHCI
0x40000000
0x50000000
0xFFFFF200
UHP EHCI
PIOA
0xFFFFF400
0x00900000
Reserved
PIOB
0xFFFFF600
0x00A00000
EBI Chip Select 4
Compact Flash Slot 0
0x60000000
Undefined (Abort)
0x70000000
Undefined (Abort)
0xFFFFF800
Internal Peripherals
0xFFFFFC00
UDPHS
+0x40
+0x80
TC0
TC0
TC0
TC0
TC1
+18
0xFFF88000
TWI1
0xFFF8C000
11
12
13
USART0
7
0xFFF90000
block peripheral
ID
(+ : wired-or)
+0x10
+0x20
0xFFF94000
USART2
0xFFF98000
USART3
0xFFF9C000
SSC0
0xFFFA0000
SSC1
0xFFFA4000
SPI0
0xFFFA8000
SPI1
0xFFFAC000
AC97C
0xFFFB0000
TSADCC
0xFFFB4000
ISI
0xFFFB8000
PWM
0xFFFBC000
EMAC
0xFFFC0000
+0x40
+0x50
+0x60
SYSC
RSTC
SYSC
SHDWC
SYSC
RTT
SYSC
PIT
SYSC
WDT
SYSC
SCKCR
SYSC
GPBR
+0x70
SYSC
USART1
offset
0xFFFFFD00
+0x30
0xFFF80000
TWI0
0xFFFFFFFF
27
+18
8
0xFFFFFDB0
9
0xFFFFFDC0
10
0;31
2
3
4
+5
+5
PMC
TC2
0xFFF84000
Internal Peripherals
0xFFFFFA00
PIOE
0xFFF78000
HSMCI0
0xF0000000
PIOD
Reserved
0xFFF7C000
DDR2-LPDDR
Chip Select
PIOC
0x0FFFFFFF
0xF0000000
EBI Chip Select 5
Compact Flash Slot 1
0x80000000
AIC
0x00800000
EBI Chip Select 3
NANDFlash
21
DBGU
1
1
1
1
1
1
1
Reserved
RTC
Reserved
0xFFFFFFFF
16
17
14
15
24
20
26
19
25
Reserved
0xFFFC4000
Reserved
0xFFFC8000
Reserved
0xFFFCC000
TRNG
6
0xFFFD0000
HSMCI1
0xFFFD4000
+0x40
+0x80
TC1
TC3
TC1
TC4
TC1
TC5
29
0xFFFD8000
Reserved
0xFFFFC000
System controller
0xFFFFFFFF
15
6438D–ATARM–13-Oct-09
6.1
Memory Mapping
A first level of address decoding is performed by the AHB Bus Matrix, i.e., the implementation of
the Advanced High performance Bus (AHB) for its Master and Slave interfaces with additional
features.
Decoding breaks up the 4 Gbytes of address space into 16 banks of 256 Mbytes. The banks 1 to
6 are directed to the EBI that associates these banks to the external chip selects NCS0 to
NCS5.
The bank 7 is directed to the DDRSDRC0 that associates this bank to DDR_NCS chip select
and so dedicated to the 4-port DDR2/ LPDDR controller.
The bank 0 is reserved for the addressing of the internal memories, and a second level of
decoding provides 1 Mbyte of internal memory area. The bank 15 is reserved for the peripherals
and provides access to the Advanced Peripheral Bus (APB).
Other areas are unused and performing an access within them provides an abort to the master
requesting such an access.
6.2
6.2.1
Embedded Memories
Internal SRAM
The AT91SAM9G45 product embeds a total of 64 Kbytes high-speed SRAM split in 4 blocks of
16 KBytes connected to one slave of the matrix. After reset and until the Remap Command is
performed, the four SRAM blocks are contiguous and only accessible at address 0x00300000.
After Remap, the SRAM also becomes available at address 0x0.
Figure 6-2.
Internal SRAM Reset
RAM
RAM
Remap
64K
0x00300000
64K
0x00000000
The AT91SAM9G45 device embeds two memory features. The processor Tightly Coupled Memory Interface (TCM) that allows the processor to access the memory up to processor speed
(PCK) and the interface on the AHB side allowing masters to access the memory at AHB speed
(MCK).
A wait state is necessary to access the TCM at 400 MHz. Setting the bit NWS_TCM in the bus
Matrix TCM Configuration Register of the matrix inserts a wait state on the ITCM and DTCM
accesses.
16
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
6.2.2
TCM Interface
On the processor side, this Internal SRAM can be allocated to two areas.
• Internal SRAM A is the ARM926EJ-S Instruction TCM. The user can map this SRAM block
anywhere in the ARM926 instruction memory space using CP15 instructions and the TCR
configuration register located in the Chip Configuration User Interface. This SRAM block is
also accessible by the ARM926 Masters and by the AHB Masters through the AHB bus
• Internal SRAM B is the ARM926EJ-S Data TCM. The user can map this SRAM block
anywhere in the ARM926 data memory space using CP15 instructions. This SRAM block is
also accessible by the ARM926 Data Master and by the AHB Masters through the AHB bus.
• Internal SRAM C is only accessible by all the AHB Masters. After reset and until the Remap
Command is performed, this SRAM block is accessible through the AHB bus at address
0x0030 0000 by all the AHB Masters. After Remap, this SRAM block also becomes
accessible through the AHB bus at address 0x0 by the ARM926 Instruction and the ARM926
Data Masters.
Within the 64 Kbyte SRAM size available, the amount of memory assigned to each block is software programmable according to Table 6-1.
Table 6-1.
ITCM and DTCM Memory Configuration
SRAM A ITCM size (KBytes)
seen at 0x100000 through AHB
SRAM B DTCM size (KBytes)
seen at 0x200000 through AHB
SRAM C (KBytes)
seen at 0x300000 through AHB
0
0
64
0
64
0
32
32
0
6.2.3
Internal ROM
The AT91SAM9G45 embeds an Internal ROM, which contains the boot ROM and SAM-BA®
program.
At any time, the ROM is mapped at address 0x0040 0000. It is also accessible at address 0x0
(BMS =1) after the reset and before the Remap Command.
6.3
6.3.1
I/O Drive Selection and Delay Control
I/O Drive Selection
The aim of this control is to adapt the signal drive to the frequency. Two bits allow the user to
select High or Low drive for memories data/address/ctrl signals.
• Setting the bit [17], EBI_DRIVE, in the EBI_CSA register of the matrix allows to control the
drive of the EBI.
• Setting the bit [18], DDR_DRIVE, in the EBI_CSA register of the matrix allows to control the
drive of the DDR.
6.3.2
Delay Control
To avoid the simultaneous switching of all the I/Os, a delay can be inserted on the different EBI,
DDR2 and PIO lines.
17
6438D–ATARM–13-Oct-09
The control of these delays is the following:
• DDR2SDRC
DDR_D[15:0] controlled by 2 registers, DELAY1 and DELAY2, located in the DDR2SDRC user
interface
– DDR_D[0] <=> DELAY1[3:0],
– DDR_D[1] <=> DELAY1[7:4],...
– DDR_D[6] <=> DELAY1[27:24],
– DDR_D[7] <=> DELAY1[31:28]
– DDR_D[8] <=> DELAY2[3:0],
– DDR_D[9] <=> DELAY2[7:4],...,
– DDR_D[14] <=> DELAY2[27:24],
– DDR_D[15] <=> DELAY2[31:28]
DDR_A[13:0] controlled by 2 registers, DELAY3 and DELAY4, located in the DDR2SDRC user
interface
– DDR_A[0] <=> DELAY3[3:0],
– DDR_A[1] <=> DELAY3[7:4], ...,
– DDR_A[6] <=> DELAY3[27:24],
– DDR_A[7] <=> DELAY3[31:28]
– DDR_A[8] <=> DELAY4[3:0],
– DDR_A[9] <=> DELAY4[7:4], ...,
– DDR_A[12] <=> DELAY4[19:16],
– DDR_A[13] <=> DELAY4[23:20]
• EBI (DDR2SDRC\HSMC3\Nandflash)
D[15:0] controlled by 2 registers, DELAY1 and DELAY2, located in the HSMC3 user interface
– D[0] <=> DELAY1[3:0],
– D[1] <=> DELAY1[7:4],...,
– D[6] <=> DELAY1[27:24],
– D[7] <=> DELAY1[31:28]
– D[8] <=> DELAY2[3:0],
– D[9] <=> DELAY2[7:4],...,
– D[14] <=> DELAY2[27:24],
– D[15] <=> DELAY2[31:28]
D[31,16]on PIOC[31:16] controlled by 2 registers, DELAY3 and DELAY4, located in the
HSMC3 user interface
– D[16] <=> DELAY3[3:0],
– D[17] <=> DELAY3[7:4],...,
– D[22] <=> DELAY3[27:24],
– PC[23] <=> DELAY3[31:28]
18
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
– D[24] <=> DELAY4[3:0],
– D[25] <=> DELAY4[7:4],...,
– D[30] <=> DELAY4[27:24],
– D[31] <=> DELAY4[31:28]
A[25:0], controlled by 4 registers, DELAY5, DELAY6, DELAY7and DELAY8, located in the
HSMC3 user interface
– A[0] <=> DELAY5[3:0],
– A[1] <=> DELAY5[7:4],...,
– A[6] <=> DELAY5[27:24],
– A[7] <=> DELAY5[31:28]
– A[8] <=> DELAY6[3:0],
– A[9] <=> DELAY6[7:4],...,
– A[14] <=> DELAY6[27:24],
– A[15] <=> DELAY6[31:28]
– A[16] <=> DELAY7[3:0],
– A[17] <=> DELAY7[7:4],
– A[18] <=> DELAY7[11:8]
A25 on PC[12] and A[24:19] on PC[7:2]
– A19 <=> DELAY7[15:12],
– A20 <=> DELAY7[19:16],...,
– A23 <=> DELAY7[31:28],
– A24 <=> DELAY8[3:0],
– A25 <=> DELAY8[7:4]
• PIOA User interface
The delay can only be inserted on the HSMCI0 and HSMCI1 I/O lines, so on PA[9:2] and
PA[30:23]. The delay is controlled by 2 registers, DELAY1 and DELAY2, located in the PIOA
user interface.
– PA[2] <=> DELAY1[3:0],
– PA[3] <=> DELAY1[7:4],...,
– PA[8] <=> DELAY1[27:24],
– PA[9] <=> DELAY1[31:28]
– PA[23] <=> DELAY2[3:0],
– PA[24] <=> DELAY2[7:4],...,
– PA[29] <=> DELAY2[27:24],
– PA[30] <=> DELAY2[31:28]
7. System Controller
The System Controller is a set of peripherals that allows handling of key elements of the system,
such as power, resets, clocks, time, interrupts, watchdog, etc.
19
6438D–ATARM–13-Oct-09
The System Controller User Interface also embeds the registers that configure the Matrix and a
set of registers for the chip configuration. The chip configuration registers configure the EBI chip
select assignment and voltage range for external memories.
7.1
System Controller Mapping
The System Controller’s peripherals are all mapped within the highest 16 KBytes of address
space, between addresses 0xFFFF E800 and 0xFFFF FFFF.
However, all the registers of the System Controller are mapped on the top of the address space.
All the registers of the System Controller can be addressed from a single pointer by using the
standard ARM instruction set, as the Load/Store instruction have an indexing mode of ±4 KB.
Figure 7-1 on page 21 shows the System Controller block diagram.
Figure 6-1 on page 15 shows the mapping of the User Interfaces of the System Controller
peripherals.
20
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
7.2
System Controller Block Diagram
Figure 7-1.
AT91SAM9G45 System Controller Block Diagram
System Controller
VDDCORE Powered
irq0-irq2
fiq
periph_irq[2..24]
nirq
nfiq
Advanced
Interrupt
Controller
pit_irq
rtt_irq
wdt_irq
dbgu_irq
pmc_irq
rstc_irq
int
ntrst
por_ntrst
MCK
periph_nreset
Debug
Unit
dbgu_irq
dbgu_txd
dbgu_rxd
MCK
debug
periph_nreset
proc_nreset
PCK
debug
Periodic
Interval
Timer
pit_irq
Watchdog
Timer
wdt_irq
jtag_nreset
SLCK
debug
idle
proc_nreset
ARM926EJ-S
Boundary Scan
TAP Controller
MCK
wdt_fault
WDRPROC
NRST
periph_nreset
Bus Matrix
rstc_irq
por_ntrst
jtag_nreset
VDDCORE
POR
Reset
Controller
periph_nreset
proc_nreset
backup_nreset
VDDBU
VDDBU
POR
VDDBU Powered
SLCK
UPLLCK
UHP48M
SLCK
backup_nreset
Real-Time
Clock
SLCK
backup_nreset
Real-Time
Timer
rtc_irq
rtc_alarm
UHP12M
rtt_irq
periph_nreset
rtt_alarm
periph_irq[25]
USB High Speed
Host Port
SLCK
SHDN
WKUP
backup_nreset
RC
OSC
XIN32
XOUT32
SLOW
CLOCK
OSC
rtt0_alarm
4 General-purpose
Backup Registers
XIN
periph_nreset
USB High Speed
Device Port
periph_irq[24]
SCKCR
SLCK
XOUT
UPLLCK
Shut-Down
Controller
int
12MHz
MAIN OSC
MAINCK
UPLL
UPLLCK
PLLA
PLLACK
Power
Management
Controller
periph_clk[2..30]
pck[0-1]
UHP48M
UHP12M
PCK
MCK
DDR sysclk
pmc_irq
idle
periph_clk[6..30]
periph_nreset
periph_nreset
periph_nreset
periph_clk[2..6]
dbgu_rxd
PA0-PA31
PB0-PB31
PC0-PC31
PD0-PD31
PIO
Controllers
periph_irq[2..6]
irq
fiq
dbgu_txd
Embedded
Peripherals
periph_irq[6..30]
in
out
enable
PE0-PE31
21
6438D–ATARM–13-Oct-09
7.3
Chip Identification
The AT91SAM9G45 Chip ID is defined in the Debug Unit Chip ID Register and Debug Unit Chip
ID Extension Register.
• Chip ID: 0x819B05A2
• Ext ID: 0x00000004
• JTAG ID: 05B2_703F
• ARM926 TAP ID: 0x0792603F
7.4
Backup Section
The AT91SAM9G45 features a Backup Section that embeds:
• RC Oscillator
• Slow Clock Oscillator
• SCKR register
• RTT
• RTC
• Shutdown Controller
• 4 backup registers
• A part of RSTC
This section is powered by the VDDBU rail.
22
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
8. Peripherals
8.1
Peripheral Mapping
As shown in Figure 6-1, the Peripherals are mapped in the upper 256 Mbytes of the address
space between the addresses 0xFFF7 8000 and 0xFFFC FFFF.
Each User Peripheral is allocated 16K bytes of address space.
8.2
Peripheral Identifiers
Table 8-1 defines the Peripheral Identifiers of the AT91SAM9G45. A peripheral identifier is
required for the control of the peripheral interrupt with the Advanced Interrupt Controller and for
the control of the peripheral clock with the Power Management Controller.
Table 8-1.
AT91SAM9G45 Peripheral Identifiers
Peripheral ID
Peripheral Mnemonic
Peripheral Name
0
AIC
1
SYSC
System Controller Interrupt
2
PIOA
Parallel I/O Controller A,
3
PIOB
Parallel I/O Controller B
4
PIOC
Parallel I/O Controller C
5
PIOD/PIOE
6
RNG
True Random Number Generator
7
US0
USART 0
8
US1
USART 1
Advanced Interrupt Controller
External Interrupt
FIQ
Parallel I/O Controller D/E
9
US2
USART 2
10
US3
USART 3
11
MCI0
High Speed Multimedia Card Interface 0
12
TWI0
Two-Wire Interface 0
13
TWI1
Two-Wire Interface 1
14
SPI0
Serial Peripheral Interface
15
SPI1
Serial Peripheral Interface
16
SSC0
Synchronous Serial Controller 0
17
SSC1
Synchronous Serial Controller 1
18
TC0..TC5
19
PWMC
20
TSADCC
21
DMA
22
UHPHS
23
LCDC
LCD Controller
24
AC97
AC97 Controller
25
EMAC
Ethernet MAC
26
ISI
Image Sensor Interface
27
UDPHS
USB Device High Speed
29
MCI1
30
Reserved
31
AIC
Timer Counter 0,1,2,3,4,5
Pulse Width Modulation Controller
Touch Screen ADC Controller
DMA Controller
USB Host High Speed
High Speed Multimedia Card Interface 1
Advanced Interrupt Controller
IRQ
23
6438D–ATARM–13-Oct-09
8.3
8.3.1
Peripheral Interrupts and Clock Control
System Interrupt
The System Interrupt in Source 1 is the wired-OR of the interrupt signals coming from:
• the DDR2/LPDDR Controller
• the Debug Unit
• the Periodic Interval Timer
• the Real-Time Timer
• the Real-Time Clock
• the Watchdog Timer
• the Reset Controller
• the Power Management Controller
The clock of these peripherals cannot be deactivated and Peripheral ID 1 can only be used
within the Advanced Interrupt Controller.
8.3.2
8.4
External Interrupts
All external interrupt signals, i.e., the Fast Interrupt signal FIQ or the Interrupt signal IRQ, use a
dedicated Peripheral ID. However, there is no clock control associated with these peripheral IDs.
Peripheral Signals Multiplexing on I/O Lines
The AT91SAM9G45 features 5 PIO controllers, PIOA, PIOB, PIOC, PIOD and PIOE, which multiplexes the I/O lines of the peripheral set.
Each PIO Controller controls up to 32 lines. Each line can be assigned to one of two peripheral
functions, A or B. The multiplexing tables in the following paragraphs define how the I/O lines of
the peripherals A and B are multiplexed on the PIO Controllers. The two columns “Function” and
“Comments” have been inserted in this table for the user’s own comments; they may be used to
track how pins are defined in an application.
Note that some peripheral function which are output only, might be duplicated within the both
tables.
The column “Reset State” indicates whether the PIO Line resets in I/O mode or in peripheral
mode. If I/O is mentioned, the PIO Line resets in input with the pull-up enabled, so that the
device is maintained in a static state as soon as the reset is released. As a result, the bit corresponding to the PIO Line in the register PIO_PSR (Peripheral Status Register) resets low.
If a signal name is mentioned in the “Reset State” column, the PIO Line is assigned to this function and the corresponding bit in PIO_PSR resets high. This is the case of pins controlling
memories, in particular the address lines, which require the pin to be driven as soon as the reset
is released. Note that the pull-up resistor is also enabled in this case.
To amend EMC, programmable delay has been inserted on PIO lines able to run at high speed.
24
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
8.4.1
Table 8-2.
PIO Controller A Multiplexing
Multiplexing on PIO Controller A (PIOA)
I/O Line
Peripheral A
Peripheral B
Reset
State
Power
Supply
PA0
MCI0_CK
TCLK3
I/O
VDDIOP0
PA1
MCI0_CDA
TIOA3
I/O
VDDIOP0
PA2
MCI0_DA0
TIOB3
I/O
VDDIOP0
PA3
MCI0_DA1
TCKL4
I/O
VDDIOP0
PA4
MCI0_DA2
TIOA4
I/O
VDDIOP0
PA5
MCI0_DA3
TIOB4
I/O
VDDIOP0
PA6
MCI0_DA4
ETX2
I/O
VDDIOP0
PA7
MCI0_DA5
ETX3
I/O
VDDIOP0
PA8
MCI0_DA6
ERX2
I/O
VDDIOP0
PA9
MCI0_DA7
ERX3
I/O
VDDIOP0
PA10
ETX0
I/O
VDDIOP0
PA11
ETX1
I/O
VDDIOP0
PA12
ERX0
I/O
VDDIOP0
PA13
ERX1
I/O
VDDIOP0
PA14
ETXEN
I/O
VDDIOP0
PA15
ERXDV
I/O
VDDIOP0
PA16
ERXER
I/O
VDDIOP0
PA17
ETXCK
I/O
VDDIOP0
PA18
EMDC
I/O
VDDIOP0
PA19
EMDIO
I/O
VDDIOP0
PA20
TWD0
I/O
VDDIOP0
PA21
TWCK0
I/O
VDDIOP0
PA22
MCI1_CDA
SCK3
I/O
VDDIOP0
PA23
MCI1_DA0
RTS3
I/O
VDDIOP0
PA24
MCI1_DA1
CTS3
I/O
VDDIOP0
PA25
MCI1_DA2
PWM3
I/O
VDDIOP0
PA26
MCI1_DA3
TIOB2
I/O
VDDIOP0
PA27
MCI1_DA4
ETXER
I/O
VDDIOP0
PA28
MCI1_DA5
ERXCK
I/O
VDDIOP0
PA29
MCI1_DA6
ECRS
I/O
VDDIOP0
PA30
MCI1_DA7
ECOL
I/O
VDDIOP0
PA31
MCI1_CK
PCK0
I/O
VDDIOP0
Function
Comments
25
6438D–ATARM–13-Oct-09
8.4.2
PIO Controller B Multiplexing
Table 8-3.
Multiplexing on PIO Controller B (PIOB)
Reset
State
Power
Supply
SPI0_MISO
I/O
VDDIOP0
PB1
SPI0_MOSI
I/O
VDDIOP0
PB2
SPI0_SPCK
I/O
VDDIOP0
PB3
SPI0_NPCS0
I/O
VDDIOP0
PB4
TXD1
I/O
VDDIOP0
PB5
RXD1
I/O
VDDIOP0
PB6
TXD2
I/O
VDDIOP0
PB7
RXD2
I/O
VDDIOP0
PB8
TXD3
ISI_D8
I/O
VDDIOP2
PB9
RXD3
ISI_D9
I/O
VDDIOP2
PB10
TWD1
ISI_D10
I/O
VDDIOP2
PB11
TWCK1
ISI_D11
I/O
VDDIOP2
PB12
DRXD
I/O
VDDIOP0
PB13
DTXD
I/O
VDDIOP0
PB14
SPI1_MISO
I/O
VDDIOP0
PB15
SPI1_MOSI
CTS0
I/O
VDDIOP0
PB16
SPI1_SPCK
SCK0
I/O
VDDIOP0
PB17
SPI1_NPCS0
RTS0
I/O
VDDIOP0
PB18
RXD0
SPI0_NPCS1
I/O
VDDIOP0
PB19
TXD0
SPI0_NPCS2
I/O
VDDIOP0
PB20
ISI_D0
I/O
VDDIOP2
PB21
ISI_D1
I/O
VDDIOP2
PB22
ISI_D2
I/O
VDDIOP2
PB23
ISI_D3
I/O
VDDIOP2
PB24
ISI_D4
I/O
VDDIOP2
PB25
ISI_D5
I/O
VDDIOP2
PB26
ISI_D6
I/O
VDDIOP2
PB27
ISI_D7
I/O
VDDIOP2
PB28
ISI_PCK
I/O
VDDIOP2
PB29
ISI_VSYNC
I/O
VDDIOP2
PB30
ISI_HSYNC
I/O
VDDIOP2
PB31
ISI_MCK
I/O
VDDIOP2
I/O Line
Peripheral A
PB0
26
Peripheral B
PCK1
Function
Comments
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
8.4.3
Table 8-4.
PIO Controller C Multiplexing
Multiplexing on PIO Controller C (PIOC)
Reset
State
Power
Supply
DQM2
DQM2
VDDIOM1
PC1
DQM3
DQM3
VDDIOM1
PC2
A19
A19
VDDIOM1
PC3
A20
A20
VDDIOM1
PC4
A21/NANDALE
A21
VDDIOM1
PC5
A22/NANDCLE
A22
VDDIOM1
PC6
A23
A23
VDDIOM1
PC7
A24
A24
VDDIOM1
PC8
CFCE1
I/O
VDDIOM1
PC9
CFCE2
RTS2
I/O
VDDIOM1
PC10
NCS4/CFCS0
TCLK2
I/O
VDDIOM1
PC11
NCS5/CFCS1
CTS2
I/O
VDDIOM1
PC12
A25/CFRNW
A25
VDDIOM1
PC13
NCS2
I/O
VDDIOM1
PC14
NCS3/NANDCS
I/O
VDDIOM1
PC15
NWAIT
I/O
VDDIOM1
PC16
D16
I/O
VDDIOM1
PC17
D17
I/O
VDDIOM1
PC18
D18
I/O
VDDIOM1
PC19
D19
I/O
VDDIOM1
PC20
D20
I/O
VDDIOM1
PC21
D21
I/O
VDDIOM1
PC22
D22
I/O
VDDIOM1
PC23
D23
I/O
VDDIOM1
PC24
D24
I/O
VDDIOM1
PC25
D25
I/O
VDDIOM1
PC26
D26
I/O
VDDIOM1
PC27
D27
I/O
VDDIOM1
PC28
D28
I/O
VDDIOM1
PC29
D29
I/O
VDDIOM1
PC30
D30
I/O
VDDIOM1
PC31
D31
I/O
VDDIOM1
I/O Line
Peripheral A
PC0
Peripheral B
Function
Comments
27
6438D–ATARM–13-Oct-09
8.4.4
PIO Controller D Multiplexing
Table 8-5.
Multiplexing on PIO Controller D (PIOD)
I/O Line
Peripheral A
Peripheral B
Reset
State
Power
Supply
PD0
TK0
PWM3
I/O
VDDIOP0
PD1
TF0
I/O
VDDIOP0
PD2
TD0
I/O
VDDIOP0
PD3
RD0
I/O
VDDIOP0
PD4
RK0
I/O
VDDIOP0
PD5
RF0
I/O
VDDIOP0
PD6
AC97RX
I/O
VDDIOP0
PD7
AC97TX
TIOA5
I/O
VDDIOP0
PD8
AC97FS
TIOB5
I/O
VDDIOP0
PD9
AC97CK
TCLK5
I/O
VDDIOP0
PD10
TD1
I/O
VDDIOP0
PD11
RD1
I/O
VDDIOP0
PD12
TK1
I/O
VDDIOP0
PD13
RK1
I/O
VDDIOP0
PD14
TF1
I/O
VDDIOP0
PD15
RF1
I/O
VDDIOP0
PD16
RTS1
I/O
VDDIOP0
PD17
CTS1
I/O
VDDIOP0
PD18
SPI1_NPCS2
IRQ
I/O
VDDIOP0
PD19
SPI1_NPCS3
FIQ
I/O
VDDIOP0
PD20
TIOA0
I/O
VDDANA
TSAD0
PD21
TIOA1
I/O
VDDANA
TSAD1
PD22
TIOA2
I/O
VDDANA
TSAD2
PD23
TCLK0
I/O
VDDANA
TSAD3
PD24
SPI0_NPCS1
PWM0
I/O
VDDANA
GPAD4
PD25
SPI0_NPCS2
PWM1
I/O
VDDANA
GPAD5
PD26
PCK0
PWM2
I/O
VDDANA
GPAD6
PD27
PCK1
SPI0_NPCS3
I/O
VDDANA
GPAD7
PD28
TSADTRG
SPI1_NPCS1
I/O
VDDIOP0
PD29
TCLK1
SCK1
I/O
VDDIOP0
PD30
TIOB0
SCK2
I/O
VDDIOP0
PD31
TIOB1
PWM1
I/O
VDDIOP0
28
PCK0
Function
Comments
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
8.4.5
Table 8-6.
PIO Controller E Multiplexing
Multiplexing on PIO Controller E (PIOE)
I/O Line
Peripheral A
Peripheral B
Reset
State
Power
Supply
PE0
LCDPWR
PCK0
I/O
VDDIOP1
PE1
LCDMOD
I/O
VDDIOP1
PE2
LCDCC
I/O
VDDIOP1
PE3
LCDVSYNC
I/O
VDDIOP1
PE4
LCDHSYNC
I/O
VDDIOP1
PE5
LCDDOTCK
I/O
VDDIOP1
PE6
LCDDEN
I/O
VDDIOP1
PE7
LCDD0
LCDD2
I/O
VDDIOP1
PE8
LCDD1
LCDD3
I/O
VDDIOP1
PE9
LCDD2
LCDD4
I/O
VDDIOP1
PE10
LCDD3
LCDD5
I/O
VDDIOP1
PE11
LCDD4
LCDD6
I/O
VDDIOP1
PE12
LCDD5
LCDD7
I/O
VDDIOP1
PE13
LCDD6
LCDD10
I/O
VDDIOP1
PE14
LCDD7
LCDD11
I/O
VDDIOP1
PE15
LCDD8
LCDD12
I/O
VDDIOP1
PE16
LCDD9
LCDD13
I/O
VDDIOP1
PE17
LCDD10
LCDD14
I/O
VDDIOP1
PE18
LCDD11
LCDD15
I/O
VDDIOP1
PE19
LCDD12
LCDD18
I/O
VDDIOP1
PE20
LCDD13
LCDD19
I/O
VDDIOP1
PE21
LCDD14
LCDD20
I/O
VDDIOP1
PE22
LCDD15
LCDD21
I/O
VDDIOP1
PE23
LCDD16
LCDD22
I/O
VDDIOP1
PE24
LCDD17
LCDD23
I/O
VDDIOP1
PE25
LCDD18
I/O
VDDIOP1
PE26
LCDD19
I/O
VDDIOP1
PE27
LCDD20
I/O
VDDIOP1
PE28
LCDD21
I/O
VDDIOP1
PE29
LCDD22
I/O
VDDIOP1
PE30
LCDD23
I/O
VDDIOP1
PE31
PWM2
I/O
VDDIOP1
PCK1
Function
Comments
29
6438D–ATARM–13-Oct-09
30
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
9. ARM926EJ-S Processor Overview
9.1
Description
The ARM926EJ-S™ processor is a member of the ARM9™ family of general-purpose microprocessors. The ARM926EJ-S implements ARM architecture version 5TEJ and is targeted at multitasking applications where full memory management, high performance, low die size and low
power are all important features.
The ARM926EJ-S processor supports the 32-bit ARM and 16-bit THUMB instruction sets,
enabling the user to trade off between high performance and high code density. It also supports
8-bit Java instruction set and includes features for efficient execution of Java bytecode, providing a Java performance similar to a JIT (Just-In-Time compilers), for the next generation of Javapowered wireless and embedded devices. It includes an enhanced multiplier design for
improved DSP performance.
The ARM926EJ-S processor supports the ARM debug architecture and includes logic to assist
in both hardware and software debug.
The ARM926EJ-S provides a complete high performance processor subsystem, including:
• an ARM9EJ-S integer core
• a Memory Management Unit (MMU)
• separate instruction and data AMBA AHB bus interfaces
• separate instruction and data TCM interfaces
31
6438D–ATARM–13-Oct-09
9.2
Embedded Characteristics
• RISC Processor Based on ARM v5TEJ Architecture with Jazelle technology for Java
acceleration
• Two Instruction Sets
– ARM High-performance 32-bit Instruction Set
– Thumb High Code Density 16-bit Instruction Set
• DSP Instruction Extensions
• 5-Stage Pipeline Architecture:
– Instruction Fetch (F)
– Instruction Decode (D)
– Execute (E)
– Data Memory (M)
– Register Write (W)
• 32-KByte Data Cache, 32-KByte Instruction Cache
– Virtually-addressed 4-way Associative Cache
– Eight words per line
– Write-through and Write-back Operation
– Pseudo-random or Round-robin Replacement
• Write Buffer
– Main Write Buffer with 16-word Data Buffer and 4-address Buffer
– DCache Write-back Buffer with 8-word Entries and a Single Address Entry
– Software Control Drain
• Standard ARM v4 and v5 Memory Management Unit (MMU)
– Access Permission for Sections
– Access Permission for large pages and small pages can be specified separately for
each quarter of the page
– 16 embedded domains
• Bus Interface Unit (BIU)
– Arbitrates and Schedules AHB Requests
– Separate Masters for both instruction and data access providing complete Matrix
system flexibility
– Separate Address and Data Buses for both the 32-bit instruction interface and the
32-bit data interface
– On Address and Data Buses, data can be 8-bit (Bytes), 16-bit (Half-words) or 32-bit
(Words)
• TCM Interface
32
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
9.3
Block Diagram
Figure 9-1.
ARM926EJ-S Internal Functional Block Diagram
CP15 System
Configuration
Coprocessor
External Coprocessors
ETM9
External
Coprocessor
Interface
Trace Port
Interface
Write Data
ARM9EJ-S
Processor Core
Instruction
Fetches
Read
Data
Data
Address
Instruction
Address
MMU
DTCM
Interface
Data TLB
Instruction
TLB
ITCM
Interface
Data TCM
Instruction TCM
Instruction
Address
Data
Address
Data Cache
AHB Interface
and
Write Buffer
Instruction
Cache
AMBA AHB
33
6438D–ATARM–13-Oct-09
9.4
9.4.1
ARM9EJ-S Processor
ARM9EJ-S Operating States
The ARM9EJ-S processor can operate in three different states, each with a specific instruction
set:
• ARM state: 32-bit, word-aligned ARM instructions.
• THUMB state: 16-bit, halfword-aligned Thumb instructions.
• Jazelle state: variable length, byte-aligned Jazelle instructions.
In Jazelle state, all instruction Fetches are in words.
9.4.2
Switching State
The operating state of the ARM9EJ-S core can be switched between:
• ARM state and THUMB state using the BX and BLX instructions, and loads to the PC
• ARM state and Jazelle state using the BXJ instruction
All exceptions are entered, handled and exited in ARM state. If an exception occurs in Thumb or
Jazelle states, the processor reverts to ARM state. The transition back to Thumb or Jazelle
states occurs automatically on return from the exception handler.
9.4.3
Instruction Pipelines
The ARM9EJ-S core uses two kinds of pipelines to increase the speed of the flow of instructions
to the processor.
A five-stage (five clock cycles) pipeline is used for ARM and Thumb states. It consists of Fetch,
Decode, Execute, Memory and Writeback stages.
A six-stage (six clock cycles) pipeline is used for Jazelle state It consists of Fetch,
Jazelle/Decode (two clock cycles), Execute, Memory and Writeback stages.
9.4.4
Memory Access
The ARM9EJ-S core supports byte (8-bit), half-word (16-bit) and word (32-bit) access. Words
must be aligned to four-byte boundaries, half-words must be aligned to two-byte boundaries and
bytes can be placed on any byte boundary.
Because of the nature of the pipelines, it is possible for a value to be required for use before it
has been placed in the register bank by the actions of an earlier instruction. The ARM9EJ-S control logic automatically detects these cases and stalls the core or forward data.
9.4.5
Jazelle Technology
The Jazelle technology enables direct and efficient execution of Java byte codes on ARM processors, providing high performance for the next generation of Java-powered wireless and
embedded devices.
The new Java feature of ARM9EJ-S can be described as a hardware emulation of a JVM (Java
Virtual Machine). Java mode will appear as another state: instead of executing ARM or Thumb
instructions, it executes Java byte codes. The Java byte code decoder logic implemented in
ARM9EJ-S decodes 95% of executed byte codes and turns them into ARM instructions without
any overhead, while less frequently used byte codes are broken down into optimized sequences
of ARM instructions. The hardware/software split is invisible to the programmer, invisible to the
application and invisible to the operating system. All existing ARM registers are re-used in
Jazelle state and all registers then have particular functions in this mode.
34
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Minimum interrupt latency is maintained across both ARM state and Java state. Since byte
codes execution can be restarted, an interrupt automatically triggers the core to switch from
Java state to ARM state for the execution of the interrupt handler. This means that no special
provision has to be made for handling interrupts while executing byte codes, whether in hardware or in software.
9.4.6
ARM9EJ-S Operating Modes
In all states, there are seven operation modes:
• User mode is the usual ARM program execution state. It is used for executing most
application programs
• Fast Interrupt (FIQ) mode is used for handling fast interrupts. It is suitable for high-speed data
transfer or channel process
• Interrupt (IRQ) mode is used for general-purpose interrupt handling
• Supervisor mode is a protected mode for the operating system
• Abort mode is entered after a data or instruction prefetch abort
• System mode is a privileged user mode for the operating system
• Undefined mode is entered when an undefined instruction exception occurs
Mode changes may be made under software control, or may be brought about by external interrupts or exception processing. Most application programs execute in User Mode. The non-user
modes, known as privileged modes, are entered in order to service interrupts or exceptions or to
access protected resources.
9.4.7
ARM9EJ-S Registers
The ARM9EJ-S core has a total of 37 registers.
• 31 general-purpose 32-bit registers
• 6 32-bit status registers
Table 9-1 shows all the registers in all modes.
Table 9-1.
ARM9TDMI Modes and Registers Layout
User and
System Mode
Supervisor
Mode
Abort Mode
Undefined
Mode
Interrupt
Mode
Fast Interrupt
Mode
R0
R0
R0
R0
R0
R0
R1
R1
R1
R1
R1
R1
R2
R2
R2
R2
R2
R2
R3
R3
R3
R3
R3
R3
R4
R4
R4
R4
R4
R4
R5
R5
R5
R5
R5
R5
R6
R6
R6
R6
R6
R6
R7
R7
R7
R7
R7
R7
R8
R8
R8
R8
R8
R8_FIQ
R9
R9
R9
R9
R9
R9_FIQ
R10
R10
R10
R10
R10
R10_FIQ
R11
R11
R11
R11
R11
R11_FIQ
35
6438D–ATARM–13-Oct-09
Table 9-1.
ARM9TDMI Modes and Registers Layout
User and
System Mode
Supervisor
Mode
Abort Mode
Undefined
Mode
Interrupt
Mode
Fast Interrupt
Mode
R12
R12
R12
R12
R12
R12_FIQ
R13
R13_SVC
R13_ABORT
R13_UNDEF
R13_IRQ
R13_FIQ
R14
R14_SVC
R14_ABORT
R14_UNDEF
R14_IRQ
R14_FIQ
PC
PC
PC
PC
PC
PC
CPSR
CPSR
CPSR
CPSR
CPSR
CPSR
SPSR_SVC
SPSR_ABOR
T
SPSR_UNDE
F
SPSR_IRQ
SPSR_FIQ
Mode-specific banked registers
The ARM state register set contains 16 directly-accessible registers, r0 to r15, and an additional
register, the Current Program Status Register (CPSR). Registers r0 to r13 are general-purpose
registers used to hold either data or address values. Register r14 is used as a Link register that
holds a value (return address) of r15 when BL or BLX is executed. Register r15 is used as a program counter (PC), whereas the Current Program Status Register (CPSR) contains condition
code flags and the current mode bits.
In privileged modes (FIQ, Supervisor, Abort, IRQ, Undefined), mode-specific banked registers
(r8 to r14 in FIQ mode or r13 to r14 in the other modes) become available. The corresponding
banked registers r14_fiq, r14_svc, r14_abt, r14_irq, r14_und are similarly used to hold the values (return address for each mode) of r15 (PC) when interrupts and exceptions arise, or when
BL or BLX instructions are executed within interrupt or exception routines. There is another register called Saved Program Status Register (SPSR) that becomes available in privileged modes
instead of CPSR. This register contains condition code flags and the current mode bits saved as
a result of the exception that caused entry to the current (privileged) mode.
In all modes and due to a software agreement, register r13 is used as stack pointer.
The use and the function of all the registers described above should obey ARM Procedure Call
Standard (APCS) which defines:
• constraints on the use of registers
• stack conventions
• argument passing and result return
For more details, refer to ARM Software Development Kit.
The Thumb state register set is a subset of the ARM state set. The programmer has direct
access to:
• Eight general-purpose registers r0-r7
• Stack pointer, SP
• Link register, LR (ARM r14)
• PC
36
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
• CPSR
There are banked registers SPs, LRs and SPSRs for each privileged mode (for more details see
the ARM9EJ-S Technical Reference Manual, revision r1p2 page 2-12).
9.4.7.1
Status Registers
The ARM9EJ-S core contains one CPSR, and five SPSRs for exception handlers to use. The
program status registers:
• hold information about the most recently performed ALU operation
• control the enabling and disabling of interrupts
• set the processor operation mode
Figure 9-2.
Status Register Format
31 30 29 28 27
24
N Z C V Q
J
7 6 5
Reserved
I F T
Jazelle state bit
Reserved
Sticky Overflow
Overflow
Carry/Borrow/Extend
Zero
Negative/Less than
0
Mode
Mode bits
Thumb state bit
FIQ disable
IRQ disable
Figure 9-2 shows the status register format, where:
• N: Negative, Z: Zero, C: Carry, and V: Overflow are the four ALU flags
• The Sticky Overflow (Q) flag can be set by certain multiply and fractional arithmetic
instructions like QADD, QDADD, QSUB, QDSUB, SMLAxy, and SMLAWy needed to achieve
DSP operations.
The Q flag is sticky in that, when set by an instruction, it remains set until explicitly cleared by
an MSR instruction writing to the CPSR. Instructions cannot execute conditionally on the
status of the Q flag.
• The J bit in the CPSR indicates when the ARM9EJ-S core is in Jazelle state, where:
– J = 0: The processor is in ARM or Thumb state, depending on the T bit
– J = 1: The processor is in Jazelle state.
• Mode: five bits to encode the current processor mode
9.4.7.2
Exceptions
9.4.7.3
Exception Types and Priorities
The ARM9EJ-S supports five types of exceptions. Each type drives the ARM9EJ-S in a privileged mode. The types of exceptions are:
• Fast interrupt (FIQ)
• Normal interrupt (IRQ)
• Data and Prefetched aborts (Abort)
• Undefined instruction (Undefined)
• Software interrupt and Reset (Supervisor)
37
6438D–ATARM–13-Oct-09
When an exception occurs, the banked version of R14 and the SPSR for the exception mode
are used to save the state.
More than one exception can happen at a time, therefore the ARM9EJ-S takes the arisen exceptions according to the following priority order:
• Reset (highest priority)
• Data Abort
• FIQ
• IRQ
• Prefetch Abort
• BKPT, Undefined instruction, and Software Interrupt (SWI) (Lowest priority)
The BKPT, or Undefined instruction, and SWI exceptions are mutually exclusive.
Note that there is one exception in the priority scheme: when FIQs are enabled and a Data Abort
occurs at the same time as an FIQ, the ARM9EJ-S core enters the Data Abort handler, and proceeds immediately to FIQ vector. A normal return from the FIQ causes the Data Abort handler to
resume execution. Data Aborts must have higher priority than FIQs to ensure that the transfer
error does not escape detection.
9.4.7.4
Exception Modes and Handling
Exceptions arise whenever the normal flow of a program must be halted temporarily, for example, to service an interrupt from a peripheral.
When handling an ARM exception, the ARM9EJ-S core performs the following operations:
1. Preserves the address of the next instruction in the appropriate Link Register that corresponds to the new mode that has been entered. When the exception entry is from:
– ARM and Jazelle states, the ARM9EJ-S copies the address of the next instruction
into LR (current PC(r15) + 4 or PC + 8 depending on the exception).
– THUMB state, the ARM9EJ-S writes the value of the PC into LR, offset by a value
(current PC + 2, PC + 4 or PC + 8 depending on the exception) that causes the
program to resume from the correct place on return.
2. Copies the CPSR into the appropriate SPSR.
3. Forces the CPSR mode bits to a value that depends on the exception.
4. Forces the PC to fetch the next instruction from the relevant exception vector.
The register r13 is also banked across exception modes to provide each exception handler with
private stack pointer.
The ARM9EJ-S can also set the interrupt disable flags to prevent otherwise unmanageable
nesting of exceptions.
When an exception has completed, the exception handler must move both the return value in
the banked LR minus an offset to the PC and the SPSR to the CPSR. The offset value varies
according to the type of exception. This action restores both PC and the CPSR.
The fast interrupt mode has seven private registers r8 to r14 (banked registers) to reduce or
remove the requirement for register saving which minimizes the overhead of context switching.
The Prefetch Abort is one of the aborts that indicates that the current memory access cannot be
completed. When a Prefetch Abort occurs, the ARM9EJ-S marks the prefetched instruction as
invalid, but does not take the exception until the instruction reaches the Execute stage in the
38
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
pipeline. If the instruction is not executed, for example because a branch occurs while it is in the
pipeline, the abort does not take place.
The breakpoint (BKPT) instruction is a new feature of ARM9EJ-S that is destined to solve the
problem of the Prefetch Abort. A breakpoint instruction operates as though the instruction
caused a Prefetch Abort.
A breakpoint instruction does not cause the ARM9EJ-S to take the Prefetch Abort exception until
the instruction reaches the Execute stage of the pipeline. If the instruction is not executed, for
example because a branch occurs while it is in the pipeline, the breakpoint does not take place.
9.4.8
ARM Instruction Set Overview
The ARM instruction set is divided into:
• Branch instructions
• Data processing instructions
• Status register transfer instructions
• Load and Store instructions
• Coprocessor instructions
• Exception-generating instructions
ARM instructions can be executed conditionally. Every instruction contains a 4-bit condition
code field (bits[31:28]).
For further details, see the ARM Technical Reference Manual.
Table 9-2 gives the ARM instruction mnemonic list.
Table 9-2.
Mnemonic
ARM Instruction Mnemonic List
Operation
Mnemonic
Operation
MOV
Move
MVN
Move Not
ADD
Add
ADC
Add with Carry
SUB
Subtract
SBC
Subtract with Carry
RSB
Reverse Subtract
RSC
Reverse Subtract with Carry
CMP
Compare
CMN
Compare Negated
TST
Test
TEQ
Test Equivalence
AND
Logical AND
BIC
Bit Clear
EOR
Logical Exclusive OR
ORR
Logical (inclusive) OR
MUL
Multiply
MLA
Multiply Accumulate
SMULL
Sign Long Multiply
UMULL
Unsigned Long Multiply
SMLAL
Signed Long Multiply Accumulate
UMLAL
Unsigned Long Multiply
Accumulate
MSR
B
BX
LDR
Move to Status Register
Branch
MRS
BL
Move From Status Register
Branch and Link
Branch and Exchange
SWI
Software Interrupt
Load Word
STR
Store Word
LDRSH
Load Signed Halfword
LDRSB
Load Signed Byte
39
6438D–ATARM–13-Oct-09
Table 9-2.
Mnemonic
Mnemonic
Operation
Load Half Word
STRH
Store Half Word
LDRB
Load Byte
STRB
Store Byte
Load Register Byte with
Translation
STRBT
Store Register Byte with
Translation
LDRT
Load Register with Translation
STRT
Store Register with Translation
LDM
Load Multiple
STM
Store Multiple
SWP
Swap Word
MCR
Move To Coprocessor
MRC
Move From Coprocessor
LDC
Load To Coprocessor
STC
Store From Coprocessor
CDP
Coprocessor Data Processing
SWPB
Swap Byte
New ARM Instruction Set
.
Table 9-3.
Mnemonic
BXJ
New ARM Instruction Mnemonic List
Operation
Mnemonic
Operation
Branch and exchange to Java
MRRC
Move double from coprocessor
Branch, Link and exchange
MCR2
Alternative move of ARM reg to
coprocessor
SMLAxy
Signed Multiply Accumulate 16 *
16 bit
MCRR
Move double to coprocessor
SMLAL
Signed Multiply Accumulate Long
CDP2
Alternative Coprocessor Data
Processing
SMLAWy
Signed Multiply Accumulate 32 *
16 bit
BKPT
Breakpoint
SMULxy
Signed Multiply 16 * 16 bit
PLD
SMULWy
Signed Multiply 32 * 16 bit
STRD
Store Double
Saturated Add
STC2
Alternative Store from
Coprocessor
Saturated Add with Double
LDRD
Load Double
Saturated subtract
LDC2
Alternative Load to Coprocessor
BLX (1)
QADD
QDADD
QSUB
QDSUB
Notes:
9.4.10
Operation
LDRH
LDRBT
9.4.9
ARM Instruction Mnemonic List (Continued)
Saturated Subtract with double
CLZ
Soft Preload, Memory prepare to
load from address
Count Leading Zeroes
1. A Thumb BLX contains two consecutive Thumb instructions, and takes four cycles.
Thumb Instruction Set Overview
The Thumb instruction set is a re-encoded subset of the ARM instruction set.
The Thumb instruction set is divided into:
• Branch instructions
• Data processing instructions
• Load and Store instructions
• Load and Store multiple instructions
40
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
• Exception-generating instruction
Table 5 shows the Thumb instruction set, for further details, see the ARM Technical Reference
Manual.
Table 9-4 gives the Thumb instruction mnemonic list.
Table 9-4.
Thumb Instruction Mnemonic List
Mnemonic
Operation
Mnemonic
Operation
MOV
Move
MVN
Move Not
ADD
Add
ADC
Add with Carry
SUB
Subtract
SBC
Subtract with Carry
CMP
Compare
CMN
Compare Negated
TST
Test
NEG
Negate
AND
Logical AND
BIC
Bit Clear
EOR
Logical Exclusive OR
ORR
Logical (inclusive) OR
LSL
Logical Shift Left
LSR
Logical Shift Right
ASR
Arithmetic Shift Right
ROR
Rotate Right
MUL
Multiply
BLX
Branch, Link, and Exchange
B
Branch
BL
Branch and Link
BX
Branch and Exchange
SWI
Software Interrupt
LDR
Load Word
STR
Store Word
LDRH
Load Half Word
STRH
Store Half Word
LDRB
Load Byte
STRB
Store Byte
LDRSH
Load Signed Halfword
LDRSB
Load Signed Byte
LDMIA
Load Multiple
STMIA
Store Multiple
PUSH
Push Register to stack
POP
Pop Register from stack
BCC
Conditional Branch
BKPT
Breakpoint
41
6438D–ATARM–13-Oct-09
9.5
CP15 Coprocessor
Coprocessor 15, or System Control Coprocessor CP15, is used to configure and control all the
items in the list below:
• ARM9EJ-S
• Caches (ICache, DCache and write buffer)
• TCM
• MMU
• Other system options
To control these features, CP15 provides 16 additional registers. See Table 9-5.
Table 9-5.
Register
0
Name
Read/Write
(1)
Read/Unpredictable
ID Code
0
(1)
Cache type
Read/Unpredictable
0
TCM status(1)
Read/Unpredictable
1
Control
Read/write
2
Translation Table Base
Read/write
3
Domain Access Control
Read/write
4
Reserved
None
5
Notes:
CP15 Registers
(1)
Read/write
Data fault Status
(1)
5
Instruction fault status
6
Fault Address
Read/write
7
Cache Operations
Read/Write
8
TLB operations
Unpredictable/Write
(2)
Read/write
9
cache lockdown
Read/write
9
TCM region
Read/write
10
TLB lockdown
Read/write
11
Reserved
None
12
Reserved
None
(1)
13
FCSE PID
Read/write
13
Context ID(1)
Read/Write
14
Reserved
None
15
Test configuration
Read/Write
1. Register locations 0,5, and 13 each provide access to more than one register. The register
accessed depends on the value of the opcode_2 field.
2. Register location 9 provides access to more than one register. The register accessed depends
on the value of the CRm field.
42
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
9.5.1
CP15 Registers Access
CP15 registers can only be accessed in privileged mode by:
• MCR (Move to Coprocessor from ARM Register) instruction is used to write an ARM register
to CP15.
• MRC (Move to ARM Register from Coprocessor) instruction is used to read the value of
CP15 to an ARM register.
Other instructions like CDP, LDC, STC can cause an undefined instruction exception.
The assembler code for these instructions is:
MCR/MRC{cond} p15, opcode_1, Rd, CRn, CRm, opcode_2.
The MCR, MRC instructions bit pattern is shown below:
31
30
29
28
cond
23
22
21
opcode_1
15
20
13
12
Rd
6
26
25
24
1
1
1
0
19
18
17
16
L
14
7
27
5
opcode_2
4
CRn
11
10
9
8
1
1
1
1
3
2
1
0
1
CRm
• CRm[3:0]: Specified Coprocessor Action
Determines specific coprocessor action. Its value is dependent on the CP15 register used. For details, refer to CP15 specific register behavior.
• opcode_2[7:5]
Determines specific coprocessor operation code. By default, set to 0.
• Rd[15:12]: ARM Register
Defines the ARM register whose value is transferred to the coprocessor. If R15 is chosen, the result is unpredictable.
• CRn[19:16]: Coprocessor Register
Determines the destination coprocessor register.
• L: Instruction Bit
0 = MCR instruction
1 = MRC instruction
• opcode_1[23:20]: Coprocessor Code
Defines the coprocessor specific code. Value is c15 for CP15.
• cond [31:28]: Condition
For more details, see Chapter 2 in ARM926EJ-S TRM.
43
6438D–ATARM–13-Oct-09
9.6
Memory Management Unit (MMU)
The ARM926EJ-S processor implements an enhanced ARM architecture v5 MMU to provide virtual memory features required by operating systems like Symbian OS®, Windows CE®, and
Linux®. These virtual memory features are memory access permission controls and virtual to
physical address translations.
The Virtual Address generated by the CPU core is converted to a Modified Virtual Address
(MVA) by the FCSE (Fast Context Switch Extension) using the value in CP15 register13. The
MMU translates modified virtual addresses to physical addresses by using a single, two-level
page table set stored in physical memory. Each entry in the set contains the access permissions
and the physical address that correspond to the virtual address.
The first level translation tables contain 4096 entries indexed by bits [31:20] of the MVA. These
entries contain a pointer to either a 1 MB section of physical memory along with attribute information (access permissions, domain, etc.) or an entry in the second level translation tables;
coarse table and fine table.
The second level translation tables contain two subtables, coarse table and fine table. An entry
in the coarse table contains a pointer to both large pages and small pages along with access
permissions. An entry in the fine table contains a pointer to large, small and tiny pages.
Table 7 shows the different attributes of each page in the physical memory.
Table 9-6.
Mapping Details
Mapping Name
Mapping Size
Access Permission By
Subpage Size
Section
1M byte
Section
-
Large Page
64K bytes
4 separated subpages
16K bytes
Small Page
4K bytes
4 separated subpages
1K byte
Tiny Page
1K byte
Tiny Page
-
The MMU consists of:
• Access control logic
• Translation Look-aside Buffer (TLB)
• Translation table walk hardware
9.6.1
Access Control Logic
The access control logic controls access information for every entry in the translation table. The
access control logic checks two pieces of access information: domain and access permissions.
The domain is the primary access control mechanism for a memory region; there are 16 of them.
It defines the conditions necessary for an access to proceed. The domain determines whether
the access permissions are used to qualify the access or whether they should be ignored.
The second access control mechanism is access permissions that are defined for sections and
for large, small and tiny pages. Sections and tiny pages have a single set of access permissions
whereas large and small pages can be associated with 4 sets of access permissions, one for
each subpage (quarter of a page).
9.6.2
44
Translation Look-aside Buffer (TLB)
The Translation Look-aside Buffer (TLB) caches translated entries and thus avoids going
through the translation process every time. When the TLB contains an entry for the MVA (Modi-
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
fied Virtual Address), the access control logic determines if the access is permitted and outputs
the appropriate physical address corresponding to the MVA. If access is not permitted, the MMU
signals the CPU core to abort.
If the TLB does not contain an entry for the MVA, the translation table walk hardware is invoked
to retrieve the translation information from the translation table in physical memory.
9.6.3
Translation Table Walk Hardware
The translation table walk hardware is a logic that traverses the translation tables located in
physical memory, gets the physical address and access permissions and updates the TLB.
The number of stages in the hardware table walking is one or two depending whether the
address is marked as a section-mapped access or a page-mapped access.
There are three sizes of page-mapped accesses and one size of section-mapped access. Pagemapped accesses are for large pages, small pages and tiny pages. The translation process
always begins with a level one fetch. A section-mapped access requires only a level one fetch,
but a page-mapped access requires an additional level two fetch. For further details on the
MMU, please refer to chapter 3 in ARM926EJ-S Technical Reference Manual.
9.6.4
MMU Faults
The MMU generates an abort on the following types of faults:
• Alignment faults (for data accesses only)
• Translation faults
• Domain faults
• Permission faults
The access control mechanism of the MMU detects the conditions that produce these faults. If
the fault is a result of memory access, the MMU aborts the access and signals the fault to the
CPU core.The MMU retains status and address information about faults generated by the data
accesses in the data fault status register and fault address register. It also retains the status of
faults generated by instruction fetches in the instruction fault status register.
The fault status register (register 5 in CP15) indicates the cause of a data or prefetch abort, and
the domain number of the aborted access when it happens. The fault address register (register 6
in CP15) holds the MVA associated with the access that caused the Data Abort. For further
details on MMU faults, please refer to chapter 3 in ARM926EJ-S Technical Reference Manual.
45
6438D–ATARM–13-Oct-09
9.7
Caches and Write Buffer
The ARM926EJ-S contains a 32K Byte Instruction Cache (ICache), a 32K Byte Data Cache
(DCache), and a write buffer. Although the ICache and DCache share common features, each
still has some specific mechanisms.
The caches (ICache and DCache) are four-way set associative, addressed, indexed and tagged
using the Modified Virtual Address (MVA), with a cache line length of eight words with two dirty
bits for the DCache. The ICache and DCache provide mechanisms for cache lockdown, cache
pollution control, and line replacement.
A new feature is now supported by ARM926EJ-S caches called allocate on read-miss commonly
known as wrapping. This feature enables the caches to perform critical word first cache refilling.
This means that when a request for a word causes a read-miss, the cache performs an AHB
access. Instead of loading the whole line (eight words), the cache loads the critical word first, so
the processor can reach it quickly, and then the remaining words, no matter where the word is
located in the line.
The caches and the write buffer are controlled by the CP15 register 1 (Control), CP15 register 7
(cache operations) and CP15 register 9 (cache lockdown).
9.7.1
Instruction Cache (ICache)
The ICache caches fetched instructions to be executed by the processor. The ICache can be
enabled by writing 1 to I bit of the CP15 Register 1 and disabled by writing 0 to this same bit.
When the MMU is enabled, all instruction fetches are subject to translation and permission
checks. If the MMU is disabled, all instructions fetches are cachable, no protection checks are
made and the physical address is flat-mapped to the modified virtual address. With the MVA use
disabled, context switching incurs ICache cleaning and/or invalidating.
When the ICache is disabled, all instruction fetches appear on external memory (AHB) (see
Tables 4-1 and 4-2 in page 4-4 in ARM926EJ-S TRM).
On reset, the ICache entries are invalidated and the ICache is disabled. For best performance,
ICache should be enabled as soon as possible after reset.
9.7.2
9.7.2.1
Data Cache (DCache) and Write Buffer
ARM926EJ-S includes a DCache and a write buffer to reduce the effect of main memory bandwidth and latency on data access performance. The operations of DCache and write buffer are
closely connected.
DCache
The DCache needs the MMU to be enabled. All data accesses are subject to MMU permission
and translation checks. Data accesses that are aborted by the MMU do not cause linefills or data
accesses to appear on the AMBA ASB interface. If the MMU is disabled, all data accesses are
noncachable, nonbufferable, with no protection checks, and appear on the AHB bus. All
addresses are flat-mapped, VA = MVA = PA, which incurs DCache cleaning and/or invalidating
every time a context switch occurs.
The DCache stores the Physical Address Tag (PA Tag) from which every line was loaded and
uses it when writing modified lines back to external memory. This means that the MMU is not
involved in write-back operations.
Each line (8 words) in the DCache has two dirty bits, one for the first four words and the other
one for the second four words. These bits, if set, mark the associated half-lines as dirty. If the
46
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
cache line is replaced due to a linefill or a cache clean operation, the dirty bits are used to decide
whether all, half or none is written back to memory.
DCache can be enabled or disabled by writing either 1 or 0 to bit C in register 1 of CP15 (see
Tables 4-3 and 4-4 on page 4-5 in ARM926EJ-S TRM).
The DCache supports write-through and write-back cache operations, selected by memory
region using the C and B bits in the MMU translation tables.
The DCache contains an eight data word entry, single address entry write-back buffer used to
hold write-back data for cache line eviction or cleaning of dirty cache lines.
The Write Buffer can hold up to 16 words of data and four separate addresses. DCache and
Write Buffer operations are closely connected as their configuration is set in each section by the
page descriptor in the MMU translation table.
9.7.2.2
Write Buffer
The ARM926EJ-S contains a write buffer that has a 16-word data buffer and a four- address buffer. The write buffer is used for all writes to a bufferable region, write-through region and writeback region. It also allows to avoid stalling the processor when writes to external memory are
performed. When a store occurs, data is written to the write buffer at core speed (high speed).
The write buffer then completes the store to external memory at bus speed (typically slower than
the core speed). During this time, the ARM9EJ-S processor can preform other tasks.
DCache and Write Buffer support write-back and write-through memory regions, controlled by C
and B bits in each section and page descriptor within the MMU translation tables.
9.7.2.3
Write-though Operation
When a cache write hit occurs, the DCache line is updated. The updated data is then written to
the write buffer which transfers it to external memory.
When a cache write miss occurs, a line, chosen by round robin or another algorithm, is stored in
the write buffer which transfers it to external memory.
9.7.2.4
Write-back Operation
When a cache write hit occurs, the cache line or half line is marked as dirty, meaning that its
contents are not up-to-date with those in the external memory.
When a cache write miss occurs, a line, chosen by round robin or another algorithm, is stored in
the write buffer which transfers it to external memory.
47
6438D–ATARM–13-Oct-09
9.8
9.8.1
Tightly-Coupled Memory Interface
TCM Description
The ARM926EJ-S processor features a Tightly-coupled Memory (TCM) interface, which enables
separate instruction and data TCMs (ITCM and DTCM) to be directly reached by the processor.
TCMs are used to store real-time and performance critical code, they also provide a DMA support mechanism. Unlike AHB accesses to external memories, accesses to TCMs are fast and
deterministic and do not incur bus penalties.
The user has the possibility to independently configure each TCM size with values within the following ranges, [0K Byte, 64K Bytes] for ITCM size and [0K Byte, 64K Bytes] for DTCM size.
TCMs can be configured by two means: HMATRIX TCM register and TCM region register (register 9) in CP15 and both steps should be performed. HMATRIX TCM register sets TCM size
whereas TCM region register (register 9) in CP15 maps TCMs and enables them.
The data side of the ARM9EJ-S core is able to access the ITCM. This is necessary to enable
code to be loaded into the ITCM, for SWI and emulated instruction handlers, and for accesses to
PC-relative literal pools.
9.8.2
Enabling and Disabling TCMs
Prior to any enabling step, the user should configure the TCM sizes in HMATRIX TCM register.
Then enabling TCMs is performed by using TCM region register (register 9) in CP15. The user
should use the same sizes as those put in HMATRIX TCM register. For further details and programming tips, please refer to chapter 2.3 in ARM926EJ-S TRM.
9.8.3
TCM Mapping
The TCMs can be located anywhere in the memory map, with a single region available for ITCM
and a separate region available for DTCM. The TCMs are physically addressed and can be
placed anywhere in physical address space. However, the base address of a TCM must be
aligned to its size, and the DTCM and ITCM regions must not overlap. TCM mapping is performed by using TCM region register (register 9) in CP15. The user should input the right
mapping address for TCMs.
48
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
9.9
Bus Interface Unit
The ARM926EJ-S features a Bus Interface Unit (BIU) that arbitrates and schedules AHB
requests. The BIU implements a multi-layer AHB, based on the AHB-Lite protocol, that enables
parallel access paths between multiple AHB masters and slaves in a system. This is achieved by
using a more complex interconnection matrix and gives the benefit of increased overall bus
bandwidth, and a more flexible system architecture.
The multi-master bus architecture has a number of benefits:
• It allows the development of multi-master systems with an increased bus bandwidth and a
flexible architecture.
• Each AHB layer becomes simple because it only has one master, so no arbitration or masterto-slave muxing is required. AHB layers, implementing AHB-Lite protocol, do not have to
support request and grant, nor do they have to support retry and split transactions.
• The arbitration becomes effective when more than one master wants to access the same
slave simultaneously.
9.9.1
Supported Transfers
The ARM926EJ-S processor performs all AHB accesses as single word, bursts of four words, or
bursts of eight words. Any ARM9EJ-S core request that is not 1, 4, 8 words in size is split into
packets of these sizes. Note that the Atmel bus is AHB-Lite protocol compliant, hence it does not
support split and retry requests.
Table 8 gives an overview of the supported transfers and different kinds of transactions they are
used for.
Table 9-7.
Supported Transfers
HBurst[2:0]
Description
Single transfer of word, half word, or byte:
• data write (NCNB, NCB, WT, or WB that has missed in DCache)
SINGLE
Single transfer
• data read (NCNB or NCB)
• NC instruction fetch (prefetched and non-prefetched)
• page table walk read
INCR4
Four-word incrementing burst
Half-line cache write-back, Instruction prefetch, if enabled. Four-word burst NCNB,
NCB, WT, or WB write.
INCR8
Eight-word incrementing burst
Full-line cache write-back, eight-word burst NCNB, NCB, WT, or WB write.
WRAP8
Eight-word wrapping burst
Cache linefill
9.9.2
Thumb Instruction Fetches
All instructions fetches, regardless of the state of ARM9EJ-S core, are made as 32-bit accesses
on the AHB. If the ARM9EJ-S is in Thumb state, then two instructions can be fetched at a time.
9.9.3
Address Alignment
The ARM926EJ-S BIU performs address alignment checking and aligns AHB addresses to the
necessary boundary. 16-bit accesses are aligned to halfword boundaries, and 32-bit accesses
are aligned to word boundaries.
49
6438D–ATARM–13-Oct-09
50
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
10. AT91SAM9G45 Debug and Test
10.1
Description
The AT91SAM9G45 features a number of complementary debug and test capabilities. A common JTAG/ICE (In-Circuit Emulator) port is used for standard debugging functions, such as
downloading code and single-stepping through programs. The Debug Unit provides a two-pin
UART that can be used to upload an application into internal SRAM. It manages the interrupt
handling of the internal COMMTX and COMMRX signals that trace the activity of the Debug
Communication Channel.
A set of dedicated debug and test input/output pins gives direct access to these capabilities from
a PC-based test environment.
10.2
Embedded Characteristics
• ARM926 Real-time In-circuit Emulator
– Two real-time Watchpoint Units
– Two Independent Registers: Debug Control Register and Debug Status Register
– Test Access Port Accessible through JTAG Protocol
– Debug Communications Channel
• Debug Unit
– Two-pin UART
– Debug Communication Channel Interrupt Handling
– Chip ID Register
• IEEE1149.1 JTAG Boundary-scan on All Digital Pins.
51
6438D–ATARM–13-Oct-09
10.3
Block Diagram
Figure 10-1. Debug and Test Block Diagram
TMS
TCK
TDI
NTRST
ICE/JTAG
TAP
Boundary
Port
JTAGSEL
TDO
RTCK
POR
Reset
and
Test
ARM9EJ-S
TST
ICE-RT
ARM926EJ-S
DBGU
PIO
DTXD
PDC
DRXD
TAP: Test Access Port
52
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
10.4
10.4.1
Application Examples
Debug Environment
Figure 10-2 on page 53 shows a complete debug environment example. The ICE/JTAG interface is used for standard debugging functions, such as downloading code and single-stepping
through the program. A software debugger running on a personal computer provides the user
interface for configuring a Trace Port interface utilizing the ICE/JTAG interface.
Figure 10-2. Application Debug and Trace Environment Example
Host Debugger PC
ICE/JTAG
Interface
ICE/JTAG
Connector
AT91SAM9G45
RS232
Connector
Terminal
AT91SAM9G45-based Application Board
53
6438D–ATARM–13-Oct-09
10.4.2
Test Environment
Figure 10-3 on page 54 shows a test environment example. Test vectors are sent and interpreted by the tester. In this example, the “board in test” is designed using a number of JTAGcompliant devices. These devices can be connected to form a single scan chain.
Figure 10-3. Application Test Environment Example
Test Adaptor
Tester
JTAG
Interface
ICE/JTAG
Chip n
AT91SAM9G45
Chip 2
Chip 1
AT91SAM9G45-based Application Board In Test
10.5
Debug and Test Pin Description
Table 10-1.
Pin Name
Debug and Test Pin List
Function
Type
Active Level
Input/Output
Low
Input
High
Low
Reset/Test
NRST
Microcontroller Reset
TST
Test Mode Select
ICE and JTAG
NTRST
Test Reset Signal
Input
TCK
Test Clock
Input
TDI
Test Data In
Input
TDO
Test Data Out
TMS
Test Mode Select
RTCK
Returned Test Clock
JTAGSEL
JTAG Selection
Output
Input
Output
Input
Debug Unit
54
DRXD
Debug Receive Data
Input
DTXD
Debug Transmit Data
Output
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
10.6
10.6.1
Functional Description
Test Pin
One dedicated pin, TST, is used to define the device operating mode. The user must make sure
that this pin is tied at low level to ensure normal operating conditions. Other values associated
with this pin are reserved for manufacturing test.
10.6.2
EmbeddedICE
The ARM9EJ-S EmbeddedICE-RT™ is supported via the ICE/JTAG port. It is connected to a
host computer via an ICE interface. Debug support is implemented using an ARM9EJ-S core
embedded within the ARM926EJ-S. The internal state of the ARM926EJ-S is examined through
an ICE/JTAG port which allows instructions to be serially inserted into the pipeline of the core
without using the external data bus. Therefore, when in debug state, a store-multiple (STM) can
be inserted into the instruction pipeline. This exports the contents of the ARM9EJ-S registers.
This data can be serially shifted out without affecting the rest of the system.
There are two scan chains inside the ARM9EJ-S processor which support testing, debugging,
and programming of the EmbeddedICE-RT. The scan chains are controlled by the ICE/JTAG
port.
EmbeddedICE mode is selected when JTAGSEL is low. It is not possible to switch directly
between ICE and JTAG operations. A chip reset must be performed after JTAGSEL is changed.
For further details on the EmbeddedICE-RT, see the ARM document:
ARM9EJ-S Technical Reference Manual (DDI 0222A).
10.6.3
JTAG Signal Description
TMS is the Test Mode Select input which controls the transitions of the test interface state
machine.
TDI is the Test Data Input line which supplies the data to the JTAG registers (Boundary Scan
Register, Instruction Register, or other data registers).
TDO is the Test Data Output line which is used to serially output the data from the JTAG registers to the equipment controlling the test. It carries the sampled values from the boundary scan
chain (or other JTAG registers) and propagates them to the next chip in the serial test circuit.
NTRST (optional in IEEE Standard 1149.1) is a Test-ReSeT input which is mandatory in ARM
cores and used to reset the debug logic. On Atmel ARM926EJ-S-based cores, NTRST is a
Power On Reset output. It is asserted on power on. If necessary, the user can also reset the
debug logic with the NTRST pin assertion during 2.5 MCK periods.
TCK is the Test ClocK input which enables the test interface. TCK is pulsed by the equipment
controlling the test and not by the tested device. It can be pulsed at any frequency. Note the
maximum JTAG clock rate on ARM926EJ-S cores is 1/6th the clock of the CPU. This gives 5.45
kHz maximum initial JTAG clock rate for an ARM9E running from the 32.768 kHz slow clock.
RTCK is the Return Test Clock. Not an IEEE Standard 1149.1 signal added for a better clock
handling by emulators. From some ICE Interface probes, this return signal can be used to synchronize the TCK clock and take not care about the given ratio between the ICE Interface clock
and system clock equal to 1/6th. This signal is only available in JTAG ICE Mode and not in
boundary scan mode.
55
6438D–ATARM–13-Oct-09
10.6.4
Debug Unit
The Debug Unit provides a two-pin (DXRD and TXRD) USART that can be used for several
debug and trace purposes and offers an ideal means for in-situ programming solutions and
debug monitor communication. Moreover, the association with two peripheral data controller
channels permits packet handling of these tasks with processor time reduced to a minimum.
The Debug Unit also manages the interrupt handling of the COMMTX and COMMRX signals
that come from the ICE and that trace the activity of the Debug Communication Channel.The
Debug Unit allows blockage of access to the system through the ICE interface.
A specific register, the Debug Unit Chip ID Register, gives information about the product version
and its internal configuration.
The AT91SAM9G45 Debug Unit Chip ID value is 0x819B 05A2 and the extended ID is
0x00000004 on 32-bit width.
For further details on the Debug Unit, see the Debug Unit section.
10.6.5
IEEE 1149.1 JTAG Boundary Scan
IEEE 1149.1 JTAG Boundary Scan allows pin-level access independent of the device packaging
technology.
IEEE 1149.1 JTAG Boundary Scan is enabled when JTAGSEL is high. The SAMPLE, EXTEST
and BYPASS functions are implemented. In ICE debug mode, the ARM processor responds
with a non-JTAG chip ID that identifies the processor to the ICE system. This is not IEEE 1149.1
JTAG-compliant.
It is not possible to switch directly between JTAG and ICE operations. A chip reset must be performed after JTAGSEL is changed.
A Boundary-scan Descriptor Language (BSDL) file is provided to set up test.
10.6.6
Access:
31
JID Code Register
Read-only
30
29
28
27
VERSION
23
22
26
25
24
PART NUMBER
21
20
19
18
17
16
10
9
8
PART NUMBER
15
14
13
12
11
PART NUMBER
7
6
MANUFACTURER IDENTITY
5
4
MANUFACTURER IDENTITY
3
2
1
0
1
• VERSION[31:28]: Product Version Number
Set to 0x0.
• PART NUMBER[27:12]: Product Part Number
Product part Number is 5B27
• MANUFACTURER IDENTITY[11:1]
Set to 0x01F.
Bit[0] required by IEEE Std. 1149.1.
Set to 0x1.
JTAG ID Code value is 05B2_703F.
56
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
11. Boot Strategies
The system always boots at address 0x0. To ensure maximum boot possibilities the memory
layout can be changed with two parameters.
• REMAP allows the user to layout the internal SRAM bank to 0x0 to ease the development.
This is done by software once the system has boot.
• BMS allows the user to layout to 0x0, when convenient, the ROM or an external memory. This
is done by hardware at reset.
Note: All the memory blocks can always be seen at their specified base addresses that are not
concerned by these parameters.
The AT91SAM9G45 manages a boot memory that depends on the level on the BMS pin at
reset. The internal memory area mapped between address 0x0 and 0x000F FFFF is reserved to
this effect.
If BMS is detected at 0, the boot memory is the memory connected on the Chip Select 0 of the
External Bus Interface.
• Boot on on-chip RC
• Boot with the default configuration for the Static Memory Controller, byte select mode, 16-bit
data bus, Read/Write controlled by Chip Select, allows boot on 16-bit non-volatile memory.
For optimization purpose, nothing else is done. To speed up the boot sequence user programmed software should perform a complete configuration:
• Enable the 32768 Hz oscillator if best accuracy is needed
• Program the PMC (main oscillator enable or bypass mode)
• Program and Start the PLL
• Reprogram the SMC setup, cycle, hold, mode timings registers for EBI CS0 to adapt them to
the new clock
• Switch the system clock to the new value
If BMS is detected at 1, the boot memory is the embedded ROM and the boot program
described below is executed.
11.1
Boot Program
The Boot Program is contained in the embedded ROM. It is also called: “Rom Code” or “First
level bootloader”. At power on, if the BMS pin is detected at 1, the boot memory is the embedded ROM and the Boot Program is executed.
The Boot Program consists of several steps. First, it performs device initialization. Then it
attempts to boot from external non volatile memories (NVM). And finally, if no valid program is
found in NVM, it executes a monitor called SAM-BA® Monitor.
57
6438D–ATARM–13-Oct-09
11.2
Flow Diagram
The Boot Program implements the algorithm shown below in Figure 11-1.
Figure 11-1. Boot Program Algorithm Flow Diagram
Device Setup
Valid boot code
found in one
NVM
Yes
Copy and run it
in internal SRAM
No
SAM-BA Monitor
58
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
11.3
Device Initialization
11.3.1
Clock at Start Up
At boot start up, the processor clock (PCK) and the master clock (MCK) are found on the slow
clock. The slow clock can be an external 32 kHz crystal oscillator or the internal RC oscillator. By
default the slow clock is the internal RC oscillator. Its frequency is not precise and is between 20
kHz and 40 kHz. Its start up is much faster than an external 32 kHz quartz. If a battery supplies
the backup power and if the external 32 kHz clock was previously started up and selected, the
slow clock at boot is the external 32 kHz quartz oscillator. Refer to the Slow Clock Crystal Oscillator description in the Clock Generator section of the datasheet.
11.3.2
Initialization Sequence
Initialization follows the steps described below:
1. Stack setup for ARM supervisor mode.
2. Main Oscillator Detection: (External crystal or external clock on XIN). The Main Oscillator is disabled at startup (MOSCEN = 0). First it is bypassed (OSCBYPASS set at 1).
Then the MAINRDY bit is polled. Since this bit is raised, the Main Clock Frequency field
is analyzed (MAINF). If the value is bigger than 16, an external clock connected on XIN
is detected. If not, an external quartz connected between XIN and XOUT (whose frequency is unknown at this moment) is detected.
3. Main Oscillator Enabling: if an external clock is connected on XIN, the Main Oscillator
does not need to be started. Otherwise, the OSCBYPASS bit is not set. The Main Oscillator is enabled (MOSCEN = 1) with the maximum start-up time and the MOSC bit is
polled to wait for stabilization.
4. Main Oscillator Selection: the Master Clock source is switched from Slow Clock to the
Main Oscillator without prescaler. The PMC Status Register is polled to wait for MCK
Ready. PCK and MCK are now the Main Oscillator clock.
5. C variable initialization: non zero-initialized data are initialized in RAM (copy from
ROM to RAM). Zero-initialized data are set to 0 in RAM.
6. PLLA initialization: PLLA is configured to allow communication on the USB link for the
SAM-BA Monitor. Its configuration depends on the Main Oscillator source (external
clock or crystal) and on its frequency.
59
6438D–ATARM–13-Oct-09
AT91SAM9G45
11.4
11.4.1
NVM Boot
NVM Bootloader Program Description
Figure 11-2. NVM bootloader program diagram
Start
Initialize NVM
Initialization OK ?
No
Restore the reset values
for the peripherals and
Jump to next boot solution
Yes
Valid code detection in NVM
NVM contains valid code
No
Yes
Copy the valid code
from external NVM to internal SRAM.
Restore the reset values for the peripherals.
Perform the REMAP and set the PC to 0
to jump to the downloaded application
End
60
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 11-3. Remap Action after Download Completion
0x0000_0000
0x0000_0000
REMAP
Internal
ROM
Internal
SRAM
0x0030_0000
0x0030_0000
Internal
SRAM
Internal
SRAM
0x0040_0000
0x0040_0000
Internal
ROM
Internal
ROM
The NVM bootloader program initializes the NVM. It initializes the required PIO. It sets the right
peripheral depending on the NVM and tries to access the memory. If the initialization fails, it
restores the reset values for the PIO and peripherals and then the next NVM bootloader program
is executed.
If the initialization is successful, the NVM bootloader program reads the beginning of the NVM
and determines if the NVM contains valid code.
If the NVM does not contain valid code, the NVM bootloader program restores the reset value for
the peripherals and then the next NVM bootloader program is executed.
If valid code is found, this code is loaded from NVM into internal SRAM and executed by branching at address 0x0000_0000 after remap. This code may be the application code or a secondlevel bootloader. All the calls to functions are PC relative and do not use absolute addresses.
11.4.2
11.4.2.1
Valid Code Detection
There are two kinds of valid code detection. Depending on the NVM bootloader, either one or
both of them is used.
ARM Exception Vectors Check
The NVM bootloader program reads and analyzes the first 28 bytes corresponding to the first
seven ARM exception vectors. Except for the sixth vector, these bytes must implement the ARM
instructions for either branch or load PC with PC relative addressing.
Figure 11-4. LDR Opcode
31
1
28 27
1
1
0
0
24 23
1
I
P
U
20 19
1
W
0
16 15
Rn
12 11
Rd
0
O set
61
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 11-5. B Opcode
31
1
28 27
1
1
0
1
24 23
0
1
0
0
O set (24 bits)
Unconditional instruction: 0xE for bits 31 to 28
Load PC with PC relative addressing instruction:
– Rn = Rd = PC = 0xF
– I==0 (12-bit immediate value)
– P==1 (pre-indexed)
– U offset added (U==1) or subtracted (U==0)
– W==1
The sixth vector, at offset 0x14, contains the size of the image to download. The user must
replace this vector with his/her own vector. This information is described below.
Figure 11-6. Structure of the ARM Vector 6
31
0
Size of the code to download in bytes
The value has to be smaller than 60 KBytes. 60 KBytes is the maximum size for a valid code.
This size is the internal SRAM size minus the stack size used by the ROM Code at the end of
the internal SRAM.
Example
An example of valid vectors follows:
11.4.2.2
00
ea000006
B
0x20
04
eafffffe
B
0x04
08
ea00002f
B
_main
0c
eafffffe
B
0x0c
10
eafffffe
B
0x10
14
00001234
B
0x14
18
eafffffe
B
0x18
<- Code size = 4660 bytes < 60kB
boot.bin file check
The NVM bootloader program looks for a boot.bin file in the root directory of a FAT12/16/32 formatted NVM Flash.
62
6438D–ATARM–13-Oct-09
AT91SAM9G45
11.4.3
NVM Bootloader Sequence
Figure 11-7. NVM Bootloader Sequence Diagram
Device
Setup
NAND Flash Boot
Yes
Copy from
NAND Flash to SRAM
Run
NAND Flash Bootloader
Yes
Copy from
SD Card to SRAM
Run
SD Card Bootloader
Yes
Copy from
SPI Flash to SRAM
Run
SPI Flash Bootloader
Yes
Copy from
TWI EEPROM to SRAM
Run
TWI EEPROM Bootloader
No
SD Card Boot
No
SPI Flash Boot
No
TWI EEPROM Boot
No
SAM-BA
Monitor
11.4.3.1
NAND Flash Boot
The NAND Flash bootloader program uses the EBI CS3. It uses both valid code detections. First
it searches a boot.bin file. Then it analyzes the ARM exception vectors.
The first block must be guaranteed by the manufacturer. There is no ECC check.
After NAND Flash interface configuration, the Manufacturer ID is read. If it is different from 0xFF,
the Device ID is read, else, the NAND Flash boot is aborted. The Boot program contains a list of
SLC small block Device ID with their characteristics (size, bus width, voltage) (see Table 11-1). If
the device ID is not found in this list, the NAND Flash device is considered as an SLC large block
and its characteristics are obtained by reading the Extended Device ID byte 3.
63
6438D–ATARM–13-Oct-09
AT91SAM9G45
Supported NAND Flash Devices
The supported SLC small block NAND Flash devices that are described below inTable 11-1.
Table 11-1.
Supported SLC Small Block NAND Flash
Device ID
Size
(MBytes)
PageSize
(Bytes)
BlockSsize
(Bytes)
Bus Width
Voltage (V)
0x6E
1
256
4096
8
5
0x64
2
256
4096
8
5
0x6B
4
512
8196
8
5
0xE8
1
256
4096
8
3.3
0xEC
1
256
4096
8
3.3
0xEA
2
256
4096
8
3.3
0xE3
4
512
8196
8
3.3
0xE5
4
512
8196
8
3.3
0xD6
8
512
8196
8
3.3
0xE6
8
512
8196
8
3.3
0x33
16
512
16384
8
1.8
0x73
16
512
16384
8
3.3
0x43
16
512
16384
16
1.8
0x53
16
512
16384
16
3.3
0x45
32
512
16384
16
1.8
0x55
32
512
16384
16
3.3
0x36
64
512
16384
8
1.8
0x76
64
512
16384
8
3.3
0x46
64
512
16384
16
1.8
0x56
64
512
16384
16
3.3
0x78
128
512
16384
8
1.8
0x79
128
512
16384
8
3.3
0x72
128
512
16384
16
1.8
0x74
128
512
16384
16
3.3
The NAND Flash boot also supports all the SLC large block NAND Flash devices.
11.4.3.2
SD Card Boot
The SD Card bootloader uses MCI0. It uses only one valid code detection. It searches a boot.bin
file.
Supported SD Card devices
SD Card Boot supports all SD Card memories compliant with SD Memory Card Specification
V2.0. This includes SDHC cards.
64
6438D–ATARM–13-Oct-09
AT91SAM9G45
11.4.3.3
SPI Flash Boot
Two kinds of SPI Flash are supported, SPI Serial Flash and SPI DataFlash.
The SPI Flash bootloader tries to boot on SPI0 Chip Select 0, first looking for SPI Serial flash,
and then for SPI DataFlash.
It uses only one valid code detection: analysis of ARM exception vectors.
The SPI Flash read is done thanks to a Continuous Read command from address 0x0. This
command is 0xE8 for DataFlash and 0x0B for Serial Flash devices.
Supported DataFlash Devices
The SPI Flash Boot program supports all Atmel DataFlash devices.
Table 11-2.
Device
DataFlash Device
Density
Page Size (bytes)
Number of Pages
AT45DB011
1 Mbit
264
512
AT45DB021
2 Mbits
264
1024
AT45DB041
4 Mbits
264
2048
AT45DB081
8 Mbits
264
4096
AT45DB161
16 Mbits
528
4096
AT45DB321
32 Mbits
528
8192
AT45DB642
64 Mbits
1056
8192
Supported Serial Flash Devices
The SPI Flash Boot program supports all Serial Flash devices.
11.4.3.4
TWI EEPROM Boot
The TWI EEPROM Bootloader uses the TWI0. It uses only one valid code detection. It analyzes
the ARM exception vectors.
Supported TWI EEPROM Devices
TWI EEPROM Boot supports all I2C-compatible TWI EEPROM memories using 7 bits device
address 0x50.
11.4.4
Hardware and Software Constraints
The NVM drivers use several PIOs in peripheral mode to communicate with devices. Care must
be taken when these PIOs are used by the application. The devices connected could be unintentionally driven at boot time, and electrical conflicts between output pins used by the NVM
drivers and the connected devices may occur.
To assure correct functionality, it is recommended to plug in critical devices to other pins not
used by NVM.
Table 11-3 contains a list of pins that are driven during the boot program execution. These pins
are driven during the boot sequence for a period of less than 1 second if no correct boot program
is found.
65
6438D–ATARM–13-Oct-09
AT91SAM9G45
Before performing the jump to the application in internal SRAM, all the PIOs and peripherals
used in the boot program are set to their reset state.
Table 11-3.
PIO Driven during Boot Program Execution
NVM Bootloader
Peripheral
Pin
PIO Line
EBI CS3 SMC
NANDCS
PIOC14
EBI CS3 SMC
NAND ALE
A21
EBI CS3 SMC
NAND CLE
A22
EBI CS3 SMC
Cmd/Addr/Data
D[16:0]
MCI0
MCI0_CK
PIOA0
MCI0
MCI0_CD
PIOA1
MCI0
MCI0_D0
PIOA2
MCI0
MCI0_D1
PIOA3
MCI0
MCI0_D2
PIOA4
MCI0
MCI0_D3
PIOA5
SPI0
MOSI
PIOB1
SPI0
MISO
PIOB0
SPI0
SPCK
PIOB2
SPI0
NPCS0
PIOB3
TWI0
TWD0
PIOA20
TWI0
TWCK0
PIOA21
DBGU
DRXD
PIOB12
DBGU
DTXD
PIOB13
NAND
SD Card
SPI Flash
TWI0 EEPROM
SAM-BA Monitor
11.5
SAM-BA Monitor
If no valid code has been found in NVM during the NVM bootloader sequence, the SAM-BA
Monitor program is launched.
The SAM-BA Monitor principle is to:
– Initialize DBGU and USB
– Check if USB Device enumeration has occurred.
– Check if characters have been received on the DBGU.
– Once the communication interface is identified, the application runs in an infinite
loop waiting for different commands as listed in Table .
66
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 11-8. SAM-BA Monitor Diagram
No valid code in NVM
Init DBGU and USB
No
USB Enumeration
Successful ?
No
Character(s) received
on DBGU ?
Yes
Yes
Run monitor
Wait for command
on the DBGU link
Run monitor
Wait for command
on the USB link
11.5.1
Command List
Table 11-4.
Commands Available through the SAM-BA Monitor
Command
Action
Argument(s)
Example
N
set Normal mode
No argument
N#
T
set Terminal mode
No argument
T#
O
write a byte
Address, Value#
O200001,CA#
o
read a byte
Address,#
o200001,#
H
write a half word
Address, Value#
H200002,CAFE#
h
read a half word
Address,#
h200002,#
W
write a word
Address, Value#
W200000,CAFEDECA#
w
read a word
Address,#
w200000,#
S
send a file
Address,#
S200000,#
R
receive a file
Address, NbOfBytes#
R200000,1234#
G
go
Address#
G200200#
V
display version
No argument
V#
• Mode commands:
– Normal mode configures SAM-BA Monitor to send / receive data in binary format,
– Terminal mode configures SAM-BA Monitor to send / receive data in ascii format.
• Write commands: Write a byte (O), a halfword (H) or a word (W) to the target.
– Address: Address in hexadecimal.
– Value: Byte, halfword or word to write in hexadecimal.
67
6438D–ATARM–13-Oct-09
– Output: ‘>’.
• Read commands: Read a byte (o), a halfword (h) or a word (w) from the target.
– Address: Address in hexadecimal
– Output: The byte, halfword or word read in hexadecimal following by ‘>’
• Send a file (S): Send a file to a specified address
– Address: Address in hexadecimal
– Output: ‘>’.
Note:
There is a time-out on this command which is reached when the prompt ‘>’ appears before the
end of the command execution.
• Receive a file (R): Receive data into a file from a specified address
– Address: Address in hexadecimal
– NbOfBytes: Number of bytes in hexadecimal to receive
– Output: ‘>’
• Go (G): Jump to a specified address and execute the code
– Address: Address to jump in hexadecimal
– Output: ‘>’once returned from the program execution. If the executed program does
not handle the link register at its entry and does not return, the prompt will not be
displayed.
• Get Version (V): Return the Boot Program version
– Output: version, date and time of ROM code followed by the prompt: ‘>’.
11.5.2
DBGU Serial Port
Communication is performed through the DBGU serial port initialized to 115200 Baud, 8 bits of
data, no parity, 1 stop bit.
11.5.2.1
Supported External Crystal/External Clocks
The SAM-BA Monitor supports a frequency of 12 MHz to allow DBGU communication for both
external crystal and external clock.
11.5.2.2
Xmodem Protocol
The Send and Receive File commands use the Xmodem protocol to communicate. Any terminal
performing this protocol can be used to send the application file to the target. The size of the
binary file to send depends on the SRAM size embedded in the product. In all cases, the size of
the binary file must be lower than the SRAM size because the Xmodem protocol requires some
SRAM memory in order to work.
The Xmodem protocol supported is the 128-byte length block. This protocol uses a two-character CRC-16 to guarantee detection of a maximum bit error.
Xmodem protocol with CRC is accurate provided both sender and receiver report successful
transmission. Each block of the transfer looks like:
<SOH><blk #><255-blk #><--128 data bytes--><checksum> in which:
– <SOH> = 01 hex
– <blk #> = binary number, starts at 01, increments by 1, and wraps 0FFH to 00H (not
to 01)
– <255-blk #> = 1’s complement of the blk#.
68
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
– <checksum> = 2 bytes CRC16
Figure 11-9 shows a transmission using this protocol.
Figure 11-9. Xmodem Transfer Example
Host
Device
C
SOH 01 FE Data[128] CRC CRC
ACK
SOH 02 FD Data[128] CRC CRC
ACK
SOH 03 FC Data[100] CRC CRC
ACK
EOT
ACK
11.5.3
USB Device Port
11.5.3.1
Supported external crystal / external clocks
The only frequency supported by SAM-BA Monitor to allow USB communication is a 12 MHz
crystal or external clock.
11.5.3.2
USB class
The device uses the USB communication device class (CDC) drivers to take advantage of the
installed PC RS-232 software to talk over the USB. The CDC class is implemented in all
releases of Windows®, from Windows 98SE® to Windows XP®. The CDC document, available at
www.usb.org, describes how to implement devices such as ISDN modems and virtual COM
ports.
The Vendor ID is Atmel’s vendor ID 0x03EB. The product ID is 0x6124. These references are
used by the host operating system to mount the correct driver. On Windows systems, the INF
files contain the correspondence between vendor ID and product ID.
69
6438D–ATARM–13-Oct-09
11.5.3.3
Enumeration Process
The USB protocol is a master/slave protocol. The host starts the enumeration, sending requests
to the device through the control endpoint. The device handles standard requests as defined in
the USB Specification.
Table 11-5.
Handled Standard Requests
Request
Definition
GET_DESCRIPTOR
Returns the current device configuration value.
SET_ADDRESS
Sets the device address for all future device access.
SET_CONFIGURATION
Sets the device configuration.
GET_CONFIGURATION
Returns the current device configuration value.
GET_STATUS
Returns status for the specified recipient.
SET_FEATURE
Used to set or enable a specific feature.
CLEAR_FEATURE
Used to clear or disable a specific feature.
The device also handles some class requests defined in the CDC class.
Table 11-6.
Handled Class Requests
Request
Definition
SET_LINE_CODING
Configures DTE rate, stop bits, parity and number of
character bits.
GET_LINE_CODING
Requests current DTE rate, stop bits, parity and number
of character bits.
SET_CONTROL_LINE_STATE
RS-232 signal used to tell the DCE device the DTE
device is now present.
Unhandled requests are STALLed.
11.5.3.4
Communication Endpoints
There are two communication endpoints and endpoint 0 is used for the enumeration process.
Endpoint 1 is a 64-byte Bulk OUT endpoint and endpoint 2 is a 64-byte Bulk IN endpoint. SAMBA Boot commands are sent by the host through endpoint 1. If required, the message is split by
the host into several data payloads by the host driver.
If the command requires a response, the host can send IN transactions to pick up the response.
70
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
12. Reset Controller (RSTC)
12.1
Description
The Reset Controller (RSTC), based on power-on reset cells, handles all the resets of the system without any external components. It reports which reset occurred last.
The Reset Controller also drives independently or simultaneously the external reset and the
peripheral and processor resets.
12.2
Embedded Characteristics
The Reset Controller is based on two Power-on-Reset cells, one on VDDBU and one on
VDDCORE.
The Reset Controller is capable to return to the software the source of the last reset, either a
general reset (VDDBU rising), a wake-up reset (VDDCORE rising), a software reset, a user
reset or a watchdog reset.
The Reset Controller controls the internal resets of the system and the NRST pin. The NRST pin
is bidirectional. It is handled by the on-chip reset controller and can be driven low to provide a
reset signal to the external components or asserted low externally to reset the microcontroller. It
will reset the Core and the peripherals except the Backup region. There is no constraint on the
length of the reset pulse and the reset controller can guarantee a minimum pulse length.
The NRST pin integrates a permanent pull-up resistor to VDDIOP0 of about 100 kOhms.
The configuration of the Reset Controller is saved as supplied on VDDBU.
12.3
Block Diagram
Figure 12-1. Reset Controller Block Diagram
Reset Controller
Main Supply
POR
Backup Supply
POR
rstc_irq
Startup
Counter
Reset
State
Manager
proc_nreset
user_reset
NRST
nrst_out
NRST
Manager
periph_nreset
exter_nreset
backup_neset
WDRPROC
wd_fault
SLCK
71
6438D–ATARM–13-Oct-09
12.4
Functional Description
12.4.1
Reset Controller Overview
The Reset Controller is made up of an NRST Manager, a Startup Counter and a Reset State
Manager. It runs at Slow Clock and generates the following reset signals:
• proc_nreset: Processor reset line. It also resets the Watchdog Timer.
• backup_nreset: Affects all the peripherals powered by VDDBU.
• periph_nreset: Affects the whole set of embedded peripherals.
• nrst_out: Drives the NRST pin.
These reset signals are asserted by the Reset Controller, either on external events or on software action. The Reset State Manager controls the generation of reset signals and provides a
signal to the NRST Manager when an assertion of the NRST pin is required.
The NRST Manager shapes the NRST assertion during a programmable time, thus controlling
external device resets.
The startup counter waits for the complete crystal oscillator startup. The wait delay is given by
the crystal oscillator startup time maximum value that can be found in the section Crystal Oscillator Characteristics in the Electrical Characteristics section of the product documentation.
The Reset Controller Mode Register (RSTC_MR), allowing the configuration of the Reset Controller, is powered with VDDBU, so that its configuration is saved as long as VDDBU is on.
12.4.2
NRST Manager
The NRST Manager samples the NRST input pin and drives this pin low when required by the
Reset State Manager. Figure 12-2 shows the block diagram of the NRST Manager.
Figure 12-2. NRST Manager
RSTC_MR
URSTIEN
RSTC_SR
URSTS
NRSTL
rstc_irq
RSTC_MR
URSTEN
Other
interrupt
sources
user_reset
NRST
RSTC_MR
ERSTL
nrst_out
12.4.2.1
External Reset Timer
exter_nreset
NRST Signal or Interrupt
The NRST Manager samples the NRST pin at Slow Clock speed. When the line is detected low,
a User Reset is reported to the Reset State Manager.
However, the NRST Manager can be programmed to not trigger a reset when an assertion of
NRST occurs. Writing the bit URSTEN at 0 in RSTC_MR disables the User Reset trigger.
72
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
The level of the pin NRST can be read at any time in the bit NRSTL (NRST level) in RSTC_SR.
As soon as the pin NRST is asserted, the bit URSTS in RSTC_SR is set. This bit clears only
when RSTC_SR is read.
The Reset Controller can also be programmed to generate an interrupt instead of generating a
reset. To do so, the bit URSTIEN in RSTC_MR must be written at 1.
12.4.2.2
NRST External Reset Control
The Reset State Manager asserts the signal ext_nreset to assert the NRST pin. When this
occurs, the “nrst_out” signal is driven low by the NRST Manager for a time programmed by the
field ERSTL in RSTC_MR. This assertion duration, named EXTERNAL_RESET_LENGTH, lasts
2(ERSTL+1) Slow Clock cycles. This gives the approximate duration of an assertion between 60 μs
and 2 seconds. Note that ERSTL at 0 defines a two-cycle duration for the NRST pulse.
This feature allows the Reset Controller to shape the NRST pin level, and thus to guarantee that
the NRST line is driven low for a time compliant with potential external devices connected on the
system reset.
As the field is within RSTC_MR, which is backed-up, this field can be used to shape the system
power-up reset for devices requiring a longer startup time than the Slow Clock Oscillator.
12.4.3
BMS Sampling
The product matrix manages a boot memory that depends on the level on the BMS pin at reset.
The BMS signal is sampled three slow clock cycles after the Core Power-On-Reset output rising
edge.
Figure 12-3. BMS Sampling
SLCK
Core Supply
POR output
BMS Signal
XXX
H or L
BMS sampling delay
= 3 cycles
proc_nreset
12.4.4
Reset States
The Reset State Manager handles the different reset sources and generates the internal reset
signals. It reports the reset status in the field RSTTYP of the Status Register (RSTC_SR). The
update of the field RSTTYP is performed when the processor reset is released.
12.4.4.1
General Reset
A general reset occurs when VDDBU and VDDCORE are powered on. The backup supply POR
cell output rises and is filtered with a Startup Counter, which operates at Slow Clock. The purpose of this counter is to make sure the Slow Clock oscillator is stable before starting up the
73
6438D–ATARM–13-Oct-09
device. The length of startup time is hardcoded to comply with the Slow Clock Oscillator startup
time.
After this time, the processor clock is released at Slow Clock and all the other signals remain
valid for 3 cycles for proper processor and logic reset. Then, all the reset signals are released
and the field RSTTYP in RSTC_SR reports a General Reset. As the RSTC_MR is reset, the
NRST line rises 2 cycles after the backup_nreset, as ERSTL defaults at value 0x0.
When VDDBU is detected low by the Backup Supply POR Cell, all resets signals are immediately asserted, even if the Main Supply POR Cell does not report a Main Supply shutdown.
VDDBU only activates the backup_nreset signal.
The backup_nreset must be released so that any other reset can be generated by VDDCORE
(Main Supply POR output).
Figure 12-4 shows how the General Reset affects the reset signals.
Figure 12-4. General Reset State
SLCK
Any
Freq.
MCK
Backup Supply
POR output
Startup Time
Main Supply
POR output
backup_nreset
Processor Startup
= 3 cycles
proc_nreset
RSTTYP
XXX
0x0 = General Reset
XXX
periph_nreset
NRST
(nrst_out)
BMS Sampling
EXTERNAL RESET LENGTH
= 2 cycles
74
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
12.4.4.2
Wake-up Reset
The Wake-up Reset occurs when the Main Supply is down. When the Main Supply POR output
is active, all the reset signals are asserted except backup_nreset. When the Main Supply powers up, the POR output is resynchronized on Slow Clock. The processor clock is then re-enabled
during 3 Slow Clock cycles, depending on the requirements of the ARM processor.
At the end of this delay, the processor and other reset signals rise. The field RSTTYP in
RSTC_SR is updated to report a Wake-up Reset.
The “nrst_out” remains asserted for EXTERNAL_RESET_LENGTH cycles. As RSTC_MR is
backed-up, the programmed number of cycles is applicable.
When the Main Supply is detected falling, the reset signals are immediately asserted. This transition is synchronous with the output of the Main Supply POR.
Figure 12-5. Wake-up State
SLCK
Any
Freq.
MCK
Main Supply
POR output
backup_nreset
Resynch.
2 cycles
proc_nreset
RSTTYP
Processor Startup
= 3 cycles
XXX
0x1 = WakeUp Reset
XXX
periph_nreset
NRST
(nrst_out)
EXTERNAL RESET LENGTH
= 4 cycles (ERSTL = 1)
12.4.4.3
User Reset
The User Reset is entered when a low level is detected on the NRST pin and the bit URSTEN in
RSTC_MR is at 1. The NRST input signal is resynchronized with SLCK to insure proper behavior of the system.
The User Reset is entered as soon as a low level is detected on NRST. The Processor Reset
and the Peripheral Reset are asserted.
The User Reset is left when NRST rises, after a two-cycle resynchronization time and a 3-cycle
processor startup. The processor clock is re-enabled as soon as NRST is confirmed high.
75
6438D–ATARM–13-Oct-09
When the processor reset signal is released, the RSTTYP field of the Status Register
(RSTC_SR) is loaded with the value 0x4, indicating a User Reset.
The NRST Manager guarantees that the NRST line is asserted for
EXTERNAL_RESET_LENGTH Slow Clock cycles, as programmed in the field ERSTL. However, if NRST does not rise after EXTERNAL_RESET_LENGTH because it is driven low
externally, the internal reset lines remain asserted until NRST actually rises.
Figure 12-6. User Reset State
SLCK
MCK
Any
Freq.
NRST
Resynch.
2 cycles
Resynch.
2 cycles
Processor Startup
= 3 cycles
proc_nreset
RSTTYP
Any
XXX
0x4 = User Reset
periph_nreset
NRST
(nrst_out)
>= EXTERNAL RESET LENGTH
12.4.4.4
Software Reset
The Reset Controller offers several commands used to assert the different reset signals. These
commands are performed by writing the Control Register (RSTC_CR) with the following bits at
1:
• PROCRST: Writing PROCRST at 1 resets the processor and the watchdog timer.
• PERRST: Writing PERRST at 1 resets all the embedded peripherals, including the memory
system, and, in particular, the Remap Command. The Peripheral Reset is generally used for
debug purposes.
Except for Debug purposes, PERRST must always be used in conjunction with PROCRST
(PERRST and PROCRST set both at 1 simultaneously.)
• EXTRST: Writing EXTRST at 1 asserts low the NRST pin during a time defined by the field
ERSTL in the Mode Register (RSTC_MR).
The software reset is entered if at least one of these bits is set by the software. All these commands can be performed independently or simultaneously. The software reset lasts 3 Slow
Clock cycles.
76
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
The internal reset signals are asserted as soon as the register write is performed. This is
detected on the Master Clock (MCK). They are released when the software reset is left, i.e.; synchronously to SLCK.
If EXTRST is set, the nrst_out signal is asserted depending on the programming of the field
ERSTL. However, the resulting falling edge on NRST does not lead to a User Reset.
If and only if the PROCRST bit is set, the Reset Controller reports the software status in the field
RSTTYP of the Status Register (RSTC_SR). Other Software Resets are not reported in
RSTTYP.
As soon as a software operation is detected, the bit SRCMP (Software Reset Command in Progress) is set in the Status Register (RSTC_SR). It is cleared as soon as the software reset is left.
No other software reset can be performed while the SRCMP bit is set, and writing any value in
RSTC_CR has no effect.
Figure 12-7. Software Reset
SLCK
MCK
Any
Freq.
Write RSTC_CR
Resynch.
1 cycle
Processor Startup
= 3 cycles
proc_nreset
if PROCRST=1
RSTTYP
Any
XXX
0x3 = Software Reset
periph_nreset
if PERRST=1
NRST
(nrst_out)
if EXTRST=1
EXTERNAL RESET LENGTH
8 cycles (ERSTL=2)
SRCMP in RSTC_SR
12.4.4.5
Watchdog Reset
The Watchdog Reset is entered when a watchdog fault occurs. This state lasts 3 Slow Clock
cycles.
When in Watchdog Reset, assertion of the reset signals depends on the WDRPROC bit in
WDT_MR:
• If WDRPROC is 0, the Processor Reset and the Peripheral Reset are asserted. The NRST
line is also asserted, depending on the programming of the field ERSTL. However, the
resulting low level on NRST does not result in a User Reset state.
77
6438D–ATARM–13-Oct-09
• If WDRPROC = 1, only the processor reset is asserted.
The Watchdog Timer is reset by the proc_nreset signal. As the watchdog fault always causes a
processor reset if WDRSTEN is set, the Watchdog Timer is always reset after a Watchdog
Reset, and the Watchdog is enabled by default and with a period set to a maximum.
When the WDRSTEN in WDT_MR bit is reset, the watchdog fault has no impact on the reset
controller.
Figure 12-8. Watchdog Reset
SLCK
MCK
Any
Freq.
wd_fault
Processor Startup
= 3 cycles
proc_nreset
RSTTYP
Any
XXX
0x2 = Watchdog Reset
periph_nreset
Only if
WDRPROC = 0
NRST
(nrst_out)
EXTERNAL RESET LENGTH
8 cycles (ERSTL=2)
12.4.5
Reset State Priorities
The Reset State Manager manages the following priorities between the different reset sources,
given in descending order:
• Backup Reset
• Wake-up Reset
• Watchdog Reset
• Software Reset
• User Reset
Particular cases are listed below:
• When in User Reset:
– A watchdog event is impossible because the Watchdog Timer is being reset by the
proc_nreset signal.
– A software reset is impossible, since the processor reset is being activated.
• When in Software Reset:
– A watchdog event has priority over the current state.
– The NRST has no effect.
78
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
• When in Watchdog Reset:
– The processor reset is active and so a Software Reset cannot be programmed.
– A User Reset cannot be entered.
12.4.6
Reset Controller Status Register
The Reset Controller status register (RSTC_SR) provides several status fields:
• RSTTYP field: This field gives the type of the last reset, as explained in previous sections.
• SRCMP bit: This field indicates that a Software Reset Command is in progress and that no
further software reset should be performed until the end of the current one. This bit is
automatically cleared at the end of the current software reset.
• NRSTL bit: The NRSTL bit of the Status Register gives the level of the NRST pin sampled on
each MCK rising edge.
• URSTS bit: A high-to-low transition of the NRST pin sets the URSTS bit of the RSTC_SR
register. This transition is also detected on the Master Clock (MCK) rising edge (see Figure
12-9). If the User Reset is disabled (URSTEN = 0) and if the interruption is enabled by the
URSTIEN bit in the RSTC_MR register, the URSTS bit triggers an interrupt. Reading the
RSTC_SR status register resets the URSTS bit and clears the interrupt.
Figure 12-9.
Reset Controller Status and Interrupt
MCK
read
RSTC_SR
Peripheral Access
2 cycle
resynchronization
2 cycle
resynchronization
NRST
NRSTL
URSTS
rstc_irq
if (URSTEN = 0) and
(URSTIEN = 1)
79
6438D–ATARM–13-Oct-09
12.5
Reset Controller (RSTC) User Interface
Table 12-1.
Register Mapping
Offset
Register
Name
0x00
Control Register
0x04
0x08
Note:
80
Access
Reset
Backup Reset
RSTC_CR
Write-only
-
Status Register
RSTC_SR
Read-only
0x0000_0001
0x0000_0000
Mode Register
RSTC_MR
Read-write
-
0x0000_0001
1. The reset value of RSTC_SR either reports a General Reset or a Wake-up Reset depending on last rising power supply.
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
12.5.1
Name:
Reset Controller Control Register
RSTC_CR
Address:
0xFFFFFD00
Access Type:
Write-only
31
30
29
28
27
26
25
24
KEY
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
8
–
7
–
6
–
5
–
4
–
3
EXTRST
2
PERRST
1
–
0
PROCRST
• PROCRST: Processor Reset
0 = No effect.
1 = If KEY is correct, resets the processor.
• PERRST: Peripheral Reset
0 = No effect.
1 = If KEY is correct, resets the peripherals.
• EXTRST: External Reset
0 = No effect.
1 = If KEY is correct, asserts the NRST pin.
• KEY: Password
Should be written at value 0xA5. Writing any other value in this field aborts the write operation.
81
6438D–ATARM–13-Oct-09
12.5.2
Name:
Reset Controller Status Register
RSTC_SR
Address:
0xFFFFFD04
Access Type:
Read-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
SRCMP
16
NRSTL
15
–
14
–
13
–
12
–
11
–
10
9
RSTTYP
8
7
–
6
–
5
–
4
–
3
–
2
–
1
–
0
URSTS
• URSTS: User Reset Status
0 = No high-to-low edge on NRST happened since the last read of RSTC_SR.
1 = At least one high-to-low transition of NRST has been detected since the last read of RSTC_SR.
• RSTTYP: Reset Type
Reports the cause of the last processor reset. Reading this RSTC_SR does not reset this field.
RSTTYP
Reset Type
Comments
0
0
0
General Reset
Both VDDCORE and VDDBU rising
0
0
1
Wake Up Reset
VDDCORE rising
0
1
0
Watchdog Reset
Watchdog fault occurred
0
1
1
Software Reset
Processor reset required by the software
1
0
0
User Reset
NRST pin detected low
• NRSTL: NRST Pin Level
Registers the NRST Pin Level at Master Clock (MCK).
• SRCMP: Software Reset Command in Progress
0 = No software command is being performed by the reset controller. The reset controller is ready for a software command.
1 = A software reset command is being performed by the reset controller. The reset controller is busy.
82
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
12.5.3
Name:
Reset Controller Mode Register
RSTC_MR
Address:
0xFFFFFD08
Access Type:
Read-write
31
30
29
28
27
26
25
24
17
–
16
9
8
1
–
0
URSTEN
KEY
23
–
22
–
21
–
20
–
19
–
18
–
15
–
14
–
13
–
12
–
11
10
7
–
6
–
5
4
URSTIEN
3
–
ERSTL
2
–
• URSTEN: User Reset Enable
0 = The detection of a low level on the pin NRST does not generate a User Reset.
1 = The detection of a low level on the pin NRST triggers a User Reset.
• URSTIEN: User Reset Interrupt Enable
0 = USRTS bit in RSTC_SR at 1 has no effect on rstc_irq.
1 = USRTS bit in RSTC_SR at 1 asserts rstc_irq if URSTEN = 0.
• ERSTL: External Reset Length
This field defines the external reset length. The external reset is asserted during a time of 2(ERSTL+1) Slow Clock cycles. This
allows assertion duration to be programmed between 60 μs and 2 seconds.
• KEY: Password
Should be written at value 0xA5. Writing any other value in this field aborts the write operation.
83
6438D–ATARM–13-Oct-09
84
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
13. Real-time Timer (RTT)
13.1
Description
The Real-time Timer is built around a 32-bit counter and used to count elapsed seconds. It generates a periodic interrupt and/or triggers an alarm on a programmed value.
13.2
Embedded Characteristics
• Real-Time Timer, allowing backup of time with different accuracies
– 32-bit Free-running back-up Counter
– Integrates a 16-bit programmable prescaler running on slow clock
– Alarm Register capable to generate a wake-up of the system through the Shut Down
Controller
13.3
Block Diagram
Figure 13-1. Real-time Timer
RTT_MR
RTTRST
RTT_MR
RTPRES
RTT_MR
SLCK
RTTINCIEN
reload
16-bit
Divider
set
0
RTT_MR
RTTRST
RTTINC
RTT_SR
1
reset
0
rtt_int
32-bit
Counter
read
RTT_SR
RTT_MR
ALMIEN
RTT_VR
reset
CRTV
RTT_SR
ALMS
set
rtt_alarm
=
RTT_AR
13.4
ALMV
Functional Description
The Real-time Timer is used to count elapsed seconds. It is built around a 32-bit counter fed by
Slow Clock divided by a programmable 16-bit value. The value can be programmed in the field
RTPRES of the Real-time Mode Register (RTT_MR).
Programming RTPRES at 0x00008000 corresponds to feeding the real-time counter with a 1 Hz
signal (if the Slow Clock is 32.768 kHz). The 32-bit counter can count up to 232 seconds, corresponding to more than 136 years, then roll over to 0.
85
6438D–ATARM–13-Oct-09
The Real-time Timer can also be used as a free-running timer with a lower time-base. The best
accuracy is achieved by writing RTPRES to 3. Programming RTPRES to 1 or 2 is possible, but
may result in losing status events because the status register is cleared two Slow Clock cycles
after read. Thus if the RTT is configured to trigger an interrupt, the interrupt occurs during 2 Slow
Clock cycles after reading RTT_SR. To prevent several executions of the interrupt handler, the
interrupt must be disabled in the interrupt handler and re-enabled when the status register is
clear.
The Real-time Timer value (CRTV) can be read at any time in the register RTT_VR (Real-time
Value Register). As this value can be updated asynchronously from the Master Clock, it is advisable to read this register twice at the same value to improve accuracy of the returned value.
The current value of the counter is compared with the value written in the alarm register
RTT_AR (Real-time Alarm Register). If the counter value matches the alarm, the bit ALMS in
RTT_SR is set. The alarm register is set to its maximum value, corresponding to 0xFFFF_FFFF,
after a reset.
The bit RTTINC in RTT_SR is set each time the Real-time Timer counter is incremented. This bit
can be used to start a periodic interrupt, the period being one second when the RTPRES is programmed with 0x8000 and Slow Clock equal to 32.768 Hz.
Reading the RTT_SR status register resets the RTTINC and ALMS fields.
Writing the bit RTTRST in RTT_MR immediately reloads and restarts the clock divider with the
new programmed value. This also resets the 32-bit counter.
Note:
86
Because of the asynchronism between the Slow Clock (SCLK) and the System Clock (MCK):
1) The restart of the counter and the reset of the RTT_VR current value register is effective only 2
slow clock cycles after the write of the RTTRST bit in the RTT_MR register.
2) The status register flags reset is taken into account only 2 slow clock cycles after the read of the
RTT_SR (Status Register).
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 13-2. RTT Counting
APB cycle
APB cycle
SCLK
RTPRES - 1
Prescaler
0
RTT
0
...
ALMV-1
ALMV
ALMV+1
ALMV+2
ALMV+3
RTTINC (RTT_SR)
ALMS (RTT_SR)
APB Interface
read RTT_SR
87
6438D–ATARM–13-Oct-09
13.5
Real-time Timer (RTT) User Interface
Table 13-1.
Register Mapping
Offset
Register
Name
Access
Reset
0x00
Mode Register
RTT_MR
Read-write
0x0000_8000
0x04
Alarm Register
RTT_AR
Read-write
0xFFFF_FFFF
0x08
Value Register
RTT_VR
Read-only
0x0000_0000
0x0C
Status Register
RTT_SR
Read-only
0x0000_0000
88
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
13.5.1
Real-time Timer Mode Register
Register Name:
RTT_MR
Address:
0xFFFFFD20
Access Type:
Read/Write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
RTTRST
17
RTTINCIEN
16
ALMIEN
15
14
13
12
11
10
9
8
3
2
1
0
RTPRES
7
6
5
4
RTPRES
• RTPRES: Real-time Timer Prescaler Value
Defines the number of SLCK periods required to increment the Real-time timer. RTPRES is defined as follows:
RTPRES = 0: The prescaler period is equal to 216.
RTPRES ≠ 0: The prescaler period is equal to RTPRES.
• ALMIEN: Alarm Interrupt Enable
0 = The bit ALMS in RTT_SR has no effect on interrupt.
1 = The bit ALMS in RTT_SR asserts interrupt.
• RTTINCIEN: Real-time Timer Increment Interrupt Enable
0 = The bit RTTINC in RTT_SR has no effect on interrupt.
1 = The bit RTTINC in RTT_SR asserts interrupt.
• RTTRST: Real-time Timer Restart
1 = Reloads and restarts the clock divider with the new programmed value. This also resets the 32-bit counter.
89
6438D–ATARM–13-Oct-09
13.5.2
Real-time Timer Alarm Register
Register Name:
RTT_AR
Address:
0xFFFFFD24
Access Type:
Read/Write
31
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
0
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
0
ALMV
23
22
21
20
ALMV
15
14
13
12
ALMV
7
6
5
4
ALMV
• ALMV: Alarm Value
Defines the alarm value (ALMV+1) compared with the Real-time Timer.
13.5.3
Real-time Timer Value Register
Register Name:
RTT_VR
Address:
0xFFFFFD28
Access Type:
Read-only
31
30
29
28
CRTV
23
22
21
20
CRTV
15
14
13
12
CRTV
7
6
5
4
CRTV
• CRTV: Current Real-time Value
Returns the current value of the Real-time Timer.
90
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
13.5.4
Real-time Timer Status Register
Register Name:
RTT_SR
Address:
0xFFFFFD2C
Access Type:
Read-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
–
6
–
5
–
4
–
3
–
2
–
1
RTTINC
0
ALMS
• ALMS: Real-time Alarm Status
0 = The Real-time Alarm has not occurred since the last read of RTT_SR.
1 = The Real-time Alarm occurred since the last read of RTT_SR.
• RTTINC: Real-time Timer Increment
0 = The Real-time Timer has not been incremented since the last read of the RTT_SR.
1 = The Real-time Timer has been incremented since the last read of the RTT_SR.
91
6438D–ATARM–13-Oct-09
92
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
14. Real-time Clock (RTC)
14.1
Description
The Real-time Clock (RTC) peripheral is designed for very low power consumption.
It combines a complete time-of-day clock with alarm and a two-hundred-year Gregorian calendar, complemented by a programmable periodic interrupt. The alarm and calendar registers are
accessed by a 32-bit data bus.
The time and calendar values are coded in binary-coded decimal (BCD) format. The time format
can be 24-hour mode or 12-hour mode with an AM/PM indicator.
Updating time and calendar fields and configuring the alarm fields are performed by a parallel
capture on the 32-bit data bus. An entry control is performed to avoid loading registers with
incompatible BCD format data or with an incompatible date according to the current
month/year/century.
14.2
Embedded Characteristics
• Low power consumption
• Full asynchronous design
• Two hundred year calendar
• Programmable Periodic Interrupt
• Alarm and update parallel load
• Control of alarm and update Time/Calendar Data In
14.3
Block Diagram
Figure 14-1. RTC Block Diagram
Crystal Oscillator: SLCK
32768 Divider
Bus Interface
Bus Interface
Time
Date
Entry
Control
Interrupt
Control
RTC Interrupt
93
6438D–ATARM–13-Oct-09
14.4
Product Dependencies
14.4.1
Power Management
The Real-time Clock is continuously clocked at 32768 Hz. The Power Management Controller
has no effect on RTC behavior.
14.4.2
Interrupt
The RTC Interrupt is connected to interrupt source 1 (IRQ1) of the advanced interrupt controller.
This interrupt line is due to the OR-wiring of the system peripheral interrupt lines (System Timer,
Real Time Clock, Power Management Controller, Memory Controller, etc.). When a system
interrupt occurs, the service routine must first determine the cause of the interrupt. This is done
by reading the status registers of the above system peripherals successively.
14.5
Functional Description
The RTC provides a full binary-coded decimal (BCD) clock that includes century (19/20), year
(with leap years), month, date, day, hours, minutes and seconds.
The valid year range is 1900 to 2099, a two-hundred-year Gregorian calendar achieving full Y2K
compliance.
The RTC can operate in 24-hour mode or in 12-hour mode with an AM/PM indicator.
Corrections for leap years are included (all years divisible by 4 being leap years, including year
2000). This is correct up to the year 2099.
After hardware reset, the calendar is initialized to Thursday, January 1, 1998.
14.5.1
Reference Clock
The reference clock is Slow Clock (SLCK). It can be driven internally or by an external 32.768
kHz crystal.
During low power modes of the processor (idle mode), the oscillator runs and power consumption is critical. The crystal selection has to take into account the current consumption for power
saving and the frequency drift due to temperature effect on the circuit for time accuracy.
14.5.2
Timing
The RTC is updated in real time at one-second intervals in normal mode for the counters of seconds, at one-minute intervals for the counter of minutes and so on.
Due to the asynchronous operation of the RTC with respect to the rest of the chip, to be certain
that the value read in the RTC registers (century, year, month, date, day, hours, minutes, seconds) are valid and stable, it is necessary to read these registers twice. If the data is the same
both times, then it is valid. Therefore, a minimum of two and a maximum of three accesses are
required.
14.5.3
Alarm
The RTC has five programmable fields: month, date, hours, minutes and seconds.
Each of these fields can be enabled or disabled to match the alarm condition:
• If all the fields are enabled, an alarm flag is generated (the corresponding flag is asserted
and an interrupt generated if enabled) at a given month, date, hour/minute/second.
• If only the “seconds” field is enabled, then an alarm is generated every minute.
94
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Depending on the combination of fields enabled, a large number of possibilities are available to
the user ranging from minutes to 365/366 days.
14.5.4
Error Checking
Verification on user interface data is performed when accessing the century, year, month, date,
day, hours, minutes, seconds and alarms. A check is performed on illegal BCD entries such as
illegal date of the month with regard to the year and century configured.
If one of the time fields is not correct, the data is not loaded into the register/counter and a flag is
set in the validity register. The user can not reset this flag. It is reset as soon as an acceptable
value is programmed. This avoids any further side effects in the hardware. The same procedure
is done for the alarm.
The following checks are performed:
1. Century (check if it is in range 19 - 20)
2. Year (BCD entry check)
3. Date (check range 01 - 31)
4. Month (check if it is in BCD range 01 - 12, check validity regarding “date”)
5. Day (check range 1 - 7)
6. Hour (BCD checks: in 24-hour mode, check range 00 - 23 and check that AM/PM flag is
not set if RTC is set in 24-hour mode; in 12-hour mode check range 01 - 12)
7. Minute (check BCD and range 00 - 59)
8. Second (check BCD and range 00 - 59)
Note:
14.5.5
If the 12-hour mode is selected by means of the RTC_MODE register, a 12-hour value can be programmed and the returned value on RTC_TIME will be the corresponding 24-hour value. The
entry control checks the value of the AM/PM indicator (bit 22 of RTC_TIME register) to determine
the range to be checked.
Updating Time/Calendar
To update any of the time/calendar fields, the user must first stop the RTC by setting the corresponding field in the Control Register. Bit UPDTIM must be set to update time fields (hour,
minute, second) and bit UPDCAL must be set to update calendar fields (century, year, month,
date, day).
Then the user must poll or wait for the interrupt (if enabled) of bit ACKUPD in the Status Register. Once the bit reads 1, it is mandatory to clear this flag by writing the corresponding bit in
RTC_SCCR. The user can now write to the appropriate Time and Calendar register.
Once the update is finished, the user must reset (0) UPDTIM and/or UPDCAL in the Control
When entering programming mode of the calendar fields, the time fields remain enabled. When
entering the programming mode of the time fields, both time and calendar fields are stopped.
This is due to the location of the calendar logic circuity (downstream for low-power considerations). It is highly recommended to prepare all the fields to be updated before entering
programming mode. In successive update operations, the user must wait at least one second
after resetting the UPDTIM/UPDCAL bit in the RTC_CR (Control Register) before setting these
bits again. This is done by waiting for the SEC flag in the Status Register before setting
UPDTIM/UPDCAL bit. After resetting UPDTIM/UPDCAL, the SEC flag must also be cleared.
95
6438D–ATARM–13-Oct-09
Figure 14-2. Update Sequence
Begin
Prepare TIme or Calendar Fields
Set UPDTIM and/or UPDCAL
bit(s) in RTC_CR
Read RTC_SR
Polling or
IRQ (if enabled)
ACKUPD
=1?
No
Yes
Clear ACKUPD bit in RTC_SCCR
Update Time andor Calendar values in
RTC_TIMR/RTC_CALR
Clear UPDTIM and/or UPDCAL bit in
RTC_CR
End
96
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
14.6
Reset Controller (RTC) User Interface
Table 14-1.
Register Mapping
Offset
Register
Name
Access
Reset
0x00
Control Register
RTC_CR
Read-write
0x0
0x04
Mode Register
RTC_MR
Read-write
0x0
0x08
Time Register
RTC_TIMR
Read-write
0x0
0x0C
Calendar Register
RTC_CALR
Read-write
0x01819819
0x10
Time Alarm Register
RTC_TIMALR
Read-write
0x0
0x14
Calendar Alarm Register
RTC_CALALR
Read-write
0x01010000
0x18
Status Register
RTC_SR
Read-only
0x0
0x1C
Status Clear Command Register
RTC_SCCR
Write-only
---
0x20
Interrupt Enable Register
RTC_IER
Write-only
---
0x24
Interrupt Disable Register
RTC_IDR
Write-only
---
0x28
Interrupt Mask Register
RTC_IMR
Read-only
0x0
0x2C
Valid Entry Register
RTC_VER
Read-only
0x0
97
6438D–ATARM–13-Oct-09
14.6.1
Name:
RTC Control Register
RTC_CR
Address:
0xFFFFFDB0
Access Type: Read-write
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
–
–
–
–
–
–
15
14
13
12
11
10
–
–
–
–
–
–
16
CALEVSEL
9
8
TIMEVSEL
7
6
5
4
3
2
1
0
–
–
–
–
–
–
UPDCAL
UPDTIM
• UPDTIM: Update Request Time Register
0 = No effect.
1 = Stops the RTC time counting.
Time counting consists of second, minute and hour counters. Time counters can be programmed once this bit is set and
acknowledged by the bit ACKUPD of the Status Register.
• UPDCAL: Update Request Calendar Register
0 = No effect.
1 = Stops the RTC calendar counting.
Calendar counting consists of day, date, month, year and century counters. Calendar counters can be programmed once
this bit is set.
• TIMEVSEL: Time Event Selection
The event that generates the flag TIMEV in RTC_SR (Status Register) depends on the value of TIMEVSEL.
0 = Minute change.
1 = Hour change.
2 = Every day at midnight.
3 = Every day at noon.
• CALEVSEL: Calendar Event Selection
The event that generates the flag CALEV in RTC_SR depends on the value of CALEVSEL.
0 = Week change (every Monday at time 00:00:00).
1 = Month change (every 01 of each month at time 00:00:00).
2, 3 = Year change (every January 1 at time 00:00:00).
98
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
14.6.2
Name:
RTC Mode Register
RTC_MR
Address:
0xFFFFFDB4
Access Type: Read-write
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
–
–
–
–
–
–
–
HRMOD
• HRMOD: 12-/24-hour Mode
0 = 24-hour mode is selected.
1 = 12-hour mode is selected.
All non-significant bits read zero.
99
6438D–ATARM–13-Oct-09
14.6.3
Name:
RTC Time Register
RTC_TIMR
Address:
0xFFFFFDB8
Access Type: Read-write
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
AMPM
15
14
10
9
8
2
1
0
HOUR
13
12
–
7
11
MIN
6
5
–
4
3
SEC
• SEC: Current Second
The range that can be set is 0 - 59 (BCD).
The lowest four bits encode the units. The higher bits encode the tens.
• MIN: Current Minute
The range that can be set is 0 - 59 (BCD).
The lowest four bits encode the units. The higher bits encode the tens.
• HOUR: Current Hour
The range that can be set is 1 - 12 (BCD) in 12-hour mode or 0 - 23 (BCD) in 24-hour mode.
• AMPM: Ante Meridiem Post Meridiem Indicator
This bit is the AM/PM indicator in 12-hour mode.
0 = AM.
1 = PM.
All non-significant bits read zero.
100
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
14.6.4
Name:
RTC Calendar Register
RTC_CALR
Address:
0xFFFFFDBC
Access Type: Read-write
31
30
–
–
23
22
29
28
27
21
20
19
DAY
15
14
26
25
24
18
17
16
DATE
MONTH
13
12
11
10
9
8
3
2
1
0
YEAR
7
6
5
–
4
CENT
• CENT: Current Century
The range that can be set is 19 - 20 (BCD).
The lowest four bits encode the units. The higher bits encode the tens.
• YEAR: Current Year
The range that can be set is 00 - 99 (BCD).
The lowest four bits encode the units. The higher bits encode the tens.
• MONTH: Current Month
The range that can be set is 01 - 12 (BCD).
The lowest four bits encode the units. The higher bits encode the tens.
• DAY: Current Day in Current Week
The range that can be set is 1 - 7 (BCD).
The coding of the number (which number represents which day) is user-defined as it has no effect on the date counter.
• DATE: Current Day in Current Month
The range that can be set is 01 - 31 (BCD).
The lowest four bits encode the units. The higher bits encode the tens.
All non-significant bits read zero.
101
6438D–ATARM–13-Oct-09
14.6.5
Name:
RTC Time Alarm Register
RTC_TIMALR
Address:
0xFFFFFDC0
Access Type: Read-write
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
21
20
19
18
17
16
10
9
8
2
1
0
23
22
HOUREN
AMPM
15
14
HOUR
13
12
MINEN
7
11
MIN
6
5
SECEN
4
3
SEC
• SEC: Second Alarm
This field is the alarm field corresponding to the BCD-coded second counter.
• SECEN: Second Alarm Enable
0 = The second-matching alarm is disabled.
1 = The second-matching alarm is enabled.
• MIN: Minute Alarm
This field is the alarm field corresponding to the BCD-coded minute counter.
• MINEN: Minute Alarm Enable
0 = The minute-matching alarm is disabled.
1 = The minute-matching alarm is enabled.
• HOUR: Hour Alarm
This field is the alarm field corresponding to the BCD-coded hour counter.
• AMPM: AM/PM Indicator
This field is the alarm field corresponding to the BCD-coded hour counter.
• HOUREN: Hour Alarm Enable
0 = The hour-matching alarm is disabled.
1 = The hour-matching alarm is enabled.
102
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
14.6.6
Name:
RTC Calendar Alarm Register
RTC_CALALR
Address:
0xFFFFFDC4
Access Type: Read-write
31
30
DATEEN
–
29
28
27
26
25
24
18
17
16
DATE
23
22
21
MTHEN
–
–
20
19
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
MONTH
7
6
5
4
3
2
1
0
–
–
–
–
–
–
–
–
• MONTH: Month Alarm
This field is the alarm field corresponding to the BCD-coded month counter.
• MTHEN: Month Alarm Enable
0 = The month-matching alarm is disabled.
1 = The month-matching alarm is enabled.
• DATE: Date Alarm
This field is the alarm field corresponding to the BCD-coded date counter.
• DATEEN: Date Alarm Enable
0 = The date-matching alarm is disabled.
1 = The date-matching alarm is enabled.
103
6438D–ATARM–13-Oct-09
14.6.7
Name:
RTC Status Register
RTC_SR
Address:
0xFFFFFDC8
Access Type: Read-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
–
–
–
CALEV
TIMEV
SEC
ALARM
ACKUPD
• ACKUPD: Acknowledge for Update
0 = Time and calendar registers cannot be updated.
1 = Time and calendar registers can be updated.
• ALARM: Alarm Flag
0 = No alarm matching condition occurred.
1 = An alarm matching condition has occurred.
• SEC: Second Event
0 = No second event has occurred since the last clear.
1 = At least one second event has occurred since the last clear.
• TIMEV: Time Event
0 = No time event has occurred since the last clear.
1 = At least one time event has occurred since the last clear.
The time event is selected in the TIMEVSEL field in RTC_CTRL (Control Register) and can be any one of the following
events: minute change, hour change, noon, midnight (day change).
• CALEV: Calendar Event
0 = No calendar event has occurred since the last clear.
1 = At least one calendar event has occurred since the last clear.
The calendar event is selected in the CALEVSEL field in RTC_CR and can be any one of the following events: week
change, month change and year change.
104
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
14.6.8
Name:
RTC Status Clear Command Register
RTC_SCCR
Address:
0xFFFFFDCC
Access Type: Write-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
–
–
–
CALCLR
TIMCLR
SECCLR
ALRCLR
ACKCLR
• ACKCLR: Acknowledge Clear
0 = No effect.
1 = Clears corresponding status flag in the Status Register (RTC_SR).
• ALRCLR: Alarm Clear
0 = No effect.
1 = Clears corresponding status flag in the Status Register (RTC_SR).
• SECCLR: Second Clear
0 = No effect.
1 = Clears corresponding status flag in the Status Register (RTC_SR).
• TIMCLR: Time Clear
0 = No effect.
1 = Clears corresponding status flag in the Status Register (RTC_SR).
• CALCLR: Calendar Clear
0 = No effect.
1 = Clears corresponding status flag in the Status Register (RTC_SR).
105
6438D–ATARM–13-Oct-09
14.6.9
Name:
RTC Interrupt Enable Register
RTC_IER
Address:
0xFFFFFDD0
Access Type: Write-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
–
–
–
CALEN
TIMEN
SECEN
ALREN
ACKEN
• ACKEN: Acknowledge Update Interrupt Enable
0 = No effect.
1 = The acknowledge for update interrupt is enabled.
• ALREN: Alarm Interrupt Enable
0 = No effect.
1 = The alarm interrupt is enabled.
• SECEN: Second Event Interrupt Enable
0 = No effect.
1 = The second periodic interrupt is enabled.
• TIMEN: Time Event Interrupt Enable
0 = No effect.
1 = The selected time event interrupt is enabled.
• CALEN: Calendar Event Interrupt Enable
0 = No effect.
• 1 = The selected calendar event interrupt is enabled.
106
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
14.6.10
Name:
RTC Interrupt Disable Register
RTC_IDR
Address:
0xFFFFFDD4
Access Type: Write-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
–
–
–
CALDIS
TIMDIS
SECDIS
ALRDIS
ACKDIS
• ACKDIS: Acknowledge Update Interrupt Disable
0 = No effect.
1 = The acknowledge for update interrupt is disabled.
• ALRDIS: Alarm Interrupt Disable
0 = No effect.
1 = The alarm interrupt is disabled.
• SECDIS: Second Event Interrupt Disable
0 = No effect.
1 = The second periodic interrupt is disabled.
• TIMDIS: Time Event Interrupt Disable
0 = No effect.
1 = The selected time event interrupt is disabled.
• CALDIS: Calendar Event Interrupt Disable
0 = No effect.
1 = The selected calendar event interrupt is disabled.
107
6438D–ATARM–13-Oct-09
14.6.11
Name:
Address:
RTC Interrupt Mask Register
RTC_IMR
0xFFFFFDD8
Access Type: Read-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
–
–
–
CAL
TIM
SEC
ALR
ACK
• ACK: Acknowledge Update Interrupt Mask
0 = The acknowledge for update interrupt is disabled.
1 = The acknowledge for update interrupt is enabled.
• ALR: Alarm Interrupt Mask
0 = The alarm interrupt is disabled.
1 = The alarm interrupt is enabled.
• SEC: Second Event Interrupt Mask
0 = The second periodic interrupt is disabled.
1 = The second periodic interrupt is enabled.
• TIM: Time Event Interrupt Mask
0 = The selected time event interrupt is disabled.
1 = The selected time event interrupt is enabled.
• CAL: Calendar Event Interrupt Mask
0 = The selected calendar event interrupt is disabled.
1 = The selected calendar event interrupt is enabled.
108
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
14.6.12
Name:
RTC Valid Entry Register
RTC_VER
Address:
0xFFFFFDDC
Access Type:
Read-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
–
–
–
–
NVCALALR
NVTIMALR
NVCAL
NVTIM
• NVTIM: Non-valid Time
0 = No invalid data has been detected in RTC_TIMR (Time Register).
1 = RTC_TIMR has contained invalid data since it was last programmed.
• NVCAL: Non-valid Calendar
0 = No invalid data has been detected in RTC_CALR (Calendar Register).
1 = RTC_CALR has contained invalid data since it was last programmed.
• NVTIMALR: Non-valid Time Alarm
0 = No invalid data has been detected in RTC_TIMALR (Time Alarm Register).
1 = RTC_TIMALR has contained invalid data since it was last programmed.
• NVCALALR: Non-valid Calendar Alarm
0 = No invalid data has been detected in RTC_CALALR (Calendar Alarm Register).
1 = RTC_CALALR has contained invalid data since it was last programmed.
109
6438D–ATARM–13-Oct-09
110
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
15. Periodic Interval Timer (PIT)
15.1
Description
The Periodic Interval Timer (PIT) provides the operating system’s scheduler interrupt. It is
designed to offer maximum accuracy and efficient management, even for systems with long
response time.
15.2
Embedded Characteristics
• Includes a 20-bit Periodic Counter, with less than 1μs accuracy
• Includes a 12-bit Interval Overlay Counter
• Real Time OS or Linux/WinCE compliant tick generator
15.3
Block Diagram
Figure 15-1. Periodic Interval Timer
PIT_MR
PIV
=?
PIT_MR
PITIEN
set
0
PIT_SR
PITS
pit_irq
reset
0
MCK
Prescaler
15.4
0
0
1
12-bit
Adder
1
read PIT_PIVR
20-bit
Counter
MCK/16
CPIV
PIT_PIVR
CPIV
PIT_PIIR
PICNT
PICNT
Functional Description
The Periodic Interval Timer aims at providing periodic interrupts for use by operating systems.
The PIT provides a programmable overflow counter and a reset-on-read feature. It is built
around two counters: a 20-bit CPIV counter and a 12-bit PICNT counter. Both counters work at
Master Clock /16.
111
6438D–ATARM–13-Oct-09
The first 20-bit CPIV counter increments from 0 up to a programmable overflow value set in the
field PIV of the Mode Register (PIT_MR). When the counter CPIV reaches this value, it resets to
0 and increments the Periodic Interval Counter, PICNT. The status bit PITS in the Status Register (PIT_SR) rises and triggers an interrupt, provided the interrupt is enabled (PITIEN in
PIT_MR).
Writing a new PIV value in PIT_MR does not reset/restart the counters.
When CPIV and PICNT values are obtained by reading the Periodic Interval Value Register
(PIT_PIVR), the overflow counter (PICNT) is reset and the PITS is cleared, thus acknowledging
the interrupt. The value of PICNT gives the number of periodic intervals elapsed since the last
read of PIT_PIVR.
When CPIV and PICNT values are obtained by reading the Periodic Interval Image Register
(PIT_PIIR), there is no effect on the counters CPIV and PICNT, nor on the bit PITS. For example, a profiler can read PIT_PIIR without clearing any pending interrupt, whereas a timer
interrupt clears the interrupt by reading PIT_PIVR.
The PIT may be enabled/disabled using the PITEN bit in the PIT_MR register (disabled on
reset). The PITEN bit only becomes effective when the CPIV value is 0. Figure 15-2 illustrates
the PIT counting. After the PIT Enable bit is reset (PITEN= 0), the CPIV goes on counting until
the PIV value is reached, and is then reset. PIT restarts counting, only if the PITEN is set again.
The PIT is stopped when the core enters debug state.
Figure 15-2. Enabling/Disabling PIT with PITEN
APB cycle
APB cycle
MCK
15
restarts MCK Prescaler
MCK Prescaler 0
PITEN
CPIV
0
1
PICNT
PIV - 1
0
PIV
1
0
1
0
PITS (PIT_SR)
APB Interface
read PIT_PIVR
112
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
15.5
Periodic Interval Timer (PIT) User Interface
Table 15-1.
Register Mapping
Offset
Register
Name
Access
Reset
0x00
Mode Register
PIT_MR
Read-write
0x000F_FFFF
0x04
Status Register
PIT_SR
Read-only
0x0000_0000
0x08
Periodic Interval Value Register
PIT_PIVR
Read-only
0x0000_0000
0x0C
Periodic Interval Image Register
PIT_PIIR
Read-only
0x0000_0000
113
6438D–ATARM–13-Oct-09
15.5.1
Periodic Interval Timer Mode Register
Register Name:
PIT_MR
Address:
0xFFFFFD30
Access Type:
Read/Write
31
–
30
–
29
–
28
–
27
–
26
–
23
–
22
–
21
–
20
–
19
18
15
14
13
12
25
PITIEN
24
PITEN
17
16
PIV
11
10
9
8
3
2
1
0
PIV
7
6
5
4
PIV
• PIV: Periodic Interval Value
Defines the value compared with the primary 20-bit counter of the Periodic Interval Timer (CPIV). The period is equal to
(PIV + 1).
• PITEN: Period Interval Timer Enabled
0 = The Periodic Interval Timer is disabled when the PIV value is reached.
1 = The Periodic Interval Timer is enabled.
• PITIEN: Periodic Interval Timer Interrupt Enable
0 = The bit PITS in PIT_SR has no effect on interrupt.
1 = The bit PITS in PIT_SR asserts interrupt.
114
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
15.5.2
Periodic Interval Timer Status Register
Register Name:
PIT_SR
Address:
0xFFFFFD34
Access Type:
Read-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
–
6
–
5
–
4
–
3
–
2
–
1
–
0
PITS
• PITS: Periodic Interval Timer Status
0 = The Periodic Interval timer has not reached PIV since the last read of PIT_PIVR.
1 = The Periodic Interval timer has reached PIV since the last read of PIT_PIVR.
15.5.3
Periodic Interval Timer Value Register
Register Name:
PIT_PIVR
Address:
0xFFFFFD38
Access Type:
Read-only
31
30
29
28
27
26
25
24
19
18
17
16
PICNT
23
22
21
20
PICNT
15
14
CPIV
13
12
11
10
9
8
3
2
1
0
CPIV
7
6
5
4
CPIV
Reading this register clears PITS in PIT_SR.
• CPIV: Current Periodic Interval Value
Returns the current value of the periodic interval timer.
• PICNT: Periodic Interval Counter
Returns the number of occurrences of periodic intervals since the last read of PIT_PIVR.
115
6438D–ATARM–13-Oct-09
AT91SAM9G45
15.5.4
Periodic Interval Timer Image Register
Register Name:
PIT_PIIR
Address:
0xFFFFFD3C
Access Type:
Read-only
31
30
29
28
27
26
19
18
25
24
17
16
PICNT
23
22
21
20
PICNT
15
14
CPIV
13
12
11
10
9
8
3
2
1
0
CPIV
7
6
5
4
CPIV
• CPIV: Current Periodic Interval Value
Returns the current value of the periodic interval timer.
• PICNT: Periodic Interval Counter
Returns the number of occurrences of periodic intervals since the last read of PIT_PIVR.
116
6438D–ATARM–13-Oct-09
AT91SAM9G45
16. Watchdog Timer (WDT)
16.1
Description
The Watchdog Timer can be used to prevent system lock-up if the software becomes trapped in
a deadlock. It features a 12-bit down counter that allows a watchdog period of up to 16 seconds
(slow clock at 32.768 kHz). It can generate a general reset or a processor reset only. In addition,
it can be stopped while the processor is in debug mode or idle mode.
16.2
Embedded Characteristics
• 16-bit key-protected only-once-Programmable Counter
• Windowed, prevents the processor to be in a dead-lock on the watchdog access
16.3
Block Diagram
Figure 16-1. Watchdog Timer Block Diagram
write WDT_MR
WDT_MR
WDV
WDT_CR
WDRSTT
reload
1
0
12-bit Down
Counter
WDT_MR
WDD
reload
Current
Value
1/128
SLCK
<= WDD
WDT_MR
WDRSTEN
= 0
wdt_fault
(to Reset Controller)
set
set
read WDT_SR
or
reset
WDERR
reset
WDUNF
reset
wdt_int
WDFIEN
WDT_MR
117
6438D–ATARM–13-Oct-09
16.4
Functional Description
The Watchdog Timer can be used to prevent system lock-up if the software becomes trapped in
a deadlock. It is supplied with VDDCORE. It restarts with initial values on processor reset.
The Watchdog is built around a 12-bit down counter, which is loaded with the value defined in
the field WDV of the Mode Register (WDT_MR). The Watchdog Timer uses the Slow Clock
divided by 128 to establish the maximum Watchdog period to be 16 seconds (with a typical Slow
Clock of 32.768 kHz).
After a Processor Reset, the value of WDV is 0xFFF, corresponding to the maximum value of
the counter with the external reset generation enabled (field WDRSTEN at 1 after a Backup
Reset). This means that a default Watchdog is running at reset, i.e., at power-up. The user must
either disable it (by setting the WDDIS bit in WDT_MR) if he does not expect to use it or must
reprogram it to meet the maximum Watchdog period the application requires.
The Watchdog Mode Register (WDT_MR) can be written only once. Only a processor reset
resets it. Writing the WDT_MR register reloads the timer with the newly programmed mode
parameters.
In normal operation, the user reloads the Watchdog at regular intervals before the timer underflow occurs, by writing the Control Register (WDT_CR) with the bit WDRSTT to 1. The
Watchdog counter is then immediately reloaded from WDT_MR and restarted, and the Slow
Clock 128 divider is reset and restarted. The WDT_CR register is write-protected. As a result,
writing WDT_CR without the correct hard-coded key has no effect. If an underflow does occur,
the “wdt_fault” signal to the Reset Controller is asserted if the bit WDRSTEN is set in the Mode
Register (WDT_MR). Moreover, the bit WDUNF is set in the Watchdog Status Register
(WDT_SR).
To prevent a software deadlock that continuously triggers the Watchdog, the reload of the
Watchdog must occur while the Watchdog counter is within a window between 0 and WDD,
WDD is defined in the WatchDog Mode Register WDT_MR.
Any attempt to restart the Watchdog while the Watchdog counter is between WDV and WDD
results in a Watchdog error, even if the Watchdog is disabled. The bit WDERR is updated in the
WDT_SR and the “wdt_fault” signal to the Reset Controller is asserted.
Note that this feature can be disabled by programming a WDD value greater than or equal to the
WDV value. In such a configuration, restarting the Watchdog Timer is permitted in the whole
range [0; WDV] and does not generate an error. This is the default configuration on reset (the
WDD and WDV values are equal).
The status bits WDUNF (Watchdog Underflow) and WDERR (Watchdog Error) trigger an interrupt, provided the bit WDFIEN is set in the mode register. The signal “wdt_fault” to the reset
controller causes a Watchdog reset if the WDRSTEN bit is set as already explained in the reset
controller programmer Datasheet. In that case, the processor and the Watchdog Timer are
reset, and the WDERR and WDUNF flags are reset.
If a reset is generated or if WDT_SR is read, the status bits are reset, the interrupt is cleared,
and the “wdt_fault” signal to the reset controller is deasserted.
Writing the WDT_MR reloads and restarts the down counter.
While the processor is in debug state or in idle mode, the counter may be stopped depending on
the value programmed for the bits WDIDLEHLT and WDDBGHLT in the WDT_MR.
118
AT91SAM9G45
6438D–ATARM–13-Oct-09
Figure 16-2. Watchdog Behavior
Watchdog Error
Watchdog Underflow
if WDRSTEN is 1
FFF
Normal behavior
if WDRSTEN is 0
WDV
Forbidden
Window
WDD
Permitted
Window
0
Watchdog
Fault
119
WDT_CR = WDRSTT
AT91SAM9G45
6438D–ATARM–13-Oct-09
16.5
Watchdog Timer (WDT) User Interface
Table 16-1.
Register Mapping
Offset
Register
Name
Access
Reset
0x00
Control Register
WDT_CR
Write-only
-
0x04
Mode Register
WDT_MR
Read-write Once
0x3FFF_2FFF
0x08
Status Register
WDT_SR
Read-only
0x0000_0000
16.5.1
Watchdog Timer Control Register
Register Name:
WDT_CR
Address:
0xFFFFFD40
Access Type:
31
Write-only
30
29
28
27
26
25
24
KEY
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
–
6
–
5
–
4
–
3
–
2
–
1
–
0
WDRSTT
• WDRSTT: Watchdog Restart
0: No effect.
1: Restarts the Watchdog.
• KEY: Password
Should be written at value 0xA5. Writing any other value in this field aborts the write operation.
120
AT91SAM9G45
6438D–ATARM–13-Oct-09
16.5.2
Watchdog Timer Mode Register
Register Name:
WDT_MR
Address:
0xFFFFFD44
Access Type:
Read-write Once
31
30
23
29
WDIDLEHLT
28
WDDBGHLT
27
21
20
19
11
22
26
25
24
18
17
16
10
9
8
1
0
WDD
WDD
15
WDDIS
14
13
12
WDRPROC
WDRSTEN
WDFIEN
7
6
5
4
WDV
3
2
WDV
• WDV: Watchdog Counter Value
Defines the value loaded in the 12-bit Watchdog Counter.
• WDFIEN: Watchdog Fault Interrupt Enable
0: A Watchdog fault (underflow or error) has no effect on interrupt.
1: A Watchdog fault (underflow or error) asserts interrupt.
• WDRSTEN: Watchdog Reset Enable
0: A Watchdog fault (underflow or error) has no effect on the resets.
1: A Watchdog fault (underflow or error) triggers a Watchdog reset.
• WDRPROC: Watchdog Reset Processor
0: If WDRSTEN is 1, a Watchdog fault (underflow or error) activates all resets.
1: If WDRSTEN is 1, a Watchdog fault (underflow or error) activates the processor reset.
• WDD: Watchdog Delta Value
Defines the permitted range for reloading the Watchdog Timer.
If the Watchdog Timer value is less than or equal to WDD, writing WDT_CR with WDRSTT = 1 restarts the timer.
If the Watchdog Timer value is greater than WDD, writing WDT_CR with WDRSTT = 1 causes a Watchdog error.
• WDDBGHLT: Watchdog Debug Halt
0: The Watchdog runs when the processor is in debug state.
1: The Watchdog stops when the processor is in debug state.
• WDIDLEHLT: Watchdog Idle Halt
0: The Watchdog runs when the system is in idle mode.
1: The Watchdog stops when the system is in idle state.
• WDDIS: Watchdog Disable
0: Enables the Watchdog Timer.
1: Disables the Watchdog Timer.
121
AT91SAM9G45
6438D–ATARM–13-Oct-09
16.5.3
Watchdog Timer Status Register
Register Name:
WDT_SR
Address:
0xFFFFFD48
Access Type:
Read-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
–
6
–
5
–
4
–
3
–
2
–
1
WDERR
0
WDUNF
• WDUNF: Watchdog Underflow
0: No Watchdog underflow occurred since the last read of WDT_SR.
1: At least one Watchdog underflow occurred since the last read of WDT_SR.
• WDERR: Watchdog Error
0: No Watchdog error occurred since the last read of WDT_SR.
1: At least one Watchdog error occurred since the last read of WDT_SR.
122
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
17. Shutdown Controller (SHDWC)
17.1
Description
The Shutdown Controller controls the power supplies VDDIO and VDDCORE and the wake-up
detection on debounced input lines.
17.2
Embedded Characteristics
The Shut Down Controller is supplied on VDDBU and allows a software-controllable shut down
of the system through the pin SHDN. An input change of the WKUP pin or an alarm releases the
SHDN pin, and thus wakes up the system power supply.
17.3
Block Diagram
Figure 17-1. Shutdown Controller Block Diagram
SLCK
Shutdown Controller
read SHDW_SR
SHDW_MR
CPTWK0
reset
WAKEUP0 SHDW_SR
WKMODE0
set
WKUP0
read SHDW_SR
Wake-up
reset
RTTWKEN
RTT Alarm
SHDW_MR
RTTWK
SHDW_SR
set
SHDW_CR
SHDW
Shutdown
Output
Controller
SHDN
Shutdown
123
6438D–ATARM–13-Oct-09
Figure 17-2. Shutdown Controller Block Diagram
SLCK
Shutdown Controller
SHDW_MR
read SHDW_SR
CPTWK0
reset
WAKEUP0 SHDW_SR
WKMODE0
set
WKUP0
read SHDW_SR
Wake-up
reset
RTTWKEN
SHDW_MR
RTT Alarm
RTTWK
Shutdown
Output
Controller
SHDW_SR
set
SHDN
SHDW_CR
read SHDW_SR
SHDW
Shutdown
reset
RTCWKEN
SHDW_MR
RTC Alarm
17.4
RTCWK
SHDW_SR
set
I/O Lines Description
Table 17-1.
I/O Lines Description
Name
Description
Type
WKUP0
Wake-up 0 input
Input
SHDN
Shutdown output
Output
17.5
17.5.1
124
Product Dependencies
Power Management
The Shutdown Controller is continuously clocked by Slow Clock. The Power Management Controller has no effect on the behavior of the Shutdown Controller.
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
17.6
Functional Description
The Shutdown Controller manages the main power supply. To do so, it is supplied with VDDBU
and manages wake-up input pins and one output pin, SHDN.
A typical application connects the pin SHDN to the shutdown input of the DC/DC Converter providing the main power supplies of the system, and especially VDDCORE and/or VDDIO. The
wake-up inputs (WKUP0) connect to any push-buttons or signal that wake up the system.
The software is able to control the pin SHDN by writing the Shutdown Control Register
(SHDW_CR) with the bit SHDW at 1. The shutdown is taken into account only 2 slow clock
cycles after the write of SHDW_CR. This register is password-protected and so the value written
should contain the correct key for the command to be taken into account. As a result, the system
should be powered down.
A level change on WKUP0 is used as wake-up. Wake-up is configured in the Shutdown Mode
Register (SHDW_MR). The transition detector can be programmed to detect either a positive or
negative transition or any level change on WKUP0. The detection can also be disabled. Programming is performed by defining WKMODE0.
Moreover, a debouncing circuit can be programmed for WKUP0. The debouncing circuit filters
pulses on WKUP0 shorter than the programmed number of 16 SLCK cycles in CPTWK0 of the
SHDW_MR register. If the programmed level change is detected on a pin, a counter starts.
When the counter reaches the value programmed in the corresponding field, CPTWK0, the
SHDN pin is released. If a new input change is detected before the counter reaches the corresponding value, the counter is stopped and cleared. WAKEUP0 of the Status Register
(SHDW_SR) reports the detection of the programmed events on WKUP0 with a reset after the
read of SHDW_SR.
The Shutdown Controller can be programmed so as to activate the wake-up using the RTT
alarm (the detection of the rising edge of the RTT alarm is synchronized with SLCK). This is
done by writing the SHDW_MR register using the RTTWKEN fields. When enabled, the detection of the RTT alarm is reported in the RTTWK bit of the SHDW_SR Status register. It is reset
after the read of SHDW_SR. When using the RTT alarm to wake up the system, the user must
ensure that the RTT alarm status flag is cleared before shutting down the system. Otherwise, no
rising edge of the status flag may be detected and the wake-up fails.
The Shutdown Controller can be programmed so as to activate the wake-up using the RTC
alarm (the detection of the rising edge of the RTC alarm is synchronized with SLCK). This is
done by writing the SHDW_MR register using the RTCWKEN field. When enabled, the detection
of the RTC alarm is reported in the RTCWK bit of the SHDW_SR Status register. It is reset after
the read of SHDW_SR. When using the RTC alarm to wake up the system, the user must
ensure that the RTC alarm status flag is cleared before shutting down the system. Otherwise, no
rising edge of the status flag may be detected and the wake-up fails fail.
125
6438D–ATARM–13-Oct-09
17.7
Shutdown Controller (SHDWC) User Interface
Table 17-2.
Register Mapping
Offset
Register
Name
Access
Reset
0x00
Shutdown Control Register
SHDW_CR
Write-only
-
0x04
Shutdown Mode Register
SHDW_MR
Read-write
0x0000_0003
0x08
Shutdown Status Register
SHDW_SR
Read-only
0x0000_0000
17.7.1
Shutdown Control Register
Register Name:
SHDW_CR
Address:
0xFFFFFD10
Access Type:
Write-only
31
30
29
28
27
26
25
24
KEY
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
–
6
–
5
–
4
–
3
–
2
–
1
–
0
SHDW
• SHDW: Shutdown Command
0 = No effect.
1 = If KEY is correct, asserts the SHDN pin.
• KEY: Password
Should be written at value 0xA5. Writing any other value in this field aborts the write operation.
126
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
17.7.2
Shutdown Mode Register
Register Name:
SHDW_MR
Address:
0xFFFFFD14
Access Type:
Read/Write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
RTCWKEN
16
RTTWKEN
15
14
13
12
11
–
10
–
9
3
–
2
–
1
–
7
6
5
4
CPTWK0
8
–
0
WKMODE0
• WKMODE0: Wake-up Mode 0
WKMODE[1:0]
Wake-up Input Transition Selection
0
0
None. No detection is performed on the wake-up input
0
1
Low to high level
1
0
High to low level
1
1
Both levels change
• CPTWK0: Counter on Wake-up 0
Defines the number of 16 Slow Clock cycles, the level detection on the corresponding input pin shall last before the wakeup event occurs. Because of the internal synchronization of WKUP0, the SHDN pin is released
(CPTWK x 16 + 1) Slow Clock cycles after the event on WKUP.
• RTTWKEN: Real-time Timer Wake-up Enable
0 = The RTT Alarm signal has no effect on the Shutdown Controller.
1 = The RTT Alarm signal forces the de-assertion of the SHDN pin.
• RTCWKEN: Real-time Clock Wake-up Enable
0 = The RTC Alarm signal has no effect on the Shutdown Controller.
1 = The RTC Alarm signal forces the de-assertion of the SHDN pin.
127
6438D–ATARM–13-Oct-09
17.7.3
Shutdown Status Register
Register Name:
SHDW_SR
Address:
0xFFFFFD18
Access Type:
Read-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
RTCWK
16
RTTWK
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
–
6
–
5
–
4
–
3
–
2
–
1
–
0
WAKEUP0
• WAKEUP0: Wake-up 0 Status
0 = No wake-up event occurred on the corresponding wake-up input since the last read of SHDW_SR.
1 = At least one wake-up event occurred on the corresponding wake-up input since the last read of SHDW_SR.
• RTTWK: Real-time Timer Wake-up
0 = No wake-up alarm from the RTT occurred since the last read of SHDW_SR.
1 = At least one wake-up alarm from the RTT occurred since the last read of SHDW_SR.
• RTCWK: Real-time Clock Wake-up
0 = No wake-up alarm from the RTC occurred since the last read of SHDW_SR.
1 = At least one wake-up alarm from the RTC occurred since the last read of SHDW_SR.
128
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
18. General Purpose Backup Registers (GPBR)
18.1
Description
The System Controller embeds Four general-purpose backup registers.
18.2
Embedded Characteristics
• Four 32-bit general-purpose backup registers
18.3
General Purpose Backup Registers (GPBR) User Interface
Table 18-1.
Register Mapping
Offset
0x0
...
0xc
Register
Name
General Purpose Backup Register 0
SYS_GPBR0
...
...
General Purpose Backup Register 3
SYS_GPBR3
Access
Reset
Read-write
–
...
...
Read-write
–
129
6438D–ATARM–13-Oct-09
18.3.0.1
Name:
General Purpose Backup Register x
SYS_GPBRx
Addresses:
0xFFFFFD60 [0], 0xFFFFFD64 [1], 0xFFFFFD68 [2], 0xFFFFFD6C [3]
Type:
Read-write
31
30
29
28
27
26
25
24
18
17
16
10
9
8
2
1
0
GPBR_VALUEx
23
22
21
20
19
GPBR_VALUEx
15
14
13
12
11
GPBR_VALUEx
7
6
5
4
3
GPBR_VALUEx
• GPBR_VALUEx: Value of GPBR x
130
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
19. Bus Matrix (MATRIX)
19.1
Description
The Bus Matrix implements a multi-layer AHB, based on the AHB-Lite protocol, that enables parallel access paths between multiple AHB masters and slaves in a system, thus increasing the
overall bandwidth. The Bus Matrix interconnects up to 16 AHB masters to up to 16 AHB slaves.
The normal latency to connect a master to a slave is one cycle except for the default master of
the accessed slave which is connected directly (zero cycle latency).
The Bus Matrix user interface is compliant with ARM Advanced Peripheral Bus and provides a
Chip Configuration User Interface with Registers that allow the Bus Matrix to support application
specific features.
19.2
Embedded Characteristics
• 11-layer Matrix, handling requests from 11 masters
• Programmable Arbitration strategy
– Fixed-priority Arbitration
– Round-Robin Arbitration, either with no default master, last accessed default master
or fixed default master
• Burst Management
– Breaking with Slot Cycle Limit Support
– Undefined Burst Length Support
• One Address Decoder provided per Master
– Three different slaves may be assigned to each decoded memory area: one for
internal ROM boot, one for internal flash boot, one after remap
• Boot Mode Select
– Non-volatile Boot Memory can be internal ROM or external memory on EBI_NCS0
– Selection is made by General purpose NVM bit sampled at reset
• Remap Command
– Allows Remapping of an Internal SRAM in Place of the Boot Non-Volatile Memory
(ROM or External Flash)
– Allows Handling of Dynamic Exception Vectors
19.2.1
Matrix Masters
The Bus Matrix of the AT91SAM9G45 manages Masters, thus each master can perform an
access concurrently with others, depending on whether the slave it accesses is available.
Each Master has its own decoder, which can be defined specifically for each master. In order to
simplify the addressing, all the masters have the same decodings.
131
6438D–ATARM–13-Oct-09
Table 19-1.
19.2.2
List of Bus Matrix Masters
Master 0
ARM926™ Instruction
Master 1
ARM926 Data
Master 2
Peripheral DMA Controller (PDC)
Master 3
USB HOST OHCI
Master 4
DMA
Master 5
DMA
Master 6
ISI Controller DMA
Master 7
LCD DMA
Master 8
Ethernet MAC DMA
Master 9
USB Device High Speed DMA
Master 10
USB Host High Speed EHCI DMA
Matrix Slaves
Each Slave has its own arbiter, thus allowing a different arbitration per Slave to be programmed.
Table 19-2.
Slave 0
List of Bus Matrix Slaves
Internal SRAM
Internal ROM
USB OHCI
Slave 1
USB EHCI
UDP High Speed RAM
LCD User Interface
19.2.3
132
Slave 2
DDR Port 0
Slave 3
DDR Port 1
Slave 4
DDR Port 2
Slave 5
DDR Port 3
Slave 6
External Bus Interface
Slave 7
Internal Peripherals
Masters to Slaves Access
All the Masters can normally access all the Slaves. However, some paths do not make sense,
such as allowing access from the Ethernet MAC to the internal peripherals. Thus, these paths
are forbidden or simply not wired, and shown as “-” in the following tables.
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Table 19-3.
AT91SAM9G45 Masters to Slaves Access
Master
0
1
2
3
6
7
8
9
10
ARM
ARM
926
Instr.
926
Data
PDC
USB
Host
OHCI
ISI
LCD
DMA
DMA
DMA
Ethernet
MAC
USB
Device
HS
USB
Host
EHCI
Internal SRAM 0
X
X
X
X
X
X
-
X
X
X
Internal ROM
X
X
X
-
-
-
-
-
X
-
UHP OHCI
X
X
-
-
-
-
-
-
-
-
UHP EHCI
X
X
-
-
-
-
-
-
-
-
LCD User Int.
X
X
-
-
-
-
-
-
-
-
1
UDPHS RAM
X
X
-
-
-
-
-
-
-
-
2
DDR Port 0
X
-
-
-
-
-
-
-
-
-
3
DDR Port 1
-
X
-
-
-
-
-
-
-
-
4
DDR Port 2
-
-
X
X
X
X
-
X
X
X
5
DDR Port 3
-
-
X
X
X
X
X
X
X
X
6
EBI
X
X
X
X
X
X
X
X
X
X
7
Internal Periph.
X
X
X
-
X
-
-
-
-
-
Slave
0
4&5
Table 19-4 summarizes the Slave Memory Mapping for each connected Master, depending on
the Remap status (RCBx bit in Bus Matrix Master Remap Control Register MATRIX_MRCR) and
the BMS state at reset.
Table 19-4.
Internal Memory Mapping
Master
19.3
RCBx = 0
Slave
Base Address
BMS = 1
BMS = 0
0x0000 0000
Internal ROM
EBI NCS0
RCBx = 1
Internal SRAM
Memory Mapping
The Bus Matrix provides one decoder for every AHB master interface. The decoder offers each
AHB master several memory mappings. In fact, depending on the product, each memory area
may be assigned to several slaves. Booting at the same address while using different AHB
slaves (i.e. external RAM, internal ROM or internal Flash, etc.) becomes possible.
The Bus Matrix user interface provides Master Remap Control Register (MATRIX_MRCR) that
performs remap action for every master independently.
19.4
Special Bus Granting Mechanism
The Bus Matrix provides some speculative bus granting techniques in order to anticipate access
requests from some masters. This mechanism reduces latency at first access of a burst or single
transfer as long as the slave is free from any other master access, but does not provide any benefit as soon as the slave is continuously accessed by more than one master, since arbitration is
pipelined and then has no negative effect on the slave bandwidth or access latency.
This bus granting mechanism sets a different default master for every slave.
133
6438D–ATARM–13-Oct-09
At the end of the current access, if no other request is pending, the slave remains connected to
its associated default master. A slave can be associated with three kinds of default masters: no
default master, last access master and fixed default master.
To change from one kind of default master to another, the Bus Matrix user interface provides the
Slave Configuration Registers, one for each slave, that set a default master for each slave. The
Slave Configuration Register contains two fields: DEFMSTR_TYPE and FIXED_DEFMSTR. The
2-bit DEFMSTR_TYPE field selects the default master type (no default, last access master, fixed
default master), whereas the 4-bit FIXED_DEFMSTR field selects a fixed default master provided that DEFMSTR_TYPE is set to fixed default master. Please refer to Section 19.7.2 “Bus
Matrix Slave Configuration Registers” on page 142.
19.4.1
No Default Master
After the end of the current access, if no other request is pending, the slave is disconnected from
all masters. No Default Master suits low-power mode.
This configuration incurs one latency clock cycle for the first access of a burst after bus Idle.
Arbitration without default master may be used for masters that perform significant bursts or several transfers with no Idle in between, or if the slave bus bandwidth is widely used by one or
more masters.
This configuration provides no benefit on access latency or bandwidth when reaching maximum
slave bus throughput whatever is the number of requesting masters.
19.4.2
Last Access Master
After the end of the current access, if no other request is pending, the slave remains connected
to the last master that performed an access request.
This allows the Bus Matrix to remove the one latency cycle for the last master that accessed the
slave. Other non privileged masters still get one latency clock cycle if they want to access the
same slave. This technique is useful for masters that mainly perform single accesses or short
bursts with some Idle cycles in between.
This configuration provides no benefit on access latency or bandwidth when reaching maximum
slave bus throughput whatever is the number of requesting masters.
19.4.3
Fixed Default Master
After the end of the current access, if no other request is pending, the slave connects to its fixed
default master. Unlike last access master, the fixed master does not change unless the user
modifies it by a software action (field FIXED_DEFMSTR of the related MATRIX_SCFG).
This allows the Bus Matrix arbiters to remove the one latency clock cycle for the fixed default
master of the slave. Every request attempted by this fixed default master will not cause any arbitration latency whereas other non privileged masters will still get one latency cycle. This
technique is useful for a master that mainly perform single accesses or short bursts with some
Idle cycles in between.
This configuration provides no benefit on access latency or bandwidth when reaching maximum
slave bus throughput whatever is the number of requesting masters.
134
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
19.5
Arbitration
The Bus Matrix provides an arbitration mechanism that reduces latency when conflict cases
occur, i.e. when two or more masters try to access the same slave at the same time. One arbiter
per AHB slave is provided, thus arbitrating each slave differently.
The Bus Matrix provides the user with the possibility of choosing between 2 arbitration types or
mixing them for each slave:
1. Round-Robin Arbitration (default)
2. Fixed Priority Arbitration
The resulting algorithm may be complemented by selecting a default master configuration for
each slave.
When a re-arbitration must be done, specific conditions apply. See Section 19.5.1 “Arbitration
Scheduling” on page 135.
19.5.1
Arbitration Scheduling
Each arbiter has the ability to arbitrate between two or more different master requests. In order
to avoid burst breaking and also to provide the maximum throughput for slave interfaces, arbitration may only take place during the following cycles:
1. Idle Cycles: When a slave is not connected to any master or is connected to a master
which is not currently accessing it.
2. Single Cycles: When a slave is currently doing a single access.
3. End of Burst Cycles: When the current cycle is the last cycle of a burst transfer. For
defined length burst, predicted end of burst matches the size of the transfer but is managed differently for undefined length burst. See “Undefined Length Burst Arbitration” on
page 135
4. Slot Cycle Limit: When the slot cycle counter has reached the limit value indicating that
the current master access is too long and must be broken. See “Slot Cycle Limit Arbitration” on page 136
19.5.1.1
Undefined Length Burst Arbitration
In order to optimize AHB burst lengths and arbitration, it may be interesting to set a maximum for
undefined length bursts (INCR). The Bus Matrix provides specific logic in order to re-arbitrate
before the end of the INCR transfer. A predicted end of burst is used as a defined length burst
transfer and can be selected from among the following Undefined Length Burst Type (ULBT)
possibilities:
1. Unlimited: No predicted end of burst is generated and therefore INCR burst transfer will
not be broken by this way, but will be able to complete unless broken at the Slot Cycle
Limit. This is normally the default and should be let as is in order to be able to allow full
1 Kilobyte AHB intra-boundary 256-beat word bursts performed by some ATMEL AHB
masters.
2. 1-beat bursts: Predicted end of burst is generated at each single transfer inside the
INCR transfer.
3. 4-beat bursts: Predicted end of burst is generated at the end of each 4-beat boundary
inside INCR transfer.
4. 8-beat bursts: Predicted end of burst is generated at the end of each 8-beat boundary
inside INCR transfer.
5. 16-beat bursts: Predicted end of burst is generated at the end of each 16-beat boundary inside INCR transfer.
135
6438D–ATARM–13-Oct-09
AT91SAM9G45
6. 32-beat bursts: Predicted end of burst is generated at the end of each 32-beat boundary inside INCR transfer.
7. 64-beat bursts: Predicted end of burst is generated at the end of each 64-beat boundary inside INCR transfer.
8. 128-beat bursts: Predicted end of burst is generated at the end of each 128-beat
boundary inside INCR transfer.
Use of undefined length 16-beat bursts or less is discouraged since this generally decreases
significantly overall bus bandwidth due to arbitration and slave latencies at each first access of a
burst.
If the master does not permanently and continuously request the same slave or has an intrinsically limited average throughput, the ULBT should be let at its default unlimited value, knowing
that the AHB specification natively limits all word bursts to 256 beats and double-word bursts to
128 beats because of its 1 Kilobyte address boundaries.
Unless duly needed the ULBT should be let to its default 0 value for power saving.
This selection can be done through the field ULBT of the Master Configuration Registers
(MATRIX_MCFG).
19.5.1.2
Slot Cycle Limit Arbitration
The Bus Matrix contains specific logic to break long accesses, such as back to back undefined
length bursts or very long bursts on a very slow slave (e.g., an external low speed memory). At
each arbitration time a counter is loaded with the value previously written in the SLOT_CYCLE
field of the related Slave Configuration Register (MATRIX_SCFG) and decreased at each clock
cycle. When the counter elapses, the arbiter has the ability to re-arbitrate at the end of the current AHB bus access cycle.
Unless some master has a very tight access latency constraint which could lead to data overflow
or underflow due to a badly undersized internal fifo with respect to its throughput, the Slot Cycle
Limit should be disabled (SLOT_CYCLE = 0) or let to its default maximum value in order not to
inefficiently break long bursts performed by some ATMEL masters.
However, the Slot Cycle Limit should not be disabled in the very particular case of a master
capable of accessing the slave by performing back to back undefined length bursts shorter than
the number of ULBT beats with no Idle cycle in between, since in this case the arbitration could
be frozen all along the bursts sequence.
In most cases this feature is not needed and should be disabled for power saving.
Warning: This feature cannot prevent any slave from locking its access indefinitely.
19.5.2
Arbitration Priority Scheme
The bus Matrix arbitration scheme is organized in priority pools.
Round-Robin priority is used inside the highest and lowest priority pools, whereas fix level priority is used between priority pools and inside the intermediate priority pools.
For each slave, each master x is assigned to one of the slave priority pools through the Priority
Registers for Slaves (MxPR fields of MATRIX_PRAS and MATRIX_PRBS). When evaluating
masters requests, this programmed priority level always takes precedence.
After reset, all the masters are belonging to the lowest priority pool (MxPR = 0) and so are
granted bus access in a true Round-Robin fashion.
136
6438D–ATARM–13-Oct-09
AT91SAM9G45
The highest priority pool must be specifically reserved for masters requiring very low access
latency. If more than one master belong to this pool, these will be granted bus access in a
biased Round-Robin fashion which allow tight and deterministic maximum access latency from
AHB bus request. In fact, at worst, any currently high priority master request will be granted after
the current bus master access is ended and the other high priority pool masters, if any, have
been granted once each.
The lowest priority pool shares the remaining bus bandwidth between AHB Masters.
Intermediate priority pools allow fine priority tuning. Typically, a moderately latency critical master or a bandwidth only critical master will use such a priority level. The higher the priority level
(MxPR value), the higher the master priority.
All combination of MxPR values are allowed for all masters and slaves. For example some masters might be assigned to the highest priority pool (round-robin) and the remaining masters to
the lowest priority pool (round-robin), with no master for intermediate fix priority levels.
If more than one master is requesting the slave bus, whatever are the respective masters priorities, no master will be granted the slave bus for two consecutive runs. A master can only get
back to back grants as long as it is the only requesting master.
19.5.2.1
Fixed Priority Arbitration
This arbitration algorithm is the first and only applied between masters from distinct priority
pools. It is also used inside priority pools other than the highest and lowest ones (intermediate
priority pools).
It allows the Bus Matrix arbiters to dispatch the requests from different masters to the same
slave by using the fixed priority defined by the user in the MxPR field for each master inside the
MATRIX_PRAS and MATRIX_PRBS Priority Registers. If two or more master requests are
active at the same time, the master with the highest priority number MxPR is serviced first.
Inside intermediate priority pools, if two or more master requests with the same priority are
active at the same time, the master with the highest number is serviced first.
19.5.2.2
Round-Robin Arbitration
This algorithm is only used inside the highest and lowest priority pools. It allows the Bus Matrix
arbiters to dispatch the requests from different masters to the same slave in a fair way. If two or
more master requests are active at the same time inside the priority pool, they are serviced in a
round-robin increasing master number order.
137
6438D–ATARM–13-Oct-09
AT91SAM9G45
19.6
Write Protect Registers
To prevent any single software error that may corrupt MATRIX behavior, the entire MATRIX
address space from address offset 0x000 to 0x1FC can be write-protected by setting the WPEN
bit in the MATRIX Write Protect Mode Register (MATRIX_WPMR).
If a write access to anywhere in the MATRIX address space from address offset 0x000 to 0x1FC
is detected, then the WPVS flag in the MATRIX Write Protect Status Register (MATRIX_WPSR)
is set and the field WPVSRC indicates in which register the write access has been attempted.
The WPVS flag is reset by writing the MATRIX Write Protect Mode Register (MATRIX_WPMR)
with the appropriate access key WPKEY.
The protected registers are:
“Bus Matrix Master Configuration Registers”
“Bus Matrix Slave Configuration Registers”
“Bus Matrix Priority Registers A For Slaves”
“Bus Matrix Priority Registers B For Slaves”
“Bus Matrix Master Remap Control Register”
138
6438D–ATARM–13-Oct-09
AT91SAM9G45
19.7
Bus Matrix (MATRIX) User Interface
Table 19-5.
Register Mapping
Offset
Register
Name
Access
Reset
0x0000
Master Configuration Register 0
MATRIX_MCFG0
Read-write
0x00000001
0x0004
Master Configuration Register 1
MATRIX_MCFG1
Read-write
0x00000000
0x0008
Master Configuration Register 2
MATRIX_MCFG2
Read-write
0x00000000
0x000C
Master Configuration Register 3
MATRIX_MCFG3
Read-write
0x00000000
0x0010
Master Configuration Register 4
MATRIX_MCFG4
Read-write
0x00000000
0x0014
Master Configuration Register 5
MATRIX_MCFG5
Read-write
0x00000000
0x0018
Master Configuration Register 6
MATRIX_MCFG6
Read-write
0x00000000
0x001C
Master Configuration Register 7
MATRIX_MCFG7
Read-write
0x00000000
0x0020
Master Configuration Register 8
MATRIX_MCFG8
Read-write
0x00000000
0x0024
Master Configuration Register 9
MATRIX_MCFG9
Read-write
0x00000000
0x0028
Master Configuration Register 10
MATRIX_MCFG10
Read-write
0x00000000
–
–
0x002C - 0x003C
Reserved
–
0x0040
Slave Configuration Register 0
MATRIX_SCFG0
Read-write
0x000001FF
0x0044
Slave Configuration Register 1
MATRIX_SCFG1
Read-write
0x000001FF
0x0048
Slave Configuration Register 2
MATRIX_SCFG2
Read-write
0x000001FF
0x004C
Slave Configuration Register 3
MATRIX_SCFG3
Read-write
0x000001FF
0x0050
Slave Configuration Register 4
MATRIX_SCFG4
Read-write
0x000001FF
0x0054
Slave Configuration Register 5
MATRIX_SCFG5
Read-write
0x000001FF
0x0058
Slave Configuration Register 6
MATRIX_SCFG6
Read-write
0x000001FF
0x005C
Slave Configuration Register 7
MATRIX_SCFG7
Read-write
0x000001FF
–
–
0x0060 - 0x007C
Reserved
–
0x0080
Priority Register A for Slave 0
MATRIX_PRAS0
Read-write
0x00000000
0x0084
Priority Register B for Slave 0
MATRIX_PRBS0
Read-write
0x00000000
0x0088
Priority Register A for Slave 1
MATRIX_PRAS1
Read-write
0x00000000
0x008C
Priority Register B for Slave 1
MATRIX_PRBS1
Read-write
0x00000000
0x0090
Priority Register A for Slave 2
MATRIX_PRAS2
Read-write
0x00000000
0x0094
Priority Register B for Slave 2
MATRIX_PRBS2
Read-write
0x00000000
0x0098
Priority Register A for Slave 3
MATRIX_PRAS3
Read-write
0x00000000
0x009C
Priority Register B for Slave 3
MATRIX_PRBS3
Read-write
0x00000000
0x00A0
Priority Register A for Slave 4
MATRIX_PRAS4
Read-write
0x00000000
0x00A4
Priority Register B for Slave 4
MATRIX_PRBS4
Read-write
0x00000000
0x00A8
Priority Register A for Slave 5
MATRIX_PRAS5
Read-write
0x00000000
0x00AC
Priority Register B for Slave 5
MATRIX_PRBS5
Read-write
0x00000000
0x00B0
Priority Register A for Slave 6
MATRIX_PRAS6
Read-write
0x00000000
0x00B4
Priority Register B for Slave 6
MATRIX_PRBS6
Read-write
0x00000000
139
6438D–ATARM–13-Oct-09
AT91SAM9G45
Table 19-5.
Register Mapping (Continued)
Offset
Register
Name
0x00B8
Priority Register A for Slave 7
0x00BC
Priority Register B for Slave 7
0x00C0 - 0x00FC
0x0100
Reserved
Master Remap Control Register
Access
Reset
MATRIX_PRAS7
Read-write
0x00000000
MATRIX_PRBS7
Read-write
0x00000000
–
–
Read-write
0x00000000
–
MATRIX_MRCR
0x0104 - 0x010C
Reserved
–
–
–
0x0110 - 0x01E0
Chip Configuration Registers
–
–
–
0x01E4
Write Protect Mode Register
MATRIX_WPMR
Read-write
0x00000000
0x01E8
Write Protect Status Register
MATRIX_WPSR
Read-only
0x00000000
140
6438D–ATARM–13-Oct-09
AT91SAM9G45
19.7.1
Name:
Bus Matrix Master Configuration Registers
MATRIX_MCFG0...MATRIX_MCFG10
Address:
0xFFFFEA00
Access:
Read-write
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
–
–
–
–
–
ULBT
This register can only be written if the WPEN bit is cleared in the “Write Protect Mode Register”.
• ULBT: Undefined Length Burst Type
0: Unlimited Length Burst
No predicted end of burst is generated and therefore INCR bursts coming from this master can only be broken if the Slave
Slot Cycle Limit is reached. If the Slot Cycle Limit is not reached, the burst is normally completed by the master, at the latest, on the next AHB 1 KByte address boundary, allowing up to 256-beat word bursts or 128-beat double-word bursts.
1: Single Access
The undefined length burst is treated as a succession of single accesses, allowing re-arbitration at each beat of the INCR
burst.
2: 4-beat Burst
The undefined length burst is split into 4-beat bursts, allowing re-arbitration at each 4-beat burst end.
3: 8-beat Burst
The undefined length burst is split into 8-beat bursts, allowing re-arbitration at each 8-beat burst end.
4: 16-beat Burst
The undefined length burst is split into 16-beat bursts, allowing re-arbitration at each 16-beat burst end.
5: 32-beat Burst
The undefined length burst is split into 32-beat bursts, allowing re-arbitration at each 32-beat burst end.
6: 64-beat Burst
The undefined length burst is split into 64-beat bursts, allowing re-arbitration at each 64-beat burst end.
7: 128-beat Burst
The undefined length burst is split into 128-beat bursts, allowing re-arbitration at each 128-beat burst end.
Unless duly needed the ULBT should be let to its default 0 value for power saving.
141
6438D–ATARM–13-Oct-09
AT91SAM9G45
19.7.2
Name:
Bus Matrix Slave Configuration Registers
MATRIX_SCFG0...MATRIX_SCFG7
Address:
0xFFFFEA40
Access:
Read-write
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
SLOT_CYCLE
7
6
5
4
3
2
1
0
FIXED_DEFMSTR
DEFMSTR_TYPE
SLOT_CYCLE
This register can only be written if the WPEN bit is cleared in the “Write Protect Mode Register”.
• SLOT_CYCLE: Maximum Bus Grant Duration for Masters
When SLOT_CYCLE AHB clock cycles have elapsed since the last arbitration, a new arbitration takes place so as to let an
other master access this slave. If an other master is requesting the slave bus, then the current master burst is broken.
If SLOT_CYCLE = 0, the Slot Cycle Limit feature is disabled and bursts always complete unless broken according to the
ULBT.
This limit has been placed in order to enforce arbitration so as to meet potential latency constraints of masters waiting for
slave access or in the particular case of a master performing back to back undefined length bursts indefinitely freezing the
arbitration.
This limit must not be small. Unreasonably small values break every burst and the Bus Matrix arbitrates without performing
any data transfer. The default maximum value is usually an optimal conservative choice.
In most cases this feature is not needed and should be disabled for power saving.
See “Slot Cycle Limit Arbitration” on page 136 for details.
• DEFMSTR_TYPE: Default Master Type
0: No Default Master
At the end of the current slave access, if no other master request is pending, the slave is disconnected from all masters.
This results in a one clock cycle latency for the first access of a burst transfer or for a single access.
1: Last Default Master
At the end of the current slave access, if no other master request is pending, the slave stays connected to the last master
having accessed it.
This results in not having one clock cycle latency when the last master tries to access the slave again.
2: Fixed Default Master
At the end of the current slave access, if no other master request is pending, the slave connects to the fixed master the
number that has been written in the FIXED_DEFMSTR field.
This results in not having one clock cycle latency when the fixed master tries to access the slave again.
142
6438D–ATARM–13-Oct-09
AT91SAM9G45
• FIXED_DEFMSTR: Fixed Default Master
This is the number of the Default Master for this slave. Only used if DEFMSTR_TYPE is 2. Specifying the number of a master which is not connected to the selected slave is equivalent to setting DEFMSTR_TYPE to 0.
143
6438D–ATARM–13-Oct-09
AT91SAM9G45
19.7.3
Name:
Bus Matrix Priority Registers A For Slaves
MATRIX_PRAS0...MATRIX_PRAS7
Addresses: 0xFFFFEA80 [0], 0xFFFFEA88 [1], 0xFFFFEA90 [2], 0xFFFFEA98 [3], 0xFFFFEAA0 [4], 0xFFFFEAA8 [5],
0xFFFFEAB0 [6], 0xFFFFEAB8 [7]
Access:
Read-write
31
30
–
–
23
22
–
–
15
14
–
–
7
6
–
–
29
28
M7PR
21
20
M5PR
13
12
M3PR
5
4
M1PR
27
26
–
–
19
18
–
–
11
10
–
–
3
2
–
–
25
24
M6PR
17
16
M4PR
9
8
M2PR
1
0
M0PR
This register can only be written if the WPEN bit is cleared in the “Write Protect Mode Register”.
• MxPR: Master x Priority
Fixed priority of Master x for accessing the selected slave. The higher the number, the higher the priority.
All the masters programmed with the same MxPR value for the slave make up a priority pool.
Round-Robin arbitration is used inside the lowest (MxPR = 0) and highest (MxPR = 3) priority pools.
Fixed priority is used inside intermediate priority pools (MxPR = 1) and (MxPR = 2).
See Section 19.5.2 “Arbitration Priority Scheme” for details.
144
6438D–ATARM–13-Oct-09
AT91SAM9G45
19.7.4
Name:
Bus Matrix Priority Registers B For Slaves
MATRIX_PRBS0...MATRIX_PRBS7
Addresses: 0xFFFFEA84 [0], 0xFFFFEA8C [1], 0xFFFFEA94 [2], 0xFFFFEA9C [3], 0xFFFFEAA4 [4], 0xFFFFEAAC [5],
0xFFFFEAB4 [6], 0xFFFFEABC [7]
Access:
Read-write
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
–
–
–
–
–
–
7
6
5
4
3
2
–
–
–
–
M9PR
8
M10PR
1
0
M8PR
This register can only be written if the WPEN bit is cleared in the “Write Protect Mode Register”.
• MxPR: Master x Priority
Fixed priority of Master x for accessing the selected slave. The higher the number, the higher the priority.
All the masters programmed with the same MxPR value for the slave make up a priority pool.
Round-Robin arbitration is used inside the lowest (MxPR = 0) and highest (MxPR = 3) priority pools.
Fixed priority is used inside intermediate priority pools (MxPR = 1) and (MxPR = 2).
See Section 19.5.2 “Arbitration Priority Scheme” for details.
145
6438D–ATARM–13-Oct-09
AT91SAM9G45
19.7.5
Name:
Bus Matrix Master Remap Control Register
MATRIX_MRCR
Address:
0xFFFFEB00
Access:
Read-write
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
RCB10
RCB9
RCB8
7
6
5
4
3
2
1
0
RCB7
RCB6
RCB5
RCB4
RCB3
RCB2
RCB1
RCB0
This register can only be written if the WPEN bit is cleared in the “Write Protect Mode Register”.
• RCB: Remap Command Bit for Master x
0: Disable remapped address decoding for the selected Master
1: Enable remapped address decoding for the selected Master
146
6438D–ATARM–13-Oct-09
AT91SAM9G45
19.7.6
Chip Configuration User Interface
Table 19-6.
Chip Configuration User Interface
Offset
Register
Name
0x0110
Bus Matrix TCM Configuration Register
CCFG_TCMR
0x0114
Reserved
0x0118
DDR Multi-Port Register
0x011C - 0x0124
0x0128
0x012C - 0x01FC
Reserved
EBI Chip Select Assignment Register
Reserved
–
CCFG_DDRMPR
–
CCFG_EBICSA
–
Access
Reset
Read-write
0x00000000
–
–
Read-write
0x00000001
–
–
Read-write
0x00010000
–
–
147
6438D–ATARM–13-Oct-09
AT91SAM9G45
19.7.6.1
Name:
Bus Matrix TCM Configuration Register
CCFG_TCMR
Access:
Read-write
Reset:
0x0000_0000
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
TCM_NWS
–
–
–
7
6
5
4
3
2
1
0
DTCM_SIZE
ITCM_SIZE
• ITCM_SIZE: Size of ITCM enabled memory block
000: 0 KB (No ITCM Memory)
110: 32 KB
Others: Reserved
• DTCM_SIZE: Size of DTCM enabled memory block
000: 0 KB (No DTCM Memory)
110: 32 KB
111: 64 KB
Others: Reserved
• TCM_NWS: TCM Wait State
0: no TCM Wait State
1: 1 TCM Wait State (only for ration 3:1 or 4:1)
148
6438D–ATARM–13-Oct-09
AT91SAM9G45
19.7.6.2
Bus Matrix DDR Multi-Port Register
Register Name:
CCFG_DDRMPR
Access Type:
Read-write
Reset:
0x0000_0001
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
–
–
–
–
–
–
–
DDRMP_DIS
• DDRMP_DIS: DDR Multi-Port Disable
0: Multi-Port is enabled
1: Multi-Port is disabled
149
6438D–ATARM–13-Oct-09
19.7.6.3
Name:
EBI Chip Select Assignment Register
CCFG_EBICSA
Access:
Read-write
Reset:
0x0001_0000
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
–
–
–
–
–
DDR_DRIVE
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
EBI_DBPUC
7
6
5
4
3
2
1
0
–
–
EBI_CS5A
EBI_CS4A
EBI_CS3A
–
EBI_CS1A
–
16
EBI_DRIVE
• EBI_CS1A: EBI Chip Select 1 Assignment
0 = EBI Chip Select 1 is assigned to the Static Memory Controller.
1 = EBI Chip Select 1 is assigned to the SDRAM Controller.
• EBI_CS3A: EBI Chip Select 3 Assignment
0 = EBI Chip Select 3 is only assigned to the Static Memory Controller and EBI_NCS3 behaves as defined by the SMC.
1 = EBI Chip Select 3 is assigned to the Static Memory Controller and the SmartMedia Logic is activated.
• EBI_CS4A: EBI Chip Select 4 Assignment
0 = EBI Chip Select 4 is only assigned to the Static Memory Controller and EBI_NCS4 behaves as defined by the SMC.
1 = EBI Chip Select 4 is assigned to the Static Memory Controller and the Compact Flash Logic Slot 0 is activated.
• EBI_CS5A: EBI Chip Select 5 Assignment
0 = EBI Chip Select 5 is only assigned to the Static Memory Controller and EBI_NCS5 behaves as defined by the SMC.
1 = EBI Chip Select 5 is assigned to the Static Memory Controller and the Compact Flash Logic Slot 1 is activated.
• EBI_DBPUC: EBI Data Bus Pull-Up Configuration
0 = EBI D0 - D15 Data Bus bits are internally pulled-up to the VDDIOM1 power supply.
1 = EBI D0 - D15 Data Bus bits are not internally pulled-up.
• EBI_DRIVE: EBI I/O Drive Configuration
This allows to avoid overshoots and give the best performances according to the bus load and external memories.
Value
Drive configuration
Conditions
00
optimized for 1.8V powered memories with High Drive
Operating frequency = 133 MHz
data bus + memory load capacitance > TBD pF
01
optimized for 3.3V powered memories with High Drive
Operating frequency = 133 MHz
data bus + memory load capacitance > TBD pF
10
optimized for 1.8V powered memories with Low Drive
Operating frequency = 133 MHz
data bus + memory load capacitance > TBD pF
11
optimized for 3.3V powered memories with Low Drive
Operating frequency = 133 MHz
data bus + memory load capacitance > TBD pF
150
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
• DDR_DRIVE: DDR2 dedicated port I/O slew rate selection
This allows to avoid overshoots and give the best performances according to the bus load and external memories.
0 = High Drive, data bus + memory load capacitance > TBD pF.
1 = Low Drive, data bus + memory load capacitance < TBD pF,
Note: This concerns only stand-alone DDR controller.
151
6438D–ATARM–13-Oct-09
19.7.7
Name:
Write Protect Mode Register
MATRIX_WPMR
Address:
0xFFFFEBE4
Access:
Read-write
31
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
WPKEY
23
22
21
20
WPKEY
15
14
13
12
WPKEY
7
6
5
4
3
2
1
0
–
–
–
–
–
–
–
WPEN
For more details on MATRIX_WPMR, refer to Section 19.6 “Write Protect Registers” on page 138.
• WPEN: Write Protect ENable
0 = Disables the Write Protect if WPKEY corresponds to 0x4D4154 (“MAT” in ASCII).
1 = Enables the Write Protect if WPKEY corresponds to 0x4D4154 (“MAT” in ASCII).
Protects the entire MATRIX address space from address offset 0x000 to 0x1FC.
• WPKEY: Write Protect KEY (Write-only)
Should be written at value 0x4D4154 (“MAT” in ASCII). Writing any other value in this field aborts the write operation of the
WPEN bit. Always reads as 0.
152
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
19.7.8
Name:
Write Protect Status Register
MATRIX_WPSR
Address:
0xFFFFEBE8
Access:
Read-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
11
10
9
8
WPVSRC
15
14
13
12
WPVSRC
7
6
5
4
3
2
1
0
–
–
–
–
–
–
–
WPVS
For more details on MATRIX_WPSR, refer to Section 19.6 “Write Protect Registers” on page 138.
• WPVS: Write Protect Violation Status
0: No Write Protect Violation has occurred since the last write of the MATRIX_WPMR.
1: At least one Write Protect Violation has occurred since the last write of the MATRIX_WPMR.
• WPVSRC: Write Protect Violation Source
When WPVS is active, this field indicates the register address offset in which a write access has been attempted.
Otherwise it reads as 0.
153
6438D–ATARM–13-Oct-09
154
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
20. External Memories
20.1
Programmable I/O Lines Power Supplies and Drive Levels
20.1.1
External Bus interface
The power supply pin VDDIOM1 accepts two voltage ranges. This allows the device to reach its
maximum speed either out of 1.8V or 3.3V external memories.
The maximum speed is 133 MHz on the SDCK pin and #SDCK signals loaded with 10 pF. The
load on data/address and control signals are 30 pF for power supply at 1.8V and 50 pF for power
supply at 3.3V. The data lines frequency reaches 133 MHz in DDR2 mode. The other signals
(control and address) do not go over 66 MHz.
The EBI I/Os accept two drive levels, HIGH and LOW. This allows to avoid overshoots and give
the best performance according to the bus load and external memories. Refer to the EBI Chip
Select Assignment Register for more details.
The voltage ranges and the drive level are determined by programming EBI_DRIVE field in the
Chip Configuration registers located in the Matrix User Interface.
At reset the selected default drive level is High.
At reset, the selected voltage defaults to 3.3V nominal and power supply pins can accept either
1.8V or 3.3V. The user must make sure to program the EBI voltage range before getting the
device out of its Slow Clock Mode. The user must make sure to program the EBI voltage range
before getting the device out of its Slow Clock Mode.
20.2
DDR2 Controller
20.2.1
Description
The DDR2 Controller is dedicated to 4-port DDR2/LPDDR support. Data transfers are performed
through a 16-bit data bus on one chip select. The DDR2 Controller operates with 1.8V Power
Supply (VDDIOM0).
20.2.2
20.2.2.1
Embedded Characteristics
DDR2/LPDDR Controller
Four AHB Interfaces, Management of All Accesses Maximizes Memory Bandwidth and Minimizes Transaction Latency.
• Supports AHB Transfers:
– Word, Half Word, Byte Access.
• Supports DDR-SDRAM 2, LPDDR
• Numerous Configurations Supported
– 2K, 4K, 8K, 16K Row Address Memory Parts
– DDR2 with Four Internal Banks
– DDR2/LPDDR with 16-bit Data Path
– One Chip Select for DDR2/LPDDR Device (256 Mbytes Address Space)
• Programming Facilities
155
6438D–ATARM–13-Oct-09
– Multibank Ping-pong Access (Up to 4 Banks Opened at Same Time = Reduces
Average Latency of Transactions)
– Timing Parameters Specified by Software
– Automatic Refresh Operation, Refresh Rate is Programmable
– Automatic Update of DS, TCR and PASR Parameters
• Energy-saving Capabilities
– Self-refresh, Power-down and Deep Power Modes Supported
• Power-up Initialization by Software
• CAS Latency of 2, 3 Supported
• Reset function supported (DDR2)
• Auto Precharge Command Not Used
• On Die Termination not supported
• OCD mode not supported
156
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
20.2.3
DDR2 Controller Block Diagram
Figure 20-1. Organization of the DDR2
DDR2
DDR_A0-DDR_A13
DDR_D0-DDR_D15
DDR_CS
Bus Matrix
DDR_CKE
DDR_RAS, DDR_CAS
DDR2
LPDDR
Controller
AHB
DDR_CLK,#DDR_CLK
DDR_DQS[0..1]
DDR_DQM[0..1]
DDR_WE
DDR_BA0, DDR_BA1
Address Decoders
DDR_VREF
User Interface
APB
157
6438D–ATARM–13-Oct-09
20.2.4
I/O Lines Description
Table 20-1.
DDR2 I/O Lines Description
Name
Function
Type
Active Level
DDR2/LPDDR Controller
DDR_D0 - DDR_D15
Data Bus
I/O
DDR_A0 - DDR_A13
Address Bus
Output
DDR_DQM0 - DDR_DQM1
Data Mask
Output
DDR_DQS0 - DDR_DQS1
Data Strobe
Output
DDR_VREF
Reference Voltage for DDR2 operations, typically 0.9V
DDR_CS
Chip Select
Output
DDR_CLK - DDR_CLK#
DDR2 Differential Clock
Output
DDR_CKE
Clock enable
Output
High
DDR_RAS
Row signal
Output
Low
DDR_CAS
Column signal
Output
Low
DDR_WE
Write enable
Output
Low
DDR_BA0 - DDR_BA1
Bank Select
Output
Input
Low
20.2.5
Product Dependencies
The pins used for interfacing the DDR2 memory are not multiplexed with the PIO lines.
20.2.6
Implementation Example
The following hardware configuration is given for illustration only. The user should refer to the
memory manufacturer web site to check current device availability.
158
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
20.2.6.1
2x8-bit DDR2
Hardware Configuration
DDR_D[0..15]
DDR_A[0..13]
MN6
DDR_A0
DDR_A1
DDR_A2
DDR_A3
DDR_A4
DDR_A5
DDR_A6
DDR_A7
DDR_A8
DDR_A9
DDR_A10
DDR_A11
DDR_A12
DDR_A13
DDR_BA0
DDR_BA1
DDR_CKE
DDR_CLK
DDR_NCLK
DDR_CS
DDR_CAS
DDR_RAS
DDR_WE
BA0
BA1
MN7
H8
H3
H7
J2
J8
J3
J7
K2
K8
K3
H2
K7
L2
L8
A0
DQ0
DDR2 SDRAM
A1
DQ1
A2 MT47H64M8CF-3 DQ2
A3
DQ3
A4
DQ4
A5
DQ5
A6
DQ6
A7
DQ7
A8
A9
DQS
A10
DQS
A11
A12
RDQS/DM
A13
RDQS/NU
G2
G3
BA0
BA1
F9
ODT
CKE
F2
CKE
CK
NCK
E8
F8
CK
CK
CS
G8
CS
CAS
RAS
G7
F7
CAS
RAS
NWE
F3
WE
G1
L3
L7
RFU1
RFU2
RFU3
C8
C2
D7
D3
D1
D9
B1
B9
DDR_D0
DDR_D1
DDR_D2
DDR_D3
DDR_D4
DDR_D5
DDR_D6
DDR_D7
B7
A8
DDR_DQS0
B3
A2
DDR_DQM0
DDR_A0
DDR_A1
DDR_A2
DDR_A3
DDR_A4
DDR_A5
DDR_A6
DDR_A7
DDR_A8
DDR_A9
DDR_A10
DDR_A11
DDR_A12
DDR_A13
1V8
VDD
VDD
VDD
VDD
A1
E9
H9
L1
C55
C57
C59
C61
100nF
100nF
100nF
100nF
VDDL
E1
C63 100nF
VDDQ
VDDQ
VDDQ
VDDQ
VDDQ
A9
C1
C3
C7
C9
C65
C67
C69
C71
C73
VREF
E2
VSS
VSS
VSS
VSS
A3
E3
J1
K9
VSSQ
VSSQ
VSSQ
VSSQ
VSSQ
A7
B2
B8
D2
D8
VSSDL
E7
100nF
100nF
100nF
100nF
100nF
DDR_VREF
C75
100nF
BA0
BA1
H8
H3
H7
J2
J8
J3
J7
K2
K8
K3
H2
K7
L2
L8
A0
DQ0
DDR2 SDRAM
A1
DQ1
A2 MT47H64M8CF-3 DQ2
A3
DQ3
A4
DQ4
A5
DQ5
A6
DQ6
A7
DQ7
A8
A9
DQS
A10
DQS
A11
A12
RDQS/DM
A13
RDQS/NU
G2
G3
BA0
BA1
F9
ODT
CKE
F2
CKE
CK
NCK
E8
F8
CK
CK
CS
G8
CS
CAS
RAS
G7
F7
CAS
RAS
NWE
F3
WE
G1
L3
L7
RFU1
RFU2
RFU3
C8
C2
D7
D3
D1
D9
B1
B9
DDR_D8
DDR_D9
DDR_D10
DDR_D11
DDR_D12
DDR_D13
DDR_D14
DDR_D15
B7
A8
B3
A2
DDR_DQS1
DDR_DQM1
1V8
A1
E9
H9
L1
C56
C58
C60
C62
VDDL
E1
C64 100nF
VDDQ
VDDQ
VDDQ
VDDQ
VDDQ
A9
C1
C3
C7
C9
C66
C68
C70
C72
C74
VDD
VDD
VDD
VDD
VREF
E2
VSS
VSS
VSS
VSS
A3
E3
J1
K9
VSSQ
VSSQ
VSSQ
VSSQ
VSSQ
A7
B2
B8
D2
D8
VSSDL
E7
100nF
100nF
100nF
100nF
100nF
100nF
100nF
100nF
100nF
DDR_VREF
C76
100nF
Software Configuration
The following configuration has to be performed:
• Initialize the DDR2 Controller depending on the DDR2 device and system bus frequency.
The DDR2 initialization sequence is described in the sub-section “DDR2 Device Initialization” of
the DDRSDRC section.
159
6438D–ATARM–13-Oct-09
20.3
External Bus Interface (EBI)
20.3.1
Description
The External Bus Interface (EBI) is designed to ensure the successful data transfer between
several external devices and the embedded Memory Controller of an ARM-based device.
The Static Memory, DDR, SDRAM and ECC Controllers are all featured external Memory Controllers on the EBI. These external Memory Controllers are capable of handling several types of
external memory and peripheral devices, such as SRAM, PROM, EPROM, EEPROM, Flash,
DDR2 and SDRAM. The EBI operates with 1.8V or 3.3V Power Supply (VDDIOM1).
The EBI also supports the CompactFlash and the NAND Flash protocols via integrated circuitry
that greatly reduces the requirements for external components. Furthermore, the EBI handles
data transfers with up to six external devices, each assigned to six address spaces defined by
the embedded Memory Controller. Data transfers are performed through a 16-bit or 32-bit data
bus, an address bus of up to 26 bits, up to six chip select lines (NCS[5:0]) and several control
pins that are generally multiplexed between the different external Memory Controllers.
20.3.2
20.3.2.1
Embedded Characteristics
The AT91SAM9G45 features an External Bus Interface to interface to a wide range of external
memories and to any parallel peripheral.
External Bus Interface
• Integrates Three External Memory Controllers:
– Static Memory Controller
– DDR2/SDRAM Controller
– SLC Nand Flash ECC Controller
• Additional logic for NAND Flash and CompactFlash
• Optional Full 32-bit External Data Bus
• Up to 26-bit Address Bus (up to 64 MBytes linear per chip select)
• Up to 6 chip selects, Configurable Assignment:
– Static Memory Controller on NCS0
– DDR2/SDRAM Controller (SDCS) or Static Memory Controller on NCS1
– Static Memory Controller on NCS2
– Static Memory Controller on NCS3, Optional NAND Flash support
– Static Memory Controller on NCS4 - NCS5, Optional CompactFlashM support
20.3.2.2
Static Memory Controller
• 8-, 16- or 32-bit Data Bus
• Multiple Access Modes supported
– Byte Write or Byte Select Lines
– Asynchronous read in Page Mode supported (4- up to 32-byte page size)
• Multiple device adaptability
– Control signals programmable setup, pulse and hold time for each Memory Bank
• Multiple Wait State Management
– Programmable Wait State Generation
– External Wait Request
160
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
– Programmable Data Float Time
• Slow Clock mode supported
20.3.2.3
DDR2/SDR Controller
• Supports DDR/LPDDR, SDR-SDRAM and LPSDR
• Numerous Configurations Supported
– 2K, 4K, 8K, 16K Row Address Memory Parts
– SDRAM with Four Internal Banks
– SDR-SDRAM with 16- or 32- bit Data Path
– DDR2/LPDDR with 16- bit Data Path
– One Chip Select for SDRAM Device (256 Mbyte Address Space)
• Programming Facilities
– Multibank Ping-pong Access (Up to 4 Banks Opened at Same Time = Reduces
Average Latency of Transactions)
– Timing Parameters Specified by Software
– Automatic Refresh Operation, Refresh Rate is Programmable
– Automatic Update of DS, TCR and PASR Parameters (LPSDR)
• Energy-saving Capabilities
– Self-refresh, Power-down and Deep Power Modes Supported
• SDRAM Power-up Initialization by Software
• CAS Latency of 2, 3 Supported
• Auto Precharge Command Not Used
• SDR-SDRAM with 16-bit Datapath and Eight Columns Not Supported
– Clock Frequency Change in Precharge Power-down Mode Not Supported
20.3.2.4
NAND Flash Error Corrected Code Controller
• Tracking the accesses to a NAND Flash device by triggering on the corresponding chip select
• Single bit error correction and 2-bit Random detection.
• Automatic Hamming Code Calculation while writing
– ECC value available in a register
• Automatic Hamming Code Calculation while reading
– Error Report, including error flag, correctable error flag and word address being
detected erroneous
– Support 8- or 16-bit NAND Flash devices with 512-, 1024-, 2048- or 4096-bytes
pages
161
6438D–ATARM–13-Oct-09
20.3.3
EBI Block Diagram
Figure 20-2. Organization of the External Bus Interface
External Bus Interface
Bus Matrix
D[15:0]
DDR2
LPDDR
SDRAM
Controller
AHB
A0/NBS0
A1/NWR2/NBS2
A[15:2], A18
A16/BA0
A17/BA1
MUX
Logic
Static
Memory
Controller
NCS0
NCS1/SDCS
NRD/CFOE
NWR0/NWE/CFWE
NWR1/NBS1/CFIOR
NWR3/NBS3/CFIOW
SDCK, SDCK#, SDCKE
DQM[1:0]
CompactFlash
Logic
DQS[1:0]
RAS, CAS
SDWE
SDA10
NANDOE
NAND Flash
Logic
NANDWE
A21/NANDALE
A22/NANDCLE
D[31:16]
ECC
Controller
A[24:19]
PIO
Address Decoders
Chip Select
Assignor
A25/CFRNW
NCS5/CFCS1
NCS4/CFCS0
NCS3/NANDCS
NCS2
User Interface
NWAIT
CFCE1
CFCE2
DQM[3:2]
APB
162
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
20.3.4
I/O Lines Description
Table 20-2.
EBI I/O Lines Description
Name
Function
Type
Active Level
EBI
EBI_D0 - EBI_D31
Data Bus
EBI_A0 - EBI_A25
Address Bus
I/O
EBI_NWAIT
External Wait Signal
Output
Input
Low
SMC
EBI_NCS0 - EBI_NCS5
Chip Select Lines
Output
Low
EBI_NWR0 - EBI_NWR3
Write Signals
Output
Low
EBI_NRD
Read Signal
Output
Low
EBI_NWE
Write Enable
Output
Low
EBI_NBS0 - EBI_NBS3
Byte Mask Signals
Output
Low
EBI for NAND Flash Support
EBI_NANDCS
NAND Flash Chip Select Line
Output
Low
EBI_NANDOE
NAND Flash Output Enable
Output
Low
EBI_NANDWE
NAND Flash Write Enable
Output
Low
DDR2/SDRAM Controller
EBI_SDCK, EBI_SDCK#
DDR2/SDRAM Differential Clock
Output
EBI_SDCKE
DDR2/SDRAM Clock Enable
Output
High
Low
EBI_SDCS
DDR2/SDRAM Controller Chip Select Line
Output
EBI_BA0 - EBI_BA1
Bank Select
Output
EBI_SDWE
DDR2/SDRAM Write Enable
Output
Low
EBI_RAS - EBI_CAS
Row and Column Signal
Output
Low
EBI_SDA10
SDRAM Address 10 Line
Output
The connection of some signals through the MUX logic is not direct and depends on the Memory
Controller in use at the moment.
Table 20-3 on page 163 details the connections between the two Memory Controllers and the
EBI pins.
Table 20-3.
EBI Pins and Memory Controllers I/O Lines Connections
EBIx Pins
SDRAM I/O Lines
SMC I/O Lines
EBI_NWR1/NBS1/CFIOR
NBS1
NWR1
EBI_A0/NBS0
Not Supported
SMC_A0
EBI_A1/NBS2/NWR2
Not Supported
SMC_A1
EBI_A[11:2]
SDRAMC_A[9:0]
SMC_A[11:2]
EBI_SDA10
SDRAMC_A10
Not Supported
EBI_A12
Not Supported
SMC_A12
EBI_A[14:13]
SDRAMC_A[12:11]
SMC_A[14:13]
EBI_A[25:15]
Not Supported
SMC_A[25:15]
EBI_D[31:0]
D[31:0]
D[31:0]
163
6438D–ATARM–13-Oct-09
20.3.5
Application Example
20.3.5.1
Hardware Interface
Table 20-4 on page 164 details the connections to be applied between the EBI pins and the
external devices for each Memory Controller.
Table 20-4.
EBI Pins and External Static Devices Connections
Pins of the Interfaced Device
8-bit Static
Device
Signals:
EBI_
2 x 8-bit
Static
Devices
16-bit Static
Device
Controller
4 x 8-bit
Static
Devices
2 x 16-bit
Static
Devices
32-bit Static
Device
SMC
D0 - D7
D0 - D7
D0 - D7
D0 - D7
D0 - D7
D0 - D7
D0 - D7
D8 - D15
–
D8 - D15
D8 - D15
D8 - D15
D8 - 15
D8 - 15
D16 - D23
–
–
–
D16 - D23
D16 - D23
D16 - D23
D24 - D31
–
–
–
D24 - D31
D24 - D31
D24 - D31
A0/NBS0
A0
–
NLB
–
NLB(3)
BE0
A1/NWR2/NBS2
A1
A0
A0
WE(2)
NLB(4)
BE2
A2 - A22
A[2:22]
A[1:21]
A[1:21]
A[0:20]
A[0:20]
A[0:20]
A23 - A25(5)
A[23:25]
A[22:24]
A[22:24]
A[21:23]
A[21:23]
A[21:23]
NCS0
CS
CS
CS
CS
CS
CS
NCS1/DDRSDCS
CS
CS
CS
CS
CS
CS
NCS2
CS
CS
CS
CS
CS
CS
NCS3/NANDCS
CS
CS
CS
CS
CS
CS
NCS4/CFCS0
CS
CS
CS
CS
CS
CS
NCS5/CFCS1
CS
CS
CS
CS
CS
CS
NRD/CFOE
OE
OE
OE
OE
OE
OE
NWR0/NWE
WE
WE(1)
WE
WE(2)
WE
WE
NWR1/NBS1
–
WE(1)
NUB
WE(2)
NUB(3)
BE1
NWR3/NBS3
–
–
–
WE(2)
NUB(4)
BE3
Notes:
164
1.
2.
3.
4.
NWR1 enables upper byte writes. NWR0 enables lower byte writes.
NWRx enables corresponding byte x writes. (x = 0,1,2 or 3)
NBS0 and NBS1 enable respectively lower and upper bytes of the lower 16-bit word.
NBS2 and NBS3 enable respectively lower and upper bytes of the upper 16-bit word.
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Table 20-5.
EBI Pins and External Device Connections
Pins of the Interfaced Device
Signals:
EBI_
Controller
DDR2/LPDDR
SDRAM
DDRC
SDRAMC
CompactFlash
CompactFlash
True IDE Mode
NAND Flash
SMC
D0 - D7
D0 - D7
D0 - D7
D0 - D7
D0 - D7
I/O0-I/O7
D8 - D15
D8 - D15
D8 - D15
D8 - 15
D8 - 15
I/O8-I/O15(4)
D16 - D31
–
D16 - D31
–
–
–
A0/NBS0
–
–
A0
A0
–
A1/NWR2/NBS2
–
–
A1
A1
–
DQM0-DQM3
DQM0-DQM3
DQM0-DQM3
–
–
–
DQS0-DQM1
DQS0-DQS1
–
–
–
–
A[0:8]
A[0:8]
A[2:10]
A[2:10]
–
A9
A9
–
–
–
–
A10
–
–
–
A2 - A10
A11
SDA10
A12
A13 - A14
A15
–
–
–
–
–
A[11:12]
A[11:12]
–
–
–
–
–
–
–
–
A16/BA0
BA0
BA0
–
–
–
A17/BA1
BA1
BA1
–
–
–
A18 - A20
–
–
–
–
–
A21/NANDALE
–
–
–
–
ALE
A22/NANDCLE
–
–
REG
REG
CLE
A23 - A24
–
–
–
–
–
A25
–
–
CFRNW(1)
CFRNW(1)
–
NCS0
NCS1/DDRSDCS
–
–
–
–
–
DDRCS
SDCS
–
–
–
NCS2
–
–
–
–
–
NCS3/NANDCS
–
–
–
–
CE(3)
NCS4/CFCS0
–
–
CFCS0(1)
CFCS0(1)
–
NCS5/CFCS1
–
–
CFCS1(1)
CFCS1(1)
–
NANDOE
–
–
–
–
OE
NANDWE
–
–
–
–
WE
NRD/CFOE
–
–
OE
–
–
NWR0/NWE/CFWE
–
–
WE
WE
–
NWR1/NBS1/CFIOR
–
–
IOR
IOR
–
NWR3/NBS3/CFIOW
–
–
IOW
IOW
–
CFCE1
–
–
CE1
CS0
–
CFCE2
–
–
CE2
CS1
–
SDCK
CK
CLK
–
–
–
SDCK#
CK#
–
–
–
–
SDCKE
CKE
CKE
–
–
–
RAS
RAS
RAS
–
–
–
CAS
CAS
CAS
–
–
–
165
6438D–ATARM–13-Oct-09
Table 20-5.
EBI Pins and External Device Connections (Continued)
Pins of the Interfaced Device
Signals:
EBI_
Controller
SDWE
(5)
DDR2/LPDDR
SDRAM
DDRC
SDRAMC
WE
WE
CompactFlash
True IDE Mode
CompactFlash
NAND Flash
SMC
–
–
–
–
–
WAIT
WAIT
–
Pxx(2)
–
–
CD1 or CD2
CD1 or CD2
–
(2)
–
–
–
–
CE(3)
Pxx(2)
–
–
–
–
RDY
NWAIT
Pxx
Notes:
1. Not directly connected to the CompactFlash slot. Permits the control of the bidirectional buffer between the EBI data bus and
the CompactFlash slot.
2. Any PIO line.
3. CE connection depends on the NAND Flash.
For standard NAND Flash devices, it must be connected to any free PIO line.
For "CE don't care" NAND Flash devices, it can be either connected to NCS3/NANDCS or to any free PIO line.
4. I/O8 - !/O15 pins used only for 16-bit NANDFlash device.
5. EBI_NWAIT signal is multiplexed with PC15.
20.3.5.2
Connection Examples
Figure 20-3 shows an example of connections between the EBI and external devices.
Figure 20-3. EBI Connections to Memory Devices
EBI
D0-D31
RAS
CAS
SDCK
SDCKE
SDWE
A0/NBS0
NWR1/NBS1
A1/NWR2/NBS2
NWR3/NBS3
NRD/NOE
NWR0/NWE
D0-D7
2M x 8
SDRAM
D8-D15
D0-D7
CS
CLK
CKE
SDWE WE
RAS
CAS
DQM
NBS0
A0-A9, A11
A10
BA0
BA1
2M x 8
SDRAM
D0-D7
CS
CLK
CKE
SDWE
WE
RAS
CAS
DQM
NBS1
A2-A11, A13
SDA10
A16/BA0
A17/BA1
A0-A9, A11
A10
BA0
BA1
A2-A11, A13
SDA10
A16/BA0
A17/BA1
SDA10
A2-A15
A16/BA0
A17/BA1
A18-A25
D16-D23
NCS0
NCS1/SDCS
NCS2
NCS3
NCS4
NCS5
D0-D7
CS
CLK
CKE
SDWE WE
RAS
CAS
DQM
2M x 8
SDRAM
A0-A9, A11
A10
BA0
BA1
D24-D31
A2-A11, A13
SDA10
A16/BA0
A17/BA1
SDWE
NBS3
2M x 8
SDRAM
D0-D7
CS
CLK
CKE
WE
RAS
CAS
DQM
A0-A9, A11
A10
BA0
BA1
A2-A11, A13
SDA10
A16/BA0
A17/BA1
NBS2
128K x 8
SRAM
D0-D7
D0-D7
CS
OE
NRD/NOE
WE
A0/NWR0/NBS0
166
A0-A16
128K x 8
SRAM
A1-A17
D8-D15
D0-D7
A0-A16
A1-A17
CS
OE
NRD/NOE
WE
NWR1/NBS1
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
20.3.6
20.3.6.1
Product Dependencies
I/O Lines
The pins used for interfacing the External Bus Interface may be multiplexed with the PIO lines.
The programmer must first program the PIO controller to assign the External Bus Interface pins
to their peripheral function. If I/O lines of the External Bus Interface are not used by the application, they can be used for other purposes by the PIO Controller.
20.3.7
Functional Description
The EBI transfers data between the internal AHB Bus (handled by the Bus Matrix) and the external memories or peripheral devices. It controls the waveforms and the parameters of the
external address, data and control buses and is composed of the following elements:
• the Static Memory Controller (SMC)
• the DDR2/SDRAM Controller (DDR2SDRAMC)
• the ECC Controller (ECC)
• a chip select assignment feature that assigns an AHB address space to the external devices
• a multiplex controller circuit that shares the pins between the different Memory Controllers
• programmable CompactFlash support logic
• programmable NAND Flash support logic
20.3.7.1
Bus Multiplexing
The EBI offers a complete set of control signals that share the 32-bit data lines, the address
lines of up to 26 bits and the control signals through a multiplex logic operating in function of the
memory area requests.
Multiplexing is specifically organized in order to guarantee the maintenance of the address and
output control lines at a stable state while no external access is being performed. Multiplexing is
also designed to respect the data float times defined in the Memory Controllers. Furthermore,
refresh cycles of the DDR2 and SDRAM are executed independently by the DDR2SDRAM Controller without delaying the other external Memory Controller accesses.
20.3.7.2
Pull-up Control
The EBI_CSA registers in the Chip Configuration User Interface permit enabling of on-chip pullup resistors on the data bus lines not multiplexed with the PIO Controller lines. The pull-up resistors are enabled after reset. Setting the EBIx_DBPUC bit disables the pull-up resistors on the D0
to D15 lines. Enabling the pull-up resistor on the D16-D31 lines can be performed by programming the appropriate PIO controller.
20.3.7.3
Static Memory Controller
For information on the Static Memory Controller, refer to the Static Memory Controller section.
20.3.7.4
DDR2SDRAM Controller
For information on the DDR2SDRAM Controller, refer to the DDR2SDRAMC section.
20.3.7.5
ECC Controller
For information on the ECC Controller, refer to the ECC section.
167
6438D–ATARM–13-Oct-09
20.3.7.6
CompactFlash Support
The External Bus Interface 0 integrates circuitry that interfaces to CompactFlash devices.
The CompactFlash logic is driven by the Static Memory Controller (SMC) on the NCS4 and/or
NCS5 address space. Programming the EBI_CS4A and/or EBI_CS5A bit of the EBI_CSA Register in the Chip Configuration User Interface to the appropriate value enables this logic. (For
details on this register, refer to the Chip Configuration User Interface in the Bus Matrix Section.)
Access to an external CompactFlash device is then made by accessing the address space
reserved to NCS4 and/or NCS5 (i.e., between 0x5000 0000 and 0x5FFF FFFF for NCS4 and
between 0x6000 0000 and 0x6FFF FFFF for NCS5).
All CompactFlash modes (Attribute Memory, Common Memory, I/O and True IDE) are supported but the signals _IOIS16 (I/O and True IDE modes) and _ATA SEL (True IDE mode) are
not handled.
I/O Mode, Common Memory Mode, Attribute Memory Mode and True IDE Mode
Within the NCS4 and/or NCS5 address space, the current transfer address is used to distinguish
I/O mode, common memory mode, attribute memory mode and True IDE mode.
The different modes are accessed through a specific memory mapping as illustrated on Figure
20-4. A[23:21] bits of the transfer address are used to select the desired mode as described in
Table 20-6 on page 168.
Figure 20-4. CompactFlash Memory Mapping
True IDE Alternate Mode Space
Offset 0x00E0 0000
True IDE Mode Space
Offset 0x00C0 0000
CF Address Space
I/O Mode Space
Offset 0x0080 0000
Common Memory Mode Space
Offset 0x0040 0000
Attribute Memory Mode Space
Offset 0x0000 0000
Note:
The A22 pin is used to drive the REG signal of the CompactFlash Device (except in True IDE
mode).
Table 20-6.
A[23:21]
168
CompactFlash Mode Selection
Mode Base Address
000
Attribute Memory
010
Common Memory
100
I/O Mode
110
True IDE Mode
111
Alternate True IDE Mode
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
CFCE1 and CFCE2 Signals
To cover all types of access, the SMC must be alternatively set to drive 8-bit data bus or 16-bit
data bus. The odd byte access on the D[7:0] bus is only possible when the SMC is configured to
drive 8-bit memory devices on the corresponding NCS pin (NCS4 or NCS5). The Chip Select
Register (DBW field in the corresponding Chip Select Register) of the NCS4 and/or NCS5
address space must be set as shown in Table 20-7 to enable the required access type.
NBS1 and NBS0 are the byte selection signals from SMC and are available when the SMC is set
in Byte Select mode on the corresponding Chip Select.
The CFCE1 and CFCE2 waveforms are identical to the corresponding NCSx waveform. For
details on these waveforms and timings, refer to the Static Memory Controller section.
Table 20-7.
CFCE1 and CFCE2 Truth Table
Mode
Attribute Memory
CFCE2
CFCE1
DBW
Comment
SMC Access Mode
NBS1
NBS0
16 bits
Access to Even Byte on D[7:0]
Byte Select
NBS1
NBS0
16bits
Access to Even Byte on D[7:0]
Access to Odd Byte on D[15:8]
Byte Select
1
0
8 bits
Access to Odd Byte on D[7:0]
NBS1
NBS0
16 bits
Access to Even Byte on D[7:0]
Access to Odd Byte on D[15:8]
1
0
8 bits
Access to Odd Byte on D[7:0]
Common Memory
I/O Mode
Byte Select
True IDE Mode
Task File
1
0
8 bits
Access to Even Byte on D[7:0]
Access to Odd Byte on D[7:0]
Data Register
1
0
16 bits
Access to Even Byte on D[7:0]
Access to Odd Byte on D[15:8]
Byte Select
Control Register
Alternate Status Read
0
1
Don’t
Care
Access to Even Byte on D[7:0]
Don’t Care
Drive Address
0
1
8 bits
Access to Odd Byte on D[7:0]
1
1
–
Alternate True IDE Mode
Standby Mode or
Address Space is not
assigned to CF
–
–
Read/Write Signals
In I/O mode and True IDE mode, the CompactFlash logic drives the read and write command
signals of the SMC on CFIOR and CFIOW signals, while the CFOE and CFWE signals are deactivated. Likewise, in common memory mode and attribute memory mode, the SMC signals are
driven on the CFOE and CFWE signals, while the CFIOR and CFIOW are deactivated. Figure
20-5 on page 170 demonstrates a schematic representation of this logic.
Attribute memory mode, common memory mode and I/O mode are supported by setting the
address setup and hold time on the NCS4 (and/or NCS5) chip select to the appropriate values.
For details on these signal waveforms, please refer to the section: Setup and Hold Cycles of the
Static Memory Controller section.
169
6438D–ATARM–13-Oct-09
Figure 20-5. CompactFlash Read/Write Control Signals
External Bus Interface
SMC
CompactFlash Logic
A23
1
1
0
1
0
0
CFOE
CFWE
1
1
A22
NRD_NOE
NWR0_NWE
0
1
1
Table 20-8.
CFIOR
CFIOW
1
CompactFlash Mode Selection
Mode Base Address
CFOE
CFWE
CFIOR
CFIOW
NRD
NWR0_NWE
1
1
I/O Mode
1
1
NRD
NWR0_NWE
True IDE Mode
0
1
NRD
NWR0_NWE
Attribute Memory
Common Memory
Multiplexing of CompactFlash Signals on EBI Pins
Table 20-9 on page 170 and Table 20-10 on page 171 illustrate the multiplexing of the CompactFlash logic signals with other EBI signals on the EBI pins. The EBI pins in Table 20-9 are strictly
dedicated to the CompactFlash interface as soon as the EBI_CS4A and/or EBI_CS5A field of
the EBI_CSA Register in the Chip Configuration User Interface is set. These pins must not be
used to drive any other memory devices.
The EBI pins in Table 20-10 on page 171 remain shared between all memory areas when the
corresponding CompactFlash interface is enabled (EBI_CS4A = 1 and/or EBI_CS5A = 1).
Table 20-9.
Dedicated CompactFlash Interface Multiplexing
CompactFlash Signals
Pins
CS4A = 1
NCS4/CFCS0
CFCS0
NCS5/CFCS1
170
CS5A = 1
EBI Signals
CS4A = 0
CS5A = 0
NCS4
CFCS1
NCS5
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Table 20-10. Shared CompactFlash Interface Multiplexing
Access to CompactFlash Device
Access to Other EBI Devices
Pins
CompactFlash Signals
EBI Signals
NRD/CFOE
CFOE
NRD
NWR0/NWE/CFWE
CFWE
NWR0/NWE
NWR1/NBS1/CFIOR
CFIOR
NWR1/NBS1
NWR3/NBS3/CFIOW
CFIOW
NWR3/NBS3
A25/CFRNW
CFRNW
A25
Application Example
Figure 20-6 on page 171 illustrates an example of a CompactFlash application. CFCS0 and
CFRNW signals are not directly connected to the CompactFlash slot 0, but do control the direction and the output enable of the buffers between the EBI and the CompactFlash Device. The
timing of the CFCS0 signal is identical to the NCS4 signal. Moreover, the CFRNW signal
remains valid throughout the transfer, as does the address bus. The CompactFlash _WAIT signal is connected to the NWAIT input of the Static Memory Controller. For details on these
waveforms and timings, refer to the Static Memory Controller Section.
Figure 20-6. CompactFlash Application Example
EBI
CompactFlash Connector
D[15:0]
D[15:0]
DIR /OE
A25/CFRNW
NCS4/CFCS0
_CD1
CD (PIO)
_CD2
/OE
A[10:0]
A[10:0]
A22/REG
_REG
NOE/CFOE
_OE
NWE/CFWE
_WE
NWR1/CFIOR
_IORD
NWR3/CFIOW
_IOWR
CFCE1
_CE1
CFCE2
_CE2
NWAIT
_WAIT
171
6438D–ATARM–13-Oct-09
20.3.7.7
NAND Flash Support
External Bus Interfaces 1 integrate circuitry that interfaces to NAND Flash devices.
External Bus Interface
The NAND Flash logic is driven by the Static Memory Controller on the NCS2 address space.
Programming the EBI_CS2A field in the EBI_CSA Register in the Chip Configuration User Interface to the appropriate value enables the NAND Flash logic. For details on this register, refer to
the Bus Matrix Section. Access to an external NAND Flash device is then made by accessing
the address space reserved to NCS4 (i.e., between 0x4000 0000 and 0x4FFF FFFF).
The NAND Flash Logic drives the read and write command signals of the SMC on the NANDOE
and NANDWE signals when the NCS3 signal is active. NANDOE and NANDWE are invalidated
as soon as the transfer address fails to lie in the NCS3 address space. See Figure 20-7 on page
172 for more information. For details on these waveforms, refer to the Static Memory Controller
section.
NAND Flash Signals
The address latch enable and command latch enable signals on the NAND Flash device are
driven by address bits A22 and A21 of the EBI address bus. The command, address or data
words on the data bus of the NAND Flash device are distinguished by using their address within
the NCSx address space. The chip enable (CE) signal of the device and the ready/busy (R/B)
signals are connected to PIO lines. The CE signal then remains asserted even when NCSx is
not selected, preventing the device from returning to standby mode.
Figure 20-7. NAND Flash Application Example
D[7:0]
AD[7:0]
A[22:21]
ALE
CLE
NCSx/NANDCS
Not Connected
EBI
NAND Flash
NANDOE
NANDWE
172
NOE
NWE
PIO
CE
PIO
R/B
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
20.3.8
Implementation Examples
The following hardware configurations are given for illustration only. The user should refer to the
memory manufacturer web site to check current device availability.
20.3.8.1
2x8-bit DDR2 on EBI
Hardware Configuration
Software Configuration
• Assign EBI_CS1 to the DDR2 controller by setting the EBI_CS1A bit in the EBI Chip Select
Register located in the bus matrix memory space.
• Initialize the DDR2 Controller depending on the DDR2 device and system bus frequency.
The DDR2 initialization sequence is described in the sub-section “DDR2 Device Initialization” of
the DDRSDRC section.
173
6438D–ATARM–13-Oct-09
20.3.8.2
16-bit LPDDR on EBI
Hardware Configuration
Software Configuration
The following configuration has to be performed:
• Assign EBI_CS1 to the DDR2 controller by setting the bit EBI_CS1A in the EBI Chip Select
Register located in the bus matrix memory space.
• Initialize the DDR2 Controller depending on the LP-DDR device and system bus frequency.
The LP-DDR initialization sequence is described in the section “Low-power DDR1-SDRAM Initialization” in “DDR/SDR SDRAM Controller (DDRSDRC)”.
174
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
20.3.8.3
16-bit SDRAM
Hardware Configuration
Software Configuration
The following configuration has to be performed:
• Assign the EBI CS1 to the SDRAM controller by setting the bit EBI_CS1A in the EBI Chip
Select Assignment Register located in the bus matrix memory space.
• Initialize the SDRAM Controller depending on the SDRAM device and system bus frequency.
The Data Bus Width is to be programmed to 16 bits.
The SDRAM initialization sequence is described in the section “SDRAM Device Initialization” in
“SDRAM Controller (SDRAMC)”.
175
6438D–ATARM–13-Oct-09
20.3.8.4
2x16-bit SDRAM
Hardware Configuration
A[1..14]
D[0..31]
SDRAM
MN1
VDDIOM
A2
A3
A4
A5
A6
A7
A8
A9
A10
A11
SDA10
A13
23
24
25
26
29
30
31
32
33
34
22
35
BA0
BA1
20
21
A14
36
40
CKE
37
CLK
38
DQM0
DQM1
15
39
CAS
RAS
17
18
WE
16
19
R1
470K
MN2
A0 MT48LC16M16A2 DQ0
A1
DQ1
A2
DQ2
A3
DQ3
A4
DQ4
A5
DQ5
A6
DQ6
A7
DQ7
A8
DQ8
A9
DQ9
A10
DQ10
A11
DQ11
DQ12
BA0
DQ13
BA1
DQ14
DQ15
A12
N.C1
VDD
VDD
CKE
VDD
VDDQ
CLK
VDDQ
VDDQ
DQML
VDDQ
DQMH
VSS
CAS
VSS
RAS
VSS
VSSQ
VSSQ
WE
VSSQ
CS
VSSQ
2
4
5
7
8
10
11
13
42
44
45
47
48
50
51
53
1
14
27
3
9
43
49
28
41
54
6
12
46
52
D0
D1
D2
D3
D4
D5
D6
D7
D8
D9
D10
D11
D12
D13
D14
D15
VDDIOM
C1
C3
C5
C7
100NF 100NF
100NF 100NF
C2
C4
C6
100NF 100NF 100NF
VDDIOM
A2
A3
A4
A5
A6
A7
A8
A9
A10
A11
SDA10
A13
23
24
25
26
29
30
31
32
33
34
22
35
BA0
BA1
20
21
A14
36
40
CKE
37
CLK
38
DQM2
DQM3
15
39
CAS
RAS
17
18
WE
16
19
A0 MT48LC16M16A2 DQ0
A1
DQ1
A2
DQ2
A3
DQ3
A4
DQ4
A5
DQ5
A6
DQ6
A7
DQ7
A8
DQ8
A9
DQ9
A10
DQ10
A11
DQ11
DQ12
BA0
DQ13
BA1
DQ14
DQ15
A12
N.C1
VDD
VDD
CKE
VDD
VDDQ
CLK
VDDQ
VDDQ
DQML
VDDQ
DQMH
VSS
CAS
VSS
RAS
VSS
VSSQ
VSSQ
WE
VSSQ
CS
VSSQ
2
4
5
7
8
10
11
13
42
44
45
47
48
50
51
53
1
14
27
3
9
43
49
28
41
54
6
12
46
52
D16
D17
D18
D19
D20
D21
D22
D23
D24
D25
D26
D27
D28
D29
D30
D31
VDDIOM
C8
C10
C12
C14
100NF 100NF
100NF 100NF
C9
C11
C13
100NF 100NF
100NF
MT48LC16M16A2P-75IT
SDCS
R2
0R
R3
470K
256 Mbits
R4
256 Mbits
0R
Software Configuration
The following configuration has to be performed:
• Assign the EBI CS1 to the SDRAM controller by setting the bit EBI_CS1A in the EBI Chip
Select Assignment Register located in the bus matrix memory space.
• Initialize the SDRAM Controller depending on the SDRAM device and system bus frequency.
The Data Bus Width is to be programmed to 32 bits. The data lines D[16..31] are multiplexed
with PIO lines and thus the dedicated PIOs must be programmed in peripheral mode in the PIO
controller.
The SDRAM initialization sequence is described in the section “SDRAM Device Initialization” in
“SDRAM Controller (SDRAMC)”.
176
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
20.3.8.5
8-bit NAND Flash
Hardware Configuration
D[0..7]
U1
CLE
ALE
NANDOE
NANDWE
(ANY PIO)
(ANY PIO)
3V3
R1
10K
R2
10K
16
17
8
18
9
CLE
ALE
RE
WE
CE
7
R/B
19
WP
1
2
3
4
5
6
10
11
14
15
20
21
22
23
24
25
26
K9F2G08U0M
N.C
N.C
N.C
N.C
N.C
N.C
N.C
N.C
N.C
N.C
N.C
N.C
N.C
N.C
N.C
N.C
N.C
I/O0
I/O1
I/O2
I/O3
I/O4
I/O5
I/O6
I/O7
29
30
31
32
41
42
43
44
N.C
N.C
N.C
N.C
N.C
N.C
PRE
N.C
N.C
N.C
N.C
N.C
48
47
46
45
40
39
38
35
34
33
28
27
VCC
VCC
37
12
VSS
VSS
36
13
2 Gb
D0
D1
D2
D3
D4
D5
D6
D7
3V3
C2
100NF
C1
100NF
TSOP48 PACKAGE
Software Configuration
The following configuration has to be performed:
• Assign the EBI CS3 to the NAND Flash by setting the bit EBI_CS3A in the EBI Chip Select
Assignment Register located in the bus matrix memory space
• Reserve A21 / A22 for ALE / CLE functions. Address and Command Latches are controlled
respectively by setting to 1 the address bit A21 and A22 during accesses.
• Configure a PIO line as an input to manage the Ready/Busy signal.
• Configure Static Memory Controller CS3 Setup, Pulse, Cycle and Mode accordingly to NAND
Flash timings, the data bus width and the system bus frequency.
177
6438D–ATARM–13-Oct-09
20.3.8.6
16-bit NAND Flash
Hardware Configuration
D[0..15]
U1
CLE
ALE
NANDOE
NANDWE
(ANY PIO)
(ANY PIO)
3V3
R1
10K
R2
10K
16
17
8
18
9
CLE
ALE
RE
WE
CE
7
R/B
19
WP
1
2
3
4
5
6
10
11
14
15
20
21
22
23
24
34
35
N.C
N.C
N.C
N.C
N.C
N.C
N.C
N.C
N.C
N.C
N.C
N.C
N.C
N.C
N.C
N.C
N.C
MT29F2G16AABWP-ET
I/O0 26
I/O1 28
I/O2 30
I/O3 32
I/O4 40
I/O5 42
I/O6 44
I/O7 46
I/O8 27
I/O9 29
I/O10 31
I/O11 33
I/O12 41
I/O13 43
I/O14 45
I/O15 47
2 Gb
N.C
PRE
N.C
39
38
36
VCC
VCC
37
12
VSS
VSS
VSS
48
25
13
D0
D1
D2
D3
D4
D5
D6
D7
D8
D9
D10
D11
D12
D13
D14
D15
3V3
C2
100NF
C1
100NF
TSOP48 PACKAGE
Software Configuration
The software configuration is the same as for an 8-bit NAND Flash except for the data bus width
programmed in the mode register of the Static Memory Controller.
178
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
20.3.8.7
NOR Flash on NCS0
Hardware Configuration
D[0..15]
A[1..22]
A1
A2
A3
A4
A5
A6
A7
A8
A9
A10
A11
A12
A13
A14
A15
A16
A17
A18
A19
A20
A21
A22
NRST
NWE
NCS0
NRD
3V3
U1
25
24
23
22
21
20
19
18
8
7
6
5
4
3
2
1
48
17
16
15
10
9
A0
A1
A2
A3
A4
A5
A6
A7
A8
A9
A10
A11
A12
A13
A14
A15
A16
A17
A18
A19
A20
A21
12
11
14
13
26
28
RESET
WE
WP
VPP
CE
OE
DQ0
DQ1
DQ2
DQ3
DQ4
DQ5
DQ6
DQ7
DQ8
DQ9
DQ10
DQ11
DQ12
DQ13
DQ14
DQ15
29
31
33
35
38
40
42
44
30
32
34
36
39
41
43
45
D0
D1
D2
D3
D4
D5
D6
D7
D8
D9
D10
D11
D12
D13
D14
D15
AT49BV6416
3V3
VCCQ
47
VCC
37
VSS
VSS
46
27
TSOP48 PACKAGE
C2
100NF
C1
100NF
Software Configuration
The default configuration for the Static Memory Controller, byte select mode, 16-bit data bus,
Read/Write controlled by Chip Select, allows boot on 16-bit non-volatile memory at slow clock.
For another configuration, configure the Static Memory Controller CS0 Setup, Pulse, Cycle and
Mode depending on Flash timings and system bus frequency.
179
6438D–ATARM–13-Oct-09
20.3.8.8
CompactFlash
Hardware Configuration
MEMORY & I/O MODE
D[0..15]
MN1A
D15
D14
D13
D12
D11
D10
D9
D8
A2
A1
B2
B1
C2
C1
D2
D1
A3
A4
A25/CFRNW
4
(CFCS0 or CFCS1)
1A1
1A2
1A3
1A4
1A5
1A6
1A7
1A8
A5
A6
B5
B6
C5
C6
D5
D6
CF_D15
CF_D14
CF_D13
CF_D12
CF_D11
CF_D10
CF_D9
CF_D8
E5
E6
F5
F6
G5
G6
H5
H6
CF_D7
CF_D6
CF_D5
CF_D4
CF_D3
CF_D2
CF_D1
CF_D0
1DIR
1OE
74ALVCH32245
MN1B
D7
D6
D5
D4
D3
D2
D1
D0
CFCSx
1B1
1B2
1B3
1B4
1B5
1B6
1B7
1B8
6
5
E2
E1
F2
F1
G2
G1
H2
H1
2B1
2B2
2B3
2B4
2B5
2B6
2B7
2B8
H3
H4
2DIR
2OE
2A1
2A2
2A3
2A4
2A5
2A6
2A7
2A8
R1
MN2A
47K
SN74ALVC32
74ALVCH32245
MN2B
SN74ALVC32
MN1C
A10
A9
A8
A7
A6
A5
A4
A3
J5
J6
K5
K6
L5
L6
M5
M6
3A1
3A2
3A3
3A4
3A5
3A6
3A7
3A8
J3
J4
3DIR
3OE
3V3
3B1
3B2
3B3
3B4
3B5
3B6
3B7
3B8
J2
J1
K2
K1
L2
L1
M2
M1
74ALVCH32245
MN1D
A2
A1
A0
N5
N6
P5
P6
R5
R6
T6
T5
A22/REG
CFWE
CFOE
CFIOW
CFIOR
T3
T4
4A1
4A2
4A3
4A4
4A5
4A6
4A7
4A8
4B1
4B2
4B3
4B4
4B5
4B6
4B7
4B8
CD2
CD1
2
A[0..10]
R2
47K
1
3
(ANY PIO)
3V3
N2
N1
P2
P1
R2
R1
T1
T2
CF_A10
CF_A9
CF_A8
CF_A7
CF_A6
CF_A5
CF_A4
CF_A3
31
30
29
28
27
49
48
47
6
5
4
3
2
23
22
21
D15
D14
D13
D12
D11
D10
D9
D8
D7
D6
D5
D4
D3
D2
D1
D0
CD2
CD1
25
26
CD2#
CD1#
CF_A10
CF_A9
CF_A8
CF_A7
CF_A6
CF_A5
CF_A4
CF_A3
CF_A2
CF_A1
CF_A0
8
10
11
12
14
15
16
17
18
19
20
A10
A9
A8
A7
A6
A5
A4
A3
A2
A1
A0
REG
44
REG#
WE
OE
IOWR
IORD
36
9
35
34
WE#
OE#
IOWR#
IORD#
CE2
CE1
CF_A2
CF_A1
CF_A0
REG
WE
OE
IOWR
IORD
J1
CF_D15
CF_D14
CF_D13
CF_D12
CF_D11
CF_D10
CF_D9
CF_D8
CF_D7
CF_D6
CF_D5
CF_D4
CF_D3
CF_D2
CF_D1
CF_D0
VCC
38
VCC
13
GND
GND
50
1
CSEL#
39
INPACK#
43
BVD2
BVD1
45
46
32
7
CE2#
CE1#
24
WP
WAIT#
42
WAIT#
VS2#
VS1#
40
33
RESET
41
RESET
RDY/BSY
37
3V3
C1
100NF
C2
100NF
RDY/BSY
N7E50-7516VY-20
4DIR
4OE
1
74ALVCH32245
2
CFCE1
5
10
4
CFCE2
CFRST
9
(ANY PIO)
CFIRQ
11
13
(ANY PIO)
MN3A
SN74ALVC125
3
CE2
MN3B
SN74ALVC125
6
CE1
MN3C
SN74ALVC125
RESET
8
MN3D
R3
SN74ALVC125
10K
RDY/BSY
12
3V3
MN4
3V3
NWAIT
5 VCC
1
4
2
GND
WAIT#
R4
10K
3V3
3
SN74LVC1G125-Q1
Software Configuration
The following configuration has to be performed:
180
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
• Assign the EBI CS4 and/or EBI_CS5 to the CompactFlash Slot 0 and/or Slot 1 by setting the
bit EBI_CS4A and/or EBI_CS5A in the EBI Chip Select Assignment Register located in the
bus matrix memory space.
• The address line A23 is to select I/O (A23=1) or Memory mode (A23=0) and the address line
A22 for REG function.
• A22, A23, CFRNW, CFS0, CFCS1, CFCE1 and CFCE2 signals are multiplexed with PIO
lines and thus the dedicated PIOs must be programmed in peripheral mode in the PIO
controller.
• Configure a PIO line as an output for CFRST and two others as an input for CFIRQ and
CARD DETECT functions respectively.
• Configure SMC CS4 and/or SMC_CS5 (for Slot 0 or 1) Setup, Pulse, Cycle and Mode
accordingly to CompactFlash timings and system bus frequency.
181
6438D–ATARM–13-Oct-09
20.3.8.9
CompactFlash True IDE
Hardware Configuration
TRUE IDE MODE
D[0..15]
MN1A
D15
D14
D13
D12
D11
D10
D9
D8
A2
A1
B2
B1
C2
C1
D2
D1
A3
A4
1A1
1A2
1A3
1A4
1A5
1A6
1A7
1A8
CF_D15
CF_D14
CF_D13
CF_D12
CF_D11
CF_D10
CF_D9
CF_D8
E5
E6
F5
F6
G5
G6
H5
H6
CF_D7
CF_D6
CF_D5
CF_D4
CF_D3
CF_D2
CF_D1
CF_D0
1DIR
1OE
74ALVCH32245
MN1B
D7
D6
D5
D4
D3
D2
D1
D0
A25/CFRNW
4
CFCSx
(CFCS0 or CFCS1)
1B1
1B2
1B3
1B4
1B5
1B6
1B7
1B8
A5
A6
B5
B6
C5
C6
D5
D6
6
5
E2
E1
F2
F1
G2
G1
H2
H1
2B1
2B2
2B3
2B4
2B5
2B6
2B7
2B8
H3
H4
2DIR
2OE
2A1
2A2
2A3
2A4
2A5
2A6
2A7
2A8
R1
MN2A
47K
SN74ALVC32
74ALVCH32245
MN2B
SN74ALVC32
A[0..10]
J5
J6
K5
K6
L5
L6
M5
M6
3A1
3A2
3A3
3A4
3A5
3A6
3A7
3A8
J3
J4
3DIR
3OE
3V3
3B1
3B2
3B3
3B4
3B5
3B6
3B7
3B8
J2
J1
K2
K1
L2
L1
M2
M1
74ALVCH32245
MN1D
A2
A1
A0
N5
N6
P5
P6
R5
R6
T6
T5
A22/REG
CFWE
CFOE
CFIOW
CFIOR
T3
T4
4A1
4A2
4A3
4A4
4A5
4A6
4A7
4A8
D15
D14
D13
D12
D11
D10
D9
D8
D7
D6
D5
D4
D3
D2
D1
D0
CD2
CD1
25
26
CD2#
CD1#
CF_A2
CF_A1
CF_A0
8
10
11
12
14
15
16
17
18
19
20
A10
A9
A8
A7
A6
A5
A4
A3
A2
A1
A0
CD1
44
REG#
IOWR
IORD
36
9
35
34
WE#
ATA SEL#
IOWR#
IORD#
CE2
CE1
32
7
CS1#
CS0#
3V3
MN1C
A10
A9
A8
A7
A6
A5
A4
A3
CD2
2
31
30
29
28
27
49
48
47
6
5
4
3
2
23
22
21
R2
47K
1
3
(ANY PIO)
3V3
4B1
4B2
4B3
4B4
4B5
4B6
4B7
4B8
N2
N1
P2
P1
R2
R1
T1
T2
CF_A10
CF_A9
CF_A8
CF_A7
CF_A6
CF_A5
CF_A4
CF_A3
CF_A2
CF_A1
CF_A0
REG
WE
OE
IOWR
IORD
3V3
J1
CF_D15
CF_D14
CF_D13
CF_D12
CF_D11
CF_D10
CF_D9
CF_D8
CF_D7
CF_D6
CF_D5
CF_D4
CF_D3
CF_D2
CF_D1
CF_D0
24
IOIS16#
IORDY
42
IORDY
RESET#
41
VCC
38
VCC
13
GND
GND
50
1
CSEL#
39
INPACK#
43
DASP#
PDIAG#
45
46
VS2#
VS1#
40
33
INTRQ
37
RESET#
C1
100NF
C2
100NF
INTRQ
N7E50-7516VY-20
4DIR
4OE
1
74ALVCH32245
2
CFCE1
5
10
4
CFCE2
CFRST
9
(ANY PIO)
CFIRQ
11
13
(ANY PIO)
MN3A
SN74ALVC125
3
CE2
MN3B
SN74ALVC125
6
CE1
MN3C
SN74ALVC125
RESET#
8
MN3D
SN74ALVC125
INTRQ
12
R3
10K
3V3
MN4
3V3
NWAIT
5 VCC
1
4
2
GND
IORDY
R4
10K
3V3
3
SN74LVC1G125-Q1
Software Configuration
The following configuration has to be performed:
182
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
• Assign the EBI CS4 and/or EBI_CS5 to the CompactFlash Slot 0 and/or Slot 1 by setting the
bit EBI_CS4A and/or EBI_CS5A in the EBI Chip Select Assignment Register located in the
bus matrix memory space.
• The address line A21 is to select Alternate True IDE (A21=1) or True IDE (A21=0) modes.
• A21, CFRNW, CFS0, CFCS1, CFCE1 and CFCE2 signals are multiplexed with PIO lines and
thus the dedicated PIOs must be programmed in peripheral mode in the PIO controller.
• Configure a PIO line as an output for CFRST and two others as an input for CFIRQ and
CARD DETECT functions respectively.
• Configure SMC CS4 and/or SMC_CS5 (for Slot 0 or 1) Setup, Pulse, Cycle and Mode
accordingly to CompactFlash timings and system bus frequency.
183
6438D–ATARM–13-Oct-09
184
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
21. Static Memory Controller (SMC)
21.1
Description
The Static Memory Controller (SMC) generates the signals that control the access to the external memory devices or peripheral devices. It has 6 Chip Selects and a 26-bit address bus. The
32-bit data bus can be configured to interface with 8-, 16-, or 32-bit external devices. Separate
read and write control signals allow for direct memory and peripheral interfacing. Read and write
signal waveforms are fully parametrizable.
The SMC can manage wait requests from external devices to extend the current access. The
SMC is provided with an automatic slow clock mode. In slow clock mode, it switches from userprogrammed waveforms to slow-rate specific waveforms on read and write signals. The SMC
supports asynchronous burst read in page mode access for page size up to 32 bytes.
21.2
I/O Lines Description
Table 21-1.
I/O Line Description
Name
Description
Type
Active Level
NCS[7:0]
Static Memory Controller Chip Select Lines
Output
Low
NRD
Read Signal
Output
Low
NWR0/NWE
Write 0/Write Enable Signal
Output
Low
A0/NBS0
Address Bit 0/Byte 0 Select Signal
Output
Low
NWR1/NBS1
Write 1/Byte 1 Select Signal
Output
Low
A1/NWR2/NBS2
Address Bit 1/Write 2/Byte 2 Select Signal
Output
Low
NWR3/NBS3
Write 3/Byte 3 Select Signal
Output
Low
A[25:2]
Address Bus
Output
D[31:0]
Data Bus
NWAIT
External Wait Signal
21.3
I/O
Input
Low
Multiplexed Signals
Table 21-2.
Static Memory Controller (SMC) Multiplexed Signals
Multiplexed Signals
Related Function
NWR0
NWE
Byte-write or byte-select access, see “Byte Write or Byte Select Access” on page 185
A0
NBS0
8-bit or 16-/32-bit data bus, see “Data Bus Width” on page 185
NWR1
NBS1
Byte-write or byte-select access see “Byte Write or Byte Select Access” on page 185
A1
NWR2
NWR3
NBS3
NBS2
8-/16-bit or 32-bit data bus, see “Data Bus Width” on page 185.
Byte-write or byte-select access, see “Byte Write or Byte Select Access” on page 185
Byte-write or byte-select access see “Byte Write or Byte Select Access” on page 185
183
6438D–ATARM–13-Oct-09
21.4
21.4.1
Application Example
Hardware Interface
Figure 21-1. SMC Connections to Static Memory Devices
D0-D31
A0/NBS0
NWR0/NWE
NWR1/NBS1
A1/NWR2/NBS2
NWR3/NBS3
D0 - D7
128K x 8
SRAM
D8-D15
D0 - D7
CS
NRD
NWR0/NWE
A2 - A25
A2 - A18
A0 - A16
NRD
OE
NWR1/NBS1
WE
128K x 8
SRAM
D16 - D23
D24-D31
D0 - D7
A0 - A16
NRD
Static Memory
Controller
21.5
21.5.1
A2 - A18
OE
WE
128K x 8
SRAM
D0-D7
CS
CS
A1/NWR2/NBS2
D0-D7
CS
A0 - A16
NCS0
NCS1
NCS2
NCS3
NCS4
NCS5
NCS6
NCS7
128K x 8
SRAM
A2 - A18
A2 - A18
A0 - A16
NRD
OE
WE
OE
NWR3/NBS3
WE
Product Dependencies
I/O Lines
The pins used for interfacing the Static Memory Controller may be multiplexed with the PIO
lines. The programmer must first program the PIO controller to assign the Static Memory Controller pins to their peripheral function. If I/O Lines of the SMC are not used by the application,
they can be used for other purposes by the PIO Controller.
184
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
21.6
External Memory Mapping
The SMC provides up to 26 address lines, A[25:0]. This allows each chip select line to address
up to 64 Mbytes of memory.
If the physical memory device connected on one chip select is smaller than 64 Mbytes, it wraps
around and appears to be repeated within this space. The SMC correctly handles any valid
access to the memory device within the page (see Figure 21-2).
A[25:0] is only significant for 8-bit memory, A[25:1] is used for 16-bit memory, A[25:2] is used for
32-bit memory.
Figure 21-2.
Memory Connections for Eight External Devices
NCS[0] - NCS[7]
NCS7
NRD
SMC
NCS6
NWE
NCS5
A[25:0]
NCS4
D[31:0]
NCS3
NCS2
NCS1
NCS0
Memory Enable
Memory Enable
Memory Enable
Memory Enable
Memory Enable
Memory Enable
Memory Enable
Memory Enable
Output Enable
Write Enable
A[25:0]
8 or 16 or 32
21.7
21.7.1
D[31:0] or D[15:0] or
D[7:0]
Connection to External Devices
Data Bus Width
A data bus width of 8, 16, or 32 bits can be selected for each chip select. This option is controlled by the field DBW in SMC_MODE (Mode Register) for the corresponding chip select.
Figure 21-3 shows how to connect a 512K x 8-bit memory on NCS2. Figure 21-4 shows how to
connect a 512K x 16-bit memory on NCS2. Figure 21-5 shows two 16-bit memories connected
as a single 32-bit memory
21.7.2
Byte Write or Byte Select Access
Each chip select with a 16-bit or 32-bit data bus can operate with one of two different types of
write access: byte write or byte select access. This is controlled by the BAT field of the
SMC_MODE register for the corresponding chip select.
185
6438D–ATARM–13-Oct-09
Figure 21-3.
Memory Connection for an 8-bit Data Bus
D[7:0]
D[7:0]
A[18:2]
A[18:2]
SMC
A0
A0
A1
A1
NWE
Write Enable
NRD
Output Enable
NCS[2]
Figure 21-4.
Memory Enable
Memory Connection for a 16-bit Data Bus
D[15:0]
D[15:0]
A[19:2]
A[18:1]
A1
SMC
A[0]
NBS0
Low Byte Enable
NBS1
High Byte Enable
NWE
Write Enable
NRD
Output Enable
NCS[2]
Memory Enable
Figure 21-5. Memory Connection for a 32-bit Data Bus
D[31:16]
SMC
D[15:0]
D[15:0]
A[20:2]
A[18:0]
NBS0
Byte 0 Enable
NBS1
Byte 1 Enable
NBS2
Byte 2 Enable
NBS3
Byte 3 Enable
NWE
Write Enable
NRD
Output Enable
NCS[2]
186
D[31:16]
Memory Enable
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
21.7.2.1
Byte Write Access
Byte write access supports one byte write signal per byte of the data bus and a single read
signal.
Note that the SMC does not allow boot in Byte Write Access mode.
• For 16-bit devices: the SMC provides NWR0 and NWR1 write signals for respectively byte0
(lower byte) and byte1 (upper byte) of a 16-bit bus. One single read signal (NRD) is provided.
Byte Write Access is used to connect 2 x 8-bit devices as a 16-bit memory.
• For 32-bit devices: NWR0, NWR1, NWR2 and NWR3, are the write signals of byte0 (lower
byte), byte1, byte2 and byte 3 (upper byte) respectively. One single read signal (NRD) is
provided.
Byte Write Access is used to connect 4 x 8-bit devices as a 32-bit memory.
Byte Write option is illustrated on Figure 21-6.
21.7.2.2
Byte Select Access
In this mode, read/write operations can be enabled/disabled at a byte level. One byte-select line
per byte of the data bus is provided. One NRD and one NWE signal control read and write.
• For 16-bit devices: the SMC provides NBS0 and NBS1 selection signals for respectively
byte0 (lower byte) and byte1 (upper byte) of a 16-bit bus.
Byte Select Access is used to connect one 16-bit device.
• For 32-bit devices: NBS0, NBS1, NBS2 and NBS3, are the selection signals of byte0 (lower
byte), byte1, byte2 and byte 3 (upper byte) respectively. Byte Select Access is used to
connect two 16-bit devices.
Figure 21-7 shows how to connect two 16-bit devices on a 32-bit data bus in Byte Select Access
mode, on NCS3 (BAT = Byte Select Access).
187
6438D–ATARM–13-Oct-09
Figure 21-6.
Connection of 2 x 8-bit Devices on a 16-bit Bus: Byte Write Option
D[7:0]
D[7:0]
D[15:8]
A[24:2]
SMC
A1
NWR0
A[23:1]
A[0]
Write Enable
NWR1
NRD
NCS[3]
Read Enable
Memory Enable
D[15:8]
A[23:1]
A[0]
Write Enable
Read Enable
Memory Enable
21.7.2.3
Signal Multiplexing
Depending on the BAT, only the write signals or the byte select signals are used. To save IOs at
the external bus interface, control signals at the SMC interface are multiplexed. Table 21-3
shows signal multiplexing depending on the data bus width and the byte access type.
For 32-bit devices, bits A0 and A1 are unused. For 16-bit devices, bit A0 of address is unused.
When Byte Select Option is selected, NWR1 to NWR3 are unused. When Byte Write option is
selected, NBS0 to NBS3 are unused.
188
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 21-7. Connection of 2x16-bit Data Bus on a 32-bit Data Bus (Byte Select Option)
D[15:0]
D[15:0]
D[31:16]
A[25:2]
SMC
A[23:0]
NWE
Write Enable
NBS0
Low Byte Enable
NBS1
High Byte Enable
NBS2
NBS3
Read Enable
NRD
Memory Enable
NCS[3]
D[31:16]
A[23:0]
Write Enable
Low Byte Enable
High Byte Enable
Read Enable
Memory Enable
Table 21-3.
SMC Multiplexed Signal Translation
Signal Name
Device Type
32-bit Bus
16-bit Bus
8-bit Bus
1x32-bit
2x16-bit
4 x 8-bit
1x16-bit
2 x 8-bit
Byte Select
Byte Select
Byte Write
Byte Select
Byte Write
NBS0_A0
NBS0
NBS0
NWE_NWR0
NWE
NWE
NWR0
NWE
NWR0
NBS1_NWR1
NBS1
NBS1
NWR1
NBS1
NWR1
NBS2_NWR2_A1
NBS2
NBS2
NWR2
A1
A1
NBS3_NWR3
NBS3
NBS3
NWR3
Byte Access Type (BAT)
NBS0
1 x 8-bit
A0
NWE
A1
189
6438D–ATARM–13-Oct-09
21.8
Standard Read and Write Protocols
In the following sections, the byte access type is not considered. Byte select lines (NBS0 to
NBS3) always have the same timing as the A address bus. NWE represents either the NWE signal in byte select access type or one of the byte write lines (NWR0 to NWR3) in byte write
access type. NWR0 to NWR3 have the same timings and protocol as NWE. In the same way,
NCS represents one of the NCS[0..5] chip select lines.
21.8.1
Read Waveforms
The read cycle is shown on Figure 21-8.
The read cycle starts with the address setting on the memory address bus, i.e.:
{A[25:2], A1, A0} for 8-bit devices
{A[25:2], A1} for 16-bit devices
A[25:2] for 32-bit devices.
Figure 21-8. Standard Read Cycle
MCK
A[25:2]
NBS0,NBS1,
NBS2,NBS3,
A0, A1
NRD
NCS
D[31:0]
NRD_SETUP
NCS_RD_SETUP
NRD_PULSE
NCS_RD_PULSE
NRD_HOLD
NCS_RD_HOLD
NRD_CYCLE
21.8.1.1
NRD Waveform
The NRD signal is characterized by a setup timing, a pulse width and a hold timing.
1. NRD_SETUP: the NRD setup time is defined as the setup of address before the NRD
falling edge;
2. NRD_PULSE: the NRD pulse length is the time between NRD falling edge and NRD
rising edge;
3. NRD_HOLD: the NRD hold time is defined as the hold time of address after the NRD
rising edge.
190
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
21.8.1.2
NCS Waveform
Similarly, the NCS signal can be divided into a setup time, pulse length and hold time:
1. NCS_RD_SETUP: the NCS setup time is defined as the setup time of address before
the NCS falling edge.
2. NCS_RD_PULSE: the NCS pulse length is the time between NCS falling edge and
NCS rising edge;
3. NCS_RD_HOLD: the NCS hold time is defined as the hold time of address after the
NCS rising edge.
21.8.1.3
Read Cycle
The NRD_CYCLE time is defined as the total duration of the read cycle, i.e., from the time where
address is set on the address bus to the point where address may change. The total read cycle
time is equal to:
NRD_CYCLE = NRD_SETUP + NRD_PULSE + NRD_HOLD
= NCS_RD_SETUP + NCS_RD_PULSE + NCS_RD_HOLD
All NRD and NCS timings are defined separately for each chip select as an integer number of
Master Clock cycles. To ensure that the NRD and NCS timings are coherent, user must define
the total read cycle instead of the hold timing. NRD_CYCLE implicitly defines the NRD hold time
and NCS hold time as:
NRD_HOLD = NRD_CYCLE - NRD SETUP - NRD PULSE
NCS_RD_HOLD = NRD_CYCLE - NCS_RD_SETUP - NCS_RD_PULSE
21.8.1.4
Null Delay Setup and Hold
If null setup and hold parameters are programmed for NRD and/or NCS, NRD and NCS remain
active continuously in case of consecutive read cycles in the same memory (see Figure 21-9).
191
6438D–ATARM–13-Oct-09
Figure 21-9. No Setup, No Hold On NRD and NCS Read Signals
MCK
A[25:2]
NBS0,NBS1,
NBS2,NBS3,
A0, A1
NRD
NCS
D[31:0]
NRD_PULSE
NCS_RD_PULSE
NRD_CYCLE
21.8.1.5
NRD_PULSE
NCS_RD_PULSE
NRD_CYCLE
NRD_PULSE
NCS_RD_PULSE
NRD_CYCLE
Null Pulse
Programming null pulse is not permitted. Pulse must be at least set to 1. A null value leads to
unpredictable behavior.
21.8.2
Read Mode
As NCS and NRD waveforms are defined independently of one other, the SMC needs to know
when the read data is available on the data bus. The SMC does not compare NCS and NRD timings to know which signal rises first. The READ_MODE parameter in the SMC_MODE register
of the corresponding chip select indicates which signal of NRD and NCS controls the read
operation.
21.8.2.1
192
Read is Controlled by NRD (READ_MODE = 1):
Figure 21-10 shows the waveforms of a read operation of a typical asynchronous RAM. The
read data is available tPACC after the falling edge of NRD, and turns to ‘Z’ after the rising edge of
NRD. In this case, the READ_MODE must be set to 1 (read is controlled by NRD), to indicate
that data is available with the rising edge of NRD. The SMC samples the read data internally on
the rising edge of Master Clock that generates the rising edge of NRD, whatever the programmed waveform of NCS may be.
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 21-10. READ_MODE = 1: Data is sampled by SMC before the rising edge of NRD
MCK
A[25:2]
NBS0,NBS1,
NBS2,NBS3,
A0, A1
NRD
NCS
tPACC
D[31:0]
Data Sampling
21.8.2.2
Read is Controlled by NCS (READ_MODE = 0)
Figure 21-11 shows the typical read cycle of an LCD module. The read data is valid tPACC after
the falling edge of the NCS signal and remains valid until the rising edge of NCS. Data must be
sampled when NCS is raised. In that case, the READ_MODE must be set to 0 (read is controlled
by NCS): the SMC internally samples the data on the rising edge of Master Clock that generates
the rising edge of NCS, whatever the programmed waveform of NRD may be.
Figure 21-11. READ_MODE = 0: Data is sampled by SMC before the rising edge of NCS
MCK
A[25:2]
NBS0,NBS1,
NBS2,NBS3,
A0, A1
NRD
NCS
tPACC
D[31:0]
Data Sampling
193
6438D–ATARM–13-Oct-09
21.8.3
21.8.3.1
Write Waveforms
The write protocol is similar to the read protocol. It is depicted in Figure 21-12. The write cycle
starts with the address setting on the memory address bus.
NWE Waveforms
The NWE signal is characterized by a setup timing, a pulse width and a hold timing.
1. NWE_SETUP: the NWE setup time is defined as the setup of address and data before
the NWE falling edge;
2. NWE_PULSE: The NWE pulse length is the time between NWE falling edge and NWE
rising edge;
3. NWE_HOLD: The NWE hold time is defined as the hold time of address and data after
the NWE rising edge.
The NWE waveforms apply to all byte-write lines in Byte Write access mode: NWR0 to NWR3.
21.8.3.2
NCS Waveforms
The NCS signal waveforms in write operation are not the same that those applied in read operations, but are separately defined:
1. NCS_WR_SETUP: the NCS setup time is defined as the setup time of address before
the NCS falling edge.
2. NCS_WR_PULSE: the NCS pulse length is the time between NCS falling edge and
NCS rising edge;
3. NCS_WR_HOLD: the NCS hold time is defined as the hold time of address after the
NCS rising edge.
Figure 21-12. Write Cycle
MCK
A[25:2]
NBS0, NBS1,
NBS2, NBS3,
A0, A1
NWE
NCS
NWE_SETUP
NCS_WR_SETUP
NWE_PULSE
NCS_WR_PULSE
NWE_HOLD
NCS_WR_HOLD
NWE_CYCLE
194
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
21.8.3.3
Write Cycle
The write_cycle time is defined as the total duration of the write cycle, that is, from the time
where address is set on the address bus to the point where address may change. The total write
cycle time is equal to:
NWE_CYCLE = NWE_SETUP + NWE_PULSE + NWE_HOLD
= NCS_WR_SETUP + NCS_WR_PULSE + NCS_WR_HOLD
All NWE and NCS (write) timings are defined separately for each chip select as an integer number of Master Clock cycles. To ensure that the NWE and NCS timings are coherent, the user
must define the total write cycle instead of the hold timing. This implicitly defines the NWE hold
time and NCS (write) hold times as:
NWE_HOLD = NWE_CYCLE - NWE_SETUP - NWE_PULSE
NCS_WR_HOLD = NWE_CYCLE - NCS_WR_SETUP - NCS_WR_PULSE
21.8.3.4
Null Delay Setup and Hold
If null setup parameters are programmed for NWE and/or NCS, NWE and/or NCS remain active
continuously in case of consecutive write cycles in the same memory (see Figure 21-13). However, for devices that perform write operations on the rising edge of NWE or NCS, such as
SRAM, either a setup or a hold must be programmed.
Figure 21-13. Null Setup and Hold Values of NCS and NWE in Write Cycle
MCK
A[25:2]
NBS0, NBS1,
NBS2, NBS3,
A0, A1
NWE,
NWR0, NWR1,
NWR2, NWR3
NCS
D[31:0]
NWE_PULSE
21.8.3.5
NWE_PULSE
NWE_PULSE
NCS_WR_PULSE
NCS_WR_PULSE
NCS_WR_PULSE
NWE_CYCLE
NWE_CYCLE
NWE_CYCLE
Null Pulse
Programming null pulse is not permitted. Pulse must be at least set to 1. A null value leads to
unpredictable behavior.
195
6438D–ATARM–13-Oct-09
21.8.4
Write Mode
The WRITE_MODE parameter in the SMC_MODE register of the corresponding chip select indicates which signal controls the write operation.
21.8.4.1
Write is Controlled by NWE (WRITE_MODE = 1):
Figure 21-14 shows the waveforms of a write operation with WRITE_MODE set to 1. The data is
put on the bus during the pulse and hold steps of the NWE signal. The internal data buffers are
turned out after the NWE_SETUP time, and until the end of the write cycle, regardless of the
programmed waveform on NCS.
Figure 21-14. WRITE_MODE = 1. The write operation is controlled by NWE
MCK
A[25:2]
NBS0, NBS1,
NBS2, NBS3,
A0, A1
NWE,
NWR0, NWR1,
NWR2, NWR3
NCS
D[31:0]
21.8.4.2
196
Write is Controlled by NCS (WRITE_MODE = 0)
Figure 21-15 shows the waveforms of a write operation with WRITE_MODE set to 0. The data is
put on the bus during the pulse and hold steps of the NCS signal. The internal data buffers are
turned out after the NCS_WR_SETUP time, and until the end of the write cycle, regardless of
the programmed waveform on NWE.
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 21-15. WRITE_MODE = 0. The write operation is controlled by NCS
MCK
A[25:2]
NBS0, NBS1,
NBS2, NBS3,
A0, A1
NWE,
NWR0, NWR1,
NWR2, NWR3
NCS
D[31:0]
21.8.5
Coding Timing Parameters
All timing parameters are defined for one chip select and are grouped together in one
SMC_REGISTER according to their type.
The SMC_SETUP register groups the definition of all setup parameters:
• NRD_SETUP, NCS_RD_SETUP, NWE_SETUP, NCS_WR_SETUP
The SMC_PULSE register groups the definition of all pulse parameters:
• NRD_PULSE, NCS_RD_PULSE, NWE_PULSE, NCS_WR_PULSE
The SMC_CYCLE register groups the definition of all cycle parameters:
• NRD_CYCLE, NWE_CYCLE
Table 21-4 shows how the timing parameters are coded and their permitted range.
Table 21-4.
Coding and Range of Timing Parameters
Permitted Range
Coded Value
Number of Bits
Effective Value
Coded Value
Effective Value
setup [5:0]
6
128 x setup[5] + setup[4:0]
0 ≤≤31
0 ≤≤128+31
pulse [6:0]
7
256 x pulse[6] + pulse[5:0]
0 ≤≤63
0 ≤≤256+63
cycle [8:0]
9
256 x cycle[8:7] + cycle[6:0]
0 ≤≤127
0 ≤≤256+127
0 ≤≤512+127
0 ≤≤768+127
21.8.6
Reset Values of Timing Parameters
Table 21-8 gives the default value of timing parameters at reset.
197
6438D–ATARM–13-Oct-09
21.8.7
Usage Restriction
The SMC does not check the validity of the user-programmed parameters. If the sum of SETUP
and PULSE parameters is larger than the corresponding CYCLE parameter, this leads to unpredictable behavior of the SMC.
For read operations:
Null but positive setup and hold of address and NRD and/or NCS can not be guaranteed at the
memory interface because of the propagation delay of theses signals through external logic and
pads. If positive setup and hold values must be verified, then it is strictly recommended to program non-null values so as to cover possible skews between address, NCS and NRD signals.
For write operations:
If a null hold value is programmed on NWE, the SMC can guarantee a positive hold of address,
byte select lines, and NCS signal after the rising edge of NWE. This is true for WRITE_MODE =
1 only. See “Early Read Wait State” on page 199.
For read and write operations: a null value for pulse parameters is forbidden and may lead to
unpredictable behavior.
In read and write cycles, the setup and hold time parameters are defined in reference to the
address bus. For external devices that require setup and hold time between NCS and NRD signals (read), or between NCS and NWE signals (write), these setup and hold times must be
converted into setup and hold times in reference to the address bus.
21.9
Automatic Wait States
Under certain circumstances, the SMC automatically inserts idle cycles between accesses to
avoid bus contention or operation conflict.
21.9.1
Chip Select Wait States
The SMC always inserts an idle cycle between 2 transfers on separate chip selects. This idle
cycle ensures that there is no bus contention between the de-activation of one device and the
activation of the next one.
During chip select wait state, all control lines are turned inactive: NBS0 to NBS3, NWR0 to
NWR3, NCS[0..5], NRD lines are all set to 1.
Figure 21-16 illustrates a chip select wait state between access on Chip Select 0 and Chip
Select 2.
198
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 21-16. Chip Select Wait State between a Read Access on NCS0 and a Write Access on NCS2
MCK
A[25:2]
NBS0, NBS1,
NBS2, NBS3,
A0,A1
NRD
NWE
NCS0
NCS2
NWE_CYCLE
NRD_CYCLE
D[31:0]
Read to Write Chip Select
Wait State
Wait State
21.9.2
Early Read Wait State
In some cases, the SMC inserts a wait state cycle between a write access and a read access to
allow time for the write cycle to end before the subsequent read cycle begins. This wait state is
not generated in addition to a chip select wait state. The early read cycle thus only occurs
between a write and read access to the same memory device (same chip select).
An early read wait state is automatically inserted if at least one of the following conditions is
valid:
• if the write controlling signal has no hold time and the read controlling signal has no setup
time (Figure 21-17).
• in NCS write controlled mode (WRITE_MODE = 0), if there is no hold timing on the NCS
signal and the NCS_RD_SETUP parameter is set to 0, regardless of the read mode (Figure
21-18). The write operation must end with a NCS rising edge. Without an Early Read Wait
State, the write operation could not complete properly.
• in NWE controlled mode (WRITE_MODE = 1) and if there is no hold timing (NWE_HOLD =
0), the feedback of the write control signal is used to control address, data, chip select and
byte select lines. If the external write control signal is not inactivated as expected due to load
capacitances, an Early Read Wait State is inserted and address, data and control signals are
maintained one more cycle. See Figure 21-19.
199
6438D–ATARM–13-Oct-09
Figure 21-17. Early Read Wait State: Write with No Hold Followed by Read with No Setup
MCK
A[25:2]
NBS0, NBS1,
NBS2, NBS3,
A0, A1
NWE
NRD
no hold
no setup
D[31:0]
write cycle
Early Read
wait state
read cycle
Figure 21-18. Early Read Wait State: NCS Controlled Write with No Hold Followed by a Read with No NCS Setup
MCK
A[25:2]
NBS0, NBS1,
NBS2, NBS3,
A0,A1
NCS
NRD
no hold
no setup
D[31:0]
write cycle
(WRITE_MODE = 0)
200
Early Read
wait state
read cycle
(READ_MODE = 0 or READ_MODE = 1)
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 21-19. Early Read Wait State: NWE-controlled Write with No Hold Followed by a Read with one Set-up Cycle
MCK
A[25:2]
NBS0, NBS1,
NBS2, NBS3,
A0, A1
internal write controlling signal
external write controlling signal
(NWE)
no hold
read setup = 1
NRD
D[31:0]
write cycle
(WRITE_MODE = 1)
21.9.3
Early Read
wait state
read cycle
(READ_MODE = 0 or READ_MODE = 1)
Reload User Configuration Wait State
The user may change any of the configuration parameters by writing the SMC user interface.
When detecting that a new user configuration has been written in the user interface, the SMC
inserts a wait state before starting the next access. The so called “Reload User Configuration
Wait State” is used by the SMC to load the new set of parameters to apply to next accesses.
The Reload Configuration Wait State is not applied in addition to the Chip Select Wait State. If
accesses before and after re-programming the user interface are made to different devices
(Chip Selects), then one single Chip Select Wait State is applied.
On the other hand, if accesses before and after writing the user interface are made to the same
device, a Reload Configuration Wait State is inserted, even if the change does not concern the
current Chip Select.
21.9.3.1
User Procedure
To insert a Reload Configuration Wait State, the SMC detects a write access to any
SMC_MODE register of the user interface. If the user only modifies timing registers
(SMC_SETUP, SMC_PULSE, SMC_CYCLE registers) in the user interface, he must validate
the modification by writing the SMC_MODE, even if no change was made on the mode
parameters.
The user must not change the configuration parameters of an SMC Chip Select (Setup, Pulse,
Cycle, Mode) if accesses are performed on this CS during the modification. Any change of the
Chip Select parameters, while fetching the code from a memory connected on this CS, may lead
201
6438D–ATARM–13-Oct-09
to unpredictable behavior. The instructions used to modify the parameters of an SMC Chip
Select can be executed from the internal RAM or from a memory connected to another CS.
21.9.3.2
21.9.4
Slow Clock Mode Transition
A Reload Configuration Wait State is also inserted when the Slow Clock Mode is entered or
exited, after the end of the current transfer (see “Slow Clock Mode” on page 213).
Read to Write Wait State
Due to an internal mechanism, a wait cycle is always inserted between consecutive read and
write SMC accesses.
This wait cycle is referred to as a read to write wait state in this document.
This wait cycle is applied in addition to chip select and reload user configuration wait states
when they are to be inserted. See Figure 21-16 on page 199.
202
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
21.10 Data Float Wait States
Some memory devices are slow to release the external bus. For such devices, it is necessary to
add wait states (data float wait states) after a read access:
• before starting a read access to a different external memory
• before starting a write access to the same device or to a different external one.
The Data Float Output Time (t DF ) for each external memory device is programmed in the
TDF_CYCLES field of the SMC_MODE register for the corresponding chip select. The value of
TDF_CYCLES indicates the number of data float wait cycles (between 0 and 15) before the
external device releases the bus, and represents the time allowed for the data output to go to
high impedance after the memory is disabled.
Data float wait states do not delay internal memory accesses. Hence, a single access to an
external memory with long t DF will not slow down the execution of a program from internal
memory.
The data float wait states management depends on the READ_MODE and the TDF_MODE
fields of the SMC_MODE register for the corresponding chip select.
21.10.1
READ_MODE
Setting the READ_MODE to 1 indicates to the SMC that the NRD signal is responsible for turning off the tri-state buffers of the external memory device. The Data Float Period then begins
after the rising edge of the NRD signal and lasts TDF_CYCLES MCK cycles.
When the read operation is controlled by the NCS signal (READ_MODE = 0), the TDF field gives
the number of MCK cycles during which the data bus remains busy after the rising edge of NCS.
Figure 21-20 illustrates the Data Float Period in NRD-controlled mode (READ_MODE =1),
assuming a data float period of 2 cycles (TDF_CYCLES = 2). Figure 21-21 shows the read operation when controlled by NCS (READ_MODE = 0) and the TDF_CYCLES parameter equals 3.
203
6438D–ATARM–13-Oct-09
Figure 21-20. TDF Period in NRD Controlled Read Access (TDF = 2)
MCK
A[25:2]
NBS0, NBS1,
NBS2, NBS3,
A0, A1
NRD
NCS
tpacc
D[31:0]
TDF = 2 clock cycles
NRD controlled read operation
Figure 21-21. TDF Period in NCS Controlled Read Operation (TDF = 3)
MCK
A[25:2]
NBS0, NBS1,
NBS2, NBS3,
A0,A1
NRD
NCS
tpacc
D[31:0]
TDF = 3 clock cycles
NCS controlled read operation
204
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
21.10.2
TDF Optimization Enabled (TDF_MODE = 1)
When the TDF_MODE of the SMC_MODE register is set to 1 (TDF optimization is enabled), the
SMC takes advantage of the setup period of the next access to optimize the number of wait
states cycle to insert.
Figure 21-22 shows a read access controlled by NRD, followed by a write access controlled by
NWE, on Chip Select 0. Chip Select 0 has been programmed with:
NRD_HOLD = 4; READ_MODE = 1 (NRD controlled)
NWE_SETUP = 3; WRITE_MODE = 1 (NWE controlled)
TDF_CYCLES = 6; TDF_MODE = 1 (optimization enabled).
Figure 21-22. TDF Optimization: No TDF wait states are inserted if the TDF period is over when the next access begins
MCK
A[25:2]
NRD
NRD_HOLD= 4
NWE
NWE_SETUP= 3
NCS0
TDF_CYCLES = 6
D[31:0]
read access on NCS0 (NRD controlled)
21.10.3
Read to Write
Wait State
write access on NCS0 (NWE controlled)
TDF Optimization Disabled (TDF_MODE = 0)
When optimization is disabled, tdf wait states are inserted at the end of the read transfer, so that
the data float period is ended when the second access begins. If the hold period of the read1
controlling signal overlaps the data float period, no additional tdf wait states will be inserted.
Figure 21-23, Figure 21-24 and Figure 21-25 illustrate the cases:
• read access followed by a read access on another chip select,
• read access followed by a write access on another chip select,
• read access followed by a write access on the same chip select,
with no TDF optimization.
205
6438D–ATARM–13-Oct-09
Figure 21-23. TDF Optimization Disabled (TDF Mode = 0). TDF wait states between 2 read accesses on different chip
selects
MCK
A[25:2]
NBS0, NBS1,
NBS2, NBS3,
A0, A1
read1 controlling signal
(NRD)
read1 hold = 1
read2 controlling signal
(NRD)
read2 setup = 1
TDF_CYCLES = 6
D[31:0]
5 TDF WAIT STATES
read 2 cycle
TDF_MODE = 0
(optimization disabled)
read1 cycle
TDF_CYCLES = 6
Chip Select Wait State
Figure 21-24. TDF Mode = 0: TDF wait states between a read and a write access on different chip selects
MCK
A[25:2]
NBS0, NBS1,
NBS2, NBS3,
A0, A1
read1 controlling signal
(NRD)
read1 hold = 1
write2 controlling signal
(NWE)
write2 setup = 1
TDF_CYCLES = 4
D[31:0]
2 TDF WAIT STATES
read1 cycle
TDF_CYCLES = 4
Read to Write Chip Select
Wait State Wait State
206
write2 cycle
TDF_MODE = 0
(optimization disabled)
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 21-25. TDF Mode = 0: TDF wait states between read and write accesses on the same chip select
MCK
A[25:2]
NBS0, NBS1,
NBS2, NBS3,
A0, A1
read1 controlling signal
(NRD)
write2 setup = 1
read1 hold = 1
write2 controlling signal
(NWE)
TDF_CYCLES = 5
D[31:0]
4 TDF WAIT STATES
read1 cycle
TDF_CYCLES = 5
Read to Write
Wait State
write2 cycle
TDF_MODE = 0
(optimization disabled)
21.11 External Wait
Any access can be extended by an external device using the NWAIT input signal of the SMC.
The EXNW_MODE field of the SMC_MODE register on the corresponding chip select must be
set to either to “10” (frozen mode) or “11” (ready mode). When the EXNW_MODE is set to “00”
(disabled), the NWAIT signal is simply ignored on the corresponding chip select. The NWAIT
signal delays the read or write operation in regards to the read or write controlling signal,
depending on the read and write modes of the corresponding chip select.
21.11.1
Restriction
When one of the EXNW_MODE is enabled, it is mandatory to program at least one hold
cycle for the read/write controlling signal. For that reason, the NWAIT signal cannot be
used in Page Mode (“Asynchronous Page Mode” on page 216), or in Slow Clock Mode
(“Slow Clock Mode” on page 213).
The NWAIT signal is assumed to be a response of the external device to the read/write request
of the SMC. Then NWAIT is examined by the SMC only in the pulse state of the read or write
controlling signal. The assertion of the NWAIT signal outside the expected period has no impact
on SMC behavior.
207
6438D–ATARM–13-Oct-09
21.11.2
Frozen Mode
When the external device asserts the NWAIT signal (active low), and after internal synchronization of this signal, the SMC state is frozen, i.e., SMC internal counters are frozen, and all control
signals remain unchanged. When the resynchronized NWAIT signal is deasserted, the SMC
completes the access, resuming the access from the point where it was stopped. See Figure 2126. This mode must be selected when the external device uses the NWAIT signal to delay the
access and to freeze the SMC.
The assertion of the NWAIT signal outside the expected period is ignored as illustrated in Figure
21-27.
Figure 21-26. Write Access with NWAIT Assertion in Frozen Mode (EXNW_MODE = 10)
MCK
A[25:2]
NBS0, NBS1,
NBS2, NBS3,
A0,A1
FROZEN STATE
4
3
2
1
1
1
1
0
3
2
2
2
2
1
NWE
6
5
4
0
NCS
D[31:0]
NWAIT
internally synchronized
NWAIT signal
Write cycle
EXNW_MODE = 10 (Frozen)
WRITE_MODE = 1 (NWE_controlled)
NWE_PULSE = 5
NCS_WR_PULSE = 7
208
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 21-27. Read Access with NWAIT Assertion in Frozen Mode (EXNW_MODE = 10)
MCK
A[25:2]
NBS0, NBS1,
NBS2, NBS3,
A0,A1
NCS
FROZEN STATE
4
1
NRD
3
2
2
2
1
0
2
1
0
2
1
0
0
5
5
5
4
3
NWAIT
internally synchronized
NWAIT signal
Read cycle
EXNW_MODE = 10 (Frozen)
READ_MODE = 0 (NCS_controlled)
NRD_PULSE = 2, NRD_HOLD = 6
NCS_RD_PULSE =5, NCS_RD_HOLD =3
Assertion is ignored
209
6438D–ATARM–13-Oct-09
AT91SAM9G45
21.11.3
Ready Mode
In Ready mode (EXNW_MODE = 11), the SMC behaves differently. Normally, the SMC begins
the access by down counting the setup and pulse counters of the read/write controlling signal. In
the last cycle of the pulse phase, the resynchronized NWAIT signal is examined.
If asserted, the SMC suspends the access as shown in Figure 21-28 and Figure 21-29. After
deassertion, the access is completed: the hold step of the access is performed.
This mode must be selected when the external device uses deassertion of the NWAIT signal to
indicate its ability to complete the read or write operation.
If the NWAIT signal is deasserted before the end of the pulse, or asserted after the end of the
pulse of the controlling read/write signal, it has no impact on the access length as shown in Figure 21-29.
Figure 21-28. NWAIT Assertion in Write Access: Ready Mode (EXNW_MODE = 11)
MCK
A[25:2]
NBS0, NBS1,
NBS2, NBS3,
A0,A1
Wait STATE
4
3
2
1
0
0
0
3
2
1
1
1
NWE
6
5
4
0
NCS
D[31:0]
NWAIT
internally synchronized
NWAIT signal
Write cycle
EXNW_MODE = 11 (Ready mode)
WRITE_MODE = 1 (NWE_controlled)
NWE_PULSE = 5
NCS_WR_PULSE = 7
210
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 21-29. NWAIT Assertion in Read Access: Ready Mode (EXNW_MODE = 11)
MCK
A[25:2]
NBS0, NBS1,
NBS2, NBS3,
A0,A1
Wait STATE
6
5
4
3
2
1
0
0
6
5
4
3
2
1
1
NCS
NRD
0
NWAIT
internally synchronized
NWAIT signal
Read cycle
EXNW_MODE = 11(Ready mode)
READ_MODE = 0 (NCS_controlled)
Assertion is ignored
Assertion is ignored
NRD_PULSE = 7
NCS_RD_PULSE =7
211
6438D–ATARM–13-Oct-09
AT91SAM9G45
21.11.4
NWAIT Latency and Read/Write Timings
There may be a latency between the assertion of the read/write controlling signal and the assertion of the NWAIT signal by the device. The programmed pulse length of the read/write
controlling signal must be at least equal to this latency plus the 2 cycles of resynchronization + 1
cycle. Otherwise, the SMC may enter the hold state of the access without detecting the NWAIT
signal assertion. This is true in frozen mode as well as in ready mode. This is illustrated on Figure 21-30.
When EXNW_MODE is enabled (ready or frozen), the user must program a pulse length of the
read and write controlling signal of at least:
minimal pulse length = NWAIT latency + 2 resynchronization cycles + 1 cycle
Figure 21-30. NWAIT Latency
MCK
A[25:2]
NBS0, NBS1,
NBS2, NBS3,
A0,A1
WAIT STATE
4
3
2
1
0
0
0
NRD
minimal pulse length
NWAIT
intenally synchronized
NWAIT signal
NWAIT latency 2 cycle resynchronization
Read cycle
EXNW_MODE = 10 or 11
READ_MODE = 1 (NRD_controlled)
NRD_PULSE = 5
212
6438D–ATARM–13-Oct-09
AT91SAM9G45
21.12 Slow Clock Mode
The SMC is able to automatically apply a set of “slow clock mode” read/write waveforms when
an internal signal driven by the Power Management Controller is asserted because MCK has
been turned to a very slow clock rate (typically 32kHz clock rate). In this mode, the user-programmed waveforms are ignored and the slow clock mode waveforms are applied. This mode is
provided so as to avoid reprogramming the User Interface with appropriate waveforms at very
slow clock rate. When activated, the slow mode is active on all chip selects.
21.12.1
Slow Clock Mode Waveforms
Figure 21-31 illustrates the read and write operations in slow clock mode. They are valid on all
chip selects. Table 21-5 indicates the value of read and write parameters in slow clock mode.
Figure 21-31. Read/write Cycles in Slow Clock Mode
MCK
MCK
A[25:2]
A[25:2]
NBS0, NBS1,
NBS2, NBS3,
A0,A1
NBS0, NBS1,
NBS2, NBS3,
A0,A1
NWE
NRD
1
1
1
1
1
NCS
NCS
NRD_CYCLE = 2
NWE_CYCLE = 3
SLOW CLOCK MODE WRITE
Table 21-5.
SLOW CLOCK MODE READ
Read and Write Timing Parameters in Slow Clock Mode
Read Parameters
Duration (cycles)
Write Parameters
Duration (cycles)
NRD_SETUP
1
NWE_SETUP
1
NRD_PULSE
1
NWE_PULSE
1
NCS_RD_SETUP
0
NCS_WR_SETUP
0
NCS_RD_PULSE
2
NCS_WR_PULSE
3
NRD_CYCLE
2
NWE_CYCLE
3
213
6438D–ATARM–13-Oct-09
AT91SAM9G45
21.12.2
Switching from (to) Slow Clock Mode to (from) Normal Mode
When switching from slow clock mode to the normal mode, the current slow clock mode transfer
is completed at high clock rate, with the set of slow clock mode parameters.See Figure 21-32 on
page 214. The external device may not be fast enough to support such timings.
Figure 21-33 illustrates the recommended procedure to properly switch from one mode to the
other.
Figure 21-32. Clock Rate Transition Occurs while the SMC is Performing a Write Operation
Slow Clock Mode
internal signal from PMC
MCK
A[25:2]
NBS0, NBS1,
NBS2, NBS3,
A0,A1
NWE
1
1
1
1
1
1
2
3
2
NCS
NWE_CYCLE = 3
NWE_CYCLE = 7
SLOW CLOCK MODE WRITE SLOW CLOCK MODE WRITE
This write cycle finishes with the slow clock mode set
of parameters after the clock rate transition
NORMAL MODE WRITE
Slow clock mode transition is detected:
Reload Configuration Wait State
214
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 21-33. Recommended Procedure to Switch from Slow Clock Mode to Normal Mode or from Normal Mode to Slow
Clock Mode
Slow Clock Mode
internal signal from PMC
MCK
A[25:2]
NBS0, NBS1,
NBS2, NBS3,
A0,A1
NWE
1
1
1
2
3
2
NCS
SLOW CLOCK MODE WRITE
IDLE STATE
NORMAL MODE WRITE
Reload Configuration
Wait State
215
6438D–ATARM–13-Oct-09
AT91SAM9G45
21.13 Asynchronous Page Mode
The SMC supports asynchronous burst reads in page mode, providing that the page mode is
enabled in the SMC_MODE register (PMEN field). The page size must be configured in the
SMC_MODE register (PS field) to 4, 8, 16 or 32 bytes.
The page defines a set of consecutive bytes into memory. A 4-byte page (resp. 8-, 16-, 32-byte
page) is always aligned to 4-byte boundaries (resp. 8-, 16-, 32-byte boundaries) of memory. The
MSB of data address defines the address of the page in memory, the LSB of address define the
address of the data in the page as detailed in Table 21-6.
With page mode memory devices, the first access to one page (tpa) takes longer than the subsequent accesses to the page (tsa ) as shown in Figure 21-34. When in page mode, the SMC
enables the user to define different read timings for the first access within one page, and next
accesses within the page.
Table 21-6.
Page Address and Data Address within a Page
Page Size
Page Address(1)
Data Address in the Page(2)
4 bytes
A[25:2]
A[1:0]
8 bytes
A[25:3]
A[2:0]
16 bytes
A[25:4]
A[3:0]
32 bytes
A[25:5]
A[4:0]
Notes:
1. A denotes the address bus of the memory device
2. For 16-bit devices, the bit 0 of address is ignored. For 32-bit devices, bits [1:0] are ignored.
21.13.1
Protocol and Timings in Page Mode
Figure 21-34 shows the NRD and NCS timings in page mode access.
Figure 21-34. Page Mode Read Protocol (Address MSB and LSB are defined in Table 21-6)
MCK
A[MSB]
A[LSB]
NRD
NCS
tpa
tsa
tsa
D[31:0]
NCS_RD_PULSE
NRD_PULSE
NRD_PULSE
The NRD and NCS signals are held low during all read transfers, whatever the programmed values of the setup and hold timings in the User Interface may be. Moreover, the NRD and NCS
216
6438D–ATARM–13-Oct-09
AT91SAM9G45
timings are identical. The pulse length of the first access to the page is defined with the
NCS_RD_PULSE field of the SMC_PULSE register. The pulse length of subsequent accesses
within the page are defined using the NRD_PULSE parameter.
In page mode, the programming of the read timings is described in Table 21-7:
Table 21-7.
Programming of Read Timings in Page Mode
Parameter
Value
Definition
READ_MODE
‘x’
No impact
NCS_RD_SETUP
‘x’
No impact
NCS_RD_PULSE
tpa
Access time of first access to the page
NRD_SETUP
‘x’
No impact
NRD_PULSE
tsa
Access time of subsequent accesses in the page
NRD_CYCLE
‘x’
No impact
The SMC does not check the coherency of timings. It will always apply the NCS_RD_PULSE
timings as page access timing (tpa) and the NRD_PULSE for accesses to the page (tsa), even if
the programmed value for tpa is shorter than the programmed value for tsa.
21.13.2
Byte Access Type in Page Mode
The Byte Access Type configuration remains active in page mode. For 16-bit or 32-bit page
mode devices that require byte selection signals, configure the BAT field of the
SMC_REGISTER to 0 (byte select access type).
21.13.3
Page Mode Restriction
The page mode is not compatible with the use of the NWAIT signal. Using the page mode and
the NWAIT signal may lead to unpredictable behavior.
21.13.4
Sequential and Non-sequential Accesses
If the chip select and the MSB of addresses as defined in Table 21-6 are identical, then the current access lies in the same page as the previous one, and no page break occurs.
Using this information, all data within the same page, sequential or not sequential, are accessed
with a minimum access time (tsa). Figure 21-35 illustrates access to an 8-bit memory device in
page mode, with 8-byte pages. Access to D1 causes a page access with a long access time
(tpa). Accesses to D3 and D7, though they are not sequential accesses, only require a short
access time (tsa).
If the MSB of addresses are different, the SMC performs the access of a new page. In the same
way, if the chip select is different from the previous access, a page break occurs. If two sequential accesses are made to the page mode memory, but separated by an other internal or external
peripheral access, a page break occurs on the second access because the chip select of the
device was deasserted between both accesses.
217
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 21-35. Access to Non-sequential Data within the Same Page
MCK
Page address
A[25:3]
A[2], A1, A0
A1
A3
A7
NRD
NCS
D[7:0]
D1
NCS_RD_PULSE
D3
NRD_PULSE
D7
NRD_PULSE
218
6438D–ATARM–13-Oct-09
AT91SAM9G45
21.14 Programmable IO Delays
The external bus interface consists of a data bus, an address bus and control signals. The simultaneous switching outputs on these busses may lead to a peak of current in the internal and
external power supply lines.
In order to reduce the peak of current in such cases, additional propagation delays can be
adjusted independently for pad buffers by means of configuration registers, SMC_DELAY1-8.
The additional programmable delays for each IO range from 0 to 4 ns (Worst Case PVT). The
delay can differ between IOs supporting this feature. Delay can be modified per programming for
each IO. The minimal additional delay that can be programmed on a PAD suppporting this feature is 1/16 of the maximum programmable delay.
When programming 0x0 in fields “Delay1 to Delay 8”, no delay is added (reset value) and the
propagation delay of the pad buffers is the inherent delay of the pad buffer. When programming
0xF in field “Delay1” the propagation delay of the corresponding pad is maximal.
SMC_DELAY1, SMC_DELAY2 allow to configure delay on D[15:0], SMC_DELAY1[3:0] corresponds to D[0] and SMC_DELAY2[3:0] corresponds to D[8].
SMC_DELAY3, SMC_DELAY4 allow to configure delay on D[31:16], SMC_DELAY3[3:0] corresponds to D[16] and SMC_DELAY4[3:0] corresponds to D[24]. In case of multiplexing through
the PIO controller, refer to the alternate function of D[31:16].
SMC_DELAY5, 6, 7 and 8 allow to configure delay on A[25:0], SMC_DELAY5[3:0] corresponds
to A[0]. In case of multiplexing through the PIO controller, refer to the alternate function of
A[25:0].
Figure 21-36. Programmable IO Delays
SMC
D_in[0]
D_out[0]
Programmable Delay Line
D[0]
Programmable Delay Line
D[1]
Programmable Delay Line
D[n]
Programmable Delay Line
A[m]
DELAY1
D_in[1]
D_out[1]
DELAY2
PIO
D_in[n]
D_out[n]
DELAYx
PIO
A[m]
DELAYy
219
6438D–ATARM–13-Oct-09
AT91SAM9G45
21.15 Static Memory Controller (SMC) User Interface
The SMC is programmed using the registers listed in Table 21-8. For each chip select, a set of 4 registers is used to program the parameters of the external device connected on it. In Table 21-8, “CS_number” denotes the chip select number.
16 bytes (0x10) are required per chip select.
The user must complete writing the configuration by writing any one of the SMC_MODE registers.
Table 21-8.
Register Mapping
Offset
Register
Name
Access
Reset
0x10 x CS_number + 0x00
SMC Setup Register
SMC_SETUP
Read-write
0x01010101
0x10 x CS_number + 0x04
SMC Pulse Register
SMC_PULSE
Read-write
0x01010101
0x10 x CS_number + 0x08
SMC Cycle Register
SMC_CYCLE
Read-write
0x00030003
0x10 x CS_number + 0x0C
SMC Mode Register
SMC_MODE
Read-write
0x10001000
0xC0
SMC Delay on I/O
SMC_DELAY1
Read-write
0x00000000
0xC4
SMC Delay on I/O
SMC_DELAY2
Read-write
0x00000000
0xC8
SMC Delay on I/O
SMC_DELAY3
Read-write
0x00000000
0xCC
SMC Delay on I/O
SMC_DELAY4
Read-write
0x00000000
0xD0
SMC Delay on I/O
SMC_DELAY5
Read-write
0x00000000
0xD4
SMC Delay on I/O
SMC_DELAY6
Read-write
0x00000000
0xD8
SMC Delay on I/O
SMC_DELAY7
Read-write
0x00000000
0xDC
SMC Delay on I/O
SMC_DELAY8
Read-write
0x00000000
0xEC-0xFC
Reserved
-
-
-
220
6438D–ATARM–13-Oct-09
AT91SAM9G45
21.15.1 SMC Setup Register
Register Name:
SMC_SETUP[0..5]
Addresses:
0xFFFFE800 [0], 0xFFFFE810 [1], 0xFFFFE820 [2], 0xFFFFE830 [3], 0xFFFFE840 [4],
0xFFFFE850 [5]
Access Type:
Read-write
31
30
–
–
23
22
–
–
15
14
–
–
7
6
–
–
29
28
27
26
25
24
18
17
16
10
9
8
1
0
NCS_RD_SETUP
21
20
19
NRD_SETUP
13
12
11
NCS_WR_SETUP
5
4
3
2
NWE_SETUP
• NWE_SETUP: NWE Setup Length
The NWE signal setup length is defined as:
NWE setup length = (128* NWE_SETUP[5] + NWE_SETUP[4:0]) clock cycles
• NCS_WR_SETUP: NCS Setup Length in WRITE Access
In write access, the NCS signal setup length is defined as:
NCS setup length = (128* NCS_WR_SETUP[5] + NCS_WR_SETUP[4:0]) clock cycles
• NRD_SETUP: NRD Setup Length
The NRD signal setup length is defined in clock cycles as:
NRD setup length = (128* NRD_SETUP[5] + NRD_SETUP[4:0]) clock cycles
• NCS_RD_SETUP: NCS Setup Length in READ Access
In read access, the NCS signal setup length is defined as:
NCS setup length = (128* NCS_RD_SETUP[5] + NCS_RD_SETUP[4:0]) clock cycles
221
6438D–ATARM–13-Oct-09
AT91SAM9G45
21.15.2 SMC Pulse Register
Register Name:
SMC_PULSE[0..5]
Addresses:
0xFFFFE804 [0], 0xFFFFE814 [1], 0xFFFFE824 [2], 0xFFFFE834 [3], 0xFFFFE844 [4],
0xFFFFE854 [5]
Access Type:
Read-write
31
30
29
28
–
23
22
21
20
–
15
26
25
24
19
18
17
16
10
9
8
2
1
0
NRD_PULSE
14
13
12
–
7
27
NCS_RD_PULSE
11
NCS_WR_PULSE
6
5
4
–
3
NWE_PULSE
• NWE_PULSE: NWE Pulse Length
The NWE signal pulse length is defined as:
NWE pulse length = (256* NWE_PULSE[6] + NWE_PULSE[5:0]) clock cycles
The NWE pulse length must be at least 1 clock cycle.
• NCS_WR_PULSE: NCS Pulse Length in WRITE Access
In write access, the NCS signal pulse length is defined as:
NCS pulse length = (256* NCS_WR_PULSE[6] + NCS_WR_PULSE[5:0]) clock cycles
The NCS pulse length must be at least 1 clock cycle.
• NRD_PULSE: NRD Pulse Length
In standard read access, the NRD signal pulse length is defined in clock cycles as:
NRD pulse length = (256* NRD_PULSE[6] + NRD_PULSE[5:0]) clock cycles
The NRD pulse length must be at least 1 clock cycle.
In page mode read access, the NRD_PULSE parameter defines the duration of the subsequent accesses in the page.
• NCS_RD_PULSE: NCS Pulse Length in READ Access
In standard read access, the NCS signal pulse length is defined as:
NCS pulse length = (256* NCS_RD_PULSE[6] + NCS_RD_PULSE[5:0]) clock cycles
The NCS pulse length must be at least 1 clock cycle.
In page mode read access, the NCS_RD_PULSE parameter defines the duration of the first access to one page.
222
6438D–ATARM–13-Oct-09
AT91SAM9G45
21.15.3 SMC Cycle Register
Register Name:
SMC_CYCLE[0..5]
Addresses:
0xFFFFE808 [0], 0xFFFFE818 [1], 0xFFFFE828 [2], 0xFFFFE838 [3], 0xFFFFE848 [4],
0xFFFFE858 [5]
Access Type:
Read-write
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
NRD_CYCLE
23
22
21
20
19
18
17
16
NRD_CYCLE
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
NWE_CYCLE
7
6
5
4
3
2
1
0
NWE_CYCLE
• NWE_CYCLE: Total Write Cycle Length
The total write cycle length is the total duration in clock cycles of the write cycle. It is equal to the sum of the setup, pulse
and hold steps of the NWE and NCS signals. It is defined as:
Write cycle length = (NWE_CYCLE[8:7]*256 + NWE_CYCLE[6:0]) clock cycles
• NRD_CYCLE: Total Read Cycle Length
The total read cycle length is the total duration in clock cycles of the read cycle. It is equal to the sum of the setup, pulse
and hold steps of the NRD and NCS signals. It is defined as:
Read cycle length = (NRD_CYCLE[8:7]*256 + NRD_CYCLE[6:0]) clock cycles
223
6438D–ATARM–13-Oct-09
AT91SAM9G45
21.15.4 SMC MODE Register
Register Name:
SMC_MODE[0..5]
Addresses:
0xFFFFE80C [0], 0xFFFFE81C [1], 0xFFFFE82C [2], 0xFFFFE83C [3], 0xFFFFE84C [4],
0xFFFFE85C [5]
Access Type:
Read-write
31
30
–
–
29
28
23
22
21
20
–
–
–
TDF_MODE
15
14
13
–
–
7
6
–
–
PS
12
DBW
5
4
EXNW_MODE
27
26
25
24
–
–
–
PMEN
19
18
17
16
TDF_CYCLES
11
10
9
8
–
–
–
BAT
3
2
1
0
–
–
WRITE_MODE
READ_MODE
• READ_MODE:
1: The read operation is controlled by the NRD signal.
– If TDF cycles are programmed, the external bus is marked busy after the rising edge of NRD.
– If TDF optimization is enabled (TDF_MODE =1), TDF wait states are inserted after the setup of NRD.
0: The read operation is controlled by the NCS signal.
– If TDF cycles are programmed, the external bus is marked busy after the rising edge of NCS.
– If TDF optimization is enabled (TDF_MODE =1), TDF wait states are inserted after the setup of NCS.
• WRITE_MODE
1: The write operation is controlled by the NWE signal.
– If TDF optimization is enabled (TDF_MODE =1), TDF wait states will be inserted after the setup of NWE.
0: The write operation is controlled by the NCS signal.
– If TDF optimization is enabled (TDF_MODE =1), TDF wait states will be inserted after the setup of NCS.
• EXNW_MODE: NWAIT Mode
The NWAIT signal is used to extend the current read or write signal. It is only taken into account during the pulse phase of
the read and write controlling signal. When the use of NWAIT is enabled, at least one cycle hold duration must be programmed for the read and write controlling signal.
EXNW_MODE
NWAIT Mode
0
0
Disabled
0
1
Reserved
1
0
Frozen Mode
1
1
Ready Mode
• Disabled Mode: The NWAIT input signal is ignored on the corresponding Chip Select.
• Frozen Mode: If asserted, the NWAIT signal freezes the current read or write cycle. After deassertion, the read/write
cycle is resumed from the point where it was stopped.
224
6438D–ATARM–13-Oct-09
AT91SAM9G45
• Ready Mode: The NWAIT signal indicates the availability of the external device at the end of the pulse of the controlling
read or write signal, to complete the access. If high, the access normally completes. If low, the access is extended until
NWAIT returns high.
• BAT: Byte Access Type
This field is used only if DBW defines a 16- or 32-bit data bus.
• 1: Byte write access type:
– Write operation is controlled using NCS, NWR0, NWR1, NWR2, NWR3.
– Read operation is controlled using NCS and NRD.
• 0: Byte select access type:
– Write operation is controlled using NCS, NWE, NBS0, NBS1, NBS2 and NBS3
– Read operation is controlled using NCS, NRD, NBS0, NBS1, NBS2 and NBS3
• DBW: Data Bus Width
DBW
Data Bus Width
0
0
8-bit bus
0
1
16-bit bus
1
0
32-bit bus
1
1
Reserved
• TDF_CYCLES: Data Float Time
This field gives the integer number of clock cycles required by the external device to release the data after the rising edge
of the read controlling signal. The SMC always provide one full cycle of bus turnaround after the TDF_CYCLES period. The
external bus cannot be used by another chip select during TDF_CYCLES + 1 cycles. From 0 up to 15 TDF_CYCLES can
be set.
• TDF_MODE: TDF Optimization
1: TDF optimization is enabled.
– The number of TDF wait states is optimized using the setup period of the next read/write access.
0: TDF optimization is disabled.
– The number of TDF wait states is inserted before the next access begins.
• PMEN: Page Mode Enabled
1: Asynchronous burst read in page mode is applied on the corresponding chip select.
0: Standard read is applied.
• PS: Page Size
If page mode is enabled, this field indicates the size of the page in bytes.
PS
Page Size
0
0
4-byte page
0
1
8-byte page
1
0
16-byte page
1
1
32-byte page
225
6438D–ATARM–13-Oct-09
AT91SAM9G45
21.15.5 SMC DELAY I/O Register
Register Name:
SMC_DELAY 1-8
Addresses:
0xFFFFE8C0 [1], 0xFFFFE8C4 [2], 0xFFFFE8C8 [3], 0xFFFFE8CC [4], 0xFFFFE8D0 [5],
0xFFFFE8D4 [6], 0xFFFFE8D8 [7], 0xFFFFE8DC [8]
Access Type:
Read-write
Reset Value:
See Table 21-8
31
30
29
28
27
26
Delay8
23
22
21
20
19
18
Delay6
15
14
13
6
24
17
16
9
8
1
0
Delay5
12
11
10
Delay4
7
25
Delay7
Delay3
5
Delay2
4
3
2
Delay1
• Delay x:
Gives the number of elements in the delay line.
226
6438D–ATARM–13-Oct-09
AT91SAM9G45
22. DDR/SDR SDRAM Controller (DDRSDRC)
22.1
Description
The DDR/SDR SDRAM Controller (DDRSDRC) is a multiport memory controller. It comprises
four slave AHB interfaces. All simultaneous accesses (four independent AHB ports) are interleaved to maximize memory bandwidth and minimize transaction latency due to SDRAM
protocol.The DDRSDRC supports a read or write burst length of 8 locations which frees the
command and address bus to anticipate the next command, thus reducing latency imposed by
the SDRAM protocol and improving the SDRAM bandwidth. Moreover it keeps track of the active
row in each bank, thus maximizing SDRAM performance, e.g., the application may be placed in
one bank and data in the other banks. So as to optimize performance, it is advisable to avoid
accessing different rows in the same bank. The DDRSDRC supports a CAS latency of 2 or 3 and
optimizes the read access depending on the frequency.
The features of self refresh, power-down and deep power-down modes minimize the consumption of the SDRAM device.
The DDRSDRC user interface is compliant with ARM Advanced Peripheral Bus (APB rev2).
Note: The term “SDRAM device” regroups SDR-SDRAM, Mobile SDR-SDRAM, Mobile DDR1SDRAM and DDR2-SDRAM devices.
227
6438D–ATARM–13-Oct-09
22.2
DDRSDRC Module Diagram
Figure 22-1.
DDRSDRC Module Diagram
DDR-SDR Controller
AHB Slave Interface 0
Input
Stage
Power Management
clk/nclk
AHB Slave Interface 1
ras,cas,we
cke
Input
Stage
Output
Stage
AHB Slave Interface 2
Input
Stage
Memory Controller
Finite State Machine
SDRAM Signal Management
Arbiter
Addr, DQM
DQS
DDR-SDR
Devices
Data
odt
AHB Slave Interface 3
Input
Stage
Asynchronous Timing
Refresh Management
Interconnect Matrix
APB
Interface APB
DDRSDRC is partitioned in two blocks (see Figure 22-1):
• An Interconnect-Matrix that manages concurrent accesses on the AHB bus between four
AHB masters and integrates an arbiter.
• A controller that translates AHB requests (Read/Write) in the SDRAM protocol.
228
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
22.3
Product Dependencies
The addresses given are for example purposes only. The real address depends on implementation in the product.
22.3.1
SDR-SDRAM Initialization
The initialization sequence is generated by software. The SDR-SDRAM devices are initialized
by the following sequence:
1. Program the memory device type into the Memory Device Register (see Section 22.7.8
on page 269).
2. Program the features of the SDR-SDRAM device into the Timing Register (asynchronous timing (trc, tras, etc.)), and into the Configuration Register (number of columns,
rows, banks, cas latency) (see Section 22.7.3 on page 260, Section 22.7.4 on page 263
and Section 22.7.5 on page 265).
3. For low-power SDRAM, temperature-compensated self refresh (TCSR), drive strength
(DS) and partial array self refresh (PASR) must be set in the Low-power Register (see
Section 22.7.7 on page 267).
A minimum pause of 200 μs is provided to precede any signal toggle.
4. A NOP command is issued to the SDR-SDRAM. Program NOP command into Mode
Register, the application must set Mode to 1 in the Mode Register (See Section 22.7.1
on page 258). Perform a write access to any SDR-SDRAM address to acknowledge
this command. Now the clock which drives SDR-SDRAM device is enabled.
5. An all banks precharge command is issued to the SDR-SDRAM. Program all banks
precharge command into Mode Register, the application must set Mode to 2 in the
Mode Register (See Section 22.7.1 on page 258). Perform a write access to any SDRSDRAM address to acknowledge this command.
6. Eight auto-refresh (CBR) cycles are provided. Program the auto refresh command
(CBR) into Mode Register, the application must set Mode to 4 in the Mode Register
(see Section 22.7.1 on page 258).Performs a write access to any SDR-SDRAM location eight times to acknowledge these commands.
7. A Mode Register set (MRS) cycle is issued to program the parameters of the SDRSDRAM devices, in particular CAS latency and burst length. The application must set
Mode to 3 in the Mode Register (see Section 22.7.1 on page 258) and perform a write
access to the SDR-SDRAM to acknowledge this command. The write address must be
chosen so that BA[1:0] are set to 0. For example, with a 16-bit 128 MB SDR-SDRAM
(12 rows, 9 columns, 4 banks) bank address, the SDRAM write access should be done
at the address 0x20000000.
Note:
This address is for example purposes only. The real address is dependent on implementation in
the product.
8. For low-power SDR-SDRAM initialization, an Extended Mode Register set (EMRS)
cycle is issued to program the SDR-SDRAM parameters (TCSR, PASR, DS). The application must set Mode to 5 in the Mode Register (see Section 22.7.1 on page 258) and
perform a write access to the SDR-SDRAM to acknowledge this command. The write
address must be chosen so that BA[1] is set to 1 and BA[0] is set to 0. For example,
with a 16-bit 128 MB SDRAM, (12 rows, 9 columns, 4 banks) bank address the SDRAM
write access should be done at the address 0x20800000.
9. The application must go into Normal Mode, setting Mode to 0 in the Mode Register (see
Section 22.7.1 on page 258) and perform a write access at any location in the SDRAM
to acknowledge this command.
229
6438D–ATARM–13-Oct-09
10. Write the refresh rate into the count field in the DDRSDRC Refresh Timer register (see
page 259). (Refresh rate = delay between refresh cycles). The SDR-SDRAM device
requires a refresh every 15.625 μs or 7.81 μs. With a 100 MHz frequency, the refresh
timer count register must to be set with (15.625 /100 MHz) = 1562 i.e. 0x061A or (7.81
/100 MHz) = 781 i.e. 0x030d
After initialization, the SDR-SDRAM device is fully functional.
22.3.2
Low-power DDR1-SDRAM Initialization
The initialization sequence is generated by software. The low-power DDR1-SDRAM devices are
initialized by the following sequence:
1. Program the memory device type into the Memory Device Register (see Section 22.7.8
on page 269).
2. Program the features of the low-power DDR1-SDRAM device into the Configuration
Register: asynchronous timing (trc, tras, etc.), number of columns, rows, banks, cas
latency. See Section 22.7.3 on page 260, Section 22.7.4 on page 263 and Section
22.7.5 on page 265.
3. Program temperature compensated self refresh (tcr), Partial array self refresh (pasr)
and Drive strength (ds) into the Low-power Register. See Section 22.7.7 on page 267.
4. An NOP command will be issued to the low-power DDR1-SDRAM. Program NOP command into the Mode Register, the application must set Mode to 1 in the Mode Register
(see Section 22.7.1 on page 258). Perform a write access to any DDR1-SDRAM
address to acknowledge this command. Now clocks which drive DDR1-SDRAM device
are enabled.
A minimum pause of 200 μs will be provided to precede any signal toggle.
5. An all banks precharge command is issued to the low-power DDR1-SDRAM. Program
all banks precharge command into the Mode Register, the application must set Mode to
2 in the Mode Register (See Section 22.7.1 on page 258). Perform a write access to
any low-power DDR1-SDRAM address to acknowledge this command
6. Two auto-refresh (CBR) cycles are provided. Program the auto refresh command
(CBR) into the Mode Register, the application must set Mode to 4 in the Mode Register
(see Section 22.7.1 on page 258). Perform a write access to any low-power DDR1SDRAM location twice to acknowledge these commands.
7. An Extended Mode Register set (EMRS) cycle is issued to program the low-power
DDR1-SDRAM parameters (TCSR, PASR, DS). The application must set Mode to 5 in
the Mode Register (see Section 22.7.1 on page 258) and perform a write access to the
SDRAM to acknowledge this command. The write address must be chosen so that
BA[1] is set to 1 BA[0] is set to 0. For example, with a 16-bit 128 MB SDRAM (12 rows,
9 columns, 4 banks) bank address, the low-power DDR1-SDRAM write access should
be done at the address 0x20800000.
Note:
This address is for example purposes only. The real address is dependent on implementation in
the product.
8. A Mode Register set (MRS) cycle is issued to program the parameters of the low-power
DDR1-SDRAM devices, in particular CAS latency, burst length. The application must
set Mode to 3 in the Mode Register (see Section 22.7.1 on page 258) and perform a
write access to the low-power DDR1-SDRAM to acknowledge this command. The write
address must be chosen so that BA[1:0] bits are set to 0. For example, with a 16-bit 128
MB low-power DDR1-SDRAM (12 rows, 9 columns, 4 banks) bank address, the
SDRAM write access should be done at the address 0x20000000
230
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
9. The application must go into Normal Mode, setting Mode to 0 in the Mode Register (see
Section 22.7.1 on page 258) and performing a write access at any location in the lowpower DDR1-SDRAM to acknowledge this command.
10. Perform a write access to any low-power DDR1-SDRAM address.
11. Write the refresh rate into the count field in the DDRSDRC Refresh Timer register (see
page 259). (Refresh rate = delay between refresh cycles). The low-power DDR1SDRAM device requires a refresh every 15.625 μs or 7.81 μs. With a 100 MHz frequency, the refresh timer count register must to be set with (15.625 /100 MHz) = 1562
i.e. 0x061A or (7.81 /100 MHz) = 781 i.e. 0x030d
12. After initialization, the low-power DDR1-SDRAM device is fully functional.
22.3.3
DDR2-SDRAM Initialization
The initialization sequence is generated by software. The DDR2-SDRAM devices are initialized
by the following sequence:
1. Program the memory device type into the Memory Device Register (see Section 22.7.8
on page 269).
2. Program the features of DDR2-SDRAM device into the Timing Register (asynchronous
timing (trc, tras, etc.)), and into the Configuration Register (number of columns, rows,
banks, cas latency and output drive strength) (see Section 22.7.3 on page 260, Section
22.7.4 on page 263 and Section 22.7.5 on page 265).
3. An NOP command is issued to the DDR2-SDRAM. Program the NOP command into
the Mode Register, the application must set Mode to 1 in the Mode Register (see Section 22.7.1 on page 258). Perform a write access to any DDR2-SDRAM address to
acknowledge this command. Now clocks which drive DDR2-SDRAM device are
enabled.
A minimum pause of 200 μs is provided to precede any signal toggle.
4. An NOP command is issued to the DDR2-SDRAM. Program the NOP command into
the Mode Register, the application must set Mode to 1 in the Mode Register (see Section 22.7.1 on page 258). Perform a write access to any DDR2-SDRAM address to
acknowledge this command. Now CKE is driven high.
5. An all banks precharge command is issued to the DDR2-SDRAM. Program all banks
precharge command into the Mode Register, the application must set Mode to 2 in the
Mode Register (See Section 22.7.1 on page 258). Perform a write access to any DDR2SDRAM address to acknowledge this command
6. An Extended Mode Register set (EMRS2) cycle is issued to chose between commercial or high temperature operations. The application must set Mode to 5 in the Mode
Register (see Section 22.7.1 on page 258) and perform a write access to the DDR2SDRAM to acknowledge this command. The write address must be chosen so that
BA[1] is set to 1 and BA[0] is set to 0. For example, with a 16-bit 128 MB DDR2SDRAM (12 rows, 9 columns, 4 banks) bank address, the DDR2-SDRAM write access
should be done at the address 0x20800000.
Note:
This address is for example purposes only. The real address is dependent on implementation in
the product.
7. An Extended Mode Register set (EMRS3) cycle is issued to set all registers to “0”. The
application must set Mode to 5 in the Mode Register (see Section 22.7.1 on page 258)
and perform a write access to the DDR2-SDRAM to acknowledge this command. The
write address must be chosen so that BA[1] is set to 1 and BA[0] is set to 1. For example, with a 16-bit 128 MB DDR2-SDRAM (12 rows, 9 columns, 4 banks) bank address,
the DDR2-SDRAM write access should be done at the address 0x20C00000.
231
6438D–ATARM–13-Oct-09
8. An Extended Mode Register set (EMRS1) cycle is issued to enable DLL. The application must set Mode to 5 in the Mode Register (see Section 22.7.1 on page 258) and
perform a write access to the DDR2-SDRAM to acknowledge this command. The write
address must be chosen so that BA[1] is set to 0 and BA[0] is set to 1. For example,
with a 16-bit 128 MB DDR2-SDRAM (12 rows, 9 columns, 4 banks) bank address, the
DDR2-SDRAM write access should be done at the address 0x20400000.
An additional 200 cycles of clock are required for locking DLL
9. Program DLL field into the Configuration Register (see Section 22.7.3 on page 260) to
high (Enable DLL reset).
10. A Mode Register set (MRS) cycle is issued to reset DLL. The application must set
Mode to 3 in the Mode Register (see Section 22.7.1 on page 258) and perform a write
access to the DDR2-SDRAM to acknowledge this command. The write address must
be chosen so that BA[1:0] bits are set to 0. For example, with a 16-bit 128 MB DDR2SDRAM (12 rows, 9 columns, 4 banks) bank address, the SDRAM write access should
be done at the address 0x20000000.
11. An all banks precharge command is issued to the DDR2-SDRAM. Program all banks
precharge command into the Mode Register, the application must set Mode to 2 in the
Mode Register (See Section 22.7.1 on page 258). Perform a write access to any DDR2SDRAM address to acknowledge this command
12. Two auto-refresh (CBR) cycles are provided. Program the auto refresh command
(CBR) into the Mode Register, the application must set Mode to 4 in the Mode Register
(see Section 22.7.1 on page 258). Performs a write access to any DDR2-SDRAM location twice to acknowledge these commands.
13. Program DLL field into the Configuration Register (see Section 22.7.3 on page 260) to
low (Disable DLL reset).
14. A Mode Register set (MRS) cycle is issued to program the parameters of the DDR2SDRAM devices, in particular CAS latency, burst length and to disable DLL reset. The
application must set Mode to 3 in the Mode Register (see Section 22.7.1 on page 258)
and perform a write access to the DDR2-SDRAM to acknowledge this command. The
write address must be chosen so that BA[1:0] are set to 0. For example, with a 16-bit
128 MB SDRAM (12 rows, 9 columns, 4 banks) bank address, the SDRAM write
access should be done at the address 0x20000000
15. Program OCD field into the Configuration Register (see Section 22.7.3 on page 260) to
high (OCD calibration default).
16. An Extended Mode Register set (EMRS1) cycle is issued to OCD default value. The
application must set Mode to 5 in the Mode Register (see Section 22.7.1 on page 258)
and perform a write access to the DDR2-SDRAM to acknowledge this command. The
write address must be chosen so that BA[1] is set to 0 and BA[0] is set to 1. For example, with a 16-bit 128 MB DDR2-SDRAM (12 rows, 9 columns, 4 banks) bank address,
the DDR2-SDRAM write access should be done at the address 0x20400000.
17. Program OCD field into the Configuration Register (see Section 22.7.3 on page 260) to
low (OCD calibration mode exit).
18. An Extended Mode Register set (EMRS1) cycle is issued to enable OCD exit. The
application must set Mode to 5 in the Mode Register (see Section 22.7.1 on page 258)
and perform a write access to the DDR2-SDRAM to acknowledge this command. The
write address must be chosen so that BA[1] is set to 0 and BA[0] is set to 1. For example, with a 16-bit 128 MB DDR2-SDRAM (12 rows, 9 columns, 4 banks) bank address,
the DDR2-SDRAM write access should be done at the address 0x20400000.
232
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
19. A mode Normal command is provided. Program the Normal mode into Mode Register
(see Section 22.7.1 on page 258). Perform a write access to any DDR2-SDRAM
address to acknowledge this command.
20. Perform a write access to any DDR2-SDRAM address.
21. Write the refresh rate into the count field in the Refresh Timer register (see page 259).
(Refresh rate = delay between refresh cycles). The DDR2-SDRAM device requires a
refresh every 15.625 μs or 7.81 μs. With a 133 MHz frequency, the refresh timer count
register must to be set with (15.625 /133 MHz) = 1175 i.e. 0x0497 or (7.81 /133 MHz) =
587 i.e. 0x024B.
After initialization, the DDR2-SDRAM devices are fully functional.
233
6438D–ATARM–13-Oct-09
22.4
22.4.1
Functional Description
SDRAM Controller Write Cycle
The DDRSDRC allows burst access or single access in normal mode (mode = 000). Whatever
the access type, the DDRSDRC keeps track of the active row in each bank, thus maximizing
performance.
The SDRAM device is programmed with a burst length equal to 8. This determines the length of
a sequential data input by the write command that is set to 8. The latency from write command to
data input is fixed to 1 in the case of DDR-SDRAM devices. In the case of SDR-SDRAM
devices, there is no latency from write command to data input.
To initiate a single access, the DDRSDRC checks if the page access is already open. If
row/bank addresses match with the previous row/bank addresses, the controller generates a
write command. If the bank addresses are not identical or if bank addresses are identical but the
row addresses are not identical, the controller generates a precharge command, activates the
new row and initiates a write command. To comply with SDRAM timing parameters, additional
clock cycles are inserted between precharge/active (t RP) commands and active/write (t RCD)
command. As the burst length is fixed to 8, in the case of single access, it has to stop the burst,
otherwise seven invalid values may be written. In the case of SDR-SDRAM devices, a Burst
Stop command is generated to interrupt the write operation. In the case of DDR-SDRAM
devices, Burst Stop command is not supported for the burst write operation. In order to then
interrupt the write operation, Dm must be set to 1 to mask invalid data (see Figure 22-2 on page
235 and Figure 22-5 on page 236) and DQS must continue to toggle.
To initiate a burst access, the DDRSDRC uses the transfer type signal provided by the master
requesting the access. If the next access is a sequential write access, writing to the SDRAM
device is carried out. If the next access is a write non-sequential access, then an automatic
access break is inserted, the DDRSDRC generates a precharge command, activates the new
row and initiates a write command. To comply with SDRAM timing parameters, additional clock
cycles are inserted between precharge/active (tRP) commands and active/write (tRCD)
commands.
For a definition of timing parameters, refer to Section 22.7.4 “DDRSDRC Timing 0 Parameter
Register” on page 263.
Write accesses to the SDRAM devices are burst oriented and the burst length is programmed to
8. It determines the maximum number of column locations that can be accessed for a given write
command. When the write command is issued, 8 columns are selected. All accesses for that
burst take place within these eight columns, thus the burst wraps within these 8 columns if a
boundary is reached. These 8 columns are selected by addr[13:3]. addr[2:0] is used to select the
starting location within the block.
In the case of incrementing burst (INCR/INCR4/INCR8/INCR16), the addresses can cross the
16-byte boundary of the SDRAM device. For example, in the case of DDR-SDRAM devices,
when a transfer (INCR4) starts at address 0x0C, the next access is 0x10, but since the burst
length is programmed to 8, the next access is at 0x00. Since the boundary is reached, the burst
is wrapping. The DDRSDRC takes this feature of the SDRAM device into account. In the case of
transfer starting at address 0x04/0x08/0x0C (DDR-SDRAM devices) or starting at address
0x10/0x14/0x18/0x1C, two write commands are issued to avoid to wrap when the boundary is
reached. The last write command is subject to DM input logic level. If DM is registered high, the
corresponding data input is ignored and write access is not done. This avoids additional writing
being done.
234
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 22-2. Single Write Access, Row Closed, Low-power DDR1-SDRAM Device
SDCLK
Row a
A[12:0]
COMMAND
PRCHG
NOP
BA[1:0]
NOP
col a
ACT
NOP
WRITE
NOP
00
DQS[1:0]
DM[1:0]
3
D[15:0]
0
Da
3
Db
Trcd = 2
Trp = 2
Figure 22-3. Single Write Access, Row Closed, DDR2-SDRAM Device
SDCLK
A[12:0]
Row a
COMMAND
BA[1:0]
NOP
PRCHG
NOP
ACT
col a
NOP
WRITE
NOP
00
DQS[1:0]
DM[1:0]
3
D[15:0]
0
Da
Trp = 2
3
Db
Trcd = 2
235
6438D–ATARM–13-Oct-09
Figure 22-4. Single Write Access, Row Closed, SDR-SDRAM Device
SDCLK
A[12:0]
COMMAND
BA[1:0]
Row a
NOP
PRCHG
NOP
ACT
Col a
NOP
WRITE
BST
NOP
00
3
DM[1:0]
0
D[31:0]
3
DaDb
Trp = 2
Trcd = 2
Figure 22-5. Burst Write Access, Row Closed, Low-power DDR1-SDRAM Device
SDCLK
A[12:0]
Row a
COMMAND
BA[1:0]
NOP
PRCHG
NOP
col a
ACT
NOP
WRITE
NOP
0
DQS[1:0]
DM[1:0]
3
0
D [15:0]
Da
Trp = 2
236
Db
Dc
Dd
3
De
Df
Dg
Dh
Trcd = 2
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 22-6. Burst Write Access, Row Closed, DDR2-SDRAM Device
SDCLK
A[12:0]
Row a
COMMAND
BA[1:0]
NOP
PRCHG
NOP
col a
ACT
NOP
WRITE
NOP
0
DQS[1:0]
DM[1:0]
3
0
D [15:0]
Da
Db
Dc
Dd
3
De
Df
Dg
Dh
Trcd = 2
Trp = 2
Figure 22-7. Burst Write Access, Row Closed, SDR-SDRAM Device
SDCLK
A[12:0]
COMMAND
Row a
NOP
BA[1:0]
0
DM[3:0]
F
PRCHG
NOP
Col a
ACT
NOP
WRITE
NOP
BST
0
D[31:0]
Da Db
Trp
Dc Dd
NOP
F
De Df
Dg Dhs
Trcd
A write command can be followed by a read command. To avoid breaking the current write
burst, Twtr/Twrd (bl/2 + 2 = 6 cycles) should be met. See Figure 22-8 on page 238.
237
6438D–ATARM–13-Oct-09
Figure 22-8. Write Command Followed By a Read Command without Burst Write Interrupt, Low-power DDR1-SDRAM
Device
SDCLK
A[12:0]
col a
COMMAND
NOP
BA[1:0]
col a
WRITE
NOP
READ
BST
NOP
0
DQS[1:0]
DM[1:0]
3
0
D[15:0]
3
Da
Db Dc
Dd
De
Df
Dg
Dh
Da Db
Twrd = BL/2 +2 = 8/2 +2 = 6
Twr = 1
In the case of a single write access, write operation should be interrupted by a read access but
DM must be input 1 cycle prior to the read command to avoid writing invalid data. See Figure 229 on page 238.
Figure 22-9. Single Write Access Followed By A Read Access Low-power DDR1-SDRAM Devices
SDCLK
A[12:0]
COMMAND
BA[1:0]
col a
Row a
NOP
PRCHG
NOP
ACT
NOP
WRITE
NOP
READ
BST
NOP
0
DQS[1:0]
DM[1:0]
3
D[15:0]
0
Da
3
Db
Da Db
Data masked
238
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 22-10. SINGLE Write Access Followed By A Read Access, DDR2 -SDRAM Device
SDCLK
A[12:0]
COMMAND
col a
Row a
NOP PRCHG NOP
BA[1:0]
ACT
NOP
WRITE
NOP
READ
NOP
0
DQS[1:0]
DM[1:0]
D[15:0]
3
0
Da
3
Da Db
Db
Data masked
twtr
22.4.2
SDRAM Controller Read Cycle
The DDRSDRC allows burst access or single access in normal mode (mode =000). Whatever
access type, the DDRSDRC keeps track of the active row in each bank, thus maximizing performance of the DDRSDRC.
The SDRAM devices are programmed with a burst length equal to 8 which determines the length
of a sequential data output by the read command that is set to 8. The latency from read command to data output is equal to 2 or 3. This value is programmed during the initialization phase
(see Section 22.3.1 “SDR-SDRAM Initialization” on page 229).
To initiate a single access, the DDRSDRC checks if the page access is already open. If
row/bank addresses match with the previous row/bank addresses, the controller generates a
read command. If the bank addresses are not identical or if bank addresses are identical but the
row addresses are not identical, the controller generates a precharge command, activates the
new row and initiates a read command. To comply with SDRAM timing parameters, additional
clock cycles are inserted between precharge/active (Trp) commands and active/read (Trcd)
command. After a read command, additional wait states are generated to comply with cas
latency. The DDRSDRC supports a cas latency of two, two and half, and three (2 or 3 clocks
delay). As the burst length is fixed to 8, in the case of single access or burst access inferior to 8
data requests, it has to stop the burst otherwise seven or X values could be read. Burst Stop
Command (BST) is used to stop output during a burst read.
To initiate a burst access, the DDRSDRC checks the transfer type signal. If the next accesses
are sequential read accesses, reading to the SDRAM device is carried out. If the next access is
a read non-sequential access, then an automatic page break can be inserted. If the bank
addresses are not identical or if bank addresses are identical but the row addresses are not
identical, the controller generates a precharge command, activates the new row and initiates a
read command. In the case where the page access is already open, a read command is
generated.
239
6438D–ATARM–13-Oct-09
To comply with SDRAM timing parameters, additional clock cycles are inserted between precharge/active (Trp) commands and active/read (Trcd) commands. The DDRSDRC supports a
cas latency of two, two and half, and three (2 or 3 clocks delay). During this delay, the controller
uses internal signals to anticipate the next access and improve the performance of the controller. Depending on the latency(2/3), the DDRSDRC anticipates 2 or 3 read accesses. In the case
of burst of specified length, accesses are not anticipated, but if the burst is broken (border, busy
mode, etc.), the next access is treated as an incrementing burst of unspecified length, and in
function of the latency(2/3), the DDRSDRC anticipates 2 or 3 read accesses.
For a definition of timing parameters, refer to Section 22.7.3 “DDRSDRC Configuration Register”
on page 260.
Read accesses to the SDRAM are burst oriented and the burst length is programmed to 8. It
determines the maximum number of column locations that can be accessed for a given read
command. When the read command is issued, 8 columns are selected. All accesses for that
burst take place within these eight columns, meaning that the burst wraps within these 8 columns if the boundary is reached. These 8 columns are selected by addr[13:3]; addr[2:0] is used
to select the starting location within the block.
In the case of incrementing burst (INCR/INCR4/INCR8/INCR16), the addresses can cross the
16-byte boundary of the SDRAM device. For example, when a transfer (INCR4) starts at
address 0x0C, the next access is 0x10, but since the burst length is programmed to 8, the next
access is 0x00. Since the boundary is reached, the burst wraps. The DDRSDRC takes into
account this feature of the SDRAM device. In the case of DDR-SDRAM devices, transfers start
at address 0x04/0x08/0x0C. In the case of SDR-SDRAM devices, transfers start at address
0x14/0x18/0x1C. Two read commands are issued to avoid wrapping when the boundary is
reached. The last read command may generate additional reading (1 read cmd = 4 DDR words
or 1 read cmd = 8 SDR words).
To avoid additional reading, it is possible to use the burst stop command to truncate the read
burst and to decrease power consumption.
240
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 22-11. Single Read Access, Row Close, Latency = 2,Low-power DDR1-SDRAM Device
SDCLK
A[12:0]
COMMAND
BA[1:0]
NOP
PRCHG
NOP
Row a
Col a
ACT
NOP
READ
BST
NOP
0
DQS[1]
DQS[0]
DM[1:0]
3
D[15:0]
Da
Trp
Trcd
Db
Latency = 2
Figure 22-12. Single Read Access, Row Close, Latency = 3, DDR2-SDRAM Device
SDCLK
A[12:0]
COMMAND
BA[1:0]
NOP
PRCHG
NOP
Row a
Col a
ACT
NOP
READ
0
DQS[1]
DQS[0]
DM[1:0]
3
D[15:0]
Da
Trp
Trcd
Db
Latency = 2
241
6438D–ATARM–13-Oct-09
Figure 22-13. Single Read Access, Row Close, Latency = 2, SDR-SDRAM Device
SDCLK
A[12:0]
COMMAND
Row a
NOP
BA[1:0]
0
DM[3:0]
3
PRCHG
NOP
ACT
col a
NOP
READ
BST
NOP
D[31:0]
DaDb
Trp
Trcd
Latency = 2
Figure 22-14. Burst Read Access, Latency = 2, Low-power DDR1-SDRAM Devices
SDCLK
Col a
A[12:0]
COMMAND
BA[1:0]
NOP
READ
NOP
0
DQS[1:0]
DM[1:0]
3
D[15:0]
Da
Db
Dc
Dd
De
Df
Dg
Dh
Latency = 2
242
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 22-15. Burst Read Access, Latency = 3, DDR2-SDRAM Devices
SDCLK
A[12:0]
Col a
COMMAND
NOP
BA[1:0]
READ
NOP
0
DQS[1:0]
DM[1:0]
3
D[15:0]
Da
Db
Dc
Dd
De
Df
Dg
Dh
Latency = 3
Figure 22-16. Burst Read Access, Latency = 2, SDR-SDRAM Devices
SDCLK
A[12:0]
COMMAND
BA[1:0]
col a
NOP
READ
NOP
BST
NOP
0
DQS[1:0]
DM[3:0]
F
D[31:0]
DaDb
DcDd
DeDf
Dg Dh
Latency = 2
243
6438D–ATARM–13-Oct-09
22.4.3
Refresh (Auto-refresh Command)
An auto-refresh command is used to refresh the DDRSDRC. Refresh addresses are generated
internally by the SDRAM device and incremented after each auto-refresh automatically. The
DDRSDRC generates these auto-refresh commands periodically. A timer is loaded with the
value in the register DDRSDRC_TR that indicates the number of clock cycles between refresh
cycles. When the DDRSDRC initiates a refresh of an SDRAM device, internal memory accesses
are not delayed. However, if the CPU tries to access the SDRAM device, the slave indicates that
the device is busy. A request of refresh does not interrupt a burst transfer in progress.
22.4.4
Power Management
22.4.4.1
Self Refresh Mode
This mode is activated by setting low-power command bits [LPCB] to ‘01’ in the
DDRSDRC_LPR Register
Self refresh mode is used to reduce power consumption, i.e., when no access to the SDRAM
device is possible. In this case, power consumption is very low. In self refresh mode, the
SDRAM device retains data without external clocking and provides its own internal clocking,
thus performing its own auto-refresh cycles. All the inputs to the SDRAM device become “don’t
care” except CKE, which remains low. As soon as the SDRAM device is selected, the DDRSDRC provides a sequence of commands and exits self refresh mode.
The DDRSDRC re-enables self refresh mode as soon as the SDRAM device is not selected. It is
possible to define when self refresh mode will be enabled by setting the register LPR (see Section 22.7.7 “DDRSDRC Low-power Register” on page 267), timeout command bit:
• 00 = Self refresh mode is enabled as soon as the SDRAM device is not selected
• 01 = Self refresh mode is enabled 64 clock cycles after completion of the last access
• 10 = Self refresh mode is enabled 128 clock cycles after completion of the last access
As soon as the SDRAM device is no longer selected, PRECHARGE ALL BANKS command is
generated followed by a SELF-REFREFSH command. If, between these two commands an
SDRAM access is detected, SELF-REFREFSH command will be replaced by an AUTOREFRESH command. According to the application, more AUTO-REFRESH commands will be
performed when the self refresh mode is enabled during the application.
This controller also interfaces low-power SDRAM. These devices add a new feature: A single
quarter, one half quarter or all banks of the SDRAM array can be enabled in self refresh mode.
Disabled banks will be not refreshed in self refresh mode. This feature permits to reduce the self
refresh current. The extended mode register controls this feature, it includes Temperature Compensated Self Refresh (TSCR), Partial Array Self Refresh (PASR) parameters and Drive
Strength (DS). These parameters are set during the initialization phase. After initialization, as
soon as PASR/DS/TCSR fields are modified, the Extended Mode Register in the memory of the
external device is accessed automatically and PASR/DS/TCSR bits are updated before entry
into self refresh mode if DDRSDRC does not share an external bus with another controller or
during a refresh command, and a pending read or write access, if DDRSDRC does share an
external bus with another controller. This type of update is a function of the UDP_EN bit (see
Section 22.7.7 “DDRSDRC Low-power Register” on page 267).
The low-power SDR-SDRAM must remain in self refresh mode for a minimum period of TRAS
periods and may remain in self refresh mode for an indefinite period. (See Figure 22-17)
244
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
The low-power DDR-SDRAM must remain in self refresh mode for a minimum of TRFC periods
and may remain in self refresh mode for an indefinite period.
The DDR2-SDRAM must remain in self refresh mode for a minimum of TCKE periods and may
remain in self refresh mode for an indefinite period.
Figure 22-17. Self Refresh Mode Entry, Timeout = 0
SDCLK
A[12:0]
COMMAND
NOP READ
BST
NOP
PRCHG
NOP
ARFSH
NOP
CKE
BA[1:0]
0
DQS[0:1]
DM[1:0]
3
D[15:0]
Da
Db
Trp
Enter Self refresh
Mode
Figure 22-18. Self Refresh Mode Entry, Timeout = 1 or 2
SDCLK
A[12:0]
COMMAND
NOP READ
BST
NOP
PRCHG
NOP
ARFSH NOP
CKE
BA[1:0]
0
DQS[1:0]
DM[1:0]
D[15:0]
3
Da
Db
64 or 128
wait states
Trp
Enter Self refresh
Mode
245
6438D–ATARM–13-Oct-09
Figure 22-19. Self Refresh Mode Exit
SDCLK
A[12:0]
COMMAND
NOP
VALID
NOP
CKE
BA[1:0]
0
DQS[1:0]
DM[1:0]
3
D[15:0]
DaDb
Exit Self Refresh mode
clock must be stable
before exiting self refresh mode
TXNRD/TXSRD
TXSR
TXSR
(DDR device)
(Low-power DDR device)
(Low-power SDR, SDR-SDRAM device)
Figure 22-20. Self Refresh and Automatic Update
SDCLK
Pasr-Tcr-Ds
A[12:0]
COMMAND
NOP
PRCHG
NOP
MRS
NOP
NOP
ARFSH
CKE
BA[1:0]
0
2
Enter Self Refresh
Mode
Trp
Tmrd
Update Extended Mode
register
246
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 22-21. Automatic Update During AUTO-REFRESH Command and SDRAM Access
SDCLK
A[12:0]
COMMAND
Pasr-Tcr-Ds
NOP PRCHALL
NOP
ARFSH
NOP
MRS
NOP
ACT
CKE
BA[1:0]
0
0
2
Trp
Trfc
Tmrd
Update Extended mode
register
22.4.4.2
Power-down Mode
This mode is activated by setting the low-power command bits [LPCB] to ‘10’.
Power-down mode is used when no access to the SDRAM device is possible. In this mode,
power consumption is greater than in self refresh mode. This state is similar to normal mode (No
low-power mode/No self refresh mode), but the CKE pin is low and the input and output buffers
are deactivated as soon the SDRAM device is no longer accessible. In contrast to self refresh
mode, the SDRAM device cannot remain in low-power mode longer than the refresh period (64
ms). As no auto-refresh operations are performed in this mode, the DDRSDRC carries out the
refresh operation. In order to exit low-power mode, a NOP command is required in the case of
Low-power SDR-SDRAM and SDR-SDRAM devices. In the case of Low-power DDR-SDRAM
devices, the controller generates a NOP command during a delay of at least TXP. In addition,
Low-power DDR-SDRAM and DDR2-SDRAM must remain in power-down mode for a minimum
period of TCKE periods.
The exit procedure is faster than in self refresh mode. See Figure 22-22 on page 248. The
DDRSDRC returns to power-down mode as soon as the SDRAM device is not selected. It is
possible to define when power-down mode is enabled by setting the register LPR, timeout command bit.
• 00 = Power-down mode is enabled as soon as the SDRAM device is not selected
• 01 = Power-down mode is enabled 64 clock cycles after completion of the last access
• 10 = Power-down mode is enabled 128 clock cycles after completion of the last access
247
6438D–ATARM–13-Oct-09
Figure 22-22. Power-down Entry/Exit, Timeout = 0
SDCLK
A[12:0]
COMMAND
READ
BST
NOP
READ
CKE
BA[1:0]
0
DQS[1:0]
DM[1:0]
3
D[15:0]
Da
Db
Exit power down mode
Entry power down mode
22.4.4.3
Deep Power-down Mode
The deep power-down mode is a new feature of the Low-power SDRAM. When this mode is
activated, all internal voltage generators inside the device are stopped and all data is lost.
This mode is activated by setting the low-power command bits [LPCB] to ‘11’. When this mode is
enabled, the DDRSDRC leaves normal mode (mode == 000) and the controller is frozen. To exit
deep power-down mode, the low-power bits (LPCB) must be set to “00”, an initialization
sequence must be generated by software. See Section 22.3.2 “Low-power DDR1-SDRAM Initialization” on page 230.
Figure 22-23. Deep Power-down Mode Entry
SDCLK
A[12:0]
COMMAND
NOP READ
BST
NOP
PRCHG
NOP
DEEPOWER
NOP
CKE
BA[1:0]
0
DQS[1:0]
DM[1:0]
3
D[15:0]
Da
Db
Trp
248
Enter Deep
Power-down
Mode
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
22.4.4.4
Reset Mode
The reset mode is a feature of the DDR2-SDRAM. This mode is activated by setting the lowpower command bits (LPCB) to 11 and the clock frozen command bit (CLK_FR) to 1.
When this mode is enabled, the DDRSDRC leaves normal mode (mode == 000) and the controller is frozen. Before enabling this mode, the end user must assume there is not an access in
progress.
To exit reset mode, the low-power command bits (LPCB) must be set to “00”, clock frozen command bit (CLK_FR) set to 0 and an initialization sequence must be generated by software. See,
Section 22.3.3 “DDR2-SDRAM Initialization” on page 231.
249
6438D–ATARM–13-Oct-09
22.4.5
Multi-port Functionality
The SDRAM protocol imposes a check of timings prior to performing a read or a write access,
thus decreasing the performance of systems. An access to SDRAM is performed if banks and
rows are open (or active). To activate a row in a particular bank, it has to de-active the last open
row and open the new row. Two SDRAM commands must be performed to open a bank: Precharge and Active command with respect to Trp timing. Before performing a read or write
command, Trcd timing must checked.
This operation represents a significative loss. (see Figure 22-24).
Figure 22-24. Trp and Trcd Timings
SDCLK
A[12:0]
COMMAND
BA[1:0]
NOP
PRCHG
NOP
ACT
NOP
READ
BST
NOP
0
DQS[1:0]
DM1:0]
3
D[15:0]
Da
Trp
Trcd
Db
Latency =2
4 cycles before performing a read command
The multi-port controller has been designed to mask these timings and thus improve the bandwidth of the system.
DDRSDRC is a multi-port controller since four masters can simultaneously reach the controller.
This feature improves the bandwidth of the system because it can detect four requests on the
AHB slave inputs and thus anticipate the commands that follow, PRECHARGE and ACTIVE
commands in bank X during current access in bank Y. This allows Trp and Trcd timings to be
masked (see Figure 22-25). In the best case, all accesses are done as if the banks and rows
were already open. The best condition is met when the four masters work in different banks. In
the case of four simultaneous read accesses, when the four banks and associated rows are
open, the controller reads with a continuous flow and masks the cas latency for each different
access. To allow a continuous flow, the read command must be set at 2 or 3 cycles (cas latency)
before the end of current access. This requires that the scheme of arbitration changes since the
round-robin arbitration cannot be respected. If the controller anticipates a read access, and thus
before the end of current access a master with a high priority arises, then this master will not
serviced.
The arbitration mechanism reduces latency when conflicts occur, i.e., when two or more masters
try to access the SDRAM device at the same time.
250
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
The arbitration type is round-robin arbitration. This algorithm dispatches the requests from different masters to the SDRAM device in a round-robin manner. If two or more master requests arise
at the same time, the master with the lowest number is serviced first, then the others are serviced in a round-robin manner. To avoid burst breaking and to provide the maximum throughput
for the SDRAM device, arbitration may only take place during the following cycles:
1. Idle cycles: When no master is connected to the SDRAM device.
2. Single cycles: When a slave is currently doing a single access.
3. End of Burst cycles: When the current cycle is the last cycle of a burst transfer. For
bursts of defined length, predicted end of burst matches the size of the transfer. For
bursts of undefined length, predicted end of burst is generated at the end of each four
beat boundary inside the INCR transfer.
4. Anticipated Access: When an anticipate read access is done while current access is
not complete, the arbitration scheme can be changed if the anticipated access is not
the next access serviced by the arbitration scheme.
Figure 22-25. Anticipate Precharge/Active Command in Bank 2 during Read Access in Bank 1
SDClK
A[12:0]
COMMAND
BA[1:0]
NOP
0
READ
PRECH
1
NOP
ACT
READ
2
NOP
1
DQS[1:0]
DM1:0]
3
D[15:0]
Da
Db
Dc
Dd
De
Df
Dg
Dh
Di
Dj
Dk
Dl
Trp
Anticipate command, Precharge/Active Bank 2
Read access in Bank 1
251
6438D–ATARM–13-Oct-09
22.4.6
Write Protected Registers
To prevent any single software error that may corrupt DDRSDRC behavior, the registers listed
below can be write-protected by setting the WPEN bit in the DDRSDRC Write Protect Mode
Register (DDRSDRC_WPMR).
If a write access in a write-protected register is detected, then the WPVS flag in the DDRSDRC
Write Protect Status Register (DDRSDRC_WPSR) is set and the field WPVSRC indicates in
which register the write access has been attempted.
The WPVS flag is automatically reset after reading the DDRSDRC Write Protect Status Register
(DDRSDRC_WPSR).
Following is a list of the write protected registers:
• “DDRSDRC Mode Register” on page 258
• “DDRSDRC Refresh Timer Register” on page 259
• “DDRSDRC Configuration Register” on page 260
• “DDRSDRC Timing 0 Parameter Register” on page 263
• “DDRSDRC Timing 1 Parameter Register” on page 265
• “DDRSDRC Timing 2 Parameter Register” on page 266
• “DDRSDRC Memory Device Register” on page 269
• “DDRSDRC High Speed Register” on page 271
252
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
22.5
Software Interface/SDRAM Organization, Address Mapping
The SDRAM address space is organized into banks, rows and columns. The DDRSDRC maps
different memory types depending on the values set in the DDRSDRC Configuration Register.
See Section 22.7.3 “DDRSDRC Configuration Register” on page 260. The following figures illustrate the relation between CPU addresses and columns, rows and banks addresses for 16-bit
memory data bus widths and 32-bit memory data bus widths.
The DDRSDRC supports address mapping in linear mode .
Linear mode is a method for address mapping where banks alternate at each last SDRAM page
of current bank.
.
The DDRSDRC makes the SDRAM devices access protocol transparent to the user. Table 22-1
to Table 22-8 illustrate the SDRAM device memory mapping seen by the user in correlation with
the device structure. Various configurations are illustrated.
22.5.1
SDRAM Address Mapping for 16-bit Memory Data Bus Width(1) and Four Banks
Table 22-1.
Linear Mapping for SDRAM Configuration, 2K Rows, 512/1024/2048/4096 Columns
CPU Address Line
27
26
25
24
23
22
21
20
19
18
17
16
Bk[1:0]
14
13
12
11
10
9
8
7
Row[10:0]
Bk[1:0]
5
4
3
2
1
M0
M0
Column[10:0]
Row[10:0]
0
M0
Column[9:0]
Row[10:0]
Bk[1:0]
6
Column[8:0]
Row[10:0]
Bk[1:0]
Table 22-2.
15
M0
Column[11:0]
Linear Mapping for SDRAM Configuration: 4K Rows, 512/1024/2048/4096 Columns
CPU Address Line
27
26
25
24
23
22
21
20
19
18
17
Bk[1:0]
Bk[1:0]
15
Row[11:0]
Bk[1:0]
Bk[1:0]
16
Row[11:0]
Row[11:0]
Row[11:0]
14
13
12
11
10
9
8
7
6
5
4
Column[8:0]
Column[9:0]
Column[10:0]
Column[11:0]
3
2
1
0
M0
M0
M0
M0
253
6438D–ATARM–13-Oct-09
Table 22-3.
Linear Mapping for SDRAM Configuration: 8K Rows, 512/1024/2048/4096 Columns
CPU Address Line
27
26
25
24
23
22
21
20
19
18
17
Bk[1:0]
16
15
14
13
12
11
10
9
8
6
Row[12:0]
Bk[1:0]
4
3
2
1
M0
M0
Column[10:0]
Row[12:0]
0
M0
Column[9:0]
Row[12:0]
Bk[1:0]
5
Column[8:0]
Row[12:0]
Bk[1:0]
Table 22-4.
7
M0
Column[11:0]
Linear Mapping for SDRAM Configuration: 16K Rows, 512/1024/2048 Columns
CPU Address Line
27
26
25
24
23
22
21
20
19
18
Bk[1:0]
16
15
14
13
12
11
10
9
8
7
6
Row[13:0]
Bk[1:0]
5
4
3
2
1
M0
Column[9:0]
Row[13:0]
0
M0
Column[8:0]
Row[13:0]
Bk[1:0]
Note:
17
M0
Column[10:0]
1. SDR-SDRAM devices with eight columns in 16-bit mode are not supported.
22.5.2
SDR-SDRAM Address Mapping for 32-bit Memory Data Bus Width
Table 22-6.
SDR-SDRAM Configuration Mapping: 2K Rows, 256/512/1024/2048 Columns
CPU Address Line
27
26
25
24
23
22
21
20
19
18
17
16
Bk[1:0]
14
13
12
11
10
9
8
7
Row[10:0]
Bk[1:0]
5
4
3
2
0
M[1:0]
Column[9:0]
Row[10:0]
1
M[1:0]
Column[8:0]
Row[10:0]
Bk[1:0]
6
Column[7:0]
Row[10:0]
Bk[1:0]
Table 22-7.
15
M[1:0]
Column[10:0]
M[1:0]
SDR-SDRAM Configuration Mapping: 4K Rows, 256/512/1024/2048 Columns
CPU Address Line
27
26
25
24
23
22
21
20
19
18
17
Bk[1:0]
Row[11:0]
Bk[1:0]
254
15
Row[11:0]
Bk[1:0]
Bk[1:0]
16
Row[11:0]
Row[11:0]
14
13
12
11
10
9
8
7
6
5
Column[7:0]
Column[8:0]
Column[9:0]
Column[10:0]
4
3
2
1
0
M[1:0]
M[1:0]
M[1:0]
M[1:0]
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Table 22-8.
SDR-SDRAM Configuration Mapping: 8K Rows, 256/512/1024/2048 Columns
CPU Address Line
27
26
25
24
23
22
21
20
19
18
17
Bk[1:0]
Row[12:0]
Bk[1:0]
Notes:
15
Row[12:0]
Bk[1:0]
Bk[1:0]
16
Row[12:0]
Row[12:0]
14
13
12
11
10
9
8
7
6
5
Column[7:0]
Column[8:0]
Column[9:0]
Column[10:0]
4
3
2
1
0
M[1:0]
M[1:0]
M[1:0]
M[1:0]
1. M[1:0] is the byte address inside a 32-bit word.
2. Bk[1] = BA1, Bk[0] = BA0
255
6438D–ATARM–13-Oct-09
22.6
Programmable IO Delays
The external bus interface consists of a data bus, an address bus and control signals. The simultaneous switching outputs on these busses may lead to a peak of current in the internal and
external power supply lines.
In order to reduce the peak of current in such cases, additional propagation delays can be
adjusted independently for pad buffers by means of configuration registers,
DDRSDRC_DELAY1-8.
The additional programmable delays for each IO range from 0 to 4 ns (Worst Case PVT). The
delay can differ between IOs supporting this feature. Delay can be modified per programming for
each IO. The minimal additional delay that can be programmed on a PAD supporting this feature
is 1/16 of the maximum programmable delay.
When programming 0x0 in fields “Delay1 to Delay8”, no delay is added (reset value) and the
propagation delay of the pad buffers is the inherent delay of the pad buffer. When programming
0xF in field “Delay1” the propagation delay of the corresponding pad is maximal.
DDRSDRC_DELAY1, DDRSDRC_DELAY2 allow to configure delay on D[15:0],
DDRSDRC_DELAY1[3:0] corresponds to D[0] and DDRSDRC_DELAY2[3:0] corresponds to
D[8].
DDRSDRC_DELAY3, DDRSDRC_DELAY4 allow to configure delay on A13:0],
DDRSDRC_DELAY3[3:0] corresponds to A[0] and DDRSDRC_DELAY4[3:0] corresponds to
A[8].
Figure 22-26. Programmable IO Delays
SMC
D_in[0]
D_out[0]
Programmable Delay Line
D[0]
Programmable Delay Line
D[1]
Programmable Delay Line
D[n]
Programmable Delay Line
A[m]
DELAY1
D_in[1]
D_out[1]
DELAY2
D_in[n]
D_out[n]
DELAYx
A[m]
DELAYy
22.7
DDR-SDRAM Controller (DDRSDRC) User Interface
The User Interface is connected to the APB bus.
256
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
The DDRSDRC is programmed using the registers listed in Table 22-9.
Table 22-9.
Register Mapping
Offset
Register
Name
Access
Reset
0x00
DDRSDRC Mode Register
DDRSDRC_MR
Read-write
0x00000000
0x04
DDRSDRC Refresh Timer Register
DDRSDRC_RTR
Read-write
0x00000000
0x08
DDRSDRC Configuration Register
DDRSDRC_CR
Read-write
0x7024
0x0C
DDRSDRC Timing0 Register
DDRSDRC_T0PR
Read-write
0x20227225
0x10
DDRSDRC Timing1 Register
DDRSDRC_T1PR
Read-write
0x3c80808
0x14
DDRSDRC Timing2 Register
DDRSDRC_T2PR
Read-write
0x2062
0x18
Reserved
–
–
–
0x1C
DDRSDRC Low-power Register
DDRSDRC_LPR
Read-write
0x10000
0x20
DDRSDRC Memory Device Register
DDRSDRC_MD
Read-write
0x10
0x24
DDRSDRC DLL Information Register
DDRSDRC_DLL
Read-only
0x00000001
0x2C
DDRSDRC High Speed Register
DDRSDRC_HS
Read-write
0x0
0x34
DDRSDRC Delay I/O Register
DDRSDRC_DELAY1
Read-write
0x00000000
0x38
DDRSDRC Delay I/O Register
DDRSDRC_DELAY2
Read-write
0x00000000
0x3C
DDRSDRC Delay I/O Register
DDRSDRC_DELAY3
Read-write
0x00000000
0x40
DDRSDRC Delay I/O Register
DDRSDRC_DELAY4
Read-write
0x00000000
0x44
Reserved
–
–
–
0x48-0x4C
Reserved
-
-
-
0x58-0xE0
Reserved
–
–
–
0xE4
DDRSDRC Write Protect Mode Register
DDRSDRC_WPMR
Read-write
0x00000000
0xE8
DDRSDRC Write Protect Status Register
DDRSDRC_WPSR
Read-only
0x00000000
257
6438D–ATARM–13-Oct-09
22.7.1
Name:
DDRSDRC Mode Register
DDRSDRC_MR
Access:
Read-write
Reset:
See Table 22-9
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
–
–
–
–
–
MODE
This register can only be written if the bit WPEN is cleared in “DDRSDRC Write Protect Mode Register” on page 273.
• MODE: DDRSDRC Command Mode
This field defines the command issued by the DDRSDRC when the SDRAM device is accessed. This register is used to initialize the SDRAM device and to activate deep power-down mode.
MODE
Description
000
Normal Mode. Any access to the DDRSDRC will be decoded normally. To activate this mode, command must be followed
by a write to the SDRAM.
001
The DDRSDRC issues a NOP command when the SDRAM device is accessed regardless of the cycle. To activate this
mode, command must be followed by a write to the SDRAM.
010
The DDRSDRC issues an “All Banks Precharge” command when the SDRAM device is accessed regardless of the cycle.
To activate this mode, command must be followed by a write to the SDRAM.
011
The DDRSDRC issues a “Load Mode Register” command when the SDRAM device is accessed regardless of the cycle.
To activate this mode, command must be followed by a write to the SDRAM.
100
The DDRSDRC issues an “Auto-Refresh” Command when the SDRAM device is accessed regardless of the cycle.
Previously, an “All Banks Precharge” command must be issued. To activate this mode, command must be followed by a
write to the SDRAM.
101
The DDRSDRC issues an “Extended Load Mode Register” command when the SDRAM device is accessed regardless of
the cycle. To activate this mode, the “Extended Load Mode Register” command must be followed by a write to the SDRAM.
The write in the SDRAM must be done in the appropriate bank.
110
Deep power mode: Access to deep power-down mode
111
Reserved
258
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
22.7.2
Name:
DDRSDRC Refresh Timer Register
DDRSDRC_RTR
Access:
Read-write
Reset:
See Table 22-9
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
7
6
5
4
1
0
COUNT
3
2
COUNT
This register can only be written if the bit WPEN is cleared in “DDRSDRC Write Protect Mode Register” on page 273.
• COUNT: DDRSDRC Refresh Timer Count
This 12-bit field is loaded into a timer which generates the refresh pulse. Each time the refresh pulse is generated, a refresh
sequence is initiated.
SDRAM devices require a refresh of all rows every 64 ms. The value to be loaded depends on the DDRSDRC clock frequency (MCK: Master Clock) and the number of rows in the device.
For example, for an SDRAM with 8192 rows and a 100 MHz Master clock, the value of Refresh Timer Count bit is programmed: (((64 x 10-3)/8192) x100 x106 = 781 or 0x030D.
259
6438D–ATARM–13-Oct-09
22.7.3
Name:
DDRSDRC Configuration Register
DDRSDRC_CR
Access:
Read-write
Reset:
See Table 22-9
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
ACTBST
–
DQMS
15
14
13
12
11
10
9
8
–
–
DIS_DLL
DIC/DS
2
1
–
7
OCD
6
5
DLL
4
CAS
3
NR
0
NC
This register can only be written if the bit WPEN is cleared in “DDRSDRC Write Protect Mode Register” on page 273.
• NC: Number of Column Bits
The reset value is 9 column bits.
SDR-SDRAM devices with eight columns in 16-bit mode (b16mode ==1) are not supported.
NC
DDR - Column bits
SDR - Column bits
00
9
8
01
10
9
10
11
10
11
12
11
• NR: Number of Row Bits
The reset value is 12 row bits.
260
NR
Row bits
00
11
01
12
10
13
11
14
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
• CAS: CAS Latency
The reset value is 2 cycles.
CAS
DDR2 CAS Latency
SDR CAS Latency
000
Reserved
Reserved
001
Reserved
Reserved
010
Reserved
2
011
3
3
100
Reserved
Reserved
101
Reserved
Reserved
110
Reserved
Reserved
111
Reserved
Reserved
• DLL: Reset DLL
Reset value is 0.
This field defines the value of Reset DLL.
0 = Disable DLL reset.
1 = Enable DLL reset.
This value is used during the power-up sequence.
Note: This field is found only in DDR1-SDRAM devices.
• DIC/DS: Output Driver Impedance Control:
Reset value is 0.
This field defines the output drive strength.
0 = Normal driver strength.
1 = Weak driver strength.
This value is used during the power-up sequence. This parameter is found in the datasheet as DIC or DS.
Note: This field is found only in DDR2-SDRAM devices.
• DIS_DLL: Disable DLL
0 = Enable DLL
1 = Disable DLL
• OCD: Off-chip Driver
Reset value is 3’b111.
Note: OCD is NOT supported by the controller, but these values MUST be programmed during the initialization sequence.
OCD
000
OCD calibration mode exit, maintain setting
111
OCD calibration default
261
6438D–ATARM–13-Oct-09
• DQMS: Mask Data is Shared
Reset value is 0.
0 = DQM is not shared with another controller.
1 = DQM is shared with another controller.
• ACTBST: ACTIVE Bank X to Burst Stop Read Access Bank Y
Reset value is 0.
0 = After an ACTIVE command in Bank X, BURST STOP command can be issued to another bank to stop current read
access.
1 = After an ACTIVE command in Bank X, BURST STOP command cannot be issued to another bank to stop current read
access.
This field is unique to SDR-SDRAM, Low-power SDR-SDRAM and Low-power DDR-SDRAM devices.
262
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
22.7.4
Name:
DDRSDRC Timing 0 Parameter Register
DDRSDRC_T0PR
Access:
Read-write
Reset:
See Table 22-9
31
30
29
28
TMRD
23
22
27
26
21
20
19
14
18
13
6
17
16
9
8
1
0
TRP
12
11
10
TRC
7
24
TWTR
TRRD
15
25
REDUCE_WRRD
TWR
5
TRCD
4
3
2
TRAS
This register can only be written if the bit WPEN is cleared in “DDRSDRC Write Protect Mode Register” on page 273.
• TRAS: Active to Precharge Delay
Reset Value is 5 cycles.
This field defines the delay between an Activate Command and a Precharge Command in number of cycles. Number of
cycles is between 0 and 15.
• TRCD: Row to Column Delay
Reset Value is 2 cycles.
This field defines the delay between an Activate Command and a Read/Write Command in number of cycles. Number of
cycles is between 0 and 15.
• TWR: Write Recovery Delay
Reset value is 2.
This field defines the Write Recovery Time in number of cycles. Number of cycles is between 1 and 15.
• TRC: Row Cycle Delay
Reset value is 7 cycles.
This field defines the delay between an Activate command and Refresh command in number of cycles. Number of cycles is
between 0 and 15
• TRP: Row Precharge Delay
Reset Value is 2 cycles.
This field defines the delay between a Precharge Command and another command in number of cycles. Number of cycles
is between 0 and 15.
• TRRD Active bankA to Active bankB
Reset value is 2.
This field defines the delay between an Active command in BankA and an active command in bankB in number of cycles.
Number of cycles is between 1 and 15.
263
6438D–ATARM–13-Oct-09
• TWTR: Internal Write to Read Delay
Reset value is 0.
This field defines the internal write to read command Time in number of cycles. Number of cycles is between 1 and 7.
In the case of low-power DDR-SDRAM device only bit 24 (TWTR[0]) is used. Bit [26:25] must be set to 0.
Bit 24 (twtr[0])
Twtr value
0
1
1
2
• REDUCE_WRRD: Reduce Write to Read Delay
Reset value is 0.
This field reduces the delay between write to read access for low-power DDR-SDRAM devices with a latency equal to 2. To
use this feature, TWTR field must be equal to 0. Important to note is that some devices do not support this feature.
• TMRD: Load Mode Register Command to Active or Refresh Command
Reset Value is 2 cycles.
This field defines the delay between a Load mode register command and an active or refresh command in number of
cycles. Number of cycles is between 0 and 15.
264
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
22.7.5
Name:
DDRSDRC Timing 1 Parameter Register
DDRSDRC_T1PR
Access:
Read-write
Reset:
See Table 22-9
31
30
29
28
–
–
–
–
23
22
21
20
27
26
25
24
TXP
19
18
17
16
11
10
9
8
2
1
0
TXSRD
15
14
13
12
TXSNR
7
6
5
–
–
–
4
3
TRFC
This register can only be written if the bit WPEN is cleared in “DDRSDRC Write Protect Mode Register” on page 273.
• TRFC: Row Cycle Delay
Reset Value is 8 cycles.
This field defines the delay between a Refresh and an Activate command or Refresh command in number of cycles. Number of cycles is between 0 and 31
• TXSNR: Exit Self Refresh Delay to Non-read Command
Reset Value is 8 cycles.
This field defines the delay between cke set high and a non Read Command in number of cycles. Number of cycles is
between 0 and 15. This field is used for SDR-SDRAM and DDR-SDRAM devices. In the case of SDR-SDRAM devices and
Low-power DDR-SDRAM, this field is equivalent to TXSR timing.
• TXSRD: ExiT Self Refresh Delay to Read Command
Reset Value is C8.
This field defines the delay between cke set high and a Read Command in number of cycles. Number of cycles is between
0 and 255 cycles.This field is unique to DDR-SDRAM devices.
• TXP: Exit Power-down Delay to First Command
Reset Value is 3.
This field defines the delay between cke set high and a Valid Command in number of cycles. Number of cycles is between
0 and 15 cycles. This field is unique to Low-power DDR-SDRAM devices and DDR2-SDRAM devices.
265
6438D–ATARM–13-Oct-09
AT91SAM9G45
22.7.6
Name:
DDRSDRC Timing 2 Parameter Register
DDRSDRC_T2PR
Access:
Read-write
Reset:
See Table 22-9
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
1
0
TRTP
7
6
5
TRPA
4
3
2
TXARDS
TXARD
This register can only be written if the bit WPEN is cleared in “DDRSDRC Write Protect Mode Register” on page 273.
• TXARD: Exit Active Power Down Delay to Read Command in Mode “Fast Exit”.
The Reset Value is 0 Cycle.
This field defines the delay between cke set high and a Read Command in number of cycles. Number of cycles is between
0 and 15.
Note: This field is found only in DDR2-SDRAM devices.
• TXARDS: Exit Active Power Down Delay to Read Command in Mode “Slow Exit”.
The Reset Value is 0 Cycle.
This field defines the delay between cke set high and a Read Command in number of cycles. Number of cycles is between
0 and 15.
Note: This field is found only in DDR2-SDRAM devices.
• TRPA: Row Precharge All Delay
The Reset Value is 0 Cycle.
This field defines the delay between a Precharge ALL banks Command and another command in number of cycles. Number of cycles is between 0 and 15.
Note: This field is found only in DDR2-SDRAM devices.
• TRTP: Read to Precharge
The Reset Value is 0 Cycle.
This field defines the delay between Read Command and a Precharge command in number of cycle.
Number of cycles is between 0 and 15.
266
6438D–ATARM–13-Oct-09
AT91SAM9G45
22.7.7
Name:
DDRSDRC Low-power Register
DDRSDRC_LPR
Access:
Read-write
Reset:
See Table 22-9
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
APDE
15
14
13
12
11
10
9
–
–
7
6
–
TIMEOUT
5
DS
4
3
PASR
8
TCR
2
CLK_FR
1
0
LPCB
• LPCB: Low-power Command Bit
Reset value is “00”.
00 = Low-power Feature is inhibited: no power-down, self refresh and Deep power mode are issued to the SDRAM device.
01 = The DDRSDRC issues a Self Refresh Command to the SDRAM device, the clock(s) is/are de-activated and the CKE
signal is set low. The SDRAM device leaves the self refresh mode when accessed and enters it after the access.
10 = The DDRSDRC issues a Power-down Command to the SDRAM device after each access, the CKE signal is set low.
The SDRAM device leaves the power-down mode when accessed and enters it after the access.
11 = The DDRSDRC issues a Deep Power-down Command to the Low-power SDRAM device.This mode is unique to
Low-power SDRAM devices.
• CLK_FR: Clock Frozen Command Bit
Reset value is “0”.
This field sets the clock low during power-down mode or during deep power-down mode. Some SDRAM devices do not
support freezing the clock during power-down mode or during deep power-down mode. Refer to the SDRAM device
datasheet for details on this.
1 = Clock(s) is/are frozen.
0 = Clock(s) is/are not frozen.
• PASR: Partial Array Self Refresh
Reset value is “0”.
This field is unique to Low-power SDRAM. It is used to specify whether only one quarter, one half or all banks of the
SDRAM array are enabled. Disabled banks are not refreshed in self refresh mode.
The values of this field are dependant on Low-power SDRAM devices.
After the initialization sequence, as soon as PASR field is modified, Extended Mode Register in the external device memory is accessed automatically and PASR bits are updated. In function of the UPD_MR bit, update is done before entering in
self refresh mode or during a refresh command and a pending read or write access.
267
6438D–ATARM–13-Oct-09
AT91SAM9G45
• TCR: Temperature Compensated Self Refresh
Reset value is “0”.
This field is unique to Low-power SDRAM. It is used to program the refresh interval during self refresh mode, depending
on the case temperature of the low-power SDRAM.
The values of this field are dependent on Low-power SDRAM devices.
After the initialization sequence, as soon as TCR field is modified, Extended Mode Register is accessed automatically and
TCR bits are updated. In function of UPD_MR bit, update is done before entering in self refresh mode or during a refresh
command and a pending read or write access.
• DS: Drive Strength
Reset value is “0”.
This field is unique to Low-power SDRAM. It selects the driver strength of SDRAM output.
After the initialization sequence, as soon as DS field is modified, Extended Mode Register is accessed automatically and
DS bits are updated. In function of UPD_MR bit, update is done before entering in self refresh mode or during a refresh
command and a pending read or write access.
• TIMEOUT
Reset value is “00”.
This field defines when low-power mode is enabled.
00
The SDRAM controller activates the SDRAM low-power mode immediately after the end of the last transfer.
01
The SDRAM controller activates the SDRAM low-power mode 64 clock cycles after the end of the last transfer.
10
The SDRAM controller activates the SDRAM low-power mode 128 clock cycles after the end of the last transfer.
11
Reserved
• APDE: Active Power Down Exit Time
Reset value is “1”.
This mode is unique to DDR2-SDRAM devices. This mode allows to determine the active power-down mode, which
determines performance versus power saving.
0 = Fast Exit
1 = Slow Exit
After the initialization sequence, as soon as APDE field is modified Extended Mode Register, located in the memory of the
external device, is accessed automatically and APDE bits are updated. In function of the UPD_MR bit, update is done
before entering in self refresh mode or during a refresh command and a pending read or write access
• UPD_MR: Update Load Mode Register and Extended Mode Register
Reset value is “0”.
This bit is used to enable or disable automatic update of the Load Mode Register and Extended Mode Register. This
update is function of DDRSDRC integration in a system. DDRSDRC can either share or not share an external bus with
another controller.
00
Update is disabled.
01
DDRSDRC shares external bus. Automatic update is done during an refresh command and a pending read or write access in SDRAM device.
10
DDRSDRC does not share external bus. Automatic update is done before entering in self refresh mode.
11
Reserved
268
6438D–ATARM–13-Oct-09
AT91SAM9G45
22.7.8
Name:
DDRSDRC Memory Device Register
DDRSDRC_MD
Access:
Read-write
Reset:
See Table 22-9
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
–
–
–
DBW
–
MD
This register can only be written if the bit WPEN is cleared in “DDRSDRC Write Protect Mode Register” on page 273.
• MD: Memory Device
Indicates the type of memory used.
Reset value is for SDR-SDRAM device.
000 = SDR-SDRAM
001 = Low-power SDR-SDRAM
010 = Reserved
011 = Low-power DDR1-SDRAM
110 = DDR2-SDRAM
• DBW: Data Bus Width
Reset value is 16 bits.
0 = Data bus width is 32 bits (reserved for SDR-SDRAM device).
1 = Data bus width is 16 bits.
269
6438D–ATARM–13-Oct-09
AT91SAM9G45
22.7.9
DDRSDRC DLL Register
Name:
DDRSDRC_DLL
Access:
Read-only
Reset:
See Table 22-9
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
MDVAL
7
6
5
4
3
2
1
0
–
–
–
–
–
MDOVF
MDDEC
MDINC
The DLL logic is internally used by the controller in order to delay DQS inputs. This is necessary to center the strobe time
and the data valid window.
• MDINC: DLL Master Delay Increment
0 = The DLL is not incrementing the Master delay counter.
1 = The DLL is incrementing the Master delay counter.
• MDDEC: DLL Master Delay Decrement
0 = The DLL is not decrementing the Master delay counter.
1 = The DLL is decrementing the Master delay counter.
• MDOVF: DLL Master Delay Overflow Flag
0 = The Master delay counter has not reached its maximum value, or the Master is not locked yet.
1 = The Master delay counter has reached its maximum value, the Master delay counter increment is stopped and the DLL
forces the Master lock. If this flag is set, it means the DDRSDRC clock frequency is too low compared to Master delay line
number of elements.
• MDVAL: DLL Master Delay Value
Value of the Master delay counter.
270
6438D–ATARM–13-Oct-09
AT91SAM9G45
22.7.10
Name:
DDRSDRC High Speed Register
DDRSDRC_HS
Access:
Read-write
Reset:
See Table 22-9
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
–
DIS_ANTICIP_RE
AD
NO_OPTI
–
–
–
–
–
This register can only be written if the bit WPEN is cleared in “DDRSDRC Write Protect Mode Register” on page 273.
• NO_OPTI: No Optimization
0 = optimization is enabled.
1 = optimization is disabled.
• DIS_ANTICIP_READ
0 = anticip read access is enabled.
1 = anticip read access is disabled.
271
6438D–ATARM–13-Oct-09
AT91SAM9G45
22.7.11
Name:
DDRSDRC DELAY I/O Register
DDRSDRC_DELAYx [x=1..4]
Access:
Read-write
Reset:
See Table 22-9
31
30
29
28
27
26
DELAY8
23
22
21
20
19
18
DELAY6
15
14
13
6
24
17
16
9
8
1
0
DELAY5
12
11
10
DELAY4
7
25
DELAY7
DELAY3
5
DELAY2
4
3
2
DELAY1
• DELAYx:
Gives the number of elements in the delay line.
272
6438D–ATARM–13-Oct-09
AT91SAM9G45
22.7.12
Name:
DDRSDRC Write Protect Mode Register
DDRSDRC_WPMR
Access:
Read-write
Reset:
See Table 22-9
31
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
WPKEY
23
22
21
20
WPKEY
15
14
13
12
WPKEY
7
6
5
4
3
2
1
0
—
—
—
—
—
—
—
WPEN
• WPEN: Write Protect Enable
0 = Disables the Write Protect if WPKEY corresponds to 0x444452 (“DDR” in ASCII).
1 = Enables the Write Protect if WPKEY corresponds to 0x444452 (“DDR” in ASCII).
Protects the registers:
• “DDRSDRC Mode Register” on page 258
• “DDRSDRC Refresh Timer Register” on page 259
• “DDRSDRC Configuration Register” on page 260
• “DDRSDRC Timing 0 Parameter Register” on page 263
• “DDRSDRC Timing 1 Parameter Register” on page 265
• “DDRSDRC Timing 2 Parameter Register” on page 266
• “DDRSDRC Memory Device Register” on page 269
• “DDRSDRC High Speed Register” on page 271
• WPKEY: Write Protect KEY
Should be written at value 0x444452 (“DDR” in ASCII). Writing any other value in this field aborts the write operation of the
WPEN bit. Always reads as 0.
273
6438D–ATARM–13-Oct-09
AT91SAM9G45
22.7.13
Name:
DDRSDRC Write Protect Status Register
DDRSDRC_WPSR
Access:
Read-only
Reset:
See Table 22-9
31
30
29
28
27
26
25
24
—
—
—
—
—
—
—
—
23
22
21
20
19
18
17
16
11
10
9
8
WPVSRC
15
14
13
12
WPVSRC
7
6
5
4
3
2
1
0
—
—
—
—
—
—
—
WPVS
• WPVS: Write Protect Violation Status
0 = No Write Protect Violation has occurred since the last read of the DDRSDRC_WPSR register.
1 = A Write Protect Violation has occurred since the last read of the DDRSDRC_WPSR register. If this violation is an unauthorized attempt to write a protected register, the associated violation is reported into field WPVSRC.
• WPVSRC: Write Protect Violation Source
When WPVS is active, this field indicates the write-protected register (through address offset or code) in which a write
access has been attempted.
Note:
Reading DDRSDRC_WPSR automatically clears all fields.
274
6438D–ATARM–13-Oct-09
AT91SAM9G45
23. Error Corrected Code Controller (ECC)
23.1
Description
NAND Flash/SmartMedia devices contain by default invalid blocks which have one or more
invalid bits. Over the NAND Flash/SmartMedia lifetime, additional invalid blocks may occur
which can be detected/corrected by ECC code.
The ECC Controller is a mechanism that encodes data in a manner that makes possible the
identification and correction of certain errors in data. The ECC controller is capable of single bit
error correction and 2-bit random detection. When NAND Flash/SmartMedia have more than 2
bits of errors, the data cannot be corrected.
The ECC user interface is compliant with the ARM Advanced Peripheral Bus (APB rev2).
23.2
Block Diagram
Figure 23-1. Block Diagram
NAND Flash
Static
Memory
Controller
SmartMedia
Logic
ECC
Controller
Ctrl/ECC Algorithm
User Interface
APB
23.3
Functional Description
A page in NAND Flash and SmartMedia memories contains an area for main data and an additional area used for redundancy (ECC). The page is organized in 8-bit or 16-bit words. The page
size corresponds to the number of words in the main area plus the number of words in the extra
area used for redundancy.
Over time, some memory locations may fail to program or erase properly. In order to ensure that
data is stored properly over the life of the NAND Flash device, NAND Flash providers recom-
275
6438D–ATARM–13-Oct-09
AT91SAM9G45
mend to utilize either 1 ECC per 256 bytes of data, 1 ECC per 512 bytes of data or 1 ECC for all
of the page.
The only configurations required for ECC are the NAND Flash or the SmartMedia page size
(528/2112/4224) and the type of correction wanted (1 ECC for all the page/1 ECC per 256 bytes
of data /1 ECC per 512 bytes of data). Page size is configured setting the PAGESIZE field in the
ECC Mode Register (ECC_MR). Type of correction is configured setting the TYPeCORRECT
field in the ECC Mode Register (ECC_MR).
ECC is automatically computed as soon as a read (00h)/write (80h) command to the NAND
Flash or the SmartMedia is detected. Read and write access must start at a page boundary.
ECC results are available as soon as the counter reaches the end of the main area. Values in
the ECC Parity Registers (ECC_PR0 to ECC_PR15) are then valid and locked until a new start
condition occurs (read/write command followed by address cycles).
23.3.1
Write Access
Once the Flash memory page is written, the computed ECC codes are available in the ECC Parity (ECC_PR0 to ECC_PR15) registers. The ECC code values must be written by the software
application in the extra area used for redundancy. The number of write accesses in the extra
area is a function of the value of the type of correction field. For example, for 1 ECC per 256
bytes of data for a page of 512 bytes, only the values of ECC_PR0 and ECC_PR1 must be written by the software application. Other registers are meaningless.
23.3.2
Read Access
After reading the whole data in the main area, the application must perform read accesses to the
extra area where ECC code has been previously stored. Error detection is automatically performed by the ECC controller. Please note that it is mandatory to read consecutively the entire
main area and the locations where Parity and NParity values have been previously stored to let
the ECC controller perform error detection.
The application can check the ECC Status Registers (ECC_SR1/ECC_SR2) for any detected
errors. It is up to the application to correct any detected error. ECC computation can detect four
different circumstances:
• No error: XOR between the ECC computation and the ECC code stored at the end of the
NAND Flash or SmartMedia page is equal to 0. No error flags in the ECC Status Registers
(ECC_SR1/ECC_SR2).
• Recoverable error: Only the RECERR flags in the ECC Status registers
(ECC_SR1/ECC_SR2) are set. The corrupted word offset in the read page is defined by the
WORDADDR field in the ECC Parity Registers (ECC_PR0 to ECC_PR15). The corrupted bit
position in the concerned word is defined in the BITADDR field in the ECC Parity Registers
(ECC_PR0 to ECC_PR15).
• ECC error: The ECCERR flag in the ECC Status Registers (ECC_SR1/ECC_SR2) are set.
An error has been detected in the ECC code stored in the Flash memory. The position of the
corrupted bit can be found by the application performing an XOR between the Parity and the
NParity contained in the ECC code stored in the Flash memory.
• Non correctable error: The MULERR flag in the ECC Status Registers
(ECC_SR1/ECC_SR2) are set. Several unrecoverable errors have been detected in the
Flash memory page.
276
6438D–ATARM–13-Oct-09
AT91SAM9G45
ECC Status Registers, ECC Parity Registers are cleared when a read/write command is
detected or a software reset is performed.
For Single-bit Error Correction and Double-bit Error Detection (SEC-DED) hsiao code is used.
24-bit ECC is generated in order to perform one bit correction per 256 or 512 bytes for pages of
512/2048/4096 8-bit words. 32-bit ECC is generated in order to perform one bit correction per
512/1024/2048/4096 8- or 16-bit words.They are generated according to the schemes shown in
Figure 23-2 and Figure 23-3.
Figure 23-2. Parity Generation for
Bit6512/1024/2048/4096
Bit5 Bit4
Bit2 Words
Bit3 8-bit
Bit1
1st byte
Bit7 Bit6 Bit5 Bit4
2nd byte
Bit3 Bit2 Bit1
3rd byte
Bit7 Bit6 Bit5 Bit4
Bit3 Bit2 Bit1
4 th byte
Bit7 Bit6 Bit5 Bit4
Bit3 Bit2 Bit1
(page size -3 )th byte
(page size -2 )th byte
(page size -1 )th byte
Page size th byte
Bit7
Bit7
Bit6
Bit6
Bit5
Bit5
Bit7
Bit7
Bit6
Bit6
Bit5
Bit5
P1
P1'
P1
P2
P8'
Bit0
Bit0
P8
P8
P8'
P16
P8
P8'
P16'
Bit3
Bit2
Bit1
Bit0
Bit4
Bit3
Bit2
Bit1
Bit0
Bit3
Bit3
Bit2
Bit2
Bit1
Bit1
Bit0
P1
P1'
P1
P1'
Bit4
Bit4
P1'
P2
P2'
P16
Bit0
Bit4
P4
Page size
Page size
Page size
Page size
Bit0
Bit0
P8'
P32
PX
P32
PX'
P16'
P2'
P4'
P1=bit7(+)bit5(+)bit3(+)bit1(+)P1
P2=bit7(+)bit6(+)bit3(+)bit2(+)P2
P4=bit7(+)bit6(+)bit5(+)bit4(+)P4
P1'=bit6(+)bit4(+)bit2(+)bit0(+)P1'
P2' bit5( )bit4( )bit1( )bit0( )P2'
To calculate P8’ to PX’ and P8 to PX, apply the algorithm that follows.
= 512
= 1024
= 2048
= 4096
Px = 2048
Px = 4096
Px = 8192
Px = 16384
Page size = 2n
for i =0 to n
begin
for (j = 0 to page_size_byte)
begin
if(j[i] ==1)
P[2i+3]=bit7(+)bit6(+)bit5(+)bit4(+)bit3(+)
bit2(+)bit1(+)bit0(+)P[2i+3]
else
P[2i+3]’=bit7(+)bit6(+)bit5(+)bit4(+)bit3(+)
bit2(+)bit1(+)bit0(+)P[2i+3]'
end
end
277
6438D–ATARM–13-Oct-09
6438D–ATARM–13-Oct-09
(Page size -3 )th word
(Page size -2 )th word
(Page size -1 )th word
Page size th word
3rd word
4th word
1st word
2nd word
(+)
AT91SAM9G45
Figure 23-3. Parity Generation for 512/1024/2048/4096 16-bit Words
278
AT91SAM9G45
To calculate P8’ to PX’ and P8 to PX, apply the algorithm that follows.
Page size = 2n
for i =0 to n
begin
for (j = 0 to page_size_word)
begin
if(j[i] ==1)
P[2i+3]= bit15(+)bit14(+)bit13(+)bit12(+)
bit11(+)bit10(+)bit9(+)bit8(+)
bit7(+)bit6(+)bit5(+)bit4(+)bit3(+)
bit2(+)bit1(+)bit0(+)P[2n+3]
else
P[2i+3]’=bit15(+)bit14(+)bit13(+)bit12(+)
bit11(+)bit10(+)bit9(+)bit8(+)
bit7(+)bit6(+)bit5(+)bit4(+)bit3(+)
bit2(+)bit1(+)bit0(+)P[2i+3]'
end
end
279
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.4
Error Corrected Code Controller (ECC) User Interface
Table 23-1.
Register Mapping
Offset
Register
Name
Access
Reset
0x00
ECC Control Register
ECC_CR
Write-only
0x0
0x04
ECC Mode Register
ECC_MR
Read-write
0x0
0x08
ECC Status1 Register
ECC_SR1
Read-only
0x0
0x0C
ECC Parity Register 0
ECC_PR0
Read-only
0x0
0x10
ECC Parity Register 1
ECC_PR1
Read-only
0x0
0x14
ECC Status2 Register
ECC_SR2
Read-only
0x0
0x18
ECC Parity 2
ECC_PR2
Read-only
0x0
0x1C
ECC Parity 3
ECC_PR3
Read-only
0x0
0x20
ECC Parity 4
ECC_PR4
Read-only
0x0
0x24
ECC Parity 5
ECC_PR5
Read-only
0x0
0x28
ECC Parity 6
ECC_PR6
Read-only
0x0
0x2C
ECC Parity 7
ECC_PR7
Read-only
0x0
0x30
ECC Parity 8
ECC_PR8
Read-only
0x0
0x34
ECC Parity 9
ECC_PR9
Read-only
0x0
0x38
ECC Parity 10
ECC_PR10
Read-only
0x0
0x3C
ECC Parity 11
ECC_PR11
Read-only
0x0
0x40
ECC Parity 12
ECC_PR12
Read-only
0x0
0x44
ECC Parity 13
ECC_PR13
Read-only
0x0
0x48
ECC Parity 14
ECC_PR14
Read-only
0x0
0x4C
ECC Parity 15
ECC_PR15
Read-only
0x0
280
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.4.1
Name:
ECC Control Register
ECC_CR
Access Type:
31
–
23
–
15
–
7
–
Write-only
30
–
22
–
14
–
6
–
29
–
21
–
13
–
5
–
28
–
20
–
12
–
4
–
27
–
19
–
11
–
3
–
26
–
18
–
10
–
2
–
25
–
17
–
9
–
1
SRST
24
–
16
–
8
–
0
RST
• RST: RESET Parity
Provides reset to current ECC by software.
1: Reset ECC Parity registers
0: No effect
• SRST: Soft Reset
Provides soft reset to ECC block
1: Resets all registers.
0: No effect.
281
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.4.2
ECC Mode Register
Register Name:
ECC_MR
Access Type:
31
–
23
–
15
–
7
–
Read-write
30
–
22
–
14
–
6
–
29
28
–
–
21
20
–
–
13
12
–
–
5
4
TYPECORRECT
27
–
19
–
11
–
3
–
26
–
18
–
10
–
2
–
25
–
17
–
9
–
1
24
–
16
–
8
–
0
PAGESIZE
• PAGESIZE: Page Size
This field defines the page size of the NAND Flash device.
Page Size
Description
00
528 words
01
1056 words
10
2112 words
11
4224 words
A word has a value of 8 bits or 16 bits, depending on the NAND Flash or SmartMedia memory organization.
• TYPECORRECT: Type of Correction
00: 1 bit correction for a page size of 512/1024/2048/4096 bytes.
01: 1 bit correction for 256 bytes of data for a page size of 512/2048/4096 bytes.
10: 1 bit correction for 512 bytes of data for a page size of 512/2048/4096 bytes.
282
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.4.3
ECC Status Register 1
Register Name:
ECC_SR1
Access Type:
31
–
23
–
15
–
7
–
Read-only
30
MULERR7
22
MULERR5
14
MULERR3
6
MULERR1
29
ECCERR7
21
ECCERR5
13
ECCERR3
5
ECCERR1
28
RECERR7
20
RECERR5
12
RECERR3
4
RECERR1
27
–
19
–
11
–
3
–
26
MULERR6
18
MULERR4
10
MULERR2
2
MULERR0
25
ECCERR6
17
ECCERR4
9
ECCERR2
1
ECCERR0
24
RECERR6
16
RECERR4
8
RECERR2
0
RECERR0
• RECERR0: Recoverable Error
0 = No Errors Detected.
1 = Errors Detected. If MUL_ERROR is 0, a single correctable error was detected. Otherwise multiple uncorrected errors
were detected.
• ECCERR0: ECC Error
0 = No Errors Detected.
1 = A single bit error occurred in the ECC bytes.
If TYPECORRECT = 0, read both ECC Parity 0 and ECC Parity 1 registers, the error occurred at the location which contains a 1 in the least significant 16 bits; else read ECC Parity 0 register, the error occurred at the location which contains a
1 in the least significant 24 bits.
• MULERR0: Multiple Error
0 = No Multiple Errors Detected.
1 = Multiple Errors Detected.
• RECERR1: Recoverable Error in the page between the 256th and the 511th bytes or the 512th and the 1023rd
bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = Errors Detected. If MUL_ERROR is 0, a single correctable error was detected. Otherwise multiple uncorrected errors
were detected.
• ECCERR1: ECC Error in the page between the 256th and the 511th bytes or the 512th and the 1023rd bytes
Fixed to 0 if TYPECORREC = 0
0 = No Errors Detected.
1 = A single bit error occurred in the ECC bytes.
Read ECC Parity 1 register, the error occurred at the location which contains a 1 in the least significant 24 bits.
• MULERR1: Multiple Error in the page between the 256th and the 511th bytes or the 512th and the 1023rd bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Multiple Errors Detected.
1 = Multiple Errors Detected.
283
6438D–ATARM–13-Oct-09
AT91SAM9G45
• RECERR2: Recoverable Error in the page between the 512th and the 767th bytes or the 1024th and the 1535th
bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = Errors Detected. If MUL_ERROR is 0, a single correctable error was detected. Otherwise, multiple uncorrected errors
were detected.
• ECCERR2: ECC Error in the page between the 512th and the 767th bytes or the 1024th and the 1535th bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = A single bit error occurred in the ECC bytes.
Read ECC Parity 2 register, the error occurred at the location which contains a 1 in the least significant 24 bits.
• MULERR2: Multiple Error in the page between the 512th and the 767th bytes or the 1024th and the 1535th bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Multiple Errors Detected.
1 = Multiple Errors Detected.
• RECERR3: Recoverable Error in the page between the 768th and the 1023rd bytes or the 1536th and the 2047th
bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = Errors Detected. If MUL_ERROR is 0, a single correctable error was detected. Otherwise multiple uncorrected errors
were detected.
• ECCERR3: ECC Error in the page between the 768th and the 1023rd bytes or the 1536th and the 2047th bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = A single bit error occurred in the ECC bytes.
Read ECC Parity 3 register, the error occurred at the location which contains a 1 in the least significant 24 bits.
• MULERR3: Multiple Error in the page between the 768th and the 1023rd bytes or the 1536th and the 2047th bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Multiple Errors Detected.
1 = Multiple Errors Detected.
• RECERR4: Recoverable Error in the page between the 1024th and the 1279th bytes or the 2048th and the 2559th
bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = Errors Detected. If MUL_ERROR is 0, a single correctable error was detected. Otherwise multiple uncorrected errors
were detected.
284
6438D–ATARM–13-Oct-09
AT91SAM9G45
• ECCERR4: ECC Error in the page between the 1024th and the 1279th bytes or the 2048th and the 2559th bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = A single bit error occurred in the ECC bytes.
Read ECC Parity 4 register, the error occurred at the location which contains a 1 in the least significant 24 bits.
• MULERR4: Multiple Error in the page between the 1024th and the 1279th bytes or the 2048th and the 2559th
bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Multiple Errors Detected.
1 = Multiple Errors Detected.
• RECERR5: Recoverable Error in the page between the 1280th and the 1535th bytes or the 2560th and the 3071st
bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = Errors Detected. If MUL_ERROR is 0, a single correctable error was detected. Otherwise multiple uncorrected errors
were detected
• ECCERR5: ECC Error in the page between the 1280th and the 1535th bytes or the 2560th and the 3071st bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = A single bit error occurred in the ECC bytes.
Read ECC Parity 5 register, the error occurred at the location which contains a 1 in the least significant 24 bits.
• MULERR5: Multiple Error in the page between the 1280th and the 1535th bytes or the 2560th and the 3071st
bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Multiple Errors Detected.
1 = Multiple Errors Detected.
• RECERR6: Recoverable Error in the page between the 1536th and the 1791st bytes or the 3072nd and the 3583rd
bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = Errors Detected. If MUL_ERROR is 0, a single correctable error was detected. Otherwise multiple uncorrected errors
were detected.
• ECCERR6: ECC Error in the page between the 1536th and the 1791st bytes or the 3072nd and the 3583rd bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = A single bit error occurred in the ECC bytes.
Read ECC Parity 6 register, the error occurred at the location which contains a 1 in the least significant 24 bits.
285
6438D–ATARM–13-Oct-09
AT91SAM9G45
• MULERR6: Multiple Error in the page between the 1536th and the 1791st bytes or the 3072nd and the 3583rd
bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Multiple Errors Detected.
1 = Multiple Errors Detected.
• RECERR7: Recoverable Error in the page between the 1792nd and the 2047th bytes or the 3584th and the 4095th
bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = Errors Detected. If MUL_ERROR is 0, a single correctable error was detected. Otherwise, multiple uncorrected errors
were detected.
• ECCERR7: ECC Error in the page between the 1792nd and the 2047th bytes or the 3584th and the 4095th bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = A single bit error occurred in the ECC bytes.
Read ECC Parity 7 register, the error occurred at the location which contains a 1 in the least significant 24 bits.
• MULERR7: Multiple Error in the page between the 1792nd and the 2047th bytes or the 3584th and the 4095th
bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Multiple Errors Detected.
1 = Multiple Errors Detected.
286
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.4.4
ECC Status Register 2
Register Name:
ECC_SR2
Access Type:
31
–
23
–
15
–
7
–
Read-only
30
MULERR15
22
MULERR13
14
MULERR11
6
MULERR9
29
ECCERR15
21
ECCERR13
13
ECCERR11
5
ECCERR9
28
RECERR15
20
RECERR13
12
RECERR11
4
RECERR9
27
–
19
–
11
–
3
–
26
MULERR14
18
MULERR12
10
MULERR10
2
MULERR8
25
ECCERR14
17
ECCERR12
9
ECCERR10
1
ECCERR8
24
RECERR14
16
RECERR12
8
RECERR10
0
RECERR8
• RECERR8: Recoverable Error in the page between the 2048th and the 2303rd bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = Errors Detected. If MUL_ERROR is 0, a single correctable error was detected. Otherwise multiple uncorrected errors
were detected
• ECCERR8: ECC Error in the page between the 2048th and the 2303rd bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = A single bit error occurred in the ECC bytes.
Read ECC Parity 8 register, the error occurred at the location which contains a 1 in the least significant 24 bits.
• MULERR8: Multiple Error in the page between the 2048th and the 2303rd bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Multiple Errors Detected.
1 = Multiple Errors Detected.
• RECERR9: Recoverable Error in the page between the 2304th and the 2559th bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = Errors Detected. If MUL_ERROR is 0, a single correctable error was detected. Otherwise multiple uncorrected errors
were detected.
• ECCERR9: ECC Error in the page between the 2304th and the 2559th bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = A single bit error occurred in the ECC bytes.
Read ECC Parity 9 register, the error occurred at the location which contains a 1 in the least significant 24 bits.
• MULERR9: Multiple Error in the page between the 2304th and the 2559th bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Multiple Errors Detected.
287
6438D–ATARM–13-Oct-09
AT91SAM9G45
1 = Multiple Errors Detected.
• RECERR10: Recoverable Error in the page between the 2560th and the 2815th bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = Errors Detected. If MUL_ERROR is 0, a single correctable error was detected. Otherwise, multiple uncorrected errors
were detected.
• ECCERR10: ECC Error in the page between the 2560th and the 2815th bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = A single bit error occurred in the ECC bytes.
Read ECC Parity 10 register, the error occurred at the location which contains a 1 in the least significant 24 bits.
• MULERR10: Multiple Error in the page between the 2560th and the 2815th bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Multiple Errors Detected.
1 = Multiple Errors Detected.
• RECERR11: Recoverable Error in the page between the 2816th and the 3071st bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = Errors Detected. If MUL_ERROR is 0, a single correctable error was detected. Otherwise, multiple uncorrected errors
were detected
• ECCERR11: ECC Error in the page between the 2816th and the 3071st bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = A single bit error occurred in the ECC bytes.
Read ECC Parity 11 register, the error occurred at the location which contains a 1 in the least significant 24 bits.
• MULERR11: Multiple Error in the page between the 2816th and the 3071st bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Multiple Errors Detected.
1 = Multiple Errors Detected.
• RECERR12: Recoverable Error in the page between the 3072nd and the 3327th bytes
Fixed to 0 if TYPECORREC = 0
0 = No Errors Detected
1 = Errors Detected. If MUL_ERROR is 0, a single correctable error was detected. Otherwise multiple uncorrected errors
were detected
• ECCERR12: ECC Error in the page between the 3072nd and the 3327th bytes
Fixed to 0 if TYPECORREC = 0
288
6438D–ATARM–13-Oct-09
AT91SAM9G45
0 = No Errors Detected
1 = A single bit error occurred in the ECC bytes.
Read ECC Parity 12 register, the error occurred at the location which contains a 1 in the least significant 24 bits.
• MULERR12: Multiple Error in the page between the 3072nd and the 3327th bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Multiple Errors Detected.
1 = Multiple Errors Detected.
• RECERR13: Recoverable Error in the page between the 3328th and the 3583rd bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = Errors Detected. If MUL_ERROR is 0, a single correctable error was detected. Otherwise multiple uncorrected errors
were detected.
• ECCERR13: ECC Error in the page between the 3328th and the 3583rd bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = A single bit error occurred in the ECC bytes.
Read ECC Parity 13 register, the error occurred at the location which contains a 1 in the least significant 24 bits.
• MULERR13: Multiple Error in the page between the 3328th and the 3583rd bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Multiple Errors Detected.
1 = Multiple Errors Detected.
• RECERR14: Recoverable Error in the page between the 3584th and the 3839th bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = Errors Detected. If MUL_ERROR is 0, a single correctable error was detected. Otherwise, multiple uncorrected errors
were detected.
• ECCERR14: ECC Error in the page between the 3584th and the 3839th bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = A single bit error occurred in the ECC bytes.
Read ECC Parity 14 register, the error occurred at the location which contains a 1 in the least significant 24 bits.
• MULERR14: Multiple Error in the page between the 3584th and the 3839th bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Multiple Errors Detected.
1 = Multiple Errors Detected.
289
6438D–ATARM–13-Oct-09
AT91SAM9G45
• RECERR15: Recoverable Error in the page between the 3840th and the 4095th bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Errors Detected.
1 = Errors Detected. If MUL_ERROR is 0, a single correctable error was detected. Otherwise, multiple uncorrected errors
were detected
• ECCERR15: ECC Error in the page between the 3840th and the 4095th bytes
Fixed to 0 if TYPECORREC = 0
0 = No Errors Detected.
1 = A single bit error occurred in the ECC bytes.
Read ECC Parity 15 register, the error occurred at the location which contains a 1 in the least significant 24 bits.
• MULERR15: Multiple Error in the page between the 3840th and the 4095th bytes
Fixed to 0 if TYPECORREC = 0.
0 = No Multiple Errors Detected.
1 = Multiple Errors Detected.
290
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.5
Registers for 1 ECC for a page of 512/1024/2048/4096 bytes
23.5.1
ECC Parity Register 0
Register Name:
ECC_PR0
Access Type:
Read-only
31
–
23
–
15
30
–
22
–
14
29
–
21
–
13
28
–
20
–
12
7
6
5
4
27
–
19
–
11
26
–
18
–
10
3
2
25
–
17
–
9
24
–
16
–
8
1
0
WORDADDR
WORDADDR
BITADDR
Once the entire main area of a page is written with data, the register content must be stored at any free location of the
spare area.
• BITADDR: Bit Address
During a page read, this value contains the corrupted bit offset where an error occurred, if a single error was detected. If
multiple errors were detected, this value is meaningless.
• WORDADDR: Word Address
During a page read, this value contains the word address (8-bit or 16-bit word depending on the memory plane organization) where an error occurred, if a single error was detected. If multiple errors were detected, this value is meaningless.
291
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.5.2
ECC Parity Register 1
Register Name:
ECC_PR1
Access Type:
Read-only
31
–
23
–
15
30
–
22
–
14
29
–
21
–
13
28
–
20
–
12
7
6
5
4
27
–
19
–
11
26
–
18
–
10
25
–
17
–
9
24
–
16
–
8
3
2
1
0
NPARITY
NPARITY
• NPARITY:
Parity N
292
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.6
Registers for 1 ECC per 512 bytes for a page of 512/2048/4096 bytes,
8-bit word
23.6.1
ECC Parity Register 0
Register Name:
ECC_PR0
Access Type:
Read-only
31
–
23
30
–
22
29
–
21
15
14
13
7
6
28
–
20
NPARITY0
12
27
–
19
26
–
18
25
–
17
24
–
16
11
10
9
8
4
3
2
1
BITADDR0
0
NPARITY0
5
WORDADDR0
WORDADD0
Once the entire main area of a page is written with data, the register content must be stored at any free location of the
spare area.
• BITADDR0: corrupted Bit Address in the page between the first byte and the 511th bytes
During a page read, this value contains the corrupted bit offset where an error occurred, if a single error was detected. If
multiple errors were detected, this value is meaningless.
• WORDADDR0: corrupted Word Address in the page between the first byte and the 511th bytes
During a page read, this value contains the word address (8-bit word) where an error occurred, if a single error was
detected. If multiple errors were detected, this value is meaningless.
• NPARITY0:
Parity N
293
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.6.2
ECC Parity Register 1
Register Name:
ECC_PR1
Access Type:
Read-only
31
–
23
30
–
22
15
14
29
–
21
13
28
–
20
NPARITY1
12
27
–
19
26
–
18
11
10
NPARITY1
7
6
5
WORDADDR1
25
–
17
24
–
16
9
8
1
BITADDR1
0
WORDADD1
4
3
2
Once the entire main area of a page is written with data, the register content must be stored at any free location of the
spare area.
• BITADDR1: corrupted Bit Address in the page between the 512th and the 1023rd bytes
During a page read, this value contains the corrupted bit offset where an error occurred, if a single error was detected. If
multiple errors were detected, this value is meaningless.
• WORDADDR1: corrupted Word Address in the page between the 512th and the 1023rd bytes
During a page read, this value contains the word address (8-bit word) where an error occurred, if a single error was
detected. If multiple errors were detected, this value is meaningless.
• NPARITY1:
Parity N
294
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.6.3
ECC Parity Register 2
Register Name:
ECC_PR2
Access Type:
Read-only
31
–
23
30
–
22
15
14
29
–
21
13
28
–
20
NPARITY2
12
27
–
19
26
–
18
11
4
3
10
9
WORDADDR2
2
1
BITADDR2
NPARITY2
7
6
5
WORDADDR2
25
–
17
24
–
16
8
0
Once the entire main area of a page is written with data, the register content must be stored at any free location of the
spare area.
• BITADDR2: corrupted Bit Address in the page between the 1023rd and the 1535th bytes
During a page read, this value contains the corrupted bit offset where an error occurred, if a single error was detected. If
multiple errors were detected, this value is meaningless.
• WORDADDR2: corrupted Word Address in the page in the page between the 1023rd and the 1535th bytes
During a page read, this value contains the word address (8-bit word) where an error occurred, if a single error was
detected. If multiple errors were detected, this value is meaningless.
• NPARITY2:
Parity N
295
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.6.4
ECC Parity Register 3
Register Name:
ECC_PR3
Access Type:
Read-only
31
–
23
30
–
22
15
14
29
–
21
13
28
–
20
NPARITY3
12
27
–
19
26
–
18
11
4
3
10
9
WORDADDR3
2
1
BITADDR3
NPARITY3
7
6
5
WORDADDR3
25
–
17
24
–
16
8
0
Once the entire main area of a page is written with data, the register content must be stored at any free location of the
spare area.
• BITADDR3: corrupted Bit Address in the page between the1536th and the 2047th bytes
During a page read, this value contains the corrupted bit offset where an error occurred, if a single error was detected. If
multiple errors were detected, this value is meaningless.
• WORDADDR3 corrupted Word Address in the page between the 1536th and the 2047th bytes
During a page read, this value contains the word address (8-bit word) where an error occurred, if a single error was
detected. If multiple errors were detected, this value is meaningless.
• NPARITY3
Parity N
296
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.6.5
ECC Parity Register 4
Register Name:
ECC_PR4
Access Type:
Read-only
31
–
23
30
–
22
15
14
29
–
21
13
28
–
20
NPARITY4
12
27
–
19
26
–
18
11
4
3
10
9
WORDADDR4
2
1
BITADDR4
NPARITY4
7
6
5
WORDADDR4
25
–
17
24
–
16
8
0
Once the entire main area of a page is written with data, the register content must be stored at any free location of the
spare area.
• BITADDR4: corrupted Bit Address in the page between the 2048th and the 2559th bytes
During a page read, this value contains the corrupted bit offset where an error occurred, if a single error was detected. If
multiple errors were detected, this value is meaningless.
• WORDADDR4: corrupted Word Address in the page between the 2048th and the 2559th bytes
During a page read, this value contains the word address (8-bit word) where an error occurred, if a single error was
detected. If multiple errors were detected, this value is meaningless.
• NPARITY4:
Parity N
297
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.6.6
ECC Parity Register 5
Register Name:
ECC_PR5
Access Type:
Read-only
31
–
23
30
–
22
15
14
29
–
21
13
28
–
20
NPARITY5
12
27
–
19
26
–
18
11
4
3
10
9
WORDADDR5
2
1
BITADDR5
NPARITY5
7
6
5
WORDADDR5
25
–
17
24
–
16
8
0
Once the entire main area of a page is written with data, the register content must be stored at any free location of the
spare area.
• BITADDR5: corrupted Bit Address in the page between the 2560th and the 3071st bytes
During a page read, this value contains the corrupted bit offset where an error occurred, if a single error was detected. If
multiple errors were detected, this value is meaningless.
• WORDADDR5: corrupted Word Address in the page between the 2560th and the 3071st bytes
During a page read, this value contains the word address (8-bit word) where an error occurred, if a single error was
detected. If multiple errors were detected, this value is meaningless.
• NPARITY5:
Parity N
298
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.6.7
ECC Parity Register 6
Register Name:
ECC_PR6
Access Type:
Read-only
31
–
23
30
–
22
15
14
29
–
21
13
28
–
20
NPARITY6
12
27
–
19
26
–
18
11
4
3
10
9
WORDADDR6
2
1
BITADDR6
NPARITY6
7
6
5
WORDADDR6
25
–
17
24
–
16
8
0
Once the entire main area of a page is written with data, the register content must be stored at any free location of the
spare area.
• BITADDR6: corrupted Bit Address in the page between the 3072nd and the 3583rd bytes
During a page read, this value contains the corrupted bit offset where an error occurred, if a single error was detected. If
multiple errors were detected, this value is meaningless.
• WORDADDR6: corrupted Word Address in the page between the 3072nd and the 3583rd bytes
During a page read, this value contains the word address (8-bit word) where an error occurred, if a single error was
detected. If multiple errors were detected, this value is meaningless.
• NPARITY6:
Parity N
299
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.6.8
ECC Parity Register 7
Register Name:
ECC_PR7
Access Type:
Read-only
31
–
23
30
–
22
15
14
29
–
21
13
28
–
20
NPARITY7
12
27
–
19
26
–
18
11
4
3
10
9
WORDADDR7
2
1
BITADDR7
NPARITY7
7
6
5
WORDADDR7
25
–
17
24
–
16
8
0
Once the entire main area of a page is written with data, the register content must be stored at any free location of the
spare area.
• BITADDR7: corrupted Bit Address in the page between the 3584h and the 4095th bytes
During a page read, this value contains the corrupted bit offset where an error occurred, if a single error was detected. If
multiple errors were detected, this value is meaningless.
• WORDADDR7: corrupted Word Address in the page between the 3584th and the 4095th bytes
During a page read, this value contains the word address (8-bit word) where an error occurred, if a single error was
detected. If multiple errors were detected, this value is meaningless.
• NPARITY7:
Parity N
300
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.7
Registers for 1 ECC per 256 bytes for a page of 512/2048/4096 bytes,
8-bit word
23.7.1
ECC Parity Register 0
Register Name:
ECC_PR0
Access Type:
Read-only
31
–
23
0
15
30
–
22
29
–
21
28
–
20
14
13
12
7
6
NPARITY0
5
WORDADDR0
4
27
–
19
NPARITY0
11
0
3
26
–
18
25
–
17
24
–
16
10
9
WORDADDR0
1
BITADDR0
8
2
0
Once the entire main area of a page is written with data, the register content must be stored at any free location of the
spare area.
• BITADDR0: corrupted Bit Address in the page between the first byte and the 255th bytes
During a page read, this value contains the corrupted bit offset where an error occurred, if a single error was detected. If
multiple errors were detected, this value is meaningless.
• WORDADDR0: corrupted Word Address in the page between the first byte and the 255th bytes
During a page read, this value contains the word address (8-bit word) where an error occurred, if a single error was
detected. If multiple errors were detected, this value is meaningless.
• NPARITY0:
Parity N
301
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.7.2
ECC Parity Register 1
Register Name:
ECC_PR1
Access Type:
31
–
23
0
15
Read-only
30
–
22
14
29
–
21
28
–
20
13
12
NPARITY1
7
6
5
WORDADDR1
4
27
–
19
NPARITY1
11
0
3
26
–
18
25
–
17
24
–
16
10
9
WORDADDR1
1
BITADDR1
8
2
0
Once the entire main area of a page is written with data, the register content must be stored at any free location of the
spare area
• BITADDR1: corrupted Bit Address in the page between the 256th and the 511th bytes
During a page read, this value contains the corrupted bit offset where an error occurred, if a single error was detected. If
multiple errors were detected, this value is meaningless.
• WORDADDR1: corrupted Word Address in the page between the 256th and the 511th bytes
During a page read, this value contains the word address (8-bit word) where an error occurred, if a single error was
detected. If multiple errors were detected, this value is meaningless.
• NPARITY1:
Parity N
302
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.7.3
ECC Parity Register 2
Register Name:
ECC_PR2
Access Type:
31
–
23
0
15
Read-only
30
–
22
14
29
–
21
28
–
20
13
12
NPARITY2
7
6
5
WORDADDR2
4
27
–
19
NPARITY2
11
0
3
26
–
18
25
–
17
24
–
16
10
9
WORDADD2
1
BITADDR2
8
2
0
Once the entire main area of a page is written with data, the register content must be stored at any free location of the
spare area.
• BITADDR2: corrupted Bit Address in the page between the 512th and the 767th bytes
During a page read, this value contains the corrupted bit offset where an error occurred, if a single error was detected. If
multiple errors were detected, this value is meaningless.
• WORDADDR2: corrupted Word Address in the page between the 512th and the 767th bytes
During a page read, this value contains the word address (8-bit word) where an error occurred, if a single error was
detected. If multiple errors were detected, this value is meaningless.
• NPARITY2:
Parity N
303
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.7.4
ECC Parity Register 3
Register Name:
ECC_PR3
Access Type:
31
–
23
0
15
Read-only
30
–
22
14
29
–
21
28
–
20
13
12
NPARITY3
7
6
5
WORDADDR3
4
27
–
19
NPARITY3
11
0
3
26
–
18
25
–
17
24
–
16
10
9
WORDADDR3
1
BITADDR3
8
2
0
Once the entire main area of a page is written with data, the register content must be stored at any free location of the
spare area.
• BITADDR3: corrupted Bit Address in the page between the 768th and the 1023rd bytes
During a page read, this value contains the corrupted bit offset where an error occurred, if a single error was detected. If
multiple errors were detected, this value is meaningless.
• WORDADDR3: corrupted Word Address in the page between the 768th and the 1023rd bytes
During a page read, this value contains the word address (8-bit word) where an error occurred, if a single error was
detected. If multiple errors were detected, this value is meaningless
• NPARITY3:
Parity N
304
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.7.5
ECC Parity Register 4
Register Name:
ECC_PR4
Access Type:
31
–
23
0
15
Read-only
30
–
22
14
29
–
21
28
–
20
13
12
NPARITY4
7
6
5
WORDADDR4
4
27
–
19
NPARITY4
11
0
3
26
–
18
10
2
25
–
17
9
WORDADDR4
1
BITADDR4
24
–
16
8
0
Once the entire main area of a page is written with data, the register content must be stored at any free location of the
spare area
• BITADDR4: corrupted bit address in the page between the 1024th and the 1279th bytes
During a page read, this value contains the corrupted bit offset where an error occurred, if a single error was detected. If
multiple errors were detected, this value is meaningless.
• WORDADDR4: corrupted word address in the page between the 1024th and the 1279th bytes
During a page read, this value contains the word address (8-bit word) where an error occurred, if a single error was
detected. If multiple errors were detected, this value is meaningless.
• NPARITY4
Parity N
305
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.7.6
ECC Parity Register 5
Register Name:
ECC_PR5
Access Type:
31
–
23
0
15
Read-only
30
–
22
14
29
–
21
28
–
20
13
12
NPARITY5
7
6
5
WORDADDR5
4
27
–
19
NPARITY5
11
0
3
26
–
18
10
2
25
–
17
9
WORDADDR5
1
BITADDR5
24
–
16
8
0
Once the entire main area of a page is written with data, the register content must be stored at any free location of the
spare area.
• BITADDR5: corrupted Bit Address in the page between the 1280th and the 1535th bytes
During a page read, this value contains the corrupted bit offset where an error occurred, if a single error was detected. If
multiple errors were detected, this value is meaningless.
• WORDADDR5: corrupted Word Address in the page between the 1280th and the 1535th bytes
During a page read, this value contains the word address (8-bit word) where an error occurred, if a single error was
detected. If multiple errors were detected, this value is meaningless.
• NPARITY5:
Parity N
306
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.7.7
ECC Parity Register 6
Register Name:
ECC_PR6
Access Type:
31
–
23
0
15
Read-only
30
–
22
14
29
–
21
28
–
20
13
12
NPARITY6
7
6
5
WORDADDR6
4
27
–
19
NPARITY6
11
0
3
26
–
18
25
–
17
24
–
16
10
9
WORDADDR6
1
BITADDR6
8
2
0
Once the entire main area of a page is written with data, the register content must be stored at any free location of the
spare area.
• BITADDR6: corrupted bit address in the page between the 1536th and the1791st bytes
During a page read, this value contains the corrupted bit offset where an error occurred, if a single error was detected. If
multiple errors were detected, this value is meaningless.
• WORDADDR6: corrupted word address in the page between the 1536th and the1791st bytes
During a page read, this value contains the word address (8-bit word) where an error occurred, if a single error was
detected. If multiple errors were detected, this value is meaningless.
• NPARITY6:
Parity N
307
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.7.8
ECC Parity Register 7
Register Name:
ECC_PR7
Access Type:
31
–
23
0
15
Read-only
30
–
22
14
29
–
21
28
–
20
13
12
NPARITY7
7
6
5
WORDADDR7
4
27
–
19
NPARITY7
11
0
3
26
–
18
25
–
17
24
–
16
10
9
WORDADDR7
1
BITADDR7
8
2
0
Once the entire main area of a page is written with data, the register content must be stored at any free location of the
spare area.
• BITADDR7: corrupted Bit Address in the page between the 1792nd and the 2047th bytes
During a page read, this value contains the corrupted bit offset where an error occurred, if a single error was detected. If
multiple errors were detected, this value is meaningless.
• WORDADDR7: corrupted Word Address in the page between the 1792nd and the 2047th bytes
During a page read, this value contains the word address (8-bit word) where an error occurred, if a single error was
detected. If multiple errors were detected, this value is meaningless.
• NPARITY7:
Parity N
308
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.7.9
ECC Parity Register 8
Register Name:
ECC_PR8
Access Type:
31
–
23
0
15
Read-only
30
–
22
14
29
–
21
28
–
20
13
12
NPARITY8
7
6
5
WORDADDR8
4
27
–
19
NPARITY8
11
0
3
26
–
18
25
–
17
24
–
16
10
9
WORDADDR8
1
BITADDR8
8
2
0
Once the entire main area of a page is written with data, the register content must be stored at any free location of the
spare area.
• BITADDR8: corrupted Bit Address in the page between the 2048th and the2303rd bytes
During a page read, this value contains the corrupted bit offset where an error occurred, if a single error was detected. If
multiple errors were detected, this value is meaningless.
• WORDADDR8: corrupted Word Address in the page between the 2048th and the 2303rd bytes
During a page read, this value contains the word address (8-bit word) where an error occurred, if a single error was
detected. If multiple errors were detected, this value is meaningless.
• NPARITY8:
Parity N.
309
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.7.10 ECC Parity Register 9
Register Name:
ECC_PR9
Access Type:
31
–
23
0
15
Read-only
30
–
22
14
29
–
21
28
–
20
13
12
NPARITY9
7
6
5
WORDADDR9
4
27
–
19
NPARITY9
11
0
3
26
–
18
25
–
17
24
–
16
10
9
WORDADDR9
1
BITADDR9
8
2
0
Once the entire main area of a page is written with data, the register content must be stored at any free location of the
spare area
• BITADDR9: corrupted bit address in the page between the 2304th and the 2559th bytes
During a page read, this value contains the corrupted bit offset where an error occurred, if a single error was detected. If
multiple errors were detected, this value is meaningless.
• WORDADDR9: corrupted word address in the page between the 2304th and the 2559th bytes
During a page read, this value contains the word address (8-bit word) where an error occurred, if a single error was
detected. If multiple errors were detected, this value is meaningless
• NPARITY9
Parity N
310
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.7.11 ECC Parity Register 10
Register Name:
ECC_PR10
Access Type:
31
–
23
0
15
7
Read-only
30
–
22
29
–
21
14
13
NPARITY10
6
5
WORDADDR10
28
–
20
12
4
27
–
19
NPARITY10
11
0
3
26
–
18
25
–
17
24
–
16
10
9
WORDADDR10
1
BITADDR10
8
2
0
Once the entire main area of a page is written with data, the register content must be stored at any free location of the
spare area.
• BITADDR10: corrupted Bit Address in the page between the 2560th and the2815th bytes
During a page read, this value contains the corrupted bit offset where an error occurred, if a single error was detected. If
multiple errors were detected, this value is meaningless.
• WORDADDR10: corrupted Word Address in the page between the 2560th and the 2815th bytes
During a page read, this value contains the word address (8-bit word) where an error occurred, if a single error was
detected. If multiple errors were detected, this value is meaningless.
• NPARITY10:
Parity N
311
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.7.12 ECC Parity Register 11
Register Name:
ECC_PR11
Access Type:
31
–
23
0
15
7
Read-only
30
–
22
29
–
21
14
13
NPARITY11
6
5
WORDADDR11
28
–
20
12
4
27
–
19
NPARITY11
11
0
3
26
–
18
25
–
17
24
–
16
10
9
WORDADDR11
1
BITADDR11
8
2
0
Once the entire main area of a page is written with data, the register content must be stored at any free location of the
spare area.
• BITADDR11: corrupted Bit Address in the page between the 2816th and the 3071st bytes
During a page read, this value contains the corrupted bit offset where an error occurred, if a single error was detected. If
multiple errors were detected, this value is meaningless.
• WORDADDR11: corrupted Word Address in the page between the 2816th and the 3071st bytes
During a page read, this value contains the word address (8-bit word) where an error occurred, if a single error was
detected. If multiple errors were detected, this value is meaningless.
• NPARITY11:
Parity N
312
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.7.13 ECC Parity Register 12
Register Name:
ECC_PR12
Access Type:
31
–
23
0
15
7
Read-only
30
–
22
29
–
21
14
13
NPARITY12
6
5
WORDADDR12
28
–
20
12
4
27
–
19
NPARITY12
11
0
3
26
–
18
25
–
17
24
–
16
10
9
WORDADDR12
1
BITADDR12
8
2
0
Once the entire main area of a page is written with data, the register content must be stored at any free location of the
spare area.
• BITADDR12; corrupted Bit Address in the page between the 3072nd and the 3327th bytes
During a page read, this value contains the corrupted bit offset where an error occurred, if a single error was detected. If
multiple errors were detected, this value is meaningless.
• WORDADDR12: corrupted Word Address in the page between the 3072nd and the 3327th bytes
During a page read, this value contains the word address (8-bit word) where an error occurred, if a single error was
detected. If multiple errors were detected, this value is meaningless.
• NPARITY12:
Parity N
313
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.7.14 ECC Parity Register 13
Register Name:
ECC_PR13
Access Type:
31
–
23
0
15
7
Read-only
30
–
22
29
–
21
14
13
NPARITY13
6
5
WORDADDR13
28
–
20
12
4
27
–
19
NPARITY13
11
0
3
26
–
18
25
–
17
24
–
16
10
9
WORDADDR13
1
BITADDR13
8
2
0
Once the entire main area of a page is written with data, the register content must be stored at any free location of the
spare area.
• BITADDR13: corrupted Bit Address in the page between the 3328th and the 3583rd bytes
During a page read, this value contains the corrupted bit offset where an error occurred, if a single error was detected. If
multiple errors were detected, this value is meaningless.
• WORDADDR13: corrupted Word Address in the page between the 3328th and the 3583rd bytes
During a page read, this value contains the word address (8-bit word) where an error occurred, if a single error was
detected. If multiple errors were detected, this value is meaningless.
• NPARITY13:
Parity N
314
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.7.15 ECC Parity Register 14
Register Name:
ECC_PR14
Access Type:
31
–
23
0
15
7
Read-only
30
–
22
29
–
21
14
13
NPARITY14
6
5
WORDADDR14
28
–
20
12
4
27
–
19
NPARITY14
11
0
3
26
–
18
25
–
17
24
–
16
10
9
WORDADDR14
1
BITADDR14
8
2
0
Once the entire main area of a page is written with data, the register content must be stored at any free location of the
spare area.
• BITADDR14: corrupted Bit Address in the page between the 3584th and the 3839th bytes
During a page read, this value contains the corrupted bit offset where an error occurred, if a single error was detected. If
multiple errors were detected, this value is meaningless.
• WORDADDR14: corrupted Word Address in the page between the 3584th and the 3839th bytes
During a page read, this value contains the word address (8-bit word) where an error occurred, if a single error was
detected. If multiple errors were detected, this value is meaningless.
• NPARITY14:
Parity N
315
6438D–ATARM–13-Oct-09
AT91SAM9G45
23.7.16 ECC Parity Register 15
Register Name:
ECC_PR15
Access Type:
31
–
23
0
15
7
Read-only
30
–
22
29
–
21
14
13
NPARITY15
6
5
WORDADDR15
28
–
20
12
4
27
–
19
NPARITY15
11
0
3
26
–
18
25
–
17
24
–
16
10
9
WORDADDR15
1
BITADDR15
8
2
0
Once the entire main area of a page is written with data, the register content must be stored at any free location of the
spare area
• BITADDR15: corrupted Bit Address in the page between the 3840th and the 4095th bytes
During a page read, this value contains the corrupted bit offset where an error occurred, if a single error was detected. If
multiple errors were detected, this value is meaningless.
• WORDADDR15: corrupted Word Address in the page between the 3840th and the 4095th bytes
During a page read, this value contains the word address (8-bit word) where an error occurred, if a single error was
detected. If multiple errors were detected, this value is meaningless.
• NPARITY15
Parity N
316
6438D–ATARM–13-Oct-09
AT91SAM9G45
24. Peripheral DMA Controller (PDC)
24.1
Description
The Peripheral DMA Controller (PDC) transfers data between on-chip serial peripherals and the
on- and/or off-chip memories. The link between the PDC and a serial peripheral is operated by
the AHB to ABP bridge.
The user interface of each PDC channel is integrated into the user interface of the peripheral it
serves. The user interface of mono directional channels (receive only or transmit only), contains
two 32-bit memory pointers and two 16-bit counters, one set (pointer, counter) for current transfer and one set (pointer, counter) for next transfer. The bi-directional channel user interface
contains four 32-bit memory pointers and four 16-bit counters. Each set (pointer, counter) is
used by current transmit, next transmit, current receive and next receive.
Using the PDC removes processor overhead by reducing its intervention during the transfer.
This significantly reduces the number of clock cycles required for a data transfer, which
improves microcontroller performance.
To launch a transfer, the peripheral triggers its associated PDC channels by using transmit and
receive signals. When the programmed data is transferred, an end of transfer interrupt is generated by the peripheral itself.
24.2
Embedded Characteristics
• Acting as one AHB Bus Matrix Master
• Allows data transfers from/to peripheral to/from any memory space without any intervention
of the processor.
• Next Pointer support, prevents strong real-time constraints on buffer management.
The Peripheral DMA Controller handles transfer requests from the channel according to the following priorities (Low to High priorities):
Table 24-1.
Peripheral DMA Controller
Instance name
Channel T/R
DBGU
Transmit
USART3
Transmit
USART2
Transmit
USART1
Transmit
USART0
Transmit
AC97
Transmit
SPI1
Transmit
SPI0
Transmit
SSC1
Transmit
SSC0
Transmit
TSDAC
Receive
DBGU
Receive
USART3
Receive
317
6438D–ATARM–13-Oct-09
Table 24-1.
318
Peripheral DMA Controller
Instance name
Channel T/R
USART2
Receive
USART1
Receive
USART0
Receive
AC97
Receive
SPI1
Receive
SPI0
Receive
SSC1
Receive
SSC0
Receive
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
24.3
Block Diagram
Figure 24-1. Block Diagram
FULL DUPLEX
PERIPHERAL
PDC
THR
PDC Channel A
RHR
PDC Channel B
Control
Status & Control
HALF DUPLEX
PERIPHERAL
Control
THR
PDC Channel C
RHR
Control
Status & Control
RECEIVE or TRANSMIT
PERIPHERAL
RHR or THR
Control
24.4
24.4.1
PDC Channel D
Status & Control
Functional Description
Configuration
The PDC channel user interface enables the user to configure and control data transfers for
each channel. The user interface of each PDC channel is integrated into the associated peripheral user interface.
The user interface of a serial peripheral, whether it is full or half duplex, contains four 32-bit
pointers (RPR, RNPR, TPR, TNPR) and four 16-bit counter registers (RCR, RNCR, TCR,
TNCR). However, the transmit and receive parts of each type are programmed differently: the
319
6438D–ATARM–13-Oct-09
transmit and receive parts of a full duplex peripheral can be programmed at the same time,
whereas only one part (transmit or receive) of a half duplex peripheral can be programmed at a
time.
32-bit pointers define the access location in memory for current and next transfer, whether it is
for read (transmit) or write (receive). 16-bit counters define the size of current and next transfers.
It is possible, at any moment, to read the number of transfers left for each channel.
The PDC has dedicated status registers which indicate if the transfer is enabled or disabled for
each channel. The status for each channel is located in the associated peripheral status register.
Transfers can be enabled and/or disabled by setting TXTEN/TXTDIS and RXTEN/RXTDIS in
the peripheral’s Transfer Control Register.
At the end of a transfer, the PDC channel sends status flags to its associated peripheral. These
flags are visible in the peripheral status register (ENDRX, ENDTX, RXBUFF, and TXBUFE).
Refer to Section 24.4.3 and to the associated peripheral user interface.
24.4.2
Memory Pointers
Each full duplex peripheral is connected to the PDC by a receive channel and a transmit channel. Both channels have 32-bit memory pointers that point respectively to a receive area and to
a transmit area in on- and/or off-chip memory.
Each half duplex peripheral is connected to the PDC by a bidirectional channel. This channel
has two 32-bit memory pointers, one for current transfer and the other for next transfer. These
pointers point to transmit or receive data depending on the operating mode of the peripheral.
Depending on the type of transfer (byte, half-word or word), the memory pointer is incremented
respectively by 1, 2 or 4 bytes.
If a memory pointer address changes in the middle of a transfer, the PDC channel continues
operating using the new address.
24.4.3
Transfer Counters
Each channel has two 16-bit counters, one for current transfer and the other one for next transfer. These counters define the size of data to be transferred by the channel. The current transfer
counter is decremented first as the data addressed by current memory pointer starts to be transferred. When the current transfer counter reaches zero, the channel checks its next transfer
counter. If the value of next counter is zero, the channel stops transferring data and sets the
appropriate flag. But if the next counter value is greater then zero, the values of the next
pointer/next counter are copied into the current pointer/current counter and the channel resumes
the transfer whereas next pointer/next counter get zero/zero as values. At the end of this transfer the PDC channel sets the appropriate flags in the Peripheral Status Register.
The following list gives an overview of how status register flags behave depending on the counters’ values:
• ENDRX flag is set when the PERIPH_RCR register reaches zero.
• RXBUFF flag is set when both PERIPH_RCR and PERIPH_RNCR reach zero.
• ENDTX flag is set when the PERIPH_TCR register reaches zero.
• TXBUFE flag is set when both PERIPH_TCR and PERIPH_TNCR reach zero.
These status flags are described in the Peripheral Status Register.
320
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
24.4.4
Data Transfers
The serial peripheral triggers its associated PDC channels’ transfers using transmit enable
(TXEN) and receive enable (RXEN) flags in the transfer control register integrated in the peripheral’s user interface.
When the peripheral receives an external data, it sends a Receive Ready signal to its PDC
receive channel which then requests access to the Matrix. When access is granted, the PDC
receive channel starts reading the peripheral Receive Holding Register (RHR). The read data
are stored in an internal buffer and then written to memory.
When the peripheral is about to send data, it sends a Transmit Ready to its PDC transmit channel which then requests access to the Matrix. When access is granted, the PDC transmit
channel reads data from memory and puts them to Transmit Holding Register (THR) of its associated peripheral. The same peripheral sends data according to its mechanism.
24.4.5
PDC Flags and Peripheral Status Register
Each peripheral connected to the PDC sends out receive ready and transmit ready flags and the
PDC sends back flags to the peripheral. All these flags are only visible in the Peripheral Status
Register.
Depending on the type of peripheral, half or full duplex, the flags belong to either one single
channel or two different channels.
24.4.5.1
Receive Transfer End
This flag is set when PERIPH_RCR register reaches zero and the last data has been transferred
to memory.
It is reset by writing a non zero value in PERIPH_RCR or PERIPH_RNCR.
24.4.5.2
Transmit Transfer End
This flag is set when PERIPH_TCR register reaches zero and the last data has been written into
peripheral THR.
It is reset by writing a non zero value in PERIPH_TCR or PERIPH_TNCR.
24.4.5.3
Receive Buffer Full
This flag is set when PERIPH_RCR register reaches zero with PERIPH_RNCR also set to zero
and the last data has been transferred to memory.
It is reset by writing a non zero value in PERIPH_TCR or PERIPH_TNCR.
24.4.5.4
Transmit Buffer Empty
This flag is set when PERIPH_TCR register reaches zero with PERIPH_TNCR also set to zero
and the last data has been written into peripheral THR.
It is reset by writing a non zero value in PERIPH_TCR or PERIPH_TNCR.
321
6438D–ATARM–13-Oct-09
24.5
Peripheral DMA Controller (PDC) User Interface
Table 24-2.
Offset
Register Mapping
Register
Name
(1)
Access
Reset
0x100
Receive Pointer Register
PERIPH _RPR
Read-write
0
0x104
Receive Counter Register
PERIPH_RCR
Read-write
0
0x108
Transmit Pointer Register
PERIPH_TPR
Read-write
0
0x10C
Transmit Counter Register
PERIPH_TCR
Read-write
0
0x110
Receive Next Pointer Register
PERIPH_RNPR
Read-write
0
0x114
Receive Next Counter Register
PERIPH_RNCR
Read-write
0
0x118
Transmit Next Pointer Register
PERIPH_TNPR
Read-write
0
0x11C
Transmit Next Counter Register
PERIPH_TNCR
Read-write
0
0x120
Transfer Control Register
PERIPH_PTCR
Write-only
0
0x124
Transfer Status Register
PERIPH_PTSR
Read-only
0
Note:
322
1. PERIPH: Ten registers are mapped in the peripheral memory space at the same offset. These can be defined by the user
according to the function and the desired peripheral.)
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
24.5.1
Name:
Receive Pointer Register
PERIPH_RPR
Access:
31
Read-write
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
0
RXPTR
23
22
21
20
RXPTR
15
14
13
12
RXPTR
7
6
5
4
RXPTR
• RXPTR: Receive Pointer Register
RXPTR must be set to receive buffer address.
When a half duplex peripheral is connected to the PDC, RXPTR = TXPTR.
323
6438D–ATARM–13-Oct-09
24.5.2
Name:
Receive Counter Register
PERIPH_RCR
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
14
13
12
11
10
9
8
3
2
1
0
RXCTR
7
6
5
4
RXCTR
• RXCTR: Receive Counter Register
RXCTR must be set to receive buffer size.
When a half duplex peripheral is connected to the PDC, RXCTR = TXCTR.
0 = Stops peripheral data transfer to the receiver
1 - 65535 = Starts peripheral data transfer if corresponding channel is active
324
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
24.5.3
Name:
Transmit Pointer Register
PERIPH_TPR
Access:
31
Read-write
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
0
TXPTR
23
22
21
20
TXPTR
15
14
13
12
TXPTR
7
6
5
4
TXPTR
• TXPTR: Transmit Counter Register
TXPTR must be set to transmit buffer address.
When a half duplex peripheral is connected to the PDC, RXPTR = TXPTR.
325
6438D–ATARM–13-Oct-09
AT91SAM9G45
24.5.4
Name:
Transmit Counter Register
PERIPH_TCR
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
14
13
12
11
10
9
8
3
2
1
0
TXCTR
7
6
5
4
TXCTR
• TXCTR: Transmit Counter Register
TXCTR must be set to transmit buffer size.
When a half duplex peripheral is connected to the PDC, RXCTR = TXCTR.
0 = Stops peripheral data transfer to the transmitter
1- 65535 = Starts peripheral data transfer if corresponding channel is active
326
6438D–ATARM–13-Oct-09
AT91SAM9G45
24.5.5
Name:
Receive Next Pointer Register
PERIPH_RNPR
Access:
31
Read-write
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
0
RXNPTR
23
22
21
20
RXNPTR
15
14
13
12
RXNPTR
7
6
5
4
RXNPTR
• RXNPTR: Receive Next Pointer
RXNPTR contains next receive buffer address.
When a half duplex peripheral is connected to the PDC, RXNPTR = TXNPTR.
327
6438D–ATARM–13-Oct-09
AT91SAM9G45
24.5.6
Name:
Receive Next Counter Register
PERIPH_RNCR
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
14
13
12
11
10
9
8
3
2
1
0
RXNCTR
7
6
5
4
RXNCTR
• RXNCTR: Receive Next Counter
RXNCTR contains next receive buffer size.
When a half duplex peripheral is connected to the PDC, RXNCTR = TXNCTR.
328
6438D–ATARM–13-Oct-09
AT91SAM9G45
24.5.7
Name:
Transmit Next Pointer Register
PERIPH_TNPR
Access:
31
Read-write
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
0
TXNPTR
23
22
21
20
TXNPTR
15
14
13
12
TXNPTR
7
6
5
4
TXNPTR
• TXNPTR: Transmit Next Pointer
TXNPTR contains next transmit buffer address.
When a half duplex peripheral is connected to the PDC, RXNPTR = TXNPTR.
329
6438D–ATARM–13-Oct-09
24.5.8
Name:
Transmit Next Counter Register
PERIPH_TNCR
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
14
13
12
11
10
9
8
3
2
1
0
TXNCTR
7
6
5
4
TXNCTR
• TXNCTR: Transmit Counter Next
TXNCTR contains next transmit buffer size.
When a half duplex peripheral is connected to the PDC, RXNCTR = TXNCTR.
330
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
24.5.9
Name:
Transfer Control Register
PERIPH_PTCR
Access:
Write-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
TXTDIS
8
TXTEN
7
–
6
–
5
–
4
–
3
–
2
–
1
RXTDIS
0
RXTEN
• RXTEN: Receiver Transfer Enable
0 = No effect.
1 = Enables PDC receiver channel requests if RXTDIS is not set.
When a half duplex peripheral is connected to the PDC, enabling the receiver channel requests automatically disables the
transmitter channel requests. It is forbidden to set both TXTEN and RXTEN for a half duplex peripheral.
• RXTDIS: Receiver Transfer Disable
0 = No effect.
1 = Disables the PDC receiver channel requests.
When a half duplex peripheral is connected to the PDC, disabling the receiver channel requests also disables the transmitter channel requests.
• TXTEN: Transmitter Transfer Enable
0 = No effect.
1 = Enables the PDC transmitter channel requests.
When a half duplex peripheral is connected to the PDC, it enables the transmitter channel requests only if RXTEN is not
set. It is forbidden to set both TXTEN and RXTEN for a half duplex peripheral.
• TXTDIS: Transmitter Transfer Disable
0 = No effect.
1 = Disables the PDC transmitter channel requests.
When a half duplex peripheral is connected to the PDC, disabling the transmitter channel requests disables the receiver
channel requests.
331
6438D–ATARM–13-Oct-09
24.5.10
Name:
Transfer Status Register
PERIPH_PTSR
Access:
Read-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
TXTEN
7
–
6
–
5
–
4
–
3
–
2
–
1
–
0
RXTEN
• RXTEN: Receiver Transfer Enable
0 = PDC Receiver channel requests are disabled.
1 = PDC Receiver channel requests are enabled.
• TXTEN: Transmitter Transfer Enable
0 = PDC Transmitter channel requests are disabled.
1 = PDC Transmitter channel requests are enabled.
332
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
25. Clock Generator
25.1
Description
The Clock Generator User Interface is embedded within the Power Management Controller
Interface and is described in Section 26.11. However, the Clock Generator registers are named
CKGR_.
25.2
Embedded Characteristics
The Clock Generator is made up of:
• One Low Power 32768 Hz Slow Clock Oscillator with bypass mode
• One Low-Power RC oscillator
• One 12 MHz Main Oscillator, which can be bypassed
• One 400 to 800 MHz programmable PLLA, capable to provide the clock MCK to the
processor and to the peripherals. This PLL has an input divider to offer a wider range of
output frequencies from the 12 MHz input, the only limitation being the lowest input frequency
shall be higher or equal to 2 MHz.
The USB Device and Host HS Clocks are provided by a the dedicated UTMI PLL (UPLL)
embedded in the UTMI macro.
Figure 25-1. Clock Generator Block Diagram
Clock Generator
RCEN
On Chip
RC OSC
XIN32
XOUT32
Slow Clock
SLCK
Slow Clock
Oscillator
OSCSEL
OSC32EN
OSC32BYP
XIN
12M Main
Oscillator
Main Clock
MAINCK
XOUT
UPLL
UPLLCK
PLLA and
Divider
Status
PLLA Clock
PLLACK
Control
Power
Management
Controller
25.3
Slow Clock Crystal Oscillator
The Clock Generator integrates a 32,768 Hz low-power oscillator. The XIN32 and XOUT32 pins
must be connected to a 32,768 Hz crystal. Two external capacitors must be wired as shown in
Figure 25-2.
333
6438D–ATARM–13-Oct-09
Figure 25-2. Typical Slow Clock Crystal Oscillator Connection
XIN32
XOUT32
GNDPLL
32,768 Hz
Crystal
25.4
Slow Clock RC Oscillator
The user has to take into account the possible drifts of the RC Oscillator. More details are given
in the section “DC Characteristics” of the product datasheet.
25.5
Slow Clock Selection
The AT91SAM9G45 slow clock can be generated either by an external 32,768 Hz crystal or by
the on-chip RC oscillator. The 32,768 Hz crystal oscillator can be bypassed by setting the bit
OSC32BYP to accept an external slow clock on XIN32.
The internal RC oscillator and the 32,768 Hz oscillator can be enabled by setting to 1, respectively, RCEN bit and OSC32EN bit in the System Controller user interface. The OSCSEL
command selects the slow clock source.
Figure 25-3. Slow Clock Selection
Clock Generator
RCEN
On Chip
RC OSC
Slow Clock
SLCK
XIN32
XOUT32
Slow Clock
Oscillator
OSCSEL
OSC32EN
OSC32BYP
RCEN, OSC32EN,OSCSEL and OSC32BYP bits are located in the Slow Clock Control Register
(SCKCR) located at address 0xFFFFFD50 in the backed up part of the System Controller and so
are preserved while VDDBU is present.
After a VDDBU power on reset, the default configuration is RCEN=1, OSC32EN=0 and OSCSEL=0, allowing the system to start on the internal RC oscillator.
The programmer controls the slow clock switching by software and so must take precautions
during the switching phase.
334
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
25.5.1
Switch from Internal RC Oscillator to the 32768 Hz Crystal
To switch from internal RC oscillator to the 32768 Hz crystal, the programmer must execute the
following sequence:
• Switch the master clock to a source different from slow clock (PLLA or PLLB or Main
Oscillator) through the Power management Controller.
• Enable the 32768 Hz oscillator by setting the bit OSCEN to 1.
• Wait 32768 Hz Startup Time for clock stabilization (software loop)
• Switch from internal RC to 32768Hz by setting the bit OSCSEL to 1.
• Wait 5 slow clock cycles for internal resynchronization
• Disable the RC oscillator by setting the bit RCEN to 0.
25.5.2
Bypass the 32768 Hz Oscillator
The following step must be added to bypass the 32768Hz Oscillator.
• An external clock must be connected on XIN32.
• Enable the bypass path OSC32BYP bit set to 1.
• Disable the 32768 Hz oscillator by setting the bit OSC32EN to 0.
25.5.3
Switch from 32768 Hz Crystal to the Internal RC oscillator
The same procedure must be followed to switch from 32768Hz crystal to the internal RC
oscillator.
• Switch the master clock to a source different from slow clock (PLLA or PLLB or Main
Oscillator)
• Enable the internal RC oscillator by setting the bit RCEN to 1.
• Wait internal RC Startup Time for clock stabilization (software loop)
• Switch from 32768Hz oscillator to internal RC by setting the bit OSCSEL to 0
• Wait 5 slow clock cycles for internal resynchronization
• Disable the 32768Hz oscillator by setting the bit OSC32EN to 0
25.5.4
Slow Clock Configuration Register
Register Name:SCKCR
Address:
0xFFFFFD50
Access Type:Read/Write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
–
6
–
5
–
4
–
3
OSCSEL
2
OSC32BYP
1
OSC32EN
0
RCEN
335
6438D–ATARM–13-Oct-09
• RCEN: Internal RC
0: RC is disabled
1: RC is enabled
• OSC32EN: 32768 Hz oscillator
0: 32768Hz oscillator is disabled
1: 32768Hz oscillator is enabled
• OSC32BYP: 32768Hz oscillator bypass
0: 32768Hz oscillator is not bypassed
1: 32768Hz oscillator is bypassed, accept an external slow clock on XIN32
• OSCSEL: Slow clock selector
0: Slow clock is internal RC
1: Slow clock is 32768 Hz oscillator
336
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
25.6
Main Oscillator
The Main Oscillator is designed for a 12 MHz fundamental crystal. The 12 MHz is an input of the
PLLA and the UPLL used to generate the 480 MHz USB High Speed Clock (UPLLCK).
Figure 25-4 shows the Main Oscillator block diagram.
Figure 25-4. Main Oscillator Block Diagram
XIN
12M Main
Oscillator
Main Clock
MAINCK
XOUT
UPLL
PLLA and
Divider
25.6.1
UPLLCK
PLLA Clock
PLLACK
Main Oscillator Connections
The typical crystal connection is illustrated in Figure 25-5. For further details on the electrical
characteristics of the Main Oscillator, see the section “DC Characteristics” of the product
datasheet.
Figure 25-5. Typical Crystal Connection
XIN
XOUT
GND
25.6.2
Main Oscillator Startup Time
The startup time of the 12 MHz Main Oscillator is given in the section “DC Characteristics” of the
product datasheet.
25.6.3
Main Oscillator Control
To minimize the power required to start up the system, the main oscillator is disabled after reset
and slow clock is selected.
The software enables or disables the main oscillator so as to reduce power consumption by
clearing the MOSCEN bit in the Main Oscillator Register (CKGR_MOR).
When disabling the main oscillator by clearing the MOSCEN bit in CKGR_MOR, the MOSCS bit
in PMC_SR is automatically cleared, indicating the main clock is off.
When enabling the main oscillator, the user must initiate the main oscillator counter with a value
corresponding to the startup time of the oscillator. This startup time depends on the crystal frequency connected to the main oscillator.
337
6438D–ATARM–13-Oct-09
When the MOSCEN bit and the OSCOUNT are written in CKGR_MOR to enable the main oscillator, the MOSCS bit in PMC_SR (Status Register) is cleared and the counter starts counting
down on the slow clock divided by 8 from the OSCOUNT value. Since the OSCOUNT value is
coded with 8 bits, the maximum startup time is about 62 ms.
When the counter reaches 0, the MOSCS bit is set, indicating that the main clock is valid. Setting the MOSCS bit in PMC_IMR can trigger an interrupt to the processor.
25.6.4
25.7
Main Oscillator Bypass
The user can input a clock on the device instead of connecting a crystal. In this case, the user
has to provide the external clock signal on the XIN pin. The input characteristics of the XIN pin
under these conditions are given in the product electrical characteristics section. The programmer has to be sure to set the OSCBYPASS bit to 1 and the MOSCEN bit to 0 in the Main OSC
register (CKGR_MOR) for the external clock to operate properly.
Divider and PLLA Block
The PLLA embeds an input divider to increase the accuracy of the resulting clock signals. However, the user must respect the PLLA minimum input frequency when programming the divider.
The PLLA embeds also an output divisor by 2.
Figure 25-6 shows the block diagram of the divider and PLLA block.
Figure 25-6. Divider and PLLA Block Diagram
DIVA
MULA
Divider
MAINCK
OUTA
PLLADIV2
/1 or /2
Divider
PLLA
PLLACK
PLLACOUNT
SLCK
25.7.1
PLLA
Counter
LOCKA
Divider and Phase Lock Loop Programming
The divider can be set between 1 and 255 in steps of 1. When a divider field (DIV) is set to 0, the
output of the corresponding divider and the PLL output is a continuous signal at level 0. On
reset, each DIV field is set to 0, thus the corresponding PLL input clock is set to 0.
The PLLA allows multiplication of the divider’s outputs. The PLLA clock signal has a frequency
that depends on the respective source signal frequency and on the parameters DIVA and
MULA. The factor applied to the source signal frequency is (MULA + 1)/DIVA. When MULA is
written to 0, the PLLA is disabled and its power consumption is saved. Re-enabling the PLLA
can be performed by writing a value higher than 0 in the MUL field.
Whenever the PLLA is re-enabled or one of its parameters is changed, the LOCKA bit in
PMC_SR is automatically cleared. The values written in the PLLACOUNT field in CKGR_PLLAR
are loaded in the PLLA counter. The PLLA counter then decrements at the speed of the Slow
338
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Clock until it reaches 0. At this time, the LOCK bit is set in PMC_SR and can trigger an interrupt
to the processor. The user has to load the number of Slow Clock cycles required to cover the
PLLA transient time into the PLLACOUNT field.
The PLLA clock can be divided by 2 by writing the PLLADIV2 bit in PMC_MCKR register.
25.8
UTMI Bias and Phase Lock Loop Programming
The multiplier is built-in to 40 to obtain the USB High Speed 480 MHz.
UPLLEN
MAINCK
UPLL
UPLLCK
PLLCOUNT
SLCK
UPLL
Counter
LOCKU
Whenever the UPLL is enabled by writing UPLLEN in CKGR_UCKR, the LOCKU bit in PMC_SR
is automatically cleared. The values written in the PLLCOUNT field in CKGR_UCKR are loaded
in the UPLL counter. The UPLL counter then decrements at the speed of the Slow Clock divided
by 8 until it reaches 0. At this time, the LOCKU bit is set in PMC_SR and can trigger an interrupt
to the processor. The user has to load the number of Slow Clock cycles required to cover the
UPLL transient time into the PLLCOUNT field. The BIAS, needed for High Speed operations, is
enabled by writing BIASEN in CKGR_UCKR once the PLL locked.
339
6438D–ATARM–13-Oct-09
26. Power Management Controller (PMC)
26.1
Description
The Power Management Controller (PMC) optimizes power consumption by controlling all system and user peripheral clocks. The PMC enables/disables the clock inputs to many of the
peripherals and the ARM Processor.
26.2
Embedded Characteristics
The Power Management Controller provides all the clock signals to the system.
PMC input clocks:
• UPLLCK: From UTMI PLL
• PLLACK From PLLA
• SLCK: slow clock from OSC32K or internal RC OSC
• MAINCK: from 12 MHz external oscillator
PMC output clocks
• Processor Clock PCK
• Master Clock MCK, in particular to the Matrix and the memory interfaces. The divider can be
1,2,3 or 4
• DDR system clock equal to 2xMCK
Note:
DDR system clock is not available when Master Clock (MCK) equals Processor Clock (PCK).
• USB Host EHCI High speed clock (UPLLCK)
• USB OHCI clocks (UHP48M and UHP12M)
• Independent peripheral clocks, typically at the frequency of MCK
• Two programmable clock outputs: PCK0 and PCK1
This allows the software control of five flexible operating modes:
• Normal Mode, processor and peripherals running at a programmable frequency
• Idle Mode, processor stopped waiting for an interrupt
• Slow Clock Mode, processor and peripherals running at low frequency
• Standby Mode, mix of Idle and Backup Mode, peripheral running at low frequency, processor
stopped waiting for an interrupt
• Backup Mode, Main Power Supplies off, VDDBU powered by a battery
340
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 26-1. AT91SAM9G45 Power Management Controller Block Diagram
PLLACK
USBS
UHP48M
USBDIV+1
USB
OHCI
UHP12M
/4
USB
EHCI
/1,/2
PCK
Processor
Clock
Controller
UPLLCK
int
Divider
MAINCK
SLCK
Prescaler
/1,/2,/4,.../64
X /1 /1.5 /2
SysClk DDR
/1 /2
MCK
/3 /4
Peripherals
Clock Controller
ON/OFF
Master Clock Controller
SLCK
MAINCK
periph_clk[..]
ON/OFF
Prescaler
/1,/2,/4,...,/64
pck[..]
UPLLCK
Programmable Clock Controller
26.2.1
26.2.1.1
Main Application Modes
The Power Management Controller provides 3 main application modes.
Normal Mode
• PLLA and UPLL are running respectively at 400 MHz and 480 MHz
• USB Device High Speed and Host EHCI High Speed operations are allowed
• Full Speed OHCI input clock is UPLLCK, USBDIV is 9 (division by 10)
• System Input clock is PLLACK, PCK is 400 MHz
• MDIV is ‘11’, MCK is 133 MHz
• DDR2 can be used at up to 133 MHz
26.2.1.2
USB HS and LP-DDR Mode
• Only UPLL is running at 480 MHz, PLLA power consumption is saved
• USB Device High Speed and Host EHCI High Speed operations are allowed
• Full Speed OHCI input clock is UPLLCK, USBDIV is 9 (division by 10)
• System Input clock is UPLLCK, Prescaler is 2, PCK is 240 MHz
• MDIV is ‘01’, MCK is 120 MHz
• Only LP-DDR can be used at up to 120 MHz
341
6438D–ATARM–13-Oct-09
26.2.1.3
No UDP HS, UHP FS and DDR2 Mode
• Only PLLA is running at 384 MHz, UPLL power consumption is saved
• USB Device High Speed and Host EHCI High Speed operations are NOT allowed
• Full Speed OHCI input clock is PLLACK, USBDIV is 7 (division by 8)
• System Input clock is PLLACK, PCK is 384 MHz
• MDIV is ‘11’, MCK is 128 MHz
• DDR2 can be used at up to 128 MHz
26.3
Master Clock Controller
The Master Clock Controller provides selection and division of the Master Clock (MCK). MCK is
the clock provided to all the peripherals and the memory controller.
The Master Clock is selected from one of the clocks provided by the Clock Generator. Selecting
the Slow Clock provides a Slow Clock signal to the whole device. Selecting the Main Clock
saves power consumption of the PLLA.
The Master Clock Controller is made up of a clock selector and a prescaler. It also contains a
Master Clock divider which allows the processor clock to be faster than the Master Clock.
The Master Clock selection is made by writing the CSS field (Clock Source Selection) in
PMC_MCKR (Master Clock Register). The prescaler supports the division by a power of 2 of the
selected clock between 1 and 64. The PRES field in PMC_MCKR programs the prescaler. The
Master Clock divider can be programmed through the MDIV field in PMC_MCKR.
Each time PMC_MCKR is written to define a new Master Clock, the MCKRDY bit is cleared in
PMC_SR. It reads 0 until the Master Clock is established. Then, the MCKRDY bit is set and can
trigger an interrupt to the processor. This feature is useful when switching from a high-speed
clock to a lower one to inform the software when the change is actually done.
Figure 26-2. Master Clock Controller
PMC_MCKR
CSS
PMC_MCKR
PRES
PMC_MCKR
MDIV
SLCK
MAINCK
PLLACK
Master Clock
Prescaler
Master
Clock
Divider
MCK
UPLLCK
Processor
Clock
Divider
26.4
To the Processor
Clock Controller (PCK)
Processor Clock Controller
The PMC features a Processor Clock Controller (PCK) that implements the Processor Idle
Mode. The Processor Clock can be disabled by writing the System Clock Disable Register
(PMC_SCDR). The status of this clock (at least for debug purpose) can be read in the System
Clock Status Register (PMC_SCSR).
342
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
The Processor Clock PCK is enabled after a reset and is automatically re-enabled by any
enabled interrupt. The Processor Idle Mode is achieved by disabling the Processor Clock, which
is automatically re-enabled by any enabled fast or normal interrupt, or by the reset of the
product.
When the Processor Clock is disabled, the current instruction is finished before the clock is
stopped, but this does not prevent data transfers from other masters of the system bus.
26.5
USB Device and Host clocks
The USB Device and Host High Speed ports clocks are controlled by the UDPHS and UHPHS
bits in PMC_PCER. To save power on this peripheral when they are is not used, the user can
set these bits in PMC_PCDR. The UDPHS and UHPHS bits PMC_PCSR gives the activity of
these clocks.
The PMC also provides the clocks UHP48M and UHP12M to the USB Host OHCI. The USB
Host OHCI clocks are controlled by the UHP bit in PMC_SCER. To save power on this peripheral when it is not used, the user can set the UHP bit in PMC_SCDR. The UHP bit in
PMC_SCSR gives the activity of this clock. The USB host OHCI requires both the 12/48 MHz
signal and the Master Clock. USBDIV field in PMC_USB register is to be programmed to 9 (division by 10) for normal operations.
To save more power consumption user can stop UTMI PLL, in this case USB high-speed operations are not possible. Nevertheless, as the USB OHCI Input clock can be selected with USBS
bit (PLLA or UTMI PLL) in PMC_USB register, OHCI full-speed operation remain possible.
The user must program the USB OHCI Input Clock and the USBDIV divider in PMC_USB register to generate a 48 MHz and a 12 MHz signal with an accuracy of ± 0.25%.
26.6
LP-DDR/DDR2 Clock
The Power Management Controller controls the clocks of the DDR memory. It provides SysClk
DDR internal clock. That clock is used by the DDR Controller to provide DDR control, data and
DDR clock signals.
The DDR clock can be enabled and disabled with DDRCK bit respectively in PMC_SCER and
PMC_SDER registers. At reset DDR clock is disabled to save power consumption.
The Input clock is the same as Master Clock. The Output SysClk DDR Clock is 2xMCK.
In the case MDIV = ‘00’, PCK = MCK and SysClk DDR and DDRCK clocks are not available.
If Input clock is PLLACK/PLLADIV2 the DDR Controller can drive DDR2 and LP-DDR at up to
133MHz with MDIV = ‘11’.
To save PLLA power consumption, the user can choose UPLLCK an Input clock for the system.
In this case the DDR Controller can drive LD-DDR at up to 120MHz.
26.7
Peripheral Clock Controller
The Power Management Controller controls the clocks of each embedded peripheral by the way
of the Peripheral Clock Controller. The user can individually enable and disable the Master
Clock on the peripherals by writing into the Peripheral Clock Enable (PMC_PCER) and Periph-
343
6438D–ATARM–13-Oct-09
eral Clock Disable (PMC_PCDR) registers. The status of the peripheral clock activity can be
read in the Peripheral Clock Status Register (PMC_PCSR).
When a peripheral clock is disabled, the clock is immediately stopped. The peripheral clocks are
automatically disabled after a reset.
In order to stop a peripheral, it is recommended that the system software wait until the peripheral
has executed its last programmed operation before disabling the clock. This is to avoid data corruption or erroneous behavior of the system.
The bit number within the Peripheral Clock Control registers (PMC_PCER, PMC_PCDR, and
PMC_PCSR) is the Peripheral Identifier defined at the product level. Generally, the bit number
corresponds to the interrupt source number assigned to the peripheral.
26.8
Programmable Clock Output Controller
The PMC controls 2 signals to be output on external pins PCKx. Each signal can be independently programmed via the PMC_PCKx registers.
PCKx can be independently selected between the Slow clock, the Master Clock, the
PLLACK/PLLADIV2, the UTMI PLL output and the main clock by writing the CSS and CSSMCK
fields in PMC_PCKx. Each output signal can also be divided by a power of 2 between 1 and 64
by writing the PRES (Prescaler) field in PMC_PCKx.
Each output signal can be enabled and disabled by writing 1 in the corresponding bit, PCKx of
PMC_SCER and PMC_SCDR, respectively. Status of the active programmable output clocks
are given in the PCKx bits of PMC_SCSR (System Clock Status Register).
Moreover, like the PCK, a status bit in PMC_SR indicates that the Programmable Clock is actually what has been programmed in the Programmable Clock registers.
As the Programmable Clock Controller does not manage with glitch prevention when switching
clocks, it is strongly recommended to disable the Programmable Clock before any configuration
change and to re-enable it after the change is actually performed.
26.9
Programming Sequence
1. Enabling the 12MHz Main Oscillator:
The main oscillator is enabled by setting the MOSCEN field in the CKGR_MOR register. In
some cases it may be advantageous to define a start-up time. This can be achieved by writing a value in the OSCOUNT field in the CKGR_MOR register.
Once this register has been correctly configured, the user must wait for MOSCS field in the
PMC_SR register to be set. This can be done either by polling the status register or by waiting the interrupt line to be raised if the associated interrupt to MOSCS has been enabled in
the PMC_IER register.
2. Setting PLLA and divider:
All parameters needed to configure PLLA and the divider are located in the CKGR_PLLAR
register.
The DIVA field is used to control divider itself. A value between 0 and 255 can be programmed. Divider output is divider input divided by DIVA parameter. By default DIVA
parameter is set to 0 which means that divider is turned off.
344
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
The OUTA field is used to select the PLLA output frequency range.
The MULA field is the PLLA multiplier factor. This parameter can be programmed between 0
and 254. If MULA is set to 0, PLLA will be turned off, otherwise the PLLA output frequency is
PLLA input frequency multiplied by (MULA + 1).
The PLLACOUNT field specifies the number of slow clock cycles before LOCKA bit is set in
the PMC_SR register after CKGR_PLLAR register has been written.
Once the PMC_PLLAR register has been written, the user must wait for the LOCKA bit to be
set in the PMC_SR register. This can be done either by polling the status register or by waiting the interrupt line to be raised if the associated interrupt to LOCKA has been enabled in
the PMC_IER register. All parameters in CKGR_PLLAR can be programmed in a single write
operation. If at some stage one of the following parameters, MULA, DIVA is modified,
LOCKA bit will go low to indicate that PLLA is not ready yet. When PLLA is locked, LOCKA
will be set again. The user is constrained to wait for LOCKA bit to be set before using the
PLLA output clock.
Code Example:
write_register(CKGR_PLLAR,0x00040805)
If PLLA and divider are enabled, the PLLA input clock is the main clock. PLLA output clock is
PLLA input clock multiplied by 5. Once CKGR_PLLAR has been written, LOCKA bit will be
set after eight slow clock cycles.
3. Setting Bias and High Speed PLL (UPLL) for UTMI
The UTMI PLL is enabled by setting the UPLLEN field in the CKGR_UCKR register. The
UTMI Bias must is enabled by setting the BIASEN field in the CKGR_UCKR register in the
same time. In some cases it may be advantageous to define a start-up time. This can be
achieved by writing a value in the PLLCOUNT field in the CKGR_UCKR register.
Once this register has been correctly configured, the user must wait for LOCKU field in the
PMC_SR register to be set. This can be done either by polling the status register or by waiting the interrupt line to be raised if the associated interrupt to LOCKU has been enabled in
the PMC_IER register.
4. Selection of Master Clock and Processor Clock
The Master Clock and the Processor Clock are configurable via the PMC_MCKR register.
The CSS field is used to select the clock source of the Master Clock and Processor Clock
dividers. By default, the selected clock source is slow clock.
The PRES field is used to control the Master/Processor Clock prescaler. The user can
choose between different values (1, 2, 4, 8, 16, 32, 64). Prescaler output is the selected clock
source divided by PRES parameter. By default, PRES parameter is set to 1 which means
that the input clock of the Master Clock and Processor Clock dividers is equal to slow clock.
The MDIV field is used to control the Master Clock divider. It is possible to choose between
different values (0, 1, 2, 3). The Master Clock output is Master/Processor Clock Prescaler
output divided by 1, 2, 4 or 3, depending on the value programmed in MDIV.
345
6438D–ATARM–13-Oct-09
The PLLADIV2 field is used to control the PLLA Clock divider. It is possible to choose
between different values (0, 1). The PMC PLLA Clock input is divided by 1 or 2, depending
on the value programmed in PLLADIV2.
By default, MDIV and PLLLADIV2 are set to 0, which indicates that Processor Clock is equal
to the Master Clock.
Once the PMC_MCKR register has been written, the user must wait for the MCKRDY bit to
be set in the PMC_SR register. This can be done either by polling the status register or by
waiting for the interrupt line to be raised if the associated interrupt to MCKRDY has been
enabled in the PMC_IER register.
The PMC_MCKR register must not be programmed in a single write operation. The preferred
programming sequence for the PMC_MCKR register is as follows:
• If a new value for CSS field corresponds to PLLA Clock,
– Program the PRES field in the PMC_MCKR register.
– Wait for the MCKRDY bit to be set in the PMC_SR register.
– Program the CSS field in the PMC_MCKR register.
– Wait for the MCKRDY bit to be set in the PMC_SR register.
• If a new value for CSS field corresponds to Main Clock or Slow Clock,
– Program the CSS field in the PMC_MCKR register.
– Wait for the MCKRDY bit to be set in the PMC_SR register.
– Program the PRES field in the PMC_MCKR register.
– Wait for the MCKRDY bit to be set in the PMC_SR register.
If at some stage one of the following parameters, CSS or PRES, is modified, the MCKRDY
bit will go low to indicate that the Master Clock and the Processor Clock are not ready yet.
The user must wait for MCKRDY bit to be set again before using the Master and Processor
Clocks.
Note:
IF PLLA clock was selected as the Master Clock and the user decides to modify it by writing in
CKGR_PLLAR, the MCKRDY flag will go low while PLLA is unlocked. Once PLLA is locked again,
LOCK goes high and MCKRDY is set.
While PLLA is unlocked, the Master Clock selection is automatically changed to Main Clock. For
further information, see Section 26.10.2. “Clock Switching Waveforms” on page 349.
Code Example:
write_register(PMC_MCKR,0x00000001)
wait (MCKRDY=1)
write_register(PMC_MCKR,0x00000011)
wait (MCKRDY=1)
The Master Clock is main clock divided by 16.
The Processor Clock is the Master Clock.
5. Selection of Programmable clocks
Programmable clocks are controlled via registers; PMC_SCER, PMC_SCDR and
PMC_SCSR.
346
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Programmable clocks can be enabled and/or disabled via the PMC_SCER and PMC_SCDR
registers. Depending on the system used, 2 programmable clocks can be enabled or disabled. The PMC_SCSR provides a clear indication as to which Programmable clock is
enabled. By default all Programmable clocks are disabled.
PMC_PCKx registers are used to configure programmable clocks.
The CSS and CSSMCK fields are used to select the programmable clock divider source. Five
clock options are available: main clock, slow clock, master clock, PLLACK, UPLLCK. By
default, the clock source selected is slow clock.
The PRES field is used to control the programmable clock prescaler. It is possible to choose
between different values (1, 2, 4, 8, 16, 32, 64). Programmable clock output is prescaler input
divided by PRES parameter. By default, the PRES parameter is set to 1 which means that
master clock is equal to slow clock.
Once the PMC_PCKx register has been programmed, The corresponding programmable
clock must be enabled and the user is constrained to wait for the PCKRDYx bit to be set in
the PMC_SR register. This can be done either by polling the status register or by waiting the
interrupt line to be raised if the associated interrupt to PCKRDYx has been enabled in the
PMC_IER register. All parameters in PMC_PCKx can be programmed in a single write
operation.
If the CSS and PRES parameters are to be modified, the corresponding programmable clock
must be disabled first. The parameters can then be modified. Once this has been done, the
user must re-enable the programmable clock and wait for the PCKRDYx bit to be set.
Code Example:
write_register(PMC_PCK0,0x00000015)
Programmable clock 0 is main clock divided by 32.
6. Enabling Peripheral Clocks
Once all of the previous steps have been completed, the peripheral clocks can be enabled
and/or disabled via registers PMC_PCER and PMC_PCDR.
Depending on the system used, 19 peripheral clocks can be enabled or disabled. The
PMC_PCSR provides a clear view as to which peripheral clock is enabled.
Note:
Each enabled peripheral clock corresponds to Master Clock.
Code Examples:
write_register(PMC_PCER,0x00000110)
Peripheral clocks 4 and 8 are enabled.
write_register(PMC_PCDR,0x00000010)
347
6438D–ATARM–13-Oct-09
Peripheral clock 4 is disabled.
26.10 Clock Switching Details
26.10.1
Master Clock Switching Timings
Table 26-1 gives the worst case timings required for the Master Clock to switch from one
selected clock to another one. This is in the event that the prescaler is de-activated. When the
prescaler is activated, an additional time of 64 clock cycles of the new selected clock has to be
added.
Table 26-1.
Clock Switching Timings (Worst Case)
From
Main Clock
SLCK
PLLA Clock
–
4 x SLCK +
2.5 x Main Clock
3 x PLLA Clock +
4 x SLCK +
1 x Main Clock
0.5 x Main Clock +
4.5 x SLCK
–
3 x PLLA Clock +
5 x SLCK
0.5 x Main Clock +
4 x SLCK +
PLLACOUNT x SLCK +
2.5 x PLLAx Clock
2.5 x PLLA Clock +
5 x SLCK +
PLLACOUNT x SLCK
2.5 x PLLA Clock +
4 x SLCK +
PLLACOUNT x SLCK
To
Main Clock
SLCK
PLLA Clock
348
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
26.10.2
Clock Switching Waveforms
Figure 26-3. Switch Master Clock from Slow Clock to PLLA Clock
Slow Clock
PLL Clock
LOCK
MCKRDY
Master Clock
Write PMC_MCKR
Figure 26-4. Switch Master Clock from Main Clock to Slow Clock
Slow Clock
Main Clock
MCKRDY
Master Clock
Write PMC_MCKR
349
6438D–ATARM–13-Oct-09
Figure 26-5. Change PLLA Programming
Main Clock
PLL Clock
LOCK
MCKRDY
Master Clock
Main Clock
Write CKGR_PLLR
Figure 26-6. Programmable Clock Output Programming
PLL Clock
PCKRDY
PCKx Output
Write PMC_PCKx
Write PMC_SCER
Write PMC_SCDR
350
PLL Clock is selected
PCKx is enabled
PCKx is disabled
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
26.11 Power Management Controllerr (PMC) User Interface
Table 26-2.
Register Mapping
Offset
Register
Name
Access
Reset Value
0x0000
System Clock Enable Register
PMC_SCER
Write-only
–
0x0004
System Clock Disable Register
PMC_SCDR
Write-only
–
0x0008
System Clock Status Register
PMC _SCSR
Read-only
0x01
0x000C
Reserved
–
–
0x0010
Peripheral Clock Enable Register
PMC _PCER
Write-only
–
0x0014
Peripheral Clock Disable Register
PMC_PCDR
Write-only
–
0x0018
Peripheral Clock Status Register
PMC_PCSR
Read-only
0x0
0x001C
UTMI Clock Register
CKGR_UCKR
Read/Write
0x1020 0800
0x0020
Main Oscillator Register
CKGR_MOR
Read/Write
0x0
0x0024
Main Clock Frequency Register
CKGR_MCFR
Read-only
0x0
0x0028
PLLA Register
CKGR_PLLAR
Read/Write
0x3F00
0x002C
Reserved
–
–
0x0030
Master Clock Register
PMC_MCKR
Read/Write
0x0
0x0038
USB Clock Register
PMC_USB
Read/Write
0x0
0x003C
Reserved
–
–
0x0040
Programmable Clock 0 Register
PMC_PCK0
Read/Write
0x0
0x0044
Programmable Clock 1 Register
PMC_PCK1
Read/Write
0x0
–
–
0x0048-0x005C
–
–
–
Reserved
–
0x0060
Interrupt Enable Register
PMC_IER
Write-only
--
0x0064
Interrupt Disable Register
PMC_IDR
Write-only
--
0x0068
Status Register
PMC_SR
Read-only
0x08
0x006C
Interrupt Mask Register
PMC_IMR
Read-only
0x0
–
–
Write-only
0x0
0x0070 - 0x007C
0x0080
Reserved
PLL Charge Pump Current Register
–
PMC_PLLICPR
351
6438D–ATARM–13-Oct-09
26.11.1 PMC System Clock Enable Register
Register Name:PMC_SCER
Address:
0xFFFFFC00
Access Type:Write-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
PCK1
PCK0
7
6
5
4
3
2
1
0
–
UHP
–
–
–
DDRCK
–
–
• DDRCK: DDR Clock Enable
0 = No effect.
1 = Enables the DDR clock.
• UHP: USB Host OHCI Clocks Enable
0 = No effect.
1 = Enables the UHP48M and UHP12M OHCI clocks.
• PCKx: Programmable Clock x Output Enable
0 = No effect.
1 = Enables the corresponding Programmable Clock output.
352
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
26.11.2 PMC System Clock Disable Register
Register Name:PMC_SCDR
Address:
0xFFFFFC04
Access Type:Write-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
PCK1
PCK0
7
6
5
4
3
2
1
0
–
UHP
–
–
–
DDRCK
–
PCK
• PCK: Processor Clock Disable
0 = No effect.
1 = Disables the Processor clock. This is used to enter the processor in Idle Mode.
• DDRCK: DDR Clock Disable
0 = No effect.
1 = Disables the DDR clock.
• UHP: USB Host OHCI Clock Disable
0 = No effect.
1 = Disables the UHP48M and UHP12M OHCI clocks.
• PCKx: Programmable Clock x Output Disable
0 = No effect.
1 = Disables the corresponding Programmable Clock output.
353
6438D–ATARM–13-Oct-09
26.11.3 PMC System Clock Status Register
Register Name:PMC_SCSR
Address:
0xFFFFFC08
Access Type:Read-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
PCK1
PCK0
7
6
5
4
3
2
1
0
–
UHP
–
–
–
DDRCK
–
PCK
• PCK: Processor Clock Status
0 = The Processor clock is disabled.
1 = The Processor clock is enabled.
• DDRCK: DDR Clock Status
0 = The DDR clock is disabled.
1 = The DDR clock is enabled.
• UHP: USB Host Port Clock Status
0 = The UHP48M and UHP12M OHCI clocks are disabled.
1 = The UHP48M and UHP12M OHCI clocks are enabled.
• PCKx: Programmable Clock x Output Status
0 = The corresponding Programmable Clock output is disabled.
1 = The corresponding Programmable Clock output is enabled.
354
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
26.11.4 PMC Peripheral Clock Enable Register
Register Name:PMC_PCER
Address:
0xFFFFFC10
Access Type:Write-only
31
30
29
28
27
26
25
24
PID31
PID30
PID29
PID28
PID27
PID26
PID25
PID24
23
22
21
20
19
18
17
16
PID23
PID22
PID21
PID20
PID19
PID18
PID17
PID16
15
14
13
12
11
10
9
8
PID15
PID14
PID13
PID12
PID11
PID10
PID9
PID8
7
6
5
4
3
2
1
0
PID7
PID6
PID5
PID4
PID3
PID2
-
-
• PIDx: Peripheral Clock x Enable
0 = No effect.
1 = Enables the corresponding peripheral clock.
Note:
PID2 to PID31 refer to identifiers as defined in the section “Peripheral Identifiers” in the product datasheet.
Note:
Programming the control bits of the Peripheral ID that are not implemented has no effect on the behavior of the PMC.
355
6438D–ATARM–13-Oct-09
26.11.5 PMC Peripheral Clock Disable Register
Register Name:PMC_PCDR
Address:
0xFFFFFC14
Access Type:Write-only
31
30
29
28
27
26
25
24
PID31
PID30
PID29
PID28
PID27
PID26
PID25
PID24
23
22
21
20
19
18
17
16
PID23
PID22
PID21
PID20
PID19
PID18
PID17
PID16
15
14
13
12
11
10
9
8
PID15
PID14
PID13
PID12
PID11
PID10
PID9
PID8
7
6
5
4
3
2
1
0
PID7
PID6
PID5
PID4
PID3
PID2
-
-
• PIDx: Peripheral Clock x Disable
0 = No effect.
1 = Disables the corresponding peripheral clock.
Note:
356
PID2 to PID31 refer to identifiers as defined in the section “Peripheral Identifiers” in the product datasheet.
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
26.11.6 PMC Peripheral Clock Status Register
Register Name:PMC_PCSR
Address:
0xFFFFFC18
Access Type:Read-only
31
30
29
28
27
26
25
24
PID31
PID30
PID29
PID28
PID27
PID26
PID25
PID24
23
22
21
20
19
18
17
16
PID23
PID22
PID21
PID20
PID19
PID18
PID17
PID16
15
14
13
12
11
10
9
8
PID15
PID14
PID13
PID12
PID11
PID10
PID9
PID8
7
6
5
4
3
2
1
0
PID7
PID6
PID5
PID4
PID3
PID2
–
–
• PIDx: Peripheral Clock x Status
0 = The corresponding peripheral clock is disabled.
1 = The corresponding peripheral clock is enabled.
Note:
PID2 to PID31 refer to identifiers as defined in the section “Peripheral Identifiers” in the product datasheet.
357
6438D–ATARM–13-Oct-09
26.11.7 PMC UTMI Clock Configuration Register
Register Name:CKGR_UCKR
Address:
0xFFFFFC1C
Access Type:Read/Write
31
30
29
28
27
–
26
–
25
–
24
BIASEN
21
20
19
–
18
–
17
–
16
UPLLEN
BIASCOUNT
23
22
PLLCOUNT
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
–
6
–
5
–
4
–
3
–
2
–
1
–
0
–
• UPLLEN: UTMI PLL Enable
0 = The UTMI PLL is disabled.
1 = The UTMI PLL is enabled.
When UPLLEN is set, the LOCKU flag is set once the UTMI PLL startup time is achieved.
• PLLCOUNT: UTMI PLL Start-up Time
Specifies the number of Slow Clock cycles multiplied by 8 for the UTMI PLL start-up time.
• BIASEN: UTMI BIAS Enable
0 = The UTMI BIAS is disabled.
1 = The UTMI BIAS is enabled.
• BIASCOUNT: UTMI BIAS Start-up Time
Specifies the number of Slow Clock cycles for the UTMI BIAS start-up time.
358
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
26.11.8 PMC Clock Generator Main Oscillator Register
Register Name:CKGR_MOR
Address:
0xFFFFFC20
Access Type:Read/Write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
14
13
12
11
10
9
8
3
–
2
–
1
OSCBYPASS
0
MOSCEN
OSCOUNT
7
–
6
–
5
–
4
–
• MOSCEN: Main Oscillator Enable
A crystal must be connected between XIN and XOUT.
0 = The Main Oscillator is disabled.
1 = The Main Oscillator is enabled. OSCBYPASS must be set to 0.
When MOSCEN is set, the MOSCS flag is set once the Main Oscillator startup time is achieved.
• OSCBYPASS: Oscillator Bypass
0 = No effect.
1 = The Main Oscillator is bypassed. MOSCEN must be set to 0. An external clock must be connected on XIN.
When OSCBYPASS is set, the MOSCS flag in PMC_SR is automatically set.
Clearing MOSCEN and OSCBYPASS bits allows resetting the MOSCS flag.
• OSCOUNT: Main Oscillator Start-up Time
Specifies the number of Slow Clock cycles multiplied by 8 for the Main Oscillator start-up time.
359
6438D–ATARM–13-Oct-09
26.11.9 PMC Clock Generator Main Clock Frequency Register
Register Name:CKGR_MCFR
Address:
0xFFFFFC24
Access Type:Read-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
MAINRDY
15
14
13
12
11
10
9
8
3
2
1
0
MAINF
7
6
5
4
MAINF
• MAINF: Main Clock Frequency
Gives the number of Main Clock cycles within 16 Slow Clock periods.
• MAINRDY: Main Clock Ready
0 = MAINF value is not valid or the Main Oscillator is disabled.
1 = The Main Oscillator has been enabled previously and MAINF value is available.
360
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
26.11.10 PMC Clock Generator PLLA Register
Register Name:CKGR_PLLAR
Address:
0xFFFFFC28
Access Type:Read/Write
31
–
30
–
29
1
28
–
23
22
21
20
27
–
26
–
25
–
24
–
19
18
17
16
10
9
8
2
1
0
MULA
15
14
13
12
11
OUTA
7
PLLACOUNT
6
5
4
3
DIVA
Possible limitations on PLL input frequencies and multiplier factors should be checked before using the PMC.
Warning: Bit 29 must always be set to 1 when programming the CKGR_PLLAR register.
• DIVA: Divider A
DIVA
Divider Selected
0
Divider output is 0
1
Divider is bypassed
2 - 255
Divider output is the selected clock divided by DIVA.
• PLLACOUNT: PLLA Counter
Specifies the number of slow clock cycles before the LOCKA bit is set in PMC_SR after CKGR_PLLAR is written.
• OUTA: PLLA Clock Frequency Range
To optimize clock performance, this field must be programmed as specified in “PLL Characteristics” in the Electrical Characteristics section of the product datasheet.
• MULA: PLLA Multiplier
0 = The PLLA is deactivated.
1 up to 254 = The PLLA Clock frequency is the PLLA input frequency multiplied by MULA+ 1.
361
6438D–ATARM–13-Oct-09
26.11.11 PMC USB Clock Register
Register Name:PMC_USB
Address:
0xFFFFFC38
Access Type:Read/Write
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
USBDIV
7
6
5
4
3
2
1
0
–
–
–
–
–
–
–
USBS
• USBS: USB OHCI Input clock selection
0 = USB Clock Input is PLLA
1 = USB Clock Input is UPLL
• USBDIV: Divider for USB OHCI Clock.
USB Clock is Input clock divided by USBDIV+1
362
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
26.11.12 PMC Master Clock Register
Register Name:PMC_MCKR
Address:
0xFFFFFC30
Access Type:Read/Write
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
–
–
–
PLLADIV2
–
–
4
3
2
7
6
5
–
–
–
8
MDIV
1
PRES
0
CSS
• CSS: Master/Processor Clock Source Selection
CSS
Clock Source Selection
0
0
Slow Clock is selected.
0
1
Main Clock is selected.
1
0
PLLA Output clock is selected.
1
1
UPLL Output clock is selected.
• PRES: Master/Processor Clock Prescaler
Master/Processor Clock Dividers
Input Clock
PRES
0
0
0
Selected clock
0
0
1
Selected clock divided by 2
0
1
0
Selected clock divided by 4
0
1
1
Selected clock divided by 8
1
0
0
Selected clock divided by 16
1
0
1
Selected clock divided by 32
1
1
0
Selected clock divided by 64
1
1
1
Reserved
363
6438D–ATARM–13-Oct-09
• MDIV: Master Clock Division
MDIV
Master Clock Division
0
0
Master Clock is Prescaler Output Clock divided by 1.
Warning: SysClk DDR and DDRCK are not available.
0
1
Master Clock is Prescaler Output Clock divided by 2.
SysClk DDR is equal to 2 x MCK.
DDRCK is equal to MCK.
1
0
Master Clock is Prescaler Output Clock divided by 4.
SysClk DDR is equal to 2 x MCK.
DDRCK is equal to MCK.
1
1
Master Clock is Prescaler Output Clock divided by 3.
SysClk DDR is equal to 2 x MCK.
DDRCK is equal to MCK.
• PLLADIV2: PLLA divisor by 2
PLLADIV2
364
PLLA Clock Division
0
PLLA clock frequency is divided by 1.
1
PLLA clock frequency is divided by 2.
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
26.11.13 PMC Programmable Clock Register
Register Name:PMC_PCKx
Address:
0xFFFFFC40
Access Type:Read/Write
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
SLCKMCK
4
3
2
1
7
6
5
–
–
–
PRES
0
CSS
• CSS: Master Clock Selection
CSS
Clock Source Selection
0
0
Slow Clock is selected.
0
1
Main Clock is selected.
1
0
PLLACK/PLLADIV2 is selected.
1
1
UPLLCK is selected.
• PRES: Programmable Clock Prescaler
PRES
Programmable Clock
0
0
0
Selected clock
0
0
1
Selected clock divided by 2
0
1
0
Selected clock divided by 4
0
1
1
Selected clock divided by 8
1
0
0
Selected clock divided by 16
1
0
1
Selected clock divided by 32
1
1
0
Selected clock divided by 64
1
1
1
Reserved
• SLCKMCK: Slow Clock or Master Clock Selection
0 = Slow clock is selected
1 = Master clock is selected
If CSS field is not programmed to ‘00’ no Output clock will be provided.
365
6438D–ATARM–13-Oct-09
26.11.14 PMC Interrupt Enable Register
Register Name:PMC_IER
Address:
0xFFFFFC60
Access Type:Write-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
PCKRDY1
PCKRDY0
7
6
5
4
3
2
1
0
–
LOCKU
–
–
MCKRDY
–
LOCKA
MOSCS
• MOSCS: Main Oscillator Status Interrupt Enable
• LOCKA: PLL Lock Interrupt Enable
• MCKRDY: Master Clock Ready Interrupt Enable
• LOCKU: UTMI PLL Lock Interrupt Enable
• PCKRDYx: Programmable Clock Ready x Interrupt Enable
0 = No effect.
1 = Enables the corresponding interrupt.
366
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
26.11.15 PMC Interrupt Disable Register
Register Name:PMC_IDR
Address:
0xFFFFFC64
Access Type:Write-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
PCKRDY1
PCKRDY0
7
6
5
4
3
2
1
0
–
LOCKU
–
–
MCKRDY
–
LOCKA
MOSCS
• MOSCS: Main Oscillator Status Interrupt Disable
• LOCKA: PLLA Lock Interrupt Disable
• MCKRDY: Master Clock Ready Interrupt Disable
• LOCKU: UTMI PLL Lock Interrupt Disable
• PCKRDYx: Programmable Clock Ready x Interrupt Disable
0 = No effect.
1 = Disables the corresponding interrupt.
367
6438D–ATARM–13-Oct-09
26.11.16 PMC Status Register
Register Name:PMC_SR
Address:
0xFFFFFC68
Access Type:Read-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
PCKRDY1
PCKRDY0
7
6
5
4
3
2
1
0
–
LOCKU
–
–
MCKRDY
–
LOCKA
MOSCS
• MOSCS: MOSCS Flag Status
0 = Main oscillator is not stabilized.
1 = Main oscillator is stabilized.
• LOCKA: PLLA Lock Status
0 = PLLA is not locked
1 = PLLA is locked.
• MCKRDY: Master Clock Status
0 = Master Clock is not ready.
1 = Master Clock is ready.
• LOCKU: UPLL Lock Status
0 = UPLL is not locked
1 = UPLL is locked.
• PCKRDYx: Programmable Clock Ready Status
0 = Programmable Clock x is not ready.
1 = Programmable Clock x is ready.
368
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
26.11.17 PMC Interrupt Mask Register
Register Name:PMC_IMR
Address:
0xFFFFFC6C
Access Type:Read-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
PCKRDY1
PCKRDY0
7
6
5
4
3
2
1
0
–
LOCKU
–
–
MCKRDY
–
LOCKA
MOSCS
• MOSCS: Main Oscillator Status Interrupt Mask
• LOCKA: PLLA Lock Interrupt Mask
• MCKRDY: Master Clock Ready Interrupt Mask
• LOCKU: UTMI PLL Lock Interrupt Mask
• PCKRDYx: Programmable Clock Ready x Interrupt Mask
0 = The corresponding interrupt is enabled.
1 = The corresponding interrupt is disabled.
369
6438D–ATARM–13-Oct-09
26.11.18 PLL Charge Pump Current Register
Register Name:PMC_PLLICPR
Address:
0xFFFFFC80
Access Type:Write-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
–
–
–
–
–
–
–
ICPLLA
• ICPLLA: Charge Pump Current
To optimize clock performance, this field must be programmed as specified in “PLL A Characteristics” in the Electrical Characteristics section of the product datasheet.
370
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
27. Advanced Interrupt Controller (AIC)
27.1
Description
The Advanced Interrupt Controller (AIC) is an 8-level priority, individually maskable, vectored
interrupt controller, providing handling of up to thirty-two interrupt sources. It is designed to substantially reduce the software and real-time overhead in handling internal and external
interrupts.
The AIC drives the nFIQ (fast interrupt request) and the nIRQ (standard interrupt request) inputs
of an ARM processor. Inputs of the AIC are either internal peripheral interrupts or external interrupts coming from the product's pins.
The 8-level Priority Controller allows the user to define the priority for each interrupt source, thus
permitting higher priority interrupts to be serviced even if a lower priority interrupt is being
treated.
Internal interrupt sources can be programmed to be level sensitive or edge triggered. External
interrupt sources can be programmed to be positive-edge or negative-edge triggered or highlevel or low-level sensitive.
The fast forcing feature redirects any internal or external interrupt source to provide a fast interrupt rather than a normal interrupt.
27.2
Embedded Characteristics
• Controls the interrupt lines (nIRQ and nFIQ) of the ARM Processor
• Thirty-two individually maskable and vectored interrupt sources
– Source 0 is reserved for the Fast Interrupt Input (FIQ)
– Source 1 is reserved for system peripherals (PIT, RTT, PMC, DBGU, etc.)
– Programmable Edge-triggered or Level-sensitive Internal Sources
– Programmable Positive/Negative Edge-triggered or High/Low Level-sensitive
• One External Sources plus the Fast Interrupt signal
• 8-level Priority Controller
– Drives the Normal Interrupt of the processor
– Handles priority of the interrupt sources 1 to 31
– Higher priority interrupts can be served during service of lower priority interrupt
• Vectoring
– Optimizes Interrupt Service Routine Branch and Execution
– One 32-bit Vector Register per interrupt source
– Interrupt Vector Register reads the corresponding current Interrupt Vector
• Protect Mode
– Easy debugging by preventing automatic operations when protect modes are
enabled
• Fast Forcing
– Permits redirecting any normal interrupt source on the Fast Interrupt of the
processor
371
6438D–ATARM–13-Oct-09
27.3
Block Diagram
Figure 27-1. Block Diagram
FIQ
AIC
ARM
Processor
IRQ0-IRQn
Up to
Thirty-two
Sources
Embedded
PeripheralEE
Embedded
nFIQ
nIRQ
Peripheral
Embedded
Peripheral
APB
27.4
Application Block Diagram
Figure 27-2. Description of the Application Block
OS-based Applications
Standalone
Applications
OS Drivers
RTOS Drivers
Hard Real Time Tasks
General OS Interrupt Handler
Advanced Interrupt Controller
External Peripherals
(External Interrupts)
Embedded Peripherals
27.5
AIC Detailed Block Diagram
Figure 27-3. AIC Detailed Block Diagram
Advanced Interrupt Controller
FIQ
PIO
Controller
Fast
Interrupt
Controller
External
Source
Input
Stage
ARM
Processor
nFIQ
nIRQ
IRQ0-IRQn
Embedded
Peripherals
Interrupt
Priority
Controller
Fast
Forcing
PIOIRQ
Internal
Source
Input
Stage
Processor
Clock
Power
Management
Controller
User Interface
Wake Up
APB
372
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
27.6
I/O Line Description
Table 27-1.
I/O Line Description
Pin Name
Pin Description
Type
FIQ
Fast Interrupt
Input
IRQ0 - IRQn
Interrupt 0 - Interrupt n
Input
27.7
27.7.1
Product Dependencies
I/O Lines
The interrupt signals FIQ and IRQ0 to IRQn are normally multiplexed through the PIO controllers. Depending on the features of the PIO controller used in the product, the pins must be
programmed in accordance with their assigned interrupt function. This is not applicable when
the PIO controller used in the product is transparent on the input path.
Table 27-2.
27.7.2
I/O Lines
Instance
Signal
I/O Line
Peripheral
AIC
FIQ
PD19
B
AIC
IRQ
PD18
B
Power Management
The Advanced Interrupt Controller is continuously clocked. The Power Management Controller
has no effect on the Advanced Interrupt Controller behavior.
The assertion of the Advanced Interrupt Controller outputs, either nIRQ or nFIQ, wakes up the
ARM processor while it is in Idle Mode. The General Interrupt Mask feature enables the AIC to
wake up the processor without asserting the interrupt line of the processor, thus providing synchronization of the processor on an event.
27.7.3
Interrupt Sources
The Interrupt Source 0 is always located at FIQ. If the product does not feature an FIQ pin, the
Interrupt Source 0 cannot be used.
The Interrupt Source 1 is always located at System Interrupt. This is the result of the OR-wiring
of the system peripheral interrupt lines. When a system interrupt occurs, the service routine
must first distinguish the cause of the interrupt. This is performed by reading successively the
status registers of the above mentioned system peripherals.
The interrupt sources 2 to 31 can either be connected to the interrupt outputs of an embedded
user peripheral or to external interrupt lines. The external interrupt lines can be connected
directly, or through the PIO Controller.
The PIO Controllers are considered as user peripherals in the scope of interrupt handling.
Accordingly, the PIO Controller interrupt lines are connected to the Interrupt Sources 2 to 31.
The peripheral identification defined at the product level corresponds to the interrupt source
number (as well as the bit number controlling the clock of the peripheral). Consequently, to simplify the description of the functional operations and the user interface, the interrupt sources are
named FIQ, SYS, and PID2 to PID31.
373
6438D–ATARM–13-Oct-09
27.8
Functional Description
27.8.1
27.8.1.1
Interrupt Source Control
Interrupt Source Mode
The Advanced Interrupt Controller independently programs each interrupt source. The SRCTYPE field of the corresponding AIC_SMR (Source Mode Register) selects the interrupt
condition of each source.
The internal interrupt sources wired on the interrupt outputs of the embedded peripherals can be
programmed either in level-sensitive mode or in edge-triggered mode. The active level of the
internal interrupts is not important for the user.
The external interrupt sources can be programmed either in high level-sensitive or low level-sensitive modes, or in positive edge-triggered or negative edge-triggered modes.
27.8.1.2
Interrupt Source Enabling
Each interrupt source, including the FIQ in source 0, can be enabled or disabled by using the
command registers; AIC_IECR (Interrupt Enable Command Register) and AIC_IDCR (Interrupt
Disable Command Register). This set of registers conducts enabling or disabling in one instruction. The interrupt mask can be read in the AIC_IMR register. A disabled interrupt does not affect
servicing of other interrupts.
27.8.1.3
Interrupt Clearing and Setting
All interrupt sources programmed to be edge-triggered (including the FIQ in source 0) can be
individually set or cleared by writing respectively the AIC_ISCR and AIC_ICCR registers. Clearing or setting interrupt sources programmed in level-sensitive mode has no effect.
The clear operation is perfunctory, as the software must perform an action to reinitialize the
“memorization” circuitry activated when the source is programmed in edge-triggered mode.
However, the set operation is available for auto-test or software debug purposes. It can also be
used to execute an AIC-implementation of a software interrupt.
The AIC features an automatic clear of the current interrupt when the AIC_IVR (Interrupt Vector
Register) is read. Only the interrupt source being detected by the AIC as the current interrupt is
affected by this operation. (See “Priority Controller” on page 377.) The automatic clear reduces
the operations required by the interrupt service routine entry code to reading the AIC_IVR. Note
that the automatic interrupt clear is disabled if the interrupt source has the Fast Forcing feature
enabled as it is considered uniquely as a FIQ source. (For further details, See “Fast Forcing” on
page 381.)
The automatic clear of the interrupt source 0 is performed when AIC_FVR is read.
27.8.1.4
Interrupt Status
For each interrupt, the AIC operation originates in AIC_IPR (Interrupt Pending Register) and its
mask in AIC_IMR (Interrupt Mask Register). AIC_IPR enables the actual activity of the sources,
whether masked or not.
The AIC_ISR register reads the number of the current interrupt (see “Priority Controller” on page
377) and the register AIC_CISR gives an image of the signals nIRQ and nFIQ driven on the
processor.
Each status referred to above can be used to optimize the interrupt handling of the systems.
374
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
27.8.1.5
Figure 27-4.
Internal Interrupt Source Input Stage
Internal Interrupt Source Input Stage
AIC_SMRI
(SRCTYPE)
Level/
Edge
Source i
AIC_IPR
AIC_IMR
Fast Interrupt Controller
or
Priority Controller
Edge
AIC_IECR
Detector
Set Clear
FF
AIC_ISCR
AIC_ICCR
AIC_IDCR
27.8.1.6
External Interrupt Source Input Stage
Figure 27-5. External Interrupt Source Input Stage
High/Low
AIC_SMRi
SRCTYPE
Level/
Edge
AIC_IPR
AIC_IMR
Source i
Fast Interrupt Controller
or
Priority Controller
AIC_IECR
Pos./Neg.
Edge
Detector
Set
AIC_ISCR
FF
Clear
AIC_IDCR
AIC_ICCR
375
6438D–ATARM–13-Oct-09
27.8.2
Interrupt Latencies
Global interrupt latencies depend on several parameters, including:
• The time the software masks the interrupts.
• Occurrence, either at the processor level or at the AIC level.
• The execution time of the instruction in progress when the interrupt occurs.
• The treatment of higher priority interrupts and the resynchronization of the hardware signals.
This section addresses only the hardware resynchronizations. It gives details of the latency
times between the event on an external interrupt leading in a valid interrupt (edge or level) or the
assertion of an internal interrupt source and the assertion of the nIRQ or nFIQ line on the processor. The resynchronization time depends on the programming of the interrupt source and on
its type (internal or external). For the standard interrupt, resynchronization times are given
assuming there is no higher priority in progress.
The PIO Controller multiplexing has no effect on the interrupt latencies of the external interrupt
sources.
27.8.2.1
External Interrupt Edge Triggered Source
Figure 27-6.
External Interrupt Edge Triggered Source
MCK
IRQ or FIQ
(Positive Edge)
IRQ or FIQ
(Negative Edge)
nIRQ
Maximum IRQ Latency = 4 Cycles
nFIQ
Maximum FIQ Latency = 4 Cycles
27.8.2.2
External Interrupt Level Sensitive Source
Figure 27-7.
External Interrupt Level Sensitive Source
MCK
IRQ or FIQ
(High Level)
IRQ or FIQ
(Low Level)
nIRQ
Maximum IRQ
Latency = 3 Cycles
nFIQ
Maximum FIQ
Latency = 3 cycles
376
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
27.8.2.3
Internal Interrupt Edge Triggered Source
Figure 27-8.
Internal Interrupt Edge Triggered Source
MCK
nIRQ
Maximum IRQ Latency = 4.5 Cycles
Peripheral Interrupt
Becomes Active
27.8.2.4
Internal Interrupt Level Sensitive Source
Figure 27-9.
Internal Interrupt Level Sensitive Source
MCK
nIRQ
Maximum IRQ Latency = 3.5 Cycles
Peripheral Interrupt
Becomes Active
27.8.3
27.8.3.1
Normal Interrupt
Priority Controller
An 8-level priority controller drives the nIRQ line of the processor, depending on the interrupt
conditions occurring on the interrupt sources 1 to 31 (except for those programmed in Fast
Forcing).
Each interrupt source has a programmable priority level of 7 to 0, which is user-definable by writing the PRIOR field of the corresponding AIC_SMR (Source Mode Register). Level 7 is the
highest priority and level 0 the lowest.
As soon as an interrupt condition occurs, as defined by the SRCTYPE field of the AIC_SMR
(Source Mode Register), the nIRQ line is asserted. As a new interrupt condition might have happened on other interrupt sources since the nIRQ has been asserted, the priority controller
determines the current interrupt at the time the AIC_IVR (Interrupt Vector Register) is read. The
read of AIC_IVR is the entry point of the interrupt handling which allows the AIC to consider
that the interrupt has been taken into account by the software.
The current priority level is defined as the priority level of the current interrupt.
If several interrupt sources of equal priority are pending and enabled when the AIC_IVR is read,
the interrupt with the lowest interrupt source number is serviced first.
377
6438D–ATARM–13-Oct-09
The nIRQ line can be asserted only if an interrupt condition occurs on an interrupt source with a
higher priority. If an interrupt condition happens (or is pending) during the interrupt treatment in
progress, it is delayed until the software indicates to the AIC the end of the current service by
writing the AIC_EOICR (End of Interrupt Command Register). The write of AIC_EOICR is the
exit point of the interrupt handling.
27.8.3.2
Interrupt Nesting
The priority controller utilizes interrupt nesting in order for the high priority interrupt to be handled
during the service of lower priority interrupts. This requires the interrupt service routines of the
lower interrupts to re-enable the interrupt at the processor level.
When an interrupt of a higher priority happens during an already occurring interrupt service routine, the nIRQ line is re-asserted. If the interrupt is enabled at the core level, the current
execution is interrupted and the new interrupt service routine should read the AIC_IVR. At this
time, the current interrupt number and its priority level are pushed into an embedded hardware
stack, so that they are saved and restored when the higher priority interrupt servicing is finished
and the AIC_EOICR is written.
The AIC is equipped with an 8-level wide hardware stack in order to support up to eight interrupt
nestings pursuant to having eight priority levels.
27.8.3.3
Interrupt Vectoring
The interrupt handler addresses corresponding to each interrupt source can be stored in the registers AIC_SVR1 to AIC_SVR31 (Source Vector Register 1 to 31). When the processor reads
AIC_IVR (Interrupt Vector Register), the value written into AIC_SVR corresponding to the current interrupt is returned.
This feature offers a way to branch in one single instruction to the handler corresponding to the
current interrupt, as AIC_IVR is mapped at the absolute address 0xFFFF F100 and thus accessible from the ARM interrupt vector at address 0x0000 0018 through the following instruction:
LDR
PC,[PC,# -&F20]
When the processor executes this instruction, it loads the read value in AIC_IVR in its program
counter, thus branching the execution on the correct interrupt handler.
This feature is often not used when the application is based on an operating system (either real
time or not). Operating systems often have a single entry point for all the interrupts and the first
task performed is to discern the source of the interrupt.
However, it is strongly recommended to port the operating system on AT91 products by supporting the interrupt vectoring. This can be performed by defining all the AIC_SVR of the interrupt
source to be handled by the operating system at the address of its interrupt handler. When doing
so, the interrupt vectoring permits a critical interrupt to transfer the execution on a specific very
fast handler and not onto the operating system’s general interrupt handler. This facilitates the
support of hard real-time tasks (input/outputs of voice/audio buffers and software peripheral handling) to be handled efficiently and independently of the application running under an operating
system.
27.8.3.4
378
Interrupt Handlers
This section gives an overview of the fast interrupt handling sequence when using the AIC. It is
assumed that the programmer understands the architecture of the ARM processor, and especially the processor interrupt modes and the associated status bits.
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
It is assumed that:
1. The Advanced Interrupt Controller has been programmed, AIC_SVR registers are
loaded with corresponding interrupt service routine addresses and interrupts are
enabled.
2. The instruction at the ARM interrupt exception vector address is required to work with
the vectoring
LDR PC, [PC, # -&F20]
When nIRQ is asserted, if the bit “I” of CPSR is 0, the sequence is as follows:
1. The CPSR is stored in SPSR_irq, the current value of the Program Counter is loaded in
the Interrupt link register (R14_irq) and the Program Counter (R15) is loaded with 0x18.
In the following cycle during fetch at address 0x1C, the ARM core adjusts R14_irq, decrementing it by four.
2. The ARM core enters Interrupt mode, if it has not already done so.
3. When the instruction loaded at address 0x18 is executed, the program counter is
loaded with the value read in AIC_IVR. Reading the AIC_IVR has the following effects:
– Sets the current interrupt to be the pending and enabled interrupt with the highest
priority. The current level is the priority level of the current interrupt.
– De-asserts the nIRQ line on the processor. Even if vectoring is not used, AIC_IVR
must be read in order to de-assert nIRQ.
– Automatically clears the interrupt, if it has been programmed to be edge-triggered.
– Pushes the current level and the current interrupt number on to the stack.
– Returns the value written in the AIC_SVR corresponding to the current interrupt.
4. The previous step has the effect of branching to the corresponding interrupt service
routine. This should start by saving the link register (R14_irq) and SPSR_IRQ. The link
register must be decremented by four when it is saved if it is to be restored directly into
the program counter at the end of the interrupt. For example, the instruction SUB PC,
LR, #4 may be used.
5. Further interrupts can then be unmasked by clearing the “I” bit in CPSR, allowing reassertion of the nIRQ to be taken into account by the core. This can happen if an interrupt with a higher priority than the current interrupt occurs.
6. The interrupt handler can then proceed as required, saving the registers that will be
used and restoring them at the end. During this phase, an interrupt of higher priority
than the current level will restart the sequence from step 1.
Note:
If the interrupt is programmed to be level sensitive, the source of the interrupt must be cleared during this phase.
7. The “I” bit in CPSR must be set in order to mask interrupts before exiting to ensure that
the interrupt is completed in an orderly manner.
8. The End of Interrupt Command Register (AIC_EOICR) must be written in order to indicate to the AIC that the current interrupt is finished. This causes the current level to be
popped from the stack, restoring the previous current level if one exists on the stack. If
another interrupt is pending, with lower or equal priority than the old current level but
with higher priority than the new current level, the nIRQ line is re-asserted, but the interrupt sequence does not immediately start because the “I” bit is set in the core.
SPSR_irq is restored. Finally, the saved value of the link register is restored directly into
the PC. This has the effect of returning from the interrupt to whatever was being executed before, and of loading the CPSR with the stored SPSR, masking or unmasking
the interrupts depending on the state saved in SPSR_irq.
379
6438D–ATARM–13-Oct-09
Note:
27.8.4
The “I” bit in SPSR is significant. If it is set, it indicates that the ARM core was on the verge of
masking an interrupt when the mask instruction was interrupted. Hence, when SPSR is restored,
the mask instruction is completed (interrupt is masked).
Fast Interrupt
27.8.4.1
Fast Interrupt Source
The interrupt source 0 is the only source which can raise a fast interrupt request to the processor
except if fast forcing is used. The interrupt source 0 is generally connected to a FIQ pin of the
product, either directly or through a PIO Controller.
27.8.4.2
Fast Interrupt Control
The fast interrupt logic of the AIC has no priority controller. The mode of interrupt source 0 is
programmed with the AIC_SMR0 and the field PRIOR of this register is not used even if it reads
what has been written. The field SRCTYPE of AIC_SMR0 enables programming the fast interrupt source to be positive-edge triggered or negative-edge triggered or high-level sensitive or
low-level sensitive
Writing 0x1 in the AIC_IECR (Interrupt Enable Command Register) and AIC_IDCR (Interrupt
Disable Command Register) respectively enables and disables the fast interrupt. The bit 0 of
AIC_IMR (Interrupt Mask Register) indicates whether the fast interrupt is enabled or disabled.
27.8.4.3
Fast Interrupt Vectoring
The fast interrupt handler address can be stored in AIC_SVR0 (Source Vector Register 0). The
value written into this register is returned when the processor reads AIC_FVR (Fast Vector Register). This offers a way to branch in one single instruction to the interrupt handler, as AIC_FVR
is mapped at the absolute address 0xFFFF F104 and thus accessible from the ARM fast interrupt vector at address 0x0000 001C through the following instruction:
LDR
PC,[PC,# -&F20]
When the processor executes this instruction it loads the value read in AIC_FVR in its program
counter, thus branching the execution on the fast interrupt handler. It also automatically performs the clear of the fast interrupt source if it is programmed in edge-triggered mode.
27.8.4.4
Fast Interrupt Handlers
This section gives an overview of the fast interrupt handling sequence when using the AIC. It is
assumed that the programmer understands the architecture of the ARM processor, and especially the processor interrupt modes and associated status bits.
Assuming that:
1. The Advanced Interrupt Controller has been programmed, AIC_SVR0 is loaded with
the fast interrupt service routine address, and the interrupt source 0 is enabled.
2. The Instruction at address 0x1C (FIQ exception vector address) is required to vector
the fast interrupt:
LDR PC, [PC, # -&F20]
3. The user does not need nested fast interrupts.
When nFIQ is asserted, if the bit “F” of CPSR is 0, the sequence is:
1. The CPSR is stored in SPSR_fiq, the current value of the program counter is loaded in
the FIQ link register (R14_FIQ) and the program counter (R15) is loaded with 0x1C. In
380
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
the following cycle, during fetch at address 0x20, the ARM core adjusts R14_fiq, decrementing it by four.
2. The ARM core enters FIQ mode.
3. When the instruction loaded at address 0x1C is executed, the program counter is
loaded with the value read in AIC_FVR. Reading the AIC_FVR has effect of automatically clearing the fast interrupt, if it has been programmed to be edge triggered. In this
case only, it de-asserts the nFIQ line on the processor.
4. The previous step enables branching to the corresponding interrupt service routine. It is
not necessary to save the link register R14_fiq and SPSR_fiq if nested fast interrupts
are not needed.
5. The Interrupt Handler can then proceed as required. It is not necessary to save registers R8 to R13 because FIQ mode has its own dedicated registers and the user R8 to
R13 are banked. The other registers, R0 to R7, must be saved before being used, and
restored at the end (before the next step). Note that if the fast interrupt is programmed
to be level sensitive, the source of the interrupt must be cleared during this phase in
order to de-assert the interrupt source 0.
6. Finally, the Link Register R14_fiq is restored into the PC after decrementing it by four
(with instruction SUB PC, LR, #4 for example). This has the effect of returning from
the interrupt to whatever was being executed before, loading the CPSR with the SPSR
and masking or unmasking the fast interrupt depending on the state saved in the
SPSR.
Note:
The “F” bit in SPSR is significant. If it is set, it indicates that the ARM core was just about to mask
FIQ interrupts when the mask instruction was interrupted. Hence when the SPSR is restored, the
interrupted instruction is completed (FIQ is masked).
Another way to handle the fast interrupt is to map the interrupt service routine at the address of
the ARM vector 0x1C. This method does not use the vectoring, so that reading AIC_FVR must
be performed at the very beginning of the handler operation. However, this method saves the
execution of a branch instruction.
27.8.4.5
Fast Forcing
The Fast Forcing feature of the advanced interrupt controller provides redirection of any normal
Interrupt source on the fast interrupt controller.
Fast Forcing is enabled or disabled by writing to the Fast Forcing Enable Register (AIC_FFER)
and the Fast Forcing Disable Register (AIC_FFDR). Writing to these registers results in an
update of the Fast Forcing Status Register (AIC_FFSR) that controls the feature for each internal or external interrupt source.
When Fast Forcing is disabled, the interrupt sources are handled as described in the previous
pages.
When Fast Forcing is enabled, the edge/level programming and, in certain cases, edge detection of the interrupt source is still active but the source cannot trigger a normal interrupt to the
processor and is not seen by the priority handler.
If the interrupt source is programmed in level-sensitive mode and an active level is sampled,
Fast Forcing results in the assertion of the nFIQ line to the core.
If the interrupt source is programmed in edge-triggered mode and an active edge is detected,
Fast Forcing results in the assertion of the nFIQ line to the core.
The Fast Forcing feature does not affect the Source 0 pending bit in the Interrupt Pending Register (AIC_IPR).
381
6438D–ATARM–13-Oct-09
The FIQ Vector Register (AIC_FVR) reads the contents of the Source Vector Register 0
(AIC_SVR0), whatever the source of the fast interrupt may be. The read of the FVR does not
clear the Source 0 when the fast forcing feature is used and the interrupt source should be
cleared by writing to the Interrupt Clear Command Register (AIC_ICCR).
All enabled and pending interrupt sources that have the fast forcing feature enabled and that are
programmed in edge-triggered mode must be cleared by writing to the Interrupt Clear Command
Register. In doing so, they are cleared independently and thus lost interrupts are prevented.
The read of AIC_IVR does not clear the source that has the fast forcing feature enabled.
The source 0, reserved to the fast interrupt, continues operating normally and becomes one of
the Fast Interrupt sources.
Figure 27-10. Fast Forcing
Source 0 _ FIQ
AIC_IPR
Input Stage
Automatic Clear
AIC_IMR
nFIQ
Read FVR if Fast Forcing is
disabled on Sources 1 to 31.
AIC_FFSR
Source n
AIC_IPR
Input Stage
Priority
Manager
Automatic Clear
AIC_IMR
nIRQ
Read IVR if Source n is the current interrupt
and if Fast Forcing is disabled on Source n.
27.8.5
Protect Mode
The Protect Mode permits reading the Interrupt Vector Register without performing the associated automatic operations. This is necessary when working with a debug system. When a
debugger, working either with a Debug Monitor or the ARM processor's ICE, stops the applications and updates the opened windows, it might read the AIC User Interface and thus the IVR.
This has undesirable consequences:
• If an enabled interrupt with a higher priority than the current one is pending, it is stacked.
• If there is no enabled pending interrupt, the spurious vector is returned.
In either case, an End of Interrupt command is necessary to acknowledge and to restore the
context of the AIC. This operation is generally not performed by the debug system as the debug
system would become strongly intrusive and cause the application to enter an undesired state.
This is avoided by using the Protect Mode. Writing PROT in AIC_DCR (Debug Control Register)
at 0x1 enables the Protect Mode.
When the Protect Mode is enabled, the AIC performs interrupt stacking only when a write access
is performed on the AIC_IVR. Therefore, the Interrupt Service Routines must write (arbitrary
data) to the AIC_IVR just after reading it. The new context of the AIC, including the value of the
Interrupt Status Register (AIC_ISR), is updated with the current interrupt only when AIC_IVR is
written.
382
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
An AIC_IVR read on its own (e.g., by a debugger), modifies neither the AIC context nor the
AIC_ISR. Extra AIC_IVR reads perform the same operations. However, it is recommended to
not stop the processor between the read and the write of AIC_IVR of the interrupt service routine
to make sure the debugger does not modify the AIC context.
To summarize, in normal operating mode, the read of AIC_IVR performs the following operations within the AIC:
1. Calculates active interrupt (higher than current or spurious).
2. Determines and returns the vector of the active interrupt.
3. Memorizes the interrupt.
4. Pushes the current priority level onto the internal stack.
5. Acknowledges the interrupt.
However, while the Protect Mode is activated, only operations 1 to 3 are performed when
AIC_IVR is read. Operations 4 and 5 are only performed by the AIC when AIC_IVR is written.
Software that has been written and debugged using the Protect Mode runs correctly in Normal
Mode without modification. However, in Normal Mode the AIC_IVR write has no effect and can
be removed to optimize the code.
27.8.6
Spurious Interrupt
The Advanced Interrupt Controller features protection against spurious interrupts. A spurious
interrupt is defined as being the assertion of an interrupt source long enough for the AIC to
assert the nIRQ, but no longer present when AIC_IVR is read. This is most prone to occur when:
• An external interrupt source is programmed in level-sensitive mode and an active level occurs
for only a short time.
• An internal interrupt source is programmed in level sensitive and the output signal of the
corresponding embedded peripheral is activated for a short time. (As in the case for the
Watchdog.)
• An interrupt occurs just a few cycles before the software begins to mask it, thus resulting in a
pulse on the interrupt source.
The AIC detects a spurious interrupt at the time the AIC_IVR is read while no enabled interrupt
source is pending. When this happens, the AIC returns the value stored by the programmer in
AIC_SPU (Spurious Vector Register). The programmer must store the address of a spurious
interrupt handler in AIC_SPU as part of the application, to enable an as fast as possible return to
the normal execution flow. This handler writes in AIC_EOICR and performs a return from
interrupt.
27.8.7
General Interrupt Mask
The AIC features a General Interrupt Mask bit to prevent interrupts from reaching the processor.
Both the nIRQ and the nFIQ lines are driven to their inactive state if the bit GMSK in AIC_DCR
(Debug Control Register) is set. However, this mask does not prevent waking up the processor if
it has entered Idle Mode. This function facilitates synchronizing the processor on a next event
and, as soon as the event occurs, performs subsequent operations without having to handle an
interrupt. It is strongly recommended to use this mask with caution.
383
6438D–ATARM–13-Oct-09
27.9
Advanced Interrupt Controller (AIC) User Interface
27.9.1
Base Address
The AIC is mapped at the address 0xFFFF F000. It has a total 4-KByte addressing space. This permits the vectoring feature, as the PC-relative load/store instructions of the ARM processor support only a ± 4-KByte offset.
Table 27-3.
Register Mapping
Offset
Register
Name
Access
Reset
0x00
0x04
Source Mode Register 0
AIC_SMR0
Read-write
0x0
Source Mode Register 1
AIC_SMR1
Read-write
0x0
---
---
---
---
---
0x7C
Source Mode Register 31
AIC_SMR31
Read-write
0x0
0x80
Source Vector Register 0
AIC_SVR0
Read-write
0x0
0x84
Source Vector Register 1
AIC_SVR1
Read-write
0x0
---
---
---
---
---
0xFC
Source Vector Register 31
AIC_SVR31
Read-write
0x0
0x100
Interrupt Vector Register
AIC_IVR
Read-only
0x0
0x104
FIQ Interrupt Vector Register
AIC_FVR
Read-only
0x0
0x108
Interrupt Status Register
AIC_ISR
Read-only
0x0
AIC_IPR
Read-only
0x0(1)
(2)
0x10C
Interrupt Pending Register
0x110
Interrupt Mask Register(2)
AIC_IMR
Read-only
0x0
0x114
Core Interrupt Status Register
AIC_CISR
Read-only
0x0
0x118 - 0x11C
Reserved
---
---
---
AIC_IECR
Write-only
---
AIC_IDCR
Write-only
---
AIC_ICCR
Write-only
---
AIC_ISCR
Write-only
---
AIC_EOICR
Write-only
---
0x120
Interrupt Enable Command Register
(2)
0x124
Interrupt Disable Command Register
0x128
Interrupt Clear Command Register(2)
(2)
0x12C
Interrupt Set Command Register
0x130
End of Interrupt Command Register
(2)
0x134
Spurious Interrupt Vector Register
AIC_SPU
Read-write
0x0
0x138
Debug Control Register
AIC_DCR
Read-write
0x0
0x13C
Reserved
---
---
---
AIC_FFER
Write-only
---
(2)
0x140
Fast Forcing Enable Register
(2)
0x144
Fast Forcing Disable Register
0x148
Fast Forcing Status Register(2)
0x14C - 0x1E0
Reserved
0x1EC - 0x1FC
Reserved
Notes:
AIC_FFDR
Write-only
---
AIC_FFSR
Read-only
0x0
---
---
---
1. The reset value of this register depends on the level of the external interrupt source. All other sources are cleared at reset,
thus not pending.
2. PID2...PID31 bit fields refer to the identifiers as defined in the Peripheral Identifiers Section of the product datasheet.
3. Values in the Version Register vary with the version of the IP block implementation.
384
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
27.9.2
AIC Source Mode Register
Register Name:
AIC_SMR0..AIC_SMR31
Address:
0xFFFFF000
Access Type:
Read-write
Reset Value:
0x0
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
–
–
–
SRCTYPE
PRIOR
• PRIOR: Priority Level
Programs the priority level for all sources except FIQ source (source 0).
The priority level can be between 0 (lowest) and 7 (highest).
The priority level is not used for the FIQ in the related SMR register AIC_SMRx.
• SRCTYPE: Interrupt Source Type
The active level or edge is not programmable for the internal interrupt sources.
SRCTYPE
Internal Interrupt Sources
External Interrupt Sources
0
0
High level Sensitive
Low level Sensitive
0
1
Positive edge triggered
Negative edge triggered
1
0
High level Sensitive
High level Sensitive
1
1
Positive edge triggered
Positive edge triggered
385
6438D–ATARM–13-Oct-09
27.9.3
AIC Source Vector Register
Register Name:
AIC_SVR0..AIC_SVR31
Address:
0xFFFFF080
Access Type:
Read-write
Reset Value:
0x0
31
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
0
VECTOR
23
22
21
20
VECTOR
15
14
13
12
VECTOR
7
6
5
4
VECTOR
• VECTOR: Source Vector
The user may store in these registers the addresses of the corresponding handler for each interrupt source.
27.9.4
AIC Interrupt Vector Register
Register Name:
AIC_IVR
Address:
0xFFFFF100
Access Type:
Read-only
Reset Value:
0x0
31
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
0
IRQV
23
22
21
20
IRQV
15
14
13
12
IRQV
7
6
5
4
IRQV
• IRQV: Interrupt Vector Register
The Interrupt Vector Register contains the vector programmed by the user in the Source Vector Register corresponding to
the current interrupt.
The Source Vector Register is indexed using the current interrupt number when the Interrupt Vector Register is read.
When there is no current interrupt, the Interrupt Vector Register reads the value stored in AIC_SPU.
386
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
27.9.5
AIC FIQ Vector Register
Register Name:
AIC_FVR
Address:
0xFFFFF104
Access Type:
Read-only
Reset Value:
0x0
31
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
0
FIQV
23
22
21
20
FIQV
15
14
13
12
FIQV
7
6
5
4
FIQV
• FIQV: FIQ Vector Register
The FIQ Vector Register contains the vector programmed by the user in the Source Vector Register 0. When there is no
fast interrupt, the FIQ Vector Register reads the value stored in AIC_SPU.
27.9.6
AIC Interrupt Status Register
Register Name:
AIC_ISR
Address:
0xFFFFF108
Access Type:
Read-only
Reset Value:
0x0
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
–
–
–
IRQID
• IRQID: Current Interrupt Identifier
The Interrupt Status Register returns the current interrupt source number.
387
6438D–ATARM–13-Oct-09
27.9.7
AIC Interrupt Pending Register
Register Name:
AIC_IPR
Address:
0xFFFFF10C
Access Type:
Read-only
Reset Value:
0x0
31
30
29
28
27
26
25
24
PID31
PID30
PID29
PID28
PID27
PID26
PID25
PID24
23
22
21
20
19
18
17
16
PID23
PID22
PID21
PID20
PID19
PID18
PID17
PID16
15
14
13
12
11
10
9
8
PID15
PID14
PID13
PID12
PID11
PID10
PID9
PID8
7
6
5
4
3
2
1
0
PID7
PID6
PID5
PID4
PID3
PID2
SYS
FIQ
• FIQ, SYS, PID2-PID31: Interrupt Pending
0 = Corresponding interrupt is not pending.
1 = Corresponding interrupt is pending.
27.9.8
AIC Interrupt Mask Register
Register Name:
AIC_IMR
Address:
0xFFFFF110
Access Type:
Read-only
Reset Value:
0x0
31
30
29
28
27
26
25
24
PID31
PID30
PID29
PID28
PID27
PID26
PID25
PID24
23
22
21
20
19
18
17
16
PID23
PID22
PID21
PID20
PID19
PID18
PID17
PID16
15
14
13
12
11
10
9
8
PID15
PID14
PID13
PID12
PID11
PID10
PID9
PID8
7
6
5
4
3
2
1
0
PID7
PID6
PID5
PID4
PID3
PID2
SYS
FIQ
• FIQ, SYS, PID2-PID31: Interrupt Mask
0 = Corresponding interrupt is disabled.
1 = Corresponding interrupt is enabled.
388
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
27.9.9
AIC Core Interrupt Status Register
Register Name:
AIC_CISR
Address:
0xFFFFF114
Access Type:
Read-only
Reset Value:
0x0
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
–
–
–
–
–
–
NIRQ
NFIQ
• NFIQ: NFIQ Status
0 = nFIQ line is deactivated.
1 = nFIQ line is active.
• NIRQ: NIRQ Status
0 = nIRQ line is deactivated.
1 = nIRQ line is active.
27.9.10 AIC Interrupt Enable Command Register
Register Name:
AIC_IECR
Address:
0xFFFFF120
Access Type:
Write-only
31
30
29
28
27
26
25
24
PID31
PID30
PID29
PID28
PID27
PID26
PID25
PID24
23
22
21
20
19
18
17
16
PID23
PID22
PID21
PID20
PID19
PID18
PID17
PID16
15
14
13
12
11
10
9
8
PID15
PID14
PID13
PID12
PID11
PID10
PID9
PID8
7
6
5
4
3
2
1
0
PID7
PID6
PID5
PID4
PID3
PID2
SYS
FIQ
• FIQ, SYS, PID2-PID31: Interrupt Enable
0 = No effect.
1 = Enables corresponding interrupt.
389
6438D–ATARM–13-Oct-09
27.9.11 AIC Interrupt Disable Command Register
Register Name:
AIC_IDCR
Address:
0xFFFFF124
Access Type:
Write-only
31
30
29
28
27
26
25
24
PID31
PID30
PID29
PID28
PID27
PID26
PID25
PID24
23
22
21
20
19
18
17
16
PID23
PID22
PID21
PID20
PID19
PID18
PID17
PID16
15
14
13
12
11
10
9
8
PID15
PID14
PID13
PID12
PID11
PID10
PID9
PID8
7
6
5
4
3
2
1
0
PID7
PID6
PID5
PID4
PID3
PID2
SYS
FIQ
• FIQ, SYS, PID2-PID31: Interrupt Disable
0 = No effect.
1 = Disables corresponding interrupt.
27.9.12 AIC Interrupt Clear Command Register
Register Name:
AIC_ICCR
Address:
0xFFFFF128
Access Type:
Write-only
31
30
29
28
27
26
25
24
PID31
PID30
PID29
PID28
PID27
PID26
PID25
PID24
23
22
21
20
19
18
17
16
PID23
PID22
PID21
PID20
PID19
PID18
PID17
PID16
15
14
13
12
11
10
9
8
PID15
PID14
PID13
PID12
PID11
PID10
PID9
PID8
7
6
5
4
3
2
1
0
PID7
PID6
PID5
PID4
PID3
PID2
SYS
FIQ
• FIQ, SYS, PID2-PID31: Interrupt Clear
0 = No effect.
1 = Clears corresponding interrupt.
390
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
27.9.13 AIC Interrupt Set Command Register
Register Name:
AIC_ISCR
Address:
0xFFFFF12C
Access Type:
Write-only
31
30
29
28
27
26
25
24
PID31
PID30
PID29
PID28
PID27
PID26
PID25
PID24
23
22
21
20
19
18
17
16
PID23
PID22
PID21
PID20
PID19
PID18
PID17
PID16
15
14
13
12
11
10
9
8
PID15
PID14
PID13
PID12
PID11
PID10
PID9
PID8
7
6
5
4
3
2
1
0
PID7
PID6
PID5
PID4
PID3
PID2
SYS
FIQ
• FIQ, SYS, PID2-PID31: Interrupt Set
0 = No effect.
1 = Sets corresponding interrupt.
27.9.14 AIC End of Interrupt Command Register
Register Name:
AIC_EOICR
Address:
0xFFFFF130
Access Type:
Write-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
–
–
–
–
–
–
–
–
The End of Interrupt Command Register is used by the interrupt routine to indicate that the interrupt treatment is complete.
Any value can be written because it is only necessary to make a write to this register location to signal the end of interrupt
treatment.
391
6438D–ATARM–13-Oct-09
27.9.15 AIC Spurious Interrupt Vector Register
Register Name:
AIC_SPU
Address:
0xFFFFF134
Access Type:
Read-write
Reset Value:
0x0
31
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
0
SIVR
23
22
21
20
SIVR
15
14
13
12
SIVR
7
6
5
4
SIVR
• SIVR: Spurious Interrupt Vector Register
The user may store the address of a spurious interrupt handler in this register. The written value is returned in AIC_IVR in
case of a spurious interrupt and in AIC_FVR in case of a spurious fast interrupt.
27.9.16 AIC Debug Control Register
Register Name:
AIC_DCR
Address:
0xFFFFF138
Access Type:
Read-write
Reset Value:
0x0
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
–
–
–
–
–
–
GMSK
PROT
• PROT: Protection Mode
0 = The Protection Mode is disabled.
1 = The Protection Mode is enabled.
• GMSK: General Mask
0 = The nIRQ and nFIQ lines are normally controlled by the AIC.
1 = The nIRQ and nFIQ lines are tied to their inactive state.
392
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
27.9.17 AIC Fast Forcing Enable Register
Register Name:
AIC_FFER
Address:
0xFFFFF140
Access Type:
Write-only
31
30
29
28
27
26
25
24
PID31
PID30
PID29
PID28
PID27
PID26
PID25
PID24
23
22
21
20
19
18
17
16
PID23
PID22
PID21
PID20
PID19
PID18
PID17
PID16
15
14
13
12
11
10
9
8
PID15
PID14
PID13
PID12
PID11
PID10
PID9
PID8
7
6
5
4
3
2
1
0
PID7
PID6
PID5
PID4
PID3
PID2
SYS
–
• SYS, PID2-PID31: Fast Forcing Enable
0 = No effect.
1 = Enables the fast forcing feature on the corresponding interrupt.
27.9.18 AIC Fast Forcing Disable Register
Register Name:
AIC_FFDR
Address:
0xFFFFF144
Access Type:
Write-only
31
30
29
28
27
26
25
24
PID31
PID30
PID29
PID28
PID27
PID26
PID25
PID24
23
22
21
20
19
18
17
16
PID23
PID22
PID21
PID20
PID19
PID18
PID17
PID16
15
14
13
12
11
10
9
8
PID15
PID14
PID13
PID12
PID11
PID10
PID9
PID8
7
6
5
4
3
2
1
0
PID7
PID6
PID5
PID4
PID3
PID2
SYS
–
• SYS, PID2-PID31: Fast Forcing Disable
0 = No effect.
1 = Disables the Fast Forcing feature on the corresponding interrupt.
393
6438D–ATARM–13-Oct-09
AT91SAM9G45
27.9.19 AIC Fast Forcing Status Register
Register Name:
AIC_FFSR
Address:
0xFFFFF148
Access Type:
Read-only
31
30
29
28
27
26
25
24
PID31
PID30
PID29
PID28
PID27
PID26
PID25
PID24
23
22
21
20
19
18
17
16
PID23
PID22
PID21
PID20
PID19
PID18
PID17
PID16
15
14
13
12
11
10
9
8
PID15
PID14
PID13
PID12
PID11
PID10
PID9
PID8
7
6
5
4
3
2
1
0
PID7
PID6
PID5
PID4
PID3
PID2
SYS
–
• SYS, PID2-PID31: Fast Forcing Status
0 = The Fast Forcing feature is disabled on the corresponding interrupt.
1 = The Fast Forcing feature is enabled on the corresponding interrupt.
394
6438D–ATARM–13-Oct-09
AT91SAM9G45
28. Debug Unit (DBGU)
28.1
Description
The Debug Unit provides a single entry point from the processor for access to all the debug
capabilities of Atmel’s ARM-based systems.
The Debug Unit features a two-pin UART that can be used for several debug and trace purposes
and offers an ideal medium for in-situ programming solutions and debug monitor communications. The Debug Unit two-pin UART can be used stand-alone for general purpose serial
communication. Moreover, the association with two peripheral data controller channels permits
packet handling for these tasks with processor time reduced to a minimum.
The Debug Unit also makes the Debug Communication Channel (DCC) signals provided by the
In-circuit Emulator of the ARM processor visible to the software. These signals indicate the status of the DCC read and write registers and generate an interrupt to the ARM processor, making
possible the handling of the DCC under interrupt control.
Chip Identifier registers permit recognition of the device and its revision. These registers inform
as to the sizes and types of the on-chip memories, as well as the set of embedded peripherals.
Finally, the Debug Unit features a Force NTRST capability that enables the software to decide
whether to prevent access to the system via the In-circuit Emulator. This permits protection of
the code, stored in ROM.
28.2
Embedded Characteristics
• Composed of two functions
– Two-pin UART
– Debug Communication Channel (DCC) support
• Two-pin UART
– Implemented features are 100% compatible with the standard Atmel USART
– Independent receiver and transmitter with a common programmable Baud Rate
Generator
– Even, Odd, Mark or Space Parity Generation
– Parity, Framing and Overrun Error Detection
– Automatic Echo, Local Loopback and Remote Loopback Channel Modes
– Support for two PDC channels with connection to receiver and transmitter
• Debug Communication Channel Support
– Offers visibility of an interrupt trigger from COMMRX and COMMTX signals from the
ARM Processor’s ICE Interface
395
6438D–ATARM–13-Oct-09
28.3
Block Diagram
Figure 28-1. Debug Unit Functional Block Diagram
Peripheral
Bridge
Peripheral DMA Controller
APB
Debug Unit
DTXD
Transmit
Power
Management
Controller
MCK
Parallel
Input/
Output
Baud Rate
Generator
Receive
DRXD
COMMRX
ARM
Processor
COMMTX
DCC
Handler
Chip ID
nTRST
ICE
Access
Handler
Interrupt
Control
dbgu_irq
Power-on
Reset
force_ntrst
Table 28-1.
Debug Unit Pin Description
Pin Name
Description
Type
DRXD
Debug Receive Data
Input
DTXD
Debug Transmit Data
Output
Figure 28-2. Debug Unit Application Example
Boot Program
Debug Monitor
Trace Manager
Debug Unit
RS232 Drivers
Programming Tool
396
Debug Console
Trace Console
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
28.4
28.4.1
Product Dependencies
I/O Lines
Depending on product integration, the Debug Unit pins may be multiplexed with PIO lines. In this
case, the programmer must first configure the corresponding PIO Controller to enable I/O lines
operations of the Debug Unit.
Table 28-2.
I/O Lines
Instance
Signal
I/O Line
Peripheral
DBGU
DRXD
PB12
A
DBGU
DTXD
PB13
A
28.4.2
Power Management
Depending on product integration, the Debug Unit clock may be controllable through the Power
Management Controller. In this case, the programmer must first configure the PMC to enable the
Debug Unit clock. Usually, the peripheral identifier used for this purpose is 1.
28.4.3
Interrupt Source
Depending on product integration, the Debug Unit interrupt line is connected to one of the interrupt sources of the Advanced Interrupt Controller. Interrupt handling requires programming of
the AIC before configuring the Debug Unit. Usually, the Debug Unit interrupt line connects to the
interrupt source 1 of the AIC, which may be shared with the real-time clock, the system timer
interrupt lines and other system peripheral interrupts, as shown in Figure 28-1. This sharing
requires the programmer to determine the source of the interrupt when the source 1 is triggered.
28.5
UART Operations
The Debug Unit operates as a UART, (asynchronous mode only) and supports only 8-bit character handling (with parity). It has no clock pin.
The Debug Unit's UART is made up of a receiver and a transmitter that operate independently,
and a common baud rate generator. Receiver timeout and transmitter time guard are not implemented. However, all the implemented features are compatible with those of a standard USART.
28.5.1
Baud Rate Generator
The baud rate generator provides the bit period clock named baud rate clock to both the receiver
and the transmitter.
The baud rate clock is the master clock divided by 16 times the value (CD) written in
DBGU_BRGR (Baud Rate Generator Register). If DBGU_BRGR is set to 0, the baud rate clock
is disabled and the Debug Unit's UART remains inactive. The maximum allowable baud rate is
Master Clock divided by 16. The minimum allowable baud rate is Master Clock divided by (16 x
65536).
MCK
Baud Rate = ---------------------16 × CD
397
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 28-3. Baud Rate Generator
CD
CD
MCK
16-bit Counter
OUT
>1
1
0
Divide
by 16
Baud Rate
Clock
0
Receiver
Sampling Clock
28.5.2
28.5.2.1
Receiver
Receiver Reset, Enable and Disable
After device reset, the Debug Unit receiver is disabled and must be enabled before being used.
The receiver can be enabled by writing the control register DBGU_CR with the bit RXEN at 1. At
this command, the receiver starts looking for a start bit.
The programmer can disable the receiver by writing DBGU_CR with the bit RXDIS at 1. If the
receiver is waiting for a start bit, it is immediately stopped. However, if the receiver has already
detected a start bit and is receiving the data, it waits for the stop bit before actually stopping its
operation.
The programmer can also put the receiver in its reset state by writing DBGU_CR with the bit
RSTRX at 1. In doing so, the receiver immediately stops its current operations and is disabled,
whatever its current state. If RSTRX is applied when data is being processed, this data is lost.
28.5.2.2
Start Detection and Data Sampling
The Debug Unit only supports asynchronous operations, and this affects only its receiver. The
Debug Unit receiver detects the start of a received character by sampling the DRXD signal until
it detects a valid start bit. A low level (space) on DRXD is interpreted as a valid start bit if it is
detected for more than 7 cycles of the sampling clock, which is 16 times the baud rate. Hence, a
space that is longer than 7/16 of the bit period is detected as a valid start bit. A space which is
7/16 of a bit period or shorter is ignored and the receiver continues to wait for a valid start bit.
When a valid start bit has been detected, the receiver samples the DRXD at the theoretical midpoint of each bit. It is assumed that each bit lasts 16 cycles of the sampling clock (1-bit period)
so the bit sampling point is eight cycles (0.5-bit period) after the start of the bit. The first sampling
point is therefore 24 cycles (1.5-bit periods) after the falling edge of the start bit was detected.
Each subsequent bit is sampled 16 cycles (1-bit period) after the previous one.
398
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 28-4. Start Bit Detection
Sampling Clock
DRXD
True Start
Detection
D0
Baud Rate
Clock
Figure 28-5. Character Reception
Example: 8-bit, parity enabled 1 stop
0.5 bit
period
1 bit
period
DRXD
Sampling
28.5.2.3
D0
D1
True Start Detection
D2
D3
D4
D5
D6
Stop Bit
D7
Parity Bit
Receiver Ready
When a complete character is received, it is transferred to the DBGU_RHR and the RXRDY status bit in DBGU_SR (Status Register) is set. The bit RXRDY is automatically cleared when the
receive holding register DBGU_RHR is read.
Figure 28-6. Receiver Ready
DRXD
S
D0
D1
D2
D3
D4
D5
D6
D7
S
P
D0
D1
D2
D3
D4
D5
D6
D7
P
RXRDY
Read DBGU_RHR
28.5.2.4
Receiver Overrun
If DBGU_RHR has not been read by the software (or the Peripheral Data Controller) since the
last transfer, the RXRDY bit is still set and a new character is received, the OVRE status bit in
DBGU_SR is set. OVRE is cleared when the software writes the control register DBGU_CR with
the bit RSTSTA (Reset Status) at 1.
Figure 28-7. Receiver Overrun
DRXD
S
D0
D1
D2
D3
D4
D5
D6
D7
P
stop
S
D0
D1
D2
D3
D4
D5
D6
D7
P
stop
RXRDY
OVRE
RSTSTA
28.5.2.5
Parity Error
Each time a character is received, the receiver calculates the parity of the received data bits, in
accordance with the field PAR in DBGU_MR. It then compares the result with the received parity
399
6438D–ATARM–13-Oct-09
AT91SAM9G45
bit. If different, the parity error bit PARE in DBGU_SR is set at the same time the RXRDY is set.
The parity bit is cleared when the control register DBGU_CR is written with the bit RSTSTA
(Reset Status) at 1. If a new character is received before the reset status command is written,
the PARE bit remains at 1.
Figure 28-8. Parity Error
DRXD
S
D0
D1
D2
D3
D4
D5
D6
D7
P
stop
RXRDY
PARE
Wrong Parity Bit
28.5.2.6
RSTSTA
Receiver Framing Error
When a start bit is detected, it generates a character reception when all the data bits have been
sampled. The stop bit is also sampled and when it is detected at 0, the FRAME (Framing Error)
bit in DBGU_SR is set at the same time the RXRDY bit is set. The bit FRAME remains high until
the control register DBGU_CR is written with the bit RSTSTA at 1.
Figure 28-9. Receiver Framing Error
DRXD
S
D0
D1
D2
D3
D4
D5
D6
D7
P
stop
RXRDY
FRAME
Stop Bit
Detected at 0
28.5.3
28.5.3.1
RSTSTA
Transmitter
Transmitter Reset, Enable and Disable
After device reset, the Debug Unit transmitter is disabled and it must be enabled before being
used. The transmitter is enabled by writing the control register DBGU_CR with the bit TXEN at 1.
From this command, the transmitter waits for a character to be written in the Transmit Holding
Register DBGU_THR before actually starting the transmission.
The programmer can disable the transmitter by writing DBGU_CR with the bit TXDIS at 1. If the
transmitter is not operating, it is immediately stopped. However, if a character is being processed into the Shift Register and/or a character has been written in the Transmit Holding
Register, the characters are completed before the transmitter is actually stopped.
The programmer can also put the transmitter in its reset state by writing the DBGU_CR with the
bit RSTTX at 1. This immediately stops the transmitter, whether or not it is processing
characters.
28.5.3.2
Transmit Format
The Debug Unit transmitter drives the pin DTXD at the baud rate clock speed. The line is driven
depending on the format defined in the Mode Register and the data stored in the Shift Register.
One start bit at level 0, then the 8 data bits, from the lowest to the highest bit, one optional parity
bit and one stop bit at 1 are consecutively shifted out as shown on the following figure. The field
400
6438D–ATARM–13-Oct-09
AT91SAM9G45
PARE in the mode register DBGU_MR defines whether or not a parity bit is shifted out. When a
parity bit is enabled, it can be selected between an odd parity, an even parity, or a fixed space or
mark bit.
Figure 28-10. Character Transmission
Example: Parity enabled
Baud Rate
Clock
DTXD
Start
Bit
28.5.3.3
D0
D1
D2
D3
D4
D5
D6
D7
Parity
Bit
Stop
Bit
Transmitter Control
When the transmitter is enabled, the bit TXRDY (Transmitter Ready) is set in the status register
DBGU_SR. The transmission starts when the programmer writes in the Transmit Holding Register DBGU_THR, and after the written character is transferred from DBGU_THR to the Shift
Register. The bit TXRDY remains high until a second character is written in DBGU_THR. As
soon as the first character is completed, the last character written in DBGU_THR is transferred
into the shift register and TXRDY rises again, showing that the holding register is empty.
When both the Shift Register and the DBGU_THR are empty, i.e., all the characters written in
DBGU_THR have been processed, the bit TXEMPTY rises after the last stop bit has been
completed.
Figure 28-11. Transmitter Control
DBGU_THR
Data 0
Data 1
Shift Register
DTXD
Data 0
S
Data 0
Data 1
P
stop
S
Data 1
P
stop
TXRDY
TXEMPTY
Write Data 0
in DBGU_THR
28.5.4
Write Data 1
in DBGU_THR
Peripheral Data Controller
Both the receiver and the transmitter of the Debug Unit's UART are generally connected to a
Peripheral Data Controller (PDC) channel.
The peripheral data controller channels are programmed via registers that are mapped within
the Debug Unit user interface from the offset 0x100. The status bits are reported in the Debug
Unit status register DBGU_SR and can generate an interrupt.
401
6438D–ATARM–13-Oct-09
AT91SAM9G45
The RXRDY bit triggers the PDC channel data transfer of the receiver. This results in a read of
the data in DBGU_RHR. The TXRDY bit triggers the PDC channel data transfer of the transmitter. This results in a write of a data in DBGU_THR.
28.5.5
Test Modes
The Debug Unit supports three tests modes. These modes of operation are programmed by
using the field CHMODE (Channel Mode) in the mode register DBGU_MR.
The Automatic Echo mode allows bit-by-bit retransmission. When a bit is received on the DRXD
line, it is sent to the DTXD line. The transmitter operates normally, but has no effect on the
DTXD line.
The Local Loopback mode allows the transmitted characters to be received. DTXD and DRXD
pins are not used and the output of the transmitter is internally connected to the input of the
receiver. The DRXD pin level has no effect and the DTXD line is held high, as in idle state.
The Remote Loopback mode directly connects the DRXD pin to the DTXD line. The transmitter
and the receiver are disabled and have no effect. This mode allows a bit-by-bit retransmission.
Figure 28-12. Test Modes
Automatic Echo
RXD
Receiver
Transmitter
Disabled
TXD
Local Loopback
Disabled
Receiver
RXD
VDD
Disabled
Transmitter
Remote Loopback
Receiver
Transmitter
28.5.6
TXD
VDD
Disabled
Disabled
RXD
TXD
Debug Communication Channel Support
The Debug Unit handles the signals COMMRX and COMMTX that come from the Debug Communication Channel of the ARM Processor and are driven by the In-circuit Emulator.
402
6438D–ATARM–13-Oct-09
AT91SAM9G45
The Debug Communication Channel contains two registers that are accessible through the ICE
Breaker on the JTAG side and through the coprocessor 0 on the ARM Processor side.
As a reminder, the following instructions are used to read and write the Debug Communication
Channel:
MRC
p14, 0, Rd, c1, c0, 0
Returns the debug communication data read register into Rd
MCR
p14, 0, Rd, c1, c0, 0
Writes the value in Rd to the debug communication data write register.
The bits COMMRX and COMMTX, which indicate, respectively, that the read register has been
written by the debugger but not yet read by the processor, and that the write register has been
written by the processor and not yet read by the debugger, are wired on the two highest bits of
the status register DBGU_SR. These bits can generate an interrupt. This feature permits handling under interrupt a debug link between a debug monitor running on the target system and a
debugger.
28.5.7
Chip Identifier
The Debug Unit features two chip identifier registers, DBGU_CIDR (Chip ID Register) and
DBGU_EXID (Extension ID). Both registers contain a hard-wired value that is read-only. The first
register contains the following fields:
• EXT - shows the use of the extension identifier register
• NVPTYP and NVPSIZ - identifies the type of embedded non-volatile memory and its size
• ARCH - identifies the set of embedded peripherals
• SRAMSIZ - indicates the size of the embedded SRAM
• EPROC - indicates the embedded ARM processor
• VERSION - gives the revision of the silicon
The second register is device-dependent and reads 0 if the bit EXT is 0.
28.5.8
ICE Access Prevention
The Debug Unit allows blockage of access to the system through the ARM processor's ICE
interface. This feature is implemented via the register Force NTRST (DBGU_FNR), that allows
assertion of the NTRST signal of the ICE Interface. Writing the bit FNTRST (Force NTRST) to 1
in this register prevents any activity on the TAP controller.
On standard devices, the bit FNTRST resets to 0 and thus does not prevent ICE access.
This feature is especially useful on custom ROM devices for customers who do not want their
on-chip code to be visible.
403
6438D–ATARM–13-Oct-09
AT91SAM9G45
28.6
Debug Unit (DBGU) User Interface
Table 28-3.
Register Mapping
Offset
Register
Name
Access
Reset
0x0000
Control Register
DBGU_CR
Write-only
–
0x0004
Mode Register
DBGU_MR
Read-write
0x0
0x0008
Interrupt Enable Register
DBGU_IER
Write-only
–
0x000C
Interrupt Disable Register
DBGU_IDR
Write-only
–
0x0010
Interrupt Mask Register
DBGU_IMR
Read-only
0x0
0x0014
Status Register
DBGU_SR
Read-only
–
0x0018
Receive Holding Register
DBGU_RHR
Read-only
0x0
0x001C
Transmit Holding Register
DBGU_THR
Write-only
–
0x0020
Baud Rate Generator Register
DBGU_BRGR
Read-write
0x0
–
–
–
0x0024 - 0x003C
Reserved
0x0040
Chip ID Register
DBGU_CIDR
Read-only
–
0x0044
Chip ID Extension Register
DBGU_EXID
Read-only
–
0x0048
Force NTRST Register
DBGU_FNR
Read-write
0x0
–
–
–
0x0100 - 0x0124
PDC Area
404
6438D–ATARM–13-Oct-09
AT91SAM9G45
28.6.1
Name:
Debug Unit Control Register
DBGU_CR
Address:
0xFFFFEE00
Access Type:
Write-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
RSTSTA
7
6
5
4
3
2
1
0
TXDIS
TXEN
RXDIS
RXEN
RSTTX
RSTRX
–
–
• RSTRX: Reset Receiver
0 = No effect.
1 = The receiver logic is reset and disabled. If a character is being received, the reception is aborted.
• RSTTX: Reset Transmitter
0 = No effect.
1 = The transmitter logic is reset and disabled. If a character is being transmitted, the transmission is aborted.
• RXEN: Receiver Enable
0 = No effect.
1 = The receiver is enabled if RXDIS is 0.
• RXDIS: Receiver Disable
0 = No effect.
1 = The receiver is disabled. If a character is being processed and RSTRX is not set, the character is completed before the
receiver is stopped.
• TXEN: Transmitter Enable
0 = No effect.
1 = The transmitter is enabled if TXDIS is 0.
• TXDIS: Transmitter Disable
0 = No effect.
1 = The transmitter is disabled. If a character is being processed and a character has been written the DBGU_THR and
RSTTX is not set, both characters are completed before the transmitter is stopped.
• RSTSTA: Reset Status Bits
0 = No effect.
1 = Resets the status bits PARE, FRAME and OVRE in the DBGU_SR.
405
6438D–ATARM–13-Oct-09
AT91SAM9G45
28.6.2
Name:
Debug Unit Mode Register
DBGU_MR
Address:
0xFFFFEE04
Access Type:
Read-write
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
14
13
12
11
10
9
–
–
15
CHMODE
8
–
PAR
7
6
5
4
3
2
1
0
–
–
–
–
–
–
–
–
• PAR: Parity Type
PAR
Parity Type
0
0
0
Even parity
0
0
1
Odd parity
0
1
0
Space: parity forced to 0
0
1
1
Mark: parity forced to 1
1
x
x
No parity
• CHMODE: Channel Mode
CHMODE
Mode Description
0
0
Normal Mode
0
1
Automatic Echo
1
0
Local Loopback
1
1
Remote Loopback
406
6438D–ATARM–13-Oct-09
AT91SAM9G45
28.6.3
Name:
Debug Unit Interrupt Enable Register
DBGU_IER
Address:
0xFFFFEE08
Access Type:
Write-only
31
30
29
28
27
26
25
24
COMMRX
COMMTX
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
RXBUFF
TXBUFE
–
TXEMPTY
–
7
6
5
4
3
2
1
0
PARE
FRAME
OVRE
ENDTX
ENDRX
–
TXRDY
RXRDY
• RXRDY: Enable RXRDY Interrupt
• TXRDY: Enable TXRDY Interrupt
• ENDRX: Enable End of Receive Transfer Interrupt
• ENDTX: Enable End of Transmit Interrupt
• OVRE: Enable Overrun Error Interrupt
• FRAME: Enable Framing Error Interrupt
• PARE: Enable Parity Error Interrupt
• TXEMPTY: Enable TXEMPTY Interrupt
• TXBUFE: Enable Buffer Empty Interrupt
• RXBUFF: Enable Buffer Full Interrupt
• COMMTX: Enable COMMTX (from ARM) Interrupt
• COMMRX: Enable COMMRX (from ARM) Interrupt
0 = No effect.
1 = Enables the corresponding interrupt.
407
6438D–ATARM–13-Oct-09
AT91SAM9G45
28.6.4
Name:
Debug Unit Interrupt Disable Register
DBGU_IDR
Address:
0xFFFFEE0C
Access Type:
Write-only
31
30
29
28
27
26
25
24
COMMRX
COMMTX
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
RXBUFF
TXBUFE
–
TXEMPTY
–
7
6
5
4
3
2
1
0
PARE
FRAME
OVRE
ENDTX
ENDRX
–
TXRDY
RXRDY
• RXRDY: Disable RXRDY Interrupt
• TXRDY: Disable TXRDY Interrupt
• ENDRX: Disable End of Receive Transfer Interrupt
• ENDTX: Disable End of Transmit Interrupt
• OVRE: Disable Overrun Error Interrupt
• FRAME: Disable Framing Error Interrupt
• PARE: Disable Parity Error Interrupt
• TXEMPTY: Disable TXEMPTY Interrupt
• TXBUFE: Disable Buffer Empty Interrupt
• RXBUFF: Disable Buffer Full Interrupt
• COMMTX: Disable COMMTX (from ARM) Interrupt
• COMMRX: Disable COMMRX (from ARM) Interrupt
0 = No effect.
1 = Disables the corresponding interrupt.
408
6438D–ATARM–13-Oct-09
AT91SAM9G45
28.6.5
Name:
Debug Unit Interrupt Mask Register
DBGU_IMR
Address:
0xFFFFEE10
Access Type:
Read-only
31
30
29
28
27
26
25
24
COMMRX
COMMTX
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
RXBUFF
TXBUFE
–
TXEMPTY
–
7
6
5
4
3
2
1
0
PARE
FRAME
OVRE
ENDTX
ENDRX
–
TXRDY
RXRDY
• RXRDY: Mask RXRDY Interrupt
• TXRDY: Disable TXRDY Interrupt
• ENDRX: Mask End of Receive Transfer Interrupt
• ENDTX: Mask End of Transmit Interrupt
• OVRE: Mask Overrun Error Interrupt
• FRAME: Mask Framing Error Interrupt
• PARE: Mask Parity Error Interrupt
• TXEMPTY: Mask TXEMPTY Interrupt
• TXBUFE: Mask TXBUFE Interrupt
• RXBUFF: Mask RXBUFF Interrupt
• COMMTX: Mask COMMTX Interrupt
• COMMRX: Mask COMMRX Interrupt
0 = The corresponding interrupt is disabled.
1 = The corresponding interrupt is enabled.
409
6438D–ATARM–13-Oct-09
AT91SAM9G45
28.6.6
Name:
Debug Unit Status Register
DBGU_SR
Address:
0xFFFFEE14
Access Type:
Read-only
31
30
29
28
27
26
25
24
COMMRX
COMMTX
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
RXBUFF
TXBUFE
–
TXEMPTY
–
7
6
5
4
3
2
1
0
PARE
FRAME
OVRE
ENDTX
ENDRX
–
TXRDY
RXRDY
• RXRDY: Receiver Ready
0 = No character has been received since the last read of the DBGU_RHR or the receiver is disabled.
1 = At least one complete character has been received, transferred to DBGU_RHR and not yet read.
• TXRDY: Transmitter Ready
0 = A character has been written to DBGU_THR and not yet transferred to the Shift Register, or the transmitter is disabled.
1 = There is no character written to DBGU_THR not yet transferred to the Shift Register.
• ENDRX: End of Receiver Transfer
0 = The End of Transfer signal from the receiver Peripheral Data Controller channel is inactive.
1 = The End of Transfer signal from the receiver Peripheral Data Controller channel is active.
• ENDTX: End of Transmitter Transfer
0 = The End of Transfer signal from the transmitter Peripheral Data Controller channel is inactive.
1 = The End of Transfer signal from the transmitter Peripheral Data Controller channel is active.
• OVRE: Overrun Error
0 = No overrun error has occurred since the last RSTSTA.
1 = At least one overrun error has occurred since the last RSTSTA.
• FRAME: Framing Error
0 = No framing error has occurred since the last RSTSTA.
1 = At least one framing error has occurred since the last RSTSTA.
• PARE: Parity Error
0 = No parity error has occurred since the last RSTSTA.
1 = At least one parity error has occurred since the last RSTSTA.
• TXEMPTY: Transmitter Empty
0 = There are characters in DBGU_THR, or characters being processed by the transmitter, or the transmitter is disabled.
1 = There are no characters in DBGU_THR and there are no characters being processed by the transmitter.
410
6438D–ATARM–13-Oct-09
AT91SAM9G45
• TXBUFE: Transmission Buffer Empty
0 = The buffer empty signal from the transmitter PDC channel is inactive.
1 = The buffer empty signal from the transmitter PDC channel is active.
• RXBUFF: Receive Buffer Full
0 = The buffer full signal from the receiver PDC channel is inactive.
1 = The buffer full signal from the receiver PDC channel is active.
• COMMTX: Debug Communication Channel Write Status
0 = COMMTX from the ARM processor is inactive.
1 = COMMTX from the ARM processor is active.
• COMMRX: Debug Communication Channel Read Status
0 = COMMRX from the ARM processor is inactive.
1 = COMMRX from the ARM processor is active.
411
6438D–ATARM–13-Oct-09
AT91SAM9G45
28.6.7
Name:
Debug Unit Receiver Holding Register
DBGU_RHR
Address:
0xFFFFEE18
Access Type:
Read-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
RXCHR
• RXCHR: Received Character
Last received character if RXRDY is set.
28.6.8
Name:
Debug Unit Transmit Holding Register
DBGU_THR
Address:
0xFFFFEE1C
Access Type:
Write-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
TXCHR
• TXCHR: Character to be Transmitted
Next character to be transmitted after the current character if TXRDY is not set.
412
6438D–ATARM–13-Oct-09
AT91SAM9G45
28.6.9
Name:
Debug Unit Baud Rate Generator Register
DBGU_BRGR
Address:
0xFFFFEE20
Access Type:
Read-write
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
3
2
1
0
CD
7
6
5
4
CD
• CD: Clock Divisor
CD
Baud Rate Clock
0
Disabled
1
MCK
2 to 65535
MCK / (CD x 16)
413
6438D–ATARM–13-Oct-09
AT91SAM9G45
28.6.10
Name:
Debug Unit Chip ID Register
DBGU_CIDR
Address:
0xFFFFEE40
Access Type:
Read-only
31
30
29
EXT
23
28
27
26
NVPTYP
22
21
20
19
18
ARCH
15
14
13
6
24
17
16
9
8
1
0
SRAMSIZ
12
11
10
NVPSIZ2
7
25
ARCH
NVPSIZ
5
4
3
EPROC
2
VERSION
• VERSION: Version of the Device
Values depend upon the version of the device.
• EPROC: Embedded Processor
EPROC
Processor
0
0
1
ARM946ES
0
1
0
ARM7TDMI
1
0
0
ARM920T
1
0
1
ARM926EJS
• NVPSIZ: Nonvolatile Program Memory Size
NVPSIZ
Size
0
0
0
0
None
0
0
0
1
8K bytes
0
0
1
0
16K bytes
0
0
1
1
32K bytes
0
1
0
0
Reserved
0
1
0
1
64K bytes
0
1
1
0
Reserved
0
1
1
1
128K bytes
1
0
0
0
Reserved
1
0
0
1
256K bytes
1
0
1
0
512K bytes
1
0
1
1
Reserved
1
1
0
0
1024K bytes
414
6438D–ATARM–13-Oct-09
AT91SAM9G45
NVPSIZ
Size
1
1
0
1
Reserved
1
1
1
0
2048K bytes
1
1
1
1
Reserved
• NVPSIZ2 Second Nonvolatile Program Memory Size
NVPSIZ2
Size
0
0
0
0
None
0
0
0
1
8K bytes
0
0
1
0
16K bytes
0
0
1
1
32K bytes
0
1
0
0
Reserved
0
1
0
1
64K bytes
0
1
1
0
Reserved
0
1
1
1
128K bytes
1
0
0
0
Reserved
1
0
0
1
256K bytes
1
0
1
0
512K bytes
1
0
1
1
Reserved
1
1
0
0
1024K bytes
1
1
0
1
Reserved
1
1
1
0
2048K bytes
1
1
1
1
Reserved
• SRAMSIZ: Internal SRAM Size
SRAMSIZ
Size
0
0
0
0
Reserved
0
0
0
1
1K bytes
0
0
1
0
2K bytes
0
0
1
1
6K bytes
0
1
0
0
112K bytes
0
1
0
1
4K bytes
0
1
1
0
80K bytes
0
1
1
1
160K bytes
1
0
0
0
8K bytes
1
0
0
1
16K bytes
1
0
1
0
32K bytes
1
0
1
1
64K bytes
415
6438D–ATARM–13-Oct-09
AT91SAM9G45
SRAMSIZ
Size
1
1
0
0
128K bytes
1
1
0
1
256K bytes
1
1
1
0
96K bytes
1
1
1
1
512K bytes
• ARCH: Architecture Identifier
ARCH
Hex
Bin
Architecture
0x19
0001 1001
AT91SAM9xx Series
0x29
0010 1001
AT91SAM9XExx Series
0x34
0011 0100
AT91x34 Series
0x37
0011 0111
CAP7 Series
0x39
0011 1001
CAP9 Series
0x3B
0011 1011
CAP11 Series
0x40
0100 0000
AT91x40 Series
0x42
0100 0010
AT91x42 Series
0x55
0101 0101
AT91x55 Series
0x60
0110 0000
AT91SAM7Axx Series
0x61
0110 0001
AT91SAM7AQxx Series
0x63
0110 0011
AT91x63 Series
0x70
0111 0000
AT91SAM7Sxx Series
0x71
0111 0001
AT91SAM7XCxx Series
0x72
0111 0010
AT91SAM7SExx Series
0x73
0111 0011
AT91SAM7Lxx Series
0x75
0111 0101
AT91SAM7Xxx Series
0x92
1001 0010
AT91x92 Series
0xF0
1111 0000
AT75Cxx Series
• NVPTYP: Nonvolatile Program Memory Type
NVPTYP
Memory
0
0
0
ROM
0
0
1
ROMless or on-chip Flash
1
0
0
SRAM emulating ROM
0
1
0
Embedded Flash Memory
0
1
1
ROM and Embedded Flash Memory
NVPSIZ is ROM size
NVPSIZ2 is Flash size
416
6438D–ATARM–13-Oct-09
AT91SAM9G45
• EXT: Extension Flag
0 = Chip ID has a single register definition without extension
1 = An extended Chip ID exists.
417
6438D–ATARM–13-Oct-09
AT91SAM9G45
28.6.11
Name:
Debug Unit Chip ID Extension Register
DBGU_EXID
Address:
0xFFFFEE44
Access Type:
Read-only
31
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
0
EXID
23
22
21
20
EXID
15
14
13
12
EXID
7
6
5
4
EXID
• EXID: Chip ID Extension
Reads 0 if the bit EXT in DBGU_CIDR is 0.
28.6.12
Name:
Debug Unit Force NTRST Register
DBGU_FNR
Address:
0xFFFFEE48
Access Type:
Read-write
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
–
–
–
–
–
–
–
FNTRST
• FNTRST: Force NTRST
0 = NTRST of the ARM processor’s TAP controller is driven by the power_on_reset signal.
1 = NTRST of the ARM processor’s TAP controller is held low.
418
6438D–ATARM–13-Oct-09
AT91SAM9G45
29. Parallel Input/Output Controller (PIO)
29.1
Description
The Parallel Input/Output Controller (PIO) manages up to 32 fully programmable input/output
lines. Each I/O line may be dedicated as a general-purpose I/O or be assigned to a function of
an embedded peripheral. This assures effective optimization of the pins of a product.
Each I/O line is associated with a bit number in all of the 32-bit registers of the 32-bit wide User
Interface.
Each I/O line of the PIO Controller features:
• An input change interrupt enabling level change detection on any I/O line.
• A glitch filter providing rejection of pulses lower than one-half of clock cycle.
• Multi-drive capability similar to an open drain I/O line.
• Control of the the pull-up of the I/O line.
• Input visibility and output control.
The PIO Controller also features a synchronous output providing up to 32 bits of data output in a
single write operation.
419
6438D–ATARM–13-Oct-09
29.2
Block Diagram
Figure 29-1. Block Diagram
PIO Controller
AIC
PIO Interrupt
PIO Clock
PMC
Data, Enable
Up to 32
peripheral IOs
Embedded
Peripheral
PIN 0
Data, Enable
PIN 1
Up to 32 pins
Embedded
Peripheral
Up to 32
peripheral IOs
PIN 31
APB
Figure 29-2. Application Block Diagram
On-Chip Peripheral Drivers
Keyboard Driver
Control & Command
Driver
On-Chip Peripherals
PIO Controller
Keyboard Driver
420
General Purpose I/Os
External Devices
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
29.3
Product Dependencies
29.3.1
Pin Multiplexing
Each pin is configurable, according to product definition as either a general-purpose I/O line
only, or as an I/O line multiplexed with one or two peripheral I/Os. As the multiplexing is hardware-defined and thus product-dependent, the hardware designer and programmer must
carefully determine the configuration of the PIO controllers required by their application. When
an I/O line is general-purpose only, i.e. not multiplexed with any peripheral I/O, programming of
the PIO Controller regarding the assignment to a peripheral has no effect and only the PIO Controller can control how the pin is driven by the product.
29.3.2
External Interrupt Lines
The interrupt signals FIQ and IRQ0 to IRQn are most generally multiplexed through the PIO
Controllers. However, it is not necessary to assign the I/O line to the interrupt function as the
PIO Controller has no effect on inputs and the interrupt lines (FIQ or IRQs) are used only as
inputs.
29.3.3
Power Management
The Power Management Controller controls the PIO Controller clock in order to save power.
Writing any of the registers of the user interface does not require the PIO Controller clock to be
enabled. This means that the configuration of the I/O lines does not require the PIO Controller
clock to be enabled.
However, when the clock is disabled, not all of the features of the PIO Controller are available.
Note that the Input Change Interrupt and the read of the pin level require the clock to be
validated.
After a hardware reset, the PIO clock is disabled by default.
The user must configure the Power Management Controller before any access to the input line
information.
29.3.4
Interrupt Generation
For interrupt handling, the PIO Controllers are considered as user peripherals. This means that
the PIO Controller interrupt lines are connected among the interrupt sources 2 to 31. Refer to the
PIO Controller peripheral identifier in the product description to identify the interrupt sources
dedicated to the PIO Controllers.
The PIO Controller interrupt can be generated only if the PIO Controller clock is enabled.
421
6438D–ATARM–13-Oct-09
29.4
Functional Description
The PIO Controller features up to 32 fully-programmable I/O lines. Most of the control logic associated to each I/O is represented in Figure 29-3. In this description each signal shown
represents but one of up to 32 possible indexes.
Figure 29-3. I/O Line Control Logic
PIO_OER[0]
PIO_OSR[0]
PIO_PUER[0]
PIO_ODR[0]
PIO_PUSR[0]
PIO_PUDR[0]
1
Peripheral A
Output Enable
0
0
Peripheral B
Output Enable
0
1
PIO_ASR[0]
PIO_PER[0]
PIO_ABSR[0]
1
PIO_PSR[0]
PIO_BSR[0]
PIO_PDR[0]
Peripheral A
Output
0
Peripheral B
Output
1
PIO_MDER[0]
PIO_MDSR[0]
PIO_MDDR[0]
0
0
PIO_SODR[0]
PIO_ODSR[0]
1
Pad
PIO_CODR[0]
1
Peripheral A
Input
PIO_PDSR[0]
PIO_ISR[0]
0
Edge
Detector
Glitch
Filter
Peripheral B
Input
(Up to 32 possible inputs)
PIO Interrupt
1
PIO_IFER[0]
PIO_IFSR[0]
PIO_IFDR[0]
PIO_IER[0]
PIO_IMR[0]
PIO_IDR[0]
PIO_ISR[31]
PIO_IER[31]
PIO_IMR[31]
PIO_IDR[31]
422
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
29.4.1
Pull-up Resistor Control
Each I/O line is designed with an embedded pull-up resistor. The pull-up resistor can be enabled
or disabled by writing respectively PIO_PUER (Pull-up Enable Register) and PIO_PUDR (Pullup Disable Resistor). Writing in these registers results in setting or clearing the corresponding bit
in PIO_PUSR (Pull-up Status Register). Reading a 1 in PIO_PUSR means the pull-up is disabled and reading a 0 means the pull-up is enabled.
Control of the pull-up resistor is possible regardless of the configuration of the I/O line.
After reset, all of the pull-ups are enabled, i.e. PIO_PUSR resets at the value 0x0.
29.4.2
I/O Line or Peripheral Function Selection
When a pin is multiplexed with one or two peripheral functions, the selection is controlled with
the registers PIO_PER (PIO Enable Register) and PIO_PDR (PIO Disable Register). The register PIO_PSR (PIO Status Register) is the result of the set and clear registers and indicates
whether the pin is controlled by the corresponding peripheral or by the PIO Controller. A value of
0 indicates that the pin is controlled by the corresponding on-chip peripheral selected in the
PIO_ABSR (AB Select Status Register). A value of 1 indicates the pin is controlled by the PIO
controller.
If a pin is used as a general purpose I/O line (not multiplexed with an on-chip peripheral),
PIO_PER and PIO_PDR have no effect and PIO_PSR returns 1 for the corresponding bit.
After reset, most generally, the I/O lines are controlled by the PIO controller, i.e. PIO_PSR
resets at 1. However, in some events, it is important that PIO lines are controlled by the peripheral (as in the case of memory chip select lines that must be driven inactive after reset or for
address lines that must be driven low for booting out of an external memory). Thus, the reset
value of PIO_PSR is defined at the product level, depending on the multiplexing of the device.
29.4.3
Peripheral A or B Selection
The PIO Controller provides multiplexing of up to two peripheral functions on a single pin. The
selection is performed by writing PIO_ASR (A Select Register) and PIO_BSR (Select B Register). PIO_ABSR (AB Select Status Register) indicates which peripheral line is currently selected.
For each pin, the corresponding bit at level 0 means peripheral A is selected whereas the corresponding bit at level 1 indicates that peripheral B is selected.
Note that multiplexing of peripheral lines A and B only affects the output line. The peripheral
input lines are always connected to the pin input.
After reset, PIO_ABSR is 0, thus indicating that all the PIO lines are configured on peripheral A.
However, peripheral A generally does not drive the pin as the PIO Controller resets in I/O line
mode.
Writing in PIO_ASR and PIO_BSR manages PIO_ABSR regardless of the configuration of the
pin. However, assignment of a pin to a peripheral function requires a write in the corresponding
peripheral selection register (PIO_ASR or PIO_BSR) in addition to a write in PIO_PDR.
29.4.4
Output Control
When the I/0 line is assigned to a peripheral function, i.e. the corresponding bit in PIO_PSR is at
0, the drive of the I/O line is controlled by the peripheral. Peripheral A or B, depending on the
value in PIO_ABSR, determines whether the pin is driven or not.
When the I/O line is controlled by the PIO controller, the pin can be configured to be driven. This
is done by writing PIO_OER (Output Enable Register) and PIO_ODR (Output Disable Register).
423
6438D–ATARM–13-Oct-09
The results of these write operations are detected in PIO_OSR (Output Status Register). When
a bit in this register is at 0, the corresponding I/O line is used as an input only. When the bit is at
1, the corresponding I/O line is driven by the PIO controller.
The level driven on an I/O line can be determined by writing in PIO_SODR (Set Output Data
Register) and PIO_CODR (Clear Output Data Register). These write operations respectively set
and clear PIO_ODSR (Output Data Status Register), which represents the data driven on the I/O
lines. Writing in PIO_OER and PIO_ODR manages PIO_OSR whether the pin is configured to
be controlled by the PIO controller or assigned to a peripheral function. This enables configuration of the I/O line prior to setting it to be managed by the PIO Controller.
Similarly, writing in PIO_SODR and PIO_CODR effects PIO_ODSR. This is important as it
defines the first level driven on the I/O line.
29.4.5
Synchronous Data Output
Controlling all parallel busses using several PIOs requires two successive write operations in the
PIO_SODR and PIO_CODR registers. This may lead to unexpected transient values. The PIO
controller offers a direct control of PIO outputs by single write access to PIO_ODSR (Output
Data Status Register). Only bits unmasked by PIO_OWSR (Output Write Status Register) are
written. The mask bits in the PIO_OWSR are set by writing to PIO_OWER (Output Write Enable
Register) and cleared by writing to PIO_OWDR (Output Write Disable Register).
After reset, the synchronous data output is disabled on all the I/O lines as PIO_OWSR resets at
0x0.
29.4.6
Multi Drive Control (Open Drain)
Each I/O can be independently programmed in Open Drain by using the Multi Drive feature. This
feature permits several drivers to be connected on the I/O line which is driven low only by each
device. An external pull-up resistor (or enabling of the internal one) is generally required to guarantee a high level on the line.
The Multi Drive feature is controlled by PIO_MDER (Multi-driver Enable Register) and
PIO_MDDR (Multi-driver Disable Register). The Multi Drive can be selected whether the I/O line
is controlled by the PIO controller or assigned to a peripheral function. PIO_MDSR (Multi-driver
Status Register) indicates the pins that are configured to support external drivers.
After reset, the Multi Drive feature is disabled on all pins, i.e. PIO_MDSR resets at value 0x0.
29.4.7
424
Output Line Timings
Figure 29-4 shows how the outputs are driven either by writing PIO_SODR or PIO_CODR, or by
directly writing PIO_ODSR. This last case is valid only if the corresponding bit in PIO_OWSR is
set. Figure 29-4 also shows when the feedback in PIO_PDSR is available.
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 29-4. Output Line Timings
MCK
Write PIO_SODR
Write PIO_ODSR at 1
APB Access
Write PIO_CODR
Write PIO_ODSR at 0
APB Access
PIO_ODSR
2 cycles
2 cycles
PIO_PDSR
29.4.8
Inputs
The level on each I/O line can be read through PIO_PDSR (Pin Data Status Register). This register indicates the level of the I/O lines regardless of their configuration, whether uniquely as an
input or driven by the PIO controller or driven by a peripheral.
Reading the I/O line levels requires the clock of the PIO controller to be enabled, otherwise
PIO_PDSR reads the levels present on the I/O line at the time the clock was disabled.
29.4.9
Input Glitch Filtering
Optional input glitch filters are independently programmable on each I/O line. When the glitch filter is enabled, a glitch with a duration of less than 1/2 Master Clock (MCK) cycle is automatically
rejected, while a pulse with a duration of 1 Master Clock cycle or more is accepted. For pulse
durations between 1/2 Master Clock cycle and 1 Master Clock cycle the pulse may or may not
be taken into account, depending on the precise timing of its occurrence. Thus for a pulse to be
visible it must exceed 1 Master Clock cycle, whereas for a glitch to be reliably filtered out, its
duration must not exceed 1/2 Master Clock cycle. The filter introduces one Master Clock cycle
latency if the pin level change occurs before a rising edge. However, this latency does not
appear if the pin level change occurs before a falling edge. This is illustrated in Figure 29-5.
The glitch filters are controlled by the register set; PIO_IFER (Input Filter Enable Register),
PIO_IFDR (Input Filter Disable Register) and PIO_IFSR (Input Filter Status Register). Writing
PIO_IFER and PIO_IFDR respectively sets and clears bits in PIO_IFSR. This last register
enables the glitch filter on the I/O lines.
When the glitch filter is enabled, it does not modify the behavior of the inputs on the peripherals.
It acts only on the value read in PIO_PDSR and on the input change interrupt detection. The
glitch filters require that the PIO Controller clock is enabled.
425
6438D–ATARM–13-Oct-09
Figure 29-5. Input Glitch Filter Timing
MCK
up to 1.5 cycles
Pin Level
1 cycle
1 cycle
1 cycle
1 cycle
PIO_PDSR
if PIO_IFSR = 0
2 cycles
PIO_PDSR
if PIO_IFSR = 1
29.4.10
up to 2.5 cycles
1 cycle
up to 2 cycles
Input Change Interrupt
The PIO Controller can be programmed to generate an interrupt when it detects an input change
on an I/O line. The Input Change Interrupt is controlled by writing PIO_IER (Interrupt Enable
Register) and PIO_IDR (Interrupt Disable Register), which respectively enable and disable the
input change interrupt by setting and clearing the corresponding bit in PIO_IMR (Interrupt Mask
Register). As Input change detection is possible only by comparing two successive samplings of
the input of the I/O line, the PIO Controller clock must be enabled. The Input Change Interrupt is
available, regardless of the configuration of the I/O line, i.e. configured as an input only, controlled by the PIO Controller or assigned to a peripheral function.
When an input change is detected on an I/O line, the corresponding bit in PIO_ISR (Interrupt
Status Register) is set. If the corresponding bit in PIO_IMR is set, the PIO Controller interrupt
line is asserted. The interrupt signals of the thirty-two channels are ORed-wired together to generate a single interrupt signal to the Advanced Interrupt Controller.
When the software reads PIO_ISR, all the interrupts are automatically cleared. This signifies that
all the interrupts that are pending when PIO_ISR is read must be handled.
Figure 29-6. Input Change Interrupt Timings
MCK
Pin Level
PIO_ISR
Read PIO_ISR
29.4.11
426
APB Access
APB Access
Write Protected Registers
To prevent any single software error that may corrupt the PIO behavior, the registers listed
below can be write-protected by setting the WPEN bit in the PIO Write Protect Mode Register
(PIO_WPMR).
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
If a write access in a write-protected register is detected, then the WPVS flag in the PIO Write
Protect Status Register (PIO_WPSR) is set and the field WPVSRC indicates in which register
the write access has been attempted.
The WPVS flag is automatically reset after reading the PIO Write Protect Status Register
(PIO_WPSR).
List of the write-protected registers:
• “PIO Enable Register” on page 432
• “PIO Disable Register” on page 432
• “PIO Output Enable Register” on page 433
• “PIO Output Disable Register” on page 434
• “PIO Input Filter Enable Register” on page 435
• “PIO Input Filter Disable Register” on page 435
• “PIO Set Output Data Register” on page 436
• “PIO Clear Output Data Register” on page 437
• “PIO Multi-driver Enable Register” on page 440
• “PIO Multi-driver Disable Register” on page 441
• “PIO Pull Up Disable Register” on page 442
• “PIO Pull Up Enable Register” on page 442
• “PIO Peripheral A Select Register” on page 443
• “PIO Peripheral B Select Register” on page 444
• “PIO Output Write Enable Register” on page 445
• “PIO Output Write Disable Register” on page 445
29.4.12
Programmable I/O Delays
The PIO interface consists of a series of signals driven by peripherals or directly by sofware. The
simultaneous switching outputs on these busses may lead to a peak of current in the internal
and external power supply lines.
In order to reduce the peak of current in such cases, additional propagation delays can be
adjusted independently for pad buffers by means of configuration registers, PIO_DELAY.
For each I/O, the additional programmable delays range from 0 to 4 ns (Worst Case PVT). The
delay can differ between IOs supporting this feature. The delay can be modified according to
programming for each I/O. The minimum additional delay that can be programmed on a PAD
supporting this feature is 1/16 of the maximum programmable delay.
Only PADs PC[12], PC[7:2], PA[30:23] and PA[9:2] can be configured.
When programming 0x0 in fields, no delay is added (reset value) and the propagation delay of
the pad buffers is the inherent delay of the pad buffer. When programming 0xF in field, the propagation delay of the corresponding pad is maximal.
427
6438D–ATARM–13-Oct-09
Figure 29-7. Programmable I/O Delays
PIO
PAin[0]
PAout[0]
Programmable Delay Line
DELAY1
PAin[1]
PAout[1]
Programmable Delay Line
DELAY2
PAin[2]
PAout[2]
Programmable Delay Line
DELAYx
29.5
I/O Lines Programming Example
The programing example as shown in Table 29-1 below is used to define the following
configuration.
• 4-bit output port on I/O lines 0 to 3, (should be written in a single write operation), open-drain,
with pull-up resistor
• Four output signals on I/O lines 4 to 7 (to drive LEDs for example), driven high and low, no
pull-up resistor
• Four input signals on I/O lines 8 to 11 (to read push-button states for example), with pull-up
resistors, glitch filters and input change interrupts
• Four input signals on I/O line 12 to 15 to read an external device status (polled, thus no input
change interrupt), no pull-up resistor, no glitch filter
• I/O lines 16 to 19 assigned to peripheral A functions with pull-up resistor
• I/O lines 20 to 23 assigned to peripheral B functions, no pull-up resistor
• I/O line 24 to 27 assigned to peripheral A with Input Change Interrupt and pull-up resistor
Table 29-1.
428
Programming Example
Register
Value to be Written
PIO_PER
0x0000 FFFF
PIO_PDR
0x0FFF 0000
PIO_OER
0x0000 00FF
PIO_ODR
0x0FFF FF00
PIO_IFER
0x0000 0F00
PIO_IFDR
0x0FFF F0FF
PIO_SODR
0x0000 0000
PIO_CODR
0x0FFF FFFF
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Table 29-1.
Programming Example (Continued)
PIO_IER
0x0F00 0F00
PIO_IDR
0x00FF F0FF
PIO_MDER
0x0000 000F
PIO_MDDR
0x0FFF FFF0
PIO_PUDR
0x00F0 00F0
PIO_PUER
0x0F0F FF0F
PIO_ASR
0x0F0F 0000
PIO_BSR
0x00F0 0000
PIO_OWER
0x0000 000F
PIO_OWDR
0x0FFF FFF0
429
6438D–ATARM–13-Oct-09
29.6
Parallel Input/Output Controller (PIO) User Interface
Each I/O line controlled by the PIO Controller is associated with a bit in each of the PIO Controller User Interface registers. Each register is 32 bits wide. If a parallel I/O line is not defined,
writing to the corresponding bits has no effect. Undefined bits read zero. If the I/O line is not multiplexed with any peripheral, the I/O line is controlled by the PIO Controller and PIO_PSR returns
1 systematically.
Table 29-2.
Register Mapping
Offset
Register
Name
Access
Reset
0x0000
PIO Enable Register
PIO_PER
Write-only
–
0x0004
PIO Disable Register
PIO_PDR
Write-only
–
PIO_PSR
Read-only
(1)
0x0008
PIO Status Register
0x000C
Reserved
0x0010
Output Enable Register
PIO_OER
Write-only
–
0x0014
Output Disable Register
PIO_ODR
Write-only
–
0x0018
Output Status Register
PIO_OSR
Read-only
0x0000 0000
0x001C
Reserved
0x0020
Glitch Input Filter Enable Register
PIO_IFER
Write-only
–
0x0024
Glitch Input Filter Disable Register
PIO_IFDR
Write-only
–
0x0028
Glitch Input Filter Status Register
PIO_IFSR
Read-only
0x0000 0000
0x002C
Reserved
0x0030
Set Output Data Register
PIO_SODR
Write-only
–
0x0034
Clear Output Data Register
PIO_CODR
Write-only
0x0038
Output Data Status Register
PIO_ODSR
Read-only
or(2)
Read/Write
–
0x003C
Pin Data Status Register
PIO_PDSR
Read-only
(3)
0x0040
Interrupt Enable Register
PIO_IER
Write-only
–
0x0044
Interrupt Disable Register
PIO_IDR
Write-only
–
0x0048
Interrupt Mask Register
PIO_IMR
Read-only
0x00000000
0x004C
Interrupt Status Register(4)
PIO_ISR
Read-only
0x00000000
0x0050
Multi-driver Enable Register
PIO_MDER
Write-only
–
0x0054
Multi-driver Disable Register
PIO_MDDR
Write-only
–
0x0058
Multi-driver Status Register
PIO_MDSR
Read-only
0x00000000
0x005C
Reserved
0x0060
Pull-up Disable Register
PIO_PUDR
Write-only
–
0x0064
Pull-up Enable Register
PIO_PUER
Write-only
–
0x0068
Pad Pull-up Status Register
PIO_PUSR
Read-only
0x00000000
0x006C
Reserved
430
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Table 29-2.
Register Mapping (Continued)
Offset
Register
0x0070
0x0074
Name
Peripheral A Select Register
(5)
Peripheral B Select Register
(5)
(5)
Access
Reset
PIO_ASR
Write-only
–
PIO_BSR
Write-only
–
PIO_ABSR
Read-only
0x00000000
0x0078
AB Status Register
0x007C-0x009C
Reserved
0x00A0
Output Write Enable
PIO_OWER
Write-only
–
0x00A4
Output Write Disable
PIO_OWDR
Write-only
–
0x00A8
Output Write Status Register
PIO_OWSR
Read-only
0x00000000
0x00AC
Reserved
0x00C0
I/O Delay Register
PIO_DELAY0R
Read/Write
0x00000000
0x00C4
I/O Delay Register
PIO_DELAY1R
Read/Write
0x00000000
0x00C8
I/O Delay Register
PIO_DELAY2R
Read/Write
0x00000000
0x00CC
I/O Delay Register
PIO_DELAY3R
Read/Write
0x00000000
0x00C4-00E0
Reserved
0x00E4
Write Protect Mode Register
PIO_WPMR
Read-write
0x00000000
0x00E8
Write Protect Status Register
PIO_WPSR
Read-only
0x00000000
0x00F0-0x00F8
Reserved
Notes:
1. Reset value of PIO_PSR depends on the product implementation.
2. PIO_ODSR is Read-only or Read/Write depending on PIO_OWSR I/O lines.
3. Reset value of PIO_PDSR depends on the level of the I/O lines. Reading the I/O line levels requires the clock of the PIO
Controller to be enabled, otherwise PIO_PDSR reads the levels present on the I/O line at the time the clock was disabled.
4. PIO_ISR is reset at 0x0. However, the first read of the register may read a different value as input changes may have
occurred.
5. Only this set of registers clears the status by writing 1 in the first register and sets the status by writing 1 in the second
register.
431
6438D–ATARM–13-Oct-09
29.6.1
Name:
PIO Enable Register
PIO_PER
Addresses:
0xFFFFF200 (PIOA), 0xFFFFF400 (PIOB), 0xFFFFF600 (PIOC), 0xFFFFF800 (PIOD),
0xFFFFFA00 (PIOE)
Access Type:
Write-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: PIO Enable
0 = No effect.
1 = Enables the PIO to control the corresponding pin (disables peripheral control of the pin).
29.6.2
Name:
PIO Disable Register
PIO_PDR
Addresses:
0xFFFFF204 (PIOA), 0xFFFFF404 (PIOB), 0xFFFFF604 (PIOC), 0xFFFFF804 (PIOD),
0xFFFFFA04 (PIOE)
Access Type:
Write-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: PIO Disable
0 = No effect.
1 = Disables the PIO from controlling the corresponding pin (enables peripheral control of the pin).
432
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
29.6.3
Name:
PIO Status Register
PIO_PSR
Addresses:
0xFFFFF208 (PIOA), 0xFFFFF408 (PIOB), 0xFFFFF608 (PIOC), 0xFFFFF808 (PIOD),
0xFFFFFA08 (PIOE)
Access Type:
Read-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: PIO Status
0 = PIO is inactive on the corresponding I/O line (peripheral is active).
1 = PIO is active on the corresponding I/O line (peripheral is inactive).
29.6.4
Name:
PIO Output Enable Register
PIO_OER
Addresses:
0xFFFFF210 (PIOA), 0xFFFFF410 (PIOB), 0xFFFFF610 (PIOC), 0xFFFFF810 (PIOD),
0xFFFFFA10 (PIOE)
Access Type:
Write-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Output Enable
0 = No effect.
1 = Enables the output on the I/O line.
433
6438D–ATARM–13-Oct-09
29.6.5
Name:
PIO Output Disable Register
PIO_ODR
Addresses:
0xFFFFF214 (PIOA), 0xFFFFF414 (PIOB), 0xFFFFF614 (PIOC), 0xFFFFF814 (PIOD),
0xFFFFFA14 (PIOE)
Access Type:
Write-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Output Disable
0 = No effect.
1 = Disables the output on the I/O line.
29.6.6
Name:
PIO Output Status Register
PIO_OSR
Addresses:
0xFFFFF218 (PIOA), 0xFFFFF418 (PIOB), 0xFFFFF618 (PIOC), 0xFFFFF818 (PIOD),
0xFFFFFA18 (PIOE)
Access Type:
Read-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Output Status
0 = The I/O line is a pure input.
1 = The I/O line is enabled in output.
434
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
29.6.7
Name:
PIO Input Filter Enable Register
PIO_IFER
Addresses:
0xFFFFF220 (PIOA), 0xFFFFF420 (PIOB), 0xFFFFF620 (PIOC), 0xFFFFF820 (PIOD),
0xFFFFFA20 (PIOE)
Access Type:
Write-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Input Filter Enable
0 = No effect.
1 = Enables the input glitch filter on the I/O line.
29.6.8
Name:
PIO Input Filter Disable Register
PIO_IFDR
Addresses:
0xFFFFF224 (PIOA), 0xFFFFF424 (PIOB), 0xFFFFF624 (PIOC), 0xFFFFF824 (PIOD),
0xFFFFFA24 (PIOE)
Access Type:
Write-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Input Filter Disable
0 = No effect.
1 = Disables the input glitch filter on the I/O line.
435
6438D–ATARM–13-Oct-09
29.6.9
Name:
PIO Input Filter Status Register
PIO_IFSR
Addresses:
0xFFFFF228 (PIOA), 0xFFFFF428 (PIOB), 0xFFFFF628 (PIOC), 0xFFFFF828 (PIOD),
0xFFFFFA28 (PIOE)
Access Type:
Read-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Input Filer Status
0 = The input glitch filter is disabled on the I/O line.
1 = The input glitch filter is enabled on the I/O line.
29.6.10
Name:
PIO Set Output Data Register
PIO_SODR
Addresses:
0xFFFFF230 (PIOA), 0xFFFFF430 (PIOB), 0xFFFFF630 (PIOC), 0xFFFFF830 (PIOD),
0xFFFFFA30 (PIOE)
Access Type:
Write-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Set Output Data
0 = No effect.
1 = Sets the data to be driven on the I/O line.
436
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
29.6.11
Name:
PIO Clear Output Data Register
PIO_CODR
Addresses:
0xFFFFF234 (PIOA), 0xFFFFF434 (PIOB), 0xFFFFF634 (PIOC), 0xFFFFF834 (PIOD),
0xFFFFFA34 (PIOE)
Access Type:
Write-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Clear Output Data
0 = No effect.
1 = Clears the data to be driven on the I/O line.
29.6.12
Name:
PIO Output Data Status Register
PIO_ODSR
Addresses:
0xFFFFF238 (PIOA), 0xFFFFF438 (PIOB), 0xFFFFF638 (PIOC), 0xFFFFF838 (PIOD),
0xFFFFFA38 (PIOE)
Access Type:
Read-only or Read/Write
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Output Data Status
0 = The data to be driven on the I/O line is 0.
1 = The data to be driven on the I/O line is 1.
437
6438D–ATARM–13-Oct-09
29.6.13
Name:
PIO Pin Data Status Register
PIO_PDSR
Addresses:
0xFFFFF23C (PIOA), 0xFFFFF43C (PIOB), 0xFFFFF63C (PIOC), 0xFFFFF83C (PIOD),
0xFFFFFA3C (PIOE)
Access Type:
Read-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Output Data Status
0 = The I/O line is at level 0.
1 = The I/O line is at level 1.
29.6.14
Name:
PIO Interrupt Enable Register
PIO_IER
Addresses:
0xFFFFF240 (PIOA), 0xFFFFF440 (PIOB), 0xFFFFF640 (PIOC), 0xFFFFF840 (PIOD),
0xFFFFFA40 (PIOE)
Access Type:
Write-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Input Change Interrupt Enable
0 = No effect.
1 = Enables the Input Change Interrupt on the I/O line.
438
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
29.6.15
Name:
PIO Interrupt Disable Register
PIO_IDR
Addresses:
0xFFFFF244 (PIOA), 0xFFFFF444 (PIOB), 0xFFFFF644 (PIOC), 0xFFFFF844 (PIOD),
0xFFFFFA44 (PIOE)
Access Type:
Write-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Input Change Interrupt Disable
0 = No effect.
1 = Disables the Input Change Interrupt on the I/O line.
29.6.16
Name:
PIO Interrupt Mask Register
PIO_IMR
Addresses:
0xFFFFF248 (PIOA), 0xFFFFF448 (PIOB), 0xFFFFF648 (PIOC), 0xFFFFF848 (PIOD),
0xFFFFFA48 (PIOE)
Access Type:
Read-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Input Change Interrupt Mask
0 = Input Change Interrupt is disabled on the I/O line.
1 = Input Change Interrupt is enabled on the I/O line.
439
6438D–ATARM–13-Oct-09
29.6.17
Name:
PIO Interrupt Status Register
PIO_ISR
Addresses:
0xFFFFF24C (PIOA), 0xFFFFF44C (PIOB), 0xFFFFF64C (PIOC), 0xFFFFF84C (PIOD),
0xFFFFFA4C (PIOE)
Access Type:
Read-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Input Change Interrupt Status
0 = No Input Change has been detected on the I/O line since PIO_ISR was last read or since reset.
1 = At least one Input Change has been detected on the I/O line since PIO_ISR was last read or since reset.
29.6.18
Name:
PIO Multi-driver Enable Register
PIO_MDER
Addresses:
0xFFFFF250 (PIOA), 0xFFFFF450 (PIOB), 0xFFFFF650 (PIOC), 0xFFFFF850 (PIOD),
0xFFFFFA50 (PIOE)
Access Type:
Write-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Multi Drive Enable.
0 = No effect.
1 = Enables Multi Drive on the I/O line.
440
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
29.6.19
Name:
PIO Multi-driver Disable Register
PIO_MDDR
Addresses:
0xFFFFF254 (PIOA), 0xFFFFF454 (PIOB), 0xFFFFF654 (PIOC), 0xFFFFF854 (PIOD),
0xFFFFFA54 (PIOE)
Access Type:
Write-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Multi Drive Disable.
0 = No effect.
1 = Disables Multi Drive on the I/O line.
29.6.20
Name:
PIO Multi-driver Status Register
PIO_MDSR
Addresses:
0xFFFFF258 (PIOA), 0xFFFFF458 (PIOB), 0xFFFFF658 (PIOC), 0xFFFFF858 (PIOD),
0xFFFFFA58 (PIOE)
Access Type:
Read-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Multi Drive Status.
0 = The Multi Drive is disabled on the I/O line. The pin is driven at high and low level.
1 = The Multi Drive is enabled on the I/O line. The pin is driven at low level only.
441
6438D–ATARM–13-Oct-09
AT91SAM9G45
29.6.21
Name:
PIO Pull Up Disable Register
PIO_PUDR
Addresses:
0xFFFFF260 (PIOA), 0xFFFFF460 (PIOB), 0xFFFFF660 (PIOC), 0xFFFFF860 (PIOD),
0xFFFFFA60 (PIOE)
Access Type:
Write-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Pull Up Disable.
0 = No effect.
1 = Disables the pull up resistor on the I/O line.
29.6.22
Name:
PIO Pull Up Enable Register
PIO_PUER
Addresses:
0xFFFFF264 (PIOA), 0xFFFFF464 (PIOB), 0xFFFFF664 (PIOC), 0xFFFFF864 (PIOD),
0xFFFFFA64 (PIOE)
Access Type:
Write-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Pull Up Enable.
0 = No effect.
1 = Enables the pull up resistor on the I/O line.
442
6438D–ATARM–13-Oct-09
AT91SAM9G45
29.6.23
Name:
PIO Pull Up Status Register
PIO_PUSR
Addresses:
0xFFFFF268 (PIOA), 0xFFFFF468 (PIOB), 0xFFFFF668 (PIOC), 0xFFFFF868 (PIOD),
0xFFFFFA68 (PIOE)
Access Type:
Read-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Pull Up Status.
0 = Pull Up resistor is enabled on the I/O line.
1 = Pull Up resistor is disabled on the I/O line.
29.6.24
Name:
PIO Peripheral A Select Register
PIO_ASR
Addresses:
0xFFFFF270 (PIOA), 0xFFFFF470 (PIOB), 0xFFFFF670 (PIOC), 0xFFFFF870 (PIOD),
0xFFFFFA70 (PIOE)
Access Type:
Write-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Peripheral A Select.
0 = No effect.
1 = Assigns the I/O line to the Peripheral A function.
443
6438D–ATARM–13-Oct-09
AT91SAM9G45
29.6.25
Name:
PIO Peripheral B Select Register
PIO_BSR
Addresses:
0xFFFFF274 (PIOA), 0xFFFFF474 (PIOB), 0xFFFFF674 (PIOC), 0xFFFFF874 (PIOD),
0xFFFFFA74 (PIOE)
Access Type:
Write-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Peripheral B Select.
0 = No effect.
1 = Assigns the I/O line to the peripheral B function.
29.6.26
Name:
PIO Peripheral A B Status Register
PIO_ABSR
Addresses:
0xFFFFF278 (PIOA), 0xFFFFF478 (PIOB), 0xFFFFF678 (PIOC), 0xFFFFF878 (PIOD),
0xFFFFFA78 (PIOE)
Access Type:
Read-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Peripheral A B Status.
0 = The I/O line is assigned to the Peripheral A.
1 = The I/O line is assigned to the Peripheral B.
444
6438D–ATARM–13-Oct-09
AT91SAM9G45
29.6.27
Name:
PIO Output Write Enable Register
PIO_OWER
Addresses:
0xFFFFF2A0 (PIOA), 0xFFFFF4A0 (PIOB), 0xFFFFF6A0 (PIOC), 0xFFFFF8A0 (PIOD),
0xFFFFFAA0 (PIOE)
Access Type:
Write-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Output Write Enable.
0 = No effect.
1 = Enables writing PIO_ODSR for the I/O line.
29.6.28
Name:
PIO Output Write Disable Register
PIO_OWDR
Addresses:
0xFFFFF2A4 (PIOA), 0xFFFFF4A4 (PIOB), 0xFFFFF6A4 (PIOC), 0xFFFFF8A4 (PIOD),
0xFFFFFAA4 (PIOE)
Access Type:
Write-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Output Write Disable.
0 = No effect.
1 = Disables writing PIO_ODSR for the I/O line.
445
6438D–ATARM–13-Oct-09
AT91SAM9G45
29.6.29
Name:
PIO Output Write Status Register
PIO_OWSR
Addresses:
0xFFFFF2A8 (PIOA), 0xFFFFF4A8 (PIOB), 0xFFFFF6A8 (PIOC), 0xFFFFF8A8 (PIOD),
0xFFFFFAA8 (PIOE)
Access Type:
Read-only
31
30
29
28
27
26
25
24
P31
P30
P29
P28
P27
P26
P25
P24
23
22
21
20
19
18
17
16
P23
P22
P21
P20
P19
P18
P17
P16
15
14
13
12
11
10
9
8
P15
P14
P13
P12
P11
P10
P9
P8
7
6
5
4
3
2
1
0
P7
P6
P5
P4
P3
P2
P1
P0
• P0-P31: Output Write Status.
0 = Writing PIO_ODSR does not affect the I/O line.
1 = Writing PIO_ODSR affects the I/O line.
29.6.30 PIO I/O Delay Register
Register Name:
PIO_DELAYxR [x=0..3]
Addresses:
0xFFFFF2C0 (PIOA), 0xFFFFF4C0 (PIOB), 0xFFFFF6C0 (PIOC), 0xFFFFF8C0 (PIOD),
0xFFFFFAC0 (PIOE)
Access Type:
Read-write
Reset Value:
See Figure 29-2
31
30
29
28
27
26
Delay7
23
22
21
20
19
18
Delay5
15
14
13
6
24
17
16
9
8
1
0
Delay4
12
11
10
Delay3
7
25
Delay6
Delay2
5
4
Delay1
3
2
Delay0
• Delay x:
Gives the number of elements in the delay line associated to pad x.
446
6438D–ATARM–13-Oct-09
AT91SAM9G45
29.6.31 PIO Write Protect Mode Register
Register Name:
PIO_WPMR
Addresses:
0xFFFFF2E4 (PIOA), 0xFFFFF4E4 (PIOB), 0xFFFFF6E4 (PIOC), 0xFFFFF8E4 (PIOD),
0xFFFFFAE4 (PIOE)
Access Type:
Read-write
Reset Value:
See Table 29-2
31
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
WPKEY
23
22
21
20
WPKEY
15
14
13
12
WPKEY
7
6
5
4
3
2
1
0
—
—
—
—
—
—
—
WPEN
• WPEN: Write Protect Enable
0 = Disables the Write Protect if WPKEY corresponds to 0x50494F (“PIO” in ASCII).
1 = Enables the Write Protect if WPKEY corresponds to 0x50494F (“PIO” in ASCII).
Protects the registers listed below:
• “PIO Enable Register” on page 432
• “PIO Disable Register” on page 432
• “PIO Output Enable Register” on page 433
• “PIO Output Disable Register” on page 434
• “PIO Input Filter Enable Register” on page 435
• “PIO Input Filter Disable Register” on page 435
• “PIO Set Output Data Register” on page 436
• “PIO Clear Output Data Register” on page 437
• “PIO Multi-driver Enable Register” on page 440
• “PIO Multi-driver Disable Register” on page 441
• “PIO Pull Up Disable Register” on page 442
• “PIO Pull Up Enable Register” on page 442
• “PIO Peripheral A Select Register” on page 443
• “PIO Peripheral B Select Register” on page 444
• “PIO Output Write Enable Register” on page 445
• “PIO Output Write Disable Register” on page 445
• WPKEY: Write Protect KEY
Should be written at value 0x534D43 (“SMC” in ASCII). Writing any other value in this field aborts the write operation of the
WPEN bit. Always reads as 0.
447
6438D–ATARM–13-Oct-09
AT91SAM9G45
29.6.32 PIO Write Protect Status Register
Register Name:
PIO_WPSR
Addresses:
0xFFFFF2E8 (PIOA), 0xFFFFF4E8 (PIOB), 0xFFFFF6E8 (PIOC), 0xFFFFF8E8 (PIOD),
0xFFFFFAE8 (PIOE)
Access Type:
Read-only
Reset Value:
See Table 29-2
31
30
29
28
27
26
25
24
—
—
—
—
—
—
—
—
23
22
21
20
19
18
17
16
11
10
9
8
WPVSRC
15
14
13
12
WPVSRC
7
6
5
4
3
2
1
0
—
—
—
—
—
—
—
WPVS
• WPVS: Write Protect Enable
0 = No Write Protect Violation has occurred since the last read of the PIO_WPSR register.
1 = A Write Protect Violation occurred since the last read of the PIO_WPSR register. If this violation is an unauthorized
attempt to write a protected register, the associated violation is reported into field WPVSRC.
• WPVSRC: Write Protect Violation Source
When WPVS is active, this field indicates the write-protected register (through address offset or code) in which a write
access has been attempted.
Note:
Reading PIO_WPSR automatically clears all fields.
448
6438D–ATARM–13-Oct-09
AT91SAM9G45
449
6438D–ATARM–13-Oct-09
AT91SAM9G45
30. Serial Peripheral Interface (SPI)
30.1
Description
The Serial Peripheral Interface (SPI) circuit is a synchronous serial data link that provides communication with external devices in Master or Slave Mode. It also enables communication
between processors if an external processor is connected to the system.
The Serial Peripheral Interface is essentially a shift register that serially transmits data bits to
other SPIs. During a data transfer, one SPI system acts as the “master”' which controls the data
flow, while the other devices act as “slaves'' which have data shifted into and out by the master.
Different CPUs can take turn being masters (Multiple Master Protocol opposite to Single Master
Protocol where one CPU is always the master while all of the others are always slaves) and one
master may simultaneously shift data into multiple slaves. However, only one slave may drive its
output to write data back to the master at any given time.
A slave device is selected when the master asserts its NSS signal. If multiple slave devices
exist, the master generates a separate slave select signal for each slave (NPCS).
The SPI system consists of two data lines and two control lines:
• Master Out Slave In (MOSI): This data line supplies the output data from the master shifted
into the input(s) of the slave(s).
• Master In Slave Out (MISO): This data line supplies the output data from a slave to the input
of the master. There may be no more than one slave transmitting data during any particular
transfer.
• Serial Clock (SPCK): This control line is driven by the master and regulates the flow of the
data bits. The master may transmit data at a variety of baud rates; the SPCK line cycles once
for each bit that is transmitted.
• Slave Select (NSS): This control line allows slaves to be turned on and off by hardware.
30.2
Embedded Characteristics
• Supports communication with serial external devices
– Four chip selects with external decoder support allow communication with up to 15
peripherals
– Serial memories, such as DataFlash and 3-wire EEPROMs
– Serial peripherals, such as ADCs, DACs, LCD Controllers, CAN Controllers and
Sensors
– External co-processors
• Master or slave serial peripheral bus interface
– 8- to 16-bit programmable data length per chip select
– Programmable phase and polarity per chip select
– Programmable transfer delays between consecutive transfers and between clock
and data per chip select
– Programmable delay between consecutive transfers
– Selectable mode fault detection
• Very fast transfers supported
– Transfers with baud rates up to MCK
– The chip select line may be left active to speed up transfers on the same device
449
6438D–ATARM–13-Oct-09
30.3
Block Diagram
Figure 30-1. Block Diagram
PDC
APB
SPCK
MISO
PMC
MOSI
MCK
SPI Interface
PIO
NPCS0/NSS
NPCS1
NPCS2
Interrupt Control
NPCS3
SPI Interrupt
Figure 30-2. Block Diagram
AHB Matrix
DMA Ch.
Peripheral Bridge
APB
SPCK
MISO
PMC
MOSI
MCK
SPI Interface
PIO
NPCS0/NSS
NPCS1
NPCS2
Interrupt Control
NPCS3
SPI Interrupt
450
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
30.4
Application Block Diagram
Figure 30-3. Application Block Diagram: Single Master/Multiple Slave Implementation
SPI Master
SPCK
SPCK
MISO
MISO
MOSI
MOSI
NPCS0
NSS
Slave 0
SPCK
NPCS1
NPCS2
NPCS3
NC
MISO
Slave 1
MOSI
NSS
SPCK
MISO
Slave 2
MOSI
NSS
451
6438D–ATARM–13-Oct-09
30.5
Signal Description
Table 30-1.
Signal Description
Type
Pin Name
Pin Description
Master
Slave
MISO
Master In Slave Out
Input
Output
MOSI
Master Out Slave In
Output
Input
SPCK
Serial Clock
Output
Input
NPCS1-NPCS3
Peripheral Chip Selects
Output
Unused
NPCS0/NSS
Peripheral Chip Select/Slave Select
Output
Input
30.6
30.6.1
Product Dependencies
I/O Lines
The pins used for interfacing the compliant external devices may be multiplexed with PIO lines.
The programmer must first program the PIO controllers to assign the SPI pins to their peripheral
functions.
Table 30-2.
30.6.2
452
I/O Lines
Instance
Signal
I/O Line
Peripheral
SPI0
SPI0_MISO
PB0
A
SPI0
SPI0_MOSI
PB1
A
SPI0
SPI0_NPCS0
PB3
A
SPI0
SPI0_NPCS1
PB18
B
SPI0
SPI0_NPCS1
PD24
A
SPI0
SPI0_NPCS2
PB19
B
SPI0
SPI0_NPCS2
PD25
A
SPI0
SPI0_NPCS3
PD27
B
SPI0
SPI0_SPCK
PB2
A
SPI1
SPI1_MISO
PB14
A
SPI1
SPI1_MOSI
PB15
A
SPI1
SPI1_NPCS0
PB17
A
SPI1
SPI1_NPCS1
PD28
B
SPI1
SPI1_NPCS2
PD18
A
SPI1
SPI1_NPCS3
PD19
A
SPI1
SPI1_SPCK
PB16
A
Power Management
The SPI may be clocked through the Power Management Controller (PMC), thus the programmer must first configure the PMC to enable the SPI clock.
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
30.6.3
Interrupt
The SPI interface has an interrupt line connected to the Advanced Interrupt Controller
(AIC).Handling the SPI interrupt requires programming the AIC before configuring the SPI.
Table 30-3.
30.6.4
30.7
30.7.1
Peripheral IDs
Instance
ID
SPI0
14
SPI1
15
Peripheral DMA Controller (PDMA) Direct Memory Access Controller (DMAC)
The SPI interface can be used in conjunction with the PDMA DMAC in order to reduce processor
overhead. For a full description of the PDMA DMAC, refer to the corresponding section in the full
datasheet.
Functional Description
Modes of Operation
The SPI operates in Master Mode or in Slave Mode.
Operation in Master Mode is programmed by writing at 1 the MSTR bit in the Mode Register.
The pins NPCS0 to NPCS3 are all configured as outputs, the SPCK pin is driven, the MISO line
is wired on the receiver input and the MOSI line driven as an output by the transmitter.
If the MSTR bit is written at 0, the SPI operates in Slave Mode. The MISO line is driven by the
transmitter output, the MOSI line is wired on the receiver input, the SPCK pin is driven by the
transmitter to synchronize the receiver. The NPCS0 pin becomes an input, and is used as a
Slave Select signal (NSS). The pins NPCS1 to NPCS3 are not driven and can be used for other
purposes.
The data transfers are identically programmable for both modes of operations. The baud rate
generator is activated only in Master Mode.
30.7.2
Data Transfer
Four combinations of polarity and phase are available for data transfers. The clock polarity is
programmed with the CPOL bit in the Chip Select Register. The clock phase is programmed with
the NCPHA bit. These two parameters determine the edges of the clock signal on which data is
driven and sampled. Each of the two parameters has two possible states, resulting in four possible combinations that are incompatible with one another. Thus, a master/slave pair must use the
same parameter pair values to communicate. If multiple slaves are used and fixed in different
configurations, the master must reconfigure itself each time it needs to communicate with a different slave.
453
6438D–ATARM–13-Oct-09
Table 30-4 shows the four modes and corresponding parameter settings.
Table 30-4.
SPI Bus Protocol Mode
SPI Mode
CPOL
NCPHA
Shift SPCK Edge
Capture SPCK Edge
SPCK Inactive Level
0
0
1
Falling
Rising
Low
1
0
0
Rising
Falling
Low
2
1
1
Rising
Falling
High
3
1
0
Falling
Rising
High
Figure 30-4 and Figure 30-5 show examples of data transfers.
Figure 30-4. SPI Transfer Format (NCPHA = 1, 8 bits per transfer)
SPCK cycle (for reference)
1
2
3
4
6
5
7
8
SPCK
(CPOL = 0)
SPCK
(CPOL = 1)
MOSI
(from master)
MISO
(from slave)
MSB
MSB
6
5
4
3
2
1
LSB
6
5
4
3
2
1
LSB
*
NSS
(to slave)
* Not defined, but normally MSB of previous character received.
454
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 30-5. SPI Transfer Format (NCPHA = 0, 8 bits per transfer)
1
SPCK cycle (for reference)
2
3
4
5
7
6
8
SPCK
(CPOL = 0)
SPCK
(CPOL = 1)
MOSI
(from master)
MISO
(from slave)
*
MSB
6
5
4
3
2
1
MSB
6
5
4
3
2
1
LSB
LSB
NSS
(to slave)
* Not defined but normally LSB of previous character transmitted.
30.7.3
Master Mode Operations
When configured in Master Mode, the SPI operates on the clock generated by the internal programmable baud rate generator. It fully controls the data transfers to and from the slave(s)
connected to the SPI bus. The SPI drives the chip select line to the slave and the serial clock
signal (SPCK).
The SPI features two holding registers, the Transmit Data Register and the Receive Data Register, and a single Shift Register. The holding registers maintain the data flow at a constant rate.
After enabling the SPI, a data transfer begins when the processor writes to the SPI_TDR (Transmit Data Register). The written data is immediately transferred in the Shift Register and transfer
on the SPI bus starts. While the data in the Shift Register is shifted on the MOSI line, the MISO
line is sampled and shifted in the Shift Register. Receiving data cannot occur without transmitting data. If receiving mode is not needed, for example when communicating with a slave
receiver only (such as an LCD), the receive status flags in the status register can be discarded.
Before writing the TDR, the PCS field in the SPI_MR register must be set in order to select a
slave.
After enabling the SPI, a data transfer begins when the processor writes to the SPI_TDR (Transmit Data Register). The written data is immediately transferred in the Shift Register and transfer
on the SPI bus starts. While the data in the Shift Register is shifted on the MOSI line, the MISO
line is sampled and shifted in the Shift Register. Transmission cannot occur without reception.
Before writing the TDR, the PCS field must be set in order to select a slave.
If new data is written in SPI_TDR during the transfer, it stays in it until the current transfer is
completed. Then, the received data is transferred from the Shift Register to SPI_RDR, the data
in SPI_TDR is loaded in the Shift Register and a new transfer starts.
455
6438D–ATARM–13-Oct-09
The transfer of a data written in SPI_TDR in the Shift Register is indicated by the TDRE bit
(Transmit Data Register Empty) in the Status Register (SPI_SR). When new data is written in
SPI_TDR, this bit is cleared. The TDRE bit is used to trigger the Transmit PDC channel.
The end of transfer is indicated by the TXEMPTY flag in the SPI_SR register. If a transfer delay
(DLYBCT) is greater than 0 for the last transfer, TXEMPTY is set after the completion of said
delay. The master clock (MCK) can be switched off at this time.
The transfer of received data from the Shift Register in SPI_RDR is indicated by the RDRF bit
(Receive Data Register Full) in the Status Register (SPI_SR). When the received data is read,
the RDRF bit is cleared.
If the SPI_RDR (Receive Data Register) has not been read before new data is received, the
Overrun Error bit (OVRES) in SPI_SR is set. As long as this flag is set, data is loaded in
SPI_RDR. The user has to read the status register to clear the OVRES bit.
Figure 30-6, shows a block diagram of the SPI when operating in Master Mode. Figure 30-7 on
page 458 shows a flow chart describing how transfers are handled.
456
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
30.7.3.1
Master Mode Block Diagram
Figure 30-6. Master Mode Block Diagram
SPI_CSR0..3
SCBR
Baud Rate Generator
MCK
SPCK
SPI
Clock
SPI_CSR0..3
BITS
NCPHA
CPOL
LSB
MISO
SPI_RDR
RDRF
OVRES
RD
MSB
Shift Register
MOSI
SPI_TDR
TD
TDRE
SPI_CSR0..3
SPI_RDR
CSAAT
PCS
PS
NPCS3
PCSDEC
SPI_MR
PCS
0
NPCS2
Current
Peripheral
NPCS1
SPI_TDR
NPCS0
PCS
1
MSTR
MODF
NPCS0
MODFDIS
457
6438D–ATARM–13-Oct-09
30.7.3.2
Master Mode Flow Diagram
Figure 30-7. Master Mode Flow Diagram
SPI Enable
- NPCS defines the current Chip Select
- CSAAT, DLYBS, DLYBCT refer to the fields of the
Chip Select Register corresponding to the Current Chip Select
- When NPCS is 0xF, CSAAT is 0.
1
TDRE ?
0
1
CSAAT ?
PS ?
0
1
0
Fixed
peripheral
PS ?
1
Fixed
peripheral
0
Variable
peripheral
Variable
peripheral
SPI_TDR(PCS)
= NPCS ?
no
NPCS = SPI_TDR(PCS)
NPCS = SPI_MR(PCS)
yes
SPI_MR(PCS)
= NPCS ?
no
NPCS = 0xF
NPCS = 0xF
Delay DLYBCS
Delay DLYBCS
NPCS = SPI_TDR(PCS)
NPCS = SPI_MR(PCS),
SPI_TDR(PCS)
Delay DLYBS
Serializer = SPI_TDR(TD)
TDRE = 1
Data Transfer
SPI_RDR(RD) = Serializer
RDRF = 1
Delay DLYBCT
0
TDRE ?
1
1
CSAAT ?
0
NPCS = 0xF
Delay DLYBCS
458
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 30-8 shows Transmit Data Register Empty (TDRE), Receive Data Register (RDRF) and
Transmission Register Empty (TXEMPTY) status flags behavior within the SPI_SR (Status Register) during an 8-bit data transfer in fixed mode and no Peripheral Data Controller involved.
Figure 30-8.
Status Register Flags Behavior
1
2
3
4
6
5
7
8
SPCK
NPCS0
MOSI
(from master)
MSB
6
5
4
3
2
1
LSB
TDRE
RDR read
Write in
SPI_TDR
RDRF
MISO
(from slave)
MSB
6
5
4
3
2
1
LSB
TXEMPTY
shift register empty
Figure 30-9 shows Transmission Register Empty (TXEMPTY), End of RX buffer (ENDRX), End
of TX buffer (ENDTX), RX Buffer Full (RXBUFF) and TX Buffer Empty (TXBUFE) status flags
behavior within the SPI_SR (Status Register) during an 8-bit data transfer in fixed mode with the
Peripheral Data Controller involved. The PDC is programmed to transfer and receive three data.
The next pointer and counter are not used. The RDRF and TDRE are not shown because these
flags are managed by the PDC when using the PDC.
459
6438D–ATARM–13-Oct-09
Figure 30-9. PDC Status Register Flags Behavior
1
3
2
SPCK
NPCS0
MOSI
(from master)
MISO
(from slave)
MSB
MSB
6
5
4
3
2
1
LSB MSB
6
5
4
3
2
1
LSB
MSB
6
5
4
3
2
1
LSB
6
5
4
3
2
1
LSB MSB
6
5
4
3
2
1
LSB
MSB
6
5
4
3
2
1
LSB
ENDTX
ENDRX
TXBUFE
RXBUFF
TXEMPTY
30.7.3.3
Clock Generation
The SPI Baud rate clock is generated by dividing the Master Clock (MCK), by a value between 1
and 255.
This allows a maximum operating baud rate at up to Master Clock and a minimum operating
baud rate of MCK divided by 255.
Programming the SCBR field at 0 is forbidden. Triggering a transfer while SCBR is at 0 can lead
to unpredictable results.
At reset, SCBR is 0 and the user has to program it at a valid value before performing the first
transfer.
The divisor can be defined independently for each chip select, as it has to be programmed in the
SCBR field of the Chip Select Registers. This allows the SPI to automatically adapt the baud
rate for each interfaced peripheral without reprogramming.
30.7.3.4
Transfer Delays
Figure 30-10 shows a chip select transfer change and consecutive transfers on the same chip
select. Three delays can be programmed to modify the transfer waveforms:
• The delay between chip selects, programmable only once for all the chip selects by writing
the DLYBCS field in the Mode Register. Allows insertion of a delay between release of one
chip select and before assertion of a new one.
• The delay before SPCK, independently programmable for each chip select by writing the field
DLYBS. Allows the start of SPCK to be delayed after the chip select has been asserted.
• The delay between consecutive transfers, independently programmable for each chip select
by writing the DLYBCT field. Allows insertion of a delay between two transfers occurring on
the same chip select
460
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
These delays allow the SPI to be adapted to the interfaced peripherals and their speed and bus
release time.
Figure 30-10. Programmable Delays
Chip Select 1
Chip Select 2
SPCK
DLYBCS
30.7.3.5
DLYBS
DLYBCT
DLYBCT
Peripheral Selection
The serial peripherals are selected through the assertion of the NPCS0 to NPCS3 signals. By
default, all the NPCS signals are high before and after each transfer.
• Fixed Peripheral Select: SPI exchanges data with only one peripheral
Fixed Peripheral Select is activated by writing the PS bit to zero in SPI_MR (Mode Register). In
this case, the current peripheral is defined by the PCS field in SPI_MR and the PCS field in the
SPI_TDR has no effect.
• Variable Peripheral Select: Data can be exchanged with more than one peripheral without
having to reprogram the NPCS field in the SPI_MR register.
Variable Peripheral Select is activated by setting PS bit to one. The PCS field in SPI_TDR is
used to select the current peripheral. This means that the peripheral selection can be defined for
each new data. The value to write in the SPI_TDR register as the following format.
[xxxxxxx(7-bit) + LASTXFER(1-bit)(1)+ xxxx(4-bit) + PCS (4-bit) + DATA (8 to 16-bit)] with PCS
equals to the chip select to assert as defined in Section 30.8.4 (SPI Transmit Data Register) and
LASTXFER bit at 0 or 1 depending on CSAAT bit. CSAAT, LASTXFER and CSNAAT bit are discussed in the Peripheral Deselection in Section 30.7.3.11.
Note:
30.7.3.6
1. Optional.
SPI Peripheral DMA Controller (PDC)
In both fixed and variable mode the Peripheral DMA Controller (PDC) can be used to reduce
processor overhead.
The Fixed Peripheral Selection allows buffer transfers with a single peripheral. Using the PDC is
an optimal means, as the size of the data transfer between the memory and the SPI is either 8
bits or 16 bits. However, changing the peripheral selection requires the Mode Register to be
reprogrammed.
The Variable Peripheral Selection allows buffer transfers with multiple peripherals without reprogramming the Mode Register. Data written in SPI_TDR is 32 bits wide and defines the real data
to be transmitted and the peripheral it is destined to. Using the PDC in this mode requires 32-bit
wide buffers, with the data in the LSBs and the PCS and LASTXFER fields in the MSBs, how-
461
6438D–ATARM–13-Oct-09
ever the SPI still controls the number of bits (8 to16) to be transferred through MISO and MOSI
lines with the chip select configuration registers. This is not the optimal means in term of memory size for the buffers, but it provides a very effective means to exchange data with several
peripherals without any intervention of the processor.
30.7.3.7
Transfer Size
Depending on the data size to transmit, from 8 to 16 bits, the PDC manages automatically the
type of pointer's size it has to point to. The PDC will perform the following transfer size depending on the mode and number of bits per data.
Fixed Mode:
• 8-bit Data:
Byte transfer,
PDC Pointer Address = Address + 1 byte,
PDC Counter = Counter - 1
• 8-bit to 16-bit Data:
2 bytes transfer. n-bit data transfer with don’t care data (MSB) filled with 0’s,
PDC Pointer Address = Address + 2 bytes,
PDC Counter = Counter - 1
Variable Mode:
In variable Mode, PDC Pointer Address = Address +4 bytes and PDC Counter = Counter - 1 for
8 to 16-bit transfer size. When using the PDC, the TDRE and RDRF flags are handled by the
PDC, thus the user’s application does not have to check those bits. Only End of RX Buffer
(ENDRX), End of TX Buffer (ENDTX), Buffer Full (RXBUFF), TX Buffer Empty (TXBUFE) are
significant. For further details about the Peripheral DMA Controller and user interface, refer to
the PDC section of the product datasheet.
30.7.3.8
SPI Direct Access Memory Controller (DMAC)
In both fixed and variable mode the Direct Memory Access Controller (DMAC) can be used to
reduce processor overhead.
The Fixed Peripheral Selection allows buffer transfers with a single peripheral. Using the DMAC
is an optimal means, as the size of the data transfer between the memory and the SPI is either 8
bits or 16 bits. However, changing the peripheral selection requires the Mode Register to be
reprogrammed.
The Variable Peripheral Selection allows buffer transfers with multiple peripherals without reprogramming the Mode Register. Data written in SPI_TDR is 32 bits wide and defines the real data
to be transmitted and the peripheral it is destined to. Using the DMAC in this mode requires 32bit wide buffers, with the data in the LSBs and the PCS and LASTXFER fields in the MSBs, however the SPI still controls the number of bits (8 to16) to be transferred through MISO and MOSI
lines with the chip select configuration registers. This is not the optimal means in term of memory size for the buffers, but it provides a very effective means to exchange data with several
peripherals without any intervention of the processor.
462
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
30.7.3.9
Peripheral Chip Select Decoding
The user can program the SPI to operate with up to 15 peripherals by decoding the four Chip
Select lines, NPCS0 to NPCS3 with 1 of up to 16 decoder/demultiplexer. This can be enabled by
writing the PCSDEC bit at 1 in the Mode Register (SPI_MR).
When operating without decoding, the SPI makes sure that in any case only one chip select line
is activated, i.e., one NPCS line driven low at a time. If two bits are defined low in a PCS field,
only the lowest numbered chip select is driven low.
When operating with decoding, the SPI directly outputs the value defined by the PCS field on
NPCS lines of either the Mode Register or the Transmit Data Register (depending on PS).
As the SPI sets a default value of 0xF on the chip select lines (i.e. all chip select lines at 1) when
not processing any transfer, only 15 peripherals can be decoded.
The SPI has only four Chip Select Registers, not 15. As a result, when decoding is activated,
each chip select defines the characteristics of up to four peripherals. As an example, SPI_CRS0
defines the characteristics of the externally decoded peripherals 0 to 3, corresponding to the
PCS values 0x0 to 0x3. Thus, the user has to make sure to connect compatible peripherals on
the decoded chip select lines 0 to 3, 4 to 7, 8 to 11 and 12 to 14. Figure 30-11 below shows such
an implementation.
If the CSAAT bit is used, with or without the PDC, the Mode Fault detection for NPCS0 line must
be disabled. This is not needed for all other chip select lines since Mode Fault Detection is only
on NPCS0.
If the CSAAT bit is used, with or without the DMAC, the Mode Fault detection for NPCS0 line
must be disabled. This is not needed for all other chip select lines since Mode Fault Detection is
only on NPCS0.
Figure 30-11. Chip Select Decoding Application Block Diagram: Single Master/Multiple Slave Implementation
SPCK
MISO
MOSI
SPCK MISO MOSI
SPCK MISO MOSI
SPCK MISO MOSI
Slave 0
Slave 1
Slave 14
NSS
NSS
SPI Master
NSS
NPCS0
NPCS1
NPCS2
NPCS3
1-of-n Decoder/Demultiplexer
463
6438D–ATARM–13-Oct-09
30.7.3.10
Peripheral Deselection without PDCDMAC
During a transfer of more than one data on a Chip Select without the PDCDMAC, the SPI_TDR
is loaded by the processor, the flag TDRE rises as soon as the content of the SPI_TDR is transferred into the internal shift register. When this flag is detected high, the SPI_TDR can be
reloaded. If this reload by the processor occurs before the end of the current transfer and if the
next transfer is performed on the same chip select as the current transfer, the Chip Select is not
de-asserted between the two transfers. But depending on the application software handling the
SPI status register flags (by interrupt or polling method) or servicing other interrupts or other
tasks, the processor may not reload the SPI_TDR in time to keep the chip select active (low). A
null Delay Between Consecutive Transfer (DLYBCT) value in the SPI_CSR register, will give
even less time for the processor to reload the SPI_TDR. With some SPI slave peripherals,
requiring the chip select line to remain active (low) during a full set of transfers might lead to
communication errors.
To facilitate interfacing with such devices, the Chip Select Register [CSR0...CSR3] can be programmed with the CSAAT bit (Chip Select Active After Transfer) at 1. This allows the chip select
lines to remain in their current state (low = active) until transfer to another chip select is required.
Even if the SPI_TDR is not reloaded the chip select will remain active. To have the chip select
line to raise at the end of the transfer the Last transfer Bit (LASTXFER) in the SPI_MR register
must be set at 1 before writing the last data to transmit into the SPI_TDR.
30.7.3.11
Peripheral Deselection with PDC
When the Peripheral DMA Controller is used, the chip select line will remain low during the
whole transfer since the TDRE flag is managed by the PDC itself. The reloading of the SPI_TDR
by the PDC is done as soon as TDRE flag is set to one. In this case the use of CSAAT bit might
not be needed. However, it may happen that when other PDC channels connected to other
peripherals are in use as well, the SPI PDC might be delayed by another (PDC with a higher priority on the bus). Having PDC buffers in slower memories like flash memory or SDRAM
compared to fast internal SRAM, may lengthen the reload time of the SPI_TDR by the PDC as
well. This means that the SPI_TDR might not be reloaded in time to keep the chip select line
low. In this case the chip select line may toggle between data transfer and according to some
SPI Slave devices, the communication might get lost. The use of the CSAAT bit might be
needed.
30.7.3.12
Peripheral Deselection with DMAC
When the Direct Memory Access Controller is used, the chip select line will remain low during
the whole transfer since the TDRE flag is managed by the DMAC itself. The reloading of the
SPI_TDR by the DMAC is done as soon as TDRE flag is set to one. In this case the use of
CSAAT bit might not be needed. However, it may happen that when other DMAC channels connected to other peripherals are in use as well, the SPI DMAC might be delayed by another
(DMAC with a higher priority on the bus). Having DMAC buffers in slower memories like flash
memory or SDRAM compared to fast internal SRAM, may lengthen the reload time of the
SPI_TDR by the DMAC as well. This means that the SPI_TDR might not be reloaded in time to
keep the chip select line low. In this case the chip select line may toggle between data transfer
and according to some SPI Slave devices, the communication might get lost. The use of the
CSAAT bit might be needed.
Figure 30-12 shows different peripheral deselction cases and the effect of the CSAAT bit.
464
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 30-12. Peripheral Deselection
CSAAT = 0
TDRE
NPCS[0..3]
CSAAT = 1
DLYBCT
DLYBCT
A
A
A
A
DLYBCS
A
DLYBCS
PCS = A
PCS = A
Write SPI_TDR
TDRE
NPCS[0..3]
DLYBCT
DLYBCT
A
A
A
A
DLYBCS
A
DLYBCS
PCS=A
PCS = A
Write SPI_TDR
TDRE
NPCS[0..3]
DLYBCT
DLYBCT
A
B
A
B
DLYBCS
PCS = B
DLYBCS
PCS = B
Write SPI_TDR
30.7.3.13
Mode Fault Detection
A mode fault is detected when the SPI is programmed in Master Mode and a low level is driven
by an external master on the NPCS0/NSS signal. In this case, multi-master configuration,
NPCS0, MOSI, MISO and SPCK pins must be configured in open drain (through the PIO controller). When a mode fault is detected, the MODF bit in the SPI_SR is set until the SPI_SR is read
and the SPI is automatically disabled until re-enabled by writing the SPIEN bit in the SPI_CR
(Control Register) at 1.
By default, the Mode Fault detection circuitry is enabled. The user can disable Mode Fault
detection by setting the MODFDIS bit in the SPI Mode Register (SPI_MR).
30.7.4
SPI Slave Mode
When operating in Slave Mode, the SPI processes data bits on the clock provided on the SPI
clock pin (SPCK).
The SPI waits for NSS to go active before receiving the serial clock from an external master.
When NSS falls, the clock is validated on the serializer, which processes the number of bits
defined by the BITS field of the Chip Select Register 0 (SPI_CSR0). These bits are processed
following a phase and a polarity defined respectively by the NCPHA and CPOL bits of the
SPI_CSR0. Note that BITS, CPOL and NCPHA of the other Chip Select Registers have no
effect when the SPI is programmed in Slave Mode.
465
6438D–ATARM–13-Oct-09
The bits are shifted out on the MISO line and sampled on the MOSI line.
(For more information on BITS field, see also, the
“SPI Chip Select Register” on page 478.)
(Note:)
below the register table; Section 30.8.9
When all the bits are processed, the received data is transferred in the Receive Data Register
and the RDRF bit rises. If the SPI_RDR (Receive Data Register) has not been read before new
data is received, the Overrun Error bit (OVRES) in SPI_SR is set. As long as this flag is set, data
is loaded in SPI_RDR. The user has to read the status register to clear the OVRES bit.
When a transfer starts, the data shifted out is the data present in the Shift Register. If no data
has been written in the Transmit Data Register (SPI_TDR), the last data received is transferred.
If no data has been received since the last reset, all bits are transmitted low, as the Shift Register resets at 0.
When a first data is written in SPI_TDR, it is transferred immediately in the Shift Register and the
TDRE bit rises. If new data is written, it remains in SPI_TDR until a transfer occurs, i.e. NSS falls
and there is a valid clock on the SPCK pin. When the transfer occurs, the last data written in
SPI_TDR is transferred in the Shift Register and the TDRE bit rises. This enables frequent
updates of critical variables with single transfers.
Then, a new data is loaded in the Shift Register from the Transmit Data Register. In case no
character is ready to be transmitted, i.e. no character has been written in SPI_TDR since the last
load from SPI_TDR to the Shift Register, the Shift Register is not modified and the last received
character is retransmitted.
Figure 30-13 shows a block diagram of the SPI when operating in Slave Mode.
Figure 30-13. Slave Mode Functional Bloc Diagram
SPCK
NSS
SPI
Clock
SPIEN
SPIENS
SPIDIS
SPI_CSR0
BITS
NCPHA
CPOL
MOSI
LSB
SPI_RDR
RDRF
OVRES
RD
MSB
Shift Register
MISO
SPI_TDR
TD
466
TDRE
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
30.8
Serial Peripheral Interface (SPI) User Interface
Table 30-5.
Register Mapping
Offset
Register
Name
Access
Reset
0x00
Control Register
SPI_CR
Write-only
---
0x04
Mode Register
SPI_MR
Read-write
0x0
0x08
Receive Data Register
SPI_RDR
Read-only
0x0
0x0C
Transmit Data Register
SPI_TDR
Write-only
---
0x10
Status Register
SPI_SR
Read-only
0x000000F0
0x14
Interrupt Enable Register
SPI_IER
Write-only
---
0x18
Interrupt Disable Register
SPI_IDR
Write-only
---
0x1C
Interrupt Mask Register
SPI_IMR
Read-only
0x0
0x20 - 0x2C
Reserved
0x30
Chip Select Register 0
SPI_CSR0
Read-write
0x0
0x34
Chip Select Register 1
SPI_CSR1
Read-write
0x0
0x38
Chip Select Register 2
SPI_CSR2
Read-write
0x0
0x3C
Chip Select Register 3
SPI_CSR3
Read-write
0x0
Reserved
–
–
–
Reserved for the PDC
–
–
–
0x004C - 0x00F8
0x100 - 0x124
467
6438D–ATARM–13-Oct-09
30.8.1
Name:
SPI Control Register
SPI_CR
Addresses:
0xFFFA4000 (0), 0xFFFA8000 (1)
Access:
Write-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
LASTXFER
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
SWRST
–
–
–
–
–
SPIDIS
SPIEN
• SPIEN: SPI Enable
0 = No effect.
1 = Enables the SPI to transfer and receive data.
• SPIDIS: SPI Disable
0 = No effect.
1 = Disables the SPI.
As soon as SPIDIS is set, SPI finishes its transfer.
All pins are set in input mode and no data is received or transmitted.
If a transfer is in progress, the transfer is finished before the SPI is disabled.
If both SPIEN and SPIDIS are equal to one when the control register is written, the SPI is disabled.
• SWRST: SPI Software Reset
0 = No effect.
1 = Reset the SPI. A software-triggered hardware reset of the SPI interface is performed.
The SPI is in slave mode after software reset.
PDC channels are not affected by software reset.
• LASTXFER: Last Transfer
0 = No effect.
1 = The current NPCS will be deasserted after the character written in TD has been transferred. When CSAAT is set, this
allows to close the communication with the current serial peripheral by raising the corresponding NPCS line as soon as TD
transfer has completed.
468
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
30.8.2
Name:
SPI Mode Register
SPI_MR
Addresses:
0xFFFA4004 (0), 0xFFFA8004 (1)
Access:
Read/Write
31
30
29
28
27
26
19
18
25
24
17
16
DLYBCS
23
22
21
20
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
PCS
7
6
5
4
3
2
1
0
LLB
–
–
MODFDIS
–
PCSDEC
PS
MSTR
• MSTR: Master/Slave Mode
0 = SPI is in Slave mode.
1 = SPI is in Master mode.
• PS: Peripheral Select
0 = Fixed Peripheral Select.
1 = Variable Peripheral Select.
• PCSDEC: Chip Select Decode
0 = The chip selects are directly connected to a peripheral device.
1 = The four chip select lines are connected to a 4- to 16-bit decoder.
When PCSDEC equals one, up to 15 Chip Select signals can be generated with the four lines using an external 4- to 16-bit
decoder. The Chip Select Registers define the characteristics of the 15 chip selects according to the following rules:
SPI_CSR0 defines peripheral chip select signals 0 to 3.
SPI_CSR1 defines peripheral chip select signals 4 to 7.
SPI_CSR2 defines peripheral chip select signals 8 to 11.
SPI_CSR3 defines peripheral chip select signals 12 to 14.
• MODFDIS: Mode Fault Detection
0 = Mode fault detection is enabled.
1 = Mode fault detection is disabled.
• LLB: Local Loopback Enable
0 = Local loopback path disabled.
1 = Local loopback path enabled
LLB controls the local loopback on the data serializer for testing in Master Mode only. (MISO is internally connected on
MOSI.)
469
6438D–ATARM–13-Oct-09
• PCS: Peripheral Chip Select
This field is only used if Fixed Peripheral Select is active (PS = 0).
If PCSDEC = 0:
PCS = xxx0
NPCS[3:0] = 1110
PCS = xx01
NPCS[3:0] = 1101
PCS = x011
NPCS[3:0] = 1011
PCS = 0111
NPCS[3:0] = 0111
PCS = 1111
forbidden (no peripheral is selected)
(x = don’t care)
If PCSDEC = 1:
NPCS[3:0] output signals = PCS.
• DLYBCS: Delay Between Chip Selects
This field defines the delay from NPCS inactive to the activation of another NPCS. The DLYBCS time guarantees non-overlapping chip selects and solves bus contentions in case of peripherals having long data float times.
If DLYBCS is less than or equal to six, six MCK periods will be inserted by default.
Otherwise, the following equation determines the delay:
Delay Between Chip Selects = DLYBCS
----------------------MCK
470
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
30.8.3
Name:
SPI Receive Data Register
SPI_RDR
Addresses:
0xFFFA4008 (0), 0xFFFA8008 (1)
Access:
Read-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
15
14
13
12
PCS
11
10
9
8
3
2
1
0
RD
7
6
5
4
RD
• RD: Receive Data
Data received by the SPI Interface is stored in this register right-justified. Unused bits read zero.
• PCS: Peripheral Chip Select
In Master Mode only, these bits indicate the value on the NPCS pins at the end of a transfer. Otherwise, these bits read
zero.
471
6438D–ATARM–13-Oct-09
AT91SAM9G45
30.8.4
Name:
SPI Transmit Data Register
SPI_TDR
Addresses:
0xFFFA400C (0), 0xFFFA800C (1)
Access:
Write-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
LASTXFER
23
22
21
20
19
18
17
16
–
–
–
–
15
14
13
12
PCS
11
10
9
8
3
2
1
0
TD
7
6
5
4
TD
• TD: Transmit Data
Data to be transmitted by the SPI Interface is stored in this register. Information to be transmitted must be written to the
transmit data register in a right-justified format.
• PCS: Peripheral Chip Select
This field is only used if Variable Peripheral Select is active (PS = 1).
If PCSDEC = 0:
PCS = xxx0
NPCS[3:0] = 1110
PCS = xx01
NPCS[3:0] = 1101
PCS = x011
NPCS[3:0] = 1011
PCS = 0111
NPCS[3:0] = 0111
PCS = 1111
forbidden (no peripheral is selected)
(x = don’t care)
If PCSDEC = 1:
NPCS[3:0] output signals = PCS
• LASTXFER: Last Transfer
0 = No effect.
1 = The current NPCS will be deasserted after the character written in TD has been transferred. When CSAAT is set, this
allows to close the communication with the current serial peripheral by raising the corresponding NPCS line as soon as TD
transfer has completed.
This field is only used if Variable Peripheral Select is active (PS = 1).
472
6438D–ATARM–13-Oct-09
AT91SAM9G45
30.8.5
Name:
SPI Status Register
SPI_SR
Addresses:
0xFFFA4010 (0), 0xFFFA8010 (1)
Access:
Read-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
SPIENS
15
14
13
12
11
10
9
8
–
–
–
–
–
–
TXEMPTY
NSSR
7
6
5
4
3
2
1
0
–
–
–
–
OVRES
MODF
TDRE
RDRF
• RDRF: Receive Data Register Full
0 = No data has been received since the last read of SPI_RDR
1 = Data has been received and the received data has been transferred from the serializer to SPI_RDR since the last read
of SPI_RDR.
• TDRE: Transmit Data Register Empty
0 = Data has been written to SPI_TDR and not yet transferred to the serializer.
1 = The last data written in the Transmit Data Register has been transferred to the serializer.
TDRE equals zero when the SPI is disabled or at reset. The SPI enable command sets this bit to one.
• MODF: Mode Fault Error
0 = No Mode Fault has been detected since the last read of SPI_SR.
1 = A Mode Fault occurred since the last read of the SPI_SR.
• OVRES: Overrun Error Status
0 = No overrun has been detected since the last read of SPI_SR.
1 = An overrun has occurred since the last read of SPI_SR.
An overrun occurs when SPI_RDR is loaded at least twice from the serializer since the last read of the SPI_RDR.
• ENDRX: End of RX buffer
0 = The Receive Counter Register has not reached 0 since the last write in SPI_RCR(1) or SPI_RNCR(1).
1 = The Receive Counter Register has reached 0 since the last write in SPI_RCR(1) or SPI_RNCR(1).
• ENDTX: End of TX buffer
0 = The Transmit Counter Register has not reached 0 since the last write in SPI_TCR(1) or SPI_TNCR(1).
1 = The Transmit Counter Register has reached 0 since the last write in SPI_TCR(1) or SPI_TNCR(1).
• RXBUFF: RX Buffer Full
0 = SPI_RCR(1) or SPI_RNCR(1) has a value other than 0.
1 = Both SPI_RCR(1) and SPI_RNCR(1) have a value of 0.
473
6438D–ATARM–13-Oct-09
AT91SAM9G45
• TXBUFE: TX Buffer Empty
0 = SPI_TCR(1) or SPI_TNCR(1) has a value other than 0.
1 = Both SPI_TCR(1) and SPI_TNCR(1) have a value of 0.
• NSSR: NSS Rising
0 = No rising edge detected on NSS pin since last read.
1 = A rising edge occurred on NSS pin since last read.
• TXEMPTY: Transmission Registers Empty
0 = As soon as data is written in SPI_TDR.
1 = SPI_TDR and internal shifter are empty. If a transfer delay has been defined, TXEMPTY is set after the completion of
such delay.
• SPIENS: SPI Enable Status
0 = SPI is disabled.
1 = SPI is enabled.
Note:
1. SPI_RCR, SPI_RNCR, SPI_TCR, SPI_TNCR are physically located in the PDC.
474
6438D–ATARM–13-Oct-09
AT91SAM9G45
30.8.6
Name:
SPI Interrupt Enable Register
SPI_IER
Addresses:
0xFFFA4014 (0), 0xFFFA8014 (1)
Access:
Write-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
TXEMPTY
NSSR
7
6
5
4
3
2
1
0
–
–
–
–
OVRES
MODF
TDRE
RDRF
0 = No effect.
1 = Enables the corresponding interrupt.
• RDRF: Receive Data Register Full Interrupt Enable
• TDRE: SPI Transmit Data Register Empty Interrupt Enable
• MODF: Mode Fault Error Interrupt Enable
• OVRES: Overrun Error Interrupt Enable
• ENDRX: End of Receive Buffer Interrupt Enable
• ENDTX: End of Transmit Buffer Interrupt Enable
• RXBUFF: Receive Buffer Full Interrupt Enable
• TXBUFE: Transmit Buffer Empty Interrupt Enable
• NSSR: NSS Rising Interrupt Enable
• TXEMPTY: Transmission Registers Empty Enable
475
6438D–ATARM–13-Oct-09
AT91SAM9G45
30.8.7
Name:
SPI Interrupt Disable Register
SPI_IDR
Addresses:
0xFFFA4018 (0), 0xFFFA8018 (1)
Access:
Write-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
TXEMPTY
NSSR
7
6
5
4
3
2
1
0
–
–
–
–
OVRES
MODF
TDRE
RDRF
0 = No effect.
1 = Disables the corresponding interrupt.
• RDRF: Receive Data Register Full Interrupt Disable
• TDRE: SPI Transmit Data Register Empty Interrupt Disable
• MODF: Mode Fault Error Interrupt Disable
• OVRES: Overrun Error Interrupt Disable
• ENDRX: End of Receive Buffer Interrupt Disable
• ENDTX: End of Transmit Buffer Interrupt Disable
• RXBUFF: Receive Buffer Full Interrupt Disable
• TXBUFE: Transmit Buffer Empty Interrupt Disable
• NSSR: NSS Rising Interrupt Disable
• TXEMPTY: Transmission Registers Empty Disable
476
6438D–ATARM–13-Oct-09
AT91SAM9G45
30.8.8
Name:
SPI Interrupt Mask Register
SPI_IMR
Addresses:
0xFFFA401C (0), 0xFFFA801C (1)
Access:
Read-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
TXEMPTY
NSSR
7
6
5
4
3
2
1
0
–
–
–
–
OVRES
MODF
TDRE
RDRF
0 = The corresponding interrupt is not enabled.
1 = The corresponding interrupt is enabled.
• RDRF: Receive Data Register Full Interrupt Mask
• TDRE: SPI Transmit Data Register Empty Interrupt Mask
• MODF: Mode Fault Error Interrupt Mask
• OVRES: Overrun Error Interrupt Mask
• ENDRX: End of Receive Buffer Interrupt Mask
• ENDTX: End of Transmit Buffer Interrupt Mask
• RXBUFF: Receive Buffer Full Interrupt Mask
• TXBUFE: Transmit Buffer Empty Interrupt Mask
• NSSR: NSS Rising Interrupt Mask
• TXEMPTY: Transmission Registers Empty Mask
477
6438D–ATARM–13-Oct-09
AT91SAM9G45
30.8.9
Name:
SPI Chip Select Register
SPI_CSR0... SPI_CSR3
Addresses:
0xFFFA4030 (0), 0xFFFA8030 (1)
Access:
Read/Write
31
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
DLYBCT
23
22
21
20
DLYBS
15
14
13
12
SCBR
7
6
5
4
BITS
Note:
3
2
1
0
CSAAT
–
NCPHA
CPOL
SPI_CSRx registers must be written even if the user wants to use the defaults. The BITS field will not be updated with the translated value unless the register is written.
• CPOL: Clock Polarity
0 = The inactive state value of SPCK is logic level zero.
1 = The inactive state value of SPCK is logic level one.
CPOL is used to determine the inactive state value of the serial clock (SPCK). It is used with NCPHA to produce the
required clock/data relationship between master and slave devices.
• NCPHA: Clock Phase
0 = Data is changed on the leading edge of SPCK and captured on the following edge of SPCK.
1 = Data is captured on the leading edge of SPCK and changed on the following edge of SPCK.
NCPHA determines which edge of SPCK causes data to change and which edge causes data to be captured. NCPHA is
used with CPOL to produce the required clock/data relationship between master and slave devices.
• CSAAT: Chip Select Active After Transfer
0 = The Peripheral Chip Select Line rises as soon as the last transfer is achieved.
1 = The Peripheral Chip Select does not rise after the last transfer is achieved. It remains active until a new transfer is
requested on a different chip select.
• BITS: Bits Per Transfer (See the (Note:) below the register table; Section 30.8.9 “SPI Chip Select Register” on page 478.)
The BITS field determines the number of data bits transferred. Reserved values should not be used.
BITS
0000
0001
0010
0011
0100
0101
0110
0111
Bits Per Transfer
8
9
10
11
12
13
14
15
478
6438D–ATARM–13-Oct-09
AT91SAM9G45
BITS
1000
1001
1010
1011
1100
1101
1110
1111
Bits Per Transfer
16
Reserved
Reserved
Reserved
Reserved
Reserved
Reserved
Reserved
• SCBR: Serial Clock Baud Rate
In Master Mode, the SPI Interface uses a modulus counter to derive the SPCK baud rate from the Master Clock MCK. The
Baud rate is selected by writing a value from 1 to 255 in the SCBR field. The following equations determine the SPCK baud
rate:
MCKSPCK Baudrate = -------------SCBR
Programming the SCBR field at 0 is forbidden. Triggering a transfer while SCBR is at 0 can lead to unpredictable results.
At reset, SCBR is 0 and the user has to program it at a valid value before performing the first transfer.
• DLYBS: Delay Before SPCK
This field defines the delay from NPCS valid to the first valid SPCK transition.
When DLYBS equals zero, the NPCS valid to SPCK transition is 1/2 the SPCK clock period.
Otherwise, the following equations determine the delay:
Delay Before SPCK = DLYBS
------------------MCK
• DLYBCT: Delay Between Consecutive Transfers
This field defines the delay between two consecutive transfers with the same peripheral without removing the chip select.
The delay is always inserted after each transfer and before removing the chip select if needed.
When DLYBCT equals zero, no delay between consecutive transfers is inserted and the clock keeps its duty cycle over the
character transfers.
Otherwise, the following equation determines the delay:
32 × DLYBCT
Delay Between Consecutive Transfers = ------------------------------------MCK
479
6438D–ATARM–13-Oct-09
480
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
31. Two-wire Interface (TWI)
31.1
Description
The Atmel Two-wire Interface (TWI) interconnects components on a unique two-wire bus, made
up of one clock line and one data line with speeds of up to 400 Kbits per second, based on a
byte-oriented transfer format. It can be used with any Atmel Two-wire Interface bus Serial
EEPROM and I²C compatible device such as Real Time Clock (RTC), Dot Matrix/Graphic LCD
Controllers and Temperature Sensor, to name but a few. The TWI is programmable as a master
or a slave with sequential or single-byte access. Multiple master capability is supported. 20
Arbitration of the bus is performed internally and puts the TWI in slave mode automatically if the
bus arbitration is lost.
A configurable baud rate generator permits the output data rate to be adapted to a wide range of
core clock frequencies.
Below, Table 31-1 lists the compatibility level of the Atmel Two-wire Interface in Master Mode and
a full I2C compatible device.
Atmel TWI compatibility with I2C Standard
Table 31-1.
I2C Standard
Atmel TWI
Standard Mode Speed (100 KHz)
Supported
Fast Mode Speed (400 KHz)
Supported
7 or 10 bits Slave Addressing
Supported
(1)
START BYTE
Not Supported
Repeated Start (Sr) Condition
Supported
ACK and NACK Management
Supported
Slope control and input filtering (Fast mode)
Not Supported
Clock stretching
Supported
Multi Master Capability
Supported
Note:
31.2
1. START + b000000001 + Ack + Sr
Embedded Characteristics
• Compatibility with standard two-wire serial memory
• One, two or three bytes for slave address
• Sequential read/write operations
• Supports either master or slave modes
• Compatible with Standard Two-wire Serial Memories
• Master, Multi-master and Slave Mode Operation
• Bit Rate: Up to 400 Kbits
• General Call Supported in Slave mode
• Connection to Peripheral DMA Controller (PDC) Channel Capabilities Optimizes Data
Transfers in Master Mode Only
– One Channel for the Receiver, One Channel for the Transmitter
– Next Buffer Support
481
6438D–ATARM–13-Oct-09
31.3
List of Abbreviations
Table 31-2.
31.4
Abbreviations
Abbreviation
Description
TWI
Two-wire Interface
A
Acknowledge
NA
Non Acknowledge
P
Stop
S
Start
Sr
Repeated Start
SADR
Slave Address
ADR
Any address except SADR
R
Read
W
Write
Block Diagram
Figure 31-1. Block Diagram
APB Bridge
TWCK
PIO
PMC
MCK
TWD
Two-wire
Interface
TWI
Interrupt
482
AIC
AT91SAM9G45
6438D–ATARM–13-Oct-09
31.5
Application Block Diagram
Figure 31-2. Application Block Diagram
VDD
Rp
Host with
TWI
Interface
Rp
TWD
TWCK
Atmel TWI
Serial EEPROM
Slave 1
I²C RTC
I²C LCD
Controller
I²C Temp.
Sensor
Slave 2
Slave 3
Slave 4
Rp: Pull up value as given by the I²C Standard
31.5.1
I/O Lines Description
Table 31-3.
I/O Lines Description
Pin Name
Pin Description
TWD
Two-wire Serial Data
Input/Output
TWCK
Two-wire Serial Clock
Input/Output
31.6
31.6.1
Type
Product Dependencies
I/O Lines
Both TWD and TWCK are bidirectional lines, connected to a positive supply voltage via a current
source or pull-up resistor (see Figure 31-2 on page 483). When the bus is free, both lines are
high. The output stages of devices connected to the bus must have an open-drain or open-collector to perform the wired-AND function.
TWD and TWCK pins may be multiplexed with PIO lines. To enable the TWI, the programmer
must perform the following step:
• Program the PIO controller to dedicate TWD and TWCK as peripheral lines.
The user must not program TWD and TWCK as open-drain. It is already done by the hardware.
Table 31-4.
483
I/O Lines
Instance
Signal
I/O Line
Peripheral
TWI0
TWCK0
PA21
A
TWI0
TWD0
PA20
A
TWI1
TWCK1
PB11
A
TWI1
TWD1
PB10
A
AT91SAM9G45
6438D–ATARM–13-Oct-09
31.6.2
Power Management
• Enable the peripheral clock.
The TWI interface may be clocked through the Power Management Controller (PMC), thus the
programmer must first configure the PMC to enable the TWI clock.
31.6.3
Interrupt
The TWI interface has an interrupt line connected to the Advanced Interrupt Controller (AIC). In
order to handle interrupts, the AIC must be programmed before configuring the TWI.
Table 31-5.
31.7
31.7.1
Peripheral IDs
Instance
ID
TWI0
12
TWI1
13
Functional Description
Transfer Format
The data put on the TWD line must be 8 bits long. Data is transferred MSB first; each byte must
be followed by an acknowledgement. The number of bytes per transfer is unlimited (see Figure
31-4).
Each transfer begins with a START condition and terminates with a STOP condition (see Figure
31-3).
• A high-to-low transition on the TWD line while TWCK is high defines the START condition.
• A low-to-high transition on the TWD line while TWCK is high defines a STOP condition.
Figure 31-3.
START and STOP Conditions
TWD
TWCK
Start
Stop
Figure 31-4. Transfer Format
TWD
TWCK
Start
31.7.2
Address
R/W
Ack
Data
Ack
Data
Ack
Stop
Modes of Operation
The TWI has six modes of operations:
• Master transmitter mode
• Master receiver mode
484
AT91SAM9G45
6438D–ATARM–13-Oct-09
• Multi-master transmitter mode
• Multi-master receiver mode
• Slave transmitter mode
• Slave receiver mode
These modes are described in the following chapters.
485
AT91SAM9G45
6438D–ATARM–13-Oct-09
31.8
Master Mode
31.8.1
Definition
The Master is the device that starts a transfer, generates a clock and stops it.
31.8.2
Application Block Diagram
Figure 31-5. Master Mode Typical Application Block Diagram
VDD
Rp
Host with
TWI
Interface
Rp
TWD
TWCK
Atmel TWI
Serial EEPROM
Slave 1
I²C RTC
I²C LCD
Controller
I²C Temp.
Sensor
Slave 2
Slave 3
Slave 4
Rp: Pull up value as given by the I²C Standard
31.8.3
Programming Master Mode
The following registers have to be programmed before entering Master mode:
1. DADR (+ IADRSZ + IADR if a 10 bit device is addressed): The device address is used
to access slave devices in read or write mode.
2. CKDIV + CHDIV + CLDIV: Clock Waveform.
3. SVDIS: Disable the slave mode.
4. MSEN: Enable the master mode.
31.8.4
Master Transmitter Mode
After the master initiates a Start condition when writing into the Transmit Holding Register,
TWI_THR, it sends a 7-bit slave address, configured in the Master Mode register (DADR in
TWI_MMR), to notify the slave device. The bit following the slave address indicates the transfer
direction, 0 in this case (MREAD = 0 in TWI_MMR).
The TWI transfers require the slave to acknowledge each received byte. During the acknowledge clock pulse (9th pulse), the master releases the data line (HIGH), enabling the slave to pull
it down in order to generate the acknowledge. The master polls the data line during this clock
pulse and sets the Not Acknowledge bit (NACK) in the status register if the slave does not
acknowledge the byte. As with the other status bits, an interrupt can be generated if enabled in
the interrupt enable register (TWI_IER). If the slave acknowledges the byte, the data written in
the TWI_THR, is then shifted in the internal shifter and transferred. When an acknowledge is
detected, the TXRDY bit is set until a new write in the TWI_THR.
While no new data is written in the TWI_THR, the Serial Clock Line is tied low. When new data is
written in the TWI_THR, the SCL is released and the data is sent. To generate a STOP event,
the STOP command must be performed by writing in the STOP field of TWI_CR.
486
AT91SAM9G45
6438D–ATARM–13-Oct-09
After a Master Write transfer, the Serial Clock line is stretched (tied low) while no new data is
written in the TWI_THR or until a STOP command is performed.
See Figure 31-6, Figure 31-7, and Figure 31-8.
Figure 31-6. Master Write with One Data Byte
STOP Command sent (write in TWI_CR)
S
TWD
DADR
W
A
DATA
A
P
TXCOMP
TXRDY
Write THR (DATA)
Figure 31-7. Master Write with Multiple Data Bytes
STOP command performed
(by writing in the TWI_CR)
TWD
S
DADR
W
A
DATA n
A
DATA n+1
A
DATA n+2
A
P
TWCK
TXCOMP
TXRDY
Write THR (Data n)
Write THR (Data n+1)
487
Write THR (Data n+2)
Last data sent
AT91SAM9G45
6438D–ATARM–13-Oct-09
Figure 31-8. Master Write with One Byte Internal Address and Multiple Data Bytes
STOP command performed
(by writing in the TWI_CR)
TWD
S
DADR
W
A
IADR
A
DATA n
A
DATA n+1
A
DATA n+2
A
P
TWCK
TXCOMP
TXRDY
Write THR (Data n)
Write THR (Data n+1)
31.8.5
Write THR (Data n+2)
Last data sent
Master Receiver Mode
The read sequence begins by setting the START bit. After the start condition has been sent, the
master sends a 7-bit slave address to notify the slave device. The bit following the slave address
indicates the transfer direction, 1 in this case (MREAD = 1 in TWI_MMR). During the acknowledge clock pulse (9th pulse), the master releases the data line (HIGH), enabling the slave to pull
it down in order to generate the acknowledge. The master polls the data line during this clock
pulse and sets the NACK bit in the status register if the slave does not acknowledge the byte.
If an acknowledge is received, the master is then ready to receive data from the slave. After data
has been received, the master sends an acknowledge condition to notify the slave that the data
has been received except for the last data, after the stop condition. See Figure 31-9. When the
RXRDY bit is set in the status register, a character has been received in the receive-holding register (TWI_RHR). The RXRDY bit is reset when reading the TWI_RHR.
When a single data byte read is performed, with or without internal address (IADR), the START
and STOP bits must be set at the same time. See Figure 31-9. When a multiple data byte read is
performed, with or without internal address (IADR), the STOP bit must be set after the next-tolast data received. See Figure 31-10. For Internal Address usage see Section 31.8.6.
Figure 31-9. Master Read with One Data Byte
TWD
S
DADR
R
A
DATA
N
P
TXCOMP
Write START &
STOP Bit
RXRDY
Read RHR
488
AT91SAM9G45
6438D–ATARM–13-Oct-09
Figure 31-10. Master Read with Multiple Data Bytes
TWD
S
DADR
R
A
DATA n
A
DATA (n+1)
A
DATA (n+m)-1
A
DATA (n+m)
N
P
TXCOMP
Write START Bit
RXRDY
Read RHR
DATA n
Read RHR
DATA (n+1)
Read RHR
DATA (n+m)-1
Read RHR
DATA (n+m)
Write STOP Bit
after next-to-last data read
31.8.6
31.8.6.1
Internal Address
The TWI interface can perform various transfer formats: Transfers with 7-bit slave address
devices and 10-bit slave address devices.
7-bit Slave Addressing
When Addressing 7-bit slave devices, the internal address bytes are used to perform random
address (read or write) accesses to reach one or more data bytes, within a memory page location in a serial memory, for example. When performing read operations with an internal address,
the TWI performs a write operation to set the internal address into the slave device, and then
switch to Master Receiver mode. Note that the second start condition (after sending the IADR) is
sometimes called “repeated start” (Sr) in I2C fully-compatible devices. See Figure 31-12. See
Figure 31-11 and Figure 31-13 for Master Write operation with internal address.
The three internal address bytes are configurable through the Master Mode register
(TWI_MMR).
If the slave device supports only a 7-bit address, i.e. no internal address, IADRSZ must be set to
0.
In the figures below the following abbreviations are used:
489
•S
Start
• Sr
Repeated Start
•P
Stop
•W
Write
•R
Read
•A
Acknowledge
•N
Not Acknowledge
• DADR
Device Address
• IADR
Internal Address
AT91SAM9G45
6438D–ATARM–13-Oct-09
Figure 31-11. Master Write with One, Two or Three Bytes Internal Address and One Data Byte
Three bytes internal address
S
TWD
DADR
W
A
IADR(23:16)
A
IADR(15:8)
A
IADR(7:0)
A
W
A
IADR(15:8)
A
IADR(7:0)
A
DATA
A
W
A
IADR(7:0)
A
DATA
A
DATA
A
P
Two bytes internal address
S
TWD
DADR
P
One byte internal address
S
TWD
DADR
P
Figure 31-12. Master Read with One, Two or Three Bytes Internal Address and One Data Byte
Three bytes internal address
S
TWD
DADR
W
A
IADR(23:16)
A
A
IADR(15:8)
IADR(7:0)
A
Sr
DADR
R
A
DATA
N
P
Two bytes internal address
S
TWD
DADR
W
A
IADR(15:8)
A
IADR(7:0)
A
Sr
W
A
IADR(7:0)
A
Sr
R
A
DADR
R
A
DATA
N
P
One byte internal address
TWD
31.8.6.2
S
DADR
DADR
DATA
N
P
10-bit Slave Addressing
For a slave address higher than 7 bits, the user must configure the address size (IADRSZ) and
set the other slave address bits in the internal address register (TWI_IADR). The two remaining
Internal address bytes, IADR[15:8] and IADR[23:16] can be used the same as in 7-bit Slave
Addressing.
Example: Address a 10-bit device (10-bit device address is b1 b2 b3 b4 b5 b6 b7 b8 b9 b10)
1. Program IADRSZ = 1,
2. Program DADR with 1 1 1 1 0 b1 b2 (b1 is the MSB of the 10-bit address, b2, etc.)
3. Program TWI_IADR with b3 b4 b5 b6 b7 b8 b9 b10 (b10 is the LSB of the 10-bit
address)
Figure 31-13 below shows a byte write to an Atmel AT24LC512 EEPROM. This demonstrates
the use of internal addresses to access the device.
Figure 31-13. Internal Address Usage
S
T
A
R
T
Device
Address
W
R
I
T
E
FIRST
WORD ADDRESS
SECOND
WORD ADDRESS
S
T
O
P
DATA
0
M
S
B
490
LR A
S / C
BW K
M
S
B
A
C
K
LA
SC
BK
A
C
K
AT91SAM9G45
6438D–ATARM–13-Oct-09
31.8.7
SMBUS Quick Command (Master Mode Only)
The TWI interface can perform a Quick Command:
1. Configure the master mode (DADR, CKDIV, etc.).
2. Write the MREAD bit in the TWI_MMR register at the value of the one-bit command to
be sent.
3. Start the transfer by setting the QUICK bit in the TWI_CR.
Figure 31-14. SMBUS Quick Command
TWD
S
DADR
R/W
A
P
TXCOMP
TXRDY
Write QUICK command in TWI_CR
31.8.8
491
Read-write Flowcharts
The following flowcharts shown in Figure 31-16 on page 493, Figure 31-17 on page 494, Figure
31-18 on page 495, Figure 31-19 on page 496 and Figure 31-20 on page 497 give examples for
read and write operations. A polling or interrupt method can be used to check the status bits.
The interrupt method requires that the interrupt enable register (TWI_IER) be configured first.
AT91SAM9G45
6438D–ATARM–13-Oct-09
Figure 31-15. TWI Write Operation with Single Data Byte without Internal Address
BEGIN
Set TWI clock
(CLDIV, CHDIV, CKDIV) in TWI_CWGR
(Needed only once)
Set the Control register:
- Master enable
TWI_CR = MSEN + SVDIS
Set the Master Mode register:
- Device slave address (DADR)
- Transfer direction bit
Write ==> bit MREAD = 0
Load Transmit register
TWI_THR = Data to send
Write STOP Command
TWI_CR = STOP
Read Status register
No
TXRDY = 1?
Yes
Read Status register
No
TXCOMP = 1?
Yes
Transfer finished
492
AT91SAM9G45
6438D–ATARM–13-Oct-09
Figure 31-16. TWI Write Operation with Single Data Byte and Internal Address
BEGIN
Set TWI clock
(CLDIV, CHDIV, CKDIV) in TWI_CWGR
(Needed only once)
Set the Control register:
- Master enable
TWI_CR = MSEN + SVDIS
Set the Master Mode register:
- Device slave address (DADR)
- Internal address size (IADRSZ)
- Transfer direction bit
Write ==> bit MREAD = 0
Set the internal address
TWI_IADR = address
Load transmit register
TWI_THR = Data to send
Write STOP command
TWI_CR = STOP
Read Status register
No
TXRDY = 1?
Yes
Read Status register
TXCOMP = 1?
No
Yes
Transfer finished
493
AT91SAM9G45
6438D–ATARM–13-Oct-09
Figure 31-17. TWI Write Operation with Multiple Data Bytes with or without Internal Address
BEGIN
Set TWI clock
(CLDIV, CHDIV, CKDIV) in TWI_CWGR
(Needed only once)
Set the Control register:
- Master enable
TWI_CR = MSEN + SVDIS
Set the Master Mode register:
- Device slave address
- Internal address size (if IADR used)
- Transfer direction bit
Write ==> bit MREAD = 0
No
Internal address size = 0?
Set the internal address
TWI_IADR = address
Yes
Load Transmit register
TWI_THR = Data to send
Read Status register
TWI_THR = data to send
No
TXRDY = 1?
Yes
Data to send?
Yes
Write STOP Command
TWI_CR = STOP
Read Status register
Yes
No
TXCOMP = 1?
END
494
AT91SAM9G45
6438D–ATARM–13-Oct-09
Figure 31-18. TWI Read Operation with Single Data Byte without Internal Address
BEGIN
Set TWI clock
(CLDIV, CHDIV, CKDIV) in TWI_CWGR
(Needed only once)
Set the Control register:
- Master enable
TWI_CR = MSEN + SVDIS
Set the Master Mode register:
- Device slave address
- Transfer direction bit
Read ==> bit MREAD = 1
Start the transfer
TWI_CR = START | STOP
Read status register
RXRDY = 1?
No
Yes
Read Receive Holding Register
Read Status register
No
TXCOMP = 1?
Yes
END
495
AT91SAM9G45
6438D–ATARM–13-Oct-09
Figure 31-19. TWI Read Operation with Single Data Byte and Internal Address
BEGIN
Set TWI clock
(CLDIV, CHDIV, CKDIV) in TWI_CWGR
(Needed only once)
Set the Control register:
- Master enable
TWI_CR = MSEN + SVDIS
Set the Master Mode register:
- Device slave address
- Internal address size (IADRSZ)
- Transfer direction bit
Read ==> bit MREAD = 1
Set the internal address
TWI_IADR = address
Start the transfer
TWI_CR = START | STOP
Read Status register
No
RXRDY = 1?
Yes
Read Receive Holding register
Read Status register
No
TXCOMP = 1?
Yes
END
496
AT91SAM9G45
6438D–ATARM–13-Oct-09
Figure 31-20. TWI Read Operation with Multiple Data Bytes with or without Internal Address
BEGIN
Set TWI clock
(CLDIV, CHDIV, CKDIV) in TWI_CWGR
(Needed only once)
Set the Control register:
- Master enable
TWI_CR = MSEN + SVDIS
Set the Master Mode register:
- Device slave address
- Internal address size (if IADR used)
- Transfer direction bit
Read ==> bit MREAD = 1
Internal address size = 0?
Set the internal address
TWI_IADR = address
Yes
Start the transfer
TWI_CR = START
Read Status register
RXRDY = 1?
No
Yes
Read Receive Holding register (TWI_RHR)
No
Last data to read
but one?
Yes
Stop the transfer
TWI_CR = STOP
Read Status register
No
RXRDY = 1?
Yes
Read Receive Holding register (TWI_RHR)
Read status register
TXCOMP = 1?
No
Yes
END
497
AT91SAM9G45
6438D–ATARM–13-Oct-09
31.9
Multi-master Mode
31.9.1
Definition
More than one master may handle the bus at the same time without data corruption by using
arbitration.
Arbitration starts as soon as two or more masters place information on the bus at the same time,
and stops (arbitration is lost) for the master that intends to send a logical one while the other
master sends a logical zero.
As soon as arbitration is lost by a master, it stops sending data and listens to the bus in order to
detect a stop. When the stop is detected, the master who has lost arbitration may put its data on
the bus by respecting arbitration.
Arbitration is illustrated in Figure 31-22 on page 499.
31.9.2
Different Multi-master Modes
Two multi-master modes may be distinguished:
1. TWI is considered as a Master only and will never be addressed.
2. TWI may be either a Master or a Slave and may be addressed.
Note:
31.9.2.1
In both Multi-master modes arbitration is supported.
TWI as Master Only
In this mode, TWI is considered as a Master only (MSEN is always at one) and must be driven
like a Master with the ARBLST (ARBitration Lost) flag in addition.
If arbitration is lost (ARBLST = 1), the programmer must reinitiate the data transfer.
If the user starts a transfer (ex.: DADR + START + W + Write in THR) and if the bus is busy, the
TWI automatically waits for a STOP condition on the bus to initiate the transfer (see Figure 3121 on page 499).
Note:
31.9.2.2
The state of the bus (busy or free) is not indicated in the user interface.
TWI as Master or Slave
The automatic reversal from Master to Slave is not supported in case of a lost arbitration.
Then, in the case where TWI may be either a Master or a Slave, the programmer must manage
the pseudo Multi-master mode described in the steps below.
1. Program TWI in Slave mode (SADR + MSDIS + SVEN) and perform Slave Access (if
TWI is addressed).
2. If TWI has to be set in Master mode, wait until TXCOMP flag is at 1.
3. Program Master mode (DADR + SVDIS + MSEN) and start the transfer (ex: START +
Write in THR).
4. As soon as the Master mode is enabled, TWI scans the bus in order to detect if it is
busy or free. When the bus is considered as free, TWI initiates the transfer.
5. As soon as the transfer is initiated and until a STOP condition is sent, the arbitration
becomes relevant and the user must monitor the ARBLST flag.
6. If the arbitration is lost (ARBLST is set to 1), the user must program the TWI in Slave
mode in the case where the Master that won the arbitration wanted to access the TWI.
7. If TWI has to be set in Slave mode, wait until TXCOMP flag is at 1 and then program the
Slave mode.
498
AT91SAM9G45
6438D–ATARM–13-Oct-09
Note:
In the case where the arbitration is lost and TWI is addressed, TWI will not acknowledge even if it
is programmed in Slave mode as soon as ARBLST is set to 1. Then, the Master must repeat
SADR.
Figure 31-21. Programmer Sends Data While the Bus is Busy
TWCK
START sent by the TWI
STOP sent by the master
DATA sent by a master
TWD
DATA sent by the TWI
Bus is busy
Bus is free
Transfer is kept
TWI DATA transfer
A transfer is programmed
(DADR + W + START + Write THR)
Bus is considered as free
Transfer is initiated
Figure 31-22. Arbitration Cases
TWCK
TWD
TWCK
Data from a Master
S
1
0 0 1 1
Data from TWI
S
1
0
TWD
S
1
0 0
1
P
Arbitration is lost
TWI stops sending data
1 1
Data from the master
P
Arbitration is lost
S
1
0
S
1
0 0 1
1
S
1
0
1
1
The master stops sending data
0 1
Data from the TWI
ARBLST
Bus is busy
Bus is free
Transfer is kept
TWI DATA transfer
A transfer is programmed
(DADR + W + START + Write THR)
Transfer is stopped
Transfer is programmed again
(DADR + W + START + Write THR)
Bus is considered as free
Transfer is initiated
The flowchart shown in Figure 31-23 on page 500 gives an example of read and write operations
in Multi-master mode.
499
AT91SAM9G45
6438D–ATARM–13-Oct-09
Figure 31-23. Multi-master Flowchart
START
Programm the SLAVE mode:
SADR + MSDIS + SVEN
Read Status Register
Yes
SVACC = 1 ?
GACC = 1 ?
No
No
No
No
SVREAD = 0 ?
EOSACC = 1 ?
TXRDY= 1 ?
Yes
Yes
Yes
No
Write in TWI_THR
TXCOMP = 1 ?
No
RXRDY= 0 ?
Yes
No
No
Yes
Read TWI_RHR
Need to perform
a master access ?
GENERAL CALL TREATMENT
Yes
Decoding of the
programming sequence
No
Prog seq
OK ?
Change SADR
Program the Master mode
DADR + SVDIS + MSEN + CLK + R / W
Read Status Register
Yes
No
ARBLST = 1 ?
Yes
Yes
No
MREAD = 1 ?
RXRDY= 0 ?
TXRDY= 0 ?
No
No
Read TWI_RHR
Yes
Yes
Data to read?
Data to send ?
Yes
Write in TWI_THR
No
No
Stop Transfer
TWI_CR = STOP
Read Status Register
Yes
500
TXCOMP = 0 ?
No
AT91SAM9G45
6438D–ATARM–13-Oct-09
31.10 Slave Mode
31.10.1
Definition
The Slave Mode is defined as a mode where the device receives the clock and the address from
another device called the master.
In this mode, the device never initiates and never completes the transmission (START,
REPEATED_START and STOP conditions are always provided by the master).
31.10.2
Application Block Diagram
Figure 31-24. Slave Mode Typical Application Block Diagram
VDD
R
Master
Host with
TWI
Interface
31.10.3
R
TWD
TWCK
Host with TWI
Interface
Host with TWI
Interface
LCD Controller
Slave 1
Slave 2
Slave 3
Programming Slave Mode
The following fields must be programmed before entering Slave mode:
1. SADR (TWI_SMR): The slave device address is used in order to be accessed by master devices in read or write mode.
2. MSDIS (TWI_CR): Disable the master mode.
3. SVEN (TWI_CR): Enable the slave mode.
As the device receives the clock, values written in TWI_CWGR are not taken into account.
31.10.4
Receiving Data
After a Start or Repeated Start condition is detected and if the address sent by the Master
matches with the Slave address programmed in the SADR (Slave ADdress) field, SVACC (Slave
ACCess) flag is set and SVREAD (Slave READ) indicates the direction of the transfer.
SVACC remains high until a STOP condition or a repeated START is detected. When such a
condition is detected, EOSACC (End Of Slave ACCess) flag is set.
31.10.4.1
Read Sequence
In the case of a Read sequence (SVREAD is high), TWI transfers data written in the TWI_THR
(TWI Transmit Holding Register) until a STOP condition or a REPEATED_START + an address
different from SADR is detected. Note that at the end of the read sequence TXCOMP (Transmission Complete) flag is set and SVACC reset.
As soon as data is written in the TWI_THR, TXRDY (Transmit Holding Register Ready) flag is
reset, and it is set when the shift register is empty and the sent data acknowledged or not. If the
data is not acknowledged, the NACK flag is set.
501
AT91SAM9G45
6438D–ATARM–13-Oct-09
Note that a STOP or a repeated START always follows a NACK.
See Figure 31-25 on page 503.
31.10.4.2
Write Sequence
In the case of a Write sequence (SVREAD is low), the RXRDY (Receive Holding Register
Ready) flag is set as soon as a character has been received in the TWI_RHR (TWI Receive
Holding Register). RXRDY is reset when reading the TWI_RHR.
TWI continues receiving data until a STOP condition or a REPEATED_START + an address different from SADR is detected. Note that at the end of the write sequence TXCOMP flag is set
and SVACC reset.
See Figure 31-26 on page 503.
31.10.4.3
Clock Synchronization Sequence
In the case where TWI_THR or TWI_RHR is not written/read in time, TWI performs a clock
synchronization.
Clock stretching information is given by the SCLWS (Clock Wait state) bit.
See Figure 31-28 on page 505 and Figure 31-29 on page 506.
31.10.4.4
General Call
In the case where a GENERAL CALL is performed, GACC (General Call ACCess) flag is set.
After GACC is set, it is up to the programmer to interpret the meaning of the GENERAL CALL
and to decode the new address programming sequence.
See Figure 31-27 on page 504.
31.10.4.5
31.10.5
31.10.5.1
Data Transfer
Read Operation
The read mode is defined as a data requirement from the master.
After a START or a REPEATED START condition is detected, the decoding of the address
starts. If the slave address (SADR) is decoded, SVACC is set and SVREAD indicates the direction of the transfer.
Until a STOP or REPEATED START condition is detected, TWI continues sending data loaded
in the TWI_THR register.
If a STOP condition or a REPEATED START + an address different from SADR is detected,
SVACC is reset.
Figure 31-25 on page 503 describes the write operation.
502
AT91SAM9G45
6438D–ATARM–13-Oct-09
Figure 31-25. Read Access Ordered by a MASTER
SADR matches,
TWI answers with an ACK
SADR does not match,
TWI answers with a NACK
TWD
S
ADR
R
NA
DATA
NA
P/S/Sr
SADR R
A
DATA
A
ACK/NACK from the Master
A
DATA
NA
S/Sr
TXRDY
Read RHR
Write THR
NACK
SVACC
SVREAD
SVREAD has to be taken into account only while SVACC is active
EOSVACC
Notes:
1. When SVACC is low, the state of SVREAD becomes irrelevant.
2. TXRDY is reset when data has been transmitted from TWI_THR to the shift register and set when this data has been
acknowledged or non acknowledged.
31.10.5.2
Write Operation
The write mode is defined as a data transmission from the master.
After a START or a REPEATED START, the decoding of the address starts. If the slave address
is decoded, SVACC is set and SVREAD indicates the direction of the transfer (SVREAD is low in
this case).
Until a STOP or REPEATED START condition is detected, TWI stores the received data in the
TWI_RHR register.
If a STOP condition or a REPEATED START + an address different from SADR is detected,
SVACC is reset.
Figure 31-26 on page 503 describes the Write operation.
Figure 31-26. Write Access Ordered by a Master
SADR does not match,
TWI answers with a NACK
TWD
S
ADR
W
NA
DATA
NA
SADR matches,
TWI answers with an ACK
P/S/Sr
SADR W
A
DATA
A
Read RHR
A
DATA
NA
S/Sr
RXRDY
SVACC
SVREAD
SVREAD has to be taken into account only while SVACC is active
EOSVACC
Notes:
1. When SVACC is low, the state of SVREAD becomes irrelevant.
2. RXRDY is set when data has been transmitted from the shift register to the TWI_RHR and reset when this data is read.
503
AT91SAM9G45
6438D–ATARM–13-Oct-09
31.10.5.3
General Call
The general call is performed in order to change the address of the slave.
If a GENERAL CALL is detected, GACC is set.
After the detection of General Call, it is up to the programmer to decode the commands which
come afterwards.
In case of a WRITE command, the programmer has to decode the programming sequence and
program a new SADR if the programming sequence matches.
Figure 31-27 on page 504 describes the General Call access.
Figure 31-27. Master Performs a General Call
0000000 + W
TXD
S
GENERAL CALL
RESET command = 00000110X
WRITE command = 00000100X
A
Reset or write DADD
A
DATA1
A
DATA2
A
New SADR
A
P
New SADR
Programming sequence
GCACC
Reset after read
SVACC
Note:
504
This method allows the user to create an own programming sequence by choosing the programming bytes and the number of them. The programming sequence has to be provided to the
master.
AT91SAM9G45
6438D–ATARM–13-Oct-09
31.10.5.4
Clock Synchronization
In both read and write modes, it may happen that TWI_THR/TWI_RHR buffer is not filled /emptied before the emission/reception of a new character. In this case, to avoid sending/receiving
undesired data, a clock stretching mechanism is implemented.
31.10.5.5
Clock Synchronization in Read Mode
The clock is tied low if the shift register is empty and if a STOP or REPEATED START condition
was not detected. It is tied low until the shift register is loaded.
Figure 31-28 on page 505 describes the clock synchronization in Read mode.
Figure 31-28. Clock Synchronization in Read Mode
TWI_THR
DATA0
S
SADR
R
DATA1
1
A
DATA0
A
DATA1
DATA2
A
XXXXXXX
DATA2
NA
S
2
TWCK
Write THR
CLOCK is tied low by the TWI
as long as THR is empty
SCLWS
TXRDY
SVACC
SVREAD
As soon as a START is detected
TXCOMP
TWI_THR is transmitted to the shift register
Notes:
Ack or Nack from the master
1
The data is memorized in TWI_THR until a new value is written
2
The clock is stretched after the ACK, the state of TWD is undefined during clock stretching
1. TXRDY is reset when data has been written in the TWI_THR to the shift register and set when this data has been acknowledged or non acknowledged.
2. At the end of the read sequence, TXCOMP is set after a STOP or after a REPEATED_START + an address different from
SADR.
3. SCLWS is automatically set when the clock synchronization mechanism is started.
505
AT91SAM9G45
6438D–ATARM–13-Oct-09
31.10.5.6
Clock Synchronization in Write Mode
The c lock is tied lo w if the shift register and the TWI_RHR is full. If a STOP or
REPEATED_START condition was not detected, it is tied low until TWI_RHR is read.
Figure 31-29 on page 506 describes the clock synchronization in Read mode.
Figure 31-29. Clock Synchronization in Write Mode
TWCK
CLOCK is tied low by the TWI as long as RHR is full
TWD
S
SADR
W
A
DATA0
TWI_RHR
A
DATA1
A
DATA0 is not read in the RHR
DATA2
DATA1
NA
S
ADR
DATA2
SCLWS
SCL is stretched on the last bit of DATA1
RXRDY
Rd DATA0
Rd DATA1
Rd DATA2
SVACC
SVREAD
TXCOMP
Notes:
As soon as a START is detected
1. At the end of the read sequence, TXCOMP is set after a STOP or after a REPEATED_START + an address different from
SADR.
2. SCLWS is automatically set when the clock synchronization mechanism is started and automatically reset when the mechanism is finished.
506
AT91SAM9G45
6438D–ATARM–13-Oct-09
31.10.5.7
Reversal after a Repeated Start
31.10.5.8
Reversal of Read to Write
The master initiates the communication by a read command and finishes it by a write command.
Figure 31-30 on page 507 describes the repeated start + reversal from Read to Write mode.
Figure 31-30. Repeated Start + Reversal from Read to Write Mode
TWI_THR
TWD
DATA0
S
SADR
R
A
DATA0
DATA1
A
DATA1
NA
Sr
SADR
W
A
DATA2
TWI_RHR
A
DATA3
DATA2
A
P
DATA3
SVACC
SVREAD
TXRDY
RXRDY
EOSACC
Cleared after read
As soon as a START is detected
TXCOMP
1. TXCOMP is only set at the end of the transmission because after the repeated start, SADR is detected again.
31.10.5.9
Reversal of Write to Read
The master initiates the communication by a write command and finishes it by a read command.Figure 31-31 on page 507 describes the repeated start + reversal from Write to Read
mode.
Figure 31-31. Repeated Start + Reversal from Write to Read Mode
DATA2
TWI_THR
TWD
S
SADR
W
A
DATA0
TWI_RHR
A
DATA1
DATA0
A
Sr
SADR
R
A
DATA3
DATA2
A
DATA3
NA
P
DATA1
SVACC
SVREAD
TXRDY
RXRDY
EOSACC
TXCOMP
Notes:
Read TWI_RHR
Cleared after read
As soon as a START is detected
1. In this case, if TWI_THR has not been written at the end of the read command, the clock is automatically stretched before
the ACK.
2. TXCOMP is only set at the end of the transmission because after the repeated start, SADR is detected again.
507
AT91SAM9G45
6438D–ATARM–13-Oct-09
31.10.6
Read Write Flowcharts
The flowchart shown in Figure 31-32 on page 508 gives an example of read and write operations
in Slave mode. A polling or interrupt method can be used to check the status bits. The interrupt
method requires that the interrupt enable register (TWI_IER) be configured first.
Figure 31-32. Read Write Flowchart in Slave Mode
Set the SLAVE mode:
SADR + MSDIS + SVEN
Read Status Register
SVACC = 1 ?
No
No
EOSACC = 1 ?
GACC = 1 ?
No
SVREAD = 0 ?
No
TXRDY= 1 ?
No
Write in TWI_THR
No
TXCOMP = 1 ?
RXRDY= 0 ?
No
END
Read TWI_RHR
GENERAL CALL TREATMENT
Decoding of the
programming sequence
Prog seq
OK ?
No
Change SADR
508
AT91SAM9G45
6438D–ATARM–13-Oct-09
31.11 Two-wire Interface (TWI) User Interface
Table 31-6.
Register Mapping
Offset
Register
Name
Access
Reset
0x00
Control Register
TWI_CR
Write-only
N/A
0x04
Master Mode Register
TWI_MMR
Read-write
0x00000000
0x08
Slave Mode Register
TWI_SMR
Read-write
0x00000000
0x0C
Internal Address Register
TWI_IADR
Read-write
0x00000000
0x10
Clock Waveform Generator Register
TWI_CWGR
Read-write
0x00000000
0x20
Status Register
TWI_SR
Read-only
0x0000F009
0x24
Interrupt Enable Register
TWI_IER
Write-only
N/A
0x28
Interrupt Disable Register
TWI_IDR
Write-only
N/A
0x2C
Interrupt Mask Register
TWI_IMR
Read-only
0x00000000
0x30
Receive Holding Register
TWI_RHR
Read-only
0x00000000
0x34
Transmit Holding Register
TWI_THR
Write-only
0x00000000
0x38 - 0xFC
Reserved
–
–
–
–
–
–
509
AT91SAM9G45
6438D–ATARM–13-Oct-09
31.11.1
Name:
TWI Control Register
TWI_CR
Addresses: 0xFFF84000 (0), 0xFFF88000 (1)
Access:
Write-only
Reset:
0x00000000
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
SWRST
6
QUICK
5
SVDIS
4
SVEN
3
MSDIS
2
MSEN
1
STOP
0
START
• START: Send a START Condition
0 = No effect.
1 = A frame beginning with a START bit is transmitted according to the features defined in the mode register.
This action is necessary when the TWI peripheral wants to read data from a slave. When configured in Master Mode with a
write operation, a frame is sent as soon as the user writes a character in the Transmit Holding Register (TWI_THR).
• STOP: Send a STOP Condition
0 = No effect.
1 = STOP Condition is sent just after completing the current byte transmission in master read mode.
– In single data byte master read, the START and STOP must both be set.
– In multiple data bytes master read, the STOP must be set after the last data received but one.
– In master read mode, if a NACK bit is received, the STOP is automatically performed.
– In master data write operation, a STOP condition will be sent after the transmission of the current data is
finished.
• MSEN: TWI Master Mode Enabled
0 = No effect.
1 = If MSDIS = 0, the master mode is enabled.
Note:
Switching from Slave to Master mode is only permitted when TXCOMP = 1.
• MSDIS: TWI Master Mode Disabled
0 = No effect.
1 = The master mode is disabled, all pending data is transmitted. The shifter and holding characters (if it contains data) are
transmitted in case of write operation. In read operation, the character being transferred must be completely received
before disabling.
510
AT91SAM9G45
6438D–ATARM–13-Oct-09
• SVEN: TWI Slave Mode Enabled
0 = No effect.
1 = If SVDIS = 0, the slave mode is enabled.
Note:
Switching from Master to Slave mode is only permitted when TXCOMP = 1.
• SVDIS: TWI Slave Mode Disabled
0 = No effect.
1 = The slave mode is disabled. The shifter and holding characters (if it contains data) are transmitted in case of read operation. In write operation, the character being transferred must be completely received before disabling.
• QUICK: SMBUS Quick Command
0 = No effect.
1 = If Master mode is enabled, a SMBUS Quick Command is sent.
• SWRST: Software Reset
0 = No effect.
1 = Equivalent to a system reset.
511
AT91SAM9G45
6438D–ATARM–13-Oct-09
31.11.2
Name:
TWI Master Mode Register
TWI_MMR
Addresses: 0xFFF84004 (0), 0xFFF88004 (1)
Access:
Read-write
Reset:
0x00000000
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
21
20
19
DADR
18
17
16
15
–
14
–
13
–
12
MREAD
11
–
10
–
9
7
–
6
–
5
–
4
–
3
–
2
–
1
–
8
IADRSZ
0
–
• IADRSZ: Internal Device Address Size
IADRSZ[9:8]
0
0
No internal device address
0
1
One-byte internal device address
1
0
Two-byte internal device address
1
1
Three-byte internal device address
• MREAD: Master Read Direction
0 = Master write direction.
1 = Master read direction.
• DADR: Device Address
The device address is used to access slave devices in read or write mode. Those bits are only used in Master mode.
512
AT91SAM9G45
6438D–ATARM–13-Oct-09
31.11.3
Name:
TWI Slave Mode Register
TWI_SMR
Addresses: 0xFFF84008 (0), 0xFFF88008 (1)
Access:
Read-write
Reset:
0x00000000
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
21
20
19
SADR
18
17
16
15
–
14
–
13
–
12
–
11
–
10
–
9
8
7
–
6
–
5
–
4
–
3
–
2
–
1
–
0
–
• SADR: Slave Address
The slave device address is used in Slave mode in order to be accessed by master devices in read or write mode.
SADR must be programmed before enabling the Slave mode or after a general call. Writes at other times have no effect.
513
AT91SAM9G45
6438D–ATARM–13-Oct-09
31.11.4
Name:
TWI Internal Address Register
TWI_IADR
Addresses: 0xFFF8400C (0), 0xFFF8800C (1)
Access:
Read-write
Reset:
0x00000000
31
–
30
–
29
–
28
–
23
22
21
20
27
–
26
–
25
–
24
–
19
18
17
16
11
10
9
8
3
2
1
0
IADR
15
14
13
12
IADR
7
6
5
4
IADR
• IADR: Internal Address
0, 1, 2 or 3 bytes depending on IADRSZ.
514
AT91SAM9G45
6438D–ATARM–13-Oct-09
31.11.5
Name:
TWI Clock Waveform Generator Register
TWI_CWGR
Addresses: 0xFFF84010 (0), 0xFFF88010 (1)
Access:
Read-write
Reset:
0x00000000
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
22
21
20
19
18
17
CKDIV
16
15
14
13
12
11
10
9
8
3
2
1
0
CHDIV
7
6
5
4
CLDIV
TWI_CWGR is only used in Master mode.
• CLDIV: Clock Low Divider
The SCL low period is defined as follows:
T low = ( ( CLDIV × 2
CKDIV
) + 4 ) × T MCK
• CHDIV: Clock High Divider
The SCL high period is defined as follows:
T high = ( ( CHDIV × 2
CKDIV
) + 4 ) × T MCK
• CKDIV: Clock Divider
The CKDIV is used to increase both SCL high and low periods.
515
AT91SAM9G45
6438D–ATARM–13-Oct-09
31.11.6
Name:
TWI Status Register
TWI_SR
Addresses: 0xFFF84020 (0), 0xFFF88020 (1)
Access:
Read-only
Reset:
0x0000F009
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
14
13
12
11
EOSACC
10
SCLWS
9
ARBLST
8
NACK
7
–
6
OVRE
5
GACC
4
SVACC
3
SVREAD
2
TXRDY
1
RXRDY
0
TXCOMP
• TXCOMP: Transmission Completed (automatically set / reset)
TXCOMP used in Master mode:
0 = During the length of the current frame.
1 = When both holding and shifter registers are empty and STOP condition has been sent.
TXCOMP behavior in Master mode can be seen in Figure 31-8 on page 488 and in Figure 31-10 on page 489.
TXCOMP used in Slave mode:
0 = As soon as a Start is detected.
1 = After a Stop or a Repeated Start + an address different from SADR is detected.
TXCOMP behavior in Slave mode can be seen in Figure 31-28 on page 505, Figure 31-29 on page 506, Figure 31-30 on
page 507 and Figure 31-31 on page 507.
• RXRDY: Receive Holding Register Ready (automatically set / reset)
0 = No character has been received since the last TWI_RHR read operation.
1 = A byte has been received in the TWI_RHR since the last read.
RXRDY behavior in Master mode can be seen in Figure 31-10 on page 489.
RXRDY behavior in Slave mode can be seen in Figure 31-26 on page 503, Figure 31-29 on page 506, Figure 31-30 on
page 507 and Figure 31-31 on page 507.
• TXRDY: Transmit Holding Register Ready (automatically set / reset)
TXRDY used in Master mode:
0 = The transmit holding register has not been transferred into shift register. Set to 0 when writing into TWI_THR register.
1 = As soon as a data byte is transferred from TWI_THR to internal shifter or if a NACK error is detected, TXRDY is set at
the same time as TXCOMP and NACK. TXRDY is also set when MSEN is set (enable TWI).
TXRDY behavior in Master mode can be seen in Figure 31-8 on page 488.
516
AT91SAM9G45
6438D–ATARM–13-Oct-09
TXRDY used in Slave mode:
0 = As soon as data is written in the TWI_THR, until this data has been transmitted and acknowledged (ACK or NACK).
1 = It indicates that the TWI_THR is empty and that data has been transmitted and acknowledged.
If TXRDY is high and if a NACK has been detected, the transmission will be stopped. Thus when TRDY = NACK = 1, the
programmer must not fill TWI_THR to avoid losing it.
TXRDY behavior in Slave mode can be seen in Figure 31-25 on page 503, Figure 31-28 on page 505, Figure 31-30 on
page 507 and Figure 31-31 on page 507.
• SVREAD: Slave Read (automatically set / reset)
This bit is only used in Slave mode. When SVACC is low (no Slave access has been detected) SVREAD is irrelevant.
0 = Indicates that a write access is performed by a Master.
1 = Indicates that a read access is performed by a Master.
SVREAD behavior can be seen in Figure 31-25 on page 503, Figure 31-26 on page 503, Figure 31-30 on page 507 and
Figure 31-31 on page 507.
• SVACC: Slave Access (automatically set / reset)
This bit is only used in Slave mode.
0 = TWI is not addressed. SVACC is automatically cleared after a NACK or a STOP condition is detected.
1 = Indicates that the address decoding sequence has matched (A Master has sent SADR). SVACC remains high until a
NACK or a STOP condition is detected.
SVACC behavior can be seen in Figure 31-25 on page 503, Figure 31-26 on page 503, Figure 31-30 on page 507 and Figure 31-31 on page 507.
• GACC: General Call Access (clear on read)
This bit is only used in Slave mode.
0 = No General Call has been detected.
1 = A General Call has been detected. After the detection of General Call, if need be, the programmer may acknowledge
this access and decode the following bytes and respond according to the value of the bytes.
GACC behavior can be seen in Figure 31-27 on page 504.
• OVRE: Overrun Error (clear on read)
This bit is only used in Master mode.
0 = TWI_RHR has not been loaded while RXRDY was set
1 = TWI_RHR has been loaded while RXRDY was set. Reset by read in TWI_SR when TXCOMP is set.
• NACK: Not Acknowledged (clear on read)
NACK used in Master mode:
0 = Each data byte has been correctly received by the far-end side TWI slave component.
1 = A data byte has not been acknowledged by the slave component. Set at the same time as TXCOMP.
517
AT91SAM9G45
6438D–ATARM–13-Oct-09
NACK used in Slave Read mode:
0 = Each data byte has been correctly received by the Master.
1 = In read mode, a data byte has not been acknowledged by the Master. When NACK is set the programmer must not fill
TWI_THR even if TXRDY is set, because it means that the Master will stop the data transfer or re initiate it.
Note that in Slave Write mode all data are acknowledged by the TWI.
• ARBLST: Arbitration Lost (clear on read)
This bit is only used in Master mode.
0: Arbitration won.
1: Arbitration lost. Another master of the TWI bus has won the multi-master arbitration. TXCOMP is set at the same time.
• SCLWS: Clock Wait State (automatically set / reset)
This bit is only used in Slave mode.
0 = The clock is not stretched.
1 = The clock is stretched. TWI_THR / TWI_RHR buffer is not filled / emptied before the emission / reception of a new
character.
SCLWS behavior can be seen in Figure 31-28 on page 505 and Figure 31-29 on page 506.
• EOSACC: End Of Slave Access (clear on read)
This bit is only used in Slave mode.
0 = A slave access is being performing.
1 = The Slave Access is finished. End Of Slave Access is automatically set as soon as SVACC is reset.
EOSACC behavior can be seen in Figure 31-30 on page 507 and Figure 31-31 on page 507
518
AT91SAM9G45
6438D–ATARM–13-Oct-09
31.11.7
Name:
TWI Interrupt Enable Register
TWI_IER
Addresses: 0xFFF84024 (0), 0xFFF88024 (1)
Access:
Write-only
Reset:
0x00000000
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
14
13
12
11
EOSACC
10
SCL_WS
9
ARBLST
8
NACK
7
–
6
OVRE
5
GACC
4
SVACC
3
–
2
TXRDY
1
RXRDY
0
TXCOMP
• TXCOMP: Transmission Completed Interrupt Enable
• RXRDY: Receive Holding Register Ready Interrupt Enable
• TXRDY: Transmit Holding Register Ready Interrupt Enable
• SVACC: Slave Access Interrupt Enable
• GACC: General Call Access Interrupt Enable
• OVRE: Overrun Error Interrupt Enable
• NACK: Not Acknowledge Interrupt Enable
• ARBLST: Arbitration Lost Interrupt Enable
• SCL_WS: Clock Wait State Interrupt Enable
• EOSACC: End Of Slave Access Interrupt Enable
0 = No effect.
1 = Enables the corresponding interrupt.
519
AT91SAM9G45
6438D–ATARM–13-Oct-09
31.11.8
Name:
TWI Interrupt Disable Register
TWI_IDR
Addresses: 0xFFF84028 (0), 0xFFF88028 (1)
Access:
Write-only
Reset:
0x00000000
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
14
13
12
11
EOSACC
10
SCL_WS
9
ARBLST
8
NACK
7
–
6
OVRE
5
GACC
4
SVACC
3
–
2
TXRDY
1
RXRDY
0
TXCOMP
• TXCOMP: Transmission Completed Interrupt Disable
• RXRDY: Receive Holding Register Ready Interrupt Disable
• TXRDY: Transmit Holding Register Ready Interrupt Disable
• SVACC: Slave Access Interrupt Disable
• GACC: General Call Access Interrupt Disable
• OVRE: Overrun Error Interrupt Disable
• NACK: Not Acknowledge Interrupt Disable
• ARBLST: Arbitration Lost Interrupt Disable
• SCL_WS: Clock Wait State Interrupt Disable
• EOSACC: End Of Slave Access Interrupt Disable
0 = No effect.
1 = Disables the corresponding interrupt.
520
AT91SAM9G45
6438D–ATARM–13-Oct-09
31.11.9
Name:
TWI Interrupt Mask Register
TWI_IMR
Addresses: 0xFFF8402C (0), 0xFFF8802C (1)
Access:
Read-only
Reset:
0x00000000
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
14
13
12
11
EOSACC
10
SCL_WS
9
ARBLST
8
NACK
7
–
6
OVRE
5
GACC
4
SVACC
3
–
2
TXRDY
1
RXRDY
0
TXCOMP
• TXCOMP: Transmission Completed Interrupt Mask
• RXRDY: Receive Holding Register Ready Interrupt Mask
• TXRDY: Transmit Holding Register Ready Interrupt Mask
• SVACC: Slave Access Interrupt Mask
• GACC: General Call Access Interrupt Mask
• OVRE: Overrun Error Interrupt Mask
• NACK: Not Acknowledge Interrupt Mask
• ARBLST: Arbitration Lost Interrupt Mask
• SCL_WS: Clock Wait State Interrupt Mask
• EOSACC: End Of Slave Access Interrupt Mask
0 = The corresponding interrupt is disabled.
1 = The corresponding interrupt is enabled.
521
AT91SAM9G45
6438D–ATARM–13-Oct-09
31.11.10 TWI Receive Holding Register
Name:
TWI_RHR
Addresses: 0xFFF84030 (0), 0xFFF88030 (1)
Access:
Read-only
Reset:
0x00000000
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
6
5
4
3
2
1
0
RXDATA
• RXDATA: Master or Slave Receive Holding Data
31.11.11 TWI Transmit Holding Register
Name:
TWI_THR
Addresses: 0xFFF84034 (0), 0xFFF88034 (1)
Access:
Read-write
Reset:
0x00000000
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
6
5
4
3
2
1
0
TXDATA
• TXDATA: Master or Slave Transmit Holding Data
522
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
32. Universal Synchronous Asynchronous Receiver Transmitter (USART)
32.1
Description
The Universal Synchronous Asynchronous Receiver Transmitter (USART) provides one full
duplex universal synchronous asynchronous serial link. Data frame format is widely programmable (data length, parity, number of stop bits) to support a maximum of standards. The receiver
implements parity error, framing error and overrun error detection. The receiver time-out enables
handling variable-length frames and the transmitter timeguard facilitates communications with
slow remote devices. Multidrop communications are also supported through address bit handling in reception and transmission.
The USART features three test modes: remote loopback, local loopback and automatic echo.
The USART supports specific operating modes providing interfaces on RS485 buses, with
ISO7816 T = 0 or T = 1 smart card slots and infrared transceivers. The hardware handshaking
feature enables an out-of-band flow control by automatic management of the pins RTS and
CTS.
The USART supports the connection to the Peripheral DMA Controller, which enables data
transfers to the transmitter and from the receiver. The PDC provides chained buffer management without any intervention of the processor.
32.2
Embedded Characteristics
• Programmable Baud Rate Generator
• 5- to 9-bit full-duplex synchronous or asynchronous serial communications
– 1, 1.5 or 2 stop bits in Asynchronous Mode or 1 or 2 stop bits in Synchronous Mode
– Parity generation and error detection
– Framing error detection, overrun error detection
– MSB- or LSB-first
– Optional break generation and detection
– By 8 or by-16 over-sampling receiver frequency
– Hardware handshaking RTS-CTS
– Receiver time-out and transmitter timeguard
– Optional Multi-drop Mode with address generation and detection
– Optional Manchester Encoding
• RS485 with driver control signal
• ISO7816, T = 0 or T = 1 Protocols for interfacing with smart cards
– NACK handling, error counter with repetition and iteration limit
• IrDA modulation and demodulation
– Communication at up to 115.2 Kbps
• Test Modes
– Remote Loopback, Local Loopback, Automatic Echo
523
6438D–ATARM–13-Oct-09
32.3
Block Diagram
Figure 32-1. USART Block Diagram
Peripheral DMA
Controller
Channel
Channel
PIO
Controller
USART
RXD
Receiver
RTS
AIC
USART
Interrupt
TXD
Transmitter
CTS
PMC
MCK
DIV
Baud Rate
Generator
SCK
MCK/DIV
User Interface
SLCK
APB
524
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.4
Application Block Diagram
Figure 32-2. Application Block Diagram
IrLAP
PPP
Serial
Driver
Field Bus
Driver
EMV
Driver
IrDA
Driver
USART
RS232
Drivers
RS485
Drivers
Serial
Port
Differential
Bus
Smart
Card
Slot
IrDA
Transceivers
525
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.5
I/O Lines Description
Table 32-1.
I/O Line Description
Name
Description
Type
Active Level
SCK
Serial Clock
I/O
TXD
Transmit Serial Data
I/O
RXD
Receive Serial Data
Input
CTS
Clear to Send
Input
Low
RTS
Request to Send
Output
Low
526
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.6
32.6.1
Product Dependencies
I/O Lines
The pins used for interfacing the USART may be multiplexed with the PIO lines. The programmer must first program the PIO controller to assign the desired USART pins to their peripheral
function. If I/O lines of the USART are not used by the application, they can be used for other
purposes by the PIO Controller.
To prevent the TXD line from falling when the USART is disabled, the use of an internal pull up
is mandatory. If the hardware handshaking feature is used, the internal pull up on TXD must
also be enabled.
Table 32-2.
32.6.2
I/O Lines
Instance
Signal
I/O Line
Peripheral
USART0
CTS0
PB15
B
USART0
RTS0
PB17
B
USART0
RXD0
PB18
A
USART0
SCK0
PB16
B
USART0
TXD0
PB19
A
USART1
CTS1
PD17
A
USART1
RTS1
PD16
A
USART1
RXD1
PB5
A
USART1
SCK1
PD29
B
USART1
TXD1
PB4
A
USART2
CTS2
PC11
B
USART2
RTS2
PC9
B
USART2
RXD2
PB7
A
USART2
SCK2
PD30
B
USART2
TXD2
PB6
A
USART3
CTS3
PA24
B
USART3
RTS3
PA23
B
USART3
RXD3
PB9
A
USART3
SCK3
PA22
B
USART3
TXD3
PB8
A
Power Management
The USART is not continuously clocked. The programmer must first enable the USART Clock in
the Power Management Controller (PMC) before using the USART. However, if the application
does not require USART operations, the USART clock can be stopped when not needed and be
restarted later. In this case, the USART will resume its operations where it left off.
Configuring the USART does not require the USART clock to be enabled.
527
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.6.3
Interrupt
The USART interrupt line is connected on one of the internal sources of the Advanced Inter-rupt
Controller. Using the USART interrupt requires the AIC to be programmed first. Note that it is not
recommended to use the USART interrupt line in edge sensitive mode.
Table 32-3.
32.7
Peripheral IDs
Instance
ID
USART0
7
USART1
8
USART2
9
USART3
10
Functional Description
The USART is capable of managing several types of serial synchronous or asynchronous
communications.
It supports the following communication modes:
• 5- to 9-bit full-duplex asynchronous serial communication
– MSB- or LSB-first
– 1, 1.5 or 2 stop bits
– Parity even, odd, marked, space or none
– By 8 or by 16 over-sampling receiver frequency
– Optional hardware handshaking
– Optional break management
– Optional multidrop serial communication
• High-speed 5- to 9-bit full-duplex synchronous serial communication
– MSB- or LSB-first
– 1 or 2 stop bits
– Parity even, odd, marked, space or none
– By 8 or by 16 over-sampling frequency
– Optional hardware handshaking
– Optional break management
– Optional multidrop serial communication
• RS485 with driver control signal
• ISO7816, T0 or T1 protocols for interfacing with smart cards
– NACK handling, error counter with repetition and iteration limit
• InfraRed IrDA Modulation and Demodulation
• Test modes
– Remote loopback, local loopback, automatic echo
528
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.7.1
Baud Rate Generator
The Baud Rate Generator provides the bit period clock named the Baud Rate Clock to both the
receiver and the transmitter.
The Baud Rate Generator clock source can be selected by setting the USCLKS field in the Mode
Register (US_MR) between:
• the Master Clock MCK
• a division of the Master Clock, the divider being product dependent, but generally set to 8
• the external clock, available on the SCK pin
The Baud Rate Generator is based upon a 16-bit divider, which is programmed with the CD field
of the Baud Rate Generator Register (US_BRGR). If CD is programmed at 0, the Baud Rate
Generator does not generate any clock. If CD is programmed at 1, the divider is bypassed and
becomes inactive.
If the external SCK clock is selected, the duration of the low and high levels of the signal provided on the SCK pin must be longer than a Master Clock (MCK) period. The frequency of the
signal provided on SCK must be at least 4.5 times lower than MCK.
Figure 32-3. Baud Rate Generator
USCLKS
MCK
MCK/DIV
SCK
Reserved
CD
CD
SCK
0
1
2
16-bit Counter
FIDI
>1
3
1
0
0
0
SYNC
OVER
Sampling
Divider
0
Baud Rate
Clock
1
1
SYNC
USCLKS = 3
32.7.1.1
Sampling
Clock
Baud Rate in Asynchronous Mode
If the USART is programmed to operate in asynchronous mode, the selected clock is first
divided by CD, which is field programmed in the Baud Rate Generator Register (US_BRGR).
The resulting clock is provided to the receiver as a sampling clock and then divided by 16 or 8,
depending on the programming of the OVER bit in US_MR.
If OVER is set to 1, the receiver sampling is 8 times higher than the baud rate clock. If OVER is
cleared, the sampling is performed at 16 times the baud rate clock.
The following formula performs the calculation of the Baud Rate.
SelectedClock
Baudrate = -------------------------------------------( 8 ( 2 – Over )CD )
This gives a maximum baud rate of MCK divided by 8, assuming that MCK is the highest possible clock and that OVER is programmed at 1.
529
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.7.1.2
Baud Rate Calculation Example
Table 32-4 shows calculations of CD to obtain a baud rate at 38400 bauds for different source
clock frequencies. This table also shows the actual resulting baud rate and the error.
Table 32-4.
Baud Rate Example (OVER = 0)
Source Clock
Expected Baud
Rate
MHz
Bit/s
3 686 400
38 400
6.00
6
38 400.00
0.00%
4 915 200
38 400
8.00
8
38 400.00
0.00%
5 000 000
38 400
8.14
8
39 062.50
1.70%
7 372 800
38 400
12.00
12
38 400.00
0.00%
8 000 000
38 400
13.02
13
38 461.54
0.16%
12 000 000
38 400
19.53
20
37 500.00
2.40%
12 288 000
38 400
20.00
20
38 400.00
0.00%
14 318 180
38 400
23.30
23
38 908.10
1.31%
14 745 600
38 400
24.00
24
38 400.00
0.00%
18 432 000
38 400
30.00
30
38 400.00
0.00%
24 000 000
38 400
39.06
39
38 461.54
0.16%
24 576 000
38 400
40.00
40
38 400.00
0.00%
25 000 000
38 400
40.69
40
38 109.76
0.76%
32 000 000
38 400
52.08
52
38 461.54
0.16%
32 768 000
38 400
53.33
53
38 641.51
0.63%
33 000 000
38 400
53.71
54
38 194.44
0.54%
40 000 000
38 400
65.10
65
38 461.54
0.16%
50 000 000
38 400
81.38
81
38 580.25
0.47%
Calculation Result
CD
Actual Baud Rate
Error
Bit/s
The baud rate is calculated with the following formula:
BaudRate = MCK ⁄ CD × 16
The baud rate error is calculated with the following formula. It is not recommended to work with
an error higher than 5%.
ExpectedBaudRate
Error = 1 – ⎛⎝ ---------------------------------------------------⎞⎠
ActualBaudRate
32.7.1.3
Fractional Baud Rate in Asynchronous Mode
The Baud Rate generator previously defined is subject to the following limitation: the output frequency changes by only integer multiples of the reference frequency. An approach to this
problem is to integrate a fractional N clock generator that has a high resolution. The generator
architecture is modified to obtain Baud Rate changes by a fraction of the reference source clock.
This fractional part is programmed with the FP field in the Baud Rate Generator Register
(US_BRGR). If FP is not 0, the fractional part is activated. The resolution is one eighth of the
530
6438D–ATARM–13-Oct-09
AT91SAM9G45
clock divider. This feature is only available when using USART normal mode. The fractional
Baud Rate is calculated using the following formula:
SelectedClock
Baudrate = ---------------------------------------------------------------⎛ 8 ( 2 – Over ) ⎛ CD + FP
⎞⎞
-----⎝
⎝
8 ⎠⎠
The modified architecture is presented below:
Figure 32-4. Fractional Baud Rate Generator
FP
USCLKS
CD
Modulus
Control
FP
MCK
MCK/DIV
SCK
Reserved
CD
SCK
0
1
2
3
16-bit Counter
glitch-free
logic
1
0
FIDI
>1
0
0
SYNC
OVER
Sampling
Divider
0
Baud Rate
Clock
1
1
SYNC
USCLKS = 3
32.7.1.4
Sampling
Clock
Baud Rate in Synchronous Mode
If the USART is programmed to operate in synchronous mode, the selected clock is simply
divided by the field CD in US_BRGR.
BaudRate = SelectedClock
-------------------------------------CD
In synchronous mode, if the external clock is selected (USCLKS = 3), the clock is provided
directly by the signal on the USART SCK pin. No division is active. The value written in
US_BRGR has no effect. The external clock frequency must be at least 4.5 times lower than the
system clock. In synchronous mode master (USCLKS = 0 or 1, CLK0 set to 1), the receive part
limits the SCK maximum frequency to MCK/4.5,
When either the external clock SCK or the internal clock divided (MCK/DIV) is selected, the
value programmed in CD must be even if the user has to ensure a 50:50 mark/space ratio on the
SCK pin. If the internal clock MCK is selected, the Baud Rate Generator ensures a 50:50 duty
cycle on the SCK pin, even if the value programmed in CD is odd.
531
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.7.1.5
Baud Rate in ISO 7816 Mode
The ISO7816 specification defines the bit rate with the following formula:
Di
B = ------ × f
Fi
where:
• B is the bit rate
• Di is the bit-rate adjustment factor
• Fi is the clock frequency division factor
• f is the ISO7816 clock frequency (Hz)
Di is a binary value encoded on a 4-bit field, named DI, as represented in Table 32-5.
Table 32-5.
Binary and Decimal Values for Di
DI field
0001
0010
0011
0100
0101
0110
1000
1001
1
2
4
8
16
32
12
20
Di (decimal)
Fi is a binary value encoded on a 4-bit field, named FI, as represented in Table 32-6.
Table 32-6.
Binary and Decimal Values for Fi
FI field
0000
0001
0010
0011
0100
0101
0110
1001
1010
1011
1100
1101
Fi (decimal
372
372
558
744
1116
1488
1860
512
768
1024
1536
2048
Table 32-7 shows the resulting Fi/Di Ratio, which is the ratio between the ISO7816 clock and the
baud rate clock.
Table 32-7.
Possible Values for the Fi/Di Ratio
Fi/Di
372
558
774
1116
1488
1806
512
768
1024
1536
2048
1
372
558
744
1116
1488
1860
512
768
1024
1536
2048
2
186
279
372
558
744
930
256
384
512
768
1024
4
93
139.5
186
279
372
465
128
192
256
384
512
8
46.5
69.75
93
139.5
186
232.5
64
96
128
192
256
16
23.25
34.87
46.5
69.75
93
116.2
32
48
64
96
128
32
11.62
17.43
23.25
34.87
46.5
58.13
16
24
32
48
64
12
31
46.5
62
93
124
155
42.66
64
85.33
128
170.6
20
18.6
27.9
37.2
55.8
74.4
93
25.6
38.4
51.2
76.8
102.4
If the USART is configured in ISO7816 Mode, the clock selected by the USCLKS field in the
Mode Register (US_MR) is first divided by the value programmed in the field CD in the Baud
Rate Generator Register (US_BRGR). The resulting clock can be provided to the SCK pin to
feed the smart card clock inputs. This means that the CLKO bit can be set in US_MR.
This clock is then divided by the value programmed in the FI_DI_RATIO field in the FI_DI_Ratio
register (US_FIDI). This is performed by the Sampling Divider, which performs a division by up
to 2047 in ISO7816 Mode. The non-integer values of the Fi/Di Ratio are not supported and the
user must program the FI_DI_RATIO field to a value as close as possible to the expected value.
The FI_DI_RATIO field resets to the value 0x174 (372 in decimal) and is the most common
divider between the ISO7816 clock and the bit rate (Fi = 372, Di = 1).
532
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 32-5 shows the relation between the Elementary Time Unit, corresponding to a bit time,
and the ISO 7816 clock.
Figure 32-5. Elementary Time Unit (ETU)
FI_DI_RATIO
ISO7816 Clock Cycles
ISO7816 Clock
on SCK
ISO7816 I/O Line
on TXD
1 ETU
32.7.2
Receiver and Transmitter Control
After reset, the receiver is disabled. The user must enable the receiver by setting the RXEN bit
in the Control Register (US_CR). However, the receiver registers can be programmed before the
receiver clock is enabled.
After reset, the transmitter is disabled. The user must enable it by setting the TXEN bit in the
Control Register (US_CR). However, the transmitter registers can be programmed before being
enabled.
The Receiver and the Transmitter can be enabled together or independently.
At any time, the software can perform a reset on the receiver or the transmitter of the USART by
setting the corresponding bit, RSTRX and RSTTX respectively, in the Control Register
(US_CR). The software resets clear the status flag and reset internal state machines but the
user interface configuration registers hold the value configured prior to software reset. Regardless of what the receiver or the transmitter is performing, the communication is immediately
stopped.
The user can also independently disable the receiver or the transmitter by setting RXDIS and
TXDIS respectively in US_CR. If the receiver is disabled during a character reception, the
USART waits until the end of reception of the current character, then the reception is stopped. If
the transmitter is disabled while it is operating, the USART waits the end of transmission of both
the current character and character being stored in the Transmit Holding Register (US_THR). If
a timeguard is programmed, it is handled normally.
32.7.3
32.7.3.1
Synchronous and Asynchronous Modes
Transmitter Operations
The transmitter performs the same in both synchronous and asynchronous operating modes
(SYNC = 0 or SYNC = 1). One start bit, up to 9 data bits, one optional parity bit and up to two
stop bits are successively shifted out on the TXD pin at each falling edge of the programmed
serial clock.
The number of data bits is selected by the CHRL field and the MODE 9 bit in the Mode Register
(US_MR). Nine bits are selected by setting the MODE 9 bit regardless of the CHRL field. The
parity bit is set according to the PAR field in US_MR. The even, odd, space, marked or none
parity bit can be configured. The MSBF field in US_MR configures which data bit is sent first. If
written at 1, the most significant bit is sent first. At 0, the less significant bit is sent first. The num533
6438D–ATARM–13-Oct-09
AT91SAM9G45
ber of stop bits is selected by the NBSTOP field in US_MR. The 1.5 stop bit is supported in
asynchronous mode only.
Figure 32-6. Character Transmit
Example: 8-bit, Parity Enabled One Stop
Baud Rate
Clock
TXD
Start
Bit
D0
D1
D2
D3
D4
D5
D6
D7
Parity
Bit
Stop
Bit
The characters are sent by writing in the Transmit Holding Register (US_THR). The transmitter
reports two status bits in the Channel Status Register (US_CSR): TXRDY (Transmitter Ready),
which indicates that US_THR is empty and TXEMPTY, which indicates that all the characters
written in US_THR have been processed. When the current character processing is completed,
the last character written in US_THR is transferred into the Shift Register of the transmitter and
US_THR becomes empty, thus TXRDY rises.
Both TXRDY and TXEMPTY bits are low when the transmitter is disabled. Writing a character in
US_THR while TXRDY is low has no effect and the written character is lost.
Figure 32-7. Transmitter Status
Baud Rate
Clock
TXD
Start
D0
Bit
D1
D2
D3
D4
D5
D6
D7
Parity Stop Start
D0
Bit Bit Bit
D1
D2
D3
D4
D5
D6
D7
Parity Stop
Bit Bit
Write
US_THR
TXRDY
TXEMPTY
32.7.3.2
Manchester Encoder
When the Manchester encoder is in use, characters transmitted through the USART are
encoded based on biphase Manchester II format. To enable this mode, set the MAN field in the
US_MR register to 1. Depending on polarity configuration, a logic level (zero or one), is transmitted as a coded signal one-to-zero or zero-to-one. Thus, a transition always occurs at the
midpoint of each bit time. It consumes more bandwidth than the original NRZ signal (2x) but the
receiver has more error control since the expected input must show a change at the center of a
bit cell. An example of Manchester encoded sequence is: the byte 0xB1 or 10110001 encodes
to 10 01 10 10 01 01 01 10, assuming the default polarity of the encoder. Figure 32-8 illustrates
this coding scheme.
534
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 32-8. NRZ to Manchester Encoding
NRZ
encoded
data
Manchester
encoded
data
1
0
1
1
0
0
0
1
Txd
The Manchester encoded character can also be encapsulated by adding both a configurable
preamble and a start frame delimiter pattern. Depending on the configuration, the preamble is a
training sequence, composed of a pre-defined pattern with a programmable length from 1 to 15
bit times. If the preamble length is set to 0, the preamble waveform is not generated prior to any
character. The preamble pattern is chosen among the following sequences: ALL_ONE,
ALL_ZERO, ONE_ZERO or ZERO_ONE, writing the field TX_PP in the US_MAN register, the
field TX_PL is used to configure the preamble length. Figure 32-9 illustrates and defines the
valid patterns. To improve flexibility, the encoding scheme can be configured using the
TX_MPOL field in the US_MAN register. If the TX_MPOL field is set to zero (default), a logic
zero is encoded with a zero-to-one transition and a logic one is encoded with a one-to-zero transition. If the TX_MPOL field is set to one, a logic one is encoded with a one-to-zero transition
and a logic zero is encoded with a zero-to-one transition.
Figure 32-9. Preamble Patterns, Default Polarity Assumed
Manchester
encoded
data
Txd
SFD
DATA
SFD
DATA
SFD
DATA
SFD
DATA
8 bit width "ALL_ONE" Preamble
Manchester
encoded
data
Txd
8 bit width "ALL_ZERO" Preamble
Manchester
encoded
data
Txd
8 bit width "ZERO_ONE" Preamble
Manchester
encoded
data
Txd
8 bit width "ONE_ZERO" Preamble
A start frame delimiter is to be configured using the ONEBIT field in the US_MR register. It consists of a user-defined pattern that indicates the beginning of a valid data. Figure 32-10
illustrates these patterns. If the start frame delimiter, also known as start bit, is one bit, (ONEBIT
at 1), a logic zero is Manchester encoded and indicates that a new character is being sent serially on the line. If the start frame delimiter is a synchronization pattern also referred to as sync
(ONEBIT at 0), a sequence of 3 bit times is sent serially on the line to indicate the start of a new
character. The sync waveform is in itself an invalid Manchester waveform as the transition
535
6438D–ATARM–13-Oct-09
AT91SAM9G45
occurs at the middle of the second bit time. Two distinct sync patterns are used: the command
sync and the data sync. The command sync has a logic one level for one and a half bit times,
then a transition to logic zero for the second one and a half bit times. If the MODSYNC field in
the US_MR register is set to 1, the next character is a command. If it is set to 0, the next character is a data. When direct memory access is used, the MODSYNC field can be immediately
updated with a modified character located in memory. To enable this mode, VAR_SYNC field in
US_MR register must be set to 1. In this case, the MODSYNC field in US_MR is bypassed and
the sync configuration is held in the TXSYNH in the US_THR register. The USART character format is modified and includes sync information.
Figure 32-10. Start Frame Delimiter
Preamble Length
is set to 0
SFD
Manchester
encoded
data
DATA
Txd
One bit start frame delimiter
SFD
Manchester
encoded
data
DATA
Txd
SFD
Manchester
encoded
data
Txd
Command Sync
start frame delimiter
DATA
Data Sync
start frame delimiter
32.7.3.3
Drift Compensation
Drift compensation is available only in 16X oversampling mode. An hardware recovery system
allows a larger clock drift. To enable the hardware system, the bit in the USART_MAN register
must be set. If the RXD edge is one 16X clock cycle from the expected edge, this is considered
as normal jitter and no corrective actions is taken. If the RXD event is between 4 and 2 clock
cycles before the expected edge, then the current period is shortened by one clock cycle. If the
RXD event is between 2 and 3 clock cycles after the expected edge, then the current period is
lengthened by one clock cycle. These intervals are considered to be drift and so corrective
actions are automatically taken.
536
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 32-11. Bit Resynchronization
Oversampling
16x Clock
RXD
Sampling
point
Expected edge
Synchro.
Error
32.7.3.4
Synchro.
Jump
Tolerance
Sync
Jump
Synchro.
Error
Asynchronous Receiver
If the USART is programmed in asynchronous operating mode (SYNC = 0), the receiver oversamples the RXD input line. The oversampling is either 16 or 8 times the Baud Rate clock,
depending on the OVER bit in the Mode Register (US_MR).
The receiver samples the RXD line. If the line is sampled during one half of a bit time at 0, a start
bit is detected and data, parity and stop bits are successively sampled on the bit rate clock.
If the oversampling is 16, (OVER at 0), a start is detected at the eighth sample at 0. Then, data
bits, parity bit and stop bit are sampled on each 16 sampling clock cycle. If the oversampling is 8
(OVER at 1), a start bit is detected at the fourth sample at 0. Then, data bits, parity bit and stop
bit are sampled on each 8 sampling clock cycle.
The number of data bits, first bit sent and parity mode are selected by the same fields and bits
as the transmitter, i.e. respectively CHRL, MODE9, MSBF and PAR. For the synchronization
mechanism only, the number of stop bits has no effect on the receiver as it considers only one
stop bit, regardless of the field NBSTOP, so that resynchronization between the receiver and the
transmitter can occur. Moreover, as soon as the stop bit is sampled, the receiver starts looking
for a new start bit so that resynchronization can also be accomplished when the transmitter is
operating with one stop bit.
Figure 32-12 and Figure 32-13 illustrate start detection and character reception when USART
operates in asynchronous mode.
537
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 32-12. Asynchronous Start Detection
Baud Rate
Clock
Sampling
Clock (x16)
RXD
Sampling
1
2
3
4
5
6
7
8
1
2
3
4
2
3
4
5
6
7
8
9 10 11 12 13 14 15 16
D0
Sampling
Start
Detection
RXD
Sampling
1
2
3
4
5
6
7
0 1
Start
Rejection
Figure 32-13. Asynchronous Character Reception
Example: 8-bit, Parity Enabled
Baud Rate
Clock
RXD
Start
Detection
16
16
16
16
16
16
16
16
16
16
samples samples samples samples samples samples samples samples samples samples
D0
32.7.3.5
D1
D2
D3
D4
D5
D6
D7
Parity
Bit
Stop
Bit
Manchester Decoder
When the MAN field in US_MR register is set to 1, the Manchester decoder is enabled. The
decoder performs both preamble and start frame delimiter detection. One input line is dedicated
to Manchester encoded input data.
An optional preamble sequence can be defined, its length is user-defined and totally independent of the emitter side. Use RX_PL in US_MAN register to configure the length of the preamble
sequence. If the length is set to 0, no preamble is detected and the function is disabled. In addition, the polarity of the input stream is programmable with RX_MPOL field in US_MAN register.
Depending on the desired application the preamble pattern matching is to be defined via the
RX_PP field in US_MAN. See Figure 32-9 for available preamble patterns.
Unlike preamble, the start frame delimiter is shared between Manchester Encoder and Decoder.
So, if ONEBIT field is set to 1, only a zero encoded Manchester can be detected as a valid start
frame delimiter. If ONEBIT is set to 0, only a sync pattern is detected as a valid start frame
delimiter. Decoder operates by detecting transition on incoming stream. If RXD is sampled during one quarter of a bit time at zero, a start bit is detected. See Figure 32-14. The sample pulse
rejection mechanism applies.
538
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 32-14. Asynchronous Start Bit Detection
Sampling
Clock
(16 x)
Manchester
encoded
data
Txd
Start
Detection
1
2
3
4
The receiver is activated and starts Preamble and Frame Delimiter detection, sampling the data
at one quarter and then three quarters. If a valid preamble pattern or start frame delimiter is
detected, the receiver continues decoding with the same synchronization. If the stream does not
match a valid pattern or a valid start frame delimiter, the receiver re-synchronizes on the next
valid edge.The minimum time threshold to estimate the bit value is three quarters of a bit time.
If a valid preamble (if used) followed with a valid start frame delimiter is detected, the incoming
stream is decoded into NRZ data and passed to USART for processing. Figure 32-15 illustrates
Manchester pattern mismatch. When incoming data stream is passed to the USART, the
receiver is also able to detect Manchester code violation. A code violation is a lack of transition
in the middle of a bit cell. In this case, MANE flag in US_CSR register is raised. It is cleared by
writing the Control Register (US_CR) with the RSTSTA bit at 1. See Figure 32-16 for an example of Manchester error detection during data phase.
Figure 32-15. Preamble Pattern Mismatch
Preamble Mismatch
Manchester coding error
Manchester
encoded
data
Preamble Mismatch
invalid pattern
SFD
Txd
DATA
Preamble Length is set to 8
Figure 32-16. Manchester Error Flag
Preamble Length
is set to 4
Elementary character bit time
SFD
Manchester
encoded
data
Txd
Entering USART character area
sampling points
Preamble subpacket
and Start Frame Delimiter
were successfully
decoded
Manchester
Coding Error
detected
When the start frame delimiter is a sync pattern (ONEBIT field at 0), both command and data
delimiter are supported. If a valid sync is detected, the received character is written as RXCHR
539
6438D–ATARM–13-Oct-09
AT91SAM9G45
field in the US_RHR register and the RXSYNH is updated. RXCHR is set to 1 when the received
character is a command, and it is set to 0 if the received character is a data. This mechanism
alleviates and simplifies the direct memory access as the character contains its own sync field in
the same register.
As the decoder is setup to be used in unipolar mode, the first bit of the frame has to be a zero-toone transition.
32.7.3.6
Radio Interface: Manchester Encoded USART Application
This section describes low data rate RF transmission systems and their integration with a Manchester encoded USART. These systems are based on transmitter and receiver ICs that support
ASK and FSK modulation schemes.
The goal is to perform full duplex radio transmission of characters using two different frequency
carriers. See the configuration in Figure 32-17.
Figure 32-17. Manchester Encoded Characters RF Transmission
Fup frequency Carrier
ASK/FSK
Upstream Receiver
Upstream
Emitter
LNA
VCO
RF filter
Demod
Serial
Configuration
Interface
control
Fdown frequency Carrier
bi-dir
line
Manchester
decoder
USART
Receiver
Manchester
encoder
USART
Emitter
ASK/FSK
downstream transmitter
Downstream
Receiver
PA
RF filter
Mod
VCO
control
The USART module is configured as a Manchester encoder/decoder. Looking at the downstream communication channel, Manchester encoded characters are serially sent to the RF
emitter. This may also include a user defined preamble and a start frame delimiter. Mostly, preamble is used in the RF receiver to distinguish between a valid data from a transmitter and
signals due to noise. The Manchester stream is then modulated. See Figure 32-18 for an example of ASK modulation scheme. When a logic one is sent to the ASK modulator, the power
amplifier, referred to as PA, is enabled and transmits an RF signal at downstream frequency.
When a logic zero is transmitted, the RF signal is turned off. If the FSK modulator is activated,
two different frequencies are used to transmit data. When a logic 1 is sent, the modulator outputs an RF signal at frequency F0 and switches to F1 if the data sent is a 0. See Figure 32-19.
From the receiver side, another carrier frequency is used. The RF receiver performs a bit check
operation examining demodulated data stream. If a valid pattern is detected, the receiver
540
6438D–ATARM–13-Oct-09
AT91SAM9G45
switches to receiving mode. The demodulated stream is sent to the Manchester decoder.
Because of bit checking inside RF IC, the data transferred to the microcontroller is reduced by a
user-defined number of bits. The Manchester preamble length is to be defined in accordance
with the RF IC configuration.
Figure 32-18. ASK Modulator Output
1
0
0
1
0
0
1
NRZ stream
Manchester
encoded
data
default polarity
unipolar output
Txd
ASK Modulator
Output
Uptstream Frequency F0
Figure 32-19. FSK Modulator Output
1
NRZ stream
Manchester
encoded
data
default polarity
unipolar output
Txd
FSK Modulator
Output
Uptstream Frequencies
[F0, F0+offset]
32.7.3.7
Synchronous Receiver
In synchronous mode (SYNC = 1), the receiver samples the RXD signal on each rising edge of
the Baud Rate Clock. If a low level is detected, it is considered as a start. All data bits, the parity
bit and the stop bits are sampled and the receiver waits for the next start bit. Synchronous mode
operations provide a high speed transfer capability.
Configuration fields and bits are the same as in asynchronous mode.
Figure 32-20 illustrates a character reception in synchronous mode.
Figure 32-20. Synchronous Mode Character Reception
Example: 8-bit, Parity Enabled 1 Stop
Baud Rate
Clock
RXD
Sampling
Start
D0
D1
D2
D3
D4
D5
D6
Stop Bit
D7
Parity Bit
541
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.7.3.8
Receiver Operations
When a character reception is completed, it is transferred to the Receive Holding Register
(US_RHR) and the RXRDY bit in the Status Register (US_CSR) rises. If a character is completed while the RXRDY is set, the OVRE (Overrun Error) bit is set. The last character is
transferred into US_RHR and overwrites the previous one. The OVRE bit is cleared by writing
the Control Register (US_CR) with the RSTSTA (Reset Status) bit at 1.
Figure 32-21. Receiver Status
Baud Rate
Clock
RXD
Start
D0
Bit
D1
D2
D3
D4
D5
D6
D7
Parity Stop Start
D0
Bit Bit Bit
D1
D2
D3
D4
D5
D6
D7
Parity Stop
Bit Bit
RSTSTA = 1
Write
US_CR
Read
US_RHR
RXRDY
OVRE
542
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.7.3.9
Parity
The USART supports five parity modes selected by programming the PAR field in the Mode
Register (US_MR). The PAR field also enables the Multidrop mode, see “Multidrop Mode” on
page 544. Even and odd parity bit generation and error detection are supported.
If even parity is selected, the parity generator of the transmitter drives the parity bit at 0 if a number of 1s in the character data bit is even, and at 1 if the number of 1s is odd. Accordingly, the
receiver parity checker counts the number of received 1s and reports a parity error if the sampled parity bit does not correspond. If odd parity is selected, the parity generator of the
transmitter drives the parity bit at 1 if a number of 1s in the character data bit is even, and at 0 if
the number of 1s is odd. Accordingly, the receiver parity checker counts the number of received
1s and reports a parity error if the sampled parity bit does not correspond. If the mark parity is
used, the parity generator of the transmitter drives the parity bit at 1 for all characters. The
receiver parity checker reports an error if the parity bit is sampled at 0. If the space parity is
used, the parity generator of the transmitter drives the parity bit at 0 for all characters. The
receiver parity checker reports an error if the parity bit is sampled at 1. If parity is disabled, the
transmitter does not generate any parity bit and the receiver does not report any parity error.
Table 32-8 shows an example of the parity bit for the character 0x41 (character ASCII “A”)
depending on the configuration of the USART. Because there are two bits at 1, 1 bit is added
when a parity is odd, or 0 is added when a parity is even.
Table 32-8.
Parity Bit Examples
Character
Hexa
Binary
Parity Bit
Parity Mode
A
0x41
0100 0001
1
Odd
A
0x41
0100 0001
0
Even
A
0x41
0100 0001
1
Mark
A
0x41
0100 0001
0
Space
A
0x41
0100 0001
None
None
When the receiver detects a parity error, it sets the PARE (Parity Error) bit in the Channel Status
Register (US_CSR). The PARE bit can be cleared by writing the Control Register (US_CR) with
the RSTSTA bit at 1. Figure 32-22 illustrates the parity bit status setting and clearing.
543
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 32-22. Parity Error
Baud Rate
Clock
RXD
Start
D0
Bit
D1
D2
D3
D4
D5
D6
D7
Bad Stop
Parity Bit
Bit
RSTSTA = 1
Write
US_CR
PARE
RXRDY
32.7.3.10
Multidrop Mode
If the PAR field in the Mode Register (US_MR) is programmed to the value 0x6 or 0x07, the
USART runs in Multidrop Mode. This mode differentiates the data characters and the address
characters. Data is transmitted with the parity bit at 0 and addresses are transmitted with the
parity bit at 1.
If the USART is configured in multidrop mode, the receiver sets the PARE parity error bit when
the parity bit is high and the transmitter is able to send a character with the parity bit high when
the Control Register is written with the SENDA bit at 1.
To handle parity error, the PARE bit is cleared when the Control Register is written with the bit
RSTSTA at 1.
The transmitter sends an address byte (parity bit set) when SENDA is written to US_CR. In this
case, the next byte written to US_THR is transmitted as an address. Any character written in
US_THR without having written the command SENDA is transmitted normally with the parity at
0.
32.7.3.11
Transmitter Timeguard
The timeguard feature enables the USART interface with slow remote devices.
The timeguard function enables the transmitter to insert an idle state on the TXD line between
two characters. This idle state actually acts as a long stop bit.
The duration of the idle state is programmed in the TG field of the Transmitter Timeguard Register (US_TTGR). When this field is programmed at zero no timeguard is generated. Otherwise,
the transmitter holds a high level on TXD after each transmitted byte during the number of bit
periods programmed in TG in addition to the number of stop bits.
As illustrated in Figure 32-23, the behavior of TXRDY and TXEMPTY status bits is modified by
the programming of a timeguard. TXRDY rises only when the start bit of the next character is
sent, and thus remains at 0 during the timeguard transmission if a character has been written in
US_THR. TXEMPTY remains low until the timeguard transmission is completed as the timeguard is part of the current character being transmitted.
544
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 32-23. Timeguard Operations
TG = 4
TG = 4
Baud Rate
Clock
TXD
Start
D0
Bit
D1
D2
D3
D4
D5
D6
D7
Parity Stop
Bit Bit
Start
D0
Bit
D1
D2
D3
D4
D5
D6
D7
Parity Stop
Bit Bit
Write
US_THR
TXRDY
TXEMPTY
Table 32-9 indicates the maximum length of a timeguard period that the transmitter can handle
in relation to the function of the Baud Rate.
Table 32-9.
32.7.3.12
Maximum Timeguard Length Depending on Baud Rate
Baud Rate
Bit time
Timeguard
Bit/sec
μs
ms
1 200
833
212.50
9 600
104
26.56
14400
69.4
17.71
19200
52.1
13.28
28800
34.7
8.85
33400
29.9
7.63
56000
17.9
4.55
57600
17.4
4.43
115200
8.7
2.21
Receiver Time-out
The Receiver Time-out provides support in handling variable-length frames. This feature detects
an idle condition on the RXD line. When a time-out is detected, the bit TIMEOUT in the Channel
Status Register (US_CSR) rises and can generate an interrupt, thus indicating to the driver an
end of frame.
The time-out delay period (during which the receiver waits for a new character) is programmed
in the TO field of the Receiver Time-out Register (US_RTOR). If the TO field is programmed at
0, the Receiver Time-out is disabled and no time-out is detected. The TIMEOUT bit in US_CSR
remains at 0. Otherwise, the receiver loads a 16-bit counter with the value programmed in TO.
This counter is decremented at each bit period and reloaded each time a new character is
received. If the counter reaches 0, the TIMEOUT bit in the Status Register rises. Then, the user
can either:
• Stop the counter clock until a new character is received. This is performed by writing the
Control Register (US_CR) with the STTTO (Start Time-out) bit at 1. In this case, the idle state
545
6438D–ATARM–13-Oct-09
AT91SAM9G45
on RXD before a new character is received will not provide a time-out. This prevents having
to handle an interrupt before a character is received and allows waiting for the next idle state
on RXD after a frame is received.
• Obtain an interrupt while no character is received. This is performed by writing US_CR with
the RETTO (Reload and Start Time-out) bit at 1. If RETTO is performed, the counter starts
counting down immediately from the value TO. This enables generation of a periodic interrupt
so that a user time-out can be handled, for example when no key is pressed on a keyboard.
If STTTO is performed, the counter clock is stopped until a first character is received. The idle
state on RXD before the start of the frame does not provide a time-out. This prevents having to
obtain a periodic interrupt and enables a wait of the end of frame when the idle state on RXD is
detected.
If RETTO is performed, the counter starts counting down immediately from the value TO. This
enables generation of a periodic interrupt so that a user time-out can be handled, for example
when no key is pressed on a keyboard.
Figure 32-24 shows the block diagram of the Receiver Time-out feature.
Figure 32-24. Receiver Time-out Block Diagram
TO
Baud Rate
Clock
1
D
Q
Clock
16-bit Time-out
Counter
16-bit
Value
=
STTTO
Character
Received
Clear
Load
TIMEOUT
0
RETTO
Table 32-10 gives the maximum time-out period for some standard baud rates.
Table 32-10. Maximum Time-out Period
Baud Rate
Bit Time
Time-out
bit/sec
μs
ms
600
1 667
109 225
1 200
833
54 613
2 400
417
27 306
4 800
208
13 653
9 600
104
6 827
14400
69
4 551
19200
52
3 413
28800
35
2 276
33400
30
1 962
546
6438D–ATARM–13-Oct-09
AT91SAM9G45
Table 32-10. Maximum Time-out Period (Continued)
32.7.3.13
Baud Rate
Bit Time
Time-out
56000
18
1 170
57600
17
1 138
200000
5
328
Framing Error
The receiver is capable of detecting framing errors. A framing error happens when the stop bit of
a received character is detected at level 0. This can occur if the receiver and the transmitter are
fully desynchronized.
A framing error is reported on the FRAME bit of the Channel Status Register (US_CSR). The
FRAME bit is asserted in the middle of the stop bit as soon as the framing error is detected. It is
cleared by writing the Control Register (US_CR) with the RSTSTA bit at 1.
Figure 32-25. Framing Error Status
Baud Rate
Clock
RXD
Start
D0
Bit
D1
D2
D3
D4
D5
D6
D7
Parity Stop
Bit Bit
RSTSTA = 1
Write
US_CR
FRAME
RXRDY
32.7.3.14
Transmit Break
The user can request the transmitter to generate a break condition on the TXD line. A break condition drives the TXD line low during at least one complete character. It appears the same as a
0x00 character sent with the parity and the stop bits at 0. However, the transmitter holds the
TXD line at least during one character until the user requests the break condition to be removed.
A break is transmitted by writing the Control Register (US_CR) with the STTBRK bit at 1. This
can be performed at any time, either while the transmitter is empty (no character in either the
Shift Register or in US_THR) or when a character is being transmitted. If a break is requested
while a character is being shifted out, the character is first completed before the TXD line is held
low.
Once STTBRK command is requested further STTBRK commands are ignored until the end of
the break is completed.
The break condition is removed by writing US_CR with the STPBRK bit at 1. If the STPBRK is
requested before the end of the minimum break duration (one character, including start, data,
parity and stop bits), the transmitter ensures that the break condition completes.
547
6438D–ATARM–13-Oct-09
AT91SAM9G45
The transmitter considers the break as though it is a character, i.e. the STTBRK and STPBRK
commands are taken into account only if the TXRDY bit in US_CSR is at 1 and the start of the
break condition clears the TXRDY and TXEMPTY bits as if a character is processed.
Writing US_CR with the both STTBRK and STPBRK bits at 1 can lead to an unpredictable
result. All STPBRK commands requested without a previous STTBRK command are ignored. A
byte written into the Transmit Holding Register while a break is pending, but not started, is
ignored.
After the break condition, the transmitter returns the TXD line to 1 for a minimum of 12 bit times.
Thus, the transmitter ensures that the remote receiver detects correctly the end of break and the
start of the next character. If the timeguard is programmed with a value higher than 12, the TXD
line is held high for the timeguard period.
After holding the TXD line for this period, the transmitter resumes normal operations.
Figure 32-26 illustrates the effect of both the Start Break (STTBRK) and Stop Break (STPBRK)
commands on the TXD line.
Figure 32-26. Break Transmission
Baud Rate
Clock
TXD
Start
D0
Bit
D1
D2
D3
D4
D5
STTBRK = 1
D6
D7
Parity Stop
Bit Bit
Break Transmission
End of Break
STPBRK = 1
Write
US_CR
TXRDY
TXEMPTY
32.7.3.15
Receive Break
The receiver detects a break condition when all data, parity and stop bits are low. This corresponds to detecting a framing error with data at 0x00, but FRAME remains low.
When the low stop bit is detected, the receiver asserts the RXBRK bit in US_CSR. This bit may
be cleared by writing the Control Register (US_CR) with the bit RSTSTA at 1.
An end of receive break is detected by a high level for at least 2/16 of a bit period in asynchronous operating mode or one sample at high level in synchronous operating mode. The end of
break detection also asserts the RXBRK bit.
32.7.3.16
Hardware Handshaking
The USART features a hardware handshaking out-of-band flow control. The RTS and CTS pins
are used to connect with the remote device, as shown in Figure 32-27.
548
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 32-27. Connection with a Remote Device for Hardware Handshaking
USART
Remote
Device
TXD
RXD
RXD
TXD
CTS
RTS
RTS
CTS
Setting the USART to operate with hardware handshaking is performed by writing the
USART_MODE field in the Mode Register (US_MR) to the value 0x2.
The USART behavior when hardware handshaking is enabled is the same as the behavior in
standard synchronous or asynchronous mode, except that the receiver drives the RTS pin as
described below and the level on the CTS pin modifies the behavior of the transmitter as
described below. Using this mode requires using the PDC channel for reception. The transmitter
can handle hardware handshaking in any case.
Figure 32-28 shows how the receiver operates if hardware handshaking is enabled. The RTS
pin is driven high if the receiver is disabled and if the status RXBUFF (Receive Buffer Full) coming from the PDC channel is high. Normally, the remote device does not start transmitting while
its CTS pin (driven by RTS) is high. As soon as the Receiver is enabled, the RTS falls, indicating
to the remote device that it can start transmitting. Defining a new buffer to the PDC clears the
status bit RXBUFF and, as a result, asserts the pin RTS low.
Figure 32-28. Receiver Behavior when Operating with Hardware Handshaking
RXD
RXEN = 1
RXDIS = 1
Write
US_CR
RTS
RXBUFF
Figure 32-29 shows how the transmitter operates if hardware handshaking is enabled. The CTS
pin disables the transmitter. If a character is being processing, the transmitter is disabled only
after the completion of the current character and transmission of the next character happens as
soon as the pin CTS falls.
Figure 32-29. Transmitter Behavior when Operating with Hardware Handshaking
CTS
TXD
549
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.7.4
ISO7816 Mode
The USART features an ISO7816-compatible operating mode. This mode permits interfacing
with smart cards and Security Access Modules (SAM) communicating through an ISO7816 link.
Both T = 0 and T = 1 protocols defined by the ISO7816 specification are supported.
Setting the USART in ISO7816 mode is performed by writing the USART_MODE field in the
Mode Register (US_MR) to the value 0x4 for protocol T = 0 and to the value 0x5 for protocol T =
1.
32.7.4.1
ISO7816 Mode Overview
The ISO7816 is a half duplex communication on only one bidirectional line. The baud rate is
determined by a division of the clock provided to the remote device (see “Baud Rate Generator”
on page 529).
The USART connects to a smart card as shown in Figure 32-30. The TXD line becomes bidirectional and the Baud Rate Generator feeds the ISO7816 clock on the SCK pin. As the TXD pin
becomes bidirectional, its output remains driven by the output of the transmitter but only when
the transmitter is active while its input is directed to the input of the receiver. The USART is considered as the master of the communication as it generates the clock.
Figure 32-30. Connection of a Smart Card to the USART
USART
SCK
TXD
CLK
I/O
Smart
Card
When operating in ISO7816, either in T = 0 or T = 1 modes, the character format is fixed. The
configuration is 8 data bits, even parity and 1 or 2 stop bits, regardless of the values programmed in the CHRL, MODE9, PAR and CHMODE fields. MSBF can be used to transmit LSB
or MSB first. Parity Bit (PAR) can be used to transmit in normal or inverse mode. Refer to
“USART Mode Register” on page 561 and “PAR: Parity Type” on page 562.
The USART cannot operate concurrently in both receiver and transmitter modes as the communication is unidirectional at a time. It has to be configured according to the required mode by
enabling or disabling either the receiver or the transmitter as desired. Enabling both the receiver
and the transmitter at the same time in ISO7816 mode may lead to unpredictable results.
The ISO7816 specification defines an inverse transmission format. Data bits of the character
must be transmitted on the I/O line at their negative value. The USART does not support this format and the user has to perform an exclusive OR on the data before writing it in the Transmit
Holding Register (US_THR) or after reading it in the Receive Holding Register (US_RHR).
32.7.4.2
Protocol T = 0
In T = 0 protocol, a character is made up of one start bit, eight data bits, one parity bit and one
guard time, which lasts two bit times. The transmitter shifts out the bits and does not drive the
I/O line during the guard time.
If no parity error is detected, the I/O line remains at 1 during the guard time and the transmitter
can continue with the transmission of the next character, as shown in Figure 32-31.
550
6438D–ATARM–13-Oct-09
AT91SAM9G45
If a parity error is detected by the receiver, it drives the I/O line at 0 during the guard time, as
shown in Figure 32-32. This error bit is also named NACK, for Non Acknowledge. In this case,
the character lasts 1 bit time more, as the guard time length is the same and is added to the
error bit time which lasts 1 bit time.
When the USART is the receiver and it detects an error, it does not load the erroneous character
in the Receive Holding Register (US_RHR). It appropriately sets the PARE bit in the Status Register (US_SR) so that the software can handle the error.
Figure 32-31. T = 0 Protocol without Parity Error
Baud Rate
Clock
RXD
Start
Bit
D0
D2
D1
D4
D3
D5
D6
D7
Parity Guard Guard Next
Bit Time 1 Time 2 Start
Bit
Figure 32-32. T = 0 Protocol with Parity Error
Baud Rate
Clock
Error
I/O
Start
Bit
D0
D1
D2
D3
D4
D5
D6
D7
Parity Guard
Bit Time 1
Guard Start
Time 2 Bit
D0
D1
Repetition
32.7.4.3
Receive Error Counter
The USART receiver also records the total number of errors. This can be read in the Number of
Error (US_NER) register. The NB_ERRORS field can record up to 255 errors. Reading US_NER
automatically clears the NB_ERRORS field.
32.7.4.4
Receive NACK Inhibit
The USART can also be configured to inhibit an error. This can be achieved by setting the
INACK bit in the Mode Register (US_MR). If INACK is at 1, no error signal is driven on the I/O
line even if a parity bit is detected, but the INACK bit is set in the Status Register (US_SR). The
INACK bit can be cleared by writing the Control Register (US_CR) with the RSTNACK bit at 1.
Moreover, if INACK is set, the erroneous received character is stored in the Receive Holding
Register, as if no error occurred. However, the RXRDY bit does not raise.
32.7.4.5
Transmit Character Repetition
When the USART is transmitting a character and gets a NACK, it can automatically repeat the
character before moving on to the next one. Repetition is enabled by writing the
MAX_ITERATION field in the Mode Register (US_MR) at a value higher than 0. Each character
can be transmitted up to eight times; the first transmission plus seven repetitions.
If MAX_ITERATION does not equal zero, the USART repeats the character as many times as
the value loaded in MAX_ITERATION.
551
6438D–ATARM–13-Oct-09
AT91SAM9G45
When the USART repetition number reaches MAX_ITERATION, the ITERATION bit is set in the
Channel Status Register (US_CSR). If the repetition of the character is acknowledged by the
receiver, the repetitions are stopped and the iteration counter is cleared.
The ITERATION bit in US_CSR can be cleared by writing the Control Register with the RSIT bit
at 1.
32.7.4.6
Disable Successive Receive NACK
The receiver can limit the number of successive NACKs sent back to the remote transmitter.
This is programmed by setting the bit DSNACK in the Mode Register (US_MR). The maximum
number of NACK transmitted is programmed in the MAX_ITERATION field. As soon as
MAX_ITERATION is reached, the character is considered as correct, an acknowledge is sent on
the line and the ITERATION bit in the Channel Status Register is set.
32.7.4.7
Protocol T = 1
When operating in ISO7816 protocol T = 1, the transmission is similar to an asynchronous format with only one stop bit. The parity is generated when transmitting and checked when
receiving. Parity error detection sets the PARE bit in the Channel Status Register (US_CSR).
32.7.5
IrDA Mode
The USART features an IrDA mode supplying half-duplex point-to-point wireless communication. It embeds the modulator and demodulator which allows a glueless connection to the
infrared transceivers, as shown in Figure 32-33. The modulator and demodulator are compliant
with the IrDA specification version 1.1 and support data transfer speeds ranging from 2.4 Kb/s to
115.2 Kb/s.
The USART IrDA mode is enabled by setting the USART_MODE field in the Mode Register
(US_MR) to the value 0x8. The IrDA Filter Register (US_IF) allows configuring the demodulator
filter. The USART transmitter and receiver operate in a normal asynchronous mode and all
parameters are accessible. Note that the modulator and the demodulator are activated.
Figure 32-33. Connection to IrDA Transceivers
USART
IrDA
Transceivers
Receiver
Demodulator
RXD
Transmitter
Modulator
TXD
RX
TX
The receiver and the transmitter must be enabled or disabled according to the direction of the
transmission to be managed.
To receive IrDA signals, the following needs to be done:
• Disable TX and Enable RX
552
6438D–ATARM–13-Oct-09
AT91SAM9G45
• Configure the TXD pin as PIO and set it as an output at 0 (to avoid LED emission). Disable
the internal pull-up (better for power consumption).
• Receive data
32.7.5.1
IrDA Modulation
For baud rates up to and including 115.2 Kbits/sec, the RZI modulation scheme is used. “0” is
represented by a light pulse of 3/16th of a bit time. Some examples of signal pulse duration are
shown in Table 32-11.
Table 32-11. IrDA Pulse Duration
Baud Rate
Pulse Duration (3/16)
2.4 Kb/s
78.13 μs
9.6 Kb/s
19.53 μs
19.2 Kb/s
9.77 μs
38.4 Kb/s
4.88 μs
57.6 Kb/s
3.26 μs
115.2 Kb/s
1.63 μs
Figure 32-34 shows an example of character transmission.
Figure 32-34. IrDA Modulation
Start
Bit
Transmitter
Output
0
Stop
Bit
Data Bits
1
0
1
0
1
0
1
0
1
TXD
3
16 Bit Period
Bit Period
32.7.5.2
IrDA Baud Rate
Table 32-12 gives some examples of CD values, baud rate error and pulse duration. Note that
the requirement on the maximum acceptable error of ±1.87% must be met.
Table 32-12. IrDA Baud Rate Error
Peripheral Clock
Baud Rate
CD
Baud Rate Error
Pulse Time
3 686 400
115 200
2
0.00%
1.63
20 000 000
115 200
11
1.38%
1.63
32 768 000
115 200
18
1.25%
1.63
40 000 000
115 200
22
1.38%
1.63
3 686 400
57 600
4
0.00%
3.26
20 000 000
57 600
22
1.38%
3.26
32 768 000
57 600
36
1.25%
3.26
553
6438D–ATARM–13-Oct-09
AT91SAM9G45
Table 32-12. IrDA Baud Rate Error (Continued)
Peripheral Clock
32.7.5.3
Baud Rate
CD
Baud Rate Error
Pulse Time
40 000 000
57 600
43
0.93%
3.26
3 686 400
38 400
6
0.00%
4.88
20 000 000
38 400
33
1.38%
4.88
32 768 000
38 400
53
0.63%
4.88
40 000 000
38 400
65
0.16%
4.88
3 686 400
19 200
12
0.00%
9.77
20 000 000
19 200
65
0.16%
9.77
32 768 000
19 200
107
0.31%
9.77
40 000 000
19 200
130
0.16%
9.77
3 686 400
9 600
24
0.00%
19.53
20 000 000
9 600
130
0.16%
19.53
32 768 000
9 600
213
0.16%
19.53
40 000 000
9 600
260
0.16%
19.53
3 686 400
2 400
96
0.00%
78.13
20 000 000
2 400
521
0.03%
78.13
32 768 000
2 400
853
0.04%
78.13
IrDA Demodulator
The demodulator is based on the IrDA Receive filter comprised of an 8-bit down counter which is
loaded with the value programmed in US_IF. When a falling edge is detected on the RXD pin,
the Filter Counter starts counting down at the Master Clock (MCK) speed. If a rising edge is
detected on the RXD pin, the counter stops and is reloaded with US_IF. If no rising edge is
detected when the counter reaches 0, the input of the receiver is driven low during one bit time.
Figure 32-35 illustrates the operations of the IrDA demodulator.
Figure 32-35. IrDA Demodulator Operations
MCK
RXD
Counter
Value
Receiver
Input
6
5
4 3
Pulse
Rejected
2
6
6
5
4
3
2
1
0
Pulse
Accepted
As the IrDA mode uses the same logic as the ISO7816, note that the FI_DI_RATIO field in
US_FIDI must be set to a value higher than 0 in order to assure IrDA communications operate
correctly.
554
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.7.6
RS485 Mode
The USART features the RS485 mode to enable line driver control. While operating in RS485
mode, the USART behaves as though in asynchronous or synchronous mode and configuration
of all the parameters is possible. The difference is that the RTS pin is driven high when the
transmitter is operating. The behavior of the RTS pin is controlled by the TXEMPTY bit. A typical
connection of the USART to a RS485 bus is shown in Figure 32-36.
Figure 32-36. Typical Connection to a RS485 Bus
USART
RXD
Differential
Bus
TXD
RTS
The USART is set in RS485 mode by programming the USART_MODE field in the Mode Register (US_MR) to the value 0x1.
The RTS pin is at a level inverse to the TXEMPTY bit. Significantly, the RTS pin remains high
when a timeguard is programmed so that the line can remain driven after the last character completion. Figure 32-37 gives an example of the RTS waveform during a character transmission
when the timeguard is enabled.
Figure 32-37. Example of RTS Drive with Timeguard
TG = 4
Baud Rate
Clock
TXD
Start
D0
Bit
D1
D2
D3
D4
D5
D6
D7
Parity Stop
Bit Bit
Write
US_THR
TXRDY
TXEMPTY
RTS
555
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.7.7
Test Modes
The USART can be programmed to operate in three different test modes. The internal loopback
capability allows on-board diagnostics. In the loopback mode the USART interface pins are disconnected or not and reconfigured for loopback internally or externally.
32.7.7.1
Normal Mode
Normal mode connects the RXD pin on the receiver input and the transmitter output on the TXD
pin.
Figure 32-38. Normal Mode Configuration
RXD
Receiver
TXD
Transmitter
32.7.7.2
Automatic Echo Mode
Automatic echo mode allows bit-by-bit retransmission. When a bit is received on the RXD pin, it
is sent to the TXD pin, as shown in Figure 32-39. Programming the transmitter has no effect on
the TXD pin. The RXD pin is still connected to the receiver input, thus the receiver remains
active.
Figure 32-39. Automatic Echo Mode Configuration
RXD
Receiver
TXD
Transmitter
32.7.7.3
Local Loopback Mode
Local loopback mode connects the output of the transmitter directly to the input of the receiver,
as shown in Figure 32-40. The TXD and RXD pins are not used. The RXD pin has no effect on
the receiver and the TXD pin is continuously driven high, as in idle state.
Figure 32-40. Local Loopback Mode Configuration
RXD
Receiver
Transmitter
1
TXD
556
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.7.7.4
Remote Loopback Mode
Remote loopback mode directly connects the RXD pin to the TXD pin, as shown in Figure 32-41.
The transmitter and the receiver are disabled and have no effect. This mode allows bit-by-bit
retransmission.
Figure 32-41. Remote Loopback Mode Configuration
Receiver
1
RXD
TXD
Transmitter
557
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.8
Universal Synchronous Asynchronous Receiver Transmitter (USART) User Interface
Table 32-14.
Register Mapping
Offset
Register
Name
Access
Reset
0x0000
Control Register
US_CR
Write-only
–
0x0004
Mode Register
US_MR
Read-write
–
0x0008
Interrupt Enable Register
US_IER
Write-only
–
0x000C
Interrupt Disable Register
US_IDR
Write-only
–
0x0010
Interrupt Mask Register
US_IMR
Read-only
0x0
0x0014
Channel Status Register
US_CSR
Read-only
–
0x0018
Receiver Holding Register
US_RHR
Read-only
0x0
0x001C
Transmitter Holding Register
US_THR
Write-only
–
0x0020
Baud Rate Generator Register
US_BRGR
Read-write
0x0
0x0024
Receiver Time-out Register
US_RTOR
Read-write
0x0
0x0028
Transmitter Timeguard Register
US_TTGR
Read-write
0x0
–
–
–
0x2C - 0x3C
Reserved
0x0040
FI DI Ratio Register
US_FIDI
Read-write
0x174
0x0044
Number of Errors Register
US_NER
Read-only
–
0x0048
Reserved
–
–
–
0x004C
IrDA Filter Register
US_IF
Read-write
0x0
0x0050
Manchester Encoder Decoder Register
US_MAN
Read-write
0x30011004
Reserved
–
–
–
Reserved for PDC Registers
–
–
–
0x5C - 0xFC
0x100 - 0x128
558
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.8.1
Name:
USART Control Register
US_CR
Addresses:
0xFFF8C000 (0), 0xFFF90000 (1), 0xFFF94000 (2), 0xFFF98000 (3)
Access:
Write-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
RTSDIS
18
RTSEN
17
–
16
–
15
RETTO
14
RSTNACK
13
RSTIT
12
SENDA
11
STTTO
10
STPBRK
9
STTBRK
8
RSTSTA
7
TXDIS
6
TXEN
5
RXDIS
4
RXEN
3
RSTTX
2
RSTRX
1
–
0
–
• RSTRX: Reset Receiver
0: No effect.
1: Resets the receiver.
• RSTTX: Reset Transmitter
0: No effect.
1: Resets the transmitter.
• RXEN: Receiver Enable
0: No effect.
1: Enables the receiver, if RXDIS is 0.
• RXDIS: Receiver Disable
0: No effect.
1: Disables the receiver.
• TXEN: Transmitter Enable
0: No effect.
1: Enables the transmitter if TXDIS is 0.
• TXDIS: Transmitter Disable
0: No effect.
1: Disables the transmitter.
559
6438D–ATARM–13-Oct-09
AT91SAM9G45
• RSTSTA: Reset Status Bits
0: No effect.
1: Resets the status bits PARE, FRAME, OVRE, MANERR and RXBRK in US_CSR.
• STTBRK: Start Break
0: No effect.
1: Starts transmission of a break after the characters present in US_THR and the Transmit Shift Register have been transmitted. No effect if a break is already being transmitted.
• STPBRK: Stop Break
0: No effect.
1: Stops transmission of the break after a minimum of one character length and transmits a high level during 12-bit periods.
No effect if no break is being transmitted.
• STTTO: Start Time-out
0: No effect.
1: Starts waiting for a character before clocking the time-out counter. Resets the status bit TIMEOUT in US_CSR.
• SENDA: Send Address
0: No effect.
1: In Multidrop Mode only, the next character written to the US_THR is sent with the address bit set.
• RSTIT: Reset Iterations
0: No effect.
1: Resets ITERATION in US_CSR. No effect if the ISO7816 is not enabled.
• RSTNACK: Reset Non Acknowledge
0: No effect
1: Resets NACK in US_CSR.
• RETTO: Rearm Time-out
0: No effect
1: Restart Time-out
• RTSEN: Request to Send Enable
0: No effect.
1: Drives the pin RTS to 0.
• RTSDIS: Request to Send Disable
0: No effect.
1: Drives the pin RTS to 1.
560
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.8.2
Name:
USART Mode Register
US_MR
Addresses:
0xFFF8C004 (0), 0xFFF90004 (1), 0xFFF94004 (2), 0xFFF98004 (3)
Access:
Read-write
31
ONEBIT
30
MODSYNC–
29
MAN
28
FILTER
27
–
26
25
MAX_ITERATION
24
23
22
VAR_SYNC
21
DSNACK
20
INACK
19
OVER
18
CLKO
17
MODE9
16
MSBF
14
13
12
11
10
PAR
9
8
SYNC
4
3
2
1
0
15
CHMODE
7
NBSTOP
6
5
CHRL
USCLKS
USART_MODE
• USART_MODE
USART_MODE
Mode of the USART
0
0
0
0
Normal
0
0
0
1
RS485
0
0
1
0
Hardware Handshaking
0
1
0
0
IS07816 Protocol: T = 0
0
1
1
0
IS07816 Protocol: T = 1
1
0
0
0
IrDA
Others
Reserved
• USCLKS: Clock Selection
USCLKS
Selected Clock
0
0
MCK
0
1
MCK/DIV (DIV = 8)
1
0
Reserved
1
1
SCK
• CHRL: Character Length.
CHRL
Character Length
0
0
5 bits
0
1
6 bits
1
0
7 bits
1
1
8 bits
561
6438D–ATARM–13-Oct-09
AT91SAM9G45
• SYNC: Synchronous Mode Select
0: USART operates in Asynchronous Mode.
1: USART operates in Synchronous Mode.
• PAR: Parity Type
PAR
Parity Type
0
0
0
Even parity
0
0
1
Odd parity
0
1
0
Parity forced to 0 (Space)
0
1
1
Parity forced to 1 (Mark)
1
0
x
No parity
1
1
x
Multidrop mode
• NBSTOP: Number of Stop Bits
NBSTOP
Asynchronous (SYNC = 0)
Synchronous (SYNC = 1)
0
0
1 stop bit
1 stop bit
0
1
1.5 stop bits
Reserved
1
0
2 stop bits
2 stop bits
1
1
Reserved
Reserved
• CHMODE: Channel Mode
CHMODE
Mode Description
0
0
Normal Mode
0
1
Automatic Echo. Receiver input is connected to the TXD pin.
1
0
Local Loopback. Transmitter output is connected to the Receiver Input.
1
1
Remote Loopback. RXD pin is internally connected to the TXD pin.
• MSBF: Bit Order
0: Least Significant Bit is sent/received first.
1: Most Significant Bit is sent/received first.
• MODE9: 9-bit Character Length
0: CHRL defines character length.
1: 9-bit character length.
• CLKO: Clock Output Select
0: The USART does not drive the SCK pin.
1: The USART drives the SCK pin if USCLKS does not select the external clock SCK.
562
6438D–ATARM–13-Oct-09
AT91SAM9G45
• OVER: Oversampling Mode
0: 16x Oversampling.
1: 8x Oversampling.
• INACK: Inhibit Non Acknowledge
0: The NACK is generated.
1: The NACK is not generated.
• DSNACK: Disable Successive NACK
0: NACK is sent on the ISO line as soon as a parity error occurs in the received character (unless INACK is set).
1: Successive parity errors are counted up to the value specified in the MAX_ITERATION field. These parity errors generate a NACK on the ISO line. As soon as this value is reached, no additional NACK is sent on the ISO line. The flag
ITERATION is asserted.
• VAR_SYNC: Variable Synchronization of Command/Data Sync Start Frame Delimiter
0: User defined configuration of command or data sync field depending on SYNC value.
1: The sync field is updated when a character is written into US_THR register.
• MAX_ITERATION
Defines the maximum number of iterations in mode ISO7816, protocol T= 0.
• FILTER: Infrared Receive Line Filter
0: The USART does not filter the receive line.
1: The USART filters the receive line using a three-sample filter (1/16-bit clock) (2 over 3 majority).
• MAN: Manchester Encoder/Decoder Enable
0: Manchester Encoder/Decoder are disabled.
1: Manchester Encoder/Decoder are enabled.
• MODSYNC: Manchester Synchronization Mode
0:The Manchester Start bit is a 0 to 1 transition
1: The Manchester Start bit is a 1 to 0 transition.
• ONEBIT: Start Frame Delimiter Selector
0: Start Frame delimiter is COMMAND or DATA SYNC.
1: Start Frame delimiter is One Bit.
563
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.8.3
Name:
USART Interrupt Enable Register
US_IER
Addresses:
0xFFF8C008 (0), 0xFFF90008 (1), 0xFFF94008 (2), 0xFFF98008 (3)
Access:
Write-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
MANE
23
–
22
–
21
–
20
–
19
CTSIC
18
–
17
–
16
–
15
–
14
–
13
NACK
12
RXBUFF
11
TXBUFE
10
ITER
9
TXEMPTY
8
TIMEOUT
7
PARE
6
FRAME
5
OVRE
4
ENDTX
3
ENDRX
2
RXBRK
1
TXRDY
0
RXRDY
• RXRDY: RXRDY Interrupt Enable
• TXRDY: TXRDY Interrupt Enable
• RXBRK: Receiver Break Interrupt Enable
• ENDRX: End of Receive Transfer Interrupt Enable
• ENDTX: End of Transmit Interrupt Enable
• OVRE: Overrun Error Interrupt Enable
• FRAME: Framing Error Interrupt Enable
• PARE: Parity Error Interrupt Enable
• TIMEOUT: Time-out Interrupt Enable
• TXEMPTY: TXEMPTY Interrupt Enable
• ITER: Iteration Interrupt Enable
• TXBUFE: Buffer Empty Interrupt Enable
• RXBUFF: Buffer Full Interrupt Enable
• NACK: Non Acknowledge Interrupt Enable
• CTSIC: Clear to Send Input Change Interrupt Enable
• MANE: Manchester Error Interrupt Enable
564
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.8.4
Name:
USART Interrupt Disable Register
US_IDR
Addresses:
0xFFF8C00C (0), 0xFFF9000C (1), 0xFFF9400C (2), 0xFFF9800C (3)
Access:
Write-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
MANE
23
–
22
–
21
–
20
–
19
CTSIC
18
–
17
–
16
–
15
–
14
–
13
NACK
12
RXBUFF
11
TXBUFE
10
ITER
9
TXEMPTY
8
TIMEOUT
7
PARE
6
FRAME
5
OVRE
4
ENDTX
3
ENDRX
2
RXBRK
1
TXRDY
0
RXRDY
• RXRDY: RXRDY Interrupt Disable
• TXRDY: TXRDY Interrupt Disable
• RXBRK: Receiver Break Interrupt Disable
• ENDRX: End of Receive Transfer Interrupt Disable
• ENDTX: End of Transmit Interrupt Disable
• OVRE: Overrun Error Interrupt Disable
• FRAME: Framing Error Interrupt Disable
• PARE: Parity Error Interrupt Disable
• TIMEOUT: Time-out Interrupt Disable
• TXEMPTY: TXEMPTY Interrupt Disable
• ITER: Iteration Interrupt Enable
• TXBUFE: Buffer Empty Interrupt Disable
• RXBUFF: Buffer Full Interrupt Disable
• NACK: Non Acknowledge Interrupt Disable
• CTSIC: Clear to Send Input Change Interrupt Disable
• MANE: Manchester Error Interrupt Disable
565
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.8.5
Name:
USART Interrupt Mask Register
US_IMR
Addresses:
0xFFF8C010 (0), 0xFFF90010 (1), 0xFFF94010 (2), 0xFFF98010 (3)
Access:
Read-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
MANE
23
–
22
–
21
–
20
–
19
CTSIC
18
–
17
–
16
–
15
–
14
–
13
NACK
12
RXBUFF
11
TXBUFE
10
ITER
9
TXEMPTY
8
TIMEOUT
7
PARE
6
FRAME
5
OVRE
4
ENDTX
3
ENDRX
2
RXBRK
1
TXRDY
0
RXRDY
• RXRDY: RXRDY Interrupt Mask
• TXRDY: TXRDY Interrupt Mask
• RXBRK: Receiver Break Interrupt Mask
• ENDRX: End of Receive Transfer Interrupt Mask
• ENDTX: End of Transmit Interrupt Mask
• OVRE: Overrun Error Interrupt Mask
• FRAME: Framing Error Interrupt Mask
• PARE: Parity Error Interrupt Mask
• TIMEOUT: Time-out Interrupt Mask
• TXEMPTY: TXEMPTY Interrupt Mask
• ITER: Iteration Interrupt Enable
• TXBUFE: Buffer Empty Interrupt Mask
• RXBUFF: Buffer Full Interrupt Mask
• NACK: Non Acknowledge Interrupt Mask
• CTSIC: Clear to Send Input Change Interrupt Mask
• MANE: Manchester Error Interrupt Mask
566
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.8.6
Name:
USART Channel Status Register
US_CSR
Addresses:
0xFFF8C014 (0), 0xFFF90014 (1), 0xFFF94014 (2), 0xFFF98014 (3)
Access:
Read-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
MANERR
23
CTS
22
–
21
–
20
–
19
CTSIC
18
–
17
–
16
–
15
–
14
–
13
NACK
12
RXBUFF
11
TXBUFE
10
ITER
9
TXEMPTY
8
TIMEOUT
7
PARE
6
FRAME
5
OVRE
4
ENDTX
3
ENDRX
2
RXBRK
1
TXRDY
0
RXRDY
• RXRDY: Receiver Ready
0: No complete character has been received since the last read of US_RHR or the receiver is disabled. If characters were
being received when the receiver was disabled, RXRDY changes to 1 when the receiver is enabled.
1: At least one complete character has been received and US_RHR has not yet been read.
• TXRDY: Transmitter Ready
0: A character is in the US_THR waiting to be transferred to the Transmit Shift Register, or an STTBRK command has been
requested, or the transmitter is disabled. As soon as the transmitter is enabled, TXRDY becomes 1.
1: There is no character in the US_THR.
• RXBRK: Break Received/End of Break
0: No Break received or End of Break detected since the last RSTSTA.
1: Break Received or End of Break detected since the last RSTSTA.
• ENDRX: End of Receiver Transfer
0: The End of Transfer signal from the Receive PDC channel is inactive.
1: The End of Transfer signal from the Receive PDC channel is active.
• ENDTX: End of Transmitter Transfer
0: The End of Transfer signal from the Transmit PDC channel is inactive.
1: The End of Transfer signal from the Transmit PDC channel is active.
• OVRE: Overrun Error
0: No overrun error has occurred since the last RSTSTA.
1: At least one overrun error has occurred since the last RSTSTA.
567
6438D–ATARM–13-Oct-09
AT91SAM9G45
• FRAME: Framing Error
0: No stop bit has been detected low since the last RSTSTA.
1: At least one stop bit has been detected low since the last RSTSTA.
• PARE: Parity Error
0: No parity error has been detected since the last RSTSTA.
1: At least one parity error has been detected since the last RSTSTA.
• TIMEOUT: Receiver Time-out
0: There has not been a time-out since the last Start Time-out command (STTTO in US_CR) or the Time-out Register is 0.
1: There has been a time-out since the last Start Time-out command (STTTO in US_CR).
• TXEMPTY: Transmitter Empty
0: There are characters in either US_THR or the Transmit Shift Register, or the transmitter is disabled.
1: There are no characters in US_THR, nor in the Transmit Shift Register.
• ITER: Max number of Repetitions Reached
0: Maximum number of repetitions has not been reached since the last RSTSTA.
1: Maximum number of repetitions has been reached since the last RSTSTA.
• TXBUFE: Transmission Buffer Empty
0: The signal Buffer Empty from the Transmit PDC channel is inactive.
1: The signal Buffer Empty from the Transmit PDC channel is active.
• RXBUFF: Reception Buffer Full
0: The signal Buffer Full from the Receive PDC channel is inactive.
1: The signal Buffer Full from the Receive PDC channel is active.
• NACKNon Acknowledge
0: No Non Acknowledge has not been detected since the last RSTNACK.
1: At least one Non Acknowledge has been detected since the last RSTNACK.
• CTSIC: Clear to Send Input Change Flag
0: No input change has been detected on the CTS pin since the last read of US_CSR.
1: At least one input change has been detected on the CTS pin since the last read of US_CSR.
• CTS: Image of CTS Input
0: CTS is at 0.
1: CTS is at 1.
• MANERR: Manchester Error
0: No Manchester error has been detected since the last RSTSTA.
1: At least one Manchester error has been detected since the last RSTSTA.
568
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.8.7
Name:
USART Receive Holding Register
US_RHR
Addresses:
0xFFF8C018 (0), 0xFFF90018 (1), 0xFFF94018 (2), 0xFFF98018 (3)
Access:
Read-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
RXSYNH
14
–
13
–
12
–
11
–
10
–
9
–
8
RXCHR
7
6
5
4
3
2
1
0
RXCHR
• RXCHR: Received Character
Last character received if RXRDY is set.
• RXSYNH: Received Sync
0: Last Character received is a Data.
1: Last Character received is a Command.
569
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.8.8
Name:
USART Transmit Holding Register
US_THR
Addresses:
0xFFF8C01C (0), 0xFFF9001C (1), 0xFFF9401C (2), 0xFFF9801C (3)
Access:
Write-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
TXSYNH
14
–
13
–
12
–
11
–
10
–
9
–
8
TXCHR
7
6
5
4
3
2
1
0
TXCHR
• TXCHR: Character to be Transmitted
Next character to be transmitted after the current character if TXRDY is not set.
• TXSYNH: Sync Field to be transmitted
0: The next character sent is encoded as a data. Start Frame Delimiter is DATA SYNC.
1: The next character sent is encoded as a command. Start Frame Delimiter is COMMAND SYNC.
570
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.8.9
Name:
USART Baud Rate Generator Register
US_BRGR
Addresses:
0xFFF8C020 (0), 0xFFF90020 (1), 0xFFF94020 (2), 0xFFF98020 (3)
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
17
FP
16
15
14
13
12
11
10
9
8
3
2
1
0
CD
7
6
5
4
CD
• CD: Clock Divider
USART_MODE ≠ ISO7816
SYNC = 0
CD
OVER = 0
0
1 to 65535
SYNC = 1
OVER = 1
USART_MODE =
ISO7816
Baud Rate Clock Disabled
Baud Rate =
Selected Clock/16/CD
Baud Rate =
Selected Clock/8/CD
Baud Rate =
Selected Clock /CD
Baud Rate = Selected
Clock/CD/FI_DI_RATIO
• FP: Fractional Part
0: Fractional divider is disabled.
1 - 7: Baudrate resolution, defined by FP x 1/8.
571
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.8.10
Name:
USART Receiver Time-out Register
US_RTOR
Addresses:
0xFFF8C024 (0), 0xFFF90024 (1), 0xFFF94024 (2), 0xFFF98024 (3)
Access:
Read-write
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
14
13
12
11
10
9
8
3
2
1
0
TO
7
6
5
4
TO
• TO: Time-out Value
0: The Receiver Time-out is disabled.
1 - 65535: The Receiver Time-out is enabled and the Time-out delay is TO x Bit Period.
572
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.8.11
Name:
USART Transmitter Timeguard Register
US_TTGR
Addresses:
0xFFF8C028 (0), 0xFFF90028 (1), 0xFFF94028 (2), 0xFFF98028 (3)
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
6
5
4
3
2
1
0
TG
• TG: Timeguard Value
0: The Transmitter Timeguard is disabled.
1 - 255: The Transmitter timeguard is enabled and the timeguard delay is TG x Bit Period.
573
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.8.12
Name:
USART FI DI RATIO Register
US_FIDI
Addresses:
0xFFF8C040 (0), 0xFFF90040 (1), 0xFFF94040 (2), 0xFFF98040 (3)
Access:
Read-write
Reset Value:
0x174
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
9
FI_DI_RATIO
8
7
6
5
4
3
2
1
0
FI_DI_RATIO
• FI_DI_RATIO: FI Over DI Ratio Value
0: If ISO7816 mode is selected, the Baud Rate Generator generates no signal.
1 - 2047: If ISO7816 mode is selected, the Baud Rate is the clock provided on SCK divided by FI_DI_RATIO.
574
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.8.13
Name:
USART Number of Errors Register
US_NER
Addresses:
0xFFF8C044 (0), 0xFFF90044 (1), 0xFFF94044 (2), 0xFFF98044 (3)
Access:
Read-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
6
5
4
3
2
1
0
NB_ERRORS
• NB_ERRORS: Number of Errors
Total number of errors that occurred during an ISO7816 transfer. This register automatically clears when read.
575
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.8.14
Name:
USART IrDA FILTER Register
US_IF
Addresses:
0xFFF8C04C (0), 0xFFF9004C (1), 0xFFF9404C (2), 0xFFF9804C (3)
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
6
5
4
3
2
1
0
IRDA_FILTER
• IRDA_FILTER: IrDA Filter
Sets the filter of the IrDA demodulator.
576
6438D–ATARM–13-Oct-09
AT91SAM9G45
32.8.15
Name:
USART Manchester Configuration Register
US_MAN
Addresses:
0xFFF8C050 (0), 0xFFF90050 (1), 0xFFF94050 (2), 0xFFF98050 (3)
Access:
Read-write
31
–
30
DRIFT
29
1
28
RX_MPOL
27
–
26
–
25
23
–
22
–
21
–
20
–
19
18
15
–
14
–
13
–
12
TX_MPOL
11
–
10
–
9
7
–
6
–
5
–
4
–
3
2
1
24
RX_PP
17
16
RX_PL
8
TX_PP
0
TX_PL
• TX_PL: Transmitter Preamble Length
0: The Transmitter Preamble pattern generation is disabled
1 - 15: The Preamble Length is TX_PL x Bit Period
• TX_PP: Transmitter Preamble Pattern
TX_PP
Preamble Pattern default polarity assumed (TX_MPOL field not set)
0
0
ALL_ONE
0
1
ALL_ZERO
1
0
ZERO_ONE
1
1
ONE_ZERO
• TX_MPOL: Transmitter Manchester Polarity
0: Logic Zero is coded as a zero-to-one transition, Logic One is coded as a one-to-zero transition.
1: Logic Zero is coded as a one-to-zero transition, Logic One is coded as a zero-to-one transition.
• RX_PL: Receiver Preamble Length
0: The receiver preamble pattern detection is disabled
1 - 15: The detected preamble length is RX_PL x Bit Period
• RX_PP: Receiver Preamble Pattern detected
RX_PP
Preamble Pattern default polarity assumed (RX_MPOL field not set)
0
0
ALL_ONE
0
1
ALL_ZERO
1
0
ZERO_ONE
1
1
ONE_ZERO
577
6438D–ATARM–13-Oct-09
AT91SAM9G45
• RX_MPOL: Receiver Manchester Polarity
0: Logic Zero is coded as a zero-to-one transition, Logic One is coded as a one-to-zero transition.
1: Logic Zero is coded as a one-to-zero transition, Logic One is coded as a zero-to-one transition.
• DRIFT: Drift compensation
0: The USART can not recover from an important clock drift
1: The USART can recover from clock drift. The 16X clock mode must be enabled.
578
6438D–ATARM–13-Oct-09
AT91SAM9G45
33. Timer Counter (TC)
33.1
Description
The Timer Counter (TC) includes three identical 16-bit Timer Counter channels.
Each channel can be independently programmed to perform a wide range of functions including
frequency measurement, event counting, interval measurement, pulse generation, delay timing
and pulse width modulation.
Each channel has three external clock inputs, five internal clock inputs and two multi-purpose
input/output signals which can be configured by the user. Each channel drives an internal interrupt signal which can be programmed to generate processor interrupts.
The Timer Counter block has two global registers which act upon all three TC channels.
The Block Control Register allows the three channels to be started simultaneously with the same
instruction.
The Block Mode Register defines the external clock inputs for each channel, allowing them to be
chained.
Table 33-1 gives the assignment of the device Timer Counter clock inputs common to Timer
Counter 0 to 2.
Table 33-1.
Timer Counter Clock Assignment
Name
Definition
TIMER_CLOCK1
MCK/2
TIMER_CLOCK2
MCK/8
TIMER_CLOCK3
MCK/32
TIMER_CLOCK4
MCK/128
TIMER_CLOCK5
Note:
33.2
(1)
SLCK
1. When Slow Clock is selected for Master Clock (CSS = 0 in PMC Master CLock Register),
TIMER_CLOCK5 input is Master Clock, i.e., Slow CLock modified by PRES and MDIV fields.
Embedded Characteristics
• Three 16-bit Timer Counter Channels
• Wide range of functions including:
– Frequency Measurement
– Event Counting
– Interval Measurement
– Pulse Generation
– Delay Timing
– Pulse Width Modulation
– Up/down Capabilities
• Each channel is user-configurable and contains:
– Three external clock inputs
– Five internal clock inputs
579
6438D–ATARM–13-Oct-09
– Two multi-purpose input/output signals
• Two global registers that act on all three TC Channels
33.3
Block Diagram
Figure 33-1. Timer Counter Block Diagram
Parallel I/O
Controller
TIMER_CLOCK1
TCLK0
TIMER_CLOCK2
TIOA1
TIOA2
TIMER_CLOCK3
XC0
TCLK1
TIMER_CLOCK4
XC1
Timer/Counter
Channel 0
TIOA
TIOA0
TIOB0
TIOA0
TIOB
TCLK2
XC2
TIMER_CLOCK5
TC0XC0S
TIOB0
SYNC
TCLK0
TCLK1
TCLK2
INT0
TCLK0
TCLK1
XC0
TIOA0
XC1
TIOA2
XC2
Timer/Counter
Channel 1
TIOA
TIOA1
TIOB1
TIOA1
TIOB
TCLK2
TC1XC1S
TCLK0
XC0
TCLK1
XC1
TCLK2
XC2
TIOB1
SYNC
Timer/Counter
Channel 2
INT1
TIOA
TIOA2
TIOB2
TIOA2
TIOB
TIOA0
TIOA1
TC2XC2S
TIOB2
SYNC
INT2
Timer Counter
Interrupt
Controller
Table 33-2.
Signal Name Description
Block/Channel
Signal Name
XC0, XC1, XC2
Channel Signal
External Clock Inputs
TIOA
Capture Mode: Timer Counter Input
Waveform Mode: Timer Counter Output
TIOB
Capture Mode: Timer Counter Input
Waveform Mode: Timer Counter Input/Output
INT
SYNC
580
Description
Interrupt Signal Output
Synchronization Input Signal
AT91SAM9G45
6438D–ATARM–13-Oct-09
33.4
Pin Name List
Table 33-3.
33.5
33.5.1
TC pin list
Pin Name
Description
Type
TCLK0-TCLK2
External Clock Input
Input
TIOA0-TIOA2
I/O Line A
I/O
TIOB0-TIOB2
I/O Line B
I/O
Product Dependencies
I/O Lines
The pins used for interfacing the compliant external devices may be multiplexed with PIO lines.
The programmer must first program the PIO controllers to assign the TC pins to their peripheral
functions.
Table 33-4.
33.5.2
581
I/O Lines
Instance
Signal
I/O Line
Peripheral
TC0
TCLK0
PD23
A
TC0
TCLK1
PD29
A
TC0
TCLK2
PC10
B
TC0
TIOA0
PD20
A
TC0
TIOA1
PD21
A
TC0
TIOA2
PD22
A
TC0
TIOB0
PD30
A
TC0
TIOB1
PD31
A
TC0
TIOB2
PA26
B
TC1
TCLK3
PA0
B
TC1
TCLK4
PA3
B
TC1
TCLK5
PD9
B
TC1
TIOA3
PA1
B
TC1
TIOA4
PA4
B
TC1
TIOA5
PD7
B
TC1
TIOB3
PA2
B
TC1
TIOB4
PA5
B
TC1
TIOB5
PD8
B
Power Management
The TC is clocked through the Power Management Controller (PMC), thus the programmer must
first configure the PMC to enable the Timer Counter clock.
AT91SAM9G45
6438D–ATARM–13-Oct-09
33.5.3
Interrupt
The TC has an interrupt line connected to the Interrupt Controller (IC). Handling the TC interrupt
requires programming the IC before configuring the TC.
33.6
Functional Description
33.6.1
TC Description
The three channels of the Timer Counter are independent and identical in operation . The registers for channel programming are listed in Table 33-5 on page 595.
33.6.2
16-bit Counter
Each channel is organized around a 16-bit counter. The value of the counter is incremented at
each positive edge of the selected clock. When the counter has reached the value 0xFFFF and
passes to 0x0000, an overflow occurs and the COVFS bit in TC_SR (Status Register) is set.
The current value of the counter is accessible in real time by reading the Counter Value Register, TC_CV. The counter can be reset by a trigger. In this case, the counter value passes to
0x0000 on the next valid edge of the selected clock.
33.6.3
Clock Selection
At block level, input clock signals of each channel can either be connected to the external inputs
TCLK0, TCLK1 or TCLK2, or be connected to the internal I/O signals TIOA0, TIOA1 or TIOA2
for chaining by programming the TC_BMR (Block Mode). See Figure 33-2 ”Clock Chaining
Selection”.
Each channel can independently select an internal or external clock source for its counter:
•
Internal clock signals: TIMER_CLOCK1, TIMER_CLOCK2, TIMER_CLOCK3,
TIMER_CLOCK4, TIMER_CLOCK5
•
External clock signals: XC0, XC1 or XC2
This selection is made by the TCCLKS bits in the TC Channel Mode Register.
The selected clock can be inverted with the CLKI bit in TC_CMR. This allows counting on the
opposite edges of the clock.
The burst function allows the clock to be validated when an external signal is high. The BURST
parameter in the Mode Register defines this signal (none, XC0, XC1, XC2). See Figure 33-3
”Clock Selection”
Note:
582
In all cases, if an external clock is used, the duration of each of its levels must be longer than the
master clock period. The external clock frequency must be at least 2.5 times lower than the master clock
AT91SAM9G45
6438D–ATARM–13-Oct-09
Figure 33-2. Clock Chaining Selection
TC0XC0S
Timer/Counter
Channel 0
TCLK0
TIOA1
XC0
TIOA2
TIOA0
XC1 = TCLK1
XC2 = TCLK2
TIOB0
SYNC
TC1XC1S
Timer/Counter
Channel 1
TCLK1
XC0 = TCLK2
TIOA0
TIOA1
XC1
TIOA2
XC2 = TCLK2
TIOB1
SYNC
Timer/Counter
Channel 2
TC2XC2S
XC0 = TCLK0
TCLK2
TIOA2
XC1 = TCLK1
TIOA0
XC2
TIOB2
TIOA1
SYNC
Figure 33-3. Clock Selection
TCCLKS
TIMER_CLOCK1
TIMER_CLOCK2
CLKI
TIMER_CLOCK3
TIMER_CLOCK4
TIMER_CLOCK5
Selected
Clock
XC0
XC1
XC2
BURST
1
583
AT91SAM9G45
6438D–ATARM–13-Oct-09
33.6.4
Clock Control
The clock of each counter can be controlled in two different ways: it can be enabled/disabled
and started/stopped. See Figure 33-4.
•
The clock can be enabled or disabled by the user with the CLKEN and the CLKDIS
commands in the Control Register. In Capture Mode it can be disabled by an RB load event
if LDBDIS is set to 1 in TC_CMR. In Waveform Mode, it can be disabled by an RC Compare
event if CPCDIS is set to 1 in TC_CMR. When disabled, the start or the stop actions have no
effect: only a CLKEN command in the Control Register can re-enable the clock. When the
clock is enabled, the CLKSTA bit is set in the Status Register.
•
The clock can also be started or stopped: a trigger (software, synchro, external or compare)
always starts the clock. The clock can be stopped by an RB load event in Capture Mode
(LDBSTOP = 1 in TC_CMR) or a RC compare event in Waveform Mode (CPCSTOP = 1 in
TC_CMR). The start and the stop commands have effect only if the clock is enabled.
Figure 33-4. Clock Control
Selected
Clock
Trigger
CLKSTA
Q
Q
S
CLKEN
CLKDIS
S
R
R
Counter
Clock
33.6.5
Stop
Event
Disable
Event
TC Operating Modes
Each channel can independently operate in two different modes:
•
Capture Mode provides measurement on signals.
•
Waveform Mode provides wave generation.
The TC Operating Mode is programmed with the WAVE bit in the TC Channel Mode Register.
In Capture Mode, TIOA and TIOB are configured as inputs.
In Waveform Mode, TIOA is always configured to be an output and TIOB is an output if it is not
selected to be the external trigger.
33.6.6
Trigger
A trigger resets the counter and starts the counter clock. Three types of triggers are common to
both modes, and a fourth external trigger is available to each mode.
584
AT91SAM9G45
6438D–ATARM–13-Oct-09
Regardless of the trigger used, it will be taken into account at the following active edge of the
selected clock. This means that the counter value can be read differently from zero just after a
trigger, especially when a low frequency signal is selected as the clock.
The following triggers are common to both modes:
•
Software Trigger: Each channel has a software trigger, available by setting SWTRG in
TC_CCR.
•
SYNC: Each channel has a synchronization signal SYNC. When asserted, this signal has
the same effect as a software trigger. The SYNC signals of all channels are asserted
simultaneously by writing TC_BCR (Block Control) with SYNC set.
•
Compare RC Trigger: RC is implemented in each channel and can provide a trigger when
the counter value matches the RC value if CPCTRG is set in TC_CMR.
The channel can also be configured to have an external trigger. In Capture Mode, the external
trigger signal can be selected between TIOA and TIOB. In Waveform Mode, an external event
can be programmed on one of the following signals: TIOB, XC0, XC1 or XC2. This external
event can then be programmed to perform a trigger by setting ENETRG in TC_CMR.
If an external trigger is used, the duration of the pulses must be longer than the master clock
period in order to be detected.
33.6.7
Capture Operating Mode
This mode is entered by clearing the WAVE parameter in TC_CMR (Channel Mode Register).
Capture Mode allows the TC channel to perform measurements such as pulse timing, frequency, period, duty cycle and phase on TIOA and TIOB signals which are considered as
inputs.
Figure 33-5 shows the configuration of the TC channel when programmed in Capture Mode.
33.6.8
Capture Registers A and B
Registers A and B (RA and RB) are used as capture registers. This means that they can be
loaded with the counter value when a programmable event occurs on the signal TIOA.
The LDRA parameter in TC_CMR defines the TIOA edge for the loading of register A, and the
LDRB parameter defines the TIOA edge for the loading of Register B.
RA is loaded only if it has not been loaded since the last trigger or if RB has been loaded since
the last loading of RA.
RB is loaded only if RA has been loaded since the last trigger or the last loading of RB.
Loading RA or RB before the read of the last value loaded sets the Overrun Error Flag (LOVRS)
in TC_SR (Status Register). In this case, the old value is overwritten.
33.6.9
Trigger Conditions
In addition to the SYNC signal, the software trigger and the RC compare trigger, an external trigger can be defined.
The ABETRG bit in TC_CMR selects TIOA or TIOB input signal as an external trigger. The
ETRGEDG parameter defines the edge (rising, falling or both) detected to generate an external
trigger. If ETRGEDG = 0 (none), the external trigger is disabled.
585
AT91SAM9G45
6438D–ATARM–13-Oct-09
586
MTIOA
MTIOB
1
If RA is not loaded
or RB is Loaded
Edge
Detector
ETRGEDG
SWTRG
Timer/Counter Channel
ABETRG
BURST
CLKI
S
R
OVF
LDRB
Edge
Detector
Edge
Detector
Capture
Register A
LDBSTOP
R
S
CLKEN
LDRA
If RA is Loaded
CPCTRG
16-bit Counter
RESET
Trig
CLK
Q
Q
CLKSTA
LDBDIS
Capture
Register B
CLKDIS
TC1_SR
TIOA
TIOB
SYNC
XC2
XC1
XC0
TIMER_CLOCK5
TIMER_CLOCK4
TIMER_CLOCK3
TIMER_CLOCK2
TIMER_CLOCK1
TCCLKS
Compare RC =
Register C
COVFS
INT
Figure 33-5. Capture Mode
CPCS
LOVRS
LDRBS
ETRGS
LDRAS
TC1_IMR
AT91SAM9G45
6438D–ATARM–13-Oct-09
33.6.10
Waveform Operating Mode
Waveform operating mode is entered by setting the WAVE parameter in TC_CMR (Channel
Mode Register).
In Waveform Operating Mode the TC channel generates 1 or 2 PWM signals with the same frequency and independently programmable duty cycles, or generates different types of one-shot
or repetitive pulses.
In this mode, TIOA is configured as an output and TIOB is defined as an output if it is not used
as an external event (EEVT parameter in TC_CMR).
Figure 33-6 shows the configuration of the TC channel when programmed in Waveform Operating Mode.
33.6.11
Waveform Selection
Depending on the WAVSEL parameter in TC_CMR (Channel Mode Register), the behavior of
TC_CV varies.
With any selection, RA, RB and RC can all be used as compare registers.
RA Compare is used to control the TIOA output, RB Compare is used to control the TIOB output
(if correctly configured) and RC Compare is used to control TIOA and/or TIOB outputs.
587
AT91SAM9G45
6438D–ATARM–13-Oct-09
588
TIOB
SYNC
XC2
XC1
XC0
TIMER_CLOCK5
TIMER_CLOCK4
TIMER_CLOCK3
TIMER_CLOCK2
TIMER_CLOCK1
1
EEVT
BURST
Timer/Counter Channel
Edge
Detector
EEVTEDG
SWTRG
ENETRG
CLKI
Trig
CLK
R
S
OVF
WAVSEL
RESET
16-bit Counter
WAVSEL
Q
Compare RA =
Register A
Q
CLKSTA
Compare RC =
Compare RB =
CPCSTOP
CPCDIS
Register C
CLKDIS
Register B
R
S
CLKEN
CPAS
INT
BSWTRG
BEEVT
BCPB
BCPC
ASWTRG
AEEVT
ACPA
ACPC
Output Controller
Output Controller
TCCLKS
TIOB
MTIOB
TIOA
MTIOA
Figure 33-6. Waveform Mode
CPCS
CPBS
COVFS
TC1_SR
ETRGS
TC1_IMR
AT91SAM9G45
6438D–ATARM–13-Oct-09
33.6.11.1
WAVSEL = 00
When WAVSEL = 00, the value of TC_CV is incremented from 0 to 0xFFFF. Once 0xFFFF has
been reached, the value of TC_CV is reset. Incrementation of TC_CV starts again and the cycle
continues. See Figure 33-7.
An external event trigger or a software trigger can reset the value of TC_CV. It is important to
note that the trigger may occur at any time. See Figure 33-8.
RC Compare cannot be programmed to generate a trigger in this configuration. At the same
time, RC Compare can stop the counter clock (CPCSTOP = 1 in TC_CMR) and/or disable the
counter clock (CPCDIS = 1 in TC_CMR).
Figure 33-7. WAVSEL= 00 without trigger
Counter Value
Counter cleared by compare match with 0xFFFF
0xFFFF
RC
RB
RA
Waveform Examples
Time
TIOB
TIOA
589
AT91SAM9G45
6438D–ATARM–13-Oct-09
Figure 33-8. WAVSEL= 00 with trigger
Counter cleared by compare match with 0xFFFF
Counter Value
0xFFFF
Counter cleared by trigger
RC
RB
RA
Time
Waveform Examples
TIOB
TIOA
33.6.11.2
WAVSEL = 10
When WAVSEL = 10, the value of TC_CV is incremented from 0 to the value of RC, then automatically reset on a RC Compare. Once the value of TC_CV has been reset, it is then
incremented and so on. See Figure 33-9.
It is important to note that TC_CV can be reset at any time by an external event or a software
trigger if both are programmed correctly. See Figure 33-10.
In addition, RC Compare can stop the counter clock (CPCSTOP = 1 in TC_CMR) and/or disable
the counter clock (CPCDIS = 1 in TC_CMR).
Figure 33-9. WAVSEL = 10 Without Trigger
Counter Value
0xFFFF
Counter cleared by compare match with RC
RC
RB
RA
Waveform Examples
Time
TIOB
TIOA
590
AT91SAM9G45
6438D–ATARM–13-Oct-09
Figure 33-10. WAVSEL = 10 With Trigger
Counter Value
0xFFFF
Counter cleared by compare match with RC
Counter cleared by trigger
RC
RB
RA
Waveform Examples
Time
TIOB
TIOA
33.6.11.3
WAVSEL = 01
When WAVSEL = 01, the value of TC_CV is incremented from 0 to 0xFFFF. Once 0xFFFF is
reached, the value of TC_CV is decremented to 0, then re-incremented to 0xFFFF and so on.
See Figure 33-11.
A trigger such as an external event or a software trigger can modify TC_CV at any time. If a trigger occurs while TC_CV is incrementing, TC_CV then decrements. If a trigger is received while
TC_CV is decrementing, TC_CV then increments. See Figure 33-12.
RC Compare cannot be programmed to generate a trigger in this configuration.
At the same time, RC Compare can stop the counter clock (CPCSTOP = 1) and/or disable the
counter clock (CPCDIS = 1).
591
AT91SAM9G45
6438D–ATARM–13-Oct-09
Figure 33-11. WAVSEL = 01 Without Trigger
Counter decremented by compare match with 0xFFFF
Counter Value
0xFFFF
RC
RB
RA
Time
Waveform Examples
TIOB
TIOA
Figure 33-12. WAVSEL = 01 With Trigger
Counter decremented by compare match with 0xFFFF
Counter Value
0xFFFF
Counter decremented
by trigger
RC
RB
Counter incremented
by trigger
RA
Time
Waveform Examples
TIOB
TIOA
33.6.11.4
WAVSEL = 11
When WAVSEL = 11, the value of TC_CV is incremented from 0 to RC. Once RC is reached, the
value of TC_CV is decremented to 0, then re-incremented to RC and so on. See Figure 33-13.
A trigger such as an external event or a software trigger can modify TC_CV at any time. If a trigger occurs while TC_CV is incrementing, TC_CV then decrements. If a trigger is received while
TC_CV is decrementing, TC_CV then increments. See Figure 33-14.
RC Compare can stop the counter clock (CPCSTOP = 1) and/or disable the counter clock
(CPCDIS = 1).
592
AT91SAM9G45
6438D–ATARM–13-Oct-09
Figure 33-13. WAVSEL = 11 Without Trigger
Counter Value
0xFFFF
Counter decremented by compare match with RC
RC
RB
RA
Time
Waveform Examples
TIOB
TIOA
Figure 33-14. WAVSEL = 11 With Trigger
Counter Value
0xFFFF
Counter decremented by compare match with RC
RC
RB
Counter decremented
by trigger
Counter incremented
by trigger
RA
Waveform Examples
Time
TIOB
TIOA
593
AT91SAM9G45
6438D–ATARM–13-Oct-09
33.6.12
External Event/Trigger Conditions
An external event can be programmed to be detected on one of the clock sources (XC0, XC1,
XC2) or TIOB. The external event selected can then be used as a trigger.
The EEVT parameter in TC_CMR selects the external trigger. The EEVTEDG parameter defines
the trigger edge for each of the possible external triggers (rising, falling or both). If EEVTEDG is
cleared (none), no external event is defined.
If TIOB is defined as an external event signal (EEVT = 0), TIOB is no longer used as an output
and the compare register B is not used to generate waveforms and subsequently no IRQs. In
this case the TC channel can only generate a waveform on TIOA.
When an external event is defined, it can be used as a trigger by setting bit ENETRG in
TC_CMR.
As in Capture Mode, the SYNC signal and the software trigger are also available as triggers. RC
Compare can also be used as a trigger depending on the parameter WAVSEL.
33.6.13
Output Controller
The output controller defines the output level changes on TIOA and TIOB following an event.
TIOB control is used only if TIOB is defined as output (not as an external event).
The following events control TIOA and TIOB: software trigger, external event and RC compare.
RA compare controls TIOA and RB compare controls TIOB. Each of these events can be programmed to set, clear or toggle the output as defined in the corresponding parameter in
TC_CMR.
594
AT91SAM9G45
6438D–ATARM–13-Oct-09
33.7
Timer Counter (TC) User Interface
Table 33-5.
Register Mapping
Offset(1)
Register
Name
Access
Reset
0x00 + channel * 0x40 + 0x00
Channel Control Register
TC_CCR
Write-only
–
0x00 + channel * 0x40 + 0x04
Channel Mode Register
TC_CMR
Read-write
0
0x00 + channel * 0x40 + 0x08
Reserved
0x00 + channel * 0x40 + 0x0C
Reserved
0x00 + channel * 0x40 + 0x10
Counter Value
TC_CV
Read-only
0
0x00 + channel * 0x40 + 0x14
Register A
TC_RA
Read-write(2)
0
0x00 + channel * 0x40 + 0x18
Register B
TC_RB
Read-write(2)
0
0x00 + channel * 0x40 + 0x1C
Register C
TC_RC
Read-write
0
0x00 + channel * 0x40 + 0x20
Status Register
TC_SR
Read-only
0
0x00 + channel * 0x40 + 0x24
Interrupt Enable Register
TC_IER
Write-only
–
0x00 + channel * 0x40 + 0x28
Interrupt Disable Register
TC_IDR
Write-only
–
0x00 + channel * 0x40 + 0x2C
Interrupt Mask Register
TC_IMR
Read-only
0
0xC0
Block Control Register
TC_BCR
Write-only
–
0xC4
Block Mode Register
TC_BMR
Read-write
0
0xD8
Reserved
0xE4
Reserved
0xFC
Reserved
–
–
–
Notes:
1. Channel index ranges from 0 to 2.
2. Read-only if WAVE = 0
595
AT91SAM9G45
6438D–ATARM–13-Oct-09
33.7.1
Name:
TC Block Control Register
TC_BCR
Addresses:
0xFFF7C0C0 (0), 0xFFFD40C0 (1)
Access:
Write-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
–
–
–
–
–
–
–
SYNC
• SYNC: Synchro Command
0 = no effect.
1 = asserts the SYNC signal which generates a software trigger simultaneously for each of the channels.
596
AT91SAM9G45
6438D–ATARM–13-Oct-09
33.7.2
Name:
TC Block Mode Register
TC_BMR
Addresses:
0xFFF7C0C4 (0), 0xFFFD40C4 (1)
Access:
Read-write
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
–
–
TC2XC2S
TC1XC1S
0
TC0XC0S
• TC0XC0S: External Clock Signal 0 Selection
TC0XC0S
Signal Connected to XC0
0
0
TCLK0
0
1
none
1
0
TIOA1
1
1
TIOA2
• TC1XC1S: External Clock Signal 1 Selection
TC1XC1S
Signal Connected to XC1
0
0
TCLK1
0
1
none
1
0
TIOA0
1
1
TIOA2
• TC2XC2S: External Clock Signal 2 Selection
TC2XC2S
597
Signal Connected to XC2
0
0
TCLK2
0
1
none
1
0
TIOA0
1
1
TIOA1
AT91SAM9G45
6438D–ATARM–13-Oct-09
33.7.3
Name:
TC Channel Control Register
TC_CCRx [x=0..2]
Addresses:
0xFFF7C000 (0)[0], 0xFFF7C040 (0)[1], 0xFFF7C080 (0)[2], 0xFFFD4000 (1)[0], 0xFFFD4040
(1)[1], 0xFFFD4080 (1)[2]
Access:
Write-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
–
–
–
–
–
SWTRG
CLKDIS
CLKEN
• CLKEN: Counter Clock Enable Command
0 = no effect.
1 = enables the clock if CLKDIS is not 1.
• CLKDIS: Counter Clock Disable Command
0 = no effect.
1 = disables the clock.
• SWTRG: Software Trigger Command
0 = no effect.
1 = a software trigger is performed: the counter is reset and the clock is started.
598
AT91SAM9G45
6438D–ATARM–13-Oct-09
33.7.4
Name:
TC Channel Mode Register: Capture Mode
TC_CMRx [x=0..2] (WAVE = 0)
Addresses:
0xFFF7C004 (0)[0], 0xFFF7C044 (0)[1], 0xFFF7C084 (0)[2], 0xFFFD4004 (1)[0],
0xFFFD4044 (1)[1], 0xFFFD4084 (1)[2]
Access:
Read-write
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
–
–
–
–
15
14
13
12
11
10
WAVE
CPCTRG
–
–
–
ABETRG
7
6
5
3
2
LDBDIS
LDBSTOP
16
LDRB
4
BURST
CLKI
LDRA
9
8
ETRGEDG
1
0
TCCLKS
• TCCLKS: Clock Selection
TCCLKS
Clock Selected
0
0
0
TIMER_CLOCK1
0
0
1
TIMER_CLOCK2
0
1
0
TIMER_CLOCK3
0
1
1
TIMER_CLOCK4
1
0
0
TIMER_CLOCK5
1
0
1
XC0
1
1
0
XC1
1
1
1
XC2
• CLKI: Clock Invert
0 = counter is incremented on rising edge of the clock.
1 = counter is incremented on falling edge of the clock.
• BURST: Burst Signal Selection
BURST
0
0
The clock is not gated by an external signal.
0
1
XC0 is ANDed with the selected clock.
1
0
XC1 is ANDed with the selected clock.
1
1
XC2 is ANDed with the selected clock.
• LDBSTOP: Counter Clock Stopped with RB Loading
0 = counter clock is not stopped when RB loading occurs.
1 = counter clock is stopped when RB loading occurs.
599
AT91SAM9G45
6438D–ATARM–13-Oct-09
• LDBDIS: Counter Clock Disable with RB Loading
0 = counter clock is not disabled when RB loading occurs.
1 = counter clock is disabled when RB loading occurs.
• ETRGEDG: External Trigger Edge Selection
ETRGEDG
Edge
0
0
none
0
1
rising edge
1
0
falling edge
1
1
each edge
• ABETRG: TIOA or TIOB External Trigger Selection
0 = TIOB is used as an external trigger.
1 = TIOA is used as an external trigger.
• CPCTRG: RC Compare Trigger Enable
0 = RC Compare has no effect on the counter and its clock.
1 = RC Compare resets the counter and starts the counter clock.
• WAVE
0 = Capture Mode is enabled.
1 = Capture Mode is disabled (Waveform Mode is enabled).
• LDRA: RA Loading Selection
LDRA
Edge
0
0
none
0
1
rising edge of TIOA
1
0
falling edge of TIOA
1
1
each edge of TIOA
• LDRB: RB Loading Selection
LDRB
600
Edge
0
0
none
0
1
rising edge of TIOA
1
0
falling edge of TIOA
1
1
each edge of TIOA
AT91SAM9G45
6438D–ATARM–13-Oct-09
33.7.5
Name:
TC Channel Mode Register: Waveform Mode
TC_CMRx [x=0..2] (WAVE = 1)
Addresses:
0xFFF7C004 (0)[0], 0xFFF7C044 (0)[1], 0xFFF7C084 (0)[2], 0xFFFD4004 (1)[0],
0xFFFD4044 (1)[1], 0xFFFD4084 (1)[2]
Access:
Read-write
31
30
29
BSWTRG
23
22
21
ASWTRG
15
28
27
BEEVT
20
19
AEEVT
14
WAVE
13
7
6
CPCDIS
CPCSTOP
24
BCPB
18
11
ENETRG
5
25
17
16
ACPC
12
WAVSEL
26
BCPC
ACPA
10
9
EEVT
4
3
BURST
CLKI
8
EEVTEDG
2
1
0
TCCLKS
• TCCLKS: Clock Selection
TCCLKS
Clock Selected
0
0
0
TIMER_CLOCK1
0
0
1
TIMER_CLOCK2
0
1
0
TIMER_CLOCK3
0
1
1
TIMER_CLOCK4
1
0
0
TIMER_CLOCK5
1
0
1
XC0
1
1
0
XC1
1
1
1
XC2
• CLKI: Clock Invert
0 = counter is incremented on rising edge of the clock.
1 = counter is incremented on falling edge of the clock.
• BURST: Burst Signal Selection
BURST
0
0
The clock is not gated by an external signal.
0
1
XC0 is ANDed with the selected clock.
1
0
XC1 is ANDed with the selected clock.
1
1
XC2 is ANDed with the selected clock.
• CPCSTOP: Counter Clock Stopped with RC Compare
0 = counter clock is not stopped when counter reaches RC.
1 = counter clock is stopped when counter reaches RC.
601
AT91SAM9G45
6438D–ATARM–13-Oct-09
• CPCDIS: Counter Clock Disable with RC Compare
0 = counter clock is not disabled when counter reaches RC.
1 = counter clock is disabled when counter reaches RC.
• EEVTEDG: External Event Edge Selection
EEVTEDG
Edge
0
0
none
0
1
rising edge
1
0
falling edge
1
1
each edge
• EEVT: External Event Selection
EEVT
Signal selected as external event
TIOB Direction
0
0
TIOB
input (1)
0
1
XC0
output
1
0
XC1
output
1
1
XC2
output
Note:
1. If TIOB is chosen as the external event signal, it is configured as an input and no longer generates waveforms and subsequently no IRQs.
• ENETRG: External Event Trigger Enable
0 = the external event has no effect on the counter and its clock. In this case, the selected external event only controls the
TIOA output.
1 = the external event resets the counter and starts the counter clock.
• WAVSEL: Waveform Selection
WAVSEL
Effect
0
0
UP mode without automatic trigger on RC Compare
1
0
UP mode with automatic trigger on RC Compare
0
1
UPDOWN mode without automatic trigger on RC Compare
1
1
UPDOWN mode with automatic trigger on RC Compare
• WAVE
0 = Waveform Mode is disabled (Capture Mode is enabled).
1 = Waveform Mode is enabled.
602
AT91SAM9G45
6438D–ATARM–13-Oct-09
• ACPA: RA Compare Effect on TIOA
ACPA
Effect
0
0
none
0
1
set
1
0
clear
1
1
toggle
• ACPC: RC Compare Effect on TIOA
ACPC
Effect
0
0
none
0
1
set
1
0
clear
1
1
toggle
• AEEVT: External Event Effect on TIOA
AEEVT
Effect
0
0
none
0
1
set
1
0
clear
1
1
toggle
• ASWTRG: Software Trigger Effect on TIOA
ASWTRG
Effect
0
0
none
0
1
set
1
0
clear
1
1
toggle
• BCPB: RB Compare Effect on TIOB
BCPB
603
Effect
0
0
none
0
1
set
1
0
clear
1
1
toggle
AT91SAM9G45
6438D–ATARM–13-Oct-09
• BCPC: RC Compare Effect on TIOB
BCPC
Effect
0
0
none
0
1
set
1
0
clear
1
1
toggle
• BEEVT: External Event Effect on TIOB
BEEVT
Effect
0
0
none
0
1
set
1
0
clear
1
1
toggle
• BSWTRG: Software Trigger Effect on TIOB
BSWTRG
604
Effect
0
0
none
0
1
set
1
0
clear
1
1
toggle
AT91SAM9G45
6438D–ATARM–13-Oct-09
33.7.6
Name:
TC Counter Value Register
TC_CVx [x=0..2]
Addresses:
0xFFF7C010 (0)[0], 0xFFF7C050 (0)[1], 0xFFF7C090 (0)[2], 0xFFFD4010 (1)[0]
0xFFFD4050 (1)[1], 0xFFFD4090 (1)[2]
Access:
Read-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
3
2
1
0
CV
7
6
5
4
CV
• CV: Counter Value
CV contains the counter value in real time.
605
AT91SAM9G45
6438D–ATARM–13-Oct-09
33.7.7
Name:
TC Register A
TC_RAx [x=0..2]
Addresses:
0xFFF7C014 (0)[0], 0xFFF7C054 (0)[1], 0xFFF7C094 (0)[2], 0xFFFD4014 (1)[0],
0xFFFD4054 (1)[1], 0xFFFD4094 (1)[2]
Access:
Read-only if WAVE = 0, Read-write if WAVE = 1
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
3
2
1
0
RA
7
6
5
4
RA
• RA: Register A
RA contains the Register A value in real time.
33.7.8
Name:
TC Register B
TC_RBx [x=0..2]
Addresses:
0xFFF7C018 (0)[0], 0xFFF7C058 (0)[1], 0xFFF7C098 (0)[2], 0xFFFD4018 (1)[0],
0xFFFD4058 (1)[1], 0xFFFD4098 (1)[2]
Access:
Read-only if WAVE = 0, Read-write if WAVE = 1
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
3
2
1
0
RB
7
6
5
4
RB
• RB: Register B
RB contains the Register B value in real time.
606
AT91SAM9G45
6438D–ATARM–13-Oct-09
33.7.9
Name:
TC Register C
TC_RCx [x=0..2]
Addresses:
0xFFF7C01C (0)[0], 0xFFF7C05C (0)[1], 0xFFF7C09C (0)[2], 0xFFFD401C (1)[0],
0xFFFD405C (1)[1], 0xFFFD409C (1)[2]
Access:
Read-write
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
3
2
1
0
RC
7
6
5
4
RC
• RC: Register C
RC contains the Register C value in real time.
607
AT91SAM9G45
6438D–ATARM–13-Oct-09
33.7.10
Name:
TC Status Register
TC_SRx [x=0..2]
Addresses:
0xFFF7C020 (0)[0], 0xFFF7C060 (0)[1], 0xFFF7C0A0 (0)[2], 0xFFFD4020 (1)[0],
0xFFFD4060 (1)[1], 0xFFFD40A0 (1)[2]
Access:
Read-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
MTIOB
MTIOA
CLKSTA
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
ETRGS
LDRBS
LDRAS
CPCS
CPBS
CPAS
LOVRS
COVFS
• COVFS: Counter Overflow Status
0 = no counter overflow has occurred since the last read of the Status Register.
1 = a counter overflow has occurred since the last read of the Status Register.
• LOVRS: Load Overrun Status
0 = Load overrun has not occurred since the last read of the Status Register or WAVE = 1.
1 = RA or RB have been loaded at least twice without any read of the corresponding register since the last read of the Status Register, if WAVE = 0.
• CPAS: RA Compare Status
0 = RA Compare has not occurred since the last read of the Status Register or WAVE = 0.
1 = RA Compare has occurred since the last read of the Status Register, if WAVE = 1.
• CPBS: RB Compare Status
0 = RB Compare has not occurred since the last read of the Status Register or WAVE = 0.
1 = RB Compare has occurred since the last read of the Status Register, if WAVE = 1.
• CPCS: RC Compare Status
0 = RC Compare has not occurred since the last read of the Status Register.
1 = RC Compare has occurred since the last read of the Status Register.
• LDRAS: RA Loading Status
0 = RA Load has not occurred since the last read of the Status Register or WAVE = 1.
1 = RA Load has occurred since the last read of the Status Register, if WAVE = 0.
• LDRBS: RB Loading Status
0 = RB Load has not occurred since the last read of the Status Register or WAVE = 1.
1 = RB Load has occurred since the last read of the Status Register, if WAVE = 0.
608
AT91SAM9G45
6438D–ATARM–13-Oct-09
• ETRGS: External Trigger Status
0 = external trigger has not occurred since the last read of the Status Register.
1 = external trigger has occurred since the last read of the Status Register.
• CLKSTA: Clock Enabling Status
0 = clock is disabled.
1 = clock is enabled.
• MTIOA: TIOA Mirror
0 = TIOA is low. If WAVE = 0, this means that TIOA pin is low. If WAVE = 1, this means that TIOA is driven low.
1 = TIOA is high. If WAVE = 0, this means that TIOA pin is high. If WAVE = 1, this means that TIOA is driven high.
• MTIOB: TIOB Mirror
0 = TIOB is low. If WAVE = 0, this means that TIOB pin is low. If WAVE = 1, this means that TIOB is driven low.
1 = TIOB is high. If WAVE = 0, this means that TIOB pin is high. If WAVE = 1, this means that TIOB is driven high.
609
AT91SAM9G45
6438D–ATARM–13-Oct-09
33.7.11
Name:
TC Interrupt Enable Register
TC_IERx [x=0..2]
Addresses:
0xFFF7C024 (0)[0], 0xFFF7C064 (0)[1], 0xFFF7C0A4 (0)[2], 0xFFFD4024 (1)[0],
0xFFFD4064 (1)[1], 0xFFFD40A4 (1)[2]
Access:
Write-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
ETRGS
LDRBS
LDRAS
CPCS
CPBS
CPAS
LOVRS
COVFS
• COVFS: Counter Overflow
0 = no effect.
1 = enables the Counter Overflow Interrupt.
• LOVRS: Load Overrun
0 = no effect.
1 = enables the Load Overrun Interrupt.
• CPAS: RA Compare
0 = no effect.
1 = enables the RA Compare Interrupt.
• CPBS: RB Compare
0 = no effect.
1 = enables the RB Compare Interrupt.
• CPCS: RC Compare
0 = no effect.
1 = enables the RC Compare Interrupt.
• LDRAS: RA Loading
0 = no effect.
1 = enables the RA Load Interrupt.
• LDRBS: RB Loading
0 = no effect.
1 = enables the RB Load Interrupt.
• ETRGS: External Trigger
0 = no effect.
1 = enables the External Trigger Interrupt.
610
AT91SAM9G45
6438D–ATARM–13-Oct-09
33.7.12
Name:
TC Interrupt Disable Register
TC_IDRx [x=0..2]
Addresses:
0xFFF7C028 (0)[0], 0xFFF7C068 (0)[1], 0xFFF7C0A8 (0)[2], 0xFFFD4028 (1)[0],
0xFFFD4068 (1)[1], 0xFFFD40A8 (1)[2]
Access:
Write-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
ETRGS
LDRBS
LDRAS
CPCS
CPBS
CPAS
LOVRS
COVFS
• COVFS: Counter Overflow
0 = no effect.
1 = disables the Counter Overflow Interrupt.
• LOVRS: Load Overrun
0 = no effect.
1 = disables the Load Overrun Interrupt (if WAVE = 0).
• CPAS: RA Compare
0 = no effect.
1 = disables the RA Compare Interrupt (if WAVE = 1).
• CPBS: RB Compare
0 = no effect.
1 = disables the RB Compare Interrupt (if WAVE = 1).
• CPCS: RC Compare
0 = no effect.
1 = disables the RC Compare Interrupt.
• LDRAS: RA Loading
0 = no effect.
1 = disables the RA Load Interrupt (if WAVE = 0).
• LDRBS: RB Loading
0 = no effect.
1 = disables the RB Load Interrupt (if WAVE = 0).
• ETRGS: External Trigger
0 = no effect.
1 = disables the External Trigger Interrupt.
611
AT91SAM9G45
6438D–ATARM–13-Oct-09
33.7.13
Name:
TC Interrupt Mask Register
TC_IMRx [x=0..2]
Addresses:
0xFFF7C02C (0)[0], 0xFFF7C06C (0)[1], 0xFFF7C0AC (0)[2], 0xFFFD402C (1)[0],
0xFFFD406C (1)[1], 0xFFFD40AC (1)[2]
Access:
Read-only
31
30
29
28
27
26
25
24
–
–
–
–
–
–
–
–
23
22
21
20
19
18
17
16
–
–
–
–
–
–
–
–
15
14
13
12
11
10
9
8
–
–
–
–
–
–
–
–
7
6
5
4
3
2
1
0
ETRGS
LDRBS
LDRAS
CPCS
CPBS
CPAS
LOVRS
COVFS
• COVFS: Counter Overflow
0 = the Counter Overflow Interrupt is disabled.
1 = the Counter Overflow Interrupt is enabled.
• LOVRS: Load Overrun
0 = the Load Overrun Interrupt is disabled.
1 = the Load Overrun Interrupt is enabled.
• CPAS: RA Compare
0 = the RA Compare Interrupt is disabled.
1 = the RA Compare Interrupt is enabled.
• CPBS: RB Compare
0 = the RB Compare Interrupt is disabled.
1 = the RB Compare Interrupt is enabled.
• CPCS: RC Compare
0 = the RC Compare Interrupt is disabled.
1 = the RC Compare Interrupt is enabled.
• LDRAS: RA Loading
0 = the Load RA Interrupt is disabled.
1 = the Load RA Interrupt is enabled.
• LDRBS: RB Loading
0 = the Load RB Interrupt is disabled.
1 = the Load RB Interrupt is enabled.
• ETRGS: External Trigger
0 = the External Trigger Interrupt is disabled.
1 = the External Trigger Interrupt is enabled.
612
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
34. Synchronous Serial Controller (SSC)
34.1
Description
The Atmel Synchronous Serial Controller (SSC) provides a synchronous communication link
with external devices. It supports many serial synchronous communication protocols generally
used in audio and telecom applications such as I2S, Short Frame Sync, Long Frame Sync, etc.
The SSC contains an independent receiver and transmitter and a common clock divider. The
receiver and the transmitter each interface with three signals: the TD/RD signal for data, the
TK/RK signal for the clock and the TF/RF signal for the Frame Sync. The transfers can be programmed to start automatically or on different events detected on the Frame Sync signal.
The SSC’s high-level of programmability and its two dedicated PDC channels of up to 32 bits
permit a continuous high bit rate data transfer without processor intervention.
The SSC’s high-level of programmability and its use of DMA permit a continuous high bit rate
data transfer without processor intervention.
Featuring connection to two PDC channels and connection to the DMA, the SSC permits interfacing with low processor overhead to the following:
• CODEC’s in master or slave mode
• DAC through dedicated serial interface, particularly I2S
• Magnetic card reader
34.2
Embedded Characteristics
• Provides serial synchronous communication links used in audio and telecom applications
(with CODECs in Master or Slave Modes, I2S, TDM Buses, Magnetic Card Reader,...)
• Contains an independent receiver and transmitter and a common clock divider
• Offers a configurable frame sync and data length
• Receiver and transmitter can be programmed to start automatically or on detection of
different event on the frame sync signal
• Receiver and transmitter include a data signal, a clock signal and a frame synchronization
signal
613
6438D–ATARM–13-Oct-09
34.3
Block Diagram
Figure 34-1. Block Diagram
System
Bus
APB Bridge
PDC
Peripheral
Bus
TF
TK
PMC
TD
MCK
SSC Interface
PIO
RF
RK
Interrupt Control
RD
SSC Interrupt
614
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 34-2. Block Diagram
System
Bus
APB Bridge
DMA
Peripheral
Bus
TF
TK
PMC
TD
MCK
PIO
SSC Interface
RF
RK
Interrupt Control
RD
SSC Interrupt
34.4
Application Block Diagram
Figure 34-3. Application Block Diagram
OS or RTOS Driver
Power
Management
Interrupt
Management
Test
Management
SSC
Serial AUDIO
Codec
Time Slot
Management
Frame
Management
Line Interface
615
6438D–ATARM–13-Oct-09
34.5
Pin Name List
Table 34-1.
I/O Lines Description
Pin Name
Pin Description
RF
Receiver Frame Synchro
Input/Output
RK
Receiver Clock
Input/Output
RD
Receiver Data
Input
TF
Transmitter Frame Synchro
Input/Output
TK
Transmitter Clock
Input/Output
TD
Transmitter Data
Output
34.6
34.6.1
Type
Product Dependencies
I/O Lines
The pins used for interfacing the compliant external devices may be multiplexed with PIO lines.
Before using the SSC receiver, the PIO controller must be configured to dedicate the SSC
receiver I/O lines to the SSC peripheral mode.
Before using the SSC transmitter, the PIO controller must be configured to dedicate the SSC
transmitter I/O lines to the SSC peripheral mode.
Table 34-2.
I/O Lines
Instance
Signal
I/O Line
Peripheral
SSC0
RD0
PD3
A
SSC0
RF0
PD5
A
SSC0
RK0
PD4
A
SSC0
TD0
PD2
A
SSC0
TF0
PD1
A
SSC0
TK0
PD0
A
SSC1
RD1
PD11
A
SSC1
RF1
PD15
A
SSC1
RK1
PD13
A
SSC1
TD1
PD10
A
SSC1
TF1
PD14
A
SSC1
TK1
PD12
A
34.6.2
Power Management
The SSC is not continuously clocked. The SSC interface may be clocked through the Power
Management Controller (PMC), therefore the programmer must first configure the PMC to
enable the SSC clock.
34.6.3
Interrupt
The SSC interface has an interrupt line connected to the Advanced Interrupt Controller (AIC).
Handling interrupts requires programming the AICbefore configuring the SSC.
616
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
All SSC interrupts can be enabled/disabled configuring the SSC Interrupt mask register. Each
Table 34-3.
Peripheral IDs
Instance
ID
SSC0
16
SSC1
17
pending and unmasked SSC interrupt will assert the SSC interrupt line. The SSC interrupt service routine can get the interrupt origin by reading the SSC interrupt status register.
617
6438D–ATARM–13-Oct-09
34.7
Functional Description
This chapter contains the functional description of the following: SSC Functional Block, Clock
Management, Data format, Start, Transmitter, Receiver and Frame Sync.
The receiver and transmitter operate separately. However, they can work synchronously by programming the receiver to use the transmit clock and/or to start a data transfer when transmission
starts. Alternatively, this can be done by programming the transmitter to use the receive clock
and/or to start a data transfer when reception starts. The transmitter and the receiver can be programmed to operate with the clock signals provided on either the TK or RK pins. This allows the
SSC to support many slave-mode data transfers. The maximum clock speed allowed on the TK
and RK pins is the master clock divided by 2.
Figure 34-4. SSC Functional Block Diagram
Transmitter
MCK
TK Input
Clock
Divider
Transmit Clock
Controller
RX clock
TXEN
RX Start Start
Selector
TF
TK
Frame Sync
Controller
TF
Data
Controller
TD
Clock Output
Controller
RK
Frame Sync
Controller
RF
Data
Controller
RD
TX clock
TX Start
Transmit Shift Register
Transmit Holding
Register
APB
Clock Output
Controller
Transmit Sync
Holding Register
User
Interface
Receiver
RK Input
Receive Clock RX Clock
Controller
TX Clock
RXEN
TX Start Start
RF
Selector
RC0R
Interrupt Control
RX Start
Receive Shift Register
Receive Holding
Register
Receive Sync
Holding Register
AIC
618
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
34.7.1
Clock Management
Transmitter
MCK
TK Input
Clock
Divider
Transmit Clock
Controller
RX clock
Clock Output
Controller
TK
Frame Sync
Controller
TF
Data
Controller
TD
Clock Output
Controller
RK
Frame Sync
Controller
RF
Data
Controller
RD
TX clock
TXEN
RX Start Start
Selector
TF
TX Start
Transmit Shift Register
Transmit Holding
Register
APB
Transmit Sync
Holding Register
User
Interface
Receiver
RK Input
Receive Clock RX Clock
Controller
TX Clock
RXEN
TX Start
Start
RF
Selector
RC0R
Interrupt Control
RX Start
Receive Shift Register
Receive Holding
Register
Receive Sync
Holding Register
NVIC
The transmitter clock can be generated by:
• an external clock received on the TK I/O pad
• the receiver clock
• the internal clock divider
The receiver clock can be generated by:
• an external clock received on the RK I/O pad
• the transmitter clock
• the internal clock divider
Furthermore, the transmitter block can generate an external clock on the TK I/O pad, and the
receiver block can generate an external clock on the RK I/O pad.
This allows the SSC to support many Master and Slave Mode data transfers.
619
6438D–ATARM–13-Oct-09
34.7.1.1
Clock Divider
Figure 34-5. Divided Clock Block Diagram
Clock Divider
SSC_CMR
MCK
/2
12-bit Counter
Divided Clock
The Master Clock divider is determined by the 12-bit field DIV counter and comparator (so its
maximal value is 4095) in the Clock Mode Register SSC_CMR, allowing a Master Clock division
by up to 8190. The Divided Clock is provided to both the Receiver and Transmitter. When this
field is programmed to 0, the Clock Divider is not used and remains inactive.
When DIV is set to a value equal to or greater than 1, the Divided Clock has a frequency of Master Clock divided by 2 times DIV. Each level of the Divided Clock has a duration of the Master
Clock multiplied by DIV. This ensures a 50% duty cycle for the Divided Clock regardless of
whether the DIV value is even or odd.
Figure 34-6.
Divided Clock Generation
Master Clock
Divided Clock
DIV = 1
Divided Clock Frequency = MCK/2
Master Clock
Divided Clock
DIV = 3
Divided Clock Frequency = MCK/6
Table 34-4.
34.7.1.2
Maximum
Minimum
MCK / 2
MCK / 8190
Transmitter Clock Management
The transmitter clock is generated from the receiver clock or the divider clock or an external
clock scanned on the TK I/O pad. The transmitter clock is selected by the CKS field in
SSC_TCMR (Transmit Clock Mode Register). Transmit Clock can be inverted independently by
the CKI bits in SSC_TCMR.
The transmitter can also drive the TK I/O pad continuously or be limited to the actual data transfer. The clock output is configured by the SSC_TCMR register. The Transmit Clock Inversion
(CKI) bits have no effect on the clock outputs. Programming the TCMR register to select TK pin
620
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
(CKS field) and at the same time Continuous Transmit Clock (CKO field) might lead to unpredictable results.
Figure 34-7. Transmitter Clock Management
TK (pin)
Clock
Output
Tri_state
Controller
MUX
Receiver
Clock
Divider
Clock
Data Transfer
CKO
CKS
34.7.1.3
INV
MUX
Tri-state
Controller
CKI
CKG
Transmitter
Clock
Receiver Clock Management
The receiver clock is generated from the transmitter clock or the divider clock or an external
clock scanned on the RK I/O pad. The Receive Clock is selected by the CKS field in
SSC_RCMR (Receive Clock Mode Register). Receive Clocks can be inverted independently by
the CKI bits in SSC_RCMR.
The receiver can also drive the RK I/O pad continuously or be limited to the actual data transfer.
The clock output is configured by the SSC_RCMR register. The Receive Clock Inversion (CKI)
bits have no effect on the clock outputs. Programming the RCMR register to select RK pin (CKS
field) and at the same time Continuous Receive Clock (CKO field) can lead to unpredictable
results.
Figure 34-8. Receiver Clock Management
RK (pin)
Tri-state
Controller
MUX
Clock
Output
Transmitter
Clock
Divider
Clock
Data Transfer
CKO
CKS
INV
MUX
Tri-state
Controller
CKI
CKG
Receiver
Clock
621
6438D–ATARM–13-Oct-09
34.7.1.4
Serial Clock Ratio Considerations
The Transmitter and the Receiver can be programmed to operate with the clock signals provided
on either the TK or RK pins. This allows the SSC to support many slave-mode data transfers. In
this case, the maximum clock speed allowed on the RK pin is:
– Master Clock divided by 2 if Receiver Frame Synchro is input
– Master Clock divided by 3 if Receiver Frame Synchro is output
In addition, the maximum clock speed allowed on the TK pin is:
– Master Clock divided by 6 if Transmit Frame Synchro is input
– Master Clock divided by 2 if Transmit Frame Synchro is output
34.7.2
Transmitter Operations
A transmitted frame is triggered by a start event and can be followed by synchronization data
before data transmission.
The start event is configured by setting the Transmit Clock Mode Register (SSC_TCMR). See
“Start” on page 623.
The frame synchronization is configured setting the Transmit Frame Mode Register
(SSC_TFMR). See “Frame Sync” on page 625.
To transmit data, the transmitter uses a shift register clocked by the transmitter clock signal and
the start mode selected in the SSC_TCMR. Data is written by the application to the SSC_THR
register then transferred to the shift register according to the data format selected.
When both the SSC_THR and the transmit shift register are empty, the status flag TXEMPTY is
set in SSC_SR. When the Transmit Holding register is transferred in the Transmit shift register,
the status flag TXRDY is set in SSC_SR and additional data can be loaded in the holding
register.
Figure 34-9. Transmitter Block Diagram
SSC_CRTXEN
SSC_SRTXEN
TXEN
SSC_CRTXDIS
SSC_TCMR.STTDLY
SSC_TFMR.FSDEN
SSC_RCMR.START SSC_TCMR.START SSC_TFMR.DATNB
SSC_TFMR.DATDEF
SSC_TFMR.MSBF
RXEN
TXEN
TX Start
TX Start
Start
RX Start
Start
RF
Selector
Selector
RF
RC0R
TX Controller
TD
Transmit Shift Register
SSC_TFMR.FSDEN
SSC_TCMR.STTDLY != 0
SSC_TFMR.DATLEN
0
SSC_THR
Transmitter Clock
1
SSC_TSHR
SSC_TFMR.FSLEN
TX Controller counter reached STTDLY
622
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
34.7.3
Receiver Operations
A received frame is triggered by a start event and can be followed by synchronization data
before data transmission.
The start event is configured setting the Receive Clock Mode Register (SSC_RCMR). See
“Start” on page 623.
The frame synchronization is configured setting the Receive Frame Mode Register
(SSC_RFMR). See “Frame Sync” on page 625.
The receiver uses a shift register clocked by the receiver clock signal and the start mode
selected in the SSC_RCMR. The data is transferred from the shift register depending on the
data format selected.
When the receiver shift register is full, the SSC transfers this data in the holding register, the status flag RXRDY is set in SSC_SR and the data can be read in the receiver holding register. If
another transfer occurs before read of the RHR register, the status flag OVERUN is set in
SSC_SR and the receiver shift register is transferred in the RHR register.
Figure 34-10. Receiver Block Diagram
SSC_CR.RXEN
SSC_SR.RXEN
SSC_CR.RXDIS
SSC_TCMR.START
SSC_RCMR.START
TXEN
RX Start
RF
Start
Selector
RXEN
RF
RC0R
Start
Selector
SSC_RFMR.MSBF
SSC_RFMR.DATNB
RX Start
RX Controller
RD
Receive Shift Register
SSC_RCMR.STTDLY != 0
load
SSC_RSHR
SSC_RFMR.FSLEN
load
SSC_RHR
Receiver Clock
SSC_RFMR.DATLEN
RX Controller counter reached STTDLY
34.7.4
Start
The transmitter and receiver can both be programmed to start their operations when an event
occurs, respectively in the Transmit Start Selection (START) field of SSC_TCMR and in the
Receive Start Selection (START) field of SSC_RCMR.
Under the following conditions the start event is independently programmable:
• Continuous. In this case, the transmission starts as soon as a word is written in SSC_THR
and the reception starts as soon as the Receiver is enabled.
• Synchronously with the transmitter/receiver
• On detection of a falling/rising edge on TF/RF
• On detection of a low level/high level on TF/RF
• On detection of a level change or an edge on TF/RF
623
6438D–ATARM–13-Oct-09
A start can be programmed in the same manner on either side of the Transmit/Receive Clock
Register (RCMR/TCMR). Thus, the start could be on TF (Transmit) or RF (Receive).
Moreover, the Receiver can start when data is detected in the bit stream with the Compare
Functions.
Detection on TF/RF input/output is done by the field FSOS of the Transmit/Receive Frame Mode
Register (TFMR/RFMR).
Figure 34-11. Transmit Start Mode
TK
TF
(Input)
Start = Low Level on TF
Start = Falling Edge on TF
Start = High Level on TF
Start = Rising Edge on TF
Start = Level Change on TF
Start = Any Edge on TF
TD
(Output)
TD
(Output)
X
BO
STTDLY
BO
X
B1
STTDLY
BO
X
TD
(Output)
B1
STTDLY
TD
(Output)
BO
X
B1
STTDLY
TD
(Output)
TD
(Output)
B1
BO
X
B1
BO
B1
STTDLY
X
B1
BO
BO
B1
STTDLY
Figure 34-12. Receive Pulse/Edge Start Modes
RK
RF
(Input)
Start = Low Level on RF
Start = Falling Edge on RF
Start = High Level on RF
Start = Rising Edge on RF
Start = Level Change on RF
Start = Any Edge on RF
RD
(Input)
RD
(Input)
X
BO
STTDLY
BO
X
B1
STTDLY
BO
X
RD
(Input)
B1
STTDLY
RD
(Input)
BO
X
B1
STTDLY
RD
(Input)
RD
(Input)
B1
BO
X
B1
BO
B1
STTDLY
X
BO
B1
BO
B1
STTDLY
624
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
34.7.5
Frame Sync
The Transmitter and Receiver Frame Sync pins, TF and RF, can be programmed to generate
different kinds of frame synchronization signals. The Frame Sync Output Selection (FSOS) field
in the Receive Frame Mode Register (SSC_RFMR) and in the Transmit Frame Mode Register
(SSC_TFMR) are used to select the required waveform.
• Programmable low or high levels during data transfer are supported.
• Programmable high levels before the start of data transfers or toggling are also supported.
If a pulse waveform is selected, the Frame Sync Length (FSLEN) field in SSC_RFMR and
SSC_TFMR programs the length of the pulse, from 1 bit time up to 256 bit time.
The periodicity of the Receive and Transmit Frame Sync pulse output can be programmed
through the Period Divider Selection (PERIOD) field in SSC_RCMR and SSC_TCMR.
34.7.5.1
Frame Sync Data
Frame Sync Data transmits or receives a specific tag during the Frame Sync signal.
During the Frame Sync signal, the Receiver can sample the RD line and store the data in the
Receive Sync Holding Register and the transmitter can transfer Transmit Sync Holding Register
in the Shifter Register. The data length to be sampled/shifted out during the Frame Sync signal
is programmed by the FSLEN field in SSC_RFMR/SSC_TFMR and has a maximum value of 16.
Concerning the Receive Frame Sync Data operation, if the Frame Sync Length is equal to or
lower than the delay between the start event and the actual data reception, the data sampling
operation is performed in the Receive Sync Holding Register through the Receive Shift Register.
The Transmit Frame Sync Operation is performed by the transmitter only if the bit Frame Sync
Data Enable (FSDEN) in SSC_TFMR is set. If the Frame Sync length is equal to or lower than
the delay between the start event and the actual data transmission, the normal transmission has
priority and the data contained in the Transmit Sync Holding Register is transferred in the Transmit Register, then shifted out.
34.7.5.2
34.7.6
Frame Sync Edge Detection
The Frame Sync Edge detection is programmed by the FSEDGE field in
SSC_RFMR/SSC_TFMR. This sets the corresponding flags RXSYN/TXSYN in the SSC Status
Register (SSC_SR) on frame synchro edge detection (signals RF/TF).
Receive Compare Modes
Figure 34-13. Receive Compare Modes
RK
RD
(Input)
CMP0
CMP1
CMP2
CMP3
Ignored
B0
B1
B2
Start
FSLEN
Up to 16 Bits
(4 in This Example)
STDLY
DATLEN
625
6438D–ATARM–13-Oct-09
34.7.6.1
34.7.7
Compare Functions
Length of the comparison patterns (Compare 0, Compare 1) and thus the number of bits they
are compared to is defined by FSLEN, but with a maximum value of 16 bits. Comparison is
always done by comparing the last bits received with the comparison pattern. Compare 0 can be
one start event of the Receiver. In this case, the receiver compares at each new sample the last
bits received at the Compare 0 pattern contained in the Compare 0 Register (SSC_RC0R).
When this start event is selected, the user can program the Receiver to start a new data transfer
either by writing a new Compare 0, or by receiving continuously until Compare 1 occurs. This
selection is done with the bit (STOP) in SSC_RCMR.
Data Format
The data framing format of both the transmitter and the receiver are programmable through the
Transmitter Frame Mode Register (SSC_TFMR) and the Receiver Frame Mode Register
(SSC_RFMR). In either case, the user can independently select:
• the event that starts the data transfer (START)
• the delay in number of bit periods between the start event and the first data bit (STTDLY)
• the length of the data (DATLEN)
• the number of data to be transferred for each start event (DATNB).
• the length of synchronization transferred for each start event (FSLEN)
• the bit sense: most or lowest significant bit first (MSBF)
Additionally, the transmitter can be used to transfer synchronization and select the level driven
on the TD pin while not in data transfer operation. This is done respectively by the Frame Sync
Data Enable (FSDEN) and by the Data Default Value (DATDEF) bits in SSC_TFMR.
626
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Table 34-5.
Data Frame Registers
Transmitter
Receiver
Field
Length
Comment
SSC_TFMR
SSC_RFMR
DATLEN
Up to 32
Size of word
SSC_TFMR
SSC_RFMR
DATNB
Up to 16
Number of words transmitted in frame
SSC_TFMR
SSC_RFMR
MSBF
SSC_TFMR
SSC_RFMR
FSLEN
Up to 16
Size of Synchro data register
SSC_TFMR
DATDEF
0 or 1
Data default value ended
SSC_TFMR
FSDEN
Most significant bit first
Enable send SSC_TSHR
SSC_TCMR
SSC_RCMR
PERIOD
Up to 512
Frame size
SSC_TCMR
SSC_RCMR
STTDLY
Up to 255
Size of transmit start delay
Figure 34-14. Transmit and Receive Frame Format in Edge/Pulse Start Modes
Start
Start
PERIOD
TF/RF
(1)
FSLEN
TD
(If FSDEN = 1)
TD
(If FSDEN = 0)
RD
Sync Data
Data
Data
From SSC_THR
From SSC_THR
Default
From SSC_TSHR FromDATDEF
Default
Sync Data
Ignored
To SSC_RSHR
STTDLY
From SSC_THR
Data
Data
To SSC_RHR
To SSC_RHR
DATLEN
DATLEN
Sync Data
FromDATDEF
Data
Data
From SSC_THR
From DATDEF
Default
Default
From DATDEF
Ignored
Sync Data
DATNB
Note:
1. Example of input on falling edge of TF/RF.
Figure 34-15. Transmit Frame Format in Continuous Mode
Start
TD
Data
From SSC_THR
Data
Default
From SSC_THR
DATLEN
DATLEN
Start: 1. TXEMPTY set to 1
2. Write into the SSC_THR
627
6438D–ATARM–13-Oct-09
AT91SAM9G45
Note:
1. STTDLY is set to 0. In this example, SSC_THR is loaded twice. FSDEN value has no effect on
the transmission. SyncData cannot be output in continuous mode.
Figure 34-16. Receive Frame Format in Continuous Mode
Start = Enable Receiver
Data
Data
To SSC_RHR
To SSC_RHR
DATLEN
DATLEN
RD
Note:
34.7.8
1. STTDLY is set to 0.
Loop Mode
The receiver can be programmed to receive transmissions from the transmitter. This is done by
setting the Loop Mode (LOOP) bit in SSC_RFMR. In this case, RD is connected to TD, RF is
connected to TF and RK is connected to TK.
34.7.9
Interrupt
Most bits in SSC_SR have a corresponding bit in interrupt management registers.
The SSC can be programmed to generate an interrupt when it detects an event. The interrupt is
controlled by writing SSC_IER (Interrupt Enable Register) and SSC_IDR (Interrupt Disable Register) These registers enable and disable, respectively, the corresponding interrupt by setting
and clearing the corresponding bit in SSC_IMR (Interrupt Mask Register), which controls the
generation of interrupts by asserting the SSC interrupt line connected to the AIC.
Figure 34-17. Interrupt Block Diagram
SSC_IMR
SSC_IER
PDC
SSC_IDR
Set
Clear
TXBUFE
ENDTX
Transmitter
TXRDY
TXEMPTY
TXSYNC
Interrupt
Control
RXBUFF
ENDRX
SSC Interrupt
Receiver
RXRDY
OVRUN
RXSYNC
628
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 34-18. Interrupt Block Diagram
SSC_IMR
SSC_IER
SSC_IDR
Set
Clear
Transmitter
TXRDY
TXEMPTY
TXSYNC
Interrupt
Control
SSC Interrupt
Receiver
RXRDY
OVRUN
RXSYNC
629
6438D–ATARM–13-Oct-09
AT91SAM9G45
34.8
SSC Application Examples
The SSC can support several serial communication modes used in audio or high speed serial
links. Some standard applications are shown in the following figures. All serial link applications
supported by the SSC are not listed here.
Figure 34-19. Audio Application Block Diagram
Clock SCK
TK
Word Select WS
I2S
RECEIVER
TF
Data SD
SSC
TD
RD
Clock SCK
RF
Word Select WS
RK
MSB
Data SD
LSB
MSB
Right Channel
Left Channel
Figure 34-20. Codec Application Block Diagram
Serial Data Clock (SCLK)
TK
Frame sync (FSYNC)
TF
Serial Data Out
SSC
CODEC
TD
Serial Data In
RD
RF
RK
Serial Data Clock (SCLK)
Frame sync (FSYNC)
First Time Slot
Dstart
Dend
Serial Data Out
Serial Data In
630
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 34-21. Time Slot Application Block Diagram
SCLK
TK
FSYNC
TF
CODEC
First
Time Slot
Data Out
TD
SSC
RD
Data in
RF
RK
CODEC
Second
Time Slot
Serial Data Clock (SCLK)
Frame sync (FSYNC)
First Time Slot
Dstart
Second Time Slot
Dend
Serial Data Out
Serial Data in
631
6438D–ATARM–13-Oct-09
AT91SAM9G45
34.9
Synchronous Serial Controller (SSC) User Interface
Table 34-6.
Offset
Register Mapping
Register
Name
Access
Reset
SSC_CR
Write-only
–
SSC_CMR
Read-write
0x0
0x0
Control Register
0x4
Clock Mode Register
0x8
Reserved
–
–
–
0xC
Reserved
–
–
–
0x10
Receive Clock Mode Register
SSC_RCMR
Read-write
0x0
0x14
Receive Frame Mode Register
SSC_RFMR
Read-write
0x0
0x18
Transmit Clock Mode Register
SSC_TCMR
Read-write
0x0
0x1C
Transmit Frame Mode Register
SSC_TFMR
Read-write
0x0
0x20
Receive Holding Register
SSC_RHR
Read-only
0x0
0x24
Transmit Holding Register
SSC_THR
Write-only
–
0x28
Reserved
–
–
–
0x2C
Reserved
–
–
–
0x30
Receive Sync. Holding Register
SSC_RSHR
Read-only
0x0
0x34
Transmit Sync. Holding Register
SSC_TSHR
Read-write
0x0
0x38
Receive Compare 0 Register
SSC_RC0R
Read-write
0x0
0x3C
Receive Compare 1 Register
SSC_RC1R
Read-write
0x0
0x40
Status Register
SSC_SR
Read-only
0x000000CC
0x44
Interrupt Enable Register
SSC_IER
Write-only
–
0x48
Interrupt Disable Register
SSC_IDR
Write-only
–
0x4C
Interrupt Mask Register
SSC_IMR
Read-only
0x0
Reserved
–
–
–
Reserved for Peripheral Data Controller (PDC)
–
–
–
0x50-0xFC
0x100- 0x124
632
6438D–ATARM–13-Oct-09
AT91SAM9G45
34.9.1
Name:
SSC Control Register
SSC_CR:
Addresses:
0xFFF9C000 (0), 0xFFFA0000 (1)
Access:
Write-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
SWRST
14
–
13
–
12
–
11
–
10
–
9
TXDIS
8
TXEN
7
–
6
–
5
–
4
–
3
–
2
–
1
RXDIS
0
RXEN
• RXEN: Receive Enable
0 = No effect.
1 = Enables Receive if RXDIS is not set.
• RXDIS: Receive Disable
0 = No effect.
1 = Disables Receive. If a character is currently being received, disables at end of current character reception.
• TXEN: Transmit Enable
0 = No effect.
1 = Enables Transmit if TXDIS is not set.
• TXDIS: Transmit Disable
0 = No effect.
1 = Disables Transmit. If a character is currently being transmitted, disables at end of current character transmission.
• SWRST: Software Reset
0 = No effect.
1 = Performs a software reset. Has priority on any other bit in SSC_CR.
633
6438D–ATARM–13-Oct-09
AT91SAM9G45
34.9.2
Name:
SSC Clock Mode Register
SSC_CMR
Addresses:
0xFFF9C004 (0), 0xFFFA0004 (1)
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
10
9
8
7
6
5
4
1
0
DIV
3
2
DIV
• DIV: Clock Divider
0 = The Clock Divider is not active.
Any Other Value: The Divided Clock equals the Master Clock divided by 2 times DIV. The maximum bit rate is MCK/2. The
minimum bit rate is MCK/2 x 4095 = MCK/8190.
634
6438D–ATARM–13-Oct-09
AT91SAM9G45
34.9.3
Name:
SSC Receive Clock Mode Register
SSC_RCMR
Addresses:
0xFFF9C010 (0), 0xFFFA0010 (1)
Access:
Read-write
31
30
29
28
27
26
25
24
19
18
17
16
10
9
8
PERIOD
23
22
21
20
STTDLY
15
–
7
14
–
13
–
12
STOP
11
6
5
CKI
4
3
CKO
CKG
START
2
1
0
CKS
• CKS: Receive Clock Selection
CKS
Selected Receive Clock
0x0
Divided Clock
0x1
TK Clock signal
0x2
RK pin
0x3
Reserved
• CKO: Receive Clock Output Mode Selection
CKO
Receive Clock Output Mode
0x0
None
0x1
Continuous Receive Clock
Output
0x2
Receive Clock only during data transfers
Output
0x3-0x7
RK pin
Input-only
Reserved
• CKI: Receive Clock Inversion
0 = The data inputs (Data and Frame Sync signals) are sampled on Receive Clock falling edge. The Frame Sync signal
output is shifted out on Receive Clock rising edge.
1 = The data inputs (Data and Frame Sync signals) are sampled on Receive Clock rising edge. The Frame Sync signal output is shifted out on Receive Clock falling edge.
CKI affects only the Receive Clock and not the output clock signal.
635
6438D–ATARM–13-Oct-09
AT91SAM9G45
• CKG: Receive Clock Gating Selection
CKG
Receive Clock Gating
0x0
None, continuous clock
0x1
Receive Clock enabled only if RF Low
0x2
Receive Clock enabled only if RF High
0x3
Reserved
• START: Receive Start Selection
START
Receive Start
0x0
Continuous, as soon as the receiver is enabled, and immediately after the end of
transfer of the previous data.
0x1
Transmit start
0x2
Detection of a low level on RF signal
0x3
Detection of a high level on RF signal
0x4
Detection of a falling edge on RF signal
0x5
Detection of a rising edge on RF signal
0x6
Detection of any level change on RF signal
0x7
Detection of any edge on RF signal
0x8
Compare 0
0x9-0xF
Reserved
• STOP: Receive Stop Selection
0 = After completion of a data transfer when starting with a Compare 0, the receiver stops the data transfer and waits for a
new compare 0.
1 = After starting a receive with a Compare 0, the receiver operates in a continuous mode until a Compare 1 is detected.
• STTDLY: Receive Start Delay
If STTDLY is not 0, a delay of STTDLY clock cycles is inserted between the start event and the actual start of reception.
When the Receiver is programmed to start synchronously with the Transmitter, the delay is also applied.
Note: It is very important that STTDLY be set carefully. If STTDLY must be set, it should be done in relation to TAG
(Receive Sync Data) reception.
• PERIOD: Receive Period Divider Selection
This field selects the divider to apply to the selected Receive Clock in order to generate a new Frame Sync Signal. If 0, no
PERIOD signal is generated. If not 0, a PERIOD signal is generated each 2 x (PERIOD+1) Receive Clock.
636
6438D–ATARM–13-Oct-09
AT91SAM9G45
34.9.4
Name:
SSC Receive Frame Mode Register
SSC_RFMR
Addresses:
0xFFF9C014 (0), 0xFFFA0014 (1)
Access:
Read-write
31
FSLEN_EXT
30
FSLEN_EXT
29
FSLEN_EXT
23
–
22
15
–
7
MSBF
28
FSLEN_EXT
27
–
26
–
21
FSOS
20
19
18
14
–
13
–
12
–
11
6
–
5
LOOP
4
3
25
–
24
FSEDGE
17
16
9
8
1
0
FSLEN
10
DATNB
2
DATLEN
• DATLEN: Data Length
0 = Forbidden value (1-bit data length not supported).
Any other value: The bit stream contains DATLEN + 1 data bits. Moreover, it defines the transfer size performed by the
PDC assigned to the Receiver. If DATLEN is lower or equal to 7, data transfers are in bytes. If DATLEN is between 8 and
15 (included), half-words are transferred, and for any other value, 32-bit words are transferred.
• LOOP: Loop Mode
0 = Normal operating mode.
1 = RD is driven by TD, RF is driven by TF and TK drives RK.
• MSBF: Most Significant Bit First
0 = The lowest significant bit of the data register is sampled first in the bit stream.
1 = The most significant bit of the data register is sampled first in the bit stream.
• DATNB: Data Number per Frame
This field defines the number of data words to be received after each transfer start, which is equal to (DATNB + 1).
• FSLEN: Receive Frame Sync Length
This field defines the number of bits sampled and stored in the Receive Sync Data Register. When this mode is selected by
the START field in the Receive Clock Mode Register, it also determines the length of the sampled data to be compared to
the Compare 0 or Compare 1 register.
This field is used with FSLEN_EXT to determine the pulse length of the Receive Frame Sync signal.
Pulse length is equal to FSLEN + (FSLEN_EXT * 16) + 1 Receive Clock periods.
637
6438D–ATARM–13-Oct-09
AT91SAM9G45
• FSOS: Receive Frame Sync Output Selection
FSOS
Selected Receive Frame Sync Signal
RF Pin
0x0
None
0x1
Negative Pulse
Output
0x2
Positive Pulse
Output
0x3
Driven Low during data transfer
Output
0x4
Driven High during data transfer
Output
0x5
Toggling at each start of data transfer
Output
0x6-0x7
Input-only
Reserved
Undefined
• FSEDGE: Frame Sync Edge Detection
Determines which edge on Frame Sync will generate the interrupt RXSYN in the SSC Status Register.
FSEDGE
Frame Sync Edge Detection
0x0
Positive Edge Detection
0x1
Negative Edge Detection
• FSLEN_EXT: FSLEN Field Extension
Extends FSLEN field. For details, refer to FSLEN bit description on page 637.
638
6438D–ATARM–13-Oct-09
AT91SAM9G45
34.9.5
Name:
SSC Transmit Clock Mode Register
SSC_TCMR
Addresses:
0xFFF9C018 (0), 0xFFFA0018 (1)
Access:
Read-write
31
30
29
28
27
26
25
24
19
18
17
16
10
9
8
PERIOD
23
22
21
20
STTDLY
15
–
7
14
–
13
–
12
–
11
6
5
CKI
4
3
CKO
CKG
START
2
1
0
CKS
• CKS: Transmit Clock Selection
CKS
Selected Transmit Clock
0x0
Divided Clock
0x1
RK Clock signal
0x2
TK Pin
0x3
Reserved
• CKO: Transmit Clock Output Mode Selection
CKO
Transmit Clock Output Mode
0x0
None
0x1
Continuous Transmit Clock
Output
0x2
Transmit Clock only during data transfers
Output
0x3-0x7
TK pin
Input-only
Reserved
• CKI: Transmit Clock Inversion
0 = The data outputs (Data and Frame Sync signals) are shifted out on Transmit Clock falling edge. The Frame sync signal
input is sampled on Transmit clock rising edge.
1 = The data outputs (Data and Frame Sync signals) are shifted out on Transmit Clock rising edge. The Frame sync signal
input is sampled on Transmit clock falling edge.
CKI affects only the Transmit Clock and not the output clock signal.
639
6438D–ATARM–13-Oct-09
AT91SAM9G45
• CKG: Transmit Clock Gating Selection
CKG
Transmit Clock Gating
0x0
None, continuous clock
0x1
Transmit Clock enabled only if TF Low
0x2
Transmit Clock enabled only if TF High
0x3
Reserved
• START: Transmit Start Selection
START
Transmit Start
0x0
Continuous, as soon as a word is written in the SSC_THR Register (if Transmit is enabled), and
immediately after the end of transfer of the previous data.
0x1
Receive start
0x2
Detection of a low level on TF signal
0x3
Detection of a high level on TF signal
0x4
Detection of a falling edge on TF signal
0x5
Detection of a rising edge on TF signal
0x6
Detection of any level change on TF signal
0x7
Detection of any edge on TF signal
0x8 - 0xF
Reserved
• STTDLY: Transmit Start Delay
If STTDLY is not 0, a delay of STTDLY clock cycles is inserted between the start event and the actual start of transmission
of data. When the Transmitter is programmed to start synchronously with the Receiver, the delay is also applied.
Note: STTDLY must be set carefully. If STTDLY is too short in respect to TAG (Transmit Sync Data) emission, data is emitted instead of the end of TAG.
• PERIOD: Transmit Period Divider Selection
This field selects the divider to apply to the selected Transmit Clock to generate a new Frame Sync Signal. If 0, no period
signal is generated. If not 0, a period signal is generated at each 2 x (PERIOD+1) Transmit Clock.
640
6438D–ATARM–13-Oct-09
AT91SAM9G45
34.9.6
Name:
SSC Transmit Frame Mode Register
SSC_TFMR
Addresses:
0xFFF9C01C (0), 0xFFFA001C (1)
Access:
Read-write
31
FSLEN_EXT
30
FSLEN_EXT
29
FSLEN_EXT
23
FSDEN
22
15
–
7
MSBF
28
FSLEN_EXT
27
–
26
–
21
FSOS
20
19
18
14
–
13
–
12
–
11
6
–
5
DATDEF
4
3
25
–
24
FSEDGE
17
16
9
8
1
0
FSLEN
10
DATNB
2
DATLEN
• DATLEN: Data Length
0 = Forbidden value (1-bit data length not supported).
Any other value: The bit stream contains DATLEN + 1 data bits. Moreover, it defines the transfer size performed by the
PDC assigned to the Transmit. If DATLEN is lower or equal to 7, data transfers are bytes, if DATLEN is between 8 and 15
(included), half-words are transferred, and for any other value, 32-bit words are transferred.
• DATDEF: Data Default Value
This bit defines the level driven on the TD pin while out of transmission. Note that if the pin is defined as multi-drive by the
PIO Controller, the pin is enabled only if the SCC TD output is 1.
• MSBF: Most Significant Bit First
0 = The lowest significant bit of the data register is shifted out first in the bit stream.
1 = The most significant bit of the data register is shifted out first in the bit stream.
• DATNB: Data Number per frame
This field defines the number of data words to be transferred after each transfer start, which is equal to (DATNB +1).
• FSLEN: Transmit Frame Syn Length
This field defines the length of the Transmit Frame Sync signal and the number of bits shifted out from the Transmit Sync
Data Register if FSDEN is 1.
This field is used with FSLEN_EXT to determine the pulse length of the Transmit Frame Sync signal.
Pulse length is equal to FSLEN + (FSLEN_EXT * 16) + 1 Transmit Clock period.
641
6438D–ATARM–13-Oct-09
AT91SAM9G45
• FSOS: Transmit Frame Sync Output Selection
FSOS
Selected Transmit Frame Sync Signal
TF Pin
0x0
None
0x1
Negative Pulse
Output
0x2
Positive Pulse
Output
0x3
Driven Low during data transfer
Output
0x4
Driven High during data transfer
Output
0x5
Toggling at each start of data transfer
Output
0x6-0x7
Reserved
Input-only
Undefined
• FSDEN: Frame Sync Data Enable
0 = The TD line is driven with the default value during the Transmit Frame Sync signal.
1 = SSC_TSHR value is shifted out during the transmission of the Transmit Frame Sync signal.
• FSEDGE: Frame Sync Edge Detection
Determines which edge on frame sync will generate the interrupt TXSYN (Status Register).
FSEDGE
Frame Sync Edge Detection
0x0
Positive Edge Detection
0x1
Negative Edge Detection
• FSLEN_EXT: FSLEN Field Extension
Extends FSLEN field. For details, refer to FSLEN bit description on page 641.
642
6438D–ATARM–13-Oct-09
AT91SAM9G45
34.9.7
Name:
SSC Receive Holding Register
SSC_RHR
Addresses:
0xFFF9C020 (0), 0xFFFA0020 (1)
Access:
Read-only
31
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
0
RDAT
23
22
21
20
RDAT
15
14
13
12
RDAT
7
6
5
4
RDAT
• RDAT: Receive Data
Right aligned regardless of the number of data bits defined by DATLEN in SSC_RFMR.
34.9.8
Name:
SSC Transmit Holding Register
SSC_THR
Addresses:
0xFFF9C024 (0), 0xFFFA0024 (1)
Access:
Write-only
31
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
0
TDAT
23
22
21
20
TDAT
15
14
13
12
TDAT
7
6
5
4
TDAT
• TDAT: Transmit Data
Right aligned regardless of the number of data bits defined by DATLEN in SSC_TFMR.
643
6438D–ATARM–13-Oct-09
AT91SAM9G45
34.9.9
Name:
SSC Receive Synchronization Holding Register
SSC_RSHR
Addresses:
0xFFF9C030 (0), 0xFFFA0030 (1)
Access:
Read-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
14
13
12
11
10
9
8
3
2
1
0
RSDAT
7
6
5
4
RSDAT
• RSDAT: Receive Synchronization Data
34.9.10
Name:
SSC Transmit Synchronization Holding Register
SSC_TSHR
Addresses:
0xFFF9C034 (0), 0xFFFA0034 (1)
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
14
13
12
11
10
9
8
3
2
1
0
TSDAT
7
6
5
4
TSDAT
• TSDAT: Transmit Synchronization Data
644
6438D–ATARM–13-Oct-09
AT91SAM9G45
34.9.11
Name:
SSC Receive Compare 0 Register
SSC_RC0R
Addresses:
0xFFF9C038 (0), 0xFFFA0038 (1)
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
14
13
12
11
10
9
8
3
2
1
0
CP0
7
6
5
4
CP0
• CP0: Receive Compare Data 0
645
6438D–ATARM–13-Oct-09
AT91SAM9G45
34.9.12
Name:
SSC Receive Compare 1 Register
SSC_RC1R
Addresses:
0xFFF9C03C (0), 0xFFFA003C (1)
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
14
13
12
11
10
9
8
3
2
1
0
CP1
7
6
5
4
CP1
• CP1: Receive Compare Data 1
646
6438D–ATARM–13-Oct-09
AT91SAM9G45
34.9.13
Name:
SSC Status Register
SSC_SR
Addresses:
0xFFF9C040 (0), 0xFFFA0040 (1)
Access:
Read-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
RXEN
16
TXEN
15
–
14
–
13
–
12
–
11
RXSYN
10
TXSYN
9
CP1
8
CP0
7
RXBUFF
6
ENDRX
5
OVRUN
4
RXRDY
3
TXBUFE
2
ENDTX
1
TXEMPTY
0
TXRDY
• TXRDY: Transmit Ready
0 = Data has been loaded in SSC_THR and is waiting to be loaded in the Transmit Shift Register (TSR).
1 = SSC_THR is empty.
• TXEMPTY: Transmit Empty
0 = Data remains in SSC_THR or is currently transmitted from TSR.
1 = Last data written in SSC_THR has been loaded in TSR and last data loaded in TSR has been transmitted.
• ENDTX: End of Transmission
0 = The register SSC_TCR has not reached 0 since the last write in SSC_TCR or SSC_TNCR.
1 = The register SSC_TCR has reached 0 since the last write in SSC_TCR or SSC_TNCR.
• TXBUFE: Transmit Buffer Empty
0 = SSC_TCR or SSC_TNCR have a value other than 0.
1 = Both SSC_TCR and SSC_TNCR have a value of 0.
• RXRDY: Receive Ready
0 = SSC_RHR is empty.
1 = Data has been received and loaded in SSC_RHR.
• OVRUN: Receive Overrun
0 = No data has been loaded in SSC_RHR while previous data has not been read since the last read of the Status
Register.
1 = Data has been loaded in SSC_RHR while previous data has not yet been read since the last read of the Status
Register.
• ENDRX: End of Reception
0 = Data is written on the Receive Counter Register or Receive Next Counter Register.
1 = End of PDCDMAC transfer when Receive Counter Register has arrived at zero.
647
6438D–ATARM–13-Oct-09
AT91SAM9G45
• RXBUFF: Receive Buffer Full
0 = SSC_RCR or SSC_RNCR have a value other than 0.
1 = Both SSC_RCR and SSC_RNCR have a value of 0.
• CP0: Compare 0
0 = A compare 0 has not occurred since the last read of the Status Register.
1 = A compare 0 has occurred since the last read of the Status Register.
• CP1: Compare 1
0 = A compare 1 has not occurred since the last read of the Status Register.
1 = A compare 1 has occurred since the last read of the Status Register.
• TXSYN: Transmit Sync
0 = A Tx Sync has not occurred since the last read of the Status Register.
1 = A Tx Sync has occurred since the last read of the Status Register.
• RXSYN: Receive Sync
0 = An Rx Sync has not occurred since the last read of the Status Register.
1 = An Rx Sync has occurred since the last read of the Status Register.
• TXEN: Transmit Enable
0 = Transmit is disabled.
1 = Transmit is enabled.
• RXEN: Receive Enable
0 = Receive is disabled.
1 = Receive is enabled.
648
6438D–ATARM–13-Oct-09
AT91SAM9G45
34.9.14
Name:
SSC Interrupt Enable Register
SSC_IER
Addresses:
0xFFF9C044 (0), 0xFFFA0044 (1)
Access:
Write-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
RXSYN
10
TXSYN
9
CP1
8
CP0
7
RXBUFF
6
ENDRX
5
OVRUN
4
RXRDY
3
TXBUFE
2
ENDTX
1
TXEMPTY
0
TXRDY
• TXRDY: Transmit Ready Interrupt Enable
0 = 0 = No effect.
1 = Enables the Transmit Ready Interrupt.
• TXEMPTY: Transmit Empty Interrupt Enable
0 = No effect.
1 = Enables the Transmit Empty Interrupt.
• ENDTX: End of Transmission Interrupt Enable
0 = No effect.
1 = Enables the End of Transmission Interrupt.
• TXBUFE: Transmit Buffer Empty Interrupt Enable
0 = No effect.
1 = Enables the Transmit Buffer Empty Interrupt
• RXRDY: Receive Ready Interrupt Enable
0 = No effect.
1 = Enables the Receive Ready Interrupt.
• OVRUN: Receive Overrun Interrupt Enable
0 = No effect.
1 = Enables the Receive Overrun Interrupt.
• ENDRX: End of Reception Interrupt Enable
0 = No effect.
1 = Enables the End of Reception Interrupt.
• RXBUFF: Receive Buffer Full Interrupt Enable
649
6438D–ATARM–13-Oct-09
AT91SAM9G45
0 = No effect.
1 = Enables the Receive Buffer Full Interrupt.
• CP0: Compare 0 Interrupt Enable
0 = No effect.
1 = Enables the Compare 0 Interrupt.
• CP1: Compare 1 Interrupt Enable
0 = No effect.
1 = Enables the Compare 1 Interrupt.
• TXSYN: Tx Sync Interrupt Enable
0 = No effect.
1 = Enables the Tx Sync Interrupt.
• RXSYN: Rx Sync Interrupt Enable
0 = No effect.
1 = Enables the Rx Sync Interrupt.
650
6438D–ATARM–13-Oct-09
AT91SAM9G45
34.9.15
Name:
SSC Interrupt Disable Register
SSC_IDR
Addresses:
0xFFF9C048 (0), 0xFFFA0048 (1)
Access:
Write-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
RXSYN
10
TXSYN
9
CP1
8
CP0
7
RXBUFF
6
ENDRX
5
OVRUN
4
RXRDY
3
TXBUFE
2
ENDTX
1
TXEMPTY
0
TXRDY
• TXRDY: Transmit Ready Interrupt Disable
0 = No effect.
1 = Disables the Transmit Ready Interrupt.
• TXEMPTY: Transmit Empty Interrupt Disable
0 = No effect.
1 = Disables the Transmit Empty Interrupt.
• ENDTX: End of Transmission Interrupt Disable
0 = No effect.
1 = Disables the End of Transmission Interrupt.
• TXBUFE: Transmit Buffer Empty Interrupt Disable
0 = No effect.
1 = Disables the Transmit Buffer Empty Interrupt.
• RXRDY: Receive Ready Interrupt Disable
0 = No effect.
1 = Disables the Receive Ready Interrupt.
• OVRUN: Receive Overrun Interrupt Disable
0 = No effect.
1 = Disables the Receive Overrun Interrupt.
• ENDRX: End of Reception Interrupt Disable
0 = No effect.
1 = Disables the End of Reception Interrupt.
• RXBUFF: Receive Buffer Full Interrupt Disable
651
6438D–ATARM–13-Oct-09
AT91SAM9G45
0 = No effect.
1 = Disables the Receive Buffer Full Interrupt.
• CP0: Compare 0 Interrupt Disable
0 = No effect.
1 = Disables the Compare 0 Interrupt.
• CP1: Compare 1 Interrupt Disable
0 = No effect.
1 = Disables the Compare 1 Interrupt.
• TXSYN: Tx Sync Interrupt Enable
0 = No effect.
1 = Disables the Tx Sync Interrupt.
• RXSYN: Rx Sync Interrupt Enable
0 = No effect.
1 = Disables the Rx Sync Interrupt.
652
6438D–ATARM–13-Oct-09
AT91SAM9G45
34.9.16
Name:
SSC Interrupt Mask Register
SSC_IMR
Addresses:
0xFFF9C04C (0), 0xFFFA004C (1)
Access:
Read-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
RXSYN
10
TXSYN
9
CP1
8
CP0
7
RXBUFF
6
ENDRX
5
OVRUN
4
RXRDY
3
TXBUFE
2
ENDTX
1
TXEMPTY
0
TXRDY
• TXRDY: Transmit Ready Interrupt Mask
0 = The Transmit Ready Interrupt is disabled.
1 = The Transmit Ready Interrupt is enabled.
• TXEMPTY: Transmit Empty Interrupt Mask
0 = The Transmit Empty Interrupt is disabled.
1 = The Transmit Empty Interrupt is enabled.
• ENDTX: End of Transmission Interrupt Mask
0 = The End of Transmission Interrupt is disabled.
1 = The End of Transmission Interrupt is enabled.
• TXBUFE: Transmit Buffer Empty Interrupt Mask
0 = The Transmit Buffer Empty Interrupt is disabled.
1 = The Transmit Buffer Empty Interrupt is enabled.
• RXRDY: Receive Ready Interrupt Mask
0 = The Receive Ready Interrupt is disabled.
1 = The Receive Ready Interrupt is enabled.
• OVRUN: Receive Overrun Interrupt Mask
0 = The Receive Overrun Interrupt is disabled.
1 = The Receive Overrun Interrupt is enabled.
• ENDRX: End of Reception Interrupt Mask
0 = The End of Reception Interrupt is disabled.
1 = The End of Reception Interrupt is enabled.
653
6438D–ATARM–13-Oct-09
AT91SAM9G45
• RXBUFF: Receive Buffer Full Interrupt Mask
0 = The Receive Buffer Full Interrupt is disabled.
1 = The Receive Buffer Full Interrupt is enabled.
• CP0: Compare 0 Interrupt Mask
0 = The Compare 0 Interrupt is disabled.
1 = The Compare 0 Interrupt is enabled.
• CP1: Compare 1 Interrupt Mask
0 = The Compare 1 Interrupt is disabled.
1 = The Compare 1 Interrupt is enabled.
• TXSYN: Tx Sync Interrupt Mask
0 = The Tx Sync Interrupt is disabled.
1 = The Tx Sync Interrupt is enabled.
• RXSYN: Rx Sync Interrupt Mask
0 = The Rx Sync Interrupt is disabled.
1 = The Rx Sync Interrupt is enabled.
654
6438D–ATARM–13-Oct-09
AT91SAM9G45
35. High Speed MultiMedia Card Interface (HSMCI)
35.1
Description
The High Speed Multimedia Card Interface (HSMCI) supports the MultiMedia Card (MMC)
Specification V4.3, the SD Memory Card Specification V2.0, the SDIO V1.1 specification and
CE-ATA V1.1.
The HSMCI includes a command register, response registers, data registers, timeout counters
and error detection logic that automatically handle the transmission of commands and, when
required, the reception of the associated responses and data with a limited processor overhead.
The HSMCI supports stream, block and multi block data read and write, and is compatible with
the DMA Controller, minimizing processor intervention for large buffers transfers.
The HSMCI operates at a rate of up to Master Clock divided by 2 and supports the interfacing of
1 slot(s). Each slot may be used to interface with a High Speed MultiMediaCard bus (up to 30
Cards) or with an SD Memory Card. Only one slot can be selected at a time (slots are multiplexed). A bit field in the SD Card Register performs this selection.
The SD Memory Card communication is based on a 9-pin interface (clock, command, four data
and three power lines) and the High Speed MultiMedia Card on a 7-pin interface (clock, command, one data, three power lines and one reserved for future use).
The SD Memory Card interface also supports High Speed MultiMedia Card operations. The
main differences between SD and High Speed MultiMedia Cards are the initialization process
and the bus topology.
HSMCI fully supports CE-ATA Revision 1.1, built on the MMC System Specification v4.0. The
module includes dedicated hardware to issue the command completion signal and capture the
host command completion signal disable.
35.2
Embedded Characteristics
• Compatibility with MultiMedia Card Specification Version 4.3
• Compatibility with SD Memory Card Specification Version 2.0
• Compatibility with SDIO Specification Version V2.0.
• Compatibility with Memory Stick PRO
• Compatibility with CE ATA
655
6438D–ATARM–13-Oct-09
35.3
Block Diagram
Figure 35-1. Block Diagram
APB Bridge
DMAC
APB
MCCK
HSMCI Interface
PMC
MCK
PIO
(1)
MCCDA
(1)
MCDA0
(1)
MCDA1 (1)
MCDA2
(1)
MCDA3
(1)
MCDA4
(1)
MCDA5
(1)
MCDA6 (1)
Interrupt Control
MCDA7
(1)
HSMCI Interrupt
Note:
656
1. When several HSMCI (x HSMCI) are embedded in a product, MCCK refers to HSMCIx_CK, MCCDA to HSMCIx_CDA,
MCDAy to HSMCIx_DAy.
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.4
Application Block Diagram
Figure 35-2. Application Block Diagram
Application Layer
ex: File System, Audio, Security, etc.
Physical Layer
HSMCI Interface
1 2 3 4 5 6 7
1 2 3 4 5 6 78
9
9 1011
1213 8
MMC
35.5
SDCard
Pin Name List
Table 35-1.
I/O Lines Description
Pin Name(2)
Pin Description
Type(1)
Comments
MCCDA
Command/response
I/O/PP/OD
CMD of an MMC or SDCard/SDIO
MCCK
Clock
I/O
CLK of an MMC or SD Card/SDIO
MCDA0 - MCDA7
Data 0..7 of Slot A
I/O/PP
DAT[0..7] of an MMC
DAT[0..3] of an SD Card/SDIO
Notes:
1. I: Input, O: Output, PP: Push/Pull, OD: Open Drain.
2. When several HSMCI (x HSMCI) are embedded in a product, MCCK refers to HSMCIx_CK, MCCDA to HSMCIx_CDA,
MCDAy to HSMCIx_DAy.
657
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.6
35.6.1
Product Dependencies
I/O Lines
The pins used for interfacing the High Speed MultiMedia Cards or SD Cards are multiplexed with
PIO lines. The programmer must first program the PIO controllers to assign the peripheral functions to HSMCI pins.
Table 35-2.
I/O Lines
Instance
Signal
I/O Line
Peripheral
HSMCI0
MCI0_CDA
PA1
A
HSMCI0
MCI0_CK
PA0
A
HSMCI0
MCI0_DA0
PA2
A
HSMCI0
MCI0_DA1
PA3
A
HSMCI0
MCI0_DA2
PA4
A
HSMCI0
MCI0_DA3
PA5
A
HSMCI0
MCI0_DA4
PA6
A
HSMCI0
MCI0_DA5
PA7
A
HSMCI0
MCI0_DA6
PA8
A
HSMCI0
MCI0_DA7
PA9
A
HSMCI1
MCI1_CDA
PA22
A
HSMCI1
MCI1_CK
PA31
A
HSMCI1
MCI1_DA0
PA23
A
HSMCI1
MCI1_DA1
PA24
A
HSMCI1
MCI1_DA2
PA25
A
HSMCI1
MCI1_DA3
PA26
A
HSMCI1
MCI1_DA4
PA27
A
HSMCI1
MCI1_DA5
PA28
A
HSMCI1
MCI1_DA6
PA29
A
HSMCI1
MCI1_DA7
PA30
A
35.6.2
Power Management
The HSMCI is clocked through the Power Management Controller (PMC), so the programmer
must first configure the PMC to enable the HSMCI clock.
35.6.3
Interrupt
The HSMCI interface has an interrupt line connected to the Advanced Interrupt Controller (AIC).
Handling the HSMCI interrupt requires programming the AIC before configuring the HSMCI.
Table 35-3.
Peripheral IDs
Instance
ID
HSMCI0
11
HSMCI1
29
658
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.7
Bus Topology
Figure 35-3. High Speed MultiMedia Memory Card Bus Topology
1 2 3 4 5 6 7
9 1011
1213 8
MMC
The High Speed MultiMedia Card communication is based on a 13-pin serial bus interface. It has
three communication lines and four supply lines.
Table 35-4.
Bus Topology
Pin
Number
Name
Type(1)
Description
HSMCI Pin
Name((2)
(Slot z)
1
DAT[3]
I/O/PP
Data
MCDz3
2
CMD
I/O/PP/OD
Command/response
MCCDz
3
VSS1
S
Supply voltage ground
VSS
4
VDD
S
Supply voltage
VDD
5
CLK
I/O
Clock
MCCK
6
VSS2
S
Supply voltage ground
VSS
7
DAT[0]
I/O/PP
Data 0
MCDz0
8
DAT[1]
I/O/PP
Data 1
MCDz1
9
DAT[2]
I/O/PP
Data 2
MCDz2
10
DAT[4]
I/O/PP
Data 4
MCDz4
11
DAT[5]
I/O/PP
Data 5
MCDz5
12
DAT[6]
I/O/PP
Data 6
MCDz6
13
DAT[7]
I/O/PP
Data 7
MCDz7
Notes:
1. I: Input, O: Output, PP: Push/Pull, OD: Open Drain.
2. When several HSMCI (x HSMCI) are embedded in a product, MCCK refers to HSMCIx_CK,
MCCDA to HSMCIx_CDA, MCDAy to HSMCIx_DAy.
659
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 35-4. MMC Bus Connections (One Slot)
HSMCI
MCDA0
MCCDA
MCCK
1 2 3 4 5 6 7
1 2 3 4 5 6 7
1 2 3 4 5 6 7
9 1011
9 1011
9 1011
1213 8
MMC1
Note:
1213 8
MMC2
1213 8
MMC3
When several HSMCI (x HSMCI) are embedded in a product, MCCK refers to HSMCIx_CK, MCCDA to HSMCIx_CDA MCDAy
to HSMCIx_DAy.
Figure 35-5. SD Memory Card Bus Topology
1 2 3 4 5 6 78
9
SD CARD
The SD Memory Card bus includes the signals listed in Table 35-5.
Table 35-5.
SD Memory Card Bus Signals
Pin
Number
Name
Type(1)
Description
HSMCI Pin
Name(2)
(Slot z)
1
CD/DAT[3]
I/O/PP
Card detect/ Data line Bit 3
MCDz3
2
CMD
PP
Command/response
MCCDz
3
VSS1
S
Supply voltage ground
VSS
4
VDD
S
Supply voltage
VDD
5
CLK
I/O
Clock
MCCK
6
VSS2
S
Supply voltage ground
VSS
7
DAT[0]
I/O/PP
Data line Bit 0
MCDz0
8
DAT[1]
I/O/PP
Data line Bit 1 or Interrupt
MCDz1
9
DAT[2]
I/O/PP
Data line Bit 2
MCDz2
Notes:
1. I: input, O: output, PP: Push Pull, OD: Open Drain.
2. When several HSMCI (x HSMCI) are embedded in a product, MCCK refers to HSMCIx_CK,
MCCDA to HSMCIx_CDA, MCDAy to HSMCIx_DAy.
660
6438D–ATARM–13-Oct-09
AT91SAM9G45
MCDA0 - MCDA3
MCCK
SD CARD
9
MCCDA
1 2 3 4 5 6 78
Figure 35-6. SD Card Bus Connections with One Slot
Note:
When several HSMCI (x HSMCI) are embedded in a product, MCCK refers to HSMCIx_CK, MCCDA to HSMCIx_CDA MCDAy
to HSMCIx_DAy.
When the HSMCI is configured to operate with SD memory cards, the width of the data bus can
be selected in the HSMCI_SDCR register. Clearing the SDCBUS bit in this register means that
the width is one bit; setting it means that the width is four bits. In the case of High Speed MultiMedia cards, only the data line 0 is used. The other data lines can be used as independent
PIOs.
35.8
High Speed MultiMedia Card Operations
After a power-on reset, the cards are initialized by a special message-based High Speed MultiMedia Card bus protocol. Each message is represented by one of the following tokens:
• Command: A command is a token that starts an operation. A command is sent from the host
either to a single card (addressed command) or to all connected cards (broadcast
command). A command is transferred serially on the CMD line.
• Response: A response is a token which is sent from an addressed card or (synchronously)
from all connected cards to the host as an answer to a previously received command. A
response is transferred serially on the CMD line.
• Data: Data can be transferred from the card to the host or vice versa. Data is transferred via
the data line.
Card addressing is implemented using a session address assigned during the initialization
phase by the bus controller to all currently connected cards. Their unique CID number identifies
individual cards.
The structure of commands, responses and data blocks is described in the High Speed MultiMedia-Card System Specification. See also Table 35-6 on page 662.
High Speed MultiMediaCard bus data transfers are composed of these tokens.
There are different types of operations. Addressed operations always contain a command and a
response token. In addition, some operations have a data token; the others transfer their information directly within the command or response structure. In this case, no data token is present
in an operation. The bits on the DAT and the CMD lines are transferred synchronous to the clock
HSMCI Clock.
Two types of data transfer commands are defined:
• Sequential commands: These commands initiate a continuous data stream. They are
terminated only when a stop command follows on the CMD line. This mode reduces the
command overhead to an absolute minimum.
661
6438D–ATARM–13-Oct-09
AT91SAM9G45
• Block-oriented commands: These commands send a data block succeeded by CRC bits.
Both read and write operations allow either single or multiple block transmission. A multiple
block transmission is terminated when a stop command follows on the CMD line similarly to the
sequential read or when a multiple block transmission has a pre-defined block count (See “Data
Transfer Operation” on page 664.).
The HSMCI provides a set of registers to perform the entire range of High Speed MultiMedia
Card operations.
35.8.1
Command - Response Operation
After reset, the HSMCI is disabled and becomes valid after setting the MCIEN bit in the
HSMCI_CR Control Register.
The PWSEN bit saves power by dividing the HSMCI clock by 2PWSDIV + 1 when the bus is
inactive.
The two bits, RDPROOF and WRPROOF in the HSMCI Mode Register (HSMCI_MR) allow
stopping the HSMCI Clock during read or write access if the internal FIFO is full. This will guarantee data integrity, not bandwidth.
All the timings for High Speed MultiMedia Card are defined in the High Speed MultiMediaCard
System Specification.
The two bus modes (open drain and push/pull) needed to process all the operations are defined
in the HSMCI command register. The HSMCI_CMDR allows a command to be carried out.
For example, to perform an ALL_SEND_CID command:
NID Cycles
Host Command
CMD
S
T
Content
CRC
E
Z
******
CID
Z
S
T
Content
Z
Z
Z
The command ALL_SEND_CID and the fields and values for the HSMCI_CMDR Control Register are described in Table 35-6 and Table 35-7.
Table 35-6.
ALL_SEND_CID Command Description
CMD Index
Type
Argument
Resp
Abbreviation
CMD2
bcr
[31:0] stuff bits
R2
ALL_SEND_CID
Note:
Command
Description
Asks all cards to send
their CID numbers on
the CMD line
bcr means broadcast command with response.
Table 35-7.
Fields and Values for HSMCI_CMDR Command Register
Field
Value
CMDNB (command number)
2 (CMD2)
RSPTYP (response type)
2 (R2: 136 bits response)
SPCMD (special command)
0 (not a special command)
OPCMD (open drain command)
1
MAXLAT (max latency for command to response)
0 (NID cycles ==> 5 cycles)
662
6438D–ATARM–13-Oct-09
AT91SAM9G45
Table 35-7.
Fields and Values for HSMCI_CMDR Command Register
Field
Value
TRCMD (transfer command)
0 (No transfer)
TRDIR (transfer direction)
X (available only in transfer command)
TRTYP (transfer type)
X (available only in transfer command)
IOSPCMD (SDIO special command)
0 (not a special command)
The HSMCI_ARGR contains the argument field of the command.
To send a command, the user must perform the following steps:
• Fill the argument register (HSMCI_ARGR) with the command argument.
• Set the command register (HSMCI_CMDR) (see Table 35-7).
The command is sent immediately after writing the command register.
As soon as the command register is written, then the status bit CMDRDY in the status register
(HSMCI_SR) is cleared.
It is released and the end of the card response.
If the command requires a response, it can be read in the HSMCI response register
(HSMCI_RSPR). The response size can be from 48 bits up to 136 bits depending on the command. The HSMCI embeds an error detection to prevent any corrupted data during the transfer.
The following flowchart shows how to send a command to the card and read the response if
needed. In this example, the status register bits are polled but setting the appropriate bits in the
interrupt enable register (HSMCI_IER) allows using an interrupt method.
663
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 35-7. Command/Response Functional Flow Diagram
Set the command argument
HSMCI_ARGR = Argument(1)
Set the command
HSMCI_CMDR = Command
Read HSMCI_SR
Wait for command
ready status flag
0
CMDRDY
1
Check error bits in the
status register (1)
Yes
Status error flags?
Read response if required
RETURN ERROR(1)
RETURN OK
Note:
35.8.2
1. If the command is SEND_OP_COND, the CRC error flag is always present (refer to R3 response in the High Speed MultiMedia Card specification).
Data Transfer Operation
The High Speed MultiMedia Card allows several read/write operations (single block, multiple
blocks, stream, etc.). These kinds of transfer can be selected setting the Transfer Type (TRTYP)
field in the HSMCI Command Register (HSMCI_CMDR).
These operations can be done using the features of the DMA Controller.
In all cases, the block length (BLKLEN field) must be defined either in the mode register
HSMCI_MR, or in the Block Register HSMCI_BLKR. This field determines the size of the data
block.
Consequent to MMC Specification 3.1, two types of multiple block read (or write) transactions
are defined (the host can use either one at any time):
• Open-ended/Infinite Multiple block read (or write):
664
6438D–ATARM–13-Oct-09
AT91SAM9G45
The number of blocks for the read (or write) multiple block operation is not defined. The card
will continuously transfer (or program) data blocks until a stop transmission command is
received.
• Multiple block read (or write) with pre-defined block count (since version 3.1 and higher):
The card will transfer (or program) the requested number of data blocks and terminate the
transaction. The stop command is not required at the end of this type of multiple block read
(or write), unless terminated with an error. In order to start a multiple block read (or write)
with pre-defined block count, the host must correctly program the HSMCI Block Register
(HSMCI_BLKR). Otherwise the card will start an open-ended multiple block read. The BCNT
field of the Block Register defines the number of blocks to transfer (from 1 to 65535 blocks).
Programming the value 0 in the BCNT field corresponds to an infinite block transfer.
35.8.3
Read Operation
The following flowchart (Figure 35-8) shows how to read a single block with or without use of
DMAC facilities. In this example, a polling method is used to wait for the end of read. Similarly,
the user can configure the interrupt enable register (HSMCI_IER) to trigger an interrupt at the
end of read.
665
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 35-8. Read Functional Flow Diagram
Send SELECT/DESELECT_CARD
(1)
command
to select the card
Send SET_BLOCKLEN command(1)
No
Yes
Read with DMAC
Reset the DMAEN bit
MCI_DMA &= ~DMAEN
Set the block length (in bytes)
HSMCI_MR l= (BlockLength<<16) (2)
Set the block count (if neccessary)
HSMCI_BLKR l= (BlockCount<<0)
Set the DMAEN bit
HSMCI_DMA |= DMAEN
Set the block length (in bytes)
(2)
HSMCI_BLKR |= (BlockLength << 16)
Configure the DMA channel X
DMAC_SADDRX = Data Address
DMAC_BTSIZE = BlockLength/4
DMACHEN[X] = TRUE
Send READ_SINGLE_BLOCK
command(1)
Number of words to read = BlockLength/4
Send READ_SINGLE_BLOCK
command(1)
Yes
Number of words to read = 0 ?
Read status register HSMCI_SR
No
Read status register HSMCI_SR
Poll the bit
XFRDONE = 0?
Poll the bit
RXRDY = 0?
Yes
Yes
No
No
RETURN
Read data = HSMCI_RDR
Number of words to read =
Number of words to read -1
RETURN
Note:
1. It is assumed that this command has been correctly sent (see Figure 35-7).
2. This field is also accessible in the HSMCI Block Register (HSMCI_BLKR).
666
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.8.4
Write Operation
In write operation, the HSMCI Mode Register (HSMCI_MR) is used to define the padding value
when writing non-multiple block size. If the bit PADV is 0, then 0x00 value is used when padding
data, otherwise 0xFF is used.
If set, the bit DMAEN in the HSMCI_DMA register enables DMA transfer.
The following flowchart (Figure 35-9) shows how to write a single block with or without use of
DMA facilities. Polling or interrupt method can be used to wait for the end of write according to
the contents of the Interrupt Mask Register (HSMCI_IMR).
667
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 35-9. Write Functional Flow Diagram
Send SELECT/DESELECT_CARD
command(1) to select the card
Send SET_BLOCKLEN command(1)
Yes
No
Write using DMAC
Reset theDMAEN bit
HSMCI_DMA &= ~DMAEN
Set the block length (in bytes)
HSMCI_MR |= (BlockLength) <<16)(2)
Set the block count (if necessary)
HSMCI_BLKR |= (BlockCount << 0)
Set the DMAEN bit
HSMCI_DMA |= DMAEN
Set the block length (in bytes)
HSMCI_BLKR |= (BlockLength << 16)(2)
Send WRITE_SINGLE_BLOCK
command(1)
Send WRITE_SINGLE_BLOCK
command(1)
Configure the DMA channel X
DMAC_DADDRX = Data Address to write
DMAC_BTSIZE = BlockLength/4
Number of words to write = BlockLength/4
DMAC_CHEN[X] = TRUE
Yes
Number of words to write = 0 ?
Read status register MCI_SR
No
Read status register HSMCI_SR
Poll the bit
XFRDONE = 0?
Poll the bit
TXRDY = 0?
Yes
Yes
No
No
RETURN
HSMCI_TDR = Data to write
Number of words to write =
Number of words to write -1
RETURN
Note:
1. It is assumed that this command has been correctly sent (see Figure 35-7).
2. This field is also accessible in the HSMCI Block Register (HSMCI_BLKR).
668
6438D–ATARM–13-Oct-09
AT91SAM9G45
The following flowchart (Figure 35-10) shows how to manage read multiple block and write multiple block transfers with the DMA Controller. Polling or interrupt method can be used to wait for
the end of write according to the contents of the Interrupt Mask Register (HSMCI_IMR).
Figure 35-10. Read Multiple Block and Write Multiple Block
Send SELECT/DESELECT_CARD
command(1) to select the card
Send SET_BLOCKLEN command(1)
Set the block length
HSMCI_MR |= (BlockLength << 16)
Set the DMAEN bit
HSMCI_DMA |= DMAEN
Send WRITE_MULTIPLE_BLOCK or
READ_MULTIPLE_BLOCK command(1)
Configure the HDMA channel X
DMAC_SADDRX and DMAC_DADDRX
DMAC_BTSIZE = BlockLength/4
DMAC_CHEN[X] = TRUE
Read status register DMAC_EBCISR
and Poll Bit CBTC[X]
New Buffer ?(2)
Yes
No
Read status register HSMCI_SR
and Poll Bit FIFOEMPTY
Send STOP_TRANSMISSION
(1)
command
Poll the bit
XFRDONE = 1
No
Yes
RETURN
Notes:
1. It is assumed that this command has been correctly sent (see Figure 35-7).
2. Handle errors reported in HSMCI_SR.
669
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.8.5
WRITE_SINGLE_BLOCK Operation using DMA Controller
1. Wait until the current command execution has successfully terminated.
a. Check that CMDRDY and NOTBUSY fields are asserted in HSMCI_SR
2. Program the block length in the card. This value defines the value block_length.
3. Program the block length in the HSMCI configuration register with block_length value.
4. Program HSMCI_DMA register with the following fields:
– OFFSET field with dma_offset.
– CHKSIZE is user defined and set according to DMAC_DCSIZE.
– DMAEN is set to true to enable DMA hardware handshaking in the HSMCI. This bit
was previously set to false.
5. Issue a WRITE_SINGLE_BLOCK command writing HSMCI_ARG then HSMCI_CMDR.
6. Program the DMA Controller.
a. Read the channel Register to choose an available (disabled) channel.
b.
Clear any pending interrupts on the channel from the previous DMAC transfer by
reading the DMAC_EBCISR register.
c.
Program the channel registers.
d. The DMAC_SADDRx register for channel x must be set to the location of the
source data. When the first data location is not word aligned, the two LSB bits
define the temporary value called dma_offset. The two LSB bits of
DMAC_SADDRx must be set to 0.
e. The DMAC_DADDRx register for channel x must be set with the starting address of
the HSMCI_FIFO address.
f.
Program DMAC_CTRLAx register of channel x with the following field’s values:
–DST_WIDTH is set to WORD.
–SRC_WIDTH is set to WORD.
–DCSIZE must be set according to the value of HSMCI_DMA, CHKSIZE field.
–BTSIZE is programmed with CEILING((block_length + dma_offset) / 4), where
the ceiling function is the function that returns the smallest integer not less than
x.
g. Program DMAC_CTRLBx register for channel x with the following field’s values:
–DST_INCR is set to INCR, the block_length value must not be larger than the
HSMCI_FIFO aperture.
–SRC_INCR is set to INCR.
–FC field is programmed with memory to peripheral flow control mode.
–both DST_DSCR and SRC_DSCR are set to 1 (descriptor fetch is disabled).
–DIF and SIF are set with their respective layer ID. If SIF is different from DIF, the
DMA controller is able to prefetch data and write HSMCI simultaneously.
h. Program DMAC_CFGx register for channel x with the following field’s values:
–FIFOCFG defines the watermark of the DMAC channel FIFO.
–DST_H2SEL is set to true to enable hardware handshaking on the destination.
–DST_PER is programmed with the hardware handshaking ID of the targeted
HSMCI Host Controller.
670
6438D–ATARM–13-Oct-09
AT91SAM9G45
i.
Enable Channel x, writing one to DMAC_CHER[x]. The DMAC is ready and waiting
for request.
7. Wait for XFRDONE in HSMCI_SR register.
35.8.6
35.8.6.1
READ_SINGLE_BLOCK Operation using DMA Controller
Block Length is Multiple of 4
1. Wait until the current command execution has successfully completed.
a. Check that CMDRDY and NOTBUSY are asserted in HSMCI_SR.
2. Program the block length in the card. This value defines the value block_length.
3. Program the block length in the HSMCI configuration register with block_length value.
4. Set RDPROOF bit in HSMCI_MR to avoid overflow.
5. Program HSMCI_DMA register with the following fields:
– ROPT field is set to 0.
– OFFSET field is set to 0.
– CHKSIZE is user defined.
– DMAEN is set to true to enable DMAC hardware handshaking in the HSMCI. This bit
was previously set to false.
6. Issue a READ_SINGLE_BLOCK command.
7. Program the DMA controller.
a. Read the channel Register to choose an available (disabled) channel.
b.
Clear any pending interrupts on the channel from the previous DMA transfer by
reading the DMAC_EBCISR register.
c.
Program the channel registers.
d. The DMAC_SADDRx register for channel x must be set with the starting address of
the HSMCI_FIFO address.
e. The DMAC_DADDRx register for channel x must be word aligned.
f.
Program DMAC_CTRLAx register of channel x with the following field’s values:
–DST_WIDTH is set to WORD.
–SRC_WIDTH is set to WORD.
–SCSIZE must be set according to the value of HSMCI_DMA, CHKSIZE field.
–BTSIZE is programmed with block_length/4.
g. Program DMAC_CTRLBx register for channel x with the following field’s values:
– DST_INCR is set to INCR.
– SRC_INCR is set to INCR.
– FC field is programmed with peripheral to memory flow control mode.
– both DST_DSCR and SRC_DSCR are set to 1 (descriptor fetch is disabled).
– DIF and SIF are set with their respective layer ID. If SIF is different from DIF, the
DMA controller is able to prefetch data and write HSMCI simultaneously.
h. Program DMAC_CFGx register for channel x with the following field’s values:
–FIFOCFG defines the watermark of the DMA channel FIFO.
–SRC_H2SEL is set to true to enable hardware handshaking on the destination.
–SRC_PER is programmed with the hardware handshaking ID of the targeted
HSMCI Host Controller.
671
6438D–ATARM–13-Oct-09
AT91SAM9G45
–Enable Channel x, writing one to DMAC_CHER[x]. The DMAC is ready and
waiting for request.
8. Wait for XFRDONE in HSMCI_SR register.
35.8.6.2
Block Length is Not Multiple of 4 and Padding Not Used (ROPT field in HSMCI_DMA register set to 0)
In the previous DMA transfer flow (block length multiple of 4), the DMA controller is configured to
use only WORD AHB access. When the block length is no longer a multiple of 4 this is no longer
true. The DMA controller is programmed to copy exactly the block length number of bytes using
2 transfer descriptors.
1. Use the previous step until READ_SINGLE_BLOCK then
2. Program the DMA controller to use a two descriptors linked list.
a. Read the channel Register to choose an available (disabled) channel.
b.
Clear any pending interrupts on the channel from the previous DMA transfer by
reading the DMAC_EBCISR register.
c.
Program the channel registers in the Memory for the first descriptor. This descriptor
will be word oriented. This descriptor is referred to as LLI_W, standing for LLI word
oriented transfer.
d. The LLI_W.DMAC_SADDRx field in memory must be set with the starting address
of the HSMCI_FIFO address.
e. The LLI_W.DMAC_DADDRx field in the memory must be word aligned.
f.
Program LLI_W.DMAC_CTRLAx with the following field’s values:
–DST_WIDTH is set to WORD.
–SRC_WIDTH is set to WORD.
–SCSIZE must be set according to the value of HSMCI_DMA, CHKSIZE field.
–BTSIZE is programmed with block_length/4. If BTSIZE is zero, this descriptor is
skipped later.
g. Program LLI_W.DMAC_CTRLBx with the following field’s values:
–DST_INCR is set to INCR
–SRC_INCR is set to INCR
–FC field is programmed with peripheral to memory flow control mode.
–SRC_DSCR is set to zero. (descriptor fetch is enabled for the SRC)
–DST_DSCR is set to one. (descriptor fetch is disabled for the DST)
–DIF and SIF are set with their respective layer ID. If SIF is different from DIF, DMA
controller is able to prefetch data and write HSMCI simultaneously.
h. Program LLI_W.DMAC_CFGx register for channel x with the following field’s
values:
–FIFOCFG defines the watermark of the DMA channel FIFO.
–DST_REP is set to zero meaning that address are contiguous.
–SRC_H2SEL is set to true to enable hardware handshaking on the destination.
–SRC_PER is programmed with the hardware handshaking ID of the targeted
HSMCI Host Controller.
i.
Program LLI_W.DMAC_DSCRx with the address of LLI_B descriptor. And set
DSCRx_IF to the AHB Layer ID. This operation actually links the Word oriented
672
6438D–ATARM–13-Oct-09
AT91SAM9G45
descriptor on the second byte oriented descriptor. When block_length[1:0] is equal
to 0 (multiple of 4) LLI_W.DMAC_DSCRx points to 0, only LLI_W is relevant.
j.
Program the channel registers in the Memory for the second descriptor. This
descriptor will be byte oriented. This descriptor is referred to as LLI_B, standing for
LLI Byte oriented.
k.
The LLI_B.DMAC_SADDRx field in memory must be set with the starting address
of the HSMCI_FIFO address.
l.
The LLI_B.DMAC_DADDRx is not relevant if previous word aligned descriptor was
enabled. If 1, 2 or 3 bytes are transferred that address is user defined and not word
aligned.
m. Program LLI_B.DMAC_CTRLAx with the following field’s values:
–DST_WIDTH is set to BYTE.
–SRC_WIDTH is set to BYTE.
–SCSIZE must be set according to the value of HSMCI_DMA, CHKSIZE field.
–BTSIZE is programmed with block_length[1:0]. (last 1, 2, or 3 bytes of the buffer).
n. Program LLI_B.DMAC_CTRLBx with the following field’s values:
–DST_INCR is set to INCR
–SRC_INCR is set to INCR
–FC field is programmed with peripheral to memory flow control mode.
–Both SRC_DSCR and DST_DSCR are set to 1 (descriptor fetch is disabled) or
Next descriptor location points to 0.
–DIF and SIF are set with their respective layer ID. If SIF is different from DIF, DMA
Controller is able to prefetch data and write HSMCI simultaneously.
o.
Program LLI_B.DMAC_CFGx memory location for channel x with the following
field’s values:
– FIFOCFG defines the watermark of the DMA channel FIFO.
– SRC_H2SEL is set to true to enable hardware handshaking on the destination.
– SRC_PER is programmed with the hardware handshaking ID of the targeted HSMCI
Host Controller.
p.
Program LLI_B.DMAC_DSCR with 0.
q. Program DMAC_CTRLBx register for channel x with 0. its content is updated with
the LLI fetch operation.
r.
Program DMAC_DSCRx with the address of LLI_W if block_length greater than 4
else with address of LLI_B.
s.
Enable Channel x writing one to DMAC_CHER[x]. The DMAC is ready and waiting
for request.
3. Wait for XFRDONE in HSMCI_SR register.
35.8.6.3
Block Length is Not Multiple of 4, with Padding Value (ROPT field in HSMCI_DMA register set to 1)
When the ROPT field is set to one, The DMA Controller performs only WORD access on the bus
to transfer a non-multiple of 4 block length. Unlike previous flow, in which the transfer size is
rounded to the nearest multiple of 4.
1. Program the HSMCI Interface, see previous flow.
– ROPT field is set to 1.
2. Program the DMA Controller
673
6438D–ATARM–13-Oct-09
AT91SAM9G45
a. Read the channel Register to choose an available (disabled) channel.
b.
Clear any pending interrupts on the channel from the previous DMA transfer by
reading the DMAC_EBCISR register.
c.
Program the channel registers.
d. The DMAC_SADDRx register for channel x must be set with the starting address of
the HSMCI_FIFO address.
e. The DMAC_DADDRx register for channel x must be word aligned.
f.
Program DMAC_CTRLAx register of channel x with the following field’s values:
–DST_WIDTH is set to WORD
–SRC_WIDTH is set to WORD
–SCSIZE must be set according to the value of HSMCI_DMA.CHKSIZE Field.
–BTSIZE is programmed with CEILING(block_length/4).
g. Program DMAC_CTRLBx register for channel x with the following field’s values:
–DST_INCR is set to INCR
–SRC_INCR is set to INCR
–FC field is programmed with peripheral to memory flow control mode.
–both DST_DSCR and SRC_DSCR are set to 1. (descriptor fetch is disabled)
–DIF and SIF are set with their respective layer ID. If SIF is different from DIF, the
DMA Controller is able to prefetch data and write HSMCI simultaneously.
h. Program DMAC_CFGx register for channel x with the following field’s values:
–FIFOCFG defines the watermark of the DMA channel FIFO.
–SRC_H2SEL is set to true to enable hardware handshaking on the destination.
–SRC_PER is programmed with the hardware handshaking ID of the targeted
HSMCI Host Controller.
–Enable Channel x writing one to DMAC_CHER[x]. The DMAC is ready and
waiting for request.
3. Wait for XFRDONE in HSMCI_SR register.
35.8.7
35.8.7.1
WRITE_MULTIPLE_BLOCK
One Block per Descriptor
1. Wait until the current command execution has successfully terminated.
a. Check that CMDRDY and NOTBUSY are asserted in HSMCI_SR.
2. Program the block length in the card. This value defines the value block_length.
3. Program the block length in the HSMCI configuration register with block_length value.
4. Program HSMCI_DMA register with the following fields:
– OFFSET field with dma_offset.
– CHKSIZE is user defined.
– DMAEN is set to true to enable DMAC hardware handshaking in the HSMCI. This bit
was previously set to false.
5. Issue a WRITE_MULTIPLE_BLOCK command.
6. Program the DMA Controller to use a list of descriptors. Each descriptor transfers one
block of data. Block n of data is transferred with descriptor LLI(n).
674
6438D–ATARM–13-Oct-09
AT91SAM9G45
a. Read the channel Register to choose an available (disabled) channel.
b.
Clear any pending interrupts on the channel from the previous DMAC transfer by
reading the DMAC_EBCISR register.
c.
Program a List of descriptors.
d. The LLI(n).DMAC_SADDRx memory location for channel x must be set to the location of the source data. When the first data location is not word aligned, the two
LSB bits define the temporary value called dma_offset. The two LSB bits of
LLI(n).DMAC_SADDRx must be set to 0.
e. The LLI(n).DMAC_DADDRx register for channel x must be set with the starting
address of the HSMCI_FIFO address.
f.
Program LLI(n).DMAC_CTRLAx register of channel x with the following field’s
values:
–DST_WIDTH is set to WORD.
–SRC_WIDTH is set to WORD.
–DCSIZE must be set according to the value of HSMCI_DMA, CHKSIZE field.
–BTSIZE is programmed with CEILING((block_length + dma_offset)/4).
g. Program LLI(n).DMAC_CTRLBx register for channel x with the following field’s
values:
–DST_INCR is set to INCR.
–SRC_INCR is set to INCR.
–DST_DSCR is set to 0 (fetch operation is enabled for the destination).
–SRC_DSCR is set to 1 (source address is contiguous).
–FC field is programmed with memory to peripheral flow control mode.
–Both DST_DSCR and SRC_DSCR are set to 1 (descriptor fetch is disabled).
–DIF and SIF are set with their respective layer ID. If SIF is different from DIF, DMA
Controller is able to prefetch data and write HSMCI simultaneously.
h. Program LLI(n).DMAC_CFGx register for channel x with the following field’s values:
–FIFOCFG defines the watermark of the DMA channel FIFO.
–DST_H2SEL is set to true to enable hardware handshaking on the destination.
–SRC_REP is set to 0. (contiguous memory access at block boundary)
–DST_PER is programmed with the hardware handshaking ID of the targeted
HSMCI Host Controller.
i.
If LLI(n) is the last descriptor, then LLI(n).DSCR points to 0 else LLI(n) points to the
start address of LLI(n+1).
j.
Program DMAC_CTRLBx for channel register x with 0. Its content is updated with
the LLI fetch operation.
k.
Program DMAC_DSCRx for channel register x with the address of the first descriptor LLI(0).
l.
Enable Channel x writing one to DMAC_CHER[x]. The DMA is ready and waiting
for request.
7. Poll CBTC[x] bit in the DMAC_EBCISR Register.
8. If a new list of buffers shall be transferred, repeat step 6. Check and handle HSMCI
errors.
9. Poll FIFOEMPTY field in the HSMCI_SR.
675
6438D–ATARM–13-Oct-09
AT91SAM9G45
10. Send The STOP_TRANSMISSION command writing HSMCI_ARG then
HSMCI_CMDR.
11. Wait for XFRDONE in HSMCI_SR register.
35.8.8
35.8.8.1
READ_MULTIPLE_BLOCK
Block Length is a Multiple of 4
1. Wait until the current command execution has successfully terminated.
a. Check that CMDRDY and NOTBUSY are asserted in HSMCI_SR.
2. Program the block length in the card. This value defines the value block_length.
3. Program the block length in the HSMCI configuration register with block_length value.
4. Set RDPROOF bit in HSMCI_MR to avoid overflow.
5. Program HSMCI_DMA register with the following fields:
– ROPT field is set to 0.
– OFFSET field is set to 0.
– CHKSIZE is user defined.
– DMAEN is set to true to enable DMAC hardware handshaking in the HSMCI. This bit
was previously set to false.
6. Issue a READ_MULTIPLE_BLOCK command.
7. Program the DMA Controller to use a list of descriptors:
a. Read the channel Register to choose an available (disabled) channel.
b.
Clear any pending interrupts on the channel from the previous DMA transfer by
reading the DMAC_EBCISR register.
c.
Program the channel registers in the Memory with the first descriptor. This descriptor will be word oriented. This descriptor is referred to as LLI_W(n), standing for LLI
word oriented transfer for block n.
d. The LLI_W(n).DMAC_SADDRx field in memory must be set with the starting
address of the HSMCI_FIFO address.
e. The LLI_W(n).DMAC_DADDRx field in the memory must be word aligned.
f.
Program LLI_W(n).DMAC_CTRLAx with the following field’s values:
–DST_WIDTH is set to WORD
–SRC_WIDTH is set to WORD
–SCSIZE must be set according to the value of HSMCI_DMA, CHKSIZE field.
–BTSIZE is programmed with block_length/4.
g. Program LLI_W(n).DMAC_CTRLBx with the following field’s values:
–DST_INCR is set to INCR.
–SRC_INCR is set to INCR.
–FC field is programmed with peripheral to memory flow control mode.
–SRC_DSCR is set to 0 (descriptor fetch is enabled for the SRC).
–DST_DSCR is set to TRUE (descriptor fetch is disabled for the DST).
–DIF and SIF are set with their respective layer ID. If SIF is different from DIF, the
DMA Controller is able to prefetch data and write HSMCI simultaneously.
676
6438D–ATARM–13-Oct-09
AT91SAM9G45
h. Program LLI_W(n).DMAC_CFGx register for channel x with the following field’s
values:
–FIFOCFG defines the watermark of the DMA channel FIFO.
–DST_REP is set to zero. Addresses are contiguous.
–SRC_H2SEL is set to true to enable hardware handshaking on the destination.
–SRC_PER is programmed with the hardware handshaking ID of the targeted
HSMCI Host Controller.
i.
Program LLI_W(n).DMAC_DSCRx with the address of LLI_W(n+1) descriptor. And
set the DSCRx_IF to the AHB Layer ID. This operation actually links descriptors
together. If LLI_W(n) is the last descriptor then LLI_W(n).DMAC_DSCRx points to
0.
j.
Program DMAC_CTRLBx register for channel x with 0. its content is updated with
the LLI Fetch operation.
k.
Program DMAC_DSCRx register for channel x with the address of LLI_W(0).
l.
Enable Channel x writing one to DMAC_CHER[x]. The DMA is ready and waiting
for request.
8. Poll CBTC[x] bit in the DMAC_EBCISR Register.
9. If a new list of buffer shall be transferred repeat step 6. Check and handle HSMCI
errors.
10. Poll FIFOEMPTY field in the HSMCI_SR.
11. Send The STOP_TRANSMISSION command writing the HSMCI_ARG then the
HSMCI_CMDR.
12. Wait for XFRDONE in HSMCI_SR register.
35.8.8.2
Block Length is Not Multiple of 4. (ROPT field in HSMCI_DMA register set to 0)
Two DMA Transfer descriptors are used to perform the HSMCI block transfer.
1. Use the previous step to configure the HSMCI to perform a READ_MULTIPLE_BLOCK
command.
2. Issue a READ_MULTIPLE_BLOCK command.
3. Program the DMA Controller to use a list of descriptors.
a. Read the channel register to choose an available (disabled) channel.
b.
Clear any pending interrupts on the channel from the previous DMAC transfer by
reading the DMAC_EBCISR register.
c.
For every block of data repeat the following procedure:
d. Program the channel registers in the Memory for the first descriptor. This descriptor
will be word oriented. This descriptor is referred to as LLI_W(n) standing for LLI
word oriented transfer for block n.
e. The LLI_W(n).DMAC_SADDRx field in memory must be set with the starting
address of the HSMCI_FIFO address.
f.
The LLI_W(n).DMAC_DADDRx field in the memory must be word aligned.
g. Program LLI_W(n).DMAC_CTRLAx with the following field’s values:
–DST_WIDTH is set to WORD.
–SRC_WIDTH is set to WORD.
–SCSIZE must be set according to the value of HSMCI_DMA, CHKSIZE field.
–BTSIZE is programmed with block_length/4. If BTSIZE is zero, this descriptor is
skipped later.
677
6438D–ATARM–13-Oct-09
AT91SAM9G45
h. Program LLI_W(n).DMAC_CTRLBx with the following field’s values:
–DST_INCR is set to INCR.
–SRC_INCR is set to INCR.
–FC field is programmed with peripheral to memory flow control mode.
–SRC_DSCR is set to 0 (descriptor fetch is enabled for the SRC).
–DST_DSCR is set to TRUE (descriptor fetch is disabled for the DST).
–DIF and SIF are set with their respective layer ID. If SIF is different from DIF, the
DMA Controller is able to prefetch data and write HSMCI simultaneously.
i.
Program LLI_W(n).DMAC_CFGx register for channel x with the following field’s
values:
–FIFOCFG defines the watermark of the DMA channel FIFO.
–DST_REP is set to zero. Address are contiguous.
–SRC_H2SEL is set to true to enable hardware handshaking on the destination.
–SRC_PER is programmed with the hardware handshaking ID of the targeted
HSMCI Host Controller.
j.
Program LLI_W(n).DMAC_DSCRx with the address of LLI_B(n) descriptor. And set
the DSCRx_IF to the AHB Layer ID. This operation actually links the Word oriented
descriptor on the second byte oriented descriptor. When block_length[1:0] is equal
to 0 (multiple of 4) LLI_W(n).DMAC_DSCRx points to 0, only LLI_W(n) is relevant.
k.
Program the channel registers in the Memory for the second descriptor. This
descriptor will be byte oriented. This descriptor is referred to as LLI_B(n), standing
for LLI Byte oriented.
l.
The LLI_B(n).DMAC_SADDRx field in memory must be set with the starting
address of the HSMCI_FIFO address.
m. The LLI_B(n).DMAC_DADDRx is not relevant if previous word aligned descriptor
was enabled. If 1, 2 or 3 bytes are transferred, that address is user defined and not
word aligned.
n. Program LLI_B(n).DMAC_CTRLAx with the following field’s values:
–DST_WIDTH is set to BYTE.
–SRC_WIDTH is set to BYTE.
–SCSIZE must be set according to the value of HSMCI_DMA, CHKSIZE field.
–BTSIZE is programmed with block_length[1:0]. (last 1, 2, or 3 bytes of the buffer).
o.
Program LLI_B(n).DMAC_CTRLBx with the following field’s values:
– DST_INCR is set to INCR.
– SRC_INCR is set to INCR.
– FC field is programmed with peripheral to memory flow control mode.
– Both SRC_DSCR and DST_DSCR are set to 1 (descriptor fetch is disabled) or Next
descriptor location points to 0.
– DIF and SIF are set with their respective layer ID. If SIF is different from DIF, the
DMA Controller is able to prefetch data and write HSMCI simultaneously.
p.
Program LLI_B(n).DMAC_CFGx memory location for channel x with the following
field’s values:
– FIFOCFG defines the watermark of the DMAC channel FIFO.
– SRC_H2SEL is set to true to enable hardware handshaking on the destination.
678
6438D–ATARM–13-Oct-09
AT91SAM9G45
– SRC_PER is programmed with the hardware handshaking ID of the targeted HSMCI
Host Controller
q. Program LLI_B(n).DMAC_DSCR with address of descriptor LLI_W(n+1). If
LLI_B(n) is the last descriptor, then program LLI_B(n).DMAC_DSCR with 0.
r.
Program DMAC_CTRLBx register for channel x with 0, its content is updated with
the LLI Fetch operation.
s.
Program DMAC_DSCRx with the address of LLI_W(0) if block_length is greater
than 4 else with address of LLI_B(0).
t.
Enable Channel x writing one to DMAC_CHER[x]. The DMAC is ready and waiting
for request.
4. Enable DMADONE interrupt in the HSMCI_IER register.
5. Poll CBTC[x] bit in the DMAC_EBCISR Register.
6. If a new list of buffers shall be transferred, repeat step 7. Check and handle HSMCI
errors.
7. Poll FIFOEMPTY field in the HSMCI_SR.
8. Send The STOP_TRANSMISSION command writing HSMCI_ARG then
HSMCI_CMDR.
9. Wait for XFRDONE in HSMCI_SR register.
35.8.8.3
Block Length is Not a Multiple of 4. (ROPT field in HSMCI_DMA register set to 1)
One DMA Transfer descriptor is used to perform the HSMCI block transfer, the DMA writes a
rounded up value to the nearest multiple of 4.
1. Use the previous step to configure the HSMCI to perform a READ_MULTIPLE_BLOCK.
2. Set the ROPT field to 1 in the HSMCI_DMA register.
3. Issue a READ_MULTIPLE_BLOCK command.
4. Program the DMA controller to use a list of descriptors:
a. Read the channel Register to choose an available (disabled) channel.
b.
Clear any pending interrupts on the channel from the previous DMAC transfer by
reading the DMAC_EBCISR register.
c.
Program the channel registers in the Memory with the first descriptor. This descriptor will be word oriented. This descriptor is referred to as LLI_W(n), standing for LLI
word oriented transfer for block n.
d. The LLI_W(n).DMAC_SADDRx field in memory must be set with the starting
address of the HSMCI_FIFO address.
e. The LLI_W(n).DMAC_DADDRx field in the memory must be word aligned.
f.
Program LLI_W(n).DMAC_CTRLAx with the following field’s values:
–DST_WIDTH is set to WORD.
–SRC_WIDTH is set to WORD.
–SCSIZE must be set according to the value of HSMCI_DMA, CHKSIZE field.
–BTSIZE is programmed with Ceiling(block_length/4).
g. Program LLI_W(n).DMAC_CTRLBx with the following field’s values:
–DST_INCR is set to INCR
–SRC_INCR is set to INCR
–FC field is programmed with peripheral to memory flow control mode.
–SRC_DSCR is set to 0. (descriptor fetch is enabled for the SRC)
679
6438D–ATARM–13-Oct-09
AT91SAM9G45
–DST_DSCR is set to TRUE. (descriptor fetch is disabled for the DST)
–DIF and SIF are set with their respective layer ID. If SIF is different from DIF, the
DMA Controller is able to prefetch data and write HSMCI simultaneously.
h. Program LLI_W(n).DMAC_CFGx register for channel x with the following field’s
values:
–FIFOCFG defines the watermark of the DMA channel FIFO.
–DST_REP is set to zero. Address are contiguous.
–SRC_H2SEL is set to true to enable hardware handshaking on the destination.
–SRC_PER is programmed with the hardware handshaking ID of the targeted
HSMCI Host Controller.
i.
Program LLI_W(n).DMAC_DSCRx with the address of LLI_W(n+1) descriptor. And
set the DSCRx_IF to the AHB Layer ID. This operation actually links descriptors
together. If LLI_W(n) is the last descriptor then LLI_W(n).DMAC_DSCRx points to
0.
j.
Program DMAC_CTRLBx register for channel x with 0. its content is updated with
the LLI Fetch operation.
k.
Program DMAC_DSCRx register for channel x with the address of LLI_W(0).
l.
Enable Channel x writing one to DMAC_CHER[x]. The DMAC is ready and waiting
for request.
5. Poll CBTC[x] bit in the DMAC_EBCISR Register.
6. If a new list of buffers shall be transferred repeat step 7. Check and handle HSMCI
errors.
7. Poll FIFOEMPTY field in the HSMCI_SR.
8. Send The STOP_TRANSMISSION command writing the HSMCI_ARG then the
HSMCI_CMDR.
9. Wait for XFRDONE in HSMCI_SR register.
35.9
SD/SDIO Card Operation
The High Speed MultiMedia Card Interface allows processing of SD Memory (Secure Digital
Memory Card) and SDIO (SD Input Output) Card commands.
SD/SDIO cards are based on the Multi Media Card (MMC) format, but are physically slightly
thicker and feature higher data transfer rates, a lock switch on the side to prevent accidental
overwriting and security features. The physical form factor, pin assignment and data transfer
protocol are forward-compatible with the High Speed MultiMedia Card with some additions. SD
slots can actually be used for more than flash memory cards. Devices that support SDIO can
use small devices designed for the SD form factor, such as GPS receivers, Wi-Fi or Bluetooth
adapters, modems, barcode readers, IrDA adapters, FM radio tuners, RFID readers, digital cameras and more.
SD/SDIO is covered by numerous patents and trademarks, and licensing is only available
through the Secure Digital Card Association.
The SD/SDIO Card communication is based on a 9-pin interface (Clock, Command, 4 x Data
and 3 x Power lines). The communication protocol is defined as a part of this specification. The
main difference between the SD/SDIO Card and the High Speed MultiMedia Card is the initialization process.
680
6438D–ATARM–13-Oct-09
AT91SAM9G45
The SD/SDIO Card Register (HSMCI_SDCR) allows selection of the Card Slot and the data bus
width.
The SD/SDIO Card bus allows dynamic configuration of the number of data lines. After power
up, by default, the SD/SDIO Card uses only DAT0 for data transfer. After initialization, the host
can change the bus width (number of active data lines).
35.9.1
SDIO Data Transfer Type
SDIO cards may transfer data in either a multi-byte (1 to 512 bytes) or an optional block format
(1 to 511 blocks), while the SD memory cards are fixed in the block transfer mode. The TRTYP
field in the HSMCI Command Register (HSMCI_CMDR) allows to choose between SDIO Byte or
SDIO Block transfer.
The number of bytes/blocks to transfer is set through the BCNT field in the HSMCI Block Register (HSMCI_BLKR). In SDIO Block mode, the field BLKLEN must be set to the data block size
while this field is not used in SDIO Byte mode.
An SDIO Card can have multiple I/O or combined I/O and memory (called Combo Card). Within
a multi-function SDIO or a Combo card, there are multiple devices (I/O and memory) that share
access to the SD bus. In order to allow the sharing of access to the host among multiple devices,
SDIO and combo cards can implement the optional concept of suspend/resume (Refer to the
SDIO Specification for more details). To send a suspend or a resume command, the host must
set the SDIO Special Command field (IOSPCMD) in the HSMCI Command Register.
35.9.2
SDIO Interrupts
Each function within an SDIO or Combo card may implement interrupts (Refer to the SDIO
Specification for more details). In order to allow the SDIO card to interrupt the host, an interrupt
function is added to a pin on the DAT[1] line to signal the card’s interrupt to the host. An SDIO
interrupt on each slot can be enabled through the HSMCI Interrupt Enable Register. The SDIO
interrupt is sampled regardless of the currently selected slot.
35.10 CE-ATA Operation
CE-ATA maps the streamlined ATA command set onto the MMC interface. The ATA task file is
mapped onto MMC register space.
CE-ATA utilizes five MMC commands:
• GO_IDLE_STATE (CMD0): used for hard reset.
• STOP_TRANSMISSION (CMD12): causes the ATA command currently executing to be
aborted.
• FAST_IO (CMD39): Used for single register access to the ATA taskfile registers, 8 bit access
only.
• RW_MULTIPLE_REGISTERS (CMD60): used to issue an ATA command or to access the
control/status registers.
• RW_MULTIPLE_BLOCK (CMD61): used to transfer data for an ATA command.
CE-ATA utilizes the same MMC command sequences for initialization as traditional MMC
devices.
35.10.1
Executing an ATA Polling Command
1. Issue READ_DMA_EXT with RW_MULTIPLE_REGISTER (CMD60) for 8kB of DATA.
2. Read the ATA status register until DRQ is set.
681
6438D–ATARM–13-Oct-09
AT91SAM9G45
3. Issue RW_MULTIPLE_BLOCK (CMD61) to transfer DATA.
4. Read the ATA status register until DRQ && BSY are set to 0.
35.10.2
Executing an ATA Interrupt Command
1. Issue READ_DMA_EXT with RW_MULTIPLE_REGISTER (CMD60) for 8kB of DATA
with nIEN field set to zero to enable the command completion signal in the device.
2. Issue RW_MULTIPLE_BLOCK (CMD61) to transfer DATA.
3. Wait for Completion Signal Received Interrupt.
35.10.3
Aborting an ATA Command
If the host needs to abort an ATA command prior to the completion signal it must send a special
command to avoid potential collision on the command line. The SPCMD field of the
HSMCI_CMDR must be set to 3 to issue the CE-ATA completion Signal Disable Command.
35.10.4
CE-ATA Error Recovery
Several methods of ATA command failure may occur, including:
• No response to an MMC command, such as RW_MULTIPLE_REGISTER (CMD60).
• CRC is invalid for an MMC command or response.
• CRC16 is invalid for an MMC data packet.
• ATA Status register reflects an error by setting the ERR bit to one.
• The command completion signal does not arrive within a host specified time out period.
Error conditions are expected to happen infrequently. Thus, a robust error recovery mechanism
may be used for each error event. The recommended error recovery procedure after a timeout
is:
• Issue the command completion signal disable if nIEN was cleared to zero and the
RW_MULTIPLE_BLOCK (CMD61) response has been received.
• Issue STOP_TRANSMISSION (CMD12) and successfully receive the R1 response.
• Issue a software reset to the CE-ATA device using FAST_IO (CMD39).
If STOP_TRANMISSION (CMD12) is successful, then the device is again ready for ATA commands. However, if the error recovery procedure does not work as expected or there is another
timeout, the next step is to issue GO_IDLE_STATE (CMD0) to the device. GO_IDLE_STATE
(CMD0) is a hard reset to the device and completely resets all device states.
Note that after issuing GO_IDLE_STATE (CMD0), all device initialization needs to be completed
again. If the CE-ATA device completes all MMC commands correctly but fails the ATA command
with the ERR bit set in the ATA Status register, no error recovery action is required. The ATA
command itself failed implying that the device could not complete the action requested, however, there was no communication or protocol failure. After the device signals an error by setting
the ERR bit to one in the ATA Status register, the host may attempt to retry the command.
35.11 HSMCI Boot Operation Mode
In boot operation mode, the processor can read boot data from the slave (MMC device) by keeping the CMD line low after power-on before issuing CMD1. The data can be read from either the
boot area or user area, depending on register setting.
682
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.11.1
Boot Procedure, Processor Mode
1. Configure the HSMCI data bus width programming SDCBUS Field in the
HSMCI_SDCR register. The BOOT_BUS_WIDTH field located in the device Extended
CSD register must be set accordingly.
2. Set the byte count to 512 bytes and the block count to the desired number of blocks,
writing BLKLEN and BCNT fields of the HSMCI_BLKR Register.
3. Issue the Boot Operation Request command by writing to the HSMCI_CMDR register
with SPCMD field set to BOOTREQ, TRDIR set to READ and TRCMD set to “start data
transfer”.
4. The BOOT_ACK field located in the HSMCI_CMDR register must be set to one, if the
BOOT_ACK field of the MMC device located in the Extended CSD register is set to one.
5. Host processor can copy boot data sequentially as soon as the RXRDY flag is
asserted.
6. When Data transfer is completed, host processor shall terminate the boot stream by
writing the HSMCI_CMDR register with SPCMD field set to BOOTEND.
35.11.2
Boot Procedure, DMA Mode
1. Configure the HSMCI data bus width by programming SDCBUS Field in the
HSMCI_SDCR register. The BOOT_BUS_WIDTH field in the device Extended CSD
register must be set accordingly.
2. Set the byte count to 512 bytes and the block count to the desired number of blocks by
writing BLKLEN and BCNT fields of the HSMCI_BLKR Register.
3. Enable DMA transfer in the HSMCI_DMA register.
4. Configure DMA controller, program the total amount of data to be transferred and
enable the relevant channel.
5. Issue the Boot Operation Request command by writing to the HSMCI_CMDR register
with SPCND set to BOOTREQ, TRDIR set to READ and TRCMD set to “start data
transfer”.
6. DMA controller copies the boot partition to the memory.
7. When DMA transfer is completed, host processor shall terminate the boot stream by
writing the HSMCI_CMDR register with SPCMD field set to BOOTEND.
683
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.12 HSMCI Transfer Done Timings
35.12.1
Definition
The XFRDONE flag in the HSMCI_SR indicates exactly when the read or write sequence is
finished.
35.12.2
Read Access
During a read access, the XFRDONE flag behaves as shown in Figure 35-11.
Figure 35-11. XFRDONE During a Read Access
CMD line
MCI read CMD
Card response
The CMDRDY flag is released 8 tbit after the end of the card response.
CMDRDY flag
Data
1st Block
Last Block
Not busy flag
XFRDONE flag
684
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.12.3
Write Access
During a write access, the XFRDONE flag behaves as shown in Figure 35-12.
Figure 35-12. XFRDONE During a Write Access
CMD line
MCI writeCMD
CMDRDY flag
Card response
The CMDRDY flag is released 8 tbit after the end of the card response.
D0 is tied by the card
D0 is released
D0
1st Block
Last Block
1st Block
Last Block
Data bus - D0
Not busy flag
XFRDONE flag
685
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.13 MultiMedia Card Interface (MCI) User Interface
Table 35-8.
Register Mapping
Offset
Register
Name
Access
Reset
0x00
Control Register
HSMCI_CR
Write
–
0x04
Mode Register
HSMCI_MR
Read-write
0x0
0x08
Data Timeout Register
HSMCI_DTOR
Read-write
0x0
0x0C
SD/SDIO Card Register
HSMCI_SDCR
Read-write
0x0
0x10
Argument Register
HSMCI_ARGR
Read-write
0x0
0x14
Command Register
HSMCI_CMDR
Write
–
0x18
Block Register
HSMCI_BLKR
Read-write
0x0
0x1C
Completion Signal Timeout Register
HSMCI_CSTOR
Read-write
0x0
(1)
HSMCI_RSPR
Read
0x0
(1)
HSMCI_RSPR
Read
0x0
0x28
(1)
Response Register
HSMCI_RSPR
Read
0x0
0x2C
Response Register(1)
HSMCI_RSPR
Read
0x0
0x30
Receive Data Register
HSMCI_RDR
Read
0x0
0x34
Transmit Data Register
HSMCI_TDR
Write
–
–
–
–
0x20
0x24
0x38 - 0x3C
Response Register
Response Register
Reserved
0x40
Status Register
HSMCI_SR
Read
0xC0E5
0x44
Interrupt Enable Register
HSMCI_IER
Write
–
0x48
Interrupt Disable Register
HSMCI_IDR
Write
–
0x4C
Interrupt Mask Register
HSMCI_IMR
Read
0x0
0x50
DMA Configuration Register
HSMCI_DMA
Read-write
0x00
0x54
Configuration Register
HSMCI_CFG
Read-write
0x00
–
–
–
0x58-0xE0
Reserved
0xE4
Write Protection Mode Register
HSMCI_WPMR
Read-write
–
0xE8
Write Protection Status Register
HSMCI_WPSR
Read-only
–
0xEC - 0xFC
Reserved
–
–
–
0x100-0x124
Reserved
–
–
–
HSMCI_FIFO
Read-write
0x0
0x200-0x3FFC
Note:
FIFO Memory Aperture
1. The response register can be read by N accesses at the same HSMCI_RSPR or at consecutive addresses (0x20 to 0x2C).
N depends on the size of the response.
686
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.13.1
Name:
HSMCI Control Register
HSMCI_CR
Addresses:
0xFFF80000 (0), 0xFFFD0000 (1)
Access:
Write-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
SWRST
6
–
5
–
4
–
3
PWSDIS
2
PWSEN
1
HSMCIDIS
0
MCIEN
• MCIEN: Multi-Media Interface Enable
0 = No effect.
1 = Enables the Multi-Media Interface if MCDIS is 0.
• MCIDIS: Multi-Media Interface Disable
0 = No effect.
1 = Disables the Multi-Media Interface.
• PWSEN: Power Save Mode Enable
0 = No effect.
1 = Enables the Power Saving Mode if PWSDIS is 0.
Warning: Before enabling this mode, the user must set a value different from 0 in the PWSDIV field (Mode Register,
HSMCI_MR).
• PWSDIS: Power Save Mode Disable
0 = No effect.
1 = Disables the Power Saving Mode.
• SWRST: Software Reset
0 = No effect.
1 = Resets the HSMCI. A software triggered hardware reset of the HSMCI interface is performed.
687
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.13.2
Name:
HSMCI Mode Register
HSMCI_MR
Addresses:
0xFFF80004 (0), 0xFFFD0004 (1)
Access:
Read-write
31
30
29
28
27
26
25
24
19
18
17
16
BLKLEN
23
22
21
20
BLKLEN
15
–
14
PADV
13
FBYTE
12
WRPROOF
11
RDPROOF
10
9
PWSDIV
8
7
6
5
4
3
2
1
0
CLKDIV
• CLKDIV: Clock Divider
High Speed MultiMedia Card Interface clock (MCCK or HSMCI_CK) is Master Clock (MCK) divided by (2*(CLKDIV+1)).
• PWSDIV: Power Saving Divider
High Speed MultiMedia Card Interface clock is divided by 2(PWSDIV) + 1 when entering Power Saving Mode.
Warning: This value must be different from 0 before enabling the Power Save Mode in the HSMCI_CR (HSMCI_PWSEN bit).
• RDPROOF Read Proof Enable
Enabling Read Proof allows to stop the HSMCI Clock during read access if the internal FIFO is full. This will guarantee data
integrity, not bandwidth.
0 = Disables Read Proof.
1 = Enables Read Proof.
• WRPROOF Write Proof Enable
Enabling Write Proof allows to stop the HSMCI Clock during write access if the internal FIFO is full. This will guarantee data
integrity, not bandwidth.
0 = Disables Write Proof.
1 = Enables Write Proof.
• FBYTE: Force Byte Transfer
Enabling Force Byte Transfer allow byte transfers, so that transfer of blocks with a size different from modulo 4 can be
supported.
Warning: BLKLEN value depends on FBYTE.
0 = Disables Force Byte Transfer.
1 = Enables Force Byte Transfer.
• PADV: Padding Value
0 = 0x00 value is used when padding data in write transfer.
1 = 0xFF value is used when padding data in write transfer.
PADV may be only in manual transfer.
688
6438D–ATARM–13-Oct-09
AT91SAM9G45
• BLKLEN: Data Block Length
This field determines the size of the data block.
This field is also accessible in the HSMCI Block Register (HSMCI_BLKR).
Bits 16 and 17 must be set to 0 if FBYTE is disabled.
Note:
In SDIO Byte mode, BLKLEN field is not used.
689
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.13.3
Name:
HSMCI Data Timeout Register
HSMCI_DTOR
Addresses:
0xFFF80008 (0), 0xFFFD0008 (1)
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
–
6
5
DTOMUL
4
3
2
1
0
DTOCYC
• DTOCYC: Data Timeout Cycle Number
• DTOMUL: Data Timeout Multiplier
These fields determine the maximum number of Master Clock cycles that the HSMCI waits between two data block transfers. It equals (DTOCYC x Multiplier).
Multiplier is defined by DTOMUL as shown in the following table:
DTOMUL
Multiplier
0
0
0
1
0
0
1
16
0
1
0
128
0
1
1
256
1
0
0
1024
1
0
1
4096
1
1
0
65536
1
1
1
1048576
If the data time-out set by DTOCYC and DTOMUL has been exceeded, the Data Time-out Error flag (DTOE) in the HSMCI
Status Register (HSMCI_SR) raises.
690
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.13.4
Name:
HSMCI SDCard/SDIO Register
HSMCI_SDCR
Addresses:
0xFFF8000C (0), 0xFFFD000C (1)
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
6
5
–
4
–
3
–
2
–
1
7
SDCBUS
0
SDCSEL
• SDCSEL: SDCard/SDIO Slot
SDCSEL
SDCard/SDIO Slot
0
0
Slot A is selected.
0
1
–
1
0
–
1
1
–
• SDCBUS: SDCard/SDIO Bus Width
SDCBUS
BUS WIDTH
0
0
1 bit
1
0
4 bit
1
1
8 bit
691
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.13.5
Name:
HSMCI Argument Register
HSMCI_ARGR
Addresses:
0xFFF80010 (0), 0xFFFD0010 (1)
Access:
Read-write
31
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
0
ARG
23
22
21
20
ARG
15
14
13
12
ARG
7
6
5
4
ARG
• ARG: Command Argument
692
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.13.6
Name:
HSMCI Command Register
HSMCI_CMDR
Addresses:
0xFFF80014 (0), 0xFFFD0014 (1)
Access:
Write-only
31
–
30
–
29
–
28
–
27
BOOT_ACK
26
ATACS
25
23
–
22
–
21
20
TRTYP
19
18
TRDIR
17
15
–
14
–
13
–
12
MAXLAT
11
OPDCMD
10
9
SPCMD
8
6
5
4
3
2
1
0
7
RSPTYP
24
IOSPCMD
16
TRCMD
CMDNB
This register is write-protected while CMDRDY is 0 in HSMCI_SR. If an Interrupt command is sent, this register is only
writeable by an interrupt response (field SPCMD). This means that the current command execution cannot be interrupted
or modified.
• CMDNB: Command Number
• RSPTYP: Response Type
RSP
Response Type
0
0
No response.
0
1
48-bit response.
1
0
136-bit response.
1
1
R1b response type
• SPCMD: Special Command
SPCMD
Command
0
0
0
Not a special CMD.
0
0
1
Initialization CMD:
74 clock cycles for initialization sequence.
0
1
0
Synchronized CMD:
Wait for the end of the current data block transfer before sending the
pending command.
0
1
1
CE-ATA Completion Signal disable Command.
The host cancels the ability for the device to return a command
completion signal on the command line.
1
0
0
Interrupt command:
Corresponds to the Interrupt Mode (CMD40).
693
6438D–ATARM–13-Oct-09
AT91SAM9G45
SPCMD
Command
1
0
1
Interrupt response:
Corresponds to the Interrupt Mode (CMD40).
1
1
0
Boot Operation Request.
Start a boot operation mode, the host processor can read boot data from
the MMC device directly.
1
1
1
End Boot Operation.
This command allows the host processor to terminate the boot operation
mode.
• OPDCMD: Open Drain Command
0 = Push pull command
1 = Open drain command
• MAXLAT: Max Latency for Command to Response
0 = 5-cycle max latency
1 = 64-cycle max latency
• TRCMD: Transfer Command
TRCMD
Transfer Type
0
0
No data transfer
0
1
Start data transfer
1
0
Stop data transfer
1
1
Reserved
• TRDIR: Transfer Direction
0 = Write
1 = Read
• TRTYP: Transfer Type
TRTYP
Transfer Type
0
0
0
MMC/SDCard Single Block
0
0
1
MMC/SDCard Multiple Block
0
1
0
MMC Stream
0
1
1
Reserved
1
0
0
SDIO Byte
1
0
1
SDIO Block
1
1
0
Reserved
1
1
1
Reserved
694
6438D–ATARM–13-Oct-09
AT91SAM9G45
• IOSPCMD: SDIO Special Command
IOSPCMD
SDIO Special Command Type
0
0
Not a SDIO Special Command
0
1
SDIO Suspend Command
1
0
SDIO Resume Command
1
1
Reserved
• ATACS: ATA with Command Completion Signal
0 = Normal operation mode.
1 = This bit indicates that a completion signal is expected within a programmed amount of time (HSMCI_CSTOR).
• BOOT_ACK: Boot Operation Acknowledge.
The master can choose to receive the boot acknowledge from the slave when a Boot Request command is issued. When
set to one this field indicates that a Boot acknowledge is expected within a programmable amount of time defined with
DTOMUL and DTOCYC fields located in the HSMCI_DTOR register. If the acknowledge pattern is not received then an
acknowledge timeout error is raised. If the acknowledge pattern is corrupted then an acknowledge pattern error is set.
695
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.13.7
Name:
HSMCI Block Register
HSMCI_BLKR
Addresses:
0xFFF80018 (0), 0xFFFD0018 (1)
Access:
Read-write
31
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
0
BLKLEN
23
22
21
20
BLKLEN
15
14
13
12
BCNT
7
6
5
4
BCNT
• BCNT: MMC/SDIO Block Count - SDIO Byte Count
This field determines the number of data byte(s) or block(s) to transfer.
The transfer data type and the authorized values for BCNT field are determined by the TRTYP field in the HSMCI Command Register (HSMCI_CMDR):
TRTYP
Type of Transfer
BCNT Authorized Values
0
0
1
MMC/SDCard Multiple Block
From 1 to 65635: Value 0 corresponds to an infinite block transfer.
1
0
0
SDIO Byte
From 1 to 512 bytes: Value 0 corresponds to a 512-byte transfer.
Values from 0x200 to 0xFFFF are forbidden.
1
0
1
SDIO Block
From 1 to 511 blocks: Value 0 corresponds to an infinite block transfer.
Values from 0x200 to 0xFFFF are forbidden.
-
Reserved.
Other values
Warning: In SDIO Byte and Block modes, writing to the 7 last bits of BCNT field, is forbidden and may lead to unpredictable results.
• BLKLEN: Data Block Length
This field determines the size of the data block.
This field is also accessible in the HSMCI Mode Register (HSMCI_MR).
Bits 16 and 17 must be set to 0 if FBYTE is disabled.
Note:
In SDIO Byte mode, BLKLEN field is not used.
696
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.13.8
Name:
HSMCI Completion Signal Timeout Register
HSMCI_CSTOR
Addresses:
0xFFF8001C (0), 0xFFFD001C (1)
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
–
6
5
CSTOMUL
4
3
2
1
0
CSTOCYC
• CSTOCYC: Completion Signal Timeout Cycle Number
• CSTOMUL: Completion Signal Timeout Multiplier
These fields determine the maximum number of Master Clock cycles that the HSMCI waits between two data block transfers. Its value is calculated by (CSTOCYC x Multiplier).
These fields determine the maximum number of Master Clock cycles that the HSMCI waits between the end of the data
transfer and the assertion of the completion signal. The data transfer comprises data phase and the optional busy phase. If
a non-DATA ATA command is issued, the HSMCI starts waiting immediately after the end of the response until the completion signal.
Multiplier is defined by CSTOMUL as shown in the following table:
CSTOMUL
Multiplier
0
0
0
1
0
0
1
16
0
1
0
128
0
1
1
256
1
0
0
1024
1
0
1
4096
1
1
0
65536
1
1
1
1048576
If the data time-out set by CSTOCYC and CSTOMUL has been exceeded, the Completion Signal Time-out Error flag
(CSTOE) in the HSMCI Status Register (HSMCI_SR) raises.
697
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.13.9
Name:
HSMCI Response Register
HSMCI_RSPR
Addresses:
0xFFF80020 (0), 0xFFFD0020 (1)
Access:
Read-only
31
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
0
RSP
23
22
21
20
RSP
15
14
13
12
RSP
7
6
5
4
RSP
• RSP: Response
Note:
1. The response register can be read by N accesses at the same HSMCI_RSPR or at consecutive addresses (0x20 to 0x2C).
N depends on the size of the response.
35.13.10 HSMCI Receive Data Register
Name:
HSMCI_RDR
Addresses:
0xFFF80030 (0), 0xFFFD0030 (1)
Access:
Read-only
31
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
0
DATA
23
22
21
20
DATA
15
14
13
12
DATA
7
6
5
4
DATA
• DATA: Data to Read
698
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.13.11 HSMCI Transmit Data Register
Name:
HSMCI_TDR
Addresses:
0xFFF80034 (0), 0xFFFD0034 (1)
Access:
Write-only
31
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
0
DATA
23
22
21
20
DATA
15
14
13
12
DATA
7
6
5
4
DATA
• DATA: Data to Write
699
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.13.12 HSMCI Status Register
Name:
HSMCI_SR
Addresses:
0xFFF80040 (0), 0xFFFD0040 (1)
Access:
Read-only
31
UNRE
30
OVRE
29
ACKRCVE
28
ACKRCV
27
XFRDONE
26
FIFOEMPTY
25
DMADONE
24
BLKOVRE
23
CSTOE
22
DTOE
21
DCRCE
20
RTOE
19
RENDE
18
RCRCE
17
RDIRE
16
RINDE
15
14
13
12
11
10
9
–
–
CSRCV
SDIOWAIT
–
–
–
8
MCI_SDIOIR
QA
7
–
6
–
5
NOTBUSY
4
DTIP
3
BLKE
2
TXRDY
1
RXRDY
0
CMDRDY
• CMDRDY: Command Ready
0 = A command is in progress.
1 = The last command has been sent. Cleared when writing in the HSMCI_CMDR.
• RXRDY: Receiver Ready
0 = Data has not yet been received since the last read of HSMCI_RDR.
1 = Data has been received since the last read of HSMCI_RDR.
• TXRDY: Transmit Ready
0 = The last data written in HSMCI_TDR has not yet been transferred in the Shift Register.
1 = The last data written in HSMCI_TDR has been transferred in the Shift Register.
• BLKE: Data Block Ended
This flag must be used only for Write Operations.
0 = A data block transfer is not yet finished. Cleared when reading the HSMCI_SR.
1 = A data block transfer has ended, including the CRC16 Status transmission.
the flag is set for each transmitted CRC Status.
Refer to the MMC or SD Specification for more details concerning the CRC Status.
• DTIP: Data Transfer in Progress
0 = No data transfer in progress.
1 = The current data transfer is still in progress, including CRC16 calculation. Cleared at the end of the CRC16 calculation.
• NOTBUSY: HSMCI Not Busy
This flag must be used only for Write Operations.
A block write operation uses a simple busy signalling of the write operation duration on the data (DAT0) line: during a data
transfer block, if the card does not have a free data receive buffer, the card indicates this condition by pulling down the data
line (DAT0) to LOW. The card stops pulling down the data line as soon as at least one receive buffer for the defined data
transfer block length becomes free.
The NOTBUSY flag allows to deal with these different states.
700
6438D–ATARM–13-Oct-09
AT91SAM9G45
0 = The HSMCI is not ready for new data transfer. Cleared at the end of the card response.
1 = The HSMCI is ready for new data transfer. Set when the busy state on the data line has ended. This corresponds to a
free internal data receive buffer of the card.
Refer to the MMC or SD Specification for more details concerning the busy behavior.
For all the read operations, the NOTBUSY flag is cleared at the end of the host command.
For the Infinite Read Multiple Blocks, the NOTBUSY flag is set at the end of the STOP_TRANSMISSION host command
(CMD12).
For the Single Block Reads, the NOTBUSY flag is set at the end of the data read block.
For the Multiple Block Reads with pre-defined block count, the NOTBUSY flag is set at the end of the last received data
block.
• SDIOIRQA: SDIO Interrupt for Slot A
0 = No interrupt detected on SDIO Slot A.
1 = An SDIO Interrupt on Slot A occurred. Cleared when reading the HSMCI_SR.
• SDIOWAIT: SDIO Read Wait Operation Status
0 = Normal Bus operation.
1 = The data bus has entered IO wait state.
• CSRCV: CE-ATA Completion Signal Received
0 = No completion signal received since last status read operation.
1 = The device has issued a command completion signal on the command line. Cleared by reading in the HSMCI_SR
register.
• RINDE: Response Index Error
0 = No error.
1 = A mismatch is detected between the command index sent and the response index received. Cleared when writing in
the HSMCI_CMDR.
• RDIRE: Response Direction Error
0 = No error.
1 = The direction bit from card to host in the response has not been detected.
• RCRCE: Response CRC Error
0 = No error.
1 = A CRC7 error has been detected in the response. Cleared when writing in the HSMCI_CMDR.
• RENDE: Response End Bit Error
0 = No error.
1 = The end bit of the response has not been detected. Cleared when writing in the HSMCI_CMDR.
• RTOE: Response Time-out Error
0 = No error.
1 = The response time-out set by MAXLAT in the HSMCI_CMDR has been exceeded. Cleared when writing in the
HSMCI_CMDR.
701
6438D–ATARM–13-Oct-09
AT91SAM9G45
• DCRCE: Data CRC Error
0 = No error.
1 = A CRC16 error has been detected in the last data block. Cleared by reading in the HSMCI_SR register.
• DTOE: Data Time-out Error
0 = No error.
1 = The data time-out set by DTOCYC and DTOMUL in HSMCI_DTOR has been exceeded. Cleared by reading in the
HSMCI_SR register.
• CSTOE: Completion Signal Time-out Error
0 = No error.
1 = The completion signal time-out set by CSTOCYC and CSTOMUL in HSMCI_CSTOR has been exceeded. Cleared by
reading in the HSMCI_SR register. Cleared by reading in the HSMCI_SR register.
• BLKOVRE: DMA Block Overrun Error
0 = No error.
1 = A new block of data is received and the DMA controller has not started to move the current pending block, a block overrun is raised. Cleared by reading in the HSMCI_SR register.
• DMADONE: DMA Transfer done
0 = DMA buffer transfer has not completed since the last read of HSMCI_SR register.
1 = DMA buffer transfer has completed.
• FIFOEMPTY: FIFO empty flag
0 = FIFO contains at least one byte.
1 = FIFO is empty.
• XFRDONE: Transfer Done flag
0 = A transfer is in progress.
1 = Command register is ready to operate and the data bus is in the idle state.
• ACKRCV: Boot Operation Acknowledge Received
0 = No Boot acknowledge received since the last read of the status register.
1 = A Boot acknowledge signal has been received. Cleared by reading the HSMCI_SR register.
• ACKRCVE: Boot Operation Acknowledge Error
0 = No error
1 = Corrupted Boot Acknowledge signal received.
• OVRE: Overrun
0 = No error.
1 = At least one 8-bit received data has been lost (not read). Cleared when sending a new data transfer command.
When FERRCTRL in HSMCI_CFG is set to 1, OVRE becomes reset after read.
702
6438D–ATARM–13-Oct-09
AT91SAM9G45
• UNRE: Underrun
0 = No error.
1 = At least one 8-bit data has been sent without valid information (not written). Cleared when sending a new data transfer
command or when setting FERRCTRL in HSMCI_CFG to 1.
When FERRCTRL in HSMCI_CFG is set to 1, UNRE becomes reset after read.
703
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.13.13 HSMCI Interrupt Enable Register
Name:
HSMCI_IER
Addresses:
0xFFF80044 (0), 0xFFFD0044 (1)
Access:
Write-only
31
UNRE
30
OVRE
29
ACKRCVE
28
ACKRCV
27
XFRDONE
26
FIFOEMPTY
25
DMADONE
24
BLKOVRE
23
CSTOE
22
DTOE
21
DCRCE
20
RTOE
19
RENDE
18
RCRCE
17
RDIRE
16
RINDE
15
–
14
–
13
CSRCV
12
SDIOWAIT
11
–
10
–
9
–
8
MCI_SDIOIRQA
7
–
6
–
5
NOTBUSY
4
DTIP
3
BLKE
2
TXRDY
1
RXRDY
0
CMDRDY
• CMDRDY: Command Ready Interrupt Enable
• RXRDY: Receiver Ready Interrupt Enable
• TXRDY: Transmit Ready Interrupt Enable
• BLKE: Data Block Ended Interrupt Enable
• DTIP: Data Transfer in Progress Interrupt Enable
• NOTBUSY: Data Not Busy Interrupt Enable
• SDIOIRQA: SDIO Interrupt for Slot A Interrupt Enable
• SDIOWAIT: SDIO Read Wait Operation Status Interrupt Enable
• CSRCV: Completion Signal Received Interrupt Enable
• RINDE: Response Index Error Interrupt Enable
• RDIRE: Response Direction Error Interrupt Enable
• RCRCE: Response CRC Error Interrupt Enable
• RENDE: Response End Bit Error Interrupt Enable
• RTOE: Response Time-out Error Interrupt Enable
• DCRCE: Data CRC Error Interrupt Enable
• DTOE: Data Time-out Error Interrupt Enable
• CSTOE: Completion Signal Timeout Error Interrupt Enable
• BLKOVRE: DMA Block Overrun Error Interrupt Enable
• DMADONE: DMA Transfer completed Interrupt Enable
• FIFOEMPTY: FIFO empty Interrupt enable
• XFRDONE: Transfer Done Interrupt enable
704
6438D–ATARM–13-Oct-09
AT91SAM9G45
• ACKRCV: Boot Acknowledge Interrupt Enable
• ACKRCVE: Boot Acknowledge Error Interrupt Enable
• OVRE: Overrun Interrupt Enable
• UNRE: Underrun Interrupt Enable
0 = No effect.
1 = Enables the corresponding interrupt.
705
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.13.14 HSMCI Interrupt Disable Register
Name:
HSMCI_IDR
Addresses:
0xFFF80048 (0), 0xFFFD0048 (1)
Access:
Write-only
31
UNRE
30
OVRE
29
ACKRCVE
28
ACKRCV
27
XFRDONE
26
FIFOEMPTY
25
DMADONE
24
BLKOVRE
23
CSTOE
22
DTOE
21
DCRCE
20
RTOE
19
RENDE
18
RCRCE
17
RDIRE
16
RINDE
15
–
14
–
13
CSRCV
12
SDIOWAIT
11
–
10
–
9
–
8
MCI_SDIOIRQA
7
–
6
–
5
NOTBUSY
4
DTIP
3
BLKE
2
TXRDY
1
RXRDY
0
CMDRDY
• CMDRDY: Command Ready Interrupt Disable
• RXRDY: Receiver Ready Interrupt Disable
• TXRDY: Transmit Ready Interrupt Disable
• BLKE: Data Block Ended Interrupt Disable
• DTIP: Data Transfer in Progress Interrupt Disable
• NOTBUSY: Data Not Busy Interrupt Disable
• SDIOIRQA: SDIO Interrupt for Slot A Interrupt Disable
• SDIOWAIT: SDIO Read Wait Operation Status Interrupt Disable
• CSRCV: Completion Signal received interrupt disable
• RINDE: Response Index Error Interrupt Disable
• RDIRE: Response Direction Error Interrupt Disable
• RCRCE: Response CRC Error Interrupt Disable
• RENDE: Response End Bit Error Interrupt Disable
• RTOE: Response Time-out Error Interrupt Disable
• DCRCE: Data CRC Error Interrupt Disable
• DTOE: Data Time-out Error Interrupt Disable
• CSTOE: Completion Signal Time out Error Interrupt Disable
• BLKOVRE: DMA Block Overrun Error Interrupt Disable
• DMADONE: DMA Transfer completed Interrupt Disable
• FIFOEMPTY: FIFO empty Interrupt Disable
• XFRDONE: Transfer Done Interrupt Disable
706
6438D–ATARM–13-Oct-09
AT91SAM9G45
• ACKRCV: Boot Acknowledge Interrupt Disable
• ACKRCVE: Boot Acknowledge Error Interrupt Disable
• OVRE: Overrun Interrupt Disable
• UNRE: Underrun Interrupt Disable
0 = No effect.
1 = Disables the corresponding interrupt.
707
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.13.15 HSMCI Interrupt Mask Register
Name:
HSMCI_IMR
Addresses:
0xFFF8004C (0), 0xFFFD004C (1)
Access:
Read-only
31
UNRE
30
OVRE
29
ACKRCVE
28
ACKRCV
27
XFRDONE
26
FIFOEMPTY
25
DMADONE
24
BLKOVRE
23
CSTOE
22
DTOE
21
DCRCE
20
RTOE
19
RENDE
18
RCRCE
17
RDIRE
16
RINDE
15
–
14
–
13
CSRCV
12
SDIOWAIT
11
–
10
–
9
–
8
MCI_SDIOIRQA
7
–
6
–
5
NOTBUSY
4
DTIP
3
BLKE
2
TXRDY
1
RXRDY
0
CMDRDY
• CMDRDY: Command Ready Interrupt Mask
• RXRDY: Receiver Ready Interrupt Mask
• TXRDY: Transmit Ready Interrupt Mask
• BLKE: Data Block Ended Interrupt Mask
• DTIP: Data Transfer in Progress Interrupt Mask
• NOTBUSY: Data Not Busy Interrupt Mask
• SDIOIRQA: SDIO Interrupt for Slot A Interrupt Mask
• SDIOWAIT: SDIO Read Wait Operation Status Interrupt Mask
• CSRCV: Completion Signal Received Interrupt Mask
• RINDE: Response Index Error Interrupt Mask
• RDIRE: Response Direction Error Interrupt Mask
• RCRCE: Response CRC Error Interrupt Mask
• RENDE: Response End Bit Error Interrupt Mask
• RTOE: Response Time-out Error Interrupt Mask
• DCRCE: Data CRC Error Interrupt Mask
• DTOE: Data Time-out Error Interrupt Mask
• CSTOE: Completion Signal Time-out Error Interrupt Mask
• BLKOVRE: DMA Block Overrun Error Interrupt Mask
• DMADONE: DMA Transfer Completed Interrupt Mask
• FIFOEMPTY: FIFO Empty Interrupt Mask
• XFRDONE: Transfer Done Interrupt Mask
708
6438D–ATARM–13-Oct-09
AT91SAM9G45
• ACKRCV: Boot Operation Acknowledge Received Interrupt Mask
• ACKRCVE: Boot Operation Acknowledge Error Interrupt Mask
• OVRE: Overrun Interrupt Mask
• UNRE: Underrun Interrupt Mask
0 = The corresponding interrupt is not enabled.
1 = The corresponding interrupt is enabled.
709
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.13.16 HSMCI DMA Configuration Register
Name:
HSMCI_DMA
Addresses:
0xFFF80050 (0), 0xFFFD0050 (1)
Access:
Read-write
31
30
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
ROPT
11
–
10
–
9
–
8
DMAEN
7
–
6
–
5
4
3
–
2
–
1
CHKSIZE
0
OFFSET
• OFFSET: DMA Write Buffer Offset
This field indicates the number of discarded bytes when the DMA writes the first word of the transfer.
• CHKSIZE: DMA Channel Read and Write Chunk Size
The CHKSIZE field indicates the number of data available when the DMA chunk transfer request is asserted.
CHKSIZE value
Number of data transferred
00
1
01
4
10
8
11
16
• DMAEN: DMA Hardware Handshaking Enable
0 = DMA interface is disabled.
1 = DMA Interface is enabled.
Note:
To avoid unpredictable behavior, DMA hardware handshaking must be disabled when CPU transfers are performed.
• ROPT: Read Optimization with padding
0: BLKLEN bytes are moved from the Memory Card to the system memory, two DMA descriptors are used when the transfer size is not a multiple of 4.
1: Ceiling(BLKLEN/4) * 4 bytes are moved from the Memory Card to the system memory, only one DMA descriptor is used.
710
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.13.17 HSMCI Configuration Register
Name:
HSMCI_CFG
Addresses:
0xFFF80054 (0), 0xFFFD0054 (1)
Access:
Read-write
31
30
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
LSYNC
11
–
10
–
9
–
8
HSMODE
7
–
6
–
5
–
4
FERRCTRL
3
–
2
–
1
–
0
FIFOMODE
• FIFOMODE: HSMCI Internal FIFO control mode
0 = A write transfer starts when a sufficient amount of data is written into the FIFO.
When the block length is greater than or equal to 3/4 of the HSMCI internal FIFO size, then the write transfer starts as soon
as half the FIFO is filled. When the block length is greater than or equal to half the internal FIFO size, then the write transfer
starts as soon as one quarter of the FIFO is filled. In other cases, the transfer starts as soon as the total amount of data is
written in the internal FIFO.
1 = A write transfer starts as soon as one data is written into the FIFO.
• FERRCTRL: Flow Error flag reset control mode
0 = When an underflow/overflow condition flag is set, a new Write/Read command is needed to reset the flag.
1 = When an underflow/overflow condition flag is set, a read status resets the flag.
• HSMODE: High Speed Mode
0 = Default bus timing mode.
1 = If set to one, the host controller outputs command line and data lines on the rising edge of the card clock. The Host
driver shall check the high speed support in the card registers.
• LSYNC: Synchronize on the last block
0 = The pending command is sent at the end of the current data block.
1 = The pending command is sent at the end of the block transfer when the transfer length is not infinite (block count shall
be different from zero).
711
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.13.18 HSMCI Write Protect Mode Register
Name:
HSMCI_WPMR
Addresses:
0xFFF800E4 (0), 0xFFFD00E4 (1)
Access:
Read-write
31
30
29
28
27
WP_KEY (0x4D => “M”)
26
25
24
23
22
21
20
19
WP_KEY (0x43 => C”)
18
17
16
15
14
13
12
11
WP_KEY (0x49 => “I”)
10
9
8
7
6
5
2
1
0
WP_EN
4
3
• WP_EN: Write Protection Enable
0 = Disables the Write Protection if WP_KEY corresponds.
1 = Enables the Write Protection if WP_KEY corresponds.
• WP_KEY: Write Protection Key password
Should be written at value 0x4D4349 (ASCII code for “HSMCI”). Writing any other value in this field has no effect.
712
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.13.19 HSMCI Write Protect Status Register
Name:
HSMCI_WPSR
Addresses:
0xFFF800E8 (0), 0xFFFD00E8 (1)
Access:
Read-only
31
-
30
-
29
-
28
-
23
22
21
20
27
-
26
-
25
-
24
-
19
18
17
16
11
10
9
8
3
2
1
0
WP_VSRC
15
14
13
12
WP_VSRC
7
-
6
-
5
-
4
-
WP_VS
• WP_VSRC: Write Protection Violation Status
WP_VS
0
0
0
0
No Write Protection Violation occurred since the last read of this
register (WP_SR)
0
0
0
1
Write Protection detected unauthorized attempt to write a control
register had occurred (since the last read.)
0
0
1
0
Software reset had been performed while Write Protection was
enabled (since the last read).
0
0
1
1
Both Write Protection violation and software reset with Write
Protection enabled had occurred since the last read.
Other value
Reserved
• WP_VSRC: Write Protection Violation SouRCe
WP_VSRC
0
0
0
0
No Write Protection Violation occurred since the last read of this
register (WP_SR)
0
0
0
1
Write access in HSMCI_MR while Write Protection was enabled
(since the last read).
0
0
1
0
Write access in HSMCI_DTOR while Write Protection was enabled
(since the last read)
0
0
1
1
Write access in HSMCI_SDCR while Write Protection was enabled
(since the last read)
0
1
0
0
Write access in HSMCI_CSTOR while Write Protection was
enabled (since the last read)
0
1
0
1
Write access in HSMCI_DMA while Write Protection was enabled
(since the last read)
0
1
1
0
Write access in HSMCI_CFG while Write Protection was enabled
(since the last read)
Other value
Reserved
713
6438D–ATARM–13-Oct-09
AT91SAM9G45
35.13.20 HSMCI FIFO Memory Aperture
Name:
HSMCI_FIFO
Access:
31
Read-write
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
0
DATA
23
22
21
20
DATA
15
14
13
12
DATA
7
6
5
4
DATA
• DATA: Data to Read or Data to Write
714
6438D–ATARM–13-Oct-09
AT91SAM9G45
36. Ethernet MAC 10/100 (EMAC)
36.1
Description
The EMAC module implements a 10/100 Ethernet MAC compatible with the IEEE 802.3 standard using an address checker, statistics and control registers, receive and transmit blocks, and
a DMA interface.
The address checker recognizes four specific 48-bit addresses and contains a 64-bit hash register for matching multicast and unicast addresses. It can recognize the broadcast address of all
ones, copy all frames, and act on an external address match signal.
The statistics register block contains registers for counting various types of event associated
with transmit and receive operations. These registers, along with the status words stored in the
receive buffer list, enable software to generate network management statistics compatible with
IEEE 802.3.
36.2
Embedded Characteristics
• Compatibility with IEEE Standard 802.3
• 10 and 100 MBits per second data throughput capability
• Full- and half-duplex operations
• MII or RMII interface to the physical layer
• Register Interface to address, data, status and control registers
• DMA Interface, operating as a master on the Memory Controller
• Interrupt generation to signal receive and transmit completion
• 128-byte transmit and 128-byte receive FIFOs
• Automatic pad and CRC generation on transmitted frames
• Address checking logic to recognize four 48-bit addresses
• Supports promiscuous mode where all valid frames are copied to memory
• Supports physical layer management through MDIO interface
• Supports Wake On Lan. The receiver supports Wake on LAN by detecting the following
events on incoming receive frames:
– Magic packet
– ARP request to the device IP address
– Specific address 1 filter match
– Multicast hash filter match
715
6438D–ATARM–13-Oct-09
36.3
Block Diagram
Figure 36-1. EMAC Block Diagram
Address Checker
APB
Slave
Register Interface
Statistics Registers
MDIO
Control Registers
DMA Interface
RX FIFO
TX FIFO
Ethernet Receive
MII/RMII
AHB
Master
Ethernet Transmit
716
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
36.4
Functional Description
The MACB has several clock domains:
•
System bus clock (AHB and APB): DMA and register blocks
•
Transmit clock: transmit block
•
Receive clock: receive and address checker block
The system bus clock must run at least as fast as the receive clock and transmit clock (25 MHz
at 100 Mbps, and 2.5 MHZ at 10 Mbps).
Figure 36-1 illustrates the different blocks of the EMAC module.
The control registers drive the MDIO interface, setup up DMA activity, start frame transmission
and select modes of operation such as full- or half-duplex.
The receive block checks for valid preamble, FCS, alignment and length, and presents received
frames to the address checking block and DMA interface.
The transmit block takes data from the DMA interface, adds preamble and, if necessary, pad
and FCS, and transmits data according to the CSMA/CD (carrier sense multiple access with collision detect) protocol. The start of transmission is deferred if CRS (carrier sense) is active.
If COL (collision) becomes active during transmission, a jam sequence is asserted and the
transmission is retried after a random back off. CRS and COL have no effect in full duplex mode.
The DMA block connects to external memory through its AHB bus interface. It contains receive
and transmit FIFOs for buffering frame data. It loads the transmit FIFO and empties the receive
FIFO using AHB bus master operations. Receive data is not sent to memory until the address
checking logic has determined that the frame should be copied. Receive or transmit frames are
stored in one or more buffers. Receive buffers have a fixed length of 128 bytes. Transmit buffers
range in length between 0 and 2047 bytes, and up to 128 buffers are permitted per frame. The
DMA block manages the transmit and receive framebuffer queues. These queues can hold multiple frames.
36.4.1
Clock
Synchronization module in the EMAC requires that the bus clock (hclk) runs at the speed of the
macb_tx/rx_clk at least, which is 25 MHz at 100 Mbps, and 2.5 MHz at 10 Mbps.
36.4.2
Memory Interface
Frame data is transferred to and from the EMAC through the DMA interface. All transfers are 32bit words and may be single accesses or bursts of 2, 3 or 4 words. Burst accesses do not cross
sixteen-byte boundaries. Bursts of 4 words are the default data transfer; single accesses or
bursts of less than four words may be used to transfer data at the beginning or the end of a
buffer.
The DMA controller performs six types of operation on the bus. In order of priority, these are:
1. Receive buffer manager write
2. Receive buffer manager read
3. Transmit data DMA read
4. Receive data DMA write
5. Transmit buffer manager read
6. Transmit buffer manager write
717
6438D–ATARM–13-Oct-09
36.4.2.1
FIFO
The FIFO depths are 128 bytes for receive and 128 bytes for transmit and are a function of the
system clock speed, memory latency and network speed.
Data is typically transferred into and out of the FIFOs in bursts of four words. For receive, a bus
request is asserted when the FIFO contains four words and has space for 28 more. For transmit,
a bus request is generated when there is space for four words, or when there is space for 27
words if the next transfer is to be only one or two words.
Thus the bus latency must be less than the time it takes to load the FIFO and transmit or receive
three words (112 bytes) of data.
At 100 Mbit/s, it takes 8960 ns to transmit or receive 112 bytes of data. In addition, six master
clock cycles should be allowed for data to be loaded from the bus and to propagate through the
FIFOs. For a 133 MHz master clock this takes 45 ns, making the bus latency requirement 8915
ns.
36.4.2.2
Receive Buffers
Received frames, including CRC/FCS optionally, are written to receive buffers stored in memory. Each receive buffer is 128 bytes long. The start location for each receive buffer is stored in
memory in a list of receive buffer descriptors at a location pointed to by the receive buffer queue
pointer register. The receive buffer start location is a word address. For the first buffer of a
frame, the start location can be offset by up to three bytes depending on the value written to bits
14 and 15 of the network configuration register. If the start location of the buffer is offset the
available length of the first buffer of a frame is reduced by the corresponding number of bytes.
Each list entry consists of two words, the first being the address of the receive buffer and the
second being the receive status. If the length of a receive frame exceeds the buffer length, the
status word for the used buffer is written with zeroes except for the “start of frame” bit and the
offset bits, if appropriate. Bit zero of the address field is written to one to show the buffer has
been used. The receive buffer manager then reads the location of the next receive buffer and
fills that with receive frame data. The final buffer descriptor status word contains the complete
frame status. Refer to Table 36-1 for details of the receive buffer descriptor list.
Table 36-1.
Receive Buffer Descriptor Entry
Bit
Function
Word 0
31:2
Address of beginning of buffer
1
Wrap - marks last descriptor in receive buffer descriptor list.
0
Ownership - needs to be zero for the EMAC to write data to the receive buffer. The EMAC sets this to one once it has
successfully written a frame to memory.
Software has to clear this bit before the buffer can be used again.
Word 1
718
31
Global all ones broadcast address detected
30
Multicast hash match
29
Unicast hash match
28
External address match
27
Reserved for future use
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Table 36-1.
Receive Buffer Descriptor Entry (Continued)
Bit
Function
26
Specific address register 1 match
25
Specific address register 2 match
24
Specific address register 3 match
23
Specific address register 4 match
22
Type ID match
21
VLAN tag detected (i.e., type id of 0x8100)
20
Priority tag detected (i.e., type id of 0x8100 and null VLAN identifier)
19:17
VLAN priority (only valid if bit 21 is set)
16
Concatenation format indicator (CFI) bit (only valid if bit 21 is set)
15
End of frame - when set the buffer contains the end of a frame. If end of frame is not set, then the only other valid status
are bits 12, 13 and 14.
14
Start of frame - when set the buffer contains the start of a frame. If both bits 15 and 14 are set, then the buffer contains a
whole frame.
13:12
Receive buffer offset - indicates the number of bytes by which the data in the first buffer is offset from the word address.
Updated with the current values of the network configuration register. If jumbo frame mode is enabled through bit 3 of the
network configuration register, then bits 13:12 of the receive buffer descriptor entry are used to indicate bits 13:12 of the
frame length.
11:0
Length of frame including FCS (if selected). Bits 13:12 are also used if jumbo frame mode is selected.
To receive frames, the buffer descriptors must be initialized by writing an appropriate address to
bits 31 to 2 in the first word of each list entry. Bit zero must be written with zero. Bit one is the
wrap bit and indicates the last entry in the list.
The start location of the receive buffer descriptor list must be written to the receive buffer queue
pointer register before setting the receive enable bit in the network control register to enable
receive. As soon as the receive block starts writing received frame data to the receive FIFO, the
receive buffer manager reads the first receive buffer location pointed to by the receive buffer
queue pointer register.
If the filter block then indicates that the frame should be copied to memory, the receive data
DMA operation starts writing data into the receive buffer. If an error occurs, the buffer is recovered. If the current buffer pointer has its wrap bit set or is the 1024th descriptor, the next receive
buffer location is read from the beginning of the receive descriptor list. Otherwise, the next
receive buffer location is read from the next word in memory.
There is an 11-bit counter to count out the 2048 word locations of a maximum length, receive
buffer descriptor list. This is added with the value originally written to the receive buffer queue
pointer register to produce a pointer into the list. A read of the receive buffer queue pointer register returns the pointer value, which is the queue entry currently being accessed. The counter is
reset after receive status is written to a descriptor that has its wrap bit set or rolls over to zero
after 1024 descriptors have been accessed. The value written to the receive buffer pointer register may be any word-aligned address, provided that there are at least 2048 word locations
available between the pointer and the top of the memory.
Section 3.6 of the AMBA 2.0 specification states that bursts should not cross 1K boundaries. As
receive buffer manager writes are bursts of two words, to ensure that this does not occur, it is
719
6438D–ATARM–13-Oct-09
best to write the pointer register with the least three significant bits set to zero. As receive buffers
are used, the receive buffer manager sets bit zero of the first word of the descriptor to indicate
used. If a receive error is detected the receive buffer currently being written is recovered. Previous buffers are not recovered. Software should search through the used bits in the buffer
descriptors to find out how many frames have been received. It should be checking the start-offrame and end-of-frame bits, and not rely on the value returned by the receive buffer queue
pointer register which changes continuously as more buffers are used.
For CRC errored frames, excessive length frames or length field mismatched frames, all of
which are counted in the statistics registers, it is possible that a frame fragment might be stored
in a sequence of receive buffers. Software can detect this by looking for start of frame bit set in a
buffer following a buffer with no end of frame bit set.
For a properly working Ethernet system, there should be no excessively long frames or frames
greater than 128 bytes with CRC/FCS errors. Collision fragments are less than 128 bytes long.
Therefore, it is a rare occurrence to find a frame fragment in a receive buffer.
If bit zero is set when the receive buffer manager reads the location of the receive buffer, then
the buffer has already been used and cannot be used again until software has processed the
frame and cleared bit zero. In this case, the DMA block sets the buffer not available bit in the
receive status register and triggers an interrupt.
If bit zero is set when the receive buffer manager reads the location of the receive buffer and a
frame is being received, the frame is discarded and the receive resource error statistics register
is incremented.
A receive overrun condition occurs when bus was not granted in time or because HRESP was
not OK (bus error). In a receive overrun condition, the receive overrun interrupt is asserted and
the buffer currently being written is recovered. The next frame received with an address that is
recognized reuses the buffer.
If bit 17 of the network configuration register is set, the FCS of received frames shall not be copied to memory. The frame length indicated in the receive status field shall be reduced by four
bytes in this case.
36.4.2.3
Transmit Buffer
Frames to be transmitted are stored in one or more transmit buffers. Transmit buffers can be
between 0 and 2047 bytes long, so it is possible to transmit frames longer than the maximum
length specified in IEEE Standard 802.3. Zero length buffers are allowed. The maximum number
of buffers permitted for each transmit frame is 128.
The start location for each transmit buffer is stored in memory in a list of transmit buffer descriptors at a location pointed to by the transmit buffer queue pointer register. Each list entry consists
of two words, the first being the byte address of the transmit buffer and the second containing
the transmit control and status. Frames can be transmitted with or without automatic CRC generation. If CRC is automatically generated, pad is also automatically generated to take frames to
a minimum length of 64 bytes. Table 36-2 on page 721 defines an entry in the transmit buffer
descriptor list. To transmit frames, the buffer descriptors must be initialized by writing an appropriate byte address to bits 31 to 0 in the first word of each list entry. The second transmit buffer
descriptor is initialized with control information that indicates the length of the buffer, whether or
not it is to be transmitted with CRC and whether the buffer is the last buffer in the frame.
After transmission, the control bits are written back to the second word of the first buffer along
with the “used” bit and other status information. Bit 31 is the “used” bit which must be zero when
720
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
the control word is read if transmission is to happen. It is written to one when a frame has been
transmitted. Bits 27, 28 and 29 indicate various transmit error conditions. Bit 30 is the “wrap” bit
which can be set for any buffer within a frame. If no wrap bit is encountered after 1024 descriptors, the queue pointer rolls over to the start in a similar fashion to the receive queue.
The transmit buffer queue pointer register must not be written while transmit is active. If a new
value is written to the transmit buffer queue pointer register, the queue pointer resets itself to
point to the beginning of the new queue. If transmit is disabled by writing to bit 3 of the network
control, the transmit buffer queue pointer register resets to point to the beginning of the transmit
queue. Note that disabling receive does not have the same effect on the receive queue pointer.
Once the transmit queue is initialized, transmit is activated by writing to bit 9, the Transmit Start
bit of the network control register. Transmit is halted when a buffer descriptor with its used bit set
is read, or if a transmit error occurs, or by writing to the transmit halt bit of the network control
register. (Transmission is suspended if a pause frame is received while the pause enable bit is
set in the network configuration register.) Rewriting the start bit while transmission is active is
allowed.
Transmission control is implemented with a Tx_go variable which is readable in the transmit status register at bit location 3. The Tx_go variable is reset when:
– transmit is disabled
– a buffer descriptor with its ownership bit set is read
– a new value is written to the transmit buffer queue pointer register
– bit 10, tx_halt, of the network control register is written
– there is a transmit error such as too many retries or a transmit underrun.
To set tx_go, write to bit 9, tx_start, of the network control register. Transmit halt does not take
effect until any ongoing transmit finishes. If a collision occurs during transmission of a multi-buffer frame, transmission automatically restarts from the first buffer of the frame. If a “used” bit is
read midway through transmission of a multi-buffer frame, this is treated as a transmit error.
Transmission stops, tx_er is asserted and the FCS is bad.
If transmission stops due to a transmit error, the transmit queue pointer resets to point to the
beginning of the transmit queue. Software needs to re-initialize the transmit queue after a transmit error.
If transmission stops due to a “used” bit being read at the start of the frame, the transmission
queue pointer is not reset and transmit starts from the same transmit buffer descriptor when the
transmit start bit is written
Table 36-2.
Transmit Buffer Descriptor Entry
Bit
Function
Word 0
31:0
Byte Address of buffer
Word 1
31
Used. Needs to be zero for the EMAC to read data from the transmit buffer. The EMAC sets this to one for the first buffer
of a frame once it has been successfully transmitted.
Software has to clear this bit before the buffer can be used again.
Note:
30
This bit is only set for the first buffer in a frame unlike receive where all buffers have the Used bit set once used.
Wrap. Marks last descriptor in transmit buffer descriptor list.
721
6438D–ATARM–13-Oct-09
Table 36-2.
Transmit Buffer Descriptor Entry
Bit
Function
29
Retry limit exceeded, transmit error detected
28
Transmit underrun, occurs either when hresp is not OK (bus error) or the transmit data could not be fetched in time or
when buffers are exhausted in mid frame.
27
Buffers exhausted in mid frame
26:17
Reserved
16
No CRC. When set, no CRC is appended to the current frame. This bit only needs to be set for the last buffer of a frame.
15
Last buffer. When set, this bit indicates the last buffer in the current frame has been reached.
14:11
Reserved
10:0
Length of buffer
36.4.3
Transmit Block
This block transmits frames in accordance with the Ethernet IEEE 802.3 CSMA/CD protocol.
Frame assembly starts by adding preamble and the start frame delimiter. Data is taken from the
transmit FIFO a word at a time. Data is transmitted least significant nibble first. If necessary,
padding is added to increase the frame length to 60 bytes. CRC is calculated as a 32-bit polynomial. This is inverted and appended to the end of the frame, taking the frame length to a
minimum of 64 bytes. If the No CRC bit is set in the second word of the last buffer descriptor of a
transmit frame, neither pad nor CRC are appended.
In full-duplex mode, frames are transmitted immediately. Back-to-back frames are transmitted at
least 96 bit times apart to guarantee the interframe gap.
In half-duplex mode, the transmitter checks carrier sense. If asserted, it waits for it to de-assert
and then starts transmission after the interframe gap of 96 bit times. If the collision signal is
asserted during transmission, the transmitter transmits a jam sequence of 32 bits taken from the
data register and then retry transmission after the back off time has elapsed.
The back-off time is based on an XOR of the 10 least significant bits of the data coming from the
transmit FIFO and a 10-bit pseudo random number generator. The number of bits used depends
on the number of collisions seen. After the first collision, 1 bit is used, after the second 2, and so
on up to 10. Above 10, all 10 bits are used. An error is indicated and no further attempts are
made if 16 attempts cause collisions.
If transmit DMA underruns, bad CRC is automatically appended using the same mechanism as
jam insertion and the tx_er signal is asserted. For a properly configured system, this should
never happen.
If the back pressure bit is set in the network control register in half duplex mode, the transmit
block transmits 64 bits of data, which can consist of 16 nibbles of 1011 or in bit-rate mode 64 1s,
whenever it sees an incoming frame to force a collision. This provides a way of implementing
flow control in half-duplex mode.
722
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
36.4.4
Pause Frame Support
The start of an 802.3 pause frame is as follows:
Table 36-3.
Start of an 802.3 Pause Frame
Destination Address
Source
Address
Type
(Mac Control Frame)
Pause
Opcode
Pause Time
0x0180C2000001
6 bytes
0x8808
0x0001
2 bytes
The network configuration register contains a receive pause enable bit (13). If a valid pause
frame is received, the pause time register is updated with the frame’s pause time, regardless of
its current contents and regardless of the state of the configuration register bit 13. An interrupt
(12) is triggered when a pause frame is received, assuming it is enabled in the interrupt mask
register. If bit 13 is set in the network configuration register and the value of the pause time register is non-zero, no new frame is transmitted until the pause time register has decremented to
zero.
The loading of a new pause time, and hence the pausing of transmission, only occurs when the
EMAC is configured for full-duplex operation. If the EMAC is configured for half-duplex, there is
no transmission pause, but the pause frame received interrupt is still triggered.
A valid pause frame is defined as having a destination address that matches either the address
stored in specific address register 1 or matches 0x0180C2000001 and has the MAC control
frame type ID of 0x8808 and the pause opcode of 0x0001. Pause frames that have FCS or other
errors are treated as invalid and are discarded. Valid pause frames received increment the
Pause Frame Received statistic register.
The pause time register decrements every 512 bit times (i.e., 128 rx_clks in nibble mode)
once transmission has stopped. For test purposes, the register decrements every rx_clk cycle
once transmission has stopped if bit 12 (retry test) is set in the network configuration register. If
the pause enable bit (13) is not set in the network configuration register, then the decrementing
occurs regardless of whether transmission has stopped or not.
An interrupt (13) is asserted whenever the pause time register decrements to zero (assuming it
is enabled in the interrupt mask register).
36.4.5
Receive Block
The receive block checks for valid preamble, FCS, alignment and length, presents received
frames to the DMA block and stores the frames destination address for use by the address
checking block. If, during frame reception, the frame is found to be too long or rx_er is asserted,
a bad frame indication is sent to the DMA block. The DMA block then ceases sending data to
memory. At the end of frame reception, the receive block indicates to the DMA block whether the
frame is good or bad. The DMA block recovers the current receive buffer if the frame was bad.
The receive block signals the register block to increment the alignment error, the CRC (FCS)
error, the short frame, long frame, jabber error, the receive symbol error statistics and the length
field mismatch statistics.
The enable bit for jumbo frames in the network configuration register allows the EMAC to receive
jumbo frames of up to 10240 bytes in size. This operation does not form part of the IEEE802.3
specification and is disabled by default. When jumbo frames are enabled, frames received with a
frame size greater than 10240 bytes are discarded.
723
6438D–ATARM–13-Oct-09
36.4.6
Address Checking Block
The address checking (or filter) block indicates to the DMA block which receive frames should
be copied to memory. Whether a frame is copied depends on what is enabled in the network
configuration register, the state of the external match pin, the contents of the specific address
and hash registers and the frame’s destination address. In this implementation of the EMAC, the
frame’s source address is not checked. Provided that bit 18 of the Network Configuration register is not set, a frame is not copied to memory if the EMAC is transmitting in half duplex mode at
the time a destination address is received. If bit 18 of the Network Configuration register is set,
frames can be received while transmitting in half-duplex mode.
Ethernet frames are transmitted a byte at a time, least significant bit first. The first six bytes (48
bits) of an Ethernet frame make up the destination address. The first bit of the destination
address, the LSB of the first byte of the frame, is the group/individual bit: this is One for multicast
addresses and Zero for unicast. The All Ones address is the broadcast address, and a special
case of multicast.
The EMAC supports recognition of four specific addresses. Each specific address requires two
registers, specific address register bottom and specific address register top. Specific address
register bottom stores the first four bytes of the destination address and specific address register
top contains the last two bytes. The addresses stored can be specific, group, local or universal.
The destination address of received frames is compared against the data stored in the specific
address registers once they have been activated. The addresses are deactivated at reset or
when their corresponding specific address register bottom is written. They are activated when
specific address register top is written. If a receive frame address matches an active address,
the frame is copied to memory.
The following example illustrates the use of the address match registers for a MAC address of
21:43:65:87:A9:CB.
Preamble 55
SFD D5
DA (Octet0 - LSB) 21
DA(Octet 1) 43
DA(Octet 2) 65
DA(Octet 3) 87
DA(Octet 4) A9
DA (Octet5 - MSB) CB
SA (LSB) 00
SA 00
SA 00
SA 00
SA 00
SA (MSB) 43
SA (LSB) 21
724
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
The sequence above shows the beginning of an Ethernet frame. Byte order of transmission is
from top to bottom as shown. For a successful match to specific address 1, the following
address matching registers must be set up:
• Base address + 0x98 0x87654321 (Bottom)
• Base address + 0x9C 0x0000CBA9 (Top)
And for a successful match to the Type ID register, the following should be set up:
• Base address + 0xB8 0x00004321
36.4.7
Broadcast Address
The broadcast address of 0xFFFFFFFFFFFF is recognized if the ‘no broadcast’ bit in the network configuration register is zero.
36.4.8
Hash Addressing
The hash address register is 64 bits long and takes up two locations in the memory map. The
least significant bits are stored in hash register bottom and the most significant bits in hash register top.
The unicast hash enable and the multicast hash enable bits in the network configuration register
enable the reception of hash matched frames. The destination address is reduced to a 6-bit
index into the 64-bit hash register using the following hash function. The hash function is an
exclusive or of every sixth bit of the destination address.
hash_index[5] = da[5] ^ da[11] ^ da[17] ^ da[23] ^ da[29] ^ da[35] ^ da[41] ^ da[47]
hash_index[4] = da[4] ^ da[10] ^ da[16] ^ da[22] ^ da[28] ^ da[34] ^ da[40] ^ da[46]
hash_index[3] = da[3] ^ da[09] ^ da[15] ^ da[21] ^ da[27] ^ da[33] ^ da[39] ^ da[45]
hash_index[2] = da[2] ^ da[08] ^ da[14] ^ da[20] ^ da[26] ^ da[32] ^ da[38] ^ da[44]
hash_index[1] = da[1] ^ da[07] ^ da[13] ^ da[19] ^ da[25] ^ da[31] ^ da[37] ^ da[43]
hash_index[0] = da[0] ^ da[06] ^ da[12] ^ da[18] ^ da[24] ^ da[30] ^ da[36] ^ da[42]
da[0] represents the least significant bit of the first byte received, that is, the multicast/unicast
indicator, and da[47] represents the most significant bit of the last byte received.
If the hash index points to a bit that is set in the hash register, then the frame is matched according to whether the frame is multicast or unicast.
A multicast match is signalled if the multicast hash enable bit is set. da[0] is 1 and the hash index
points to a bit set in the hash register.
A unicast match is signalled if the unicast hash enable bit is set. da[0] is 0 and the hash index
points to a bit set in the hash register.
To receive all multicast frames, the hash register should be set with all ones and the multicast
hash enable bit should be set in the network configuration register.
36.4.9
Copy All Frames (or Promiscuous Mode)
If the copy all frames bit is set in the network configuration register, then all non-errored frames
are copied to memory. For example, frames that are too long, too short, or have FCS errors or
rx_er asserted during reception are discarded and all others are received. Frames with FCS
errors are copied to memory if bit 19 in the network configuration register is set.
725
6438D–ATARM–13-Oct-09
36.4.10
Type ID Checking
The contents of the type_id register are compared against the length/type ID of received frames
(i.e., bytes 13 and 14). Bit 22 in the receive buffer descriptor status is set if there is a match. The
reset state of this register is zero which is unlikely to match the length/type ID of any valid Ethernet frame.
Note:
36.4.11
A type ID match does not affect whether a frame is copied to memory.
VLAN Support
An Ethernet encoded 802.1Q VLAN tag looks like this:
Table 36-4.
802.1Q VLAN Tag
TPID (Tag Protocol Identifier) 16 bits
TCI (Tag Control Information) 16 bits
0x8100
First 3 bits priority, then CFI bit, last 12 bits VID
The VLAN tag is inserted at the 13th byte of the frame, adding an extra four bytes to the frame. If
the VID (VLAN identifier) is null (0x000), this indicates a priority-tagged frame. The MAC can
support frame lengths up to 1536 bytes, 18 bytes more than the original Ethernet maximum
frame length of 1518 bytes. This is achieved by setting bit 8 in the network configuration register.
The following bits in the receive buffer descriptor status word give information about VLAN
tagged frames:
• Bit 21 set if receive frame is VLAN tagged (i.e. type id of 0x8100)
• Bit 20 set if receive frame is priority tagged (i.e. type id of 0x8100 and null VID). (If bit 20 is
set bit 21 is set also.)
• Bit 19, 18 and 17 set to priority if bit 21 is set
• Bit 16 set to CFI if bit 21 is set
36.4.12
PHY Maintenance
The register EMAC_MAN enables the EMAC to communicate with a PHY by means of the MDIO
interface. It is used during auto-negotiation to ensure that the EMAC and the PHY are configured for the same speed and duplex configuration.
The PHY maintenance register is implemented as a shift register. Writing to the register starts a
shift operation which is signalled as complete when bit two is set in the network status register
(about 2000 MCK cycles later when bit ten is set to zero, and bit eleven is set to one in the network configuration register). An interrupt is generated as this bit is set. During this time, the MSB
of the register is output on the MDIO pin and the LSB updated from the MDIO pin with each
MDC cycle. This causes transmission of a PHY management frame on MDIO.
Reading during the shift operation returns the current contents of the shift register. At the end of
management operation, the bits have shifted back to their original locations. For a read operation, the data bits are updated with data read from the PHY. It is important to write the correct
values to the register to ensure a valid PHY management frame is produced.
The MDIO interface can read IEEE 802.3 clause 45 PHYs as well as clause 22 PHYs. To read
clause 45 PHYs, bits[31:28] should be written as 0x0011. For a description of MDC generation,
see the network configuration register in the “Network Control Register” on page 733.
726
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
36.4.13
Media Independent Interface
The Ethernet MAC is capable of interfacing to both RMII and MII Interfaces. The RMII bit in the
EMAC_USRIO register controls the interface that is selected. When this bit is set, the RMII interface is selected, else the MII interface is selected.
The MII and RMII interface are capable of both 10Mb/s and 100Mb/s data rates as described in
the IEEE 802.3u standard. The signals used by the MII and RMII interfaces are described in
Table 36-5.
Table 36-5.
Pin Configuration
Pin Name
ETXCK_EREFCK
MII
RMII
ETXCK: Transmit Clock
EREFCK: Reference Clock
ECRS
ECRS: Carrier Sense
ECOL
ECOL: Collision Detect
ERXDV
ERXDV: Data Valid
ECRSDV: Carrier Sense/Data Valid
ERX0 - ERX3: 4-bit Receive Data
ERX0 - ERX1: 2-bit Receive Data
ERXER
ERXER: Receive Error
ERXER: Receive Error
ERXCK
ERXCK: Receive Clock
ETXEN
ETXEN: Transmit Enable
ETXEN: Transmit Enable
ETX0 - ETX3: 4-bit Transmit Data
ETX0 - ETX1: 2-bit Transmit Data
ERX0 - ERX3
ETX0-ETX3
ETXER
ETXER: Transmit Error
The intent of the RMII is to provide a reduced pin count alternative to the IEEE 802.3u MII. It
uses 2 bits for transmit (ETX0 and ETX1) and two bits for receive (ERX0 and ERX1). There is a
Transmit Enable (ETXEN), a Receive Error (ERXER), a Carrier Sense (ECRS_DV), and a 50
MHz Reference Clock (ETXCK_EREFCK) for 100Mb/s data rate.
36.4.13.1
RMII Transmit and Receive Operation
The same signals are used internally for both the RMII and the MII operations. The RMII maps
these signals in a more pin-efficient manner. The transmit and receive bits are converted from a
4-bit parallel format to a 2-bit parallel scheme that is clocked at twice the rate. The carrier sense
and data valid signals are combined into the ECRSDV signal. This signal contains information
on carrier sense, FIFO status, and validity of the data. Transmit error bit (ETXER) and collision
detect (ECOL) are not used in RMII mode.
727
6438D–ATARM–13-Oct-09
36.5
Programming Interface
36.5.1
36.5.1.1
Initialization
Configuration
Initialization of the EMAC configuration (e.g., loop-back mode, frequency ratios) must be done
while the transmit and receive circuits are disabled. See the description of the network control
register and network configuration register earlier in this document.
To change loop-back mode, the following sequence of operations must be followed:
1. Write to network control register to disable transmit and receive circuits.
2. Write to network control register to change loop-back mode.
3. Write to network control register to re-enable transmit or receive circuits.
Note:
36.5.1.2
These writes to network control register cannot be combined in any way.
Receive Buffer List
Receive data is written to areas of data (i.e., buffers) in system memory. These buffers are listed
in another data structure that also resides in main memory. This data structure (receive buffer
queue) is a sequence of descriptor entries as defined in “Receive Buffer Descriptor Entry” on
page 718. It points to this data structure.
Figure 36-2. Receive Buffer List
Receive Buffer 0
Receive Buffer Queue Pointer
(MAC Register)
Receive Buffer 1
Receive Buffer N
Receive Buffer Descriptor List
(In memory)
(In memory)
To create the list of buffers:
1. Allocate a number (n) of buffers of 128 bytes in system memory.
2. Allocate an area 2n words for the receive buffer descriptor entry in system memory and
create n entries in this list. Mark all entries in this list as owned by EMAC, i.e., bit 0 of
word 0 set to 0.
3. If less than 1024 buffers are defined, the last descriptor must be marked with the wrap
bit (bit 1 in word 0 set to 1).
4. Write address of receive buffer descriptor entry to EMAC register receive_buffer
queue pointer.
5. The receive circuits can then be enabled by writing to the address recognition registers
and then to the network control register.
728
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
36.5.1.3
Transmit Buffer List
Transmit data is read from areas of data (the buffers) in system memory These buffers are listed
in another data structure that also resides in main memory. This data structure (Transmit Buffer
Queue) is a sequence of descriptor entries (as defined in Table 36-2 on page 721) that points to
this data structure.
To create this list of buffers:
1. Allocate a number (n) of buffers of between 1 and 2047 bytes of data to be transmitted
in system memory. Up to 128 buffers per frame are allowed.
2. Allocate an area 2n words for the transmit buffer descriptor entry in system memory
and create N entries in this list. Mark all entries in this list as owned by EMAC, i.e. bit 31
of word 1 set to 0.
3. If fewer than 1024 buffers are defined, the last descriptor must be marked with the wrap
bit — bit 30 in word 1 set to 1.
4. Write address of transmit buffer descriptor entry to EMAC register transmit_buffer
queue pointer.
5. The transmit circuits can then be enabled by writing to the network control register.
36.5.1.4
Address Matching
The EMAC register-pair hash address and the four specific address register-pairs must be written with the required values. Each register-pair comprises a bottom register and top register,
with the bottom register being written first. The address matching is disabled for a particular register-pair after the bottom-register has been written and re-enabled when the top register is
written. See “Address Checking Block” on page 724. for details of address matching. Each register-pair may be written at any time, regardless of whether the receive circuits are enabled or
disabled.
36.5.1.5
Interrupts
There are 14 interrupt conditions that are detected within the EMAC. These are ORed to make a
single interrupt. Depending on the overall system design, this may be passed through a further
level of interrupt collection (interrupt controller). On receipt of the interrupt signal, the CPU
enters the interrupt handler (Refer to the AIC programmer datasheet). To ascertain which interrupt has been generated, read the interrupt status register. Note that this register clears itself
when read. At reset, all interrupts are disabled. To enable an interrupt, write to interrupt enable
register with the pertinent interrupt bit set to 1. To disable an interrupt, write to interrupt disable
register with the pertinent interrupt bit set to 1. To check whether an interrupt is enabled or disabled, read interrupt mask register: if the bit is set to 1, the interrupt is disabled.
36.5.1.6
Transmitting Frames
To set up a frame for transmission:
1. Enable transmit in the network control register.
2. Allocate an area of system memory for transmit data. This does not have to be contiguous, varying byte lengths can be used as long as they conclude on byte borders.
3. Set-up the transmit buffer list.
4. Set the network control register to enable transmission and enable interrupts.
5. Write data for transmission into these buffers.
6. Write the address to transmit buffer descriptor queue pointer.
7. Write control and length to word one of the transmit buffer descriptor entry.
729
6438D–ATARM–13-Oct-09
8. Write to the transmit start bit in the network control register.
36.5.1.7
Receiving Frames
When a frame is received and the receive circuits are enabled, the EMAC checks the address
and, in the following cases, the frame is written to system memory:
• if it matches one of the four specific address registers.
• if it matches the hash address function.
• if it is a broadcast address (0xFFFFFFFFFFFF) and broadcasts are allowed.
• if the EMAC is configured to copy all frames.
The register receive buffer queue pointer points to the next entry (see Table 36-1 on page 718)
and the EMAC uses this as the address in system memory to write the frame to. Once the frame
has been completely and successfully received and written to system memory, the EMAC then
updates the receive buffer descriptor entry with the reason for the address match and marks the
area as being owned by software. Once this is complete an interrupt receive complete is set.
Software is then responsible for handling the data in the buffer and then releasing the buffer by
writing the ownership bit back to 0.
If the EMAC is unable to write the data at a rate to match the incoming frame, then an interrupt
receive overrun is set. If there is no receive buffer available, i.e., the next buffer is still owned by
software, the interrupt receive buffer not available is set. If the frame is not successfully
received, a statistic register is incremented and the frame is discarded without informing
software.
730
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
36.6
Ethernet MAC 10/100 (EMAC) User Interface
Table 36-6.
Register Mapping
Offset
Register
Name
Access
Reset
0x00
Network Control Register
EMAC_NCR
Read-write
0
0x04
Network Configuration Register
EMAC_NCFG
Read-write
0x800
0x08
Network Status Register
EMAC_NSR
Read-only
-
0x0C
Reserved
0x10
Reserved
0x14
Transmit Status Register
EMAC_TSR
Read-write
0x0000_0000
0x18
Receive Buffer Queue Pointer Register
EMAC_RBQP
Read-write
0x0000_0000
0x1C
Transmit Buffer Queue Pointer Register
EMAC_TBQP
Read-write
0x0000_0000
0x20
Receive Status Register
EMAC_RSR
Read-write
0x0000_0000
0x24
Interrupt Status Register
EMAC_ISR
Read-write
0x0000_0000
0x28
Interrupt Enable Register
EMAC_IER
Write-only
-
0x2C
Interrupt Disable Register
EMAC_IDR
Write-only
-
0x30
Interrupt Mask Register
EMAC_IMR
Read-only
0x0000_3FFF
0x34
Phy Maintenance Register
EMAC_MAN
Read-write
0x0000_0000
0x38
Pause Time Register
EMAC_PTR
Read-write
0x0000_0000
0x3C
Pause Frames Received Register
EMAC_PFR
Read-write
0x0000_0000
0x40
Frames Transmitted Ok Register
EMAC_FTO
Read-write
0x0000_0000
0x44
Single Collision Frames Register
EMAC_SCF
Read-write
0x0000_0000
0x48
Multiple Collision Frames Register
EMAC_MCF
Read-write
0x0000_0000
0x4C
Frames Received Ok Register
EMAC_FRO
Read-write
0x0000_0000
0x50
Frame Check Sequence Errors Register
EMAC_FCSE
Read-write
0x0000_0000
0x54
Alignment Errors Register
EMAC_ALE
Read-write
0x0000_0000
0x58
Deferred Transmission Frames Register
EMAC_DTF
Read-write
0x0000_0000
0x5C
Late Collisions Register
EMAC_LCOL
Read-write
0x0000_0000
0x60
Excessive Collisions Register
EMAC_ECOL
Read-write
0x0000_0000
0x64
Transmit Underrun Errors Register
EMAC_TUND
Read-write
0x0000_0000
0x68
Carrier Sense Errors Register
EMAC_CSE
Read-write
0x0000_0000
0x6C
Receive Resource Errors Register
EMAC_RRE
Read-write
0x0000_0000
0x70
Receive Overrun Errors Register
EMAC_ROV
Read-write
0x0000_0000
0x74
Receive Symbol Errors Register
EMAC_RSE
Read-write
0x0000_0000
0x78
Excessive Length Errors Register
EMAC_ELE
Read-write
0x0000_0000
0x7C
Receive Jabbers Register
EMAC_RJA
Read-write
0x0000_0000
0x80
Undersize Frames Register
EMAC_USF
Read-write
0x0000_0000
0x84
SQE Test Errors Register
EMAC_STE
Read-write
0x0000_0000
0x88
Received Length Field Mismatch Register
EMAC_RLE
Read-write
0x0000_0000
731
6438D–ATARM–13-Oct-09
Table 36-6.
Register Mapping (Continued)
Offset
Register
Name
Access
Reset
0x90
Hash Register Bottom [31:0] Register
EMAC_HRB
Read-write
0x0000_0000
0x94
Hash Register Top [63:32] Register
EMAC_HRT
Read-write
0x0000_0000
0x98
Specific Address 1 Bottom Register
EMAC_SA1B
Read-write
0x0000_0000
0x9C
Specific Address 1 Top Register
EMAC_SA1T
Read-write
0x0000_0000
0xA0
Specific Address 2 Bottom Register
EMAC_SA2B
Read-write
0x0000_0000
0xA4
Specific Address 2 Top Register
EMAC_SA2T
Read-write
0x0000_0000
0xA8
Specific Address 3 Bottom Register
EMAC_SA3B
Read-write
0x0000_0000
0xAC
Specific Address 3 Top Register
EMAC_SA3T
Read-write
0x0000_0000
0xB0
Specific Address 4 Bottom Register
EMAC_SA4B
Read-write
0x0000_0000
0xB4
Specific Address 4 Top Register
EMAC_SA4T
Read-write
0x0000_0000
0xB8
Type ID Checking Register
EMAC_TID
Read-write
0x0000_0000
0xC0
User Input/Output Register
EMAC_USRIO
Read-write
0x0000_0000
0xC8 - 0xFC
Reserved
–
–
–
732
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
36.6.1
Name:
Network Control Register
EMAC_NCR
Address:
0xFFFBC000
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
THALT
9
TSTART
8
BP
7
WESTAT
6
INCSTAT
5
CLRSTAT
4
MPE
3
TE
2
RE
1
LLB
0
LB
• LB: LoopBack
Asserts the loopback signal to the PHY.
• LLB: Loopback local
Connects txd to rxd, tx_en to rx_dv, forces full duplex and drives rx_clk and tx_clk with pclk divided by 4.
rx_clk and tx_clk may glitch as the EMAC is switched into and out of internal loop back. It is important that receive
and transmit circuits have already been disabled when making the switch into and out of internal loop back.
• RE: Receive enable
When set, enables the EMAC to receive data. When reset, frame reception stops immediately and the receive FIFO is
cleared. The receive queue pointer register is unaffected.
• TE: Transmit enable
When set, enables the Ethernet transmitter to send data. When reset transmission, stops immediately, the transmit FIFO
and control registers are cleared and the transmit queue pointer register resets to point to the start of the transmit descriptor list.
• MPE: Management port enable
Set to one to enable the management port. When zero, forces MDIO to high impedance state and MDC low.
• CLRSTAT: Clear statistics registers
This bit is write only. Writing a one clears the statistics registers.
• INCSTAT: Increment statistics registers
This bit is write only. Writing a one increments all the statistics registers by one for test purposes.
• WESTAT: Write enable for statistics registers
Setting this bit to one makes the statistics registers writable for functional test purposes.
• BP: Back pressure
If set in half duplex mode, forces collisions on all received frames.
733
6438D–ATARM–13-Oct-09
• TSTART: Start transmission
Writing one to this bit starts transmission.
• THALT: Transmit halt
Writing one to this bit halts transmission as soon as any ongoing frame transmission ends.
734
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
36.6.2
Name:
Network Configuration Register
EMAC_NCFG
Address:
0xFFFBC004
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
IRXFCS
18
EFRHD
17
DRFCS
16
RLCE
14
13
PAE
12
RTY
11
10
9
–
8
BIG
5
NBC
4
CAF
3
JFRAME
2
–
1
FD
0
SPD
15
RBOF
7
UNI
6
MTI
CLK
• SPD: Speed
Set to 1 to indicate 100 Mbit/s operation, 0 for 10 Mbit/s. The value of this pin is reflected on the speed pin.
• FD: Full Duplex
If set to 1, the transmit block ignores the state of collision and carrier sense and allows receive while transmitting. Also controls the half_duplex pin.
• CAF: Copy All Frames
When set to 1, all valid frames are received.
• JFRAME: Jumbo Frames
Set to one to enable jumbo frames of up to 10240 bytes to be accepted.
• NBC: No Broadcast
When set to 1, frames addressed to the broadcast address of all ones are not received.
• MTI: Multicast Hash Enable
When set, multicast frames are received when the 6-bit hash function of the destination address points to a bit that is set in
the hash register.
• UNI: Unicast Hash Enable
When set, unicast frames are received when the 6-bit hash function of the destination address points to a bit that is set in
the hash register.
• BIG: Receive 1536 bytes frames
Setting this bit means the EMAC receives frames up to 1536 bytes in length. Normally, the EMAC would reject any frame
above 1518 bytes.
• CLK: MDC clock divider
Set according to system clock speed. This determines by what number system clock is divided to generate MDC. For conformance with 802.3, MDC must not exceed 2.5MHz (MDC is only active during MDIO read and write operations)
735
6438D–ATARM–13-Oct-09
.
CLK
MDC
00
MCK divided by 8 (MCK up to 20 MHz)
01
MCK divided by 16 (MCK up to 40 MHz)
10
MCK divided by 32 (MCK up to 80 MHz)
11
MCK divided by 64 (MCK up to 160 MHz)
• RTY: Retry test
Must be set to zero for normal operation. If set to one, the back off between collisions is always one slot time. Setting this
bit to one helps testing the too many retries condition. Also used in the pause frame tests to reduce the pause counters
decrement time from 512 bit times, to every rx_clk cycle.
• PAE: Pause Enable
When set, transmission pauses when a valid pause frame is received.
• RBOF: Receive Buffer Offset
Indicates the number of bytes by which the received data is offset from the start of the first receive buffer.
RBOF
Offset
00
No offset from start of receive buffer
01
One-byte offset from start of receive buffer
10
Two-byte offset from start of receive buffer
11
Three-byte offset from start of receive buffer
• RLCE: Receive Length field Checking Enable
When set, frames with measured lengths shorter than their length fields are discarded. Frames containing a type ID in
bytes 13 and 14 — length/type ID = 0600 — are not be counted as length errors.
• DRFCS: Discard Receive FCS
When set, the FCS field of received frames are not be copied to memory.
• EFRHD:
Enable Frames to be received in half-duplex mode while transmitting.
• IRXFCS: Ignore RX FCS
When set, frames with FCS/CRC errors are not rejected and no FCS error statistics are counted. For normal operation, this
bit must be set to 0.
736
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
36.6.3
Name:
Network Status Register
EMAC_NSR
Address:
0xFFFBC008
Access:
Read-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
–
6
–
5
–
4
–
3
–
2
IDLE
1
MDIO
0
–
• MDIO
Returns status of the mdio_in pin. Use the PHY maintenance register for reading managed frames rather than this bit.
• IDLE
0 = The PHY logic is running.
1 = The PHY management logic is idle (i.e., has completed).
737
6438D–ATARM–13-Oct-09
36.6.4
Name:
Transmit Status Register
EMAC_TSR
Address:
0xFFFBC014
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
–
6
UND
5
COMP
4
BEX
3
TGO
2
RLE
1
COL
0
UBR
This register, when read, provides details of the status of a transmit. Once read, individual bits may be cleared by writing 1
to them. It is not possible to set a bit to 1 by writing to the register.
• UBR: Used Bit Read
Set when a transmit buffer descriptor is read with its used bit set. Cleared by writing a one to this bit.
• COL: Collision Occurred
Set by the assertion of collision. Cleared by writing a one to this bit.
• RLE: Retry Limit exceeded
Cleared by writing a one to this bit.
• TGO: Transmit Go
If high transmit is active.
• BEX: Buffers exhausted mid frame
If the buffers run out during transmission of a frame, then transmission stops, FCS shall be bad and tx_er asserted. Cleared
by writing a one to this bit.
• COMP: Transmit Complete
Set when a frame has been transmitted. Cleared by writing a one to this bit.
• UND: Transmit Underrun
Set when transmit DMA was not able to read data from memory, either because the bus was not granted in time, because
a not OK hresp(bus error) was returned or because a used bit was read midway through frame transmission. If this
occurs, the transmitter forces bad CRC. Cleared by writing a one to this bit.
738
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
36.6.5
Name:
Receive Buffer Queue Pointer Register
EMAC_RBQP
Address:
0xFFFBC018
Access:
Read-write
31
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
–
0
–
ADDR
23
22
21
20
ADDR
15
14
13
12
ADDR
7
6
5
4
ADDR
This register points to the entry in the receive buffer queue (descriptor list) currently being used. It is written with the start
location of the receive buffer descriptor list. The lower order bits increment as buffers are used up and wrap to their original
values after either 1024 buffers or when the wrap bit of the entry is set.
Reading this register returns the location of the descriptor currently being accessed. This value increments as buffers are
used. Software should not use this register for determining where to remove received frames from the queue as it constantly changes as new frames are received. Software should instead work its way through the buffer descriptor queue
checking the used bits.
Receive buffer writes also comprise bursts of two words and, as with transmit buffer reads, it is recommended that bit 2 is
always written with zero to prevent a burst crossing a 1K boundary, in violation of section 3.6 of the AMBA specification.
• ADDR: Receive buffer queue pointer address
Written with the address of the start of the receive queue, reads as a pointer to the current buffer being used.
739
6438D–ATARM–13-Oct-09
36.6.6
Name:
Transmit Buffer Queue Pointer Register
EMAC_TBQP
Address:
0xFFFBC01C
Access:
Read-write
31
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
–
0
–
ADDR
23
22
21
20
ADDR
15
14
13
12
ADDR
7
6
5
4
ADDR
This register points to the entry in the transmit buffer queue (descriptor list) currently being used. It is written with the start
location of the transmit buffer descriptor list. The lower order bits increment as buffers are used up and wrap to their original
values after either 1024 buffers or when the wrap bit of the entry is set. This register can only be written when bit 3 in the
transmit status register is low.
As transmit buffer reads consist of bursts of two words, it is recommended that bit 2 is always written with zero to prevent a
burst crossing a 1K boundary, in violation of section 3.6 of the AMBA specification.
• ADDR: Transmit buffer queue pointer address
Written with the address of the start of the transmit queue, reads as a pointer to the first buffer of the frame being transmitted or about to be transmitted.
740
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
36.6.7
Name:
Receive Status Register
EMAC_RSR
Address:
0xFFFBC020
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
–
6
–
5
–
4
–
3
–
2
OVR
1
REC
0
BNA
This register, when read, provides details of the status of a receive. Once read, individual bits may be cleared by writing 1
to them. It is not possible to set a bit to 1 by writing to the register.
• BNA: Buffer Not Available
An attempt was made to get a new buffer and the pointer indicated that it was owned by the processor. The DMA rereads
the pointer each time a new frame starts until a valid pointer is found. This bit is set at each attempt that fails even if it has
not had a successful pointer read since it has been cleared.
Cleared by writing a one to this bit.
• REC: Frame Received
One or more frames have been received and placed in memory. Cleared by writing a one to this bit.
• OVR: Receive Overrun
The DMA block was unable to store the receive frame to memory, either because the bus was not granted in time or
because a not OK hresp(bus error) was returned. The buffer is recovered if this happens.
Cleared by writing a one to this bit.
741
6438D–ATARM–13-Oct-09
36.6.8
Name:
Interrupt Status Register
EMAC_ISR
Address:
0xFFFBC024
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
PTZ
12
PFR
11
HRESP
10
ROVR
9
–
8
–
7
TCOMP
6
TXERR
5
RLE
4
TUND
3
TXUBR
2
RXUBR
1
RCOMP
0
MFD
• MFD: Management Frame Done
The PHY maintenance register has completed its operation. Cleared on read.
• RCOMP: Receive Complete
A frame has been stored in memory. Cleared on read.
• RXUBR: Receive Used Bit Read
Set when a receive buffer descriptor is read with its used bit set. Cleared on read.
• TXUBR: Transmit Used Bit Read
Set when a transmit buffer descriptor is read with its used bit set. Cleared on read.
• TUND: Ethernet Transmit Buffer Underrun
The transmit DMA did not fetch frame data in time for it to be transmitted or hresp returned not OK. Also set if a used bit
is read mid-frame or when a new transmit queue pointer is written. Cleared on read.
• RLE: Retry Limit Exceeded
Cleared on read.
• TXERR: Transmit Error
Transmit buffers exhausted in mid-frame - transmit error. Cleared on read.
• TCOMP: Transmit Complete
Set when a frame has been transmitted. Cleared on read.
• ROVR: Receive Overrun
Set when the receive overrun status bit gets set. Cleared on read.
• HRESP: Hresp not OK
Set when the DMA block sees a bus error. Cleared on read.
• PFR: Pause Frame Received
Indicates a valid pause has been received. Cleared on a read.
• PTZ: Pause Time Zero
Set when the pause time register, 0x38 decrements to zero. Cleared on a read.
742
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
36.6.9
Name:
Interrupt Enable Register
EMAC_IER
Address:
0xFFFBC028
Access:
Write-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
PTZ
12
PFR
11
HRESP
10
ROVR
9
–
8
–
7
TCOMP
6
TXERR
5
RLE
4
TUND
3
TXUBR
2
RXUBR
1
RCOMP
0
MFD
• MFD: Management Frame sent
Enable management done interrupt.
• RCOMP: Receive Complete
Enable receive complete interrupt.
• RXUBR: Receive Used Bit Read
Enable receive used bit read interrupt.
• TXUBR: Transmit Used Bit Read
Enable transmit used bit read interrupt.
• TUND: Ethernet Transmit Buffer Underrun
Enable transmit underrun interrupt.
• RLE: Retry Limit Exceeded
Enable retry limit exceeded interrupt.
• TXERR
Enable transmit buffers exhausted in mid-frame interrupt.
• TCOMP: Transmit Complete
Enable transmit complete interrupt.
• ROVR: Receive Overrun
Enable receive overrun interrupt.
• HRESP: Hresp not OK
Enable Hresp not OK interrupt.
• PFR: Pause Frame Received
Enable pause frame received interrupt.
• PTZ: Pause Time Zero
Enable pause time zero interrupt.
743
6438D–ATARM–13-Oct-09
36.6.10
Name:
Interrupt Disable Register
EMAC_IDR
Address:
0xFFFBC02C
Access:
Write-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
PTZ
12
PFR
11
HRESP
10
ROVR
9
–
8
–
7
TCOMP
6
TXERR
5
RLE
4
TUND
3
TXUBR
2
RXUBR
1
RCOMP
0
MFD
• MFD: Management Frame sent
Disable management done interrupt.
• RCOMP: Receive Complete
Disable receive complete interrupt.
• RXUBR: Receive Used Bit Read
Disable receive used bit read interrupt.
• TXUBR: Transmit Used Bit Read
Disable transmit used bit read interrupt.
• TUND: Ethernet Transmit Buffer Underrun
Disable transmit underrun interrupt.
• RLE: Retry Limit Exceeded
Disable retry limit exceeded interrupt.
• TXERR
Disable transmit buffers exhausted in mid-frame interrupt.
• TCOMP: Transmit Complete
Disable transmit complete interrupt.
• ROVR: Receive Overrun
Disable receive overrun interrupt.
• HRESP: Hresp not OK
Disable Hresp not OK interrupt.
• PFR: Pause Frame Received
Disable pause frame received interrupt.
• PTZ: Pause Time Zero
Disable pause time zero interrupt.
744
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
36.6.11
Name:
Interrupt Mask Register
EMAC_IMR
Address:
0xFFFBC030
Access:
Read-only
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
PTZ
12
PFR
11
HRESP
10
ROVR
9
–
8
–
7
TCOMP
6
TXERR
5
RLE
4
TUND
3
TXUBR
2
RXUBR
1
RCOMP
0
MFD
• MFD: Management Frame sent
Management done interrupt masked.
• RCOMP: Receive Complete
Receive complete interrupt masked.
• RXUBR: Receive Used Bit Read
Receive used bit read interrupt masked.
• TXUBR: Transmit Used Bit Read
Transmit used bit read interrupt masked.
• TUND: Ethernet Transmit Buffer Underrun
Transmit underrun interrupt masked.
• RLE: Retry Limit Exceeded
Retry limit exceeded interrupt masked.
• TXERR
Transmit buffers exhausted in mid-frame interrupt masked.
• TCOMP: Transmit Complete
Transmit complete interrupt masked.
• ROVR: Receive Overrun
Receive overrun interrupt masked.
• HRESP: Hresp not OK
Hresp not OK interrupt masked.
• PFR: Pause Frame Received
Pause frame received interrupt masked.
• PTZ: Pause Time Zero
Pause time zero interrupt masked.
745
6438D–ATARM–13-Oct-09
36.6.12
Name:
PHY Maintenance Register
EMAC_MAN
Address:
0xFFFBC034
Access:
Read-write
31
30
29
SOF
28
27
26
RW
23
PHYA
22
15
14
21
13
25
24
PHYA
20
REGA
19
18
17
16
CODE
12
11
10
9
8
3
2
1
0
DATA
7
6
5
4
DATA
• DATA
For a write operation this is written with the data to be written to the PHY.
After a read operation this contains the data read from the PHY.
• CODE:
Must be written to 10. Reads as written.
• REGA: Register Address
Specifies the register in the PHY to access.
• PHYA: PHY Address
• RW: Read-write
10 is read; 01 is write. Any other value is an invalid PHY management frame
• SOF: Start of frame
Must be written 01 for a valid frame.
746
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
36.6.13
Name:
Pause Time Register
EMAC_PTR
Address:
0xFFFBC038
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
14
13
12
11
10
9
8
3
2
1
0
PTIME
7
6
5
4
PTIME
• PTIME: Pause Time
Stores the current value of the pause time register which is decremented every 512 bit times.
747
6438D–ATARM–13-Oct-09
36.6.14
Name:
Hash Register Bottom
EMAC_HRB
Address:
0xFFFBC090
Access:
Read-write
31
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
0
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
0
ADDR
23
22
21
20
ADDR
15
14
13
12
ADDR
7
6
5
4
ADDR
• ADDR:
Bits 31:0 of the hash address register. See “Hash Addressing” on page 725.
36.6.15
Name:
Hash Register Top
EMAC_HRT
Address:
0xFFFBC094
Access:
Read-write
31
30
29
28
ADDR
23
22
21
20
ADDR
15
14
13
12
ADDR
7
6
5
4
ADDR
• ADDR:
Bits 63:32 of the hash address register. See “Hash Addressing” on page 725.
748
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
36.6.16
Name:
Specific Address 1 Bottom Register
EMAC_SA1B
Address:
0xFFFBC098
Access:
Read-write
31
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
0
ADDR
23
22
21
20
ADDR
15
14
13
12
ADDR
7
6
5
4
ADDR
• ADDR
Least significant bits of the destination address. Bit zero indicates whether the address is multicast or unicast and corresponds to the least significant bit of the first byte received.
36.6.17
Name:
Specific Address 1 Top Register
EMAC_SA1T
Address:
0xFFFBC09C
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
14
13
12
11
10
9
8
3
2
1
0
ADDR
7
6
5
4
ADDR
• ADDR
The most significant bits of the destination address, that is bits 47 to 32.
749
6438D–ATARM–13-Oct-09
36.6.18
Name:
Specific Address 2 Bottom Register
EMAC_SA2B
Address:
0xFFFBC0A0
Access:
Read-write
31
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
0
ADDR
23
22
21
20
ADDR
15
14
13
12
ADDR
7
6
5
4
ADDR
• ADDR
Least significant bits of the destination address. Bit zero indicates whether the address is multicast or unicast and corresponds to the least significant bit of the first byte received.
36.6.19
Name:
Specific Address 2 Top Register
EMAC_SA2T
Address:
0xFFFBC0A4
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
14
13
12
11
10
9
8
3
2
1
0
ADDR
7
6
5
4
ADDR
• ADDR
The most significant bits of the destination address, that is bits 47 to 32.
750
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
36.6.20
Name:
Specific Address 3 Bottom Register
EMAC_SA3B
Address:
0xFFFBC0A8
Access:
Read-write
31
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
0
ADDR
23
22
21
20
ADDR
15
14
13
12
ADDR
7
6
5
4
ADDR
• ADDR
Least significant bits of the destination address. Bit zero indicates whether the address is multicast or unicast and corresponds to the least significant bit of the first byte received.
36.6.21
Name:
Specific Address 3 Top Register
EMAC_SA3T
Address:
0xFFFBC0AC
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
14
13
12
11
10
9
8
3
2
1
0
ADDR
7
6
5
4
ADDR
• ADDR
The most significant bits of the destination address, that is bits 47 to 32.
751
6438D–ATARM–13-Oct-09
36.6.22
Name:
Specific Address 4 Bottom Register
EMAC_SA4B
Address:
0xFFFBC0B0
Access:
Read-write
31
30
29
28
27
26
25
24
19
18
17
16
11
10
9
8
3
2
1
0
ADDR
23
22
21
20
ADDR
15
14
13
12
ADDR
7
6
5
4
ADDR
• ADDR
Least significant bits of the destination address. Bit zero indicates whether the address is multicast or unicast and corresponds to the least significant bit of the first byte received.
36.6.23
Name:
Specific Address 4 Top Register
EMAC_SA4T
Address:
0xFFFBC0B4
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
14
13
12
11
10
9
8
3
2
1
0
ADDR
7
6
5
4
ADDR
• ADDR
The most significant bits of the destination address, that is bits 47 to 32.
752
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
36.6.24
Name:
Type ID Checking Register
EMAC_TID
Address:
0xFFFBC0B8
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
14
13
12
11
10
9
8
3
2
1
0
TID
7
6
5
4
TID
• TID: Type ID checking
For use in comparisons with received frames TypeID/Length field.
753
6438D–ATARM–13-Oct-09
36.6.25
Name:
User Input/Output Register
EMAC_USRIO
Address:
0xFFFBC0C0
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
–
6
–
5
–
4
–
3
–
2
–
1
CLKEN
0
RMII
• RMII
When set, this bit enables the RMII operation mode. When reset, it selects the MII mode.
• CLKEN
When set, this bit enables the transceiver input clock.
Setting this bit to 0 reduces power consumption when the treasurer is not used.
754
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
36.6.26 EMAC Statistic Registers
These registers reset to zero on a read and stick at all ones when they count to their maximum value. They should be read
frequently enough to prevent loss of data. The receive statistics registers are only incremented when the receive enable bit
is set in the network control register. To write to these registers, bit 7 must be set in the network control register. The statistics register block contains the following registers.
36.6.26.1
Name:
Pause Frames Received Register
EMAC_PFR
Address:
0xFFFBC03C
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
14
13
12
11
10
9
8
3
2
1
0
FROK
7
6
5
4
FROK
• FROK: Pause Frames received OK
A 16-bit register counting the number of good pause frames received. A good frame has a length of 64 to 1518 (1536 if bit
8 set in network configuration register) and has no FCS, alignment or receive symbol errors.
36.6.26.2
Name:
Frames Transmitted OK Register
EMAC_FTO
Address:
0xFFFBC040
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
22
21
20
19
18
17
16
11
10
9
8
3
2
1
0
FTOK
15
14
13
12
FTOK
7
6
5
4
FTOK
• FTOK: Frames Transmitted OK
A 24-bit register counting the number of frames successfully transmitted, i.e., no underrun and not too many retries.
755
6438D–ATARM–13-Oct-09
36.6.26.3
Name:
Single Collision Frames Register
EMAC_SCF
Address:
0xFFFBC044
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
14
13
12
11
10
9
8
3
2
1
0
SCF
7
6
5
4
SCF
• SCF: Single Collision Frames
A 16-bit register counting the number of frames experiencing a single collision before being successfully transmitted, i.e.,
no underrun.
36.6.26.4
Name:
Multicollision Frames Register
EMAC_MCF
Address:
0xFFFBC048
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
14
13
12
11
10
9
8
3
2
1
0
MCF
7
6
5
4
MCF
• MCF: Multicollision Frames
A 16-bit register counting the number of frames experiencing between two and fifteen collisions prior to being successfully
transmitted, i.e., no underrun and not too many retries.
756
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
36.6.26.5
Name:
Frames Received OK Register
EMAC_FRO
Address:
0xFFFBC04C
Access:
Read-write
31
–
30
–
29
–
28
–
23
22
21
20
27
–
26
–
25
–
24
–
19
18
17
16
11
10
9
8
3
2
1
0
FROK
15
14
13
12
FROK
7
6
5
4
FROK
• FROK: Frames Received OK
A 24-bit register counting the number of good frames received, i.e., address recognized and successfully copied to memory. A good frame is of length 64 to 1518 bytes (1536 if bit 8 set in network configuration register) and has no FCS,
alignment or receive symbol errors.
36.6.26.6
Name:
Frames Check Sequence Errors Register
EMAC_FCSE
Address:
0xFFFBC050
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
6
5
4
3
2
1
0
FCSE
• FCSE: Frame Check Sequence Errors
An 8-bit register counting frames that are an integral number of bytes, have bad CRC and are between 64 and 1518 bytes
in length (1536 if bit 8 set in network configuration register). This register is also incremented if a symbol error is detected
and the frame is of valid length and has an integral number of bytes.
757
6438D–ATARM–13-Oct-09
36.6.26.7
Name:
Alignment Errors Register
EMAC_ALE
Address:
0xFFFBC054
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
6
5
4
3
2
1
0
ALE
• ALE: Alignment Errors
An 8-bit register counting frames that are not an integral number of bytes long and have bad CRC when their length is truncated to an integral number of bytes and are between 64 and 1518 bytes in length (1536 if bit 8 set in network configuration
register). This register is also incremented if a symbol error is detected and the frame is of valid length and does not have
an integral number of bytes.
36.6.26.8
Name:
Deferred Transmission Frames Register
EMAC_DTF
Address:
0xFFFBC058
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
14
13
12
11
10
9
8
3
2
1
0
DTF
7
6
5
4
DTF
• DTF: Deferred Transmission Frames
A 16-bit register counting the number of frames experiencing deferral due to carrier sense being active on their first attempt
at transmission. Frames involved in any collision are not counted nor are frames that experienced a transmit underrun.
758
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
36.6.26.9
Name:
Late Collisions Register
EMAC_LCOL
Address:
0xFFFBC05C
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
6
5
4
3
2
1
0
LCOL
• LCOL: Late Collisions
An 8-bit register counting the number of frames that experience a collision after the slot time (512 bits) has expired. A late
collision is counted twice; i.e., both as a collision and a late collision.
36.6.26.10
Name:
Excessive Collisions Register
EMAC_ECOL
Address:
0xFFFBC060
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
6
5
4
3
2
1
0
EXCOL
• EXCOL: Excessive Collisions
An 8-bit register counting the number of frames that failed to be transmitted because they experienced 16 collisions.
759
6438D–ATARM–13-Oct-09
36.6.26.11
Name:
Transmit Underrun Errors Register
EMAC_TUND
Address:
0xFFFBC064
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
6
5
4
3
2
1
0
TUND
• TUND: Transmit Underruns
An 8-bit register counting the number of frames not transmitted due to a transmit DMA underrun. If this register is incremented, then no other statistics register is incremented.
36.6.26.12
Name:
Carrier Sense Errors Register
EMAC_CSE
Address:
0xFFFBC068
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
6
5
4
3
2
1
0
CSE
• CSE: Carrier Sense Errors
An 8-bit register counting the number of frames transmitted where carrier sense was not seen during transmission or where
carrier sense was deasserted after being asserted in a transmit frame without collision (no underrun). Only incremented in
half-duplex mode. The only effect of a carrier sense error is to increment this register. The behavior of the other statistics
registers is unaffected by the detection of a carrier sense error.
760
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
36.6.26.13
Name:
Receive Resource Errors Register
EMAC_RRE
Address:
0xFFFBC06C
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
14
13
12
11
10
9
8
3
2
1
0
RRE
7
6
5
4
RRE
• RRE: Receive Resource Errors
A 16-bit register counting the number of frames that were address matched but could not be copied to memory because no
receive buffer was available.
36.6.26.14
Name:
Receive Overrun Errors Register
EMAC_ROV
Address:
0xFFFBC070
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
6
5
4
3
2
1
0
ROVR
• ROVR: Receive Overrun
An 8-bit register counting the number of frames that are address recognized but were not copied to memory due to a
receive DMA overrun.
761
6438D–ATARM–13-Oct-09
36.6.26.15
Name:
Receive Symbol Errors Register
EMAC_RSE
Address:
0xFFFBC074
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
6
5
4
3
2
1
0
RSE
• RSE: Receive Symbol Errors
An 8-bit register counting the number of frames that had rx_er asserted during reception. Receive symbol errors are also
counted as an FCS or alignment error if the frame is between 64 and 1518 bytes in length (1536 if bit 8 is set in the network
configuration register). If the frame is larger, it is recorded as a jabber error.
36.6.26.16
Name:
Excessive Length Errors Register
EMAC_ELE
Address:
0xFFFBC078
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
6
5
4
3
2
1
0
EXL
• EXL: Excessive Length Errors
An 8-bit register counting the number of frames received exceeding 1518 bytes (1536 if bit 8 set in network configuration
register) in length but do not have either a CRC error, an alignment error nor a receive symbol error.
762
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
36.6.26.17
Name:
Receive Jabbers Register
EMAC_RJA
Address:
0xFFFBC07C
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
6
5
4
3
2
1
0
RJB
• RJB: Receive Jabbers
An 8-bit register counting the number of frames received exceeding 1518 bytes (1536 if bit 8 set in network configuration
register) in length and have either a CRC error, an alignment error or a receive symbol error.
36.6.26.18
Name:
Undersize Frames Register
EMAC_USF
Address:
0xFFFBC080
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
6
5
4
3
2
1
0
USF
• USF: Undersize frames
An 8-bit register counting the number of frames received less than 64 bytes in length but do not have either a CRC error, an
alignment error or a receive symbol error.
763
6438D–ATARM–13-Oct-09
36.6.26.19
Name:
SQE Test Errors Register
EMAC_STE
Address:
0xFFFBC084
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
6
5
4
3
2
1
0
SQER
• SQER: SQE test errors
An 8-bit register counting the number of frames where col was not asserted within 96 bit times (an interframe gap) of
tx_en being deasserted in half duplex mode.
36.6.26.20
Name:
Received Length Field Mismatch Register
EMAC_RLE
Address:
0xFFFBC088
Access:
Read-write
31
–
30
–
29
–
28
–
27
–
26
–
25
–
24
–
23
–
22
–
21
–
20
–
19
–
18
–
17
–
16
–
15
–
14
–
13
–
12
–
11
–
10
–
9
–
8
–
7
6
5
4
3
2
1
0
RLFM
• RLFM: Receive Length Field Mismatch
An 8-bit register counting the number of frames received that have a measured length shorter than that extracted from its
length field. Checking is enabled through bit 16 of the network configuration register. Frames containing a type ID in bytes
13 and 14 (i.e., length/type ID ≥ 0x0600) are not counted as length field errors, neither are excessive length frames.
764
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
37. USB High Speed Host Port (UHPHS)
37.1
Description
The USB High Speed Host Port (UHPHS) interfaces the USB with the host application. It handles Open HCI protocol (Open Host Controller Interface) as well as Enhanced HCI protocol
(Enhanced Host Controller Interface).
37.2
Embedded Characteristics
The AT91SAM9G45 features USB communication ports as follows:
• 2 Ports USB Host full speed OHCI and High speed EHCI
• 1 Device High speed
USB Host Port A is directly connected to the first UTMI transceiver.
The Host Port B is multiplexed with the USB device High speed and connected to the second
UTMI port. The selection between Host Port B and USB device high speed is controlled by a the
UDPHS enable bit located in the UDPHS_CTRL control register.
Figure 37-1. USB Selection
HS
Transceiver
HS
Transceiver
EN_UDPHS
1
0
PA
PB
HS EHCI
FS OHCI
DMA
HS
USB
DMA
• Compliant with Enhanced HCI Rev 1.0 Specification
– Compliant with USB V2.0 High-speed and Full-speed Specification
– Supports Both High-speed 480Mbps and Full-speed 12 Mbps USB devices
• Compliant with Open HCI Rev 1.0 Specification
– Compliant with USB V2.0 Full-speed and Low-speed Specification
– Supports Both Low-speed 1.5 Mbps and Full-speed 12 Mbps USB devices
• Root Hub Integrated with 2 Downstream USB Ports
• Shared Embedded USB Transceivers
37.2.1
EHCI
The USB Host Port controller is fully compliant with the Enhanced HCI specification. The USB
Host Port User Interface (registers description) can be found in the Enhanced HCI Rev 1.0
Specification available on http://www.intel.com/technology/usb/ehcispec.htm. The standard
EHCI USB stack driver can be easily ported to Atmel’s architecture in the same way all existing
class drivers run, without hardware specialization.
765
6438D–ATARM–13-Oct-09
37.2.2
OHCI
The USB Host Port integrates a root hub and transceivers on downstream ports. It provides several Full-speed half-duplex serial communication ports at a baud rate of 12 Mbit/s. Up to 127
USB devices (printer, camera, mouse, keyboard, disk, etc.) and the USB hub can be connected
to the USB host in the USB “tiered star” topology.
The USB Host Port controller is fully compliant with the Open HCI specification. The USB Host
Port User Interface (registers description) can be found in the Open HCI Rev 1.0 Specification
available on http://h18000.www1.hp.com/productinfo/development/openhci.html. The standard
OHCI USB stack driver can be easily ported to Atmel’s architecture, in the same way all existing
class drivers run without hardware specialization.
This means that all standard class devices are automatically detected and available to the user’s
application. As an example, integrating an HID (Human Interface Device) class driver provides a
plug & play feature for all USB keyboards and mouses.
37.3
Block Diagram
Figure 37-2. Block Diagram
HCI
Slave Block
AHB
Slave
OHCI
Registers
Root
Hub Registers
List Processor
Block
Control
ED & TD
Regsisters
PORT S/M 1
Root Hub
and
Host SIE
AHB
Master
HCI
Master Block
Data
PORT S/M 0
FIFO 64 x 8
SOF
generator
HCI
Slave Block
AHB
Slave
EHCI
Registers
Embedded USB
v2.0 High-speed Transceiver
USB transceiver
HFSDPB
HFSDMB
HHSDPB
HHSDMB
USB transceiver
HFSDPA
HFSDMA
HHSDPA
HHSDMA
Packet
Buffer
FIFO
Control
List
Processor
AHB
Master
766
HCI
Master Block
Data
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Access to the USB host operational registers is achieved through the AHB bus slave interface.
The Open HCI host controller and Enhanced HCI host controller initialize master DMA transfers
through the AHB bus master interface as follows:
• Fetches endpoint descriptors and transfer descriptors
• Access to endpoint data from system memory
• Access to the HC communication area
• Write status and retire transfer descriptor
Memory access errors (abort, misalignment) lead to an “Unrecoverable Error” indicated by the
corresponding flag in the host controller operational registers.
The USB root hub is integrated in the USB host. Several USB downstream ports are available.
The number of downstream ports can be determined by the software driver reading the root
hub’s operational registers. Device connection is automatically detected by the USB host port
logic.
USB physical transceivers are integrated in the product and driven by the root hub’s ports.
Over current protection on ports can be activated by the USB host controller. Atmel’s standard
product does not dedicate pads to external over current protection.
37.4
37.4.1
Product Dependencies
I/O Lines
HFSDPs, HFSDMs, HHSDPs and HHSDMs are not controlled by any PIO controllers. The
embedded USB High Speed physical transceivers are controlled by the USB host controller.
37.5
I/O Lines
HFSDPs, HFSDMs, HHSDPs and HHSDMs are not controlled by any PIO controllers. The
embedded USB High Speed physical transceivers are controlled by the USB host controller.
One transceiver is shared with USB Device (UDP) High Speed. In this case USB Host High
Speed Controller uses only Port A, ie, the signals HFSDPA, HFSDMA, HHSDPA and HHSDMA.
The port B is driven by the UDP High Speed, the output signals are DFSDP, DFSDM, DHSDP
and DHSDM.
The transceiver is automatically selected for Device operation once the UDP High Speed is
enabled.
37.5.1
Power Management
The USB Host High Speed requires a 48 MHz clock for the embedded High-speed transceivers.
This clock is provided by the UTMI PLL, it is UPLLCK.
In case power consumption is saved by stopping the UTMI PLL, high-speed operations are not
possible. Nevertheless, OHCI Full-speed operations remain possible by selecting PLLACK as
the input clock of OHCI.
The High-speed transceiver returns a 30 MHz clock to the USB Host controller.
The USB Host controller requires 48 MHz and 12 MHz clocks for OHCI full-speed operations.
These clocks must be generated by a PLL with a correct accuracy of ± 0.25% thanks to USBDIV
field.
767
6438D–ATARM–13-Oct-09
Thus the USB Host peripheral receives three clocks from the Power Management Controller
(PMC): the Peripheral Clock (MCK domain), the UHP48M and the UHP12M (built-in UHP48M
divided by four) used by the OHCI to interface with the bus USB signals (Recovered 12 MHz
domain) in Full-speed operations.
For High-speed operations, the user has to perform the following:
• Enable UHP peripheral clock, bit (1 << AT91C_ID_UHPHS) in PMC_PCER register.
• Write CKGR_PLLCOUNT field in PMC_UCKR register.
• Enable UPLL, bit AT91C_CKGR_UPLLEN in PMC_UCKR register.
• Wait until UTMI_PLL is locked. LOCKU bit in PMC_SR register
• Enable BIAS, bit AT91C_CKGR_BIASEN in PMC_UCKR register.
• Select UPLLCK as Input clock of OHCI part, USBS bit in PMC_USB register.
• Program the OHCI clocks (UHP48M and UHP12M) with USBDIV field in PMC_USB register.
USBDIV must be 9 (division by 10) if UPLLCK is selected.
• Enable OHCI clocks, UHP bit in PMC_SCER register.
For OHCI Full-speed operations only, the user has to perform the following:
• Enable UHP peripheral clock, bit (1 << AT91C_ID_UHPHS) in PMC_PCER register.
• Select PLLACK as Input clock of OHCI part, USBS bit in PMC_USB register.
• Program the OHCI clocks (UHP48M and UHP12M) with USBDIV field in PMC_USB register.
USBDIV value is to calculated regarding the PLLACK value and USB Full-speed accuracy.
• Enable the OHCI clocks, UHP bit in PMC_SCER register.
768
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
Figure 37-3. UHP Clock trees
UPLL (480 MHz)
AHB
EHCI
Master
Interface
30 MHz
UTMI transceiver
USB 2.0 EHCI
Host Controller
Port
Router
30 MHz
EHCI
User
Interface
UTMI transceiver
MCK
OHCI
Master
Interface
Root Hub
and
Host SIE
UHP48M
UHP12M
OHCI
User
Interface
USB 1.1 OHCI
Host Controller
OHCI clocks
37.5.2
Interrupt
The USB host interface has an interrupt line connected to the Advanced Interrupt Controller
(AIC).
Handling USB host interrupts requires programming the AIC before configuring the UHP HS.
769
6438D–ATARM–13-Oct-09
37.6
Typical Connection
Figure 37-4. Board Schematic to Interface UHP High-speed Device Controller
PIO (VBUS DETECT)
15k Ω
(1)
"A" Receptacle
1 = VBUS
2 = D3 = D+
4 = GND
HHSDM
39 ± 5% Ω
HFSDM
3 4
(1)
22k Ω
Shell = Shield
HHSDP
CRPB
1 2
39 ± 5% Ω
CRPB: 1µF to 10µF
HFSDP
6K8 ± 1% Ω
VBG
10 pF
GND
Note:
770
1. The values shown on the 22k Ω and 15k Ω resistors are only valid for 3v3 supplied PIOs.
AT91SAM9G45
6438D–ATARM–13-Oct-09
AT91SAM9G45
38. USB High Speed Device Port (UDPHS)
38.1
Description
The USB High Speed Device Port (UDPHS) is compliant with the Universal Serial Bus (USB),
rev 2.0 High Speed device specification.
Each endpoint can be configured in one of several USB transfer types. It can be associated with
one, two or three banks of a dual-port RAM used to store the current data payload. If two or
three banks are used, one DPR bank is read or written by the processor, while the other is read
or written by the USB device peripheral. This feature is mandatory for isochronous endpoints.
38.2
Embedded Characteristics
The AT91SAM9G45 features USB communication ports as follows:
• 2 Ports USB Host full speed OHCI and High speed EHCI
• 1 Device High speed
USB Host Port A is directly connected to the first UTMI transceiver.
The Host Port B is multiplexed with the USB device High speed and connected to the second
UTMI port. The selection between Host Port B and USB device high speed is controlled by a the
UDPHS enable bit located in the UDPHS_CTRL control register.
Figure 38-1. USB Selection
HS
Transceiver
HS
Transceiver
EN_UDPHS
0
PA
1
PB
HS EHCI
FS OHCI
DMA
HS
USB
DMA
• USB V2.0 high-speed compliant, 480 MBits per second
• Embedded USB V2.0 UTMI+ high-speed transceiver shared with UHP HS.
• Embedded 4-KByte dual-port RAM for endpoints
• Embedded 6 channels DMA controller
• Suspend/Resume logic
• Up to 2 or 3 banks for isochronous and bulk endpoints
• Seven endpoints:
– Endpoint 0: 64 bytes, 1 bank mode
– Endpoint 1 & 2: 1024 bytes, 2 banks mode, High Bandwidth, DMA
– Endpoint 3 & 4: 1024 bytes, 3 banks mode, DMA
– Endpoint 5 & 6: 1024 bytes, 3 banks mode, High Bandwidth, DMA
771
6438D–ATARM–13-Oct-09
Table 38-1.
UDPHS Endpoint Description
Mnemonic
Nb Bank
DMA
High
BandWidth
Max. Endpoint Size
Endpoint Type
0
EPT_0
1
N
N
64
Control
1
EPT_1
2
Y
Y
1024
Ctrl/Bulk/Iso(1)/Interrupt
2
EPT_2
2
Y
Y
1024
Ctrl/Bulk/Iso(1)/Interrupt
3
EPT_3
3
Y
N
1024
Ctrl/Bulk/Iso(1)/Interrupt
4
EPT_4
3
Y
N
1024
Ctrl/Bulk/Iso(1)/Interrupt
5
EPT_5
3
Y
Y
1024
Ctrl/Bulk/Iso(1)/Interrupt
6
EPT_6
3
Y
Y
1024
Ctrl/Bulk/Iso(1)/Interrupt
Endpoint #
Note:
1. In Isochronous Mode (Iso), it is preferable that High Band Width capability is available.
The size of internal DPRAM is 4 KB.
Suspend and resume are automatically detected by the UDPHS device, which notifies the processor by raising an interrupt.
772
AT91SAM9G45
6438D–ATARM–13-Oct-09
38.3
Block Diagram
Figure 38-2. Block Diagram
APB
Interface
APB bus
ctrl
status
DHSDP
DHSDM
AHB1
AHB bus
Rd/Wr/Ready
DMA
AHB0
APB bus
UTMI
USB2.0
CORE
DFSDP
DP
DFSDM
DM
Master
AHB
Multiplexeur
Slave
Local
AHB
Slave
interface
EPT
Alloc
32 bits
DPRAM
System Clock
Domain
16/8 bits
USB Clock
Domain
PMC
Notes:
1. System clock, bit (1 << AT91C_ID_UDPHS) in PMC_PCER register.
2. Enable UDPHS clock (peripheral clock) bit AT91C_CKGR_UPLLEN in PMC_UCKR register.
3. Enable BIAS bit AT91C_CKGR_BIASEN in PMC_UCKR register.
773
AT91SAM9G45
6438D–ATARM–13-Oct-09
38.4
Typical Connection
Figure 38-3. Board Schematic
PIO (VBUS DETECT)
15k Ω
(1)
"B" Receptacle
1 = VBUS
2 = D3 = D+
4 = GND
1
2
3
4
DHSDM
39 ± 5% Ω
DFSDM
Shell = Shield
(1)
22k Ω
CRPB
DHSDP
39 ± 5% Ω
CRPB:1µF to 10µF
DFSDP
6K8 ± 1% Ω
VBG
10 pF
GND
Notes:
774
1. The values shown on the 22kΩ and 15kΩ resistors are only valid with 3V3 supplied PIOs.
AT91SAM9G45
6438D–ATARM–13-Oct-09
38.5
Functional Description
38.5.1
USB V2.0 High Speed Device Port Introduction
The USB V2.0 High Speed Device Port provides communication services between host and
attached USB devices. Each device is offered with a collection of communication flows (pipes)
associated with each endpoint. Software on the host communicates with a USB Device through
a set of communication flows.
38.5.2
USB V2.0 High Speed Transfer Types
A communication flow is carried over one of four transfer types defined by the USB device.
A device provides several logical communication pipes with the host. To each logical pipe is
associated an endpoint. Transfer through a pipe belongs to one of the four transfer types:
• Control Transfers: Used to configure a device at attach time and can be used for other devicespecific purposes, including control of other pipes on the device.
• Bulk Data Transfers: Generated or consumed in relatively large burst quantities and have
wide dynamic latitude in transmission constraints.
• Interrupt Data Transfers: Used for timely but reliable delivery of data, for example, characters
or coordinates with human-perceptible echo or feedback response characteristics.
• Isochronous Data Transfers: Occupy a prenegotiated amount of USB bandwidth with a
prenegotiated delivery latency. (Also called streaming real time transfers.)
As indicated below, transfers are sequential events carried out on the USB bus.
Endpoints must be configured according to the transfer type they handle.
Table 38-2.
USB Communication Flow
Transfer
Direction
Bandwidth
Endpoint Size
Error Detection
Retrying
Bidirectional
Not guaranteed
8,16,32,64
Yes
Automatic
Isochronous
Unidirectional
Guaranteed
8-1024
Yes
No
Interrupt
Unidirectional
Not guaranteed
8-1024
Yes
Yes
Bulk
Unidirectional
Not guaranteed
8-512
Yes
Yes
Control
38.5.3
USB Transfer Event Definitions
A transfer is composed of one or several transactions;
Table 38-3.
USB Transfer Events
CONTROL
(bidirectional)
IN
(device toward host)
775
Control Transfers (1)
• Setup transaction →Data IN transactions →Status OUT transaction
• Setup transaction →Data OUT transactions →Status IN transaction
• Setup transaction →Status IN transaction
Bulk IN Transfer
• Data IN transaction →Data IN transaction
Interrupt IN Transfer
• Data IN transaction →Data IN transaction
Isochronous IN Transfer (2)
• Data IN transaction →Data IN transaction
AT91SAM9G45
6438D–ATARM–13-Oct-09
Table 38-3.
USB Transfer Events (Continued)
CONTROL
(bidirectional)
OUT
(host toward device)
Notes:
Control Transfers (1)
• Setup transaction →Data IN transactions →Status OUT transaction
• Setup transaction →Data OUT transactions →Status IN transaction
• Setup transaction →Status IN transaction
Bulk OUT Transfer
• Data OUT transaction →Data OUT transaction
Interrupt OUT Transfer
• Data OUT transaction →Data OUT transaction
Isochronous OUT Transfer (2)
• Data OUT transaction →Data OUT transaction
1. Control transfer must use endpoints with one bank and can be aborted using a stall handshake.
2. Isochronous transfers must use endpoints configured with two or three banks.
An endpoint handles all transactions related to the type of transfer for which it has been
configured.
38.5.4
USB V2.0 High Speed BUS Transactions
Each transfer results in one or more transactions over the USB bus.
There are five kinds of transactions flowing across the bus in packets:
1. Setup Transaction
2. Data IN Transaction
3. Data OUT Transaction
4. Status IN Transaction
5. Status OUT Transaction
Figure 38-4. Control Read and Write Sequences
Setup Stage
Control Write
Setup TX
Setup Stage
Control Read
No Data
Control
Setup TX
Data Stage
Data OUT TX
Status Stage
Data OUT TX
Data Stage
Data IN TX
Setup Stage
Status Stage
Setup TX
Status IN TX
Data IN TX
Status IN TX
Status Stage
Status OUT TX
A status IN or OUT transaction is identical to a data IN or OUT transaction.
38.5.5
776
Endpoint Configuration
The endpoint 0 is always a control endpoint, it must be programmed and active in order to be
enabled when the End Of Reset interrupt occurs.
AT91SAM9G45
6438D–ATARM–13-Oct-09
To configure the endpoints:
• Fill the configuration register (UDPHS_EPTCFG) with the endpoint size, direction (IN or
OUT), type (CTRL, Bulk, IT, ISO) and the number of banks.
• Fill the number of transactions (NB_TRANS) for isochronous endpoints.
Note: For control endpoints the direction has no effect.
• Verify that the EPT_MAPD flag is set. This flag is set if the endpoint size and the number of
banks are correct compared to the FIFO maximum capacity and the maximum number of
allowed banks.
• Configure control flags of the endpoint and enable it in UDPHS_EPTCTLENBx according to
“UDPHS Endpoint Control Register” on page 822.
Control endpoints can generate interrupts and use only 1 bank.
All endpoints (except endpoint 0) can be configured either as Bulk, Interrupt or Isochronous. See
Table 38-1. UDPHS Endpoint Description.
The maximum packet size they can accept corresponds to the maximum endpoint size.
Note: The endpoint size of 1024 is reserved for isochronous endpoints.
The size of the DPRAM is 4 KB. The DPR is shared by all active endpoints. The memory size
required by the active endpoints must not exceed the size of the DPRAM.
SIZE_DPRAM = SIZE _EPT0
+ NB_BANK_EPT1 x SIZE_EPT1
+ NB_BANK_EPT2 x SIZE_EPT2
+ NB_BANK_EPT3 x SIZE_EPT3
+ NB_BANK_EPT4 x SIZE_EPT4
+ NB_BANK_EPT5 x SIZE_EPT5
+ NB_BANK_EPT6 x SIZE_EPT6
+... (refer to 38.6.11 UDPHS Endpoint Configuration Register)
If a user tries to configure endpoints with a size the sum of which is greater than the DPRAM,
then the EPT_MAPD is not set.
The application has access to the physical block of DPR reserved for the endpoint through a 64
KB logical address space.
The physical block of DPR allocated for the endpoint is remapped all along the 64 KB logical
address space. The application can write a 64 KB buffer linearly.
777
AT91SAM9G45
6438D–ATARM–13-Oct-09
Figure 38-5. Logical Address Space for DPR Access:
DPR
x banks
Logical address
8 to 64 B
8 to 64 B
8 to 64 B
8 to 64 B
...
8 to1024 B
64 KB
EP0
8 to1024 B
64 KB 8 to1024 B
8 to1024 B
EP1
...
64 KB
EP2
y banks
z banks
8 to1024 B
8 to1024 B
64 KB
EP3
...
Configuration examples of UDPHS_EPTCTLx (UDPHS Endpoint Control Register) for Bulk IN
endpoint type follow below.
• With DMA
– AUTO_VALID: Automatically validate the packet and switch to the next bank.
– EPT_ENABL: Enable endpoint.
• Without DMA:
– TX_BK_RDY: An interrupt is generated after each transmission.
– EPT_ENABL: Enable endpoint.
Configuration examples of Bulk OUT endpoint type follow below.
• With DMA
– AUTO_VALID: Automatically validate the packet and switch to the next bank.
– EPT_ENABL: Enable endpoint.
• Without DMA
– RX_BK_RDY: An interrupt is sent after a new packet has been stored in the endpoint
FIFO.
– EPT_ENABL: Enable endpoint.
778
AT91SAM9G45
6438D–ATARM–13-Oct-09
38.5.6
Transfer With DMA
USB packets of any length may be transferred when required by the UDPHS Device. These
transfers always feature sequential addressing.
Packet data AHB bursts may be locked on a DMA buffer basis for drastic overall AHB bus bandwidth performance boost with paged memories. These clock-cycle consuming memory row (or
bank) changes will then likely not occur, or occur only once instead of dozens times, during a
single big USB packet DMA transfer in case another AHB master addresses the memory. This
means up to 128-word single-cycle unbroken AHB bursts for Bulk endpoints and 256-word single-cycle unbroken bursts for isochronous endpoints. This maximum burst length is then
controlled by the lowest programmed USB endpoint size (EPT_SIZE bit in the
UDPHS_EPTCFGx register) and DMA Size (BUFF_LENGTH bit in the
UDPHS_DMACONTROLx register).
The USB 2.0 device average throughput may be up to nearly 60 MBytes. Its internal slave average access latency decreases as burst length increases due to the 0 wait-state side effect of
unchanged endpoints. If at least 0 wait-state word burst capability is also provided by the external DMA AHB bus slaves, each of both DMA AHB busses need less than 50% bandwidth
allocation for full USB 2.0 bandwidth usage at 30 MHz, and less than 25% at 60 MHz.
The UDPHS DMA Channel Transfer Descriptor is described in “UDPHS DMA Channel Transfe