ETC STMP35XX

Integrated Mixed-Signal Solutions
PRODUCT DATA SHEET
STMP35xx
D-Major™Audio System on Chip
with USB 2.0, LCD, Voice Record and Battery Charger
Third Generation Audio Decoder
Version 1.08 June 17, 2005
Host Processor
(Optional)
90
FM Tuner
94
98
102
106
Rechargeable
Battery
LED/LCD Screen
SDRAM
Hi-Speed USB
Microphone
Voice Record
Flash Memory
Headphones
Buttons/Switches
CD Pickup
Hard Drive
OFFICIAL PRODUCT DOCUMENTATION 6/17/05
5-35xx-D1-1.08-061705
Copyright © 2004 SigmaTel, Inc.
All rights reserved.
SigmaTel, Inc. makes no warranty for the use of its products, assumes no responsibility for any errors which may appear in this document,
and makes no commitment to update the information contained herein. SigmaTel reserves the right to change or discontinue this product at
any time, without notice. There are no express or implied licenses granted hereunder to design or fabricate any integrated circuits based on
information in this document.
The following are trademarks of SigmaTel, Inc., and may be used to identify SigmaTel products only: SigmaTel, the SigmaTel Logo, C Major,
D Major and Go-Chip. Other products and company names contained herein may be trademarks of their respective owners.
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
1.
TABLE OF CONTENTS
1. TABLE OF CONTENTS ..................................................................................................................... 2
2. PRODUCT OVERVIEW ..................................................................................................................... 3
3. CHARACTERISTICS/SPECIFICATIONS ........................................................................................ 21
4. DSP CORE ....................................................................................................................................... 25
5. ON-CHIP MEMORY SUBSYSTEM .................................................................................................. 30
6. CHIP WIDE PROGRAMMABLE CONTROL REGISTERS ............................................................. 38
7. INTERRUPT SUBSYSTEM .............................................................................................................. 49
8. USB CONTROLLER ........................................................................................................................ 60
9. INTEGRATED USB 2.0 PHY (HS,FS) ............................................................................................. 79
10. PARALLEL EXTERNAL MEMORY CONTROLLER (EMC) ....................................................... 103
11. GENERAL PURPOSE FLASH CONTROLLER .......................................................................... 118
12. FLASH ECC ACCELERATOR .................................................................................................... 134
13. FILTER COPROCESSOR (FILCO) ............................................................................................. 154
14. PULSE WIDTH MODULATOR (PWM) CONTROLLER .............................................................. 185
15. I2C INTERFACE ........................................................................................................................... 196
16. ENHANCED SPI INTERFACE ..................................................................................................... 209
17. SPI INTERFACE .......................................................................................................................... 217
18. TIMERS ........................................................................................................................................ 220
19. SDRAM INTERFACE ................................................................................................................... 227
20. SWIZZLE ...................................................................................................................................... 239
21. REAL-TIME CLOCK/ALARM/WATCHDOG RESET & PERSISTENT BITS .............................. 247
22. I2S SERIAL AUDIO INTERFACE ................................................................................................ 258
23. GENERAL PURPOSE INPUT/OUTPUT (GPIO) ......................................................................... 265
24. DAC .............................................................................................................................................. 276
25. ADC .............................................................................................................................................. 286
26. MIXER .......................................................................................................................................... 297
27. HEADPHONE DRIVER ................................................................................................................ 310
28. LOW RESOLUTION ADC ............................................................................................................ 315
29. BOOT MODES ............................................................................................................................. 329
30. DC-DC CONVERTER .................................................................................................................. 339
31. PIN DESCRIPTION ..................................................................................................................... 377
32. PACKAGE DRAWINGS ............................................................................................................... 389
33. STMP35XX FAMILY MEMBER PART NUMBERS & ORDERING INFORMATION ................... 391
34. INDEX OF REGISTERS ............................................................................................................... 393
ADDITIONAL SUPPORT
Additional product and company information can be obtained by going to the SigmaTel website at:www.sigmatel.com. Additional product and design information is
available for authorized customers at: extranet.sigmatel.com
2
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
2.
PRODUCT OVERVIEW
2.1.
Features
•
•
•
•
•
•
•
•
•
•
•
5-35xx-D1-1.08-061705
Decodes MP3 and WMA and is upgradeable to other digital music formats
Supports WMA Digital Rights Management (DRM) and other security schemes
Includes on-chip read only unique ID for digital rights management algorithms
USB High Speed Device Interface (up to 480Mb/s transfers)
• Enables file transfer and firmware upgrade using USB Mass Storage Class
• Both Windows and Macintosh drivers available
• Integrated USB High Speed PHY
• Direct connection to USB 5V power for operation and battery charging
96K Words (288K Bytes) of on-chip RAM
Hardware support for flexible external storage options
• NAND Flash, MMC, Secure Digital, SmartMedia, CompactFlash
• Five byte address support for new 1Gb/die (128KB block) NAND Flash
• MLC NAND Flash support
• 1.8V NAND Interface Support
• 16 bit wide NAND support
• Hardware accelerated ECC off-loads DSP bit error correction
• SDRAM
• ATA/IDE Hard Disk digital devices.
Optimized for very long battery life
• 50 hours of operation on a single AA battery
• Flexible, efficient on-chip DC-DC converter
• Flexible battery configurations, including 1xAA, 1xAAA, 2xAA, 2xAAA, LiIon
• Pulse frequency modulation mode for low standby power
• Energy saving dynamic power management
• Typical off current is 250µA (crystal oscillator & real time clock only)
• More than 1 year battery life in “off” mode on one AA Alkaline battery
• Integrated battery charger for LiIon and NiMH
• Battery temperature sensor support for safest charging protocols
• Real time clock with alarm function wakes up from powerdown/standby modes
High quality integrated audio mixed signal sub-system
• <0.05% THD direct drive headphone amplifier
– Eliminates DC blocking capacitors
– Including anti-pop and short-circuit protection
• High performance 18-bit Σ∆ technology stereo D/A and A/D converters
• Full analog mixer configuration
• Line-in to Headphone/Line-out SNR >90 dB
• Two analog line-level inputs: Line1 In (stereo), Line2 In (stereo, 144-pin package)
• Mic(mono) input with integrated pre-amp and microphone biasing circuit
• Volume control
GPIO, button I/O controls, and LCD/LED Display Compatible Interface
Pulse Width Modulators for EL backlights
Integrated 75MHz DSP with Filter Coprocessor for power optimization
• Optimized for audio applications
• Field upgradeable firmware
3
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
•
•
•
•
•
STMP35xx Block Diagram
90
LED/LCD
Buttons/
Switches
E.L. Backlight
MMC/SD
Card
I2C
Peripherals
SmartMedia
CompactFlash
NOR Flash
NAND Flash
MLC Flash
SDRAM
I/OPurpose
Pin Multiplexer
General
Input/Output
Hard Drive
InterruptControl,
Control,Timers,
Timers,Bit
Interrupt
Bit Manipulation
Unit,
Manipulation
Unit, RTC,
Trace
Trace
Unit
Debug
Unit,Debug
Reed-Solomon
USB 2.0
USB
Device
I2S & CD
I2S CDation
Synchroniz
Synchronization
Interface
Memory Bus
Peripheral Bus
Filter &
ECC
USB
engines
FM in
USB PHY
(HSUSB
& FS)
STMP35xx
Σ
Pulse
Width
SPI Interface
DAC
DAC
24-bit
DSP
DSP
I2C
I2C Interface
Interface
Headphones
Cap-less
Direct Drive
ADC
ADC
Flash/IDE
EMC
Interface
SDRAM
SDRAM
Interface
Interface
106
Amp
DAC
GPIO
Interface
I2C Interface
SPI
SPI Interface
Interface
102
FM Radio
Line in
CDInterface
Control
IDE
Interface
98
Microphone
Mic in
USB High Speed
USB Full Speed
94
HeadphoneAmplifier
Amplifier
Headphone
2.2.
• Integrated Development Environment, SDK, and debugger
• Application and support libraries
• Bass and Treble control; configurable multiple band EQ control
• Voice record in ADPCM format (upgradable to other formats)
FM tuner input and control support
Optional interface to a host chip/processor for cell phone & PDA applications, etc.
Application notes, reference schematics, sample PCB layouts are available.
Offered in 100-pin TQFP, and 144-pin fpBGA packages
Backward pin and firmware compatible with STMP3410
Te mpe ra ture
On-Chip
On-Chip ROM
ROM
16K
8K x 24bits
On-Chip
On-Chip RAM
RAM
96K
96K xx 24bits
24bits
Low
DCDC
Resolution
Converter
ADC x3
Low
Battery
Resoluti
Charger
on ADC
RTC,
ALARM,
PLL
PLL,
xtal
XTAL
DCDC
DCDC
Converter
Converter
Rechargable Battery
5V
Input
Crystal
Figure 1. Chip Block Diagram
Pin 1
Pin A1
10 mm
14 mm
10 mm
14 mm
100-pin TQFP
144-pin fpBGA
For additional package measurements, please see 32. “PACKAGE DRAWINGS” on page 389.
Figure 2. Chip Package Photos
4
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
2.3.
Description
SigmaTel's STMP35xx is a third generation single-chip highly-integrated digital music
system solution for devices such as digital audio players, PDAs, voice recorders, MP3encode recorders, and cell phones. It includes a high performance DSP, 288KBytes of
on-chip SRAM, and a USB 2.0 interface (including High speed 480Mb/second transfers)
for downloading music and uploading voice and MP3 recordings. The chip also includes
a mixer, DAC, ADC and provides interfaces to IDE Hard Drives, CD-DSPs, Flash memory, LCD/LEDs, button & switch inputs, headphone driver, FM tuner input & controls and
a microphone. The chip’s highly programmable architecture supports MP3, WMA, and
other digital audio standards. WMA digital rights management and other security
schemes are also supported. For devices like PDAs and cell phones, the STMP35xx
can act as a slave chip to a host chip/processor.
The DAC includes a headphone driver to directly drive low impedance headphones. The
ADC includes inputs for both microphone and analog audio in to support voice recording &
FM radio integration and MP3 encode features. SigmaTel's proprietary Sigma-Delta (Σ∆)
technology achieves a DAC SNR in excess of 90 dB for high-quality audio playback.
The STMP35xx has low power consumption to allow long battery life and includes an
efficient flexible on-chip DC-DC converter that allows many different battery configurations, including 1xAA, 1xAAA, 2xAA, 2xAAA and LiIon. The chip includes an integrated
intelligent charger for NiMH and LiIon batteries. In addition, the single-chip design and
low pin count enables very small digital audio devices to be designed.
2.3.1.
DSP Core
The on-chip DSP core is modeled exactly after the Motorola DSP56004. It supports
the identical instruction set, registers, addressing modes, etc., as the DSP56000
family of digital signal processors. Figure 3 shows a high level view of the DSP core.
This architecture is highly optimized for battery operated audio applications. Its 24bit intrinsic data size provides sufficient precision for high quality audio algorithms
while minimizing the number of register and data path signals that must be toggled
for any operation. The term “WORD”, as used in this data sheet, refers to a 24-bit
unit of storage unless otherwise noted.
The functionality that defines the on-chip DSP, is the memory map, interrupt processing, and peripherals it offers.
The integrated DSP comprises three execution units, an interrupt controller and a
debug interface. It connects to the rest of the STMP35xx chip via three memory busses, a set of interrupt input signals and various reset and clock inputs. It implements
a 3 memory space Harvard architecture, simultaneously referencing an X data element, a Y data element and a program element. These references are conveyed
over the program or “P” bus, the X bus and the Y bus. Each bus comprises a 24 bit
wide data path and a 16 bit address bus. Program accessible I/O registers reside in
the top 4K word addresses on the X-bus. The DSP architecture has special programmed I/O support for the top 64 words of this space but SigmaTel has extended
this space to the top 4K words, i.e. addresses $F000 through $FFFF, inclusive.
The DSP Core also implements the OnCE debugger that is the norm for this DSP
architecture. The OnCE interface connects to an external debugger over four I/O
signal pins on the STMP35xx.
Using an industry standard instruction set architecture and debugger interface for
the integrated DSP means that development tools and debuggers are in the highly
evolved and stable portion of their life cycle. In addition, it means that system devel-
5-35xx-D1-1.08-061705
5
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
DSP Core
Program
Address
Generator
To/from
P space
Address to
X space
Address Generation Unit
Debug Interface
R0
R1
N0
N1
M0
M1
R2
N2
M2
R3
N3
M3
R4
R5
N4
N5
M4
M5
R6
N6
M6
R7
N7
M7
ProgramDecoder
Address to
Y space
DATA ALU
X0
Instruction Latch
Y0
X1
Data to/from
X space
Y1
x
Decoder
&
State Machines
+
Registered Control
A
Bit Manipulation
Unit
Data to/from
Y space
B
Interrupt
Controller
Interrupts
Figure 3. DSP Core at a Glance
opers with experience developing on this DSP can be found. The SigmaTel software
developers kit (SDK) provides an excellent integrated development environment
with an assembler, C compiler, debugger and other requisite tools.
2.3.2.
On-chip RAM and ROM
The STMP35xx includes 96K words of on-chip RAM. This amounts to 2.25Mbits of
on-chip SRAM in six 16K Word blocks. The RAM is split into two 48K word banks
with one bank attached to the X bus and one attached to the Y bus. The P bus is
Y-BUS
X-BUS
P-BUS
DSP CORE
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
Figure 4. 6 x 16K Word On-chip SRAM Blocks
connected to both RAM banks so that program space can be allocated from the
same two banks that hold X and Y data values. An adjustable switching mechanism
6
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
is provided so that a portion of the X bus RAM or Y bus RAM can be allocated to the
P bus in units of 8K words from 0K to the full 64K words.
A typical application will allocate a portion of the X bus RAM and another portion of
the Y bus RAM to the P bus. In normal operation, this switching mechanism will
present a contiguous block of RAM beginning at location zero in the P bus address
space, or P:$0000 as it is written in assembler syntax
Suppose we allocate 24K Words from the X bus RAM to the P bus and another 24K
Words from the Y bus RAM to the P Bus. This gives us 24K Words of X RAM, 24K
Words of Y RAM and 48K Words of P RAM for our application. All 96K Words of onchip SRAM are allocated, as shown in Figure 5.
P
Address
Space
X
Address
Space
Y
Address
Space
$FFFF
PIO
Regs.
on-chip
ROM
$C000
No
SRAM
$6000
$5FFF
24K
Words
X
RAM
48K Words Physical SRAM
$BFFF
$BFFF
24K
Words
from
Y
RAM
24K
Words
from
X
RAM
No
SRAM
48K Words Physical SRAM
$F000
$5FFF
24K
Words
Y
RAM
$0000
Figure 5. On-Chip RAM Allocation Example
The STMP35xx contains an on-chip 16K Word ROM which holds the Bootstrap
code. At power-on time, the first instruction executed by the DSP comes from this
ROM. Power-on reset causes the on-chip ROM to be placed at P:$0000. The reset
interrupt vector is located at P:$0000, thus the first instructions executed come from
this ROM. Software in this ROM offers a large number of BOOT configuration
options, including manufacturing boot modes for “burn-in” and “tester” operation.
Other boot modes are responsible for loading application code from off-chip into the
on-chip RAM. Off chip sources for application bootstrapping include:
• External NAND FLASH
• Host (PC) controlled bootstrapping using USB,
5-35xx-D1-1.08-061705
7
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
• Host (PC) controlled bootstrapping using I2C slave.
• I2C Master transfers from serial EEPROM
Once the on-chip boot code has loaded the application code into on-chip RAM, it
can relocate the 16K Word on-chip ROM to the very top of the P address space, see
Figure 5. “On-Chip RAM Allocation Example” on page 7. The on-chip ROM can be
disabled entirely so that all 64K words of P space is available for on-chip RAM.
The on-chip boot code includes a firmware recovery mode. If the device fails to boot
from NAND flash, for example, the device will boot from a PC host connected to its
USB port. This firmware recovery mode can be invoked at anytime by holding the
PSWITCH or “play” button for at least five seconds during power up.
The on-chip RAM serves as one end of all DMA transfers, e.g either the source or
destination. Every SRAM block has three potential accessors: P-BUS, DMA-BUS,
and its respective X-BUS or Y-BUS. A number of the integrated peripheral controllers use a distributed DMA implementation to transfer data to or from on-chip
SRAM.
Y-BUS
X-BUS
P-BUS
DSP CORE
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
DMA-BUS
External Flash/IDE
External SDRAM
USB
FLASH ECC
CD
D/A
A/D
Trace Buffer
Filter Coproc.
Figure 6. On-Chip RAM & Distributed DMA
In this distributed DMA architecture, all of the peripheral controllers that use DMA
share a common DMA address and data bus path to and from on-chip RAM. Each
peripheral controller implements its own address generator. Address generation can
be highly sequential as in the case of the D/A converter or fairly random as in the
case of the USB controller. Thus each device that uses the distributed DMA will
have at least one base address register (HW_xxxBAR) and various address modifying registers. Most of the distributed DMA devices implement some form of circular
buffering in their addressing modes. There is a centralized arbiter that selects which
of the distributed DMA peripheral controllers has access to the DMA bus on any
given clock cycle.
8
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
With a DSP core clock of 65MHz, a single block of on-chip RAM can provide 65MHz
times 3 bytes or 195 MByte/second of bandwidth. There are four 24 bit data busses
connected to the on-chip SRAM blocks. Furthermore, each SRAM block is single
ported and has its own independent address and data busses. Thanks to the arbitration logic in this memory subsystem, all four busses (P, X, Y, DMA) can be made
to cycle on every clock. Thus the peak bandwidth available from the on-chip RAM is
four times 195 MByte/second or 780 MByte/second.
Of course, there are times when more than one data bus needs to transfer into or
out of the same SRAM block. When conflicts occur, the arbiter will “stall” the DSP for
one (or more) clock(s) to resolve the conflict.
The reader should not be surprised to see devices like A/D or D/A converters using
DMA transfers. Some readers may be surprised to learn that external FLASH and
external SDRAM are only accessible via the DMA. The external memories are not
mapped into the “load/store” space of the DSP’s instruction set.
2.3.3.
Power Subsystem
The STMP35xx contains a sophisticated power subsystem including two integrated
DC to DC converters to produce a very cost effective product with flexible battery
configurations. In addition, it contains power monitoring circuits for battery brownout
detection as well as system overload brownout detection. The chip also contains
detection circuits for battery installation and removal. It manages power state
changes caused by battery changes or from monitoring the on/off power switch circuit.
The chip has two programmable integrated DC-DC converters that can be used to
provide power for the device as well as the entire application. The converters can be
configured to operate from standard battery chemistries in the range of 0.9-4.2 volts
including alkaline cells, NiMH, LiIon etc. These converters use off chip reactive components (L/C) in a pulse width or frequency modulated DC to DC converter.
The DC to DC converter circuit consists of the off-chip reactive components, an integrated controller and integrated low resistance FET switches. The DC-DC converter
#1, as shown in Figure 7, has one n-channel FET and three independently controlled p-channel FETs generating three independent channels of separately controlled voltages. For the case shown, the battery is a single AA alkaline battery in the
range 0.9 to 1.5 volts. DC-DC converter #1 is used to “boost” this input voltage to
3.3 volts for use in driving the I/O VDD rail and two separate 1.8 volt sources for
driving the analog VDD rail and the digital VDD rail. This case is shown in the first
row of Table 1, “Flexible Battery Configurations,” on page 11. Other rows show different configurations supported by the DC to DC converters. For example, when the
battery chemistry provides an input voltage that is higher than that desired for the
I/O rail, digital rail or analog rail, then the DC to DC converters can operate in “buck”
mode which provides a regulated output that is lower than its input.
One obvious use for the DCDC converter is in boosting the output of a nearly
depleted alkaline battery delivering 0.9 volts up to the regulated 3.3 volt I/O rail voltage and the regulated 1.8 volt digital and core rails. The DCDC converter can also
be used to lower the voltage of a 4.2 volt LiIon battery down to the 1.8V digital core
and analog rails. Table 1, “Flexible Battery Configurations,” on page 11 shows various battery configurations that can be supported.1
1.Note VddA3 == VddHP, VddA4== VddPLL. The analog power pair formerly known, in the STMP3410,
as VddA2 and VssA2 have been redefined for the STMP35xx as Vdd5V and LRADC2.
5-35xx-D1-1.08-061705
9
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
In addition, the DCDC converter can regulate these voltages to lower the standard
core and I/O values to extend the battery life. Recall that energy consumed in a
CMOS AC circuit is proportional to V2 so this reduction can be quite significant.
These program controlled reductions in operating voltage are used in various SigmaTel software applications to provide very long battery life products. The
STMP35xx also contains a silicon speed sensor so that each device can tailor its
operating voltage to the minimum required for safe operation as constrained by its
individual silicon process parameters and junction temperature.
The DC to DC converters control the power up sequence of the device and hold the
rest of the chip in reset until the power supplies have stabilized at the correct voltages. The power up sequence begins when the battery is connected to the BATT
pin. As shown in Figure 7, the crystal oscillator will begin running as soon as the battery is connected and the pswitch is asserted. The crystal oscillator and the real
time clock (RTC) can be programmed to continue to operate even when the player
is in the off state. The crystal oscillator and RTC are the only drains on the battery in
the off state and designed for very small energy consumption. The RTC module
includes an alarm function that can be used to “wake-up” the DC to DC converters
which will then wake up the rest of the system.
VddIO
1/2/3*/4*
VddD
1/2/3
VddA
1/2/3
NC
DCDC_VddIO DCDC_VddD DCDC_VddA*
battery
battery
Low
Resolution
ADC
BATT
DC-DC
#1 Control System
(boost mode)
DCDC_Batt
DCDC_Gnd
Regulator
VddXTAL=BATT
DSP_RESET
VddXTAL
NC
DCDC2_Vout*
RTC
DC-DC
#2 Control System
medium V (2 Alkaline/NimH)
or hi V LiIon Applications
XTAL
OSC
DCDC_mod0*
DCDC_mod1*
Power Button
DCDC2_Gnd*
DCDC_mod2
Pswitch
NC
DCDC2_Batt*
* only available on 144-pin package
NC
NC
NC
Figure 7. Lowest Cost 1xAA 100-pin Configuration1
The power down sequence is also controlled by the DC to DC converters. When a
power down event is detected, they return the player to the power off state. In the
power off state with non-LiIon chemistries, the I/O Vdd rail is connected to the Battery and the internal VddD and VddA rails are pulled down to ground to minimize
10
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
leakage currents. For LiIon mode, the I/O Vdd rail is connected to ground instead of
connecting to BATT.
In addition to the various voltage detectors, a power up or power down event can be
signaled by the special power switch circuitry in the DC to DC converters. A simple
resistor network and momentary contact push button switch is sufficient for player
on/off control.
There is a special three channel low resolution A/D converter on-chip to help with
battery based applications. One channel is dedicated to measuring the voltage on
the BATT pin and is used to monitor the battery condition to estimate its remaining
life. All low resolution channel also have digital trip point comparator functions that
can be used to generate interrupts to the DSP. The trip point can be programmatically set at one of 512 levels for battery brown out detection on the Battery LRADC
or for threshold detection on the other two LRADCs. NOTE: ONLY the battery can
be connected directly to the BATT pin for correct operation of the device, thus the
battery channel of the low resolution A/D converter is not available for any other purpose. The second and third low resolution A/D converters are uncommitted and
available for application use. An optional current source can be enabled to either the
second or third LRADC pin to support external temperature sensors with minimal
external components.
In addition, the DC to DC converters have comparators to monitor their output voltages. They can report “brownout” conditions resulting from over drawing their power
capabilities. These conditions are reported either on a normal interrupt level or as a
non-maskable interrupt (NMI).
The device contains an integrated PLL which is referenced to the 24.0MHz crystal
oscillator. It can generate clock sources from 39.6MHz to 120.0MHz in steps of
1.2MHz. It includes a post divide stage for the digital clock from a divide by one to a
divide by 2048. With the PLL turned off and the post divider set to 2048, one can
achieve a low power 11.7KHz operating point.
POWER SOURCE
1 Alkaline or 1 NiMH
(0.9V-1.5V)
VDD I/O
DCDC1
DCDC_VddIO
Boost
3.3V
1 Alkaline or 1 NiMH
(0.9V-1.5V)
DCDC1
DCDC_VddIO
Boost
3.3V
LiIon, (3.0-3.6V)
2 Alkaline or 2 NiMH
(1.8V-3.0V)
LiIon (3.3V-4.2V)
5-35xx-D1-1.08-061705
VDD D
DCDC1
DCDC_VddD
Boost
1.8V
DCDC1
DCDC_VddD
Boost
1.8V
lowest cost (shared
passives & 100-pin)
LiIon Battery
DCDC1
Buck
1.8V
DCDC2
DCDC1
Boost
Buck
3.3V
1.8v
DCDC2
DCDC1
Buck
Buck
3.3V
1.8V
Table 1. Flexible Battery Configurations
VDD A
DCDC1
DCDC_VddA
Boost
1.8V
better noise floor
(144-pin)
DCDC1
DCDC_VddD
Boost
1.8V
DCDC1
Buck
1.8V
DCDC1
Buck
1.8V
DCDC1
Buck
1.8V
11
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
There is an integrated watchdog reset timer available for automatic recovery from
catastrophic software errors. If programmed by software, this circuit will generate a
reset sequence if its timer is ever allowed to reach zero. Normally functioning software will reload the watchdog count before expiration of the count. The maximum
delay until a watchdog reset is greater than four hours.
2.3.4.
Battery Charger
The STMP35xx integrates support for LiIon charging protocols in USB or AC line
attached environments. When the 5 V source is detected on the VDD5V pin the
power management system automatically reconfigures to use the integrated linear
regulators to supply the core and I/O rails. Software can then enable the integrated
current source to provide battery charge current, as shown by the bold path in Figure 8. The variable current sources tappers of the charge current as it approaches
the maximum LiIon battery voltage. Software can then take over to control the final
“topping-off” algorithm, as desired.
USB Vbus 5v
Line 5V input
>4.35V
battery
VddIO
1/2/3*/4*
5V
detect
VDD5V
V Sense
battery
BATT
LRADC1
Linear Reg.
VddD
1/2/3
VddA
1/2/3
NC
NC
DCDC_VddIO DCDC_VddD DCDC_VddA*
Linear
linear charger
controller
Charge Current Path
3 Channel
Low Resolution ADC
DC-DC
#1 Control System
(boost mode)
DCDC_Batt
DCDC_Gnd
LRADC2
Regulator
Tem p.
VddXTAL=BATT
DSP_RESET
VddXTAL
battery
disabled during linear
battery charge
DCDC2_Vout*
RTC
DC-DC
#2 Control System
XTAL
OSC
DCDC2_Batt*
DCDC_mod0*
DCDC_mod2
DCDC_mod1*
DCDC2_Gnd*
* only available on 144-pin packages
MODE = 000, for LiIon, both converters in buck mode
Figure 8. Integrated LiIon Battery Charger
2.3.5.
USB Interface
The chip includes a Universal Serial Bus (USB) version 2.0 controller and integrated
UTMI PHY. The STMP35xx device interface can be attached to USB 2.0 hosts and
12
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
hubs running in the USB 2.0 High Speed mode at 480Mbit/second. It can be
attached to USB 2.0 Full Speed interfaces at 12Mbit/second. Of course, the USB
2.0 Full Speed mode allows the STMP35xx to attach to USB 1.1 compliant hosts
and hubs.
The USB interface is used to download digital music data or program code into
external memory and to upload voice recordings or MP3 encoded recordings from
external memory to the PC. Program updates can also be loaded into the flash
memory area using the USB interface.
Y-BUS
X-BUS
P-BUS
DSP CORE
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
DMA-BUS
USB System
Programmable
Registers
USB DMA Interface
Bus Interface
USB Controller
Programmable
Registers
USB Config
State Machine
End
Point
Info
End Point
Controller
Protocol Layer
ARC USB 2.0
Device Controller
Serial Interface Engine
PHY
Regs.
USB Xcvr
480MHz PLL
Integrated
USB 2.0 PHY
External USB 2.0
UTMI PHY
Figure 9. USB Interface Block Diagram
The Universal Serial Bus (USB) is a cable bus that supports data exchange
between a host computer and a wide range of simultaneously accessible peripherals. The attached peripherals share USB bandwidth through a host-scheduled,
5-35xx-D1-1.08-061705
13
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
token based protocol. The bus allows peripherals to be attached, configured, used
and detached while the host and other peripherals are in operation.
The USB subsystem is designed to make efficient use of system resources within
the SMTP35xx. It contains a random access DMA engine that reduces the interrupt
load on the DSP and reduces the total bus bandwidth that must be dedicated to servicing the eight on-chip physical endpoints
It is a dynamically configured port which can support up to 6 general use physical
endpoints and 8 logical endpoints, each of which may be configured for bulk, interrupt or isochronous transfers. The USB configuration information is read from onchip memory via the USB controller’s DMA.
Figure 9 shows a block diagram of the USB controller. This device makes extensive
use of the DMA to read and write the multiple buffers associated with all of the endpoints that it can have open at one time
2.3.6.
External Memory Interfaces.
Y-BUS
X-BUS
P-BUS
DSP CORE
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
DMA-BUS
EMC
Programmable
Registers
EMC
NAND Flash/
SmartMedia
State Machine
EMC
DMA Engine
EMC
Compact Flash
State Machine
Figure 10. External Memory Controller
The chip includes an external memory controller that has two major functional
modes: SmartMedia/NAND and CompactFlash. The SmartMedia/NAND flash interface provides a state machine that provides all of the logic necessary to perform
DMA functions between on-chip RAM and the flash. The CompactFlash interface
supports the CompactFlash Memory mode. This mode can be used to communicate
with standard CompactFlash (CF) devices such as CF Flash and the IBM MicroDrive. The CF Memory mode can be used to communicate with standard
ATA/ATAPI devices like CD-ROM and hard drives
14
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
The external memory controller can be described as three fairly independent
devices in one: a SmartMedia/NAND flash interface based on the STMP3410 flash
controller, a CompactFlash/NOR flash/IDE interface, and a new general purpose
flash state machine that can support the new 1Gb/die NAND flash devices with
128KByte block erase modes. These interfaces share the same device pins, some
registers and the DMA engine.
The interfaces use memory mapped registers to setup and control the transactions.
Data is always sent through DMA – there are no data registers that correspond to
the interface data bus. Transactions are always started with a kick bit. The interface
sets up the control lines and transfers data to/from the internal RAM. Once the
transaction is complete the interface signals the DSP with either a polled flag or an
interrupt.
2.3.7.
Hardware Acceleration for ECC for Robust External Storage
The forward error correction module is used to provide STMP35xx applications with
a reliable interface to various storage media, especially storage media that would
otherwise have unacceptable bit error rates. The ECC module comprises two different error correcting code processors:
• 1-bit correcting Samsung SSFDC (Hamming-code) encoder/decoder.
• 4-symbol correcting (9-bits/symbol) Reed-Solomon encoder/decoder.
The 1-bit hamming code is defined by Samsung for use with all SSSFDC compliant
NAND flash memories. This code is capable of correcting a single incorrect bit over
the block for which the ECC is valid (256 bytes per page).
The purpose of the Reed-Solomon decoder is to process a coded block (data block
followed by “parity” check data) to determine if there is an error and, if there are
errors, where they are located and how to correct them. The purpose of the ReedSolomon encoder is to read a block of 503-symbols from RAM, calculate and
append 8-parity symbols to form a 512-symbol RS-codeword.
The Hamming code error corrector is strong enough to detect two bits in error in 256
bytes and to correct 1-bit/256 byte errors. Both of these error correction
encoder/decoders use DMA transfers to move data to and from on-chip RAM completely in parallel with the DSP performing other useful work.
2.3.8.
Mixed Signal Audio Subsystem
The STMP35xx contains an integrated high quality mixed signal audio subsystem,
including high quality sigma delta D/A and A/D converters. The D/A is of course the
mainstay of the Audio Decoder/Player product application while the A/D is used for
Voice Record and MP3 Encode applications.
The chip includes a low noise headphone driver that allows it to directly drive low
impedance (8Ω or 16Ω) headphones. The direct drive, or “cap-less” mode removes
the need for large expensive DC blocking capacitors in the headphone circuit. The
headphone power amplifier can detect headphone shorts and report them via the
DSP interrupt system. A digitally programmable master volume control allows user
control of the headphone volume. Annoying clicks and pops are eliminated by zero
crossing updates in the volume/mute circuits and by headphone driver startup and
shutdown circuits.
There is an integrated analog mixer that drives the master volume control programmable gain amplifier. The chip provides for two stereo line level inputs and a mono
microphone input. The microphone circuit has a mono to stereo programmable gain
5-35xx-D1-1.08-061705
15
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
pre-amp and an optional microphone bias generator. The line inputs have programmable gain/attenuation and balance capability. The integrated sigma delta DAC has
a programmable gain/attenuation analog amplifier. The programmable gain/attenuation stage outputs from all three stereo inputs and from the DAC are mixed together
to drive the master volume control. There is an analog mux in front of the ADC that
can select any of the three input sources or the mixer output. The selected source is
then sent to the ADC Gain stage and from there to the ADC. The mixer can be independently powered down. In this configuration, the mixer is bypassed so that the
DAC can still play audio through the headphone driver saving power consumption
and improving the SNR and THD performance.
Line 1 In Gain
LINEIN
Mixer
Master
Volume
Mic In Gain
MICIN
+
FM In Gain
OUTPUT
Headphone
Driver
FMIN
DAC Gain
FROM
DAC
DMA
DAC
EN
CLK
ADC Input Mux
ADC GAIN
LINEIN
FMIN
MICIN
TO
ADC
DMA
ADC
Mic Bias
CLK
EN
Figure 11. Mixed Signal Audio Elements
2.3.9.
Filter Coprocessor
A filter coprocessor has been added to the STMP35xx to reduce the DSP load associated with filter calculations. Additional enhancements in the ADC and DAC buffer
management have greatly reduced the DSP work load as compared with
STMP3410 based applications. The concomitant reduction in DSP overhead yields
more available MIPS for more intensive software applications or allows the reduction in clock frequency/voltage and thereby dramatically extends the battery life.
The filter coprocessor is a DMA based engine that overlaps execution with the DSP.
2.3.10.
IDE/ATA Hard Drive Interface
The external memory controller interface supports the attachment of an ATA/IDE
hard drive device. This is particularly useful for one inch 2GByte hard drive and 2.5
inch 10GByte hard drive MP3 players. Hard drive and external SDRAM configurations are supported in the same application, i.e. large blocks of hard drive data can
be copied to SDRAM leaving the hard drive unused for most of the MP3 play time.
2.3.11.
SDRAM Interface
The STMP35xx contains an SDRAM controller that can be used to connect external
SDRAM memory chips. The controller is designed to work with 8 bit wide memory
16
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
systems. It supports SDRAM products from the 64Mbit, 128Mbit and 256MBit
JEDEC families. SDRAM memory systems as small as 8MBytes can be configured.
SDRAM memory subsystem are useful for applications that include CD-ROM or IDE
hard drives.
2.3.12.
Serial Peripheral Control Interface and I2C
The chip contains a four wire SPI bus. It can act as a master for this bus to control
other chips in the system, such as EEPROMs. It can also act as a slave on this bus
to allow a host processor to communicate with the STMP35xx. The STMP35xx
includes an enhanced SPI interface that provides DMA transfer support. In addition,
the chip contains a two wire SMB/I2C bus interface. It can act as either a slave or
master on the SMB interface.
2.3.13.
LCD/LED and GPIO
The STMP35xx contains 85 GPIO pins in the 144-pin package. Most digital pins that
are available for specific functions, e.g. SDRAM interface are also available as
GPIO pins if they are not otherwise used in a particular application.
Most LCD and LED displays can be directly controlled from the GPIO interface.
2.3.14.
PULSE WIDTH MODULATOR (PWM) CONTROLLER
The STMP35xx contains four PWM output controllers that can be used in place of
GPIO pins. Applications include LED brightness control and high voltage generators
for electroluminescent lamp (E.L.) display back lights. Independent output control of
each phase allows zero, one or hi-Z to be independently selected for the active and
inactive phases. Individual outputs can be run in lock step with guaranteed nonoverlapping portions for differential drive applications.
The controller does not use the DMA. Initial values of Period, Active, and Inactive
widths are set for each desired channel. The outputs are selected by phase and
then the desired PWM channels are simultaneously enabled. This effectively
launches the PWM outputs to autonomously drive their loads without further DSP
intervention.
Each PWM channel has a dedicated internal 12 bit counter which increments once
for each divided clock period presented from the clock divider. The internal counter
resets when it reaches the value stored in the channel control registers. The Active
flip flop is set to one when the internal counter reaches the value stored in a register.
It remains high until the internal counter exceeds the value stored in another register. These two value define the starting and ending points for the logically “active”
portion of the waveform. The actual state on the output for each phase, e.g. active
or inactive, is completely controlled by the active and inactive state values in the
channel control registers and can be: HIGH, LOW, or TRI-STATE.
5-35xx-D1-1.08-061705
17
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
2.4.
STMP35xx Family Members
The STMP35xx is available in a number of ordering options whose function content
is represented in Figure 12,
also see 33. “STMP35xx FAMILY MEMBER PART
NUMBERS & ORDERING INFORMATION” on page 391..
STMP3501
STMP3502†
STMP3505
STMP3506‡
STMP3510
STMP3520‡
MP3
MP3
WMA, WMA w/DRM
WMA, WMA w/DRM
‡
MP3 Encode
‡
MP3 Encode
LED/LCD Interface
LED/LCD Interface
EL Backlight
EL Backlight
USB Mass Storage
USB Mass Storage
RTC [Janus DRM]
RTC [Janus DRM]
Voice Record
Voice Record
FM Tuner
FM Tuner
Headphone Amp
& Driver
Headphone Amp
& Driver
50 Hour
50 Hour
Battery Life
Battery Life
1xAA, 1xAAA Batt.
1,2xAA, 1,2xAAA,
LiIon batteries
1,2xAA, 1,2xAAA,
LiIon batteries
USB 2.0 (Full Speed)
USB 2.0 (Full Speed)
High-Speed USB 2.0
SLC, MLC NAND
SLC, MLC NAND
SLC, MLC NAND
100-pin TQFP
100-pin TQFP
100-pin TQFP
144-pin BGA
144-pin BGA
MP3
WMA
LED/LCD Interface
USB Mass Storage
Voice Record
†FM Tuner
Headphone Amp
& Driver
20-25 Hour
Battery Life
STMP3550
STMP3560‡
MP3
WMA, WMA w/DRM
‡MP3 Encode
LED/LCD Interface
EL Backlight
USB Mass Storage
SDRAM Interface
RTC [Janus DRM]
Voice Record
FM Tuner
Headphone Amp
& Driver
50 Hour
Battery Life
1,2xAA, 1,2xAAA,
LiIon batteries
High-Speed USB 2.0
Battery Charging
SLC, MLC NAND
100-pin TQFP
144-pin BGA
Figure 12. STMP35xx Family Members
18
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
2.5.
Signal Pin Sharing Among Various Application Configurations
EMC NAND & SmartMedia
#2 #1
EMC CompactFlash memory & ATA/IDE #2
#1
SDRAM
#1 #1
CapLess Mode Headphone
Line In 1
#3
Table 2. Pin Sharing Constraints by Subsystem
Line In 1)
CapLess Mode Headphone
SDRAM
EMC CompactFlash Memory & ATA/IDE
EMC NAND & SmartMedia
A large number of the chips I/.O pins are shared between various functions. The
exact conflicts can be found in Table 492, “Pin Definition Table,” on page 377.
#3
NOTES:
#1: The EMC and SDRAM interfaces share a number of pins, including address and
data busses. While precluding exactly simultaneous accesses, careful attention to
chip selects and controller programing allows them to be used within the same
application, e.g. reading CompactFlash in IDE mode and writing SDRAM for HDD
MP3 player applications. Recommendation: use driver level mutual exclusion semaphores.
#2: Within the EMC devices, conflicts can occur between shared pins in the
NAND/SmartMedia interface and the CompactFlash interface. Use driver level
mutual exclusion semaphores.
#3: Capless headphone mode common amplifier output shares a pin with the analog line 1 Right input. In addition, the headphone common mode sense input shares
a pin with analog line 1 Left input. Only one of these uses can be designed into a
specific application.
2.6.
Additional Documentation
Additional documentation and information is available from SigmaTel, including an
extensive software development kit (SDK), application notes, reference schematics,
sample PCB board layouts, sample bill of materials, etc.
It is specifically recommended that the reader refer to the peripheral device include
files from the SDK. These files provide constant declarations for address offsets to
the registers defined in the following sections. Note that the name of each programmable register defined in this data sheet corresponds to a C language #define or
assembly language equate of the exact same name. In addition, these files contain
5-35xx-D1-1.08-061705
19
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
declarations that allow symbolic access to the individual bit fields within these registers. User programs can include all of these peripheral include files by simply including the file hw_equ.inc into their assembly files and hw_equ.h into their C files.
20
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
3.
CHARACTERISTICS/SPECIFICATIONS
3.1.
Absolute Maximum Ratings
PARAMETER
MIN
MAX
UNITS
Storage temperature
Battery Pin (BATT) DCDCMODE 000, 001, 010,011
Battery Pin (BATT) DCDCMODE 101,111
5 Volt Source Pin (VDD5V)
PSWITCH DCDCMODE 101,111
PSWITCH DCDCMODE 000,001,010,011(Note 1).
Analog supply voltage (VddA1, 2, VddHP, VddPLL)
Digital supply voltage (VddD1, 2, 3)
I/O Supply (VddIO1, 2, 3, 4)
DCDC converter #1 (DCDC_VddD)
DCDC converter #1 (DCDC_VddA)
DCDC converter #1 (DCDC_VddIO) DCDCMODE 000, 011
DCDC converter #1 (DCDC_VddIO) all other DCDCMODEs
DCDC converter #1 (DCDC_Batt)
-40
-0.3
-0.3
-0.3
-0.3
-0.3
Digital Supply
-0.3
-0.3
-0.3
-0.3
-0.3
-0.3
-0.3
125
4.242
1.98
5.25
VddIO
BATT
1.98
Analog Supply
3.63
VddD Rail
VddA Rail
4.242
3.6
max (VddIO,
BATT)
4.242
3.6
BATT
°C
V
V
V
V
V
V
V
V
V
V
V
V
V
VDDIO+0.3
V
VDDD+0.3
V
3.6
V
VDDA+0.3
V
DCDC converter #2 (DCDC2_Vout) DCDCMODE 000
-0.3
DCDC converter #2 (DCDC2_Vout) all other DCDCMODEs
-0.3
Input voltage on any DCDC_mod input pin relative to ground
-0.3
(DCDCMODE) (Note 2)
Input voltage on any digital I/0 pin relative to ground (DIO3)
-0.3
(Note 3)
Input voltage on any digital I/O pin in 1.8V mode relative to
-0.3
ground (DIO18) (Note 2)
Input voltage on USB D+, D- pins relative to ground (USBIO)
-0.3
(Note 3)
Input voltage on any analog pin relative to ground (AIO)
-0.3
(Note 3)
Table 3. Absolute Maximum Ratings
Note:
V
V
V
1. Pin sets for DCDCMODE, DIO3, DIO18, and AIO are defined in the pin list, see Table 492. “Pin Definition
2.
3.
Table” on page 377.
The maximum voltage limit on the PSWITCH pin can be achieved in DCDCMODE 000, 001, 010, and
011 by connecting the power switch to battery. A 20kohm resistor is placed between the switch and the
PSWITCH pin to limit the current into the pin. In DCDCMODE 111 and 101, the power switch is tied to
VDDIO and a 20kohm resistor is placed between the switch and the PSWITCH pin to limit the current
into the pin. The ESD protection diode limits the input voltage to an acceptable level as long as a 20kohm
resistor is placed in series with PSWITCH pin to limit the current.
Contact SigmaTel for extended temperature range options. In most systems designs, battery
and display specifications will limit the operating range to well within these specifications.
5-35xx-D1-1.08-061705
21
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
3.2.
Recommended Operating Conditions
PARAMETER
MIN
TYP
Ambient operating temperature (Note 3)
-10
Digital core supply voltage – VddD1, VddD2, VddD3 (Note 4)
1.35
Specification dependent on DSP frequency
Digital I/O supply voltage – VddIO1, VddIO2
2.9
3.0
Analog supply voltage – VddA(VddA1, VddA2, VddA3) for
Digital
player type applications where the mixer is powered down.
Supply
Specification dependent on maximum output power
Analog supply voltage – VddA(VddA1, VddA2, VddA3) for FM
1.45
or Voice Record type applications which utilize the mixer.
Specification dependent on maximum output power
Battery startup input voltage in 1xAA or 1xAAA mode
0.9
Full Scale Input Voltage:
Line Inputs (Note 5)
0.6
Mic Input
With 20 dB boost
0.06
Without 20 dB boost
0.6
Max Undistorted Full Scale Output Voltage with 16Ω load:
Headphone/Line Outputs (VddA = 1.8 V) (Note 6)
0.54
Headphone/Line Outputs (VddA = 1.38 V) (Note 6)
0.41
0.42
Crosstalk between output channels (16Ω loads at 1Khz)
-75
THD+N (16Ω headphone at 1 Khz) except STMP3501/3502
-70
(Note 10)
THD+N (10KΩ load at 1 Khz)
-87
THD+N (16Ω headphone at 1 Khz) for STMP3501/3502
-64
Analog line input resistance (Note 7)
25
Microphone input resistance
100
Analog output resistance
DAC SNR Idle Channel (Note 8)
96
DAC -60dB dynamic range (Note 8)
92
94
ADC SNR Idle Channel (Note 8)
90
ADC -60dB dynamic range (Note 8)
90
Line SNR (Note 8) except STMP3501/3502
92
94
Line SNR (Note 8) for STMP3501/3502
87
89
Standby Current (Note 9)
150
Table 4. Recommended Operating Conditions
Note:
22
MAX
70
Analog
Supply
3.6
1.98
UNITS
°C
V
1.98
V
-
V
-
Vrms
Vrms
-66
Vrms
Vrms
dB
dB
<1
-
250
V
V
dB
dB
kΩ
kΩ
Ω
dB
dB
dB
dB
dB
dB
uA
4.
Recommended operating voltages for DCLK can be found in Table 5 In all cases, design must allow for
board and bypass design variations.
5. At 1.38VddA max input is 0.45Vrms
6. Undistorted is considered to be THD+N < -64dB with a 1kHz 0dB sine wave.
7. Input resistance changes with volume setting:
10KΩ at +12dB, 25KΩ at 0dB, 50KΩ at -34.5dB
8. Measured “A weighted” over a 20 Hz to a 20 kHz bandwidth, relative to full scale output voltage (1.8V)
9. The chip consumes current when in the “OFF” mode to keep the crystal oscillator and the real time clock
running. With a typical 2850mAh AA battery, the standby current would take more than 1 year to drain the
battery fully. It also is possible to design a system that disables the crystal oscillator and real time clock to
achieve a much lower OFF current.
10. The BGA package reduces THD performance by approximately 4dB.
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
The following table can be used to select a proper setting for VddD and VddD
brownout voltages based on standard analysis of worst case design and characterization data.
MAX
DCLK
TARGET
75.6MHz
69.6MHz
64.8MHz
60.0MHz
48.0MHz
39.6MHz
Min.
VddD
HW_DCDC_VDDD_
VOLTAGE_LEVEL
Corresponding
HW_DCDC_VDDD_
VddD
BROWNOUT_LEVEL
Brownout Voltage
1.92 V
11100
1.85 V
11010
1.82 V
11001
1.76 V
10111
1.73 V
10110
1.66 V
10100
1.63 V
10011
1.54 V
10000
1.4 V
01100
1.34 V
01010
1.37 V
01011
1.28 V
01000
Table 5. Recommended Operating Conditions for specific dclk targets
.
5-35xx-D1-1.08-061705
23
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
3.3.
DC Characteristics
PARAMETER
MIN
Power Dissipation, VddD = 1.37 V, VddA = 1.37 V, VddIO =
3.05 V DCDC mode = 2-Channel Boost, DCLK = 36 MHz on
PLL, USB off, Application = MP3 Play, minimum power
configuration selected.
ViH (DIO3) - Input high voltage for DIO3 digital I/O pin set, in
2.0
3.3 Volt mode.
VIL (DIO3) - Input low voltage for DIO3 digital I/O pin set in 3.3
Volt mode.
VIH (DIO18) - Input high voltage for DIO18 digital I/O pin set in 0.7*VddD
1.8 Volt mode.
VIL (DIO18) - Input low voltage for DIO18 digital I/O pin set in
1.8 Volt mode.
VOH (DIO3) - Output high voltage for DIO3 digital I/O pin set in 0.8*VddIO
3.3 Volt mode, 4mA mode.
VOH (DIO3) - Output high voltage for DIO3 digital I/O pin set in 0.8*VddIO
3.3 Volt mode, 8mA mode.
VOL (DIO3) - Output low voltage for DIO3 digital I/O pin set in
3.3 Volt mode.
VOH (DIO18) - Output high voltage for DIO18 digital I/O pin
VddD - 0.4
set in 1.8 Volt mode.
VOL (DIO18) - Output low voltage for DIO18 digital I/O pin set
in 1.8 Volt mode.
Table 6. DC Characteristics
3.4.
TYP
80
MAX
UNITS
mW
V
0.8
V
V
0.3*VddD
V
V
V
0.4
V
V
0.4
V
Restrictions on Approved Usage of SigmaTel Parts
SigmaTel Products are not designed or intended for use in life support appliances,
or systems where malfunction of a SigmaTel product can reasonably be expected to
result in personal injury or death nor are they intended for use in any application
where malfunction of a SigmaTel product can reasonably be expected to result in
environmental or other subsequent damage.
24
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
4.
DSP CORE
The on-chip DSP core is modeled exactly after the Motorola DSP56004. It supports
the identical instruction set, registers, addressing modes, etc., as the DSP56000
family of digital signal processors. Figure 13 shows the DSP architecture.The DSP
core is a general-purpose 24-bit DSP especially suited to high fidelity digital audio
applications for very low power/energy environments.
ON-CHIP
PROGRAM
RAM/ROM
ON-CHIP
Y
RAM
ON-CHIP
X
RAM
XAB
ADDRESS
GENERATION
UNIT
YAB
PAB
OnCE
Debugger
Interface
Peripherals
XDB
YDB
Data
Bus
Switch
PDB
GDB
PROGRAM
Interrupt
Controller
PROGRAM
DECODE
CONTROLLER
PROGRAM
ADDRESS
GENERATOR
DATA ALU
24 x 24 +56 --> 56-bit MAC
Two 56 bit Accumulators
Mode GPIO bits
IVL[6:0],IRQA, IRQB, NMI
RESET
Figure 13. DSP Architecture
5-35xx-D1-1.08-061705
25
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
The central components are:
• Data Busses (XD,YD,PD)
• Address Busses (XA,YA,PA)
• Data Arithmetic Logic Unit (Data ALU)
• Address Generation Unit (AGU)
• Program Control Unit (PCU)
• On-Chip Program ROM
• On-Chip X,Y,P RAM
• On-Chip Emulation circuitry
The DSP is organized around the registers of three independent execution units: the
PCU, the AGU and the data ALU. Data movement between the execution units
occurs over four bidirectional 24-bit busses: the X data bus (XDB), the Y data bus
(YDB), the program data bus (PDB) and the global data bus (GDB). Certain instructions treat the X and Y data buses as one 48-bit data by concatenating them. Data
transfers between the data ALU and the X data memory or Y data memory occur
over the XDB and YDB respectively.
The bus structure supports general register-to-register, register-to-memory, and
memory-to-register data movement. It can transfer up to two 24-bit words or one 56bit word in the same instruction cycle.
Transfers between busses occur in the internal bus switch. The internal bus switch,
which is similar to a switch matrix, can connect any two internal busses without adding pipeline delays. Thus greatly simplifying the programming model.
The bit manipulation unit is located in the bus switch so that it can access each
memory space. The bit manipulation unit performs bit operations on memory locations, address registers, control registers and data registers over the XDB, YDB and
GDB.
The data ALU performs all of the arithmetic and logical operations on data operands. It consists of four 24 bit input registers, two 48-bit accumulators and two 8-bit
accumulator extension registers, an accumulator shifter, two data bus shifter/limiter
circuits, and a parallel single cycle, non-pipelined multiply-accumulator (MAC) unit.
The address generation unit (AGU) performs all of the address storage and address
generation computations necessary to indirectly address data operands in memory.
It operates in parallel with other DSP resources to minimize address generation
overhead and keep the supply of data operands fed to the data ALU. The AGU has
two identical address arithmetic units that can generate two 16-bit addresses every
instruction cycle. Each of the arithmetic units can perform one of three types of
address arithmetic: linear, modulo and reverse-carry.
The program control unit performs instruction prefetch, instruction decode, hardware DO loop control, and interrupt/exception processing. It consists of three components: the program address generator, the program decode controller, and the
program interrupt controller. It contains a 15-level by 32 bit system stack memory
and the following directly addressable registers: the program counter (PC), loop
address (LA), loop counter (LC), status register (SR), and the operating mode register (OMR).
The DSP core responds to 7 interrupt vector level inputs (IVL[6:0], two peripheral
interrupts (IRQA, IRQB) and a non-maskable interrupt (NMI).
26
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
4.1.
Operating Mode Register
The organization of the operating mode register is shown below. The operating
mode register determines chip configuration including boot modes, and memory
configuration. The HW_OMR is a core register that is accessible by special DSP
instructions. It therefore has no address.
HW_OMR
BITS LABEL RW RESET
23:8
7
6
RSRVD R
0
RSRVD R
0
SD
RW 0
5
4
RSRVD R
0
MC
RW GP0
3:2
1
RSRVD R
0
MB
RW GP1
0
MA
RW 1
SPECIAL
DEFINITION
Reserved – Must be written with 0.
Reserved – Must be written with 0.
Stop Delay – This bit is exported from the core as an output. It can be used when
waking up from the STOP low power standby mode. If this bit is set, then when an
IRQA interrupt occurs to wake up the core from the STOP state, the clock control
circuitry will wait a time period (e.g. 65536 clock cycles) before allowing the clocks
back in to the DSP core. This can be used, for example, to restabilize a PLL clock
oscillator. If this bit has a zero value, then the clocks will be allowed back into the core
immediately after the occurrence of the IRQA interrupt, thus implementing a “warm
boot” from the STOP low power standby state.
Reserved – Must be written with 0.
Operating Mode C – This bit is used to configure the boot mode for the STMP35xx.
When the hardware reset is active, this bit samples the state of GP0 pin. Once the
boot code executes, it can check the state of this bit in order to make decisions about
what type of boot mode to perform.
Reserved – Must be written with 0.
Operating Mode B – This bit is used to configure the boot mode for theSTMP35xx.
When the hardware reset is active, this bit samples the state of the GP1 pin. Once the
boot code executes, it can check the state of this bit in order to make decisions about
what type of boot mode to perform.
Operating Mode A – This bit is used to choose between Boot ROM and Program
Memory for instruction fetches and read accesses. When this bit is set, as it is after
hardware reset, the Boot ROM space is activated and any fetches or read accesses to
the P: space will refer to the on-chip ROM. When this bit is a zero, the Program RAM
Memory space is enabled instead of the Boot ROM memory space and any fetch or
read access to P: space will refer to the on-chip RAM. Writes to P: space always
access the program RAM regardless of the state of the MA bit. It is not possible to
write to the program ROM. This bit affects ROM access in region P:$0000 through
P:$3FFF. Accesses to the high mapped region of ROM, P:$C000 through P:$FFFF is
controlled by HW_RAM_ROM_CFG_PROMIE.
Table 7. Operating Mode Register Description
4.2.
General Debug Register
The HW_GDBR Register is also mapped into the X Peripheral I/O space. This register is used as a gateway between the DSP and the Debug port. For instance, when
displaying the states of the internal registers and memory of the DSP core, the DSP
moves the data to this register and the data is then shifted out the OnCE_DSO pin.
The HW_GDBR register operation is controlled automatically by the emulator and
the debug circuitry within the core. An added feature of the Debug Unit is that the
emulator cannot access the debug unit unless a write to the HW_GDBR Register is
executed by the DSP (normally in the boot code).
5-35xx-D1-1.08-061705
27
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
HW_GDBR
2
3
2
2
2
1
X:$FFFC
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0 0
1 0
VALUE
Table 8. HW_GDBR
BITS LABEL RW RESET
DEFINITION
23:0 VALUE RW 0
Value to be read by the debugger or debugger value to be read by a DSP instruction
from the debugger.
Table 9. General Debug Register (GDBR)
4.3.
OnCE (On-Chip Emulator) Debug Interface
The DSP on-chip emulation (OnCE) circuitry provides a sophisticated debugging
tool that allows simple, inexpensive, and speed independent access to the processor’s internal registers, memories and peripherals. OnCE provides software engineers with access to the internal state including the addresses of the last five
instructions and provides the ability to modify that state, and single step the processor. OnCE capabilities are accessed through a four pin interface
1. Debug Serial input (OnCE_DSI)
2. Debug Serial Clock (OnCE_DSK)
3. Debug Serial Output (OnCE_DSO)
4. Debug Request Input (OnCE_DRN)
The OnCE controller and serial interface consists of the following blocks: OnCE
command register, bit counter, OnCE decoder and the status/control register. For a
block diagram, see Figure 14. “OnCE Interface Block Diagram” on page 28.
OnCE_DSI
OnCE COMMAND REGISTER
OnCE_DSK
ISBKPT
ISDR
ISTRACE
OnCE
DECODER
bit 7
Bit Counter
bit 23
OnCE_DRN
ISSWDBG
Status and Control
Register
REG
READ
REG
WRITE
OnCE_DSO
Figure 14. OnCE Interface Block Diagram
28
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
The OnCE Command Register is an 8 bit shift register that receives the serial data
from the OnCE_DSI pin. It holds the 8 bit commands to be used as input for the
OnCE controller.
OnCE Command Register
BITS LABEL RW RESET
DEFINITION
7
RW
RW
The read/write bit specifies the direction of the data transfer. For zero, write the data
associated with the command into the register specified in the RS field. For one, read
the data contained in the register specified in the RS field.
6
GO
RW
If the GO bit is set, the chip will execute the instruction which resides in the PIL
register. To execute the instruction, the processor leaves the debug mode, and the
status is reflected on the OS0,OS1 pins. The processor will return to the debug mode
immediately after executing the instruction.
5
EX
RW
If the Exit Command bit is set, the processor will leave the debug mode and resume
normal operation. The Exit command is executed only if the Go command was issued
and the operation is a write to OPDBR or a read/write to” No Register Selected”.
4:0
RS
RW
Register Select field
Table 10. OnCE Command Register
The Register Select field (RS[4:0]) selects one of 32 OnCE debug registers to be
read or written.
RS[4:0]
00000
00001
00010
00011
0010X
00110
00111
01000
01001
01010
01011
01100
01101
01110
01111
10000
10001
10010
10011
101XX
11XX0
11111
REGISTER SELECTED
OnCE Status and Control Register
Memory Breakpoint register
Reserved
Trace Counter
Reserved
Memory Upper Limit
Memory Lower Limit
GDB Register NOTE: this register can be read or written by the DSP instructions
PDB Register
PAB Register for Fetch
PIL register, next instruction from debugger comes from here
Clear Memory Breakpoint Counter
Reserved
Clear Trace Counter
Reserved
Reserved
Program Address Bus FIFO and Increment Counter
Reserved
PAB Register for Decode
Reserved
Reserved
No Register Selected
Table 11. OnCE Register Selects
5-35xx-D1-1.08-061705
29
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
5. ON-CHIP MEMORY SUBSYSTEM
The chip includes 96 kwords of on-chip RAM (96k x 24 bits = 2.25 Mbits) that is
used for program and data storage, and 16K words of on-chip ROM (16K x 24 bits =
384kbits) that is used for the code that boots the device (see Section 20 for more
details on boot modes and the contents of the on-chip ROM). The on-chip ROM is
mapped at the address range P:$0000-$3FFF at reset, it can also be configured to
be mapped at address range P:$CFFF-$FFFF, or it can be disabled.
The on-chip RAM is organized into two banks of 48K words each, called PXRAM
and PYRAM. PXRAM can be mapped into the DSP P memory space, starting at
P:$0000, or into the DSP X memory space, starting at X:$0000. PYRAM can be
mapped into the DSP P memory space, starting immediately after the end of the
PXRAM memory, or into the DSP Y memory space, starting at Y:$0000. Both
PXRAM and PYRAM memory can be allocated to the DSP P, X or Y memory
spaces in 8K word increments, from a minimum of 0K words to all available memory.
The memory configuration is controlled by the PX & PY Memory Configuration registers documented below. There are no hardware safeguards against improper programming of these registers. It is possible to allocate less than all of the on-chip
RAM, unallocated memory will then be invisible to the DSP.
PXRAM Configuration Register
HW_PXCFG
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
PXXSIZE
2
3
X:$FFE8
0
7
0
6
0
5
0
4
0
3
0
2
0 0
1 0
PXPSIZE
5.1.
Table 12. HW_PXCFG
BITS
LABEL
23:14
13:8
RSRVD
PXXSIZE
R
RW
RW
0
011000
RESET
7:6
5:0
RSRVD
PXPSIZE
R
RW
0
011000
DEFINITION
Reserved – Must be written with 0.
Number of kwords of PXRAM that is mapped in the DSP X memory
space. Initialize 24KW to X RAM. This six bit field represents 1K
increments of RAM. It must be allocated in 8K word chunks.
Reserved – Must be written with 0.
Number of kwords of PXRAM that is mapped in the DSP P memory
space. Initialize 24KW to Y RAM. This six bit field represents 1K
increments of RAM. It must be allocated in 8K word chunks.
Table 13. PXRAM Configuration Register Description
30
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
PYRAM Configuration Register
HW_PYCFG
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
PYYSIZE
2
3
X:$FFE9
0
7
0
6
0
5
0
4
0
3
0
2
0 0
1 0
PYPSIZE
5.2.
Table 14. HW_PYCFG
BITS
LABEL
RW
RESET
23:14
13:8
RSRVD
PYYSIZE
R
RW
0
011000
7:6
5:0
RSRVD
PYPSIZE
R
RW
0
011000
DEFINITION
Reserved – Must be written with 0.
Number of kwords of PYRAM that is mapped in the DSP Y memory
space. Initialized to 24KW Y RAM.This six bit field represents 1K
increments of RAM. It must be allocated in 8K word chunks.
Reserved – Must be written with 0.
Number of kwords of PYRAM that is mapped in the DSP P memory
space. Initialized to 24KW to P RAM. This six bit field represents 1K
increments of RAM. It must be allocated in 8K word chunks.
Table 15. PYRAM Configuration Register Description
The PXRAM bank is accessible to DSP P space accesses, DSP X space accesses,
and DMA accesses. The PYRAM bank is accessible to DSP P space accesses,
DSP Y space accesses, and DMA accesses. PXRAM & PYRAM are made up of 3
physical blocks of 16K words each for a total of 48K words each. Since the allocation of on-chip RAM to DSP P space, X space, & Y space is in 8K Word increments,
it is possible for a single physical memory block in the PXRAM bank to be accessed
by the DSP P, DSP X, and DMA busses at the same time, similarly for the PYRAM
bank. When this happens, the memory interface control logic steals one or more
clock cycles from the DSP to allow all of the accesses to complete in separate clock
cycles. In case of conflict, DMA accesses take priority over DSP accesses, and DSP
program accesses take priority over DSP data accesses.
It is possible to eliminate cycle steals because of DSP P space and DSP X or Y conflicts by allocating PXRAM & PYRAM in increments of 16K words, this means that
each physical memory block is allocated to DSP P, X or Y space. It is not possible to
eliminate the cycle steal that happens because a DMA access conflicts with a DSP
access to memory. In order to allow the programmer to monitor total number of
cycle steals are happening, the contents of the Cycle Steal Count Register
(HW_CYCSTLCNT) will increment whenever a cycle is stolen for a memory access
conflict.
Figure 15 below shows an example of how the PXRAM and PYRAM memory banks
can be allocated. In this example, PXRAM is allocated as 24K words P, &
5-35xx-D1-1.08-061705
31
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
24K words X, and PYRAM is allocated as 24K words P, & 24K words Y, for a total of
48K words of P, 24K words of X, and 24K words of Y.
PYMEM
PXMEM
X:$0000
physical:$BFFF
Y:$0000
8 kwords
per
instance
Y:$1FFF
Y:$2000
X:$1FFF
X:$2000
24 kwords
X memory
from PXRAM
8 kwords
per
instance
Y:$3FFF
Y:$4000
X:$3FFF
X:$4000
X:$5FFF
P:$5FFF
24kwords
Y memory
from PYRAM
48 kwords
total per
memory
bank
8 kwords
per
instance
Y:$5FFF
P:$BFFF
8 kwords
per
instance
P:$A000
P:$4000
8 kwords
per
instance
24 kwords
P memory
from PXRAM
24 kwords
P memory
from PYRAM
P:$8000
P:$2000
8 kwords
per
instance
P:$0000
physical:$0000
P:$6000
Figure 15. On-Chip RAM Organization
When the memory is configured as shown in Figure 15, the DSP’s view of the memory map will be as shown below in Figure 16. The DSP P address space has 48K
words of RAM as shown in the range P:$0000 to P:$BFFF. If the ROM is enabled
then it appears in the DSP P address space at P:$C000 to P:$FFFF. The DSP X
space has 24K words of RAM at X:$0000 to X:$5FFF, likewise the DSP Y space has
24K words of RAM at Y:$0000 to Y:$5FFF. The address ranges X:$6000 to
X:$EFFF, and Y:$6000 to Y:$FFFF are not populated with any memory. The
address range P:$C000 to P:$FFFF is unpopulated if the ROM is not enabled. The
address range X:$F000 to X:$FFFF is reserved for on-chip peripherals.
In the above example memory configuration, it is possible for cycle steals to happen
because of either DSP access conflicts, or because of DMA access conflicts. The
areas in the memory map where DSP access conflicts can occur are cross hatched
in Figure 16. If the DSP attempts to access an address in the range P:$4000 to
P:$5FFF at the same time as an address in the range X:$4000 to X:$5FFF, a stall
cycle will occur. If the DSP attempts to access an address in the range P:$A000 to
32
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
P:$BFFF at the same time as an address in the range Y:$4000 to Y:$5FFF, a stall
cycle will also occur.
DSP
P space
DSP
Y space
DSP
X Space
X:$FFFF
peripherals
X:$F000
P:$FFFF
Y:$FFFF
16 kwords
from PROM
P:$C000
P:$BFFF
not
populated
not
populated
P:$A000
P:$9FFF
24 kwords
from PYMEM
64 kwords
total
P:$8000
P:$7FFF
P:$6000
P:$5FFF
X:$6000
X:$5FFF
Y:$6000
Y:$5FFF
P:$4000
P:$3FFF
X:$4000
X:$3FFF
Y:$4000
Y:$3FFF
24 kwords
from PXMEM
24 kwords
from PXMEM
24 kwords
from PYMEM
P:$2000
P:$1FFF
X:$2000
X:$1FFF
Y:$2000
Y:$1FFF
P:$0000
X:$0000
Y:$0000
Figure 16. DSP Memory MAP
It is possible to reallocate on-chip RAM at any time, however doing so is dangerous
and should be done carefully. In particular, the DSP P space memory that is allocated from the PYMEM memory bank will move within the P address space if the
allocation of PXRAM memory is changed.
5.3.
On-Chip ROM
When the chip comes out of reset, the OMR MODE A bit is set to one, see 4.1.
“Operating Mode Register” on page 27. This bit, when set to one, enables the OnChip ROM to be read instead of whatever PXRAM or PYRAM is assigned to the first
16KW of the P address space, i.e. P:$0000 through P:$3FFF. For the default
PXRAM/PYRAM allocation described above, the cross hatched region of Figure 17
shows the 16KW of PXRAM space that are overlaid by ROM. Whenever a P space
access is made in the range P:$0000 through P:$3FFF, AND OMR_MODE_A =1
AND a read cycle is performed, then ROM data is presented to the DSP P-bus. If
mode A is not set then reads in this range access the allocated PXRAM data. Whenever a P-bus write cycle is performed within this address range it is always directed
5-35xx-D1-1.08-061705
33
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
to PXRAM. Thus at power on time, Instructions and P-space read data are read
from the On-Chip ROM while P-space write data is written to the PXRAM.
OMR_MODE_A =0 & READ
& HW_RAM_ROM_CFG_ROMIE =1
DSP
P space
DSP
Y space
DSP
X Space
P:$FFFF
Y:$FFFF
X:$FFFF
peripherals
X:$F000
16 kwords
from PROM
P:$C000
P:$BFFF
not
populated
not
populated
P:$A000
P:$9FFF
24 kwords
from PYMEM
64 kwords
total
P:$8000
P:$7FFF
P:$6000
P:$5FFF
P:$4000
P:$3FFF
24 kwords
from PXMEM
P:$3FFF
X:$6000
X:$5FFF
Y:$6000
Y:$5FFF
X:$4000
X:$3FFF
Y:$4000
Y:$3FFF
24 kwords
from PXMEM
24 kwords
from PYMEM
P:$2000
P:$1FFF
16KW
ROM
X:$2000
X:$1FFF
Y:$2000
Y:$1FFF
P:$0000
P:$0000
X:$0000
Y:$0000
OMR_MODE_A =0 | WRITE
OMR_MODE_A =1 & READ
Figure 17. DSP Memory MAP with On-Chip ROM
The boot strap code, which begins execution in the on-chip ROM takes advantage
of these mode settings as follows: The boot code reads the new program code in
from the NAND Flash, or other source. It copies the data, as described in the boot
mode section below, to the appropriate memory locations. If a block is copied to Pspace in the range P:$0000 through P:$3FFF then it is written to PXRAM as if the
ROM were not there. When the boot load is complete, the DSP software must turn
off the ROM and begin execution of the loaded code. All code loaded by the ROM
Boot begins execution at location P:$0000 after the ROM is turned off. The ROM
boot loader calls a routine to copy the program and branch to P:$0000, as follow:
jsr (R5)
; call the boot copy program
bclr #0,HW_OMR ; turn off the ROM
jmp $0
; begin execution at the first location in RAM
34
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
Recall that the DSP is pipelined and that the reset of the OMR bit happens when the
bclr instruction reaches the execute stage. By that time, the jmp $0 has already
been fetched into the pipeline and will fetch a RAM location instead of a ROM location for its next instruction.
5.4.
On-Chip Memory Configuration Register
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0 0
1 0
PXRAMCT
PROMIE
1
7
PYRAMCT
1
8
PROMCT
1
9
RAMRM
2
0
RAMAWT
2
1
ROM_CLK_EN
2
2
PXRAM_CLK_EN
2
3
PYRAM_CLK_EN
HW_RAM_ROM_CFG X:$FFED
Table 16. HW_RAM_ROM_CFG
BITS
LABEL
RW RESET
DEFINITION
23:22 RSRVD
R 0
Reserved – Must be written with 0.
21
PYRAM_CLK_EN RW 1
PYRAM Clock Enable – This bit enables or disables the clock to the PYRAM
array. This clock must be turned on for normal operation.
0
Disable clock for the PYRAM array
1
Enable clock for the PYRAM array
20
PXRAM_CLK_EN RW 1
PXRAM Clock Enable – This bit enables or disables the clock to the PXRAM
array. This clock must be turned on for normal operation.
0
Disable clock for the PXRAM array
1
Enable clock for the PXRAM array
19
ROM_CLK_EN
RW 1
PROM Clock Enable – This bit enables or disables the clock to the on-chip
ROM. This clock should be turned off to save power if the ROM is not being
used. It must be turned on for the correct operation of the on-chip ROM.
0
Disable clock for the on-chip ROM
1
Enable clock for the on-chip ROM
18
PROMIE
RW 1
PROM Image Enable – After reset, the on-chip ROM is located at the address
range P:$0000-$3FFF. Once the bootloader in the ROM clears the MODEA bit
in the Operating Mode Register (HW_OMR), this address range reverts to onchip RAM. This mode bit enables the contents of the ROM to be viewed at the
address range P:$C000-$FFFF, irrespective of the state of the MODEA bit of
the Operating Mode Register. Since this address range can also be used by onchip RAM, it is necessary to be able to disable the ROM in this address range.
0
Disable P:$C000-$FFFF image for on-chip ROM
1
Enable P:$C000-$FFFF image for on-chip ROM
NOTE: This bit does not affect ROM accesses in the lower region of P: space
between P:$0000 and P:3FFF which is controlled solely by HW_OMR_MA.
17
RSRVD
R 0
Reserved – Must be written with 0.
16
RAMAWT
RW 0
RAM AWT Mode – This register is used to select the asynchronous write
through (AWT) mode for the on-chip RAM. This is a test mode that should not
be used in normal operation.
0
Disable AWT mode
1
Enable AWT mode
Table 17. On-Chip Memory Configuration Register Description
5-35xx-D1-1.08-061705
35
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
15:12 RAMRM
11:8
PROMCT
7:4
PYRAMCT
3:0
PXRAMCT
RW RESET
DEFINITION
RW 0000
RAM Read Margin – This register is used to optimize the read performance of
all of the on-chip RAM. The optimum value of this register will be determined
by SigmaTel, and should not be modified.
RW 1000
PROM clock tune – This register is used to optimize the placement of the
clock to the on-chip ROM. The optimum value of this register will be
determined by SigmaTel, and should not be modified.
RW 1000
PYRAM Clock Tune – This register is used to optimize the placement of the
clock to the PYRAM block. The optimum value of this register will be
determined by SigmaTel, and should not be modified.
RW 1000
PXRAM Clock Tune – This register is used to optimize the placement of the
clock to the PXRAM block. The optimum value of this register will be
determined by SigmaTel, and should not be modified.
Table 17. On-Chip Memory Configuration Register Description (Continued)
5.5.
PXRAM0 Repair Register
HW_PXRAM0_RPR
BITS
23:0
LABEL
REPAIR
RW
RW
X:$F5A3
RESET
$000000
DEFINITION
Repair register, each bit controls a 2:1 mux on the input and output of the
PXRAM0 array. If the bit is set to one, then the corresponding bit in the
array is substituted by the next higher significant bit in the array. When
set to zero, there is a 1:1 correspondence between the control bit
position and the array bit position. This register controls the data bus for
the 16K Word SRAM bank covering X:$8000 through X:$BFFF.
Table 18. PXRAM0 Repair Register Description
5.6.
PXRAM1 Repair Register
HW_PXRAM1_RPR
BITS
23:0
LABEL
REPAIR
RW
RW
X:$F5A4
RESET
$000000
DEFINITION
Repair register, each bit controls a 2:1 mux on the input and output of the
PXRAM1 array. If the bit is set to one, then the corresponding bit in the
array is substituted by the next higher significant bit in the array. When
set to zero, there is a 1:1 correspondence between the control bit
position and the array bit position. This register controls the data bus for
the 16K Word SRAM bank covering X:$4000 through X:$7FFF.
Table 19. PXRAM1 Repair Register Description
5.7.
PXRAM2 Repair Register
HW_PXRAM2_RPR
BITS
23:0
LABEL
REPAIR
RW
RW
RESET
$000000
X:$F5A5
DEFINITION
Repair register, each bit controls a 2:1 mux on the input and output of the
PXRAM2 array. If the bit is set to one, then the corresponding bit in the
array is substituted by the next higher significant bit in the array. When
set to zero, there is a 1:1 correspondence between the control bit
position and the array bit position. This register controls the data bus for
the 16K Word SRAM bank covering X:$0000 through X:$3FFF.
Table 20. PXRAM2 Repair Register Description
36
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
5.8.
PYRAM0 Repair Register
HW_PYRAM0_RPR
BITS
23:0
LABEL
REPAIR
RW
RW
X:$F5A6
RESET
$000000
DEFINITION
Repair register, each bit controls a 2:1 mux on the input and output of the
PYRAM0 array. If the bit is set to one, then the corresponding bit in the
array is substituted by the next higher significant bit in the array. When
set to zero, there is a 1:1 correspondence between the control bit
position and the array bit position. This register controls the data bus for
the 16K Word SRAM bank covering Y:$8000 through Y:$BFFF.
Table 21. PYRAM0 Repair Register Description
5.9.
PYRAM1 Repair Register
HW_PYRAM1_RPR
BITS
23:0
LABEL
REPAIR
RW
RW
X:$F5A7
RESET
$000000
DEFINITION
Repair register, each bit controls a 2:1 mux on the input and output of the
PYRAM1 array. If the bit is set to one, then the corresponding bit in the
array is substituted by the next higher significant bit in the array. When
set to zero, there is a 1:1 correspondence between the control bit
position and the array bit position. This register controls the data bus for
the 16K Word SRAM bank covering Y:$4000 through Y:$7FFF.
Table 22. PYRAM1 Repair Register Description
5.10. PYRAM2 Repair Register
HW_PYRAM2_RPR
BITS
23:0
LABEL
REPAIR
RW
RW
RESET
$000000
X:$F5A8
DEFINITION
Repair register, each bit controls a 2:1 mux on the input and output of the
PYRAM0 array. If the bit is set to one, then the corresponding bit in the
array is substituted by the next higher significant bit in the array. When
set to zero, there is a 1:1 correspondence between the control bit
position and the array bit position. This register controls the data bus for
the 16K Word SRAM bank covering Y:$0000 through Y:$3FFF.
Table 23. PYRAM2 Repair Register Description
5-35xx-D1-1.08-061705
37
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
6.
CHIP WIDE PROGRAMMABLE CONTROL REGISTERS
6.1.
Revision Register
The Revision Register reports the Device ID and revision to the software. In addition, it shows the strap state of the DCDCMODE[2:0] bits. This register is read only.
The organization of the Revision Register is shown below.
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
RMN
2
2
RMJ
2
3
X:$FA02
DCDCMODE
HW_REVR
Table 24. HW_REVR
BITS
LABEL
RW
RESET
23:8
RMJ
R
$3500
7:5
DCDCMODE
R
4:0
RMN
R
depends on
DCDC_mod pin
strapping
depends on silicon
revision
DEFINITION
Revision Major ID – This is the device part number
in binary coded decimal:
STMP35xx
$3500
DCDCMODE[2:0] pin state, see 30. “DC-DC
CONVERTER” on page 339.
Revision Minor ID – Device revision number
$00 TA1 revision
Table 25. Revision Register Description
38
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
6.2.
Reset Control Register
The organization of the Reset Control Register is shown below.
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
STKLVL
1
6
SRST
1
7
IRQA
SOVFLEN
1
8
IRQB
IRQB2NMI
1
9
NMI
2
0
SUNFLLVL
2
1
SUNFLEN
2
2
SOVFLLVL
2
3
SUNFL
X:$FA01
SOVFL
HW_RCR
Table 26. HW_RCR
BITS
LABEL
RW RESET
23
SOVFL
R
0
22
SUNFL
R
0
21
IRQB2NMI
RW 0
20
SOVFLEN
RW 0
19:16 SOVFLLVL RW 1111
15
SUNFLEN RW 0
14:11 SUNFLLVL RW 0000
10
NMI
R
1
DEFINITION
Stack Overflow Status Bit – This bit indicates that the current stack depth is
equal to or greater than the value of the SOVFLLVL Stack Overflow Interrupt
Level field. This is set or cleared independently of the SOVFLEN Stack Overflow
Interrupt Enable field. In other words, once set this bit can only be cleared by
fixing the stack overflow event by removing words from the stack.
Stack Underflow Status Bit – This bit indicates that the current stack depth is
equal to or less than the value of the SUNFLLVL Stack Underflow Interrupt Level
field. This is set or cleared independently of the SUNFLEN Stack Underflow
Interrupt Enable field. In other words, once set this bit can only be cleared by
fixing the stack underflow event by adding words to the stack.
Redirect battery+VddD+VddIO Brownout Interrupt to NMI
0
battery+VddD+VddIO brownout interrupt on IRQB only
1
battery+VddD+VddIO brownout interrupt on IRQB & NMI
(see brownout Figure 132 for details)
Stack Overflow Interrupt Enable
0
Stack overflow interrupt disabled
1
Stack overflow interrupt enabled
Stack Overflow Interrupt Level
Stack Underflow Interrupt enable
0
Stack underflow interrupt disabled
1
Stack underflow interrupt enabled
Stack Underflow Interrupt Level
NMI Interrupt – An NMI interrupt will be generated by a stack over-/underflow
event. If the IRQB2NMI control bit above is set, an NMI interrupt will also be
generated when a brownout event is detected. This bit will be cleared by
hardware when a DSP hardware stack over- or underflow is detected. A falling
edge on this bit causes an NMI interrupt in the DSP.
0
Stack over-/underflow (or brownout if the IRQB2NMI bit is set) detected
1
No Stack over-/underflow (or brownout if the IRQB2NMI bit is set)
detected
Table 27. Reset Control Register Description
5-35xx-D1-1.08-061705
39
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
RW RESET
9
IRQB
R
1
8
IRQA
R
1
7:4
SRST
RW 0000
3:0
STKLVL
R
0000
DEFINITION
IRQB Interrupt – An IRQB interrupt will be generated when a brownout event is
detected. This bit will be cleared by hardware to indicate that a brownout event
has been detected.
0
Brownout detected
1
No brownout detected
IRQA Interrupt – An IRQA interrupt will be generated when a headphone short is
detected. This bit will be cleared by hardware to indicate that a headphone short
event has been detected.
0
Headphone short detected
1
No headphone short detected
Software Reset – Writing 1101 will cause a full chip hardware reset, writing any
other value will have no effect.
NOTE: This soft reset generates a “digital wide” reset condition. As a result, it will
return the DCDC converter digital registers to their default states. This will result
in an immediate change in the DCDC conversion loop. Software should use the
normal DCDC voltage change algorithm to restore the DCDC state to its default
state before issuing this soft reset. Failure to properly change the DCDC state can
lead to unstable DCDC converter operation.
Stack Level – The current position of the DSP hardware stack.
Table 27. Reset Control Register Description (Continued)
Note that when the IRQB2NMI bit is set, the ISR that services the NMI will need to
do a few special things. When this bit is set, if both interrupts happen at the same
time, only one NMI will be received. Also, if one interrupt occurs while the other is
being serviced, a new NMI interrupt will not be triggered. The NMI routine will need
to check each possible NMI event and take appropriate action to clear the interrupt
event, and then recheck that all NMI interrupt sources are clear before returning. If
the NMI ISR returns with a NMI interrupt event still pending, that event (and all subsequent NMI interrupts) will be lost. Finally, the NMI ISR needs to be able to deal
with the case where no interrupt event is pending when the ISR checks it, since the
latency of the NMI interrupt is several cycles long, and the stack over-/underflow
could have been cleared by the time that the NMI runs. In fact, this is guaranteed to
happen in the case of the stack underflow interrupt, as the very fact of calling the
NMI will push items onto the stack.
6.3.
Hardware Profiling Support
Two clock counters are implemented to allow accurate code profiling and cycle
counting. The HW_DCLKCNTL & HW_DCLKCNTU registers together form a 48 bit
counter that will increment once per clock cycle. There is no hardware support for
reading and writing these registers atomically. It is possible to read the lower register before it wraps around, and then read the upper register after it has been
wrapped around, and get an erroneous value. Various software techniques can be
used to avoid this problem. The HW_DCLKCNTL & HW_DCLKCNTU registers will
increment every clock cycle, including cycles that the DSP doesn’t see because
they are stolen for DMA accesses or because of an access conflict in the on-chip
RAMs. The HW_CYCSTLCNT register counts these cycles, and if you want to calculate the number of cycles actually seen by the DSP during a period of time, you
must subtract the number of stolen clock cycles by referring to this register.
40
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
6.3.1.
DCLK Count Lower Register
HW_DCLKCNTL
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
HW_DCLKCNTL
2
3
X:$FFEA
Table 28. HW_DCLKCNTL
BITS
23:0
LABEL
RW RESET
HW_DCLKCNTL RW 0
DEFINITION
DCLK Counter Lower – This counter will increment once per DSP clock
cycle.
Table 29. DCLK Count Lower Register Description
6.3.2.
DCLK Count Upper Register
HW_DCLKCNTU
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
HW_DCLKCNTU
2
3
X:$FFEB
Table 30. HW_DCLKCNTU
BITS
23:0
LABEL
RW RESET
HW_DCLKCNTU RW 0
DEFINITION
DCLK Counter Upper – This counter will increment every time that the
HW_DCLKCNTL counter overflows.
Table 31. DCLK Count Upper Register Description
5-35xx-D1-1.08-061705
41
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
6.3.3.
Cycle Steal Count Register
HW_CYCSTLCNT
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
HW_CYCSTLCNT
2
3
X:$FFEC
Table 32. HW_CYCSTLCNT
BITS
23:0
LABEL
RW RESET
HW_CYCSTLCNT RW 0
DEFINITION
Cycle Steal Counter – This counter will increment every time a cycle is
stolen from the DSP for DMA or because of an access conflict in the RAMs
Table 33. Cycle Steal Count Register Description
6.4.
Clock Control Register
C rysta l
O sc
24 .0 M H z
C rysta l
÷2 0
PW DN
P LL E N
φ
P D IV
D D IV _
MSB:
D D IV
SYSTEM PLL
φ
P LL C LK D IV S E L 1
U S B 2 .0 P LL
P L LP W D V C O
P o st
D ivider
P L LC K D IV C T L 2
P LLP W D C P 3
DCLK
CKSRC
P LL_S O U R C E _ S E L
3
R T C & A larm
(D igital C lock)
D A C D IV
C lo ck to D A C
D A C D ivide r
ACKEN
X T LE N
A D C D IV
C lo ck to A D C
A D C D ivide r
PW M
R e al T im e C lock
1
U S B P H Y A nalog P LL C ontro l R egister (H W _U S B P H Y P L L )
U S B P H Y A nalog R X C ontrol R eg ister (H W _U S B P H Y R X )
3
U S B P H Y A nalog P ow er C ontrol R e gister (H W _U S B P H Y P W D )
2
Figure 18. Clock Control Register (HW_CCR)
The Clock Control Register configures the system clock sources, including the Analog clock and PLL. It is also used to shut down system power by turning off the DCDC converter. Note that none of the bits in this register have any effect unless the
CLKRST bit (bit [0]) is set to one.
42
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
The digital clock can be set to nearly any value between 12 kHz and 77 MHz using
the PLL and post divider.
The various clock multiplexors shown in Figure 18 are fully synchronous “glitch-free”
switches. In order to successfully change a clock multiplexor “switch setting”, the
clock inputs to each leg must be operating. In particular to throw the switch controlled by HW_CCR_PLL_SOURCE_SEL, both the system PLL and the USB 2.0
PLL must be operating. In addition, dividers like HW_CCR_DDIV are glitch free
upon change. This implies that it takes several clocks to transition to new divider
settings. This requires certain rules for changing various controls in the HW_CCR
register.
1. HW_CCR_DDIV must not be changed in the same clock change window as
either HW_CCR_CKSRC or HW_CCR_PLL_SOURCE_SEL. The clock change
window is satisfied with four DSP NOPS.
2. If DDIV is going to a larger value, then change it first followed by CKSRC. If
DDIV is going to a smaller value, then change it after CKSRC, again observing
clock change window.
3. Changing HW_CCR_PLL_SOURCE_SEL and DDIV have similar constraints.
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
CKRST
1
2
LTC
1
3
PLLEN
1
4
XTLEN
1
5
PLL_SOURCE_SEL
1
6
ADCDIV
1
7
CKSRC
1
8
DDIV
1
9
PDIV
2
0
PWDN
2
1
ACKEN
2
2
DACDIV
DDIV_MSB
2
3
X:$FA00
LOCK
HW_CCR
Table 34. HW_CCR
BITS
LABEL
23
DDIV_MSB
22:20 DACDIV
RW RESET
DEFINITION
RW 0
Extension to digital clock divider, used in conjunction with DDIV field
below.
RW 000
Analog clock divider for DAC – Changes the clock rate used by the
DAC. This is used to scale down the frequencies used in the DAC
when using audio sample rates less than the maximum. These are
calculated for a 24.0MHz crystal (not 24.576MHz).
000 xtal/4 = 6.0 MHz = 128*Fssc = 128*46.875 kHz
001 xtal/6 = 4.0 MHz = 128*Fssc = 128*31.25 kHz
010 xtal/8 = 3.0 MHz = 128*Fssc = 128*23.4375 kHz
011 xtal/12 = 2.0 MHz = 128*Fssc = 128*15.625 kHz
100 xtal/8 = 3.0 MHz = 128*Fssc = 128*23.4375 kHz
101 xtal/12 = 2.0 MHz = 128*Fssc = 128*15.625 kHz
110 xtal/16 = 1.5 MHz = 128*Fssc = 128*11.71875 kHz
111 xtal/24 = 1.0 MHz = 128*Fssc = 128*7.8125 kHz
Table 35. Clock Control Register Description
5-35xx-D1-1.08-061705
43
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
19
LOCK
18
ACKEN
17
PWDN
16:12 PDIV
11:9
DDIV
8
CKSRC
RW RESET
DEFINITION
R
PLL lock status
0
PLL not locked
1
PLL locked
RW 0
Analog clock enable – The crystal clock runs digital circuitry for the
ADC, DAC, PWM and for the TIMERs when in crystal clock mode.This
bit enables clocks to the analog circuitry. This bit must be set before
using the RTC, Alarm, PWM, Timers, DAC or ADC. In addition to this
mode bit, the ADC or DAC power down bits in the Mixer Power Down
Control Status Register must not be asserted for the ADC or DAC to
operate.
0
Analog clocks disabled
1
Analog clocks enabled
RW 0
System power-down
0
No action
1
Power down system
Note: bit 9 of HW_MIXTBR must be set low to enable this functionality
Note: When the once port is attached, the DCDC converter does not
accept a power down signal from the HW_CCR_PWDN bit.
Note: HW_CCR_PLLEN must be cleared before using the
HW_CCR_PWDN bit functionality.
RW 00000 PLL frequency divider – Assuming a 24.0 MHz crystal, the PLL can
be programmed in 1.2 MHz steps, from a minimum frequency of
39.6 MHz to a maximum frequency of 76.8 MHz. The reset value is
00000, which yields a PLL frequency of 39.6 MHz. When used in
combination with DDIV post-divider, it is possible to reach frequencies
below 39.6 MHz with smaller granularities.
PLL output frequency = (33 + PDIV) * (Crystal frequency/20)
RW 000
Digital clock post-divider – The post divider one cycle to update. It is
recommended to change the clock source to be the crystal (via
CKSRC) before altering DDIV to limit the maximum frequency the
divider will output during the update time. To achieve the minimum
power mode, select the crystal as the source for the digital clocks
(CKSRC=0), turn off the PLL (PLLEN=0), and set this divider to the
maximum divide rate. The digital clock will then be set to
24.0 MHz/2048 = 11.7 kHz.
if DDIV_MSB = 0:
000 divide by 1
011 divide by 8
110 divide by 64
001 divide by 2
100 divide by 16
111
divide by 128
010 divide by 4
101 divide by 32
if DDIV_MSB = 1:
000 divide by 256 001 divide by 512
010 divide by 1024
011 divide by 2048 1XX undefined
RW 0
Clock source – This bit may only be set from 0 to 1 if the PLL is
enabled and locked. The new clock source may not be available to the
system until two periods of the crystal clock have elapsed.
0
Digital clock generated from crystal
1
Digital clock generated from PLL
Table 35. Clock Control Register Description (Continued)
44
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
7:5
ADCDIV
4
PLL_SOURCE_SEL
3
XTLEN
2
PLLEN
1
LTC
0
CKRST
RW RESET
DEFINITION
RW 000
Analog clock divider for ADC – Changes the clock rate used by the
ADC. This is used to scale down the frequencies used in the ADC
when using audio sample rates less than the maximum.These are
calculated for a 24.0MHz crystal (not 24.576MHz).
000 xtal/4 = 6.0 MHz = 128*Fssc = 128*46.875 kHz
001 xtal/6 = 4.0 MHz = 128*Fssc = 128*31.25 kHz
010 xtal/8 = 3.0 MHz = 128*Fssc = 128*23.4375 kHz
011 xtal/12 = 2.0 MHz = 128*Fssc = 128*15.625 kHz
100 xtal/8 = 3.0 MHz = 128*Fssc = 128*23.4375 kHz
101 xtal/12 = 2.0 MHz = 128*Fssc = 128*15.625 kHz
110 xtal/16 = 1.5 MHz = 128*Fssc = 128*11.71875 kHz
111 xtal/24 = 1.0 MHz = 128*Fssc = 128*7.8125 kHz
RW 0
PLL_SOURCE_SEL – This bit selects either the low speed system PLL
or the USB 2 PHY high speed PLL as the source for DCLK, see Figure
33. “USB 2.0 PHY PLL Block Diagram” on page 89.
0
select system PLL
1
select USB 2.0 PHY high speed PLL
RW 0
Crystal clock enable – The crystal clock runs digital circuitry for the
ADC, DAC, PWM and for the TIMERs when in crystal clock mode. This
clock should be disabled when all of these blocks are not in use. The
crystal clock must enabled for these blocks to function correctly.
0
Crystal clock disabled
1
Crystal clock enabled
RW 0
PLL enable
0
PLL disabled
1
PLL enabled
RW 0
Lock timer reset – Resets the PLL lock timer. Must be written as 0,
then 1 to start the PLL lock timer.
RW 0
Clock reset – Must be written with 1 for all writes to this register. If set
to 0, the entire clock logic block stays in its reset state.
Table 35. Clock Control Register Description (Continued)
5-35xx-D1-1.08-061705
45
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
6.5.
Misc./Spare Register
The Misc./Spare Register is used for system updates. It is used to detect the state of
the analog PSWITCH pin and to configure the I2S pins onto alternate pins. It also
indicates when the ONCE port is plugged in and activated. The organization of the
Misc./Spare Register is shown below.
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
I2S_SELECT
2
1
SPARE_BITS
2
2
PSWITCH
2
3
X:$FA16
ONCE_MODE
HW_SPARER
Table 36. HW_SPARER
BITS
LABEL
RW RESET
23:10 RSRVD
10
ONCE_MODE
R
R
0
0
9
PSWITCH
R
8
7:2
RSRVD
SPARE_BITS
R
0
RW 0
1
0
RSRVD
I2S_SELECT
R
0
RW 0
DEFINITION
Reserved – Must be written with 0.
When the once port debugger is attached to the chip several special
operating modes are forced, e.g. power down is suppressed, etc.
This bit is set to one to indicate when the once port debugger is
attached and enabled. It is set to zero for normal operation.
PSWITCH status – This bit indicates the state of the PSWITCH pin.
Normally, this pin will be a 1 when the power switch is pressed.
0
PSWITCH pin is 0
1
PSWITCH pin is 1
Reserved – Must be written with 0.
Spare bits – These bits can be written or read normally but have no
function in this revision of the chip.
Reserved – Must be written with 0.
I2S select – This bit is used to switch the I2S functions onto
alternate pins. One setting allows a subset of the I2S functionality to
be supported in a 100-pin package, the other setting allows the full
I2S functionality but is only supported in a 144-pin package. (see
22.1. “I2S External Pins” on page 258.)
0
Subset of I2S functionality for the 100-pin package
1
Full I2S functionality for the 144-pin package
Table 37. Misc./Spare Register Description
46
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
6.6.
PIN Control Register
Special controls related to the optional low voltage pin mode. Also see 31. “PIN
DESCRIPTION” on page 377.
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
PIN1_8V
2
2
ESD_TEST
2
3
X:$FA30
KEEPER_EN
HW_PIN_CTRL
Table 38. HW_PIN_CTRL
BITS
LABEL
RW
R
W
RESET
23:13
12
RSRVD
ESD_TEST
0
0
11:9
8
RSRVD
R
000
KEEPER_EN RW 0
7:0
PIN1_8V
DEFINITION
Reserved – Must be written with 0.
ESD Test Mode – Set to one to deactivate the ESD clamp devices.
There is an internal delay of 400 to 600 nSeconds after this bit is set to
one until the clamp is deactivated. WARNING: this bit is WRITE ONLY
on the STMP35xx TA1-TA4. It will always read as a zero.
Reserved – Must be written with 0.
Keeper Enable – Each of the 1.8V capable pads have an optional small
keeper that can be enabled when this bit is set to one. WARNING: do
NOT set KEEPER_EN to one until AFTER PAD1_8V bits have been set
to one.
1.8Volt Pin Control – These bits select between 3.3V and 1.8V PAD I/O
functions for the 8 pad groups, see Table 40. “Pin Control to PIn
Mapping” on page 47. Set these pins to one BEFORE setting
KEEPER_EN to one.
RW $00
Table 39. Pin Control Register Description
100
TQFP
PIN #
144
FPBGA
PIN #
GPIO
PIN
#
1
2
3
16
17
18
20
21
23
24
25
26
M2
L2
K4
K7
J7
K8
L9
L10
M11
K10
M12
L11
14
13
12
36
37
38
40
41
53
45
56
54
PIN NAME
SPI_MOSI
SPI_MISO
SPI_SCK
CF/RAM A4, SM CE3n
CF/RAM A5, SM CE2n
CF/RAM A6, SM CE0n
CF/RAM A8, SM CLE
CF/RAM A9, SM ALE
CF OEn, SM REn
CF CE0n, SM CE1n
CF WAITn, SM READY
CF WEn, SM WEn
1.8 V ENABLE BIT
HW_PIN_CTRL_PIN1_8V[7]
HW_PIN_CTRL_PIN1_8V[7]
HW_PIN_CTRL_PIN1_8V[7]
HW_PIN_CTRL_PIN1_8V[0]
HW_PIN_CTRL_PIN1_8V[1]
HW_PIN_CTRL_PIN1_8V[2]
HW_PIN_CTRL_PIN1_8V[4]
HW_PIN_CTRL_PIN1_8V[4]
HW_PIN_CTRL_PIN1_8V[4]
HW_PIN_CTRL_PIN1_8V[3]
HW_PIN_CTRL_PIN1_8V[4]
HW_PIN_CTRL_PIN1_8V[4]
Table 40. Pin Control to PIn Mapping
5-35xx-D1-1.08-061705
47
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
100
TQFP
PIN #
144
FPBGA
PIN #
GPIO
PIN
#
27
30
31
32
33
34
35
36
37
-
K12
J10
J12
H9
J11
H12
H11
H10
G11
G10
G12
G9
F9
F11
F12
F10
E9
55
24
79
25
78
26
77
27
76
28
75
29
74
30
73
31
72
PIN NAME
SM WPn
CF/RAM/SM D0
CF/RAM/SM D15
CF/RAM/SM D1
CF/RAM/SM D14
CF/RAM/SM D2
CF/RAM/SM D13
CF/RAM/SM D3
CF/RAM/SM D12
CF/RAM/SM D4
CF/RAM/SM D11
CF/RAM/SM D5
CF/RAM/SM D10
CF/RAM/SM D6
CF/RAM/SM D9
CF/RAM/SM D7
CF/RAM/SM D8
1.8 V ENABLE BIT
HW_PIN_CTRL_PIN1_8V[4]
HW_PIN_CTRL_PIN1_8V[5]
HW_PIN_CTRL_PIN1_8V[6]
HW_PIN_CTRL_PIN1_8V[5]
HW_PIN_CTRL_PIN1_8V[6]
HW_PIN_CTRL_PIN1_8V[5]
HW_PIN_CTRL_PIN1_8V[6]
HW_PIN_CTRL_PIN1_8V[5]
HW_PIN_CTRL_PIN1_8V[6]
HW_PIN_CTRL_PIN1_8V[5]
HW_PIN_CTRL_PIN1_8V[6]
HW_PIN_CTRL_PIN1_8V[5]
HW_PIN_CTRL_PIN1_8V[6]
HW_PIN_CTRL_PIN1_8V[5]
HW_PIN_CTRL_PIN1_8V[6]
HW_PIN_CTRL_PIN1_8V[5]
HW_PIN_CTRL_PIN1_8V[6]
Table 40. Pin Control to PIn Mapping (Continued)
48
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
7.
INTERRUPT SUBSYSTEM
7.1.
Interrupt Priority Register
The DSP core has seven main interrupt lines, IVL[6:0]. This register is used to
enable each line and set its priority. Some peripherals connect directly to this interrupt bus, but most interrupt sources go through the Interrupt collector, which multiplexes many interrupt sources onto 4 interrupts of these 7 interrupt lines.
If an interrupt with a higher priority level occurs while the DSP core is servicing
another interrupt, the higher priority interrupt will preempt the lower priority interrupt.
If the new interrupt is of the same or lower priority level, then it will not preempt the
interrupt that is currently being serviced.
HW_IPR
BITS LABEL RW RESET
23:22 L6P
21:20 L5P
19:18 L4P
17:16 L3P
15:14 L2P
13:12 L1P
11:10 L0P
RW 0
RW 0
RW 0
RW 0
RW 0
RW 0
RW 0
9:6
5
RSRVD R
0
IRQBT RW 0
4:3
IRQBP
RW 0
2
IRQAT
RW 0
1:0
IRQAP
RW 0
X:$FFFF
DEFINITION
Interrupt line 6 priority level
00 Disabled
01 Priority Level 0 (lowest priority level)
Interrupt line 5 priority level – SPI
00 Disabled
01 Priority Level 0 (lowest priority level)
Interrupt line 4 priority level – I2C
00 Disabled
01 Priority Level 0 (lowest priority level)
Interrupt line 3 priority level
00 Disabled
01 Priority Level 0 (lowest priority level)
Interrupt line 2 priority level
00 Disabled
01 Priority Level 0 (lowest priority level)
Interrupt line 1 priority level
00 Disabled
01 Priority Level 0 (lowest priority level)
Interrupt line 0 priority level – I2S
00 Disabled
01 Priority Level 0 (lowest priority level)
Reserved – Must be written with 0.
IRQB Type
0 Level
1 Negative Edge
IRQB Priority Level
00 Disable
01 Enable
IRQA Type
0 Level
1 Negative Edge
IRQA Priority Level
00 Disable
01 Enable
10 Priority Level 1
11 Priority Level 2 (highest priority level)
10 Priority Level 1
11 Priority Level 2 (highest priority level)
10 Priority Level 1
11 Priority Level 2 (highest priority level)
10 Priority Level 1
11 Priority Level 2 (highest priority level)
10 Priority Level 1
11 Priority Level 2 (highest priority level)
10 Priority Level 1
11 Priority Level 2 (highest priority level)
10 Priority Level 1
11 Priority Level 2 (highest priority level)
10 Enable
11 Enable
10 Enable
11 Enable
Table 41. Interrupt Priority Register Description
5-35xx-D1-1.08-061705
49
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
7.2.
Interrupt Collector
The DSP core provides seven interrupt lines, IVL[6:0], for individual interrupt
requests. The peripheral interrupt count exceeds the seven interrupt request lines,
so the Interrupt Collector (ICOLL) muxes all sources to the seven lines. Three of the
seven interrupt lines are reserved for certain peripherals, so the ICOLL steers 36
interrupt sources to four of the interrupt request lines: IVL[6,3,2,1]. Within an individual interrupt request line, the ICOLL offers an 8-level priority for each of it is interrupt
sources, although preemption of a lower priority interrupt by a higher priority is not
supported. If preemption is required, the interrupt source that needs to preempt
must be routed to a higher priority interrupt request line.
0
7
INT
Sources
0
HW_ICLPRIOR0R
(Priority Register)
HW_ICLSTEER0R
(Steering Register)
8
0
.
.
.
23
HW_ICLENABLE0R
(Enable Register)
15
IVL1
IVL4
IVL5
23
HW_ICLENABLE1R
(Enable Register)
IVL0
12
HW_ICLSTEER1R
(Steering Register)
16
24
.
.
.
30
IVL3
IVL6
11
HW_ICLPRIOR1R
(Priority Register)
IVL2
HW_ICLPRIOR2R
(Priority Register)
30
HW_ICLPRIOR3R
(Priority Register)
DSP
IRQA
IRQB
NMI
23
24
24
HW_IPR
(Enable and
Priority Control
Register)
HW_ICLSTEER2R
(Steering Register)
30
Figure 19. Interrupt Collector Diagram
Interrupt response time is dependent on the interrupt priority. The minimum interrupt
time is as follows:
• 2 clock cycles for interrupt to appear at DSP.
• 2 clock cycles to respond and get vector to execute interrupt service routine.
50
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
7.2.1.
Interrupt Sources
Table 42 shows all the interrupt sources.
INTERRUPT SOURCE
SRC
DAC Refill
0
DAC Underflow
1
ADC Refill
2
ADC Overflow
3
GP Flash Done
4
EMC-CompactFlash Card IRQ
5
EMC-SmartMedia/NAND Timeout
6
EMC-SmartMedia/NAND Interface
7
Invalid Programming
EMC-CompactFlash No Card
8
EMC-CompactFlash Status Change
9
GPIO0
10
GPIO1
11
GPIO2
12
Timer0
13
Timer1
14
Timer2
15
Timer3
16
GPIO3
17
SDRAM
18
CDI
19
Vdd5V Connected
20
USB Controller
21
USB Wakeup
22
Vdd5V Disconnected
23
ESPI
24
FILCO
25
LRADC1
26
RTC Alarm
27
LRADC2
28
Flash ECC
29
30
CDSync Interrupt
31
CDSync Exception
32
CD-RS Interrupt
33
I2C Rx Ready
I2C Rx Overflow
I2C Tx Empty
I2C Tx Underflow
SPI Complete
I2S Rx Overflow
I2S Tx Underflow
I2S Rx Ready
I2S Tx Empty
IRQA - Headphone Short
IRQB - Battery LRADC
NMI
-
IVL BIT #
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
VECTOR
$003C
$003E
$0042
$0044
$006E
$0070
$0072
$0074
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
6,3,2,1
4
4
4
4
5
0
0
0
0
IRQA
IRQB
NMI
$0076
$0078
$0024
$0020
$0022
$0026
$0028
$002A
$0048
$004A
$004C
$007E
$0050
$0052
$0054
$0056
$0058
$005A
$005C
$005E
$0060
$0062
$0066
$0068
$006A
$0030
$0032
$0034
$0036
$000E
$0016
$0012
$0014
$0010
$0008
$000A
$001E
DESCRIPTION
DAC request to fill DAC FIFO buffer
DAC FIFO buffer underflow
ADC request to empty ADC FIFO buffer
ADC FIFO buffer overflow
Flash transaction complete
CompactFlash card interrupt
SmartMedia/NAND card WAIT timeout
Bad programming of SmartMedia/NAND
interface
CompactFlash card remove/absence
CompactFlash card status change
GPIO module 0 interrupt
GPIO module 1 interrupt
GPIO module 2 interrupt
Timer module 0 interrupt
Timer module 1 interrupt
Timer module 2 interrupt
Timer module 3 interrupt
GPIO module 3 interrupt
SDRAM interrupt
CDI interface interrupt
Vdd5V Connected(Vdd5V > VddIO+0.5V)
USB Controller Interrupt
USB Wakeup (Resume from Suspend)
Vdd5V pin disconnected from 5V
ESPI (enhanced SPI)
Filter Coprocessor
Low Resolution ADC
Real Time Clock Alarm
Low Resolution ADC #2
Hardware ECC accelerator for ECC
reserved
CD synchronizer/formatter interrupt
CD synchronizer/formatter exception
Reed-Solomon error corrector interrupt
I2C receiver data ready
I2C receiver data overflow
I2C transmitter data empty
I2C transmitter data underflow
SPI transfer complete
I2S receiver data overflow
I2S transmitter data underflow
I2S receiver data ready
I2S transmitter data empty
Headphone Short
Battery Brown Out
Non-Maskable Interrupt
Table 42. Interrupt Sources
5-35xx-D1-1.08-061705
51
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
7.2.2.
Interrupt Vectors
Table 43 below shows the interrupt vectors used for each possible interrupt source.
ADDRESS
P:$0000
P:$0002
P:$0004
P:$0006
P:$0008
P:$000A
P:$000C
P:$000E
P:$0010
P:$0012
P:$0014
P:$0016
P:$0018
P:$001A
P:$001C
P:$001E
P:$0020
P:$0022
P:$0024
P:$0026
P:$0028
P:$002A
P:$002C
P:$002E
P:$0030
P:$0032
P:$0034
P:$0036
P:$0038
P:$003A
P:$003C
P:$003E
7.2.3.
INTERRUPT SOURCE
Hardware Reset
Stack Error
Unused (formerly Trace)
SWI
IRQA/Headphone Short Detect
IRQB/Battery Brownout Detect
SPI Complete
I2S Tx Data Empty
I2S Tx Underflow
I2S Rx Data Full
I2S Rx Overflow
NMI
GPIO 1
GPIO 2
GPIO 0
Timer 0
Timer 1
Timer 2
I2C Rx Data Ready
I2C Rx Overflow
ADDRESS
P:$0040
P:$0042
P:$0044
P:$0046
P:$0048
P:$004A
P:$004C
P:$004E
P:$0050
P:$0052
P:$0054
P:$0056
P:$0058
P:$005A
P:$005C
P:$005E
P:$0060
P:$0062
P:$0064
P:$0066
P:$0068
P:$006A
P:$006C
P:$006E
P:$0070
P:$0072
INTERRUPT SOURCE
ADC Full
ADC Overflow
Timer 3
GPIO3
SDRAM
Vdd5V Pin Connected to 5V
USB Controller
USB Wakeup (Resume)
Vdd5V Pin Disconnected
ESPI (enhanced SPI)
FILCO (filter coprocessor)
Low Resolution ADC 1
RTC Alarm
Low Resolution ADC 2
Flash ECC Accelerator
CDSync Interrupt
CDSync Exception
CD-RS Interrupt
GPFlash Done
EMC-CompactFlash Card IRQ
EMC-SmartMedia/NAND
Timeout
I2C Tx Data Empty
P:$0074
EMC-SmartMedia/NAND Invalid
Programming
I2C Tx Underflow
P:$0076
EMC-CompactFlash No Card
Invalid DSP instruction
P:$0078
EMC-CompactFlash Status
Change
P:$007A
DAC Empty
P:$007C
DAC Underflow
P:$007E
CDI Interrupt
Table 43. Interrupt Vector Map
Interrupt Collector Registers
7.2.3.1.
ICOLL Enable 0 Register
The enable registers provide bits to enable/disable interrupts for each source. Each
enable corresponds to a specific interrupt source listed in Table 42. A 1 enables the
relevant interrupt, a 0 disables the relevant interrupt.
52
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
HW_ICLENABLE0R
BITS
LABEL
23:0 SEN23:SEN0
X:$F300
RW RESET
RW 0
DEFINITION
0
Disables this interrupt
1
Enables this interrupt
Table 44. ICOLL Enable 0 Register Description
7.2.3.2.
ICOLL Enable 1 Register
The enable registers provide bits to enable/disable interrupts for each source. Each
enable corresponds to a specific interrupt source listed in Table 42. A 1 enables the
relevant interrupt, a 0 disables the relevant interrupt.
HW_ICLENABLE1R
BITS
LABEL
23:10 RSRVD
9:0
SEN33:SEN24
X:$F301
RW RESET
DEFINITION
R 0
Reserved – Must be written with 0.
RW 0
0
Disables this interrupt
1
Enables this interrupt
Table 45. ICOLL Enable 1 Register Description
7.2.3.3.
ICOLL Status 0 Register
The status registers reflect the interrupt state of each source. Each enable corresponds to a specific interrupt source listed in Table 42. A 1 indicates an active interrupt, a 0 indicates an inactive interrupt. This register is read only.
HW_ICLSTATUS0R
BITS
LABEL
23:0 SST23:SST0
X:$F302
RW RESET
R
DEFINITION
0
Interrupt is not active
1
Interrupt is active
Table 46. ICOLL Status 0 Register Description
7.2.3.4.
ICOLL Status 1 Register
The status registers reflect the interrupt state of each source. Each enable corresponds to a specific interrupt source listed in Table 42. A 1 indicates an active interrupt, a 0 indicates an inactive interrupt. This register is read only.
HW_ICLSTATUS1R
BITS
LABEL
23:10 RSRVD
9:0
SST33:SST24
X:$F303
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
RW 0
0
Interrupt is not active
1
Interrupt is active
Table 47. ICOLL Status 1 Register Description
7.2.3.5.
ICOLL Priority 0 Register
The priority registers set the priority for each source. Each enable corresponds to a
specific interrupt source listed in Table 42. Lowest priority is 111 and highest is 000.
5-35xx-D1-1.08-061705
53
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
HW_ICLPRIOR0R
BITS
23:21
20:18
17:15
14:12
11:9
8:6
5:3
2:0
LABEL
S7P
S6P
S5P
S4P
S3P
S2P
S1P
S0P
RW
RW
RW
RW
RW
RW
RW
RW
RW
RESET
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
X:$F304
DEFINITION
Priority level for interrupt source 0
000 Highest priority level
.
.
.
111 Lowest priority level
Table 48. ICOLL Priority 0 Register Description
7.2.3.6.
ICOLL Priority 1 Register
The priority registers set the priority for each source. Each enable corresponds to a
specific interrupt source listed in Table 42. Lowest priority is 111 and highest is 000.
HW_ICLPRIOR1R
BITS
23:21
20:18
17:15
14:12
11:9
8:6
5:3
2:0
LABEL
S15P
S14P
S13P
S12P
S11P
S10P
S9P
S8P
RW
RW
RW
RW
RW
RW
RW
RW
RW
X:$F305
RESET
DEFINITION
Unknown Priority level for interrupt source 1
000 Highest priority level
Unknown
.
Unknown
.
Unknown
.
Unknown
111 Lowest priority level
Unknown
Unknown
Unknown
Table 49. ICOLL Priority 1 Register Description
7.2.3.7.
ICOLL Priority 2 Register
The priority registers set the priority for each source. Each enable corresponds to a
specific interrupt source listed in Table 42. Lowest priority is 111 and highest is 000.
HW_ICLPRIOR2R
BITS
23:21
20:18
17:15
14:12
11:9
8:6
5:3
2:0
LABEL
S23P
S22P
S21P
S20P
S19P
S18P
S17P
S16P
RW
RW
RW
RW
RW
RW
RW
RW
RW
X:$F306
RESET
DEFINITION
Unknown Priority level for interrupt source 2
000 Highest priority level
Unknown
.
Unknown
.
Unknown
.
Unknown
111 Lowest priority level
Unknown
Unknown
Unknown
Table 50. ICOLL Priority 2 Register Description
7.2.3.8.
ICOLL Priority 3 Register
The priority registers set the priority for each source. Each enable corresponds to a
specific interrupt source listed in Table 42. Lowest priority is 111 and highest is 000.
54
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
HW_ICLPRIOR3R
BITS
23:21
20:18
17:15
14:12
11:9
8:6
5:3
2:0
LABEL
S31P
S30P
S29P
S28P
S27P
S26P
S25P
S24P
RW
RW
RW
RW
RW
RW
RW
RW
RW
X:$F307
RESET
DEFINITION
Unknown Priority level for interrupt source 3
000 Highest priority level
Unknown
.
Unknown
.
Unknown
.
Unknown
111 Lowest priority level
Unknown
Unknown
Unknown
Table 51. ICOLL Priority 3 Register Description
7.2.3.9.
ICOLL Priority 4 Register
The priority registers set the priority for each source. Each enable corresponds to a
specific interrupt source listed in Table 42. Lowest priority is 111 and highest is 000.
HW_ICLPRIOR4R
BITS
23:6
5:3
2:0
LABEL
RSRVD
S33P
S32P
RW
R
RW
RW
X:$F311
RESET
DEFINITION
0
Reserved – Must be written with 0.
Unknown Priority level for interrupt source 4
000 Highest priority level
Unknown
.
.
.
111 Lowest priority level
Table 52. ICOLL Priority 4 Register Description
7.2.3.10.
ICOLL Steering 0 Register
The steering registers are used to steer a given source to a given IVL as follows:
SETTING
00
01
10
11
IVL
1
2
3
6
Each steering value corresponds to a specific interrupt source listed in Table 42.
HW_ICLSTEER0R
BITS
23:22
21:20
19:18
17:16
15:14
13:12
11:10
9:8
7:6
5:4
LABEL
S11S
S10S
S9S
S8S
S7S
S6S
S5S
S4S
S3S
S2S
5-35xx-D1-1.08-061705
RW
RW
RW
RW
RW
RW
RW
RW
RW
RW
RW
RESET
X:$F308
DEFINITION
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Table 53. ICOLL Steering 0 Register Description
55
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
3:2
1:0
56
LABEL
S1S
S0S
RW
RW
RW
RESET
DEFINITION
Unknown
Unknown
Table 53. ICOLL Steering 0 Register Description
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
7.2.3.11.
ICOLL Steering 1 Register
The steering registers are used to steer a given source to a given IVL as follows:
SETTING
00
01
10
11
IVL
1
2
3
6
Each steering value corresponds to a specific interrupt source listed in Table 42.
HW_ICLSTEER1R
BITS
23:22
21:20
19:18
17:16
15:14
13:12
11:10
9:8
7:6
5:4
3:2
1:0
LABEL
S23S
S22S
S21S
S20S
S19S
S18S
S17S
S16S
S15S
S14S
S13S
S12S
RW
X:$F309
RESET
RW
RW
RW
RW
RW
RW
RW
RW
RW
RW
RW
RW
DEFINITION
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Table 54. ICOLL Steering 1 Register Description
7.2.3.12.
ICOLL Steering 2 Register
The steering registers are used to steer a given source to a given IVL as follows:
SETTING
00
01
10
11
IVL
1
2
3
6
Each steering value corresponds to a specific interrupt source listed in Table 42.
HW_ICLSTEER2R
BITS
23:20
19:18
17:16
15:14
13:12
11:10
9:8
7:6
5:4
3:2
1:0
LABEL
RSRVD
S33S
S32S
S31S
S30S
S29S
S28S
S27S
S26S
S25S
S24S
RW
R
RW
RW
RW
RW
RW
RW
RW
RW
RW
RW
RESET
X:$F30A
DEFINITION
Reserved – Must be written with 0.
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Unknown
Table 55. ICOLL Steering 2 Register Description
5-35xx-D1-1.08-061705
57
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
7.2.4.
Interrupt Collector Debug Registers
7.2.4.1.
ICOLL Debug Force 0 Register
The debug force value registers will force an interrupt for a given source. The
enable registers enable the forcing mechanism. Each enable corresponds to a specific interrupt source listed in Table 42.
HW_ICLFORCE0R
BITS
23:0
LABEL
S23FV:S0FV
X:$F30B
RW RESET
DEFINITION
RW 0
Table 56. ICOLL Force Value 0 Register Description
7.2.4.2.
ICOLL Debug Force 1 Register
The debug force value registers will force an interrupt for a given source. The
enable registers enable the forcing mechanism. Each enable corresponds to a specific interrupt source listed in Table 42.
HW_ICLFORCE1R
BITS
LABEL
23:10 RSRVD
9:0
S33FV:S24FV
X:$F30C
RW RESET
R
0
DEFINITION
Reserved – Must be written with 0.
Table 57. ICOLL Force Value 1 Register Description
7.2.4.3.
ICOLL Force Enable 0 Register
To generate a forced interrupt you have to write a 1 into the relevant position in both
the force and force enable registers. Writing a 1 to the force enable register will
block any interrupts from the normal interrupt source for the relevant bit. Each force
bit corresponds to a specific interrupt source listed in Table 42.
HW_ICLFENABLE0R X:$F30D
BITS
LABEL
23:0
S23FE:S0FE
RW RESET
DEFINITION
RW 0
Table 58. ICOLL Force Enable 0 Register Description
7.2.4.4.
ICOLL Force Enable 1 Register
To generate a forced interrupt you have to write a 1 into the relevant position in both
the force and force enable registers. Writing a 1 to the force enable register will
block any interrupts from the normal interrupt source for the relevant bit. Each force
bit corresponds to a specific interrupt source listed in Table 42.
HW_ICLFENABLE1R X:$F30E
BITS
LABEL
23:10 RSRVD
9:0
S33FE:S24FE
RW RESET
R
0
RW 0
DEFINITION
Reserved – Must be written with 0.
Table 59. ICOLL Force Enable 1 Registers Description
7.2.5.
58
ICOLL Observation Registers (HW_ICLOBSV0R/1R)
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
The observation registers make visible the state of the interrupt request vector that
the ICOLL is sending out on IVL[6:0] and the winning (prioritized) interrupt vectors
for destinations A, B, C, and D.
A IVL1
B IVL2
C IVL3
D IVL6
Each observation bit corresponds to a specific interrupt source listed in Table 42.
This register is read only and is primarily for debug purposes.
7.2.5.1.
Interrupt Collector Observe 0 Register
HW_ICLOBSVZ0R
BITS
23:21
20:14
13:7
6:0
LABEL
RSRVD
IVB
IVA
REQ
X:$F30F
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
R
R
R
Table 60. ICOLL Observe 0 Register Description
7.2.5.2.
Interrupt Collector Observe 1 Register
HW_ICLOBSVZ1R
BITS
23:14
13:7
6:0
LABEL
RSRVD
IVD
IVC
X:$F310
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
R
R
Table 61. ICOLL Observe 1 Register Description
5-35xx-D1-1.08-061705
59
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
8.
USB CONTROLLER
The STMP35xx includes a Universal Serial Bus (USB) version 2.0 device controller.
The USB controller is used to download digital music data or program code into
external memory and to upload voice recordings from memory to the PC. Program
updates can also be loaded into the flash memory area using the USB interface.
The reader should refer to the USB Implementer’s Forum website www.usb.org for
detailed specifications and information on the USB protocol, timing and electrical
characteristics.
The USB 2.0 controller comprises both a programmed I/O (PIO) interface and a
DMA interface, see Figure 9. “USB Interface Block Diagram” on page 13. Both of
these interfaces, as implemented in the ARC High Speed USB core, are designed to
meet a VSI Alliance Basic Virtual Component Interface BVCI, see www.vsi.org. The
BVCI used by the USB controller has a target (PIO) and initiator (DMA) data bus of
32-bits. In addition, the initiator address bus is also 32 bits wide.
In integrating this core into the STMP35xx, the 24 bit nature of the DSP is mapped
onto the 32 bit nature of the BVCI by a SigmaTel developed USB interface block. In
addition, the STMP35xx only makes use of a 16 bit subset of the BVCI initiator’s 32
bit address. For the following discussion, the USB interface block is broken into a
target mapping function and an initiator mapping function. A third portion of the USB
interface block maps the ARC core USB controller to the UTMI PHY interface. see
Figure 20. “USB 2.0 Device Controller” on page 61.
8.1.
USB Programmed I/O (PIO) Target Interface
The programmed I/O interface uses three 24 bit registers, HW_USBARCACCESS,
HW_USBARCDATALOW, and HW_USBARCDATAHIGH. A hardware state
machine then translates PIO requests from the DSP into target bus cycles for the
ARC USB 2.0 Device Controller core. see Figure 21. “USB 2.0 PIO Target Interface”
on page 62. To write data to a register in the USB device controller, software first
loads the DATAHIGH and DATALOW registers with the upper and lower 16 bit parts
of the 32 bit data to be written. It then loads the ARC access register,
HW_USBARCACCESS, fields with the address, the Read/Write bit and sets the
Kick bit to one. Setting the Kick bit starts the state machine which transfers the 32 bit
data into a USB controller PIO register. All transfers to and from the USB controller
are exactly 32 bits wide.
The Kick bit will remain set until the state machine finishes the target write cycle.
While the Kick bit is high, software must not modify any field in the
HW_USBARCACCESS, HW_USBARCDATALOW, or HW_USBARCDATAHIGH
registers. Fortunately the maximum number of cycles the Kick bit can be high is 18
DCLKs. As long as the software path length is guaranteed to be at least this long
until the next PIO operation is initiated, software does not have to poll the kick bit.
In order to read a PIO register from the USB controller, software loads the ARC
access register, HW_USBARCACCESS, fields with the address, the Read/Write bit
60
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
and sets the Kick bit to one. The Read/Write bit is set to one for reads and zero for
writes.
DMA-BUS
X-BUS
USB System
Programmable Registers
USB DMA Interface
USB PIO Interface
VSIA BVCI Target Bus
Programmed I/O
Target Inteface
SigmaTel USB
Interface Block
VSIA BVCI Initiator Bus
DMA Engine
!
!
!
Bus Interface
Endpoint Priming state machine
Data Movement
Dual Port RAM Controller
!
!
!
Bus Interface
Control & Status
Interrupts
!
!
Virtual FIFO channels
DMA Contexts
On-Chip Dual Port
Synchronous SRAM
Protocol Engine
ARC USB 2.0
Device Controller
!
!
!
Interval Timers
Error Handling
CRC Handling
!
!
Asynchronous clock domain crossing
Transceiver Interface Logic
Port Controller
USB UTMI Interface
PHY
Regs.
480MHz PLL
USB Xcvr
Integrated
USB 2.0 PHY
External USB 2.0
UTMI PHY
Figure 20. USB 2.0 Device Controller
The target bus and the intitiator busses emanating from the USB Device Controller
do not share any resources in common. Thus no arbitration or other startup delay is
encounter in performing PIO operations to/from the USB Device Controller.
5-35xx-D1-1.08-061705
61
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
HW_USBARCACCESS:KICK
State
Machine
t_data_r[31:0]
31
0
HW_USBARCDATAHIGH:
DATA
15
0
HW_USBARCDATALOW:
DATA
15
31
0
15
HW_USBARCACCESS:
ADD
0
t_cmd,t_cmdack, etc.
HW_USBARCACCESS:RWB
0
t_data_w[31:0]
t_address[8:0]
ARC USB 2.0 Device Controller
Figure 21. USB 2.0 PIO Target Interface
C versions of generalized PIO read and PIO write routines are given below, see Figure 22. “USB 2.0 PIO Target Interface Sample Code” on page 62. These routines
are provided to facilitate understanding. Actual software uses tightly coded assembly routines to manipulate the USB Controller’s programmable registers.
//This function unpacks the 48-bit double word into two 16-bit words and writes
// it back to the ARC register
void _reentrant write_usb_reg(USHORT usRegAdd, DWORD dwData)
{
HW_USBARCDATALOW.B.DATA =
(WORD)(dwData & (WORD)(HW_USBARCDATALOW_DATA_SETMASK));
HW_USBARCDATAHIGH.B.DATA =
(WORD)((dwData >> 16) & (DWORD)(HW_USBARCDATAHIGH_DATA_SETMASK));
usRegAdd &= (USHORT)HW_USBARCACCESS_ADD_SETMASK;
HW_USBARCACCESS.B.ADD = usRegAdd;
HW_USBARCACCESS.B.RWB = 0;
HW_USBARCACCESS.B.KICK = 1;
}
// This function reads the ARC 32-bit register and packs it in a 48-bit double word
// Use this function in combination with write_usb_reg() to read-modify-write
// an ARC register
void _reentrant read_usb_reg(USHORT usRegAdd, DWORD * dwData)
{
usRegAdd &= HW_USBARCACCESS_ADD_SETMASK;
HW_USBARCACCESS.B.ADD = usRegAdd;
HW_USBARCACCESS.B.RWB = 1;
HW_USBARCACCESS.B.KICK = 1;
while(HW_USBARCACCESS.B.KICK); // wait for state machine
*dwData = (DWORD)(HW_USBARCDATALOW.B.DATA & HW_USBARCDATALOW_DATA_SETMASK);
*dwData |= (DWORD)(HW_USBARCDATAHIGH.B.DATA << 16);
}
Figure 22. USB 2.0 PIO Target Interface Sample Code
62
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
8.2.
USB DMA Interface
Most software interactions with the USB device controller occur with data or control
structures that have been DMA transferred to the on-chip RAM.
time order of payload bytes on USB wire
23
0
P
L
H
D
0
O
K
G
C
N
J
F
B
M
I
E
A
i_wdata[31:0]
from USB Controller
N
K
H
E
B
P
M
J
G
D
A
ARC & DSP
Addresses Align
31
O
L
I
F
C
N
K
H
E
B
P O
ML
J I
GF
DC
A -
ARC & DSP Addresses
do NOT Align
A B C D E F GH I J K L MN OP
Contents
of on-chip RAM
Figure 23. USB 2.0 32 bit to 24 bit Packing Example
The USB DMA interface has to handle data transfers to and from a 32 bit interface
to the ARC USB Device Controller. It has to map these data transfers to and from a
24-bit on-chip RAM DMA bus. Figure 23, above shows one example of a stream of
bytes coming across the USB cable, being DMAed across the 32-bit BVCI initiator
bus and from there being packed into the 24 bit wide on-chip memory. The DMA
interface data flow that performs this packing is shown below, see Figure 24. “USB
2.0 DMA Data Flow” on page 64.
From these figures one sees that all DMA address pointers are referenced to the 32
bit data bus world of the ARC USB DMA engine. When the data lands in on-chip
RAM or is taken from on-chip RAM, then software has to access it with full knowledge of how the USB DMA transferred the bytes. The figure above shows that
depending on where the starting address for the data is assigned in the ARC
address space determines how the data aligns with 24 bit DSP RAM. In particular,
one divides the ARC byte address by three to obtain a 24 bit word address in the
DSP RAM. The remainder of this division determines the alignment in the DSP
RAM. A remainder of zero implies a perfectly aligned transfer point. A remainder of
one yields the alignment shown in the lower right hand portion of Figure 23. Software must be constantly aware of this alignment and must deal with it as it arises.
5-35xx-D1-1.08-061705
63
OFF ICIAL
PRODUCT
DOCUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
For example, the eight end points implemented in the USB Device controller require
16 Queue Head structures to be accessed by the DMA from on-chip RAM. Each
queue head is 64 bytes long or 21 remainder 1 DSP words in length. The remainder
of one causes the alignment each successive queue head to shift by one byte.
ram_dma_db[23:0]
usb2_ram_dma_db[23:0]
DMA Bus Interface to/from on-chip RAM
0
[RPTR]
[RPTR+1]
[RPTR+2]
[RPTR+3]
[WPTR]
[WPTR+1]
[WPTR+2]
32-bit USB to RAM [WPTR+3]
24-bit RAM to USB
USB to 24-bit RAM
RAM to 32-bit USB
1 byte wide multi-tap FIFO
15 7
0
Interface to BVCI Initiator in USB Controller
i_rdata{31:0]
i_rdata{31:0]
USB Controller Initiator & DMA
Figure 24. USB 2.0 DMA Data Flow
The general strategy used by software is have pack and unpack routines to access
data structures shared with the USB controller, i.e. queue heads, transfer descriptors. In addition, software uses pack and unpack routines to access the packet
headers, etc.
8.3.
ERRATA: USB DMA WRITE BUFFER ERROR AFFECTING ADJACENT
SRAM
There is an error in the DMA write command processing for certain combinations of
USB Device Core 32-bit address pointers and 24-bit SRAM buffer targets. When the
Device core generates a 32-bit address with byte enables set to modify less than 3
bytes on the first transfer to a buffer, then a NULL read modify write (RMW) cycle is
generated for the 24-bit SRAM word immediately ahead of the targeted USB DMA
Buffer. This NULL RMW cycle reads the word, does no modification and then writes
it back out. This normally have no effect on the SRAM location accessed by the read
modify write cycle, UNLESS, another DMA device such as the hardware ECC tries
to perform a DMA write to the same word. If the ECC write occurs between the time
the USB interface has done the read portion of the RMW cycle and the time it does
the write back portion of the RMW then the data written by the ECC will be over writ-
64
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DOCUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
ten by the USB with the location’s original contents. This will leave the SRAM in
state it was in BEFORE the hardware ECC wrote its result value.
To circumvent this problem in software, do not place SRAM write buffers for other
DMA devices such that they end immediately before any USB buffer or descriptor or
queue head. This will keep the other DMA from intermittently having its last word
written being destroyed by this USB interface error.
8.4.
USB UTMI Interface
The SigmaTel developed UTMI interface logic allows multiplexing between the integrated USB 2.0 High Speed PHY and an off-chip UTMI compliant USB 2.0 PHY, see
Figure 28. “USB 2.0 PHY at the System on Chip Level” on page 79. The
HW_USBCSR_UTMI_EXT bit selects the external PHY when set to one.
8.5.
USB Device Controller Core
The USB Device Controller is an instantiation of the ARC VUSBHS-DEV High
Speed USB Device Controller Core. This proprietary core, the intellectual property it
represents and the copyrighted documentation for the core are the property of ARC
International. For detailed information about the device controller core, the reader is
referred to the appropriate ARC documentation. The core synthesized as a device
controller only and implements eight endpoints, including endpoint zero.
8.6.
USB Controller Flowcharts
5-35xx-D1-1.08-061705
65
OFF ICIAL
PRODUCT
DOCUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
turn on the USB
Plugged In detector.
Check_USB_Plugged_In()
Set HW_USBCSR_PLUGGEDIN_EN=1
Check for 5V input
NO
HW_USBCSR_
VDD5VSENSE == 1
Wait for light pull-ups to settle
NO
1 mSecond
Expired ?
YES
NO
HW_USBCSR_
PLUGGED_IN == 1
YES
Turn off plugged in detector
Set HW_USBCSR_PLUGGEDIN_EN=0
Set HW_USBCSR_USBEN=1
Turn on USB Device Controller Clocks
Set HW_USBCSR_ARCCONNECT=1
RETURN
NOT_CONNECTED
RETURN
CONNECTED
ARC Core has a Virtual
Plugged In Indicator
Plugged In Detector Remains ON
Figure 25. USB 2.0 Check_USB_Plugged_In Flowchart
66
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DOCUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
Called with or without 5V
InitVDD5VAtStartup
turn on the USB
Plugged In detector.
Set HW_USBCSR_PLUGGEDIN_EN=1
Preferred Automatic
Power down on
5V disconnect
Turn off current limit
HW_USB_PWR_CHARGE_DISABLE_ILIMIT=1
Enable (optional)
5V Disconnect IRQ $0056
This mode is typically not used.
Check for 5V input from USB or Wall
YES
HW_USBCSR_
VDD5VSENSE == 1
Enable 5V Connect
IRQ $0050
NO
HW_USBCSR_VDD5VDISCXIE=1
HW_USBCSR_VDD5VCXIE=1
HW_USB_PWR_CHARGE_PWDN_ON_IOBRNOUT =1
HW_PERSIST_CFG_AUTO_RESTART =1
HW_PERSIST_CFG_UPDATE = 0
STOP
HW_PERSIST_CFG_UPDATE = 1
call USB
Connect Check
USB Cable
Connected?
Enable Auto Restart
Update Persistant Bits: note UPDATE bit normally
left high, taken low on same cycle as data setup to
persistent regs, raised high to update in registers.
NO
Start Regular App
Inform App that wall current is
available for charging.
YES
Start USB
Connect Sequence
STOP
STOP
Figure 26. USB 2.0 InitializeVDD5VSense Flowchart
5-35xx-D1-1.08-061705
67
OFF ICIAL
PRODUCT
DOCUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
$0050
$0056
VDD5V_CONN_ISR
Enable 5V Disconnect
IRQ $0056
HW_USBCSR_VDD5VDISCXIE=1
HW_USBCSR_VDD5VCXIE=0
Disable Conn IRQ
Opitonal Automatic
Power down on 5V
disconnect
VDD5V_DISCONN_ISR
If HW_USB_PWR_CHARGE_PWDN_ON_IOBRNOUT
is set then this interrupt routine will never be entered,
since the DCDC converter will power down before it can
run. If not set, DSP can execute for a few hundred
nanoseconds before the core fails due to low voltage.
HW_CCR_PWDN = 1
Enable Auto Restart
HW_USB_PWR_CHARGE_PWDN_ON_IOBRNOUT =1
HW_PERSIST_CFG_AUTO_RESTART =1
HW_PERSIST_CFG_UPDATE = 0
HW_PERSIST_CFG_UPDATE = 1
STOP
Force Power Down
Update Persistant Bits: note UPDATE bit normally
left high, taken low on same cycle as data setup to
persistent regs, raised high to update in registers.
call USB
Connect Check
USB Cable
Connected?
NO
YES
Stop Current App &
Start USB Connect Sequence
STOP
Inform App that wall current is
available for charging.
STOP
Figure 27. USB 2.0 VDD5V Conn, Disconn ISR Flowchart
8.7.
USB Interface Registers
The following subsections describe the programmable registers of the SigmaTel
USB Interface Block.
8.7.1.
USB Control Status Register
This register handles the overall configuration and control of the USB interface. It
provides specific interrupt status and interrupt enables for interrupt events arising
from the USB Controller as well as interrupts arising from monitoring the 5V sense
comparator and the cable plugged in monitoring circuit in the integrated PHY.
68
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DOCUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
USBEN
1
0
WAKEUPIE
1
1
WAKEUPIRQ
1
2
VDD5VCXIE
1
3
VDD5VCXIRQ
1
4
VDD5VDISCXIRQ
1
5
CLKOFF
1
6
VDD5VDISCXIE
1
7
SUSP
1
8
SUSPF
1
9
UTMI_EXT
2
0
ARCCONNECT
HOSTDISCONNECT
2
1
PLUGGED_IN
2
2
X:$F200
PLUGGEDIN_EN
2
3
VDD5VSENSE
HW_USBCSR
Table 62. HW_USBCSR
BITS
LABEL
23
VDD5VSENSE
22
HOSTDISCONNECT
21:14 RSRVD
13
PLUGGED_IN
12
PLUGGEDIN_EN
11
ARC_CONNECT
10
UTMI_EXT
9
SUSPF
8
SUSP
RW RESET
DEFINITION
R
0
USB 5Volt Vdd5V Sense State – This read only bit returns a one
when the VDD5V pin is more than one Vt above the VDDIO
voltage. This situation indicates that a 5V power source either from
a wall transformer or from the USB VBUS has been connected to
the STMP35xx. It returns a zero otherwise, refer to the power
charger, see Section 30.5.11. on page 367
R
0
Host Disconnect – This disconnect signal reports the state of the
Hub Disconnect detector in the integrated PHY.
R
0
Reserved – Must be written with 0.
R
0
Cable Plugged In – This read only bit returns a zero when the
integrated PHY’s plugged-in detector reports that both DP and DN
are high. This can only occur when the cable is unplugged and the
200KΩ integrated pull-ups are enabled onto DP and DN. This
detector is powered down at reset. Power control is located in bit
HW_USBCSR_PLUGGEDIN_EN and must be turned on for at
least 1mSecond prior to relying on the state of this bit.
RW 0
USB plugged-in detector Enable – Set this bit to one to power up
the plugged in detector and to switch in the two 200KΩ pull-up
resistors.
RW 0
ARC Connect – The ARC core has an input that is designed to be
hooked up to a VBUS Sensor to tell it when to go to powered state.
In this implementation, the DSP detects this situation using the
VDD5VSENSE bit above. The DSP therefore virtualizes this
indication by controlling the ARCCONNECT bit. When the DSP
sets this bit to one, the ARC core moves its internal state machine
from the “detached” state to the “powered” state.
RW 0
UTMI EXTERNAL Interface Enable – When set to one, this bit
causes the external USB 2.0 PHY to be selected instead of the
integrated PHY. Set this bit to zero for normal operation.
RW 0
Force USB Suspend mode – This bit controls a mux which
determines the drive source for the utmi_suspendm signal. When
set to one, the mux select the SUSP bit below as the source for
driving utmi_suspendm. Set this bit to zero for normal operation.
RW 0
USB Suspend – This bit provides direct programmable control of the
utmi_suspendm signal when enabled by the SUSPF bit above.
Table 63. USB Control Status Register Description
5-35xx-D1-1.08-061705
69
OFF ICIAL
PRODUCT
DOCUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
7
CLKOFF
6
VDD5VDISCXIE
5
VDD5VDISCXIRQ
4
VDD5VCXIE
3
VDD5VCXIRQ
2
WAKEUPIE
1
WAKEUPIRQ
0
USBEN
RW RESET
DEFINITION
RW 1
USB clock off – When set to one, clocks to the USB module will be
set to zero. Set CLKOFF to zero for normal operation of the USB
interface.
RW 0
USB 5V Vdd5V pin Disconnect Interrupt Enable– When this bit
is set to one, then VDD5VDISCXIRQ below will cause an interrupt
at vector $0056 when it is set to one.
RW 0
USB 5V Vdd5V Disconnect Interrupt – This bit is set to one
whenever a falling edge is detected on the VDD5SENSE
comparator. This bit can be reset by writing a one directly to it.
RW 0
USB 5V Vdd5V Connect Interrupt Enable– When this bit is set to
one, then VDD5VCXIRQ below will cause an interrupt at vector
$0050 when it is set to one.
RW 0
USB 5V Vdd5V Connect Interrupt – This bit is set to one
whenever a rising edge is detected on the VDD5SENSE
comparator. This bit can be reset by writing a one directly to it.
RW 0
Wakeup Interrupt Enable – When this bit is set to one, then
WAKEUPIRQ below will cause an interrupt at vector $0054 when it
is set to one.
RW 0
Wakeup Interrupt – This bit is set to one whenever the
synchronized versions of utmi_linestate indicate a K-state on the
line, i.e linestate[0] == DP ==0 & linestate[1] == DN ==1. This bit
can be reset by writing a one directly to it. Activity on. NOTE: this
interrupt bit monitors the raw state of the USB signals and will
continue to be set as transitions occur on the USB. Software
should reset WAKEUPIE and then clear WAKEUPIRQ as soon as
this interrupt is detected.
RW 0
USB Enable Bit – The USBEN bit enables the USB port. This bit
must be set before any other USB registers are written to.
Table 63. USB Control Status Register Description (Continued)
8.7.2.
USB DMA Offset Register
All ARC core DMA addresses have an offset added to them. This addition takes
place after the 32-bit ARC byte address is divided by three to obtain a DSP 24-bit
word address. USB DMA buffers can be placed in any one of the three processor
memory spaces (P,X,Y).
2
2
2
1
2
0
1
9
1
8
1
7
1
6
MEM
2
3
X:$F201
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ADD
HW_USBDMAOFF
Table 64. HW_USBDMAOFF
70
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DOCUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
23:18 RSRVD
17:16 MEM
15:0
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
RW 0
MEM – Selects the memory space (P,X or Y) to which or from
which all DMA transfers occur.
RW 0
USB DMA ADDRESS OFFSET – All DMA addresses generated in
the USB controller have this 16 bit offset added to them, after the
division by 3.
ADD
Table 65. USB DMA Offset Address Register Description
8.7.3.
USB ARC ACCESS
This register is used in programmed I/O (PIO) access to ARC core internal registers, see 8.1. “USB Programmed I/O (PIO) Target Interface” on page 60.
HW_USBARCACCESS X:$F202
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ADD
2
2
RWB
KICK
2
3
Table 66. HW_USBARCACCESS
BITS
23
KICK
LABEL
22:17 RSRVD
16
RWB
15:9
8:0
RSRVD
ADD
RW RESET
DEFINITION
RW 0
KICK – This bit is set to a one to initiate either a read or write
programmed I/O cycle to the ARC Core BVCI bus target. This bit
will be automatically reset when the transfer is complete. Software
must not read or modify the data or address registers until the KICK
bit has been reset by hardware.
R
0
Reserved – Must be written with 0.
RW 0
Read Write Bit – This bit determines whether a programmed I/O
cycle to the ARC core will be a read or a write transfer. When this
bit is a one, an ARC core PIO register will be read into the data
registers. When this bit is a zero, the data registers will be written to
an ARC core PIO register.
RWB = 1 =READ
RWB = 0 =WRITE
R
0
Reserved – Must be written with 0.
RW 0
ARC Register Address – The ARC core target interface receives
ADD[8:0] as an address for its internal programmed I/O registers.
Table 67. USB ARC Access Register Description
8.7.4.
USB ARC Data Low Register
All programmed I/O data transfers to and from the ARC core’s internal registers are
32 bits wide. This register holds the lower order 16 data bits for such transfers, see
8.1. “USB Programmed I/O (PIO) Target Interface” on page 60.
5-35xx-D1-1.08-061705
71
OFF ICIAL
PRODUCT
DOCUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
HW_USBARCDATALOW X:$F203
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
DATA
2
3
Table 68. HW_USBARCDATALOW
BITS
LABEL
23:16 RSRVD
15:0 DATA
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
RW 0
DATA – The lower 16 bits of a data read/write operation to an ARC
internal register.
Table 69. USB ARC Data Low Register Description
8.7.5.
USB ARC Data High Register
All programmed I/O data transfers to and from the ARC core’s internal registers are
32 bits wide. This register holds the higher order 16 data bits for such transfers, see
8.1. “USB Programmed I/O (PIO) Target Interface” on page 60.
HW_USBARCDATAHIGH X:$F204
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
DATA
2
3
Table 70. HW_USBARCDATAHIGH
BITS
LABEL
23:16 RSRVD
15:0 DATA
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
RW 0
DATA – The higher 16 bits of a data read/write operation to an ARC
internal register.
Table 71. USB ARC Data High Register Description
72
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DOCUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
8.7.6.
USB UTMI Test Control Status Register
HW_USBUTCSR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
UTMI_XFER_SIZE
UTMI_TEST_ENABLE
2
3
X:$F205
Table 72. HW_USBUTCSR
BITS
LABEL
23:9 RSRVD
8
UTMI_TEST_ENABLE
7:0
UTMI_XFER_SIZE
RW
R
RW
RW
RESET
DEFINITION
0
Reserved – Must be written with 0.
0
UTMI TEST ENABLE –
0
UTMI TRANSFER SIZE – Load this field with the number of DMA
transfers to make during UTMI loop back test mode.
Table 73. USB UTMI Test Control Status Register Description
8.7.7.
USB UTMI1 Register
HW_USBUTMI1
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
UNUSED
2
3
X:$F206
Table 74. HW_USBUTMI1
LABEL
20:16 RSRVD
15:0 UNUSED
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
RW 0
UNUSED Bit – This field is currently unused in the design.
Table 75. USB UTMI1 Register Description
5-35xx-D1-1.08-061705
73
OFF ICIAL
PRODUCT
DOCUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
8.7.8.
USB UTMI2 Register
HW_USBUTMI2
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
UNUSED
2
3
X:$F207
Table 76. HW_USBUTMI2
BITS
LABEL
20:16 RSRVD
15:0 UNUSED
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
RW 0
UNUSED Bit – This field is currently unused in the design.
Table 77. USB UTMI2 Register Description
74
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DOCUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
8.7.9.
USB UTMI Sense Register
This register provides hardware debug and design verification access to various signals on the internal UTMI interface.
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
UTMI_DATABUS16_8
1
1
UTMI_DATAOE
1
2
UTMI_RXERROR
1
3
UTMI_RXVALID
1
4
UTMI_RXACTIVE
1
5
UTMI_RXVALIDH
1
6
UTMI_TXVALID
1
7
UTMI_TXREADY
1
8
UTMI_TXVALIDH
1
9
UTMI_OPMODE
2
0
UTMI_LINESTATE
2
1
UTMI_TERM_SELECT
2
2
UTMI_RESET
2
3
UTMI_XCVR_SELECT
HW_USBUTMISENSE X:$F208
Table 78. HW_USBUTMISENSE
BITS
20:16
15
14
13
12:11
10:9
8
7
6
5
4
3
2
1
0
LABEL
RW RESET
DEFINITION
R
Reserved – Must be written with 0.
R
UTMI Reset Signal – T
R
UTMI High Speed Transceiver Select – T
R
UTMI High Speed Terminator Select – T
R
UTMI Linestate[1:0] – T
R
UTMI Opmode[1:0] – T
R
UTMI Transmit Valid High – T
R
UTMI Transmit Valid Low – T
R
UTMI Transmit Ready – T
R
UTMI Receive Valid High – T
R
UTMI Receive Valid low – T
R
UTMI Receive Active –
R
UTMI Receive Error –
R
UTMI Data Output Enable –
R
UTMI Data Bus 16 or 8 – The USBEN bit enables the USB port.
This bit must be set before any other USB registers are written to.
RSRVD
UTMI_RESET
UTMI_XCVR_SELECT
UTMI_TERM_SELECT
UTMI_LINESTATE
UTMI_OPMODE
UTMI_TXVALIDH
UTMI_TXVALID
UTMI_TXREADY
UTMI_RXVALIDH
UTMI_RXVALID
UTMI_RXACTIVE
UTMI_RXERROR
UTMI_DATAOE
UTMI_DATABUS16_8
Table 79. USB UTMI Sense Register Description
8.7.10.
USB Read Test Register
HW_USBREADTEST X:$F209
2
3
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
TESTVALUE
Table 80. HW_USBREADTEST
5-35xx-D1-1.08-061705
75
OFF ICIAL
PRODUCT
DOCUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
15:0
LABEL
TESTVALUE
RW
RESET
R
$ABCDEF
DEFINITION
Read Only Test Value – This field always reads back the same
constant value $ABCDEF
Table 81. USB UTMI1 Register Description
8.7.11.
USB State Machine Sense Register
This register provides hardware debug and design verification access to various
state machines in the ARC core interface.
HW_USBSTATEMACHINES X:$F20A
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
CMD_STATE
2
1
IRESP_STATE
2
2
DMA_STATE
2
3
Table 82. HW_USBSTATEMACHINES
BITS
20:18
17:12
11
10:8
7:5
4:0
LABEL
RSRVD
DMA_STATE
RSRVD
IRESP_STATE
RSRVD
CMD_STATE
RW RESET
DEFINITION
R
Reserved – Must be written with 0.
R
DMA State Machine State
R
Reserved – Must be written with 0.
R
Interrupt Response State Machine State
R
Reserved – Must be written with 0.
R
PIO Command State Machine State
Table 83. USB State Machine Sense Register Description
8.7.12.
USB ARC Unused Signals Sense Register
This register provides hardware debug and design verification access to various
unused signals of the ARC core.
HW_USBARCUNUSED X:$F20B
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
PHILIPS_SIGS
2
2
SERIAL_SIGS
2
3
Table 84. HW_USBARCUNUSED
76
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DOCUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
20:11 RSRVD
10:5 SERIAL_SIGS
4:0
PHILIPS_SIGS
RW RESET
DEFINITION
R
Reserved – Must be written with 0.
R
Unconnected Serial PHY Interface Signals
R
Unconnected Philips PHY Interface Signals
Table 85. USB ARC UNUSED Signal Sense Register Description
8.7.13.
USB Laser Fuse Sense Register
This register provides hardware debug and design verification access to various
unused signals of the ARC core.
HW_USBLASERFUSE X:$F20C
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ARC_UNCONNECTED
2
2
LASER_FUSE
2
3
Table 86. HW_USBSTATEMACHINES
BITS
LABEL
20:11 RSRVD
21:20 LASER_FUSE
RW RESET
DEFINITION
R
Reserved – Must be written with 0.
R
USB Laser Fuse – When LASER_FUSE[1] is a one, then the USB
controller core is limited to Full Speed operation only. When
LASER_FUSE[0] is a one, then the USB controller clock is gated
off and the function is not available.
19:15 RSRVD
R
Reserved – Must be written with 0.
14:0 ARC_UNCONNECTED R
Various Unconnected Signals from ARC USB 2.0
Table 87. USB State Machine Sense Register Description
8.8.
Known Chip Defects with USB Device
8.8.1.
Clear Quest Entry STMP00004471
All revisions of the chip are known to have a defect in which the USB device controller incorrectly handles short packets that are multiples of 128 bytes.
Simulations revealed a bug in the USB gasket for OUT transfers. The bug can corrupt the end of the data buffer. It occurs when the Host sends out a short packet that
is multiple of 128 Bytes. During this scenario the USB core does writes that are
voided (byte enable == all zero), which were an undefined capability and thus the
5-35xx-D1-1.08-061705
77
OFF ICIAL
PRODUCT
DOCUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
gasket did not apply a proper response. This does not affect Mass Storage applications because data transfers are 512 Bytes and SCSI commands are <16 Bytes.
78
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DOCUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
9.
INTEGRATED USB 2.0 PHY (HS,FS)
DCLK
DSP X-Bus
External USB 2.0
UTMI PHY
DMA (P,X,Y)
USB 2.0
CONTROLLER
On-Chip
RAM
Bus Interface
The STMP35xx contains an integrated USB 2.0 PHY macro-cell capable of connecting to PC host systems at the USB Full Speed (FS) rate of 12Mbits/Second or at the
USB 2.0 High Speed (HS) rate of 480Mbits/Second. The integrated PHY provides a
standard UTMI interface. This allows the STMP35xx to alternatively connect to an
external UTMI compliant USB 2.0 PHY chip.
DSP
UTMI Digital
System PLL
Crystal
Oscillator
High
Speed
PLL
(480MHz)
Digital
RX
Analog
RX/TX
DP
Digital
TX
DN
Integrated USB 2.0 PHY
(UTMI macro cell)
Figure 28. USB 2.0 PHY at the System on Chip Level
The following subsections describe the external interfaces, internal interfaces, major
blocks, and programable registers that comprise the integrated USB 2.0 PHY.
9.1.
External Signals
DP,DN – These pins connect directly to a USB device connector.
Precision Calibration Resistor. This pin connects a 620Ω +/- 1% resistor to ground.
The key on chip resistors, i.e. the 45Ω high speed termination resistor and the
1500Ω pull up resistor contain digitally controlled trimming and calibration circuits to
match their impedance to the external precision resistor for USB 2.0 specification
compliance.
5-35xx-D1-1.08-061705
79
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
9.2.
UTMI Internal Signals
utmi_clk – Used to clock receive and transmit data to and from the USB controller.
The internal UTMI interface is 16 bits wide resulting in a 30MHz UTMI_CLK sourced
by the PHY. This clock can either be derived from a divide by sixteen off of the USB
2.0 PHY PLL or can be divided by two from DCLK. see Figure 33. “USB 2.0 PHY
PLL Block Diagram” on page 89.
utmi_reset – From chip wide master reset distribution.
utmi_xcvr_select – Selects between the High Speed and Full Speed transceivers
(0 = High Speed, 1 = Full Speed).
utmi_term_select – Selects between High Speed and Full Speed terminations.
(0 = High Speed Termination, 1 = Full Speed Termination).
utmi_suspend – Not implemented in the internal UTMI interface. The availability of
PHY control registers that are directly accessible by DSP instructions obviates the
need for this signal.
utmi_line_state[1:0] – These signals directly reflect the current state of the single
ended receivers (DP and DN) in the PHY.
DN
LS[1]
DP
LS[0]
UTMI_LINE_STATE[1:0]
0
0
00 = SE0, single ended zero
0
1
1
1
0
1
01 = ‘j’ State
10 = ‘K’ State
11 = SE1, single ended one
Table 88. USB PHY line_state[1:0]l
utmi_op_mode[1:0] – These input signals tell the PHY whether to encode and
decode transmissions with bit stuffing and NRZI or to simply pass the data straight
through.
OP_MODE
UTMI_OP_MODE[1:0]
00
Normal operation
01
10
11
Non-driving
Disable bit stuffing and NRZI encoding
Reserved
Table 89. USB PHY op_ode[1:0]l
utmi_tx_data[15:0] – These PHY inputs convey the transmit data from the USB controller to the PHY.
utmi_tx_valid – This signal indicates when tx_data[7:0] has a valid data output byte.
utmi_tx_validh – This signal indicates when tx_data[15:0] has a valid data output
byte.
utmi_tx_ready – The USB controller uses this signal in conjunction with the current
state of tx_valid to determine when a data cycle has been accepted into the PHY
serializer. When utmi_tx_rdy and utmi_tx_valid are both asserted at the rising edge
of utmi_clk then a new data word should be presented by the USB controller to the
PHY.
80
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
utmi_rx_data[16:0] – These signals convey the received data from the PHY to the
USB controller.
utmi_rx_valid – When high, this signal tells the USB controller that utmi_rx_data[7:0]
contains a valid byte of data.
utmi_rx_validh – When high, this signal tells
utmi_rx_data[15:8] contains a valid byte of data.
the
USB
controller that
utmi_rx_active – Indicates that the USB PHY receive machine has detected SYNC
and is active.
utmi_rx_error – Zero indicates no error detected. One indicates that a receive error
has been detected.
usb_plugged_in_detect – Digital output from the receiver’s analog plugged-in detector circuit. This signal is captured within the USB controller and made available from
there to the DSP.
9.3.
UTMI and Digital Circuits
The UTMI provides a 16 bit interface to the USB Controller. This interface is clocked
at 30MHz. There are four parts to the UTMI/Digital circuits block. The UTMI block,
the Digital Transmitter, the Digital Receiver and the Programmable Registers block.
9.3.1.
UTMI Block
This block handles the line_state bits, reset buffering, suspend distribution, transceiver speed selection, and transceiver termination selection. The PLL supplies a
60MHz signal to all of the digital logic. The UTMI block does a final divide by two to
develop the 30MHz clock used in the interface.
9.3.2.
Digital Transmitter Block
The digital transmitter block receives the 16 bit transmit data from the USB controller, and handles the tx_valid, tx_validh and tx_ready handshake. In addition, it contains the transmit serializer which converts the 16 bit parallel words at 30MHz to a
single bit stream at 480Mbit for High Speed or 12Mbit for Full Speed. It does this
while implementing the bit stuffing algorithm and the NRZI encoder that are used to
remove the DC component from the serial bit stream. The output of this encoder is
sent to the Full Speed (FS) or High Speed (HS) drivers in the analog transceiver
section’s transmitter block.
9.3.3.
Digital Receiver Block
The digital receiver block receives the raw serial bit stream either from the HS differential transceiver or from the FS differential transceiver. The HS input goes to a very
fast DLL which uses one of eight identically spaced phases of the 480MHz clock to
pick a sample point. As the phase of the USB host transmitter shifts relative to the
local PLL, the receiver section’s HS DLL tracks these changes to give a reliable
sample of the incoming 480Mbit/Second bit stream. Since this sample point shifts
relative to the PLL phase used by the digital logic, a rate matching elastic buffer is
provided to cross this clock domain boundary. Once the bit stream is in the local
clock domain, an NRZI decoder and Bit Unstuffer restores the original payload data
bit stream and passes it to a de-serializer and holding register. The Receive state
machine handles the rx_valid, rx_validh and handshake with the USB controller.
The handshake is not interlocked in that there is no rx_ready signal coming from the
controller. The controller must take each 16 bit value as presented by the PHY. The
5-35xx-D1-1.08-061705
81
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
Receive state machine provides an rx_active signal to the controller that indicates
when it is inside a valid packet (SYNC detected, etc.).
9.3.4.
Programmable Registers Block
The PHY contains four 24 bit programmable registers that are read and written via
the DSP’s X bus, see 9.7. “Programmable Registers” on page 93.
9.4.
Analog Transceiver
The analog transceiver section comprises an analog receiver and an analog transmitter, see Figure 29 below.
Transmitter
RPU Enable
HS Current Source Enable
VDDIO
(3.3V)
HS Drive Enable
HS Data Drive
FS Driver Output Enable
1500Ω
FS DataDrive
Assert SE0
FS Edge Mode Select
test, calibration,
& discrete powerdown contols
DP
HS Differential RCVR
DN
Squelch
USB
Cable
FS Differential RCVR
HS_Disconnect_Detect
USB_Plugged_In_Detect
Single Ended Detector SE_DP
Single Ended Detector SE_DM
Receiver
Figure 29. USB 2.0 PHY Analog Transceiver Block Diagram
82
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
9.4.1.
Analog Receiver
The analog receiver comprises five differential receivers and two single ended
receivers, described below.
9.4.1.1.
HS Differential Receiver
The high speed differential receiver is both a differential analog receiver and threshold comparator. Its output is a one if the differential signal is greater than a 0V
threshold. Its output is zero otherwise. Its purpose is to discriminate the +/- 400mV
differential voltage resulting from the high speed drivers current flow into the dual
45Ω terminations found on each leg of the differential pair. The envelope or squelch
detector, below, ensures that the differential signal has sufficient magnitude to be
valid. The HS differential receiver tolerates up to 500mV of common mode offset.
9.4.1.2.
Squelch Detector
The squelch detector is a differential analog receiver and threshold comparator. Its
output is a one if the differential magnitude is less than a nominal 100mV threshold.
Its output is zero otherwise. Its purpose is to invalidate the HS differential receiver
when the incoming signal is simply to low to receive reliably.
9.4.1.3.
FS Differential Receiver
The full speed differential receiver is both a differential analog receiver and threshold comparator. The crossover voltage falls between 1.3V and 2.0V. Its output is a
one when the DP line is above the crossover point and the DN line is below the
crossover point.
9.4.1.4.
HS Disconnect detector
This Host side function is not used in STMP35xx applications but is included to
make a complete UTMI macro-cell. It is a differential analog receiver and threshold
comparator. Its output is a one if the differential magnitude is greater than a nominal
575mV threshold. Its output is zero otherwise.
9.4.1.5.
USB Plugged-In Detector
The USB Plugged-In detector looks for both DP and DN to be high. There is a pair of
large on-chip pull-up resistors (200KΩ) that hold both DP and DN high when the
USB cable not attached. The USB Plugged-In detector signals a zero in this case.
The host/hub interface that is upstream from the STMP35xx contains a 15KΩ pulldown resistor that easily over-rides the 200KΩ pull-up. When plugged in, at least
one signal in the pair will be low which will force the Plugged-In detector’s output
high.
9.4.1.6.
Single Ended DP Receiver
The single ended DP receiver output is high whenever the DP input is above its
nominal 1.8V threshold.
9.4.1.7.
Single Ended DN Receiver
The single ended DN receiver output is high whenever the DN input is above its
nominal 1.8V threshold.
5-35xx-D1-1.08-061705
83
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
9.4.2.
Analog Transmitter
The analog transmitter comprises two differential drivers, one for high speed signaling and one for full speed signalling. It also contains the switchable1500Ω pull up
resistor.
9.4.2.1.
Switchable High Speed 45Ω Termination Resistors
High speed current mode differential signalling requires good 90Ω differential termination at each end of the USB cable. This results from switching in 45Ω terminating
resistors from each signal line to ground at each end of the cable. Since each signal
is parallel terminated with 45Ω at each end, each driver sees a 22.5Ω load. This is
much too low of a load impedance for full speed signalling levels hence the need for
switchable high speed terminating resistors. Switchable trimming resistors are provided to tune the actual termination resistance of each device, see Figure 30. “USB
2.0
PHY
Transmitter
Block
Diagram”
on
page 85.
The
HW_USBPHYTX_TXCAL45DP bit field, for example, allows one of 16 trimming
resistor values to be placed in parallel with the 45Ω terminator on the DP signal. The
calibration operation is described below, see 9.4.2.5. “Resistor Calibration Mode” on
page 86.
9.4.2.2.
Full Speed Differential Driver
The full speed differential drivers are essentially “open drain” low impedance pull
down devices which are switched in a differential mode for full speed signalling, i.e.
either one or the other device is turned on to signal the “J” state or the “K” state.
These drivers are both turned on, simultaneously, for high speed signalling. This
has the effect of switching in both 45Ω terminating resistors. The tx_fs_hiz signal
originates in the digital transmitter section. The hs_term signal which also controls
these drivers comes from the UTMI.
9.4.2.3.
High Speed Differential Driver
The high speed differential driver receives a 17.78mA current from the constant current source and essentially steers it down either the DP signal or the DN signal or
alternatively to ground. This current will produce approximately a 400mV drop
across the 22.5Ω termination seen by the driver when it is steered onto one of the
signal lines. The approximately 17.78mA current source is referenced back to the
integrated voltage band gap circuit. The Iref, IBias and V to I circuits are shared
with the integrated battery charger.
9.4.2.4.
Switchable 1500Ω DP Pull up resistor
The STMP35xx contains a switchable 1500Ω pull-up resistor on the DP signal. This
resistor is switched on to tell the Host/Hub controller that a full speed capable device
is on the USB cable, powered on, and ready. This resistor is switched off at power
on reset so the host doesn’t recognize a USB device until DSP software enables the
announcement of a full speed device. This pull-up also includes 16 switchable parallel trimming resistors, see HW_USBPHYTX_TXCAL1500.
84
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
To Battery
Charger
REF_RES
Vbg
Ibias
Iref
force
disconn
V to I
HW_USBPHYTPWD:
TXPWDVBG,
TXPWDV2I,
TXPWDIBIAS
hs_term
17.78mA
620Ω
+/- 1%
1500ΩPU
&
Calibration
HW_USBPHYTX:
TXCAL1500,
TXENCAL1500
HW_USBPHYPWD:
TXDISCON1500
Current Steering
current
switch
current
switch
data_p,hs_xcvr
data_n,hs_xcvr
DP
USB
Cable
DN
45Ω
45Ω
HW_USBPHYTX:
TXCAL45DN
HW_USBPHYTX:
TXENCAL45DP,DN
FS
DRVR
Calibration
Comparator
HW_USBPHYTX:
TXCAL45DP
FS
DRVR
data_p,data_n,
fs_hiz, hs_term
HW_USBPHYTX:
TXCOMPOUT
HW_USBPHYTX:
TXENCAL45DP,
TXENCAL45DN,
TXENCAL1500,
TXCALIBRATE
HW_USBPHYTX:TXHSXCVR
HW_USBPHYTX:TXHSTERM
~utmi_hsterm
hs_term
~utmi_hsxcvr
hs_xcvr
HW_USBPHYRX:XCVRTERMSELECTEN
HW_USBPHYTX:TXFSHIZ
~phy_tx_fs_hiz
fs_hiz
HW_USBPHYRX:TXFSHIZEN
Figure 30. USB 2.0 PHY Transmitter Block Diagram
5-35xx-D1-1.08-061705
85
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
The following table summarizes the response of the PHY analog transmitter to various states of UTMI input and key transmit/receive state machine states.
UTMI
OPMODE
00=Normal
01=NoDrive
10=NoNRZI
NoBitStuff
11= Invalid
UTMI
TERM
UTMI
XCVR
T/R
FUNCTION
45 Ω
HIZ
0
0
X
HS
0
1
1
1
1
1
0
0
0
1
1
0
0
1
1
0
0
1
0
0
1
0
1
0
T
R
R
T
X
T
R
X
X
X
X
FS
FS
CHIRP
CHIRP
DISCONNECT
HS
HS
FS
CHIRP
DISCONNECT
HS
0
1
1
1
1
1
1
1
1
1
0
0
0
0
0
1
1
1
1
1
1
1
1
1
1
0
0
0
1
1
0
1
1
0
0
1
0
0
1
0
1
T
R
R
T
X
T
R
X
X
X
FS
FS
CHIRP
CHIRP
DISCONNECT
HS
HS
FS
CHIRP
DISCONNECT
0
1
1
1
1
1
1
1
1
1
0
0
0
0
1
1
1
1
1
1
1500Ω
HIZ
SUSPEND
POR
1
Table 90. USB PHY Terminator States
9.4.2.5.
Resistor Calibration Mode
The analog transmitter section includes a calibration comparator that can monitor
the DP or DN voltage as desired. Setting HW_USBPHYTX_ENCAL45DP selects the
DP signal. The comparator output is visible in HW_USBPHYTX_TXCOMPOUT. To
calibrate the 45Ω DP terminator, first set the field HW_USBPHYTX_TXCAL45DP to
all ones ($F). The flowchart of Figure 31, below, shows how to search for the proper
trimming resistor to calibrate the DP terminator.
Essentially one first puts the chip into termination resistor calibration mode for the
DP terminator. One starts with the largest value of trimming select, i.e. all ones. One
has to make several precise minimum delay calculations to allow the mixed signal
components to stabilize. The comparator output is sampled and then checked. If the
comparator has not tripped, then one reduces the value of the trimming select field
and tries again. This repeats until the trip point is reached.
While this flowchart shows how to calibrate the DP terminator, calibration of the DN
terminator is accomplished in a similar manner, substituting *DN bit fields for *DP
backfields.
Calibrating the 1500Ω pull-up is done in a similar manner. The flow chart of Figure
32 shows how to accomplish this task.
86
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
START
Set TXPWDVBG=0.
Set TXPWDV2I=0.
Set TXPWDIBIAS=0.
Set TXPWDFS=0.
Set TXPWDCOMP=0.
Set TXCAL45DP=1111.
Set TXENCAL45DP=1.
FLOWCHART FOR
CALIBRATING THE
45-OHM HIGH SPEED
TERMINATION
RESISTOR.
Allow 1.5us (forty-five 30MHz cycles) for
analog circuitry to stabilize.
Set TXCALIBRATE=1.
Allow 1us (thirty 30MHz cycles) for
analog circuitry to stabilize.
Set TXCALIBRATE=0.
Decrement TXCAL45DP.
NO
NO
TXCOMP0UT == 1?
TXCAL45DP == 0000?
YES
YES
Set TXENCAL45DP=0.
Set TXPWD bits for normal operation.
STOP
Figure 31. 45Ω Calibration Flow Chart
5-35xx-D1-1.08-061705
87
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
START
Set TXPWDVBG=0.
Set TXPWDV2I=1.
Set Battery Charge
and Temperature
Sensor PWDs = 1.
Set TXPWDIBIAS=0.
Set TXPWDFS=0.
Set TXPWDCOMP=0.
FLOWCHART FOR
CALIBRATING THE
1500-OHM RESISTOR.
Set TXCAL1500=1111.
Set TXENCAL1500=1.
Allow 500ns (fifteen 30MHz cycles) for
analog circuitry to stabilize.
Set TXCALIBRATE=1.
Allow 100ns (three 30MHz cycles)
for analog circuitry to stabilize.
Decrement TXCAL1500.
Set TXCALIBRATE=0.
NO
TXCOMP0UT == 1?
YES
NO
TXCAL1500 == 0000?
YES
Set TXENCAL1500=0.
Set TXPWD bits for
normal operation.
STOP
Figure 32. 1500Ω pull up resistor Calibration Flow
88
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
9.5.
USB 2.0 PHY 480MHz PLL
The STMP35xx includes a 480MHz PLL to clock the high speed transceiver. This
PLL can also be used for generating the system wide DCLK. In addition, it can be
left powered down and the system wide DCLK can be used to drive 60MHz into the
Full Speed PHY to operate the USB.
HW_CCR:CKSRC
DCLK
HW_CCR:
PLL_SOURCE_SEL
System PLL Block
HW_CCR:PDIV,DDIV
dclk_u
24MHz
HW_USBPHYRX:
PLLCKDIVCTL
HW_USBPHYPLL:
PLLCLKDIVSEL
8
2
Phase/Freq Det.
HW_USBPHYPLL:
FSCKSOURCESEL
30MHz
UTMI CLK
60MHz
FS XCVR
CLK
VCO
480MHz Phase 2
480MHz Phase 2Z
HW_USBPHYPLL:
PLLCPNSEL
USB 2.0 PHY PLL
Phase Followers
Charge
Pump
480MHz Phase 3
480MHz Phase 3Z
480MHz Phase 4
480MHz Phase 4Z
480MHz Phase 1
480MHz Phase 1Z
Figure 33. USB 2.0 PHY PLL Block Diagram
9.5.1.
480MHz VCO and Phase Followers
The heart of the PLL is the VCO which can operate from 120MHz to 720MHz. The
VCO frequency is determined by the output of the charge pump, in standard fashion. The VCO produces a 480MHz clock and its exact out of phase component. In
the design, these are identified as vco_clk2 and vco_clk2z.
In addition, three
5-35xx-D1-1.08-061705
89
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
phase followers are included to produce a precise eight phase clock at 480MHz.
These eight phases are used in the high speed digital receiver to operate the DLL
that tracks the incoming 480Mbit/second USB receive digital stream. vco_clk2 is
also used as a single phase 480MHz digital clock for various clock dividers and
other circuits. The VCO and various of its phase followers can be selectively powered down to reduce the overall energy requirements of the STMP35xx.
9.5.2.
PFD and Charge Pump
The phase/frequency detector (PFD) and charge pump (CP) are used to lock the
VCO to the reference oscillator. For the STMP35xx the reference is the integrated
crystal oscillator. The most common reference crystal frequencies are 24MHz and
20MHz. Selective power down and control of the PFD, the CP and various loop filter
parameters can be controlled in HW_USBPLL. The charge pump gain (current)
should be adjusted for different feedback settings, see
HW_USBPLL_PLLCPNSEL.
9.5.3.
Feedback Divider
The feedback divider is connected to dived the VCO frequency by one of several
constant dividers. The loop goal is to lock to 480MHz from several different reference frequencies. If a 24MHz crystal is used with a divide by 20, then the loop will
stabilize to 480MHz. Similarly if a 20MHz crystal is used with a divide by 24, then
the loop will stabilize again at 480MHz. With a 20MHz crystal and a divide by 6 the
loop will stabilize to 120MHz. Of course the charge pump gain must be set appropriately. The feedback divider is programmed in HW_USBPLL_PLLCLKDIVSEL.
9.5.4.
DCLK_U Generation
The high speed PLL supplies a post divider output of the VCO that can be used to
drive the DCLK chip wide clock net. The divider value is set in
HW_USBRX_PLLCKDIVCTL which defaults to a divide by eight. With the VCO
locked to 480MHz, this produces a 60MHz clock which is driven out on the signal
dclk_u. Dclk_u is wired to the system PLL where a 2:1 mux selects either the system PLL or the high speed PLL to drive DCLK. The HW_CCR_PLLSOURCESEL bit
selects the desired PLL. Of course, the HW_CCR_CKSRC bit selects either the
crystal oscillator or the selected PLL to drive the DCLK clock net.
9.5.5.
60MHz Full Speed USB PHY Clock Generation
A fixed divide by eight post divider is included to generate 60MHz from a 480MHz
VCO. This clock is used within the full speed USB transceiver. It is also further
divided by two to produce the 30MHz clock used in the UTMI interface for the integrated USB 2.0 PHY. For test purposes, the chip wide DCLK can be selected
instead of this high speed PLL version of the 60MHz signal, see
HW_USBPHYPLL_FSCKSOURCESEL.
90
5-35xx-D1-1.08-061705
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
9.6.
Integrated USB 2.0 PHY Initialization Flow Charts
Turn USB Device Controller Clocks
Set HW_USBCSR_USBEN=1
Only Start the PHY when 5V
sensed
NO
HW_USBCSR_
VDD5VSENSE == 1
5V power is present so turn on the
USB Plugged In detector.
Set HW_SPARER_USB_PLUGGEDIN_EN=1
NO
Only Start the PHY when
cable is plugged in
HW_USBCSR_
PLUGGED_IN == 1
YES
Flow chart for starting Internal PHY
START_INT_PHY
if not already done
Set HW_CCR_CLKRST = 1
system PLL must be on for mux change
Set HW_CCR_PLL_EN = 1
force all PLL regs to default state
Set HW_USBPHYPWD = $80000000
Set HW_USBPHYPWD = $00000000
Set HW_USBPHYRX.PLLLOCKED = 0
turn off powerdown bits that were
just forced to a default of one.
Write zero to PLLLOCKED bit to start
PLL lock sequence. At the end of this
sequence, PHY will be on.
NO
Wait for PLL Lock Sequence
Completion
HW_USBPHYRX_
PLLLOCKED == 1
Set HW_CCR_CKSRC = 1
Set HW_CCR_PLL_SOURCE_SEL = 1
Set HW_CCR_PLL_EN = 0
Switch to PHY PLL source for DCLK
PHY STARTED, now start the ARC USB Device Controller
YES
STOP
Figure 34. USB 2.0 PHY PLL START_PHY Flow Chart
5-35xx-D1-1.08-061705
91
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
ARC IRQ
Only suspend the phy when
the ARC knows it is OK within
the USB protocol.
To Other
ARC ISR
NO
Arc Suspend
Interrupt?
YES
Flow chart for Suspending
the Internal PHY
START_INT_PHY
Switch back to XTAL DCLK
Set HW_CCR_PLLEN = 0
start PHY PLL lock sequence to
switch PHY internals back to DCLK
Set HW_USBPHYRX_PLLLOCKED = 0
Set HW_USBPHYPWD = $7FFFDF
Power down PHY
Bit 23 -> do not establish defautlts,
this preserves calibration info
Bit 5 TXDISCON1500 =0
-> keep ARC in charge of pull up
Call SHUTDOWN_ARC
Set HW_USBCSR_WAKEUPIE =1
Wake Up for Resume
Call OTHER_LOW_POWER_SETTINGS
STOP
PHY powered down, ARC
powered down, other low power
clocking, etc in effect
Figure 35. USB 2.0 PHY PLL Suspend Flowchart
92
5-35xx-D1-1.08-061705
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
9.7.
Programmable Registers
The USB 2.0 Integrated PHY contains four registers that are directly programmable
from the DSP, as follows:
USB PHY Analog Power Control
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
TXDISCON1500
RXPWDDISCONDET
1
4
PLLPWDCP
RXPWDENV
1
5
PLLPWDVCO
1
6
TXPWDFS
1
7
TXPWDIBIAS
1
8
TXPWDV2I
1
9
TXPWDVBG
2
0
RXPWD1PT1
PWDIBIAS
2
1
RXPWDDIFF
2
2
X:$F210
RXPWDRX
2
3
REGRESET
HW_USBPHYPWD
TXPWDCOMP
9.7.1.
0
4
0
3
0
2
0
1
0
0
Table 91. HW_USBPHYPWD
BITS
LABEL
RW RESET
23
REGRESET
RW 0
22
PWDIBIAS
RW 1
21
20
RSRVD
RXPWDRX
R
0
RW 1
19
RXPWDDIFF
RW 1
18
RXPWD1PT1
RW 1
17
RXPWDENV
RW 1
16
RXPWDDISCONDET
RW 1
15
14
RSRVD
TXPWDCOMP
R
0
RW 1
13
TXPWDVBG
RW 1
DEFINITION
Writing a one to this bit resets the Analog Control register to its default
(reset) state. This bit always reads a Zero.
Powers down the bias generation circuit for the USB 2.0 PHY. This
should be set only when all of the other power down bits (PWD) for the
USB PHY are set to zero. This bit must be set to zero (low) during
battery charge operation.
RESERVED -- must be written with zero.
Set to one to power down the entire USB PHY receiver block except
for the full speed differential receiver. Set to zero for normal operation.
Set to one to power down the USB PHY high speed differential
receiver. Set to zero for normal operation.
Set to one to power down the USB 1.1 style full speed differential
receiver. Set to zero for normal operation.
Set to one to power down the USB 2.0 PHY receiver envelope
detector (squelch signal). Set to zero for normal operation.
Set to one to power down the USB 2.0 PHY receiver disconnect
detector. Set to zero for normal operation.
RESERVED -- must be written with zero.
Set to one to power down the USB 2.0 PHY transmit calibration
comparator. Set to zero during calibration and set to one after
calibration is complete.
Set to one to power down the USB 2.0 PHY transmit voltage band gap
buffer amp as well as the V-to-I converter and the current mirror. Note
these circuits are shared with the battery charge circuit, setting this bit
to one will not power down these circuits unless the corresponding bit
in the battery charger is also set for power down, see 30.5.11. “Power
Charger Register” on page 367. Set to zero for normal operation and
for calibration.
Table 92. USB PHY Analog Power Control
5-35xx-D1-1.08-061705
93
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
RW RESET
12
TXPWDV2I
RW 1
11
TXPWDIBIAS
RW 1
10
TXPWDFS
RW 1
9:8
7
RSRVD
PLLPWDCP
R
00
RW 1
6
PLLPWDVCO
RW 1
5
TXDISCON1500
RW 0
4:0
RSRVD
R
00000
DEFINITION
Set to one to power down the USB 2.0 PHY transmit V-to-I converter
and current mirror.Note these circuits are shared with the battery
charge circuit, see 30.5.11. “Power Charger Register” on page 367.
Setting this bit to one will not power down these circuits unless the
corresponding bit in the battery charger is also set for power down. Set
to zero for normal operation and for calibration.
Set to one to power down the USB 2.0 PHY current bias block for the
transmitter. This bit should only be set when the USB is in suspend
mode. This effectively powers down the entire USB transmit path.
Note these circuits are shared with the battery charge circuit, see
30.5.11. “Power Charger Register” on page 367. Setting this bit to one
will not power down these circuits unless the corresponding bit in the
battery charger is also set for power down. Set to zero for normal
operation and for calibration.
Set to one to power down the USB 2.0 PHY full speed drivers. This
turns off the current starvation sources and puts the drivers into a Hi-Z
output.
RESERVED -- must be written with zero.
Set to one to power down the USB 2.0 PHY charge pump in the PLL.
Should be used in conjunction with PLLPWDVCO to completely power
down the PLL. Set to zero for normal operation.
Set to one to power down the USB 2.0 PHY VCO in the PLL. This bit
only powers down the VCO section, use in conjunction with
PLLPWDCP. Set to zero for normal operation.
Set to zero to connect the integrated 1500Ω pull up resistor tied to the
DPLUS USB pad. This defaults to “disconnected” to allow the DSP
and USB controller to perform the appropriate initialization before
connecting to the USB DPLUS & DMINUS.
RESERVED -- must be written with zero.
Table 92. USB PHY Analog Power Control
94
5-35xx-D1-1.08-061705
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
USB PHY Analog Transmit Control
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
TXCAL1500
1
6
TXENCAL1500
1
7
TXHSXCVR
1
8
TXCALIBRATE
TXENCAL45DP
1
9
TXCAL45DN
TXFSHIZ
2
0
TXENCAL45DN
2
1
TXSKEW
2
2
X:$F211
TXCAL_45DP
2
3
TXCOMPOUT
HW_USBPHYTX
TXHSTERM
9.7.2.
Table 93. HW_USBPHYTX
BITS
LABEL
RW RESET
23
TXCOMPOUT
RW 0
22
TXFSHIZ
RW 1
21
TXENCAL45DP
RW 0
20
RSVRD
19:16 TXCAL45DP
R
0
RW 0110
15
RW 0
TXSKEW
DEFINITION
The calibration comparator output is latched to this bit. This bit should
be erased for every new calibration. This bit can set to zero or set to
one by a normal write from the DSP. In addition, it is loaded with the
state of the calibration comparator’s output whenever
HW_USBPHT_TXCALIBRATE is set to one. It continuously copies
the comparator to this bit as long TXCALIBRATE is set to one, i.e.
when TXCALIBRATE is one, writing to this bit has no effect.
Set to zero to tri-state the full speed driver. The actual tristate control is
muxed between the normal source coming from the PHY transmitter
logic and this bit. The mux is controlled by
HW_USBPHYRX_TXFSHIZEN. Set to one for normal operation.
Set to one for the time you wish to compare the 45Ω DP termination
resistor to the reference resistor. This bit should be set to one each
time a new value of HW_USBPHYTX_TXCAL45DP is set in order to
compare the resulting resistance. NOTE: only one of the following bits
can be set to one for any calibration operation:
HW_USBPHYTX_TXENCAL1500, HW_USBPHYTX_TXCAL45DN &
HW_USBPHYTX_TXENCAL45P. Set to zero when DP calibration is
completed. The result of a comparison can be seen in
HW_USBPHYTX_TXCOMPOUT.
RESERVED -- must be written with zero.
Decode to select a 45Ω resistance for the DP output pin.
0000= Maximum resistance. Resistance is centered by design at 0110.
Perform calibration routine by initially setting to 1111 and counting
down until the comparator trips.
Test mode bit to skew the transmit signal in order to test the receiver
sensitivity.
Table 94. USB PHY Analog Transmit Control
5-35xx-D1-1.08-061705
95
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
RW RESET
14
TXHSTERM
RW 0
13
TXENCAL45DN
RW 0
12
11:8
RSVRD
TXCAL45DN
R
0
RW 0110
7
TXCALIBRATE
RW 0
6
TXHSXCVR
RW 0
5
TXENCAL1500
RW 0
4
3:0
RSVRD
TXCAL1500
R
0
RW 1000
DEFINITION
Set to one to connect the high speed 45Ω terminations. This forces
both of the FS drivers to zero so that the 45Ω resistors are connected
between DP/DN and ground. It also disconnects the 1500Ω resistor
from VDD so that the pull-up is not there. When first communicating
with a hub, the PHY needs to be in Full Speed mode, i.e. leaving this
bit at zero. Only after the proper high speed handshaking should this
bit be set to one. When the PHY enters Suspend mode, this bit needs
to revert to zero so that Full Speed terminations are re-established. 0-> full speed termination, 1 --> high speed termination.
This bit is muxed with the inverse of UTMI_TERM. The mux is
controlled by HW_USBRX_TERMSELECTEN.
Set to one for the time you wish to compare the 45Ω DN termination
resistor to the reference resistor. This bit should be set to one each
time a new value of HW_USBTX_TXCAL45DN is set in order to
compare the resulting resistance. NOTE: only one of the following bits
can be set to one for any calibration operation:
HW_USBTX_TXENCAL1500, HW_USBPYTX_TXCAL45DN &
HW_USBPHYTX_TXENCAL45P. Set to zero when DN calibration is
completed. The result of a comparison can be seen in
HW_USBPHYTX_TXCOMPOUT.
RESERVED -- must be written with zero.
Decode to select a 45Ω resistance for the DN output pin.
0000= Maximum resistance. Resistance is centered by design at 0110.
Perform calibration routine by initially setting to 1111 and counting
down until the comparator trips.
Set to one to effect calibration of any of the three precision resistances
and set back to zero to read the result of calibration in
HW_USBPHYTX_TXCOMPOUT. When set to one, it causes the
calibration comparator output to continuously update the state of
HW_USBPHYTX_TXCOMPOUT. Set to zero for normal operation.
NOTE: only one of the following bits can be set to one for any
calibration operation: HW_USBTX_TXENCAL1500,
HW_USBPYTX_TXCAL45DN & HW_USBPHYTX_TXENCAL45P.
Set to one to enable the high speed transceiver. This enables the data
lines to control the current steer block. Set to zero for full speed
operation. This bit is muxed with the inverse of UTMI_XCVR_SELECT.
The mux is controlled by HW_USBRX_XCVRSELECTEN.
Set to one for the time you wish to compare the 1500Ω RPU resistor to
the reference resistor. This bit should be set to one each time a new
value of HW_USBTX_TXCAL1500 is set in order to compare the
resulting resistance. NOTE: only one of the following bits can be set to
one for any calibration operation: HW_USBTX_TXENCAL1500,
HW_USBPYTX_TXCAL45DN & HW_USBPHYTX_TXENCAL45P.
Set to zero when RPU calibration is completed. The result of a
comparison can be seen in HW_USBPHYTX_TXCOMPOUT.
RESERVED -- must be written with zero.
Decoded to select a 1500Ω resistance for the RPU Output.
0000= Maximum resistance. Resistance is centered by design at
1000. Perform calibration routine by initially setting to 1111 and
counting down until the comparator trips.
Table 94. USB PHY Analog Transmit Control
9.7.3.
96
USB PHY Analog PLL Control
5-35xx-D1-1.08-061705
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
PLLPFDRST
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
PLLV2ISEL
PLLCPSHORTLFR
1
8
PLLCPDBLIP
PLLVCOKSTART
1
9
PLLVCOCLK2
2
0
PLLVCOCLK24
2
1
PLLCPNSEL
2
2
X:$F212
PLLCLKDIVSEL
2
3
PLLCLKDIVRSTZ
HW_USBPHYPLL
Table 95. HW_USBPHYPLL
BITS
LABEL
RW RESET
23
PLLCLKDIVRSTZ
RW 1
22
PLLVCOKSTART
RW 0
21
PLLCPSHORTLFR
RW 0
20
PLLPFDRST
RW 0
19:16 RSVRD
15:12 PLLCLKDIVSEL
R
0000
RW 0000
DEFINITION
This bit should normally be set to one. It can be momentarily set to
zero and then back to one to provide a pulse to load the clock divider
bits whenever HW_USBPHPLL_PLLCLKDIVSEL is changed.
DCLK should always be switch back to the 24MHz crystal oscillator
source before changing the PLL frequency.
This test bit is provided for the very unlikely event that the VCO does
not start oscillation. This theoretically possible but highly unlikely event
can only happen in a noiseless system-- an unlikely scenario. This bit
is normally set to zero. To kick start the VCO, perform a zero to one
transition on this bit followed by a one to zero transition.
This normally low test mode bit is used to short the charge pump
resistor for a highly under-damped response. Set to one to short the
resistor. The resistor should only be shorted in test mode.
This bit can be used to reset the PFD. This bit is set to zero for normal
operation. To reset the PFD, perform a zero to one transition on this bit
followed by a one to zero transition. This transition is not needed for
normal operation.
RESERVED -- must be written with zero.
The PLLCLKDIVSEL bit-field is used to select the PLL feedback divide
ratios. See Table 97, “PLL Clock Divider Values,” on page 99 for
recommend values for 24MHz and 20MHz crystals.
0000 DIV20
0001 DIV24
0010 DIV30
0011 DIV16
0100 DIV12
0101 DIV10
0110 DIV8
0111 DIV6
1XXX reserved, do not use
Table 96. USB PHY Analog PLL Control
5-35xx-D1-1.08-061705
97
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
RW RESET
11:8
PLLCPNSEL
RW 0000
7
PLLVCOCLK24
RW 0
6
PLLVCOCLK2
RW 0
5
PLLCPDBLIP
RW 0
4
3:0
RSVRD
PLLV2ISEL
R
0
RW 0000
DEFINITION
These bits are set in conjunctions with the
HW_USBPHYPLL_PLLCLKDIVSEL bits to maintain a constant loop
filter damping factor for the different divide ratios. They can also be
used independently to speed up or slow down the activity of the PLL.
Recommended settings are found in Table 97, “PLL Clock Divider
Values,” on page 99.
0000 ip current (default)
0001 ip current
0010 1.50 * ip current
0011 0.75 * ip current
0100 0.50 * ip current
0101 0.50 * ip current
0110 0.40 * ip current
0111 0.40 * ip current
1XXX reserved do not use
Set to one to disable four of the eight phases of PLL clock output, i.e.
turn off vco_clk1, vco_clk1z, vco_clk3, and vco_clk3z. Only vco_clk2,
vco_clk2z, vco_clk4, and vco_clk4z remain enabled when this bit is
set. Disabling these clock phases when they are not needed can save
approximately 1mA. Set to zero to enable all eight phases out of the
PLL. Note this bit overlaps with PLLVCOCLK2 in disabling these
phases.
Set to one to disable six of the eight phases of PLL clock output, i.e.
turn off vco_clk1, vco_clk1z, vco_clk3, vco_clk3z, vco_clk4, and
vco_clk4z. Only vco_clk2 and vco_clk2z remain enabled when this bit
is set to one. vco_clk2 is also used as the digital clock for the 480MHz
digital clock domain and the various digital clock domains resulting
from divisions of the 480MHz oscillator. Disabling these clock phases
when they are not needed can save approximately 1.5mA. Set to zero
to enable driving all phases out of the PLL. Note: this bit overlaps with
PLLVCOCLK24 in disabling some phases. Setting this bit provides
useful power reductions when the high speed PLL is used to drive
DCLK in applications that do not use the USB.
Set this bit to one to double the charge pump current to speed up lock
time. It can be used in conjunction with
HW_USBPHYPLL_PLLCPNSEL to change the loop performance. At
start up time it can be set to one to shorten the lock time. During
normal operation, this be should be set to zero for lowest overall
tracking jitter.
RESERVED -- must be written with zero.
These bits can be used to extend the frequency range of the PLL.
0000 Nominal frequency range (default)
0001 Lower the useful frequency range
0010 Lowest useful frequency range
0011 Highest useful frequency range
01XX Reserved, do not use
1XXXf Reserved, do not use
Table 96. USB PHY Analog PLL Control
.
9.7.4.
98
USB PHY Analog Receive Control
5-35xx-D1-1.08-061705
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
PLLCLKDIVSEL DIVIDE
R
0000
0001
0010
0011
0100
0101
0110
0111
1XXX
FREQUENCY
FOR 24MHZ
CRYSTAL
OSCILLATOR
FREQUENCY
FOR 20MHZ
CRYSTAL
OSCILLATOR
PLLCPNSEL
480MHz
576MHz
720MHz
384MHz
288MHz
240MHz
192MHz
144MHz
unused
400MHz
480MHz
600MHz
320MHz
240MHz
200MHz
160MHz
120MHz
unused
0000,0001
0000,0001
0010
0011
0100
0101
0110
0111
DIV20
DIV24
DIV30
DIV16
DIV12
DIV10
DIV8
DIV6
unused
Table 97. PLL Clock Divider Values
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ENVADJ
HOSTMODETEST
1
7
DISCONADJ
FSCKSOURCESEL
1
8
DEBUGMODE
REGRXDBYPASS
1
9
PLLLKTIMECTL
2
0
TXFSHIZEN
2
1
XCVRTERMSELECTEN
2
2
X:$F213
PLLCKDIVCTL
2
3
PLLLOCKED
HW_USBPHYRX
Table 98. HW_USBPHYRX
BITS
LABEL
23
PLLLOCKED
22
REGRXDBYPASS
RW RESET
DEFINITION
R
PLL
This bit is set whenever the PLL achieves its lock state, i.e.
LOCK whenever the PLL lock counter counts down to zero. This is
STATE determined by design and from characterization to be a maximum
time period. The PLL includes a lock counter to time this period. The
initial value of the lock counter is taken from the
HW_USBPHYRX_PLLLKTIMECTL bit field. What is written to this
bit is indirectly related to what is read.
W N/A
Software must write a zero to this bit position to initiate a new PLL
lock count. After powering up the USB PHY or changing the PLL
feedback divider, software must write a zero to this bit to restart the
lock count.
WARNING: To modify other bits in this register without initiating a
PLL lock cycle, software must write a ONE to this bit position.
RW 0
Set this bit to one to use the output of the DP single ended receiver
in place of the full speed differential receiver. This test mode is
intended for lab use only. For normal operation, set this bit to zero.
Table 99. USB PHY Analog RX Control
5-35xx-D1-1.08-061705
99
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
RW RESET
21
FSCKSOURCESEL
RW 0
20
HOSTMODETECT
RW 0
19:16 PLLCKDIVCTL
RW 0111
15
TXFSHIZEN
RW 0
14
XCVRTERMSELECTEN RW 0
DEFINITION
The UTMI interface of the PHY and the digital full speed transceiver
receive a 60MHz clock from the PHY’s PLL, see Figure 33. “USB
2.0 PHY PLL Block Diagram” on page 89. The source of this clock is
either a divide by eight from the 480MHz PLL or from the chip wide
DCLK. The FSCKSOURCESEL bit selects the 480MHz divided by
eight or the DCLK as the source.
Set this bit to zero to use the 480MHz PLL divided by eight to drive
the Full Speed transceiver and the divide by two that generates the
UTMI clock. Set this bit to one to use the chip wide DCLK to drive
the Full Speed transceiver and the divide by two that generates the
UTMI clock.
Set this bit to one to put the USB 2.0 PHY into host mode. This
mode is not supported in the STMP35xx, however the functionality
is built into the PHY and will be characterized for subsequent
product use of the USB 2.0 PHY in host mode. Set this bit to zero
for normal operation.
The USB 2.0 PHY PLL can be used as the clock generator to drive
DCLK for the entire chip. These bits select the divide ratio used for
driving DCLK. The USB 2.0 PHY PLL frequency is adjusted using
HW_USBPHYPLL_PLLCLKDIVSEL. For some of the values of
PLLCKDIVCTL, the PLL must be running at 120MHz maximum
frequency, these values are marked with a †.
0000 DIV1† (PLL freakiness ≤ 120MHz pass through)
0001 DIV2† (PLL freakiness ≤ 120MHz)
0010 DIV3† (PLL freakiness ≤ 120MHz)
0011 DIV4
(480MHz ÷ 4 = 120MHz
= DCLK)
0100 DIV5
(480MHz ÷ 5 = 96MHz
= DCLK)
0101 DIV6
(480MHz ÷ 6 = 80MHz
= DCLK)
0110 DIV7
(480MHz ÷ 7 = 68.57MHz = DCLK)
= DCLK)
0111 DIV8
(480MHz ÷ 8 = 60MHz
1000 DIV9
(480MHz ÷ 9 = 53.33MHz = DCLK)
1001 DIV10 (480MHz ÷ 10 = 48MHz
= DCLK)
1010 DIV11 (480MHz ÷ 11 = 43.64MHz = DCLK)
1011 DIV12 (480MHz ÷ 12 = 40MHz
= DCLK)
11XX reserved, do not use
This bit is set to one to override the normal control by the PHY
transmitter of the full speed driver’s Hi Z control signal. Set to zero
for normal operation, i.e. controlled by the PHY transmitter.
This bit is set to one to override the normal control by the ARC core
of the xcvr_select and the term_select signals. Set to zero for
normal operation.
Table 99. USB PHY Analog RX Control
100
5-35xx-D1-1.08-061705
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
RW RESET
DEFINITION
13:12 PLLLKTIMECTL
RW 10
11:8
DEBUGMODE
RW 0000
7:4
DISCONADJ
RW 0000
These bits determine the number of clocks counted by the lock
timer before it reports the locked condition in
HW_USBPHYRX_PLLLOCKED. The maximum lock time across
temperature and process variations is 90 µseconds with register
default settings. Lock is defined here as stabilizing within 1% of final
frequency. This field selects a 10-bit, 11-bit, 12-bit or 13-bit counter.
The clock input to this counter comes from the PLL reference clock
source, i.e. the crystal oscillator. For a 24MHz crystal, the lock times
selected are as follows:
00 1024 clocks or 42.67 µS
01 2048 clocks or 85.33 µS
10 4096 clocks or 170.67 µS
11 8192 clocks or 341.33 µS
These bits are used for debug only. These bits must be set to zero
for normal operation.
0000 Normal operation.
0001 HS DLL bypass
0010 HS elastic buffer bypass
0011 Sync detect count reduction
0100 511 test pattern generation and detection
01XX undefined
1XXX undefined
This bit field adjusts the trip point for the disconnect detector.
RW 0000
FIELD
TRIP LEVEL VOLTAGE
0000 0.57500 volts
0001 0.56875 volts
0010 0.58125 volts
0011 0.58750 volts
01XX reserved, do not use
1XXX reserved, do not use
This bit field adjusts the trip point for the envelope detector.
3:0
ENVADJ
FIELD
TRIP
0000 0.10000
0001 0.10625
0010 0.11225
0011 0.12500
LEVEL VOLTAGE
volts
volts
volts
volts
Table 99. USB PHY Analog RX Control
The HW_USBPHYRX_PLLCKDIVCTL bit field selects the divide ratio used to generate
DCLK when the USB 2.0 PHY PLL is used to drive DCLK to the entire chip. The PHY PLL is
of course adjustable and must be set to less than 120MHz for several of the DIVCTL settings.
The following table shows the DCLK values available when the PLL is set at 480MHz and
when it is set at 120MHz. The divider value DIV1 should only be used in test modes.
5-35xx-D1-1.08-061705
101
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
PLLCKDIVCTL
DIVIDE
BY
DCLK
FREQUENCY
FOR 480MHZ
USB 2.0 PHY
PLL
DCLK
FREQUENCY
FOR 120MHZ
USB 2.0 PHY
PLL
0000
0001
0010
0011
0100
0101
0110
0111
1000
1001
1010
1011
11XX
DIV1
DIV2
DIV3
DIV4
DIV5
DIV6
DIV7
DIV8
DIV9
DIV10
DIV11
DIV12
unused
N/A
N/A
N/A
120MHz
96MHz
80MHz
68.57MHz
60MHz
53.33MHz
48MHz
43.64MHz
40MHz
unused
120MHz
60MHz
40MHz
30MHz
24MHz
20MHz
17.1MHz
15MHz
13.33MHz
12MHz
10.9MHz
10MHz
unused
reset
Table 100. PLL Clock Divider Values
102
5-35xx-D1-1.08-061705
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
10. PARALLEL EXTERNAL MEMORY CONTROLLER (EMC)
The chip includes an external memory controller that has two major functional
modes: SmartMedia/NAND and CompactFlash. The SmartMedia/NAND flash interface provides a state machine that provides all of the logic necessary to perform
DMA functions between on-chip RAM and the flash. The CompactFlash interface
supports CompactFlash Memory mode. This mode can be used to communicate
with standard CompactFlash (CF) devices such as CF Flash and the IBM MicroDrive. The CF Memory mode can be used to communicate with standard
ATA/ATAPI devices like CD-ROM and hard drives.
This documentation will detail the configuration and operation of the external memory interface. It will discuss the standard use of the interface for flash applications.
Additional information about other specific applications (CD-ROM, DRAM, NOR
Flash, etc.) will be available in future application notes.
10.1. EMC Overview
The external memory controller has several major features:
• DMA data transfers allow minimal CPU overhead
• 32-bit SmartMedia/NAND addressing supports future devices up to 4Gbyte
• Multiple device support with four SmartMedia/NAND and two CF chip selects
• Configured timing can be set to support various devices
The external memory controller can be described as two fairly independent devices
in one: a SmartMedia/NAND flash interface and a CompactFlash/NOR flash/IDE
interface. Both interfaces share the same device pins, some registers and the DMA
engine.
Both interfaces uses memory mapped registers to setup and control the transactions. Data is always sent through the DMA – there are no data registers that correspond to the interface data bus. Transactions are always started with a kick bit. The
interface sets up the control lines and transfers data to/from the internal RAM. Once
the transaction is complete the interface signals the DSP with either a polled flag or
an interrupt. The kick bit, polling bit and interrupt configurations are all contained in
the registers.
5-35xx-D1-1.08-061705
103
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
10.1.1.
EMC External Pins
The EMC supports several different modes of operation, and the pins used in each
mode overlap substantially. The following table shows how the pins relate to each
other.
SMARTMEDIA/NAND
PIN NAME
SM_D15 - SM_D8
SM_D7 - SM_D0
SM_WEn
SM_REn
*
SM_ALE
SM_CLE
SM_SEn
SM_CE0n
SM_CE2n
SM_CE3n
*
*
*
*
SM_CE1n
*
*
*
*
*
*
*
SM_WPn
*
*
SM_READY
CF+/COMPACTFLASH
PIN NAME,
MEMORY MODE
CF_D15 - CF_D8
CF_D7 - CF_D0
CF_WEn
CF_OEn
CF-A23 - CF_A10
CF_A9
CF_A8
CF_A7
CF_A6
CF_A5
CF_A4
CF_A3
CF_A2
CF_A1
CF_A0
CF_CE0n
CF_CE1n
*
*
CF_WPn
CF_REGn
CF_RESETn
CF_BVD1/IREQ
*
CF_READY
CF_CDn
CF_WAITn
Table 101. Mapping of pins to CF+/CompactFlash and SmartMedia/NAND Card/Device Pins
Note: Pins marked with (*) are not used by the module and are available for use in GPIO mode. See the GPIO module
documentation for more information. General Use of the External Memory Interface
The external memory interface has some functions that are shared between both
the SmartMedia/NAND and the CompactFlash modes. In particular, common registers control the functions that communicate with the DSP such as the DMA and
transaction start/done functions. Each of the memory interfaces also have their own
registers that set up their specific parameters. This section will focus on the common
registers.
10.1.2.
104
Common Flash Registers
5-35xx-D1-1.08-061705
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
Flash Control Register
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
KICK
1
9
RW
2
0
TCIE
2
1
IRQP
2
2
SRST
2
3
X:$F000
MMD
HW_FLCR
NB
10.1.2.1.
Table 102. HW_FLCR
BITS LABEL RW RESET
23:22 RSRVD R
0
21
SRST
RW 0
20:19 RSRVD R
0
18:6 NB
RW $001
5:4
MMD
R
0
3
IRQP
RW 0
2
1
0
TCIE
RW
KICK
RW 0
RW 0
RW 0
DEFINITION
Reserved – Must be written with 0.
Flash module software reset – This bit is itself reset by a hardware reset only; a
software reset does not clear the bit.
Reserved – Must be written with 0.
Number of bytes to transfer
External interface type
00
SmartMedia/NAND
10
Reserved
01
CompactFlash
11
Reserved
Reading a one indicates a pending interrupt; writing back a one de-asserts the
interrupt; writing zero has no effect.
Setting this bit enables the transaction-complete interrupt.
Setting this bit initiates an external memory transfer is a read; otherwise it is a write.
Setting this bit initiates an external memory transfer; it automatically clears when the
transfer completes.
Table 103. Flash Control Register Description
The Flash Control Register sets up basic operating parameters. After the transaction has been set up in all of the appropriate registers (transaction type, timing,
address, etc.) the software triggers the kick bit. This will initiate the transaction. The
status bit will indicate when the transaction has been completed. The DSP can
either poll the status bit or the register can be configured to send an interrupt on
transaction completion.
The Flash Start Address Low and High Registers determine the addresses of both
the internal XRAM, YRAM and PRAM and the external device. Multiple cycle transactions (reading/writing multiple bytes sequentially) will pack/unpack data from the
DSP XRAM, YRAM and PRAM’s 24-bit words into the external device’s 8-bit interface.
The LSB is packed/unpacked first. Each byte is packed with its MSB at bit 7, 15, or 23 for
bytes 1, 2 or 3.
The address bus’s behavior depends on the specific external device type. SmartMedia/NAND devices don’t use the address bus. They are externally selected with chip
select lines (CE0-3) and internally with multi-byte accesses over the I/O lines (see
more in the SmartMedia/NAND section, below). CompactFlash devices may use the
Address pins in Memory or I/O mode or with a combination of control pins (CE0,
CE1, CF_A0, A1, A2) in IDE Mode. In all cases the mode configuration registers
determine how the target device is addressed.
5-35xx-D1-1.08-061705
105
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
Flash Control 2 Register
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
RA
2
2
CKGT
2
3
X:$F004
ASEL
HW_FLCR2
NDMA
10.1.2.2.
0
0
Table 104. HW_FLCR2
BITS
23:7
6
5
4
3
2
LABEL
RSRVD
CKGT
RSRVD
NDMA
RSRVD
RA
RW
RW
RW
R
RW
R
RW
RESET
0
0
10
0
00
0
1:0
ASEL
RW 00
DEFINITION
Reserved – Must be written with 0.
Turns clocks to Flash module off.
Reserved – Must be written with 0.
Inverts data from the Flash to the system.
Reserved – Must be written with 0.
Right align 16-bit word data from bits from 15:0. The MSByte will be zeroed. This is
only associated with CompactFlash16-bit data mode in the CFControl register bit 23. If
the Right align bit is not set then the CompactFlash 16-bit data mode will ‘pack’ the
word data into the 24-bit memory. NOTE: The left align mode of the STMP3410 is not
supported in the STMP35xx.
Memory space to use for DMA transfers
00
X space
10
P space
01
Y space
11
Reserved
Table 105. Flash Control 2 Register Description
The Flash Control 2 Register sets up additional operating parameters.
10.1.2.3.
Flash Start Address Low Register
HW_FLSALR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
XA
2
3
X:$F001
Table 106. HW_FLSALR
BITS LABEL RW RESET
23:0
XA
DEFINITION
Lower 24 bits of external memory starting address
Table 107. Flash Start Address Low Register Description
106
5-35xx-D1-1.08-061705
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
10.1.2.4.
Flash Start Address High Register
HW_FLSAHR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
XA
2
3
X:$F002
Table 108. HW_FLSAHR
BITS LABEL RW
23:8
7:0
DA
XA
R
R
DEFINITION
Starting address for DMA transfer.
Upper 8 bits of external memory starting address
Table 109. Flash Start Address High Register Description
10.2. External Memory Interface with SmartMedia/NAND Flash Devices
The external memory interface will commonly be used with SmartMedia/NAND flash
devices. SmartMedia are small removable cards that contain one or two NAND
Flash devices. Alternatively, the system designer can use non-removable NAND
flash chips. Both devices use the same pins. The SmartMedia/NAND electrical interface uses an 8-bit data/address bus and 8 control lines. Multiple devices share the
same bus, with chip selects used for selection.
10.2.1.
SmartMedia/NAND Pins
• SM_D0 - SM_D7: 8-bit I/O interface. This bus is used to send addresses
(multiple bytes required to send a complete address), data and commands to the
SmartMedia/NAND device. Data, device ID and status information are received
from the Flash over this bus.
• SM_WEn: Write Enable. This active low output is used to indicate a write cycle to
the Flash. The external memory interface will write to the bus during a WE~
cycle.
• SM_REn: Read Enable. This active low output indicates a read cycle to the
Flash. The flash device writes to the bus during a RE~ cycle.
• SM_ALE: Address Latch Enable. This active high output indicates that the data
on the I/O bus is part of an address. The chip outputs one byte of a Flash
memory address during this cycle.
• SM_CLE: Command Latch Enable. This active high output indicates that the data
on the I/O bus is a flash command. The chip outputs a SmartMedia/NAND
command during this cycle. See the data sheet for your SmartMedia/NAND
device for more information on the valid commands.
• SM_SEn: Spare Enable Select. This active low output is used to indicate that the
ECC data should be sent with a regular 512 Byte Flash Block. This command is
not typically used since the ECC should always be used. It is recommended that
the SE~ line be tied to ground (asserted) to ensure that the spare area is always
used.
• SM_CE0n – SM_CE3n: Chip Enable Lines. Active low outputs that select each
SmartMedia or NAND Flash. Connect one to each of the flash devices.
• SM_WPn: Write Protect. This active low output is used prevent accidental data
corruption during non-write cycles.
5-35xx-D1-1.08-061705
107
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
• SM_READY: Ready/ Busy~. This active low input indicates that the flash device
is busy. No new transactions to any devices should occur unless this line is not
asserted. The Ready/Busy~ line uses open-drain drivers and requires an
external pull-up resistor. All SmartMedia/NAND Flash devices busy pins should
be connected to this pin.
The basic unit of access of a SmartMedia/NAND device is the 8 bit byte. All bytes
must be written in complete 528 byte “pages”. A “block” is made up of 32 pages. A
64MByte NAND Flash has 4096 blocks, see Figure 36. “NAND Flash Device Block
Diagram” on page 109. When an address is sent to a NAND Flash it is sent in three
or four bytes depending on the size of the device.
CYCLE
ADDR 0
ADDR 1
ADDR 2
ADDR 3
I/O 7
A7
A16
A24
0
I/O 6
A6
A15
A23
0
I/O 5
A5
A14
A22
0
I/O 4
A4
A13
A21
0
I/O 3
A3
A12
A20
0
I/O 2
A2
A11
A19
0
I/O 1
A1
A10
A18
0
I/O 0
A0
A9
A17
A25
Table 110. SmartMedia/NAND Device Address Bytes
Transfers to and from a NAND Flash device are made in packets, starting with a
control byte which may be followed by one, three or four bytes of address. The command and address portion can be followed by up to 528 bytes of data for a read or
program (write) command, see Figure 37. “NAND Flash Device Command & Data
Sequences” on page 110. When a NAND Flash contains more than 32MBytes of
storage the optional four address byte format has to be used. Pay careful attention
to the device address bit mapping in Table 110, “SmartMedia/NAND Device
Address Bytes,” on page 108. There is no address bit 8. This reflects the growth of
page sizes from 256 bytes to 512 bytes as NAND flash technology evolved. Three
read commands are provided to allow starting at offsets of zero bytes, 256 bytes or
512 bytes into a 528 byte page.
The reader is cautioned to pay careful attention to the mapping of Flash Start
Address Low and High register bits to the four address bytes. Recall the “missing”
address bit 8 in the NAND device’s interpretation of these address bytes.
CYCLE
ADDR 0
ADDR 1
ADDR 2
ADDR 3
I/O 7
AL7
AL15
AL23
AH7
I/O 6
AL6
AL14
AL22
AH6
I/O 5
AL5
AL13
AL21
AH5
I/O 4
AL4
AL12
AL20
AH4
I/O 3
AL3
AL11
AL19
AH3
I/O 2
AL2
AL10
AL18
AH2
I/O 1
AL1
AL9
AL17
AH1
I/O 0
AL0
AL8
AL16
AH0
Table 111. Flash Address Low/High mapping to SmartMedia/NAND
108
5-35xx-D1-1.08-061705
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
sense amps & I/Os
sm_ad[0]
512
sm_ad[7]
16
} 32 pages = 1 block
528 byte page
Figure 36. NAND Flash Device Block Diagram
The EMC NAND FLASH controller supports six initial commands (iCMD) and two
packet ending commands (eCMD) as shown in the figure below. Initial commands
include:
• The RESET command initializes the NAND flash internal state machine to idle.
• The Read Status command is useful to tell when a page has been programmed
or a block has been erased.
• The Read Data command is used to fetch a page of data from the Flash. One
should generally fetch all 528 bytes and apply any error correction stored in the
extra 16 bytes to the 512 real data bytes.
• The Program Data command writes 528 data bytes to a page in the NAND Flash.
• Each device has a built in vendor ID and device ID that can be read from the
device with a READ ID command.
Ending commands include:
The AUTO ERASE START command $60.
The AUTO PAGE PROGRAM command $10.
The appropriate ending command is automatically sent by hardware at the end of
the program sequence and the erase block commands.
5-35xx-D1-1.08-061705
109
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
reset
iCMD
Read Status
iCMD
D
Block Erase
iCMD
A1
A2
A3
eCMD
required for devices larger than 32MB
Read Data
iCMD
A0
A1
A2
A3
D0
D1
Program Data
iCMD
A0
A1
A2
A3
D0
D1
Read ID
iCMD
A
D
data byte from ROM
D1
byte from DSP3410
Figure 37. NAND Flash Device Command & Data Sequences
As the above figure shows, all packets sent to the NAND device start with a command byte followed by the appropriate number of address bytes. Either data bytes
are then sent to the device or data./status/ID bytes are received from the device.
Note the first address byte sent on the Block Erase command is Address Byte 1.
All timing references in the NAND Flash controller are based on DCLK. Since DCLK
is quite variable, the timing parameters must be recalculated and reloaded whenever the frequency of DCLK changes. For a timing references, see Figure 38.
“NAND Flash Write Timing” on page 111.
110
5-35xx-D1-1.08-061705
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
sm_cle
tRWSU
tRWH
tWPW
tWPW
sm_wen
tWPW
tRWH
tRWSU
sm_ale
tRWH
tRWSU
tRWSU
tRWH+tRWSU
read cmd
sm_d[7:0]
$00
AL[7:0]
AL[15:8]
Figure 38. NAND Flash Write Timing
.
sm_cle
sm_ale
tRWSU
tRWH
tRPW
tRPW
sm_ren
tRPW
tRWSU
tRWH
tRWSU
tRWH
tRWSU
tRWH+tRWSU
sm_d[7:0]
data byte 0
dat byte 1
data byte 2
Figure 39. NAND Flash Read Timing
5-35xx-D1-1.08-061705
111
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
SmartMedia/NAND Registers
SmartMedia/NAND Control Register
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
CS
1
8
SE
1
9
WP
2
0
SIZE
2
1
ICMD
2
2
TOIE
2
3
X:$F010
BPIE
HW_FLSMCR
TOIRQ
10.2.2.1.
BPIRQ
10.2.2.
Table 112. HW_FLSMCR
BITS LABEL RW RESET
23:17 RSRVD R
16
BPIRQ R
DEFINITION
0
0
Reserved – Must be written with 0.
Bad programming interrupt pending – Reading a one indicates a pending interrupt;
writing back a one de-asserts the interrupt; writing zero has no effect.
15
TOIRQ RW 0
Timeout interrupt pending – Reading a one indicates a pending interrupt; writing
back a one de-asserts the interrupt; writing zero has no effect.
14
BPIE
RW 0
Bad Programming interrupt enable – Generate BPIRQ if SmartMedia/NAND
interface is kicked with bad programming.
13
TOIE
RW 0
SmartMedia/NAND Timeout interrupt enable
12:5 ICMD
R
$FF
Initial standard SmartMedia/NAND command for transaction
$FF Card Reset
$70 Card Status Register Read
$60 Card Block Erase. Second command byte is always $D0.
$00 Read/Program Data. Start address is in first 256 bytes of 528-byte page.
$01 Read/Program Data. Start address is in second 256 bytes of 528-byte
page.
$50 Read/Program Data. Start address is in redundant area (last 16-bytes of
page).
$90 Car ID Read
4
SIZE
R
0
Target SmartMedia/NAND device size
0
<= 32 MBytes
1
> 32 MBytes
3
WP
RW 0
Write Protect – Controls the output of the SM_WPn pin (active-low)
0
Assert -WP pin
1
De-assert -WP pin
2
SE
RW 0
-SE pin control – Allows the spare (redundant) area (last 16 bytes of page) to be
skipped.
0
assert -SE pin. Redundant area enabled.
1
de-assert -SE pin. Redundant area disabled.
1:0
CS
R
0
External Chip Select
00
Select Chip 0
01
Select Chip 1
10
Select Chip 2
11
Select Chip 3
Note: The SmartMedia/NAND interface's state machine logic makes a number of checks on the legitimacy of DSP
programming before allowing the master state machine out of its IDLE state. If these checks fail, the state
machine remains IDLE, the BPIRQ gets set and the transaction is considered done (Kick bit in the HW_FLSMCR
register is de-asserted). Optionally, a bad-programming interrupt can be enabled by setting the BPIE bit.
Table 113. SmartMedia/NAND Control Register Description
112
5-35xx-D1-1.08-061705
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
The SmartMedia/NAND Control Register configures the interface and determines
the type of transaction. The SmartMedia/NAND portion of the interface was
designed to very closely interface to external Flash device. The interface knows how
to set the appropriate control and data bits at a device command level. Operations
like Reset, accesses to the Status Register and data operations are automatically
processed by the interface. The user code just needs to set up some basic timing
and device information and make a high level request.
SmartMedia/NAND Timer 1 Register
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
TWPW
2
2
TRWH
2
3
X:$F011
0
7
0
6
0
5
0
4
0
3
0
6
0
5
0
4
0
3
0
2
0
1
0
0
0
1
0
0
TRWSU
HW_FLSMTMR1R
TRPW
10.2.2.2.
Table 114. HW_FLSMTMR1R
BITS
23:22
21:17
16:11
10:5
4:0
LABEL
RSRVD
TRWH
TWPW
TRPW
TRWSU
RW
R
RW
RW
RW
RW
RESET
0
00010
000110
000110
00010
DEFINITION
Reserved – Must be written with 0.
Read/Write Pulse Hold in terms of dclks
Write Pulse Width in terms of dclks
Read Pulse Width in terms of dclks
Read/Write Strobe Setup in terms of dclks
Table 115. SmartMedia/NAND Timer 1 Register Description
SmartMedia/NAND Timer 2 Register
HW_FLSMTMR2R
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
TWTO
2
3
X:$F012
0
2
TWT
10.2.2.3.
Table 116. HW_FLSMTMR2R
BITS LABEL RW RESET
23:6 TWTO RW $001E8
5:0
TWT
RW 010000
DEFINITION
Delay before timing out on RDY pin in terms of dclks
Delay before examining RDY pin in terms of (dclks x 1024)
Table 117. SmartMedia/NAND Timer 2 Register Description
5-35xx-D1-1.08-061705
113
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
The SmartMedia/NAND timing registers default to conservative settings on reset.
TIMING PARAMETER
Trwh
Twpw
Trpw
Trwsu
Twto
Twait
DEFAULT IN DCLK CYCLES
TIME @ 48MHZ (21NS CYCLE)
2
42ns
6
125ns
6
125ns
2
42ns
488 * 1024
10.5ms
16
336ns
Table 118. SmartMedia/NAND Timing Specifications
SAMSUNG SPEC
40ns
80ns
80ns
30ns
4ms
200ns
The firmware designer can change the timing to accommodate higher performance
or newer flash devices, and should check that the timing values are acceptable if the
dclk clock frequency is changed.
10.3. External Memory Interface in CompactFlash Mode
Some applications will require an interface to either a CompactFlash device such as
a standard CF memory card or IBM Microdrive. Other applications will interface to a
standard IDE/ATAPI device like a CD-ROM or hard disk. The system designer may
also want to interface to an external SDRAM, DRAM or NOR Flash for bulk data
storage. All of these are possible with the CompactFlash mode of the external Memory Interface and, occasionally, some external glue logic like a CPLD.
This interface mode differs significantly from the SmartMedia/NAND mode in that it
doesn’t have the intelligence to closely control any device family at a high level. All
of the control pins must be configured at the register level. High-level command
state machines are implemented in software rather than the memory interface. This
requires additional system software development. The benefit is significantly
increased flexibility. The CompactFlash mode covers a large number of interfaces.
10.3.1.
CompactFlash Modes
The STMP35xx supports the 8 bit CompactFlash (CF) Memory Mode in the 100-pin
package. In the 144-pin fpBGA, the STMP35xx supports 16 bit CompactFlash (CF)
Memory Mode The CompactFlash (CF) Memory mode can be used to connect hard
disk drives and CD-ROM drives.
114
5-35xx-D1-1.08-061705
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
CompactFlash Registers
CompactFlash Control Register
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
WP
0
8
CDP
0
9
SM
1
0
XATTR
1
1
CRST
1
2
RI
1
3
XWT
1
4
IFCE
1
5
ISCE
1
6
INCE
1
7
IFCS
1
8
ISCS
1
9
INCS
2
0
CS
DASP
2
1
X:$F008
CRE
2
2
VS
2
3
MODE16
HW_FLCFCR
CFAI
10.3.2.1.
XDDI
10.3.2.
Table 119. HW_FLCFCR
BITS
23
22
21:20
19
LABEL
MODE16
DASP
VS
CRE
RW
RW
R
R
RW
18:17 CS
16
XDDI
15:14 CFAI
RW
R
RW
RESET
0
unknown
unknown
EMC5600_
CARDRSTEN_OUT
00
unknown
10
13
INCS
RW
0
12
ISCS
RW
0
11
IFCS
RW
0
10
9
INCE
ISCE
RW
RW
0
0
8
IFCE
RW
0
7
RI
R
unknown
6
XWT
R
unknown
5
CRST
RW
4
XATTR
RW
EMC5600_
CARDRST_OUT
0
DESCRIPTION
CompactFlash data bus is 16-bits
DASP pin of the CompactFlash card, if connected.
VS2, VS1 voltage sense inputs (if connected)
Output enable for RESET/-RESET pin
Active-high versions of outgoing -CE2,-CE1/-CS1,CS0 chip selects
Incoming -PDIAGN/-STSCHG pin, sampled into dclk domain
Multi-byte transfer card address increment/toggle
00
next A10-A0 same as previous
01
next A10-A0 equal to previous plus one
10
next A10-A0 equal to previous plus two (CIS)
11
next A10-A0 equal to previous, but A0 toggles
Reading a one indicates a pending interrupt; writing back a one deasserts the interrupt; writing zero has no effect
Reading a one indicates a pending interrupt; writing back a one deasserts the interrupt; writing zero has no effect
Reading a one indicates a pending interrupt; writing back a one deasserts the interrupt; writing zero has no effect
Setting this bit enables the card absence/removal interrupt
Setting this bit asserts an interrupt based on the -PDIAGN/-STSCHG
pin of the CompactFlash card going low in any PC Card Mode
Setting this bit passes on a card-based interrupt from the RDY/BSY/-IREQ/INTRQ pin of the CompactFlash card. The interrupt
occurs if the input pin transitions (either direction) in any PC Card
socket mode (Memory, I/O, True IDE)
RDY/-BSY/-IREQ/INTRQ pin of the CompactFlash card, sampled
into dclk domain
-WAIT active-low card pin of the CompactFlash card, sampled into
dclk domain
0
wait/card is busy
1
card/data is available
Output value to RESET/-RESET pin of the CompactFlash card.
Active-high in Memory or I/O Mode; Active-low in True IDE Mode.
0
Access to Attribute Memory or I/O space
1
Access to Common Memory
Table 120. CompactFlash Control Register Description
5-35xx-D1-1.08-061705
115
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS LABEL
3:2
SM
RW
RESET
RW 2'b00
1
0
RW
R
CDP
WP
DESCRIPTION
Card access mode
00
PC Card Memory Mode
01
reserved
10
reserved
11
reserved
Card Detect pin (-CD1) pull-up enable/disable
Write Protect (WP) pin of CompactFlash card, sampled into dclk domain
EMC5600_PUP
unknown
Table 120. CompactFlash Control Register Description (Continued)
CompactFlash Timer1 Register
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
TWPW
2
2
TRWH
2
3
X:$F009
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
TRWSU
HW_FLCFTMR1R
TRPW
10.3.2.2.
Table 121. HW_FLCTMR1R
BITS
LABEL
RW
RESET
23:19 TRWH
RW
00001
18:12 TWPW
11:5
TRPW
4:0
TRWSU
RW
RW
RW
0001010
0001010
00001
DESCRIPTION
Hold time for -CE, An, -REG pins of the CompactFlash card, relative to -WE, RE, -IOWR or -IORD de-asserting, in dclk cycles
-WE, -IOWR pins of the CompactFlash card pulse width, in dclk cycles
-OE, -IORD pins of the CompactFlash card pulse width, in dclk cycles
Read or write pin of the CompactFlash card setup time, in dclk cycles
Table 122. CompactFlash Timer1 Register Description
CompactFlash Timer2 Register
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
TWW
2
2
TRAQ
2
3
X:$F00A
TWTO
HW_FLCFTMR2R
THW
10.3.2.3.
Table 123. HW_FLCFTMR2R
BITS LABEL RW
RESET
DESCRIPTION
23:19 TRAQ
RW
000101
18:14 THW
RW
00001
13:4
3:0
RW
RW
$0C8
0100
Delay between -OE/-IORD de-assertion and EMC interface re-acquiring the data
bus, in dclk cycles
Hold time between -WAIT de-asserting and read or write strobe (-OE, -WE, etc.) deasserting, dclk cycles
Timeout waiting for card-imposed wait (-WAIT low) period, in dclk cycles
Wait-for-wait time, in dclk cycles. Time for -WAIT to assert, from -OE/ -WE/-IOWR/IORD assertion edge.
TWTO
TWW
Table 124. CompactFlash Timer2 Register Description
116
5-35xx-D1-1.08-061705
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
10.4. Known Chip Defects with EMC Device
10.4.1.
Clear Quest Entry STMP00004728
The Left Align mode feature of the STMP3410 was removed from the STMP35xx
design. This CQ Entry was closed as a works as designed entry. However, several
users have discovered empirically instead of from the documentation.
5-35xx-D1-1.08-061705
117
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
11. GENERAL PURPOSE FLASH CONTROLLER
The General Purpose Flash Interface controller (GP Flash) is a DSP configurable
interface to external NAND Flash. This highly configurable and flexible interface can
attach to and utilize most readily available NAND Flash devices, including the newer
large block erase 1Gbit per die devices. The GP Flash is a DMA device that can
autonomously transfer NAND Flash pages to or from on-chip RAM as shown in Figure 40.
Y-BUS
X-BUS
P-BUS
DSP CORE
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
DMA-BUS
GPFLASH
Programmable
Registers
GPFLASH
DMA Engine
General Purpose Flash
State Machine
NAND Flash/
SmartMedia
Figure 40. General Purpose Flash Controller
The GP Flash interface provides automatic timing control for the critical data read
and write access signal lines. The interface automatically maintains proper CLE and
ALE setup and hold time as well as proper read/write strobe to data setup and hold
times making automatic transfers via DMA practical. see Figure 41. “GP Flash Command/Address/Write Data Timing” on page 119. Also see Figure 42. “GP Flash
Read Timing” on page 119. These diagrams show the fundamental constraints
maintained by the interface controller. In addition, it shows the programmable registers and bit fields that control the various critical times in the NAND Flash cycle.
The GP Flash DMA will transfer the data to/form the system X/Y/P memory. The
external data bus can be configured to be 8-bit or 16-bit access.
118
5-35xx-D1-1.08-061705
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
HW_GPFLASH_TIMING1R
SM_CLE
SM_ALE
SM_CE0n
SM_REN
TSU
THOLD
SM_WEN
TDS
TDH
SM_D[7:0]
SM_D[15:8]
HW_GPFLASH_TIMING2R
no command or address
Figure 41. GP Flash Command/Address/Write Data Timing
HW_GPFLASH_TIMING1R
SM_CLE
SM_ALE
SM_CE0n
SM_WEN
HW_GPFLASH_TIMING2R
TSU
THOLD
SM_REN
TDS
TDH
TDS
from flash device
SM_D[7:0]
SM_D[15:8]
TDS
Data Capture
TDS
Figure 42. GP Flash Read Timing
Every transfer over the external Flash data bus travels over the DMA bus to or from
the on-chip RAM. This includes the address bytes and it includes the command byte
or bytes sent with CLE asserted. The GPFlash interface controller separates the
three phases of Flash device access into three separate DMA transactions, one for
the command byte, one for the address bytes and a third one for moving the actual
data bytes to or from the device. Thus one must “kick-off” a DMA transaction to send
the read command byte itself and another DMA to send the address bytes for the
read. see Figure 43. “GP Flash Command Timing” on page 120.
5-35xx-D1-1.08-061705
119
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
There are two sets of setup and hold control counters. One set is for maintaining
setup and hold constraints on the command latch enable (CLE)/address latch
enable (ALE). A second set of counters is for managing constraints for the data
transfer setup and hold time. The ALE/CLE setup and hold times are controlled in
HW_GPFLASH_TIMING1R while the data setup and hold times are controlled in
HW_GPFLASH_TIMING2R.
HW_GPFLASH_TIMING1R_THOLD
KICK
SM_CE0n
SM_CLE
TSU
TH
TDS
TDS
SM_WEN
TDH
TDS
TDH
TSU
SM_ALE
read cmd
SM_D[7:0]
$00
AL[7:0]
AL[15:8]
Figure 43. GP Flash Command Timing
The kick bit in HW_GPFLASHCSR0R is used to kick off a DMA transfer. Bits in the
CSR0R and CSR1R registers permit the definition of a command transfer, an
address transfer or a data read or write transfer. All are initiated or kicked-off by setting the KICK bit. Software monitors the state of the KICK bit either by polling or by
waiting on a transfer done interrupt. Figure 43 shows the KICK bit being set by software for the command transfer phase and being reset by hardware at the completion of the single byte command transfer. It shows the KICK bit being set again to
transfer the address bytes. Of course, the HW_GPFLASHCSR0R_XFER_TYPE
bits
were
changed
from
COMMAND
to
ADDRESS,
and
the
HW_GPFLASH_DMA_ADDR_DMA_ADD field was set to point to the first address
byte before this second “kick” was performed. In addition, the
HW_GPFLASH_XFER_SIZE_NUM_BYTES field was changed from one to four or
whatever was appropriate for the device being read.
We see that even to send a single command byte, CSR0, CSR1, have to be setup,
the NUM_BYTES field has to be set to one, the DMA_ADD field must point to an onchip RAM byte containing the proper command and a DMA must be kicked-off.
Figure 44 shows a complete transaction consisting of a one byte command DMA
transfer, a two byte address DMA transfer and a two byte data read DMA transfer.
Many NAND Flash devices expect the chip enable to remain low during all three
phases, as in Figure 44. The GP Flash state machines respond to the
HW_GPFLASH_CSR0R_CEB_CTRL bit by leaving the selected chip enable pin in
the state indicated by this bit at the end of a DMA transfer. Figure 44 shows the
effect of leaving this bit low for the command and address DMA transfers but setting
it before the data read DMA transfer. Thus it ends up high (device disabled) after the
complete three phase transaction. CEB is brought high after its hold time constraint
is met, as specified in HW_GPFLASH_TIMER1R_THOLD. More than one data
phase DMA transaction can be used while keeping the flash device selected.
120
5-35xx-D1-1.08-061705
5-35xx-D1-1.08-061705
SM_D[15:8]
SM_D[7:0]
SM_ALE
$00
read cmd
TDS
TH
TSU
TDH
AL[7:0]
TDS
TDH
AL[10:9]
TDS
ADDRESS
AL[18:11]
Figure 44. GP Flash Command, Address and Read Data Operations
TSU
COMMAND
AL[26:19]
read data
read data
read data
read data
DATA
PRODUCT
SM_WEN
SM_CLE
SM_CE0n
KICK
OFF ICI AL
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
121
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
11.1. Fractional Word Writes
Whenever a write to on-chip RAM ends at a non-word boundary then a fractional
word write occurs. This can arise from setting HW_GPFLASH_CSR1_START_
BYTE to start transferring either to the Intermediate or most significant byte (byte
one or byte two) while setting HW_GPFLASH_XFERSIZER to an even word size,
i.e. byte count is a multiple of three. It can also arise from setting
HW_GPFLASH_CSR1_START_BYTE to start transferring with the least significant
byte (byte zero is the aligned case) while setting HW_GPFLASH_XFERSIZER to a
value that is not evenly divisible by three.
In either of these cases, the last word of the transfer is only partially filled from the
GP FLASH, see Figure 45. A read-modify-write cycle is performed by the GP
FLASH hardware for the last word so that the one or two bytes that are not targeted
by the GP FLASH transfer remain unmodified in on-chip RAM. Similarly for a partial first word write, i.e. with an unaligned start byte, a read-modify-write cycle is performed to maintain the values of the untargeted bytes.
Byte from
GPFLASH
MSB
(BYTE 2)
ISB
(BYTE 1)
LSB
(BYTE 0)
GPFLASH Hardware
performs read-modify-write to
keep original values here
Figure 45. GP Flash First or Last Partial Word Transfer
11.2. Programming Example
There are certain settings that must be made when the GP FLASH module is first
initialized. The initialization flow chart provides an example of such an initialization
sequence, see Figure 46. “GP Flash Example Initialization Flowchart” on page 123.
First, clocks are turned on and software reset is removed to enable the use of the
GP Flash. Next the timing control bit fields are initialized. The NAND Flash device
specification should be consulted to determine the timing specifications for the
device in absolute time, i.e. nano-seconds. Next these parameters must be converted to DCLK cycle counts by dividing each timing specification by the period of
the intended DCLK frequency. The ceiling of this number must be used for minimum
specifications. The resulting integer value is then loaded into the appropriate bit
field.
122
5-35xx-D1-1.08-061705
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
GPLASH Init
Turn on clocks, remove reset
HW_GPFLASH_CSR0R_CLK_DISABLE = 0
HW_GPFLASH_CSR0R_SOFT_RESET = 0
Initialize Timing Values based
on DCLK frequency
HW_GPFLASH_TIMING1R_TSU = ?
HW_GPFLASH_TIMING1R_HOLD = ?
HW_GPFLASH_TIMING2R_TDS = ?
HW_GPFLASH_TIMING2R_TDH = ?
HW_GPFLAS_TIMINGBUSYR_TWTO = ?
enable writes at device
HW_GPFLASH_CSR1R_WP0 = 0
HW_GPFLASH_CSR1R_MUX = 1
use GPFLASH not EMC
STOP
Figure 46. GP Flash Example Initialization Flowchart
Consider the full transfer timing diagram shown above, see Figure 44. “GP Flash
Command, Address and Read Data Operations” on page 121. The flowchart of Figure 47 outlines a function designed to generate this sequence and read a page of
data bytes from a NAND Flash device and write them into a buffer in on-chip RAM.
Note that the one-byte read command is read from a location in on-chip RAM via
DMA and written to the NAND Flash device with the command latch enable (CLE)
signal asserted. As shown in the flowchart, three separate transfers are “kicked-off”
to effect reading data bytes from a NAND Flash page.
The four byte NAND Flash address is read from a buffer in on-chip RAM and written
to the device. For this transfer, the CLE signal has been negated and the address
latch enable signal (ALE) is asserted. The ALE/CLE setup and hold timings are
automatically guaranteed by the hardware. Finally, the device is configured for a
data read transfer, depositing the NAND Flash data into a buffer in on-chip RAM.
The example provided in the flowchart of Figure 47 uses polling to determine when
the GP FLASH hardware has completed a transfer that was initiated by setting the
KICK bit. While this method functions correctly, it wastes significant time in busy
waiting using polling.
It is much better to use the transfer done (XFERDONE) interrupt to monitor completion of each phase of the transaction, see Figure 48. “GP Flash Send Command
using XFERDONE Interrupt” on page 125.
This flowchart shows only the command byte transfer. Address and data transfers
using XFERDONE interrupts are implemented in a similar manner. Notice that the
ISR will have to have a small amount of state information to determine what action
to take on a completion interrupt, e.g. after a command transfer, initiate the address
transfer, etc.
5-35xx-D1-1.08-061705
123
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
Run GPFlash Init First
GPFLASH Example Read
send cmd & addr in 8 bit mode
HW_GPFLASH_CSR1R_MODE16 =0
HW_GPFLASH_CSR1R_CEB = n
HW_GPFLASH_CSR1R_START_BYTE = ?
HW_GPFLASH_CSR1R_MEM = 10
select one device's chip enable
select starting byte alignment
get cmd/addr bytes from P-MEM
Write transfer for command byte
HW_GPFLASH_CSR0R_RW = 0
HW_GPFLASH_CSR0R_CEB_CTRL = 0
HW_GPFLASH_CSR0R_XFER_TYPE = 00
Leave chip enable low after cmd
Select COMMAND transfer
point to 1 byte cmd in on-chip RAM
HW_GPFLASH_DMA_ADDR_DMA_ADD = #>cmd_read
HW_GPFLASH_XFER_SIZE = 1
HW_GPFLASH_CSR0R_KICK = 1
YES
Kick off 1 byte CMD xfer
Wait for Transfer Complete
KICK == 1?
point DMA at 4 byte address
in P-Mem, Set xfer type to
ADDR, and size to 4 bytes
NO
HW_GPFLASH_CSR1R_START_BYTE = 0
HW_GPFLASH_CSR0R_XFER_TYPE = 01
HW_GPFLASH_DMA_ADDR_DMA_ADD = #>addr
HW_GPFLASH_XFER_SIZE = 4
HW_GPFLASH_CSR0R_KICK = 1
YES
1 byte
Kick off 4 byte ADDR xfer
Wait for Transfer Complete
KICK == 1?
16-bit READ DATA
TRANSFER
NO
HW_GPFLASH_CSR0R_XFER_TYPE = 11
HW_GPFLASH_CSR0R_RW =1
HW_GPFLASH_CSR0R_MODE16 = 1
HW_GPFLASH_CSR0R_CEB_CTRL = 1
HW_GPFLASH_CSR1R_MEM = 0
HW_GPFLASH_CSR1R_START_BYTE = ?
HW_GPFLASH_DMA_ADDR_DMA_ADD = #>buffer
HW_GPFLASH_XFER_SIZE = n
HW_GPFLASH_CSR0R_KICK = 1
return chip enable high,
after transfer complete
point to read data buffer
transfer size
Kick off DATA read transfer
YES
Wait for Transfer Complete
KICK == 1?
NO
STOP
Figure 47. GP Flash Read Data Example Flowchart
124
5-35xx-D1-1.08-061705
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
Run GPFlash Init First
Uses xfer done interrutpt
GPFLASH_send_cmd
set up GPFLASH xferdone interrupt vector ($006E)
HW_GPFLASH_CSR0R_INT_XFERDONE_STATUS = 1
HW_GPFLASH_CSR0R_INT_XFERDONE_ENA = 1
Clear any previous IRQ
Enable XFER DONE IRQ
send cmd & addr in 8 bit mode
HW_GPFLASH_CSR1R_MODE16 =0
HW_GPFLASH_CSR1R_CEB = n
HW_GPFLASH_CSR1R_START_BYTE = ?
HW_GPFLASH_CSR1R_MEM = 10
select one device's chip enable
select starting byte alignment
get cmd/addr bytes from P-MEM
Write transfer for command byte
HW_GPFLASH_CSR0R_RW = 0
HW_GPFLASH_CSR0R_CEB_CTRL = 0
HW_GPFLASH_CSR0R_XFER_TYPE = 00
Leave chip enable low after cmd
Select COMMAND transfer
point to 1 byte cmd in on-chip RAM
HW_GPFLASH_DMA_ADDR_DMA_ADD = #>cmd_read
HW_GPFLASH_XFER_SIZE = 1
HW_GPFLASH_CSR0R_KICK = 1
STOP
1 byte
Kick off 1 byte CMD xfer
Return and wait for XFERDONE interrupt
Figure 48. GP Flash Send Command using XFERDONE Interrupt
11.3. Monitoring the NAND Flash Ready/Busy Signal
The GP FLASH module contains several circuits to aid in monitoring the relatively
long interval transitions of the NAND Flash device’s Ready/Busy signal. This
includes an interrupt that can monitor either the rising or falling edge of the busy signal and a TWTO timer that can be set to generate a timeout interrupt if the Flash
device hangs and never completes a block erase, etc.
The busy interrupt monitoring interrupt should only be scheduled after a command
that will report its availability via the R/B signal long after the command is started.
WARNING: not all NAND Flash devices have reliable busy signaling. For some
devices, only read status polling from a timer interrupt schedule is practical.
5-35xx-D1-1.08-061705
125
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
11.4. GP FLASH Programmable Registers
All GP Flash interface controller programmable registers exist in the DSP’s X memory space as shown below.
11.4.1.
General Purpose Flash Control Status Register 0
This register provides control of the General Purpose Flash Controller.
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
KICK
0
9
RW
1
0
INT_XFERDONE_ENA
1
1
INT_XFERDONE_STATUS
1
2
BUSY_INT_TYPE
1
3
BUSY_TIMEOUT_EDGE
1
4
BUSY_INT_ENA
1
5
BUSY_INT_STATUS
1
6
BUSY_TIMEOUT_INT_ENA
1
7
CEB_CTRL
1
8
SOFT_RESET
1
9
CLK_DISABLE
2
0
XFER_TYPE
2
1
SPARE
2
2
EXT_BUSY_VAL
2
3
BUSY_TIMEOUT_INT_STAT
HW_GPFLASH_CSR0R X:$F0C0
Table 125. HW_GPFLASH_CSR0R
BITS
LABEL
23:21 RSRVD
20
EXT_BUSY_VAL
19:18 SPARE
17:16 XFER_TYPE
15
14
RSRVD
BUSY_TIMEOUT_INT_STAT
13
CLK_DISABLE
RW
R
R
RW
RESET
DEFINITION
000
Reserved – Must be written with 0.
0
This read only bit samples the state of the external busy pin.
00
These bits, while present in the register, are currently
unconnected in the design and are reserved as patch gates.
RW 00
The GP FLASH breaks up the normal transaction flow to the
FLASH into one transaction for sending the command, one
transaction for sending any address bytes and one or more
commands for transferring read or write data. This field
determines which type of transaction is being “kicked off”.
00 = command
01 = address
10 = data
11 = reserved
R
0
Reserved – Must be written with 0.
RW 0
Reading a one indicates a pending busy completion TIMED
OUT instead of completing normally. This is a sticky bit which
is cleared by writing back a one in this bit position. Writing
zero has no effect. This bit is ANDed with
BUSY_TIMEOUT_IN_ENA to generate a
BUSY_TIMEOUT_IRQ. BUSY_TIMEOUT_IRQ is ORed with
BUSY_INT_IRQ to generate the GPFLASH busy interrupt on
source 6 and vector $0072.
RW 1
When this bit is set to one, the GP FLASH controller clocks
are gated off to reduce power.
Table 126. GP FLASH Command/Status Register 0 Description
126
5-35xx-D1-1.08-061705
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
12
CEB_CTRL
11
SOFT_RESET
10
BUSY_INT_STATUS
9
BUSY_TIMEOUT_INT_ENA
8
BUSY_INT_ENA
7:6
BUSY_TIMEOUT_EDGE
5:4
BUSY_INT_TYPE
3
INT_XFERDONE_STATUS
2
1
INT_XFERDONE_ENA
RW
0
KICK
RW RESET
DEFINITION
RW 1
Setting the chip enable control bit to one allows the chip
enable to return to the high state at the end of a transfer.
Setting this bit to zero will cause the chip enable to remain low
after a transfer. Keeping it low allows multiple DMA
transactions to look like a single contiguous transfer to the
NAND FLASH device.
RW 0
Writing a one to this bit position forces a reset to most of the
counters and state machines in the General Purpose Flash
Controller. This bit must be reset to zero by software for
normal operation of the GP FLASH. Programmable registers
are NOT returned to their default state by SOFT_RESET.
RW 0
Reading a one indicates a pending busy completion or busy
timeout interrupt has occurred. This is a sticky bit which is
cleared by writing back a one in this bit position. Writing zero
has no effect. This bit is ANDed with BUSY_IN_ENA to
generate a BUSY_IRQ. BUSY_IRQ is ORed with
BUSY_TIMEOUT_IRQ to generate the GP FLASH busy
interrupt on source 6 and vector $0072.
RW 0
Setting this bit to one enables the busy timeout state machine
and counter. If the trailing edge of busy is not detected then
BUSY_TIMEOUT_INT_STAT is set.
RW 0
Setting this bit to one enables a busy interrupt in
BUSY_INT_STATUS
RW 00
This field selects the edge behavior of the busy time out
interrupt state machine. The 13 bit internal timer counter is
reloaded from TXTO at the leading edge. The timer counts
down the width of the busy pulse. If the trailing edge is seen
then the timeout is suppressed. If the internal timer times out
then the BUSY_TIMEOUT_INT_STAT bit is set.
The leading/trailing edged definition depends on the polarity
of the busy signal.
00 = Reload timer on both edges
01 = Reload timer on rising edges, trailing is falling.
10 = Reload timer on falling edge, trailing is rising (NAND).
11 = Test mode
RW 00
This field selects the edge behavior of the busy interrupt, as
follows:
00 = trigger interrupt on both edges
01 = trigger interrupt on rising edge only
10 = trigger interrupt on falling edge only
11 = undefined behavior
RW 0
Reading a one indicates a pending transfer complete interrupt
has occurred. This is a sticky bit which is cleared by writing
back a one in this bit position. Writing zero has no effect.
RW 0
Setting this bit to one enables the transfer-complete interrupt.
RW 0
Setting this bit to one indicates that the external memory
transfer is a read; otherwise it is a write.
RW 0
Setting this bit to one initiates an external memory transfer; it
automatically clears when the transfer completes. It can be
polled by software to detect transfer completion or more
typically,
HW_GPFLASH_CSR0R_INT_XFERDONE_STATUS
interrupt can be used to detect this condition.
Table 126. GP FLASH Command/Status Register 0 Description (Continued)
5-35xx-D1-1.08-061705
127
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
11.4.2.
General Purpose Flash Control Status Register 1
This register provides control of the General Purpose Flash Controller.
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
MEM
1
6
START_BYTE
1
7
CEB
1
8
WP0
1
9
WP1
2
0
MODE16
2
1
MUX
2
2
SPARE_AREA_EN
2
3
WP_CTL
HW_GPFLASH_CSR1 X:$F0C1
Table 127. HW_GPFLASH_CSR1
BITS
LABEL
23:20 RSRVD
19
WP_CTL
18
SPARE_AREA_EN
17
MUX
16
MODE16
15:14 RSRVD
13
WP1
12
WP0
11:10 RSRVD
RW RESET
DEFINITION
R
0000
Reserved – Must be written with 0.
RW 0
This bit controls a multiplexor that selects the source for the SPI
MOSI output pin. In order to use the WP1 write protect signal,
this bit must be set to one. Setting it to zero returns control of the
MOSI driver to the SPI.
RW 1
The spare area enable pin is essentially a GPIO pin that is
intended to drive the SA input pin on NAND FLASH devices that
provide this option. It drives SDRAM Address pin 8 when the GP
Flash controls the external memory interface pins.
RW 0
GP Flash pins are shared with a number of other integrated
interface controllers. This bit controls a multiplexor that selects
between the GP Flash controller and the External Memory
Controller (EMC), as follows:
0 = HW_GPFLASH_MUX_EMC_ACCESS
1 = HW_GPFLASH_MUX_GPFLASH_ACCESS
RW 0
The GPFLASH controller supports either eight or sixteen bit
Flash devices. Set this pin to one for 16 bit devices.
0 = HW_GPFLASH_8BIT_ACCESS
1 = HW_GPFLASH_16BIT_ACCESS
R
00
Reserved – Must be written with 0.
0
The write protect 1 bit is essentially a GPIO pin that is intended to
connect to the write protect pin of a second NAND Flash device.
This pin is shared with the MOSI pin, see WP_CTL.
0
The write protect 0 bit is essentially a GPIO pin, it is intended to
connect to the write protect pin on the NAND Flash.
R
00
Reserved – Must be written with 0.
Table 128. GP FLASH Command/Status Register 1 Description
128
5-35xx-D1-1.08-061705
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
9:8
CEB
LABEL
7:6
5:4
RSRVD
START_BYTE
3:2
1:0
RSRVD
MEM
RW RESET
DEFINITION
RW 00
The interface allows access to four external NAND Flash
devices. The chip enable selector field is decoded to pick one of
the four external chip enable pins. Only the selected pin will be
driven to zero.
00 = CE0
01 = CE1
10 = CE2
11 = CE3
R
00
Reserved – Must be written with 0.
00
This field the byte alignment of the starting byte
00 = HW_GPFLASH_START_BYTE_LSB
01 = HW_GPFLASH_START_BYTE_ISB
10 = HW_GPFLASH_START_BYTE_MSB
11 = reserved, undefined behavior
R
00
Reserved – Must be written with 0.
00
This field defines the DSP memory space used for DMA transfers
as follows:
00 = X Memory HW_GPFLASH_USE_X_MEMORY
01 = Y Memory HW_GPFLASH_USE_Y_MEMORY
10 = P Memory HW_GPFLASH_USE_P_MEMORY
11 = reserved, undefined behavior
Table 128. GP FLASH Command/Status Register 1 Description (Continued)
11.4.3.
General Purpose Flash DMA Address Register
This register holds the DMA address used by the General Purpose Flash Controller
to access on-chip RAM.
HW_GPFLASH_DMA_ADDR X:$F0C2
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
DMA_ADD
2
3
Table 129. HW_GPFLASH_DMA_ADDR
BITS
LABEL
23:16 RSRVD
15:0 DMA_ADD
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
RW $0000 Begin the DMA transfer at this on-chip RAM address in the
address space specified in HW_GPFLASH_CSR0R_MEM.
Table 130. GP FLASH DMA Address Register Description
11.4.4.
General Purpose Flash DMA Transfer Size Register
This register holds the transfer count for DMA transactions kicked off for the General
Purpose Flash Controller.
5-35xx-D1-1.08-061705
129
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
HW_GPFLASH_XFER_SIZE X:$F0C3
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
NUM_BYTES
2
3
Table 131. HW_GPFLASH_XFER_SIZE
BITS
LABEL
23:13 RSRVD
12:0 NUM_BYTES
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
RW $001
This field contains the number of bytes to be transferred by the
GP Flash controller, when “kicked”.
Table 132. GP FLASH DMA Transfer Size Register Description
11.4.5.
General Purpose Flash Timing 1 Register
This register holds part of the timing values for specifying read or write cycles to or
from an external NAND Flash when using the General Purpose Flash Controller.
HW_GPFLASH_TIMING1 X:$F0C4
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
TSU
2
2
THOLD
2
3
Table 133. HW_GPFLASH_TIMING1
BITS
LABEL
23:13 RSRVD
12:8 THOLD
7:5
4:0
RSRVD
TSU
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
RW $01
After the last byte of command, address or data is transferred,
the CLE or ALE signals are returned to zero, if they were set. The
CLE/ALE hold time constrained is guaranteed by the GPFlash
state machines counting THOLD DCLKs before releasing the
ALE or CLE signals. THOLD can not be set to zero.
R
00
Reserved – Must be written with 0.
RW $00
Immediately after a transfer is kicked off, CSR0R is examined
and the CLE/ALE signals are set as required for the transfer. The
GPFlash then counts TSU DCLKs before allowing the first read
or write enable (REN/WEN) pulse.
Table 134. GP FLASH Timing 1 Register Description
130
5-35xx-D1-1.08-061705
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
11.4.6.
General Purpose Flash Timing 2 Register
This register holds part of the timing values for specifying read or write cycles to or
from an external NAND Flash when using the General Purpose Flash Controller.
The GPFlash state machines drive the write enable signal low at the same time they
drive new data onto the eight or sixteen bit data bus. Setup time for the flash device
is guaranteed by holding the write enable low for a sufficient number of DCLKs.
Flash device hold time is guaranteed after the write enable is returned to one by
keeping the data bus value stable for a sufficient number of DCLKs. These counts
are specified in the HW_GPFLASH_TIMING2R register.
In addition, setup time for read data arriving back to the STMP35xx and its necessary hold time is also controlled in this register, see Figure 42. “GP Flash Read Timing” on page 119.
HW_GPFLASH_TIMING2 X:$F0C5
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
TDS
2
2
TDH
2
3
Table 135. HW_GPFLASH_TIMING2
BITS
LABEL
23:14 RSRVD
13:8 TDH
7:6
5:0
RSRVD
TDS
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
RW 01
Number of DCLKS to hold off starting another write or read
enable low sequence. In the case of writes, this keeps the data
bus stable, as driven to the flash device. In the case of reads, this
keeps the device driving the same data to the STMP35xx until its
hold time is satisfied. TDH can not be set to a value of zero.
R
00
Reserved – Must be written with 0.
RW 01
Number of DCLKs to hold the write enable or read enable low.
Can not be set to zero.
Table 136. GP FLASH Timing 2 Register Description
11.4.7.
General Purpose Flash TIMING BUSY Register
This register holds part of the timing values for the General Purpose Flash Controller. Flash devices have a signal called ready/not busy that indicates when the
device is busy erasing a block of NAND Flash, when it is programming new values
into the NAND Flash array and in some cases, when it is reading the next page of
data from the NAND Flash array. There are two ways to determine when the flash
device is finished, i.e. no longer busy. One is to perform a device status read command. This is most reliable method across all flash devices. The second method is
to monitor the state of the ready/not-busy line. The flash devices have open drain
drivers with pull-ups on these signals so that not-BUSY can be wire ORed between
all the devices. After the device is commanded to erase a block, the not-BUSY signal will be pulled low. When the block erase is complete, the flash device releases
its driver and the signal is pulled up to indicate “READY” for the next operation. The
STMP35xx has an interrupt based monitor circuit for this ready/not-busy signal. The
not-busy signal is synchronized into the GP Flash interface controller and then both
edges are detected. The HW_GPFLASH_CSR0R_BUSY_INT_STATUS bit indi-
5-35xx-D1-1.08-061705
131
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
cates the arrival of an edge on the ready/not-busy signal. This bit is “sticky”, so once
it is set software must write a one to it to reset it.
The HW_GPFLASH_CSR0R_BUSY_INT_TYPE field tells the busy monitor circuit
which edge it should monitor. Thus software can choose to be interrupted when the
device first goes busy or when it leaves the busy state. Not all devices report busy in
robust manner, by design. In addition, broken devices may also fail to either go into
the busy state or return from it or to even mark these transitions on the ready/notbusy signal. As a result, the busy monitor circuit includes a time out counter that will
ultimately interrupt if the DSP on the BUSY interrupt, even if the device failed to do
so. The number of DCLKs to wait for the device to go from busy to not busy is specified in the TWTO field below. It tracks the time from the fall of the not-busy signal.
The field HW_GPFLASH_CSR0R_BUSY_TIMEOUT_EDGE can be used to start
timing on the opposite edge.
HW_GPFLASH_TIMINGBUSY X:$F0C6
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
TWTO
2
3
Table 137. HW_GPFLASH_TIMINGBUSY
BITS
LABEL
23:18 RSRVD
17:0 TWTO
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
RW $00001 Number of DCLKs to wait from the falling edge of ready/not-busy
until a time-out should be reported.
Table 138. GP FLASH Timing BUSY Register Description
132
5-35xx-D1-1.08-061705
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
11.4.8.
General Purpose Flash Debug Register
This register provides a read only path to view various state machines in the General Purpose Flash Controller.
HW_GPFLASH_DEBUG X:$F0C7
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
RD_WR_STATE
2
1
DMA_STATE
2
2
CMD_STATE
2
3
Table 139. HW_GPFLASH_DEBUG
BITS
23:11
10:8
7:4
3:0
LABEL
RSRVD
CMD_STATE
DMA_STATE
RD_WR_STATE
RW RESET
DEFINITION
R
$0000 Reserved – Must be written with 0.
R
Read only view of the Command State Machine
R
Read only view of the DMA State Machine
R
Read only view of the Read/Write State Machine.
Table 140. GP FLASH Debug Register Description
11.4.9.
General Purpose Flash Busy Timeout Counter Register
This register provides a read only path to view the current value in the Busy Timeout
Counter. This counter is preloaded from the HW_GPFLASH_TIMINGBUSYR register. It decrements to zero whenever HW_GPFLASH_CSR0R_BUSY_INT_ENA is
set to one and the leading edge of the busy pulse has been encountered. If the trailing edge of the busy pulse is detected then the counter is reloaded form
HW_GPFLASH_TIMINGBUSYR and no timeout will be reported.
HW_GPFLASH_TWTOCNT X:$F0C8
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
COUNTER
2
3
Table 141. HW_GPFLASH_TWTOCNT
BITS
LABEL
23:18 RSRVD
17:0 COUNTER
RW RESET
DEFINITION
R
$0000 Reserved – Must be written with 0.
R
$00001 Read only view of the Busy Timeout Counter
Table 142. GP FLASH Busy Timeout Counter Register Description
5-35xx-D1-1.08-061705
133
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
12. FLASH ECC ACCELERATOR
The FLASH ECC Accelerator provides a forward error correction function for
improving the reliability of various storage media that can be attached to the
STMP35xx. Modern high density NAND Flash devices, for example, presume the
existence of forward error correction algorithms. This allows much higher yield and
therefore lower cost storage devices by allowing some soft or hard bit errors within
the bits of a flash page. The FLASH ECC block comprises two forward error correction algorithms, one based on the Samsung SSFDC hamming code algorithm for
single bit error correction and a second, more robust, algorithm for multi-bit error
correction Reed-Solomon block codes, see Figure 49. Having a DMA based hardware accelerator for this function allows the DSP to focus on more signal processing
algorithms for enhanced functionality or to operate at lower clock frequencies and
voltages for improved battery life.
Y-BUS
X-BUS
P-BUS
DSP CORE
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
DMA-BUS
FLASH ECC
Programmable
Registers
FLASH ECC
DMA Engine
Reed-Solomon ECC Engine
Syndrome
Calculation
Key
Equation
Solver
SSFDC Hamming ECC Engine
1-bit
Hamming
Encoder
1-bit
Hamming
Decoder
Chien
Search
Forney Evaluator
Figure 49. FLASH ECC ACCELERATOR ENGINES
The 1-bit hamming ECC algorithm, as defined by Samsung, is for use with all
SSFDC compliant NAND Flash memories. The error coding is capable of correcting
a single bit in error over a 256 byte (2048 bit) data block. It generates a 3 byte parity
field which is stored in the spare area at the end of a flash page. The Reed-Solomon
algorithm is capable of correcting up to 4 9-bit symbols in a 512 byte block. Thus up
to 36 bits in error can be corrected in a 512 byte block, provided they are clustered
134
5-35xx-D1-1.08-061705
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
within no more than four 9 bit symbols. This algorithm generates 9 bytes (8 symbols) of error code or parity (sometimes called syndromes) per 512 byte block. The
parity bytes are stored in the spare area at the end of each NAND Flash page.
The Reed-Solomon algorithm takes a significantly larger number of cycles to correct
an error than the SSFDC Hamming algorithm but can correct more errors per page.
In either case, the DSP is not directly involved in checking for the errors nor is it
directly involved in correcting errors that may be found.
12.1. Reed-Solomon ECC Accelerator
Consider the case where there is a 512 byte data block located in the on-chip RAM
that needs to be written to a NAND Flash device. Further, assume that a 9 byte
Reed-Solomon parity field is to be written into the 16 byte spare area of the 528 byte
NAND Flash page.
528 byte NAND Flash Page
HW_ECC_BLKSTRTADDR_ADDR
HW_ECC_BLKSTRTINDEX_INDEX
16 Byte
Spare Area
512 Byte Data Area
DATA
455 1/9 Data Symbols (9-bit)
47 8/9 Zero Pad
Symbols
8 Parity
Symbols
503 Symbol Polynomial Dividend
511 Symbol Reed-Solomon Block Code
Galois Field
Polynomial Divider
HW_ECC_PARSTRTADDR_ADDR
HW_ECC_PARSTRINDEX_INDEX
512 Byte Data Area
DATA
8 Parity
Symbol
Remainder
16 Byte
Spare Area
9 Byte
Parity
Figure 50. FLASH ECC Reed-Solomon Block Coding- Encoder
5-35xx-D1-1.08-061705
135
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
Assume that the GPFLASH module will be used to write the resultant 521 bytes of
data and parity from on-chip memory to the NAND Flash device. The FLASH ECC
module HW_ECC_BLKSTRTADDR register is used to point to the data block in onchip RAM, see Figure 50. “FLASH ECC Reed-Solomon Block Coding- Encoder” on
page 135. The Reed-Solomon (RS) algorithm uses 9 bit symbols. Thus a 512 byte
data block encompasses 455 1/9 symbols. As the data is read from on-chip RAM,
the hardware appends 47 8/9 zero pad symbols to form the basic 511 symbol RS
block. This block is treated as a large polynomial and is divided by the hardware
using the mathematics of Galois Fields1. The hardware retains the 8 symbol (72 bits
or 9 bytes) remainder from this division which it then stores as the parity for the
block. The parity is stored into on-chip RAM at the address specified in
HW_ECC_PARSTRTADDR. The GPFLASH DMA can then be started to copy the
521 bytes to the NAND Flash device. Of course, both units can be fully overlapped.
Note that there is an HW_ECC_BLKSTRTINDEX register and an
HW_ECC_PARSTRTINDEX register which allows the data symbols and parity symbols to start on arbitrary bit boundaries within a 24 bit on-chip RAM word.
The RS encoder flowchart shows the detailed steps involved in programming and
using the FLASH ECC’s Reed-Solomon encoder, see Figure 51. “FLASH ECC
Reed-Solomon Encode Flowchart” on page 137. To use the encoder, one must first
turn off the module wide soft reset bit. One then programs the control status register
0 to select the Reed-Solomon encode algorithm, to remove the encoder reset, to
enable an interrupt to signal completion of the encode stage, and to specify the
DMA wait cycle count. Since the FLASH ECC is a memory to memory DMA
device, its DMA utilization is nominally limited only by the encoder’s demand for
data. This natural limit may utilize too much DMA bus bandwidth over its command
time. As a result, the HW_ECC_CSR0_DMAWAIT bit field can specify additional
wait cycles to insert between the DMA cycle requests to reduce the FLASH ECC’s
short term utilization.
The block start address/index values are next set to point to the data block in onchip RAM that is to be encoded. The parity start address/index values are set to
point to the on-chip RAM buffer that will hold the nine generated bytes until they are
written to Flash. Any previous Done Interrupt status is cleared by writing a one to the
interrupt bit and the KICK bit is set to one. Software can then poll the KICK bit waiting for it to return to zero, however, this typically takes hundreds of clock cycles. To
get full overlap of the DSP and the FLASH ECC module, one uses the “done” interrupt which vectors to $0062, source bit 29. When this interrupt is received, the
GPFLASH module can be scheduled to write the entire page to the NAND Flash
device.
1.Oliver Pretzel, “Error-Correction Codes and Finite Fields,” Oxford Univ. Press, 1992 ISBN 0-19-269067-1.
136
5-35xx-D1-1.08-061705
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
FLASH ECC
ENCODE
HW _ECC_CSR0_SFTRST = 0
HW _ECC_CSR0_ECCSEL = 2
HW _ECC_CSR0_DMAW AIT = n
HW _ECC_CSR0_DONEINTEN =1
HW _ECC_RSCFG_ENCSFTRST = 0
HW _ECC_RSCFG_BLOCKSIZE = 456
HW _ECC_RSCFG_BYTEN =
(((# data bytes)*8 % 9) !=0)
# wait states between
DMA transfers
NOT "even" bit boundary
Data block read from here
HW _ECC_BLKSTRTADDR_ADDR = word address of data block
HW _ECC_BLKSTRINDEX_INDEX = starting bit in word
parity results written here
HW _ECC_PARSTRTADDR_ADDR = word address of parity block
HW _ECC_PARSTRINDEX_INDEX = starting bit in word
force interrupt bit to zero
HW _ECC_CSR1_DONEINT = 1
HW _ECC_CSR0_KICK = 1
YES
HW _ECC_CSR0_
KICK==1
wait for Reed-Solomon Encode
or return and wait for done interrupt
NO
STOP
FLASH ECC
DONEINT
clear DONEINT
HW _ECC_CSR1_DONEINT = 1
STOP
Figure 51. FLASH ECC Reed-Solomon Encode Flowchart
5-35xx-D1-1.08-061705
137
OFF ICI AL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
When a page is read from NAND Flash, its RS parity must be checked and if correctable errors are found they must be corrected. This decoding process can also
be fully overlapped with DSP execution, see Figure 52. “FLASH ECC ReedSolomon Block Coding- Decoding Phase 1” on page 138. Notice that in this case,
both the block start/index registers and the parity start/index registers point into the
on-chip RAM buffers where bytes were read from the NAND Flash device. The
decoder can be initialized to read the data block, append the zero pad and perform
the polynomial division, this time with the supplied parity bytes. If the resulting division yields a zero remainder then no errors are present and the FLASH ECC block
can immediately report via its “done” interrupt. If the remainder was non-zero then it
further examines the syndrome bytes to correct any errors that may be present, if
possible.
528 byte NAND Flash Page
HW_ECC_BLKSTRTADDR_ADDR
HW_ECC_BLKSTRTINDEX_INDEX
HW_ECC_PARSTRTADDR_ADDR
HW_ECC_PARSTRINDEX_INDEX
16 Byte
Spare Area
512 Byte Data Area
9 Byte
Parity
DATA
455 1/9 Data Symbols (9-bit)
47 8/9 Zero Pad
Symbols
8 Parity
Symbols
503 Symbol Polynomial Dividend
511 Symbol Reed-Solomon Block Code
8 SYNDROMES
8 Parity
Symbol
Remainder
Galois Field
Polynomial Divider
OK
== 0?
Y
N
ERROR!
Figure 52. FLASH ECC Reed-Solomon Block Coding- Decoding Phase 1
The RS decoder processes the 511 symbol block code in three phases. Not all of
the phases may be necessary, for example when no errors are found or when
uncorrectable errors are found. The three phases are:
138
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
1. Syndrome Calculation Phase (SC): This is the process of reading in all of the
symbols of the block and continuously dividing by the generator polynomial for
the field. The eight syndromes are calculated as the remainder of this division
and must examined, as described above. This phase takes approximately 700
cycles for a 512 byte data block, with no planned DMA wait states added.
2. Key Equation Solver Phase (KES): Once the eight syndromes have been calculated, a set of eight linear equations in eight unknowns is formed. The process of solving these equations and selecting from the numerous possible
solutions constitutes the KES phase. The partial solution is obtained by dividing
a polynomial based on the syndromes by a Euclidian polynomial. This division,
again using the mathematics of Galois Fields, yields two polynomials, the Error
Evaluator (EE) polynomial and the Error Locator (EL) polynomial. The EE polynomial is the remainder of this division and is zero if an uncorrectable, i.e nonsolvable case exists. The hardware terminates with an uncorrectable error in
this case. This phase takes up to 560 DCLKs, with no planned DMA wait states
added.
3. Chien Search and Forney Evaluator Phase (EVAL): This phase takes the EEand EL polynomials from the KES phase and uses Chien’s algorithm for finding
the locations of the errors based on the EE-polynomial. The method basically
involves substituting all 512 9-bit symbols into the EE-polynomial. All non-zero
results of these substitutions represent the locations of the various errors.
Another GF division is performed at this point to determine the error value or the
correction to apply at the symbol in error location. This phase consumes
approximately 550 DCLKs, with no planned DMA wait states added. The EVAL
phase terminates either with an uncorrectable error interrupt or simply a “done”
interrupt. Done is reported in either case.
The flowchart shows that the initialization of the decoder is quite similar to that for
the encoder, see Figure 53. “FLASH ECC Reed-Solomon Decode Flowchart” on
page 140. Basic differences are setting ECCSEL to 3 for RS decode and turning off
the decoder soft reset instead of the encoder soft reset.
There are a few critical differences, however, first one also enables the uncorrectable error interrupt. In addition one sets the auto correct bit in the RS configuration
register. This setting is important because the FLASH ECC module state machines
have several stall states, notably at each phase boundary. If the autocorrect bit is
not set and errors are detected then the decoder will stall at the end of the SC
phase, set the ErrInt bit and wait for it to be cleared. A similar stall occurs at the end
of the KES phase. Finally, within the EVAL phase, it will stall as each error location is
computed so that the location can be verified. For normal operation, these stalls
are bypassed by setting the autocorrect bit.
Finally, one should note that the data buffer pointed to by the block start
address/index registers and the parity buffer pointed to by the parity start
address/index registers will be read for the purposes of detecting and locating errors
and will be written to correct up to four symbols in error. Note that correctable errors
in the parity storage will be modified as needed.
The kick operation and done interrupt ISR invocation work as in the encoder case.
Note that for the decoder, the ISR must monitor the uncorrectable error status bit to
determine that a good data buffer resulted from the decode operation. If uncorrectable errors occurred, it is up to software to determine how to deal with a bad block.
One strategy might be to re-read the data from NAND Flash in the hope that enough
soft errors will have been removed to make correction successful on a second pass.
5-35xx-D1-1.08-061705
139
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
FLASH EC C
D EC O D E
H W _EC C _C SR 0_SFT R ST = 0
H W _EC C _C SR 0_EC C SEL = 3
H W _EC C _C SR 0_D M AW AIT = n
H W _EC C _C SR 0_U N C O R R IN T EN = 1
H W _EC C _C SR 0_D O N EIN T EN =1
H W _EC C _C SR 0_AU T O C O R R = 1
H W _EC C _R SC FG _D EC SFT R ST = 0
H W _EC C _R SC FG _BLO C KSIZE =
456H W _EC C _R SC FG _BYT EN =
(((# data bytes)*8 % 9) ! =0)
# wait states between
D M A transfers
N O T "even" bit boundary
D ata block read from
here and possibly
corrected here
H W _EC C _BLKST R T AD D R _AD D R = word address of data block
H W _EC C _BLKST R IN D EX_IN D EX = starting bit in word
H W _EC C _PAR ST R T AD D R _AD D R = word address of parity block
parity bytes read
from here and
possibly corrected
here
H W _EC C _PAR ST R IN D EX_IN D EX = starting bit in word
FLASH EC C
D O N EIN T
H W _EC C _C SR 1_U N C O R R IN T = 1
H W _EC C _C SR 1_D O N EIN T = 1
force interrupt bits to zero
clear D O N EIN T
H W _EC C _C SR 0_KIC K = 1
H W _EC C _C SR 1
_D O N EIN T = 1
YES
H W _EC C _C SR 1_
U N C O R R IN T ==1
wait for R eed-Solom on D ecode
or return and wait for done or
uncorrectable error interrupt
H W _EC C _C SR 0_
KIC K==1
NO
ST O P
ST O P
clear U N C O R R IN T
H W _EC C _C SR 1_U N C O R R IN T = 1
D eal with
U ncorrectable Page
from Flash
ST O P
Figure 53. FLASH ECC Reed-Solomon Decode Flowchart
140
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
12.2. SSFDC Hamming ECC Accelerator
Consider the case where a 1Gbit large block NAND Flash device is attached to the
STMP35xx. These devices have pages with 2048 bytes of data and 64 bytes of
spare area. The Hamming code accelerator is designed to work on exactly 256 byte
blocks. Therefore, one must treat the NAND Flash page as eight 256 byte blocks,
computing eight separate three byte parity codes to cover the entire NAND Flash
page as shown in Figure 54. This figure shows the computation of one 256 byte
block parity out of the eight blocks needed.
2112 byte NAND Flash Page
HW_ECC_BLKSTRTADDR_ADDR
HW_ECC_BLKSTRTINDEX_INDEX
HW_ECC_PARSTRTADDR_ADDR
HW_ECC_PARSTRINDEX_INDEX
64 byte
spare
2048 Byte Data Area
256 byte
Data
256 byte
Data
256 byte
Data
256 byte
Data
3 Byte
Parity
256 Byte Data Block
22 bit parity Generator
3 Byte Parity
Figure 54. FLASH ECC SSFDC Encode Operation
As in the Reed-Solomon case, the HW_ECC_BLKSTRTADDR and
HW_ECC_BLKSTRTINDEX registers are set to point to the beginning of the
desired data block in on-chip RAM. The parity start address and index registers are
set to point to the three byte buffer in on-chip RAM that will receive the generated
parity bytes. When started, the DMA will fetch all 256 bytes from on-chip RAM and
accumulate the appropriate 22 bit Hamming code for this fixed block size. The DMA
will then write these bytes to the on-chip RAM.
Assume that the GPFLASH module will be used to write the results to the NAND
Flash device in an overlapped fashion. While several options exists, the best compromise between buffer size and organization is to combine two 256 byte blocks as
a unit and generate 6 bytes of parity. These can then be written to NAND Flash as a
512 byte data block followed immediately by a 16 byte spare area into which 6 bytes
of parity are copied. With this scheme good overlap is established between the
GPFLASH module and the ECC module while using only 2 X 518 byte buffers.
5-35xx-D1-1.08-061705
141
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
The encoder flow chart for the SSFDC looks similar to and behaves like the ReedSolomon encoder. The differences are that the ECCSEL is set to zero for SSFDC
encode and the SSFDC soft reset bit is turned off in the HW_ECC_SSFDCCFG register.
FLASH ECC
SSFDC ENCODE
HW_ECC_CSR0_SFTRST = 0
HW_ECC_CSR0_ECCSEL = 0
HW_ECC_CSR0_DMAWAIT = n
HW_ECC_CSR0_DONEINTEN =1
# wait states between
DMA transfers
HW_ECC_SSFDCCFG_SFTRST=0
Data block read from here
HW_ECC_BLKSTRTADDR_ADDR = word address of data block
HW_ECC_BLKSTRINDEX_INDEX = starting bit in word
parity results written here
HW_ECC_PARSTRTADDR_ADDR = word address of parity block
HW_ECC_PARSTRINDEX_INDEX = starting bit in word
HW_ECC_CSR1_DONEINT = 1
force interrupt bit to zero
HW_ECC_CSR0_KICK = 1
YES
HW_ECC_CSR0_
KICK==1
wait for SSFDC Hamming Encode
or return and wait for done interrupt
NO
STOP
FLASH ECC
DONEINT
clear DONEINT
HW_ECC_CSR1_DONEINT = 1
STOP
start next block or
start GPFLASH
Figure 55. FLASH ECC SSFDC Encode Flowchart
The ECC module knows that it will write three bytes of parity instead of nine. The
other difference is that software will set up the block start address and parity start
address for 256 byte buffers instead of 512 byte as in the Reed-Solomon case. The
ECC module takes 266 dclks to encode a 256 byte block.
When a page is read from NAND Flash, its SSFDC hamming code parity must also
be read from the NAND Flash device. Each 256 byte block must be checked for
errors and if a correctable error is found it must be corrected, either in the 256 byte
142
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
data portion or the 3 byte parity portion. This decoding process can also be fully
overlapped with the NAND Flash transfers, as shown in Figure 56.
2112 byte NAND Flash Page
HW_ECC_BLKSTRTADDR_ADDR
HW_ECC_BLKSTRTINDEX_INDEX
HW_ECC_PARSTRTADDR_ADDR
HW_ECC_PARSTRINDEX_INDEX
64 byte
spare
2048 Byte Data Area
256 byte
Data
256 byte
Data
256 byte
Data
256 Byte Data Block
256 byte
Data
3 Byte
Parity
3 Byte
Parity
22 bit parity Generate &
single bit error correct
Figure 56. FLASH ECC SSFDC Decode Operation
Notice that in this case, the block start and the parity start address/index registers
are set to point into the on-chip RAM buffers which received the bytes from the
NAND Flash device. The decoder reads the data bytes and computes a new 22 bit
parity value. It then reads the stored parity bytes and computes an error syndrome
value by XORing the two parity values.
If all 22 bits of this error syndrome are zero then no errors are present. Otherwise
the bits are examined to determine whether a single bit is in error. If not then an
uncorrectable error interrupt is generated and further processing is terminated. For
a single bit error, if the auto correct bit is set then the bit is corrected and a done
interrupt is generated. The decoder takes approximately 266 DCLKs to complete,
with or without error correction, provided no DMA WAIT cycles were added.
The SSFDC decoder flow chart of Figure 57 is very similar to the Reed-Solomon
decoder flow chart. The major difference is setting the ECC SEL field to one for
SSFDC decode. In addition, the SSFDC configuration register soft reset bit is reset.
5-35xx-D1-1.08-061705
143
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
FLASH ECC
DECODE
HW_ECC_CSR0_SFTRST = 0
HW_ECC_CSR0_ECCSEL = 1
HW_ECC_CSR0_DMAWAIT = n
HW_ECC_CSR0_UNCORRINTEN = 1
HW_ECC_CSR0_DONEINTEN =1
HW_ECC_CSR0_AUTOCORR = 1
# wait states between
DMA transfers
Data block read from
here and possibly
corrected here
HW_ECC_SSFDCCFG_SFTRST=0
HW_ECC_BLKSTRTADDR_ADDR = word address of data block
HW_ECC_BLKSTRINDEX_INDEX = starting bit in word
HW_ECC_PARSTRTADDR_ADDR = word address of parity block
parity bytes read from
here and possibly
corrected here
HW_ECC_PARSTRINDEX_INDEX = starting bit in word
FLASH ECC
DONEINT
HW_ECC_CSR1_UNCORRINT = 1
HW_ECC_CSR1_DONEINT = 1
force interrupt bits to zero
clear DONEINT
HW_ECC_CSR0_KICK = 1
HW_ECC_CSR1
_DONEINT = 1
YES
HW_ECC_CSR1_
UNCORRINT==1
wait for SSFDC Hamming Decode
or return and wait for done or
uncorrectable error interrupt
HW_ECC_CSR0_
KICK==1
YES
NO
NO
STOP
STOP
clear UNCORRINT
HW_ECC_CSR1_UNCORRINT = 1
STOP
Deal with
Uncorrectable Page
from Flash
Figure 57. FLASH ECC SSFDC Decode Flowchart
144
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
12.3. FLASH ECC Programmable Registers
The following registers are available for DSP programmer access and control of the
FLASH ECC accelerator.
12.3.1.
Flash ECC Control/Status Register 0
This register provides overall control of the FLASH ECC accelerator for transaction
and interrupts.
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ECCKICK
1
6
AUTOCORR
1
7
DONEINTEN
1
8
KESINTEN
1
9
CORRINTEN
2
0
UNCORRINTEN
2
1
ERRINTEN
2
2
ECCSEL
SFTRST
2
3
X:$F780
DMAWAIT
HW_ECC_CSR0
Table 143. HW_ECC_CSR0
BITS
LABEL
23
SFTRST
22:18 RSRVD
17:16 ECCSEL
15:12 DMAWAIT
11:9
8
RSRVD
ERRINTEN
7
UNCORRINTEN
6
CORRINTEN
5
KESINTEN
4
DONEINTEN
3:2
1
0
RSRVD
AUTOCORR
ECCKICK
RW RESET
DEFINITION
RW 1
Set to one for software reset and low-power mode. Set to zero for
normal operation.
R
$0
Reserved – Must be written with 0.
RW 00
Select and enable desired ECC mode;
00 = HW_ECC_SSFDC_ENCODE
01 = HW_ECC_SSFDC_DECODE
10 = HW_ECC_RS_ENCODE
11 = HW_ECC_RS_DECODE
RW 0000
DMA Wait States - Specifies the number of cycles the ECC DMA
will wait between consecutive memory cycle requests.
R
$0
Reserved – Must be written with 0.
RW 0
Error Interrupt Enable - Set to one to enable the ECC to generate
an interrupt when errors are detected in the current codeword.
RW 0
Set the uncorrectable interrupt enable bit to one to enable the
ECC to generate an interrupt when an uncorrectable error
situation is found and that further processing is not possible.
RW 0
Set the correctable interrupt enable bit to one to enable the
generation of interrupts when a set of correctable errors is found.
The interrupt (and stall if AUTOCORR is not set) is generated
before initiating the corrections.
RW 0
Set the key equation interrupt enable bit to one to enable the
generation of an interrupt once KES processing is complete.
RW 0
Set the done interrupt enable bit to one to enable the generation
of interrupts when processing is complete.
R
$0
Reserved – Must be written with 0.
RW 0
Allow the ECC to automatically correct any correctable errors.
RW 0
Set to one to start processing the currently configured block. This
bit will automatically reset to zero when processing is complete.
Table 144. FLASH ECC Control/Status Register 0 Description
5-35xx-D1-1.08-061705
145
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
12.3.2.
Flash ECC RS Configuration Register
This register provides configuration control for the various phases of the ReedSolomon encoder/decoder.
2
1
2
0
DECSFTRST
RSCFGERR
BYTEEN
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
SYMBOLSIZE
2
2
X:$F782
BLOCKSIZE
2
3
ENCSFTRST
HW_ECC_RSCFG
Table 145. HW_ECC_RSCFG
BITS
LABEL
23
ENCSFTRST
22
DECSFTRST
21
RSCFGERR
20
BYTEEN
19:13 RSRVD
12:4 BLOCKSIZE
3:0
SYMBOLSIZE
RW RESET
DEFINITION
RW 1
Software reset and low-power mode for the RS-Encoder. Must be
set to zero for normal operation of the encoder.
RW 1
Software reset and low-power mode for the RS-Decoder. Must be
set to zero for normal operation of the decoder.
R
0
When the RS configuration error bit reads back a value of one, it
indicates that an invalid combination of block size and symbol
size has been written to the BLOCKSIZE field.
RW 1
The RS encode/decode engine works on 9-bit symbols. The
DMA behavior must be different if the last symbol perfectly aligns
with the last word boundary. Let N equal the number of data
bytes to transfer, if ((N*8)%9) is not zero then set BYTEEN to
one.
R
0
Reserved – Must be written with 0.
RW $1C8
The RS algorithm always implements a block code of 511 RS
symbols, however, the programmer can choose to compute a 9
byte parity value for a source block smaller than 512 bytes. If 256
bytes are chosen, for example, then set this field to 228. The
hardware will simply increase the number of zero pad bytes as
appropriate.
R
$9
This read only value indicates the symbol size for which the
hardware was designed.
Table 146. FLASH RS Configuration Register Description
146
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
12.3.3.
Flash ECC Control/Status Register 1
This register provides a continuation of control bits for the FLASH ECC module.
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
DONEINT
1
8
KESINT
1
9
CORRINT
2
0
UNCORRINT
2
1
ERRINT
2
2
ECCERROR
2
3
X:$F781
ECCEXCEPTION
HW_ECC_CSR1
0
3
0
2
0
1
0
0
Table 147. HW_ECC_CSR1
BITS
LABEL
23:19 RSRVD
18:16 ECCERRORS
15:12 ECCEXCEPTION
11:9
8
RSRVD
ERRINT
7
UNCORRINT
6
CORRINT
5
KESINT
4
DONEINT
3:0
RSRVD
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
R
$0
Number of errors detected/corrected. See ECCEXCEPTION field
for details on uncorrectable errors.
R
0000
Reason for termination of algorithm or uncorrectable error:
0: no exceptions
1: RS --> degree of “lambda” exceeds 4
2: RS --> “lambda” == all zeroes
4: RS --> degree of “lambda”!= number of roots of “lambda”, i.e.
duplicate roots indicated.
8: SSFDC --> More than 1 Error.
R
0
Reserved – Must be written with 0.
RW 0
When set to one, the error interrupt status bit indicates that the
current block contains errors and further processing is needed to
determine correctability. This indicates that the syndrome
computation is complete and the result is non-zero, i.e. indicating
errors are present. Write a one to this bit to clear it.
RW 0
When set to one, the uncorrectable error interrupt status bit
indicates that uncorrectable errors have been detected in the
most recently processed block. If this interrupt is enabled,
processing will stop until this bit is cleared. Write a one to this bit
to clear it.
RW 0
When set to one, the Correctable errors interrupt status bit
indicates that a correctable error has been detected. Processing
will stop when this bit is set until it is cleared. It will become valid
each time the Chien Search module finds an error location. This
can happen no more than 4 times per block. Write a one to this
bit to clear it.
RW 0
When set to one, the Key Equation Solver interrupt status bit
indicates that the KES module has completed processing and a
solution is available. Write a one to this bit to clear it.
RW 0
When set to one, the done interrupt status bit indicates that the
current block has been processed and all errors are identified
and corrected. Write a one to this bit clear it.
R
0
Reserved – Must be written with 0.
Table 148. FLASH ECC Control/Status Register 1 Description
5-35xx-D1-1.08-061705
147
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
12.3.4.
Flash ECC SSFDC ECC Configuration Register
This register provides configuration control for the SSFDC encoder and decoder.
HW_ECC_SSFDCCFG X:$F783
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
SYM;BOLSIZE
2
1
BLOCKSIZE
2
2
SFTRST
2
3
Table 149. HW_ECC_SSFDCCFG
BITS
LABEL
23
SFTRST
RW RESET
DEFINITION
RW 1
Software reset and low-power mode for SSFDC. Common reset
for both encode and decode functions.
R
0
Reserved – Must be written with 0.
R
$100
Only block size 256 ($100) is supported.
R
$8
Only symbols size 8 is supported.
22:13 RSRVD
12:4 BLOCKSIZE
3:0
SYMBOLSIZE
Table 150. FLASH ECC SSFDC Configuration Register Description
12.3.5.
Flash ECC Block Start Address Register
This register provides a pointer into on-chip RAM to the first byte of the data block.
HW_ECC_BLKSTRTADDR X:$F784
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ADDR
2
3
Table 151. HW_ECC_BLKSTRTADDR
•
BITS
LABEL
23:16 RSRVD
15:0 ADDR
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
RW $0000 Start address for the data block.
Table 152. FLASH Block Start Address Register Description
148
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
12.3.6.
Flash ECC Block Start Index Register
This register provides the starting bit offset into the first 24 bit word pointed to by
HW_ECC_BLKSTRTADDR. This register also specifies the memory space that
contains the data block.
HW_ECC_BLKSTRTINDEX X:$F785
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
INDEX
MEMSPACE
2
3
Table 153. HW_ECC_BLKSTRTINDEX
BITS
LABEL
23:10 RSRVD
9:8
MEMSPACE
7:5
4:0
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
RW 00
Memory space in which the data block is stored.
00 = X-Memory
01 = Y-Memory
10 = P-Memory
R
0
Reserved – Must be written with 0.
RW 00000 Bit index into the 24-bit memory word where the 1st symbol
occurs in the data block.
RSRVD
INDEX
Table 154. FLASH ECC Block Start Index Register Description
12.3.7.
Flash ECC Parity Start Address Register
This register provides a pointer into on-chip RAM for the buffer containing the parity
bytes.
HW_ECC_PARSTRTADDR X:$F786
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ADDR
2
3
Table 155. HW_ECC_PARSTRTADDR
BITS
LABEL
23:16 RSRVD
15:0 ADDR
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
RW $000
Start address for the parity storage buffer.
Table 156. FLASH ECC Parity Start Address Register Description
12.3.8.
5-35xx-D1-1.08-061705
Flash ECC Parity Start Index Register
149
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
This register provides the starting bit offset into the first 24 bit word pointed to by
HW_ECC_PARSTRTADDR. This register also specifies the memory space that
contains the parity storage block.
HW_ECC_PARSTRTINDEX X:$F787
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
0
1
0
0
INDEX
MEMSPACE
2
3
Table 157. HW_ECC_PARSTRTINDEX
BITS
LABEL
23:10 RSRVD
9:8
MEMSPACE
7:5
4:0
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
RW 00
Memory space in which the parity is stored.
00 = X-Memory
01 = Y-Memory
10 = P-Memory
R
0
Reserved – Must be written with 0.
RW 00000 Bit index into the 24-bit memory word where the 1st symbol
occurs in the data block.
RSRVD
INDEX
Table 158. FLASH ECC Parity Start Index Register Description
12.3.9.
Flash ECC Error Location Address Register
This register provides a read only view of the location of an error.
HW_ECC_LOCADDR X:$F788
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
ADDR
2
3
Table 159. HW_ECC_LOCADDR
BITS
LABEL
23:16 RSRVD
15:0 ADDR
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
R
$0000 Word address of the error. This field is only valid while the Corrint
status bit is set.
Table 160. FLASH ECC Error Location Address Register Description
150
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
12.3.10. Flash ECC Error Location Index Register
This register provides a read only view of the memory space indicator and bit index
for the error that is currently being reported.
HW_ECC_LOCINDEX X:$F789
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
INDEX
2
2
MEMSPACE
2
3
Table 161. HW_ECC_LOCINDEX
BITS
LABEL
23:10 RSRVD
9:8
MEMSPACE
7:5
4:0
RSRVD
INDEX
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
R
00
Memory space for the error. This field is only valid while the
Corrint status bit is set.
00 = X-Memory
01 = Y-Memory
10 = P-Memory
R
0
Reserved – Must be written with 0.
R
0000
Contains the bit index into the 24-bit memory word where the 1st
bit of the erroneous symbol occurs. This can either be a 9-bit RS
symbol or a 1-bit SSFDC symbol. This field is only valid while the
Corrint status bit is set.
Table 162. FLASH ECC Error Location Index Register Description
5-35xx-D1-1.08-061705
151
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
12.3.11. Flash ECC Error Value Register
This register provides a read only view of the ECC error value to be used to correct
a symbol in error.
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
BLOCKLOC
2
3
X:$F78A
0
5
0
4
0
3
0
2
0
1
0
0
ERRORVALUE
HW_ECC_ERRVAL
Table 163. HW_ECC_ERRVAL
BITS
LABEL
23:21 RSRVD
20:12 BLOCKLOC
11:9
8:0
RSRVD
ERRORVALUE
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
R
$000
Block location of an Error Value. Provides the location within the
data block or parity block where the error is located. To be used
in of Search/Replace function of Chien search module. This field
is only valid when the Corrint field is set.
R
0
Reserved – Must be written with 0.
R
$000
The Error Value is the 9-bit value to be XORed with the existing
data at the Error Location. For SSFDC data, only the LSB should
be used to correct the single-bit error. This field is only valid when
the Corrint field is set.
Table 164. FLASH ECC Error Value Register Description
12.4. Known Chip Defects with ECC block
12.4.1.
Clear Quest Entry STMP00004063
All revisions of the chip are known to incorrectly handle the erased block case.
The software ECC code for an erased block of Flash memory doesn't match the
code that the HWECC block calculates for the same block. In the C program there is
a final inversion of the ECC code that is not being done in our HW engine.
There is a software workaround proposed by CV as follows:
The 3500 hardware SSFDC ECC needs a couple of extra operations to get it to
match the standard software SSFDC ECC algorithm:
1) Do the 3500 hardware ECC generation
2)
Clear ecc bits 16 and 17.
3)
Invert all 24 ecc bits.
To do the 3500 SSFDC ECC correction requires the following:
152
1)
Invert all 24 ecc bits
2)
Set ecc bits 16 and 17
3)
Do the 3500 hardware ECC correction
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
4)
Clear ecc bits 16 and 17
5)
Invert all 24 ecc bits
Users might not need to do steps 2 and 4 as bits 16 and 17 are constant and are not
used to do error correction.
5-35xx-D1-1.08-061705
153
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
13. FILTER COPROCESSOR (FILCO)
Previous generations of the D-MajorTM family used a significant amount of processor time to perform basic FIR filtering. The DAC output flow requires a 2:1 interpolating FIR filter on the stereo pairs to upsample the audio, see Section 24. “DAC” on
page 276. The ADC flow requires an 8:1 decimating FIR filter on the stereo pairs to
downsample the audio stream and to derive the appropriate quantization, see Section 25. “ADC” on page 286. In addition, a multi-band equalizer utilizes a number of
bi-quadratic IIR filters to adjust the various bands.
All of these activities made extensive use of the DSP’s multiply accumulate features. The STMP35xx filter coprocessor (FILCO) is used to off load significant
amount of very standard signal processing operations from the DSP. The reduction
in DSP overhead yields more available MIPS for additional software. This reduction
in DSP workload can also be used to lower the DCLK and in some cases reduce
operating voltages to greatly extend battery life.
Y-BUS
X-BUS
P-BUS
DSP CORE
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
DMA-BUS
FILCO
Programmable
Registers
FILCO
DMA Engine & Control
Accumulators
(8x56)
+
Coefficient
RegFile
(8x24)
56-bit
24x24
MAC
Sample
RegFile
(6x24)
Figure 58. Filter Coprocessor (FILCO)
Figure 58 shows the high level block diagram for the FILCO. The FILCO is essentially a MAC engine that is directly fed from a DMA engine. To perform a simple 8tap FIR, for example, the DMA is pointed at a 44.1KHz stereo sample buffer to be
FIR low pass filtered. These samples are fetched via DMA cycles into the 24 bit
wide Sample Register File. The eight coefficients are fetched into the coefficient register file, as needed. The accumulator register file holds the intermediate summa154
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
tions for the multiply accumulates that are occurring for both the left channel output
samples and the right channel output samples, as shown in Figure 59, below.
HW_FILCO_WCNT
HW_FILCO_CBAR
HW_FILCO_SBAR
HW_FILCO_TAPCNT
HW_FILCO_SMR
Coeff0
Coeff1
Coeff2
Coeff3
Coeff4
Coeff5
Coeff6
Coeff7
# Coeff
HW_FILCO_DBAR
HW_FILCO_DMR
# samples
in
buffer
ACCUMULATORS
MAC
SAMPLE REGFILE
COEFF. REGFILE
DMA
Left0
Right0
Left1
Right1
Left2
Right2
Left3
Right3
Left4
Right4
Left5
Right5
Left6
Right6
Left7
Right7
Left8
Right8
Left9
Right9
Left10
Right10
Left11
Right11
Left0
Right0
Left1
Right1
Left2
Right2
Left3
Right3
Left4
Right4
Left5
Right5
Left6
Right6
Left7
Right7
Left8
Right8
Left9
Right9
Left10
Right10
Left11
Right11
# Out
Samples
in
Buffer
# Out
Samples
to
Process
Figure 59. FILCO FIR EXAMPLE
In setting up this simple example, we set HW_FILCO_DBAR to point to the stereo
input buffer and set the HW_FILCO_DMR modulo register to the number of words
(not word pairs) in the input buffer. We set the HW_FILCO_CBAR base address
register to point the buffer containing the eight coefficients to be used for this filter
and we set HW_FILCO_TAPCNT tap count register to a value of eight. Next we set
the HW_FILCO_SBAR output sample base address register to point to the circular
buffer to receive the filtered output. We set the HW_FILCO_SMR sample modulo
register to the number of words in the output buffer. Next we set the
HW_FILCO_WCNT word count register to the number of output sample words to
produce for the next DMA KICK. Notice that we can schedule much fewer words
than the entire output circular buffer for each DMA KICK, (technically we could also
schedule more). Once these basics are completed along with other more detailed
5-35xx-D1-1.08-061705
155
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
control register settings we can KICK the DMA and cause the FIR output samples to
be computed without further DSP involvement. As the FILCO processes this operation, it reads a set of 8 left channel inputs and multiply accumulates them with their
corresponding coefficients to produce the first left output sample. It then reads eight
right channel samples and performs the coefficient MACs to produce the first right
output sample. This process goes on until all the desired output samples have been
generated at which time the KICK bit is reset and an interrupt is generated to the
DSP. See the discussion below which describes the FILCOS use of all six sample
registers, all eight coefficient registers and all eight accumulator registers to make
this process much more parallel and efficient. In essence, the FILCO can sustain a
rate of one MAC per DCLK indefinitely without saturating the DMA bus (it uses
approximately 66% of the bus bandwidth at full speed). DSP applications rarely
exceed 25% MAC utilizations over long time periods. Thus a significantly more powerful signal processing capability is available for a few specific algorithms, namely
FIR and bi-quadratic IIR stereo filters.
13.1. FIR FILTER MODE
Typical DSPs are Harvard architectures with multiple memory banks, e.g. the 56K
DSP has a P memory for instructions and X and Y memory for simultaneous access
to coefficients and samples. For normal multiply accumulate instructions, the DSP
simultaneously reads the next sample and coefficient values from X and Y memories. It then performs a 24 bit by 24 bit multiply and adds the result to a 56 bit accumulator. This allows the 56K DSP to execute a MAC every clock. For the specific
algorithms supported by the FILCO, very simple but effective data reuse strategies
exist. Consider the basic equation of an FIR filter as it would be executed for one
FILCO KICK as shown here.
∀j, ( j ∈ { 0… ( S – 1 ) } ), Accumulator j=
T–1
∑ ai + j × ci (for left channel)
i=0
∀j, ( j ∈ { 0… ( S – 1 ) } ), Accumulator j=
T–1
∑ bi + j × ci (for right channel)
i=0
In these equations, T is the number of taps and S is the number of output samples
to be computed for a single FILCO KICK. If one examines the partial computation in
the middle of processing a pair of FIRs, one for each stereo channel, one sees the
computations depicted in Figure 60. Thus the opportunity exists to compute four
separate output sample results for the left channel and another four separate output
sample results for the right channel. Each of these partial results is accumulated in
one of the eight accumulators in the accumulator register file. From Figure 60, we
see that each left input sample is fetched once and used four times, each right input
sample is fetched once and used four times and each coefficient is used 8 times for
each fetch, for a sustained utilization of 1/4 + 1/4 + 1/8 or 63%. It takes 24 multiplies
to compute the partial filter of figure 60 during this time 12 coefficient and samples
are fetched so that the utilization ends up being 12/24 or approximately 50% of the
DMA bus. This is a steady state number for long filters; for startup and rundown, the
FILCO will use closer to 100% of the DMA bus utilization for short periods. The
HW_FILCO_CSR_DMADLY value can be used to throttle the DMA utilization of the
156
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
FILCO. This field indicates the number of additional DCLKs to wait before asserting
a DMA request to the on-chip RAM.
Because the FILCO is running eight separate FIR filter computations at once, there
is a four deep pipeline running for each channel. As a result, three words of zero
must be appended to the end of the coefficient array for FIR filters (plus three more
for FIR decimation mode). These zero words pad out the run down of the four deep
pipeline.,
Accumulator0 += an * cm
+ an+1 * cm+1 + an+2 * cm+2
Accumulator1 += an * cm+1 + an+1 * cm+2 + an+2 * cm+3
Accumulator2 += an * cm+2 + an+1 * cm+3 + an+2 * cm+4
Accumulator3 += an * cm+3 + an+1 * cm+4 + an+2 * cm+5
Accumulator4 += bn * cm
+ bn+1 * cm+1 + bn+2 * cm+2
Accumulator5 += bn * cm+1 + bn+1 * cm+2 + bn+2 * cm+3
Accumulator6 += bn * cm+2 + bn+1 * cm+3 + bn+2 * cm+4
Accumulator7 += bn * cm+3 + bn+1 * cm+4 + bn+2 * cm+5
Figure 60. FILCO FIR Mode DMA Fetch Re-use
The FILCO performs 24-bit by 24-bit, parallel, twos-complement fractional multiples.
The “fixed point” for the fractional number occurs between the sign bit and the first
data bit, i.e. between bit 23 and bit 22 with the sign bit located in bit 23. The result of
the multiply is a 47-bit twos-complement fractional value to which a zero is
appended on the right to make a 48-bit number before accumulation. The most negative number that can be represented in 24 bit fractional representation is -1.0 while
the most positive is approximately +0.999998 which is ((2^23)-1)/(2^23), i.e. just
less than +1.0.
Multiplying two numbers whose value is -1.0 should yield a value of +1.0 but this
value is not representable in 24-bit twos-complement fractional numbers. An overflow occurs in this case and the result is -1.0. This overflow is neither detected nor
corrected in the FILCO. Users should avoid FIR filters whose coefficient sets contain
the number -1.0. If this coefficient is used in an FIR filter, then the user should insure
that no corresponding sample with a value of -1.0 will be seen.
When the output sample is extracted from the accumulator, the DMA pulls bits 47
down through bit 24 to write to on-chip RAM as the 24 bit FIR result. For a fractional
number representation, this has the effect of discarding precision without changing
the magnitude. For example, a 48 bit twos-complement fractional number with a
value of +0.5 becomes a 24-bit twos-complement fractional number with a value of
+0.5. The accumulator value is saturated so that the 24 bit result has the most positive number substituted for any value that is larger or the most negative number is
substituted for any value that is smaller. Otherwise convergent rounding is used on
the accumulator value to produce the 24-bit result.
The HW_FILCO_SAT_SAT bit-field is an 8-bit field with a sticky bit for each of the
eight accumulators. Anytime a saturation correction is applied to a specific accumulator value to store it to on-chip RAM during an FIR operation, then its correspond-
5-35xx-D1-1.08-061705
157
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
ing SAT bit is set. Once software has examined a saturation status bit, it can be
cleared by writing a one to it.
The FILCO module includes a completion interrupt to notify software that a kicked
off operation is complete. To facilitate context switching, a currently running filter can
be “unkicked” so that it will quiesce to a safe state even before its word count
reaches zero.
Consider a simple FIR with 10 taps whose data reference patterns can be seen in
Figure 61, below. For this FIR, all ten coefficients are accessed from the coefficient
register file, sequentially one after the other. For the “mth” output sample, whose
basis is shown starting at left sample Left0, all of the left samples from Left0 through
Left 9, inclusive, are read and multiplied by their corresponding coefficient.
HW_FILCO_CBAR
smaller address
larger address
Coeff0
Coeff1
Coeff2
Coeff3
Coeff4
Coeff5
Coeff6
Coeff7
Coeff8
Coeff9
zero pad
zero pad
zero pad
newer
older
pad with three
words of zero
HW_FILCO_DBAR
smaller address
older
basis for left output sample m
samplen+1
samplen+2
samplen+3
samplen+4
samplen+9
larger address
Left0
Right0
Left1
Right1
Left2
Right2
Left3
Right3
Left4
Right4
Left5
Right5
Left6
Right6
Left7
Right7
Left8
Right8
Left9
Right9
Left10
Right10
Left11
Right11
samplen
samplen+1
samplen+2
basis for left output sample m+1
samplen
samplen+9
newer
HW_FILCO_CSR_FIRMODE = 00
Figure 61. FILCO Normal FIR Mode
For output sample m+1, the basis begins at the next left sample after Left0 or Left1.
So the coefficient vector is multiplied by a basis vector that slides along the sample
158
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
stream, moving one sample down the input sample stream for each filter output
sample.
This
sample
reference
pattern
results
when
the
HW_FILCO_CSR_FIRMODE bit field is set to a value of 00, for the 1:1 or normal
FIR mode
13.1.1.
FIR Decimation Filter
When the HW_FILCO_CSR_FIRMODE bit field is set to 10 then it enables the 2:1
decimation mode. In this mode, FIR filter calculations for each output sample are
made, as in the normal case, except that the filter basis slides down the input sample buffer differently, see Figure 62. “FILCO FIR Decimation Mode” on page 159. In
this mode, after sample m has been computed with a basis that began at sample
Left0, the sample m+1 basis is fetched beginning at sample Left2. Notice that sample Left1 was “skipped” and never used as the first sample of any basis. Thus the
2:1 decimation mode produces one output sample for every two input samples.
smaller address
larger address
smaller address
samplen
basis for left output sample m
samplen+1
samplen+2
samplen+3
samplen+4
samplen+10
larger address
Coeff0
Coeff1
Coeff2
Coeff3
Coeff4
Coeff5
Coeff6
Coeff7
Coeff8
Coeff9
Coeff10
zero pad
zero pad
zero pad
zero pad
zero pad
zero pad
Left0
Right0
Left1
Right1
Left2
Right2
Left3
Right3
Left4
Right4
Left5
Right5
Left6
Right6
Left7
Right7
Left8
Right8
Left9
Right9
Left10
Right10
Left11
Right11
HW _FILCO_CSR_FIRMODE = 10
newer
older
pad with six
words of zero
HW _FILCO_DBAR
older
SKIPPED
samplen
samplen+1
samplen+2
basis for left output sample m+1
HW _FILCO_CBAR
samplen+10
newer
Figure 62. FILCO FIR Decimation Mode
5-35xx-D1-1.08-061705
159
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
13.1.2.
FIR Interpolation Filter
When the HW_FILCO_CSR_FIRMODE bit field is set to 01 then it enables the 1:2
interpolation mode. In this mode, FIR filter calculations for each output sample are
made, as in the normal case, except that the filter basis slides down the input sample buffer differently, see Figure 63. “FILCO FIR Interpolation Mode” on page 160.
HW_FILCO_CBAR
Exact Phase n
Exact Phase n+1
Exact Phase n+2
Exact Phase n+3
Exact Phase n+10
larger address
newer
Even Coeff0
Odd Coeff0
Even Coeff1
Odd Coeff1
Even Coeff2
Odd Coeff2
Even Coeff3
Odd Coeff3
Even Coeff4
Odd Coeff4
Even Coeff5
Odd Coeff5
Even Coeff6
Odd Coeff6
Even Coeff7
Odd Coeff7
Even Coeff8
Odd Coeff8
Even Coeff9
Odd Coeff9
Even Coeff10
Odd Coeff10
zero pad
zero pad
zero pad
Interp.Phase n
Interp.Phase n+1
Interp.Phase n+2
Interp.Phase n+3
Interp.Phase n+2
older
pad with three
words of zero
HW_FILCO_DBAR
smaller address
samplen+1
samplen+2
samplen+3
samplen+4
samplen+10
larger address
Left0
Right0
Left1
Right1
Left2
Right2
Left3
Right3
Left4
Right4
Left5
Right5
Left6
Right6
Left7
Right7
Left8
Right8
Left9
Right9
Left10
Right10
Left11
Right11
HW_FILCO_CSR_FIRMODE = 01
older
samplen
samplen+1
samplen+2
basis for left output sample m+2 and m+3
basis for left output sample m and m+1
samplen
Polyphase Filter Coefficients for left output sample m+1, m+3
Polyphase Filter Coefficients for left output sample m, m+2
smaller address
samplen+10
newer
Figure 63. FILCO FIR Interpolation Mode
This setting enables a special polyphase filter mode in which the same basis is read
twice and convolved with two separate sets of coefficients, an even set and an odd
160
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
set. For the case shown in Figure 63, output sample m is computed using the input
sample basis starting at sample Left 0. EvenCoeff 0, 1, 2, etc. is used to compute
output sample m. To compute output sample m+1, the same input sample basis
starting at sample Left0 is read and convolved with the Odd coefficients, starting
with OddCoeff0, 1, 2, etc. To compute output sample m+2, the basis is stepped so
that it starts at input sample Left1 which is then convolved with the Even coefficients. Finally the same input sample basis, beginning with input sample Left1 is
used to generate output sample m+3. Pay careful attention to the four output samples shown in Figure 63, noting which coefficient sets and which input sample basis
are used for each one.
13.2. Bi-Quadratic IIR Filter Cascade Mode
FILCO supports Bi-Quadratic IIR filter modes for both additive and subtractive
graphic equalizer implementations. The basic bi-quad filter is shown in Figure 64.
The bi-quad mode is selected by setting HW_FILCO_CSR_FILCOSEL to one. For
this example, we set HW_FILCO_CSR_IIRMODE to zero for cascade mode.
This filter uses exactly five user supplied coefficients, a1, a2, b0, b1 and b2. These
are pointed to by the HW_FILCO_CBAR coefficient base address register. Since
exactly five coefficients are used every time, FILCO ignores any value that may be
in the HW_FILCO_TAPCNT register at the time a bi-quad filter is “KICKed” off. The
bi-quad inner loop scale factor is controlled by HW_FILCO_CSR_IIRSCALE and
can be set to multiply by either 1.0 or 2.0, recall that neither of these values could
have been represented in a 24-bit twos complement fractional number.
The bi-quad filter processes interleaved stereo sample streams so there are two biquad filters running at the same time with both filters using the same coefficient
sets. Figure 64 and the following discussion focuses only on the left channel but
right channel filtering is identical.
HW_FILCO_DBAR points to the input sample data buffer and HW_FILCO_DMR
specifies its modulo size in words as it did for FIR filters. For every left sample taken
from the input buffer, one sample is produced and written to the output buffer. The
output buffer is pointed to by HW_FILCO_SBAR and its modulo size is given in
HW_FILCO_SMR as it was for FIR filters. The HW_FILCO_WCNT register specifies the number of samples to compute for each kick, as id did for the FIR case.
Finally notice that each bi-quad output value has a final gain multiplier applied to it
before it is written to the output sample buffer. Various zero crossing decisions and
gain value substitutions can be automatically handled by FILCO, see Section 13.4.
“IIR Mode Zero Crossing Detector and Gain Substitution” on page 165.
As with the FIR filter, 56 bit accumulator values are saturated and/or rounded before
storing to the output buffer. In addition, feed back values derived from the accumulator are also saturated and rounded before being used in subsequent multiplications.
Also as in the FIR case, multiplication of -1.0 by -1.0 is neither detected nor saturated to +1.0 and will supply -1.0 to the accumulation.
FILCO is used to processes sub-buffers of long continuous streams of samples.
When the available samples have been processed then FILCO must have its context switched to a new filter segment. When switching to or from an IIR filter then the
accumulated state of the filter must saved and restored. Figure 65 shows the seven
registers that must be saved and restored at an IIR context switch. When an IIR filter is first started, these registered should be loaded with zeroes.
5-35xx-D1-1.08-061705
161
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
HW_FILCO_CSR_IIRSCALE
b0
HW_FILCO_IIRGAIN_IIRGAIN
1 or 2
Σ
Left(n+2)
OutL(n+1)
a1
b1
OutL(n)
Left(n+1)
b2
a2
OutL(n-1)
Left(n)
HW_FILCO_DBAR
HW_FILCO_SBAR
Left0
Right0
Left1
Right1
Left2
Right2
Left n
Right n
Left n+1
Right n+1
Left n+2
Right n+2
Left n+3
Right n+3
Left n+4
Right n+4
HW_FILCO_CBAR
b0
a1
b1
a2
b2
samplen+3
HW_FILCO_CSR_IIRMODE = 0
Left0
Right0
Left1
Right1
Left2
Right2
Left n
Right n
Left n+1
Right n+1
Left n+2
Right n+2
Left n+3
Right n+3
Left n+4
Right n+4
HW_FILCO_CSR_FILCOSEL = 1
Figure 64. FILCO BIQUAD IIR Cascade Filter
162
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
HW_FILCO_CSR_IIRSCALE
b0
HW_FILCO_IIRGAIN_IIRGAIN
1 or 2
Σ
Left(n+2)
OutL(n+1)
Zero
Crossing
Detect
a1
b1
OutL(n)
Left(n+1)
b2
a2
OutL(n-1)
Left(n)
HW_FILCO_CTXT_S0,1
HW_FILCO_CTXT_SW
HW_FILCO_CTXT_A0,1
HW_FILCO_CTXT_S2,3
HW_FILCO_CTXT_WP
HW_FILCO_CTXT_A2,3
HW_FILCO_CTXT_S4,5
HW_FILCO_CSR_IIRMODE = 0
HW_FILCO_CSR_FILCOSEL = 1
Figure 65. Context Switching an IIR Filter
Figure 66 shows the construction of a K band subtractive mode graphics equalizer.
With a subtractive mode equalizer, a single buffer is used for multiple passes of
each stage/band of the equalizer. Each time the bi-quad filter is run to process a
band, both HW_FILCO_DBAR and HW_FILCO_SBAR are set to point the same
circular buffer. HW_FILCO_DMR and HW_FILCO_SMR are both set to the size of
the buffer. The filter coefficients are set for the first band and the filter is kicked off.
The filter coefficients for this case are set for notch filtering, i.e. potentially reducing
the amplitude of any frequency components within its passband. Out-of-band frequency components are essentially unmodified. Thus, frequency components within
the band are effectively “subtracted” from the composite signal. This is counter intuitive to what is displayed on the user interface of a graphic equalizer but an effective
implementation technique. For the final filter/channel bank run for the equalizer, one
5-35xx-D1-1.08-061705
163
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
can choose to point the results back in to the same buffer for in-place computation
or one can use this filter operation to also copy the result to another circular buffer.
X(0)
...
X(0)
Y(n)
Bi-Quad #k
X(n)
Bi-Quad #2
X(n)
Bi-Quad #1
X(n)
X(0)
in-place in single buffer
Y(0)
Final
Result
Buffer
Figure 66. FILCO BiQuad Cascade EQ Application
13.3. Bi-Quadratic IIR Filter Parallel Mode
The K-band additive graphic equalizer model shown in Figure 67 uses a different
and more intuitively pleasing strategy. In this case, each filter is set as a bandpass
filter so that the frequency components outside of its passband are strongly suppressed while the components lying inside its passband are copied to the output
buffer where the components from all banks are superpositioned with the summation operator.
This form of the equalizer is used when all bands are computed in parallel and
results are simultaneously available from all filter banks. Thanks to the associativity
of the addition operation, the STMP35xx implements a final addition stage, see Figure 68. “FILCO BIQUAD IIR Parallel Filter” on page 166. In this parallel mode, each
filter bank is run as a separate kick. For a five bank equalizer there are five separate
filters to run and each runs to completion before the next one is started. In parallel
mode, just as a filter output sample is computed, it is added to the value in the output sample array. Thus the superposition of all filter banks is formed in the output
buffer where all filter results for a given sample are added together.
To simplify the number of operations, set the HW_FILCO_CSR_IIRMODE to cascade (zero) for the first filter bank and then set it to parallel (one) for all subsequent
filters.
164
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
G1
X(n)
Y(n)
Bi-Quad #0, parallel
G2
Bi-Quad #1, parallel
G3
Σ
Bi-Quad #2, parallel
...
X(0)
GK
Bi-Quad #N, parallel
Y(0)
Feed Forward
Final
Result
Buffer
Figure 67. FILCO BiQuad Parallel EQ Application
Setting up for the parallel mode bi-quad filter of Figure 68 is essentially the same as
setting up the cascade case, except for setting HW_FILCO_CSR_IIRMODE to one.
Of
course
different
final
gain
values
are
indicated
for
HW_FILCO_IIRGAIN_IIRGAIN. The coefficients are computed for a very different
filter shape, namely bandpass filters instead of notch filters.
Note that no additional zero padding words are required for the IIR filter modes as
opposed to the FIR modes.
The addition of the current output buffer value to the freshly computed IIR output is
performed with overflow detection, saturation and convergent rounding.
13.4. IIR Mode Zero Crossing Detector and Gain Substitution
FILCO contains two zero crossing detect status bits, one for the left channel and
one for the right channel. These bits are cleared at the time an IIR filter operation is
kicked off. These bits can be seen in:
HW_FILCO_ZC_STATUS_ZERO_CROSS_DETECT_STATUS.
When a bit is set it indicates that a zero crossing detect was enabled for the corresponding channel, and the zero crossing has been detected within that filter run.
Once the status bit is set then the 24-bit zero crossing gain from register
HW_FILCO_ZC_GAIN is substituted for the value in HW_FILCO_IIRGAIN in the biquad filter loop.
Thus one can implement a mute function by setting a very low gain value in the zero
crossing gain register. When the zero crossing is detected, then the filter output is
essentially muted while the signal is in the vicinity of its zero crossing. Zero crossing
are detected at the output of the Bi-Quad IIR before the gain is multiplied. Thus one
5-35xx-D1-1.08-061705
165
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
can use the zero crossing detector to go from play to mute and also from mute to
play.
HW_FILCO_CSR_IIRSCALE
b0
1 or 2
Σ
Left(n+2)
HW_FILCO_IIRGAIN_IIRGAIN
OutL(n+1)
a1
b1
OutL(n)
Left(n+1)
b2
a2
OutL(n-1)
Left(n)
HW_FILCO_SBAR
Left0
Right0
Left1
Right1
Left2
Right2
HW_FILCO_DBAR
Left0
Right0
Left1
Right1
Left2
Right2
Left n
Right n
Left n+1
Right n+1
Left n+2
Right n+2
Left n+3
Right n+3
Left n+4
Right n+4
Σ
HW_FILCO_CBAR
b0
a1
b1
a2
b2
Left n
Right n
Left n+1
Right n+1
Left n+2
Right n+2
Left n+3
Right n+3
Left n+4
Right n+4
samplen+3
HW_FILCO_CSR_IIRMODE = 1
HW_FILCO_CSR_FILCOSEL = 1
Figure 68. FILCO BIQUAD IIR Parallel Filter
166
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
13.5. FILCO Programmable Registers
The following registers are available for DSP programmer access and control of the
FILCO Filter Coprocessor.
13.5.1.
FILCO Control/Status Register
This register provides overall control of the filter coprocessor for transactions and
interrupts.
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
DMAOUT
1
7
DMASAMP
1
8
DMACOEF
1
9
IIRMODE
2
0
FILCOSEL
2
1
IIRSCALE
2
2
DMADLY
CLKGT
2
3
X:$FC00
FIRMODE
HW_FILCO_CSR
Table 165. HW_FILCO_CSR
BITS
23
CLKGT
LABEL
22:20 RSRVD
19:16 DMADLY
15:14 FIRMODE
13
IIRSCALE
12
IIRMODE
11
FILCOSEL
10:8
7:6
RSRVD
DMACOEF
5:4
DMASAMP
RW RESET
DEFINITION
RW 0
Clock Gate for the Filco functions
1 = Clocks enabled
0 = Clocks gated off
WARNING: this the inverse of most other STMP3500 blocks.
R
000
Reserved – Must be written with 0.
RW $0
Delay each DMA request by a number of clocks
$0 = 1 clock delay
...
$F = 16 clock delays.
RW 00
FIR Mode filter type
00 = FIR 1:1 normal mode
01 = FIR 1:2 interpolation mode
10 = FIR 2:1 decimation mode
RW 0
IIR final scale mode
0 = Multiply IIR filter result by 1x
1 = Multiply IIR filter result by 2x
RW 0
IIR Mode
0 = Cascade
1 = Parallel (add to output buffer)
RW 0
Filter Type:
0 = FIR
1 = IIR
R
$0
Reserved – Must be written with 0.
RW 00
00 - DMA input coefficients from on-chip XRAM
01 - DMA input coefficients from on-chip YRAM
10 - DMA input coefficients from on-chip PRAM
11- Reserved
RW 00
00 - DMA input samples from on-chip XRAM
01 - DMA input samples from on-chip YRAM
10 - DMA input samples from on-chip PRAM
11- Reserved
Table 166. FILCO Control/Status Register Description
5-35xx-D1-1.08-061705
167
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
3:2
DMAOUT
1:0
RW RESET
DEFINITION
RW 00
00 - DMA into on-chip XRAM
01 - DMA Into on-chip YRAM
10 - DMA into on-chip PRAM
11- Reserved
R
$0
Reserved – Must be written with 0.
RSRVD
Table 166. FILCO Control/Status Register Description (Continued)
13.5.2.
FILCO SAT Register
This register provides a debug view of the saturation state of each of the eight accumulators.
HW_FILCO_SAT
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
SAT
2
3
X:$FC01
Table 167. HW_FILCO_SAT
BITS
LABEL
23:8 RSRVD
7:0
SAT
RW RESET
DEFINITION
R
$0
Reserved – Must be written with 0.
RW $00
Overflow status for each accumulator.These bits will be set if an
accumulator store referenced an accumulator with an overflow
(out range condition). Once these sticky bits are set they will
remain set unto cleared. Set to one to clear these bits.
Table 168. FILCO Accumulator Status Register Description
13.5.3.
FILCO Coefficient Base Address Register
This register provides a pointer to the array of coefficients to be used for the current
filter.
HW_FILCO_CBAR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
CBAR
2
3
X:$FC02
Table 169. HW_FILCO_CBAR
BITS
LABEL
23:16 RSRVD
15:0 CBAR
RW RESET
DEFINITION
R
$0
Reserved – Must be written with 0.
RW $0000 Address Location of first coefficient.These bits work in
conjunction with HW_FILCO_CSR_DMACOEF bit field
which determines the source on-chip RAM.
Table 170. FILCO Coefficient Base Address Register Description
168
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
13.5.4.
FILCO TAP COUNT Register
This register contains the number of coefficients in an FIR filter. Remember that
either three or six words of zero padding must be appended to the end of the coefficient array, however, they should be included in the tap count value stored here. In
addition, this register contains the number of even or odd phase taps for a 1:2 interpolating FIR filter, not the sum of the two. Finally, all bi-quad IIR filters have exactly
five coefficients and therefore ignore any value written to this register.
HW_FILCO_TAPCNT X:$FC03
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
CMR
2
3
Table 171. HW_FILCO_TAPCNT
BITS
LABEL
23:8 RSRVD
7:0
CMR
RW RESET
DEFINITION
R
$0
Reserved – Must be written with 0.
RW $00
Size of filter, i.e. number of coefficients (256 taps). This field is
ignored for IIR registers which always have exactly five
coefficients.
Table 172. FILCO Coefficient Modulo Register Description
13.5.5.
FILCO FIR Coefficient Current Position Register
This register provides a diagnostic view of the offset to add to the CBAR to find the
address from which the current coefficient will be fetched.
HW_FILCO_CCPR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
CCPR
2
3
X:$FC04
Table 173. HW_FILCO_CCPR
BITS
LABEL
23:13 RSRVD
12:0 CCPR
RW RESET
DEFINITION
R
$0
Reserved – Must be written with 0.
R
$0000 Current offset from coefficient base address.
Table 174. FILCO Coefficient Current Position Register Description
5-35xx-D1-1.08-061705
169
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
13.5.6.
FILCO Spare Register
This register provides bits that can read or written from DSP software for use in generating metal mask patches to the silicon.
HW_FILCO_SPARE
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
SPARE
2
3
X:$FC05
Table 175. HW_FILCO_SPARE
BITS
23:0 SPARE
LABEL
RW RESET
RW $000000
DEFINITION
Spare bits for patch gates
Table 176. FILCO SPARE Register Description
13.5.7.
FILCO Interrupt Configuration Register
This register holds the completion interrupt status bit and the completion interrupt
enable for the FILCO.
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
IRQ
2
3
X:$FC0B
IRQEN
HW_FILCO_INTR
Table 177. HW_FILCO_INTR
BITS
LABEL
23:2 RSRVD
1
IRQ
0
IRQEN
RW RESET
DEFINITION
R
$0
Reserved – Must be written with 0.
RW 0
FILCO interrupt is a status bit indicating a completion interrupt
has been generated. IRQ occurs when the filter has finished
processing a buffer, i.e. WORDCOUNT = 0.
RW 0
FILCO interrupt enable. Set to one to enable interrupt generation
whenever IRQ is a one.
Table 178. FILCO Interrupt Configuration Register Description
170
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
13.5.8.
FILCO Kick Register
This register provides access to the bits that control kicking off a transaction as well
as stopping that transaction.
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
KICK
2
2
PAUSE
2
3
X:$FC0C
UNKICK
HW_FILCO_KICK
Table 179. HW_FILCO_KICK
BITS
23:21 PAUSE
20:2
1
RSRVD
UNKICK
0
KICK
LABEL
RW RESET
DEFINITION
RW 000
101- Pause the FILCO at any point of the operation. All other
values allow normal operation.
R
$0
Reserved – Must be written with 0.
RW 0
Write a one to this bit to temporarily freeze states to allow for
processing new samples. When read, a one indicates that the
FILCO is busy; a zero indicates that the FILCO is idle.
RW 0
Set the KICK bit to one to start the filter process in the FILCO.
Kick is automatically cleared when the word count reaches zero.
Table 180. FILCO Kick Register Description
13.5.9.
FILCO Input Sample (Data) Base Address Register
This register contains the base address pointing to the input Data sample buffer.
HW_FILCO_DBAR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
DBAR
2
3
X:$FC0D
Table 181. HW_FILCO_DBAR
BITS
LABEL
23:16 RSRVD
15:0 DBAR
RW RESET
DEFINITION
R
$0
Reserved – Must be written with 0.
RW $0000 Address of the first input sample data word. These bits work in
conjunction with HW_FILCO_CSR_DMADATA bit field
which determines the source on-chip RAM.
Table 182. FILCO Input Sample Data Base Address Register Description
5-35xx-D1-1.08-061705
171
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
13.5.10. FILCO Sample Data Buffer Modulo Register
This register provides the modulo address for the input sample data circular buffer.
This register holds the count of the maximum number words in the buffer. All data
references from the sample data buffer fall within the range HW_FILCO_DBAR to
HW_FILCO_DBAR plus HW_FILCO_DMR.
HW_FILCO_DMR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
DMR
2
3
X:$FC0E
Table 183. HW_FILCO_DMR
BITS
LABEL
23:13 RSRVD
12:0 DMR
RW RESET
DEFINITION
R
$0
Reserved – Must be written with 0.
RW $000
Size of data input buffer (up to 8192 words)
Table 184. FILCO Sample Data Buffer Modulo Register Description
13.5.11. FILCO Sample Count Register
FILCO uses base plus index addressing to access samples in the input sample
buffer. The 13 bit index is constrained to wrap from the modulo register value back
to zero as successive input samples are fetched. There are two sample index register in the FILCO. The first one, called the start index is used to keep track of where
the next filter basis will begin. The second one, called the running index, is used to
step backward in time, fetching each sample from the filter’s basis.
Samples are fetched from locations determined by adding the value in
HW_FILCO_DBAR to the running index register.
The writes to this register are directed to the start index register. Reads come from
the running index register. When FILCO is not in the “KICKED” they should be equal
so that saving the read value as part of a context switch will provide the correct
value to write to HW_FILCO_DCPR when the context is switched back in.
Values written to the HW_FILCO_DCPR will not be visible when read back. Values
written to the start index register are copied to the running index register after a kick
start.
172
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
HW_FILCO_DCPR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
DCPR
2
3
X:$FC0F
Table 185. HW_FILCO_DCPR
BITS
LABEL
23:13 RSRVD
12:0 DCPR
RW RESET
DEFINITION
R
$0
Reserved – Must be written with 0.
RW $0001 Current input sample counter. Reads return the value of the
running index register. Reads indicate the first sample to be read
on the next set of output samples. Reads should only occur while
the FILCO is not in the KICKED state or while it is paused. Writes
are directed to the start index register. Only odd values can be
written to this start index register.
Table 186. FILCO Sample Count Register Description
13.5.12. FILCO Output Sample Base Address Register
This register holds the base address pointing to the on-chip RAM buffer which will
receive the filtered output samples.
HW_FILCO_SBAR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
SBAR
2
3
X:$FC10
Table 187. HW_FILCO_SBAR
BITS
LABEL
23:16 RSRVD
15:0 SBAR
RW RESET
DEFINITION
R
$0
Reserved – Must be written with 0.
RW $0000 Address of first filtered output sample word. These bits work in
conjunction with HW_FILCO_CSR_DMAOUT bit field
which determines the destination on-chip RAM.
Table 188. FILCO Output Sample Base Address Register Description
5-35xx-D1-1.08-061705
173
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
13.5.13. FILCO Output Sample Modulo Register
This register provides the modulo size for the output sample circular buffer.
HW_FILCO_SMR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
SMR
2
3
X:$FC11
Table 189. HW_FILCO_SMR
BITS
LABEL
23:13 RSRVD
12:0 SMR
RW RESET
DEFINITION
R
$0
Reserved – Must be written with 0.
RW $0000 Size of output sample data buffer.
Table 190. FILCO Output Sample Modulo Register Description
13.5.14. FILCO Output Sample Current Position Register
This register provides a diagnostic and context switch view of the offset to add to the
SBAR to find the address to which the next output sample word will be written.
HW_FILCO_SCPR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
SCPR
2
3
X:$FC12
Table 191. HW_FILCO_SCPR
BITS
LABEL
23:13 RSRVD
12:0 SCPR
RW RESET
DEFINITION
R
$0
Reserved – Must be written with 0.
RW $0000 Current output sample pointer.
Table 192. FILCO Output Sample Current Position Register Description
13.5.15. FILCO Tail Pointer Register
This register provides a read only view of the address of the last input data sample
fetched for an FIR filter.
HW_FILCO_TPTR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
TAIL_PNTR
2
3
X:$FC13
Table 193. HW_FILCO_TPTR
174
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
23:16 RSRVD
15:0 TAIL_PNTR
RW RESET
DEFINITION
R
$0
Reserved – Must be written with 0.
R
$0000 DMA address (pointer) of last FIR filter sample fetched for current
input data sample basis.
Table 194. FILCO Tail Pointer Register Description
13.5.16. FILCO Unity Gain Register
This register provides a unity gain feed forward path for one or both channels.
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
PER_CHANNEL
2
3
X:$FC17
0
0
UNITY_GAIN
HW_FILCO_UGAIN
Table 195. HW_FILCO_UGAIN
BITS
LABEL
23:13 RSRVD
1
PER_CHANNEL
0
UNITY_GAIN
RW RESET
DEFINITION
R
$0
Reserved – Must be written with 0.
RW 0
Use each channel zero cross detect bit to enable unity gain on a
channel by channel basis.
RW 0
Force unity gain in IIR mode. Since +1.0 is not representable in a
24-bit twos complement fractional number but is a highly
desirable value for the final IIR gain multiplier, this bit provides a
way to specify an exact multiply by +1.0.
Table 196. FILCO Unity Gain Register Description
5-35xx-D1-1.08-061705
175
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
13.5.17. FILCO IIR Gain Register
This register provides an overall gain control for the output of an IIR filter step. This
is a convenient place to set the attenuation for each equalizer band in a parallel IIR
implementation of an equalizer. If the HW_FILCO_UGAIN_UNITY_GAIN bit is set
then a value of 1.0 is used in place of this value. If the zero cross detector is in use
and a zero crossing has been detected then the HW_FILCO_ZC_GAIN value will be
used in place of this registers value to set the final gain of an IIR filter.
HW_FILCO_IIRGAIN X:$FC18
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
IIRGAIN
2
3
Table 197. HW_FILCO_IIRGAIN
BITS
LABEL
23:0 IIRGAIN
RW RESET
RW $000000
DEFINITION
IIR mode volume register, same register as coefficient register
5.
Table 198. FILCO IIR Volume Register Description
13.5.18. FILCO Word Count Register
This register holds the count of the number of output samples to process (count of
individual output sample words, not word pairs).
HW_FILCO_WORDCOUNT X:$FC1B
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
WCNT
2
3
Table 199. HW_FILCO_WORDCOUNT
BITS
LABEL
23:13 RSRVD
12:0 WCNT
RW RESET
R
$0
RW $0000
DEFINITION
Reserved – Must be written with 0.
Number of single samples to be processed.
Table 200. FILCO IIR Volume Register Description
176
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
13.5.19. FILCO Zero Cross Status Register (IIRMODE ONLY)
This register provides controls for the IIR zero crossing detector mode.
HW_FILCO_ZC_STATUSX:$FC1C
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ZERO_CROSS_DETECT_EN
2
2
ZERO_CROSS_DETECT_STATUS
2
3
Table 201. HW_FILCO_ZC_STATUS
BITS
LABEL
23:4 RSRVD
3:2
ZERO_CROSS_DETECT_STATUS
1:0
ZERO_CROSS_DETECT_EN
RW RESET
DEFINITION
R
$0
Reserved – Must be written with 0.
RW 00
Zero Cross Detect Status
00 - Zero cross not detected.
01 - even channel detected.
10 - odd channel detected.
11 - both channels detected.
RW 00
Zero Cross Detect Enable
00 - Zero cross detect disabled.
01 - right channel, i.e. odd addresses enabled.
10 - left channel, i.e. even addresses enabled.
11 - zero crossing detector enabled for both channels.
Table 202. FILCO Zero Cross Status Register Description
5-35xx-D1-1.08-061705
177
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
13.5.20. FILCO Zero Cross Gain Setting Register (IIRMODE ONLY)
This register provides the gain to be substituted when a channel goes through a
zero crossing.
HW_FILCO_ZC_GAIN X:$FC1D
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ZCGAIN
2
3
Table 203. HW_FILCO_ZC_GAIN
BITS
LABEL
23:0 ZCGAIN
RW RESET
RW $000000
DEFINITION
Gain setting to be used once a zero cross is detected.
Table 204. FILCO Zero Cross Gain Setting Register Description
178
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
13.5.21. FILCO Context Acc0 Register (IIRMODE ONLY)
This register provides access to accumulator zero for context switching between IIR
sub-buffers.
HW_FILCO_CTXT_A0R X:$FC20
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ACCUM
2
3
Table 205. HW_FILCO_CTXT_A0R
BITS
LABEL
23:0 ACCUM
RW RESET
RW $000000
DEFINITION
Accumulator 0, save and restore this register between IIR
sub buffers. Initialize to zero for first sub-buffer.
Table 206. FILCO Context Acc0 Register Description
13.5.22. FILCO Context Acc1 Register (IIRMODE ONLY)
This register provides access to accumulator one for context switching between IIR
sub-buffers.
HW_FILCO_CTXT_A1R X:$FC21
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ACCUM
2
3
Table 207. HW_FILCO_CTXT_A1R
BITS
LABEL
23:0 ACCUM
RW RESET
RW $000000
DEFINITION
Accumulator 1, save and restore this register between IIR
sub buffers. Initialize to zero for first sub-buffer.
Table 208. FILCO Context Acc1 Register Description
13.5.23. FILCO Context Acc2 Register (IIRMODE ONLY)
This register provides access to accumulator two for context switching between IIR
sub-buffers.
5-35xx-D1-1.08-061705
179
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
HW_FILCO_CTXT_A2R X:$FC22
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ACCUM
2
3
Table 209. HW_FILCO_CTXT_A2R
BITS
LABEL
23:0 ACCUM
RW RESET
RW $000000
DEFINITION
Accumulator 2, save and restore this register between IIR
sub buffers.
Table 210. FILCO Context Acc0 Register Description
13.5.24. FILCO Context Acc3 Register (IIRMODE ONLY)
This register provides access to accumulator three for context switching between
IIR sub-buffers.
HW_FILCO_CTXT_A3R X:$FC23
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ACCUM
2
3
Table 211. HW_FILCO_CTXT_A3R
BITS
LABEL
23:0 ACCUM
RW RESET
RW $000000
DEFINITION
Accumulator 3, save and restore this register between IIR
sub buffers.
Table 212. FILCO Context Acc3 Register Description
13.5.25. FILCO Switch Register (IIRMODE ONLY)
This register provides access to the ping pong switch bit for context switching
between IIR sub-buffers.
HW_FILCO_ CTXT_SWX:$FC24
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
BIT
2
2
xx
2
3
Table 213. HW_FILCO_CTXT_SW
180
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
23:1 RSRVD
0
BIT
RW RESET
R
$000000
RW 0
DEFINITION
Reserved – Must be written with 0.
Ping Pong Switch bit. Save and restore this bit at IIR
context switches. Initialize it to zero at the beginning of a
filter run.
Table 214. FILCO Switch Register Description
13.5.26. FILCO Sample 0 Register (IIRMODE ONLY)
This register provides access to sample 0 of the sample register file for context
switching between IIR sub-buffers.
HW_FILCO_ CTXT_S0X:$FC25
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
SAMPLE
2
3
Table 215. HW_FILCO_CTXT_S0
BITS
LABEL
23:0 SAMPLE
RW RESET
RW $000000
DEFINITION
Sample 0 of the sample register file. Save and restore this
value at IIR context switches. Initialize it to zero at the
beginning of a filter run.
Table 216. FILCO Sample 0 Register Description
13.5.27. FILCO Sample 1 Register (IIRMODE ONLY)
This register provides access to sample 1 of the sample register file for context
switching between IIR sub-buffers.
HW_FILCO_ CTXT_S1X:$FC26
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
SAMPLE
2
3
Table 217. HW_FILCO_CTXT_S1
BITS
LABEL
23:0 SAMPLE
RW RESET
RW $000000
DEFINITION
Sample 1 of the sample register file. Save and restore this
value at IIR context switches. Initialize it to zero at the
beginning of a filter run.
Table 218. FILCO Sample Register Description
5-35xx-D1-1.08-061705
181
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
13.5.28. FILCO Sample 2 Register (IIRMODE ONLY)
This register provides access to sample 2 of the sample register file for context
switching between IIR sub-buffers.
HW_FILCO_ CTXT_S2X:$FC27
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
SAMPLE
2
3
Table 219. HW_FILCO_CTXT_S2
BITS
LABEL
23:0 SAMPLE
RW RESET
RW $000000
DEFINITION
Sample 2 of the sample register file. Save and restore this
value at IIR context switches. Initialize it to zero at the
beginning of a filter run.
Table 220. FILCO Sample 2 Register Description
13.5.29. FILCO Sample 3 Register (IIRMODE ONLY)
This register provides access to sample 3 of the sample register file for context
switching between IIR sub-buffers.
HW_FILCO_ CTXT_S3X:$FC28
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
SAMPLE
2
3
Table 221. HW_FILCO_CTXT_S3
BITS
LABEL
23:0 SAMPLE
RW RESET
RW $000000
DEFINITION
Sample 3 of the sample register file. Save and restore this
value at IIR context switches. Initialize it to zero at the
beginning of a filter run.
Table 222. FILCO Sample 3 Register Description
13.5.30. FILCO Sample 4 Register (IIRMODE ONLY)
This register provides access to sample 4 of the sample register file for context
switching between IIR sub-buffers.
182
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
HW_FILCO_ CTXT_S4X:$FC29
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
SAMPLE
2
3
Table 223. HW_FILCO_CTXT_S4
BITS
LABEL
23:0 SAMPLE
RW RESET
RW $000000
DEFINITION
Sample 4 of the sample register file. Save and restore this
value at IIR context switches. Initialize it to zero at the
beginning of a filter run.
Table 224. FILCO Sample 4 Register Description
13.5.31. FILCO Sample 5 Register (IIRMODE ONLY)
This register provides access to sample 5 of the sample register file for context
switching between IIR sub-buffers.
HW_FILCO_ CTXT_S5X:$FC2A
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
SAMPLE
2
3
Table 225. HW_FILCO_CTXT_S5
BITS
LABEL
23:0 SAMPLE
RW RESET
RW $000000
DEFINITION
Sample 5 of the sample register file. Save and restore this
value at IIR context switches. Initialize it to zero at the
beginning of a filter run.
Table 226. FILCO Sample 5 Register Description
13.5.32. FILCO Write Pointer Register (IIRMODE ONLY)
This register provides to the write pointer bits for context switching between IIR subbuffers.
HW_FILCO_CTXT_WPX:$FC2B
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
WPTR
2
3
Table 227. HW_FILCO_CTXT_WP
5-35xx-D1-1.08-061705
183
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
23:2 RSRVD
1:0
WPTR
RW RESET
R
$000000
RW $000000
DEFINITION
Reserved – Must be written with 0.
Holds the write pointer value at the end of a sub-buffer.
Save and restore this value at IIR context switches. Initialize
it to zero at the beginning of a filter run.
Table 228. FILCO Write Pointer Register Description
184
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
14. PULSE WIDTH MODULATOR (PWM) CONTROLLER
The STMP35xx contains four PWM output controllers that can be used in place of
GPIO pins. Applications include LED brightness control and high voltage generators
for electroluminescent lamp (E.L.) display back lights. Independent output control of
each phase allows zero, one or hi-Z to be independently selected for the active and
inactive phases. Individual outputs can be run in lock step with guaranteed nonoverlapping portions for differential drive applications.
16KWord SRAM
Y-BUS
P-BUS
X-BUS
DSP CORE
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
Active3,Inactive3
count3
resets at Period3
Divide
BY :
16,
64,
256,
1024
Active3 >= Count3
Count3 < Inactive3
Act.,Inact. State0
Period0
PWM Programmable Registers
ACT
PWM3
GP7
Active2,Inactive2
count2
resets at Period2
PWM2
Active2 >= Count2
Count2 < Inactive2
ACT
GP9
Active1,Inactive1
24MHz
XTAL
Osc.
count1
resets at Period1
PWM1
Active1 >= Count1
Count1 < Inactive1
ACT
TIO1
Active0,Inactive0
count0
resets at Period0
PWM0
Active0 >= Count0
Count0 < Inactive0
ACT
TIO0
Figure 69. Pulse Width Modulation Controller (PWM) Block Diagram
Figure 69 shows the block diagram of the PWM controller. The controller does not
use the DMA. Initial values of Period, Active, and Inactive widths are set for each
desired channel. The outputs are selected by phase and then the desired PWM
5-35xx-D1-1.08-061705
185
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
channels are simultaneously enabled. This effectively launches the PWM outputs to
autonomously drive their loads without further DSP intervention. In backlit high voltage applications, a feed forward control can be periodically used to change the
count parameters based on LRADC evaluation of the battery state. Feedback control can be provided by assigning one LRADC channel to monitor the integrating
capacitor voltage. Care must be taken to protect the LRADC from catastrophic over
voltage in this case. For most EL back light applications, open loop control with precision PWM timers based on a stable crystal oscillator is sufficient.
Each PWM channel has a dedicated internal 12 bit counter which increments once
for each divided clock period presented from the clock divider. The internal counter
resets when it reaches the value stored in the channel control registers, e.g.
HW_PWM_CH0BR_PERIOD. The Active flip flop is set to one when the internal
counter reaches the value stored in HW_PWM_CH0AR_ACTIVE. It remains high
until
the
internal
counter
exceeds
the
value
stored
in
HW_PWM_CH0AR_INACTIVE. These two value define the starting and ending
points for the logically “active” portion of the waveform. As shown in Figure 70, the
actual state on the output for each phase, e.g. active or inactive, is completely controlled by the active and inactive state values in the channel control registers.
HW_PWM_CH0AR_ACTIVE
HW_PWM_CH0AR_INACTIVE
HW_PWM_CH0BR_PERIOD
PERIOD
INACTIVE
ACTIVE
ACTIVE_FF
PWM0 Output
inactive_state[1:0]
active_state[1:0]
inactive_state[1:0]
active_state[1:0]
HW_PWM_CH0BR_ACTIVE_STATE
HW_PWM_CH0BR_INACTIVE_STATE
Figure 70. PWM Output Example
The actual values obtainable on the output are shown below, see Figure 72. “PWM
Output Driver” on page 187. Notice that one possible state is to turn off the output
driver to provide a hi-Z output. This is useful for external circuits that drive E.L. backlights and for direct drive of LEDs.
By setting up two channels in lock step and by setting their low and high states to
opposite values, one can generate a differential signal pair that alternates between
pulling to Vss and floating to Hi-Z. By creating an appropriate offset in the settings of
the two channels with the same period and the same enable instant one can generate differential drive pulses with digitally guaranteed non-overlapping intervals suitable for controlling high voltage switches,
186
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
ACTIVE_FF0
600
680
40
40
600
ACTIVE_FF1
640
PWM0 Output
hi-Z
Vss
PWM1 Output
digitally
guaranteed
HW_PWMCH0BR_PERIOD = 1280
HW_PWMCH1BR_PERIOD = 1280
HW_PWMCH0AR_ACTIVE = 0
HW_PWMCH1AR_ACTIVE = 640
HW_PWMCH0AR_INACTIVE = 600
HW_PWMCH1AR_INACTIVE = 1240
HW_PWMCH0BR_ACTIVE_STATE = 10
HW_PWMCH1BR_ACTIVE_STATE = 10
HW_PWMCH0BR_INACTIVE_STATE = 00
HW_PWMCH1BR_INACTIVE_STATE = 00
Figure 71. PWM Differential Output Pair Example
In the above example, a differential pair is established using channel zero and channel one. The period is set for 1280 divided clocks for both channels. All active
phases are set for 600 divided clocks. There is a 40 divided clock guaranteed offtime between each active phase. Since this is based on a crystal oscillator it is a
very stable non-overlapping period. The total period is also a very stable crystal
oscillator based time interval. In this example, the active phases are pulled to Vss
(ground) while the inactive phases are allowed to float to a hi-Z state.
i2s_sdo2_out
HW_PWM_CH3R_ACTIVE_STATE[0]
1
PWM3
1
HW_PWM_CH3R_INACTIVE_STATE[0]
GP7
I2S_SDO[2]
Active FF
HW_PWM_CH3R_ACTIVE_STATE[1]
1
1
HW_PWM_CH3R_INACTIVE_STATE[1]
i2s_sdo2_oe
HW_PWM_CSR_PWM3_EN
Figure 72. PWM Output Driver
5-35xx-D1-1.08-061705
187
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
Figure 72 shows the generation of the PWM Channel three output. This channel
controls the GP7 output pin when HW_PWM_CSR_PWM3_EN is set to one. The
output pin can be set to a “ZERO”, a “ONE”, or left to float in the high impedance
state. These choices can be independently made for either the active or inactive
phase of the output, see Figure 72. “PWM Output Driver” on page 187.
14.1. PWM Programmable Registers
The following registers are available for DSP programmer access and control of the
PWM controller.
14.1.1.
PWM Configuration Register
This register provides overall control of the four PWM channels.
1
9
1
8
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
CDIV
1
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
PWM0_EN
2
0
PWM1_EN
2
1
PWM2_EN
2
2
MSTR_EN
2
3
X:$FA31
PWM3_EN
HW_PWM_CSR
Table 229. HW_PWM_CSR
BITS
LABEL
23
MSTR_EN
RW RESET
RW 0
22:10 RSRVD
9:8
CDIV
R
$000
RW 00
7:4
3
RSRVD
PWM3_EN
R
$0
Rw 0
2
PWM2_EN
RW 0
DEFINITION
Master Enable is set to one to start the clock divider and to
enable any PWM channel operation.
Reserved – Must be written with 0.
Clock divider ratio to apply to crystal clock frequency.
00 = divide by 16
01 = divide by 64
10 = divide by 256
11 = divide by 1024
NOTE: Beginning with revision TA2 of the 3500 (and all
subsequent revisions), the crystal clock is used for PWM 0
instead of the output of this clock divider. As a result, PWM
0 can not be set for a pulse frequency less than
approximately 6KHz.
NOTE: Be sure to set HW_CCR_XTLEN to use the PWM.
Reserved – Must be written with 0.
Enables PWM Channel 3 to begin cycling when set to one.
Setting this bit to one also enables PWM3 onto the output
pin in place of the other uses for that pin, see Section 31.
“PIN DESCRIPTION” on page 377.
Enables PWM Channel 2 to begin cycling when set to
one.Setting this bit to one also enables PWM2 onto the
output pin in place of the other uses for that pin, see Section
31. “PIN DESCRIPTION” on page 377.
Table 230. PWM Configuration and Status Register Description
188
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
1
PWM1_EN
RW RESET
RW 0
0
RW 0
PWM0_EN
DEFINITION
Enables PWM Channel 1 to begin cycling when set to one.
Setting this bit to one also enables PWM1 onto the output
pin in place of the other uses for that pin, see Section 31.
“PIN DESCRIPTION” on page 377.
Enables PWM Channel 0 to begin cycling when set to one.
Setting this bit to one also enables PWM0 onto the output
pin in place of the other uses for that pin, see Section 31.
“PIN DESCRIPTION” on page 377.
Table 230. PWM Configuration and Status Register Description (Continued)
14.1.2.
PWM Channel 0 A Register
This register controls PWM channel 0.
HW_PWM_CH0AR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ACTIVE
INACTIVE
2
3
X:$FA32
Table 231. HW_PWM_CH0AR
BITS
LABEL
23:12 INACTIVE
RW RESET
RW $000
11:0
RW $000
ACTIVE
DEFINITION
Number of divided xtal clocks to count before resetting the
ACTIVE flip flop for this channel. The internal count of the
channel is compared for greater than this value to reset the
ACTIVE FF.
Number of divided xtal clocks to count before setting the
ACTIVE flip flop for this channel. The internal count of the
channel is compared for greater than or equal to this value.
If the internal count is greater than or equal then the
ACTIVE FF is set to one.
Table 232. PWM Channel 0 A Register Description
14.1.3.
PWM Channel 0 B Register
This register controls PWM channel 0.
5-35xx-D1-1.08-061705
189
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
PERIOD
2
2
ACTIVE_STATE
2
3
X:$FA33
INACTIVE_STATE
HW_PWM_CH0BR
Table 233. HW_PWM_CH0BR
BITS
LABEL
15:14 ACTIVE_STATE
RW RESET
RW 00
13:12 IN_ACTIVE_STATE
RW 00
11:0
RW $000
PERIOD
DEFINITION
The logical active state is mapped to a physical state at the
output pins, as follows:
00,01 = hi-Z
10 = 0
11 = 1
The logical inactive state is mapped to a physical state at
the output pins, as follows:
00,01 = hi-Z
10 = 0
11 = 1
Number of divided xtal clocks in entire period of the PWM
waveform minus one, i.e. to obtain six clocks in the actual
period then set this field to five.
Table 234. PWM Channel 0 B Register Description
14.1.4.
PWM Channel 1 A Register
This register controls PWM channel 1.
HW_PWM_CH1AR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ACTIVE
INACTIVE
2
3
X:$FA34
Table 235. HW_PWM_CH1AR
190
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
23:12 INACTIVE
RW RESET
RW $000
11:0
RW $000
ACTIVE
DEFINITION
Number of divided xtal clocks to count before resetting the
ACTIVE flip flop for this channel. The internal count of the
channel is compared for greater than this value to reset the
ACTIVE FF.
Number of divided xtal clocks to count before setting the
ACTIVE flip flop for this channel. The internal count of the
channel is compared for greater than or equal to this value.
If the internal count is greater than or equal then the
ACTIVE FF is set to one.
Table 236. PWM Channel 1 A Register Description
5-35xx-D1-1.08-061705
191
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
14.1.5.
PWM Channel 1 B Register
This register controls PWM channel 1.
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
PERIOD
2
2
ACTIVE_STATE
2
3
X:$FA35
INACTIVE_STATE
HW_PWM_CH1BR
Table 237. HW_PWM_CH1BR
BITS
LABEL
15:14 ACTIVE_STATE
RW RESET
RW 00
13:12 IN_ACTIVE_STATE
RW 00
11:0
RW $000
PERIOD
DEFINITION
The logical active state is mapped to a physical state at the
output pins, as follows:
00,01 = hi-Z
10 = 0
11 = 1
The logical inactive state is mapped to a physical state at
the output pins, as follows:
00,01 = hi-Z
10 = 0
11 = 1
Number of divided xtal clocks in entire period of the PWM
waveform.
Table 238. PWM Channel 1 B Register Description
14.1.6.
PWM Channel 2 A Register
This register controls PWM channel 2.
HW_PWM_CH2AR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ACTIVE
INACTIVE
2
3
X:$FA36
Table 239. HW_PWM_CH2AR
192
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
23:12 INACTIVE
RW RESET
RW $000
11:0
RW $000
ACTIVE
DEFINITION
Number of divided xtal clocks to count before resetting the
ACTIVE flip flop for this channel. The internal count of the
channel is compared for greater than this value to reset the
ACTIVE FF.
Number of divided xtal clocks to count before setting the
ACTIVE flip flop for this channel. The internal count of the
channel is compared for greater than or equal to this value.
If the internal count is greater than or equal then the
ACTIVE FF is set to one.
Table 240. PWM Channel 2 A Register Description
14.1.7.
PWM Channel 2 B Register
This register controls PWM channel 2.
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
PERIOD
2
2
ACTIVE_STATE
2
3
X:$FA37
INACTIVE_STATE
HW_PWM_CH2BR
Table 241. HW_PWM_CH2BR
BITS
LABEL
15:14 ACTIVE_STATE
RW RESET
RW 00
13:12 IN_ACTIVE_STATE
RW 00
11:0
RW $000
PERIOD
DEFINITION
The logical active state is mapped to a physical state at the
output pins, as follows:
00,01 = hi-Z
10 = 0
11 = 1
The logical inactive state is mapped to a physical state at
the output pins, as follows:
00,01 = hi-Z
10 = 0
11 = 1
Number of divided xtal clocks in entire period of the PWM
waveform.
Table 242. PWM Channel 2 B Register Description
5-35xx-D1-1.08-061705
193
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
14.1.8.
PWM Channel 3 A Register
This register controls PWM channel 3.
HW_PWM_CH3AR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ACTIVE
INACTIVE
2
3
X:$FA38
Table 243. HW_PWM_CH3AR
BITS
LABEL
23:12 INACTIVE
RW RESET
RW $000
11:0
RW $000
ACTIVE
DEFINITION
Number of divided xtal clocks to count before resetting the
ACTIVE flip flop for this channel. The internal count of the
channel is compared for greater than this value to reset the
ACTIVE FF.
Number of divided xtal clocks to count before setting the
ACTIVE flip flop for this channel. The internal count of the
channel is compared for greater than or equal to this value.
If the internal count is greater than or equal then the
ACTIVE FF is set to one.
Table 244. PWM Channel 3 A Register Description
14.1.9.
PWM Channel 3 B Register
This register controls PWM channel 3.
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
PERIOD
2
2
ACTIVE_STATE
2
3
X:$FA39
INACTIVE_STATE
HW_PWM_CH3BR
Table 245. HW_PWM_C30BR
BITS
LABEL
23:16 RSRVD
15:14 ACTIVE_STATE
RW RESET
R
$0
RW 00
DEFINITION
Reserved – Must be written with 0.
The logical active state is mapped to a physical state at the
output pins, as follows:
00,01 = hi-Z
10 = 0
11 = 1
Table 246. PWM Channel 3 B Register Description
194
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
13:12 IN_ACTIVE_STATE
RW RESET
RW 00
11:0
RW $000
PERIOD
DEFINITION
The logical inactive state is mapped to a physical state at
the output pins, as follows:
00,01 = hi-Z
10 = 0
11 = 1
Number of divided xtal clocks in entire period of the PWM
waveform.
Table 246. PWM Channel 3 B Register Description (Continued)
5-35xx-D1-1.08-061705
195
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
15. I2C INTERFACE
The I2C is a standard 2 wire serial interface used to connect the chip with peripherals or host controllers. This interface provides a standard speed (up to 100kbps) and
high speed (up to 400kbps) I2C connection to multiple devices with the chip acting in
either I2C master or I2C slave mode. Typical applications for the I2C bus include:
EEPROM, LED/LCD, FM Tuner, Real Time Clock, etc.
The I2C port supports multi-master configurations where peripheral devices can
send messages without being queried.
15.1.
I2C-Specific Implementation
• The I2C block can be configured as either a master or slave device. In master
mode it generates the clock (I2C_SCL) and initiates transactions on the data
line (I2C_SDA).
• The I2C block can be configured to pack data into 8, 16 or 24 bit words. Data on
the I2C bus is always byte oriented.
• The I2C block has a programmable device address for master transactions. It
has a fixed seven bit address of $43 = 7’b1000011 for slave transactions. As
seen in the eight bit device address byte, this address corresponds to $86
where the least significant bit is the R/W bit.
• When the interface is enabled, it immediately goes into slave mode and searches
for a start event. It then looks for a match on its hardwired device address. If an
STMP35xx master transaction is started, then the slave is forced to remain in its
idle state until the master transaction completes.
15.1.1.
I2C Interface External Pins
I2C_SDA – I2C Serial Data. This pin carries all address and data bits.
I2C_SCL – I2C Serial Clock. This pin carries the clock used to time the address &
data.
Pull-up resistors are required on both of the I2C lines as all of the I2C drivers are
open drain (pull-down only). Typically, external 2k-Ohm resistors are used to pull the
signals up to VddIO.
Note: See Table 498 for I2C pin placement.
196
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
I2C Interface Control/Status Register
0
5
0
4
0
3
0
2
0
1
RIE
0
6
0
0
I2C_EN
0
7
BUSY
0
8
TIE
0
9
ARBLOST
1
0
MODE
1
1
TDE
1
2
RDR
1
3
RWN
1
4
WL
1
5
ROFL
1
6
TREQ
1
7
TUFL
1
8
ACKF
1
9
SUBA
2
0
LWORD
2
1
ROFLCL
2
2
TUFLCL
2
3
X:$FFE7
ONEBYTE
HW_I2CCSR
BCNT
15.1.2.
Table 247. HW_I2CCSR
BITS
LABEL RW RESET
DEFINITION
23:22 RSRVD
R
0
Reserved – Must be written with 0.
21
ONEBYTE RW 1
Master One Byte Transfer Mode – When set to one, this bit enables a special
one payload byte transfer mode in which the sub address bytes are not generated.
This master only mode allows the transmission of a single I2C address byte from
HW_I2CDAT[23:16] and one data byte taken from HW_I2CDAT[15:8].
WARNING: There is a defect in all revisions of the silicon that precludes the use of
the one byte transfer mode. The SDK has an effective circumvention for one byte
devices using GPIO.
20
TUFLCL
RW 0
Transmitter Underflow Clear bit – Setting this bit clears both the TDE and TUFL
bits. When an underflow condition occurs the correct procedure is to write data to
the HW_I2CDATR register and then send a high pulse on the TUFLCL to clear the
TUFL bit. The TUFLCL bit must be manually reset by software after setting it to
clear the TUFL or TDE bit.
19
ROFLCL
RW 0
Receive Overflow Clear Flag – Setting this bit clears both the RDR and ROFL
bits. The correct procedure is to read the HW_I2CDATR register followed by a high
pulse on the ROFLCL bit. The ROFLCL bit must be manually reset by software
after setting it to clear the ROFL or RDR bit.
18
SUBA
R
0
Sub Address – This bit is set when the Master is transmitting two bytes of sub
address following the slave address byte.
17
LWORD
RW 0
Last Word – This bit is set in master mode when the processor has written its last
word to the I2C transmit register or upon receiving the second last word. The WL
field of this register decides the size of the last block of data received by the
master. This bit is cleared when the stop is generated. If a receive overflow
condition occurs in the master, at least two bytes must be received before a stop is
generated i.e. if the WL field is 0 then the last word bit must not be set on this
receive.
16:15 BCNT
R
00
Byte Count – These bits hold the number of bytes received by the device in either
master or slave mode.
BCNT Bytes received
01
1 byte
10
2 bytes
00
3 bytes
14
ACKF
RW 0
Acknowledge Failure – When this bit is set, no acknowledge has been returned.
13
TUFL
R
0
Transmitter Underflow – An underflow occurs when TDE = 1 and another transfer is
triggered copying the contents of the data register into the shift register.
Table 248. I2C Interface Control/Status Register Description
5-35xx-D1-1.08-061705
197
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
12
ROFL
11
TREQ
RW RESET
DEFINITION
R
0
Receiver Overflow – An overflow occurs when RDR = 1 and the shift register
performs another parallel load of the data register. The shift register is 8 bits wide
for the master receiver, 24 bits wide for the slave receiver. Cleared by a read of the
HW_I2CDATR register followed by a high pulse on the ROFLCL bit.
RW 0
Master Transaction Request – This bit is used to request a transfer with the I2C
in master mode. It is pulsed high (bset,bclr) to request the transfer and the mode
bit must already be driven for either standard or fast mode. Upon receiving the
pulse the I2C master section will generate a start condition and transmit the slave
address.
10:9
WL
RW 00
8
RWN
R
0
7
TDE
R
0
6
RDR
R
0
5
MODE
RW 0
4
TIE
RW 0
3
ARBLOST RW 0
2
BUSY
R
0
This bit is also used to request a repeated start to the master device. A pulse must
be sent, when the processor has written its last word to the I2C transmit register or
upon receiving the last word, to generate a repeated start at the end of the
subsequent receive or transmit. To request a repeated start immediately after the
slave address and two bytes of sub address have been transmitted it is necessary
to send a pulse first to start the transfer of the slave address and then another
pulse a maximum of ns later to request the repeated start.
Word Length – Defines the word length being used by the interface for either I2C
master or slave modes.
WL Word Length
00
8 bit
10
16 bit
01
24 bit
These bits are also used with the LWORD bit so that the device can decide whether
the last block of data the master receives is 8, 16 or 24 bits, i.e. wl0, wl1 are set
before the transfer and then the LWORD bit is set on the receive of the second last
block of data. These bits should not be changed while a transfer is in progress.
Read/Not Write – If equal to zero the Master is writing to the Slave, when equal to
one the Master is reading from the slave.
Transmitter Data Empty – This bit is set when the contents of the transmitter data
register are loaded into the shift register. When this bit is set, the processor can
write to the Transmit Data register. If transmit interrupts are enabled, when this bit
becomes set, an interrupt will be asserted to the processor.
Receiver Data Ready – This bit is set when the shift register performs a broadside
load to the Receive Data Register. It is cleared by hardware when the processor
reads the Receive Data Register. If receive interrupts are enabled, when this bit
becomes set, an interrupt will be asserted to the processor.
Operating Mode Bit – When set to one, the I2C is in Fast mode (400Kbits/sec.)
and when cleared, the I2C is in Standard mode (100 Kbits/sec). There is noise
suppression for both modes. If the device is to be a master, the mode must be set
or cleared before TREQ (bit 11) becomes active.
Transmitter Interrupt Enable – When set, the transmit interrupts are enabled to
the processor.
Arbitration lost – After every transfer using the master I2C device this bit should
be checked to ensure arbitration has not been lost to another master. This bit is
cleared by a write to the control/status register.
I2C Bus Busy – When this bit is set the I2C bus is busy. It is set after a START
condition is detected and remains set until a STOP condition is detected. If there is
arbitration on the I2C_SDA line and our device loses, then the busy bit will remain
set until our device detects a stop condition from the winning device.
Table 248. I2C Interface Control/Status Register Description (Continued)
198
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
1
RIE
0
I2C_EN
RW RESET
DEFINITION
RW 0
RIE Receiver Interrupt – Enable When set, the receive interrupts (Receiver Data
Ready and Receiver Overflow) are enabled. Both interrupts are processed by
reading data from the I2C data register (HW_I2CDATR). If these interrupts are not
processed they will continuously reoccur.
RW 0
Peripheral Enable – When set, the I2C receive and transmit channel are enabled.
Table 248. I2C Interface Control/Status Register Description (Continued)
5-35xx-D1-1.08-061705
199
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
15.1.3.
I2C Data Registers
There are two 24-bit Data Registers located at the same address, Receive and
Transmit Data Registers. The Receive Data Register contents are read over the
data bus if the processor performs a read of the address X:$FFE6. The Transmit
Data Register is written to the external data bus if the processor writes to this same
address. The Transmit Data Register should be filled before setting the TREQ bit to
initiate a transfer.
HW_I2CDAT
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
DATA
2
3
X:$FFE6
Table 249. HW_I2CDAT
BITS LABEL RW
23:0
DATA
RW 0
RESET
DEFINITION
Address/DATA register
Table 250. I2C Data Register Description
On the first write to the transmit data register, the I2C master transmits
HW_I2CDAT[23:16] first. Therefore this byte is always the slave address targeted
for a master read or write transaction. When the HW_I2CCSR_ONEBYTE bit is set
to one in master mode, then HW_I2CDAT_DATA[23:16] is transmitted followed by
HW_I2CDAT_DATA[15:8]. Again, the upper byte holds the slave address. For master
read
transactions,
the
target
slave
address
is
written
to
HW_I2CDAT_DATA[23:16] before the read transaction is started with the TREQ bit.
15.1.4.
I2C Clock Divider Register
The I2C Clock Divider Register controls the division ratio between the system clock
(dclk) and the I2C serial clock (I2C_SCL). The value in the FACT field is used to
divide the system clock to generate the I2C clock. This clock is further divided by 4 if
the MODE bit (bit 5) in the I2C Interface Control/Status register is clear. The I2C
clock frequency can be calculated as follows:
Fast mode (HW_I2CCSR:MODE = 1): I2C clock = DCLK / (FACT*2 + 3)
Slow mode (HW_I2CCSR:MODE = 0):I2C clock = DCLK / (FACT*8 + 22)
If the system clock is set to the crystal oscillator with a 24.0 MHz crystal, then the
reset value for this register will divide this clock by 550 to give an I2C clock of
~43.6 kHz. During I2C boot modes, the bootrom sets the HW_I2CCSR:MODE bit to
switch the I2C block into fast mode — the clock speed in this case will become
~178 kHz.
Note that the fastest master on the bus will determine the high period of the clock
while the slowest master OR slave determines the low period of the clock. The
HW_I2CDIV counter determines the minimum high period as controlled by the
STMP35xx when it is in master mode. This counter controls the minimum low period
when the STMP35xx is enabled in master mode. In slave mode, the STMP35xx follows the clock timing established by the controlling master.
200
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
HW_I2CDIV
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
FACT
2
3
X:$FFE5
Table 251. HW_I2CDIV
BITS
23:9
8:1
0
LABEL
RSRVD
FACT
RSRVD
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
W 01000010 Clock divider value bits, NOTE This is a write only register.
R
0
Reserved – Must be written with 0.
Table 252. I2C Clock Divider Register Description
15.1.5.
I2C Interrupt Sources
The I2C port can be used in either interrupt driven or polled modes. If interrupts are
enabled, a level-sensitive interrupt will be signaled to the processor upon one of the
following events.
ADDRESS
P:$0030
P:$0032
P:$0034
P:$0036
BIT
6
12
7
13
Table 253.
LABEL
INTERRUPT SOURCE
2C
Receiver Data Ready
RDR
I
ROFL
I2C
Receiver Overflow
TDE
I2C
Transmitter Data Empty
TUFL
I2C
Transmitter Underflow
I2C
Interrupt Address Map
The interrupt lines are tied directly to the status bits of the Control/Status Register.
Clearing these bits through software will remove the interrupt request. The Receiver
Data Ready and Transmitter Data Empty requests are automatically removed via
hardware when the Receive Data Register is read or the Transmit Data Register is
written respectively. The overflow and underflow error signals need to be cleared
through software by writing directly to the status bits of the Control/Status Register.
15.2. I2C Bus Protocol
With reference to the clocking scheme shown in the figure below, the I2C interface
operates in the following manner:
A START condition is defined as a HIGH to LOW transition on the data line while the
I2C_SCL line is held HIGH. After this has been transmitted by the Master, the bus is
considered busy. The next byte of data transmitted after the start condition contains
the address of the slave in the first 7 bits and the eighth bit tells whether the Master
is receiving data from the slave or transmitting data to the slave. When an address
is sent, each device in the system compares the first seven bits after a start condition with its address. If they match, the device considers itself addressed by the
Master. In slave mode, the I2C write address is 86h, its read address is 87h.
Data transfer with acknowledge is obligatory. The transmitter must release the
I2C_SDA line during the acknowledge pulse. The receiver must then pull the data
5-35xx-D1-1.08-061705
201
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
Standard Mode
SCL
SDA
8 bits
8 bits
SLAVE Address & r/w
DATA
Acknowledge
signal
START
Condition
Acknowledge
signal
STOP
Condition
Figure 73. I2C data and clock timing
line LOW so that it remains stable low during the HIGH period of the acknowledge
clock pulse. A receiver which has been addressed is obliged to generate an
acknowledge after each byte of data has been received.
15.2.1.
Slave Mode Protocol
The flow chart for the finite state machine is shown in Figure 74. At device start-up
all the registers are reset so that the state is known from that time onward. Once the
I2C is enabled the slave will wait to detect a start condition on the I2C_SCL and
I2C_SDA lines. Once this is detected the slave will read in 8 bits and check against
its own device address which is $86 == 7’b1000011 to see if a master device is trying to start a transfer with our device. If it is our address an acknowledge is sent,
otherwise our slave will not acknowledge and will return to state IDLE.
Next the RW is checked. If it is a write operation then the two sub address bytes
must be received and acknowledged and then the data is received with a check of
receive overflow before data is overwritten into the DSP_rx register.
If the master is requesting a read operation then the slave must start sending data
on the I2C_SDA bus immediately after acknowledging the Slave address and RW
bit. After each byte the acknowledge from the master must be checked. When the
master has received the last byte it will not send an acknowledge and the slave will
return to state IDLE. As long as the slave continues transmitting it must first check
whether the DSP_tx register has been fully transmitted and if so request more data.
The underflow condition is also checked.
Data is now transmitted in byte format. Each data transfer has to contain 8 bits. The
number of bytes transferred per transfer is unlimited. Data is transferred with the
Most Significant Bit (MSB) first. If a receiver can’t receive another complete byte of
data until it has performed some other function, it can hold the clock line, I2C_SCL
LOW to force the transmitter into a wait state. Data transfer only continues when the
receiver is ready for another byte and releases the clock line.
If a slave receiver doesn’t acknowledge the slave address (e.g. it is unable to
receive because it is performing some real time function) the data line must be left
HIGH by the slave. The Master can then abort the transfer.
A LOW to HIGH transition on the I2C_SDA line while the I2C_SCL line is HIGH is
defined as a STOP condition. Each data transfer must be terminated by the generation of a STOP condition.
202
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
W ait for start
Yes
O ur
Master
TR EQ
Receive sub
address 0
No
Receive slave
address
No
Setup tx data
O ur address
Send Ack
Yes
Yes
Ack address
Yes
Rofll
rdrx
No
Read
No
No
Receive
8 bits
Yes
Transmit
8 bits
No
24 bits or
start stop
Yes
No
ACK
Setup tx data
Yes
No
tdefx
No
TX reg
empty
Yes
tdefx
interrupt req
Yes
Yes
Rdr
interrupt req
TUFL
No
Figure 74. I2C Slave Mode Flow Chart
5-35xx-D1-1.08-061705
203
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
Tables 254 - 257 show the first 2 bytes of data as a sub-address for purposes of
illustration. The sub-address is used to address the memory space inside the
device. Table 258 defines each sub-address shown.
ST
SAD+W SAK SUB SAK SUB SAK DATA SAK SP
Table 254. I2C transfer when Master is writing one byte of data to a slave
ST SAD+W SAK SUB SAK SUB SAK DATA SAK DATA SAK SP
Table 255. I2C transfer when Master is writing multiple bytes to a slave
ST SAD+W SAK SUB SAK SUB SR SAD+R SAK DATA NMAK SP
Table 256. I2C transfer when Master is receiving one byte of data from a slave
ST SAD+W SAK SUB SAK SUB SR SAD+R SAK DATA MAK DATA MAK DATA NMAK SP
Table 257. I2C transfer when Master is receiving multiple bytes of data from a slave
BIT
ST
SR
SAD
SAK
SUB
DATA
SP
MAK
NMAK
DESCRIPTION
START condition
Repeated START condition
Slave address
Slave acknowledge
Sub address
Data
STOP condition
Master Acknowledge
No Master Acknowledge
Table 258. I2C Slave and Master mode address definitions
15.2.2.
Master Mode Protocol
In Master Mode the I2C interface generates the clock and initiates all transfers.
15.2.2.1.
Clock Generation
2
The I C clock is generated from the system clock as described above in the register
description.
If another device pulls the clock low before the I2C block has counted the high
period, then the I2C block immediately pulls the clock low as well and starts counting
its the low period. Once the low period has been counted the I2C block releases the
clock line high but must then check to see if another device stills holds the line low in
which case it enters a high wait-state.
In this way the I2C_SCL clock is generated with its low period determined by the
device with the longest clock low period and its high period determined by the one
with the shortest clock high period.
204
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
15.2.2.2.
Master Mode Operation
The finite state machine for master mode operation is shown in Figure 75 and Figure 76. Figure 75 shows the start condition and transmission of the slave address
and two sub address bytes. Figure 76 shows the read and write states.
Tables 259 - 262 show examples of Master Mode I2C transactions. Table 258
defines each sub-address shown. The following read after write transactions are
performed using the restart technique described in the description of
HW_I2CCSR_TREG.
ST SAD+W SAK SUB SAK SUB SAK DATA SAK SP
Table 259. I2C transfer when the interface as master is transmitting one byte of data
ST SAD+R SAK DATA MAK DATA MAK DATA NMAK SP
Table 260. I2C transfer when the interface as master is >1 byte of data from slave
ST SAD+W SAK SUB SAK SUB SAK SR SAD+R SAK DATA NMAK SP
Table 261. I2C transfer when Master is receiving one byte of data from slave internal sub-address
ST SAD+W SAK SUB SAK SUB SAK SR SAD+R SAK DATA MAK DATA MAK DATA NMAK SP
Table 262. I2C transfer when Master is receiving >1 byte of data from slave internal sub-address
To receive data bytes from non-EEPROM devices that do not require the two subaddress bytes, perform an isolated read without using the read after write restart
technique. Make sure that the HW_I2CCSR_ONEBYTE bit is a zero and the following read transactions occur::
ST
SAD+R SAK DATA MAK SP
Table 263. I2C transfer “FM Tuner” read of one byte.
ST
SAD+R SAK DATA MAK DATA MAK DATA
Table 264.
I 2C
NMAK SP
transfer “FM Tuner” read of three bytes.
15.2.2.3.
Special One Byte Master Mode Transmission
There is a special “one byte mode” which is enabled when the
HW_I2CCSR_ONEBYTE bit equals one. In this mode, a start condition is transmitted, followed by the device address byte followed by a single byte of write data. This
sequence always ends with a stop condition.
ST SAD+W SAK DATA SAK SP
Table 265. I2C transfer when the interface as master is transmitting special ONEBYTE mode
The following C code is used to send a one byte transmission:
// Power up the I2C bus pins
HW_GP0PWR.B.B16 = 1;
HW_GP0PWR.B.B17 = 1;
// Load the data register with 1 byte of address[23:16] = 0xC0,
// 1 byte data [15:8] = 0x55 and 1 byte of don’t care [7:0]
HW_I2CDAT.U = 0xC05500;
// Enable I2C
HW_I2CCSR.B.I2C_EN = 1;
5-35xx-D1-1.08-061705
205
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
// Set word length to 16 bits (1 addr byte & 1 data byte)
HW_I2CCSR.B.WL = 2;
// Set the last word bit
HW_I2CCSR.B.LWORD = 1;
// Enable special one byte functionality
HW_I2CCSR.B.ONEBYTE = 1;
// Start the transaction
HW_I2CCSR.B.TREQ = 1;
HW_I2CCSR.B.TREQ = 0;
// Wait for busy to clear
while(HW_I2CCSR.B.BUSY == 1);
// a frame with one byte of address and one byte of data was just sent
READ
STATES
Read
WRITE
STATES
Receive 8 bits
Enable tx data
No
8 bits tx
trans_en <= 1
Check
last_word
rec_count
Transmit 8 bits
8 bits
Decide on ack
Yes
Check ack
last_word
No
Yes
Ack received
NMACK
NMACK
End ack
Check tx reg
ROFL
One byte left
No
STOP
STATE
Yes
ROFL
No
No
SETUP TX
STATE
Yes
No
rdrx
last_word
tdef <= 1
STOP
STATE
Figure 76. I2C Master Mode Flow Chart – Read and Write States
206
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
Wait for dspt
i2c_sda_en=1
stop condition
i2c_sda_en=0
start condition
Setup stop
Suba byte 0
Yes
suba 0 just
transmitted
No
Setup address
Setup TX
TX addr/suba
Underflow
8 bits tx
No
No
Yes
Yes
stop_clk <= 1
Chk ack
No
ACK
Yes
ONEBYTE
&&
(ADDR+SUBA0)
&& WL1=1
WRITE
STATES
Yes
No
rep_start
rep_start
/last_word
last_word
No
Chk read or
write
Write
No
slave addr
just
transmitted
Yes
Read
or Write
Read
READ
STATES
Figure 75. I2C Master Mode Flow Chart
Start condition and transmission of slave address and two sub address bytes or
special device address plus one byte mode (HW_I2CCSR_ONEBYTE =1).
15.3. Known Chip Defects with I2C
15.3.1.
Clear Quest Entry STMP00003940
All revisions of the chip are known to have a defect in which the I2C one byte feature fails for back-to-back writes. If the I2C block writes back-to-back to a slave in
ONE_BYTE mode, only the first transfer actually sends out the data byte. The frame
for the first transfer looks correct and stops correctly. All subsequent transfers (until
5-35xx-D1-1.08-061705
207
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
chip reset) look as follows: start; send address; stop. The frames look correct except
that the data byte is missing.
The workarounds for past I2C bugs are ineffective. Disabling and re-enabling
ONE_BYTE and the I2C block itself have no effect.
The software workaround available in SDK for the FM Tuner suffices for other one
byte devices.
15.3.2.
Clear Quest Entry STMP00002736
All revisions of the chip are known to have a defect in which the I2C gets confused if
reset while still busy.
In the ROM code, there is a section that waits for the port to be no longer busy after
the last word is read in the boot process. Then it writes a 0 to the I2C control register, which should reset it. What is actually happening is that the I2C peripheral goes
into a bad state when the 0 is written. It seems to work okay when a delay is added
between when it is no longer "busy" and when the 0 can be written to the CSR. But,
as this is ROM code, there's no potential for adding the delay. It was found, that the
peripheral can be reset by:
Set TREQ Bit
Clear TREQ Bit
Set Enable bit
Clear ACKF Bit
208
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
16. ENHANCED SPI INTERFACE
The SPI is a standard 4-pin Serial Peripheral Interface for inter-IC control and communication. It interfaces on one side to the SPI bus and on the other has a standard
register data and interrupt interface. Although an SPI system can be configured as a
master or a slave device, the STMP35xx only supports a master interface via the
eSPI.
During an SPI transfer, data is shifted out and shifted in (transmitted and received)
simultaneously. The SPI_SCK line synchronizes the shifting and sampling of the
information. It is an output when the SPI is configured as a master and an input
when the SPI is configured as a slave. Selection of an individual slave SPI device is
performed on the slave select line and slave devices that are not selected do not
interfere with the SPI buses.
Y-BUS
X-BUS
P-BUS
DSP CORE
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
DMA-BUS
ESPI
Programmable
Registers
match
ESPI
DMA Engine & Control
X
MISO
MOSI
SCLK
shift_in[7:0]
shift_out[7:0]
Byte
Processor
Clock
Generation
Figure 77. Enhanced SPI Block Diagram
5-35xx-D1-1.08-061705
209
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
16.1. SPI Pins
SPI_MISO – Master In/Slave Out. Serial input in master mode and output in slave
mode.
SPI_MOSI – Master Out/Slave In. Serial output in master mode and input in slave
mode.
SPI_SCK – Serial Clock. Bit clock output in master mode and input in slave mode.
SPI_SSn – Slave Select. Selects SPI CS0 in master mode, chip select input in slave
mode.
In master mode the SPI_SSn pin is controlled via software through the GPIO module. In slave mode the pin is in SPI mode and connects directly to the SPI module.
Note: See Table 502 for SPI pin placements.
espi_init
HW_ESPI_CSR_SFTRST = 0
HW_ESPI_CSR_MPSPSEL = 0
HW_ESPI_CSR_DMAWAIT = n
# wait states between
DMA transfers
send/receive MSB first
HW_ESPI_CNFG_MSB_FIRST = 1
HW_ESPI_CNFG_ESPI_EN = 1
HW_ESPI_CLKCNTRL_CLKDIV = [2:255]
HW_ESPI_CLKCNTRL_CLKPOL = [0:1]
HW_ESPI_CLKCNTRL_CLKPHA = [0:1]
ESPI controls pins
set clock divider
Set clock polarity & phase
STOP
Figure 78. Enhanced SPI Common Initialization Sequence
210
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
espi_dma_read
call espi_init first
Set number of words
HW_ESPI_CNFG_XFER_SIZE = [0:65535]
HW_ESPI_CNFG_READ_RAM = 1
HW_ESPI_CNFG_SS = [0:1]
HW_ESPI_ADDR_ADDR = [0:65535]
HW_ESPI_INDEX_MEM_SPACE = [0:2]
HW_ESPI_INDEX_BYTE = [0:2]
HW_ESPI_CSR_DONE_INT = 1
HW_ESPI_CSR_DONE_INTEN = 1
HW_ESPI_CSR_KICK = 1
YES
HW_ESPI_CSR_
KICK==1
Set for READ from on-chip RAM
Set ESPI Slave Select
point to source buffer
Select memory space
Set starting bytre position
Clear previous done int
Use done interrupts
Kick off a DMA read to SPI
wait for ESPI read DMA or return
and wait for done interrupt
NO
STOP
Figure 79. Enhanced SPI DMA Read Flowchart
5-35xx-D1-1.08-061705
211
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
espi_dma_write
call espi_init first
Set number of words
HW_ESPI_CNFG_XFER_SIZE = [0:65535]
HW_ESPI_CNFG_READ_RAM = 0
HW_ESPI_CNFG_SS = [0:1]
Set ESPI Slave Select
HW_ESPI_ADDR_ADDR = [0:65535]
point to destination buffer
Set for write to on-chip RAM
Select memory
space
Set starting byte position
HW_ESPI_INDEX_MEM_SPACE = [0:2]
HW_ESPI_INDEX_BYTE = [0:2]
Clear previous done int
HW_ESPI_CSR_DONE_INT = 1
Use done interrupts
HW_ESPI_CSR_DONE_INTEN = 1
HW_ESPI_CSR_KICK = 1
YES
Kick off a DMA write to SPI
wait for ESPI write DMA or return
and wait for done interrupt
HW_ESPI_CSR_
KICK==1
NO
STOP
Figure 80. Enhanced SPI DMA Write Flowchart
16.2. Enhanced SPI Programmable Registers
The following registers are available for DSP programmer access and control of the
enhanced SPI controller.
16.2.1.
ESPI Configuration and Status Register
This register provides overall control and status information for the ESPI, including
interrupt control and DMA overhead control.
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
KICK
1
8
DONE_INT
1
9
PIOMATCH_INT
2
0
DONE_INTEN
2
1
PIOMATCH_INTEN
2
2
MPSPSEL
SFTRST
2
3
X:$FF00
DMAWAIT
HW_ESPI_CSR
Table 266. HW_ESPI_CSR
212
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
23
SFTRST
22
RSRVD
21:20 MPSPSEL
RW RESET
RW 1
R
0
RW 00
19:16
15:12
11:10
9
8
7:6
5
RSRVD
DMAWAIT
RSRVD
PIOMATCH_INTEN
DONE_INTEN
RSRVD
PIOMATCH_INT
R
RW
R
RW
RW
R
RW
4
DONE_INT
RW 0
3:1
0
RSRVD
KICK
R
000
RW 0
0000
0000
00
0
0
00
0
DEFINITION
Software Reset
Reserved – Must be written with 0.
Mode Select bits. In the current design these bits have no
effect.
Reserved – Must be written with 0.
Number dclk wait states to insert between DMA requests.
Reserved – Must be written with 0.
PIO Match Interrupt Enable.
Done Interrupt Enable.
Reserved – Must be written with 0.
When SPI_MODE == MATCH and PIO Read Register
matches received data then this interrupt status bit will be
set to one. Writing a one to this bit will clear it.
When the SPI interface has completed transfer that was
kicked off then this bit is set to one. Writing a one to this
sticky bit will clear it.
Reserved – Must be written with 0.
Set to one to process the command that has been
established in the various HW_ESPI registers.
Table 267. ESPI Configuration and Status Register Description
16.2.2.
ESPI Configuration Register 1
This register provides configuration and control specific to an ESPI transfer.
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ESPI_EN
1
9
READ_RAM
2
0
SS
2
1
XFER_SIZE
2
2
SPI_MODE
2
3
X:$FF01
MSB_FIRST
HW_ESPI_CNFG
Table 268. HW_ESPI_CNFG
BITS
LABEL
23:20 SPI_MODE
RW RESET
RW $0
19:4
XFER_SIZE
RW $0200
3
MSB_FIRST
RW 1
DEFINITION
00 = DMA Mode, ESPI will transmit entire buffer.
01 = PIO Mode, ESPI will transmit each byte written to PIO DATA register.
10 = PIO Match Mode, ESPI continuously transmits byte in PIO register until
received data matches PIO_MATCH register.
011,1XX = reserved
Number of bytes to transfer when the KICK bit is set in DMA or PIO MATCH
mode.
Transmit serial data from the Most Significant Bit (MSB) first when this bit is
set to one. Transfer LSB first when this bit is set to zero.
Table 269. ESPI Configuration Register Description
5-35xx-D1-1.08-061705
213
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
2
SS
RW RESET
RW 0
1
READ_RAM
RW 0
0
ESPI_EN
RW 0
DEFINITION
Slave Select pin is controlled by HW_ESPI_SPICNFG_SS when
HW_ESPI_SPICNFG_ESPI_EN is a one.
Set to one to “read” from on-chip RAM and write to the slave SPI device. Set
to zero to transfer from the external SPI device TO the on-chip RAM buffer.
Same directions relative to SPI device apply for the PIO case.
Set to one to enable ESPI Control of chip pins instead of older SPI device
interface.
Table 269. ESPI Configuration Register Description (Continued)
16.2.3.
ESPI Clock Control Register
This register controls the SPI clock signal frequency, phase and idle polarity.
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
CLKPOL
2
2
CLKDIV
2
3
CLKPHA
HW_ESPI_CLKCNTRL X:$FF02
Table 270. HW_ESPI_CLKCNTRL
BITS
LABEL
23:12 RSRVD
11:4 CLKDIV
RW RESET
R
00
RW $02
3:2
1
RSRVD
CLKPOL
R
00
RW 0
0
CLKPHA
RW 0
DEFINITION
Reserved – Must be written with 0.
DCLK divider for SPI-clock. Maximum frequency is dclk/2,
minimum frequency is dclk/256. Set to 0 or 1 to disable.
Reserved – Must be written with 0.
Defines the idle state of the clock.
1: clock idle at logic high voltage
0: clock idle at logic low voltage
In conjunction with CLKPOL, determines the edge on which
data is transmitted.
Table 271. SPI Clock Control Register Description
16.2.4.
ESPI Programmed I/O Data Register
This register provides PIO read and write access to the serial port.
HW_ESPI_PIODATA X:$FF03
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
MATCH
2
1
IN
2
2
OUT
2
3
Table 272. HW_ESPI_PIODATA
214
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
23:16 OUT
LABEL
RW RESET
RW $00
15:8
IN
R
$00
7:0
MATCH
RW $00
DEFINITION
Data to output on the SPI interface during a PIO transfer or
during a PIO_MATCH mode transfer.
Received data from the SPI interface can be read here.
Valid only while in PIO-Mode and DONE_INT is set to one.
Data used to compare against input from slave to generate
interrupts.
Table 273. ESPI Programmed I/O Data Register Description
16.2.5.
ESPI DMA Address Register
This register holds the DMA starting address.
HW_ESPI_ADDR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ADDR
2
3
X:$FF04
Table 274. HW_ESPI_ADDR
BITS
LABEL
23:16 RSRVD
15:0 ADDR
RW RESET
R
00
RW $0000
DEFINITION
Reserved – Must be written with 0.
Address location of the first data word to be transferred by
the DMA.
Table 275. ESPI DMA Address Register Description
16.2.6.
ESPI DMA Index Register
This register holds the memory space selector and starting byte number for DMA
transfers.
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
MEM_SPACE
2
3
X:$FF05
0
3
0
2
0
1
0
0
OFFSET
HW_ESPI_INDEX
Table 276. HW_ESPI_INDEX
5-35xx-D1-1.08-061705
215
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
23:6 RSRVD
5:4
MEM_SPACE
RW RESET
R
$00000
RW 00
3:2
1:0
R
00
RW 00
RSRVD
OFFSET
DEFINITION
Reserved – Must be written with 0.
Memory space in which the data-block is located
00 = X-Memory
01 = Y-Memory
10 = P-Memory
11 = reserved
Reserved – Must be written with 0.
Byte index into 24-bit on-chip RAM word where the first SPI
byte will be written or read. This bitfield provides byte level
addressability to DMA transfers targeted at the 24 bit wide
on-chip RAM. It specifies the starting byte within the first
word addressed by the DMA.
Table 277. ESPI DMA Index Register Description
216
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
17. SPI INTERFACE
The SPI is a standard 4-pin Serial Peripheral Interface for inter-IC control and communication. It interfaces on one side to the SPI bus and on the other has a standard
register data and interrupt interface. The SPI system can be configured as a master
or a slave device.
During an SPI transfer, data is shifted out and shifted in (transmitted and received)
simultaneously. The SPI_SCK line synchronizes the shifting and sampling of the
information. It is an output when the SPI is configured as a master and an input
when the SPI is configured as a slave. Selection of an individual slave SPI device is
performed on the slave select line and slave devices that are not selected do not
interfere with the SPI buses.
17.1. SPI Pins
SPI_MISO – Master In/Slave Out. Serial input in master mode and output in slave
mode.
SPI_MOSI – Master Out/Slave In. Serial output in master mode and input in slave
mode.
SPI_SCK – Serial Clock. Bit clock output in master mode and input in slave mode.
SPI_SSn – Slave Select. Selects SPI CS0 in master mode, chip select input in slave
mode.
In master mode the SPI_SSn pin is controlled via software through the GPIO module. In slave mode the pin is in SPI mode and connects directly to the SPI module.
Note: See Table 502 for SPI pin placements.
17.2. SPI Registers
17.2.1.
SPI Control Register
HW_SPCSR
BITS LABEL RW
RESET
X:$FFF9
DEFINITION
23:16 RSRVD R
0
Reserved – Must be written with 0.
15:8 DIV
RW 00001010 Divide factor bits – These bits are used to control the frequency of the SPI serial
clock with relation to the device clock. The number must be an even number.
These bits must be set before the SPE or MSTR bits i.e. before the SPI is
configured into master mode.
Note: When the DIV field is written to, if the value written is less than the current
value in the internal counter register, then the counter will continue to
increment to its maximum value, thereby causing a single very long clock
cycle. The new DIV value will then not take effect until the next SPI clock.
7
MODF RW 0
Mode fault flag – Mode fault occurs if the SPI system is configured as a master
and the SPI_SSn input line goes to active low. This happens if a second SPI device
becomes a master and selects this device as if it were a slave. Reading the
HW_SPCSR with MODF set automatically clears the MODF flag.
Note: This condition should never exist since the SPI_SSn line is not used as an
input in master mode.
6
WCOL RW 0
Write Collision Error Flag – The WCOL Flag is set if the HW_SPDR is written
before the end of transfer is signaled. WCOL is cleared by reading the
HW_SPCSR with WCOL set, followed by an access of the HW_SPDR.
Table 278. SPI Control Register Description
5-35xx-D1-1.08-061705
217
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS LABEL RW
RESET
5
SPIF
RW 0
4
CPHA
RW 0
3
CPOL
RW 0
2
SPIE
RW 0
1
MSTR
RW 0
0
SPE
RW 0
DEFINITION
SPI Transfer Complete Flag – The SPIF Flag is set to a one at the end of an SPI
transfer. SPIF is cleared by an access of the HW_SPDR or by a write to the
HW_SPDR register. The definition of an end of transfer depends on who is the
master and who is the slave.
Clock Phase Select –
Note: There must be a delay of 1/2 the clock period of SCLK, after a toggle of
CPHA or CPOL, before there is a write to the HW_SPDR. This is to
prevent glitches on the SCLK.
Clock Polarity Select –
0
Active high clocks selected; SCLK idles low.
1
Active low clocks selected; SCLK idles high.
Note: The SPI_SSn line must be de-asserted and reasserted for a change in the
CPOL bit.
SPI interrupts enabled – When the SPIE bit is set SPI Interrupts are enabled and
triggered when the SPIF bit is set.
Master/Slave Select – When the MSTR bit is set the SPI is configured as a master
and when 0 a slave.
SPI System Enable – When SPE is set the SPI system is enabled and, when
cleared, disabled.
Table 278. SPI Control Register Description (Continued)
17.2.2.
SPI Data Register
The SPI interface runs in PIO mode only. Its 8-bit data register is mapped to the X
data address space. Writes to the lower byte of this register are shifted out
SPI_MOSI in Master Mode and SPI_MISO in slave mode. This register is read on
the transaction completion to retrieve the incoming data from SPI_MISO in master
mode and SPI_MOSI in slave mode.
HW_SPDR
BITS
23:8
7:0
LABEL
RW
X:$FFFA
RESET
DEFINITION
RSRVD
R
0
Reserved – Must be written with 0.
SPIDATA RW 00000000
Table 279. SPI Data Register Description
17.3. Transferring Data over SPI
17.3.1.
Master Mode
In master mode the SPI block must generate the SPI_SCK and send appropriate
data/commands to the slave device(s).
17.3.1.1.
Clock Generation
The clock is the main chip digital clock divided by the divide factor (see above). Out
of reset the SPI_SCK is 24.0/10 ~= 2.4 MHz.
The CPOL (clock polarity) and CPHA (clock phase) bits of the HW_SPCSR are
used to select any of the four combinations of serial clock. These bits must be the
same for both the master and slave SPI devices. The clock polarity bit, selects
either an active high or active low clock but does not affect transfer format. The
clock phase bit selects the transfer format. See Figure 81.
218
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
SPI_SSn
SPI_SCK
CPOL=0, CPHA=0
SPI_SCK
CPOL=0, CPHA=1
SPI_SCK
CPOL=1, CPHA=0
SPI_SCK
CPOL=1, CPHA=1
SPI_MISO
SPI_MOSI
MSB
6
5
4
3
2
1
LSB
Internal Strobe for Data Capture
Figure 81. SPI signal timing, including the various SCLK phases
17.3.1.2.
Slave Selects in Master Mode
Multiple peripheral slave devices can be accessed by the chip, using GPIO pins as
chip selects. The chip select pins must be configured as GPIO outputs by the onchip DSP. The system code should contain a setup routine that sets each chip select
to non-asserted (high). Before any peripheral is accessed its chip select must be
asserted (low). The chip select should be de-asserted again before another peripheral is accessed. The SPI_SSn pin should not be in a low state while the SPI is used
or a mode fault will occur.
17.3.2.
Slave Mode
In slave mode the master device (another chip) generates the clock and slave select
for the SPI block. The SPI block will only receive and transmit data on the
SPI_MISO and SPI_MOSI lines when the SPI_SSn input is asserted low. In this
mode the SPI_SSn line is controlled by the SPI peripheral block.
To initialize the SPI into slave mode, the SPI interrupts are enabled (SPIE=1) and
the Master Mode is left off (MSTR=0). The clock polarity is set and finally the system
is enabled (SPE=1). The master device should start sending a slave select, clock
pulses data. After each byte of data is transferred the SPI will interrupt the processor
to receive the incoming data and send a new transmit byte.
5-35xx-D1-1.08-061705
219
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
18. TIMERS
Four timers are implemented, TIO0-TIO3. Two are connected to off-chip pins (TIO0,
TIO1) and the others are used for on-chip functions only. The timers are independently configured through PIO mode registers mapped to the X memory space at
base address X:$F100, X:$F140, X:$F180, and X:$F1C0. All timers are identical
(except for external connections) and provide the function shown in Figure.
Y-BUS
P-BUS
X-BUS
DSP CORE
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
Timer I/O
= = 0,
etc.
24 bit Count
Timer Out
running count
mode
controller
up, down, load, copy
Timer In
dclk_by_2
Divide
by 2
tick
Divide by
4
16
64
24MHz
XTAL
Osc.
either
PLL
Figure 82. One of Four Timers
18.1. Using the Timer Modules
18.1.1.
General Programming Guidelines
If you are re-starting a timer from scratch, all relevant fields of the Timer Count and
Timer Control registers should be re-programmed.
220
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
The Timer Interrupt Service Routine resets the TimerStatus bit in the Control Register below. The timer will wait until it has expired to assert the interrupt. Otherwise,
the Timer Interrupt Service Routine would run continuously.
Throughout the time when an interrupt occurs, the internal, software-invisible running counter within the module, continues to run (otherwise the timer events will not
occur at periodic intervals). Thus, if an interrupt is not serviced in a timely fashion, it
can become lost - i.e., it merges with another interrupt event before the handler can
get to it.
In the default timer clock mode, the timer counts time in terms of DSP clocks. Since
the DSP clock frequency can change based on system activity, this means that the
delays in the timer will adjust accordingly. To allow the timers to operate independently of the DSP clock speed, use one of the crystal clock modes controlled by the
TimerMode field of the Timer Control register. Note, that when using one of the crystal clock modes, the DSP clock must still be running at a minimum of twice the clock
source rate for correct operation. This means that the DSP clock must be running at
a speed of at least crystal clock divided by 2 (or 12 MHz when using a 24.0 MHz
crystal) when in crystal clock/4 mode. Similarly DSP clock must run at a speed of at
least crystal clock/8 (3.0 MHz) when in crystal clock/16 mode, and must run at a
speed of at least crystal clock/32 (750kHz) when in crystal clock/64 mode.
5-35xx-D1-1.08-061705
221
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
18.1.2.
Software-Visible Programmable I/O (PIO) Register
All register reserved fields should be written with zeroes.
18.1.2.1.
Timer Control Register
This is the module's control and status register. Note that the timer can be enabled
(TimerEnable bit) and configured in the same PIO write to this register
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
TIMER_ENABLE
1
8
TIMER_INT_ENABLE
1
9
INVERT
2
0
TIMER_CONTROL
2
1
TIMER_MODE
2
2
CLKGT
2
3
X:$F100
X:$F140
X:$F180
X:$F1C0
TIMER_STATUS
HW_TMR0CSR
HW_TMR1CSR
HW_TMR2CSR
HW_TMR3CSR
Table 280. HW_TMR*CSR
BITS
23
LABEL
RW RESET
CLKGT
RW 0
22:10 RSRVD
9:8
TIMER_MODE
R
0
RW 00
7
TIMER_STATUS
R
0
6
RSRVD
R
0
DEFINITION
Clock gate – Used to disable the clocks to the timer module to
conserve power when the timer is not in use. Must be set to 0 before
writing to any other timer registers.
0
Clocks not gated, i.e. normal operation
1
Clocks are gated, i.e. clocks disabled.
Reserved – Must be written with 0.
Timer clock mode
00 - Clock source = DSP clock/2
01 - Clock source = crystal clock/4 (requires a DSP clock of at
least crystal clock /2)
10 - Clock source = crystal clock/16 (requires a DSP clock of at
least crystal clock /8)
11 - Clock source = crystal clock/64 (requires a DSP clock of at
least crystal clock /32)
NOTE: Be sure to set HW_CCR_XTLEN to use the timers in crystal
clock modes.
Timer status notification The TimerStatus bit is set to one in different
ways depending on the value in TIMER_CONTROL field. See the
sections below which describe the conditions that set the
TIMER_STATUS bit to one. The TIMER_STATUS bit is reset to zero as
a side effect of reading this register.
Reserved – Must be written with 0.
Table 281. Timer Control Register Description
222
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
RW RESET
5:3
TIMER_CONTROL
RW 000
2
1
Invert
TIMER_INT_EN
RW 0
RW 0
DEFINITION
Timer Control (mode)
000 internal clock, downcount, no output
001 internal clock, downcount, output pulse
010 internal clock, downcount, output toggle
011 (reserved)
100 internal clock, external pulse width measurement
101 internal clock, external period measurement
110 external clock, upcount
111 external clock, downcount
Edge/level inversion selection
Timer interrupt enable
Note: 1. The Interrupt must also be enabled in the Interrupt
Collector.
2.
0
TIMER_ENABLE
HW_CCR XTLEN must be set to one for some timer
modes.
3. Set HW_TMR*CSR_TIMER_CONTROL and
HW_TMR*CSR_MODE before enable interrupts.
Timer enable
RW 0
Table 281. Timer Control Register Description (Continued)
18.1.2.2.
Timer Count Register
This register is used both to program initial count values and to monitor ongoing
counts, depending on the timer mode.
HW_TMR0CNTR
HW_TMR1CNTR
HW_TMR2CNTR
HW_TMR3CNTR
2
3
2
2
2
1
2
0
1
9
X:$F101
X:$F141
X:$F181
X:$F1C1
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
COUNT
Table 282. HW_TMR*CNTR
BITS LABEL RW RESET
23:0
COUNT RW
not
reset
DEFINITION
Count value expressed as the number of periods specified by the TimerMode field of
the HW_TMR0CSR/HW_TMR1CSR/HW_TMR2CSR/HW_TMR3CSR registers.
WARNING: these register are not reset to a known value.
Table 283. Timer Count Register Description
18.1.3.
Timer Modes
18.1.3.1.
Internal Clock Decrement, No Output Clock
(TimerControl = 000)
With the timer enabled (TimerEnable = 1), the running counter is loaded with the
value contained in the Count register. The running counter is decremented every two
system clock cycles (dclk/2). During the dclk cycle following the cycle when the run-
5-35xx-D1-1.08-061705
223
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
ning counter decrements to zero, the TimerStatus bit is set and the module asserts an
interrupt (tio_interrupt output) if the TimerIntEnable field is one. Upon reaching zero,
the running counter is automatically reloaded with the value from the Count register
and the process is repeated until the timer is disabled (TimerEnable = 0).
18.1.3.2.
Internal Clock Decrement, Output Pulse
(TimerControl = 001)
This mode is a super-set of the Internal Clock, No Output mode (TimerControl =
000). In addition to its functionality, this mode causes the TIO pin to be pulsed by the
timer. The duration of the pulse is two dclk cycles. The pulse begins in the dclk cycle
after the running counter has reached zero. The Invert bit determines the polarity of
the output pulse. If Invert is zero, the TIO pulses high for two cycles; if Invert is one,
TIO pulses low for two cycles.
18.1.3.3.
Internal Clock, Output Toggle
(TimerControl = 010)
This mode is a super-set of the Internal Clock, No Output mode (TimerControl =
000). In addition to its functionality, this mode causes the TIO pin to be toggled by
the timer. When the chip is reset or when the timer is disabled (TimerEnable zero),
the state of the TIO output is identical to the Invert programming field. A toggle of
the TIO output occurs off the dclk rising edge after the running counter has reached
zero.
18.1.3.4.
External Pulse Width Measurement
(TimerControl = 100)
Pulse Width
kick
input
clr_running_cnt
inc_running
copy_running_cnt
Figure 83. Pulse Width Measurement Mode
In this mode the timer samples an external event/clock on the TIO pin. The internal
sampling clock runs at half the internal clock frequency, or dclk/2. This mode mea-
224
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
sures the duration of high or low pulses on TIO, in terms of dclk/2 periods. If the
Invert bit in the TimerControl register is zero, a TIO-high pulse is measured. If the
Invert bit is set, TIO's low phase is measured. When the timer is enabled, the running counter is held at zero until a leading timing edge is detected on TIO. Thus if
the timer is enabled in the during TIO's active phase, no counting begins (this eliminates start-up run counts). At the count-terminating edge of TIO, the running count
value is transferred to the Count PIO register, the TimerStatus bit is set, the running
count register internal to the timer is cleared and an interrupt is optionally asserted
(if the interrupt enable is set). The running count register re-starts on the next leading edge of TIO.
Note that this and the following modes should only be used on the externally accessible timers, TIO0 and TIO1.
18.1.3.5.
External Period Measurement
(TimerControl = 101)
In this mode the timer samples an external event/clock on the TIO pin. The internal
sampling clock runs at half the internal clock frequency, or dclk/2. This mode measures the period TIO, in dclk/2 periods. The user programs an initial count value in
the Count PIO register. When the timer is enabled, this count value is transferred to
the timer's running count register. The running count register runs continuously, as
long as the timer is enabled, overflowing to zero without event. At each significant
edge of TIO (rising edge if the Invert bit is zero; falling edge otherwise), the running
count is written to the count register, the TimerStatus bit in the Control register is set
and the timer will optionally interrupt (if the TimerIntEnable bit is set). The user program can read successive values in the Count register to get a running count of
TIO's clock period, in dclk/2 clock cycles. A user program must discard the first count
value, since, unlike the pulse width mode, the timer is enabled at an arbitrary point in the
TIO clock cycle.
18.1.3.6.
External Clock Increment
(TimerControl = 110)
In this mode, the timer counts external clock cycles. When the timer is enabled (TimerEnable = 1), the one's complement of the value in the Count register is transferred
to the running count register. The running counter is will increment by transitions on
the relevant TIO0 or TIO1 pin. If the Invert bit is zero, the running count register
increments on 0-to-1 transitions of the TIO0 or TIO1 pin; otherwise it increments on
1-to-0 transitions. At each update of the running count register, the software-visible
Count register is also updated. The running counter counts continuously through its
overflow condition (all ones to all zeroes). At the transition after the running counter
has reached zero, the TimerStatus bit is set; and if interrupts are enabled, an interrupt is asserted.
18.1.3.7.
External Clock Decrement
(TimerControl = 111)
In this mode, the timer counts external clock cycles. When the timer is enabled (TimerEnable = 1), the value in the Count register is transferred to the running count register. The running counter is decremented by transitions on the relevant TIO0 or
TIO1 pin. If the Invert bit is zero, the running count register decrements on 0-to-1
transitions of the relevant TIO0 or TIO1 pin; otherwise it decrements on 1-to-0 transitions. On the transition after the running count register has reached zero, the Tim-
5-35xx-D1-1.08-061705
225
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
erStatus bit is set, the tio_interrupt is asserted (if the TimerIntEnable bit is set) and
the running count register is re-loaded with the value from the Count register.
18.1.4.
AC Timing Considerations
When a timer module counts off an external clock, its running counter increments
every other dclk cycle. Design reuse methodology forces an external clock (TIO)
sampling scheme whereby the external clock is sampled into the dclk domain and
its edges are then detected. This combination of Nyquist and dclk/2 counting means
that the external clock must run at most one fourth as fast as the system clock, or at
<=dclk/4 frequency. In practice, the external clock frequency needs to be even lower
than this because of the timer interrupt service or polling overhead. For instance if
the timer is placed in a mode to measure the pulse width of an external clock or
event, the period of such events must be long enough to give the interrupt handler
enough time to read the timer's Count register and record the sampled count.
226
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
19. SDRAM INTERFACE
The SDRAM interface is a DSP configurable interface to external SDRAM. The
interface will connect to various combinations of SDRAMs. The addressing scheme
provided by the interface will allow connections to 64 Mbit, 128 Mbit, 256 Mbit, or
512Mbit SDRAMs. The data connection allows only 8-bit transfers. The rate of data
transfers and all SDRAM activities is programmable from a divide by one to fifteen
of the system clock. The SDRAM timing specifications are also programmable
based on this rate. Read or write accesses to SDRAM data can be done through linear or modulo addressing of the system addresses and linear addressing of the
SDRAM addresses.
Y-BUS
X-BUS
P-BUS
DSP CORE
16KWord SRAM
16KWord SRAM
8KWord SRAM
16KWord SRAM
8KWord SRAM
16KWord SRAM
DMA-BUS
SDRAM
Programmable Registers
SDRAM
DMA Engine
sdram_we
sdram_cas
sdram_ras
sdram_dqm[0]
sdram_data[7:0]
sdram_clk
sdram_cke, ckb
sdram_bank[1:0]
sdram_addr[13:0]
SDRAM Controller
x8 SDRAM
Figure 84. SDRAM Interface
Internally the SDRAM interface will transfer data to/from the system X/Y/P memory
(aka system memory) only via DMA and is accessible in a linear or modulo address
mode. Data transfers can be programmable to various combinations of data packing
and unpacking to/from system memory.
5-35xx-D1-1.08-061705
227
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
In 8-bit transfers of SDRAM data to/from system memory (24-bit bus), the interface
is programmable to pack the data into 3-byte words, 2-byte words, or 1-byte words
during reads from SDRAM and is programmable to unpack with the same capability
as during writes to SDRAM. Transferring 2-byte words into system memory can be
programmed to be left aligned or right aligned to the 24-bit memory. When left
aligned the least significant byte is zeroed and when right aligned the most significant byte can be zeroed or sign extended. Transferring 1-byte words into a 24-bit
system memory will result in the data residing in the most significant byte and other
two bytes will be zeroed. The interface can be programmed to pick the byte to start
transferring data to/from system memory. Writes to SDRAM can be configured to
read from the system memory in a Big Endian or Little Endian (default) pattern.
sdram_clk
sdram_cke
sdram_ras
sdram_cas
sdram_we
sdram_cs
sdram_addr[13:0]
sdram_bank[1:0]
sdram_dqm[0]
sdram_data[7:0]
DESCRIPTION
Inverted system clock to SDRAM
SDRAM clock enable to SDRAM
SDRAM row address strobe
SDRAM column address strobe
SDRAM write enable
SDRAM chip select
SDRAM multiplexed column and row addresses
SDRAM bank selection code
SDRAM data mask
SDRAM data byte for 8 bit interface, lower data byte for 16 bit
interface.
Table 284. SDRAM External Signal Pins
The SDRAM controller state machine generates one of six cycle types
• Row Precharge is used to deactivate the open row in a particular bank or the
open row in all banks. The SDRAM controller precharges all banks at the same
time. The banks will be available after a precharge interval tRP. The controller
drives sdram_addr[10] to a one during the precharge command to force all banks
to precharge at once. Once a bank is precharge, it is in an idle state and must be
activated before it can be read or written.
• Row Activate is used to open (or activate) a row in a particular bank for
subsequent read or write access. Sdram_bank[1:0] select the bank to be
activated while the sdram_addr[13:11] and sdram_addr[9:0] select the row to be
activated or opened. This row remains open for accesses until a precharge
command is issued to that bank. A precharge command must be issued before
opening a row in the same bank. Since the SDRAM controller works with all
banks at once, it precharges all banks then opens the same row in each bank for
access.
• Row Refresh causes the SDRAM to read a row from its DRAM array and write it
back to make up for any charge leakage in the DRAM cells. This is really an auto
refresh command so the address and bank address busses are don’t cares for
this cycle. It must be issued to each row in the SDRAM within a refresh period.
The SDRAM chip keeps track of the address but the controller has to schedule it.
Since the controller runs on the system DCLK which is quite variable, users must
be careful to program the controller to keep the SDRAM refresh cycles within the
manufacturers specifications.
• The Load Mode Register command is issued once after reset to place the
SDRAM in the proper CAS latency mode, etc. For the STMP35xx, the value
228
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
written to the SDRAM mode register is fixed at $0027 which specifies sequential
full page burst mode with a CAS latency of two.
• Column Read commands are used by the DMA engine to load data from the
SDRAM to the on-chip system RAM. The SDRAM controller reads up to an entire
row across all accessed banks with strictly sequential column read commands.
• Column Write commands are used by the DMA engine to copy data from the onchip system RAM to the SDRAM. The SDRAM controller writes up to an entire
row across all accessed banks with strictly sequential column write commands.
The SDRAM controller works on all four banks of an SDRAM chip at a time. It precharges all of them for every transfer and activates all of them for every transfer. It
then goes into a page size burst read or write transferring data across all four banks.
This is true for 64Mb, 128Mb and 256Mb SDRAM. The following figure shows this
effect for the 64Mbit SDRAM case. If two 8 bit wide SDRAMs are attached then the
common bank size grows from 2048 byte “pages” to 4096 byte “pages”.
SDRAM controller prechareges,
and activates all banks to the
same row for each transfer.
SDRAM
BANK 0
SDRAM
BANK 1
SDRAM
BANK 2
SDRAM
BANK 3
4096 rows
x 512 byte column
4096 rows
x 512 byte column
4096 rows
x 512 byte column
4096 rows
x 512 byte column
sense amps
sense amps
sense amps
sense amps
Column Mux
for 64Mbit by 8 SDRAMs, the SDRAM controller
reads and writes 2048 bytes as a single row
Figure 85. SDRAM accesses all banks as a group
All flip flops int the SDRAM controller are clocked off of DCLK, the same clock used
for DMA and DSP functions. The clock sent off-chip to the SDRAM itself can be
divided down from the DCLK. In the discussion that follows, we will use sdram_clk
for the clock signal sent off chip. Refer to the four bit DIV field in Table 286,
“SDRAM Control Status Register Description,” on page 230 below.
This field
selects the number of DCLKs to count for one full sdram_clk cycle. The duty cycle is
50/50 for even cycles and has one additional DCLK in the down phase for odd
cycles. All time events in the SDRAM controller are synchronized to the DCLK
state where sdram_clk has just gone high. The sdram_clk is inverted as it is driven
off chip so that the SDRAM sees a clock that is exactly out of phase with the
sdram_clk used within the SDRAM controller. The controller has 16 bit refresh and
initialization and a four bit shared delay counter used to guarantee the various timing parameters of and SDRAM. Both of these counters are clocked by DCLK but
count the rising edge of the sdram_clk. In the discussion of the various timing register fields below, these two counters are used to implement the required programmable delays for each parameter. All SDRAM specifications give their required timing
parameters in terms of nano/micro seconds. All timing parameter fields in the
SDRAM controller are given in terms of divided DCLK counts. The DCLK period
5-35xx-D1-1.08-061705
229
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
depends on the settings in the clock control register, Table 35, “Clock Control Register Description,” on page 43. Obviously, if the SDRAM is used with more than one
setting of the DCLK frequency in an application then SDRAM timing parameter bit
fields will have to be modified.
Commands given to the SDRAM chips are encoded on the three interface signals
sdram_ras, sdram_cas, and sdram_we. On any rising edge of the SDRAM chips
clock, it captures these 3 lines and decodes them. When a read command is
encoded, the data is available a fixed number of clocks later, depending on how the
SDRAM’s mode register is set. For the STMP35xx, the mode register is always
loaded with a CAS latency of two which tells the SDRAM to return data on the second rising edge after a read command. Many of the timing parameters specified in
the SDRAM timing registers set the minimum time between various commands.
SDRAM No-op commands are used to fill the gaps and keep all the timing within
specification.
sdram_clk
command
NOP
PRE
REF
ACT
READ
sdram_addr
row
row
col
sdram_bank
bank
bank
bank
all banks
tRP
tRFC
tRCD
Figure 86. SDRAM Programmable Timing Parameters
19.1. SDRAM Interface Programmable Registers
SDRAM Control Status Register
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
IE
1
2
SDRAMEN
1
3
RNW
1
4
LM
1
5
KICK
1
6
ISAT
1
7
PWDN
1
8
SBYTE
MULTI
1
9
MEM
SIGN
2
0
BIGE
2
1
X:$F900
ASIZE
2
2
DIV
2
3
SDRAM
HW_SDRAM_CSR
UNKICK
19.1.1.
Table 285. HW_SDRAM_CSR
BITS
LABEL
RW RESET
23
22
SIGN
SDRAM
RW 0
RW 0
21
MULTI
RW 0
20:17 DIV
RW 0001
DEFINITION
Sign Extend bit for SDRAM right align reads.
External bus muxing control – SDRAM address, control, and data buses are
shared with the EMC5600. Setting this signal allows the SDRAM to have control.
Default is EMC control.
Multiple external SDRAM mode – This gives up control of the CSB signal after
initialization so that it can be controlled by the GPIO’s when there is multiple SDRAMs.
Clock Divide – Allows clock to SDRAM to be divisible from 1 to 15 of system clock.
Table 286. SDRAM Control Status Register Description
230
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
16
LABEL
RW RESET
UKICK
RW 0
15:13 ASIZE
RW 0
DEFINITION
Unkick – Writing a one will cleanly stop the current transfer. A transfer halted by
Unkick will not interrupt the DSP. After the current transfer stops successfully Unkick
will clear.
Align and Connection size – Controls how bytes or 2-byte words are read or
written into system memory. During write to the system memory the byte data can
packed onto the 24-bit bus as 3 bytes or 2 bytes or 1 byte. Packing 3 bytes will
default starting at the least significant byte. Packing 2 bytes can have two options:
left aligned or right aligned. Left aligned will result in the least significant byte
assigned equal to zero. The default starting byte is the middle byte (byte 1). Right
aligned will result in the most significant bytes assigned a sign extension of bit[15] of
the data if the sign extend bit the HW_SDRAM_CSR is set else the most significant
bytes will be zeroed. The default starting byte is the least significant byte (byte 0).
Packing 1 byte will place the data at the most significant byte of the bus and the
lower two bytes will be zeroed. During read from the system memory, data written to
the SDRAM will be taken in the same placement as previously described and the
zeroed and sign extended data will not be transferred to the SDRAM.
For 2-byte words (or 16-bit connection to SDRAM data bus) there is the two options
of left aligned or right aligned. For left aligned write to system memory the data will
occupy the two most significant bytes of the bus while the least significant byte will
be zeroed. For right aligned write to system memory the data will occupy the two
least significant bytes of the bus while the most significant byte will have the option
of sign extending bit[15] or zeroing. For reads from system memory the word data
transferred to the 16-bit SDRAM data bus will have the same alignment mentioned
above. The zeroed and sign extended data will not be transferred to the SDRAM. Bit
ASIZE2 will be used to differentiate between 8-bit or 16-bit data connection to
SDRAM.
12
BIGE
RW 0
11:10 MEM
RW 00
9:8
RW 00
SBYTE
ASIZE[2:0] = 000 8-bit connection, pack 3 bytes onto system bus (DEFAULT).
ASIZE[2:0] = 001 8-bit connection, pack 2 bytes onto system bus with left alignment.
ASIZE[2:0] = 010 8-bit connection, pack 2 bytes onto system bus with right
alignment.
ASIZE[2:0] = 011 8-bit connection, pack 1 byte onto system bus.
ASIZE[2:0] =100 16-bit connection, pack 2 bytes onto system with left alignment.
ASIZE[2:0] = 101 16-bit connection, pack 2 bytes onto system with right alignment.
Note: ASIZE[2] will differentiate between 8-bit or 16-bit connection.
Big Endian – Controls data is read from system memory and written to SDRAM
memory. In big endian mode the most significant byte is transferred first, followed
byte the next most significant bytes. Little Endian is the DEFAULT and the least
significant byte is transferred first, followed byte the next least significant byte. Only
is valid for 2 or 3 byte transfers.
Memory select – Chooses X/Y/P memory to write or read data.
MEM = 00 X memory
MEM = 01 Y memory
MEM = 10 P memory
Start Byte – Start read or write transfer on bytes 0 or 1 or 2. These control bits are
closely tied to the Align and Connection Size (ASIZE) control bits.
If ASIZE = 0 start byte can be 0,1,2
If ASIZE = 1 start byte can be 1,2
If ASIZE = 2 start byte can be 0,1
If ASIZE = 3 start byte can 2
If ASIZE = 4 start byte can 1
If ASIZE = 5 start byte can 0
Table 286. SDRAM Control Status Register Description (Continued)
5-35xx-D1-1.08-061705
231
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
RW RESET
7
6
RSRVD
PWDN
RW 0
RW 0
5
ISTAT
RW 0
4
LM
RW 0
3
KICK
RW 0
2
RNW
RW 0
1
0
IE
RW 0
SDRAMEN RW 0
DEFINITION
Reserved.
Powerdown – Will set the SDRAM into self refresh mode then shut off clocks to the
SDRAM and this interface.
Interrupt Status – Reading a one indicates pending interrupt, writing back a one will
clear interrupt and writing a zero has no effect.
Load Mode – loads value of LOAD MODE register into Mode register of SDRAM.
After the SDRAMMODE value is loaded into SDRAM LM will clear.
Kick writing one will start transfer – After successful completion kick will clear and
interrupt the DSP.
Read not write – writing one will read data from SDRAM to system memory, writing
zero will write data to SDRAM from system memory
Interrupt Enable
SDRAM Enable Bit – The SDRAM bit enables the SDRAM port. This bit must be set
after the DIV bits are set and any other SDRAM registers are written to. This allows
the divided clocks to the SDRAM to stabilize.
Table 286. SDRAM Control Status Register Description (Continued)
19.1.2.
SDRAM Type Register
HW_SDRAM_TYPE configures the interface to handle the SDRAM types. The different SDRAM configuration differ from each other by the number of address bits
used during column and row commands. This interface can access 64 Mb, 128 Mb,
and 256 Mb SDRAM or larger SDRAMs with 8 bit data busses.
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
ROW
2
3
X:$F90F
0
3
0
2
0
1
0
0
COLUMN
HW_SDRAM_TYPE
Table 287. HW_SDRAM_TYPE
BITS
LABEL
23:6 RSRVD
5:4
ROW
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
RW 0000
This field specifies the number of address bits in the Row Address
00 - ROW_WIDTH_11_0 twelve bit row address [11:0]
01 - ROW_WIDTH_12_0 thirteen bit row address [12:0]
10 - ROW_WIDTH_13_0 fourteen bit row address [13:0]
11 - reserved
Table 288. SDRAM Mode Register Description
232
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
3
RSRVD
2:0
COLUMN
RW RESET
DEFINITION
RW g
Reserved – Must be written with 0.
RW 0001
This field specifies the number of address bits in the Column Address
0000 - COL_WIDTH_7_0 eight bit column address [7:0]
0001 - COL_WIDTH_8_0 nine bit column address [8:0]
0010 - COL_WIDTH_9_0 ten bit column address [9:0]
0011 - COL_WIDTH_10_0 eleven bit column address [10:0]
0100 - COL_WIDTH_11_0 twelve bit column address [11:0]
0101 - COL_WIDTH_12_0 thirteen bit column address [12:0]
011x - reserved
NOTE: column address bit 10 is not used as an address bit in any JEDEC
SDRAM.
Table 288. SDRAM Mode Register Description
The SDRAM controller multiplexes the 29 bit SDRAM address onto the 14 bit multiplexed row/column address bus and the 2-bit bank address bus differently depending on the type of SDRAM that is connected to the STMP35xx. For example, a
64Mbit SDRAM has a different number of column address bits than a 16Mbit
SDRAM. The controller must adjust the address mapping appropriately. For the
64MBit “By 8” SDRAM there are 9 column address bits. So SDRAM address bits
[8:0] are muxed out as column addresses. The controller selects SDRAM address
bits [10:9] to drive out the sdram_bank[1:0] signals in this case, finally SDRAM
address bits [25:11] are muxed out as row address bits on sdram_addr[13:0]. Thus
one selects ROW_WIDTH_11_0 and COL_WIDTH_8_0 for a 64Mbit SDRAM.
SDRAM
TYPE
2
8
2
7
2
6
2
5
2
4
00,000
00,001
01,001
10,010
10,101
unused
unused
unused
unused
2
3
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
row[11:0]
row[11:0]
row[12:0]
row[13:0]
row[13:0]
bank
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
bank
column[7:0]
bank
column[8:0]
bank
column[8:0]
bank
column[9:0]
column[12:0]
Table 289. 29 Bit SDRAM Address Mapping to ROW/COLUMN
For a 16Mbit SDRAM, one loads the type register with ROW_WIDTH_11_0 and
COL_WIDTH_7_0, as shown in the table above. The largest SDRAM that could be
supported by the STMP35xx is 4Gbit (2^29 addressable locations, 8 bits wide). One
obtains this mode by setting the type register to ROW_WIDTH_13_0 and
COL_WIDTH_12_0.
Complicating all this muxing is the JDEC decision to reserve column address bit 10
to indicate auto precharge for the last cycle on a page. This leaves a hole in the mux
patterns as seen on the external pins. A thirteen bit column address
(COL_WIDTH_12_0) uses all fourteen bits of the multiplexed row/column address,
skipping over bit ten on the multiplexed address bus.
To set these fields, look at the SDRAM you have selected and determine the number of bits in its column address and row address and set the type register field
based on these values. Note that some configurations may define more address bits
than are actually used in the ROW address.
To connect a 128Mbit “By 8” Micron MT48LC16M8A2 SDRAM to the STMP35xx,
use ROW_WIDTH_11_0 and COL_WIDTH_9_0. To connect a 64 Mbit “By 8”
Micron MT48LC8M8A2 to the STMP35xx, use ROW_WIDTH_11_0 and
5-35xx-D1-1.08-061705
233
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
COL_WIDTH_8_0. NOTE: the STMP35xx only supports SDRAMs with exactly four
banks, thus two bank SDRAMS such as the 16Mbit “By 8” Micron MT48LC2M8A2
are not supported.
19.1.3.
SDRAM Address Pointer 1 Register
This register holds the lower 24 bits of the 29 bit SDRAM address. The column
address bits which are multiplexed out during CAS cycles always come from this
register. Similarly, the bank address bits always come from this register. The interface will select the appropriate portion of the row address from this register during
RAS cycles. At the end of a transfer, an incremented address will be placed in this
register.
HW_SDRAM_ADDR1 X:$F901
2
3
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ADDR_LOW
Table 290. HW_SDRAM_ADDR1
BITS
23:0
LABEL
RW RESET
ADDR_LOW RW 0
DEFINITION
Bits 23 through 0 of the 29 bit SDRAM address.
Table 291. SDRAM Address Pointer 1 Register Description
19.1.4.
SDRAM Address Pointer 2 Register
This register holds the upper 5 bits of the 29 bit SDRAM address. The interface will
select the appropriate portion of the row address from this register during RAS
cycles. At the end of a transfer, an incremented address will be placed in this register.
HW_SDRAM_ADDR2 X:$F902
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ADDR_HIGH
2
3
Table 292. HW_SDRAM_ADDR2
BITS
23:5
4:0
234
LABEL
RSRVD
ADDR_HIGH
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
RW $00
Bits 28 through 24 of the 29 bit SDRAM address.
Table 293. SDRAM Address Pointer 2 Register Description
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
19.1.5.
SDRAM System Address Pointer Register
SDRAM system address contains the system memory address. This is used with the
MEM field of the HW_SDRAM_CSR register to choose reading/writing from X/Y/P
memories. At the end of a transfer, the next address will be placed in this register.
HW_SDRAM_SYSADDR X:$F903
2
3
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ADDR
Table 294. HW_SDRAM_SYSADDR
BITS LABEL RW RESET
DEFINITION
23:16 RSRVD R
0
Reserved – Must be written with 0.
15:0 ADDR RW 0
DMA transfer address in on-chip SRAM.
Table 295. SDRAM System Address Pointer Register1 Description
19.1.6.
SDRAM Size Register
SDRAM size register contains the number of transfers to/from system memory
from/to the SDRAM.
HW_SDRAM_SIZE
2
3
2
2
2
1
2
0
1
9
1
8
X:$F904
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
SIZE
Table 296. HW_SDRAM_SIZE
BITS LABEL RW RESET
DEFINITION
23:18 RSRVD R
0
Reserved – Must be written with 0.
17:0 SIZE
RW 0
Number of transfers to make, i.e. number of bytes.
Table 297. SDRAM Size Register Description
19.1.7.
SDRAM Timer 1 Register
SDRAM timer register programs the delays as specified in the SDRAM specifications in relation to the number of cycles of SDRAM clocks. SDRAM clock is a
divided version of system clock. HW_SDRAM_TIMER1 programs the number of
SDRAM clock cycles for the initialization delay for SDRAM, delay from a precharge
command to the next command (tRP), and delay from refresh command to the next
command (tRFC).
HW_SDRAM_TIMER1 X:$F905
2
3
2
2
2
1
TRFC
2
0
1
9
1
8
1
7
TRP
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
INIT
Table 298. HW_SDRAM_TIMER1
5-35xx-D1-1.08-061705
235
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS LABEL RW RESET
DEFINITION
23:20 TRFC RW 6/$6
Refresh to next command – Value should be greater or equal to 70ns/(SDRAM
period).
19:16 TRP
RW 2/$2
Precharge to next command – Value should be greater or equal to 20ns/(SDRAM period).
15:0 INIT
RW 15000/$ Initialization of SDRAM from when SDRAM enable is activated in
3A98
HW_SDRAM_CSR[0]. Value should be greater or equal to 200us/(SDRAM period).
Table 299. SDRAM Timer 1 Register Description
The value in the INIT field is used to delay the initialization sequence after the controller is enabled (SDRAMEN bit in the HW_SDRAM_CSR) until all voltages are stabilized and the SDRAM clock has been running long enough to take the SDRAM
chip(s) through their initialization sequence. The inti sequence includes precharge
all banks to their idle state issuing an autorefresh cycle and then loading the mode
register. For most SDRAMs, this is specified as 200us. If we assume for the
moment that the DCLK is derived directly from the 24.0MHz crystal oscillator and
that the DIV field is set to 8 then we can calculate the value as follows:
The period of the sdram_clk is running at 1/8th the DCLK so it is running at 3.0MHz
with a period of 333.3ns. In this case a count of 600 is required to comply. Had the
sdram_clk been running at the DCLK (41.6ns), we set this field to 200us divided by
41.6ns or 4807.69. We round this value up to 4808 and load it into the INIT field.
19.1.8.
SDRAM Timer 2 Register
SDRAM timer register programs the delays as specified in the SDRAM specifications in relation to the number of cycles of SDRAM clocks. HW_SDRAM_TIMER2
programs the number of cycles in between the row and bank activate (ACTIVE)
command to the next command (tRCD), the number of cycles in between each
refresh commands (tREF/4096), and the number of cycles in between exiting low
power mode SELF REFRESH to the next ACTIVE command (tXSR).
HW_SDRAM_TIMER2 X:$F906
2
3
2
2
2
1
2
0
1
9
1
8
1
7
TRCD
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
TREF
0
6
0
5
0
4
0
3
0
2
0
1
0
0
TXSR
Table 300. HW_SDRAM_TIMER2
BITS LABEL RW
RESET
DEFINITION
23:20 RSRVD R
0
19:16 TRCD RW 2/$2
15:4
TREF
3:0
TXSR
Reserved – Must be written with 0.
ACTIVE to next command – Value should be greater or equal to 20ns/(system
period).
RW 375/$117 Refresh interval – Within SDRAM spec the refresh interval for 4096 rows is 64ms
thus a refresh must be done every 15us (64ms/4096). Thus this value should be
less than or equal to 15us/(system period).
RW 6/$6
Exiting self refresh mode to next command – Value should be greater or equal to
80ns/(system period).
Table 301. SDRAM Timer 2 Register Description
For TRCD, see Figure 86. “SDRAM Programmable Timing Parameters” on
page 230.
236
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
TXSR is specified from the time the SDRAM chip has been given its clock enable
and sdram_clk has been restarted before it can accept a new command. See the
SDRAM manufactures specifications for this value.
19.1.9.
System Memory Modulo Base Address Register
HW_SDRAM_BAR programs the base address for system modulo access. The
companion register is the DSP modulo register HW_SDRAM_MR. Once the system
access matches the address created by adding the HW_SDRAM_BAR and
HW_SDRAM_MR it returns to the HW_SDRAM_BAR address for the next access.
HW_SDRAM_BAR
2
3
2
2
2
1
2
0
1
9
1
8
X:$F907
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
BAR
Table 302. HW_SDRAM_BAR
BITS LABEL RW RESET
23:16 RSRVD R
0
15:0 BAR
RW 0
DEFINITION
Reserved – Must be written with 0.
Table 303. SDRAM System Memory Modulo Base Address Register Description
19.1.10. System Memory Modulo Register
HW_SDRAM_MR programs the modulo offset for system modulo access. The companion register is the DSP modulo base address register HW_SDRAM_BAR. The offset
value is added to the HW_SDRAM_BAR to the determine the limits of the modulo buffer.
HW_SDRAM_MR
2
3
2
2
2
1
2
0
1
9
X:$F908
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
MOD
Table 304. HW_SDRAM_MR
BITS LABEL RW RESET
DEFINITION
23:16 RSRVD R
0
Reserved – Must be written with 0.
15:0 MOD
RW $00FFFF Modulo offset 1-FFFE – If 0000 or FFFF is programmed system access will be
linear.
Table 305. SDRAM System Memory Modulo Register Description
5-35xx-D1-1.08-061705
237
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
19.1.11. SDRAM Transfer Count Register
HW_SDRAM_CNT holds the value of the number of transfers to or from the
SDRAM at the end of the transfer. (READ ONLY)
HW_SDRAM_CNT
2
3
2
2
2
1
2
0
1
9
1
8
X:$F90D
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
COUNT
Table 306. HW_SDRAM_CNT
BITS LABEL RW RESET
DEFINITION
23:16 RSRVD R
0
Reserved – Must be written with 0.
15:0 COUNT R
0
Number of transfers completed.
Table 307. SDRAM Transfer Count Register Description
19.1.12. SDRAM Mode Register
HW_SDRAM_MODE holds the value of the Mode register to be programmed into the
SDRAM during initialization. This register is to accommodate future SDRAM capabilities but is defaulted to a sequential full page burst mode with a CAS latency of two.
HW_SDRAM_MODE X:$F90E
2
3
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
VALUE
Table 308. HW_SDRAM_MODE
BITS LABEL RW RESET
DEFINITION
23:14 RSRVD R
0
Reserved – Must be written with 0.
13:0 VALUE R
$0027 Holds the value to be place into the Mode Register of the SDRAM.
Defaults to sequential full page burst mode with a CAS latency of 2
Table 309. SDRAM Mode Register Description
238
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
20. SWIZZLE
The SWIZZLE block is a DSP configurable module that is used to manipulate programmed I/O data or data within memory. One or two immediate words can be programmed into the module for manipulation. The resulting data can be read, after
various manipulations have been performed, in the SWIZZLE registers. SWIZZLE
features include Endianess, bit reversing, returning only specific bytes or words with
the data sign extended or zeroed, barrel shifting left or right, and division by the
fixed integer 3.
The SWIZZLE block also can be used to manipulate data within X/Y/P memory.
Data can be changed in place or changed then moved to a different memory location.
Y-BUS
X-BUS
P-BUS
DSP CORE
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
SWIZZLE
Programmable Registers
SWIZZLE
DMA Engine
USING DMA
USING PIO
DMA-BUS
swizzle block bit manipulation logic
Figure 87. Swizzle PIO & DMA data flows
For DMA operation, a memory to memory move paradigm is used, with a source
address pointer, a destination address pointer and a transfer size. Once kicked, the
source stream is read, the specified manipulations are performed and the resultant
data is written back to on-chip RAM.
In addition to the data manipulations available in previous generations of the SWIZZLE block, a divide by the fixed integer 3 is available. This division is only available
as a DSP PIO operation.
5-35xx-D1-1.08-061705
239
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
20.1. SWIZZLE Registers
20.1.1.
SWIZZLE Control and Status Register 1
SWIZZLE CSR1 controls the basic function of the module and the DSP programmable manipulation.
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0 0
1 0
LA
1
7
EN
1
8
LNR
1
9
SIGN
2
0
SHIFT
2
1
MEM
2
2
CLK_OFF
2
3
NEWADD
HW_SWIZZLE_CS1R X:$F380
Table 310. HW_SWIZZLE_CS1R
BITS
LABEL
RW RESET
23:11 RSRVD
R
0
10
NEWADD RW 0
9
CLK_OFF RW 0
8
MEM
RW 0
7:4
SHIFT
RW 0000
3
SIGN
RW 0
2
LNR
RW 0
1
LA
RW 0
0
EN
RW 0
DEFINITION
Reserved – Must be written with 0.
Place data into new memory location – If this bit is cleared, the value in the
HW_SWIZZLEDESTADDRR register is ignored.
0
Manipulate data in memory without moving it
1
Manipulate data in memory and put it in a new location
Turn clocks off – Clocks must be turned on before any other registers can be
accessed.
0
SWIZZLE clocks turned on
1
SWIZZLE clocks turned off
Manipulate data in memory – The SWIZZLE block can be used to manipulate
data written into the HW_SWIZZLEDATAR0 & HW_SWIZZLEDATA2R registers, or
can be used to manipulate blocks of data in on-chip memory. If this bit is clear, then
the data in the HW_SWIZZLESOURCER, HW_SWIZZLEDESTADDRR, &
HW_SWIZZLESIZER registers are ignored.
0
Manipulate data in registers
1
Manipulate data in memory
Barrel Shift from 0 to 15 – This field can be used for 16 bit shifts, however it is
recommend that all new programs use the new shift value in
HW_SWIZZLE_SIZER_NEW_SHIFT.
Sign extend data of pass_lsB,isB,msB modes
0
Data is not sign-extended
1
Data is sign-extended
Left barrel shift – Used in conjunction with the SHIFT field of this register.
0
Barrel shift to the right
1
Barrel shift to the left
Left Align data of pass_LSB, ISB, MSB modes – Default is the data is right
aligned with the 24-bit memory bus.
SWIZZLE enable
Table 311. SWIZZLE Control and Status Register 1 Description
240
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
20.1.2.
SWIZZLE Control and Status Register 2
SWIZZLE CSR2 is the rest of the controls to manipulate data in memory.
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0 0
1 0
KICK
1
2
SASEL
1
3
DESASEL
1
4
BIGE
1
5
BITREV
1
6
PLSB
1
7
PISB
1
8
PMSB
1
9
P16I
2
0
BS_EN
2
1
SBYTEDEST
2
2
UNKICK
2
3
P16L
HW_SWIZZLE_CS2R X:$F381
Table 312. HW_SWIZZLE_CS2R
BITS
LABEL
23:16 RSRVD
15
UNKICK
RW RESET
R
0
RW 0
14:13 SBYTEDEST RW 00
12
BS_EN
RW 0
11
P16I
RW 0
10
P16L
RW 0
9
PMSB
RW 0
8
PISB
RW 0
7
PLSB
RW 0
6
BITREV
RW 0
5
BIGE
RW 0
DEFINITION
Reserved – Must be written with 0.
Will halt memory swizzling – Will return to zero after successful halt. Address
of the next SWIZZLE will be located in HW_SWIZZLESOURCER and
HW_SWIZZLEDESTADDRR
Start Byte of Destination – When the destination is a different location.
SBYTEDEST can be used to place the word starting at an offset byte of the
destination. SBYTEDEST = 00 will place the word starting at the LSByte,
SBYTEDEST = 01 will place the word starting at the ISByte, and SBYTEDEST
= 10 will place the word starting at the MSBYTE of the destination.
Barrel shift enable – In memory swizzling mode this used in conjunction CSR1
SHIFT bits.
Pass Intermediate Significant word enable – In memory SWIZZLE mode this
will take the input word and only return the most significant and middle byte.
Used in conjunction with SIGN bit of CSR1.
Pass Least Significant word enable – In memory SWIZZLE mode this will
take the input word and only return the middle and least significant byte. Used
in conjunction with SIGN bit of CSR1.
Pass Most Significant byte enable – In memory SWIZZLE mode this will take
the input word and only return the most significant byte. The end location will
have this byte in the least significant byte position of the 24-bit word. Used in
conjunction with the SIGN bit of the CSR1.
Pass Intermediate byte enable – In memory SWIZZLE mode this will take the
input word and only return the middle byte. The end location will have this byte
in the least significant byte position of the 24-bit word. Used in conjunction with
the SIGN bit of the CSR1.
Pass Least Significant byte enable – In memory SWIZZLE mode this will take
the input word and only return the least significant byte. The end location will
have this byte in the least significant byte position of the 24-bit word. Used in
conjunction with the SIGN bit of the CSR1.
Bit Reversed enable – In memory SWIZZLE mode this will take the input word
and 24-bit reverse data to the resulting location.
Big Endian enable – In memory SWIZZLE mode this will take the input word
and change to Big Endian word.
Table 313. SWIZZLE Control and Status Register 2 Description
5-35xx-D1-1.08-061705
241
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
RW RESET
4:3
DESASEL
RW 00
2:1
SASEL
RW 00
0
KICK
RW 0
DEFINITION
Destination Memory select – Chooses X/Y/P memory to write or read data.
00
X space
01
Y space
10
P space
11
Reserved
Source Memory select – Chooses X/Y/P memory to write or read data.
00
X space
01
Y space
10
P space
11
Reserved
Kick bit – Writing one will start transfer. After successful completion kick will
clear.
Table 313. SWIZZLE Control and Status Register 2 Description (Continued)
20.1.3.
SWIZZLE Transfer Size
Number of words to be swizzled. This register also contains the new shift amount
field.
HW_SWIZZLE_SIZER X:$F382
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0 0
1 0
SIZE
NEW_SHIFT
2
3
Table 314. HW_SWIZZLE_SIZER
BITS
LABEL
RW RESET
23:21 RSRVD
R
000
20:16 NW_SHIFT RW $00
15:0
SIZE
RW $0000
DEFINITION
Reserved – Must be written with 0.
This five bit field can specify shift amounts up to 24 bits. The old shift field in
HW_SWIZZLE_CS1R_SHIFT is used when this field is set to zero. When
NEW_SHIFT is non-zero then it over-rides any value in the old shift field.
Number of words of memory to be manipulated by the SWIZZLE module.
Table 315. SWIZZLE Transfer Size Description
20.1.4.
SWIZZLE Source Address Register
Source address for memory manipulation mode. Used in conjunction with CSR1
Memory mode enable.
HW_SWIZZLE_SOURCER X:$F383
BITS LABEL RW RESET
23:16 RSRVD R
0
15:0 ADD
RW $0000
DEFINITION
Reserved – Must be written with 0.
Source address of the data in memory to be manipulated by the SWIZZLE module.
Table 316. SWIZZLE Source Address Register Description
20.1.5.
242
SWIZZLE DATA1 Register
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
DSP programmed data to be swizzled and placed on output SWIZZLE registers
HW_SWIZZLE_DATA1R X:$F384
BITS LABEL RW RESET
23:0
DAT
DEFINITION
RW $000000
Table 317. SWIZZLE DATA1 Register Description
20.1.6.
SWIZZLE DATA2 Register
DSP programmed data to be swizzled and placed on output SWIZZLE registers.
This is used only for resulting data for the HW_SWIZZLEPASSMSWR register.
HW_SWIZZLE_DATA2R X:$F385
BITS LABEL RW RESET
23:0
DAT
DEFINITION
RW $000000
Table 318. SWIZZLE DATA2 Register Description
20.1.7.
SWIZZLE Destination Address Register
SWIZZLE destination address in memory mode.
HW_SWIZZLE_DESTADDRR X:$F386
BITS LABEL RW RESET
23:16 RSVD
15:0 ADR
DEFINITION
R
0
RW $0000
Reserved
Destination address for memory based SWIZZLE operations.
Table 319. SWIZZLE Destination Address Register Description
20.1.8.
SWIZZLE Big Endian Register
Results of data programed into the HW_SWIZZLEDATA1R register. Used in conjunction with the memmode of the HW_SWIZZLECS1R.
HW_SWIZZLE_BIGENDIANR X:$F387
BITS LABEL RW RESET
23:0
DAT
DEFINITION
R
Table 320. SWIZZLE Big Endian Register Description
20.1.9.
SWIZZLE Bit Reversed Register
Results of data programed into the HW_SWIZZLEDATA1R register. Used in conjunction with the memmode of the HW_SWIZZLECS1R.
HW_SWIZZLE_BITREVR X:$F388
BITS LABEL RW RESET
23:0
DAT
R
DEFINITION
$000000
Table 321. SWIZZLE Bit Reversed Register Description
20.1.10. SWIZZLE Pass Least Significant Byte Register
Results of data programed into the HW_SWIZZLEDATA1R register. Used in conjunction with the memmode of the HW_SWIZZLECS1R.
5-35xx-D1-1.08-061705
243
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
HW_SWIZZLE_PASSLSBR X:$F389
BITS LABEL RW RESET
23:0
DAT
R
DEFINITION
$000000
Table 322. SWIZZLE Pass Least Significant Byte Register Description
20.1.11. SWIZZLE Pass Intermediate Significant Byte Register
Results of data programed into the HW_SWIZZLEDATA1R register. Used in conjunction with the memmode of the HW_SWIZZLECS1R.
HW_SWIZZLE_PASSISBR X:$F38A
BITS LABEL RW RESET
23:0
DAT
R
DEFINITION
$000000
Table 323. SWIZZLE Pass Intermediate Significant Byte Register Description
20.1.12. SWIZZLE Pass Most Significant Byte Register
Results of data programed into the HW_SWIZZLEDATA1R register. Used in conjunction with the memmode of the HW_SWIZZLECS1R.
HW_SWIZZLE_PASSMSBR X:$F38B
BITS LABEL RW RESET
23:0
DAT
R
DEFINITION
$000000
Table 324. SWIZZLE Pass Most Significant Byte Register Description
20.1.13. SWIZZLE Pass Least Significant Word Register
Results of data programed into the HW_SWIZZLEDATA1R register. Used in conjunction with the memmode of the HW_SWIZZLECS1R.
HW_SWIZZLE_PASSLSWR X:$F38C
BITS LABEL RW RESET
23:0
DAT
R
DEFINITION
$000000
Table 325. SWIZZLE Pass Least Significant Word Register Description
20.1.14. SWIZZLE Pass Intermediate Significant Word Register
Results of data programed into the HW_SWIZZLEDATA1R register. Used in conjunction with the memmode of the HW_SWIZZLECS1R.
HW_SWIZZLE_PASSISWR X:$F38D
BITS LABEL RW RESET
23:0
DAT
R
DEFINITION
$000000
Table 326. SWIZZLE Pass Intermediate Significant Word Register Description
20.1.15. SWIZZLE Pass Most Significant Word Register
Results of data programed into the HW_SWIZZLEDATA1R register. Used in conjunction with the memmode of the HW_SWIZZLECS1R and the HW_SWIZZLEDATA2R
register.
244
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
HW_SWIZZLE_PASSMSWR X:$F38E
BITS LABEL RW RESET
23:0
DAT
R
DEFINITION
$000000
Table 327. SWIZZLE Pass Most Significant Word Register Description
20.1.16. SWIZZLE Barrel Shift Register
Results of data programed into the HW_SWIZZLEDATA1R register. Used in conjunction with SHIFT field of the HW_SWIZZLECS1R register.
HW_SWIZZLE_BARRELR X:$F38F
BITS LABEL RW RESET
23:0
DAT
R
DEFINITION
$000000
Table 328. SWIZZLE Barrel Shift Register Description
20.1.17. SWIZZLE DIV3 Lower Register
Lower 24 bits of swizzle results from divide by three.
HW_SWIZZLE_DIV3L X:$F390
2
3
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0 0
1 0
DIV3_LOWER
Table 329. HW_SWIZZLE_DIV3L
BITS
23:0
LABEL
DIV3_LOWER
RW RESET
R
000
DEFINITION
Lower 24 bit result of dividing the 32 bit value in data reg 2 concatenated with
data reg 1 by the fixed integer 3.
Table 330. SWIZZLE Transfer Size Description
20.1.18. SWIZZLE DIV3 Upper Register
Remainder and upper 8 bits of swizzle results from divide by three.
HW_SWIZZLE_DIV3U X:$F391
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0 0
1 0
DIV3_UPPER
2
2
REMAINDER
2
3
Table 331. HW_SWIZZLE_SIZER
5-35xx-D1-1.08-061705
245
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
RW RESET
23:22 RSRVD
R
000
21:20 REMAINDER RW $00
19:8
7:0
DEFINITION
Reserved – Must be written with 0.
Two-bit remainder resulting from dividing the 32 bit value in data reg 2
concatenated with data reg 1 by the fixed integer 3.
Reserved – Must be written with 0.
Upper 8 bit result of dividing the 32 bit value in data reg 2 concatenated
with data reg 1 by the fixed integer 3.
RSRVD
R
$000
DIV3_UPPER RW $00
Table 332. SWIZZLE Transfer Size Description
The dividend for the fixed integer divide by three swizzle operation comes from the
concatenation of Data Register 2 with Data Register 1, as shown in Table 333.
3
1
3
0
2
9
2
8
2
7
2
6
2
5
2
4
2
3
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
DATA_REG_2[7:0]
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0 0
1 0
DATA_REG_1[23:0]
Table 333. SWIZZLE 32 bit divide by fixed integer 3
20.2. SWIZZLE Data Manipulation Examples
The following example shows how the SWIZZLE block can be used in a programmed I/O application. Data written to the Swizzle data registers and to the control and status registers causes nine different views of the source data to be
readable through the various pass register views, the big endian register view, and
the barrel shifter view. Data can also be modified in on-chip RAM in similar ways by
setting up the SWIZZLE DMA registers.
Example 1
Setting these register values
HW_SWIZZLEDATA1R
HW_SWIZZLEDATA2R
HW_SWIZZLECS1R
Yields these register values
HW_SWIZZLEBIGENDIANR
HW_SWIZZLEBITREVR
HW_SWIZZLEPASSLSBR
HW_SWIZZLEPASSISBR
HW_SWIZZLEPASSMSBR
HW_SWIZZLEPASSLSWR
HW_SWIZZLEPASSISWR
HW_SWIZZLEPASSMSWR
HW_SWIZZLEBARRELR
C
2
0
D
3
0
Example 2
A
0
0
B
1
0
E
4
0
F
5
1
A
0
0
E
F
0
0
0
0
0
0
A
F
C
D
A
7
B
3
D
0
0
0
E
0
0
0
C
0
0
0
A
0
C
D
E
0
A
B
C
0
4
5
A
B
C
D
E
No sign extension
No barrel shift
B
5
F
D
B
F
D
B
F
E
F
F
F
F
F
F
0
F
B
1
0
C
2
0
D
3
0
E
4
4
F
5
9
F
C
B
A
B
7
B
3
D
5
F
F
F
E
F
F
F
F
C
D
F
F
F
A
B
F
C
D
E
F
F
A
B
C
D
0
4
5
A
B
A
B
C
D
E
Sign extension on
Barrel shift to the right by 4 bits
Table 334. SWIZZLE Data Manipulations Examples
246
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
21. REAL-TIME CLOCK/ALARM/WATCHDOG RESET & PERSISTENT BITS
These features share a common source of one millisecond time pulses and utilize
persistent storage when the chip is in its powered down state. The one-millisecond
time base is derived from the 24.0MHz crystal oscillator, even during chip power
down states. The crystal oscillator power net is the only circuit powered up during
these chip-wide power-down states.
16KWord SRAM
Y-BUS
P-BUS
X-BUS
DSP CORE
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
RTC/WD/ALARM/PERSIST I/O
copy
controller
24 bit Shadow Persistent 1
24 bit Master Persistent 1
24 bit Shadow Persistent 0
24 bit Master Persistent 0
16 bit Shadow Divider
16 bit Master Divider
48 bit Alarm Shadow
48 bit Alarm Master
==
48 bit RTC Shadow Counter
24 bit Watchdog Counter
Alarm Event
48 bit RTC Master Counter
1KHz
variable divider
375KHz
fixed divide by 64
24MHz
XTAL
Osc.
Crystal
Power
&
Clock
Domain
Figure 88. RTC, Watchdog, Alarm, Persistent Bits
5-35xx-D1-1.08-061705
247
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
As a result, the real time clock count, alarm functions and persistent bit storage are
kept in the crystal oscillator clock and power domain. Shadow versions of these values are maintained in the DSP’s power and dclk clock domain when the chip is in a
power-up state. When the chip transitions from power-off to power-on, the master
values are copied to shadow values by the copy controller. Whenever software
writes to a shadow register then the copy controller copies the new value into the
master register.
Some of the persistent bits are used to control features which can continue to operate after power-down, such as the milli-second counter and the alarm function.
Other persistent bits are available to store application state information over power
downs. NOTE: the term power-down, as used refers to a state in which the DCDC
converter and various parts of the crystal power domain are still powered-up but the
rest of the chip is powered-down. If the battery is removed, then the persistent bits,
the alarm value and the milli-second counter value will be lost.
Immediately after reset it can take several hundred clocks for the copy controller to
complete the copy process from the analog domain to the digital domain. Software
can not rely on the contents of the milliseconds counter, alarm or persistent bits until
this copy is complete. Therefore, software must wait until all bits in the
HW_RTCCSR_STALE_REGS field have been reset to zero by the copy controller
before reading the initial state of these values.
HW_CCR_XTLEN
must be set to one
before using the RTC
rtc_first_init
call during initialization
HW_RTC_CSR_SOFT_RESET = 0;
return
rtc_second_init
remove soft reset, this
releses the copy controller.
call at some point in
initialization, after copy
controller has had time
to complete.
wait for copy controller
to complete.
NO
HW_RTC_CSR_
STALE_REGS == 0
YES
copy controller is finished,
so it's safe to write to
divider.
HW_RTC_DIVIDE_DIVIDEND = 375;
return
(24.0MHz /64)/375 = 1KHz
Figure 89. RTC initialization sequence
Before writing a new value to a shadow register, software must first confirm that the
corresponding bit of HW_RTCCSR_NEW_REGS is a zero. This will insure that a
value previously written to the register has been completely handled by the copy
state machine. Failure to obey this constraint could cause a newer updated value to
be lost.
248
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
rtc_write2master
how to write any
master register on
analog side. Do not
call until after
rtc_second_init.
TestValue = HW_RTC_CSR_NEW_REGS & WHICH_REG;
extract corresponding bit
NO
TestValue == 0
YES
wait for previous write
to this register to
complete
HW_RTC_DIVIDE = 375;
return
write new value to shadow
which will be automatically
copied to analog side. Use
this same technique for
milliseconds, alarm, or
persistent registers
Figure 90. RTC Writing to a Master Register from DSP
21.1. Real Time Clock
The real-time clock is a DSP-accessible, continuously running 48-bit counter that
increments every millisecond. A 48-bit millisecond counter has enough resolution to
count up to 8896 years with millisecond accuracy. The RTC will continue to count
time as long as a voltage is applied to the BATT pin, irrespective of whether the rest
of the chip is powered up. Reset has no effect on the RTC registers. This hardware
assumes a 24.0 MHz crystal is being used. If a different speed crystal is used, then
the times measured by this block should be scaled up or down according to the
crystal speed or the crystal oscillator divider should be changed.
Note that the hardware contains a fixed divide by 64 that generates the clock to the
RTC. For a 24MHz crystal this produces a 375KHz clock for use in the analog portion
of
the
RTC
hardware.
A
variable
divider
specified
in
HW_RTC_DIVIDE_DIVIDEND register determines how much to further divide this
value to obtain a 1millisecond increment signal for the RTC master milliseconds,
shadow milliseconds counter and the watchdog timer. Warning, the default value of
this divider register is not correct and must be changed at first power on.
For consistency across applications, it is recommended that the milli-second timer
should be referenced to January 1, 1980 at a 48-bit value of zero (same epoch reference as PC) in applications that use it as a time-of-day clock.
The register-set is composed of two 24-bit registers which must be read in a specific
order to prevent inconsistencies between the upper and lower words. Re-read the
upper word a second time (after reading the lower word) and compare to insure consistency. The upper data word (24-bits) must be written before the lower data-word
(24-bits). Consistency is protected for writes. Writing to the upper word freezes the
shadow counter until a corresponding write to the lower word occurs.
5-35xx-D1-1.08-061705
249
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
21.2. Watchdog Reset Registers
The watchdog reset is a DSP configurable device. Programmed by software to generate a chip wide RESET after HW_RTC_WATCHDOG milliseconds, the module
will generate this reset if software does not re-write this register before this time
elapses. The watchdog timer decrements once for every tick of the 1KHz clock generated by HW_RTC_DIVIDE_DIVIDEND, see Figure 88. “RTC, Watchdog, Alarm,
Persistent Bits” on page 247.
The WATCHDOG TIMER is initially disabled and set to count 16,777,215 milliseconds before generating a watchdog reset.
21.3. Alarm Clock
The alarm clock function allows an application to specify a future instant at which
the chip should be “awakened”, i.e. if powered down it can be powered up and the
DSP can be interrupted. The alarm clock setting is a DSP-accessible, 48-bit value
that is continuously matched against the 48-bit milli-second counter. When the two
values are equal, an alarm event is triggered. Persistent bits indicate whether an
alarm event should power-up the chip from its powered down state. In addition to or
instead of powering up the chip, the alarm event can also cause a DSP interrupt.
21.4. Real Time Clock Programmable Registers
This section describes the programmable registers of the real time clock, including
the watch dog registers, alarm registers and persistent registers.
21.4.1.
RTC Control and Status Register
Contains the control and status bits for the RTC.
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ALARM_EN
1
8
ALARM_INT
1
9
WATCHDOG_EN
2
0
STALE_REGS
2
1
CLKTUNE
2
2
FORCED_UPDATE
SOFT_RESET
2
3
X:$F500
NEW_REGS
HW_RTC_CSR
Table 335. HW_RTC_CSR
BITS
23
LABEL
SOFT_RESET
22:21 RSRVD
20
FORCE_UPDATE
19:17 RSRVD
RW RESET RESET
DEFINITION
1
2
RW 1
1
Set to zero to remove the soft reset and allow normal
operation.
R
000
000
Reserved – Must be written with 0.
RW 0
0
Set to one to force an update of all digital shadow registers.
R
000
000
Reserved – Must be written with 0.
Table 336. RTC Control and Status Register Description
250
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
16:12 NEW_REGS
11:9
CLKTUNE
8:4
STALE_REGS
3
2
RSRVD
WATCHDOG_EN
1
ALARM_INT
0
ALARM_EN
RW RESET RESET
DEFINITION
1
2
R
$00
$00
These read only bits indicate that more current data has
been written to a shadow register and that a write to the
persistent area is pending.
1xxxx = persistent register 1 is newer
x1xxx = persistent register 0 is newer
xx1xx = alarm count is newer
xxx1x = milli-seconds count is newer
xxxx1 = crystal divide register is newer
R
000
000
Each bit is connected to a delay line tap, setting the MSB
bypasses the other two switches, etc. The following skew
adjustments are approximate. Contact SigmaTel before
setting any of these bits.
000 -2.3ns
001 -729ps
01X -97ps
1XX +831ps
R
$1F
$00
These read only bits indicate that more current data is
available in the persistent area. An update of the shadow
register will be required before accurate data can be
obtained.
1xxxx = persistent register 1 is stale
x1xxx = persistent register 0 is stale
xx1xx = alarm count is stale
xxx1x = milli-seconds count is stale
xxxx1 = crystal divide register is stale
R
0
0
Reserved – Must be written with 0.
RW
Set to one to enable the watch dog timer function. If the
watch dog timer then counts down to zero, a chip wide reset
will be issued.
RW 0
0
The alarm interrupt bit is set to one if the alarm goes off on
the analog side. Write a one to this bit position to clear the
interrupt bit. When this bit is set to one and ALARM_EN is
set to one then interrupt 27 is signaled on vector $005E.
RW 0
0
Set this bit to one to enable an alarm interrupt if the alarm
“buzzes”.
Table 336. RTC Control and Status Register Description
5-35xx-D1-1.08-061705
251
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
21.4.2.
RTC Milli-Seconds 0 Register
Contains the lower data word (24 bits) of the RTC millisecond count
HW_RTC_MSECONDS0 X:$F501
2
3
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
RTC
Table 337. HW_RTC_MSECONDS0
BITS LABEL RW RESET
DEFINITION
23:0 RTC
RW 0
Bits [23:0] of the RTC millisecond counter – This field initializes to a random value,
and continues to count millisecond as long as the battery power is applied and the
crystal oscillator is powered up, regardless of whether the rest of the chip is powered
up or down. This register is a shadow of the master counter maintained in the crystal
clock/power domain. The master counter continues to count as long as battery power
is applied and the crystal oscillator is powered up. When power is restored to the rest
of the chip and the chip wide reset is removed, then the master value is copied over to
the shadow. At every milli-second tick thereafter, both counters increment by one in
their respective clock domains. NOTE: writes to HW_RTC_MSECONDS1 freeze the
shadow counter until a corresponding write to HW_RTC_MSECONS0 occurs.
WARNING: on STMP3501/3502, the milliseconds counter is cleared at every power
up state transition, causing a loss of the “real-time” epoch normally held in this counter.
Table 338. RTC Milli-Seconds 0 Register Description
21.4.3.
RTC Milli-Seconds 1 Register
Contains the upper data word (24 bits) of the RTC millisecond count
HW_RTC_MSECONDS1 X:$F502
2
3
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
RTC
Table 339. HW_RTC_MSECONDS1
BITS LABEL RW RESET
DEFINITION
23:0 RTC
RW 0
Bits [47:24] of the RTC millisecond counter – This field initializes to a random value,
and continues to count millisecond as long as the battery power is applied and the
crystal oscillator is powered up, regardless of whether the rest of the chip is powered
up or down. This register is a shadow of the master counter maintained in the crystal
clock/power domain. The master counter continues to count as long as battery power
is applied and the crystal oscillator is powered up. When power is restored to the rest
of the chip and the chip wide reset is removed, then the master value is copied over to
the shadow. At every milli-second tick thereafter, both counters increment by one in
their respective clock domains.
NOTE: writes to HW_RTC_MSECONDS1 freeze the shadow counter until a
corresponding write to HW_RTC_MSECONS0 occurs.
Table 340. RTC Milli-Seconds 1Register Description
252
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
21.4.4.
Watchdog Reset Count Register
This register contains the 24-bit delay (expressed in milliseconds) after which the
watchdog timer will reset the device when the watchdog timer is enabled. The watch
dog timer count value is not shadowed. After a power-up it is always reset to
$FFFFFF and the enable is set to zero.
HW_RTC_WATCHDOG X:$F503
2
3
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
WATCHDOG
Table 341. HW_RTC_WATCHDOG
BITS
LABEL
23:0 WATCHDOG
RW
RESET
RW $FFFFFF
DEFINITION
The number of milliseconds before a watchdog reset is initiated – This
counter only decrements when the watchdog timer is enabled.
Table 342. Watchdog Reset Count Register Description
21.4.5.
RTC Divider Register
This register contains the dividend which is used to divide the analog domain clock
to obtain a one milli-second “tick” for use in the real time clock milli-second counter
and the watch dog timer. The RTC contains a fixed divide by 64 ahead of
HW_RTC_DIVIDE_DIVIDEND. For a 24.0MHz crystal, this yields a 375KHz clock.
Setting HW_RTC_DIVIDE_DIVIDEND to decimal 375 ($177) yields a 1KHz tick.
375KHz = 24.0MHz / 64
1KHz = (24.0MHz / 64) / 375
HW_RTC_DIVIDE
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
DIVIDEND
2
3
X:$F506
Table 343. HW_RTC_DIVIDE
BITS
23:16
15:0
LABEL RW RESET
DEFINITION
RSRVD
R
0
Reserved – Must be written with 0.
DIVIDEND RW $5DC0 375KHz divider. WARNING: default value is wrong for all applications. Set this to
$0177 for a 24.0MHz oscillator.
Table 344. Watchdog Reset Enable Register Description
5-35xx-D1-1.08-061705
253
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
21.4.6.
RTC Alarm 0 Register
Contains the lower data word (24 bits) of the Alarm millisecond value.
HW_RTC_ALARM0
2
3
2
2
2
1
2
0
1
9
1
8
X:$F504
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ALARM0
Table 345. HW_RTC_ALARM0
BITS
LABEL
23:0 ALARM0
RW RESET
DEFINITION
RW 0
Bits [23:0] of the RTC alarm value – This register is a shadow of the master
alarm value maintained in the crystal clock/power domain. The master alarm value
retains its value as long as battery power is applied. When power is restored to the
rest of the chip and the chip wide reset is removed, then the master alarm value is
copied over to the shadow. At every milli-second tick thereafter, the master alarm
value is compared against the master milli-second counter. If there is an exact
match between all 48 bits of the master milli-second counter and all 48 bits of the
master alarm value then the alarm event occurs.
When a new value is written to the shadow register it is copied to the master alarm
value register. Always write the upper bits (ALARM1) first.
Table 346. RTC Alarm 0 Register Description
21.4.7.
RTC Alarm 1 Register
Contains the upper data word (24 bits) of the Alarm millisecond value.
HW_RTC_ALARM1
2
3
2
2
2
1
2
0
1
9
1
8
X:$F505
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ALARM1
Table 347. HW_RTC_ALARM1
BITS
LABEL
23:0 ALARM1
RW RESET
DEFINITION
RW 0
Bits [47:24] of the RTC alarm value – This register is a shadow of the master
alarm value maintained in the crystal clock/power domain. The master alarm value
retains its value as long as battery power is applied. When power is restored to the
rest of the chip and the chip wide reset is removed, then the master alarm value is
copied over to the shadow. At every milli-second tick thereafter, the master alarm
value is compared against the master milli-second counter. If there is an exact
match between all 48 bits of the master milli-second counter and all 48 bits of the
master alarm value then the alarm event occurs.
When a new value is written to the shadow register it is copied to the master alarm
value register. Always write the upper bits (ALARM1) first.
Table 348. RTC Alarm10 Register Description
254
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
21.4.8.
RTC Persistent Register 0
This register contains the 24 persistent bits of Persistent Register 0. These bits
retain their value through the various power phase of the DCDC converter controller.
As long as the battery is not removed, they retain their state.
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
ALARM_EN
2
0
ALARM_WAKE_EN
2
1
ALARM_WAKE
2
2
PERSISTENT
2
3
XTAL_PDN
HW_RTC_PERSISTENT0 X:$F507
Table 349. HW_RTC_PERSISTENT0
BITS
LABEL
23:4
PERISTENT
3
XTAL_PDN
2
ALARM_WAKE
RW RESET RESET
DEFINITION
1
2
RW 0
0
Additional Persistent Bits. These bits retain their value as
long as the battery remains connected.
RW 0
1
XTAL Power Down. Default state upon first power-up is to
turn off the crystal oscillator during subsequent power down
states of the DCDC converters. This bit affects the state of the
crystal oscillator when the chip is placed in its various powerdown states. During power up modes and entrance to such
states the crystal oscillator is automatically restarted. The
crystal power down bit, when set to one, removes the last
vestige of AC power when the chip is powered down. The
power switch and the DC circuits that monitor it are sufficient to
restart the chip. The alarm function can not be used when the
crystal is powered down.
NOTE: HW_DCDC_PERSIST_SLEEP_XTAL_ENABLE must
also be set to zero put the crystal to sleep.
WARNING: powerdown of the crystal also causes a reset
of the persistent bits to their default state.
RW 0
0
This bit is set to one whenever the ALARM causes a wake up
event. It can be reset by software by writing a zero directly to
the bit. If this bit is not reset before a subsequent power down,
then the part will immediately power up again as if a new alarm
had happened.
Table 350. RTC Persistent Register 0 Description
5-35xx-D1-1.08-061705
255
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
1
ALARM_WAKE_EN
0
ALARM_EN
RW RESET RESET
DEFINITION
1
2
RW 0
0
Set to one to wake up the chip from a power down state upon
the arrival of the alarm event. The alarm event must be
enabled by ALARM_EN and the crystal oscillator must not be
in its powered down state.
RW 0
0
Set to one to enable the alarm clock function. The alarm event
will occur whether the chip is in the powered-on or powered-off
state. In the powered off state, it will wake up the chip and
return it to a powered-on state, iff the ALARM_WAKE bit is
also set to one. The 48 bit alarm register is compared to the 48
bit milli-second counter, if an exact match occurs, and the
ALARM_EN bit is set to one then an alarm event is triggered.
Table 350. RTC Persistent Register 0 Description
256
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
21.4.9.
RTC Persistent Register 1
This register contains the 24 persistent bits of Persistent Register 1. These bits
retain their value through the various power phases of the DCDC converter controller. As long as the battery is not removed, they retain their state.
HW_RTC_PERSISTENT1 X:$F508
2
3
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
PERSISTENT
Table 351. HW_RTC_PERSISTENT1
BITS
LABEL
RW RESET
DEFINITION
23:0 PERSISTENT RW 000000 Additional Persistent Bits. These bits retain their value as long as the battery
remains connected, even if the chip is in one of its powered down states.
Table 352. RTC Persistent Register 1 Description
5-35xx-D1-1.08-061705
257
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
22. I2S SERIAL AUDIO INTERFACE
The chip includes a standard 3-channel I2S Serial Audio Interface (SAI). This allows
the chip to receive data from an external host or A/D or transmit data to an external
D/A. The chip must be configured as slave device on the I2S bus. Data can be transferred between the DSP core and the SAI by polling status flags or by servicing
interrupts.
X Peripheral Data Bus (xdb_per)
RCS
TCS
TO xdb_per
Status
0
Control 23
RX2 Shift Register
Status
Receiver
Controler
Status
Status
SDI1
Status
SDI0
SCKR
LRCKR
SDO2
TO xdb_per
23
0
TX1 Data RegisterR
TX1 Data RegisterL
Transmitter
23
0
Controler Control TX1 Shift Register
TO xdb_per
23
0
RX0 Data RegisterL
RX0 Data RegisterR
0
Control 23
RX0 Shift Register
TX2 Data RegisterR
TX2 Data RegisterL
0
Control 23
TX2 Shift Register
SDI2
TO xdb_per
23
0
RX1 Data RegisterL
RX1 Data RegisterR
0
Control 23
RX1 Shift Register
Status
FROM xdb_per
23
0
RX2 Data RegisterL
RX2 Data RegisterR
SDO1
TO xdb_per
23
0
TX0 Data RegisterR
TX0 Data RegisterL
0
Control 23
TX0 Shift Register
SDO0
SCKT
LRCKT
Figure 91. I2S Block Diagram
22.1. I2S External Pins
Pin #
Name
I2S_BCLK
I2S_WCLK
I2S_DataI0
I2S_DataI1
I2S_DataI2
I2S_DataO0
I2S_DataO1
I2S_DataO2
Description
Input, I2S serial bit clock
Input, I2S left/right word clock
Input, I2S input data 0
Input, I2S input data 1
Input, I2S input data 2
Output, I2S output data 0
Output, I2S output data 1
Output, I2S output data 2
I2S_SELECT=0
91/G3
92/H3
95/J1
94/H2
93/H4
90/G4
89/F3
88/F4
I2S_SELECT=1
NA/K3
NA/K1
NA/J2
94/H2
93/H4
90/G4
89/F3
88/F4
Note: The I2S_Select bit of the HW_SPARER register shown on page 46, controls
the pinout of these pins.
258
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
22.2. I2S Receive and Transmit Registers
22.2.1.
Receivers
The SAI contains three 2-channel receivers. These receivers shift serial data from
the I2S_DataIx pins using the bit and word clocks. It fills the data register
(SAIRX0,1,2R) until the word clock or RCS word length indicate that the word has
ended. The next word is placed in the other receiver channel.
After both words have been clocked into the data registers the Read Data Ready
(RDR) Flag is set. The DSP should read the data registers and clear the RDR flag
before the next word is ready to move from the shift register to the data register. The
DSP can either poll the RDR flag or receive an interrupt if the RXIE Bit is set.
Both channels of the read data registers are muxed into the same DSP address.
The first read will retrieve the data from the left channel data register. The second
will retrieve the right channel data. Subsequent reads will alternately retrieve the left
and right channel data.
22.2.1.1.
I2S Receive Status Control Register
This register is used both to program initial count values and to monitor ongoing
counts, depending on the timer mode.
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
REN0
1
1
REN1
1
2
REN2
1
3
RMME
1
4
RWL
1
5
RDIR
1
6
RLRS
1
7
RCKP
1
8
RREL
1
9
RXIE
2
0
RDWJ
2
1
RDR
2
2
ROFCL
2
3
X:$FFF0
ROFL
HW_SAIR_CSR
Table 353. HW_SAIR_CSR
BITS LABEL RW RESET
DEFINITION
23:17 RSRVD R
0
Reserved – Must be written with 0.
16
ROFCL RW 0
Receiver Data Overflow Clear – When this bit is set by the DSP both the ROFL and
RDR flags are cleared. This bit is implemented as a one-shot and always reads a zero.
Note: if ROFL is cleared in the last instruction of an Interrupt Service
Routine triggered by this bit, then the Interrupt will be executed twice.
15
RDR
R
0
Receiver Data Ready Flag – Indicates that both left and right data words have been
received. This bit is cleared when both left and right data are read from the receiver
data registers of all enabled receivers, or by writing a one to the ROFCL bit.
14
ROFL
R
0
Receiver Data Overflow – Indicates that an overflow condition was detected. This
condition occurs when the RDR Flag is set and a new data word is transferred from the
shift register to the receiver data register. Writing a one to the ROFCL bit clears this bit.
13
RSRVD R
0
Reserved – Must be written with 0.
12
RXIE
RW 0
Receiver Interrupt Enable for DSP – Interrupt enable for DSP.
Table 354. I2S Receive Status Control Register Description
5-35xx-D1-1.08-061705
259
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS LABEL RW RESET
DEFINITION
11
RDWJ RW 0
Receiver Data Word Justification – Determines which portion of the 24-bit portion
of the received 32-bit word will be transferred from the shift register to the data
register when programmed to receive 32 bit words.
0
The first 24 bits received are transferred to the data register.
1
The last 24 bits received are transferred to the data register.
When the receiver is programmed to receive 24 or 16 bit words this bit functions as follows:
0
Means that either the SCKR bit-counter (when it reaches 16 or 24) or a
LRCKR transition, whichever comes first, can transfer data to the data
registers.
1
Means that a LRCKR transition only will terminate the transfer and the last
16 or 24 bits received before the LRCKR transition gets transferred to the
data registers.
10
RREL
RW 0
Receiver Relative Timing – Determines the relative timing of the LRCKR signal as
referred to the serial data inputs.
0
The transition of LRCKR occurs together with the first bit data.
1
The transition of LRCKR occurs one SCKR earlier (I2S format).
9
RCKP RW 0
Receiver Clock Polarity – Defines the polarity of the receiver clock.
1
Positive Clock Polarity. Means the LRCKR and SDI lines change
synchronously with the positive edge of the clock, and are considered valid
during negative transitions.
0
Negative Clock Polarity. Means the LRCKR and SDI lines change
synchronously with the negative edge of the clock, and are considered valid
during positive transitions
8
RLRS
RW 0
Receiver Left Right Selection – Defines the polarity of the LRCKR.
0
If LRCKR = 0 then Left Word, if LRCKR = 1 then Right Word.
1
If LRCKR = 0 then Right Word, if LRCKR = 1 then Left Word.
7
RDIR
RW 0
Receiver Data Shift Direction – Determines the shift direction of the received data.
0
MSB First
1
LSB First.
6:5
RWL
RW 0
Receiver World Length Control – Selects the length of the data word being
received by the SAI.
00 16 Bits
01 24 Bits
10 32 Bits
11 Reserved.
4
RSRVD R
0
Reserved – Must be written with 0.
3
RMME RW 0
Receiver master mode enable
1
Receiver in master mode.
0
Receiver in slave mode.
Enable Master Mode for the receiver. i.e. Enables use of the Transmitter clocks from
the PLL to shift in data. A `1' in this bit configures the receiver into Master Mode.
2
REN2
RW 0
Receiver Enable – Enable bit for Receiver channel 2.
1
REN1
RW 0
Receiver Enable – Enable bit for Receiver channel 1.
0
REN0
RW 0
Receiver Enable – Enable bit for Receiver channel 0.
Table 354. I2S Receive Status Control Register Description (Continued)
22.2.1.2.
I2S Receive Status DataI0 Register
HW_SAIRX0R
BITS
23:0
LABEL
X:$FFF1
RW RESET
RECEIVE0 RW 0
DEFINITION
Data received from I2S_DATAI0
Table 355. I2S Receive Status DataI0 Register Description
260
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
22.2.1.3.
I2S Receive Status DataI1 Register
HW_SAIRX1R
BITS
23:0
LABEL
X:$FFF2
RW RESET
RECEIVE1 RW 0
DEFINITION
Data received from I2S_DATAI1
Table 356. I2S Receive Status DataI1 Register Description
22.2.1.4.
I2S Receive Status DataI2 Register
HW_SAIRX2R
BITS
23:0
LABEL
X:$FFF3
RW RESET
RECEIVE2 RW 0
DEFINITION
Data received from I2S_DATAI2
Table 357. I2S Receive StatusDataI2 Register Description
22.2.1.5.
Handling Different Word Lengths
It is possible that an external SAI and the chip’s SAI could be programmed to different size word lengths. The normal case is handled by terminating (performing a parallel load of the data register) the transfer when a LRCKR transition occurs.
However, when the chip’s SAI is programmed to receive 16 bit words and the external SAI is programmed to transmit 24 bit words (for example) the parallel load
should occur as soon as the 16 bits are serially shifted in. This implies that a bit
counter must be used to terminate the transfer.
Another case would be when the external SAI is programmed to transmit fewer bits
than the chip’s SAI is programmed to receive. For example, if the external SAI is
programmed to transmit 16 bits and the chip’s SAI is programmed to receive 24 bits
the transfer should be terminated by the LRCKR.
22.2.2.
Transmitters
Each I2S channel has a transmitter. The transmitter serially outputs data from the
left/right data registers (SAITX0,1,2R). Each I2S channel can carry two data streams
(i.e. audio channels). The data can be arranged in 16, 24 or 32 bit words. The chip’s
I2S transmitters must operate in slave mode (clocks generated by the master). The
transmit data must be loaded into the transmit registers by the DSP before the master begins clocking. After the master clock copies the second word into the transmit
shift register it sets the Transmit Data Empty (TDE) flag to signal the DSP for more
data. The Transmit Status/Control Register (TCS) configures the I2S transmitters.
The left and right channel transmit registers also share the same DSP memory. The
first word write will go to the left channel data register. The second will go to the
right. Subsequent rights will alternate between left and right data registers.
22.2.2.1.
I2S Transmit Status Control Register
This register is used both to program initial count values and to monitor ongoing
counts, depending on the timer mode.
5-35xx-D1-1.08-061705
261
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
TEN0
1
1
TEN1
1
2
TEN2
1
3
TMME
1
4
TWL
1
5
TDIR
1
6
TLRS
1
7
TCKP
1
8
TREL
1
9
TXIE
2
0
TDWE
2
1
TDE
2
2
TUFL
2
3
X:$FFF5
TUFCL
HW_SAITCSR
Table 358. HW_SAITCSR
BITS LABEL RW RESET
DEFINITION
23:17 RSRVD R
0
Reserved – Must be written with 0.
16
TUFCL RW 0
Transmitter Data Underflow Clear – When this bit is set by the DSP both the TUFL and
TDE flags are cleared. This bit is implemented as a one-shot and always reads a zero.
15
TDE
R
0
14
TUFL
R
0
13
12
11
RSRVD R
0
TXIE
RW 0
TDWE RW 0
10
TREL
RW 0
9
TCKP
RW 0
8
TLRS
RW 0
7
TDIR
RW 0
Note: if TUFL is cleared in the last instruction of an Interrupt Service Routine triggered
by this bit, then the Interrupt will be executed twice.
Transmitter Data Empty Flag – Indicates that both left and right data words have
been down loaded to the shift register. This bit is cleared by writing a one to the
TUFCL bit (as explained above).
Transmitter Data Underflow – Indicates that an underflow condition was detected.
This condition occurs if all the transmit data registers are not written to in time (i.e.
TDE is still set when the transmitter loads the shift register with new Left Data,
implying that old data is re-transmitted.) This bit is cleared by writing a one to the
TUFCL bit.
Reserved – Must be written with 0.
Transmitter Interrupt Enable for DSP – Interrupt enable for DSP.
Transmitter Data Word Expansion – Determines the way that the 24-bit data word
to be transmitted is expanded to 32 bits during transmission.
0
The last bit is transmitted 8 times AFTER transmitting the 24-bit data word
from the transmit data register.
1
The first bit is transmitted eight times BEFORE transmitting the 24-bit data
word from the transmit data register is transmitted eight times.
Transmitter Relative Timing – Determines the relative timing of the LRCKT signal
as referred to the serial data inputs.
0
The transition of LRCKT occurs together with the first bit data.
1
The transition of LRCKT occurs one SCKT earlier (I 2 S format).
Transmitter Clock Polarity – Defines the polarity of the Transmitter clock.
1
Positive Clock Polarity. Means the LRCKT and SDO lines change
synchronously with the positive edge of the clock, and are considered valid
during negative transitions.
0
Negative Clock Polarity. Means the LRCKT and SDO lines change
synchronously with the negative edge of the clock, and are considered valid
during positive transitions
Transmitter Left Right Selection – Defines the polarity of the LRCKT.
0
If LRCKT = 0 then Left Word, if LRCKT = 1 then Right Word.
1
If LRCKT = 0 then Right Word, if LRCKT = 1 then Left Word.
Transmitter Data Shift Direction – Determines the shift direction of the transmitted data.
0
MSB First
1
LSB First.
Table 359. I2S Transmit Status Control Register Description
262
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS LABEL RW RESET
DEFINITION
6:5
TWL
RW 00
Transmitter World Length Control – Selects the length of the data word being
transmitted by the SAI.
00
16 Bits
01
24 Bits
10
32 Bits
11
Reserved.
4
RSRVD R
0
Reserved – Must be written with 0.
3
TMME RW 0
Transmitter master mode enable – Reserved, must be set to 0.
2
TEN2
RW 0
Transmitter Enable – Enable bit for Transmitter channel 2.
1
TEN1
RW 0
Transmitter Enable – Enable bit for Transmitter channel 1.
0
TEN0
RW 0
Transmitter Enable – Enable bit for Transmitter channel 0.
Table 359. I2S Transmit Status Control Register Description (Continued)
22.2.2.2.
I2S Transmit Status DataO0 Register
HW_SAITX0R
BITS
23:0
LABEL
TRANSMIT0
X:$FFF6
RW RESET
RW 0
DEFINITION
Data to be sent I2S_DataO0 – Most significant bit is sent first. 16 bit transfer
data should be placed at most significant bytes (bits 23:16)
Table 360. I2S Transmit Status DataI0 Register Description
22.2.2.3.
I2S Transmit Status DataO1 Register
HW_SAITX1R
BITS
23:0
LABEL
TRANSMIT1
X:$FFF7
RW RESET
RW 0
DEFINITION
Data to be sent I2S_DataO1 – Most significant bit is sent first. 16 bit transfer
data should be placed at most significant bytes (bits 23:16)
Table 361. I2S Transmit Status DataI1 Register Description
22.2.2.4.
Transmit Status I2S DataO2 Register
HW_SAITX2R
BITS
23:0
LABEL
TRANSMIT2
RW RESET
RW 0
X:$FFF8
DEFINITION
Data to be sent I2S_DataO2 – Most significant bit is sent first. 16 bit transfer
data should be placed at most significant bytes (bits 23:16)
Table 362. I2S Transmit Status DataI2 Register Description
22.2.3.
Timing
Figure 92 shows the SAI timings for both the RDR and TDE flags. Internal flags
allow some leeway in reading and writing the data registers as shown on the diagrams. The incoming and outgoing data is shifted in/out through a buffer shift register. This gives a full LRCKR half-period to service the RDR interrupt and read the left
word received before it is overwritten by the next left word received. The right word
will not be overwritten until the following LRCKR transition. Similarly for transmit
there is a half-period of LRCKT from TDE being enabled to service the interrupt and
write the next left/right pair to be transmitted.
22.2.4.
Setting SAI Mode
To enable the pins into SAI mode, set the I2S_SELECT field of the HW_SPARER
register. Unless over-ridden by the GPIO, etc., the I2S interface is always available.
The I2S_SELECT bit, when set to one, invokes an alternate pin-out mode for channel 0 input in the larger 144-pin fpBGA package.
5-35xx-D1-1.08-061705
263
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
LRCKT
Left
Right
Left
Right
TDE
Internal Flag is set when left data is written to
all enabled transmitters. If this internal flag is
set then right data must be written to data
registers before the next falling edge of
LRCKT.
LRCKR
Left
TDE is cleared
when right data
is written to all
enabled
transmitters
Right
Left
Right
RDR
Internal Flag is set when left data is read
from all enabled receivers. If this internal flag
is set then right data must be read from data
registers before the next rising edge of
LRCKR.
RDR is cleared
when right data
is read from all
enabled
transmitters
Figure 92. I2S Receive and Transmit Data Timing
22.2.5.
Interrupts
The SAI interrupt vectors are located at the addresses shown in Table 363:
PRIORITY
Receiver Overflow
Transmitter Underflow
Receiver Data Ready
Transmitter Data Empty
PROGRAM ADDRESS
P:$0016
P:$0012
P:$0014
P:$0010
Table 363. SAI interrupts and Priorities
22.3. Known Chip Defects with SAI
22.3.1.
Clear Quest Entry STMP00003941
All revisions of the chip are known to incorrectly handle overflow and underflow
error reporting for the SAI block. When the RX data is not read in time, causing an
overflow condition the chip does not generate an Rx Overflow interrupt. Instead it
generates two RxReady interrupts back-to back.
When TX data is not written in a timely manner, causing an underflow condition, the
chip does not generate a TX Underflow interrupt. Instead it generates TX Empty
interrupts repeatedly until cleared by software.
264
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
23. GENERAL PURPOSE INPUT/OUTPUT (GPIO)
The chip contains several GPIO modules. These modules provide flexible software
control of each pin. Each digital pin can either be controlled by a hardware interface
(SPI, Flash, etc.) or by a GPIO module. Every digital pin on the chip except for
TESTMODE, ONCE_DRN, ONCE_DSI, ONCE_DSO and ONCE_DSK can be used
as a GPIO pin. The GPIO configuration registers control the pin connection (GPIO
or normal interface) and the pin function (if it is in GPIO mode). The GPIO configuration is independent for all of the pins. For example, the SmartMedia/NAND
SM_CE3n pin can be configured as a GPIO pin while the other SmartMedia/NAND
pins are connected to the Flash interface. If a pin is switched to be a GPIO, then the
internal module connected to that pin will see a logic 0 on that pin, irrespective of the
actual status of that pin.
The GPIO module is the only means of communicating with devices that don’t use
one of the chip’s hardware interfaces. For example, LED’s, buttons and other
peripherals will require software that uses pins in GPIO mode. The pins are also
used in GPIO mode for functions such as Key Scan, Backlights, SmartMedia card
insert, etc.
23.1. GPIO Interface
There are four GPIO Pin Registers (4 banks) on the chip used to configure digital
pins as GPIO or their designated function: GPIO0, GPIO1, GPIO2 and GPIO3 (See
Section 23.2.12. on page 273 for details). The following registers exist within each
of these four banks to configure the chips digital pins. Some pins only exist in the
144-pin package options - the registers that control these pins exist but perform no
useful function on the 100-pin devices.
GPIO Output Enable
HW_GPxDOER
OE#
Power
HW_GPxPWR
PERIPH_DOE
Assigned Hardware
Function Output Enable
Drive Strength
(4 or 8mA)
HW_GPx8MAR
HW_GPxENR
GPIO Output
PAD
HW_GPxDOR
PERIPH_DO
Assigned Hardware
Function Output
GPIO Input
HW_GPxDIR
PERIPH_DI
Assigned Hardware
Function Input
OUTPUT
INPUT
Gate Keeper (20-30uA Drive
Strength). Note that GPIOs
require 47k (or smaller) pull-ups
or pull-downs to overdrive Gate
Keepers.
HW_GPxPWR
AND gate ensures a
deterministic input value
Figure 93. GPIO PAD Diagram
5-35xx-D1-1.08-061705
265
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BEGIN
HW_GPxPWR
Turns pins ON/OFF. Pins must be turned on
irrespective of whether the pin is configured as a
GPIO or not. Set to 1 to power up the pin, which
can then be used as a GPIO input, a GPIO output,
or for assigned hardware functionality (SPI, Flash,
etc.). Set to 0 to power down and tri-state the pin.
HW_GPx8MAR
Sets Drive Strength for output pins. Note that
drive strength may be set regardless of whether
the pin is used as a GPIO output or as an
assigned hardware ouput.
HW_GPxENR
Enable/Disable pins as GPIOs. Set to 1 to use
the pin as a GPIO. Set to 0 to use the assigned
hardware functionality (SPI, Flash, etc.)
GPIO INPUT
GPIO OUTPUT
Set output value
for GPIO pin
Output Enable. Set to
1 to drive out the value
in HW_GPxDOR. Set
to 0 to use the pin as
an input or to tri-state
the pin
HW_GPxDOR
HW_GPxDOER
GPIO INPUT
w/o Interrupt
Output Enable. Set to 0 to
use the pin as an input.
GPIO INPUT
with Interrupt
HW_GPxDOER
END
HW_GPxILVLR
Set interrupt
level/edge
sensitivity
HW_GPxIPOLR
Set interrupt
polarity
HW_GPxIPENR
Select GPIO
pins that can
generate an
interrupt
HW_GPxIENR
HW_GPxDIR
Enable
interrupt
Read the current value
on the pin input
END
Figure 94. GPIO Setup Flow Chart
266
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
.
1
dsp_data[b]
HW_GPxISTATb
DSP write to GPIOx
edge
HW_GPxISTATb
from other
23 bits
GPIOxIRQ
to DSP
FALL
RISE
HW_GPxIPENRb
LOW
HIGH
HW_GPxISTATb
HW_GPxILVLb
HW_GPxIENRb
level
DIRb
HW_GPxIPOLRb
pin GPxBb
sync
HW_GPxDIRb
POLRb
Figure 95. GPIO Interrupt Generation
5-35xx-D1-1.08-061705
267
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
23.2. GPIO Registers
23.2.1.
GPIO Enable Register
This register enables most of the digital pins on the chip to be GPIOs or perform
assigned functionality (I2C, SPI, etc.).
HW_GP0ENR
HW_GP1ENR
HW_GP2ENR
HW_GP3ENR
X:$F400
X:$F410
X:$F420
X:$F430
BIT LABEL RW RESET
DESCRIPTION
23:0 EN
RW 0
Module Enable bits – These pins reset to 0.
0
Pin configured to perform assigned function
1
Pin configured as GPIO
Table 364. GPIO Enable Register Description
23.2.2.
GPIO Data Out Register
This register allows GPIO to send data out on a pin-by-pin basis. Always set this
register before GPIO Data Out Enable because as soon as the enable bit is set, the
current data on the pin is sent out.
HW_GP0DOR
HW_GP1DOR
HW_GP2DOR
HW_GP3DOR
X:$F401
X:$F411
X:$F421
X:$F431
BIT LABEL RW RESET
DESCRIPTION
23:0 DO
RW 0
Output bits if GPIO is set in GPIO Enable and the GPIO Output Enable bit is set.
Table 365. GPIO Data Out Register Description
23.2.3.
GPIO Data In Register
This register allows GPIO to input data on a pin-by-pin basis.
HW_GP0DIR
HW_GP1DIR
HW_GP2DIR
HW_GP3DIR
X:$F402
X:$F412
X:$F422
X:$F432
BIT LABEL RW RESET
DESCRIPTION
23:0
DI
R
0
Input bits if pin is in GPIO Mode and set as an input.
Table 366. GPIO Data In Register Description
23.2.4.
GPIO Data Out Enable Register
This register allows GPIO to enable data output on a pin-by-pin basis.
HW_GP0DOER
HW_GP1DOER
HW_GP2DOER
HW_GP3DOER
X:$F403
X:$F413
X:$F423
X:$F433
BIT LABEL RW RESET
23:0
DOE RW 0
Output enable bits
0
Input / Tri-State
1
Output
DESCRIPTION
Table 367. GPIO Data Out Enable Register Description
268
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
23.2.5.
GPIO Interrupt Pin Enable Register
This register allows GPIO to define specific pins to be interrupt pins.
HW_GP0IPENR
HW_GP1IPENR
HW_GP2IPENR
HW_GP3IPENR
X:$F404
X:$F414
X:$F424
X:$F434
BIT LABEL RW RESET
23:0 IPEN
RW
0
DESCRIPTION
Interrupt enable bits
0
Corresponding pin is not interrupt pin
1
Corresponding pin is an interrupt pin
Table 368. GPIO Interrupt Pin Enable Register Description
23.2.6.
GPIO Interrupt Enable Register
This register allows GPIO pins to enable specific interrupts to assert a DSP interrupt. Note that this register distinguishes between polling and interrupt driven routines. If a particular pin functions as an interrupt pin (as defined by GPIO Interrupt
Pin Enable register), it's assertion will be reported in the interrupt status register but
the interrupt signal to DSP will not be asserted if GPIO Interrupt Enable bit is also
not asserted. Once an interrupt has been received in the interrupt collector and it
has been determined that it is GPIO bank 0, 1, 2, or 3, then this register is used to
determine which pin asserted the interrupt. See Table 372 for Interrupt Options.
HW_GP0IENR
HW_GP1IENR
HW_GP2IENR
HW_GP3IENR
X:$F405
X:$F415
X:$F425
X:$F435
BIT LABEL RW RESET
23:0 EN
RW
0
DESCRIPTION
Interrupt enable bits
0
Interrupt Disable
1
Interrupt Enable
Table 369. GPIO Interrupt Enable Register Description
23.2.7.
GPIO Interrupt Level Register
This register allows GPIO to define specific pins to be level or edge sensitive if they
are enabled in GPIO Interrupt Enable register. See Table 372 for Interrupt Options.
HW_GP0ILVLR
HW_GP1ILVLR
HW_GP2ILVLR
HW_GP3ILVLR
X:$F406
X:$F416
X:$F426
X:$F436
BIT LABEL RW RESET
23:0 ILVL
RW 0
Interrupt level bits
0
Edge Sensitive
1
Level Sensitive
DESCRIPTION
Table 370. GPIO Interrupt Level Register Description
5-35xx-D1-1.08-061705
269
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
23.2.8.
GPIO Interrupt Polarity Register
This register allows GPIO to define specific pins to Active High/Low or Positive/Negative edge interrupt based on programming of corresponding bit in GPIO Interrupt
Enable and GPIO Interrupt Level register. See Table 372 for Interrupt Options.
HW_GP0IPOLR
HW_GP1IPOLR
HW_GP2IPOLR
HW_GP3IPOLR
X:$F407
X:$F417
X:$F427
X:$F437
BIT LABEL RW RESET
DESCRIPTION
23:0 IPOL
RW 0
Interrupt Polarity bits
0
Active low level / falling edge
1
Active high level / rising edge
Table 371. GPIO Interrupt Polarity Register Description
HW_GPXIENR
(INT ENABLE)
0
1
1
1
1
HW_GPXILVLR
(INT LEVEL)
X
1
1
0
0
HW_GPXIPOLR
(INT POLARITY)
X
0
1
0
1
DESCRIPTION
Disable Interrupt Pin Functionality
Active Low Interrupt
Active High Interrupt
Falling Edge Interrupt
Rising Edge Interrupt
Table 372. GPIO Interrupt Options Table
23.2.9.
GPIO Interrupt Status Register
This register allows GPIO
grammed as an interrupt
interrupt to DSP.
HW_GP0ISTATR
HW_GP1ISTATR
HW_GP2ISTATR
HW_GP3ISTATR
to monitor and clear interrupts if corresponding bit is propin. Combination of interrupt status and enables assert
X:$F408
X:$F418
X:$F428
X:$F438
BIT LABEL RW RESET
DESCRIPTION
23:0 ISTAT RW 0
Interrupt status bits – Reading 1 indicates a pending interrupt. The bits in interrupt
status register are set irrespective of Interrupt Enable bits. To clear an interrupt, set the
proper bit to 1. This method of clearing is needed for both edge and level interrupts, see
Figure 95.
Table 373. GPIO Interrupt Status Register Description
23.2.10. GPIO Pin Power Register
This register controls whether each pin is powered up irrespective of whether one
pin is configured as a GPIO pin or not. When a pin is powered down its input will
read as 0 irrespective of the voltage on the pin. Similarly, a pin that is powered down
will always be high impedance (i.e. tristated) even if the pin is configured as an output. All pins are powered down after a chip reset.
HW_GP0PWR
X:$F409
HW_GP1PWR
X:$F419
HW_GP2PWR
X:$F429
HW_GP3PWR
X:$F439
270
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BIT LABEL RW RESET
23:0 PWR
RW 0
5-35xx-D1-1.08-061705
DESCRIPTION
0
Pin is powered down (reset value)
1
Pin is powered up (i.e. active)
Table 374. GPIO Pin Power Register Description
271
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
23.2.11. GPIO Pin Drive Strength Register
This register controls the output drive strengths of groups of pins, when the pins are
configured as outputs. Note that unlike other GPIO registers, each pin does not get
its own control bit. Instead pins are assigned in groups of 8 to an individual control
bit. Each group of pins can be configured as 4mA drivers, or 8mA drivers. The
required setting will depend on the application, but in general it is recommended
that as many pins as possible are configured in 4mA mode. Doing so will increase
the number of simultaneously pins that can be supported without adverse effects on
the rest of the system, and will reduce the momentary peak load on the DCDC converters, and will reduce EMI emissions. A conservative limit on the maximum number of simultaneously switching output pins for the chip would be 24 4mA pins, or 12
8mA pins, or an equivalent combination of 4mA and 8mA pins.
It is possible to save some power by gating the clocks in the GPIO modules, this
functionality is controlled by the CLKGATE field of the GPIO Pin Drive Strength register. When the clocks are gated, the DSP will no longer be able to write to any of
the GPIO control registers, except for this one. When the clocks are gated, GPIO
interrupts will continue to be generated, and the DSP will have full read access to all
registers, including the GPIO Data Input registers. The DSP will need to clear the
CLKGATE bit before it can change the value of pin configured as a GPIO output.
HW_GP08MAR
HW_GP18MAR
HW_GP28MAR
HW_GP38MAR
X:$F40A
X:$F41A
X:$F42A
X:$F43A
BIT
LABEL
RW RESET
23
CLKGATE RW 0
22:4 RSRVD
2
PDS2
R
RW
0
0
1
PDS1
RW
0
0
PDS0
RW
0
272
DESCRIPTION
0 Clocks not gated
1 Clocks gated
Reserved – Must be written with 0.
Controls bits [23:15]
0
4 mA driver
1
8 mA driver
Controls bits [16:8]
0
4 mA driver
1
8 mA driver
Controls bits [7:0]
0
4 mA driver
1
8 mA driver
Table 375. GPIO Pin Drive Strength Register Description
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
23.2.12. GPIO Register Pin Assignments
23.2.12.1. GPIO0 (Bank 0)
GPIO0
BIT
LABEL
23:20
19
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
Note:
RSRVD
GP0B19
GP0B18
GP0B17
GP0B16
GP0B15
GP0B14
GP0B13
GP0B12
GP0B11
GP0B10
GP0B9
GP0B8
GP0B7
GP0B6
GP0B5
GP0B4
GP0B3
GP0B2
GP0B1
GP0B0
X:$F400–$F40A
PIN NAME (NOTE 1)
DESCRIPTION
Reserved – Must be written with 0.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
GP19
GP18
GP17
GP16
GP15
GP14
GP13
GP12
GP11
GP10
GP9
GP8
GP7
GP6
GP5
GP4
GP3
GP2
GP1
GP0
1. Please see Table 497. “General Purpose Input/Output Pins” on page 384.
Table 376. GPIO0 Pin Register (Bank 0) Description
23.2.12.2. GPIO1 (Bank 1)
GPIO1
BIT
23
22
21
20
19
18
17
16
15
14
13
12
11
10
LABEL
GP1B23
GP1B22
GP1B21
GP1B20
GP1B19
GP1B18
GP1B17
GP1B16
GP1B15
GP1B14
GP1B13
GP1B12
GP1B11
GP1B10
PIN NAME (NOTE 1)
GP47
GP46
GP45
GP44
GP43
GP42
GP41
GP40
GP39
GP38
GP37
GP36
GP35
GP34
X:$F410–$F41A
DESCRIPTION
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
Table 377. GPIO1 Pin Register (Bank 1) Description
5-35xx-D1-1.08-061705
273
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BIT
LABEL
PIN NAME (NOTE 1)
DESCRIPTION
9
GP1B9
GP33
This pin is available in all packages.
8
GP1B8
GP32
This pin is available in all packages.
7
GP1B7
GP31
This pin is available in all packages.
6
GP1B6
GP30
This pin is available in all packages.
5
GP1B5
GP29
This pin is available in all packages.
4
GP1B4
GP28
This pin is available in all packages.
3
GP1B3
GP27
This pin is available in all packages.
2
GP1B2
GP26
This pin is available in all packages.
1
GP1B1
GP25
This pin is available in all packages.
0
GP1B0
GP24
This pin is available in all packages.
Note: 1. Please see Table 497. “General Purpose Input/Output Pins” on page 384.
Table 377. GPIO1 Pin Register (Bank 1) Description (Continued)
23.2.12.3. GPIO2 (Bank 2)
GPIO2
BIT
23
22
21
20
19
18
17
16
15
14
13
12:9
8
7
6
5
4
3
2
1
0
Note:
LABEL
RSRVD
GP2B22
GP2B21
GP2B20
GP2B19
GP2B18
GP2B17
GP2B16
GP2B15
GP2B14
GP2B13
RSRVD
GP2B8
GP2B7
GP2B6
GP2B5
GP2B4
GP2B3
GP2B2
GP2B1
GP2B0
PIN NAME (NOTE 1)
GP70
GP69
GP68
GP67
GP66
GP65
GP64
GP63
GP62
GP61
GP56
GP55
GP54
GP53
GP52
GP51
GP50
GP49
GP48
X:$F420 – $F42A
DESCRIPTION
Reserved – Must be written with 0.
This pin is only available in 144-pin packages.
This pin is only available in 144-pin packages.
This pin is only available in 144-pin packages.
This pin is only available in 144-pin packages.
This pin is only available in 144-pin packages.
This pin is only available in 144-pin packages.
This pin is only available in 144-pin packages.
This pin is only available in 144-pin packages.
This pin is only available in 144-pin packages.
This pin is only available in 144-pin packages.
Reserved – Must be written with 0.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
This pin is available in all packages.
1. Please see Table 497. “General Purpose Input/Output Pins” on page 384.
Table 378. GPIO2 Pin Register (Bank 2) Description
274
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
23.2.12.4. GPIO3 (Bank 3)
GPIO3
BIT
LABEL
23:22
21
20
19
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
Note:
RSRVD
GP3B21
GP3B20
GP3B19
GP3B18
GP3B17
GP3B16
GP3B15
GP3B14
GP3B13
GP3B12
GP3B11
GP3B10
GP3B9
GP3B8
GP3B7
GP3B6
GP3B5
GP3B4
GP3B3
GP3B2
GP3B1
GP3B0
PIN NAME (NOTE 1)
GP93
GP92
GP91
GP90
GP89
GP88
GP87
GP86
GP85
GP84
GP83
GP82
GP81
GP80
GP79
GP78
GP77
GP76
GP75
GP74
GP73
GP72
X:$F430 – $F43A
DESCRIPTION
Reserved – Must be written with 0.
This pin is only available in 144-pin packages
This pin is only available in 144-pin packages
This pin is only available in 144-pin packages
This pin is only available in 144-pin packages
This pin is only available in 144-pin packages
This pin is only available in 144-pin packages
This pin is only available in 144-pin packages
This pin is only available in 144-pin packages
This pin is only available in 144-pin packages
This pin is only available in 144-pin packages
This pin is only available in 144-pin packages
This pin is only available in 144-pin packages
This pin is only available in 144-pin packages
This pin is only available in 144-pin packages
This pin is only available in 144-pin packages
This pin is only available in 144-pin packages
This pin is only available in 144-pin packages
This pin is only available in 144-pin packages
This pin is only available in 144-pin packages
This pin is only available in 144-pin packages
This pin is only available in 144-pin packages
This pin is only available in 144-pin packages
1. Please see Table 497. “General Purpose Input/Output Pins” on page 384..
Table 379. GPIO2 Pin Register (Bank 3) Description
5-35xx-D1-1.08-061705
275
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
24. DAC
The DAC behaves as a DMA device, outputting data from a modulo (circular) buffer
in X/Y/P memory specified by software configuration. Software configuration begins
by programming the DAC Base Address Register (HW_DACBAR), the DAC Modulo
Register (HW_DACMR), and the DAC Current Position Register (HW_DACCPR) to
specify the base address and modulo of the buffer to be used by the DAC as well as
the current position in the DAC buffer. Next, the buffer is filled by software. The
sample rate of the DAC may be specified in the DAC Sample Rate Register
(HW_DACSRR). The sample rate of the DAC can be adjusted with fine resolution
using this register. The number of samples available is specified in the DAC Word
Count Register (HW_DACWCR). The interrupt threshold word count is specified in
the DAC interrupt control Register (HW_DACICR). Finally, transmit is enabled, and
the DAC begins to output data. As each DSP word is processed, the DAC Word
Count Register (HW_DACWCR) is decremented and the DAC Current Position
Register (HW_DACCPR) is incremented. When the word count decrements to the
value in the interrupt control register, an interrupt will occur if interrupts have been
enabled in the DAC Control Status Register (HW_DACCSR). Software is then
responsible for writing to HW_DACWCR register indicating more data is available. If
the HW_DACWCR is not updated before the last sample in the buffer is needed by
the DAC, an exception will occur, and a DAC underflow interrupt will be generated.
Y-BUS
X-BUS
P-BUS
DSP CORE
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
DMA-BUS
DAC
Programmable Registers
DAC
DMA Engine
stereo Sigma Delta D/A Converter
Figure 96. Stereo Sigma Delta D/A and DMA
DAC data in stored in X/Y/P memory with the left channel sample in the lowest
address, followed by the corresponding right channel sample in the next memory
location. The DAC data should always consist of an even number of samples to
allow left and right channels, it is not possible to play mono data unless the mono
samples are each repeated twice in memory, once for the left channel and once for
the right channel. The data is stored as 24 bit 2’s compliment values, where the full
scale value depends on the programming of the sample rate converter. The DAC
276
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
expects data samples that already have been 2x over-sampled by code in the DSP,
i.e. to play 44.1k sample/sec data, there should be 44.1k * 2 = 88.2k samples/sec
per channel in the buffer in on-chip memory.
Filco/DSP Write Pointer
HW_DACBAR_BAR
HW_DACMR_MR
# Out
Samples
to
transmit
MP3
Decoder
1:2 Interpolation
in DSP or FILCO
44.1KHz
Samples
Left0
Right0
Left1
Right1
Left2
Right2
Left3
Right3
Left4
Right4
Left5
Right5
Left6
Right6
Left7
Right7
Left8
Right8
Left9
Right9
Left10
Right10
Left11
Right11
# Empty
Samples
# Out
Samples
to
transmit
Left0
Right0
Left1
Right1
Left2
Right2
Left3
Right3
Left4
Right4
Left5
Right5
Left6
Right6
Left7
Right7
Left8
Right8
Left9
Right9
Left10
Right10
Left11
Right11
Max #
Samples
in
Buffer
HW_DACCPR_CPR
HW_DACWCR_WCR
==
DSP Interrupt
HW_DACICR_ICR
Figure 97. Stereo Sigma Delta D/A DMA Buffer Registers
A producer/consumer double buffer mechanism can be defined by setting
HW_DACICR_ICR to half the size of HW_DACMR_MR. This will let the DAC consume half the buffer before the next DAC interrupt is generated.
Before the DAC can be turned on and used, the crystal clock used by the DAC and
mixer must be enabled using the ACKEN bit of the Clock Control Register
(HW_CCR). Also the DAC and Mixer analog circuitry must be powered up using the
PR1 & PR2 bits of the Mixer Powerdown Control/Status Register
(HW_MIXPWRDNR).
5-35xx-D1-1.08-061705
277
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
24.1. DAC Registers
24.1.1.
DAC Base Address Register
The DAC Base Address Register is used to specify the base address of the modulo
buffer used by the DAC port. The buffer’s base address must zero the k LSBs,
where 2k >= HW_DACMR.
HW_DACBAR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
BAR
2
3
X:$F805
Table 380. HW_DACBAR
BIT
LABEL RW RESET
23:16 RSRVD R
15:0 BAR
RW
0
0
DESCRIPTION
Reserved – Must be written with 0.
Base address for the DAC output sample buffer in memory.
Table 381. DAC Base Address Register Description
24.1.2.
DAC Modulo Register
The DAC Modulo Register specifies the modulus of the DAC buffer. The modulus is
specified in the same manner as the Mn modulo registers of the Address Generation Unit of the DSP core. For example, writing $000001 to the HW_DACMR indicates a modulo buffer of size $000002.
HW_DACMR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
MR
2
3
X:$F804
Table 382. HW_DACMR
BIT
LABEL RW RESET
23:10 RSRVD R
9:0
MR
RW
0
0
DESCRIPTION
Reserved – Must be written with 0.
Modulo (i.e. size) of the DAC output sample buffer in memory.
Table 383. DAC Modulo Register Description
24.1.3.
DAC Current Position Register
The DAC Current Position Register indicates the address offset from the address
specified in the HW_DACBAR DAC Base Address Register where the DAC will read
the next output sample.
278
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
HW_DACCPR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
CPR
2
3
X:$F803
Table 384. HW_DACCPR
BIT
LABEL RW RESET
23:10 RSRVD R
9:0
CPR
RW
0
0
DESCRIPTION
Reserved – Must be written with 0.
Current read position of the DAC output sample buffer in memory.
Table 385. DAC Current Position Register Description
24.1.4.
DAC Word Count Register
The DAC Word Count Register specifies the number of words remaining in the onchip memory buffer to be output. Software should fill the memory buffer with sample
data, then program the HW_DACWCR to indicate the number of words available to
the DAC to be output. For each sample output, the HW_DACWCR is decremented
twice, one each for the left and right samples. When the HW_DACWCR_WCR ==
HW_DACICR_IPT, a transmit interrupt is generated. If the HW_DACWCR_WCR
reaches zero and is not updated before the DAC requires another sample (i.e.,
under-run), then and exception occurs (HW_DACCSR_TXEXC is set).
The hardware avoids the race condition between the DMA decrementing
HW_DACWCR_WCR and software attempting to indicate the addition of more samples to the DAC buffer. When the DAC is not enabled (HW_DACCSR_TXEN==0),
then writes to HW_DACWCR_WCR directly replace the current contents. When the
DAC is enabled (HW_DACCSR_TXEN==1), then writes to HW_DACWCR_WCR
add to the current contents of HW_DACWCR_WCR, as follows:
HW_DACWCR_WCR = HW_DACWCR_WCR + DSP_WRITE_VALUE
If the DMA attempts to decrement the word count on the same cycle that the DSP
attempts to write to it then the new value becomes:
HW_DACWCR_WCR = HW_DACWCR_WCR + DSP_WRITE_VALUE -1
This register counts words in memory, not samples. Since the DAC operates on stereo data, the number of words will be twice the number of samples.
HW_DACWCR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
WCR
2
3
X:$F802
Table 386. HW_DACWCR
5-35xx-D1-1.08-061705
279
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BIT
LABEL RW RESET
23:10 RSRVD R
9:0
WCR
RW
DESCRIPTION
0
0
Reserved – Must be written with 0.
Number of DAC samples remaining until a DAC interrupt will be generated.
Table 387. DAC Word Count Register Description
24.1.5.
DAC INTERRUPT CONTROL REGISTER
The DAC interrupt control register specifies the number of samples remaining in the
DAC output buffer at which to trigger an interrupt. This allows the interrupt to be
generated well ahead of a completely empty buffer.
HW_DACICR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
IPT
2
3
X:$F806
Table 388. HW_DACICR
BIT
LABEL RW RESET
23:10 RSRVD R
9:0
IPT
RW
DESCRIPTION
0
0
Reserved – Must be written with 0.
DAC interrupt will be generated on the sample when HW_DACWCR_WCR ==
HW_DACICR_IPT. Since this register initializes to zero, the default behavior is the
same as for the STMP3410 where an interrupt is generated on the sample when
HW_DACWCR_WCR reaches zero.
Table 389. DAC Interrupt Control Register Description
24.1.6.
DAC Sample Rate Register
The DAC Sample Rate Register is programmed to specify the sample rate of the DAC.
HW_DACSRR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
SR
2
3
X:$F801
Table 390. HW_DACSRR
BIT LABEL RW RESET
23
RSRVD R
22:0 SR
RW
0
0
DESCRIPTION
Reserved – Must be written with 0.
Sample Rate to use for DAC playback.
Table 391. DAC Sample Rate Register Description
Most readers can use the values in Table 392 and simply skip the rest of the
detailed discussion in this section. The following table contains the required value of
the HW_DACSRR register for various common sample rates. If the crystal is changed
from the normal 24.0 MHz rate, the FanalogDAC frequency will change, and the
HW_DACSRR register must be changed to compensate. Values for both 24.0MHz
280
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
and 24.576MHz are given in the table. The table assumes HW_CCR_DACDIV is
set to zero to cause the crystal clock to be divided by four (FanalogDAC = 6.0MHz).
.
24.0 MHZ
SAMPLE
RATE
24.576 MHZ
HW_DACSRR_SR HW_DACSRR_SR
VALUE
VALUE
FsampleDAC
48,000 Hz
44,100 Hz
32,000 Hz
24,000 Hz
22,050 Hz
16,000 Hz
12,000 Hz
11,025 Hz
8,000 Hz
$06D000
$0780DF
$0AB800
$0EA000
$1001BE
$167000
$1E4000
$21037C
$2DE000
$070000
$07B51E
$0B0000
$0F0000
$106A3B
$170000
$1F0000
$21D476
$5F0000
Table 392. Values for the HW_DACSRR registers
For any of the desired sample rates, the value of the HW_DACSRR register is calculated according to the following formula:
OSRDAC = (65536 * 8* FanalogDAC) / (128 * FsampleDAC)
= (4096 * FanalogDAC) / FsampleDAC
If computed with the above explicit operator precedence, the resulting over sample rate
(OSRDAC) will be a 24 bit scaled fixed point representation of the desired decimation
rate. In order to load the HW_DACSRR_SR field, we need to account for the interpolator’s pos_zero comparison. To do this, we subtract one from the whole number portion of
the OSRDAC. We load the SR bit field with:
HW_DACSRR_SR = (OSRDAC - $010000)
The value stored in the HW_DACSRR is interpreted as a 24 bit unsigned value.
WARNING: The computation of the scaled fixed point OSRDAC value has an intermediate result precision of 35 bits before the division. Long Long (64-bit) C variables
should be used to avoid errors due to loss of precision. These are static values that
are hard coded into applications for the nine sample rates of interest. Results should
be rounded not truncated.
NOTE: While the OSR calculation, above, uses the desired sample rate, the buffer
must actually contain audio samples at twice this sample rate. Therefore to play a
44.1KHz audio stream through the DAC, one must present an 88.2KHz sample rate
interpolated version of the original samples in the buffer that will be fetched by the
DAC’s DMA, see Figure 97. “Stereo Sigma Delta D/A DMA Buffer Registers” on
page 277. Thus a 2:1 interpolation algorithm must have been run on the 44.1KHz
stream before it can be passed to the DMA. This algorithm can run either in DSP
software or in the Filter coprocessor, see Section 13. “FILTER COPROCESSOR
(FILCO)” on page 154.
The 1-bit sigma delta D/A converter is always sampled on a submultiple of the
24.0MHz crystal oscillator frequency as specified in the HW_CCR_DACDIV
register, see Figure 98. “Stereo Sigma Delta D/A Converter” on page 282. This
divider generates sample strobes at FanalogDAC where the divisors available come
5-35xx-D1-1.08-061705
281
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
from the set {4,6,8,12,16,24}. With HW_CCR_DACDIV set to zero, to divide by 4,
FanalogDAC becomes 6.0MHz for a 24.0 MHz crystal. The sample strobe derived
from this divider is used to interpolate the 1 bit D/A values. The 1-bit sigma delta
modulator is effectively running at FanalogDAC. As shown in Figure 98, the 24 bit 2x
sample rate D/A values are extracted from on-chip RAM via the DMA. They are
filtered to band limit the audio stream. This filter runs on xtal_clk but filters samples
at the source sample rate which is slower than the output D/A sample. In the
process, this filter performs a fixed 1:8 interpolation or up-sample of the already 2X
up-sampled input stream. Thus the output of the 1:8 fixed interpolation filter is
running at 16X FsampleDAC. A single 24 bit sample at the output of the fixed filter is
further interpolated up to the 1-bit D/A rate. The variable rate sample, hold and
interpolate block performs this function.
16 x FsampleDAC
Handshake
Right Channel
from DMA
1:8 Fixed
interp. Filter
Sample Hold
& Interpolate
Sigma
dac_rsamp
Delta
Modulator
1-Bit
analog
D/A
R
Handshake
Left Channel
from DMA
1:8 Fixed
interp. Filter
Sample Hold
& Interpolate
Sigma
Delta
Modulator
1-Bit
analog
D/A
L
HW_CCR_DACDIV
handshake
2 x FsampleDAC
÷DA_DIV
dac_lsamp
samp_strobe
FanalogDAC
24.0MHz
XTAL_CLK
XTAL
OSC
position_reg[31:0]
HW_DACSRR_SR
24'hFF0000
1
+
1
31
samp_strobe
16 15
Position Reg
whole # fraction
pos_zero=
(position_reg[23:16] == 8'h00)
frac[15:0]
pos_zero
variable rate interpolator
Figure 98. Stereo Sigma Delta D/A Converter
It stalls the filter pipeline and DMA source, using the handshake lines that connect
with the previous filter stage to supply samples at the correct over-sample ratio. The
1-bit DAC runs at the fixed sample rate of FanalogDAC while the DMA fetches samples in burst at irregular intervals to maintain the required input to the modulator.
In this case, the 1-bit D/A is running at 136.054 times the desired sample rate of
44.1KHz or 6MHz. The sample hold and interpolate block accepts a new sample
from the filter at a 44.1KHz times 2 times 8 or 705.6KHz. It passes interpolated samples to the modulator at a 6.0MHz rate. The sample, hold and interpolate block
passes a source sample from the fixed 1:8 interpolation filter to the sigma delta
modulator corresponding to every 8.503 FanalogDAC samples. Recall that this is a
282
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
variable rate interpolation stage which changes for every Over Sample Rate (OSR)
setting in use.
If the desired sample rate FsampleDAC= 44.1KHz, for example, the sample hold and
interpolate block will accept samples from fixed interpolation filter at 705.6KHz, i.e.
16 times the desired sample rate. There is a handshake pair (request/ack) between
the variable rate sample hold and interpolate block and the fixed interpolating filter
block. This hand shake is used to pace the samples from the filter and DMA buffer
to 88.2KHz (2x the desired 44.1KHz) rate. An interpolation filter routine was used
within the DSP to generate the 88.2KHz sample stream from the 44.1KHz source.
Of course, this is not all the sample rates of interest! There are the members of the
48KHz family, whose members satisfy the property:
24.576MHz = Q*FsampleDAC.
These sample rates include 48KHz, 32KHz, 24KHz, 16KHz, and 12KHz. There are
also the members of the 44.1KHz family, whose members satisfy the property:
16.9344MHz = Q*FsampleADC
where Q comes from the set of integers. These sample rates include 44.1KHz,
22.05KHz and 11.025KHz. Since 24.0MHz and 16.9344MHz are relatively prime,
members of the 44.1KHz family are related to the 24.0MHz source clock by the relationship 24.0MHz=P*FsampleADC, where P is a rational number.
The D/A converter block includes a variable rate or rational interpolator as shown in
Figure 98 to accommodate these sample rates. Rational numbers in the DAC are
approximated with a scaled fixed point 24 bit value. In this case, the decimal point
falls between bit 15 and bit 16. Therefore the lower two bytes hold the fractional part
while the upper byte holds the whole number portion of the scaled fixed point. The
position register uses this scaled fixed point representation for the number of 1-bit
samples to be interpolated (generated) to find the next sample to be sent to the
sigma delta modulator. Whenever the whole number part (viz. bits 23:16) are zero,
then the next DMA sample is consumed. For playback at 44.1KHz, we have to set
the interpolator to generate 67.027 new interpolated samples between every DMA
sample.
To accomplish this we load HW_DACSRR_SR with a value of $0780DF in the 24 bit
scaled fixed point representation or 7.503 decimal. Referring to the bottom of Figure
98, at reset the position register is loaded with zero so the pos_zero boolean is true.
Thus the value $0780DF will be continuously added to the position_reg. After the
first addition, pos_zero is no longer true so the value $FF0000 is added to the position-reg. This is effectively a value of minus 1 in the whole number portion resulting
in a position register value of $0680DF (6.503). Thus seven 1-bit samples will be
generated. At the eighth sample, the DAC will consume the next 24-bit value from
the fixed 1:8 filter. Since the upper part of the position register is once again zero,
the $0780DF is added to it. The lower 16-bits still contain $00DF so the new position
register is $781BE or 7.507decimal. After enough samples are fetched, the fractional part summation will overflow into an integral sample increase causing an additional interpolation cycle. Thus the fractional part of the rational interpolator
periodically corrects the diffused interpolation error and corrects for it.
Note: Although the DAC can operate at a maximum sampe rate of 48 kHz, the digital filter is limited to 44.1 kHz when
both the DAC and ADC are operationg. Sampling above 44.1 kHz may result in an elevated noise floor when the
ADV and DAC are operating simultaneously.
5-35xx-D1-1.08-061705
283
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
DAC Control Status Register
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
TXEN
1
5
TXIEN
1
6
TXI
1
7
LPBK
1
8
TXEXC
1
9
DMASEL
2
0
DAC_HPPOP
2
1
LVUP
2
2
CLKGT
2
3
X:$F800
RVUP
HW_DACCSR
VUP
24.1.7.
Table 393. HW_DACCSR
BITS
23
LABEL
CLKGT
22:11 RSRVD
10
VUP
RW RESET
RW 0
R
R
0
0
R
0
9
RVUP
8
LVUP
7
DAC_HPPOP_EN RW 0
6:5
DMASEL
RW 00
4
LPBK
RW 0
3
TXEXC
RW 0
2
TXI
RW 0
DEFINITION
Clock gate – Used to disable the clocks to the DAC module to conserve
power when the DAC is not in use. Must be set to 0 before writing to any
other DAC registers.
0Clocks not gated
1Clocks are gated off
Reserved – Must be written with 0.
Volume Update Pending on Either Channel– This bit is set one if a
volume update is pending, i.e. waiting for zero crossing, on either the left or
right channels or both.
Volume Update Pending on Right Channel– This bit is set one if a
volume update is pending, i.e. waiting for zero crossing, on the right
channel.
Volume Update Pending on Left Channel– This bit is set one if a volume
update is pending, i.e. waiting for zero crossing, on the left channel.
DAC POP Suppressor Enable – Set this bit to one to enable the pop
suppression algorithm. Set this bit to zero to disable the pop suppressor
and cause the DAC to mimic the STMP3410 behavior. The pop suppressor
works by sending a continuous alternating zero and one stream of samples
(101010101010) whenever the DAC is disabled (HW_DACCSR_TXEN
=0).
Memory space to use for DMA transfers
00 - X space
01 - Y space
10 - P space
11 - Reserved
Loopback – When set, the LPBK bit connects the ADC single bit data from
the ADC analog circuitry directly to the DAC analog circuitry to perform an
analog loopback test without utilizing the DSP core.
Transmit Exception – The TXEXC bit is a status bit indicating a transmit
exception has occurred. A transmit exception occurs when the DAC needs
to read a sample but the HW_DACWCR is zero indicating no samples are
available. The TXEXC bit is cleared by writing a 0 to this location.
Transmit Interrupt – The TXI bit is a status bit indicating a transmit
interrupt has occurred. The TXI bit is cleared by writing a 0 to this location.
The Transmit Interrupt can also be cleared by writing to the HW_DACWCR
register.
Table 394. DAC Control Status Register Description
284
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
RW RESET
1
TXIEN
RW 0
0
TXEN
RW 0
DEFINITION
Transmit Interrupt Enable – The TXIEN bit enables interrupt generation
for this port. When the HW_DACWCR reaches zero, an interrupt is
generated.
Transmit Enable – Setting the TXEN bit causes the DAC Port to begin
outputting samples from the modulo buffer specified by the HW_DACBAR,
HW_DACMR, and HW_DACWCR. When TXEN is cleared, the TXI and
TXEXC bits are cleared. No other DAC registers are cleared.
Table 394. DAC Control Status Register Description (Continued)
5-35xx-D1-1.08-061705
285
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
25. ADC
The ADC behaves as a DMA device which places data in a modulo (circular) buffer
in X/Y/P memory specified by software configuration. Software configuration begins
by programming the ADC Base Address Register (HW_ADCBAR), the ADC Modulo
Register (HW_ADCMR), and the ADC Current Position Register (HW_ADCCPR) to
specify the base address and modulo of the buffer to be used by the ADC as well as
the current position in the ADC buffer. The sample rate of the ADC must be specified in the ADC Sample Rate Register (HW_ADCSRR). The number of available
sample entries are specified in the ADC Word Count Register (HW_ADCWCR). The
ADC interrupt threshold word count is specified in the ADC Interrupt Control Register (HW_ADCICR). At this point sample capture is enabled and the ADC begins to
input stereo data. As each DSP word is input, the HW_ADCWCR is decremented
and the ADC Current Position Register (HW_ADCCPR) is incremented. When the
word count decrements to the value in the HW_ADCICR_IPT, an interrupt will occur
if interrupts have been enabled in the ADC Control Status Register (HW_ADCCSR).
Software is then responsible for writing to HW_ADCWCR register indicating space
for more data is available. If the HW_ADCWCR is not updated before the last free
sample location needs to be written to on-chip RAM by the ADC, an exception will
occur and an ADC overflow interrupt will be generated.
Y-BUS
X-BUS
P-BUS
DSP CORE
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
16KWord SRAM
DMA-BUS
ADC
Programmable Registers
ADC
DMA Engine
stereo Sigma Delta A/D Converter
Figure 99. Stereo Sigma Delta A/D and DMA
ADC data is stored in X/Y/P memory with the left channel sample in the lowest
address, followed by the corresponding right channel sample in the next memory
location. The ADC data always consists of an even number of words to allow left
and right channels, it is not possible to record mono data unless the stereo data is
recorded and one of the channels is thrown away by the DSP software. The data is
stored as 24 bit 2’s compliment values, where the full scale value depends on the
programming of the sample rate converter. The ADC generates data samples that
are 8x over-sampled, and that need to be decimated to the base rate by code in the
DSP. For example, to record 44.1k sample/sec data, the ADC will generate 44.1k * 8
286
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
= 352.8k samples/sec per channel in memory, which must be filtered and decimated
down to 44.1k samples/sec by code executing on the DSP or by utilizing three 2:1
decimation stages in the Filter Coprocessor.
WARNING: While the ADC is designed to produce 8X over sampled streams relative to the desired sample rate, these are 8X over sampled at a lower precision.
DSP filtering should always filter and decimate not just decimate the result.
HW_ADCWCR_WCR
DSP Interrupt
==
HW_ADCICR_ICR
HW_ADCBAR_BAR
HW_ADCMR_MR
# of
Samples
from
ADC
# Empty
Samples
# of
Samples
from
ADC
Left0
Right0
Left1
Right1
Left2
Right2
Left3
Right3
Left4
Right4
Left5
Right5
Left6
Right6
Left7
Right7
Left8
Right8
Left9
Right9
Left10
Right10
Left11
Right11
44.1KHz
Samples
Max #
Samples
in
Buffer
8:1 Decimation
in DSP or FILCO
Left0
Right0
Left1
Right1
Left2
Right2
Left3
Right3
Left4
Right4
Left5
Right5
Left6
Right6
Left7
Right7
Left8
Right8
Left9
Right9
Left10
Right10
Left11
Right11
HW_ADCCPR_CPR
Filco/DSP Read Pointer
Figure 100. Stereo Sigma Delta A/D and DMA Buffers
A producer/consumer double buffer mechanism can be defined by setting
HW_ADCICR_IPT to half the size of HW_ADCWCR_WCR. The ADC can then fill
half the buffer before generating an interrupt. Figure 100 shows that ADC utilizes a
circular buffer whose base address is specified in HW_ADCBAR_BAR. The total
number of words in the circular is specified in HW_ADCMR_MR. The word that will
next by written with a new ADC sample value is pointed to by the current position
register, HW_ADCCPR_CPR. The decimation software or the FILCO has a read
pointer that points to the next sample to be read. The difference in these two positions corresponds to the total number of unused locations in the buffer. The software
is responsible for notifying the ADC of the number of words it or FILCO have read
5-35xx-D1-1.08-061705
287
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
from the buffer. It does this by writing the number consumed to
HW_ADCWCR_WCR. The ADC adds the write value to its internal word count. The
total word count maintained by the ADC represents the number of unfilled locations
available. If that size falls to zero an overflow is reported on the next sample captured. As that word count falls below the HW_ADCICR_IPT then an interrupt is generated, notifying software that more samples are available in the buffer.
ADC write addresses are generated by logically ORing the BAR and current position
registers together. Therefore the BAR must be aligned with next larger power of two
than the modulo value. The modulo value does not have to be a power of two.
Before the ADC can be turned on and used, the crystal clock used by the ADC must
be enabled using the ACKEN bit of the HW_CCR Clock Control register. Also the
ADC analog circuitry must be powered up using the PR0 bit of the
HW_MIXPWRDNR Mixer Powerdown Control/Status register. The value in
HW_CCR_ADCDIV may have to be modified for some sample rates and crystal frequency combinations.
25.1. ADC Registers
25.1.1.
ADC Base Address Register
The HW_ADCBAR is used to specify the base address of the modulo buffer used
by the ADC. The buffer’s base address must zero the k LSBs, where 2k >=
HW_ADCMR.
HW_ADCBAR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
BAR
2
3
X:$FB05
Table 395. HW_ADCBAR
BIT LABEL RW RESET
DESCRIPTION
23:16 RSRVD R
0
Reserved – Must be written with 0.
15:0 BAR
RW 0
Base address for the ADC input sample buffer in memory.
Table 396. ADC Base Address Register Description
25.1.2.
ADC Modulo Register
The ADC Modulo Register specifies the modulus of the ADC buffer. The modulus
specified in the same manner as the Mn modulo registers of the Address Generation Unit. For example, writing $0001 to the HW_ADCMR_MR indicates a modulo
buffer of size $0002.
HW_ADCMR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
MR
2
3
X:$FB04
Table 397. HW_ADCMR
288
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BIT LABEL RW RESET
DESCRIPTION
23:10 RSRVD R
0
Reserved – Must be written with 0.
9:0
MR
RW 0
Modulo (i.e. size) of the ADC input sample buffer in memory.
Table 398. ADC Modulo Register Description
25.1.3.
ADC Current Position Register
The ADC Current Position Register indicates the address offset from the address
specified by the ADC Base Address Register (HW_ADCBAR_BAR) where the ADC
will write the next output sample.
HW_ADCCPR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
CPR
2
3
X:$FB03
Table 399. HW_ADCCPR
BIT LABEL RW RESET
DESCRIPTION
23:10 RSRVD R
0
Reserved – Must be written with 0.
9:0
CPR
RW 0
Current write position of the ADC input sample buffer in memory.
Table 400. ADC Current Position Register Description
25.1.4.
ADC Word Count Register
The ADC Word Count Register specifies the number of words that are available in
the circular buffer to hold new samples. Software should allocate a memory buffer to
hold the sample data, then program the HW_ADCWCR_WCR to indicate the
number of words available to the ADC to hold the samples being captured. For each
sample captured, the HW_ADCWCR_WCR is decremented twice, once each for
the left and right samples. When the HW_ADCWCR_WCR == HW_ADCICR_IPT, a
transmit interrupt is generated. If the HW_ADCWCR_WCR reaches zero before the
ADC needs to write another sample to on-chip RAM (i.e., overrun), then an
exception occurs (HW_ADCCSR_TXEXC is set).
The hardware avoids the race condition between the DMA decrementing
HW_ADCWCR_WCR and software attempting to indicate the availability of more
sample space by writing to HW_ADCWCR_WCR. When the ADC is not enabled
(HW_ADCCSR_TXEN == 0) then writes to the HW_ADCWCR_WCR replace the
current contents. When the ADC is enabled, (HW_ADCCSR_TXEN = 1) then writes
to the HW_ADCWCR_WCR register add to the current contents as follows:
HW_ADCWCR_WCR = HW_ADCWCR_WCR + DSP_WRITE_VALUE
If the DMA attempts to decrement the word count on the same cycle that the DSP
attempts to write to it, then the new value becomes:
HW_ADCWCR_WCR = HW_ADCWCR_WCR + DSP_WRITE_VALUE - 1
This register counts words in memory, not samples. Since the ADC operates on
stereo data, the number of words will be twice the number of samples.
5-35xx-D1-1.08-061705
289
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
HW_ADCWCR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
WCR
2
3
X:$FB02
Table 401. HW_ADCWCR
BIT
LABEL RW RESET
23:10 RSRVD R
9:0
WCR
RW
DESCRIPTION
0
0
Reserved – Must be written with 0.
Number of ADC samples remaining until an ADC interrupt will be generated.
Table 402. ADC Word Count Register Description
25.1.5.
ADC INTERRUPT CONTROL REGISTER
The ADC interrupt control register specifies the number of empty sample entries
remaining in the ADC input buffer at which to trigger an interrupt. This allows the
interrupt to be generated well ahead of a completely full buffer.
HW_ADCICR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
IPT
2
3
X:$FB06
Table 403. HW_ADCICR
BIT
LABEL RW RESET
23:10 RSRVD R
9:0
IPT
RW
0
0
DESCRIPTION
Reserved – Must be written with 0.
DAC interrupt will be generated on the sample when HW_ADCWCR_WCR ==
HW_ADCICR_IPT. Since this register initializes to zero, the default behavior is the
same as for the STMP3410 where an interrupt is generated on the sample when
HW_ADCWCR_WCR reaches zero.
Table 404. ADC Interrupt Control Register Description
25.1.6.
ADC Sample Rate Register
The ADC Sample Rate Register is programmed to specify the sample rate of the ADC.
HW_ADCSRR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
SR
2
3
X:$FB01
Table 405. HW_ADCSRR
290
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BIT LABEL RW RESET
23
RSRVD R
22:0 SR
RW
0
0
DESCRIPTION
Reserved – Must be written with 0.
Sample Rate to use for ADC recording.
Table 406. ADC Sample Rate Register Description
Most readers can simply use the values in Table 407 and skip the following detailed discussion. The following table contains the required value of the HW_ADCSSR_SR register for various common sample rates. If the crystal is changed from the normal 24.0 MHz
rate, the Fanalog frequency will change, and the HW_ADCSSR_SR register must be
changed to compensate. The table assumes that HW_CCR_ADCDIV is either set to
zero, i.e. FanalogADC = 6.0MHz or FanalogADC = 1.0MHz. Optional low pass filter is
enabled at HW_ADCCSR_LPF_MODE.
Fanalog = 6.0MHz
SAMPLE
RATE
HW_ADCSRR_SR
FsampleADC
Value
SCALE FACTOR
with LPF
no LPF
48,000 Hz
$0EA000
16
128
44,100 Hz
$1001BE
13
108
32,000 Hz
$167000
7
57
24,000 Hz
$1E4000
4
32
22,050 Hz
$21037C
3.37
27
16,000 Hz
$2DE000
1.77 *
14
1.0 *
8.0
12,000 Hz
$3D8000
0.844 *
6.75
11,025 Hz
$4306F7
8,000 Hz
$5CC000
0.44 *
3.55
* The OSR for these sample rates at Fanalog=6MHz is too high,
use Fanalog=1.0MHz to avoid clipping.
Table 407. Values for the HW_ADCSRR_SR register at 6.0MHz
Fanalog = 1.0MHz
SAMPLE
RATE
HW_ADCSRR_SR
SCALE FACTOR
FsampleADC
Value
no LPF
with LPF
48,000 Hz
44,100 Hz
32,000 Hz
24,000 Hz
22,050 Hz
16,000 Hz
12,000 Hz
11,025 Hz
8,000 Hz
$019AAA
$01D59F
$02E800
$043555
$04AB3F
$06D000
$096AAA
$0A567E
$0EA000
463
392
206
116
99
52
29
24
13
3704
3136
1648
928
792
416
232
192
104
Table 408. Values for the HW_ADCSRR_SR register at 1.0MHz
5-35xx-D1-1.08-061705
291
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
For any of the desired sample rates, the value of the HW_ADCSSR_SR register is calculated according to the following formula:
OSRADC = (65536 * 16 * FanalogADC) / (128 * FsampleADC)
= (8192* FanalogADC) / FsampleADC
If computed with the above explicit operator precedence, the resulting over sample rate
(OSRADC) will be a 24 bit scaled fixed point representation of the desired decimation
rate. In order to load the HW_ADCSRR_SR field, we need to account for the decimators
pos_zero comparison. To do this, we subtract one from the whole number portion of the
OSRADC. We load the SR bit field with:
HW_ADCSSR_SR = (OSRADC - $010000)
The value stored in the HW_ADCSRR_SR is interpreted as a 24 bit unsigned value.
WARNING: The computation of the scaled fixed point OSRADC value has an intermediate result precision of 35 bits before the division. These are static values that
are hard coded into applications for the nine sample rates of interest. Long Long
(64-bit) C variables should be used to avoid errors due to loss of precision.
NOTE: While the OSR calculation, above, uses the desired sample rate, the buffer
will actually contain audio samples at 8 times this sample rate. Therefore to record a
44.1KHz audio stream from the ADC, one must accept the 352.8KHz sample
stream that is DMAed to on-chip memory. To record it at 44.1KHz, one must run an
8:1 decimation algorithm in the DSP or run three stages of the FILCO’s 2:1 decimation FIR filter, see Figure 100. “Stereo Sigma Delta A/D and DMA Buffers” on
page 287.
The 1-bit sigma delta A/D converter is always sampled on a submultiple of the
24.0MHz crystal oscillator frequency as specified in the HW_CCR_ADCDIV register, see Figure 101. “Variable Rate A/D Converter” on page 294. This divider generates sample strobes at FanalogADC where the divisors available come from the set
{4,6,8,12,16,24}. It is recommended that ADCDIV always be set to 000 so that a
6.0MHz 1-bit A/D sample rate is used. The sample strobe is used to integrate the 1bit A/D values. As shown in Figure 101, these integrated values are filtered and then
delivered to the ADC DMA to write into on-chip RAM.
Notice that the integrators run continuously while the filters produce samples at the
decimated rate. Depending on the decimation or over-sample ratio of the CIC filter
engine, the integrators will produce samples of various precisions and scale factors.
The filtered values written to RAM are signed 24 bit numbers with the conversion
data LSB justified, i.e. down-scaled in the lower end of the 24 bit word. The scale
factor column of Table 407, “Values for the HW_ADCSRR_SR register at 6.0MHz,”
on page 291 gives the approximate values to multiply the resultant samples to
achieve a normalize fixed point scaled fixed point with the fixed point set between
the sign bit and the first data bit.
If the variable rate decimator is set to decimate by exactly 16, then this setting
causes it to produce samples that are 8X over-sampled with respect to the desired
frequency (8x16=128 total over-sample rate). A filter and decimation routine in the
DSP is then used to produce stereo or mono audio samples at the desired FsampleADC for use in various recording applications. With FanalogADC at 6.0MHz and an
oversample rate of exactly 128 gives us an FsampleADC of 46.875KHz. Thus there
are no standard audio stream sample rates that are exactly interpolated with a
24.0MHz crystal.
292
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
The 48KHz family of sample rates satisfy the property:
24.576MHz = Q*FsampleADC where Q comes from the set of integers.
These sample rates include 48KHz, 32KHz, 24KHz, 16KHz, 12KHz and 8KHz.
There are also the members of the 44.1KHz family, whose members satisfy the
property:
16.9344MHz = Q*FsampleADC where Q comes from the set of integers.
These sample rates include 44.1KHz, 22.05KHz and 11.025KHz. Since 24.576KHz
and 16.9344MHz are relatively prime to 24.0MHz, members of the 48KHz family
and 44.1KHz family are related to the 24.0MHz source clock by the relationship
24.0MHz=P*FsampleADC, where P is a rational number.
The A/D block includes a variable rate or rational decimator as shown in Figure 101
to accommodate these sample rates. Rational numbers in the ADC are approximated with a scaled fixed point 24 bit value. In this case, the decimal point falls
between bit 15 and bit 16. Therefore the lower two bytes hold the fractional part
while the upper byte holds the whole number portion of the scaled fixed point. The
position register uses this scaled fixed point representation to hold the number of 1bit samples to be dropped (decimated) to find the next sample at which to produce a
filtered multi-bit sigma delta A/D value to send to the DMA. Whenever the whole
number part (viz. bits 23:16) are zero, then a sample is produced. For recording at
44.1KHz, we have to set the decimator to skip 17.0068 (P=2500/147) 1-bit samples
and produce a filtered multi-bit result. Recall that all 17.0068 1-bit samples are integrated before producing the result. To accomplish this we load HW_ADCSRR_SR
with a value of $1001BE or 16 as the whole number portion. Referring to the bottom
of Figure 101, at reset the position_reg is loaded with zero so the pos_zero boolean
is true. Thus the value $1001BE will be continuously added to the position_reg.
After the first addition, pos_zero is no longer true so the value $FF0000 is added to
the position-reg. This is effectively a value of minus 1 in the whole number portion
so sixteen 1-bit samples will be skipped. On the seventeenth sample, the ADC will
produce a 24 bit result and will forward it to the on-chip RAM, via the DMA.
Notice that the lower 16 bits of the HW_ADCSRR_SR value will be accumulate into
the position register and will eventually overflow into the whole number portion
causing an additional sample to be skipped.
Note: Although the ADC can operate at a maximum sample rate of 48kHz, the digital filter is limited to 44.1KHz when
both teh ASC and DAC are operating. Sampling above 44.1kHz may result in an elevated noise floor when the
ADC and DAC are operationg simultaneously.
5-35xx-D1-1.08-061705
293
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
analog_R
analog_L
1-Bit
A/D
rsamp_in
1-Bit
A/D
lsamp_in
+/- 1
Integrator
CIC Filter &
Interpolators
rsamp_out[23:0]
+/- 1
Integrator
CIC Filter &
Interpolators
lsamp_out[23:0]
HW_CCR_ADC_DIV
Variable Rate
Decimator
AD_DIV
table
÷AD_DIV
24.0MHz
6.0MHz
sample strobe
CIC
State Matchine
DMA_request
XTAL_CLK
XTAL
OSC
position_reg[31:0]
HW_ADCSRR_SR
24'hFF0000
1
+
samp_strobe
1
31
16 15
Position Reg
whole # fraction
pos_zero=
(position_reg[31:16] == 16'h0000)
frac[15:0]
pos_zero
variable rate decimator
Figure 101. Variable Rate A/D Converter
As shown in Table 392, “Values for the HW_DACSRR registers,” on page 281, at
FsampleADC =44.1KHz, FanalogADC is 6.00MHz so the ratio of these two frequencies is 136.0544. This is the over-sample ratio at which the 1-bit A/D over samples
the desired sample rate. Since we have set the decimation rate to 1:17.0068 we will
see 352.8KSamples/second in the on-chip buffer or 8X over sample of our desired
44.1KHz sample rate. Either DSP software or the FILCO coprocessor is used to low
pass filter and decimate this resultant sample stream to 44.1KHz.
The range of values of the samples stored into the on-chip RAM is proportional to
the square of the over sample rate (OSR) used in the capture process. The larger
the OSR, the longer period the integrators run in the ADC. As result the range of
values scene for the same signal wave form captured at the same sample rate but
with two different OSR will be different. For example, an 8KHz microphone captured
at FADC = 6.0MHz will be 36 times larger than the values resulting from capturing
the same source signal at FADC = 1.0MHz. The peak range of values seen in a
capture of a signal at 44.1KHz with FanalogADC = 6.0MHz is +/-3200 decimal. The
oversimple ratio in this case is OSR= 136.054. We can calculate a magnitude constant, Kfilter for ADC’s filter from this as Kfilter = OSR2/Peak Value = (136.054)2/3200
294
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
= 5.7846. For any OSR in any sample rate, the peak value can be approximated by
Valuepeak = OSR2 /Kfilter.
In signal processing, one frequently normalizes the range of values to +/-1.0 as
seen in a fixed point scaled integer1. For a 24 bit DSP, the fixed point is placed
between bit 23 and the sign bit (bit 24) (bit 1 = 20). So the desired maximum excursion is then +/- 223 or +/-8388608, decimal. One can calculate a normalization constant to multiply all incoming samples for each sampling condition from the following
equation:
ScaleFactor = 223 * Kfilter / OSR2
If the incoming sample stream is multiplied, sample by sample, by ScaleFactor, then
normalized +/-1.0 samples result. The reader may wish to reduce the size of ScaleFactor slightly to allow for out or range samples.
The STMP35xx implements an optional low pass filter which is applied to the 1-bit
sigma delta stream, see HW_ADCCSR_LPFMODE. Selecting the LPF option
increases the magnitude of the signal going into the decimation by a factor of eight.
As a result, the scale factors are one eighth as large when using the LPF than without the LPF enabled, see Table 407. “Values for the HW_ADCSRR_SR register at
6.0MHz” on page 291.
NOTE: Because of the effect of scaling, sample rates at 8KHz, 11.0225KHz, 12KHz,
and 16KHz should be performed at Fanalog = 1.0MHz. One can empirically determine the value of Kfilter for a design and then adjust the scale factor table.
1.A normalized two’s complement 24 bit number can not actually express a value of +1.0 without overflowing.
5-35xx-D1-1.08-061705
295
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
ADC Control Status Register
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
TXEN
1
9
TXIEN
2
0
TXI
2
1
LPBK
2
2
LPFMODE
CLKGT
2
3
X:$FB00
TXEXC
HW_ADCCSR
DMASEL
25.1.7.
Table 409. HW_ADCCSR
BITS
23
LABEL
CLKGT
RW RESET
RW 0
22:12 RSRVD
R 0
11:8 LPFMODE RW 0000
7
6:5
RSRVD
DMASEL
R 0
RW 00
4
LPBK
RW 0
3
TXEXC
RW 0
2
TXI
RW 0
1
TXIEN
RW 0
0
TXEN
RW 0
DEFINITION
Clock gate – Used to disable the clocks to the ADC module to conserve power
when the ADC is not in use. This bit must be set to 0 before writing to any other
ADC registers, and while the ADC is operating.
0Clocks enabled
1Clocks gated off
Reserved – Must be written with 0.
Low Pass Filter Mode – The ADC implements a hardware low pass filter on the
one bit values coming from the 1-bit sigma delta A/D converter. This field controls
the low pass filter mode for both the left and the right channel.
0000 - Low pass filtering disabled (3410 compatibility mode)
1111 - input data summed across 16 samples, then divide by 16
Other LPFMODE settings are reserved.
Reserved – Must be written with 0.
Memory space to use for DMA transfers
00 - X space
01 - Y space
10 - P space
11 - Reserved
Loopback – When set, the LPBK bit connects the DAC single bit data output from
the DAC digital circuitry directly to the ADC digital circuitry to perform a digital
loopback test without utilizing the analog circuitry.
Transmit Exception – The TXEXC bit is a status bit indicating a transmit
exception has occurred. A transmit exception occurs when the ADC needs to
write a sample but the HW_ADCWCR is zero indicating no room is available in the
module buffer.The TXEXC bit is cleared by writing a 0 to this location.
Transmit Interrupt – The TXI bit is a status bit indicating a transmit interrupt has
occurred. This register can also be cleared by writing to the HW_ADCWCR
register. The TXI bit is cleared by writing a 0 to this location.
Transmit Interrupt Enable – This bit enables interrupt generation for the ADC.
When the HW_ADCWCR register reaches zero, an interrupt is generated.
Transmit Enable – Setting the TXEN bit causes the ADC Port to begin outputting
samples into the modulo buffer specified by the HW_ADCBAR, HW_ADCMR, and
HW_ADCWCR registers. When TXEN is cleared, the TXI and TXEXC bits are
cleared. No other ADC registers are cleared.
Table 410. ADC Control Status Register Description
296
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
26. MIXER
26.1. Mixer Address Registers
ADDRESS
DESCRIPTION
X:$FA03
X:$FA04
X:$FA05
X:$FA06
X:$FA07
X:$FA08
X:$FA09
X:$FA0A
X:$FA0B
X:$FA1C
Codec/mixer test register
Mixer master volume register
Mixer Microphone in volume register
Mixer Line In volume register
Mixer FM In volume register
Mixer DAC In volume register
Mixer Record Select register
Analog ADC gain register
Mixer Power down/control stat register
Mixer Test Register
Table 411. Mixer Address Registers
26.2. Mixer Block Diagram
The analog level of the signals mixed throughout the chip can be controlled through
registers. The block diagram is shown in Figure 103. The microphone channel is a
mono source, so its signal is available on both the left and right channels.
HW_MIXLINE1INVR
LINEIN
HW_MIXMICINV
HW_MIXMASTERVR
MICIN
+
HW_MIXLINE2INVR
OUTPUT
HW_HPCTRL
FMIN
HW_MIXDACINVR
FROM
DAC
BUFFER
DAC
CLK
EN
HW_DACCSR[0]
HW_DACSRR
CLK
CRYSTAL
HW_CCR[22:20]
HW_CCR[3]
HW_CCR[18]
HW_MIXRECSELR
HW_MIXADCGAINR
LINEIN
FMIN
ADC
MICIN
CLK
CLK
CRYSTAL
HW_CCR[7:5]
HW_CCR[18]
HW_ADCSRR
TO ADC
BUFFER
EN
HW_ADCCSR[0]
HW_CCR[3]
Figure 102. Mixer Flow Diagram
Note that when the mixer is powered down (using the PR2 field of the
HW_MIXPWRDNR register), the DAC can still play audio through the headphone
amplifier to the output. This mode has the dual advantage of saving the power
5-35xx-D1-1.08-061705
297
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
consumption of the analog mixer, as well as eliminating the mixer as a source of
SNR or THD loss.
LINEIN
LINEIN
MICIN
MIC+20dB
bits of
MICIN reg
LINEIN
MICIN
Headphone
(HPH)
+
FMIN
FMIN
MICIN
Headphone
(HPH)
Output
Output
FMIN
Master Volume Stage
(MASTERV)
DACIN
DACIN
MIC+20dB
bits of
MICIN reg
MICIN
DACIN
ADC Mux.
(RECSELECT)
Master Volume Stage
(MASTERV)
DACIN
ADC Mux.
(RECSELECT)
ADC In
ADCGAIN
ADC In
ADCGAIN
(PR2 = 0)
(PR2 = 1)
Mixer Powered Up
Mixer Powered Down
Figure 103. Mixer Block Diagram
26.3. Mixer Programming Model
The various mixer control registers are summarized below.
26.3.1.
Mixer Master Volume Register
The ML field adjusts the level for the left channel master output, while the MR field
adjusts the level for the right channel master output. Each increment of the ML/MR
fields represents 2.0dB1 of attenuation of the master volume output on the mixer.
The Mute bit will silence the output regardless of the current settings of the ML/MR
bits.
HW_MIXMASTERVR X:$FA04
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
MR
2
1
ML
2
2
MUTE
2
3
Table 412. HW_MIXMASTERVR
1.The master volume level at a value of ML or MR=00010 on the STMP35xx corresponds to the same level
as a value of ML or MR=00000 on the STMP3410. NOTE: dB levels are approximate values at each code
point.
298
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS LABEL RW RESET
23:16 RSRVD R
0
15
MUTE RW 1
14:13
12:8
7:5
4:0
RSRVD
ML
RSRVD
MR
R
RW
R
RW
0
00010
0
00010
DEFINITION
Reserved – Must be written with 0.
Mixer Master mute control
0
unmuted
1
muted
Reserved – Must be written with 0.
Mixer Master volume control for left channel
Reserved – Must be written with 0.
Mixer Master volume control for right channel
Table 413. Mixer Master Volume Register Description
MUTE
ML/MR
LEVEL
0
0
1
0 0000
1 1111
X XXXX
+6.0 dB
-56 dB
-infinity
Table 414. Mixer Master Volume Register values
26.3.2.
Analog Mixer Volume Registers
This section refers to the MicIn, Line-In, Line-In 2 (also know as FM-In), and DAC
registers listed below. The GL fields adjust the gain for the left channel of the analog
input, while the GR fields adjust the gain for the right channel of the analog input.
For the MicIn, there is only one channel adjusted by the GN field, although the
results of this volume stage are available on both channels of the mixer. Each increment of the GL/GR/GN fields represents -1.5 dB of adjustment in the gain level. All
analog mixer inputs have a mute bit which will silence this analog input regardless of
the settings of the GL/GR/GN bits. The MicIn additionally has a 20 dB boost bit
which provides a fixed 20 dB boost to the microphone signal when set. Unlike the
master volume, each analog mixer input has the ability to apply gain or attenuation
to the analog signal.
MUTE
0
0
0
1
GL/GR/GN
0 0000
0 1000
1 1111
X XXXX
LEVEL
+12 dB
0 dB
-34.5 dB
-infinity
Table 415. Mixer Volume Register values
Mixer Microphone-In Volume Register
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
MUTE
2
3
X:$FA05
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
GN
HW_MIXMICINVR
20DB
26.3.2.1.
Table 416. HW_MIXMICINVR
5-35xx-D1-1.08-061705
299
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
23:16
15
14:7
6
LABEL
RSRVD
MUTE
RSRVD
20DB
RW
R
RW
R
RW
RESET
0
1
0
0
5
4:0
RSRVD R
0
GN
RW 01000
DEFINITION
Reserved – Must be written with 0.
Mixer Microphone In mute
Reserved – Must be written with 0.
Mixer Microphone In boost
0
no boost
1
20 dB boost
Reserved – Must be written with 0.
Mixer Microphone-In volume
Table 417. Mixer Microphone-In Volume Register Description
The external microphone needs a bias voltage to enable it to operate. This bias voltage can be generated externally using discrete components as shown in
Figure 104, or if either the LRADC1 or LRADC2 pins are available, one of them can
be used to supply a bias voltage from an on-chip generator, as shown in Figure 105.
To enable the generation of the microphone bias voltage on pin LRADC1 or
LRADC2, the two MICBIAS bits in the HW_MIXTBR need to be written with
required values for desired internal resistor selection. To select either pin LRADC1
or LRADC2 as mic bias source, write the MIC_BIAS bit (SDK bit name is
MIC_BIAS_OUT_SEL) in register HW_MIXTBR as follows: 0 for pin LRADC1, 1 for
pin LRADC2.
MIC
0.1µF
Microphone
2.2k
VddIO 2.2k
0.1µF
10µF
Figure 104. External Microphone Bias Generation
300
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
0.1µF
Microphone
2.2K
MIC
HW_MIXTBR_MICBIAS_OUT_SEL
LRADC1 (opt.)
LRADC2 (opt.)
HW_MIXTBR_
MICBIAS
10µF
2k
4k
8k
2.0V
Figure 105. Internal Microphone Bias Generation
Mixer Line-In Volume Register
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
MUTE
2
3
X:$FA06
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
GR
HW_MIXLINE1INVR
GL
26.3.2.2.
Table 418. HW_MIXLINE1INVR
BITS
23:16
15
14:13
12:8
7:5
4:0
LABEL
RSRVD
MUTE
RSRVD
GL
RSRVD
GR
RW
R
RW
R
RW
R
RW
RESET
0
1
0
01000
0
01000
DEFINITION
Reserved – Must be written with 0.
Mixer Line-In mute
Reserved – Must be written with 0.
Mixer Line-In left channel volume
Reserved – Must be written with 0.
Mixer Line-In right channel volume
Table 419. Mixer Line-In Volume Register Description
26.3.2.3.
Mixer Line-In 2 Volume Register
Line-In 2 is also know as FM-In, although nothing about this input restricts it to only
be used with an FM source, it can also be used with other sources. This input channel is only available in 144-pin package versions of the chip.
5-35xx-D1-1.08-061705
301
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
6
0
5
0
4
0
3
GL
MUTE
2
3
X:$FA07
0
2
0
1
0
0
0
1
0
0
0
1
0
0
GR
HW_MIXLINE2INVR
Table 420. HW_MIXLINE2INVR
BITS LABEL RW RESET
23:16
15
14:13
12:8
7:5
4:0
RSRVD
MUTE
RSRVD
GL
RSRVD
GR
R
RW
R
RW
R
RW
0
1
0
01000
0
01000
DEFINITION
Reserved – Must be written with 0.
Mixer Line-in 2 mute
Reserved – Must be written with 0.
Mixer Line-in 2 left channel volume
Reserved – Must be written with 0.
Mixer Line-in 2 right channel volume
Table 421. Mixer Line-in 2 Volume Register Description
Mixer DAC In Volume Register
HW_MIXDACINVR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
GL
MUTE
2
3
X:$FA08
0
2
GR
26.3.2.4.
Table 422. HW_MIXDACINVR
BITS LABEL RW RESET
23:16
15
14:13
12:8
7:5
4:0
RSRVD
MUTE
RSRVD
GL
RSRVD
GR
R
RW
R
RW
R
RW
0
1
0
01000
0
01000
DEFINITION
Reserved – Must be written with 0.
Mixer DAC mute – This bit is only valid when PR2 in HW_MIXPWRDNR = 0.
Reserved – Must be written with 0.
Mixer DAC Left channel volume
Reserved – Must be written with 0.
Mixer DAC Right channel volume
Table 423. Mixer DAC In Volume Register Description
Mixer Record Select Register
HW_MIXRECSELR
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
SL
2
3
X:$FA09
0
6
0
5
0
4
0
3
0
2
SR
26.3.2.5.
Table 424. HW_MIXRECSELR
302
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS LABEL RW RESET
23:12
10:8
7:3
2:0
RSRVD
SL
RSRVD
SR
R
RW
R
RW
0
000
0
000
DEFINITION
Reserved – Must be written with 0.
Mixer Record Left channel select
Reserved – Must be written with 0.
Mixer Record Right channel select
Table 425. Mixer Record Select Register Description
SR
RIGHT RECORD SELECT
000
001
010
011
100
101
110
111
SL
Mic-in
Reserved
Reserved
Line-in2 (AKA FM-in)
Line-in
StereoMix
Reserved
Reserved
LEFT RECORD SELECT
000
001
010
011
100
101
110
111
Mic-in
Reserved
Reserved
Line-in2 (AKA FM-in)
Line-in
StereoMix
Reserved
Reserved
Table 426. Mixer ADC Select Register
26.3.3.
Mixer ADC Gain Register
The record gain register provides gain only to the analog signal being input into the
ADC. The GL field adjusts the left channel, while the GR field adjusts the right
channel. Each increment of these fields represents -1.5 dB of gain. The mute bit will
silence the input regardless of the settings of the GL/GR fields.
HW_MIXADCGAINR X:$FA0A
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
GR
2
1
GL
2
2
MUTE
2
3
Table 427. HW_MIXADCGAINR
BITS LABEL RW RESET
23:16
15
14:12
11:8
7:4
3:0
RSRVD
MUTE
RSRVD
GL
RSRVD
GR
R
RW
R
RW
R
RW
0
1
0
0000
0
0000
DEFINITION
Reserved – Must be written with 0.
Mixer ADC Gain mute
Reserved – Must be written with 0.
Mixer ADC Gain left channel
Reserved – Must be written with 0.
Mixer ADC Gain right channel
Table 428. Mixer ADC Gain Register Description
5-35xx-D1-1.08-061705
303
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
MUTE
GL/GR
LEVEL
0
0
1
1111
0000
XXXX
+22.5 dB
0 dB
-infinity
Table 429. Mixer ADC Gain Register
Mixer Power Down Control/Status Register
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
PR0
2
3
X:$FA0B
PR1
HW_MIXPWRDNR
PR2
26.3.4.
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
Table 430. HW_MIXPWRDNR
BITS LABEL RW RESET
23:12 RSRVD R
0
11
PR2
RW 1
10
PR1
RW 1
9
PR0
RW 1
8:0
RSRVD R
DEFINITION
Reserved – Must be written with 0.
Power down analog mixer
0
Analog mixer powered up
1
Analog mixer powered down
When the analog mixer is powered down, it is still possible to play audio through the
DAC. The reference voltage generator is still powered up, and if the DAC is still
powered up and playing audio, the audio is routed directly into the master volume
stage, bypassing the analog mixer. Aside from saving power, this mode of operation
has the added advantage of providing better SNR/THD performance.
Power down DAC analog circuitry
0
DAC & input mux powered up
1
DAC & input mux powered down
Power down ADC & input mux analog circuitry
0
ADC & input mux powered up
1
ADC & input mux powered down
Reserved – Must be written with 0.
0
Table 431. Mixer Power Down Control/Stat Register Description
BIT
PR0
PR1
PR2
FUNCTION
PCM in ADCs and Input Mux Powerdown
PCM out DACs Powerdown
Analog Mixer Powerdown (Vref still on)
(Please see Figure 103.)
Table 432. Mixer Power Down Register
26.3.5.
Codec/Mixer Test Register
The Codec/Mixer test register contains some bits that control functionality that was
designed as options, and that was not expected to be used. It also contains control
bits for functions that are available for general use. Please pay careful attention to
the notes in each bit field which indicate whether that field is available for general
usage. If it is marked unavailable then be certain to only write its reset value when
304
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
modifying other bits in the register. The programmer shouldn’t change the
unavailable bit fields in this register without being directed to do so by SigmaTel, as
undesired operation may result. The organization of the Codec/Mixer Test register is
shown below.
ADTHD
XBGC
XBCO
VCOS
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
PLL_ZERO_VT
MICBIAS0
0
9
HOLD_GND
PWDADC
1
0
ACKI
1
1
ASD2X
1
2
PCPCU
1
3
DCKI
1
4
PCPCD
1
5
PSRN
1
6
MIC_BIAS_OUT_SEL
1
7
FX2
1
8
EZD
1
9
MICBIAS1
2
0
DZCDA
DZCMI
2
1
DZCLI
2
2
X:$FA03
DZCFM
2
3
DZCMA
HW_MIXTBR
0
0
Table 433. HW_MIXTBR
BITS
LABEL
RW
RESET
DEFINITION
23
DZCMA
RW
0
Enable zero crossing volume update for master volume
stage. This bit field is available for general usage.
22
DZCMI
RW
0
Enable zero crossing volume update for microphone
volume stage. This bit field is available for general usage.
21
DZCLI
RW
0
Enable zero crossing volume update for line-in volume
stage. This bit field is available for general usage.
20
DZCFM
RW
0
Enable zero crossing volume update for line-in 2 (also
known as FM-in) volume stage. This bit field is available for
general usage.
19
DZCDA
RW
0
Enable zero crossing volume update for DAC volume stage
– (Use only when mixer is powered up, HW_MIXPWRDNR PR2
= 0. DAC zero crossing volume updates are done digitally when
the mixer is powered down, HW_MIXPWRDNR PR2 = 1). This
bit field is available for general usage.
18
EZD
RW
0
Enable zero crossing detect for the volume stage
determined by bits 23-19 – Note that one and only one of the
bits 23-19 must be set high for the zero crossing detect to
properly function. Also, there is a 10Hz timeout used, so if a
zero crossing is not detected within a 10Hz period, the volume
is automatically updated. Only one bit in the range 23-19 can be
enable at a time for zero crossing detection. This bit field is
available for general usage.
17
MICBIAS1
RW
0
SEE MICBIAS field below. WARNING this is a split field with
PWDADC in the middle.
Table 434. Codec/Mixer Test Register Description
5-35xx-D1-1.08-061705
305
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
RW
RESET
DEFINITION
16
PWDADC
RW
0
Power down ADC right channel – This powers down one ADC
channel when the ADC is being used to record only a mono
channel. When set, some power will be saved in the analog
portion of the ADC. The ADC will still generate stereo data,
however only the samples for the left channel will be valid. This
bit reserved for use only by SigmaTel, always set it to zero.
17,15
MICBIAS
RW
00
Microphone resistor select – Provides an option for reducing
board complexity by integrating some bias circuitry for the
microphone. This circuit provides a regulated supply connected
via a selectable resistor value to the line-in right pin. This allows
users to connect the microphone to the LRADC1 or LRADC2
pins and then externally place a capacitor to the microphone
input to implement the microphone function. This functionality
cannot be used at the same time that the selected LRADC input
is needed. See the microphone volume register for a figure
detailing the circuit. WARNING this two bit field is split by
PWDADC in bit 16.
00
01
Mic bias off
2kohm
10
11
4kohm
8kohm
14
ADTHD
RW
0
Turn off ADC dither. This bit is reserved for use only by
SigmaTel, always set it to zero.
13
XBGC
RW
0
Causes the xtal oscillator to use the band gap bias current,
instead of its self-generated bias current. This bit is reserved
for use only by SigmaTel, always set it to zero.
12
XBCO
RW
0
Turns off xtal internal bias current. This bit is reserved for use
only by SigmaTel, always set it to zero.
11
VCOS
RW
0
Speed up PLL VCO. This bit is reserved for use only by
SigmaTel, always set it to zero.
10
FX2
RW
0
Double frequency reference to the PLL. This bit is reserved
for use only by SigmaTel, always set it to zero.
9
PSRN
RW
0
Disable fast falling edge power down function of PSWITCH
pin. This bit is reserved for use only by SigmaTel, always set it
to zero. NOTE: setting this bit to one overrides all other sources
of reset to the chip as well.
8
MIC_BIAS
RW
0
Enable Mic Bias. This bit controls the mic bias source
selection.
0 - mic bias from LRADC1
1 - mic bias from LRADC2
7
DCKI
RW
0
Invert DAC clock. This bit is reserved for use only by SigmaTel,
always set it to zero.
6
PCPCD
RW
0
Decrease PLL charge pump current 2X. This bit is reserved
for use only by SigmaTel, always set it to zero.
5
PCPCU
RW
0
Increase PLL charge pump current 2X. This bit is reserved for
use only by SigmaTel, always set it to zero.
4
ASD2X
RW
0
Slow ADC dither. This bit is reserved for use only by SigmaTel,
always set it to zero.
3
ACKI
RW
0
Invert ADC clock. This bit is reserved for use only by SigmaTel,
always set it to zero.
Table 434. Codec/Mixer Test Register Description (Continued)
306
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
RW
RESET
DEFINITION
2
HOLD_GND
RW
0
Hold at Ground. This bit can be used to reduce “pops” at
startup. It causes the headphone output to be held at ground.
1
PLL_ZERO_VT
RW
0
Select Zero Vt Devices. This bit controls the special PLL zero
VT mode.
0 - PLL Normal VT
1 - PLL Zero VT
0
RSRVD
R
0
Reserved – Must be written with 0.
Table 434. Codec/Mixer Test Register Description (Continued)
Reference Control Register
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
DACVBGVAL
1
6
ADJDAC
1
7
VAGVAL
1
8
ADJV
1
9
ADCREFV
2
0
ADJADC
2
1
PWRDWNS
2
2
LWREF
2
3
X:$FA19
LOW_PWR
HW_REF_CTRL
BIASC
26.3.6.
Table 435. HW_REF_CTRL
BITS
LABEL
RW RESET
23:20 RSRVD
19
LOW_PWR
R
0
RW 0
18
LWREF
17:16 BIASC
RW 0
RW 0
15
PWRDWNS
RW 0
14
ADJADC
RW 0
13:10 ADCREFV
RW 0
DEFINITION
Reserved – Must be written with 0.
Lowers power in the band gap amplifier. This mode is useful in USB suspend or
standby when absolute band gap accuracy is not critical.
lower voltage in 13:10, 8:5, 3:0 by 20%
Bias current control
00
nominal
10
-10%
01
-20%
11
+10%
These bits control the bias currents sent to all the analog circuits from the band
gap generator.
Powers down selfbias circuit – The reference uses a self bias circuit during
powerup that can be turned off with this bit. However, care must be taken that
the DC-DC converter must be using the bandgap-generated current before the
selfbias is powered down (bit [15] in $fa14).
Adjust ADC reference using bits [13:10], default 1.5V derived from the
bandgap
ADC reference value
1111
1.600V
1001
1.450V
0100
1.325V
1110
1.575V
1000
1.425V
0011
1.300V
1101
1.550V
0111
1.400V
0010
1.275V
1100
1.525V
0110
1.375V
0001
1.250V
1011
1.500V
0101
1.350V
0000
1.225V
1010
1.475V
Table 436. Reference Control Register Description
5-35xx-D1-1.08-061705
307
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
RW RESET
9
ADJV
RW 0
8:5
VAGVAL
RW 0
4
ADJDAC
RW 0
3:0
DACVBGVAL RW 0000
DEFINITION
These bits adjust the value of the reference used by the ADC, and can be used
to allow acceptable analog performance at low supply voltages. Voltages
shown can be lowered 20% lower by setting bit 18.
Adjust Vag – Default is a resistor divider from the analog power supply. These
bits adjust the value of the analog reference used throughout the codec. This
value should be programmed to be near half of the analog target supply
voltage.
Vag Value
1111
1.000V
1001
0.850V
0100
0.725V
1110
0.975V
1000
0.825V
0011
0.700V
1101
0.950V
0111
0.800V
0010
0.675V
1100
0.925V
0110
0.775V
0001
0.650V
1011
0.900V
0101
0.750V
0000
0.625V
1010
0.875V
Voltages shown can be lowered 20% lower by setting bit 18.
Adjust DAC reference value – Default is the band gap voltage. These bits
adjust the value of the reference used by the d/a converters, and can be used
to allow acceptable analog performance at low supply voltages.
NOTE: These adjustments also affect the VBG used by the USB 2.0 PHY. If the
PHY is transmitting, this bit should be set to zero.
DAC Vbg value
1111
1.300V
1001
1.150V
0011
1.000V
1110
1.275V
1000
1.125V
0010
0.975V
1101
1.250V
0111
1.100V
0001
0.950V
1100
1.225V
0110
1.075V
0000
0.925V
1011
1.200V
0101
1.050V
1010
1.175V
0100
1.025V
Voltages shown can be lowered 20% lower by setting bit 18.
Table 436. Reference Control Register Description (Continued)
MIXER TEST REGISTER
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
DAC_CHOP_CLK
2
1
DAC_DISABLE_RTZ
2
2
TMPPWD
2
3
X:$FA1C
DAC_MORE_AMP_I
HW_MIX_TEST
TMP_CFG
26.3.7.
Table 437. HW_MIX_TEST
308
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
RW RESET
23:12 RSRVD
12
TMPPWD
R
0
RW 1
11:8
TMP_CFG
RW 0000
7:6
5
4
RSRVD
DAC_MORE_AMP_I
DAC_DISABLE_RTZ
R
00
RW 0
RW 0
3:2
1:0
RSRVD
DAC_CHOP_CLK
R
00
RW 00
DEFINITION
Reserved – Must be written with 0.
Set to one to power down the temperature sensor current
source. This current source is only available on LRADC2.
Each bit turns on a separate current source and the source
currents are additive.
0001 = 20uA
0010 = 40uA
0100 = 80uA
1000 = 160uA
1111 = 300uA
NOTE: set HW_USBPHYPWD_PWDIBIAS to zero and
set HW_REF_CTRL_LWREF to zero for proper operation.
Reserved – Must be written with 0.
Set to one to increase the current in the DAC I amplifier
Set to one to disable DAC RTZ mode. Set to zero for
normal operation with RTZ enabled.
Reserved – Must be written with 0.
Enable a clock that may improve signal-to-noise
performance by chopping the DAC opamp input.
00 - chop clock disabled
01 - 384KHz chop clock
10 - 192KHz chop clock
11 - 96KHz chop clock
Table 438. Mixer Test Register Description
5-35xx-D1-1.08-061705
309
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
27. HEADPHONE DRIVER
The headphone driver is designed to directly drive low impedance (16 ohm) headphones. It has built in pop-suppression circuits. It can automatically detect a headphone short situation and report it to the DSP via maskable or non-maskable
interrupts.
The STMP35xx includes a common mode amplifier which can present a headphone
common node, eliminating the need for large expensive DC blocking capacitors in
the headphone circuit.
27.1. Headphone Control Register
The organization of the Headphone Control Register is shown below.
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
TESTIALLUP
1
0
TESTIALLDWN
1
1
TESTI1UP
1
2
POP0
1
3
TESTI1DWNP
1
4
POP1
1
5
POP2
1
6
HPHPWD
1
7
CAPLESS
1
8
CLASSAB
1
9
SHORTMODE_LR
2
0
SHORTMODE_CM
SHORT_CM
2
1
X:$FA15
SHORT_LVLADJ
2
2
CHOP_CLK
2
3
_SHORT_LR
HW_HPCTRL
Table 439. HW_HPCTRL
BITS
LABEL
23
SHORT_CM
22
SHORT_LR
21:20
CHOP_CLK
19
RSRVD
RW RESET
DEFINITION
R
0
Status of common mode amplifier short detection.
0 = no short detected
1 = short present
To clear this interrupt, set HW_HPCTRL_SHORTMODE_CM to 00. This has
the effect of resetting the latch and holding it in reset.
If HW_HPCTRL_SHORTMODE_CM is changed back to 10 or 11 and the
short still exists then interrupts will continue on IRQA, see Figure 108. “Stereo
Headphone Common Short Detection & Powerdown Circuit” on page 313.
R
0
Status of right/left differential amplifier short detection.
0 = no short detected
1 = short present
To clear this interrupt, set HW_HPCTRL_SHORTMODE_LR to 00. This has
the effect of resetting the latch and holding it in reset.
If HW_HPCTRL_SHORTMODE_LR is changed back to 10 or 11 and the
short still exists then interrupts will continue on IRQA.
RW 00
Enable a chop clock to improve signal-to-noise performance by chopping the
head phone amplifier input.
00 = chop clock disabled
01 = 48KHz chop clock
10 = 96KHz chop clock
11 = 192KHz chop clock
R
0
Reserved – Must be written with 0.
Table 440. Headphone Control Register Description
310
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
18:16 SHORT_LVLA
DJ
RW RESET
DEFINITION
RW 000
Adjusts headphone current short detect trip point
000 = Nominal
001 = Nominal * 0.75
010 = Nominal * 0.5
011 = Nominal * 0.25
100 = Nominal * 1.25
101 = Nominal * 1.5
110 = Nominal * 1.75
111 = Nominal * 2
Increasing trip point current makes the detection less sensitive, while
decreasing the trip point current makes it more sensitive.
15:14 SHORTMODE_ RW 00
Headphone Common Amplifier Short Control Mode
00 - SW view direct short, reset latched short, HW pwdn on direct short
CM
01 - SW view latched short HW pwdn on latched short
10 - sw view direct short signal, HW pwdn on direct short
11 - sw view direct short signal, HW pwdn disabled
The state of the common mode short detector, or its latched version can be
seen in HW_HPCTRL_SHORT_CM.
13:12 SHORTMODE_ RW $00
Headphone L/R Short Control Mode
00 - short function held in reset
LR
01 - sw view latched short signal, HW pwd on latched signal
10 - sw view latched short signal, HW pwd disabled
11 - sw view direct short signal, HW pwd disabled
The state of the left/right mode short detector, or its latched version can be
seen in HW_HPCTRL_SHORT_LR.
11
RSRVD
R
0
Reserved – Must be written with 0.
10
CAPLESS
RW 0
Capless – Set to one to enable the direct drive or “cap-less” headphone
mode. Set to zero for conventional operation.
9
CLASSAB
RW 0
ClassAB – ClassA mode is intended only for powerup anti-pop. ClassAB
mode should be set before a signal is applied. ClassAB mode should be used
to drive both a low impedance (e.g. headphone) or high impedance (e.g. linein to another device) load.
8
HPHPWD
RW 1
Headphone Powerdown
0 Headphone amplifier powered up
1 Headphone amplifier powered down (reset value)
7
RSRVD
R
0
Reserved – Must be written with 0.
6
POP2
RW 0
pop2
5
POP1
RW 0
pop1
4
POP0
RW 0
pop0
hphpwd classab pop2
pop1 pop0
ramp result
expected output pop
[bit 10]
[bit 7]
[bit 6] [bit 5] [bit 4]
0
0
0
0
0
1*84uA PMOS source current
1.3mV
0
0
0
0
1
2*84uA PMOS source current
1.3mV
0
0
0
1
0
3*84uA PMOS source current
1.3mV
0
0
1
1
1
8*84uA PMOS source current
1.3mV
1
x
0
0
0
11.0k pull-down resistor
1.3mV
1
x
0
0
1
5.6k pull-down resistor
depends on timing
1
x
0
1
0
3.7k pull-down resistor
depends on timing
1
x
0
1
1
2.8k pull-down resistor
depends on timing
1
x
1
0
0
1.9k pull-down resistor
depends on timing
1
x
1
0
1
1.2k pull-down resistor
depends on timing
1
x
1
1
0
0.8k pull-down resistor
depends on timing
1
x
1
1
1
0.4k pull-down resistor
depends on timing
Table 440. Headphone Control Register Description (Continued)
5-35xx-D1-1.08-061705
311
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
3
TESTI1DWN
2
TESTI1UP
1
0
RW RESET
RW 0
test i1down
RW 0
test i1up
I1up
I1dn
0
0
0
1
1
0
1
1
TESTIALLUP
RW 0
test iall up
TESTIALLDWN RW 0
test iall down
Iallup
Ialldown
0
0
0
1
1
0
1
1
DEFINITION
Change to 1st stage current
nominal
-50%
+100%
+50%
Change to 1st stage current
nominal
-50%
+50%
-40%
Table 440. Headphone Control Register Description (Continued)
27.2. Headphone Driver
The STM35xx supports a conventional stereo headphone drive as shown in Figure
106.
.01µF
100 Ω
16 Ω
16 Ω
100 Ω
16 Ω
Headphones
16 Ω
.01µF
220 µF
65
HPR
220 µF
62
HPL
Figure 106. Conventional Stereo Headphone Application Circuit
In addition, the chip can generate an optional headphone common node circuit for
the headphones which eliminates the need for the large and expensive DC blocking
capacitors. It also improves the anti-pop performance. These benefits are obtained
at a slight increase in power consumptions, i.e. at 30 mV rms output, the resultant
increase in power consumption is approximately 2.7 mW.
312
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
16 Ω
.01µF
.01µF
65
59
HPR
16 Ω
.01µF
57
HP_COMMON
16 Ω
16 Ω
100 Ω
tie at headphone jack
100 Ω
16 Ω
Headphones
62
HP_SENSE
HPL
Figure 107. Stereo Headphone Application Circuit with Common Node
HP
AMP
Headphone Common
HW_CTRL_SHORTMODE_CM != `00'
HW_CTRL_SHORTMODE_CM == `01'
IRQA
SHORT
DETECT
0
HW_CTRL_SHORT_CM
HW_CTRL_SHORTMODE_CM == `00'
1
Latch
S
R
0
1
Powerdown
Headphone Amp
HW_CTRL_SHORTMODE_CM == `01'
HW_CTRL_SHORTMODE_CM != `11'
Figure 108. Stereo Headphone Common Short Detection & Powerdown Circuit
5-35xx-D1-1.08-061705
313
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
HP
AMP
Headphone Common
HW_CTRL_SHORTMODE_LR != `00'
HW_CTRL_SHORTMODE_LR == `01' ||
HW_CTRL_SHORTMODE_LR == `10'
SHORT
DETECT
0
HW_CTRL_SHORT_LR
1
Latch
S
HW_CTRL_SHORTMODE_LR == `00'
IRQA
R
HW_CTRL_SHORTMODE_LR != `01'
Powerdown
Headphone Amp
Figure 109. Stereo Headphone L/R Short Detection & Powerdown Circuit
314
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
28. LOW RESOLUTION ADC
The low resolution ADC (LRADC) block is used for voltage measurement. Three
channels are available. One channel is dedicated to measure the voltage on the
BATT pin. The other channels, LRADC1 and LRADC2, measure the voltage on the
two application dependent LRADC pins. The battery channel can be used to sense
the amount of battery life remaining. The auxiliary channels can be used for a variety of different uses, including a resistor based wired remote control, temperature
sensing, etc. The LRADC is accurate to 8 bits of resolution, and samples at a
divided rate from the 24.0MHz crystal clock
H W _ B A T T _ T H R S H [2 0 :1 2 ]
EVENT 1
B A N D G A P 2 .7 V
V D D IO
H W _ B A T T _ C T R L [1 7 :1 6 ]
REF
OUT
BATT
2
IN
H W _ B A T T _ R E S U L T [2 ]
LT
H W _ B A T T _ R E S U L T [4 ]
[8 :0 ]
H W _ B A T T _ R E S U L T [0 ]
H W _ B A T T _ R E S U L T [1 6 :8 ]
ADC
OFFSET
PW D
H W _ B A T T _ C T R L [6 :0 ]
EVENT 0
H W _ B A T T _ C T R L [9 ]
GT
EQ
H W _ B A T T _ R E S U L T [3 ]
GT
EQ
H W _ B A T T _ R E S U L T [1 ]
LT
H W _ B A T T _ R E S U L T [5 ]
H W _ B A T T _ C T R L [1 1 ]
H W _ B A T T _ T H R S H [8 :0 ]
H W _ L R A D C 1 _ T H R S H [2 0 :1 2 ]
EVENT 1
B A N D G A P 2 .7 V
V D D IO
H W _ L R A D C 1 _ C T R L [1 7 :1 6 ]
REF
OUT
LRADC1
2
IN
H W _ L R A D C 1 _ R E S U L T [0 ]
LT
H W _ L R A D C 1 _ R E S U L T [4 ]
[8 :0 ]
H W _ L R A D C 1 _ R E S U L T [1 6 :8 ]
ADC
OFFSET
PW D
H W _ L R A D C 1 _ C T R L [6 :0 ]
EVENT 0
H W _ L R A D C 1 _ C T R L [9 ]
H W _ L R A D C 1 _ R E S U L T [2 ]
GT
EQ
GT
EQ
H W _ L R A D C 1 _ R E S U L T [3 ]
LT
H W _ L R A D C 1 _ R E S U L T [5 ]
H W _ L R A D C 1 _ C T R L [1 1 ]
H W _ L R A D C 1 _ R E S U L T [1 ]
H W _ L R A D C 1 _ T H R S H [8 :0 ]
H W _ L R A D C 2 _ T H R S H [2 0 :1 2 ]
EVENT 1
B A N D G A P 2 .7 V
V D D IO
H W _ L R A D C 2 _ C T R L [1 7 :1 6 ]
REF
OUT
LRADC2
2
IN
H W _ L R A D C 2 _ C T R L [1 1 ]
H W _ L R A D C 2 _ R E S U L T [4 ]
H W _ L R A D C 2 _ R E S U L T [1 6 :8 ]
ADC
OFFSET
H W _ L R A D C 2 _ C T R L [6 :0 ]
H W _ L R A D C 2 _ R E S U L T [0 ]
LT
[8 :0 ]
PW D
EVENT 0
H W _ L R A D C 2 _ C T R L [9 ]
H W _ L R A D C 2 _ R E S U L T [2 ]
GT
EQ
GT
H W _ L R A D C 2 _ R E S U L T [3 ]
EQ
H W _ L R A D C 2 _ R E S U L T [1 ]
LT
H W _ L R A D C 2 _ R E S U L T [5 ]
H W _ L R A D C 2 _ T H R S H [8 :0 ]
Figure 110. Low Resolution ADC
5-35xx-D1-1.08-061705
315
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
As shown in Figure 110, the LRADC module consists of three essentially identical
analog to digital converters. These converters run in the crystal clock domain for
sample interval precision. Programmable registers for each of the three converters
are declared as HW_BATT_*, HW_LRADC1_* and HW_LRADC2_*. In the discussion below, the battery LRADC (HW_BATT_*) is used to describe the general operation of all three converters.
Consider the battery LRADC channel, the converter can take its voltage reference
either from the bandgap or I/O voltage rail (VddIO). Thus high voltage LiIon batteries
can be monitored with the appropriate settings of HW_BATT_CTRL_REF_VAL.
Once an eight bit A/D conversion is made, a seven bit digital offset can be added to
it to slide the eight bit value up and down within a nine bit value. Thus the converter
presents an eight bit precision result in a nine bit range. The offset can be used to
calibrate an LRADC instance if desired.
2.7V
(from BG)
decoder
256
8
Average 4
Samples
50KΩ
UP/DN Counter
BATT
R-LADDER
HW_BATT_CTRL[17:16]
VddIO
+
50KΩ
HW_BATT_CTRL[9]
HW_BATT_CTRL[10]
HW_BATT_CTRL[6:0]
clear
+
Σ/4
9
LRADC Output
digital offset
Figure 111. Low Resolution ADC Detail
The nine bit result is compared against two independent threshold values, Event
Detector 0 and Event Detector 1. Each event detector reports boolean less than,
greater than and equal bits which can be seen directly in the result register. These
event booleans can be used to trigger DSP interrupts as discussed below. The current
value
of
the
LRADC
conversion
can
be
read
in
HW_BATT_RESULT_DATA_OUT.
WARNING, the conversion happens in the crystal clock domain while the result register is read by the DSP in the dclk domain. These clock domains are asynchronous
when the DSP dclk is derived from a PLL. The clock domain crossing can lead to
uncertainty in the correct reading of individual bits when the conversion value
316
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
changes. As a result, software should read this register repeatedly until two successive readings are identical.
Figure 111 shows additional detail of the LRADC itself. Notice that the bias resistor
at the bottom of the resistive ladder can be bypassed by
HW_BATT_CTRL_REF_VAL. This has the effect of shifting the R-Ladder taps
down in voltage and making their step sizes larger. When the bias resistor is
enabled, one can place a calibration offset into HW_BATT_CTRL_INPUT_OFFSET
to account for the voltage dropped across the bias resistor. Also notice that the input
voltage can be divided by two by setting HW_BATT_CTRL_INPUT_DIV2 to one.
This has the side effect of changing the input resistance from essentially infinite to
100KΩ. The following table summarizes the input ranges and step size resolution
for various settings of REF_VAL and INPUT_DIV2.
REF_VAL
11
11
10
INPUT_DIV2
1
0
1
INPUT RANGE
2*VDDIO - 0V
VDDIO - 0V
5.12V - 0V
RESOLUTION
VDDIO/128
VDDIO/256
0.020V
10
01
01
00
0
1
0
1
2.56V - 0V
5.2V - 1.256V
2.6V - 0.628V
5.4V - 1.304V
0.010V
00
0
2.7V - 0.652V
PROG. OFFSET
0x00
0x00
0x00
BATTERY
Quad NiMH
(4.8V - 3.6V)
LiIon
(4.4V - 2.6V)
0x00
This mode is not recommended
This mode is not recommended
0.016V
0x51
Double Cell
(3.4V - 1.5V)
0.008V
0x51
Single Cell
(1.7V - 0.8V)
Table 441. LRADC Input Ranges and Resolutions
Note:
1. Do not exceed Maximum Pin voltage on any input pin, see Section 3.
“CHARACTERISTICS/SPECIFICATIONS” on page 21.
WARNING: The pad ESD protection limits maximum voltage on all LRADC inputs.
The BATT LRADC is specifically designed to handle higher voltages, but LRADC1
and LRADC2 inputs are limited to 3.3V.
Two other controls of interest are HW_BATT_CTRL_CLEAR which holds the
up/down counter at zero and HW_BATT_CTRL_PWD which powers down the comparator and the resistive ladder.
A linear convergence A/D architecture was selected for cost and power reasons
while maintaining sufficient slew rate performance for brownout detection. This
allows the DSP enough time to perform a graceful shutdown. To reduce the linear
convergence tracking noise, four successive samples are averaged in the hardware
to produce the final A/D value that is used by the event detectors.
Figure 112 shows the generation of an LRADC interrupt from event detector 0 of the
battery LRADC. The interrupt request flip flop can be seen by DSP software in
HW_BATT_RESULT_IRQ_EVENT0 (HW_BATT_RESULT[20]). This bit is “sticky”
in that once it is set it remains set until cleared by software. The polarity of the event
trigger comparison is controlled by HW_BATT_CTRL_POLARITY_EVENT0. The
event 0 interrupt is enabled by HW_BATT_CTRL_IRQ_EN_EVENT0. Notice that
the equality detector is not available as an interrupt source.
5-35xx-D1-1.08-061705
317
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
Because of the four sample averaging, there is a slight hysteresis effect in the event
detectors,
While the above discussion has focused on the Battery LRADC, the operation of
LRADC1 and LRADC2 are essentially the same.
WARNING: The interrupt requests (IRQ) to the DSP is generated somewhat differently in the low resolution ADC, as shown in Figure 112. The IRQ comes from the
unlatched outputs of the event detectors, not from the latched status bit. The status
bit is only a sticky sample of a signal that could come and go many times, i.e. it is
possible for the DSP interrupt to be asserted and then removed before the DSP can
determine the actual interrupt source. It is also possible for the DSP to be interrupted many times even after the sticky bit is set.
EVENT0 GT
ADC
1
HW_BATT_RESULT[20]
EVENT0 LT
EVENT0 EQ
IRQ
FF
X
HW_BATT_CTRL[22]
HW_BATT_CTRL[20]
Event 0 IRQ
Figure 112. Low Resolution ADC Interrupt Generation
28.1. Low Resolution ADC Programmable Registers
The following programmable registers are available to DSP software for controlling
and using the low resolution analog to digital converters.
28.1.1.
LRADC Battery Control Register
This register controls the overall operation of the battery LRADC.
318
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
INPUT_OFFSET
IRQ_EN_EVENT0
1
7
INPUT_DIV2
IRQ_EN_EVENT1
1
8
HALF_CMP_PWR
POLARITY_EVENT0
1
9
CLEAR
2
0
PWD
2
1
CLK_DIV
2
2
X:$FA20
REF_VAL
2
3
POLARITY_EVENT1
HW_BATT_CTRL
Table 442. HW_BATT_CTRL
BITS
LABEL
23
POLARITY_EVENT1
RW RESET
RW 0
22
POLARITY_EVENT0
RW 0
21
IRQ_EN_EVENT1
RW 0
20
IRQ_EN_EVENT0
RW 0
21:18 RSRVD
17:16 REF_VAL
R
00
RW 00$
15:14 RSRVD
13:12 CLK_DIV
R
00
RW 00
11
PWD
RW 1
10
CLEAR
RW 1
9
INPUT_DIV2
RW 0
8
HALF_CMP_PWR
RW 0
DEFINITION
Set to one to trigger event 1 interrupts when filtered A/D
value is greater than the event 1 threshold. Set to zero for
interrupt when the filtered A/D value is less than the event 1
threshold.
Set to one to trigger event 0 interrupts when filtered A/D
value is greater than the event 0 threshold. Set to zero for
interrupt when the filtered A/D value is less than the event 0
threshold.
Set to one to enable an interrupt from the event 1
comparator.
Set to one to enable an interrupt from the event 0
comparator.
Reserved – Must be written with 0.
Reference voltage selection, see Table 441. “LRADC Input
Ranges and Resolutions” on page 317. These are the
primary mode control bits. They affect the range and
resolution of the converter.
Reserved – Must be written with 0.
Clock divider selection.
00 - divide by 4
01 - divide by 8
10 - divide by 16
11 - divide by 32
Set this bit to one to power down the comparator, the RLadder and stop the internal clocks to minimize power
consumption. Set to zero for normal operation.
Set this bit to one to force the up/down counter to all zeroes.
Set to zero for normal operation.
Set to one to divide the input voltage by two before applying
it to the comparator. The division by two changes the input
resistance from essentially infinite to 100KΩ.
Set to one to reduce the analog current for low power
operation. This is accomplished at the expense of
conversion performance. This bit should only be set to one
if static measurements are needed.
Table 443. LRADC Battery Control Register Description
5-35xx-D1-1.08-061705
319
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
7
RSRVD
6:0
INPUT_OFFSET
RW RESET
R
0
RW $00
DEFINITION
Reserved – Must be written with 0.
These bits are added to the final converter output to add a
digital offset to the converted value. This may be useful for
LRADC calibration and to compensate for additional bias
resistor when it is switched into the R-Ladder.
Table 443. LRADC Battery Control Register Description (Continued)
320
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
28.1.2.
LRADC Battery Threshold Register
This register sets the thresholds used by the digital comparators in the battery
LRADC.
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
THRESHOLD_EVENT1
2
3
X:$FA21
0
4
0
3
0
2
0
1
0
0
THRESHOLD_EVENT0
HW_BATT_THRSH
Table 444. HW_BATT_THRSH
BITS
LABEL
23:21 RSRVD
20:12 THRESHOLD_EVENT1
RW RESET
R
000
RW $000
11:9
8:0
R
000
RW $000
RSRVD
THRESHOLD_EVENT0
DEFINITION
Reserved – Must be written with 0.
This nine bit field defines the set point for event detector
one’s threshold comparator. The nine bit post offset
conversion value is compared to this threshold.
Reserved – Must be written with 0.
This nine bit field defines the set point for event detector
zero’s threshold comparator. The nine bit post offset
conversion value is compared to this threshold.
Table 445. LRADC Battery Threshold Register Description
28.1.3.
LRADC Battery Result Register
This register controls the interrupts reported by the battery LRADC and provides
read only views of the LRADC data value and the results of the event detector
threshold comparators.
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
EQ_EVENT1
1
6
EQ_EVENT0
1
7
GT_EVENT1
1
8
LT_EVENT1
1
9
GT_ENVET0
2
0
LT_EVENT0
2
1
X:$FA22
DATA_OUT
2
2
IRQ_EVENT0
2
3
IRQ_EVENT1
HW_BATT_RESULT
Table 446. HW_BATT_RESULT
5-35xx-D1-1.08-061705
321
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
23:22 RSRVD
21
IRQ_EVENT1
RW RESET
R
00
RW 0
20
RW 0
IRQ_EVENT0
19:17 RSRVD
16:8 DATA_OUT
R
R
00
$000
7:6
5
RSRVD
LT_EVENT0
R
R
00
0
4
LT_EVENT1
R
0
3
GT_EVENT0
R
0
2
GT_EVENT1
R
0
1
0
EQ_EVENT0
EQ_EVENT1
R
R
0
0
DEFINITION
Reserved – Must be written with 0.
Interrupt request status for event detector one. This sticky
bit is set whenever event detector one’s comparator trips
the set point. See HW_BATT_CTRL_POLARITY_EVENT1
and HW_BATT_THRSH_THRESHOLD1. Software clears
this bit by writing a one to it.
Interrupt request status for event detector zero. This sticky
bit is set whenever event detector zero’s comparator trips
the set point. See HW_BATT_CTRL_POLARITY_EVENT0
and HW_BATT_THRSH_THRESHOLD0. Software clears
this bit by writing a one to it.
Reserved – Must be written with 0.
The current nine bit value of the BATT LRADC is read here.
WARNING: due to clock domain crossings, software must
repeatedly read this value until two successive values are
read which match bit for bit.
Reserved – Must be written with 0.
This read only bit indicates the current state of the event
detector 0 Less Than result.
This read only bit indicates the current state of the event
detector 1Less Than result.
This read only bit indicates the current state of the event
detector 0 Greater Than result.
This read only bit indicates the current state of the event
detector 1 Greater Than result.
The “excluded muddle”.
The “excluded muddle”.
Table 447. LRADC Battery Result Register Description
322
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
28.1.4.
LRADC1 Control Register
This register controls the overall operation of LRADC 1.
WARNING: The pad ESD protection limits maximum voltage on all LRADC inputs.
The BATT LRADC is specifically designed to handle higher voltages, but LRADC1
and LRADC2 inputs are limited to 3.3V.
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
INPUT_OFFSET
IRQ_EN_EVENT0
1
7
INPUT_DIV2
IRQ_EN_EVENT1
1
8
HALF_CMP_PWR
POLARITY_EVENT0
1
9
CLEAR
2
0
PWD
2
1
CLK_DIV
2
2
REF_VAL
2
3
POLARITY_EVENT1
HW_LRADC1_CTRL X:$FA23
Table 448. HW_LRADC1_CTRL
BITS
LABEL
23
POLARITY_EVENT1
RW RESET
RW 0
22
POLARITY_EVENT0
RW 0
21
IRQ_EN_EVENT1
RW 0
20
IRQ_EN_EVENT0
RW 0
21:18 RSRVD
17:16 REF_VAL
R
00
RW 00$
15:14 RSRVD
13:12 CLK_DIV
R
00
RW 00
11
PWD
RW 1
10
CLEAR
RW 1
DEFINITION
Set to one to trigger event 1 interrupts when filtered A/D
value is greater than the event 1 threshold. Set to zero for
interrupt when the filtered A/D value is less than the event 1
threshold.
Set to one to trigger event 0 interrupts when filtered A/D
value is greater than the event 0 threshold. Set to zero for
interrupt when the filtered A/D value is less than the event 0
threshold.
Set to one to enable an interrupt from the event 1
comparator.
Set to one to enable an interrupt from the event 0
comparator.
Reserved – Must be written with 0.
Reference voltage selection, see Table 441. “LRADC Input
Ranges and Resolutions” on page 317. These are the
primary mode control bits. They affect the range and
resolution of the converter.
Reserved – Must be written with 0.
Clock divider selection.
00 - divide by 4
01 - divide by 8
10 - divide by 16
11 - divide by 32
Set this bit to one to power down the comparator, the RLadder and stop the internal clocks to minimize power
consumption. Set to zero for normal operation.
Set this bit to one to force the up/down counter to all zeroes.
Set to zero for normal operation.
Table 449. LRADC 1 Control Register Description
5-35xx-D1-1.08-061705
323
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
9
INPUT_DIV2
RW RESET
RW 0
8
HALF_CMP_PWR
RW 0
7
6:0
RSRVD
INPUT_OFFSET
R
0
RW $00
DEFINITION
Set to one to divide the input voltage by two before applying
it to the comparator. The division by two changes the input
resistance from essentially infinite to 100KΩ.
Set to one to reduce the analog current for low power
operation. This is accomplished at the expense of
conversion performance. This bit should only be set to one
if static measurements are needed.
Reserved – Must be written with 0.
These bits are added to the final converter output to add a
digital offset to the converted value. This may be useful for
LRADC calibration and to compensate for additional bias
resistor when it is switched into the R-Ladder.
Table 449. LRADC 1 Control Register Description (Continued)
28.1.5.
LRADC1 Threshold Register
This register sets the thresholds used by the digital comparators in LRADC 1.
HW_LRADC1_THRSHX:$FA24
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
THRESHOLD_EVENT0
2
2
THRESHOLD_EVENT1
2
3
Table 450. HW_LRADC1_THRSH
BITS
LABEL
23:21 RSRVD
20:12 THRESHOLD_EVENT1
RW RESET
R
000
RW $000
11:9
8:0
R
000
RW $000
RSRVD
THRESHOLD_EVENT0
DEFINITION
Reserved – Must be written with 0.
This nine bit field defines the set point for event detector
one’s threshold comparator. The nine bit post offset
conversion value is compared to this threshold.
Reserved – Must be written with 0.
This nine bit field defines the set point for event detector
zero’s threshold comparator. The nine bit post offset
conversion value is compared to this threshold.
Table 451. LRADC 1 Threshold Register Description
324
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
28.1.6.
LRADC1 Result Register
This register controls the interrupts reported by LRADC 1 and provides read only
views of the LRADC data value and the results of the event detector threshold comparators.
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
EQ_EVENT1
1
6
EQ_EVENT0
1
7
GT_EVENT1
1
8
LT_EVENT1
1
9
GT_ENVET0
2
0
LT_EVENT0
2
1
DATA_OUT
2
2
IRQ_EVENT0
2
3
IRQ_EVENT1
HW_LRADC1_RESULTX:$FA25
Table 452. HW_LRADC1_RESULT
BITS
LABEL
23:22 RSRVD
21
IRQ_EVENT1
RW RESET
R
00
RW 0
20
RW 0
IRQ_EVENT0
19:17 RSRVD
16:8 DATA_OUT
R
R
00
$000
7:6
5
RSRVD
LT_EVENT0
R
R
00
0
4
LT_EVENT1
R
0
3
GT_EVENT0
R
0
2
GT_EVENT1
R
0
1
0
EQ_EVENT0
EQ_EVENT1
R
R
0
0
DEFINITION
Reserved – Must be written with 0.
Interrupt request status for event detector one. This sticky
bit is set whenever event detector one’s comparator trips
the set point.
See HW_LRADC1_CTRL_POLARITY_EVENT1 and
HW_LRADC1_THRSH_THRESHOLD1. Software clears
this bit by writing a one to it.
Interrupt request status for event detector zero. This sticky
bit is set whenever event detector zero’s comparator trips
the set point.
See HW_LRADC1_CTRL_POLARITY_EVENT0 and
HW_LRADC1_THRSH_THRESHOLD0. Software clears
this bit by writing a one to it.
Reserved – Must be written with 0.
The current nine bit value of the LRADC is read here.
WARNING: due to clock domain crossings, software must
repeatedly read this value until two successive values are
read which match bit for bit.
Reserved – Must be written with 0.
This read only bit indicates the current state of the event
detector 0 Less Than result.
This read only bit indicates the current state of the event
detector 1Less Than result.
This read only bit indicates the current state of the event
detector 0 Greater Than result.
This read only bit indicates the current state of the event
detector 1 Greater Than result.
The “excluded muddle”.
The “excluded muddle”.
Table 453. LRADC 1 Result Register Description
5-35xx-D1-1.08-061705
325
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
28.1.7.
LRADC2 Control Register
This register controls the overall operation of LRADC 2.
WARNING: The pad ESD protection limits maximum voltage on all LRADC inputs.
The BATT LRADC is specifically designed to handle higher voltages, but LRADC1
and LRADC2 inputs are limited to 3.3V.
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
INPUT_OFFSET
IRQ_EN_EVENT0
1
7
INPUT_DIV2
IRQ_EN_EVENT1
1
8
HALF_CMP_PWR
POLARITY_EVENT0
1
9
CLEAR
2
0
PWD
2
1
CLK_DIV
2
2
REF_VAL
2
3
POLARITY_EVENT1
HW_LRADC2_CTRL X:$FA26
Table 454. HW_LRADC2_CTRL
BITS
LABEL
23
POLARITY_EVENT1
RW RESET
RW 0
22
POLARITY_EVENT0
RW 0
21
IRQ_EN_EVENT1
RW 0
20
IRQ_EN_EVENT0
RW 0
21:18 RSRVD
17:16 REF_VAL
R
00
RW 00$
15:14 RSRVD
13:12 CLK_DIV
R
00
RW 00
11
PWD
RW 1
10
CLEAR
RW 1
DEFINITION
Set to one to trigger event 1 interrupts when filtered A/D
value is greater than the event 1 threshold. Set to zero for
interrupt when the filtered A/D value is less than the event 1
threshold.
Set to one to trigger event 0 interrupts when filtered A/D
value is greater than the event 0 threshold. Set to zero for
interrupt when the filtered A/D value is less than the event 0
threshold.
Set to one to enable an interrupt from the event 1
comparator.
Set to one to enable an interrupt from the event 0
comparator.
Reserved – Must be written with 0.
Reference voltage selection, see Table 441. “LRADC Input
Ranges and Resolutions” on page 317. These are the
primary mode control bits. They affect the range and
resolution of the converter.
Reserved – Must be written with 0.
Clock divider selection.
00 - divide by 4
01 - divide by 8
10 - divide by 16
11 - divide by 32
Set this bit to one to power down the comparator, the RLadder and stop the internal clocks to minimize power
consumption. Set to zero for normal operation.
Set this bit to one to force the up/down counter to all zeroes.
Set to zero for normal operation.
Table 455. LRADC 2 Control Register Description
326
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
9
INPUT_DIV2
RW RESET
RW 0
8
HALF_CMP_PWR
RW 0
7
6:0
RSRVD
INPUT_OFFSET
R
0
RW $00
DEFINITION
Set to one to divide the input voltage by two before applying
it to the comparator. The division by two changes the input
resistance from essentially infinite to 100KΩ.
Set to one to reduce the analog current for low power
operation. This is accomplished at the expense of
conversion performance. This bit should only be set to one
if static measurements are needed.
Reserved – Must be written with 0.
These bits are added to the final converter output to add a
digital offset to the converted value. This may be useful for
LRADC calibration and to compensate for additional bias
resistor when it is switched into the R-Ladder.
Table 455. LRADC 2 Control Register Description (Continued)
28.1.8.
LRADC2 Threshold Register
This register sets the thresholds used by the digital comparators in LRADC 2.
HW_LRADC2_THRSHX:$FA27
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
THRESHOLD_EVENT0
2
2
THRESHOLD_EVENT1
2
3
Table 456. HW_LRADC2_THRSH
BITS
LABEL
23:21 RSRVD
20:12 THRESHOLD_EVENT1
RW RESET
R
000
RW $000
11:9
8:0
R
000
RW $000
RSRVD
THRESHOLD_EVENT0
DEFINITION
Reserved – Must be written with 0.
This nine bit field defines the set point for event detector
one’s threshold comparator. The nine bit post offset
conversion value is compared to this threshold.
Reserved – Must be written with 0.
This nine bit field defines the set point for event detector
zero’s threshold comparator. The nine bit post offset
conversion value is compared to this threshold.
Table 457. LRADC 2 Threshold Register Description
5-35xx-D1-1.08-061705
327
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
28.1.9.
LRADC2 Result Register
This register controls the interrupts reported by LRADC 2 and provides read only
views of the LRADC data value and the results of the event detector threshold comparators.
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
EQ_EVENT1
1
6
EQ_EVENT0
1
7
GT_EVENT1
1
8
LT_EVENT1
1
9
GT_ENVET0
2
0
LT_EVENT0
2
1
DATA_OUT
2
2
IRQ_EVENT0
2
3
IRQ_EVENT1
HW_LRADC2_RESULTX:$FA28
Table 458. HW_LRADC2_RESULT
BITS
LABEL
23:22 RSRVD
21
IRQ_EVENT1
RW RESET
R
00
RW 0
20
RW 0
IRQ_EVENT0
19:17 RSRVD
16:8 DATA_OUT
R
R
00
$000
7:6
5
RSRVD
LT_EVENT0
R
R
00
0
4
LT_EVENT1
R
0
3
GT_EVENT0
R
0
2
GT_EVENT1
R
0
1
0
EQ_EVENT0
EQ_EVENT1
R
R
0
0
DEFINITION
Reserved – Must be written with 0.
Interrupt request status for event detector one. This sticky
bit is set whenever event detector one’s comparator trips
the set point.
See HW_LRADC2_CTRL_POLARITY_EVENT1 and
HW_LRADC2_THRSH_THRESHOLD1. Software clears
this bit by writing a one to it.
Interrupt request status for event detector zero. This sticky
bit is set whenever event detector zero’s comparator trips
the set point.
See HW_LRADC2_CTRL_POLARITY_EVENT0 and
HW_LRADC2_THRSH_THRESHOLD0. Software clears
this bit by writing a one to it.
Reserved – Must be written with 0.
The current nine bit value of the LRADC is read here.
WARNING: due to clock domain crossings, software must
repeatedly read this value until two successive values are
read which match bit for bit.
Reserved – Must be written with 0.
This read only bit indicates the current state of the event
detector 0 Less Than result.
This read only bit indicates the current state of the event
detector 1 Less Than result.
This read only bit indicates the current state of the event
detector 0 Greater Than result.
This read only bit indicates the current state of the event
detector 1 Greater Than result.
The “excluded muddle”.
The “excluded muddle”.
Table 459. LRADC 2 Result Register Description
328
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
29. BOOT MODES
29.1. General Information on Boot Modes
The on-chip ROM contains program code that is responsible for loading code from
outside the chip into on-chip RAM, and transferring control to that code. The DSP
determines which mode to use for booting by examining the signal level on some
pins immediately after reset. Table 460 shows the pins used for determining the boot
mode
100-PIN TIFFS
PIN #
95
84
94
93
92
144-PIN FPBGA
PIN #
J1
D1
H2
H4
H3
PIN LABEL
BOOT FUNCTION
GP0
GP8
GP1
GP2
GP3
Set to one to enable POST operation
Specifies boot mode bit 0
Specifies boot mode bit 1
Specifies boot mode bit 2
Specifies boot mode bit 3
Table 460. Boot Control Pins
NOTE: The Power on self test (POST) must be run in all applications of the
STMP35xx. Therefore GP0 (pin 95/J1) must be tied high through a high impedance
resistance (47KΩ) on ALL customer boards.
These pins can and are used for other functions besides configuring the boot mode;
this is achieved by specifying the high or low state on these pins with a high value
resistor (typically 47kΩ) pulling the pin either up to VddIO or down to VssIO. During
boot mode, these pins are not driven by the chip, and the pull-up/down resistors
define the values. Table 461 shows the different boot modes supported.
GP0
POST
0
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
GP3
GP2
GP1
GP8
MODE 3 MODE 2 MODE 1 MODE 0
X
X
X
X
0
0
0
X
0
0
1
0
0
0
1
1
0
1
0
0
0
1
0
1
0
1
1
0
0
1
1
1
1
0
0
0
1
0
0
1
1
0
1
0
1
0
1
1
1
1
0
0
1
1
0
1
1
1
1
0
1
1
1
1
PORT
BOOT MODE
INVALID
RESERVED
NAND 1.8 Volt
NAND 1.8 Volt
RESERVED
SPI
RESERVED
UTMI
USB
I2C
NAND 3.3 Volt
NAND 3.3 Volt
I2C
SPI
Tester Loader
Burn In
Not Valid for customer use.
Reserved for future use
NAND with Play recovery
NAND with PSWITCH recovery
Reserved for future use
Master
Reserved for future use
UTMI Self Test Mode
STMP Boot Class - Bulk out Endpoint
Slave
NAND with Play recovery
NAND with PSWITCH recovery
Master
Slave
INTERNAL USE ONLY
INTERNAL USE ONLY
Table 461. Boot Modes
A Power On Self Test (POST) is available during the boot process. This function
runs a self-test and repair function on the on-chip RAMs and swaps in spare bits
5-35xx-D1-1.08-061705
329
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
where necessary/possible to repair any defects found. Note that chips shipped by
SigmaTel are guaranteed to have no un-repaired defects in the on-chip RAMs, however, they may contain repairable errors. This means that the POST MUST be run
as part of the boot process for correct operation of the chip. The POST process
takes about 95msec to complete, and is designed to consume as little power as
possible. The POST process does not stop on errors, and does not give any indication as to success or failure; the application loaded by the boot loader should
assume that the on-chip RAM has been repaired correctly. The POST operation will
overwrite any data stored in on-chip RAM, so any data in on-chip RAM will be lost
during the boot process.
The USB, I2C master & slave, NAND Flash and SPI slave boot modes are intended
to be used in applications of the chip; in other words, they are user boot modes. The
INTERNAL boot modes are for internal SigmaTel use only. Any changes to the
above boot mode table will preserve all of the entries for user boot modes to preserve compatibility for existing designs.
During reset, and while the POST process is running, all digital pins on the chip are
in tristate mode. Once an individual boot mode starts, the boot mode will enable and
configure the pins required for that boot mode. It is the responsibility of the application loaded by the boot ROM to enable & configure all of the pins it requires, the
application should do this without making any assumptions about how the pins were
configured by the boot ROM.
There is one additional implicit boot mode, over and above the boot modes listed in
the above table, that is OnCE (On Chip Emulation) boot mode. This boot mode
occurs when the external debug hardware pulls the ONCE_DRn pin low during the
boot process to trigger the on-chip debugger. When this happens the DSP will execute the first few instructions in the on-chip ROM before stopping and allowing the
external debugger to take control. In this circumstance, the actual boot mode
selected on the boot pins will have no effect. OnCE boot mode will only be used by
developers, however PCB boards should support the OnCE boot mode wherever
possible, to enable system debug.
29.2. Bootloader Code Format
For the user boot modes, the boot loader expects the code read from the external
storage to be organized in a specific format. This format is able to load blocks of
data into X, Y, P, or L memory. It is also capable of initializing a block of memory to a
specified value. Loading a block of code or data into memory is achieved by first
sending a two-word command header, followed by the data. The command header
is shown in Table 462.
WORD 0
WORD 1
WORD 2-N
BITS 23-20
Mc[3:0]
BITS 19-16
BITS [15:0]
0000
Address[15:0]
00000000
Length[15:0]
Data[23:0]
Table 462. Bootloader Command Header + Data
For Word 0, the Memory Control bits, Mc[3:0], are used to specify the target memory
space for the code that is being loaded, and whether this is a block memory load, or
an initialization of a range of addresses to one value. Table 463 specifies the meaning of these bits in more detail.
The Address[15:0] field specifies the base address in memory where data is to be
written. The Length[15:0] specifies the length of the block of data in memory to be
loaded or initialized. When loading into P, X or Y memory, Words 2 through N will
330
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
MC3
MC2 (P)
MC1 (X)
MC0 (Y)
0
0
0
0
0
0
0
0
1
1
1
1
1
1
1
1
0
0
0
0
1
1
1
1
0
0
0
0
1
1
1
1
0
0
1
1
0
0
1
1
0
0
1
1
0
0
1
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
0
1
DESCRIPTION
Unused
Load into Y Memory
Load into X Memory
Load into L memory (XY memory)
Load into P memory
Unused
Unused
Unused
Unused
Initialize Y memory with data value
Initialize X memory with data value
Initialize L (i.e. X & Y) memory with data value
Initialize P memory with data value
Unused
Unused
Boot load complete, exit and jump to P:$0.
Must be a complete Command Header
($F00000 $000000).
Table 463. Bootloader Memory Control Bits
contain one word of data for each P, X, or Y memory location to be loaded. When
loading into L memory, Words 2 through N will contain two words of data for each L
memory location to be loaded, words 2, 4, etc., will go into X memory, words 3, 5,
etc., will go into Y memory. When initializing P, X, or Y memory, Word 2 will specify
the value to be used in initializing memory. When initializing L memory, Words 2
through 3 will specify the value to be used in initializing memory.
This data structure allows the initialization of hardware control registers as part of
the boot process, this could be achieved by using the above data structure to write
directly to a memory mapped hardware register. Most frequently, this is used to write
to the HW_PXCFG & HW_PYCFG registers that control the configuration of on-chip
RAM.
5-35xx-D1-1.08-061705
331
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
29.3. Encryption
The chip expects the code to be loaded by the boot loader to be encrypted. This
encryption serves two purposes; it makes it harder for an end user to hack the
device, and it includes a check-sum to prevent corrupted code from being used to
boot the device. The boot loader will fall back to USB boot mode if the code being
loaded fails the check-sum test, this allows a player with corrupted code to be recovered by an end user with the use of a recovery program on their PC.
The details of the encryption algorithm are not covered in this document; please
contact SigmaTel for more information if needed. The SigmaTel SDK (Software
Development Kit) includes everything needed to encrypt code that is to be loaded
into the chip.
29.4. Bootloader Examples
This section provides a few examples of data delivered to the DSP and how it will be
copied into memory. Note that these examples show the data before it is encrypted
using the encryption mentioned above.
29.4.1.
Boot Example #1
The following example of data presented to the DSP would result in 8 words being
loaded into Y memory.
$100210
$000008
$000000
$111111
$222222
$333333
$444444
$555555
$666666
$777777
$F00000
$000000
29.4.2.
; load into Y memory starting at Y:$0210
; load 8 words of data into Y memory
; data word #0, written to Y:$0210
; data word #1, written to Y:$0211
; data word #2, written to Y:$0212
; data word #3, written to Y:$0213
; data word #4, written to Y:$0214
; data word #5, written to Y:$0215
; data word #6, written to Y:$0216
; data word #7, written to Y:$0217
; end of boot, execute program
; this word IS required
Boot Example #2
The following example of data presented to the DSP would result in 8 contiguous P
memory locations being initialized to a specified value.
$C00400
$000008
$CCCCCC
$F00000
$000000
332
; initialize P memory starting at P:$0400
; initialize 8 locations, i.e. P:$0400-$0407
; initialize with the value $CCCCCC
; end of boot, execute program
; this word IS required
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
29.4.3.
Boot Example #3
The following example of data presented to the DSP would result in 11 contiguous L
memory (XY memory) locations being initialized to specified values.
$300230
$00000B
$000001
$000002
$000004
$000008
$000010
$000020
$000040
$000080
$000100
$000200
$000400
$000800
$001000
$002000
$004000
$008000
$010000
$020000
$040000
$080000
$100000
$200000
$F00000
$000000
; load L memory at address L:$0230
; load 11 L words
; data word #0 upper, loaded to X:$0230
; data word #0 lower, loaded to Y:$0230
; data word #1 upper, loaded to X:$0231
; data word #1 lower, loaded to Y:$0231
; data word #2 upper, loaded to X:$0232
; data word #2 lower, loaded to Y:$0232
; data word #3 upper, loaded to X:$0233
; data word #3 lower, loaded to Y:$0233
; data word #4 upper, loaded to X:$0234
; data word #4 lower, loaded to Y:$0234
; data word #5 upper, loaded to X:$0235
; data word #5 lower, loaded to Y:$0235
; data word #6 upper, loaded to X:$0236
; data word #6 lower, loaded to Y:$0236
; data word #7 upper, loaded to X:$0237
; data word #7 lower, loaded to Y:$0237
; data word #8 upper, loaded to X:$0238
; data word #8 lower, loaded to Y:$0238
; data word #9 upper, loaded to X:$0239
; data word #9 lower, loaded to Y:$0239
; data word #10 upper, loaded to X:$023A
; data word #10 lower, loaded to Y:$023A
; end of boot, execute program
; this word IS required
29.5. Boot Procedure
The flow chart in Figure 113 shows the initial boot sequence and Figure 114 shows
the subsequent general boot procedure.
29.5.1.
USB boot mode
The USB boot mode enables the USB interface, attempts to enumerate on the USB
bus, and then expects USB Bulk-Out data from the host PC containing the data to
be use for booting. This boot mode is intended to be used at the end of the production line for EOL (end of line) tests, and for personalizing the players. Together with
a recovery application running on a PC, this mode may also be used by an end user
to recover if the code on a player somehow becomes corrupted.
One issue with using the USB Boot Class MODE is that the on-chip ROM does not
have a unique ID and therefore multiple devices built using the chip may not be connected simultaneously to a single computer. To get around this limitation all software
implementations provided by SigmaTel will populate the boot component with a Boot
Manager program. The Boot Manager is used to decide whether to load code for a
player or for USB. The Boot Manager will check to see if USB is connected and if so
5-35xx-D1-1.08-061705
333
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
Power
Up
p:$0 = jmp
p:BootLoader
BootLoader
Entry
Enable
OnCE
Init
Variables
Power
BootMode
up
GPIO
Pins
Enable, Sense, &
Disable
BootMode
GPIO pins
BURNIN
Mode?
Y
jmp BURNIN
Y
jmp
TESTERLOADE
R
N
Run
POST
Y
POST
Enabled?
N
TESTERLOADE
mode
R
?
N
Execute
BootMode
Figure 113. Initial Boot Sequence
it will load the Device Control Class (DCC) Image, and if not it will load the Player
Image. The DCC Image will be specific to a given product hence it may have unique
USB IDs and allow multiple devices to co-exist on a computer. In this manner the
generic USB Boot Class is completely by-passed. Please refer to the chip Boot
Manager document for more details. USB boot mode differs from all the other boot
modes in that it leaves the PLL enabled and the DSP running at 68 MHz when
done.
334
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
Continue Boot
Each BootMode is
responsible for powering
all required pads
BootMode ==
NAND Flash?
N
BootMode ==
USB?
The UsbInit Routine will wait
up to 5 minutes for USB to
be connected.
BootMode = USB
N
USB
Connected?
Y
N
N
Play/PSWITCH
Button
Pressed > 5s ?
Y
Y
BootMode = USB
Generic BootClass
N
Copy ROM to
RAM
Y
N
Set OMR[0] == 0
Run BootMode Init
Routine
jmp p:RunInit
Got
CipherKey?
Y
N
Got CheckSum
Target?
N
Y
Version Okay?
N
N
Get Command
All pads powered for a
given BootMode will
remain powered
Valid
Command?
N
Y
Run BootMode
Exit Routine
y
CheckSum
Okay?
Reboot?
Y
N
jmp p:$0
Execute
Command
Okay?
Figure 114. Boot Procedure
5-35xx-D1-1.08-061705
335
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
29.5.1.1.
USB Boot Mode Pin Power
USB Boot Mode pins are analog pins and are powered at system startup up. No
other digital pins are powered for this boot mode.
Note: See Table 506 for USB Interface pin placement.
29.5.2.
NAND Flash Boot Mode
The NAND Flash boot mode will read boot commands from NAND Flash or SmartMedia flash devices. The Boot Mode Init routine will search up to four NAND/SmartMedia devices and boot from the first device/card that is found with valid boot code.
The NAND Flash boot mode searches for and boots from the chip Boot component.
Please refer to the chip SmartMedia/NAND Blocks document for more details on
components. The search will start with chip select 0 and move on to 1, 2, and 3. If
an error is encountered then the device on the next chip select is searched. If
desired, the entire code set may reside in the chip Boot component and the Boot
Manager and other components may be left out of the system.
If all NAND/SmartMedia devices are searched and no valid boot code is found the
boot mode is changed to USB, and the part attempts to boot from the USB bus. This
situation will most commonly be encountered at the end of a production line where
the devices being manufactured will not contain any code. This situation will also be
encountered if the code on the NAND/SmartMedia device becomes corrupted for
some reason, the end user will be able to reload the code onto the device by using a
recovery program on a PC.
Finally, the end user can manually force a boot from USB by holding down the Play
button (NAND Flash boot mode with play recovery), or power button (NAND Flash
boot mode with PSWITCH recovery) for 5 seconds during the boot process. See
Figure 114, Boot Procedures, for details.
29.5.2.1.
29.5.3.
NAND Flash Boot Mode Pin Power
The NAND Flash boot mode will power the following pins:I2C Slave Boot Mode
SM_D0
SM_D5
SM_ALE
SM_D1
SM_D6
SM_CLE
SM_D2
SM_D7
SM_D3
SM_REn
SM_CE0n
SM_D4
SM_WEn
SM_CE1n
Note: See Table 496 for SmartMedia/NAND pin placement.
SM_CE2n
SM_CE3n
SM_WPn
SM_READY
The I2C slave boot mode enables the I2C port as a slave device, and waits for data
to be written to the I2C port using the bootloader code format documented above.
29.5.3.1.
I2C Slave Boot Mode Pin Power
The I2C Slave boot mode will power the following pins:
I2C_SCL
I2C_SDA
Note: See Table 498 for I2C Interface pin placement.
29.5.4.
336
I2C Master Boot Mode
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
The I2C master boot mode enables the I2C port as a master device, and then
attempts to read data formatted using the bootloader code format documented
above from an external I2C device at I2C address $A0. If a standard 24.576 MHz
crystal is used, this boot mode will use an I2C clock speed of ~180 kHz.
29.5.4.1.
I2C Master Boot Mode Pin Power
The I2C Master boot mode will power the following pins:
I2C_SCL
I2C_SDA
Note: See Table 498 for I2C Interface pin placement.
29.5.5.
SPI Slave Boot Mode
The SPI slave boot mode enables the SPI port as a slave device, and waits for data
to be written to the SPI port using the bootloader code format documented above.
29.5.5.1.
SPI Slave Boot Mode Pin Power
The SPI Slave boot mode will power the following pins:
SPI_MOSI
SPI_MISO
SPI_SCK
SPI_SSn
Note: See Table 498 for SPI Interface pin placement.
29.5.6.
TESTERLOADER Boot Mode
This boot mode provides a simple, fast interface to testers for loading test code.
This mode is intended for internal SigmaTel use only, and no further documentation
is provided here.
29.5.7.
BURNIN Boot Mode
This boot mode provides a simple way to exercise large portions of the on-chip
hardware. Note that the power consumed by the chip will be significantly higher than
normal when the part is operating in BURNIN mode. This boot mode is intended for
internal SigmaTel use, and no further documentation is provided here.
5-35xx-D1-1.08-061705
337
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
29.5.8.
System Recovery Mode
This is not an explicit boot mode but is entered in a special case of the two NAND
modes. If the chip is booted with MODE = NAND with play recovery and USB is connected, and the Play button is held for 5 seconds the player will automatically go to
the USB MODE. This will cause the device to show up as a USB Boot Class device.
If a host driver detects this device it may take action to reformat the SmartMedia/NAND. Please refer to Figure 114.
29.6. Memory Maps
Various memory maps are employed in the different boot modes. Figure 115 shows
the memory map for the USB boot modes and Figure 116 shows the map for all
other modes. Shaded areas are available for loading code.
$9FFF
$9FFF
$4FFF
$4FFF
$4FFF
$4FFF
$2000
$1FFF
USB
Boot
Code
$0000
$01FF
$0000
$0200
$1FFF
$0200
Boot
Params
$01FF
Boot
ROM
$0000
$0000
Boot
Params
$0000
$0000
PRAM
XRAM
YRAM
Figure 115. USB Boot Mode Memory Map
PRAM
XRAM
YRAM
Figure 116. All Other Boot Modes Memory Map
In USB Boot Class p:$0..1 are available for writing the Reboot vector. All other interrupt vectors must be created by code at run-time.
Although the interrupt vector table is available (P:$0000-P:$007F) it is required that
all vectors must be created by code at run-time.
338
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
30. DC-DC CONVERTER
The chip has 2 programmable integrated DC-DC converters that can be used to
provide power for the device as well as the entire application, including external
NAND flash memories, SDRAM, FM tuner etc. Please use the efficiency curves to
determine if the efficiency will be acceptable for high power applications. The converters can be configured to operate from standard battery chemistries in the range
of 0.9-3.6 volts including alkaline cells, NiMH, LiIon, etc. If multiple batteries are
used, a series configuration is strongly preferred to optimize efficiency and battery
life. Also, for applications that do not need integrated power management, the converters can be disabled and external supplies provided.
Board layout and external inductor/capacitor quality are critical to maximize analog performance and as well as DC-DC converter efficiency. Users should read and understand SigmaTel's printed circuit board layout application notes for guidance before
beginning PCB layout. Further, users should refer to SigmaTel's reference designs for
assistance in selecting the inductor and large external decoupling capacitors.
30.1. DC-DC Converter Theory of Operation
The DC-DC power conversion and charging circuitry must support a wide range of battery configurations, chemistries, and ultimately input voltages to the BATT pin, from a
weak single cell alkaline battery at 0.9 volts to a fully charged LiIon battery at 4.2 volts.
In addition, it must operate off of a 5 volt DC input supplied either from the USB’s
VBUS or from a DC power source derived from the AC mains. In all of these cases, it
must supply a steady reliable 3.3 volts to the I/O rail and 1.8 volts to the core digital
power rail and to the analog power rail.
In addition to supplying the core 1.8 volt analog and digital sources, the DC-DC converter must supply power to the other chips or devices in the application, e.g. NAND
Flash power, SDRAM power, IDE power, etc. The DC-DC converter must also support
operating the system at reduced rail voltages for power saving modes.
When generating 3.3 volts for the I/O rail from a single alkaline battery, the DC-DC converter must operate in a “boost” mode, as shown in Figure 117, below. In this case, the
0.9 volts applied to the battery pin must be boosted to 3.3V. For boost mode operation,
the battery is connected through an inductor to the DCDC_Batt pin. This pin has both a
low resistance N-FET and a low resistance P-FET. When the N-FET switches on, it
pulls the inductor to ground. When it switches off, the inductor voltage goes quite positive as the field collapses. During this high voltage phase, the P-FET device is
switched on to charge the capacitance on the VDD I/O rail. The proportion of the duty
cycle that the P-FET device is active determines how much charge is transferred to
VDD I/O.
The VDD I/O rail has a tapped voltage divider (R-ladder) which is fed to a comparator.
The comparator references the band gap voltage (VBG). The comparator effectively
indicates whether the VDD I/O rail is above or below the desired voltage. If it is below
the desired voltage then the P-FET should be switched on to charge the rail capacitance. If the rail droops to a lower voltage, then the N-FET on-time is increased to
increase the inductor field, making more back EMF available for the next P-FET transfer. The DC-DC converter operates with a feedback loop filter to determine the desired
duty cycle for the P-FET. The loop is constrained at both positive and negative limits,
see 30.5.1. “DCDC1 Control Register 0” on page 353.
The HW_DCDC_VDDIO_VOLTAGE_LEVEL bit field allows one of 24 taps to be
selected on the VDD I/O voltage divider, allowing a DC range from 2.05 V to 4.03 V.
5-35xx-D1-1.08-061705
339
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
Thus one can set a variety of voltage targets for the VDD I/O rail optimizing either
energy consumption or performance, as required. The loop filter output can be
checked in the HW_SPEED register for design validation purposes.
3.3 Volts
VDDIO
VDDIO RAIL
to rest of the chip
DCDC_VDDIO
decoder
5
DSP interrupt
DCDC_BATT
. .24. .
Duty Cycle Driver
5
HW_DCDC_VDDIO[12:8]
. .24. .
P-FET
decoder
HW_DCDC_VDDIO[4:0]
N-FET
DCDC_GND
LIMIT
- +
16
0.9 Volts
+
7
BATT
+
Digital
Filter
VDDIO
Brownout
BandGap
HW_DCDC_VDDIO[20]
to RTC
& Crystal
HW_SPEED[10:4]
HW_DCDC_VDDIO[16]
HW_DCDC1_CTRL0
P-FET charges Cap.
DCDC MODE = 101
N-FET cuts off current flow, inductor field collapses
N-FET "pumps" the inductor
Figure 117. DC-DC Converter Boost Theory of Operation
There is a second set of taps and a switch matrix on the VDD I/O voltage divider controlled by HW_DCDC_VDDIO_BROWNOUT_LEVEL. This tap is connected to a second VBG referenced comparator which detects under-voltage or brown-out conditions.
The comparator results can generate a brown-out alert so that software can take
appropriate action, i.e load shedding or graceful power-down as appropriate. The comparator result can be seen in HW_DCDC_VDDIO_BROWNOUT_STATUS. The interrupt is enabled by the HW_DCDC_VDDIO_BROWNOUT_ENABLE bit.
A buck mode is provided for dropping the appropriate voltage across the inductor to
regulate a higher voltage battery input down to the desired VDD I/O rail. For a fully
charged LiIon battery at 4.2 V, one must drop 0.9 volts across the inductor to keep a
steady 3.3 V on the I/O rail, see Figure 118. “DC-DC Converter Buck Theory of Operation” on page 341. In this mode, the P-FET switches on to add charge to the VDD I/O
340
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
rail capacitance and to store energy in the inductor field, while the N-FET switches on
to remove energy from the inductor field. Again, it is the duty cycle that controls the
target voltage. The duty cycle is controlled by a feedback loop that tracks the error
between the desired rail voltage and VBG, similar to what we saw in the boost mode
loop.
3.3 Volts
VDDIO
VDDIO RAIL
to the rest of the chip
DCDC_VDDIO
DCDC_BATT
decoder
. .24. .
Duty Cycle Driver
5
HW_DCDC_VDDIO[12:8]
. .24. .
P-FET
decoder
HW_DCDC_VDDIO[4:0]
5
N-FET
DCDC_GND
+
LIMIT
- +
16
+
7
4.2 Volts BATT
Digital
Filter
VDDIO
Brownout
IRQ
BandGap
HW_DCDC_VDDIO[20]
To RTC &
Crystal
HW_SPEED[10:4]
HW_DCDC_VDDIO[16]
HW_DCDC1_CTRL0
DCDC MODES = 000,001,011
Figure 118. DC-DC Converter Buck Theory of Operation
NOTE: the DCDC converter pin names can be misleading in buck mode. For example
the pin named “VddIO” is actually connected to the battery while the pin named
“DCDC_Batt” is actually connected to the VDD I/O rail.
The brownout detection circuitry works in exactly the same way as it does in boost
mode. Notice that board level wiring has to change between these two modes. One
must determine the desired mode before designing the application board schematic.
The above discussion focused on generating a single regulated voltage for a single
VDD rail. The STMP35xx has three distinct power planes or rails: 3.3 V VDD I/O rail,
1.8 V VDD rail for the digital core and a 1.8 V VDD rail for the analog circuitry, see
5-35xx-D1-1.08-061705
341
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
Table 464. “Power Pin to Power Plane Mapping” on page 342. This table shows which
power pins must be attached to which power planes on the application board. The discussion below shows the various ways these power planes can be supplied for the battery chemistry and configuration desired.
POWER PLANE
PIN NAME
100-PIN
144-PIN
DESIGNATION
TQFP
FPBGA
VddIO1
VddIO1
29
G8
VddIO2
VddIO2
96
G5
VddIO3
VddIO3
F7
VddIO4
VddIO4
E7
VddA1
VddA1
73
B2
VddA2
VddPLL
76
B1
VddA3
VddHP
64
D5
VddD1
VddD1
39
E8
VddD2
VddD2
11
H7
VddD3
VddD3
86
F6
Table 464. Power Pin to Power Plane Mapping
Figure 119, below, shows all three regulator channels of DC-DC converter #1 running
in boost mode. As shown in the figure, there is one N-FET that generates the bulk high
voltage, boosted across the inductor. There are three P-FET devices that can be
switched on to charge the three separate voltage planes attached to the converter.
The DC-DC converter selects one P-FET per cycle to charge one of the VDD rail channels. Each rail has its own voltage divider, comparator, converter feedback loop and
duty cycle control. Thus the N-FET will turn on once per cycle and exactly one of the PFETs will be turned on. There is a PFM mode that only enables FET switching to keep
the load voltage within a specified tolerance for light load applications, see
$HW_DCDCTBR.
Figure 119 also shows some additional circuitry related to the DC-DC converter. Notice
that the BATT pin is connected to a linear regulator which supplies the crystal oscillator,
the real time clock and the portion of the DC-DC converter that must remain on even
when the STMP35xx is in a “powered-down” state. Thus these circuits are powered up
as soon as the battery is installed and remain powered until the battery is removed or
discharged. This power domain is referred to elsewhere in the data sheet as the real
time clock or crystal power domain. The power switch (PSWITCH) monitoring circuitry
is also powered from the crystal power domain so that it can awaken the DCDC controller. NOTE: it is not necessary for the crystal oscillator and RTC to be enabled and
operational before the power switch triggers a power up event. The power switch
causes the oscillator to awaken which counts out a delay and awakens the DCDC converter controller to begin to ramp the rail voltages. Thus in the lowest energy consuming powered off state, all clocks are stopped and a DC circuit is waiting for the
PSWITCH transition.
342
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
VddD
1/2/3
VddA
1/2/3
DCDC_VddA*
1
2
3*
4*
DCDC_VddD
VddIO
VddIO
VddIO
VddIO
DCDC_VddIO
D-Major™Audio System on Chip
battery
BATT
Low
Resolution
ADC
battery
DC-DC
#1 Control System
DCDC_Batt
DCDC_Gnd
Regulator
VddXTAL
DSP_RESET
RTC
* these pins are not available in the 100 pin package.
Vbg
DCDC_mod0*
DCDC_mod2
DCDC_mod1*
PSWITCH
XTAL
OSC
Figure 119. DC-DC #1 Converter, 3 output Channels
30.1.1.
Defining Battery Configuration
The outputs of both DC-DC converters are isolated from the power rails of the chip.
Therefore, the connection between the DC-DC converter outputs and the voltage rails
of the chip must be done on the PCB. However, the specific PCB connections and
battery connections are unique to each battery configuration as described below.
The battery configuration of the player is determined by the three mode pins
DCDC_mod2, DCDC_mod1, DCDC_mod0. These pins each have an internal
100 kΩ pull-up resistor to the BATT pin (either the battery must always be connected to the BATT pin for the device to power up or the device must have power
supplied to its 5 V input pin, VDD5V), so each mode pin is default high. Low value
resistors should be added on the board to pull the desired mode pin low. Note that
the configuration of these pins cannot be changed dynamically. In 100-pin configurations, only DCDC_mod2 is available on a pin, therefore only modes 111 and 011
can be used in this configuration. Also, the pins of DCDC Converter #2 are not available in a 100-pin configuration.
The selection of the three DC-DC mode pins determines whether each of the two
DC-DC converters is operating in Buck Mode, Boost Mode, or is powered off. As the
5-35xx-D1-1.08-061705
343
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
following table shows, each converter operates in Boost Mode when its output voltages > battery voltage, and in Buck Mode when output voltage < battery voltage.
Table 465 details the decode of the three DC-DC mode select pins:
DCDCMODE2:0 POWER SOURCE DCDC CONVERTER #1 DCDC CONVERTER #2
111
1 Alkaline or 1 NiMH
2-channel boost
off
(0.9V-1.5V)
(1.8/3.3 V)
110*
reserved
reserved
reserved
101*
1 Alkaline or 1 NiMH
3-channel boost
off
(0.9V-1.5V)
(1.8/1.8/3.3 V)
100*
reserved
reserved
reserved
011
LiIon, (3.3-4.2V)
1-channel buck (1.8 V)
off
010*
External supplies
off
off
001*
2 Alkaline or 2 NiMH 1-channel buck (1.8 V) 1-channel boost (3.3 V)
(1.8V-3.0V)
000*
LiIon (3.3V-4.2V)
1-channel buck (1.8 V) 1-channel buck (3.3 V)
* Only available in 144-pin package
Table 465. Decode of the DC-DC Mode Select Pins
As Table 465 shows, DC-DC converter #2 is only used to generate the 3.3 V rail
when higher voltage battery configurations are selected. Additionally, DC-DC converter #2 has a lower resistance PMOS FET that can perform well in high current
applications such as rotating media. In DCDC mode 011, the DC-DC converter only
generates the 1.8 V rail, so an external regulator must be used to generate the 3.3 V
I/O rail voltage.
30.1.1.1.
DC-DC Converter Configuration
The two integrated DC-DC converters provide several low resistance FETs for use
in power conversion as shown in the following Figure 120:
VddIO
1/2/3*/4*
VddD
1/2/3
VddA
1/2/3
DCDC_VddD
DCDC_VddA*
DCDC_VddIO
battery
BATT
Low
Resolution
ADC
DC-DC
#1 Control System
DCDC_Batt
DCDC_Gnd
Regulator
DSP_RESET
VddXTAL
DCDC2_Vout*
RTC
DC-DC
#2 Control System
XTAL
OSC
DCDC2_Batt*
Vbg
DCDC_mod0*
DCDC_mod2
DCDC_mod1*
PSWITCH
DCDC2_Gnd*
* only available on 144-pin packages
Figure 120. DC-DC Converter Control System
344
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
The configuration of the three DC-DC mode pins described in the previous section
determines how these two groups of FETs are switched to provide the desired output voltage values. The FETs can be switched to either create a Buck Mode converter (battery >= output voltage) or Boost Mode converter (battery <= output
voltage). These two different operating modes require different connectivity between
the battery and inductor on the PCB. The various arrangements are shown in Figures 121 to 126.
VddIO
1/2/3*/4*
VddD
1/2/3
NC
battery
VddA
NC
DCDC_VddD
1/2/3
DCDC_VddA*
DCDC_VddIO
battery
BATT
Low
Resolution
ADC
DC-DC
#1 Control System
VddD
DCDC_Batt
DCDC_Gnd
Regulator
VddXTAL =
BATT/2
battery
VddXTAL
RTC
DC-DC
#2 Control System
DCDC2_Vout*
VddIO
DCDC2_Batt*
DCDC2_Gnd*
DCDC_mod0*
DCDC_mod2
DCDC_mod1*
XTAL
OSC
30.1.1.2.
* only available on 144-pin packages
Figure 121. DC-DC Converter Control System (Mode 000)
144-Pin only DCDC1
1 channel buck
DCDC2
1 channel buck
Power Up Sequence
The DC-DC converters control the power up sequence of the device and holds the
rest of the chip in reset until after the power up sequence is complete. This is necessary to prevent operation of the DSP and system before stable power supply voltages are present. It is also important the entire application not draw static current
during the powerup sequence to guarantee correct startup behavior. High value
pull-up and pull-down resistors are acceptable as long as the total current << 1mA.
The Power Up sequence begins when the battery is connected to the BATT pin of
the device (or a 5 V source is connected to the VDD5V pin). As shown in Figure
120, the BATT pin provides the pull-up on the three DC-DC mode select pins, as
well as supplying power to the crystal oscillator and the real-time clock. This means
that the crystal oscillator can be running, if desired, whenever a battery is connected
to BATT pin. This feature allows the real time clock to operate when the chip is in the
off state. The crystal oscillator/RTC is the only power drain on the battery in this
state and consumes only a very small amount of power. During this time, the digital
(VddD) and analog (VddA) supplies are held at ground, while the VddIO rail is
shorted to either the battery (mode111, 101,001) or ground (mode000). This is the
off state that continues until the PSWITCH pin is asserted high.
5-35xx-D1-1.08-061705
345
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
VddD
1/2/3
VddIO
1/2/3*/4*
NC
battery
VddA
NC
DCDC_VddD
1/2/3
DCDC_VddA*
DCDC_VddIO
battery
BATT
Low
Resolution
ADC
VddD
DC-DC
#1 Control System
DCDC_Batt
DCDC_Gnd
Regulator
VddXTAL =
BATT/2
VddXTAL
DCDC2_Vout*
RTC
DC-DC
#2 Control System
battery
DCDC2_Batt*
XTAL
OSC
DCDC_mod0*
DCDC_mod2
DCDC_mod1*
DCDC2_Gnd*
* only available on 144-pin packages
NC
Figure 122. DC-DC Converter Control System (Mode 001)
144-Pin only DCDC1
1 channel buck
DCDC2
1 channel boost
driven externally
VddD
VddA
1/2/3
1/2/3
VddIO
1/2/3*/4*
BATT
Low
Resolution
ADC
NC
NC
NC
DCDC_VddD
DCDC_VddA*
DCDC_VddIO
DC-DC
#1 Control System
NC
DCDC_Batt
DCDC_Gnd
Regulator
VddXTAL =
BATT/2
VddXTAL
NC
DCDC2_Vout*
RTC
DC-DC
#2 Control System
XTAL
OSC
DCDC_mod0*
DCDC_mod1*
DCDC_mod2
DCDC2_Gnd*
* only available on 144-pin packages
NC
346
NC
DCDC2_Batt*
Figure 123. DC-DC Converter Control System (Mode 010)
144-Pin only DCDC1
Off
DCDC2
Off
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
linear regulator or
converter**
Batt
VddD
1/2/3
VddIO
1/2/3*/4*
BATT
Low
Resolution
ADC
Batt
VddA
1/2/3
DCDC_VddIO
DC-DC
#1 Control System
NC
NC
DCDC_VddD
DCDC_VddA*
VddD
DCDC_Batt
DCDC_Gnd
Regulator
VddXTAL =
BATT/2
VddXTAL
NC
DCDC2_Vout*
RTC
DC-DC
#2 Control System
NC
DCDC2_Batt*
XTAL
OSC
DCDC_mod0*
DCDC_mod2
DCDC_mod1*
DCDC2_Gnd*
* only available on 144-pin packages
NC
NC
** VddIO is powered using a linear regulator or converter. This
same power source is connected to the BATT and DCDC_Batt pins.
The primary application for this mode is 100 pin devices that use
series batteries or high voltage batteries such as LiIon.
Figure 124. DC-DC Converter Control System (Mode 011)
100-Pin
DCDC1
1 channel buck
or 144-Pin
DCDC2
Off
When the PSWITCH pin is asserted, for at least 100ms, the DC-DC converters are
enabled and the device attempts to power up. Depending on the DC-DC mode
select pins, one or both of the DC-DC converters begin switching and drive the voltage outputs toward the target values. When both of the DC-DC converters’ control
systems sense that the voltage rails have reached their default target values, the
DSP reset is de-asserted and the DSP begins executing code. If the power supplies
do not reach the target values by the time PSWITCH is de-asserted, then the player
returns to the off state.
The startup time for the DC-DC converters are dependent on battery voltage, but
should be less than 100 milliseconds.
The crystal oscillator and RTC/ALARM can optionally remain off to minimize the
standby power drain, see HW_DCDC_PERSIST_SLEEP_XTAL_ENABLE. Note
that the default state of the crystal oscillator is powered down.
There is an integrated 5KΩ resistor that can be switched in between the VDDXTAL
pin and the PSWITCH pin. If enabled, see HW_DCDC_PERSIST_AUTORESTART,
then the device will immediately go through a power up sequence after power down,
as if the PSWITCH had been pressed.
30.1.1.3.
Power Down Sequence
Power Down is also controlled by the DC-DC converters. When the DC-DC converters detect a power down event, they return the player to the off state described
above that holds the internal VddD and VddA supplies at ground, and holds the
VddIO rail at ground in DCDC MODE 000; otherwise VddIO is connected to the bat-
5-35xx-D1-1.08-061705
347
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
VddD
1/2/3
VddIO
1/2/3*/4*
VddA
1/2/3
DCDC_VddIO DCDC_VddD DCDC_VddA*
battery
BATT
battery
Low
Resolution
ADC
DC-DC
#1 Control System
DCDC_Batt
DCDC_Gnd
Regulator
VddXTAL=BATT
VddXTAL
NC
DCDC2_Vout*
RTC
NC
DCDC2_Batt*
DC-DC
#2 Control System
XTAL
OSC
DCDC_mod0*
DCDC_mod2
DCDC_mod1*
DCDC2_Gnd*
* only available on 144-pin packages
NC
NC
Figure 125. DC-DC Converter Control System (Mode 101)
144-Pin only DCDC1
3 channel boost
DCDC2
Off
VddD
1/2/3
VddIO
1/2/3*/4*
VddA
1/2/3
DCDC_VddIO DCDC_VddD
battery
BATT
Low
Resolution
ADC
NC
DCDC_VddA*
battery
DC-DC
#1 Control System
DCDC_Batt
DCDC_Gnd
Regulator
VddXTAL=BATT
VddXTAL
NC
DCDC2_Vout*
RTC
DC-DC
#2 Control System
XTAL
OSC
DCDC_mod0*
DCDC_mod1*
DCDC_mod2
DCDC2_Gnd*
* only available on 144-pin packages
NC
NC
NC
348
NC
DCDC2_Batt*
Figure 126. DC-DC Converter Control System (Mode 111)
100-Pin
DCDC1
2 channel boost
or 144-pin
DCDC2
Off
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
tery. A Power Down event is triggered by writing a 1 to the Clock Control Register
PWDN bit (HW_CCR, bit [17]). A Power Down event can also be triggered by a fast
falling edge on the PSWITCH pin. The Power Down via PSWITCH is an edge sensitive event and requires < 15ns falling edge to cause a Power Down to be detected.
An external capacitor on PSWITCH can be used to prevent unwanted Power Down
events in the final application, but this fast falling edge Power Down mechanism can
also be disabled by writing a 1 to the Mixer Test Register PSRN field
(HW_MIXTBR_PSRN).
30.1.1.4.
Powered Down State
While the chip is powered down, the VddA & VddD rails are shorted to ground. The
VddIO rail is either shorted to ground (DCDC MODE 000) or to the battery (DCDC
MODE 111,101, & 001), or controlled by an external linear regulator (DCDC MODE
011). The following table shows the power down state in each of the DCDC configurations.:
DCDC
MODE2:0
111
101
011
001
000
DESCRIPTION
VDDD & VDDA
VDDIO
POWER DOWN STATE
POWER DOWN STATE
2-channel Boost
Shorted to GND
Shorted to Battery
3-channel Boost
Shorted to GND
Shorted to Battery
1-channel Buck
Shorted to GND
Controlled by an ext. regulator
Buck/Boost
Shorted to GND
Shorted to Battery
2-channel Buck
Shorted to GND
Shorted to GND
Table 466. DCDC Power Supply Powerdown States
For the DCDC Boost Modes (modes 101 & 111), the DCDC_VddD & DCDC_VddA
pins are shorted to digital ground, which will short the VddA & VddD rails to ground.
In the DCDC Buck Modes (modes 000, 001 & 011), the DCDC_Batt pin is shorted to
digital ground, which will also short the VddA & VddD rails to ground. In mode 000,
the DCDC2_BATT pin is pulled to ground which will short VddIO to ground as well.
For the DCDC external supply mode (mode 010), the DCDC_Batt pin is also
shorted to digital ground, but since this pin is not connected, it will have no effect on
the VddA & VddD rails.
For the DCDC Boost Modes (modes 101 & 111), the DCDC_VddIO & DCDC_Batt
pins are shorted to each other, which will short the VddIO rail to the battery voltage.
For the DCDC Buck Modes, (mode001) the DCDC2_Batt and DCDC2_Vout pins
are shorted to each other, which will also short the VddIO rail to the battery. In
Mode000 DCDC2_Batt is held at ground, which holds VddIO at ground. In Mode
011 it is assumed that VddIO is supplied by an external linear regulator as long as a
battery is present. For the DCDC External Supply Mode (mode 010), the
DCDC2_Batt and DCDC2_Vout pins are also shorted to each other, but since these
pins are not connected, it will have no effect on the VddIO rail.
In all modes, the crystal oscillator and RTC are powered through the BATT pin
(unless 5V is present) and are decoupled by a capacitor on the VddXTAL pin. If
desired, the crystal oscillator and the RTC can continue to operate when the device
is powered down by drawing power from the BATT pin.
30.1.1.5.
Reset Sequence
A reset event can be triggered by writing the binary value 1101 to the Reset Control
Register SRST field (HW_RCR, bits [7:4]). This reset only affects the digital logic,
although the digital logic also includes all of the registers that control the analog portions of the chip. The DCDC converters continue to maintain the power supply rails
5-35xx-D1-1.08-061705
349
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
during the reset, although the target voltages for the DCDC converters may change
once the digital registers go to their reset values.
30.1.1.6.
PSWITCH Recovery Mode Support
The Boot loader in the on-chip ROM contains several recovery mode options to
allow new firmware to be downloaded to the device from the USB host. One method
of invoking the recovery mode is to hold the PSWITCH closed for an extended
period of time after power on. Currently, the ROM code looks for the PSWITCH to
be held for 5 seconds after power on to force it into recovery mode. When the user
first presses the PSWITCH, the rising edge seen on the PSWITCH pin triggers a
power up, as described above. Once the ROM code is executing it looks at the Boot
Mode pins to determine the type of boot to perform, see Section 29. “BOOT
MODES” on page 329. If one of the PSWITCH recovery boot modes is selected
then the state of the PSWITCH pin is read from HW_SPARER_PSWITH to determine whether to enter the recovery mode.
30.2. Linear Mode Battery Charger
Several new circuits have been added to the STMP35xx to integrate battery charging
for LiIon and NiMH batteries from a 5V source connected to the VDD5V pin. The
source can come either from the USB’s VBUS or from a transformer. The first new circuit of interest is a five volt presence detector which trips when a voltage on the VDD5V
pin exceeds the VDDIO rail by more than 0.6 V. Detection of this condition causes the
activation of a linear regulator which drops the supplied 5 V down to 3.3 V. This regulator is tied to the VDD IO rail so it immediately begins to supply power to the rail. A second linear regulator is activated which drops the 3.3 V I/O rail to 1.8 V for the digital
core VDD, finally a P-FET shorts the analog and digital core rails together so they both
receive current from the linear regulator In addition, the presence of a 5 V source automatically stops the DC-DC converter(s) so that all chip power is supplied from the
VDD5V pin.
If one sets HW_VDD5V_POWER_CHARGE_BATT_CURRENT to a non-zero value
then the variable current shown in Figure 127 provides charging current to a LiIon battery. There is a voltage sensor monitoring the battery voltage which will automatically
shut down the current source when the battery voltage reaches 4.1 V (4.2 V with
HW_VDD5V_POWER_CHARGE_LI_TYPE set to one). One can programatically
monitor the battery voltage using the Battery LRADC, see Section 28.1.1. “LRADC
Battery Control Register” on page 318. LRADC2 can be used to monitor a temperature
sensor for optimum battery charge profiles. There is an integrated current source for
the external temperature sensor which can be enabled by setting
HW_MIX_TEST_TMP_CFG to the desired bias current.
30.3. Silicon Speed Sensor
The STMP35xx contains two silicon speed sensors to measure the performance
characteristics of an individual die at its ambient temperature and process parametrics. Each sensor consists of a ring oscillator and a frequency counter. The ring
oscillator runs on the VddD power rail. Therefore its frequency tracks the silicon performance as it changes in response to changes in the DCDC converter target voltage. The crystal oscillator is directly used as the precision time base for measuring
the frequency of a ring oscillator. The ring oscillator is normally disabled. There is a
seven bit counter connected to the ring oscillator which performs the frequency
measurement. When a zero to one transition is detected on
350
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
HW_DCDCTBR_OSC1_START_COUNT, then the counter is reset. The ring oscillator is next enabled for exactly one crystal clock period then disabled.
Thus the counter holds the number of cycles the ring oscillator was able to generate
during one crystal clock period. The natural frequency of the ring oscillator strongly
tracks the silicon process parametrics, i.e. faster silicon processes yield ring oscillators that run faster and thereby yield larger count values. The natural frequency
tracks junction temperature effects on silicon speed as well. By taking a quick snapshot of the silicon sensor frequency one can lower the core digital rail voltages to a
point more closely approaching the actual performance limit.
USB Vbus 5v
Line 5V input
>4.35V
battery
VddIO
1/2/3*/4*
5V
detect
VDD5V
VddA
1/2/3
NC
NC
DCDC_VddIO DCDC_VddD DCDC_VddA*
Linear
linear charger
controller
V Sense
battery
Linear Reg.
VddD
1/2/3
Charge Current Path
BATT
3 Channel
Low Resolution ADC
LRADC1
DC-DC
#1 Control System
(boost mode)
DCDC_Batt
DCDC_Gnd
LRADC2
Regulator
Temp.
VddXTAL=BATT
DSP_RESET
VddXTAL
disabled during linear
battery charge
battery
DCDC2_Vout*
RTC
DC-DC
#2 Control System
XTAL
OSC
DCDC2_Batt*
DCDC_mod0*
DCDC_mod2
DCDC_mod1*
DCDC2_Gnd*
* only available on 144-pin packages
MODE = 000, for LiIon, both converters in buck mode
Figure 127. DC-DC Converter Linear Converter Theory of Operation
The application work load will define a minimum frequency of operation for the DSP
and other devices on the DCLK. After device characterization, a calibration curve
will be provided for the silicon speed sensor, see 30.6.7. “Clock Speed vs. Voltage”
on page 374. Software can then translate this curve into a linear approximation represented by the following table:
WARNING: the use of the speed sensor in this fashion is still under investigation by
SigmaTel. Such usage by customer is deprecated, i.e. customers should not
attempt to use this feature at this time.
5-35xx-D1-1.08-061705
351
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
DESIRED
FREQUENCY
STARTING
VDDD
12KHZ - 35MHZ
35MHZ - 50MHZ
50MHZ - 60MHZ
65MHZ - 70MHZ
75MHZ - 85MHZ
1.65 V (TBD)
1.75 V (TBD)
1.8 V (TBD)
1.8 V (TBD)
1.8 V (TBD)
MINIMUM
COUNTER
VALUE
TBD
TBD
TBD
TBD
TBD
Table 467. Silicon Speed Sensor
One heuristic for using the speed sensor is to select the desired new frequency of
operation and set the VddD to the appropriate entry in the table. This value is conservative in that after characterization, there will still be sufficient margin that all die
from across the process and temperature design will work at that voltage.
The software then goes into a slow adaptation loop to track the voltage to the lowest
safe point for the process/temperature of each die. At 50ms intervals, the software
will read the silicon speed sensor and compare it to the minimum safe value. If it is
still above the safe threshold, it will implement a feedback loop filter to gradually
lower the voltage to the threshold. As the temperature will change over time, the
loop filter will also raise the voltage as required to stay in the safe region for the
desired frequency.
30.4. Summary of Major DC-DC Features
The unique architecture of the integrated DC-DC converters on the device allows
access for many control features via registers. Some of the most commonly used
features are described below:
1. Digital adjustment of the voltage rails: The target values of the voltage rails can
be adjusted during operation to optimize power consumption and extend battery
life. A suggested clock speed vs. voltage curve to allow for the optimum settings
to be chosen is included in 30.6.7 “Clock Speed vs. Voltage” on page 374.
Note: In most DC-DC operating modes (except mode 101) the analog supply is
shorted externally to the digital supply. In these cases analog performance may
suffer if the supply voltage is reduced below 1.46 volts or so. If the supply voltage is affecting analog performance, alternate settings for the analog reference
voltage and bandgap voltage may provide relief. The control for the analog reference voltages resides in the Reference Control Register (HW_REF_CTRL).
2. Power supply brownout detection: Digital comparators in the DC-DC converters
provide immediate input to the DSP via an IRQB or NMI interrupt (controlled by
the HW_RCR Reset Control Register) if the supplies drop below the brownout
level specified in the HW_DCDC_VDDIO, HW_DCDC_VDDD,
HW_DCDC_VDDA Control Registers. This feature is generally used to alert the
DSP when the power supplies are reaching dangerously low levels and allow a
controlled system shutdown. These comparators may trip on brief transients
below the target brownout voltage, so enough margin between the output voltage target and the brownout voltage target must be provided to eliminate false
triggers. The magnitude of the difference between the output voltage target and
brownout voltage target will depend on system power requirements and board
layout, should typically be set to 100 mV.
352
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
3. Battery brownout detection. Another comparator is present outside of the DCDC converters to provide battery brownout information to the DSP via the Reset
Control Register (HW_RCR). This feature is intended to provide indication to
the DSP of a low battery voltage that requires a controlled system shutdown
due to an event such as the battery falling out during application operation. The
programming of the battery brownout level is via the Battery Low Resolution
ADC Control Register (HW_BATT_CTRL).
4. DC-DC clock rate. The DC-DC converter clock rate can be lowered by 2x, 4x, or
8x from the default value of crystal frequency divided by 16 (1.536 MHz for a
24.576 MHz crystal or 1.5 MHz for a 24.0 MHz crystal). The clock rate can be
lowered to reduce power consumption in the DC-DC converter if the corresponding increase in supply transients can be tolerated in the application. See
bit 4, 5 in DC-DC Test Bit Register (HW_DCDCTBR).
5. Controlled linear charge current source supplying LiIon or NiMH battery charge
current from five volt source.
6. Temperature sensor support for improved battery charge protocols.
7. Optional PFM mode for low load applications, see the DC-DC Test Bit Register
(HW_DCDCTBR).
8. Silicon speed sensor to allow the supply voltage to be dynamically adjust to take
advantage of silicon speed and temperature variations.
30.5. DC-DC Programmable Registers
The DC-DC control registers are used to control DC-DC low-level functionality. It is
ill-advised to change these registers from their defaults unless
SigmaTel specifically indicates otherwise.
30.5.1.
DCDC1 Control Register 0
The organization of the DCDC1 Control Register 0 is shown below.
HW_DCDC1_CTRL0 X:$FA0C
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
PLVBU
2
1
PLVBO
2
2
NLEV
2
3
Table 468. HW_DCDC1_CTRL0
5-35xx-D1-1.08-061705
353
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS LABEL RW RESET
23:21 RSRVD R
0
20:16 NLEV
RW 01001
15:13 RSRVD R
000
12:8 PLVBO RW 01000
7:5
4:0
RSRVD R
000
PLVBU RW 10111
DEFINITION
Reserved – Must be written with 0.
Negative Digital Loop Clip Level – This value represents the negative value at
which the digital loop filter clips to prevent illegal values. These five bits represent bits
18:14 of the 20-bit loop filter value. Bit 19 of the negative clip value is set to binary 1,
while bits 13:0 are high. Altering the clip value should be done with caution since it
may cause the sigma delta output to be non-monotonic. Default decodes to -376832.
It is not recommended to increase the NLVL value when the value on the loop filter is
equal to NLVL. This case occurs when the battery voltage is greater than the VddD
voltage in mode 5 or mode7.
Reserved – Must be written with 0.
Positive Digital Loop Clip Level in Boost Mode – This value represents the
positive value at which the digital loop filter clips to prevent illegal values while
operating in boost converter mode. These five bits represent bits 18:14 of the 20-bit
loop filter value. Bit 19 of the positive clip is set to binary 0, while bits 13:0 are low.
The optimum value of this register depends on the battery voltage and acts
effectively to limit the battery current when the load exceeds the capabilities of the
converter. Default decodes to 131072.
Reserved – Must be written with 0.
Positive Digital Loop Clip Level in Buck Mode – This value represents the
positive value at which the digital loop filter clips to prevent illegal values while
operating in buck converter mode. These five bits represent bits 18:14 of the 20-bit
loop filter value. Bit 19 of the positive clip is set to binary 0, while bits13:0 are low.
The optimum value of this register depends on the battery voltage and acts
effectively to limit the battery current when the load exceeds the capabilities of the
converter. Default decodes to 376832.
Table 469. DCDC1 Control Register 0 Description
354
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
30.5.2.
DCDC1 Control Register 1
The organization of the DCDC1 Control Register 1 is shown below.
HW_DCDC1_CTRL1 X:$FA0D
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
C
2
1
R
2
2
FFOR
2
3
Table 470. HW_DCDC1_CTRL1
BITS LABEL RW RESET
23:11 RSRVD R
0
10:8 FFOR
RW 000
7:4
R
RW 1000
DEFINITION
Reserved – Must be written with 0.
Feed Forward to loop filter value – This two's complement value steps the loop
filter value by the amount DDD once on a 000->DDD transition. Thus, this value
must be rewritten to 000 every time before use. The feed forward feature is used to
help the control loop react under heavy, but well-known, transient loads.
Resistor Value – This value represents the R value of the “resistor” in the digital
loop filter in the common mode control loop in the DC-DC converter. The reset value
of decimal 16384 is chosen in conjunction with the value for bits [3:0], the actual
decoupling capacitor, and the clock rate to provide a stable control system. The
value will rarely be altered, but conditions may arise in the actual application for
which a different value is more optimum. Values range 2048 to 30720 (Only bits
14:11 are shown here, others bits are always 0). Values for NLVL, R, PLVU, and
PLVD must satisfy the following restrictions:
In modes 111 or 101 only, setting bit 4 improves the ability of the dcdc converter to
output a VDD/VDDA voltage that is less than the battery voltage. In other modes, bit
4 has no effect.
NLVL – 4R ≥ – 475137
PLVU + 4R ≤ 475136
PLVD + 4R ≤ 475136
3:0
C
RW 1000
As of chip revision TA2, and beyond, the LSB of this bit field was commandeered for
an additional purpose in addition to its usage above. When the chip is configured for
mode 101 or 111, then setting this bit to one allows VDD to be less than the battery
voltage by using a PFM mode switching algorithm.
Capacitor 1/C Value – This value represents 1/C value of the “capacitor” in the
digital loop filter in the common mode control loop in the DC-DC converter. The reset
value of decimal 32 is chosen in conjunction with the value for the R field in the
DCDC1 Control Register 0 (HW_DCDC1_CTRL0_R), the actual external decoupling
capacitor, and the clock rate to provide a stable control system. This value will rarely
be altered, but conditions may arise in the actual application for which a different
value is more optimum. Values can range between 8 and 120.
Table 471. DCDC1 Control Register 1 Description
5-35xx-D1-1.08-061705
355
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
30.5.3.
DCDC2 Control Register 0
The organization of the DCDC2 Control Register 0 is shown below.
HW_DCDC2_CTRL0 X:$FA11
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
PLVBU
2
1
PLVBO
2
2
NLEV
2
3
Table 472. HW_DCDC2_CTRL0
BITS LABEL RW RESET
23:21 RSRVD R
000
20:16 NLEV
RW 01001
15:13 RSRVD R
000
12:8 PLVBO RW 01000
7:5
4:0
RSRVD R
000
PLVBU RW 10111
DEFINITION
Reserved – Must be written with 0.
Negative Digital Loop Clip Level – This value represents the negative value at
which the digital loop filter clips to prevent illegal values. These five bits represent bits
18:14 of the 20-bit loop filter value. Bit 19 of the negative clip value is set to binary 1,
while bits 13:0 are high. Altering the clip value should be done with caution since it
may cause the sigma delta output to be non-monotonic. Default decodes to -376832.
Reserved – Must be written with 0.
Positive Digital Loop Clip Level in Boost Mode – This value represents the
positive value at which the digital loop filter clips to prevent illegal values while
operating in boost converter mode. These five bits represent bits 18:14 of the 20-bit
loop filter value. Bit 19 of the positive clip is set to binary 0, while bits 13:0 are low.
The optimum value of this register depends on the battery voltage and acts
effectively to limit the battery current when the load exceeds the capabilities of the
converter. Default decodes to 131072.
Reserved – Must be written with 0.
Positive Digital Loop Clip Level in Buck Mode – This value represents the
positive value at which the digital loop filter clips to prevent illegal values while
operating in buck converter mode. These five bits represent bits 18:14 of the 20-bit
loop filter value. Bit 19 of the positive clip is set to binary 0, while bits 13:0 are low.
The optimum value of this register depends on the battery voltage and acts
effectively to limit the battery current when the load exceeds the capabilities of the
converter. Default decodes to 376832.
Table 473. DCDC2 Control Register 0 Description
356
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
30.5.4.
DCDC2 Control Register 1
The organization of the DCDC2 Control Register 1 is shown below.
HW_DCDC2_CTRL1 X:$FA12
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
C
2
1
R
2
2
FFOR
2
3
Table 474. HW_DCDC2_CTRL1
BITS LABEL RW RESET
23:11 RSRVD R
0
10:8 FFOR
RW 000
7:4
R
RW 1000
DEFINITION
Reserved – Must be written with 0.
Feed Forward to loop filter value – This two's complement value steps the loop
filter value by the amount DDD once on a 000->DDD transition. Thus, this value
must be rewritten to 000 every time before use. The feed forward feature is used to
help the control loop react under heavy, but well-known, transient loads.
Resistor Value – This value represents the R value of the “resistor” in the digital
loop filter in the common mode control loop in the DC-DC converter. The reset value
of decimal 4096 is chosen in conjunction with the value for bits [21:15], the actual
decoupling capacitor, and the clock rate to provide a stable control system. The
value will rarely be altered, but conditions may arise in the actual application for
which a different value is more optimum. Values range 256 to 32767 (Only bits 14:8
are shown here, others bits are always 0). Values for NLVL, R, PLVU, and PLVD
must satisfy the following restrictions:
NLVL – 4R ≥ – 475137
PLVU + 4R ≤ 475136
PLVD + 4R ≤ 475136
3:0
C
RW 1000
Capacitor 1/C Value – This value represents 1/C value of the “capacitor” in the
digital loop filter in the common mode control loop in the DC-DC converter. The reset
value of decimal 64 is chosen in conjunction with the value for
HW_DCDC2_CTRL1_R, the actual external decoupling capacitor, and the clock rate
to provide a stable control system. This value will rarely be altered, but conditions
may arise in the actual application for which a different value is more optimum.
Values can range between 8 and 120.
Table 475. DCDC2 Control Register 1 Description
5-35xx-D1-1.08-061705
357
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
30.5.5.
DC-DC VddIO Control Register
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
VOLTAGE_LEVEL
2
0
BROWNOUT_LEVEL
2
1
BROWNOUT_ENABLE
2
2
VDDIO_OK
2
3
BROWNOUT_STATUS
The organization of the DC-DC VddIO Control Register is shown below.
HW_DCDC_VDDIO X:$FA0E
Table 476. HW_DCDC_VDDIO
BITS
LABEL
23:22 RSRVD
21
VDDIO_OK
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
R
Convergence State
0
converter still converging to new setting
1
converged on new voltage setting
20
BROWNOUT_STATUS R
VddIO brownout detect status
0
No brownout detected on VddIO rail
1
Brown out detected on VddIO rail
19:17 RSRVD
R
00
Reserved – Must be written with 0.
16
BROWNOUT_ENABLE RW 0
VddIO brownout detect enable
0
VddIO brownout detect disabled
1
VddIO brownout detect enabled
15:13 RSRVD
R
00
Reserved – Must be written with 0.
12:8 BROWNOUT_LEVEL
RW 01100
On-chip VddIO brownout detect level
11111 4.03 V 10111 3.52 V 01111 3.01 V
00111 2.50 V
11110 3.96 V 10110 3.45 V 01110 2.94 V 00110 2.43 V
11101 3.90 V 10101 3.39 V 01101 2.88 V 00101 2.37 V
11100 3.84 V 10100 3.33 V 01100 2.82 V 00100 2.30 V
11011 3.78 V 10011 3.26 V 01011 2.75 V 00011 2.24 V
11010 3.71 V 10010 3.20 V 01010 2.69 V 00010 2.18 V
11001 3.65 V 10001 3.14 V 01001 2.62 V 00001 2.11 V
11000 3.58 V 10000 3.07 V 01000 2.56 V 00000 2.05 V
7:5
RSRVD
R
000
Reserved – Must be written with 0.
4:0
VOLTAGE_LEVEL
RW 10000
On-chip VddIO voltage level – The voltage level settings for this
field are the same as for the BROWNOUT_LEVEL field above.
Default Voltage is 3.07 Volts.
Note: BROWNOUT_LEVEL and VOLTAGE_LEVEL represent nominal on-chip voltages. Due to random offsets, board
layout, and ohmic drops, the measured voltages on the board will not always measure exactly as the above table
decodes.
Table 477. DC-DC VddIO Control Register Description
358
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
30.5.6.
DC-DC VddD Control Register
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
VOLTAGE_LEVEL
2
0
BROWNOUT_LEVEL
2
1
BROWNOUT_ENABLE
2
2
VDDD_OK
2
3
BROWNOUT_STATUS
The organization of the DC-DC VddD Control Register is shown below.
HW_DCDC_VDDD
X:$FA0F
Table 478. HW_DCDC_VDDD
BITS
LABEL
23:22 RSRVD
21
VDDD_OK
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
R
Convergence State
0
converter still converging to new setting
1
converged on new voltage setting
20
BROWNOUT_STATUS R
VddD brownout detect status
0
No brownout detected on VddD rail
1
Brown out detected on VddD rail
19:17 RSRVD
R
00
Reserved – Must be written with 0.
16
BROWNOUT_ENABLE RW 0
VddD brownout detect enable
0
VddD brownout detect disabled
1
VddD brownout detect enabled
15:13 RSRVD
R
00
Reserved – Must be written with 0.
12:8 BROWNOUT_LEVEL
RW 10000
On-chip VddD brownout detect level
11111 2.02 V
10111 1.76 V
01111 1.50 V
00XXX RSRVD
11110 1.98 V
10110 1.73 V
01110 1.47 V
11101 1.95 V
10101 1.70 V
01101 1.44 V
11100 1.92 V
10100 1.66 V
01100 1.40 V
11011 1.89 V
10011 1.63 V
01011 1.37 V
11010 1.85 V 10010 1.60 V
01010 1.34 V
11001 1.82 V
10001 1.57 V
01001 1.31 V
11000 1.79 V 10000 1.54 V
01000 1.28 V
7:5
RSRVD
R
000
Reserved – Must be written with 0.
4:0
VOLTAGE_LEVEL
RW 10110
On-chip VddD voltage level – The voltage level settings for this
field are the same as for the BROWNOUT_LEVEL field above.
Default Voltage is 1.73 Volts.
Note: BROWNOUT_LEVEL and VOLTAGE_LEVEL represent nominal on-chip voltages. Due to random offsets, board
layout, and ohmic drops, the measured voltages on the board will not always measure exactly as the above table
decodes.
Table 479. DC-DC VddD Control Register Description
5-35xx-D1-1.08-061705
359
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
30.5.7.
DC-DC VddA/Battery Brownout Enable Control Register
The organization of the DC-DC VddA Control Register is shown below.
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
VOLTAGE_LEVEL
2
0
BROWNOUT_LEVEL
2
1
X:$FA10
BROWNOUT_ENABLE
2
2
VDDA_OK
2
3
BROWNOUT_STATUS
HW_DCDC_VDDA
Table 480. HW_DCDC_VDDA
BITS
LABEL
23:22 RSRVD
21
VDDA_OK
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
R
Convergence State
0
converter still converging to new setting
1
converged on new voltage setting
20
BROWNOUT_STATUS R
VddA brownout detect status
0
No brownout detected on VddA rail
1
Brown out detected on VddA rail
19:17 RSRVD
R
00
Reserved – Must be written with 0.
16
BROWNOUT_ENABLE RW 0
VddA brownout detect enable
0
VddA brownout detect disabled
1
VddA brownout detect enabled
15:13 RSRVD
R
00
Reserved – Must be written with 0.
12:8 BROWNOUT_LEVEL
RW 10000
On-chip VddA brownout detect level
11111 2.02 V
10111 1.76 V
01111 1.50 V
00XXX RSRVD
11110 1.98 V
10110 1.73 V
01110 1.47 V
11101 1.95 V
10101 1.70 V
01101 1.44 V
11100 1.92 V
10100 1.66 V
01100 1.40 V
11011 1.89 V
10011 1.63 V
01011 1.37 V
11010 1.85 V 10010 1.60 V
01010 1.34 V
11001 1.82 V
10001 1.57 V
01001 1.31 V
11000 1.79 V 10000 1.54 V
01000 1.28 V
WARNING: The STMP3501/3502 limits the minimum brownout
level setting to 1.82 V.
7:5
RSRVD
R
000
Reserved – Must be written with 0.
4:0
VOLTAGE_LEVEL
RW 10110
On-chip VddA voltage level – The voltage level settings for this
field are the same as for the BROWNOUT_LEVEL field above.
Default Voltage is 1.73 Volts.
WARNING: The STMP3501/3502 should not have its
VOLTAGE_LEVEL set below 1.92V.
Note: BROWNOUT_LEVEL and VOLTAGE_LEVEL represent nominal on-chip voltages. Due to random offsets, board
layout, and ohmic drops, the measured voltages on the board will not always measure exactly as the above table
decodes.
Table 481. DC-DC VddA Control Register Description
360
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
30.5.8.
Silicon Speed Sensor Register
The organization of the Silicon Speed Sensor Register is shown below.
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
OSC1_PWRUP
1
9
OSC1_START_COUNT
2
0
RINGOSC1
2
1
OSC2_PWRUP
2
2
RINGOSC2
2
3
X:$FA13
OSC2_START_COUNT
HW_SISPEED
Table 482. HW_SISPEED
BITS
LABEL
RW RESET
23
22:16
RSRVD
RINGOSC2
R
R
0
15:14
13
RSRVD
OSC2_START_COUNT
R
00
RW 0
12
OSC2_PWRUP
RW 0
11
RSRVD
R
0000
DEFINITION
Reserved – Must be written with 0.
If OSC2_PWRUP is set to one then this read only bit field returns
the frequency count value from ring oscillator 2. If OSC2_PWRUP
is set to zero then it returns the upper bits of DCDC2 converter
loop filter Duty Cycle value.
Frequency Count – The crystal oscillator is used as a time base to
measure the frequency of ring oscillator 2. The measured
frequency is returned here.
Duty Cycle – Digital loop filter value representing the duty cycle of
charging /discharging the inductor.
100000
DC-DC duty cycle at minimum (small charge
time/large discharge time)
......
011111
DC-DC duty cycle at maximum (large charge time/
small discharge time)
This value will be limited according to the NLEV and PLBO/PLVBU
values set for this DC-DC Converter.
Reserved – Must be written with 0.
Execute ring oscillator counting on low to high transition of this bit
and change meaning of HW_SISPEED_RINGOSC2. Note that
OSC2_PWRUP must be set to one to power on the oscillator
before counting.
Set to one to power up ring oscillator 2 circuitry. Set to zero for low
power mode.
Reserved – Must be written with 0.
Table 483. Silicon Speed Sensor Register Description
5-35xx-D1-1.08-061705
361
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
RW RESET
10:4
RINGOSC1
R
3:2
1
RSRVD
OSC1_START_COUNT
R
00
RW 0
0
OSC1_PWRUP
RW 0
DEFINITION
If OSC1_PWRUP is set to one then this read only bit field returns
the frequency count value from ring oscillator 1. If OSC1_PWRUP
is set to zero then it returns the upper bits of DCDC1 converter
loop filter Duty Cycle value.
Frequency Count – The crystal oscillator is used as a time base to
measure the frequency of ring oscillator 1. The measured
frequency is returned here.
Duty Cycle – Digital loop filter value representing the duty cycle of
charging /discharging the inductor.
100000
DC-DC duty cycle at minimum (small charge
time/large discharge time)
......
011111
DC-DC duty cycle at maximum (large charge time/
small discharge time)
This value will be limited according to the NLEV and PLBO/PLVBU
values set for this DC-DC Converter.
Reserved – Must be written with 0.
Execute ring oscillator counting on low to high transition of this bit
and change meaning of HW_SISPEED_RINGOSC1. Note that
OSC1_PWRUP must be set to one to power on the oscillator
before counting.
Set to one to power up ring oscillator 1 circuitry. Set to zero for low
power mode.
Table 483. Silicon Speed Sensor Register Description (Continued)
362
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
30.5.9.
DC-DC Test Bit Register
The organization of the DC-DC Test Bit Register is shown below.
1
3
1
2
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
DCDC1_VDD5V_ACTIVE
DCDC1_NOZERO
DCDC1_CLK2X
DCDC1_CLK4X
DCDC1_BAT_ADJ
DCDC1_DIS_5BIT
0
3
0
2
0
1
0
0
DCDC1_ADJ_TN
1
4
DCDC1_PFM
DCDC2_CLK4X
1
5
DCDC1_STOPCLK
DCDC2_CLK2X
1
6
PWRUP_VDDIO_BRNOUT
DCDC2_VDD5V_ACTIVE
1
7
DCDC_ANA_LESSI
DCDC2_PFM
1
8
DCDC_MORE_CAP
1
9
DCDC_ANA_BGR_BIAS
2
0
DCDC1_NEW_SCHEME
2
1
DCDC2_DIS_5BIT
2
2
X:$FA14
DCDC1_HALF_FETS
2
3
DCDC2_STOPCLK
HW_DCDCTBR
Table 484. HW_DCDCTBR
BITS
LABEL
23
DCDC2_STOPCLK
22
DCDC2_PFM
21
DCDC2_VDD5V_ACTIVE
20
DCDC2_CLK2X
19
DCDC2_CLK4X
18
DCDC2_DIS_5BIT
17
DCDC1_HALF_FETS
16
DCDC1_NEW_SCHEME
15
DCDC_ANA_BGR_BIAS
14
DCDC_MORE_CAP
13
DCDC_ANA_LESSI
RW RESET
DEFINITION
RW 0
Set to one to disable the clock to the DCDC2 converter.
Only use this when power comes from the VDD5V pin to
reach minimum system power or DCDC2 is unused.
RW 0
Set to one to enable a pulse skip mode (pulse frequency
modulation in DCDC #2) that is useful in very lightly
loaded conditions to minimize system power.
RW 0
Set to one to force DCDC #2 to remain active, even with
5V attached to the device. This mode may help during the
5V unplug transition back to battery power supplied by
the DCDC converter. When 5V is not present and the
mode is NOT 101 or 11, this bit prevents pass through
behavior when the battery voltage is approximately equal
to the output voltage.
RW 0
Set to one to slow down the DCDC #2 clock frequency by
a factor of 1/2. Can be combined with DCDC2CLK4X.
RW 0
Set to one to slow down the DCDC #2 clock frequency by
a factor of 1/4. Can be combined with DCDC2CLK2X.
RW 0
Set to one to reduce the quantization of DC-DC switching
times from 5 bits to 3 bits for DCDC #2.
RW 0
Set to one to reduce the size of the FETs by half (disable
half of the FETS) for DCDC converter #1. This mode may
be useful to minimize total system power in lightly loaded
applications.
RW 0
Set to one to use the new switching scheme for switching
FETS in DCDC converter #1 buck mode.
RW 0
Set to one to switch the DCDC analog bias current from
the selfbias current to a bandgap derived current.
RW 0
Set to one to add additional capacitance to common
mode sense node.
RW 0
Set to one to reduce the bias current to the DCDC
analog.
Table 485. DC-DC Test Bit Register Description
5-35xx-D1-1.08-061705
363
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
RW RESET
DEFINITION
12
PWRUP_VDDIO_BRNOUT RW 0
Set to one to enable the VDDIO brownout comparator.
NOTE: the VDDD brownout comparator is on by default.
11
DCDC1_STOPCLK
RW 0
Set to one to disable the clock to the DCDC1 converter.
Only use this when power comes from the 5V pin to reach
minimum system power.
10
DCDC1_PFM
RW 0
Set to one to enable a pulse skip mode (pulse frequency
modulation for DCDC #1) that is useful in very lightly
loaded conditions to minimize system power.
9
DCDC1_VDD5V_ACTIVE RW 0
Set to one to force DCDC #1 to remain active, even with
5V attached to the device. This mode may help during the
5V unplug transition back to battery power supplied by
the DCDC converter. When 5V is not present and the
mode is NOT 101 or 111, this bit prevents pass through
behavior when the battery voltage is approximately equal
to the output voltage.
8
DCDC1_NOZERO
RW 0
Set to one to prevent DCDC converter #1 from a duty
cycle of zero.
7
DCDC1_CLK2X
RW 0
Set to one to slow down the DCDC #1 clock frequency by
a factor of 1/2. Can be combined with DCDC1CLK4X.
6
DCDC1_CLK4X
RW 0
Set to one to slow down the DCDC #1 clock frequency by
a factor of 1/4. Can be combined with DCDC1CLK2X.
5
DCDC1_BAT_ADJ
RW 0
Set to one to use battery LRADC information to scale
VDDIO duty cycle in DCDC #1. This is only used in mode
101 or 111.
4
DCDC1_DIS_5BIT
RW 0
Set to one to reduce the quantization of DC-DC switching
times from 5 bits to 3 bits for DCDC #1. Clearing the
DCDC_1_DIS_5BIT forces the DCDC clock to operate at
1.5 Mhz, regardless of the state of the CLK2X and CLK4X
bits
3:0
DCDC1_ADJ_TN
RW 0000
Constant adjust duty cycle of DCDC #1 VDDIO output in
mode 111 or 101.
The optimum value for this field is based on VDDD and
VDDIO. When VDDIO is set to a typical 3.07 Volts, the
following is a suggested setting for DCDC1_ADJ_TN
based on VDDD value:
VDDD
DCDC1_ADJ_TN
DCDC1_BATADJ
0xa - 0xb
2
1
0xc - 0xe
1
1
0xf - -x11
0
1
>= x12
0
0
Table 485. DC-DC Test Bit Register Description (Continued)
364
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
30.5.10. DC-DC Persistent Bit Register
The organization of the DC-DC Persistent Register is shown below. This register
has a shadow copy that is in the DCLK clock and DSP power domain. It accessible
by the DSP. There is a master copy of this register in the VDDXTAL clock and power
domain. The persistent bits control various function states during power down
modes of operation. Writing the shadow copy from the DSP does not automatically
update the master copy. One must preload the shadow bits with desired values and
then toggle the UPDATE bit to copy the values from the shadow version to the master version.
1
1
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
XTAL_BIAS_DOWN0
1
2
0
0
XTAL_TRIM_ENABLE
1
3
XTAL_BIAS_DOWN1
1
4
XTAL_TRIM0
1
5
XTAL_TRIM1
1
6
SLEEP_XTAL_ENABLE
1
7
LOW_BATT_TYPE
1
8
LOW_BATTERY_ENABLE
1
9
AUTO_RESTART
2
0
DELAY_5V_AUTO_RESTART
2
1
UDPATE
2
2
AUTO_RESTART_STAT
2
3
DELAY_5V_AUTO_RESTART_STAT
HW_DCDC_PERSIST X:$FA1B
Table 486. HW_DCDC_PERSIST
BITS
LABEL
23:13 RSRVD
12
AUTO_RESTART_STAT
11
10
RW RESET
DEFINITION
R
0
Reserved – Must be written with 0.
R
0
This read only bit shows the state of the
AUTO_RESTART bit as it was last copied (by toggling
UPDATE) into the master register bit in the XTAL clock
and power domain. NOTE: Immediately after power up,
AUTO_RESTART_STAT reflects the persistent state,
whereas AUTO_RESTART will have just been reset to
zero.
DELAY_5V_AUTO_RESTART_STAT R
0
This read only bit shows the state of the
DELAY_5V_AUTO_RESTART bit as it was last copied
(by toggling UPDATE) into the master register bit in the
XTAL clock and power domain. NOTE: Immediately
after power up, DELAY_5V_AUTO_RESTART_STAT
reflects the persistent state, whereas
DELAY_5V_AUTO_RESTART will have just been
reset to zero.
UPDATE
RW 0
A zero to one transition on this bit causes the shadow
values of the HW_DCDC_PERSIT bits to be copied to
the master register in the VDDXTAL clock and power
domain.
Table 487. DC-DC Persistent Bit Register Description
5-35xx-D1-1.08-061705
365
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
9
AUTO_RESTART
8
DELAY_5V_AUTO_RESTART
7
LOW_BATT_TYPE
6
LOW_BATTERY_ENABLE
5
SLEEP_XTAL_ENABLE
4
XTAL_TRIM1
3
XTAL_TRIM0
2
1
XTAL_BIAS_DOWN1
XTAL_BIAS_DOWN0
0
XTAL_TRM_ENABLE
RW RESET
DEFINITION
RW 0
Set this bit to one to enable a 5KΩ resistor between the
VDDXTAL pin and the PSWITCH pin. When the DCDC
converter powers down the chip either because the
power down bit was set in HW_CCR or because the 5V
source was removed, it waits for the power switch to be
pressed. Enabling the 5KΩ resistor over-rides the
power switch and makes it behave as if it had just been
pressed.
RW 0
The DCDC converter uses two primary methods to
trigger a DSP restart. The first waits for the power
switch to be pressed (or the 5KΩ resistor switched in
by AUTO_RESTART). The second trigger is generated
when a voltage greater than VDDIO appears on the 5V
pin. Customers should not set this bit to one without
explicit advice from SigmaTel.
RW 0
Set to one to select LiIon specific threshold for low
battery detection. Set to zero for other battery
chemistries.
RW 0
Set to one to enable the low battery detect function.
The crystal oscillator and RTC can operate on the
VDDXTAL power domain even while the rest of the
chip is powered down. Setting this bit to one enables a
brownout detector for the VDDXTAL domain so that the
RTC and Persistent bits will not be scrambled in an
under voltage event.
RW 0
Set to one to allow crystal oscillator to remain on during
power down state. NOTE: default turns the crystal off. It
must be turned on to use the Real Time Clock and
Alarm.
NOTE: HW_RTC_PERSIST0_XTAL_PDN must also
be set to zero to enable crystal oscillator operation
during power down states.
RW 0
Set to one to add four pF capacitance to XTAL0/XTALI
pins.
RW 0
Set to one to add four more pF capacitance to
XTAL0/XTALI pins. Be sure to turn on XTAL_TRIM1
first.
RW 0
Set to one to shrink the crystal current 30%.
RW 0
Set to one to shrink the crystal current another 30%. Be
sure to turn on XTAL_BIAS_DOWN1 first.
RW 0
Set to one to enable a resistor between the crystal
capacitors. This allows the caps. added by
XTAL_TRIM0 and XTAL_TRIM1 to start at the correct
DC level to avoid a clock glitch when adding the caps.
Table 487. DC-DC Persistent Bit Register Description (Continued)
366
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
30.5.11. Power Charger Register
This register controls the 5 Volt power source mode. In this mode, the power
charger can take power from the 5 Volt input from the USB cable or from a wall
transformer for normal operation of the chip. The STMP35xx can optionally supply
energy to recharge the battery in 5 Volt power mode.
1
3
1
2
1
1
DISABLE_ILIMIT
LI_TYPE
1
0
0
9
0
8
0
7
0
6
0
5
0
4
0
3
0
2
0
1
0
0
BATT_CURRENT
1
4
RES
1
5
PWD
1
6
\
1
7
PWDN_ON_IOBRNOUT
1
8
DCANA_LP
1
9
TEST_USBREGS
2
0
DRV_BAT
2
1
X:$FA1D
SWCHRG_BAT
2
2
BATT_INFO
2
3
VDD5V_PRESENT
HW_VDD5V_PWR_CHARGE
Table 488. HW_VDD5V_PWR_CHARGE
BITS
LABEL
RW RESET
23:22 RSRVD
21
BATT_INFO
R
R
0
-
20
R
-
VDD5V_PRESENT
19:18 RSRVD
17
SWCHRG_BAT
R
00
RW 0
16
DRV_BAT
RW 0
15
TEST_USBREGS
RW 0
14
DCANA_LP
RW 0
DEFINITION
Reserved – Must be written with 0.
This read only test is intended for chip validation and software
debug but should not be used in software charging algorithms.
This bit is set to one when the linear charger is supplying current to
the battery. It is set to zero when the charging current falls below
approximately 0.5 mA.
This read-only status bit is set to one when the 5V input pin voltage
is greater than the VDDIO rail voltage.
Reserved – Must be written with 0.
Set to one to enable the DCDC converter to supply energy to the
NiMH battery using its switching mode for higher efficiency
charging.
WARNING: This function has been deprecated and is no longer
supported.
Set this SigmaTel test mode bit to one to connect the VDDD rail to
the BATT pin.
Set to one to enable the 5V sourced linear regulators. Set to zero
for normal operation, wherein the 5V sensor controls the linear
regulators. NOTE: this is a SigmaTel only test function.
Set to one to power down unused circuitry in the DCDC converter.
This is useful for minimizing current in USB standby mode.
Table 489. VDD5V Power Charger Register Description
5-35xx-D1-1.08-061705
367
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
BITS
LABEL
RW RESET
13
PWDN_ON_IOBRNOUT RW 0
12
DISABLE_ILIMIT
RW 0
11
LI_TYPE
RW 0
10
9
RSRVD
RES
R
000
RW 0
8
PWD
RW 01
7:5
4:0
RSRVD
BATT_CURRENT
R
000
RW 00000
DEFINITION
Reset the system when a brownout event is detected on the
VDDIO rail. This will commonly occur when the system which has
been drawing current from the USB 5V bus is suddenly unplugged.
Set to one to enable system reset on VDDIO brownout detection.
WARNING: setting this bit to one enables the linear regulators and
shuts down the DCDC converter, even if VDD5V is not connected
to a power source. Always check for 5 V presence on VDD5V
before setting this bit to one. It is required to clear the PSRN
field of MIXTBR register when this functionality is enabled.
Set to one to disable In-rush current limiter on the linear charger.
Set to zero to enable the current limiter. This circuit is present to
meet the USB 2.0 in-rush current specification.
Linear Charger LiIon protocol: Set to zero for a 4.2 volt LiIon
battery. Set it to one for a 4.1 volt LiIon battery.
Reserved – Must be written with 0.
When this bit is set to zero the charging current is derived using the
accurate off-chip resistor shared with the integrated USB 2.0 PHY.
When set to one, an on-chip bias is used.
Battery charger powerdown mode, turn off all battery charge
current except those needed to avoid over voltage when 5V is
sensed.
Reserved – Must be written with 0.
These bits control the magnitude of the battery charge current. For
the linear regulator, these values are:
00000 = OFF
00001 = 10 mA
00010 = 25 mA
00011 = 35 mA
00100 = 50 mA
01000 = 100 mA
10000 = 200 mA
11111 = 385 mA
Currents are additive for other combinations.
NOTE: set HW_USBPHYPWD_PWDIBIAS to zero and set
HW_REF_CTRL_LWREF to zero for proper operation.When
switched mode charger is used (NiMH)
(HW_VDD5V_PWR_CHARGE_SWCHRG_BAT = 1) then the
battery current = (charge current from table above) * 0.8 * (charger
efficiency) * VDDIO/Vbattery.
WARNING: The switched mode charging function has been
deprecated and is no longer supported.
Table 489. VDD5V Power Charger Register Description (Continued)
Charger current limitations due to power dissipation, electromigration limits and or
temperature limits will be determined during chip characterization.
368
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
30.6. DC-DC Converter Efficiency
The following table shows the typical on resistances of the power FETs in both DCDC converters. These are nominal values that include package parasitic resistance
in the TQFP package. The fpBGA package has significantly more parasitic resistance than the TQFP, so resistance values may be approximately 100-200 mΩ
higher in the fpBGA.
DC-DC #1
DC-DC #1
DC-DC #1
DC-DC #1
DC-DC #2
DC-DC #2
NFET
PFET connected to VddA
PFET connected to VddD
PFET connected to VddIO
NFET
PFET
0.5 Ω
0.7 Ω
0.7 Ω
1.1 Ω
0.5 Ω
0.7 Ω
Table 490. Typical Resistance of Power Fets on DC-DC Converters
The following graphs show typical efficiency plots vs. the battery voltage for the DCDC converters configured in different modes. These graphs show measurements
made with typical devices at room temperature with specific static loads. Therefore,
these graphs should not be interpreted as specifications, but as estimates of typical
efficiencies under nominal conditions.
30.6.1.
DCDC1 Mode111 Efficiency
DcDc#1 Dual Boost Mode efficiency curves for various VDD loads, VDD=1.52V (VDDIO pwr = 3mW)
100
"dcdc1_boost_VDD_75mW" using 1:2
"dcdc1_boost_VDD_150mW" using 1:2
"dcdc1_boost_VDD_225mW" using 1:2
"dcdc1_boost_VDD_300mW" using 1:2
"dcdc1_boost_VDD_375mW" using 1:2
95
90
85
80
75
70
1.6
1.5
1.4
1.3
1.2
1.1
1
0.9
Figure 128. Efficiency of DC-DC #1 in mode 111 with VddIO=3mW
5-35xx-D1-1.08-061705
369
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
DcDc#1 Dual Boost Mode efficiency curves for various loads, VDDIO=3.01 (VDD pwr = 75mW)
100
"dcdc1_boost_VDDIO_75mW" using 1:2
"dcdc1_boost_VDDIO_150mW" using 1:2
"dcdc1_boost_VDDIO_225mW" using 1:2
"dcdc1_boost_VDDIO_300mW" using 1:2
95
90
85
80
75
70
1.6
1.5
1.4
1.3
1.2
1.1
1
0.9
Figure 129. Efficiency of DC-DC #1 in mode 111 with VddD load=75mW
These two graphs show the efficiency of the DC-DC converter (#1 in mode 111) versus battery voltage for several values of output power. The test conditions for the
two graphs are identical except for the load the power is delivered to. Because the
power FET that connects the inductor to the DCDC_VDDIO output has a higher
resistance than the FET that connects the inductor to the DCDC_VDDD output, the
efficiency of the DC-DC converter is improved when the same power is delivered to
the low voltage output, VddD, relative to the high voltage output, VddIO.
It should be noted that the boost configuration of DC-DC #1 (mode 101 and mode
111) could be the least efficient configurations for two reasons. First, Boost Mode
power conversion only transfers power to the load during one phase of the converters charge/discharge cycle. Therefore, conservation of charge requires an inductor
current that is higher than the load current, so I2R losses in the power FETs that
switch the inductor are increased relative to a Buck Mode configuration where current is transferred to the load during the entire charge/discharge cycle. Secondly,
since mode 101 and mode 111 generate multiple outputs using one external inductor, the inductor current must increase to support multiple load currents. Thus, I2R
losses are also increased when using the multiple output configurations relative to
the configurations that have one inductor per output voltage. Therefore, mode 101
and mode 111 are primarily intended for lower power applications that use a single
battery and single inductor to achieve an ultra small form factor.
370
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
30.6.2.
DCDC2 Mode001 Efficiency
DcDc#2 Boost Mode efficiency curves for various loads, VDDIO=3.07
100
"dcdc2_boost_75mW" using 1:2
"dcdc2_boost_450mW" using 1:2
"dcdc2_boost_800mW" using 1:2
"dcdc2_boost_1000mW" using 1:2
95
90
85
80
75
70
3
2.8
2.6
2.4
2.2
2
1.8
Figure 130. Efficiency of DC-DC #2 in mode 001 for various current loads
This graph shows the efficiency of DC-DC #2 for two values of output power. Since
DC-DC #2 has a PFET with less on-resistance than the comparable DC-DC #1
PFET, and this configuration is a single load converter, this mode is preferred for
higher power VddIO loads than mode 111 or mode 101.
30.6.3.
Max Power Out in Boost Mode
For a Boost Mode converter, the power delivered to the load can be estimated using
the following equation:
2
V OUT ( V BAT ( 1 – D ) – V OUT ( 1 – D ) )
P OUT = ------------------------------------------------------------------------------------------RP + RL + D ( RN – RP )
In this equation “D” represents the duty cycle of DC-DC converter, Vout represents
the output voltage of the converter, Rn and Rp represent the on resistance of the
power fets appropriate for the particular DC-DC converter configuration, and Rl represents the on-resistance of the DC-DC inductor. For multiple output modes 111 and
101, this equation is a reasonable approximation if it is assumed that all the power is
transferred to one of the loads. This equation leads to the following expression for
the duty cycle that gives to the maximum power transferred to the load:
D MAX
V BAT ( R P – R N )
R P – R N 1 + ------------- × -----------------------V OUT
RN
= ------------------------------------------------------------------------------RP – RN
Substituting Dmax into the equation for Pout gives a maximum power that can be
theoretically delivered to the load. Note that this equation was derived using only
5-35xx-D1-1.08-061705
371
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
V=Ldi/dt and conservation of charge, and thus represents a theoretical maximum
that neglects many significant sources of loss including frequency dependent core
losses, bond wire ringing, switch non-overlap times, etc.
From an application point of view, it is probably not desirable to operate the converter at the maximum power output, since I2R losses will also be maximized.
Therefore, it is recommended to constrain system power requirements to loads that
can be supplied with efficiency ≥ 70% as shown in the graphs in Figures 128 and
130.
30.6.4.
DCDC1 Mode011/Mode001/Mode000 Efficiency
DcDc#1 Buck Mode efficiency curves for various loads, VDD=1.52V
100
"dcdc1_buck_VDD_75mW" using 1:2
"dcdc1_buck_VDD_150mW" using 1:2
"dcdc1_buck_VDD_225mW" using 1:2
"dcdc1_buck_VDD_300mW" using 1:2
"dcdc1_buck_VDD_375mW" using 1:2
95
90
85
80
75
70
4
3.5
3
2.5
2
Figure 131. Efficiency of DC-DC #1 in mode 011/mode 001/mod000 vs. load current
This graph shows the efficiency for DC-DC #1 configured as a single channel buck
converter. Since DC-DC #1 is configured identically in modes 011, 001, 000, this
graph is applicable in all three modes. The positive slope of the 75mW curve and
150mW curve is due to the fact that the current required for the PMOS driver dominates the loss at these load levels. Since the PMOS driver power is proportional to
the battery voltage squared, these losses decrease and the efficiency improves as
the battery voltage falls.
372
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
30.6.5.
DCDC2 Mode000 Efficiency
DcDc#2 Buck Mode efficiency curves for various loads, VDDIO = 3.01
100
"dcdc2_buck_75mW" using 1:2
"dcdc2_buck_450mW" using 1:2
"dcdc2_buck_800mW" using 1:2
"dcdc2_buck_1000mW" using 1:2
95
90
85
80
75
70
4.2
4
3.8
3.6
3.4
3.2
3
This graph shows the efficiency for DC-DC #2 configured as a single channel buck
converter. The positive slope of the 75mW curve is due to the fact that the current
required for the PMOS driver dominates the loss at this load levels. Since the PMOS
driver power is proportional to the battery voltage squared, these losses decrease
and efficiency improves as the battery voltage falls.
Also note that the 1000mW curve does not extend below a battery voltage of 3.6V.
This is because converter has stopped switching and the PMOS device is in dropout
(i.e. on 100% of the duty cycle), and the load voltage is not well regulated. It is not
recommended to use the integrated dc-dc converters when load currents and output
voltages cause the converter to drop out.
30.6.6.
Max Power Out in Buck Mode
Unlike the boost case, the maximum output power from a Buck Mode configuration
is limited simply by voltage difference between the load and the battery divided by
the Pfet on resistance.
( Vbat – Vout ) × Vout
P OUT = -----------------------------------------------------Rp + RL
Thus a Buck Mode configuration can better supply large currents to the load than
Boost Mode configurations.
5-35xx-D1-1.08-061705
373
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
30.6.7.
Clock Speed vs. Voltage
The following table can be used to select a proper setting for VddD and VddD
brownout voltages based on standard analysis of worst case design and characterization data.
MAX
DCLK
TARGET
75.6MHz
69.6MHz
64.8MHz
60.0MHz
48.0MHz
39.6MHz
Min.
VddD
HW_DCDC_VDDD_
VOLTAGE_LEVEL
Corresponding
HW_DCDC_VDDD_
VddD
BROWNOUT_LEVEL
Brownout Voltage
1.92 V
11100
1.85 V
11010
1.82 V
11001
1.76 V
10111
1.73 V
10110
1.66 V
10100
1.63 V
10011
1.54 V
10000
1.4 V
01100
1.34 V
01010
1.37 V
01011
1.28 V
01000
Table 491. Recommended Operating Conditions for specific dclk targets
30.7. System Brownout
The chip also contains circuitry to sense when the power requirements of the system are more than the power the DC-DC converters can provide. The entrance into
the state of power required > power available is called a brownout event. Detection
of a brownout event is important to provide a controlled shutdown and acceptable
system behavior in the event of battery falling out, weak battery, short circuit, etc.
Typically, these brownout events are low frequency events (<100 kHz) due to the
large capacitors on the battery and the supply rails in the application, and the software can shutdown the application in a controlled manner. Figure 132 shows the circuitry available to detect a brownout event:
As shown in Figure 132, there are three brownout sources that can be used to interrupt the DSP via a NMI or IRQB. All brownouts are detected with comparators which
detect when a crucial system voltages (battery, VddD, VddIO) falls below a software
defined target voltage. These brownout detection comparators have a resolution
time of approximately 1us. This finite resolution time combined with the internal lowpass filter pole at 1 MHz, limit the frequency range for which a brownout event will
be detected. In the audio frequency range, the brownout events will be detected
when any instantaneous system voltage falls below the target voltage as defined in
the software control of the Voltage Generator blocks. In the 100 kHz frequency
range, the comparator resolution time will act to reduce brownout sensitivity, and the
instantaneous system voltage will need to fall below the software control value to
detect a brownout event. As the frequency continues to increase, this effect continues to worsen and by 500kHz, a brownout event cannot be detected. Thus, it is necessary to decouple the supplies well on the board to filter any high frequency
transients that could cause an undetected brownout event.
30.8. Known Chip Defects with Power Management Subsystem
30.8.1.
Clear Quest Entry STMP00005058
All revisions of the chip are known to have a defect in which the USB 5V detect line
may not transition low when 5V removed while driving heavy loads.
This issue is related to applications that perform a hand off between the linear regulators and the switching dcdc converters. The detect circuit that senses 5V is not
374
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
HW_VDD5V_PWR_CHARGE_
PWDN_ON_IOBRNOUT
HW_DCDC_VDDIO_
BROWNOUT_LEVEL
Comparator
resolution time ~1µs
HW_DCDC_VDDIO_
BROWNOUT_STATUS
STOP
DCDC
Converter
Voltage Generator
HW_CCR_PWDN
VddD
Internal Pole
@ ~1MHz
HW_DCDC_VDDIO_
BROWNOUT_ENABLE
HW_DCDC_VDDD_
BROWNOUT_LEVEL
HW_DCDC_VDDD_
BROWNOUT_STATUS
stack over/
under flow
Voltage Generator
NMI to
DSP
VddIO
HW_DCDC_VDDD_
BROWNOUT_ENABLE
HW_DCDC_VDDA_
BROWNOUT_LEVEL
HW_DCDC_RCR_
IRQB2NMI
HW_DCDC_VDDA_
BROWNOUT_STATUS
IRQB
to DSP
Voltage Generator
VddA
HW_IPR_IRQBP[0]
HW_DCDC_VDDA_
BROWNOUT_ENABLE
LRADC_BATT_EVENT_0_1_IRQ
Figure 132. Brownout Event Detect Available Circuitry
present detects when 5V is less than a Vt above VDDIO. Unfortunately, when the
VDDIO linear regulator is sourcing fairly large currents then dropout voltage of the
linear regulator is greater than a Vt. This means that the 5v detect circuit will not
sense that 5v is removed. Therefore, dc-dc hand off applications the software
should sense when a VDDIO brownout occurs instead of when a 5v unplug is
detected.
30.8.2.
Clear Quest Entry STMP00004681
All revisions of the chip are known to have a latch up issue in DCDC #1 with a LiIon
battery that will cause the battery voltage to momentarily drop and thus cause the
VDD voltage to drop. The latchup issue appears to be in a floating well with the battery charger. The pin inductance initiates current into the substrate which will create
an SCR path to the VDD5V pin. The SCR is only active for a brief time and will not
cause destructive currents to flow. There is a software work around. Changing the
speed of the DCDC #1 clock to 6MHz will reduce the ripple of the inductor current
4X, and thus reduce the voltage transient caused by the pin inductance. Testing
shows that this completely eliminates the problem for VDD loads of at least 200mA.
Customers must also follow layout guidelines to place decoupling capacitors close
to the dcdc1_vddio pin.
5-35xx-D1-1.08-061705
375
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
30.8.3.
Clear Quest Entry STMP00004530
All revisions of the chip are known to have a defect in which the chip will reset at
high LiIon battery voltages and low core voltage when 5V is connected. This is only
an issue for applications that want a invisible transfer between the switching dc-dc
converter and integrated linear regulators.
This problem should be invisible to customers. The software that implements to
transfer between dc-dc converters and linear regulators contains the necessary
code to eliminate this issue.
30.8.4.
Clear Quest Entry STMP00004413
All revisions of the chip are known to have a defect in which the switch logic of the
dcdc converter that allows a large current to flow through these switches when the
battery voltage is a Vt above VDDIO. The only case in which this should be a problem is LiIon. All capacitors should be removed from the dcdc_vddd and dcdc_vdda
pins when operating in buck mode. These outputs should be left floating in non-alkaline/NiMH modes.
30.8.5.
Clear Quest Entry STMP00004201
All revisions of the chip are known to have a defect in which an analog mux that
switches the xtal/rtc current from the battery to a power source created from the
VDD5V input. This leakage current is in the 200uA range, and will decrease as the
LiIon battery voltage drops.
This defect is only an issue when a player is left connected to 5V with battery charge
turned off.
30.8.6.
Clear Quest Entry STMP00003530
All revisions of the chip are known to have a defect in which there is a possibility for
charge flow across the DC-DC #2 P-FET when the following conditions are met:
1) External Power (i.e. 5V power)
2) DC-DC is configured in LiIon Buck mode 000
3) Battery voltage is lower than VDDIO-Vt
For example, if a Li-ION battery is at 2.5V and the part is powered up from Vdd5v
then the body diode in the DC-DC#2 P-FET will be forward biased and dump current
into the battery. Since the linear regulators are current limited to ~50mA, this current
will flow into the battery and effectively trickle charge it. Only when the voltage on
the battery rises enough to allow VDDIO to rise above the it's startup voltage will the
player complete the startup sequence and begin operating.
The behavior can be eliminated, if desired, with external FET switches that disconnect the battery from the dcdc converters when 5V is present.
376
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
31. PIN DESCRIPTION
10 x 10 mm Pin A1
Pin 1
14 x 14 mm
100 TQFP
144 fpBGA
For additional package measurements, please see 32. “PACKAGE DRAWINGS” on page 389.
Figure 133. Chip Package Photos
31.1. Pin Placement and Definitions
100
TQFP
144
fpBGA
1
M2
2
L2
3
K4
4
M3
5
L3
6
7
H6
M4
8
J5
9
L4
10
11
12
G6
H7
M5
NA
K5
13
L5
NA
M6
14
K6
NA
J6
PIN NAME
MODULE
TYPE
GP14
SPI_MOSI
GP13
SPI_MISO
GP12
SPI_SCK
GPIO
SPI
GPIO
SPI
GPIO
SPI
I/O
I/O
I/O
I/O
I/O
I/O
GP16
I2C_SCL
GP17
I2C_SDA
TESTMODE
CF_CE1n
GP44
CF_IORDn
GP52
CF_IOWRn
GP51
VssD2
VddD2
CF_A0
GP32
RAM_A0
CF_A22
GP69
UTMI_RXVALID
CF_A1
GP33
RAM_A1
CF_A21
GP68
UTMI_RXACTIVE
CF_A2
GP34
RAM_A2
CF_A20
GP67
UTMI_RXVALIDH
GPIO
I2 C
GPIO
I2C
SYSTEM
EMC-CF
GPIO
EMC-CF
GPIO
EMC-CF
GPIO
POWER
POWER
EMC-CF
GPIO
SDRAM
EMC-CF
GPIO
UTMI
EMC-CF
GPIO
SDRAM
EMC-CF
GPIO
UTMI
EMC-CF
GPIO
SDRAM
EMC-CF
GPIO
UTMI
I/O
I/O
I/O
I/O
I
O
I/O
O
I/O
O
I/O
P
P
O
I/O
O
O
I/O
I
O
I/O
O
O
I/O
I
O
I/O
O
O
I/O
I
PIN
SET
DIO3,
DIO18
DIO3,
DIO18
D,
DIO3,
DIO18
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DESCRIPTION
GP0B14
SPI Master Output/Slave Input
GP0B13
SPI Master Input/Slave Output
GP0B12
SPI Serial Clock
GP0B16
I2C Serial Clock
GP0B17
I2C Serial Data
Test Mode Pin
CompactFlash Chip Enable 1
GP1B20
CompactFlash I/O Read Data Strobe
GP2B4
CompactFlash I/O Write Data Strobe
GP2B3
Digital Core Ground 2
Digital Core Power 2
CompactFlash Address 0
GP1B8
SDRAM Address 0
CompactFlash Address 22
GP2B21
UTMI receive valid
CompactFlash Address 1
GP1B9
SDRAM Address 1
CompactFlash Address 21
GP2B20
UTMI
CompactFlash Address 2
GP1B10 (play switch recovery if used)
SDRAM Address 2
CompactFlash Address 20
GP2B19
UTMI
Table 492. Pin Definition Table
5-35xx-D1-1.08-061705
377
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
100
TQFP
144
fpBGA
15
L6
NA
L7
16
K7
NA
NA
NA
F7
G7
M7
17
J7
NA
L8
18
K8
NA
M8
19
J8
NA
M9
20
L9
NA
K9
21
L10
22
M10
23
M11
24
K10
25
M12
PIN NAME
CF_A3
GP35
RAM_A3
CF_A19
GP66
UTMI_RXERR
CF_A4
SM_CE3n
GP36
RAM_A4
VddIO3
VssIO3
CF_A18
GP65
UTMI_DATA0
CF_A5
SM_CE2n
GP37
RAM_A5
CF_A17
GP64
UTMI_DATA1
CF_A6
SM_CE0n
GP38
RAM_A6
CF_A16
GP63
UTMI_DATA2
CF_A7
SM_SEn
GP39
RAM_A7
CF_A15
GP62
RAM_A13
UTMI_DATA3
CF_A8
SM_CLE
GP40
RAM_A8
CF_A14
GP61
RAM_A12
UTMI_DATA4
CF_A9
SM_ALE
GP41
RAM_A9
CF_A10
GP42
RAM_A10
CF_OEn
SM_REn
GP53
CF_CE0n
SM_CE1n
GP45
CF_WAITn
SM_READY
GP56
MODULE
TYPE
EMC-CF
GPIO
SDRAM
EMC-CF
GPIO
UTMI
EMC-CF
EMC-SM
GPIO
SDRAM
POWER
POWER
EMC-CF
GPIO
UTMI
EMC-CF
EMC-SM
GPIO
SDRAM
EMC-CF
GPIO
UTMI
EMC-CF
EMC-SM
GPIO
SDRAM
EMC-CF
GPIO
UTMI
EMC-CF
EMC-SM
GPIO
SDRAM
EMC-CF
GPIO
SDRAM
UTMI
EMC-CF
EMC-SM
GPIO
SDRAM
EMC-CF
GPIO
SDRAM
UTMI
EMC-CF
EMC-SM
GPIO
SDRAM
EMC-CF
GPIO
SDRAM
EMC-CF
EMC-SM
GPIO
EMC-CF
EMC-SM
GPIO
EMC-CF
EMC-SM
GPIO
O
I/O
O
O
I/O
I
O
O
I/O
O
P
P
O
I/O
I/O
O
O
I/O
O
O
I/O
I/O
O
O
I/O
O
O
I/O
I/O
O
O
I/O
O
O
I/O
O
I/O
O
O
I/O
O
O
I/O
O
I/O
O
O
I/O
O
O
I/O
O
O
O
I/O
O
O
I/O
I
I
I/O
PIN
SET
DIO3
DIO3
DIO3,
DIO18
DIO3
DIO3,
DIO18
DIO3
DIO3,
DIO18
DIO3
DIO3
DIO3
DIO3,
DIO18
DIO3
DIO3,
DIO18
DIO3
DIO3,
DIO18
DIO3,
DIO18
DIO3,
DIO18
DESCRIPTION
CompactFlash Address 3
GP1B11
SDRAM Address 3
CompactFlash Address 19
GP2B18
UTMI receive error
CompactFlash Address 4
SmartMedia/NAND Chip Enable 3
GP1B12
SDRAM Address 4
Digital I/O Power 3
Digital I/O Ground 3
CompactFlash Address 18
GP2B17
UTMI data bus
CompactFlash Address 5
SmartMedia/NAND Chip Enable 2
GP1B13
SDRAM Address 5
CompactFlash Address 17
GP2B16
UTMI data bus
CompactFlash Address 6
SmartMedia/NAND Chip Enable 0
GP1B14
SDRAM Address 6
CompactFlash Address 16
GP2B15
UTMI Data Bus
CompactFlash Address 7
SmartMedia/NAND Spare Area Enable
GP1B15
SDRAM Address 7
CompactFlash Address 15
GP2B14
SDRAM Address 13
UTMI Data Bus
CompactFlash Address 8
SmartMedia/NAND Command Latch Enable
GP1B16
SDRAM Address 8
CompactFlash Address 14
GP2B13
SDRAM Address 12
UTMI Data Bus
CompactFlash Address 9
SmartMedia/NAND Address Latch Enable
GP1B17
SDRAM Address 9
CompactFlash Address 10
GP1B18
SDRAM Address 10
CompactFlash Output Enable Strobe
SmartMedia/NAND Read Enable Strobe
GP2B5
CompactFlash Chip Enable 0
SmartMedia/NAND Chip Enable 1
GP1B21
CompactFlash Wait (not)
SmartMedia/NAND Ready/Busy
GP2B8
Table 492. Pin Definition Table (Continued)
378
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
100
TQFP
144
fpBGA
NA
L12
26
L11
NA
K11
27
K12
NA
J9
28
29
30
H8
G8
J10
NA
J12
31
H9
NA
J11
32
H12
NA
H11
33
H10
NA
G11
34
G10
NA
G12
35
G9
PIN NAME
CF_A11
GP80
RAM_A11
UTMI_DATA5
CF_WEn
SM_WEn
GP54
CF_A12
GP81
RAM_BA0
UTMI_DATA6
SM_WPn
GP55
CF_A13
GP82
RAM_BA1
UTMI_DATA7
VssIO1
VddIO1
CF_D0
SM_D0
GP24
RAM_D0
CF_D15
GP79
RAM_D15
UTMI_DATA8
CF_D1
SM_D1
GP25
RAM_D1
CF_D14
GP78
RAM_D14
UTMI_DATA9
CF_D2
SM_D2
GP26
RAM_D2
CF_D13
GP77
RAM_D13
UTMI_DATA10
CF_D3
SM_D3
GP27
RAM_D3
CF_D12
GP76
RAM_D12
UTMI_DATA11
CF_D4
SM_D4
GP28
RAM_D4
CF_D11
GP75
RAM_D11
UTMI_DATA12
CF_D5
SM_D5
GP29
RAM_D5
MODULE
TYPE
EMC-CF
GPIO
SDRAM
UTMI
EMC-CF
EMC-SM
GPIO
EMC-CF
GPIO
SDRAM
UTMI
EMC-SM
GPIO
EMC-CF
GPIO
SDRAM
UTMI
POWER
POWER
EMC-CF
EMC-SM
GPIO
SDRAM
EMC-CF
GPIO
SDRAM
UTMI
EMC-CF
EMC-SM
GPIO
SDRAM
EMC-CF
GPIO
SDRAM
UTMI
EMC-CF
EMC-SM
GPIO
SDRAM
EMC-CF
GPIO
SDRAM
UTMI
EMC-CF
EMC-SM
GPIO
SDRAM
EMC-CF
GPIO
SDRAM
UTMI
EMC-CF
EMC-SM
GPIO
SDRAM
EMC-CF
GPIO
SDRAM
UTMI
EMC-CF
EMC-SM
GPIO
SDRAM
O
I/O
O
I/O
O
O
I/O
O
I/O
O
I/O
O
I/O
O
I/O
O
I/O
P
P
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
PIN
SET
DIO3
DIO3,
DIO18
DIO3
DIO3,
DIO18
DIO3
DIO3,
DIO18
DIO3,
DIO18
DIO3,
DIO18
DIO3,
DIO18
DIO3,
DIO18
DIO3,
DIO18
DIO3,
DIO18
DIO3,
DIO18
DIO3,
DIO18
DIO3,
DIO18
DIO3,
DIO18
DESCRIPTION
CompactFlash Address 11
GP3B8
SDRAM Address 11
UTMI Data Bus
CompactFlash Write Enable Strobe
SmartMedia/NAND Write Enable Strobe
GP2B6
CompactFlash Address 12
GP3B9
SDRAM Bank Address 0
UTMI Data Bus
SmartMedia/NAND Write Protect
GP2B7
CompactFlash Address 13
GP3B10
SDRAM Bank Address 1
UTMI Data Bus
Digital I/O Ground 1
Digital I/O Power 1
CompactFlash Data 0
SmartMedia/NAND I/O 0
GP1B0
SDRAM Data 0
CompactFlash Data 15
GP3B7
SDRAM Data 15
UTMI Data Bus
CompactFlash Data 1
SmartMedia/NAND I/O 1
GP1B1
SDRAM Data 1
CompactFlash Data 14
GP3B6
SDRAM Data 14
UTMI Data Bus
CompactFlash Data 2
SmartMedia/NAND I/O 2
GP1B2
SDRAM Data 2
CompactFlash Data 13
GP3B5
SDRAM Data 13
UTMI Data Bus
CompactFlash Data 3
SmartMedia/NAND I/O 3
GP1B3
SDRAM Data 3
CompactFlash Data 12
GP3B4
SDRAM Data 12
UTMI Data Bus
CompactFlash Data 4
SmartMedia/NAND I/O 4
GP1B4
SDRAM Data 4
CompactFlash Data 11
GP3B3
SDRAM Data 11
UTMI Data Bus
CompactFlash Data 5
SmartMedia/NAND I/O 5
GP1B5
SDRAM Data 5
Table 492. Pin Definition Table (Continued)
5-35xx-D1-1.08-061705
379
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
100
TQFP
144
fpBGA
NA
F9
36
F11
NA
F12
37
F10
38
39
NA
F8
E8
E9
40
E12
41
E10
42
E11
43
C11
44
D11
45
D10
46
47
48
49
50
NA
NA
NA
NA
NA
NA
51
52
53
NA
B11
D12
C12
B12
A12
A11
E7
E6
B10
A10
A9
C10
D9
C9
B9
54
55
56
57
D8
A7
D7
B8
58
A8
59
C8
60
C7
PIN NAME
CF_D10
GP74
RAM_D10
UTMI_DATA13
CF_D6
SM_D6
GP30
RAM_D6
CF_D9
GP73
RAM_D9
UTMI_DATA14
CF_D7
SM_D7
GP31
RAM_D7
VssD1
VddD1
CF_D8
GP72
RAM_D8
CF_CDn
GP46
CF_READY
GP47
CF_WPn
GP48
CF_RESETn
GP50
CF_REGn
GP43
CF_BVD1
GP49
ONCE_DSI
DCDC_VddIO
DCDC_VddD
DCDC_Batt
DCDC_Gnd
DCDC_VddA
VddIO4
VssIO4
DCDC2_Gnd
DCDC2_Batt
DCDC2_Vout
ONCE_DSK
ONCE_DSO
ONCE_DRN
GP83
RAM_DQM0
UTMI_DATA15
DCDC_mod2
MIC
BATT
LINE1L
HP_SENSE
LRADC1
MIC_BIAS
LINE1R
HP_COMMON
LRADC2
TEMP_SENSE
MIC_BIAS
MODULE
TYPE
EMC-CF
GPIO
SDRAM
UTMI
EMC-CF
EMC-SM
GPIO
SDRAM
EMC-CF
GPIO
SDRAM
UTMI
EMC-CF
EMC-SM
GPIO
SDRAM
POWER
POWER
EMC-CF
GPIO
SDRAM
EMC-CF
GPIO
EMC-CF
GPIO
EMC-CF
GPIO
EMC-CF
GPIO
EMC-CF
GPIO
EMC-CF
GPIO
SYSTEM
DCDC
DCDC
DCDC
DCDC
DCDC
POWER
POWER
DCDC
DCDC
DCDC
SYSTEM
SYSTEM
SYSTEM
GPIO
SDRAM
UTMI
DCDC
CODEC
POWER
CODEC
CODEC
SYSTEM
CODEC
CODEC
CODEC
SYSTEM
SYSTEM
CODEC
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
P
P
I/O
I/O
I/O
I
I/O
I
I/O
I
I/O
I
I/O
O
I/O
I
I/O
I
P
P
P
P
P
P
P
P
P
P
O
O
I
I/O
O
I/O
P
A
P
A
A
A
A
A
A
A
A
A
PIN
SET
DIO3,
DIO18
DIO3,
DIO18
DIO3,
DIO18
DIO3,
DIO18
DIO3,
DIO18
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
AIO
AIO
AIO
AIO
AIO
DESCRIPTION
CompactFlash Data 10
GP3B2
SDRAM Data 10
UTMI Data Bus
CompactFlash Data 6
SmartMedia/NAND I/O 6
GP1B6
SDRAM Data 6
CompactFlash Data 9
GP3B1
SDRAM Data 9
UTMI Data Bus
CompactFlash Data 7
SmartMedia/NAND I/O 7
GP1B7
SDRAM Data 7
Digital Core Ground 1
Digital Core Power 1
CompactFlash Data 8
GP3B0
SDRAM Data 8
CompactFlash Card Detect
GP1B22
CompactFlash Ready
GP1B23
CompactFlash Write Protect
GP2B0
CompactFlash Reset
GP2B2
CompactFlash Register Select
GP1B19
CompactFlash Bad Voltage Detect
GP2B1
Debug Data In
DCDC VddIO
DCDC VddD
DCDC Battery
DCDC Ground
DCDC VddA
Digital I/O Power 4
Digital I/O Ground 4
DCDC2 Ground
DCDC2 Battery
DCDC2 Vout
Debug Clock
Debug Data Out
Debug Reset
GP3B11
SDRAM DQM0
UTMI Data Bus
DCDC mode pin 2
Microphone Input
Battery Input
Line-in 1 Left
Direct coupled headphone sense
Low Resolution ADC channel 1 Input
Microphone bias
Line-in 1 Right
Direct Coupled Headphone common amp.
Low Resolution ADC channel 2 input
Temperature sensor bias current
Alternate Microphone Bias
Table 492. Pin Definition Table (Continued)
380
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
100
TQFP
144
fpBGA
61
62
63
NA
64
65
NA
NA
66
67
68
69
70
71
NA
72
73
74
75
76
77
B7
B6
C6
D6
D5
A6
D4
C5
A5
B4
A4
B3
C4
C3
A3
B5
B2
A2
A1
B1
E4
78
79
E5
D3
80
81
82
83
C2
C1
E3
D2
NA
E2
84
85
86
87
88
D1
E1
F6
F5
F4
89
F3
NA
F2
90
G4
NA
F1
91
G3
NA
G2
92
H3
PIN NAME
MODULE
TYPE
VDD5V
HPL
VssHP
DCDC_mod1
VddHP
HPR
LINE2L
LINE2R
Vbg
Vag
ADCL
ADCR
REF_RES
REFp
DCDC_mod0
VssA1
VddA1
XTALO
XTALI
VddPLL
VddXTAL
POWER
CODEC
POWER
DCDC
POWER
CODEC
CODEC
CODEC
CODEC
CODEC
CODEC
CODEC
USB
CODEC
DCDC
POWER
POWER
SYSTEM
SYSTEM
POWER
SYSTEM
P
A
P
P
P
A
A
A
A
A
A
A
A
A
P
P
P
A
A
P
A
VssPLL
FRESET
POWER
SYSTEM
P
A
PSWITCH
USB_DP
USB_DM
GP11
GP9
PWM2
GP90
RAM_CLK
UTMI_TXREADY
GP8
GP10
VddD3
VssD3
GP7
I2S_DataO2
PWM3
GP6
I2S_DataO1
GP84
RAM_DQM1
UTMI_TXVALID
GP5
I2S_DataO0
GP89
RAM_CKE
UTMI_TXVALIDH
GP4
I2S_BCLK
GP88
RAM_CSn
UTMI_OPMODE0
GP3
I2S_LRCLK
SYSTEM
USB
USB
GPIO
GPIO
PWM
GPIO
SDRAM
UTMI
GPIO
GPIO
POWER
POWER
GPIO
I2S
PWM
GPIO
I2S
GPIO
SDRAM
UTMI
GPIO
I2S
GPIO
SDRAM
UTMI
GPIO
I2S
GPIO
SDRAM
UTMI
GPIO
I2 S
A
A
A
I/O
I/O
O
I/O
O
O
I/O
I/O
P
P
I/O
O
O
I/O
O
I/O
O
I
I/O
O
I/O
O
O
I/O
I
I/O
O
O
I/O
I
PIN
SET
AIO
AIO
AIO
AIO
AIO
AIO
AIO
AIO
AIO
AIO
AIO
USBIO
USBIO
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DESCRIPTION
USB/External 5V power source
Headphone/Line-out Left
Headphone Ground
DCDC mode pin 1
Headphone Power
Headphone/Line-out Right
Line-in 2 Left (AKA: FM-in Left)
Line-in 2 Right (AKA: FM-in Right)
Bandgap Decoupling Capacitor
Analog Ground Decoupling Capacitor
ADC Left Filter Capacitor
ADC Right Filter Capacitor
USB Precision Resistor
ADC Positive Reference Capacitor
DCDC mode pin 0
Analog Ground 1
Analog Power 1
Crystal Out
Crystal In
PLL Power
Power for XTAL oscillator (generated onchip)
PLL Ground
Fast Reset, Fast Falling Edge (<15ns)
Powers Down Device
Power Switch
USB Positive Data Line
USB Negative Data Line
GP0B11
GP0B9
Pulse Width Modulator #2
GP3B18
SDRAM Clock
UTMI transmit data ready
GP0B8
GP0B10
Digital Core Power 3
Digital Core Ground 3
GP0B7
I2S Data Out 2
Pulse Width Modulator #3
GP0B6
I2S Data Out 1
GP3B12
SDRAM DQM1
UTMI transmit valid low byte
GP0B5
I2S Data Out 0
GP3B17
SDRAM CKE
UTMI transmit valid high byte
GP0B4
I2S Bit Clock
GP3B16
SDRAM CSn
UTMI
GP0B3
I2S Word Clock
Table 492. Pin Definition Table (Continued)
5-35xx-D1-1.08-061705
381
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
100
TQFP
144
fpBGA
NA
G1
93
H4
NA
J3
94
H2
NA
H1
95
J1
NA
J2
96
97
98
G5
H5
J4
NA
K3
99
K2
NA
K1
100
L1
NA
M1
PIN NAME
GP85
RAM_RASn
UTMI_OPMODE1
GP2
I2S_DataI2
GP86
RAM_CASn
UTMI_XVER_SEL
GP1
I2S_DataI1
GP87
RAM_WEn
UTMI_LINESTATE0
GP0
I2S_DataI0
GP92
I2S_DataI0
UTMI_LINESTATE1
VddIO2
VssIO2
GP19
TIO1
PWM1
GP93
I2S_BCLK
UTMI_CLK
GP18
TIO0
PWM0
GP91
I2S_WCLK
UTMI_RESET
GP15
SPI_SSn
CF_A23
GP70
RAM_A23
MODULE
TYPE
GPIO
SDRAM
UTMI
GPIO
I2S
GPIO
SDRAM
UTMI
GPIO
I2S
GPIO
SDRAM
UTMI
GPIO
I2S
GPIO
I2S
UTMI
POWER
POWER
GPIO
TIMER
PWM
GPIO
I2S
UTMI
GPIO
TIMER
PWM
GPIO
I2S
UTMI
GPIO
SPI
EMC-CF
GPIO
SDRAM
I/O
O
O
I/O
I
I/O
O
O
I/O
I
I/O
O
I
I/O
I
I/O
I
I
P
P
I/O
I/O
O
I/O
I
O
I/O
I/O
O
I/O
I
O
I/O
I
O
I/O
O
PIN
SET
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DIO3
DESCRIPTION
GP3B13
SDRAM RASn
UTMI
GP0B2
I2S Data In 2
GP3B14
SDRAM CASn
UTMI
GP0B1
I2S Data In 1
GP3B15
SDRAM WEn
UTMI
GP0B0
I2S Data In 0
GP3B20
I2S Data In 0 (Alternate pin)
UTMI
Digital I/O Power 2
Digital I/O Ground 2
GP0B19
Timer 1 Pin
Pulse Width Modulator #1
GP3B21
I2S Bit Clock (Alternate pin)
UTMI
GP0B18
Timer 0 Pin
Pulse Width Modulator #0
GP3B19
I2S Word Clock (Alternate pin)
UTMI
GP0B15 (Do not use as GPIO if using SPI)
SPI Slave Select
CompactFlash Address 23
GP2B22
SDRAM Address 23
Table 492. Pin Definition Table (Continued)
382
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
31.1.1.
Analog Pins
100TQFP
68
69
62
65
144fpBGA
A4
B3
B6
A6
57
B8
59
C8
NA
NA
55
70
71
67
66
D4
C5
A7
C4
C3
B4
A5
PIN
ADCL
ADCR
HPL
HPR
LINE1L
HP_SENSE
LINE1R
HP_COMMON
LINE2L
LINE2R
MIC
REF_RES
REFp
Vag
Vbg
TYPE
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
DESCRIPTION
ADC Left Filter Capacitor
ADC Right Filter Capacitor
Headphone/Line-out Left
Headphone/Line-out Right
Line-in 1 Left
Direct coupled headphone sense
Line-in 1 Right
Direct coupled headphone common amplifier
Line-in 2 Left (AKA: FM-in Left)
Line-in 2 Right (AKA: FM-in Right)
Microphone Input
USB Precision Resistor
ADC Positive Reference Capacitor
Analog Ground Decoupling Capacitor
Bandgap Decoupling Capacitor
Table 493. Analog Pins
31.1.2.
100TQFP
49
50
NA
NA
54
NA
48
47
NA
NA
NA
DCDC Converter Pins
144fpBGA
B12
A12
A3
D6
D8
A11
C12
D12
A10
B10
A9
PIN
DCDC_Batt
DCDC_Gnd
DCDC_mod0
DCDC_mod1
DCDC_mod2
DCDC_VddA
DCDC_VddD
DCDC_VddIO
DCDC2_Batt
DCDC2_Gnd
DCDC2_Vout
TYPE
P
P
P
P
P
P
P
P
P
P
P
DESCRIPTION
DCDC Battery
DCDC Ground
DCDC mode pin 0
DCDC mode pin 1
DCDC mode pin 2
DCDC VddA
DCDC VddD
DCDC VddIO
DCDC2 Battery
DCDC2 Ground
DCDC2 Vout
Table 494. DCDC Converter Pins
31.1.3.
100TQFP
12
13
14
15
16
17
18
19
20
21
22
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
45
External Memory Interface (CompactFlash) Pins
144fpBGA
M5
L5
K6
L6
K7
J7
K8
J8
L9
L10
M10
L12
K11
J9
K9
M9
M8
L8
M7
L7
J6
M6
K5
M1
D10
PIN
CF_A0
CF_A1
CF_A2
CF_A3
CF_A4
CF_A5
CF_A6
CF_A7
CF_A8
CF_A9
CF_A10
CF_A11
CF_A12
CF_A13
CF_A14
CF_A15
CF_A16
CF_A17
CF_A18
CF_A19
CF_A20
CF_A21
CF_A22
CF_A23
CF_BVD1
TYPE
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
I
DESCRIPTION
CompactFlash Address 0
CompactFlash Address 1
CompactFlash Address 2
CompactFlash Address 3
CompactFlash Address 4
CompactFlash Address 5
CompactFlash Address 6
CompactFlash Address 7
CompactFlash Address 8
CompactFlash Address 9
CompactFlash Address 10
CompactFlash Address 11
CompactFlash Address 12
CompactFlash Address 13
CompactFlash Address 14
CompactFlash Address 15
CompactFlash Address 16
CompactFlash Address 17
CompactFlash Address 18
CompactFlash Address 19
CompactFlash Address 20
CompactFlash Address 21
CompactFlash Address 22
CompactFlash Address 23
CompactFlash Bad Voltage Detect
Table 495. External Memory Interface (CompactFlash)
5-35xx-D1-1.08-061705
383
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
100TQFP
40
24
7
30
31
32
33
34
35
36
37
NA
NA
NA
NA
NA
NA
NA
NA
8
9
23
41
44
43
25
26
42
144fpBGA
E12
K10
M4
J10
H9
H12
H10
G10
G9
F11
F10
E9
F12
F9
G12
G11
H11
J11
J12
J5
L4
M11
E10
D11
C11
M12
L11
E11
PIN
CF_CDn
CF_CE0n
CF_CE1n
CF_D0
CF_D1
CF_D2
CF_D3
CF_D4
CF_D5
CF_D6
CF_D7
CF_D8
CF_D9
CF_D10
CF_D11
CF_D12
CF_D13
CF_D14
CF_D15
CF_IORDn
CF_IOWRn
CF_OEn
CF_READY
CF_REGn
CF_RESETn
CF_WAITn
CF_WEn
CF_WPn
TYPE
I
O
O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
O
O
O
I
O
I
I
O
I
DESCRIPTION
CompactFlash Card Detect
CompactFlash Chip Enable 0
CompactFlash Chip Enable 1
CompactFlash Data 0
CompactFlash Data 1
CompactFlash Data 2
CompactFlash Data 3
CompactFlash Data 4
CompactFlash Data 5
CompactFlash Data 6
CompactFlash Data 7
CompactFlash Data 8
CompactFlash Data 9
CompactFlash Data 10
CompactFlash Data 11
CompactFlash Data 12
CompactFlash Data 13
CompactFlash Data 14
CompactFlash Data 15
CompactFlash I/O Read Data Strobe
CompactFlash I/O Write Data Strobe
CompactFlash Output Enable Strobe
CompactFlash Ready
CompactFlash Register Select
CompactFlash Reset
CompactFlash Wait (not)
CompactFlash Write Enable Strobe
CompactFlash Write Protect
Table 495. External Memory Interface (CompactFlash) (Continued)
31.1.4.
100TQFP
21
18
24
17
16
20
30
31
32
33
34
35
36
37
25
23
19
26
27
External Memory Interface (SmartMedia/NAND) Pins
144fpBGA
L10
K8
K10
J7
K7
L9
J10
H9
H12
H10
G10
G9
F11
F10
M12
M11
J8
L11
K12
PIN
SM_ALE
SM_CE0n
SM_CE1n
SM_CE2n
SM_CE3n
SM_CLE
SM_D0
SM_D1
SM_D2
SM_D3
SM_D4
SM_D5
SM_D6
SM_D7
SM_READY
SM_REn
SM_SEn
SM_WEn
SM_WPn
TYPE
O
O
O
O
O
O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I
O
O
O
O
DESCRIPTION
SmartMedia/NAND Address Latch Enable
SmartMedia/NAND Chip Enable 0
SmartMedia/NAND Chip Enable 1
SmartMedia/NAND Chip Enable 2
SmartMedia/NAND Chip Enable 3
SmartMedia/NAND Command Latch Enable
SmartMedia/NAND I/O 0
SmartMedia/NAND I/O 1
SmartMedia/NAND I/O 2
SmartMedia/NAND I/O 3
SmartMedia/NAND I/O 4
SmartMedia/NAND I/O 5
SmartMedia/NAND I/O 6
SmartMedia/NAND I/O 7
SmartMedia/NAND Ready/Busy
SmartMedia/NAND Read Enable Strobe
SmartMedia/NAND Spare Area Enable
SmartMedia/NAND Write Enable Strobe
SmartMedia/NAND Write Protect
Table 496. External Memory Interface (SmartMedia/NAND) Pins
31.1.5.
100TQFP
95
94
93
92
91
90
General Purpose Input/Output Pins
144fpBGA
J1
H2
H4
H3
G3
G4
PIN
GP0
GP1
GP2
GP3
GP4
GP5
TYPE
I/O
I/O
I/O
I/O
I/O
I/O
DESCRIPTION
GP0B0
GP0B1
GP0B2
GP0B3
GP0B4
GP0B5
Table 497. General Purpose Input/Output Pins
384
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
100TQFP
89
88
84
83
85
82
3
2
1
100
4
5
99
98
30
31
32
33
34
35
36
37
12
13
14
15
16
17
18
19
20
21
22
44
7
24
40
41
42
45
43
9
8
23
26
27
25
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
144fpBGA
F3
F4
D1
D2
E1
E3
K4
L2
M2
L1
M3
L3
K2
J4
J10
H9
H12
H10
G10
G9
F11
F10
M5
L5
K6
L6
K7
J7
K8
J8
L9
L10
M10
D11
M4
K10
E12
E10
E11
D10
C11
L4
J5
M11
L11
K12
M12
K9
M9
M8
L8
M7
L7
J6
M6
K5
M1
E9
F12
F9
G12
G11
H11
J11
J12
PIN
GP6
GP7
GP8
GP9
GP10
GP11
GP12
GP13
GP14
GP15
GP16
GP17
GP18
GP19
GP24
GP25
GP26
GP27
GP28
GP29
GP30
GP31
GP32
GP33
GP34
GP35
GP36
GP37
GP38
GP39
GP40
GP41
GP42
GP43
GP44
GP45
GP46
GP47
GP48
GP49
GP50
GP51
GP52
GP53
GP54
GP55
GP56
GP61
GP62
GP63
GP64
GP65
GP66
GP67
GP68
GP69
GP70
GP72
GP73
GP74
GP75
GP76
GP77
GP78
GP79
TYPE
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
DESCRIPTION
GP0B6
GP0B7
GP0B8
GP0B9
GP0B10
GP0B11
GP0B12
GP0B13
GP0B14
GP0B15 (Do not use as GPIO if using SPI)
GP0B16
GP0B17
GP0B18
GP0B19
GP1B0
GP1B1
GP1B2
GP1B3
GP1B4
GP1B5
GP1B6
GP1B7
GP1B8
GP1B9
GP1B10
GP1B11
GP1B12
GP1B13
GP1B14
GP1B15
GP1B16
GP1B17
GP1B18
GP1B19
GP1B20
GP1B21
GP1B22
GP1B23
GP2B0
GP2B1
GP2B2
GP2B3
GP2B4
GP2B5
GP2B6
GP2B7
GP2B8
GP2B13
GP2B14
GP2B15
GP2B16
GP2B17
GP2B18
GP2B19
GP2B20
GP2B21
GP2B22
GP3B0
GP3B1
GP3B2
GP3B3
GP3B4
GP3B5
GP3B6
GP3B7
Table 497. General Purpose Input/Output Pins (Continued)
5-35xx-D1-1.08-061705
385
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
100TQFP
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
144fpBGA
L12
K11
J9
B9
F2
G1
J3
H1
G2
F1
E2
K1
J2
K3
PIN
TYPE
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
GP80
GP81
GP82
GP83
GP84
GP85
GP86
GP87
GP88
GP89
GP90
GP91
GP92
GP93
DESCRIPTION
GP3B8
GP3B9
GP3B10
GP3B11
GP3B12
GP3B13
GP3B14
GP3B15
GP3B16
GP3B17
GP3B18
GP3B19
GP3B20
GP3B21
Table 497. General Purpose Input/Output Pins (Continued)
31.1.6.
100TQFP
4
5
I2C Interface Pins
144fpBGA
M3
L3
PIN
I2C_SCL
I2C_SDA
TYPE
I/O
I/O
DESCRIPTION
I2C Serial Clock
I2C Serial Data
Table 498. I2C Pins
31.1.7.
100TQFP
91
NA
95
NA
94
93
90
89
88
92
NA
I2S Interface Pins
144fpBGA
G3
K3
J1
J2
H2
H4
G4
F3
F4
H3
K1
PIN
I2S_BCLK
I2S_BCLK
I2S_DataI0
I2S_DataI0
I2S_DataI1
I2S_DataI2
I2S_DataO0
I2S_DataO1
I2S_DataO2
I2S_WCLK
I2S_WCLK
TYPE
I
I
I
I
I
I
O
O
O
I
I
DESCRIPTION
I2S Bit Clock
I2S Bit Clock (Alternate pin)
I2S Data In 0
I2S Data In 0 (Alternate pin)
I2S Data In 1
I2S Data In 2
I2S Data Out 0
I2S Data Out 1
I2S Data Out 2
I2S Word Clock
I2S Word Clock (Alternate pin)
Table 499. I2S Pins
31.1.8.
100TQFP
56
73
61
39
11
86
64
29
96
NA
NA
76
72
38
10
87
63
28
97
NA
NA
78
Power Pins
144fpBGA
D7
B2
B7
E8
H7
F6
D5
G8
G5
F7
E7
B1
B5
F8
G6
F5
C6
H8
H5
G7
E6
E5
PIN
BATT
VddA1
VddA2
VddD1
VddD2
VddD3
VddHP
VddIO1
VddIO2
VddIO3
VddIO4
VddPLL
VssA1
VssD1
VssD2
VssD3
VssHP
VssIO1
VssIO2
VssIO3
VssIO4
VssPLL
TYPE
P
P
P
P
P
P
P
P
P
P
P
P
P
P
P
P
P
P
P
P
P
P
DESCRIPTION
Battery Input
Analog Power 1
Analog Power 2
Digital Core Power 1
Digital Core Power 2
Digital Core Power 3
Headphone Power
Digital I/O Power 1
Digital I/O Power 2
Digital I/O Power 3
Digital I/O Power 4
PLL Power
Analog Ground 1
Digital Core Ground 1
Digital Core Ground 2
Digital Core Ground 3
Headphone Ground
Digital I/O Ground 1
Digital I/O Ground 2
Digital I/O Ground 3
Digital I/O Ground 4
PLL Ground
Table 500. Power Pins
386
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
31.1.9.
100TQFP
12
13
14
15
16
17
18
19
20
21
22
NA
NA
NA
NA
NA
NA
NA
NA
NA
30
31
32
33
34
35
36
37
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
SDRAM Interface Pins
144fpBGA
M5
L5
K6
L6
K7
J7
K8
J8
L9
L10
M10
L12
K11
J9
K9
M9
J3
F1
E2
G2
J10
H9
H12
H10
G10
G9
F11
F10
E9
F12
F9
G12
G11
H11
J11
J12
B9
F2
G1
H1
PIN
RAM_A0
RAM_A1
RAM_A2
RAM_A3
RAM_A4
RAM_A5
RAM_A6
RAM_A7
RAM_A8
RAM_A9
RAM_A10
RAM_A11
RAM_BA0
RAM_BA1
RAM_A12
RAM_A13
RAM_CASn
RAM_CKE
RAM_CLK
RAM_CSn
RAM_D0
RAM_D1
RAM_D2
RAM_D3
RAM_D4
RAM_D5
RAM_D6
RAM_D7
RAM_D8
RAM_D9
RAM_D10
RAM_D11
RAM_D12
RAM_D13
RAM_D14
RAM_D15
RAM_DQM0
RAM_DQM1
RAM_RASn
RAM_WEn
TYPE
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
I/O
O
O
O
O
DESCRIPTION
SDRAM Address 0
SDRAM Address 1
SDRAM Address 2
SDRAM Address 3
SDRAM Address 4
SDRAM Address 5
SDRAM Address 6
SDRAM Address 7
SDRAM Address 8
SDRAM Address 9
SDRAM Address 10
SDRAM Address 11
SDRAM Bank Address 0
SDRAM Bank Address 1
SDRAM Address 12
SDRAM Address 13
SDRAM CASn
SDRAM CKE
SDRAM Clock
SDRAM CSn
SDRAM Data 0
SDRAM Data 1
SDRAM Data 2
SDRAM Data 3
SDRAM Data 4
SDRAM Data 5
SDRAM Data 6
SDRAM Data 7
SDRAM Data 8
SDRAM Data 9
SDRAM Data 10
SDRAM Data 11
SDRAM Data 12
SDRAM Data 13
SDRAM Data 14
SDRAM Data 15
SDRAM DQM0
SDRAM DQM1
SDRAM RASn
SDRAM WEn
Table 501. SDRAM Pins
31.1.10. SPI Pins
100TQFP
2
1
3
100
144fpBGA
L2
M2
K4
L1
PIN
SPI_MISO
SPI_MOSI
SPI_SCK
SPI_SSn
TYPE
I/O
I/O
I/O
I
DESCRIPTION
SPI Master Input/Slave Output
SPI Master Output/Slave Input
SPI Serial Clock
SPI Slave Select
Table 502. SPI Pins
31.1.11. System Pins
100TQFP
79
144fpBGA
D3
58
A8
60
C7
53
46
C9
B11
PIN
FRESET
TYPE
A
PSWITCH
LRADC1
MIC_BIAS
LRADC2
MIC_BIAS
ONCE_DRn
ONCE_DSI
A
A
A
A
A
I
I
DESCRIPTION
Fast Reset, Fast Falling Edge (<15ns) Powers
Down Device
Power Switch
Low Resolution ADC Input
microphone bias
Low Resolution ADC Input & Temp sensor bias
alternate mic bias
Debug Reset
Debug Data In
Table 503. System Pins
5-35xx-D1-1.08-061705
387
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
100TQFP
51
52
6
77
75
74
144fpBGA
C10
D9
H6
E4
A1
A2
PIN
ONCE_DSK
ONCE_DSO
TESTMODE
VddXTAL
XTALI
XTALO
TYPE
O
O
I
A
A
A
DESCRIPTION
Debug Clock
Debug Data Out
Test Mode Pin
Power for XTAL oscillator (generated on-chip)
Crystal In
Crystal Out
Table 503. System Pins (Continued)
31.1.12. Timer Pins
100TQFP
99
98
144fpBGA
K2
J4
PIN
TYPE
I/O
I/O
TIO0
TIO1
DESCRIPTION
Timer 0 Pin
Timer 1 Pin
Table 504. Timer Pins
31.1.13. Pulse Width Modulator Pins
100TQFP
99
98
83
88
144fpBGA
K2
J4
D2
F4
PIN
PWM0
PWM1
PWM2
PWM3
TYPE
O
O
O
O
DESCRIPTION
Pulse Width Modulator #0
Pulse Width Modulator #1
Pulse Width Modulator #2
Pulse Width Modulator #3
Table 505. PWM Pins
31.1.14. USB Pins
100TQFP
81
80
144fpBGA
C1
C2
PIN
USB_DM
USB_DP
TYPE
A
A
DESCRIPTION
USB Negative Data Line
USB Positive Data Line
Table 506. USB Pins
TYPE
MODULE
DESCRIPTION
CODEC
Analog pins
DCDC
DCDC Converter pins
EMC-CF
External Memory Interface pins (CompactFlash)
EMC-SM
External Memory Interface pins (SmartMedia/NAND)
GPIO
General Purpose Input/Output pins
I2C pins
I2C
I2S pins
I2S
POWER
Power pins
SDRAM
SDRAM pins
SPI
SPI pins
SYSTEM
System pins
TIMER
Timer pins
USB
USB pins
PWM
Pulse Width Modulator
Almost all digital pins are powered down (i.e. high impedance) at reset, until reprogrammed by the DSP. The only exceptions are:
TESTMODE, ONCE_DSI, ONCE_DSK, ONCE_DSO, ONCE_DRN; these pins are always active.
A
I
I/O
O
P
DESCRIPTION
Analog pin
Input pin
Input/output pin
Output pin
Power pin
Table 507. Notes on Pin Placement and Definitions
388
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
32. PACKAGE DRAWINGS
32.1. 100-Pin TQFP
D
D
1
BODY + 2.00 mm FOOTPRINT
LEADS
100L-1.4 THK.
L/F
DIMS.
.127 .152
TOL.
A
MAX.
1.60
A1
.05 MIN./.15 MAX.
A2
1.40
±.05
16.00
D
±.20
14.00
D1
±.05
E
16.00
±.20
14.00
E1
±.05
.60
L
+.15/-.10
e
.50
BASIC
b
±.05
.22
O
0°-7D
MAX.
.08
ddd
ccc
MAX.
.08
N
1
A
E
B
E
1
D
12° TYP.
A2
A
e
A1
12° TYP.
ANOTHER VARIATION
OF PIN 1 VISUAL AID
N
.20 RAD. TYP.
NOTES: 1) ALL DIMENSIONS IN MM.
1
2) DIMENSIONS SHOWN ARE NOMINAL WITH TOL.
AS INDICATED.
3) L/F: EFTEC 64T COPPER OR EQUIVALENT, 0.127 MM
(.005") OR 0.15 MM (.006") THICK.
4) FOOT LENGTH "L" IS MEASURED AT GAGE PLANE,
AT 0.25 ABOVE THE SEATING PLANE.
.20 RAD. TYP.
6°P4°
A
STANDOFF
A
1
.25
.17 MAX.
O
L
b
ddd M C A-B S D S
SEATING
PLANE
C
LEAD COPLANARITY
100 Pin 14x14 TQFP
ccc C
Figure 134. 100-Pin TQFP Package Drawing
5-35xx-D1-1.08-061705
389
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
32.2. 144-Pin fpBGA
DEATIL B
0.10
-A-
D
12
11
10
9
8
6
7
5
4
3
2
DIMENSIONAL REFERENCES
1
-B-
A
B
C
D
E
F
G
H
J
K
L
M
(e)
E2
E
(E1)
REF.
MIN.
NOR.
MAX.
A
1.25
1.35
1.45
A1
0.25
0.30
0.35
A2
0.65
0.70
0.75
10.00
10.20
D
(D1)
D2
9.80
10.00
10.20
E
9.80
10.00
10.20
8.80 BSC
E1
BOTTOM VIEW
TOP VIEW
8.80 BSC.
D1
(e)
D2
9.80
E2
9.80
10.00
10.20
b
0.35
0.40
0.45
0.35
c
f
DETAIL A
NX b
0.15 s
f
0.075 s
A s
0.15
0.20
B s
C
4
e
SIDE VIEW
C
aaa
bbb
0.25
ccc
e
0.725
f
0.50
0.80
0.875
0.60
0.70
M
12
N
144
NOTES:
DETAIL B
1. ALL DIMENSIONS ARE IN MILLIMETERS.
2. 'e' REPRESENTS THE BASIC SOLDER BALL GRID PITCH.
ccc
c
bbb C
-CA
A2
DETAIL A
C
AND SYMBOL 'N' IS THE MAXIMUM ALLOWABLE NUMBER OF
BALLS AFTER DEPOPULATING.
4. 'b' IS MEASURABLE AT THE MAXIMUM SOLDER BALL DIAMETER
PARALLEL TO PRIMARY DAIUM -C- .
6
SEATING PLANE
5. DIMENSION 'aaa' IS MEASURED PARALLEL TO PRIMARY DATUM -C- .
6. PRIMARY DATUM -C- AND SEATING PLANE ARE DEFINED BY THE SPHERICAL
CROWNS OF THE SOLDER BALLS.
A1
aaa C
3. 'M' REPRESENTS THE BASIC SOLDER BALL MATRIX SIZE.
7. PACKAGE SURFACE SHALL BE MATTE FINISH CHARMILLES 24 TO 27.
5
8. PACKAGE CENTERING TO SUBSTRATE SHALL BE 0.0760 MM MAXIMUM
FOR BOTH X AND Y DIRECTION RESPECTIVELY.
9. PACKAGE WARP SHALL BE 0.050MM MAXIMUM.
144 fpBGA (10 x 10 mm)
10. SUBSTRATE MATERIAL BASE IS BT RESIN.
11. THE OVERALLPACKAGE THICKNESS "A" ALREADY CONSIDERS COLLAPSE BALLS
Figure 135. 144-Pin fpBGA Package Drawing
390
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
33. STMP35XX FAMILY MEMBER PART NUMBERS & ORDERING INFORMATION
The STMP35xx family comprises a large set of parts targeted at specific applications and customers. The following table summarizes the family members and provides part numbers for order placement.
Customers with prepaid royalties or other royalty arrangements for certain
intellectual property items can order parts without the corresponding royalty
fees included in the purchase price. At the moment, the only royalty options
are for certain MP3 items, see www.mp3licensing.com. The letter N at the
correct location in the part number signifies a part that does not have the
royalty payment included in the purchase price.
PART NUMBER
ROYALTY
STMP3502XXLANA6N No MP3 decode royalty
STMP3502XXBANA6N
PACKAGE
100 pin QFP
144 pin BGA
STMP3503XXLANA6N No MP3 decode royalty
STMP3503XXBANA6N
100 pin QFP
144 pin BGA
STMP3505XXBANA6M
STMP3505XXBANA6N
STMP3505XXLANA6M
STMP3505XXLANA6N
STMP3506XXBANA6N
STMP3506XXLANA6N
MP3 decode royalty
No MP3 decode royalty
MP3 decode royalty
No MP3 decode royalty
No MP3 decode royalty
No MP3 decode royalty
144 pin BGA
STMP3510XXBANA6M
STMP3510XXBANA6N
STMP3510XXLANA6M
STMP3510XXLANA6N
STMP3520XXBANA6N
STMP3520XXLANA6N
MP3 decode royalty
No MP3 decode royalty
MP3 decode royalty
No MP3 decode royalty
No MP3 decode royalty
No MP3 decode royalty
144 pin BGA
STMP3550XXBANA6M
STMP3550XXBANA6N
STMP3550XXLANA6M
STMP3560XXLANA6N
STMP3560XXBANA6N
STMP3560XXLANA6N
MP3 decode royalty
No MP3 decode royalty
MP3 decode royalty
No MP3 decode royalty
No MP3 decode royalty
No MP3 decode royalty
144 pin BGA
100 pin QFP
144 pin BGA
100 pin QFP
100 pin QFP
144 pin BGA
100 pin QFP
100 pin QFP
144 pin BGA
100 pin QFP
DESCRIPTION
Full Speed USB2.0
No Battery Charging
No MP3 encode
No Digital Rights
Management/RTC
No SDRAM Controller
Full Speed USB 2.0
No Battery Charging
No MP3 encode
PD-DRM only
No SDRAM Controller
Full Speed USB 2.0
No Battery Charging
No MP3 encode
No SDRAM Controller
Full Speed USB 2.0
No Battery Charging
No SDRAM Controller
High Speed USB 2.0
No Battery Charging
No MP3 encode
No SDRAM Controller
High Speed USB 2.0
No Battery Charging
No SDRAM Controller
High Speed USB 2.0
Battery Charging
No MP3 Encode
SDRAM Controller
High Speed USB 2.0
Battery Charging
SDRAM Controller
Table 508. STMP35xx Family Members
In addition, SigmaTel also offers fully RoHS and WEEE compliant lead (Pb) free
environmental packaging. All of the products listed above are also available in the
environmental packaging. The part numbers for these packages are listed below.
5-35xx-D1-1.08-061705
391
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
Part Number
STMP3502XXLAEA6N
Royalty
No MP3 decode royalty
STMP3502XXBAEA6N
STMP3503XXLAEA6N
No MP3 decode royalty
STMP3503XXBAEA6N
STMP3505XXBAEA6M MP3 decode royalty
STMP3505XXBAEA6N No MP3 decode royalty
STMP3505XXLAEA6M MP3 decode royalty
STMP3505XXLAEA6N No MP3 decode royalty
STMP3506XXBAEA6N No MP3 decode royalty
STMP3506XXLAEA6N
No MP3 decode royalty
STMP3510XXBAEA6M MP3 decode royalty
STMP3510XXBAEA6N No MP3 decode royalty
STMP3510XXLAEA6M MP3 decode royalty
STMP3510XXLAEA6N No MP3 decode royalty
STMP3520XXBAEA6N No MP3 decode royalty
STMP3520XXLAEA6N
No MP3 decode royalty
STMP3550XXBAEA6M MP3 decode royalty
STMP3550XXBAEA6N No MP3 decode royalty
STMP3550XXLAEA6M MP3 decode royalty
STMP3560XXLAEA6N No MP3 decode royalty
STMP3560XXBAEA6N No MP3 decode royalty
Package
100 pin
Environmental
QFP
144 pin
Environmental
BGA
100 pin
Environmental
QFP
144 pin
Environmental
BGA
144 pin
Environmental
BGA
100 pin
Environmental
QFP
144 pin
Environmental
BGA
100 pin
Environmental
QFP
144 pin
Environmental
BGA
100 pin
Environmental
QFP
144 pin
Environmental
BGA
100 pin
Environmental
QFP
144 pin
Environmental
BGA
100 pin
Environmental
QFP
144 pin
Environmental
BGA
Description
Full Speed USB 2.0
No Battery Charging
No MP3 encode
No Digital Rights
Mangement/RTC
No SDRAM Controller
Full Speed USB 2.0
No Battery Charging
No MP3 encode
PD-DRM only
No SDRAM Controller
Full Speed USB 2.0
No Battery Charging
No MP3 encode
No SDRAM Controller
Full Speed USB 2.0
No Battery Charging
No SDRAM Controller
High Speed USB 2.0
No Battery Charging
No MP3 encode
No SDRAM Controller
High Speed USB 2.0
No Battery Charging
No SDRAM Controller
High Speed USB 2.0
Battery Charging
No MP3 encode
SDRAM Controller
High Speed USB 2.0
Battery Charging
SDRAM Controller
Table 509. STMP35xx Family Members (Continued)
392
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
34. INDEX OF REGISTERS
HW_ADCBAR
HW_ADCCPR
HW_ADCCSR
HW_ADCICR
HW_ADCMR
HW_ADCSRR
HW_ADCWCR
HW_BATT_CTRL
HW_BATT_RESULT
HW_BATT_THRSH
HW_CCR
HW_CYCSTLCNT
HW_DACBAR
HW_DACCPR
HW_DACCSR
HW_DACICR
HW_DACMR
HW_DACSRR
HW_DACWCR
HW_DCDC_PERSIST
HW_DCDC_VDDA
HW_DCDC_VDDD
HW_DCDC_VDDIO
HW_DCDC1_CTRL0
HW_DCDC1_CTRL1
HW_DCDC2_CTRL0
HW_DCDC2_CTRL1
HW_DCDCTBR
HW_DCLKCNTL
HW_DCLKCNTU
HW_ECC_BLKSTRTADDR
HW_ECC_BLKSTRTINDEX
HW_ECC_CSR0
HW_ECC_CSR1
HW_ECC_ERRVAL
HW_ECC_LOCADDR
HW_ECC_LOCINDEX
HW_ECC_PARSTRTADDR
HW_ECC_PARSTRTINDEX
HW_ECC_RSCFG
HW_ECC_SSFDCCFG
HW_ESPI_ADDR
HW_ESPI_CLKCNTRL
HW_ESPI_CNFG
HW_ESPI_CSR
HW_ESPI_INDEX
HW_ESPI_PIODATA
HW_FILCO_ CTXT_S0
HW_FILCO_ CTXT_S1
HW_FILCO_ CTXT_S2
HW_FILCO_ CTXT_S3
HW_FILCO_ CTXT_S4
HW_FILCO_ CTXT_S5
HW_FILCO_ CTXT_SW
HW_FILCO_CBAR
HW_FILCO_CCPR
HW_FILCO_CSR
HW_FILCO_CTXT_A0R
HW_FILCO_CTXT_A1R
HW_FILCO_CTXT_A2R
5-35xx-D1-1.08-061705
X:$FB05 ................................................................................288
X:$FB03 ................................................................................289
X:$FB00 ................................................................................296
X:$FB06 ................................................................................290
X:$FB04 ................................................................................288
X:$FB01 ................................................................................290
X:$FB02 ................................................................................290
X:$FA20 ................................................................................319
X:$FA22 ................................................................................321
X:$FA21 ................................................................................321
X:$FA00 ..................................................................................43
X:$FFEC .................................................................................42
X:$F805 ................................................................................278
X:$F803 ................................................................................279
X:$F800 ................................................................................284
X:$F806 ................................................................................280
X:$F804 ................................................................................278
X:$F801 ................................................................................280
X:$F802 ................................................................................279
X:$FA1B ................................................................................365
X:$FA10 ................................................................................360
X:$FA0F ................................................................................359
X:$FA0E ................................................................................358
X:$FA0C ...............................................................................353
X:$FA0D ...............................................................................355
X:$FA11 ................................................................................356
X:$FA12 ................................................................................357
X:$FA14 ................................................................................363
X:$FFEA .................................................................................41
X:$FFEB .................................................................................41
X:$F784 ................................................................................148
X:$F785 ................................................................................149
X:$F780 ................................................................................145
X:$F781 ................................................................................147
X:$F78A ................................................................................152
X:$F788 ................................................................................150
X:$F789 ................................................................................151
X:$F786 ................................................................................149
X:$F787 ................................................................................150
X:$F782 ................................................................................146
X:$F783 ................................................................................148
X:$FF04 ................................................................................215
X:$FF02 ................................................................................214
X:$FF01 ................................................................................213
X:$FF00 ................................................................................212
X:$FF05 ................................................................................215
X:$FF03 ................................................................................214
X:$FC25 ................................................................................181
X:$FC26 ................................................................................181
X:$FC27 ................................................................................182
X:$FC28 ................................................................................182
X:$FC29 ................................................................................183
X:$FC2A ...............................................................................183
X:$FC24 ................................................................................180
X:$FC02 ................................................................................168
X:$FC04 ................................................................................169
X:$FC00 ................................................................................167
X:$FC20 ................................................................................179
X:$FC21 ................................................................................179
X:$FC22 ................................................................................180
393
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
HW_FILCO_CTXT_A3R
HW_FILCO_CTXT_WP
HW_FILCO_DBAR
HW_FILCO_DCPR
HW_FILCO_DMR
HW_FILCO_IIRGAIN
HW_FILCO_INTR
HW_FILCO_KICK
HW_FILCO_SAT
HW_FILCO_SBAR
HW_FILCO_SCPR
HW_FILCO_SMR
HW_FILCO_SPARE
HW_FILCO_TAPCNT
HW_FILCO_TPTR
HW_FILCO_UGAIN
HW_FILCO_WORDCOUNT
HW_FILCO_ZC_GAIN
HW_FILCO_ZC_STATUS
HW_FLCFCR
HW_FLCFTMR1R
HW_FLCFTMR2R
HW_FLCR
HW_FLCR2
HW_FLSAHR
HW_FLSALR
HW_FLSMCR
HW_FLSMTMR1R
HW_FLSMTMR2R
HW_GDBR
HW_GP08MA
HW_GP0DIR
HW_GP0DOER
HW_GP0DOR
HW_GP0ENR
HW_GP0IENR
HW_GP0ILVLR
HW_GP0IPENR
HW_GP0IPOLR
HW_GP0ISTATR
HW_GP0PWR
HW_GP18MA
HW_GP1DIR
HW_GP1DOER
HW_GP1DOR
HW_GP1ENR
HW_GP1IENR
HW_GP1ILVLR
HW_GP1IPENR
HW_GP1IPOLR
HW_GP1ISTATR
HW_GP1PWR
HW_GP28MA
HW_GP2DIR
HW_GP2DOER
HW_GP2DOR
HW_GP2ENR
HW_GP2IENR
HW_GP2ILVLR
HW_GP2IPENR
HW_GP2IPOLR
HW_GP2ISTATR
HW_GP2PWR
394
X:$FC23 ................................................................................180
X:$FC2B ...............................................................................183
X:$FC0D ...............................................................................171
X:$FC0F ................................................................................173
X:$FC0E ...............................................................................172
X:$FC18 ................................................................................176
X:$FC0B ...............................................................................170
X:$FC0C ...............................................................................171
X:$FC01 ................................................................................168
X:$FC10 ................................................................................173
X:$FC12 ................................................................................174
X:$FC11 ................................................................................174
X:$FC05 ................................................................................170
X:$FC03 ................................................................................169
X:$FC13 ................................................................................174
X:$FC17 ................................................................................175
X:$FC1B ...............................................................................176
X:$FC1D ...............................................................................178
X:$FC1C ...............................................................................177
X:$F008 ................................................................................115
X:$F009 ................................................................................116
X:$F00A ................................................................................116
X:$F000 ................................................................................105
X:$F004 ................................................................................106
X:$F002 ................................................................................107
X:$F001 ................................................................................106
X:$F010 ................................................................................112
X:$F011 ................................................................................113
X:$F012 ................................................................................113
X:$FFFC .................................................................................28
X:$F40A ................................................................................272
X:$F402 ................................................................................268
X:$F403 ................................................................................268
X:$F401 ................................................................................268
X:$F400 ................................................................................268
X:$F405 ................................................................................269
X:$F406 ................................................................................269
X:$F404 ................................................................................269
X:$F407 ................................................................................270
X:$F408 ................................................................................270
X:$F409 ................................................................................270
X:$F41A ................................................................................272
X:$F412 ................................................................................268
X:$F413 ................................................................................268
X:$F411 ................................................................................268
X:$F410 ................................................................................268
X:$F415 ................................................................................269
X:$F416 ................................................................................269
X:$F414 ................................................................................269
X:$F417 ................................................................................270
X:$F418 ................................................................................270
X:$F419 ................................................................................270
X:$F42A ................................................................................272
X:$F422 ................................................................................268
X:$F423 ................................................................................268
X:$F421 ................................................................................268
X:$F420 ................................................................................268
X:$F425 ................................................................................269
X:$F426 ................................................................................269
X:$F424 ................................................................................269
X:$F427 ................................................................................270
X:$F428 ................................................................................270
X:$F429 ................................................................................270
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
HW_GP38MA
HW_GP3DIR
HW_GP3DOER
HW_GP3DOR
HW_GP3ENR
HW_GP3IENR
HW_GP3ILVLR
HW_GP3IPENR
HW_GP3IPOLR
HW_GP3ISTATR
HW_GP3PWR
HW_GPFLASH_CSR0R
HW_GPFLASH_CSR1
HW_GPFLASH_DEBUG
HW_GPFLASH_DMA_ADDR
HW_GPFLASH_TIMING1
HW_GPFLASH_TIMING2
HW_GPFLASH_TIMINGBUSY
HW_GPFLASH_TWTOCNT
HW_GPFLASH_XFER_SIZE
HW_HPCTRL
HW_I2CCSR
HW_I2CDAT
HW_I2CDIV
HW_ICLENABLE0R
HW_ICLENABLE1R
HW_ICLFENABLE0R
HW_ICLFENABLE1R
HW_ICLFORCE0R
HW_ICLFORCE1R
HW_ICLOBSVZ0R
HW_ICLOBSVZ1R
HW_ICLPRIOR0R
HW_ICLPRIOR1R
HW_ICLPRIOR2R
HW_ICLPRIOR3R
HW_ICLPRIOR4R
HW_ICLSTATUS0R
HW_ICLSTATUS1R
HW_ICLSTEER0R
HW_ICLSTEER1R
HW_ICLSTEER2R
HW_IPR
HW_LRADC1_CTRL
HW_LRADC1_RESULT
HW_LRADC1_THRSH
HW_LRADC2_CTRL
HW_LRADC2_RESULT
HW_LRADC2_THRSH
HW_MIX_TEST
HW_MIXADCGAINR
HW_MIXDACINVR
HW_MIXLINE1INVR
HW_MIXLINE2INVR
HW_MIXMASTERVR
HW_MIXMICINVR
HW_MIXPWRDNR
HW_MIXRECSELR
HW_MIXTBR
HW_OMR
HW_PIN_CTRL
HW_PWM_CH0AR
5-35xx-D1-1.08-061705
X:$F43A ................................................................................272
X:$F432 ................................................................................268
X:$F433 ................................................................................268
X:$F431 ................................................................................268
X:$F430 ................................................................................268
X:$F435 ................................................................................269
X:$F436 ................................................................................269
X:$F434 ................................................................................269
X:$F437 ................................................................................270
X:$F438 ................................................................................270
X:$F439 ................................................................................271
X:$F0C0 ................................................................................126
X:$F0C1 ................................................................................128
X:$F0C7 ................................................................................133
X:$F0C2 ................................................................................129
X:$F0C4 ................................................................................130
X:$F0C5 ................................................................................131
X:$F0C6 ................................................................................132
X:$F0C8 ................................................................................133
X:$F0C3 ................................................................................130
X:$FA15 ................................................................................310
X:$FFE7 ................................................................................197
X:$FFE6 ................................................................................200
X:$FFE5 ................................................................................201
X:$F300 ..................................................................................53
X:$F301 ..................................................................................53
X:$F30D ..................................................................................58
X:$F30E ..................................................................................58
X:$F30B ..................................................................................58
X:$F30C ..................................................................................58
X:$F30F ..................................................................................59
X:$F310 ..................................................................................59
X:$F304 ..................................................................................54
X:$F305 ..................................................................................54
X:$F306 ..................................................................................54
X:$F307 ..................................................................................55
X:$F311 ..................................................................................55
X:$F302 ..................................................................................53
X:$F303 ..................................................................................53
X:$F308 ..................................................................................55
X:$F309 ..................................................................................57
X:$F30A ..................................................................................57
X:$FFFF ..................................................................................49
X:$FA23 ................................................................................323
X:$FA25 ................................................................................325
X:$FA24 ................................................................................324
X:$FA26 ................................................................................326
X:$FA28 ................................................................................328
X:$FA27 ................................................................................327
X:$FA1C ...............................................................................308
X:$FA0A ................................................................................303
X:$FA08 ................................................................................302
X:$FA06 ................................................................................301
X:$FA07 ................................................................................302
X:$FA04 ................................................................................298
X:$FA05 ................................................................................299
X:$FA0B ................................................................................304
X:$FA09 ................................................................................302
X:$FA03 ................................................................................305
SPECIAL .................................................................................27
X:$FA30 ..................................................................................47
X:$FA32 ................................................................................189
395
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
HW_PWM_CH0BR
HW_PWM_CH1AR
HW_PWM_CH1BR
HW_PWM_CH2AR
HW_PWM_CH2BR
HW_PWM_CH3AR
HW_PWM_CH3BR
HW_PWM_CSR
HW_PXCFG
HW_PXRAM0_RPR
HW_PXRAM1_RPR
HW_PXRAM2_RPR
HW_PYCFG
HW_PYRAM0_RPR
HW_PYRAM1_RPR
HW_PYRAM2_RPR
HW_RAM_ROM_CFG
HW_RCR
HW_REF_CTRL
HW_REVR
HW_RTC_ALARM0
HW_RTC_ALARM1
HW_RTC_CSR
HW_RTC_DIVIDE
HW_RTC_MSECONDS0
HW_RTC_MSECONDS1
HW_RTC_PERSISTENT0
HW_RTC_PERSISTENT1
HW_RTC_WATCHDOG
HW_SAIR_CSR
HW_SAIRX0R
HW_SAIRX1R
HW_SAIRX2R
HW_SAITCSR
HW_SAITX0R
HW_SAITX1R
HW_SAITX2R
HW_SDRAM_ADDR1
HW_SDRAM_ADDR2
HW_SDRAM_BAR
HW_SDRAM_CNT
HW_SDRAM_CSR
HW_SDRAM_MODE
HW_SDRAM_MR
HW_SDRAM_SIZE
HW_SDRAM_SYSADDR
HW_SDRAM_TIMER1
HW_SDRAM_TIMER2
HW_SDRAM_TYPE
HW_SISPEED
HW_SPARER
HW_SPCSR
HW_SPDR
HW_SWIZZLE_BARRELR
HW_SWIZZLE_BIGENDIANR
HW_SWIZZLE_BITREVR
HW_SWIZZLE_CS1R
HW_SWIZZLE_CS2R
HW_SWIZZLE_DATA1R
HW_SWIZZLE_DATA2R
HW_SWIZZLE_DESTADDRR
HW_SWIZZLE_DIV3L
HW_SWIZZLE_DIV3U
396
X:$FA33 ................................................................................190
X:$FA34 ................................................................................190
X:$FA35 ................................................................................192
X:$FA36 ................................................................................192
X:$FA37 ................................................................................193
X:$FA38 ................................................................................194
X:$FA39 ................................................................................194
X:$FA31 ................................................................................188
X:$FFE8 ..................................................................................30
X:$F5A3 ..................................................................................36
X:$F5A4 ..................................................................................36
X:$F5A5 ..................................................................................36
X:$FFE9 ..................................................................................31
X:$F5A6 .................................................................................. 37
X:$F5A7 .................................................................................. 37
X:$F5A8 .................................................................................. 37
X:$FFED .................................................................................35
X:$FA01 ..................................................................................39
X:$FA19 ................................................................................307
X:$FA02 ..................................................................................38
X:$F504 ................................................................................255
X:$F505 ................................................................................255
X:$F500 ................................................................................251
X:$F506 ................................................................................254
X:$F501 ................................................................................253
X:$F502 ................................................................................253
X:$F507 ................................................................................256
X:$F508 ................................................................................258
X:$F503 ................................................................................254
X:$FFF0 ................................................................................260
X:$FFF1 ................................................................................261
X:$FFF2 ................................................................................262
X:$FFF3 ................................................................................262
X:$FFF5 ................................................................................263
X:$FFF6 ................................................................................264
X:$FFF7 ................................................................................264
X:$FFF8 ................................................................................264
X:$F901 ................................................................................234
X:$F902 ................................................................................234
X:$F907 ................................................................................237
X:$F90D ................................................................................238
X:$F900 ................................................................................230
X:$F90E ................................................................................238
X:$F908 ................................................................................237
X:$F904 ................................................................................235
X:$F903 ................................................................................235
X:$F905 ................................................................................235
X:$F906 ................................................................................236
X:$F90F ................................................................................232
X:$FA13 ................................................................................361
X:$FA16 ..................................................................................46
X:$FFF9 ................................................................................217
X:$FFFA ................................................................................218
X:$F38F ................................................................................246
X:$F387 ................................................................................244
X:$F388 ................................................................................244
X:$F380 ................................................................................240
X:$F381 ................................................................................241
X:$F384 ................................................................................243
X:$F385 ................................................................................244
X:$F386 ................................................................................244
X:$F390 ................................................................................246
X:$F391 ................................................................................246
5-35xx-D1-1.08-061705
OFF ICIAL
PRODUCT
DO CUMENTATION
6/17/05
STMP35xx
D-Major™Audio System on Chip
HW_SWIZZLE_PASSISBR
HW_SWIZZLE_PASSISWR
HW_SWIZZLE_PASSLSBR
HW_SWIZZLE_PASSLSWR
HW_SWIZZLE_PASSMSBR
HW_SWIZZLE_PASSMSWR
HW_SWIZZLE_SIZER
HW_SWIZZLE_SOURCER
HW_TMR0CNTR
HW_TMR0CSR
HW_TMR1CNTR
HW_TMR1CSR
HW_TMR2CNTR
HW_TMR2CSR
HW_TMR3CNTR
HW_TMR3CSR
HW_USBARCACCESS
HW_USBARCDATAHIGH
HW_USBARCDATALOW
HW_USBARCUNUSED
HW_USBCSR
HW_USBDMAOFF
HW_USBLASERFUSE
HW_USBPHYPLL
HW_USBPHYPWD
HW_USBPHYRX
HW_USBPHYTX
HW_USBREADTEST
HW_USBSTATEMACHINES
HW_USBUTCSR
HW_USBUTMI1
HW_USBUTMI2
HW_USBUTMISENSE
HW_VDD5V_PWR_CHARGE
X:$F38A ................................................................................245
X:$F38D ................................................................................245
X:$F389 ................................................................................244
X:$F38C ................................................................................245
X:$F38B ................................................................................245
X:$F38E ................................................................................245
X:$F382 ................................................................................243
X:$F383 ................................................................................243
X:$F101 ................................................................................223
X:$F100 ................................................................................222
X:$F141 ................................................................................223
X:$F140 ................................................................................222
X:$F181 ................................................................................223
X:$F180 ................................................................................222
X:$F1C1 ................................................................................223
X:$F1C0 ................................................................................222
X:$F202 ..................................................................................71
X:$F204 ..................................................................................72
X:$F203 ..................................................................................72
X:$F20B ..................................................................................76
X:$F200 ..................................................................................69
X:$F201 ..................................................................................70
X:$F20C ..................................................................................77
X:$F212 ..................................................................................97
X:$F210 ..................................................................................93
X:$F213 ..................................................................................99
X:$F211 ..................................................................................95
X:$F209 ..................................................................................75
X:$F20A ..................................................................................76
X:$F205 ..................................................................................73
X:$F206 ..................................................................................73
X:$F207 ..................................................................................74
X:$F208 ..................................................................................75
X:$FA1D ..............................................................................367
MPEG Layer 3 audio coding technology from Fraunhofer IIS and THOMSON Multimedia.
5-35xx-D1-1.08-061705
397