NSC LM9823CCWM

N
LM9823 3 Channel 48-Bit Color Scanner Analog Front End
General Description
Key Specifications
The LM9823 is a high performance Analog Front End (AFE) for
image sensor processing systems. It performs all the analog and
mixed signal functions (correlated double sampling, color specific gain and offset correction, and analog to digital conversion)
necessary to digitize the output of a wide variety of CIS and
CCD sensors. The LM9823 has a 16 bit 6MHz ADC.
•
•
•
•
•
Features
Applications
•
•
•
•
•
•
•
•
•
•
•
6 million pixels/s conversion rate
Digitally programmed gain and offset for red, green and blue
color balancing
Correlated Double Sampling for lowest noise from CCD
sensors
Compatible with CCD and CIS type image sensors
Internal Voltage Reference Generation
TTL/CMOS compatible input/output
Output Data Resolution
Pixel Conversion Rate
Analog Supply Voltage
I/O Supply Voltage
Power Dissipation (typical)
16 Bits
6MHz
5V±5%
3.3V±10% or 5V±5%
375mW
Color Flatbed Document Scanners
Color Sheetfed Scanners
Multifunction Imaging Products
Digital Copiers
General Purpose Linear Array Imaging
Connection Diagram
VBANDGAP
VREFMID
VA
AGND
OS R
VREF+
OSG
VREFOSB
VA
AGND
SEN
SDI
SDO
1
2
3
4
5
6
7
8
9
10
11
12
13
14
LM9823
28 pin
SOIC
28
27
26
25
24
23
22
21
20
19
18
17
16
15
D7
D6
D5
D4
D3
D2
D1
D0
VD
DGND
CLMP
VSMP
MCLK
SCLK
Ordering Information
Temperature Range
0°C ≤ TA ≤ +70°C
Order Number
Device Marking
LM9823CCWM1
LM9823CCWMX2
LM9823CCWM
LM9823CCWM
NS Package
Number
M28B
M28B
Notes: 1 - Rail transport media, 26 parts per rail, 2 - Tape and reel transport media, 1000 parts per reel
©2000 National Semiconductor Corporation
www.national.com
LM9823 3 Channel 48-Bit Color Scanner Analog Front End
October 2000
2
SEN
SDO
SDI
SCLK
BLUE OS
from CCD
GREEN OS
from CCD
RED OS
from CCD
OSB
OSG
OSR
Serial
Interface
VCLAMP
VA
CDS
CDS
CDS
6
Offset
DACB
+
+
Offset
DACG
+
+
Offset
DACR
+
+
Static
Offset
DACs
Timing and Control
1
x1or x3
x1or x3
x1or x3
Gain
Boost
MCLK CLMP VSMP
AGND
x0.93
to x3
x0.93
to x3
x0.93
to x3
5
Coarse Color
Balance PGAs
VA
Internal
Bandgap
Reference
16-Bit
ADC
AGND
16
VD
16 bits
to 8 bit
Bytes
DGND
VBANDGAP
VREF-
VREFMID
VREF+
D7 - D0
LM9823
Block Diagram
www.national.com
+
Positive Supply Voltage (V =VA=VD)
With Respect to GND=AGND=DGND
Voltage On Any Input or Output Pin
Input Current at any pin (Note 3)
Package Input Current (Note 3)
Package Dissipation at TA = 25°C
ESD Susceptibility (Note 5)
Human Body Model
Machine Model
Soldering Information
Infrared, 10 seconds (Note 6)
Storage Temperature
6.5V
-0.3V to V +0.3V
±25mA
±50mA
(Note 4)
+
7000V
450V
Operating Ratings (Notes 1& 2)
Operating Temperature Range
LM9823CCWM
VA Supply Voltage
VD Supply Voltage
VD-VA
OSR, OSG, OSB
Input Voltage Range
SCLK, SDI, SEN, MCLK, VSMP, CLMP
Input Voltage Range
TMIN≤TA≤TMAX
0°C≤TA≤+70°C
+4.75V to +5.25V
+3.0V to +5.25V
≤ 100mV
-0.05V to VA + 0.05V
-0.05V to VD + 0.05V
235°C
-65°C to +150°C
Electrical Characteristics
The following specifications apply for AGND=DGND=0V, VA=+5.0VDC, VD=+3.0 or +5.0VDC, f MCLK=12MHz. Boldface limits apply
for TA=TJ=TMIN to TMAX; all other limits TA=TJ=25°C. (Notes 7, 8, 12 & 16)
Symbol
Parameter
Conditions
Typical
(Note 9)
Limits
(Note 10)
Units
(Limits)
CCD/CIS Source Requirements for Full Specified Accuracy and Dynamic Range (Note 12)
VOS PEAK
Sensor’s Maximum Peak Differential
Signal Range
Gain = 0.933
Gain = 3.0
Gain = 9.0
2.1
0.65
0.21
V
V
V
Full Channel Linearity (in units of 12 bit LSBs) (Note 14)
DNL
Differential Non-Linearity
+0.9
-0.4
+2
-0.9
LSBs (max)
INL
Integral Non-Linearity Error (Note 11)
±2.2
+5
-7
LSBs (max)
Analog Input Characteristics
OSR, OSG, OSB Input Capacitance
OSR, OSG, OSB Input Leakage Current
5
Measured with OS = 3.5VDC
CDS disabled
20
CDS enabled
10
pF
25
µA (max)
nA
Coarse Color Balance PGA Characteristics
Monotonicity
5
bits (min)
V/V (min)
V/V (max)
G0 (Minimum PGA Gain)
PGA Setting = 0
0.93
.90
.96
G31 (Maximum PGA Gain)
PGA Setting = 31
3.0
2.95
3.07
V/V (min)
V/V (max)
x3 Boost Gain
x3 Boost Setting On
(Bit 5 of Gain Register is set)
3.0
2.86
3.08
V/V (min)
V/V (max)
±0.3
1.6
% (max)
Gain Error at any gain (Note 13)
3
www.national.com
LM9823
Absolute Maximum Ratings (Notes 1& 2)
LM9823
Electrical Characteristics (Continued)
The following specifications apply for AGND=DGND=0V, VA=+5.0VDC, VD=+3.0 or +5.0VDC, f MCLK=12MHz. Boldface limits apply
for TA=TJ=TMIN to TMAX; all other limits TA=TJ=25°C. (Notes 7, 8, 12 & 16)
Symbol
Parameter
Conditions
Typical
(Note 9)
Limits
(Note 10)
Units
(Limits)
Internal Reference Characteristics
VREFMID
Mid Reference Output Voltage
2.5
V
VREF+ OUT
Positive Reference Output Voltage
3.5
V
VREF- OUT
Negative Reference Output Voltage
1.5
V
Differential Reference Voltage
VREF+ OUT - VREF- OUT
2.0
V
∆VREF
Static Offset DAC Characteristics (in units of 12 bit LSBs)
Monotonicity
6
bits (min)
Offset DAC LSB size
PGA gain = 1
18.9
13
24
LSB (min)
LSB (max)
Offset DAC Adjustment Range
PGA gain = 1
±585
±570
LSB (min)
2107
1934
2281
LSB (min)
LSB (max)
System Characteristics (in units of 12 bit LSBs) (see section 5.1, Internal Offsets)
C
Analog Channel Gain Constant
(ADC Codes/V)
Includes voltage reference
variation, gain setting = 1
VOS1
Pre-Boost Analog Channel Offset Error,
CCD Mode
17.3
-61
+94
LSB (min)
LSB (max)
VOS1
Pre-Boost Analog Channel Offset Error,
CIS Mode
27
-49
+103
LSB (min)
LSB (max)
VOS2
Pre-PGA Analog Channel Offset Error
-40
-124
+44
LSB (min)
LSB (max)
VOS3
Post-PGA Analog Channel Offset Error
-38
-130
+55
LSB (min)
LSB (max)
DC and Logic Electrical Characteristics
The following specifications apply for AGND=DGND=0V, VA=+5.0VDC, VD=+3.0 or +5.0VDC, fMCLK=12MHz. Boldface limits apply
for TA=TJ=TMIN to TMAX; all other limits TA=TJ=25°C. (Notes 7 & 8)
Symbol
Parameter
Conditions
Typical
(Note 9)
Limits
(Note 10)
Units
(Limits)
SCLK, SDI, SEN, MCLK, VSMP, CLMP Digital Input Characteristics
VIN(1)
Logical “1” Input Voltage
VA=5.25V
2.0
V (max)
VIN(0)
Logical “0” Input Voltage
VA=4.75V
0.8
V (min)
IIN
Input Leakage Current
VIN=VA
VIN=DGND
CIN
Input Capacitance
0.1
-0.1
µA (max)
µA (max)
5
pF
D0-D7 Digital Output Characteristics
VOUT(1)
Logical “1” Output Voltage
IOUT=-360µA
0.8*VD
V (min)
VOUT(0)
Logical “0” Output Voltage
IOUT=1.6mA
0.2*VD
V (max)
4
www.national.com
The following specifications apply for AGND=DGND=0V, VA=+5.0VDC, VD=+3.0 or +5.0VDC, fMCLK=12MHz. Boldface limits apply
for TA=TJ=TMIN to TMAX; all other limits TA=TJ=25°C. (Notes 7 & 8)
Symbol
Typical
(Note 9)
Limits
(Note 10)
Units
(Limits)
75
108
mA (max)
Power Down
675
900
µA (max)
Operating
210
475
µA (max)
2
25
µA (max)
Parameter
Conditions
Power Supply Characteristics
Operating
IA
ID
Analog Supply Current
Digital Supply Current (Note 15)
Power Down
AC Electrical Characteristics
The following specifications apply for AGND=DGND=0V, VA=+5.0VDC, VD=+3.0 or +5.0VDC, fMCLK=12MHz, except where noted
otherwise. Boldface limits apply for TA=TJ=TMIN to TMAX; all other limits TA=TJ=25°C. (Notes 7& 8)
Symbol
Parameter
f MCLK
Maximum MCLK frequency
tMCLK
MCLK period
Conditions
Typical
(Note 9)
Limits
(Note 10)
Units
(Limits)
12
MHz (min)
83
ns (min)
MCLK duty cycle
40
60
%(min)
%(max)
tSCLK
Serial Clock Period
1
tMCLK(min)
tSEN
Serial Enable high time
3
tMCLK(min)
tSSU
SDI setup time
1
ns (min)
tSH
SDI hold time
3
ns (min)
tSDDO
SCLK edge to new valid data
VD = 5.0V
VD = 3.3V
8.5
19
20
ns (max)
tVSU
VSMP setup time
1
ns (min)
tVH
VSMP hold time
3
ns (min)
tCSU
CLMP setup time
1
ns (min)
tCH
CLMP hold time
tDDO
MCLK edge to new valid data
VD = 5.0V
VD = 3.3V
16
25
3
ns (min)
25
ns (max)
ns (max)
Note 1: Absolute Maximum Ratings indicate limits beyond which damage to the device may occur. Operating Ratings indicate conditions for which the device is functional,
but do not guarantee specific performance limits. For guaranteed specifications and test conditions, see the Electrical Characteristics. The guaranteed specifications apply
only for the test conditions listed. Some performance characteristics may degrade when the device is not operated under the listed test conditions.
Note 2: All voltages are measured with respect to GND=AGND=DGND=0V, unless otherwise specified.
Note 3: When the input voltage (VIN) at any pin exceeds the power supplies (V IN<GND or VIN >VA or VD ), the current at that pin should be limited to 25mA. The 50mA maximum package input current rating limits the number of pins that can simultaneously safely exceed the power supplies with an input current of 25mA to two.
Note 4: The maximum power dissipation must be derated at elevated temperatures and is dictated by T Jmax, Θ JA and the ambient temperature, TA . The maximum allowable power dissipation at any temperature is PD = (T Jmax - TA ) / Θ JA. TJmax = 150°C for this device. The typical thermal resistance (Θ JA) of this part when board mounted
is 69°C/W for the M28B SOIC package.
Note 5: Human body model, 100pF capacitor discharged through a 1.5kΩ resistor. Machine model, 200 pF capacitor discharged through a 0Ω resistor.
Note 6: See AN450 “Surface Mounting Methods and Their Effect on Product Reliability” or the section titled “Surface Mount” found in any National Semiconductor Linear
Data Book for other methods of soldering surface mount devices.
5
www.national.com
LM9823
DC and Logic Electrical Characteristics (Continued)
LM9823
AC Electrical Characteristics (Continued)
Note 7: Two diodes clamp the OS analog inputs to AGND and VA as shown below. This input protection, in combination with the external clamp capacitor and the output
impedance of the sensor, prevents damage to the LM9823 from transients during power-up.
VA
TO INTERNAL
CIRCUITRY
OS Input
AGND
Note 8: To guarantee accuracy, it is required that VA and VD be connected to clean, low noise power supplies, with separate bypass capacitors at each supply pin. When
both VA and VD are operated at 5.0V, they must be powered by the same regulator, with separate power planes or traces and separate bypass capacitors at each supply pin.
Note 9: Typicals are at TJ =TA=25°C, f MCLK = 12MHz, and represent most likely parametric norm.
Note 10: Tested limits are guaranteed to National's AOQL (Average Outgoing Quality Level).
Note 11: Full channel integral non-linearity error is defined as the deviation of the analog value, expressed in LSBs, from the straight line that best fits the actual transfer
function of the AFE.
Note 12: The sensor’s maximum peak differential signal range is defined as the peak sensor output voltage for a white (full scale) image, with respect to the dark reference
level.
CIS Output Signal
CCD Output Signal
VRFT
VREF
V WHITE
Black Level
V WHITE
Note 13:
PGA Gain Error is the maximum difference between the measured gain for any PGA code and the ideal gain calculated by using the formula
V
32
PGA code
Gain PGA  ---- = G 0 + X --------------------------- where X = ( G 31 – G 0 ) ------ .
V
31
32
Note 14: Full Channel INL and DNL are tested with CDS disabled, negative signal polarity, DOE = 0, and a single OS input with a gain register setting of 1 (000001b) and
an offset register setting of 0 (000000b).
Note 15: The digital supply current (ID ) does not include the load, data and switching frequency dependent current required to drive the digital output bus on pins (D7 - D0).
The current required to switch the digital data bus can be calculated from: ISW = 2*ND *P SW*C L*VD/tMCLK where N D is total number of data pins, PSW is the probability of
each data bit switching, CL is the capacitive loading on each data pin, VD is the digital supply voltage and tMCLK is the period of the MCLK input. For most applications, ND
is 8, PSW is ≈ 0.5, and VD is 5V, and the switching current can be calculated from: ISW = 40*C L/t MCLK . (With VD at 3.3V, the equation becomes: ISW = 26.4*CL/tMCLK.) For
example, if the capacitive load on each digital output pin (D7 - D0) is 20pF and the period of t MCLK is 1/12MHz or 83ns, then the digital switching current would be 9.6mA.
The calculated digital switching current will be drawn through the VD pin and should be considered as part of the total power budget for the LM9823.
Note 16: All specifications quoted in LSBs are based on 12 bit resolution.
6
www.national.com
LM9823
Typical Performance Characteristics
Full Channel DNL and INL
(Divide by 2, Monochrome Mode, 6 MHz Pixel Rate)
Typical 16 Bit DNL
2
Typical 16 Bit INL
60
1.5
40
1
20
0.5
LSBs 0
LSBs 0
-20
-0.5
-40
-1
-60
0
16384
32768
49152
Output Code
65536
0
Typical 12 Bit DNL
1
16384
32768
49152
Output Code
65536
Typical 12 Bit INL
3
2
0.5
1
LSBs 0
0
LSBs
-1
-0.5
-2
-1
-3
0
1024
2048
3072
Output Code
4096
0
1024
2048
3072
Output Code
4096
Note: The LM9823 provides 16-bit data for high resolution imaging applications. The typical full channel device performance is shown
in the above graphs. In many applications, particularly those where high speed is important, or where lower cost CCD and CIS sensors
are used, the signal source is only accurate to 12 bits. In these applications, only 12-bit of data may be used. 12-bit DNL and INL plots
have also been provided to illustrate the performance of the LM9823 in these applications.
7
www.national.com
LM9823
Pin Descriptions
Analog Power
VA
The two VA pins are the analog supply pins.
They should be connected to a voltage
source of +5V and bypassed to AGND with a
0.1µF monolithic capacitor in parallel with a
10µF tantalum capacitor.
AGND
These two pins are the ground returns for
the analog supplies.
VD
This is the positive supply pin for the digital
I/O pins. It should be connected to a voltage
source between +3.3V and +5.0V and be
bypassed to DGND with a 0.1µF monolithic
capacitor in parallel with a 10µF tantalum
capacitor.
DGND
Data Output
D7-D0
Data Output pins. The 16 bit conversion
results of the ADC are multiplexed in 8 bit
bytes to D7-D0 synchronous with MCLK.
The MSB consists of data bits d15-d8 on
pins D7-D0 and the LSB consists of d7-d0
on pins D7-D0.
Serial Input/Output
SCLK
Serial Shift Clock. Input data on SDI is valid
on the rising edge of SCLK. The minimum
SCLK period is 1 tMCLK.
SDO
Serial Data Output. Data bits are shifted out
of SDO on falling edges of SCLK. The first
eight falling edges of SCLK after SEN goes
low will shift out eight data bits (MSB first)
from the configuration register addressed
during the previous SEN low time.
SDI
Serial Data Input. A read/write bit, followed
by a four address bits and eight data bits is
shifted into SDI, MSB first. Data should be
valid on the rising edge of SCLK. If the
read/write bit is a “0” (a write), then the
shifted data bits will be stored. If the
read/write bit is a “1” (a read), then the data
bits will be ignored, and SDO will shift out
the addressed register’s contents during the
next SEN low time.
SEN
Shift enable and load signal. When SEN is
low, data is shifted into SDI. When SEN
goes high, the last thirteen bits (one
read/write, four address and eight data)
shifted into SDI will be used to program the
addressed configuration register. SEN must
be high for at least 3 MCLK cycles between
SEN low times.
This is the ground return for the digital supply.
Analog Input/Output
OSR, OSG, OSB
VREF+, VREFMID,
VREF-
Analog Inputs. These inputs (for Red,
Green, and Blue) should be tied to the sensor’s OS (Output Signal) through DC blocking capacitors.
Voltage reference bypass pins. VREF+,
VREFMID, and VREF- should each be
bypassed to AGND through a 0.1µF monolithic capacitor.
Timing Control
MCLK
Master clock input. The ADC conversion
rate will be 1/2 of MCLK. 12MHz is the maximum frequency for MCLK.
VSMP
Sample timing input signal. If VSMP is high
on the rising edge of MCLK, the input is
sampled on the rising edge of the next
MCLK. The reference signal for the next
pixel will be sampled one to four MCLKs
later, depending on the value in the
CDSREF configuration bits. If CDS is not
enabled, the internal reference will be sampled during the reference sample time.
Connection Diagram
The number of MCLK cycles between
VSMP pulses determines the pixel rate.
Timing Diagrams 1 through 6 illustrate the
VSMP timings for all the valid pixel rates.
VBANDGAP
VREFMID
VA
AGND
OSR
Note: See the applications section of the
datasheet for the proper timing relationships
between VSMP and MCLK.
CLMP
VREF+
OSG
VREF-
Clamp timing input. If CLMP and VSMP are
high on the rising edge of MCLK, all three
OS inputs will be internally connected to
VCLAMP during the next pixel. VCLAMP is either
VREF+ or VREF- depending on the state of
the Signal Polarity bit in the Sample Mode
register (Reg. 0, Bit 4).
OSB
VA
AGND
SEN
SDI
SDO
8
1
2
3
4
5
6
7
8
9
10
11
12
13
14
LM9823
28 pin
SOIC
28
27
26
25
24
D7
23
22
21
20
19
18
17
16
15
D2
D6
D5
D4
D3
D1
D0
VD
DGND
CLMP
VSMP
MCLK
SCLK
www.national.com
LM9823
Timing Diagrams
OSR, OSG,
OSB
MCLK
VSMP
ADC Clock
(internal)
Sample Signal
Level
Sample Reference
Level
N-1
N
N
D7 -D0
GH
GL
BH
N+1
BL
RH
N-5
N-5
DOE (Register 0, Bit 4) = 0
RL
GH
N-4
GL
BH
N-4
BL
RH
N-4
H = d15-d8
L = d7-d0
Diagram 1: Divide by 6 Color Mode Sample and Data Output Timing
OSG
MCLK
VSMP
ADC Clock
(internal)
Sample Signal
Level
Sample Reference
Level
N-1
N
N
D7 -D0
XX
XX
XX
N+1
XX
GH
GL
XX
XX
XX
XX
GH
N-4
H = d15-d8
DOE (Register 0, Bit 4) = 0
L = d7-d0
Diagram 2: Divide by 6 Monochrome Mode Sample and Data Output Timing (Green Input shown)
OSR, OSG,
OSB
MCLK
VSMP
ADC Clock
(internal)
Sample Signal
Level
Sample Reference
Level
D7 -D0
N-1
N
N
XX
RH
RL
GH
N+1
GL
BH
N-4
N-4
DOE (Register 0, Bit 4) = 0
BL
XX
XX
RH
N-4
RL
GH
N-3
H = d15-d8
L = d7-d0
Diagram 3: Divide by 8 Color Mode Sample and Data Output Timing
9
www.national.com
LM9823
Timing Diagrams (continued)
OSG
MCLK
VSMP
ADC Clock
(internal)
Sample Signal
Level
Sample Reference
Level
D7 -D0
N
N-1
N
XX
GH
GL
N+1
XX
XX
XX
XX
XX
XX
GH
N-4
DOE (Register 0, Bit 4) = 0
GL
XX
N-3
H = d15-d8
L = d7-d0
Diagram 4: Divide by 8 Monochrome Mode Sample and Data Output Timing (Green Input Shown)
OSG
MCLK
VSMP
ADC Clock
(internal)
Sample Signal
Level
Sample Reference
Level
D7 -D0
N-1
N+1
N
N
N+1
N+2
N+2
CDSREF = 00
GH
GL
GH
GL
N-11
DOE (Register 0, Bit 4) = 0
GH
GL
N-10
GH
GL
N-9
H = d15-d8
L = d7-d0
Diagram 5: Divide by 3 Monochrome Mode Sample and Data Output Timing (Green Input shown)
OSG
MCLK
VSMP
ADC Clock
(internal)
Sample Signal
Level
Sample Reference
Level
D7 -D0
N-1
N
N-1
N+2
N+1
N
N+1
N+2
N+3
N+3
N+4
CDSREF = 00
GH
GL
N-11
GH
GL
GH
N-10
GL
N-9
GH
GL
N-8
GH
GL
N-7
H = d15-d8
DOE (Register 0, Bit 4) = 0
L = d7-d0
Diagram 6: Divide by 2 Monochrome Mode Sample and Data Output Timing (Green Input shown)
10
www.national.com
LM9823
Timing Diagrams (continued)
OSR, OSG,
OSB
MCLK
VSMP
Sample Signal
Sample Reference
CDSREF=00
D3 -D0
CDSREF=01
CDSREF=10
CDSREF=11
Diagram 7: Programmable Reference Sample Timing
OS
MCLK
VSMP
CLMP
N-1
Sample Signal
N+1
N
N
Sample Reference
N+1
N+2
N+2
CDSREF = 00
Clamp On
(internal signal)
Diagram 8: Clamp Timing With SMPCL = 0
OS
MCLK
VSMP
CLMP
N-1
Sample Signal
N+1
N
N
Sample Reference
N+1
N+2
N+2
CDSREF = 00
Clamp On
(internal signal)
Diagram 9: Clamp Timing With SMPCL = 1
SCLK
SEN
SDI
XX
0
R/W bit
A3
A2
A1
A0
D7
D6
D5
D4
D3
D2
D1
D0
XX
Diagram 10: Configuration Register Serial Write Timing
SCLK
SEN
SDI
XX
1
SDO
A3 - A0
XX
XX
1
A3 - A0
b7 - b0
R/W bit
Diagram 11: Configuration Register Serial Read Timing
11
www.national.com
LM9823
Timing Diagrams (continued)
tSCLK
1/2 tSCLK
1/2 tSCLK
SCLK
tSEN
SEN
tSSU
SDI
tSH
D2
tSSU
D1
D0
XX
R/W
A3
A2
tSH
A1
A0
D7
D4
D3
D2
tSDDO
SDO
XX
D7
D6
D5
Diagram 12: Serial Input and Output Timing
tMCLK
MCLK
tVSU
tVH
tCSU
tCH
VSMP
CLMP
tDDO
DOE = 1
D7 - D0
tDDO
DOE = 0
D7 - D0
Diagram 13: MCLK, VSMP and CLMP Input Timing and Data Output Timing
12
www.national.com
LM9823
Table 1: Configuration Register Address Table
Address
(Binary)
Register Name and Bit Definitions
A3
A2
A1
A0
0
0
0
0
B7
B6
B5
B4
B3
B2
B1
B0
Sample Mode (Power Up Default = 62h)
I/O Mode
DOE
CDS
Polarity
SMPCL
CDSREF1 CDSREF0
PD
Red Offset Setting (Power Up Default = 00h)
0
0
0
1
N/A
N/A
Polarity
MSB
LSB
Green Offset Setting (Power Up Default = 00h)
0
0
1
0
N/A
N/A
Polarity
MSB
LSB
Blue Offset Setting (Power Up Default = 00h)
0
0
1
1
N/A
N/A
Polarity
MSB
LSB
Red Gain Setting (Power Up Default = 00h)
0
1
0
0
N/A
N/A
x3
MSB
LSB
Green Gain Setting (Power Up Default = 00h)
0
1
0
1
N/A
N/A
x3
MSB
LSB
Blue Gain Setting (Power Up Default = 00h)
0
1
1
0
N/A
N/A
x3
MSB
LSB
Color Mode (Power Up Default = 00h)
0
1
1
1
N/A
N/A
N/A
N/A
N/A
N/A
CM1
CM0
0
0
0
0
0
0
Test Register 0 (Power Up Default = 00h)
1
0
0
0
0
0
0
0
0
0
Test Register 1 (Power Up Default = 10h)
1
0
0
1
0
0
0
1
0
0
Test Register 2 (Power Up Default = 00h)
1
0
1
0
0
0
0
13
0
0
0
www.national.com
LM9823
Table 2: Configuration Register Parameters
Note: Power-Up Default Register Settings are shown in Bold Italics
Parameter
(Address)
Control Bits
Result
Sample Mode (0)
I/O Mode
(0)
DOE
(Data Output Edge)
(0)
B7
0
1
B6
0
Normal Output Driver Operation
Reduced Slew Rate Output Driver Operation
1
D7-D0 are clocked out (change) on the falling edge of MCLK - Recommended setting for
lowest noise and best overall performance.
D7-D0 are clocked out (change) on the rising edge of MCLK
CDS (Enable)
(0)
B5
0
1
CDS disabled (CIS)
CDS Enabled (CCD)
Signal Polarity
(0)
B4
0
1
Negative Polarity (CCD) Clamping to VREF+
Positive Polarity (CIS) Clamping to VREF-
SMPCL
(0)
B3
0
1
Clamp is on for 1 MCLK before reference sampled
Clamp is on between the reference and the signal sample points
CDSREF
(0)
B2
0
0
1
1
PD
(Power Down)
(0)
B0
0
1
B1
0
1
0
1
Reference
Reference
Reference
Reference
(for pixel N+1) sampled 1 MCLK cycle after
(for pixel N+1) sampled 2 MCLK cycle after
(for pixel N+1) sampled 3 MCLK cycle after
(for pixel N+1) sampled 4 MCLK cycle after
signal (for pixel N) sampled
signal (for pixel N) sampled
signal (for pixel N) sampled
signal (for pixel N) sampled
Operating
Low Power Standby
14
www.national.com
LM9823
Table 2: Configuration Register Parameters
(Continued)
Note: Power-Up Default Register Settings are shown in Bold Italics
Parameter
(Address)
Control Bits
Result
Red, Green and Blue Offset DAC Settings (1, 2 & 3)
Offset Polarity
B5
0
1
Offset Value
B4(MSB)
B3
B2
B1
B0(LSB)
B4
(MSB)
0
0
0
•••
1
1
0
0
0
•••
1
1
B3
B2
B1
Typical Offset
Values
B5
(SIGN)
0
0
0
•••
0
0
1
1
1
•••
1
1
0
0
0
•••
1
1
0
0
0
•••
1
1
0
0
0
•••
1
1
0
0
0
•••
1
1
0
0
1
•••
1
1
0
0
1
•••
1
1
Positive Offset
Negative Offset
Typical Offset = 20LSBs * Offset Value * PGA Gain
Typical Offset (with PGA Gain = 1)
B0
(LSB)
0
1
0
•••
0
1
0
1
0
•••
0
1
12 bit LSBs
0.00
+20
+40
•••
+600
+620
0
-20
-40
•••
-600
-620
Red, Green and Blue Gain Settings (4, 5 & 6)
Boost Gain Enable
B5
0
1
PGA Gain Value
B4(MSB)
Boost Gain = 1V/V
Boost Gain = 3V/V
B3
B2
Gain
Typical Gain Values
B1
B0(LSB)
PGA Gain (V/V) =.933 + 0.0667 * (PGA Gain Value)
Gain = Boost Gain * PGA Gain
B5
(x3)
0
0
0
•••
0
0
0
•••
1
1
1
•••
1
1
1
B4
(MSB)
0
0
0
•••
1
1
1
•••
0
0
0
•••
1
1
1
B3
B2
B1
0
0
0
•••
1
1
1
•••
0
0
0
•••
1
1
1
0
0
0
•••
1
1
1
•••
0
0
0
•••
1
1
1
0
0
1
•••
0
1
1
•••
0
0
1
•••
0
1
1
15
B0
(LSB)
0
1
0
•••
1
0
1
•••
0
1
0
•••
1
0
1
Typical Gain
(V/V)
0.93
1.00
1.07
•••
2.87
2.93
3.00
•••
2.79
3.00
3.20
•••
8.60
8.80
9.00
www.national.com
LM9823
Table 2: Configuration Register Parameters
(Continued)
Note: Power-Up Default Register Settings are shown in Bold Italics
Parameter
(Address)
Control Bits
Result
Color Mode (7)
Color Mode
B1
0
0
1
1
B0
0
1
0
1
Color
Monochrome - Red
Monochrome - Green
Monochrome - Blue
Reserved Register 0 (8)
Reserved Register 0
00000000
Reserved, always set to 00h.
Reserved Register 1 (9)
Reserved Register 1
00010000
Reserved, always set to 10h.
Reserved Register 2 (A)
Reserved Register 2
00000000
Reserved, always set to 00h.
16
www.national.com
of Q1, and other sources of error. When the shift register clock
(Ø1) makes a low to high transition (period 4), the electrons from
the next pixel flow into C1. The charge across C1 now contains
the voltage proportional to the number of electrons plus VRESIDUAL, an error term. If OS is sampled at the end of period 3 and
that voltage is subtracted from the OS at the end of period 4, the
VRESIDUAL term is canceled and the noise on the signal is
reduced ([VSIGNAL+VRESIDUAL]-VRESIDUAL = VSIGNAL). This is the
principal of Correlated Double Sampling.
1.0 Introduction
The LM9823 is a high performance scanner Analog Font End
(AFE) for image sensor processing systems. It is designed to
work with color CCD and CIS image sensors and provides a full 3
channel sampling, gain and offset correction system, coupled
with a 16 bit high speed analog to digital converter. A typical
application of the LM9823 is in a color flatbed document scanner.
The image sensing and processing portion of the system would
be configured similar to that shown in Figure 1.
3.0 CIS Mode (CDS Off, Selectable Signal Polarity)
The also LM9823 supports CIS (Contact Image Sensor) devices.
The output signal of a CIS sensor (Figure 3) differs from a CCD
signal in two primary ways: its output usually increases with
increasing signal strength, and it does not usually have a reference level as an integral part of the output waveform of every
pixel.
To Host
ASIC
OSR
AFE Control
LM9823
CCD
CCD Control
OSG
OSB
8
8 Output Data
OS (CIS)
RAM
Other scanner elements
omitted for simplicity.
Figure 1. LM9823 in Basic Color Scanner
OS (CCD)
2.0 CDS Correlated Double Sampler
1
The LM9823 uses a high-performance CDS (Correlated Double
Sampling) circuit to remove many sources of noise and error from
the image sensor output signal. It also supports CIS image sensors with a single ended sampling mode.
2
3
4
5
Figure 3. CIS
When the LM9823 is in CIS (CDS off) mode (Register 0, B5=1), it
uses either VREF+ or VREF- as the reference (or black) voltage for
each pixel (depending on the signal polarity setting (Register 0,
Bit 4)). If the signal polarity is set to one, then VREF- will be sampled as the reference level. If it is set to zero, then VREF+ will be
sampled as the reference level.
Figure 2 shows the output stage of a typical CCD and the resulting output waveform:
VDD
Q1
C1
RS (RESET)
4.0 Programmable Gain
e-
Q2
The output of the Sampler drives the input of the x3 Boost gain
stage. The gain of each x3 Boost gain is 3V/V if bit B5 of that
color’s gain register (register 4,5, or 6) is set, or 1V/V if bit B5 is
cleared. The output of each x3 gain stage is the input an offset
DAC and the output of each offset DAC is the input to a PGA
(Programmable Gain Amplifier). Each PGA provides 5 bits of gain
correction over a 0.93V/V to 3V/V (-0.6 to 9.5dB) range. The x3
Boost gain stage and the PGA can be combined for an overall
gain range of 0.93V/V to 9.0V/V (-.6 to 19dB). The gain setting for
each color (registers 4, 5 and 6) should be set during calibration
to bring the maximum amplitude of the strongest pixel to a level
just below the desired maximum output from the ADC. The PGA
gain is determined by the following equation:
V
PGA Gain  ---- = 0.933 + .0667 (value in bits B4-B0)
V
Equation 1. PGA Gain
OS
(from shift register)
VSS
Ø1
RS
OS
1
2
3
4
5
Figure 2. CDS
Capacitor C1 converts the electrons coming from the CCD’s shift
register to an analog voltage. The source follower output stage
(Q2) buffers this voltage before it leaves the CCD. Q1 resets the
voltage across capacitor C1 between pixels at intervals 2 and 5.
When Q1 is on, the output signal (OS) is at its most positive voltage. After Q1 turns off (period 3), the OS level represents the
residual voltage across C1 (VRESIDUAL). VRESIDUAL includes
charge injection from Q1, thermal noise from the ON resistance
If the x3 Boost gain is enabled then the overall signal gain will be
three times the PGA gain.
17
www.national.com
LM9823
Applications Information
LM9823
Applications Information (Continued)
beginning of every line, the LM9823 implements a clamping function. The clamping function is initiated by asserting the CLMP
input. If CLMP and VSMP are both high on a rising edge of
MCLK, all three OS inputs will be internally connected to VREF+
or VREF- during the next pixel, depending on bit 4 of register 0. If
bit 4 is set to one (positive signal polarity), then the OS input will
be connected to VREF-. If bit 4 is set to zero (negative signal
polarity), then it will be connected to VREF+.
5.0 Offset DAC
The Offset DACs remove the DC offsets generated by the sensor
and the LM9823’s analog signal chain (see section 5.1, Internal
Offsets). The DAC value for each color (registers 1,2 and 3)
should be set during calibration to the lowest value that still
results in an ADC output code greater than zero for all the pixels
when scanning a black line. With a PGA gain of 1V/V, each LSB
of the offset DAC typically adds the equivalent of 20 ADC LSBs,
providing a total offset adjustment range of ±590 ADC LSBs. The
Offset DAC’s output voltage is given by:
6.1 Clamp Capacitor Selection
The output signal of many sensors rides on a DC offset (greater
than 5V for many CCDs) which is incompatible with the LM9823’s
5V operation. To eliminate this offset without resorting to additional higher voltage components, the output of the sensor is AC
coupled to the LM9823 through a DC blocking capacitor, C CLAMP.
The sensor’s DOS output, if available, is not used. The value of
this capacitor is determined by the leakage current of the
LM9823’s OS input and the output impedance of the sensor. The
leakage through the OS input determines how quickly the capacitor value will drift from the clamp value of VREF+ or VREF-, which
then determines how many pixels can be processed before the
droop causes errors in the conversion (±0.1V is the recommended limit for CDS operation). The output impedance of the
sensor determines how quickly the capacitor can be charged to
the clamp value during the black reference period at the beginning of every line.
V DAC = 9.75mV • (value in B4 - B0)
Equation 2. Offset DAC Output Voltage
In terms of 12 bit output codes, the offset is given by:
Offset = 20LSBs • (value in B4 - B0) • PGA Gain
Equation 3. Offset in ADC Output Codes
The offset is positive if bit B5 is cleared and negative if B5 is set.
Since the analog offset is added before the PGA gain, the value
of the PGA gain must be considered when selecting the offset
DAC values.
5.1 Internal Offsets
Figure 4 is a model of the LM9823’s internal offsets. Equation 4
shows how to calculate the expected output code given the input
voltage (VIN), the LM9823 internal offsets (VOS1, VOS2, VOS3),
the programmed offset DAC voltage (VDAC), the programmed
gains (GB, GPGA) and the analog channel gain constant C.
The minimum clamp capacitor value is determined by the maximum droop the LM9823 can tolerate while converting one sensor
line. The minimum clamp capacitor value is much smaller for CDS
mode applications than it is for CIS mode applications.
C is a constant that combines the gain error through the AFE, reference voltage variance, and analog voltage to digital code conversion into one constant. Ideally, C = 2048 codes/V (4096
codes/2V) in 12 bit LSBs. Manufacturing tolerances widen the
range of C (see Electrical Specifications).
x3 Boost
1V/V or
3V/V
VIN
+
Σ GB
+
VOS1
Inside LM9823
VIN
OS CS
CCLAMP
+
+
Σ + Σ GPGA Σ
+
+
+
VOS3
VOS2
VDAC
ADC
P2
P1
PGA
0.93V/V to
3V/V
CI
+
CDS Mode Input Circuitry
DOUT
Inside LM9823 C
I
Offset
DAC
VIN
Figure 4. Internal Offset Model
OS P2
CCLAMP
CS
P1
+
VREF
D OUT = ( ( ( VIN + V O S1 )GB + VDAC + V OS2 )G PG A + VOS3 )C
Equation 4. Output code calculation with internal offsets
CIS Mode Input Circuitry
Equation 5 is a simplification of the output code calculation,
neglecting the LM9823’s internal offsets.
Figure 5. Input Circuitry
The LM9823 input current is considerably less when the LM9823
is operating in CDS mode. In CDS mode, the LM9823 average
input current is no more than 25nA. With CDS disabled, which will
likely be the case when CIS sensors are used, the LM9823 input
impedance will be 1/(f Sample*CS). where fSample is the sample
rate of the analog input and CS is 2pF.
D O UT = ( VIN G B + VDAC )G PGA C
Equation 5. Simplified output code calculation
6.0 Clamping
6.1.1 CDS mode Minimum Clamp Capacitor Calculation:
To perform a DC restore across the AC coupling capacitors at the
The following equation takes the maximum leakage current into
18
www.national.com
CLMP is high. In this case the available charge time per line can
be calculated using:
the OS input, the maximum allowable droop, the number of pixels
on the sensor, and the pixel conversion rate, fVSMP, and provides
the minimum clamp capacitor value:
Number of optical black pixels
t CLAMP = ------------------------------------------------------------------------------2f VSMP
i
C CLAMP MIN = --------- dt
dV
leakage current (A) number of pixels
= --------------------------------------------------- -------------------------------------------max droop(V)
f VSMP
Equation 12. Clamp Time Per Line Calculation
For example, if a sensor has 18 black reference pixels and fVSMP
is 2MHz with a 50% duty cycle, then tCLAMP is 4.5µs. Other
“divide by” modes will have lower or higher clamp duty cycles
accordingly, depending on the SMPCL setting. See Diagram 8,
Clamp Timing With SMPCL = 0 and Diagram 9, Clamp Timing
With SMPCL = 1.
The following equation takes the number of optical black pixels,
the amount of time (per pixel) that the clamp is closed, the sensor’s output impedance, and the desired accuracy of the final
clamp voltage and provides the maximum clamp capacitor value
that allows the clamp capacitor to settle to the desired accuracy
within a single line:
Equation 6. CDS mode CCLAMP MIN Calculation
For example, if the OS input leakage current is 25nA worst-case,
the sensor has 2700 active pixels, the conversion rate is 2MHz
(tVSMP = 500ns), and the max droop desired is 0.1V, the minimum
clamp capacitor value is:
25nA 2700
= -------------- --------------0.1V 2MHz
= 340pF
Equation 7. CDS mode C CLAMP MIN Example
C
CLAMP MIN
t
1
= ------ -------------------------------R ln(accuracy)
t
CLAMP
1
= -------------------------- -------------------------------R CLAMP ln(accuracy)
Equation 13. CCLAMP MAX for a single line of charge time
C
6.1.2 CIS mode Minimum Clamp Capacitor Calculation:
If CDS is disabled, then the maximum LM9823 OS input leakage
current can be calculated from:
I leakage = VSAT f SampCLK C SAMP
CLAMP MAX
Where tCLAMP is the amount of time (per line) that the clamp is
on, RCLAMP is the output impedance of the CCD plus 50Ω for the
LM9823 internal clamp switch, and accuracy is the ratio of the
worst-case initial capacitor voltage to the desired final capacitor
voltage. If t CLAMP is 4.5µs, the output impedance of the sensor is
1500Ω, the worst case voltage change required across the capacitor (before the first line) is 5V, and the desired accuracy after
clamping is to within 0.1V (accuracy = 5/0.1 = 50), then:
4.5µs 1
C CLAMP MAX = ------------------ --------------1550Ω ln(50)
= 728pF
Equation 14. CCLAMP MAX Example
The final value for CCLAMP should be less than or equal to
CCLAMP MAX, but no less than C CLAMP MIN.
Equation 8. CIS mode Input Leakage Current Calculation
where VSAT is the peak pixel signal swing of the CIS OS output
and CSAMP is the capacitance of the LM9823 internal sampling
capacitor (2pF). Inserting this into Equation 6 results in:
i
C CLAMP MIN = --------- dt
dV
t SampCLK
V SAT
= ---------------------------C
----------------------------------- num pixels
t SampCLK SAMP max droop(V)
Equation 9. CIS mode CCLAMP MIN Calculation
with CSAMP equal to 2pF and VSAT equal to 2V (the LM9823 maximum input signal), then Equation 9 reduces to:
4p(F)(V)
CCLAMP MIN = ------------------------------------ num pixels
max droop(V)
In some cases, depending primarily on the choice of sensor,
CCLAMP MAX may actually be less than CCLAMP MIN, meaning that
the capacitor can not be charged to its final voltage during the
black pixels at the beginning of a line and hold it’s voltage without
drooping for the duration of that line. This is usually not a problem
because in most applications the sensor is clocked continuously
as soon as power is applied. In this case, a larger capacitor can
be used (guaranteeing that the CCLAMP MIN requirement is met),
and the final clamp voltage is forced across the capacitor over
multiple lines. This equation calculates how many lines are
required before the capacitor settles to the desired accuracy:
C

Initial Error Voltage
CLAMP
lines =  R CLAMP -------------------------  ln  ----------------------------------------------------
 Final Error Voltage
t

CLAMP 
Equation 10. CIS mode C CLAMP MIN Calculation
In CIS mode (CDS disabled), the max droop limit must be much
more carefully chosen, since any change in the clamp capacitor’s
DC value will affect the LM9823 conversion results. If a droop of
one 10 bit LSB across a line is considered acceptable, then the
allowed droop voltage is calculated as: 2V/1024, or approximately
2mV. If there are 2700 active pixels on a line then:
4p(F)(V)
C CLAMP MIN = ---------------------- 2700
2mV
= 5.4uF
Equation 11. CIS mode CCLAMP MIN Calculation Example
Equation 15. Number of Lines Required for Clamping
6.1.3 Maximum Clamp Capacitor Calculation:
Using the values shown before and a clamp capacitor value of
0.01µF, this works out to be:
5V
0.01µF
lines =  1550 ------------------- ln  ------------ = 13.5 lines

4.5µs   0.1V
Equation 16. Clamping Lines Required Example
The maximum size of the clamp capacitor is determined by the
amount of time available to charge it to the desired value during
the optical black portion of the sensor output. The internal clamp
occurs when CLMP and VSMP are both high on a rising edge of
MCLK. If SMPCL=0, the clamps are on immediately before the
sample reference time, if SMPCL=1, the clamps are on immediately after the sample reference time. If the LM9823 is operated
in divide by 2 mode, then the clamp is on 50% of the time when
In this example, a 0.01µF capacitor takes 14 lines after power-up
to charge to its final value. On subsequent lines, the only error will
be the droop across a single line which should be significantly
19
www.national.com
LM9823
Applications Information (Continued)
LM9823
Applications Information (Continued)
to 1 to reduce the slew rate of the output drivers.
less than the initial error. If the LM9823 is operating in CDS
mode and multiple lines are used to charge up the clamping
capacitors after power-up, then a clamp capacitor value of
0.01µF should be significantly greater than the calculated
CCLAMP MIN value and can virtually always be used.
9.2 DOE (Data Output Edge) Setting
The Data Output Edge bit selects which edge of MCLK is used to
clock output data onto the output pins. For lowest noise performance, this bit should be set to 0. With this setting, new data is
placed on the D7-D0 pins on every falling edge of MCLK. See
Diagrams 1 through 6 and Diagram 13 for more information on
data output timing for the different Divide By modes, and detailed
timing of the output data signals.
If the LM9823 is operating in CIS mode, then significantly larger
clamp capacitors must be used. Fortunately, the output impedance of most CIS sensors is significantly smaller than the output
impedance of CCD sensors, and RCLAMP will be dominated by
the 50Ω from the LM9823 internal clamp switch. With a smaller
RCLAMP value, the clamp capacitors will charge faster.
The bit can be set to 1 to adjust the data output timing for some
applications, but the noise performance of the LM9823 may be
somewhat degraded.
7.0 Power Supply Considerations
The LM9823 analog supplies (VA) should be powered by a single
+5V source. The two analog supplies are brought out individually
to allow separate bypassing for each supply input. They should
not be powered by two or more different supplies.
9.3 CDS Enable
The CDS Enable bit determines whether the sampling section of
the LM9823 operates in Correlated Double Sampling mode or in
Single Ended Sampling mode. CDS mode is normally used with
CCD type sensors, while Single Ended mode is normally used
with CIS type sensors.
Each supply input should be bypassed to its respective ground
with a 0.1µF capacitor located as close as possible to the supply
input pin. A single 10µF tantalum capacitor should be placed near
the VA supply pins to provide low frequency bypassing.
9.4 Signal Polarity
The VD input can be powered at 3.3V or 5.0V. Power should be
supplied by a clean, low noise linear power supply, with a 0.1µF
ceramic capacitor and a 10µF tantalum capacitor placed near the
VD and DGND pins. If possible, a separate power and ground
plane should be provided to isolate the noisy digital output signals
from the sensitive analog supply pins. If the VD voltage is lower
than VA, a separate linear regulator should be used. If VD and VA
are both at 5.0V, then they should be supplied by a common linear regulator, with separate analog and digital power and ground
planes.
Whether the LM9823 is operating in Correlated Double Sampling
Mode, or Single Ended Sampling mode, the basic sampling operation is the same. First a reference level is sampled, then a signal
level is sampled. For CDS mode operation, if the signal level is
lower in voltage than the reference level, the Signal Polarity bit
should be set to 0. This is the normal setting for CCD type sensors. If the signal level is more positive than the reference level,
the Signal Polarity bit would be set to 1 for Positive Polarity mode.
When Single Ended Mode is selected, the Signal Polarity bit
determines which internal reference voltage is used to compare
with the input signal. Most CIS type sensors have a positive polarity type output, and in this case the Signal Polarity Bit should be
set to 1. In this case, the internal VREF- is used as the reference
level during the Reference Sampling period.
To minimize noise, keep the LM9823 and all analog components
as far as possible from noise generators, such as switching power
supplies and high frequency digital busses. If possible, isolate all
the analog components and signals (OS, reference inputs and
outputs, VA, AGND) on an analog ground plane, separate from
the digital ground plane. The two ground planes should be tied
together at a single point, preferably the point where the power
supply enters the PCB.
In addition, the Signal Polarity bit determines which internal reference voltage is used during the Clamping interval. If Signal Polarity = 0, VREF+ is used for clamping, if Signal Polarity = 1, VREF- is
used.
8.0 Serial Interface and Configuration Registers
The serial interface is used to program the configuration registers
which control the operation of the LM9823. The SEN, SCLK, SDI
and SDO signals are used to set and verify configuration register
settings. In addition, MCLK must be active during all serial interface activity. MCLK is used to register the level of the SEN input
and drives the logic that process information input on the SDI line.
9.5 SMPCL
The SMPCL setting controls when the clamping action occurs
during the acquisition cycle. If SMPCL is set to 0, the Clamp will
be on for 1 MCLK before the reference sampling point. If SMPCL
is set to 1, clamping will occur in the interval after the reference
sampling point, and before the signal sampling point. In this case,
the clamping time is dependent on the present “divide by” mode,
and the settings of the CDSREF bits.
9.0 Sample Mode Register Settings
A brief overview of the sample mode register and the bit locations
is give in Table 2: Configuration Register Parameters on page
14. The function of each bit is summarized in the following sections.
9.6 CDSREF
The CDSREF setting is provided to allow adjustable sampling
points for the reference sample at the higher “divide by” modes.
This may be useful to optimize the timing of the Reference Sampling point for particular CCD sensors. Diagram 7 shows how the
various settings of CDSREF can be used to delay the Reference
Sampling point. Care must be taken to avoid setting CDSREF to
9.1 Output Driver Mode
The Output Driver Mode bit is normal set to 0. This bit can be set
20
www.national.com
To clamp across multiple pixels in a row, CLMP can be set high
and remain there for the entire number of pixels to be clamped,
then returned to the low state for normal (signal) operation. This
may simplify the timing required to generate the CLMP signal.
an inappropriate value when operating in the lower “divide by”
modes.
Valid CDSREF settings are:
“Divide By” Mode
Valid CDSREF
/8
00,01,10,11
/6
00,01,10,11
/3
00,01
/2
00
10.2 MCLK and VSMP Timing
The relationship between VSMP and MCLK is used to determine
the 'divide by' mode that is presently being used with the part.
Valid 'divide by' settings are:
Color - /8, /6
Monochrome - /8, /6, /3, /2
When entering a new mode, it is important to provide consistent
MCLK/VSMP timing signals that meet the following condition.
When switching to a new 'divide by' mode, VSMP should be held
low for a minimum of 3 MCLK cycles, then valid timing according
to the datasheet diagrams for the particular mode should be
started. This ensures that all internal circuitry is properly synchronized to the new conversion 'divide by' mode being used. If the
timing relationship between VSMP and MCLK is disturbed for any
reason, the same procedure should be used before restarting
operation in the chosen 'divide by' mode.
9.7 PD (Power Down) Mode
A Power Down bit is provided to configure the LM9823 in a lower
power Standby mode. In this mode, typical power consumption is
reduced to less than 1% of normal operating power. The serial
interface is still active, but the majority of the analog and digital
circuitry is powered down.
10.0 LM9823 Basic Operation
The normal operational sequence when using the LM9823 is as
follows:
Immediately after applying power, all configuration registers are
reset to default settings. MCLK should be applied, and the appropriate values written to the registers using the procedure discussed in section 8.0 Serial Interface and Configuration Registers
on page 20 and detailed in Diagrams 10, 11 and 12. Once the
configuration registers are loaded, the timing control signals can
be applied at the proper rates for the mode of conversion desired.
MCLK is applied initially with VSMP and CLMP low. After at least
3 MCLKS, VSMP and CLMP signals can begin. The divide by
mode is determined by the ratio of MCLK to VSMP frequency as
described in section 10.2.
For example: To change from monochrome divide by 3 mode to
monochrome divide by 2 mode, VSMP should be held low for at
least 3 MCLK cycles, then VSMP can be brought high using
'divide by 2' timing. If VSMP is not low for at least 3 MCLKs, the
LM9823 may enter an unknown mode.
MCLK
VSMP
Divide by 3
Transition
(≥ 3 MCLK)
Divide by 2
Diagram 6: Timing of Transitions between ‘Divide By’ Modes
16-Bit conversion results are placed on the data output pins as
follows: The upper 8 bits are output first with bit 15 of the ADC on
D7 and bit 8 of the ADC on D0. The lower 8 bits are then output
with bit 7 of the ADC on D7 and bit 0 of the ADC on D0. The exact
timing and conversion latency of the output data is affected by the
settings of the DOE variable in the Sample Mode register, and the
divide by mode of operation. If DOE = 0 (recommended setting
for best performance), output data will change on the falling edge
of MCLK. If DOE = 1, output data is updated on the rising edge of
MCLK. See Diagrams 1 through 6 and Diagram 13 for more information on data output timing.
10.1 CLMP Operation
The CLMP signal is used to engage the LM9823 internal clamp
circuits at the proper time during the CCD or CIS data output
cycle. If both CLMP and VSMP are high on a rising edge of
MCLK, then CLMP will be applied during the next pixel. The exact
timing of the internal Clamp signal is determined by the divide by
mode of operation and the setting of the SMPCL variable in the
Sample Mode register. If SMPCL = 0, then the Clamp is on for 1
MCLK before the reference is sampled. If SMPCL = 1 then the
clamp is on between the reference and the signal sample points.
Please see Diagram 8 and Diagram 9 for a graphic example of
this timing.
21
www.national.com
LM9823
Applications Information (Continued)
LM9823 3 Channel 48-Bit Color Scanner Analog Front End
Physical Dimensions inches (millimeters) unless otherwise noted
28-Lead (0.300" wide) Molded Small Outline Package (JEDEC)
Order Number LM9823CCWM
NS Package Number M28B
LIFE SUPPORT POLICY
NATIONAL'S PRODUCTS ARE NOT AUTHORIZED FOR USE AS CRITICAL COMPONENTS IN LIFE SUPPORT DEVICES OR
SYSTEMS WITHOUT THE EXPRESS WRITTEN APPROVAL OF THE PRESIDENT AND GENERAL COUNSEL OF NATIONAL
SEMICONDUCTOR CORPORATION. As used herein:
1. Life support devices or systems are devices or systems which,
(a) are intended for surgical implant into the body, or (b) support
or sustain life, and whose failure to perform when properly used
in accordance with instructions for use provided in the labeling,
can be reasonably expected to result in a significant injury to the
user.
N
National Semiconductor
National Semiconductor Europe
Corporation
Fax: +49 (0) 1 80-530 85 86
Americas
Tel: 1-800-272-9959
2. A critical component is any component of a life support device
or system whose failure to perform can be reasonably expected
to cause the failure of the life support device or system, or to
affect its safety or effectiveness.
National Semiconductor
Ltd. Japan Ltd.
Email: europe.support @ nsc.com
National Semiconductor Asia
Pacific Customer Response
Group
Deutsch Tel: +49 (0) 69 9508 6208
Tel: 65-2544466
Fax: 81-3-5639-7507
Fax: 1-800-737-7018
English Tel: +44 (0) 870 24 0 2171
Fax: 65-2504466
Email: support @ nsc.com
Francais Tel: +33 (0) 1 41 91 8790
Email: [email protected]
Tel: 81-3-5639-7560
http://www.national.com
National does not assume any responsibility for use of any circuitry described, no circuit patent licenses are implied and National reserves the right at any time without notice to change said circuitry and specifications.