The ADCS-1021 and ADCS-2021
CMOS Image Sensors capture
high quality, low noise images
while consuming very low power.
These parts integrate a highly
sensitive active pixel photodiode
array with timing control and
onboard A/D conversion. Available in either VGA (640x480) or
CIF (352x288) resolution image
arrays, the devices are ideally
suited for a wide variety of
applications.
The ADCS-2021 and ADCS-1021,
when coupled with compatible
image processors from either
Agilent or selected Agilent
partners, provide a complete
imaging system to enable rapid
end-product development.
Designed for low-cost consumer
electronic applications, the
ADCS-2021 and ADCS-1021
sensors deliver unparalleled
performance for mainstream
imaging applications.
ADCS-2021 (VGA) and
ADCS-1021 (CIF) are CMOS active
pixel image sensors with integrated A/D conversion and full
timing control. They provide
random access of sensor pixels,
which allows windowing and
panning capabilities. The sensor
is designed for video conferencing
Agilent ADCS-1021, ADCS-2021
CMOS Image Sensors
Data Sheet
Key Specifications and Features
• High quality, low cost CMOS image
sensors
• Industry-standard 32-pin CLCC
package
• VGA resolution (640H x 480V)–
applications and still image
capabilities. The ADCS family
achieves excellent image quality
with very low dark current, high
sensitivity, and superior antiblooming characteristics. The
devices operate from a single DC
bias voltage, are easy to configure
and control, and feature low
power consumption.
Programmable Features
• Programmable window size ranging
from the full array down to a 4 x 4
pixel window
• Programmable panning capability
which allows a specified window
(minimum 4x4 pixels) to be located
anywhere on the sensor array
• Integrated programmable gain
amplifiers with independent gain
control for each color (R, G, B)
• Internal register set programmable
via either the UART or Synchronous
serial interface
• Integrated timing controller with
rolling electronic shutter, row/
column addressing, and operating
mode selection with programmable
exposure control, frame rate, and
data rate
• Programmable horizontal, vertical,
and shutter synchronization signals
• Programmable horizontal and
vertical blanking intervals
ADCS-2021
• CIF resolution (352H x 288V)–
ADCS-1021
• High frame rates for digital video
VGA: 15 frames/second
CIF: 30 frames/second
• High sensitivity, low noise design
ideal for capturing high-quality
images in a variety of lighting
conditions
• Automatic subtraction of column
fixed pattern noise
• Still image capability
• Synchronous serial or UART
interface
• Integrated voltage references
Applications
• Digital still camera
• PC camera
• Handheld computers
• Cellular phones
• Notebook computers
• Toys
Page 2
Brief Introduction
The Agilent ADCS-2021 and
Agilent ADCS-1021 image sensors act as normal CMOS digital
devices from the outside. Internal circuits are a combination of
sensitive analog and timing
circuits. Therefore, the designer
must pay attention to the PC
board layout and power supply
design. Writing to registers via an
2
I
C compatible two-wire interface
provides control of the sensor.
Sensor data is normally output
via an 8 or 10 bit parallel interface (serial data output is also
available). Once the registers are
programmed the sensor is selfclocking and all timing is internally generated. On chip programmable amplifiers provide a
way to separately adjust the red
green and blue pixels for a good
white balance. Analog to digital
conversion is also on chip and 8
or 10 bit digital data is output. A
data ready pulse follows each
valid pixel output. An end of row
signal follows each row and an
end of frame signal follows each
frame.
PCB Layout
Analog Vdd and analog ground
need to be routed separately
from digital Vdd and digital
ground. Noisy circuits or ICs
should not be placed on the
opposite side of the PC board.
Heat producing circuits such as
microprocessors or LCD displays
should not be placed next to or
opposite from the sensor to
reduce noise in the image.
Power Supply
The sensor operates at 3.3 VDC.
There are two power supplies for
the sensor. Analog V
dd
and
Digital Vdd. The two supplies and
grounds must be kept separate.
Two separate regulators provide
the best isolation. Any noise on
the analog supply will result in
noise in the image. Analog and
digital ground should be tied
together at a single point of
lowest impedance and noise.
Master Clock
The part requires a 50% duty cycle
master clock. Maximum clock
rates are 25 MHz for ADCS-2021
and 32 MHZ for ADCS-1021.
Reset
A hard reset is required before
the sensor will function properly.
Once the master clock is running,
assert nRST_nSTBY for 40 clock
cycles.
Register Communication
Communication (read/write) to
the sensor registers is via a two
wire serial interface — either a
synchronous I2C compatible or
half duplex UART (9600 baud
default). nTristate (pin 3
ADCS-1021 only) must be pulled
high for normal operation. The
ADCS-2021 does not have
nTristate.
Parallel Data Output
8 or 10 bit parallel data is output
from the sensor. A data ready
line (DRDY) is asserted when the
data is valid. The sensor acts as a
master in the way it outputs
data. There is no flow control or
data received handshake. Once
the RUN bit (CONTROL register)
is set, the image processor must
be ready to accept data at the
sensor rate and when the data is
presented.
Serial Data Output
In this mode, output data lines
D0 and D1 (the lower two bits of
the parallel data port) act as a
two wire serial interface.
Handshaking
At the end of one row of data, the
nROW line is asserted. At the end
of one frame of data, the
nFRAME_nSYNC line is asserted.
Registers
On the next page is a table of
sample register settings (see
Figure 1). These values are a
good starting point.
2
Page 3
Table 1. Register Set Declaration for Agilent ADCS-1021 and ADCS-2021 Image Sensors.
Register NameMnemonicAddress (hex)Sample Value (hex)
Identifications RegisterIDENT0x00
Status RegisterSTATUS0x010x7F
Interrupt Mask RegisterIMASK0x020x00
Pad Control RegisterPCTRL0x030x03
Pad Drive Control RegisterPDRV0x040x00
Interface Control RegisterICTRL0x050x20
Interface Timing RegisterITMG0x060x00
Baud Fraction RegisterBFRAC0x070x00
Baud Rate RegisterBRATE0x080x00
ADC Control RegisterADCCTRL0x090x08
First Window Row RegisterFWROW0x0A0x00
First Window Column RegisterFWCOL0x0B0x07
Last Window Row RegisterLWROW0x0C0x79
Last Window Column RegisterLWCOL0x0D0xA8
Timing Control RegisterTCTRL0x0E0x04
PGA Gain Register: GreenERECPGA0x0F0x00
PGA Gain Register: RedEROCPGA0x100x00
PGA Gain Register: BlueORECPGA0x110x00
PGA Gain Register: GreenOROCPGA0x120x00
Row Exposure Low RegisterROWEXPL0x130x00
Row Exposure High RegisterROWEXPH0x140x02
Sub-Row Exposure RegisterSROWEXP0x150x00
Error Control RegisterERROR0x160x00
Interface Timing 2 RegisterITMG20x170x4B
Interface Control 2 RegisterICTRL20x180x00
Horizontal Blank RegisterHBLANK0x190x00
Vertical Blank RegisterVBLANK0x1A0x00
Configuration RegisterCONFIG0x1B0x0C
Control RegisterCONTROL0x1C0x04
Reserved0x1D—
Reserved0x1E—
Reserved0x1F—
Reserved0x20—
3
Page 4
Setting Exposure and Gain
The exposure of an image is a
function of the exposure and
gain registers. Exposure sets the
length of time each pixel integrates the light (shutter speed).
Gain settings allow pixel values
to be amplified. Gain values from
1x to 40x are allowed, but higher
gain settings amplify noise (much
like higher ISO film speeds are
grainier). It is best to use the
lower gain settings for better
images. Gains from 1x to 10x are
generally recommended.
Note in Table 2, there are two
green gain registers, one for the
odd number row green pixels and
one for the even number row
green pixels. The green color
filters can be slightly different
between rows and this allows
fine-tuning. Using the same gain
setting for both green registers is
usually enough. Since the blue
channel is not as sensitive, using
blue gains approximately double
that of red and green will allow
the A/D full range on all three
channels.
Using a MacBeth Color Checker is
a good way to judge exposure and
color balance. A good raw image
will have a good grey scale (the
bottom patches on the chart).
Gain settings should be adjusted
so the red, green, and blue values
are equal on any one grey patch.
After setting gain, the exposure
registers should be adjusted for a
good exposure. There are three
exposure registers (see Table 3).
Table 2. PGA Gain Register Settings.
Register NameMnemonicAddress (hex)
PGA Gain Register: GreenERECPGA0x0F
PGA Gain Register: RedEROCPGA0x10
PGA Gain Register: BlueORECPGA0x11
PGA Gain Register: GreenOROCPGA0x12
Table 3. Row Exposure Register Settings.
Register NameMnemonicAddress (hex)
Row Exposure Low RegisterROWEXPL0x13
Row Exposure High RegisterROWEXPH0x14
Sub-Row Exposure RegisterSROWEXP0x15
The row exposure high register
(upper 8 bits) and row exposure
low register (lower 8 bits) act as
a single 16 bit register. This
16 bit register sets the integration time (shutter speed) of the
Image Processing
The raw data from the sensor
requires image processing before
a digital image is ready for
viewing. Some standard steps of
image processing are as follows:
sensor. The sub-row exposure
register is used for very small
changes to exposure and allow
fine-tuning for exact shutter
speeds.
1. Defective pixel correction
2. Lens flare subtraction
3. Auto-exposure
4. Auto-white balance
5. Color filter array interpolation
Proper exposure will result in
black values near 0x00 and white
values near 0xFF (assuming
8 bits). All six grey patches on the
MacBeth chart should have
different average intensity values
(demosaic)
6. Color correction (3x3 matrix)
7. Gamma correction
8. Color space correction (3x3
matrix)
9. Data compression
in the image. If the two brightest
patches both appear white then
the exposure is too long. If the
two darkest patches both appear
black then the exposure is too
short. Remember that the raw
image does not have gamma
correction applied yet. The final
Image processing is not part of
the sensor and must be supplied
separately. Image processors
that are compatible with these
sensors are available from Agilent
Technologies and selected Agilent
partners.
grey scale image needs to be
evaluated after gamma correction.
4
Page 5
Typical Application
30 MHz
Clock
27
Clk
Vdd
2
DRDY
nRST_nSTBY
nROW
Digital
Vdd
24, 31, 5
3.3V
D0
1
D1
30
D2
29
D3
28
D4
21
D5
20
D6
19
D7
26
13
14
9
18
17
12
IMODE0
IMODE1
11
3
nTRISTATE
10K
10
nIRQ
NC
4
NCClock
NC
Analog
Vdd
23, 8,
16
3.3V
Regulator
ADCS-1021
Analog
GND
22, 7,
15
nFRAME_nSYNC
Digital
GND
25, 32,
6
Regulator
Star Ground
D0
D1
D2
D3
D4
D5
D6
D7
DATA READY
Reset
End of Row
End of Frame
TxD/RxD
Parallel Interface
Serial Interface
Host System
Typical Electrical Specifications
Part NumberADCS-2021 (VGA)ADCS-1021 (CIF)
Pixel size7.4 x 7.4 µm7.4 x 7.4 µm
Maximum Clock Rate25 MHz (VGA)32 MHz (CIF)
Effective Sensor Dynamic Range65 dB (VGA)61 dB (CIF)
Effective Noise Floor43 e-43 e-
Dark Signal
Saturation Voltage1.22 V1.22 V
Full Well Capacity68,000 e-68,000 e-
Conversion Gain
Programmable Gain Range1 –40 (8 bit resolution)1– 40 (8 bit resolution)