![](http://datasheet.mmic.net.cn/290000/HV7161SPA2_datasheet_16136939/HV7161SPA2_12.png)
Confidential
HV7161SPA2
CMOS Image Sensor
With Image Signal Processing
This document has a general product description and is subject to change without notice.
MagnaChip Semiconductor Ltd. does not assume any responsibility for use of circuits described
and no patent licenses are implied.
- 12 -
2005 MagnaChip Semiconductor Ltd.
Integration
plane
frame 0
Integration
plane
frame 1
Scan
plane
frame 0
Scan
plane
frame 1
Time
Integration
time
Frame 0
time
Frame 1
time
Correlated Double Sampling(CDS) and Programmable Gain Amplifier(PGA)
Correlated double sampling is a circuit for reducing any correlated noise between reset signal and
video data signal. First, reset signal is taken and held at the reset sampler. Second, video signal is
taken and held at the data sampler. Final, these two signal is subtracted to null any common
signal(fixed pattern noise), and thereby the correlated noise that exits at both the reset signal and
video signal is minimized.
The above subtracted video signal is properly scaled by each R, G, and B programmable gain
amplifier and sequentially transmitted from PGA to ADC via the serial line. The amplifier's gain is
mutually controlled by auto-exposure control function(AE) and auto-white balance control
function(AWB). Any internal offset or noise within the amplifier is removed by internal offset
elimination circuit. The scaling range of pre-amplifier for AE is from 0.5X to 16.5X, and the scaling
range of color-amplifier for AWB is from 0.5X to 3.5X.
11bit on-chip ADC
On-chip ADC converts analog pixel voltage to 11bit digital data. On-chip ADC has a low power, a
high resolution, and a high conversion speed to be suitable at an imaging system. Internally, to null
the parasitic offset and the fixed pattern noise of pixel, on-chip ADC has an offset adjustment
circuit. Also, to increase the conversion rate on-chip ADC has a bias control circuit.
Dark Noise Cancellation(DNC)
When an interesting center pixel has abnormally large value (decided to the 'dark noise') the
center pixel is corrected by using neighbor pixels.
Gamma Correction
Piecewise linear approximation method is implemented. Ten-piece linear segments are supported
and user-programmable.