A self-adaptive correlated multiple sampling technique based on single-slope ADC for low-noise CMOS image sensors

Abstract

This paper presents a self-adaptive correlated multiple sampling (SACMS) technique for low-noise CMOS image sensors. Compared to the traditional correlated multiple sampling (TCMS), SACMS technique adds a pre-determination stage to determine the brightness of the pixel, reducing the 4 large-range ramps to 3 small-range ramps and 1 large-range ramp when quantizing the signal. For dark pixels, the signal is quantized using small-range ramps with short sampling interval to reduce low-frequency noise. For bright pixels, the CMS operation is turned off and the signal is quantized only once in the large-range ramp to improve the frame rate. Simulation results demonstrate that, for dark pixels, SACMS achieves a minimum readout noise of 84.22μVrms at eight times analog gain, compared to 108.76μVrms with TCMS, representing an approximately 21.7 % improvement in noise suppression. Additionally, the sampling rate of the SSADC employing SACMS technique is 36.36 kHz, while the sampling rate of the SSADC using TCMS technique is 21 kHz.