Computational SAR ADC for a 3D CMOS image sensor

Abstract

The architecture and simulation of a Computational SAR ADC (C-SAR) dedicated to the processing of image descriptors for a 3D CMOS image sensor are reported here. The differential charge sharing architecture enables to A/D convert the convolution of multiple binary weighted pixels signals on multi-scale kernels. The CMOS image sensor is constituted of two tiers. An array of C-SAR is implemented on the bottom layer. Each C-SAR is associated to a square of 8×8 pixels on the top layer, with a pitch of 10μm and a fill factor of 80%. With regard to a standard differential SAR ADC, only multiplexing facilities are added in the C-SAR. This area over cost is 10 times lower than the surface induced by the memory required for a counterpart digital architecture. The total noise of 458μVRMS simulated at transistor level on a 65nm technology enables to reach a processing resolution of 9 signed bits on 0.5V pixels dynamic. As the processing is done within the conversion stage, no additional time is needed. With a power consumption of 400μW and a bandwidth of 1 mega-convolution per second, this processing architecture outputs a FOM of 6.25pJ/pixel.