A Split-Based Neural Network Calibrator for SAR-Pipelined ADC on FPGA

This paper proposes an FPGA implementation scheme for the calibrator for SAR-Pipelined ADC based on split architecture and neural network. Different from the traditional calibrator, this calibrator is prior-knowledge free. It adopts two fully connected neural networks with single hidden layer. Binary input makes the hidden layer multiplier-free. The experiment indicates that the SFDR (Spurious-Free Dynamic Range) of the ADC is increased by 44.1 dB, from 49.5 dB to 93.6 dB with the assistance of the calibrator.