LMS-based calibration of pipelined ADCs including linear and nonlinear errors

A least mean square (LMS) based calibration algorithm is proposed to calibrate most known error sources in 1.5 bit/stage pipelined ADCs, known to be immune to moderate comparator offsets. The error sources include linear gain errors, reference voltage errors, systematic offset errors and amplifier non-linear errors of each pipeline stage. LMS is used to estimate the error parameters in the digital domain. After estimation the proposed algorithm calibrate the pipelined ADC using the estimated parameter errors. Simulation results show that the proposed algorithm can improve the ENOB from 6.6 bits to 13.9 bits for a 14 bit 1.5 bits/stage pipelined ADC.