Radix-based digital calibration method for pipelined ADCs with signal-dependent dithering

Abstract

This paper presents a new digital calibration method for a 14-bit 140MS/s pipelined analog-to-digital converter (ADC). Signal-dependent Pseudo-random Noise (PN) sequences are injected into the first 3.5-bit multiplying digital-to-analog converter (MDAC) without interrupting its normal conversion operation to extract the radix term, which includes finite opamp gain, settling error and capacitor mismatch. Compared to conventional fixed magnitude PN dithering, signal-dependent dithering greatly improves the signal-to-noise plus distortion (SNDR) of the ADC and reduces the calibration time. By using proposed method, the ADC achieves a peak SNDR of 81.5dB and spurious-free dynamic range (SFDR) of 102.8dB with 31.6MHz input at 140MS/s and the ENOB is improved from 6.5 bits to 13.25 bits.