Noise-Shaping SAR ADCs.

Abstract

This work investigates hybrid analog-to-digital converters (ADCs) that combine the phenomenal energy efficiency of successive-approximation (SAR) ADCs with the resolution enhancement strategies used by noise-shaping converters. Because charge-redistribution SAR ADCs contain few active components and rely on highly digital controllers, SAR ADCs demonstrate the best energy efficiencies of all low bandwidth, moderate resolution converters (~10 bits).

SAR ADCs achieve remarkable power efficiency at low resolution, but as the resolution of the SAR ADC increases, the specifications for input-referred comparator noise become more stringent and total DAC capacitance becomes too large, which degrades both power efficiency and bandwidth. For these reasons, lower resolution, lower bandwidth applications tend to favor traditional SAR ADC architectures, while higher bandwidth, higher resolution applications tend to favor pipeline-SARs. Although the use of amplifiers in pipeline-assisted SARs relaxes the comparator noise requirements and improves bandwidth, amplifier design becomes more of a challenge in highly scaled processes with reduced supply voltages.

In this work, we explore the use of feedback and noise-shaping to enhance the resolution of SAR ADCs. Unlike pipeline-SARs, which require high-gain, linear amplifiers, noise-shaping SARs can be constructed using passive FIR filter structures. Furthermore, the use of feedback and noise-shaping reduces the impact of thermal kT/C noise and comparator noise. This work details and explores a new class of noise-shaping SARs.