An adaptive digital domain calibration technique for pipelined Analog-To-Digital Converters

Abstract

An adaptive digital calibration technique for enhancing the accuracy of Pipelined Analog-to-Digital Converters (ADC) is discussed in this paper. In contrast to the traditional approach that uses ideal bit-weights for calculation of digital output, the approach adopted here is based on calculating digital output using actual bit-weights. Actual bit-weights are extracted by measuring discontinuities at points where the bit transition occurs and applying the Least Mean Square (LMS) algorithm in the sense that mean-squared error is minimized to improve the linearity of the ADC. In contrast to the traditional MDAC residue characteristics that introduce errors in both stage-gain and reference terms, the modified MDAC architecture used here has only one error term in residue characteristic. Simulation results demonstrate the extension of inherent accuracy of ADC to 16-Bits and more.