摘要

This brief presents a novel mismatch shaping technique for multibit delta-sigma digital-to-analog converters (DACs). It uses the intrinsic quantization noise to randomize the element selection. Different from most existing mismatch shaping techniques that increase the element transition activity, the proposed technique keeps the same transition rate as that for the basic thermometer coding scheme. As a result, it produces much lower intersymbol interference (ISI)-induced distortions. Moreover, it does not produce tones and can high-pass shape the mismatch errors, unlike thermometer coding that produces large distortions due to static mismatch. An efficient hardware implementation based on the vector-quantizer mismatch shaping framework is also presented. Simulations show that the proposed technique can significantly improve DAC linearity in the presence of both ISI and mismatch errors.

  • 出版日期2014-7