Design and implementation on high-order mismatch-shaped multibit delta-sigma d/a converters

Abstract

As the rapid evolution in semiconductor technology, transistors’ feature size has reached to 22nm and below, which brings great impact to analog and mixed-signal circuits. As the significant bridge connecting the analog world and digital system, data converter suffers from nonlinearity resulting from mismatch among its unit components. The smaller transistors are, the larger relative mismatch among them becomes. However, using larger transistors leads to more area cost and power consumption. Therefore, researchers have been working hard on how to alleviate the mismatch issue. In recent years, Dynamic Element Matching (DEM) becomes a popular approach that can significantly improve linearity, especially Spurious-free Dynamic Range (SFDR), of a data converter system. The basic idea of DEM is to shuffle the usage pattern of unit elements so that the mismatch error is no longer correlated to the input signal. Thus, DAC’s linearity will be improved. Generally, DEM Nyquist-rate DAC does mismatch scrambling, which smooths distortions resulting from mismatch into white noise. DEM Delta-Sigma DAC does mismatch shaping, which pushes distortions away from the signal band, typically lower frequencies. In this thesis, we focused on mismatch-shaping Delta-Sigma DACs. Two of those various algorithms are implemented logically and physically. With placement and routing information, we got more accurate result on the speed and power dissipation. The comparison shows the tradeoff among number of quantization levels, mismatch-shaping order, and hardware complexity.