Inverse Modelling Techniques as Dynamic Digital Pre-Distortion for Broadband Systems for CS-DACs Based on Volterra Series Expansion

This paper presents a dynamic digital pre-distortion (DPD) method for a current-steering digital-to-analog converter (CS-DAC). The novel proposed method is based on the inverse modelling of the system with a Volterra series expansion. It uses a memory structure often only used for DPD in power amplifiers (PA). The goal is to find a model to counteract the signal dependent non-linearities of charge injection from the switching activity of the converter with a limited parameter count. A priori knowledge of the non-linear behavior and architecture of the DAC is exploited. The proposed Volterra system consists of a first and third order kernel with a respective memory depth of one. This results in a total number of only 6 parameters that need to be estimated. This is sufficient to significantly attenuate the non-linear distortions and efficient in terms of memory usage and computational complexity. The paper compares three different methods for generating the inverse of the Volterra system and determines the performance difference between them. The evaluation of the proposed DPD is done via Spice simulations of a 14-bit CS-DAC using a lumped circuit model and Spectre simulations of a real 14-bit 16 nm DAC design. Narrow-band and broad-band discrete multi-tone (DMT) signals with a frequency of up to 684 MHz are considered. The results show a performance gain of the missing-tone-power-ratio (MTPR) of up to 8.4 dB for both model-based and circuit-based simulations based analysis, respectively.