Integral Nonlinearity and Area Parameters Optimization of Segmented Digital to Analog Converter

This paper introduces a novel prototyping algorithm for integrated current-steering digital-to-analog converter segmentation. The proposed algorithm aims to accelerate design and replace time-consuming direct calculations using computer-aided design tools. It leverages converter's integral nonlinearity and occupied area parameters. The paper focuses specifically on the impact of device mismatch effects on integral nonlinearity, a significant performance-limiting factor in digital-to-analog converter design. While acknowledging the existence of other contributing factors, this work demonstrates the algorithm's effectiveness in capturing and predicting integral nonlinearity degradation due to mismatch. The prototyping approach employs a Monte-Carlo algorithm applied to a simplified mathematical model of the digital-to-analog converter, implemented in MATLAB. The electrical parameters of this model are randomly selected within boundaries defined by relative variations extracted from simulations of unit cells (current sources) using computer-aided design tools. To demonstrate the algorithm's practical applicability, the paper presents a detailed example of its implementation on choosing optimal segmentation of 12-bit digital-to-analog converter in 0.18 um Bipolar-CMOS-DMOS process. Furthermore, a comparison with direct computer-aided design tools simulations validates the accuracy and efficiency of the proposed approach, showcasing its potential for significantly reducing design time and complexity.