Computer Aided Design of Sigma-Delta ADCs

Since no accurate analytical model is available for ΣΔ ADCs in general, there is need to simulate different possible solutions before deciding which one has the potential to be the optimal one. Even if a local optimum is targeted, for example if the constraint is to use a fourth order design, the lack of a good analytical model undermines the use of gradient-based methods. This Chapter presents a systematic procedure to exhaustively explore the design space of ΣΔ ADCs, by automatically designing and evaluating a large number of candidate solutions based on behavioral simulations. Section 5.1 presents the first level of abstraction used in this procedure, the filter level, where the linear model is used (with all its limitations) to reduce the number of possible solutions. Section 5.2 presents the second abstraction level, the architecture level, where the ΣΔ ADC loops are actually designed and simulated to measure their performance. Two design examples are shown in Section 5.3.