An efficient method of Pareto-optimal front generation for analog circuits

摘要

In this paper, an efficient deterministic methodology for generating Pareto-optimal front (PoF) of analog circuits is proposed. The proposed methodology utilizes modified epsilon constraint method along with geometric programming based circuit sizer to determine the Pareto-optimal points (PoPs) of the analog circuits. The generated PoPs are then modeled to generate the PoFs. The efficiency of the proposed methodology and the accuracy of the generated PoF has been verified with respect to that of the commonly used stochastic approach. It is found that to generate PoF having similar spread, the proposed methodology takes only 20 min whereas the stochastic approach takes 60 h. It also has been observed that the accuracy of the generated PoF using the proposed methodology improved by more than 10% with respect to that of the stochastic approach. The proposed methodology has been implemented to generate the PoF of a two-stage Op-Amp, fully differential folded cascode Op-Amp, fully differential single-stage and two-stage Op-Amps. The generated PoFs have been utilized for (1) feasibility checking for a user given specification of an Op-Amp, (2) performance prediction and (3) topology selection of the analog circuit. As per authors' knowledge, this is the first paper which deals with all the aspects, i.e., generation, modeling and application of the PoF of the analog circuits.