A comparative study of design of analog CMOS circuits using numerical optimization strategies

Abstract

Design of analog CMOS circuits is time consuming and very complex. Analog circuits need to often be handcrafted as opposed to software and digital design which can simply be recompiled following a change of parameters. Further, minor changes in design parameters can have unexpected outcomes on system performance, due to which manual design space exploration can be difficult except for most non-trivial analog circuit configurations. In this paper we explore different methods for numerical optimization based tools which are available to automate analog design so that the design in a pseudo sense can simply be recompiled following some changes. We explore genetic algorithms, geometric optimization as well as finding the minimum of constrained nonlinear multivariable functions. The findings provide guidelines to Op-Amp designers looking to automate the parameter selection process.