Simulation Based Automated Analog Design using Statistical Optimization

Abstract

A cost function structure and innovative heuristics are given for the automated design of analog cells using statistical optimization in a simulation based approach. DC solution strategies are used to reduce CPU time. The technique allows to size complex circuits starting from scratch within reasonable CPU time. A 2¿m CMOS prototype of a 48 transistors fully differential opamp for a second order over-sampled modulator is presented demonstrating suitability of the technique.