A behavioral signal path modeling methodology for qualitative insight in and efficient sizing of CMOS opamps

Abstract

This paper describes a new modeling methodology that allows to derive systematically behavioral signal path models of operational amplifiers. Combined with symbolic simulation, these models provide high qualitative insight in the small-signal functioning of a circuit. The behavioral signal path model provides compact interpretable expressions for the poles and zeros that constitute the signal path. These expressions show which design parameters have dominant influence on the position of a pole/zero and thus enable a designer to control a manual interactive sizing process. The methodology consists of the application of a sequence of abstractions, so that one gradually progresses from a full device to a full behavior circuit representation. During this translation, qualitative insight and design requirements are obtained. The methodology is implemented in an open tool called \EFtoef. The behavioral signal path model is also used for optimization based sizing in order to achieve pole placement in an efficient way. For optimization based sizing, a new strategy for hierarchical penalty function composition is proposed, which allows sequential pruning of the design space. Combined with an operating point driven \DC formulation and local minimax optimization, a fast sizing method is obtained which can be used for interactive design space exploration. Experimental results of both modeling and sizing are shown.