Automatic behavioral model generation of nonlinear analog circuits.

摘要

The growing complexity of analog and mixed-signal systems continues to stimulate the need for behavioral models of analog components. While things have matured nicely in the modeling of linear circuits and systems, the present methodologies for nonlinear circuit modeling are circuit-specific. Developing modeling methodologies for circuits with strong nonlinearities remains a challenge. In this work, a behavioral modeling method for nonlinear analog circuits was developed.;The behavioral modeling method starts from the netlist description of the circuit and generates differential algebraic equation (DAE) based behavioral models. Consider a circuit as being comprised of a set of nonlinear ordinary differential equations obtained by applying Kirchhoff's Current Law (KCL) at each node. This method seeks to determine a subset of those equations for the behavioral model. The basic idea behind this approach is to analyze the original circuit for (a) signal path information, (b) linear frequency response, and (c) nonlinear time-domain response to identify a subset of nodes in the original circuit. Each node is physically represented by a collection of elements such as voltage-controlled current sources and linear and nonlinear passives. KCL is applied to these nodes to derive a reduced set of DAEs that represent the behavioral model.;This behavioral modeling method has been fully automated. Simulation times for the models compare quite favorably to those of the original circuits, while high degrees of accuracy are maintained.