There was a recent article in EE times (Oct 13) that talked about a technique for extracting the frequency domain behavior from nonlinear devices using spice simulations with special excitations. Apparently, this technique utilizes an algorithm to extract a "kernal" formula for the device that yields the frequency domain response.

This technique apparently bypasses the need to do multi-dimensional FFTs on the time simulations of the non-linear devices. A fast algorithm on the time simulation data yields the frequency domain behavior (apparently) very fast.

Work on this method is currently being persued by none other than Paul Tuinenga (founder of Microsim). He is currently presenting his idea to chip makers and EDA vendors.

I was just wondering if Eagleware had been in contact with Mr. Tuinenga to talk about his proposed methods (or at least caught wind of them). It sounds like sort of a "harmonic balance holy grail", if the method works....

(Ref: Electronic Engineering Times, October 13, 2003, pp 51, 52 "Analog Vet Revamps Circuit Simulation")

later, Don

This technique apparently bypasses the need to do multi-dimensional FFTs on the time simulations of the non-linear devices. A fast algorithm on the time simulation data yields the frequency domain behavior (apparently) very fast.

Work on this method is currently being persued by none other than Paul Tuinenga (founder of Microsim). He is currently presenting his idea to chip makers and EDA vendors.

I was just wondering if Eagleware had been in contact with Mr. Tuinenga to talk about his proposed methods (or at least caught wind of them). It sounds like sort of a "harmonic balance holy grail", if the method works....

(Ref: Electronic Engineering Times, October 13, 2003, pp 51, 52 "Analog Vet Revamps Circuit Simulation")

later, Don

Rob Lefebvre