A nonlinear statistical MESFET model using low order statistics of equivalent circuit model parameter sets

Abstract

A nonlinear statistical MESFET model is presented which shows good agreement with measured results. A single stage GaAs power amplifier tuned at 13.5 GHz is simulated with the model and accurately predicts output power, input return loss, and power added efficiency. Not only is nominal performance good, but the spreads of amplifier performance seen over many wafers and several lots is represented well. The model consists of capturing low order statistics (means, variances, and correlations) of equivalent circuit model parameter sets and using these to produce a continuous distribution of devices representative of those seen from manufacturing. Complete nonlinear and statistical modeling extraction software packages are also presented which allow easy investigation of many aspects of statistical models such as sensitivity of data set statistics as a function of sample size and how model parameters vary as a function of time with a given fabrication process.