An accurate neural network model of FET for intermodulation and power analysis

Abstract

An accurate nonlinear FET model based on a Neural Network representation of the drain current source has been developed to predict intermodulation distortion as well as output power performance. This new neural network model has been implemented in a commercial harmonic balance simulator and its efficiency is evidenced by a comparison with an empirical model (Tajima). The accuracy of the proposed model is verified by active load-pull measurements on a Texas HFET at 10 GHz.