Nonlinear HEMT Modeling Using Artificial Neural Network Technique

Abstract

An improved nonlinear modeling technique for high electron mobility transistors (HEMT) based on the combination of the conventional equivalent circuit and artificial neural network (ANN) modeling techniques is presented. Effective initial values of the artificial neural network for each nonlinear element in HEMT model are evaluated from a semi-analytical parameter extraction technique. A multi-goal DC, S-parameter, and harmonic (DC/S/HB) training process has been formulated. Good agreement is obtained between the model and data of the DC, S parameter, and harmonic performance for a 200um gate width 0.25μm PHEMT (FHX04LG) over a wide range of bias points.