GaN power amplifier design based on artificial neural network modelling

Abstract

GaN field effect transistors (FETs) have a strong potential for high-power applications. However the RF performance of these devices often experiences limitation due to trapping effects and self-heating. These complicate the development of accurate large-signal models for GaN FETs. To simplify this process, a state-space modelling technique using an artificial neural network (ANN) is used in this work to model the large signal behaviour of the GaN device. In this way, the model is constructed directly from large-signal measurement data collected while the device is in an operating mode close to its application, i.e., class AB power amplifier (PA). To demonstrate the approach, a hybrid power amplifier based on GaN FETs was designed and fabricated. The good agreement between measurements and simulation results verifies the proposed approach. It is the first time that this modelling approach is used in circuit design.