A Hammerstein-Like Broadband ANN-Based Transistor Behavioral Model

This work presents a broadband artificial neural network (ANN)-based transistor behavioral model (BM). The model is based on load—pull and
$S$
-parameter measurements of a gallium nitride (GaN) high-electron-mobility transistor (HEMT) transistor over a predetermined frequency band. The incorporation of frequency dependence in the model follows a Hammerstein-inspired approach, which can be implemented and extracted using conventional ANN training techniques. The proposed model is tested in a commercial harmonic balance (HB) simulator, demonstrating its suitability for power amplifier (PA) design. The model is then validated with a sequential load-modulated balanced amplifier (LMBA) that was experimentally implemented and measured.