Global modeling of GaN HEMT resistive current including charge trapping and self-heating for multi-bias multi-class PA performance prediction

Abstract

An empirical Gallium Nitride (GaN) HEMT model, suitable for multi-bias and multi-class power amplifier (PA) performance prediction, is formulated. In addition to the fast dynamically-nonlinear capture mechanisms normally considered for local modeling, dynamically-linear charge trapping is taken into account here. A straightforward empirical identification procedure based on tailored double-pulsed IV measurements is described. Validation experiments carried out on a 8×125 pm (gate length: 0.25 pm) GaN-on-SiC HEMT show good model prediction capabilities under different drain bias conditions and class AB, B, and C large-signal PA operation at both low-frequency (f = 4 MHz) and RF (f = 2.5 GHz).