Characterization, Modeling, and Compensation of the Dynamic Self-Biasing Behavior of GaN HEMT-Based Power Amplifiers

Charge-trapping phenomena in radio-frequency (RF) power amplifiers (PAs) based on GaN high-electron-mobility transistor (HEMT) technology are understood to be responsible for the dynamic self-biasing behavior that leads to a seemingly intractable slow dynamic residual nonlinearity in communications applications. For this reason, and based on recent developments in the characterization and modeling of charge-trapping phenomena, in this article we demonstrate how the dynamic self-biasing behavior of GaN HEMT-based PAs can be characterized, modeled, and compensated. First, we describe a method for the accurate characterization of the capture and emission dynamics of charge-trapping phenomena using transient two-tone large-signal RF measurements. Then, we demonstrate that the accurate modeling of these phenomena is contingent on the capture process being described by a state-variable time constant, rather than a fixed near-instantaneous time constant as is typically assumed. Finally, we propose a fully analog electronic circuit that implements an approximation of the Shockley;Read;Hall (SRH) statistics-based physical model of charge trapping to compensate the dynamic self-biasing behavior of a 15 W GaN HEMT-based PA.