Analytical Switching Loss Model for GaN Gate Injection Transistors

Gallium nitride (GaN) high electron mobility transistors (HEMTs) are increasingly adopted in highly efficient power converters. In the design and optimization of GaN-based converters, accurate analytical models with low computational effort are required for predicting the switching losses. This paper proposes a novel equivalent-circuit-based switching loss model for GaN gate injection transistors (GITs). Due to the inserted p-GaN layer and the ohmic-contact gate structure, the gate capacitance of GaN GITs depends on the gate-voltage bias, and a specialized gate drive circuit is required to maintain a leakage gate current during conduction, both of which are incorporated in the proposed model. While most conventional loss models rely on curve fitting and numerical solvers, making them device-specific and prone to convergence issues, the proposed model avoids these shortcomings by using piecewise linearization and thus achieves higher generality and reliability. The switching waveforms by the proposed model match closely with the experimental measurements, with average switching loss errors below 15% for both turn-on and turn-off over a wide operating range.