A Postprocessing-Technique-Based Switching Loss Estimation Method for GaN Devices

Gallium nitride (GaN) high electron mobility transistor (HEMT) is a promising candidate for the high-density power converter applications. Due to the low switching loss, the GaN HEMT may lead to a new horizon in the applications, such as fast charger, wireless charging, and 5G power amplifier. However, the fast switching speed of GaN HEMT makes it extremely sensitive to parasitic parameters, especially for the low-voltage GaN devices with small packages. Conventional switching loss estimation methods, which are normally based on the calibration fixture or modification of test circuit, cannot be effectively utilized for GaN HEMT. This article presents an accurate switching loss estimation method using the postprocessing technique. The proposed switching loss technique consists of three parts of signal processing, which are the wavelet denoising process to minimize the influence of background noise, the voltage-current (V-I) alignment process to reduce errors from probe propagation delay, and the linear interpolation process for further improvement of alignment accuracy. In addition, a modified SPICE model is proposed to investigate the influence of parasitic parameters on switching losses by the circuit simulation. Based on the theoretical switching characteristics, the switching loss estimation method is finally validated experimentally on a commercial 40-V/10-A GaN HEMT in a double-pulse test.