Precise modeling of silicon carbide-based power switches

Abstract

The Silicon Carbide Metal-Oxide-Semiconductor Field-Effect Transistor (SiC MOSFET) is a new wide-bandgap semiconductor device that has high working temperature, high breakdown voltage capabilities, and low on-resistance. This paper presents an in-depth study of SiC MOSFETs and provides a detailed evaluation of their switching performance.

First, the industrial background and research significance of SiC MOSFETs are introduced, summarizing their development history and current research status, including how to choose suitable simulation models to assess power devices' switching capability. Then, a specific device is chosen, and its static and dynamic properties are thoroughly examined, including output characteristics, transfer characteristics, and factors affecting the on-state resistance. Based on this analysis, a semi-physical model of the SiC MOSFET is built using Saber to obtain fitting curves of its relevant characteristics.

Subsequently, utilizing this model, LTspice is used to do simulations based on a double-pulse test circuit. The switching performance of the module is examined in relation to changes in the parameters of parasitic inductance and gate resistance. A comparative study is conducted on the changes in SiC MOSFET switching performance under these parameter variations, analyzing the impact of different parasitic parameters on the switching waveforms. The experimental results demonstrate that parasitic parameters not only reduce the device's switching speed and increase losses during the turn-on process but also cause oscillations, voltage and current overshoot, and crosstalk issues in the circuit.

Finally, the principles underlying the main circuit switching oscillations and the driver circuit oscillations and crosstalk issues caused by the high switching speed of SiC MOSFETs are analyzed. Methods to mitigate the impact on SiC MOSFET switching transients by selecting appropriate gate resistance and reducing parasitic parameters are explored to optimize the simulation.