一种基于物理机理的SiC MOSFET与GaN HEMT变换器通用开关过程预测简易模型

A simple and physically insightful model for predicting the switching transients of SiC MOSFETs and GaN HEMTs in power electronic half-bridges is proposed in this research. The proposed model has a hybrid structure: It combines an artificial neural network (ANN) based prediction of device currents and capacitances with a state-space model representing the dominant parasitics of the switching cell and gate drive circuitry. The ANN-based device model facilitates the representation of different devices in a simple model structure, which is demonstrated using both, SiC MOSFETs and GaN HEMTs. The state-space model is derived based on state-of-the-art model equations of a switching cell but features a unique physically insightful format that decouples capacitive and inductive system dynamics. The few explicit equations that make up the model facilitate its implementation in simulation algorithms and code and can contribute to fast simulation speed. In addition, the model format allows the derivation of a state-block diagram that, for the first time, represents the entire system dynamics of a half bridge, including all cause-and-effect relationships and state cross-couplings. For SiC MOSFETs and GaN e-mode HEMTs, it is shown that the model prediction, after parameterization with derived parasitics and static device data, e.g. from data sheets, curve tracer or TCAD data, matches the switching transients of double-pulse measurements with small deviations. On the basis of the switching prediction, a procedure is proposed for the generation of a system-level behavioral loss model for half-bridges in hard, partially-hard and soft switching operation. This can be rapidly computed and applied in system-level converter simulation, thus linking the switching cell and system-level design domain. Overall, the methodology presented in this paper enables fast and effective analysis of different devices in different packages and system environments, which is critical for data-driven design of future converters.