ABSTRACT An accurate simulation model can guide applications such as loss estimation and key parameter evaluation for power devices. Existing SPICE‐compatible models suffer from compromised accuracy‐efficiency trade‐offs, inadequate characterization of temperature effects, and convergence limitations. To address the limitations of conventional behavioural models, which often require numerous parameters and exhibit poor extrapolation capability, this paper proposes a novel behavioural model utilizing a tanh(x)‐based channel current expression. This formulation not only inherently ensures smoothness and continuous differentiability, mitigating convergence issues, but also significantly reduces the number of core characterization parameters to just five. A stepwise parameter extraction method is given via Levenberg–Marquardt optimization to effectively prevent overfitting‐induced spurious points in the output characteristics during multiparameter fitting. Recognizing the critical impact of temperature on SiC MOSFET performance, temperature effects are embedded through second‐order polynomial fittings across the full operational range. In the meantime, temperature effects on critical parameters during switching transients are considered through theoretical analysis. Experimental validation via double‐pulse tests across a wide temperature range confirms the model's high fidelity, with static characteristics deviation below 3%, switching loss error within 8%, and transient oscillation discrepancy under 2%, demonstrating its value for precise simulation in SiC‐based converter design.
Ye et al. (Thu,) studied this question.