An automated SPICE modeling procedure utilizing static and dynamic characterization of power FETs

This paper presents a procedure for autonomously generating behavioral SPICE models for wide bandgap power FETs using static and dynamic device behavioral data. The procedure applies a modified exhaustive corner search optimization algorithm to tune model parameters. This work utilizes parameters adopted from the Berkeley Level-3 MOSFET model to match static device characteristics, though the proposed methodology is modular and capable of accepting any set of model equations. Dynamic behavior optimization is performed using function calls to LTSPICE as an objective function in an optimization loop. The process presented yields a computationally-compact, automatically generated model which can be used to predict switch-mode behavior of various types of FETs in power electronics applications.