A Double Pulse Test Based IGBT On-State Capacitance Extrac-tion of ANN-assisted Hybrid Model

Abstract:
An accurate semiconductor IGBT model is crucial for application power design, significantly influencing overall system performance, particularly in electrical switching behavior. We introduce an Artificial Neural Network (ANN) body diode model to capture reverse recovery current overshoot and present an improved Angelov-based IGBT model to extract on-state reverse capacitance (Cres) using Double Pulse Test (DPT) results, overcoming hardware power compliance limits in the on-state. An automatic parameter extraction workflow with an AI/ML optimizer efficiently covers device transient switching behavior under multiple bias conditions. Comprehensive fitting of transient current and voltage waveforms is verified for both switch-on and switch-off, with currents ranging from 30 to 600 A and voltages from 100 to 820 V.