Turn-Off Voltage Overshoot Sensing for SiC MOSFETs Based on Electroluminescence Spectra Detection and Support Vector Regression

Sensing the turn-off voltage overshoot in silicon carbide (SiC) metal-oxide-semiconductor field-effect transistors (MOSFETs) is critically important for enhancing the reliability of converters. However, existing peak sampling measurement schemes face numerous challenges, such as high static errors and slow dynamic response. Given the correlation between the electroluminescence spectral characteristics and device operation states, this work proposes a method for sensing turn-off voltage overshoots based on spectral detection results and support vector regression (SVR), with an average sensing error of 5.01V.