SiC MOSFET devices are increasing in popularity in high-power converter applications. Device on-resistance (Rdson) is an important indicator for SiC MOSFET health status. Large increments in Rdson are indicative of device failure and decreased system efficiency. Directly measuring device Rdson in high power applications is difficult and little work has been done on predicting Rdson. In this work, data collected from accelerated lifetime tests of high power SiC MOSFETs are used to train and test machine learning regression models to predict device Rdson from thermal cycle count and instantaneous temperature. The developed models, coupled with cycle counting algorithms, and device case thermal measurements, provide an accurate live estimation of Rdson and may be used to predict changes in Rdson over expected mission profiles during power converter design.
