Multidimensional Characterization of Temperature Sensitive Electrical Parameters in SiC Mosfets using Automation of Measurement Data Analysis

An automation program extracting and analyzing experimental data to obtain multidimensional dependencies of 15 temperature sensitive electrical parameters that characterize the thermal and health state of SiC MOSFETs is developed. Experimental data is collected on a 50 kW three-phase AC power cycling setup. Experiments are carried out on two SiC MOSFET 120 $A$, 1200 V modules from the same batch, with one module being new and the other aged in 3 years of operation. About 30,000 inline measurements of 15 TSEPs for two modules are obtained as a result of the automatic analysis. Turn-on and turn-off delay times, threshold voltage, and peak gate current are found to be the most affected by aging in the investigated devices. Experimental measurements of the TSEPs have been compared with manufacturer's data to validate the automated method. The analysis automation program makes it possible to reduce the characterization time of TSEPs under different conditions, to minimize the engineer's involvement to a minimum in this process, and to eliminate human errors in repetitive actions during data analysis. The resulting dataset of extracted data can be used in further analysis using a neural network to find correlative relationships between TSEPs and use their combination to improve state-of-health prediction.