Power Module Thermal Characterization Considering Aging Towards Online State-of-Health Monitoring

Semiconductor advancement towards high power density and high temperature operation call for compact design; however, this accelerates material degradation due to additional stress in high operating conditions, which leads to failure. The rate of failure of a power module (PM) is directly related to operating temperature, its' variation and distribution. As a result, temperature dependent offline reliability estimation with thermal network or finite element (FEA) based analysis has received much attention in the recent past. However, these conventional approaches only estimate degradation rating based on a constant mission profile, applied load, and initial condition of the PM, which can increase estimation error due to random load profile and aging of PM material substances. Therefore, a novel thermal network model has been developed considering the aging factor for online state-of-health (SOH) monitoring. Towards this objective, an advanced loss model is developed in a non-invasive method with semiconductors' change in electro-thermal properties with temperature. The resultant heat loss is used to track junction temperature using Cauer thermal model considering PM geometry and material properties. The model is improved with online thermal characterization in healthy and degraded conditions. Subsequently, the model has been updated with cross-coupling effect between semiconductors using FEA. The developed model is validated with an electric vehicle (EV) traction inverter module.