Junction Temperature Prediction Method of Discrete IGBT Based on PIN Temperature

This article proposes a non-intrusive online prediction method for the junction temperature of discrete IGBTs, based on the temperature of the device's PIN terminals to construct a thermal transmission model, addressing the limitation of traditional temperature-sensitive electrical parameter methods that require interference with circuit operation. Firstly, the linear relationship between IGBT case temperature and PIN terminal temperature was calibrated through experiments, and the consistency between devices of the same model was verified. Furthermore, the junction temperature was calculated in conjunction with the junction-case thermal resistance model. The thermal resistance values were compared using measured values (based on the transient dual interface method) and data sheet values. To validate robustness, tests were conducted under natural cooling, forced air cooling, and different thermal interface materials (with/without thermal grease), showing that the predicted junction temperature highly matches the measurement results from the temperature-sensitive electrical parameter method (average error of about 5%), maintaining good accuracy under various thermal resistance and heat dissipation conditions. This method does not require damage to the packaging or modification of the drive circuit, supports adaptive compensation for aging conditions, and provides a cost-effective solution for IGBT reliability monitoring in scenarios such as photovoltaic inverters and motor drives.