Prediction of device temperatures in an electric vehicle battery charger system by analysis of device thermal cross-coupling

This paper presents a practical implementation of the measurement and use of thermal cross-coupling data. Thermal cross-coupling, the thermal transfer impedance between the heat generation of one power device and the temperature of another, is characterised using pseudo-random binary sequences for a typical H-bridge arrangement. Knowledge of the cross-coupling between each pair of devices allows temperature profiles to be predicted for arbitrary input power waveforms. The technique is verified by applying arbitrary power waveforms to all devices in the experimental arrangement and measuring the temperature response. Good agreement is shown between predicted and practical device temperature waveforms, demonstrating the viability of the technique.