In Situ Condition Monitoring of High-Voltage Discrete Power MOSFET in Boost Converter Through Software Frequency Response Analysis

The efforts for more reliable power conversion systems have been gaining momentum in recent years. The majority of the studies concerning reliability of power switches focus on the package-related failures, mainly caused by the cyclic thermal stress. The basic failure precursor for this type of stress has been identified as increased on-state resistance for power MOSFETs in the recent literature. On-state resistance monitoring during converter operation is a challenging and costly task as it requires current and voltage sensing circuits, which can block the high voltage across the switch during off-state to protect the measurement or control unit. This paper proposes a software frequency response analysis method to determine the health status of high-voltage power MOSFETs with high on-state resistance. This is achieved by analyzing and evaluating the variation in the plant model at double pole frequency using the same DSP that is used for control purposes. The proposed concept is analyzed for boost converter; however, it can be used to detect the on-state resistance variation in other types of converters operating in continuous condition mode (CCM). The proposed algorithm is embedded in a low cost DSP and experimentally verified on a dc/dc boost converter.