An Overview to Chip-based Power Transistor and Prognostics Techniques

Power ratings and switching capability have been the main performance characteristics in the development of power transistors. To safely secure the performance reliability of its operation, a prognostics study for power transistor was introduced. The prognostics of power transistors enable the assessment of its health condition and prediction of remaining useful lifetime (RUL), given the current characteristics and loading condition. This paper presents the classification of small-signal transistor and power transistor, and the applications of power transistors. Three types of prognostics method: Model-driven, Data-driven and Hybrid method are summarized and compared. Subsequently, a new prognostics methodology for RUL prediction of power MOSFET due to active and passive thermal stress is proposed. The proposed method is based on the data-driven methodology that will utilize the characteristics of voids, and ON-State resistance, 𝑹𝒅𝒔,𝒐𝒏 as input for the proposed algorithm. The algorithm will be termed as RULPOV (Remaining Useful Life Prediction based on Voids). The proposed method is expected to improve the RUL prediction as well as to minimize the reliance on junction temperature measurement.