Power Module Packaging Insulation Aging Based on Partial Discharge Multispectral Characteristics Under Square Wave Voltage

Partial discharge (PD) under high dV/dt square wave voltage is one of the main causes resulting in power module packaging insulation aging and failure. However, the evolution of PD during aging and the mechanism of its effects on package insulation are still unknown. The optical spectral distribution of PD could characterize various microscopic physical processes, whereas the topic is seldom investigated for power module packaging insulation. Therefore, this paper focuses on the PD multispectral characteristics of the power module packaging insulation during PD. A multispectral PD detection system based on the silicon photomultiplier (SiPM) is established. The multispectral signals of PDs during aging at different temperatures and square wave voltage frequencies are investigated. The results indicate that PD mainly produces optical signals in the near-ultraviolet (UV) band, and the magnitudes of the visible and IR lights are lower. The aging process can be divided into initial, developing, and late stages based on the generation situation of long-wavelength visible and near-infrared (IR) optical signals. In the initial stage, UV and near-UV visible lights are generated. In the developing stage, after the formation of the gas cavity, long-wavelength visible and near-IR optical signals are generated. In the late stage, the magnitude of each band increases, and the signal magnitude and generation probability of the long-wavelength visible and near-IR optical signals increase significantly until the insulation breakdown and failure. The average magnitude of light signals in all bands increases with rising temperature, and the generation probability of photons in the B3 and B4 bands increases significantly. At higher voltage frequencies, photons in the B3 and B4 bands are more likely to be generated. The evolution of multispectral characteristics can act as helpful indicators for packaging insulation diagnosis.