Statistical Method of Estimating Semiconductor Switching Transition Time Enabling Condition Monitoring of Megawatt Converters

Abstract

Detection of insulated-gate bipolar transistor (IGBT) switching transition time ( $t_{\text {tr}}$ ) is a promising temperature-sensitive electrical parameter (TSEP) for condition monitoring of IGBT power modules in converters for wind turbines. The accuracy required on $t_{\text {tr}}$ detection is in nanoseconds, which typically requires precision timing circuitry, or analog-to-digital converter (ADC) rates in hundreds of megasamples per second (MS/s). A method of calculating the statistically estimated switching transition time ( $\hat {t}_{\text {tr}}$ ) is proposed in this article. During multiple switching transitions of an IGBT, the IGBT collector–emitter voltage (vce) may be sampled with sample rates in sub-MS/s rates, yet with statistical accuracy on estimated $\hat {t}_{\text {tr}}$ in nanoseconds. A standalone converter monitoring unit (CMU), which samples vce and IGBT current ( $i_{c}$ ), was mounted in the converter of a multi-megawatt (MW) field test wind turbine. The proposed method is validated by testing it on field data from the CMU and by testing it on synthetically generated data. The proposed method is an enabler for low-cost monitoring of converters for wind-turbine field applications.