During Inductive Turn Off

Insulated gate bipolar transistor (IGBT)-based pulsewidth modulation (PWM) inverters are commonly used in inductive load circuits such as motor control. During clamped inductive load turn off of the IGBT, high-power losses occur during two phases. Due to the large inductive motor load, the voltage across the IGBT rises to the bus voltage while carrying the full-rated current. In the second phase, the current decreases as the IGBT goes into its forward blocking mode. In this paper, the turn-off process during the first phase is analyzed in detail for the first time. A simple analytical model has been derived, based upon the initial steady-state minority carrier distribution, which allows predicting the rate of rise of the voltage during this time period where the collector current remains constant. The predictions of the analytical model are in excellent agreement with results obtained from two-dimensional (2-D) numerical simulations over a broad range of minority carrier lifetime values. This analytical model provides a good estimate (within 10%) of the power losses incurred during the first phase of turn off. lifetimes ( ranging from 8 to 0.5 s) indicate that the losses occurring during the first phase are a significant contributor to the total turn-off losses. The losses occurring during this period are a function of the rate of rise of voltage across the device ( ), with higher ( )'s resulting in lower losses. Previous work by Heffner et al. (3)-(5) provides an an- alytical model for the ( ) during turn off. This model is a complex function of the redistribution capacitance and displacement capacitance not easily amenable to analytical calculations. Work by Shenai et al. (7) provides a numerical solution to obtain the ( ) during turn off for hard- switching applications. In this paper, we present a simple analytical model for the first time, which accurately predicts the ( ) of the IGBT during the current boundary condition of turn off. The variation of the ( ) with different minority carrier lifetimes is predicted using this model and is in excellent agreement with the ( )'s extracted from two-dimensional (2-D) numerical simulations performed using MEDICI.