Research Of Non-switch Junction Temperature Monitoring Methods And Systems For Insulated Gate Bipolar Transistor

Compared with other conventional semiconductor devices,power semiconductor devices represented by IGBTs carry higher power and generate a large amount of heat during operation.The semiconductor device can ensure a normal operation within a certain temperature range.Safety problems can be caused when the junction temperature exceeds the normal temperature range,so it is necessary to find a way to monitor the junction temperature of the device online.On the one hand,it can monitor the junction temperature in time,and it is convenient to control the junction temperature before the junction temperature exceeds the normal temperature range.On the other hand,it can also help improve the accuracy of junction temperature control,which can serve for the reliability test of the device.The existing junction temperature monitoring methods are mostly switching methods,which need to switch between large current and small current,and an inevitable temperature difference will be introduced when switching the current.In this thesis,IGBT is taken as the research object,and a non-switching method for monitoring the junction temperature of IGBT is explored.Models are established to predict the junction temperature.Then it is applied to the junction temperature control in the power cycle test to provide a new solution of the reliability study of IGBT.The saturation voltage drop at high current is selected as the temperature sensitivity parameter,and the IGBT junction temperature online monitoring system suitable for high current conditions is built to obtain the relationship between junction temperature,collector current and saturation voltage drop automatically.Based on the semiconductor physics theory,the relationship between the obtained junction temperature,saturation voltage drop and collector current is analyzed.The junction temperature prediction model is established by using the polynomial model and the prediction error of the model is analyzed.The correctness of the model is verified by physical contact detection.At the same time,the neural network model is introduced into the junction temperature prediction model.The BP neural network and the RBF neural network are selected to establish the junction temperature prediction model respectively.The error analysis of the model verifies the feasibility of the neural network in the field of IGBT junction temperature prediction.The predictive effects of the polynomial model and the neural network model are statistically evaluated.The IGBT power cycle system is built with reference to the power cycle standard.The junction temperature prediction model is applied to the junction temperature control.The finite element simulation method is used to design a suitable heat sink for heat dissipation during power cycling.The thermal simulation of heat sinks of different structures are carried out to choose the heat sink with the best heat dissipation effect.The software scheme corresponding to the built power cycle system is proposed.Finally,the life prediction scheme after IGBT power cycle is proposed.