Real-Time Monitoring of SiC Power Modules via Spectral-Electroluminescence Sensing

• Applicant: Professor Dr.-Ing. Rik W. De Doncker, since 9/2020
• Subject Area: Electrical Energy Systems, Power Management, Power Electronics, Electrical Machines and Drives ; Electronic Semiconductors, Components and Circuits, Integrated Systems, Sensor Technology, Theoretical Electrical Engineering
• Term: from 2020 to 2024
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 445288041

The reliable operation of power electronic converters is of great importance in many applications. This is especially true for the electric drive train of autonomous driving vehicles in which the supervision and controllability provided by a human driver does not exist anymore. Thus, a malfunction of the converter resulting from thermal overload, an undetected short circuit or excessive thermal cycling can lead to a complete failure of the vehicle and cause severe accidents. For this reason, real-time monitoring systems that supervise the operational stress and diagnose the state of health of power converters are key components in future drive trains. Their development becomes even more critical in silicon carbide (SiC) based power modules. These achieve a higher power density and are the preferable choice for highly compact drive systems in electric vehicles. However, the compactness of the semiconductors and, therefore, lower thermal capacitance makes them more vulnerable to overload.This research proposal aims to develop precise and minimal invasive real-time monitoring technologies for SiC power modules that use the electroluminescence spectrum of the SiC body diode for device junction temperature and current sensing. After characterizing the electroluminescence spectrum of the SiC body diode for a wide operation range, different design options for sensors that extract the optical spectral information in real-time are developed and experimentally evaluated. An important research question in this context is how the device current and junction temperature can be sensed independently from each other. For this purpose, system identification techniques, observers and machine learning technologies are combined to create new multi-physics monitoring solutions that obtain both variables with high bandwidth, low noise and the required precision. Finally, this research develops and investigates methods for power-module-integrated electroluminescence sensing based on fiber optics to monitor a multi-chip power module with minimized sensing effort.The envisaged research shall lead to a potent monitoring solution for SiC power modules. It potentially enables precise high-bandwidth electrothermal monitoring of future power electronics to protect SiC power modules from failures due to excessive thermal loading or overcurrents.

• DFG Programme: Research Grants
• Ehemaliger Antragsteller: Dr.-Ing. Christoph van der Broeck, until 9/2020