Project Details
Projekt Print View

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
 
Final Report Year 2023

Final Report Abstract

The primary objective of the project ElluSense was to investigate new real-time monitoring technologies for power modules that utilize the electroluminescence (EL) emitted by SiC MOSFETs. The EL occurs during conduction of the internal body diode and is caused by radiative recombination at the p-n junction. A typical spectrum of the EL consists of two major peaks in the visible spectrum that exhibit distinct current and temperature sensitivities. By measuring the EL and extracting these sensitivities, the project ElluSense aimed to enable high-bandwidth current and temperature monitoring of power electronic modules. Within the scope of the project, the EL spectrum of several SiC MOSFETs were characterized and modeled over a wide range of currents and temperature as well as certain degradation effects. Three spectral sensitivities were identified: opposing temperature sensitivity of the two major peaks, uniform current sensitivity of the entire spectrum and a temperature-dependent redshift of the spectrum. To utilize these EL sensitivities for the proposed monitoring approach, a multi-sensor circuitry was developed that measures the spectral characteristics of the EL by means of three photodiode sensors and optical filters. The photodiodes were interfaced with a sensing circuitry that amplifies and samples the photocurrent with high dynamics. As the EL typically occurs only during the dead time of power electronic converters, the developed circuitry is able to measure light pulse duration of about 500 ns. To extract current and temperature information from the EL measurements, the project researched various signal processing algorithms. It was shown that artificial neural networks are particularly suitable for this purpose as they provide easy calibration and high accuracy while requiring only little complexity and computational effort. Moreover, the project investigated methods to integrate the sensing approach into power modules. The power modules were equipped with optical fibers to capture the EL from the SiC dies and transfer the light to the photodiodes-based sensing circuitry. Using optical fibers provides intrinsic galvanic isolation and high immunity to electromagnetic interference as the sensing circuitry can be spatially separated from the power module. A 3D-printed, customized housing ensured correct alignment of the optical fibers as well as reproducible measurement conditions. Finally, the EL-based condition monitoring concept was validated by means of real-time measurements on an industry-standard SiC power module, showing that current and temperature can be measured simultaneously with high bandwidth and accuracy.

Publications

 
 

Additional Information

Textvergrößerung und Kontrastanpassung