Open student thesis
Simulation-to-Measurement Correlation and ML Model for Bond-Wire Condition Monitoring in SiC MOSFETs
Bond-wire degradation is a major failure mechanism in power MOSFETs. Time Domain Reflectometry (TDR) enables non-destructive detection of bond-wire lift-off and degradation, but real datasets often cover only limited scenarios. In this work, a CST or ANSYS simulation model will be developed to generate TDR responses for defined bond-wire defect states and compared with experimental measurements. Building on this, a machine learning model will be designed and trained to classify bond-wire states, combining simulated and real data to enhance robustness and generalization of condition monitoring approaches.
Type of Thesis:
BA❌ FA ✅ MA ✅
Relevant Experience:
- Proficiency in CST Studio
- Knowledge of power electronics
- Knowledge of signal processing and machine learning
Contact
The emergence of Wide Bandgap (WBG) semiconductors such as Gallium Nitride (GaN) and Silicon Carbide (SiC) has enabled significant improvements in switching speed, efficiency, and power density in power electronic systems. Their robustness under extreme conditions is strongly influenced by their behavior during short-circuit events. This thesis focuses on the repetitive short-circuit (RSC) testing of GaN and SiC devices. In addition to acquiring fundamental knowledge of WBG semiconductor physics and test methodologies, the thesis involves the software extension and the integration of advanced thermal management in the existing test bench to simulate realistic operating conditions. The aim is to characterize device limits under repetitive stress and to investigate degradation mechanisms.
Type of Thesis:
BA❌ FA ✅ MA ✅
Relevant Experience:
- Basic knowledge of power electronics and semiconductor devices
- Interest in experimental work and data analysis
- Independent and structured approach to work
- Prior knowledge of microcontroller programming desirable
Contact:
Increasing switching frequencies are placing ever greater demands on signal generation and processing for power electronics applications.Higher switching frequencies allow the use of smaller passive components, but require faster control systems. In order to realize such a control system, a control and actuation system is required that can cope with the increased requirements. In order to be able to to operate at these elevated frequencies, microcontrollers no longer offer sufficient performance. Thus, an FPGA based alternative is to be developed based on the Diligent Cmod A7 35T evaluation board.
Type of Thesis:
BA ✅ FA ✅ MA ✅
Relevant Experience:
- Control Therory
- Modulation schemes for power electronics
- Programming skills ideally in connection with FPGAs desirable
Contact:
This thesis aims to systematically investigate the aging behavior of SiC MOSFETs. To this end, control strategies will first be developed and implemented that allow the temperature rise to be controlled on the basis of TSEP measurements. Subsequently, static load cycling tests with a constant temperature rise will be performed. The aim is to identify aging mechanisms, quantify their influence on electrical parameters, and derive possible correlations between load profile and degradation.
Type of Thesis:
BA❌ FA ✅ MA ✅
Relevant Experience:
- Basic knowledge of power electronics and semiconductor devices
- Interest in experimental work and data analysis
- Independent and structured approach to work
- Prior knowledge of microcontroller programming desirable
Contact:
This work aims to improve the accuracy of temperature estimation in power semiconductor devices using Temperature-Sensitive Electrical Parameters (TSEPs), which are often affected by noise, anomalies, and drift due to device degradation. The project will investigate advanced techniques for denoising, anomaly detection, and drift correction to enhance estimation accuracy with minimal sensor inputs. Key tasks include selecting and evaluating suitable denoising and anomaly detection algorithms for TSEP-based temperature estimation, comparing aspects such as accuracy, response time, and algorithm complexity. An additional goal is to minimize the number of TSEPs needed for accurate junction temperature estimation using machine learning methods.
Relevant Experience:
- MATLAB or Python
- Understanding of power electronics
- Knowledge in signal processing and machine learning
Contact:
This project focuses on developing a driving simulator connected to a digital twin of a three-phase inverter. The simulator enables real-time interaction with the inverter model, allowing analysis of system behavior under different driving conditions. By integrating the digital twin, it becomes possible to monitor aging effects of the transistors, optimize performance, and implement predictive control strategies. This approach supports improved reliability, extended lifetime, and smarter energy management in power electronic systems.
Type of Thesis:
BA ✅ FA ✅ MA ✅
Relevant Experience::
- Experience with MQTT communication protocols for real-time data exchange
- Proficiency in MATLAB/Simulink for modeling and simulation of dynamic systems
- Knowledge of power electronics and control of three-phase inverters
- Familiarity with digital twin concepts and predictive maintenance strategies
Contact:
Contact
Dominik Koch
M.Sc.Group Leader Power Electronics / Research Assistant
Benjamin Schoch
M.Sc.Group Leader High Frequency Electronics / Research Assistant