Open student theses
Gallium nitride (GaN)-based high electron mobility transistors (HEMTs) play a key role in modern power electronics, especially in applications such as high-frequency circuits, power converters and renewable energy systems. Precise modeling of these transistors is crucial to maximize their performance and efficiency. An established model for this is the ASM GaN HEMT model, which was developed specifically to describe the complex physical properties of GaN transistors. However, traditional approaches to parameter extraction, such as manual fitting, often reach their limits, especially for nonlinear systems such as HEMTs. Reinforcement learning (RL) offers a promising alternative by enabling adaptive, data-driven optimization strategies that could improve the accuracy and efficiency of parameter extraction for the ASM model.
Type of Thesis:
BA ❌ FA ✅ MA ✅
Relevant Experience:
- Basic knowledge of transistor modeling
- Experience with programming (ideally Python) and Machine Learning
- Interest in the application of modern AI methods in Power Electronics
Contact:
Together with the wide-bandgap semicondcutor based power transistors, such as SiC power MOSFETs and GaN power HEMTs, ceramic capacitors form the basic building blocks of fast-switching high current and voltage commutation cells and power modules in high power density switched-mode power converters. The power capacitors must follow suit with the power transistors in high-frequency capabilities, in order not to become the bottleneck for switching loss energy and thermal management, and accurate simulation models of the capacitors‘ high-frequency characteristcs must be developed for the reliable design of fast-switching power modules. Together with the wide-bandgap semicondcutor based power transistors, such as SiC power MOSFETs and GaN power HEMTs, ceramic capacitors form the basic building blocks of fast-switching high current and voltage commutation cells and power modules in high power density switched-mode power converters. The work is carried out in collaboration at University of Stuttgart, Germany, and at ENSI Caen, France. The final work programme is determined in trilateral planification between the tutors and the student and can comprise: Survey of the prevailing state of the art, Microwave characterisation of ceramic capacitors at ILH, including the design of dedicated test breadboards, Model development, implementation in VerilogA and verification at ENSI Caen, Optional: power module breadboard design and experimental validation at ILH.
Type of Thesis:
BA ✅ FA ✅ MA ✅
Relevant Experience:
- Theoretical expertise in power semiconductor devices and power electronics circuits is recommended.
Contact:
More Information:
The work is carried out in collaboration and joint tutorship with ENSI Caen, France. Parts of the work may be carried out at ENSI Caen.
Gallium Nitride (GaN) power semiconductors offer high switching speeds, low losses, and compact designs, making them ideal for next-generation brushless DC (BLDC) motor inverters. However, achieving optimal performance requires targeted filter design to enhance efficiency, improve electromagnetic compatibility (EMC), and enable soft switching.
In this thesis, suitable LC filter concepts will be developed, simulated, and evaluated to optimize the switching behavior and overall system performance of GaN-based BLDC inverters. The goal is to achieve Zero Voltage Switching (ZVS) or Zero Current Switching (ZCS) through appropriately designed resonant or commutation networks.
Type of Thesis:
BA ✅ FA ✅ MA ✅
Relevant Experience:
- Power Electronics 1
- Basic knowledge of circuit simulation and electrical measurement techniques
- Experience with MATLAB/Simulink, LTspice, or PLECS is an advantage
Contact:
Modern high-frequency and power electronics require highly optimised PCB layouts that meet both electrical and thermal requirements. Parameterised layouts make it possible to automatically generate, systematically evaluate and optimise design variants. In combination with FEM simulations and algorithmic optimisation (e.g. parameter sweeps or genetic algorithms), this creates a powerful development workflow that shortens design cycles.
The aim of this work is to develop a complete workflow that enables:
- automated generation of parameterised PCB layouts in Keysight ADS
- simulate these layouts in an FEM environment,
- evaluate the simulation results, and
- optimise the layout for a specific target using parameter sweeps or genetic algorithms.
The workflow should be reproducible, modular, and transferable to selected layout topologies.
Type of Thesis:
BA ✅ FA ✅ MA ✅
Relevant Experience:
- PE1
- RPSS1
- Good knowledge in Finite-Element Methods
Contact:
Contact
Dominik Koch
M.Sc.Group Leader Power Electronics / Research Assistant
Benjamin Schoch
M.Sc.Group Leader High Frequency Electronics / Research Assistant