Micromagnetic Modeling and Sensor Design for High-Frequency Power Electronics
Date Issued
May 2026
Author(s)
Sohid, Sadia Binte
The University of Tennessee
Advisor(s)
Leon M. Tolbert
Additional Advisor(s)
Daniel Costinett
Gong Gu
Helen Cui
Abstract
The rapid advancement of wide-bandgap semiconductor devices has enabled high-frequency power electronic converters with substantially improved power density and efficiency. However, these advances have also introduced critical challenges in magnetic component design and current sensing, which remain major bottlenecks to achieving optimal system performance. In particular, accurate modeling of nonlinear magnetic materials and the development of high-bandwidth, compact current sensors are essential for next-generation power electronics operating in the hundreds of kilohertz to megahertz range.
This dissertation addresses these challenges through two complementary research thrusts: (1) the development of high-bandwidth current sensing solutions for double-pulse test applications, and (2) the formulation of a physics-based multi-domain modeling framework for nonlinear magnetic materials.
First, a combined Rogowski coil-based current sensing architecture with enhanced direct current measurement capability is presented. By integrating inductive sensing with a direct current sensing mechanism through frequency-domain signal fusion, the proposed approach overcomes the inherent bandwidth limitations of conventional sensors while maintaining non-invasive operation and a compact form factor. The design methodology is validated through circuit-level simulations and experimental measurements, demonstrating accurate wideband current reconstruction under high-speed switching conditions.
Second, a physics-based circuit modeling framework is developed to capture the nonlinear behavior of magnetic materials. Starting from the Landau-Lifshitz-Gilbert equation, a generalized equivalent circuit model is proposed to represent magnetization dynamics at the domain level. Building on this foundation, a simplified spherical Landau-Lifshitz-Gilbert formulation is introduced to reduce computational complexity while preserving physical accuracy. The model is further extended to a multi-domain system by incorporating domain-wall coupling and magnetostatic interactions, enabling the prediction of hysteresis behavior and core loss under dynamic excitation.
The proposed modeling framework is validated against experimental data for manganese-zinc ferrite materials, demonstrating strong agreement in hysteresis characteristics and core-loss estimation. Compared to conventional empirical and semi-empirical models, the presented approach offers improved interpretability and scalability by directly linking material-level physics to circuit-level behavior.
Overall, this dissertation establishes a unified framework that bridges sensor design and physics-based magnetic modeling for high-frequency power electronics. The results contribute to the advancement of compact, efficient, and reliable power electronic systems by enabling accurate current measurement and predictive magnetic modeling under realistic operating conditions.
Degree
Doctor of Philosophy
Major
Electrical Engineering
Embargo Date
May 1, 2027