Modeling and Mitigation of EMI from Near-Field Coupling Effects in Front-End Power Supplies

Abstract

Switch-mode power converters are essential in many emerging and rapidly evolving industries including renewable energy, electrified transporation, telecommunications, and datacenters. Increasing performance demands in these markets have driven significant advancements in the development of high-density, high-efficiency power supplies. Many of these advancements have been enabled by the widespread commercialization of wide-bandgap (WBG) power semiconductors. The improved performance of WBG devices over traditional silicon devices has enabled significant increases in switching frequency and power density without compromising power efficiency. However, some challenges remain which complicate high-density converter design. Electronic products are required to meet certain electromagnetic compatability (EMC) standards, yet EMC is often not addressed until late in the design cycle through trial-and-error methods. Even when EMC is considered earlier, models can underestimate the levels of electromagnetic interference (EMI) observed in hardware testing. These discrepancies are often caused by parasitic electromagnetic couplings within the hardware, which create additional propagation paths for high-frequency noise that may not be attenuated by the EMI filter. This dissertation proposes a framework to model and mitigate EMI from capacitive and magnetic (near-field) coupling effects within front-end AC/DC power supplies. Equivalent noise models are developed, and the key couplings that most significantly impact conducted EMI are identified. Three-dimensional (3D) finite-element analysis (FEA) simulations are used to predict the values of these couplings based on the converter's layout and mechanical design. By combining analytical and FEA models, the conducted EMI spectra can be accurately predicted before building a hardware prototype. The most significant near-field coupling effects are caused by inductive (magnetic field) and capacitive (electric-field) interactions between the high-frequency switching power stage and the passive EMI filter. These couplings can increase both the common-mode (CM) and differential-mode (DM) conducted noise. Therefore, separate CM and DM noise models are developed to model both the magnetic and capacitive coupling effects. Existing models are often incomplete or rely on physical hardware measurements to fully characterize the conducted noise. The combination of anaytical and FEA models in this dissertation enables quantitative EMI prediction based on the physical arrangement of electrical components and mechanical structures. This allows EMC to be optimized alongside electrical, thermal, and other system requirements earlier in the design cycle. There are many techniques which can be used to mitigate parasitic couplings and reduce the conducted noise. In this dissertation, the proposed EMI modeling framework is used to describe the mechanism and effectiveness of various mitigation techniques including shielding, magnetic design, component rotation, and coupling cancellation. Additionally, a novel integrated shield structure is proposed for PCB-winding magnetics. This structure is applied to a PFC inductor to contain its electric field and reduce the converter's CM noise by as much as 28 dB, without causing any significant increase in converter loss. The modeling and mitigation strategies are first demonstrated using a relatively simple bridgeless PFC converter in order to facilitate an easier understanding of the concepts. Then, these techniques are applied to a two-channel interleaved bridgeless PFC using a coupled inductor structure. The EMI modeling and mitigation framework proposed in this dissertation can ultimately be extended to predict and reduce the conducted noise of any front-end power supply with EMI filter.