Conducted-EMI prediction for AC converter systems using an equivalent modular-terminal-behavioral (MTB) source model

This paper presents a new frequency-domain method for predicting conducted electromagnetic-interference (EMI) noise in ac switching power converters whose switching conditions vary during an operating period. Based on an equivalent modular-terminal-behavioral (MTB) frequency-domain EMI source model developed for one switching period at a given operating point, the proposed approach superposes MTB models for different operating zones in the frequency domain to predict EMI noise for the entire operating period. Compared with the various existing modeling approaches, including physics-based or behavioral time-domain and frequency-domain methods, the proposed method is efficient, accurate, and more suitable for the system-level EMI study. The use of the methodology for both differential-mode and common-mode EMI-noise prediction is investigated. Verification was carried out through simulations and experiments using a half-bridge ac converter. This modular modeling approach can be applied to many other types of converters, including three-phase pulse-width-modulation inverters, as well as converters with different switching control schemes.