Analysis of the Influence of Measurement Circuit Asymmetries on Three-Phase CM/DM Conducted EMI Separation

Abstract

Electromagnetic interference (EMI) conducted emissions (CE) are of increasing concern in power electronics due to the high switching frequency and fast switching speeds of the latest generation of wide-bandgap semiconductors. The decomposition of the total conducted EMI noise into its common-mode (CM) and differential-mode (DM) part by means of a CM/DM noise separator is a useful tool that allows for a systematic EMI filter design. Carefully designed realizations achieve a CM rejection ratio and DM rejection ratio of 50 dB at 30 MHz. However, a very high-performance CM/DM noise separator is not sufficient. It is theoretically analyzed and experimentally proven that asymmetries in the EMI test setup result in an unwanted conversion between CM and DM EMI noise, and therefore significantly influence the CM/DM EMI separation. In particular, three main influences are identified: the line impedance stabilization network (LISN), the connection cables between LISN and the equipment under test (EUT), and the converter EMI filter. The unwanted noise conversion is pronounced for frequencies in the MHz range, where parasitic resonances occur. Experimental results show a CM-to-DM conversion of up to -30 dB at 30 MHz (a degradation by 20 dB or a factor of 10 compared to a high-performance separator alone) considering a connection cable length mismatch of roughly 5 cm. Values as high as -21 dB result when standard commercial LISNs are used for the measurement. The impact of asymmetries in the EMI filter is most severe, and clearly limits the EMI noise splitting at high frequencies. A high-performance noise separator can, however, be used to investigate such filter asymmetries (component tolerances and/or layout), and therefore helps to improve the filter design process and facilitates the modeling of EMI noise sources. [COMP: Change bullet list to numbered].