Examining the Design of Different Types of DM EMI Filters and Their Effect on EMI Noise and Control Characteristics for Cuk DC-DC Converter

Abstract

In this paper, DM EMI filters including LC, LC with damping, and Pi and LCL filters are designed for DC-DC Cuk converter operated in BCM. Although, DM EMI is so important for any power converter, the effect of DM filter is not investigated in detail in literature with respect to control characteristics and DM noise reduction. The main contribution of this paper is comparisons of DM EMI filter topologies regarding control characteristics and EMI noise reduction which is not presented in the literature. Another contribution is to give a practical design methodology for EMI filter design avoiding complex mathematical procedures. Besides, the paper designs an LCL filter, popular for inverter applications, for Cuk converter that is not introduced in the literature. Examining the filter effect on the converter from the control point of view is a very complicated issue because a linearized model of the converter which is naturally nonlinear should be obtained. To do so, the converter's small signal analysis must be derived. Therefore, for examining the effect of filters, a small signal analysis of the Cuk converter is made through an average state-space model. After that, transfer functions of each filter topology are obtained. Lastly, the entire transfer functions of the system using both the Cuk converter and each filter are acquired, separately. By using the transfer functions of overall systems, the effect of each filter topology is investigated by root locus and bode plots. Moreover, experimental verification of all filters is realized over a 15W Cuk converter with a 25kHz switching frequency. It is validated by applications that the LCL filter provides better control characteristics on the Cuk converter, the total inductor value is reduced by 30% and the total capacitor value is reduced by 25% compared to other filters. However, with respect to DM noise Pi and LC filters give better results.