Selecting the right EMI/EMC filter is a major challenge when system noise levels exceed compliance or pre-compliance limits. Inline PCB filters are designed to mitigate noise in standalone conditions, but their behavior changes when integrated into a larger system due to unknown parasitics. These parasitics can disrupt electromagnetic compatibility (EMC), leading to non-compliance. To address this, engineers often use off-the-shelf EMI filters, but determining their real-world effectiveness remains complex. Even with simulation-based methods, accurately predicting insertion loss and attenuation is difficult due to limitations in conventional modeling approaches.
Traditional SPICE-based simulations rely on static models defined at specific frequency points, with interpolated values for intermediate frequencies. This interpolation introduces inaccuracies, affecting the precision of simulated results. To overcome these limitations, we propose a methodology that reconstructs a realistic EMI/EMC filter model based on insertion loss characteristics under symmetrical and unsymmetrical conditions.
Our approach involves deconstructing the EMI/EMC filter into its subcomponents—X-capacitance, Y-capacitance, common mode choke (CMC), busbar, and PCB traces—and parameterizing them using CST simulations. Instead of relying on physical measurements, which are prone to parasitic influences, we extract subcomponent values from datasheets. We focus on dominant parasitic elements exceeding 1 pF and 1 nH, as lower values predominantly affect GHz-range frequencies rather than the MHz-range compliance limits. By analyzing the impact of parasitics on insertion loss, resonance, and damping characteristics, we develop a filter model that accurately represents real-world behavior.
This methodology results in a high-fidelity EMI/EMC filter model with minimal deviation from actual performance. It enables precise pre-compliance conducted emissions simulations and facilitates optimized filter selection and tuning based on system-specific noise conditions.