Parasitic Study Of Damped DM EMI Filter And The Impact Analysis On The System Stability

As the Switching Mode Power Supply (SMPS) are used more and morefrequently in the daily life, and the high power electronic devices must meet theElectromagnetic Compatibility (EMC) standard, the EMC design has been a veryimportant part during the SMPS design. EMI filter is an effective way to suppress theconducted interference, but due to the parasitic effects, EMI filter performances willbe deteriorated in high frequency, and the conducted interference suppression will benot effective. In addition, because of the “negative impedance” feature of the filter,the system loop gain is affected and the system become instable with the input filter.In order to keep system stable and improve the performance of the EMI filter, thispaper added a damping network to a traditional π-shaped filter, extracted the parasiticparameters of the damping differential mode (DM) filter, proposed a high frequencymodel of the filter, and analyzed the parasitic effects on the filter insertion loss basedon this model.First, this paper analyzes the facts that may cause the instability of SPMS whenadding the input filter and gives the stability condition of the filter theoretically. Thepaper proposed that the EMI π-shaped filter can achieve stability by adding RCsnubber network, and simulated the output impedance and the filter transfer functionvia Matlab. The simulated results show that the RC snubber network can successfullyeliminate spikes and stabilize the filter. The paper also analyzed how differentdamping resistors affect the maximum output impedance, and proposed the method toobtain the most optimal damping resistor to make the maximum output impedanceminimum.Then, under the condition of stable system, the parasitic parameters of EMI filterare identified, and how the single parasitic parameter affect the EMI filter insertionloss is discussed. Using an inductor to replace the mutual inductance that can’t beignored in the filter, the filter model was simplified and the value of the inductor wasextracted by using finite element analysis (FEA) software. Based on the proposed filter model, the filter with measured60dB insertion loss was simulated. Simulationresults follow experimental results well, the maximum absolute error of insertion lossis less than6dB and the relative error is less than10%. The agreement betweenmeasurement results and simulation results validates the model and the method ofparasitic extraction.Finally, using the proposed high frequency EMI filter model, the effects of themutual-parasitic and self-parasitic on the insertion loss of EMI filter were analyzed,and some approaches reducing mutual coupling and improving filter insertion lossgain were provided.