Comparison of FIR Filter Lengths and Adaptation Methods in Digital Active EMI Filters based on Delay-Compensated Digital Gate Control Signals

Modern motor inverters use steep switching waveforms to reduce losses and high switching frequencies for a high power density. This makes inverters potential sources of electromagnetic interference (EMI), which is usually reduced by large and heavy passive EMI filters. The idea of Active EMI Filters (AEF) is to reduce size and weight of the passive filter circuit. Digital Active EMI Filters (DAEF) based on FIR (finite impulse response) filter structures can achieve very high performance when used in feed-forward mode with an estimated, predicted anti-noise signal. Hereby, the anti-noise signal is calculated by a FIR filter, whose input signal is the superposition of the digital gate control signals. The time required to calculate and inject the anti-noise signal is compensated by adding a short delay to the digital gate control signals before arriving at the gate driver. In this work, a comparison of different FIR filter lengths and adaptation methods for Common Mode (CM) EMI reduction in a motor drive system is presented. A further development of the delayed digital control (DDC)-DAEF implementation is introduced, in which the filter weights are calculated iteratively using the Wiener-Hopfequation. Results are presented and discussed for a freewheeling and a mechanically loaded inverter-motor system.