FPGA-based Adaptive Notch Filters for the Active Cancellation of Varying Electromagnetic Emissions of Power Electronic Inverter Systems

In many modern power electronic systems, fast-switching semiconductor devices are used to reduce switching losses. Due to steep switching waveforms and high switching frequencies, significant electromagnetic disturbances can be emitted. In contrast to conventional passive filter components, active cancellation methods are based on the controlled destructive interference between a noise signal and a corresponding anti-noise signal. Adaptive notch filters revealed themselves as a promising active EMI cancellation concept for periodic noise signals. In this work, adaptive notch filters are regarded in context of slowly time-varying periodic noise signals, e.g. pulse-width modulated signals as common in inverters. The corresponding noise signals consist of switching harmonics and adjacent sideband harmonics. Therefore, the notch filter’s bandwidth comes into focus and an analytical approximation for the ideal adaptive notch filter’s bandwidth is discussed. With help of this approximation, the adaptive notch filter can be parametrized specifically to a given noise spectrum and other requirements. The capability of the parametrization strategy and the adaptive notch filter itself are shown by simulation and measurement.