Adaptive signal processing system for accurate zero-crossing detection of cycloconverter phase currents

We introduce a multistage adaptive digital filtering system for processing of cycloconverter phase current signals. In cycloconverter drives, the zero current periods, related to changing of thyristor bridges, cause torque distortion, which impairs the performance of speed control. The proposed filtering system is used for accurate zero-crossing detection in order to shorten the mandatory zero current periods. Our filtering system combines a low-pass prefilter and an adaptive predictive filter making it possible to extract the sinusoidal primary signal from noise and disturbances without harmfully phase shifting the fundamental component even when the frequency or amplitude alters rapidly. The predictive filtering is performed at time-varying sampling rate, so that the relative frequency of the fundamental component remains almost constant.