NOISE-REDUCTION IN ZERO-CROSSING DETECTION BY PREDICTIVE DIGITAL FILTERING

A simple combination of nonlinear and linear digital signal processing methods is proposed for efficient noise reduction in zero crossing de tectors. The method is very robust against strong impulsive noise, typ ically encountered in thyristor power converters, where reliable zero crossing detection is required for firing synchronization. A systemati c design procedure is described for the proposed filter-based synchron ization method, taking into account the specified line frequency toler ance. The fully digital signal processing approach allows compact impl ementations, and supports flexible interfacing to digital motor contro l systems.