Real-time digital waveform averaging with sub-sampling resolution

Abstract

Noise suppression is achieved by averaging a sequence of repetitive waveforms with correction of frequency distortions, establishing a real time processing of signals. First, the waveforms are processed seriatim, and saved in partitioned memory. Then, the memory contents are merged to form an output digital signal. Initially, an input repetitive signal is sampled with a sampling period T, and divided into K sections along the sampling period so that a ksection, where 0≤k<K, coincides with segment [k·T/K, (k−1)·T/K]. Time displacement detection determines relative positions of trigger pulses and edges of the sampling clock, and the number k of a sampling period section where the trigger pulse appears, keeping k unchanged thereafter. A resultant stream of N·K samples is transmitted through a lowpass filter, followed by decimation by K, to complete the averaging.