Digital (DL)/sup N/ shaping with reduced quantization error penalty

Abstract

AbstraclMultiple delay line shaping (DL)N is a powerful method for implementing weighting functions WF characterized by a time-limited pulse response. Two major problems are encountered in the implementation of such shapings: quantization error and lack of flexibility in of the WF parameters once the analog prefilter is chosen. Methods aimed at eliminating such problems are discussed.For what concerns the quantization error, new WF allow the use of A/D convertem with two or even three bits less than those necessary in previous implementations. This is obtained with little penalty on the required A/D speed.WF with durations variable on a 8:l scale are also shown to be possible with simple analog prefilters made only of three real poles; flexibility is obtained by including only three switches in these prefilters.