We all deal with sampled data from data acquisition systems, but there is one thing that must be done to the data before it is sampled in order to ensure its integrity. The penalty for noncompliance is not just noise or distortion or a scaling error; it is much more fundamental than that. The values of the time-domain samples can become systematically corrupted in such a way that frequencies and the results of FFT-analysis can be convincingly misleading. This article deals primarily with aliasing - one of the implications of sampling - but it also touches many other practical aspects of digital signal processing (DSP), including the use of Delta-Sigma converters for data acquisition.
展开▼
