LAUSR

Averaging techniques are frequently applied when sampling quasi-periodic signals to enhance signal fidelity. This article investigates the use of spectral envelope preprocessing for evaluating power consumption and harmonic content drawn by ac loads. A statistical characterization of a preprocessor system’s resolution given analog input noise is derived. Resolution is defined here as the ability to detect transient step changes in power consumption. With these results, the concept of effective number of bit (ENOB) is extended to this averaging system. Design trade-offs in the number of quantization bits, sampling frequency, noise level, and minimum resolvable transient are explored. The conclusions reached here generalize to any system in which spectral envelopes are extracted.