The way the TAW works is, it's intended to measure voltage and current for exactly one 60Hz period. It assumes a sample rate of exactly 1mS per sample, and therefore 16 or 17 samples will cover a 16.7mS period. Makes perfect sense. However, looking at the documented waveforms and examples, 17 samples always returns 1 and a fraction of the next 60Hz cycle which, if all samples are used for RMS calculations, will result in erratic readings depending on where the first sample starts especially with low power factor loads.
I'm seeing 12 samples per 60Hz period is sufficient and acceptably accurate since the 1mS sample rate setting is nowhere close to 1mS.
Anything more or less results in wildly erratic measurements just like these college kids were seeing, but they simply dismissed the errors as being caused by a 'low sample rate.'http://www.ee.duke.edu/files/ece/GWDD2010_Hamner.pdf