If I have a complex periodic signal I want to characterize assuming an irregular peak amplitude (within 20% deviation from peak value) and constant frequency, how badly will including an additional partial period affect the result?
To be more specific, if I have discrete data which is sampled at 400 times the Nyquist frequency, it is comprised of 20.5 periods (for example) and I have used that to calculate my RMS value discretely (assume basic rms equation for discrete data in time domain, $\sqrt{\frac{\sum_{i=1}^n x_{i}^2}{n}}$); is there a significant deviation in RMS results from a dataset that has exactly 20 periods, and does that scale with period count and sample density?
I have attempted to estimate this functionally by generating sine waves over full and partial periods and taking the rms values, but received the same value regardless of periodicity. I have difficulty believing the same principle would apply to a discrete irregular signal, however, and would like outside input.