I have a sine wave plot of 1 Hz of amplitude 5V peak to peak.

The waveform is not clean and has lots of glitches that have a amplitude of say +/- 0.5V.

I am sampling this waveform at a rate of 2.5KHz and acquiring about 125 samples in a scan loop that is firing once every 50ms.For a 1 sec period I will have 2500 samples.

I thus have enough data for analysis. Question is : If I need to represent the 1 Hz sine wave truly, stripping all the noise glitches what would be the best way to do it ?

I have tried with a simple mean value of the 125 samples but there is an inherent offset and reduction of amplitude. Would something else like a median or moving average be better ?

Or a discrete function that compares the current sample to the previous sample and only records if the difference is more than a specified magnitude will be OK ?

( My waveform is not a pure sine, but just took the example to explain the case )

Thanks

raghu