…and it’s Almost Always Line Noise
50/60 Hz line sources abound in conductive and radiated states to create noise in your most carefully-crafted measurements, but sometimes it isn’t obvious that they’re the culprit. At low sample rates line noise can morph into alias frequencies that bear no resemblance to their true origin. This can hide an otherwise obvious noise source and send you off chasing ghosts. The image on the left in the nearby graphic in an example of this. Sampled at a rate of 23 Hz, the signal obviously suffers from noise intrusion, but it isn’t clear that 60 Hz conducted noise is the source. To test that possibility, all that’s necessary is a simple adjustment to sample rate so it’s equal to the line frequency. As shown by the image on the right, the noise disappears like magic to reveal true transitions in the underlying signal. What happened?
By setting the sample rate equal to the line frequency we ensure that analog-to-digital conversion occurs at precisely the same time slot on each cycle of the 60 Hz noise waveform, essentially cancelling its affects. The same technique will work for 50 Hz line noise by setting sample rate to 50 Hz. Note that the noise isn’t eliminated when you apply this approach. It’s still there wreaking havoc with your signal-of-interest, but synchronizing ADC conversions to it temporarily masks its affects. You’ve only proven that the source of the noise is line-related so you can begin tracking down the real cause of the problem.
Before you apply this technique to your next noisy measurement remember that you have to set sample rate equal to line frequency per measured channel. If you have four channels enabled and you suspect 60 Hz noise, you have to set throughput rate to 240 Hz to achieve the desired result. If the required multiple isn’t possible, then disable or enable channels as required so the sample rate per channel is precisely equal to line frequency.