Spurious peak

Having problems ? let us know the details here
Post Reply
jcclq
Active User
Active User
Posts: 12
Joined: Thu Jul 22, 2021 1:05 pm

Spurious peak

Post by jcclq »

Hi all

I have a spurious peak on my FFT of the signal from a light beam. The peak looks optical because when I block the beam the peaks goes away. Strangely its frequency depends on the frequency range. I also checked with a spectrum analyser, and the peak is not there.

It is probably related to the way I am doing the measurement, but I don't what is wrong. I attached some example files with settings.

Any help will be much appreciated! Thanks a lot!

JC
Attachments
picoscope settings 199kHz.pssettings
(3.33 KiB) Downloaded 165 times
picoscope settings 20kHz.pssettings
(3.32 KiB) Downloaded 140 times
20kHz screenshot.png
199kHz screenshot.png
picoscope settings 199kHz.pssettings
(3.33 KiB) Downloaded 176 times
picoscope settings 20kHz.pssettings
(3.32 KiB) Downloaded 144 times

hexamer
Advanced User
Advanced User
Posts: 3
Joined: Tue Aug 12, 2014 10:09 pm

Re: Spurious peak

Post by hexamer »

JC,

It's probably just aliasing, which is to be expected if you have frequency content above half the FFT's sample rate. See this thread for examples and explanation: viewtopic.php?p=145832#p145832

If you could tell us about the frequency content of your signal, we could confirm.

Thanks,

Aaron.

jcclq
Active User
Active User
Posts: 12
Joined: Thu Jul 22, 2021 1:05 pm

Re: Spurious peak

Post by jcclq »

Yes you are right, this is probably the reason, as we do have some higher frequencies. Thanks for your help, I will make sure the range and number of points are big enough. I guess that will sort it?

Best regards
JC

hexamer
Advanced User
Advanced User
Posts: 3
Joined: Tue Aug 12, 2014 10:09 pm

Re: Spurious peak

Post by hexamer »

jcclq wrote:
Mon Jul 26, 2021 8:27 am
I guess that will sort it?
Generally, yes.

Here's a good overview of the application of the FFT in the PicoScope's Spectrum Mode, including some good advice on how to pick FFT parameters: https://www.picotech.com/download/broch ... _paper.pdf

A quick excerpt from the summary:

When analyzing a signal using an FFT, the best frequency resolution is achieved by setting the frequency span to the minimum necessary, thus increasing the sample period. The number of spectrum or frequency bins should be as large as possible, whilst maintaining acceptable acquisition performance.

Minimum necessary means high enough to cover the highest frequency in the signal being analyzed. The more samples you have, the longer duration the FFT record. And since FFT bandwidth is inversely proportional to record length, with more samples you'll get lower bandwidth (finer resolution). The reason you don't want to have any higher frequency span than necessary is that higher span means faster sample rate, and so you need more samples to achieve the record length required for a given bandwidth/resolution.

If you go to the View -> View Properties menu, the Spectrum Mode will show you many of these parameters.

Post Reply