Author Topic: Boxcar FFT ripple  (Read 1588 times)

0 Members and 1 Guest are viewing this topic.

Offline fonographTopic starter

  • Frequent Contributor
  • **
  • Posts: 369
  • Country: at
Boxcar FFT ripple
« on: May 10, 2017, 09:58:11 am »
If you use oscilloscope FFT while in enhanced resolution mode that is using boxcar/rectangular window function,will you get distorted reading? Boxcar funcion have perfect time domain behavior,no ringing but it have freqency domain ripple,ERES/Gaussian window have tiny bit of ringing but its freqency response is smooth curve,no ripple.

I wonder,does it matter what type of window funcion is the scope using to get more bits as tradeoff for less samplerate? Is it possible that the ERES/Gaussian is better for FFT than HiRes/Boxcar/Rectangle due to freqency ripple?
 

Offline dmills

  • Super Contributor
  • ***
  • Posts: 2093
  • Country: gb
Re: Boxcar FFT ripple
« Reply #1 on: May 10, 2017, 10:23:08 am »
Depends on what you are doing.

Boxcar incidentally does NOT have perfect time domain behaviour, it is in fact shockingly poor if the input has any harmonic components above the output rate Shannon limit.

To enhance resolution at lower rates properly means low pass filtering followed by decimation which is not the same as a boxcar average (That has very poor attenuation out of the new passband so will be prone to aliasing causing weird spurious responses when you take the FFT), which impulse to chose for that LPF is again something dependent on what you are trying to do.

Iff the scope gets this right then you get to do the FFT on the reduced rate signal, and at this point the question of FFT window comes up, window choice depends on what you are trying to measure (Which is why there are so many of the damn things). 

There is a reason RF folks laugh at scope FFT mode ('spectrum analysers').

Regards, Dan.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf