EEVblog #1188 - $10 DIY EMC Probe using Scope FFT

[EEVblog]

How good is your existing oscilloscopes FFT function with the $10 DIY EMC H-field probe compared with a dedicated spectrum analyser?
Dave tries the DIY EMC probe with half a dozen different low cost scopes, so it turned into a bit of a practical FFT shootout:
Rigol DS1054Z
Siglent SDS1202X-E
Keysight DSO1204X
Rohde & Schwarz HMO1202
GW-Instek GDS-1104S
Owon XDS3202A
and the higher end Rohde & Schwarz RTB2004

[Bud]

 Dave you give a poor advice by moving the input trace beyond the screen area to hide it (@10:50 in the video). That is because the ADC in a typical modern scope has only 10% margin below and above screen area. If the signal is larger than that the ADC will start clipping the peaks which creates harmonics which are added to the FFT trace. This may not be much of an issue for looking at very noisy signals like the use case in this video but in general you should not do that or be very careful to not clip the signal. Try it with sinewave and watch the harmonic content increasing in the FFT trace as you move the signal trace outside of the screen. You can actually demonstrate this as a tip in one of your next videos.

[EEVblog]

Dave you give a poor advice by moving the input trace beyond the screen area to hide it (@10:50 in the video). That is because the ADC in a typical modern scope has only 10% margin below and above screen area. If the signal is larger than that the ADC will start clipping the peaks which creates harmonics which are added to the FFT trace. This may not be much of an issue for looking at very noisy signals like the use case in this video but in general you should not do that or be very careful to not clip the signal. Try it with sinewave and watch the harmonic content increasing in the FFT trace as you move the signal trace outside of the screen. You can actually demonstrate this as a tip in one of your next videos.

I demonstrated it in this video, seems you didn't watch far enough!

[Brumby]

What if you don't have FFT on your scope?

Edit: Oh, that's right ... you did mention SDR units.  Will be interested to see how these fare.

[nctnico]

Regarding the GW Instek: uphack it to get the spectrum analyser mode from the MDO version.

[johnlsenchak]

Sorry Dave  ,  this video didn't interest me, but  I did  enjoy watching  the live  "question and  answer"  video  tonight

[David Hess]

FFT resolution on a DSO by itself is not comparable to RBW (resolution bandwidth) on a spectrum analyzer.  EDN had a good article about this several years ago:

https://www.edn.com/electronics-blogs/the-practicing-instrumentation-engineer/4427466/DSOs-and-noise-

The article also mentions the problem with combining average mode with FFT which many DSOs have; generally you want to average the FFT and not the input.  If you do average the input, then the trigger needs to be valid and the signals need to be synchronous.

I suspect the Rohde & Schwarz results are showing detail which does not exist.  I wonder if the RBW on the R&S displayed changes with FFT window selection like it should.

[nctnico]

If you enable averaging in FFT mode then the FFT results get averaged, not the input signal. Ofcourse you can average the input signal as well but that is an acquisition setting. I get the feeling the author of the article isn't intimately familiar with how FFT on a DSO works.

[David Hess]

If you enable averaging in FFT mode then the FFT results get averaged, not the input signal. Ofcourse you can average the input signal as well but that is an acquisition setting. I get the feeling the author of the article isn't intimately familiar with how FFT on a DSO works.

The point is that many DSOs can only average during acquisition and not after the FFT and unless the trigger is correct, averaging during the acquisition will screw up the results which was easy to see with the R&S.  Averaging during the acquisition will *always* invalidate noise marker results.  It is generally the wrong thing to do.

[nctnico]

Sorry but that seems incorrect to me. On all the DSOs I have used which have averaging in FFT mode, enabling averaging averages the FFT result and not the input signal. If the user enables averaging on the acquisition settings instead on the FFT trace then that is a user error.
I checked the video again, but on the R&S there is no averaging on the input signal. The signal isn't exactly stable so if the input signal was averaged you'd see the peaks and other high frequency content being filtered out.

[EEVblog]

Sorry Dave  ,  this video didn't interest me, but  I did  enjoy watching  the live  "question and  answer"  video  tonight

John, please stop telling me what videos interest you and what ones don't. I appreciate the desire to give feedback, but doing it constantly is getting very tiring. Thanks.

[EEVblog]

What if you don't have FFT on your scope?

Is there a single DSO on the market in the last decade or more that doesn't have FFT?

[Brumby]

The most modern scope I have is a TDS220 - and there's my Hantek 6022BL.  Neither really useful for this.

I will be looking for your take on the SDR route.

[nctnico]

What if you don't have FFT on your scope?

Is there a single DSO on the market in the last decade or more that doesn't have FFT?

I think you have to define that in terms of useability. On several oscilloscopes the FFT is nothing more than a checkbox feature (it is there but not useful at all).

[thm_w]

You can always send the trace data off to a PC and perform the FFT there (matlab, octave, something custom).
But again, if its a scope that is old or doesn't have a useful FFT, the chance it has good live data streaming capability is low. So update rate will be horrific.

[dimtass]

Dave, thanks to your video I've build my own probes and I've used a RTL-SDR.
Works great, kudos.

https://www.stupid-projects.com/emc-probe-using-rtl-sdr/