Author Topic: Tektronix MSO54 - why do I see a signal in the FFT?  (Read 1158 times)

0 Members and 1 Guest are viewing this topic.

Offline PlasmateurTopic starter

  • Regular Contributor
  • *
  • Posts: 204
  • Country: us
Tektronix MSO54 - why do I see a signal in the FFT?
« on: January 05, 2023, 05:12:04 am »
Hello,

I am using a MSO54 O-scope and a Siglent SDG6000X to test for an application where a small amplitude signal is buried in a large amplitude signal. The MSO54 has 8-bit, 12-bit, and up to 16-bit acquisition.

Considering I have the range of a channel on the O-scope set to +/- 5 Volts, then at 8-bits - the smallest digitized increment should be 10*2^-8 = 0.0390625 volts, about 40mV correct? - (I'm not considering ENOB here)

So I send two sinusoidal signals from the SDG6000X ARB in one channel of the MSO54 terminated at 50Ohm. Signal 1 is 5 Vpp and signal 2 is under 40mVpp. Because signal 2 is under the smallest digitized increment, I thought I shouldn't be able to see signal 2.

And yet when I used the FFT on the MSO54 I can still see signal 2 in the FFT, even down to 1mVpp. I'm confused as to why. Does the FFT on the MSO54 do something similar to a signal analyzer? I wouldn't think it did.

Perhaps it might have something to do with isolation of the channels on the SDG6000X? I tried many different configurations, but the small signal is still seen in the FFT. I'm not understanding why I see it there. Perhaps my knowledge of test equipment needs a bit of help.

If anyone knows why I might be seeing this, I'd love to know why.

Thanks.

 

Offline ataradov

  • Super Contributor
  • ***
  • Posts: 11906
  • Country: us
    • Personal site
Re: Tektronix MSO54 - why do I see a signal in the FFT?
« Reply #1 on: January 05, 2023, 05:44:24 am »
It is hard to tell for sure without eliminating other setup issues, but one way I see this can happen is dithering. That small signal may be big enough to make large signal cross the bit threshold.

Imagine your large signal perfect with no noise at all and is just below the ADC count threshold. The ADC would read the constant value. Now add very small signal (or noise) on top of that. This is enough for the big signal to cross the threshold and be discriminated. And if the small signal is periodic, then this period would be clearly visible in the digitized data.

This is often used intentionally in DSP to reduce ADC quantization noise.
Alex
 
The following users thanked this post: Plasmateur

Offline rf-loop

  • Super Contributor
  • ***
  • Posts: 4135
  • Country: cn
  • Born in Finland with DLL21 in hand
Re: Tektronix MSO54 - why do I see a signal in the FFT?
« Reply #2 on: January 05, 2023, 05:51:16 am »
Hello,

I am using a MSO54 O-scope and a Siglent SDG6000X to test for an application where a small amplitude signal is buried in a large amplitude signal. The MSO54 has 8-bit, 12-bit, and up to 16-bit acquisition.

Considering I have the range of a channel on the O-scope set to +/- 5 Volts, then at 8-bits - the smallest digitized increment should be 10*2^-8 = 0.0390625 volts, about 40mV correct? - (I'm not considering ENOB here)

So I send two sinusoidal signals from the SDG6000X ARB in one channel of the MSO54 terminated at 50Ohm. Signal 1 is 5 Vpp and signal 2 is under 40mVpp. Because signal 2 is under the smallest digitized increment, I thought I shouldn't be able to see signal 2.

And yet when I used the FFT on the MSO54 I can still see signal 2 in the FFT, even down to 1mVpp. I'm confused as to why. Does the FFT on the MSO54 do something similar to a signal analyzer? I wouldn't think it did.

Perhaps it might have something to do with isolation of the channels on the SDG6000X? I tried many different configurations, but the small signal is still seen in the FFT. I'm not understanding why I see it there. Perhaps my knowledge of test equipment needs a bit of help.

If anyone knows why I might be seeing this, I'd love to know why.

Thanks.

It seems like you're also asking "Where Does FFT Process Gain Come From?"

https://www.designnews.com/aerospace/where-does-fft-process-gain-come

Using Google or Baidu you can find lot of answers.

This is mandatory to understand least partially:
https://www.analog.com/media/en/training-seminars/tutorials/MT-001.pdf

Also here in forum is some threads what handle these things directly or indirectly
Example here (FFT things):
https://www.eevblog.com/forum/testgear/comparison-between-siglent-sdg1000x-and-2000x/

and here (FFT things):
https://www.eevblog.com/forum/testgear/siglent-sdg1032x-sine-distortion-at-1-khz/

Keep always in mind that oscilloscope itself can also produce lot of "fake" signals what you can see when high FFT processing gain and it do not so much matter if brand is Tektronix, Siglent, Keyshitg, R&S ...etc..  or chungfunhang.
« Last Edit: January 05, 2023, 05:54:44 am by rf-loop »
EV of course. Cars with smoke exhaust pipes - go to museum.
Wises must compel the mad barbarians to stop their crimes against humanity. Where have the (strong)wises gone?
 
The following users thanked this post: 2N3055, Plasmateur

Offline PlasmateurTopic starter

  • Regular Contributor
  • *
  • Posts: 204
  • Country: us
Re: Tektronix MSO54 - why do I see a signal in the FFT?
« Reply #3 on: January 05, 2023, 05:57:21 am »
I could actually code this up and test it just for fun. Also going to look up - "DSP to reduce ADC quantization noise."

Very helpful response! Thank you so much!
 

Offline PlasmateurTopic starter

  • Regular Contributor
  • *
  • Posts: 204
  • Country: us
Re: Tektronix MSO54 - why do I see a signal in the FFT?
« Reply #4 on: January 05, 2023, 05:58:59 am »
Thank you for all of these links, diving in now.
 

Online nctnico

  • Super Contributor
  • ***
  • Posts: 28429
  • Country: nl
    • NCT Developments
Re: Tektronix MSO54 - why do I see a signal in the FFT?
« Reply #5 on: January 05, 2023, 01:21:59 pm »
Hello,

I am using a MSO54 O-scope and a Siglent SDG6000X to test for an application where a small amplitude signal is buried in a large amplitude signal. The MSO54 has 8-bit, 12-bit, and up to 16-bit acquisition.

Considering I have the range of a channel on the O-scope set to +/- 5 Volts, then at 8-bits - the smallest digitized increment should be 10*2^-8 = 0.0390625 volts, about 40mV correct? - (I'm not considering ENOB here)
Simplistic: With the FFT function you are also getting more bits. If you turn on high-res mode, you'll likely be able to see the signal as well if you use vertical zoom.
« Last Edit: January 05, 2023, 01:23:36 pm by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline EC8010

  • Regular Contributor
  • *
  • Posts: 205
  • Country: gb
Re: Tektronix MSO54 - why do I see a signal in the FFT?
« Reply #6 on: December 25, 2024, 09:25:56 am »
Beware that although your MSO54 claims 12-16 bit quantisation, its ADCs are not perfect and you will see that in an FFT of a low distortion sine wave with harmonics typically at around -54dB due to the ADCs. If you want to see the small signal, it's best to add an external filter that can attenuate the large signal, allowing you to wind in some more sensitivity.
 

Offline TheoB

  • Regular Contributor
  • *
  • Posts: 140
  • Country: nl
Re: Tektronix MSO54 - why do I see a signal in the FFT?
« Reply #7 on: December 25, 2024, 10:09:42 am »
It's like averaging multiple measurements. Suppose you want to measure a 12V signal with a multimeter that has limited resolution.  You measure 12.00V or 12.01V. Repeat that 16 times and average the values. You now measure with a 1/16*10mV=0.625mV resolution instead of 10mV. That adds 16 more levels or 4 bits of resolution. This only works if the noise is larger than the quantization level as mentioned before.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf