Author Topic: EEVblog #845 - Oscilloscope FFT Comparison  (Read 44640 times)

0 Members and 2 Guests are viewing this topic.

Online ADT123

  • Regular Contributor
  • *
  • Posts: 56
  • Country: gb
    • Pico Technology
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #25 on: January 27, 2016, 12:41:44 pm »
Yeah, I know, everyone hates USB oscilloscopes, but leveraging a modern PC's processing power for FFTs could really improve things. My 2015 ultrabook i7 can easily do a 10M-point FFT (using libfftw) every second. No need for FPGA or ASIC wizardry. Compare this to the MDO3000 where a 1M point FFT took like 10 seconds.


Bear in mind Dave was doing an FFT over a frequency of DC just over 1MHz (all FFTs are from DC, the center / span displays in these scopes are zooms of the full data).  A 1 million point FFT over a 1MHz span will require 1 seconds to acquire the data and then an extra amount of time to do the FFT.   

The collection time depends on the frequency range you are measuring and the number of bins.

The time to process the FFT depends on how fast the hardware can crunch the numbers.

Suspect the above is why the Tek was so slow with 2 million points.

Update - just tested this on a PicoScope 5000 PC scope
1 million point FFT measuring up to 1MHz takes 1.1 second per screen update
128K FFT measuring up to 1MHz takes 0.3 seconds per screen update
1 million point FFT measuring up to 200MHz takes 0.2 seconds per screen update
128K FFT measuring up to 200MHz takes 0.15 seconds per screen update

If I get a chance I will put in a similar signal to the one Dave used - suspect the PicoScope would have been the winner in terms of low noise / speed of update etc
« Last Edit: January 27, 2016, 12:57:36 pm by ADT123 »
Disclaimer:  I have worked for Pico Technology for over 25 years and designed some of their early oscilloscopes. 

Happy to answer Pico related questions when I have time but I am on eevblog in a personal capacity as electronics is a hobby.
 

Online ADT123

  • Regular Contributor
  • *
  • Posts: 56
  • Country: gb
    • Pico Technology
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #26 on: January 27, 2016, 01:14:00 pm »
OK, similar FM modulated signal from a R&S sig gen.  PicoScope 5444B with a million point FFT.  One advantage this scope has is that you can change the resolution in hardware, I also enabled averaging (after all Dave did when available).  Combined this with the noise reduction you get from 1 million point FFTs and the noise floor is literally off the chart.



« Last Edit: January 27, 2016, 01:20:18 pm by ADT123 »
Disclaimer:  I have worked for Pico Technology for over 25 years and designed some of their early oscilloscopes. 

Happy to answer Pico related questions when I have time but I am on eevblog in a personal capacity as electronics is a hobby.
 

Offline dr.diesel

  • Super Contributor
  • ***
  • Posts: 2200
  • Country: us
  • Cramming the magic smoke back in...
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #27 on: January 27, 2016, 02:29:09 pm »
The DS/MSO4000s are just as horrible as their little brothers.   :--

Offline nixfu

  • Supporter
  • ****
  • Posts: 344
  • Country: us
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #28 on: January 27, 2016, 02:50:40 pm »
A fellow forum member has written software you can use to run live FFT against up to the entire sample memory of the Rigols using the much more powerful PC CPU.

https://www.youtube.com/watch?v=twRke3suKpE&feature=youtu.be

Project site:
http://rheslip.blogspot.com/

« Last Edit: January 27, 2016, 02:58:50 pm by nixfu »
 
The following users thanked this post: pfm

Offline Kalvin

  • Super Contributor
  • ***
  • Posts: 1836
  • Country: fi
  • Embedded SW/HW.
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #29 on: January 27, 2016, 05:50:02 pm »
Even a poor FFT of a cheap oscilloscope is better than nothing if you do not have a proper analyzer at hand to do the spectral analysis.
 

Offline Neganur

  • Supporter
  • ****
  • Posts: 1128
  • Country: fi
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #30 on: January 27, 2016, 06:00:35 pm »
Dave, my 30+ year old TEK 7L12 will do better than any of those scopes, 70dB to 80dB dynamic range and 1.8GHz, albiet very very very slowly. Try an Analog Discovery or Red Pitaya, 14-bit A-D, I don't why manufacturers bother putting FFT software in scopes with 8-bit or 10-bit digitizers.

OK, to be fair FFTs are a damn sight faster than a swept 7L12 and you get much better frequency resolution, but I'm thinking 16-bit A-D plumbed into 1st IF (105 MHz) or 2nd IF (10MHz).  ;)

Comparing apples to oranges. And an ADC behind the IF of the 7L is still a full blown SA compared to the FFT of those oscilloscopes.

(Edit: also, the Pitaya is only 125MSa/s)
« Last Edit: January 27, 2016, 06:04:27 pm by Neganur »
 

Offline Karel

  • Super Contributor
  • ***
  • Posts: 1688
  • Country: 00
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #31 on: January 27, 2016, 06:37:46 pm »
I did a quick check with DSRemote and EDFbrowser.
The signal source is the 1KHz calibration signal from the scope itself.
(I don't have a signal generator at hand at home.)

I downloaded 24 Mpts. Then I downsampled the data from 1GHz to 1MHz with the builtin reducer tool of EDFbrowser.
(It makes no sense to use 1GHz samplefrequency when you are interested in a 1KHz signal only.)



 

Offline Lightages

  • Supporter
  • ****
  • Posts: 4303
  • Country: ca
  • Canadian po
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #32 on: January 27, 2016, 06:56:48 pm »
I was hoping to see a comparison between the DS1052E and the DS1054Z in this video as well.
 

Offline mxmxmx

  • Contributor
  • Posts: 12
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #33 on: January 27, 2016, 06:57:21 pm »
By the way: Given that the FFT window in question was named in honor of Julius von Hann, LeCroy is actually doing the right thing in naming the window "von Hann". "Hanning window" is a misnomer, albeit a quite popular one, so LeCroy deserves a little bit of praise for using the correct name.
 

Offline karoru

  • Regular Contributor
  • *
  • Posts: 196
  • Country: pl
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #34 on: January 27, 2016, 08:26:28 pm »
By the way: Given that the FFT window in question was named in honor of Julius von Hann, LeCroy is actually doing the right thing in naming the window "von Hann". "Hanning window" is a misnomer, albeit a quite popular one, so LeCroy deserves a little bit of praise for using the correct name.

I've been living my entire life in error - I've always thought it's "Hamming window"...
 

Offline Neganur

  • Supporter
  • ****
  • Posts: 1128
  • Country: fi
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #35 on: January 27, 2016, 08:42:23 pm »
That's the thing, there is Hamming. There isn't Hanning tough, it's supposed to be Hann. German names confused the hell out of the English speaking people (or so my Signals & Systems prof claimed :P)
 

Offline janoc

  • Super Contributor
  • ***
  • Posts: 3346
  • Country: fr
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #36 on: January 27, 2016, 08:53:04 pm »
A fellow forum member has written software you can use to run live FFT against up to the entire sample memory of the Rigols using the much more powerful PC CPU.

https://www.youtube.com/watch?v=twRke3suKpE&feature=youtu.be

Project site:
http://rheslip.blogspot.com/


Or, an even simpler solution if you don't need realtime analysis - capture the waveform on an USB stick and then analyze it on a PC in something like Matlab or using Python. It certainly works much better.

https://github.com/mabl/pyRigolWFM

« Last Edit: January 27, 2016, 08:55:52 pm by janoc »
 

Offline max666

  • Frequent Contributor
  • **
  • Posts: 367
  • Country: at
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #37 on: January 27, 2016, 09:40:54 pm »
By the way: Given that the FFT window in question was named in honor of Julius von Hann, LeCroy is actually doing the right thing in naming the window "von Hann". "Hanning window" is a misnomer, albeit a quite popular one, so LeCroy deserves a little bit of praise for using the correct name.

That nobiliary particle "von" can suck my 'dick n' balls'. I will not play along with that nonsense, it's Hann window period.
 

Offline TinkerFan

  • Regular Contributor
  • *
  • Posts: 93
  • Country: de
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #38 on: January 27, 2016, 09:42:36 pm »
I'm not an owner of one of those scopes (above my budget), but I like to dream  :)

Is it possible to use the scopes in FFT-Mode and in a delayed timebase mode simultanously? That would be such a great feature...
But I'm just curious...
"A good scientist is a person with original ideas. A good engineer is a person who makes a design that works with as few original ideas as possible. There are no prima donnas in engineering." - Freeman Dyson
 

Offline orion242

  • Supporter
  • ****
  • Posts: 745
  • Country: us
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #39 on: January 28, 2016, 03:32:15 am »
I think we need an audio guy to make some of Dave's famous phrases into ring tones.

This video definitely has one that I could assign to a few contacts....lol.
 

Offline Maxlor

  • Frequent Contributor
  • **
  • Posts: 565
  • Country: ch
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #40 on: January 28, 2016, 03:54:53 pm »
Actually, the Rigol isn't quite as horrible as it appears in Dave's video, once you update it.



This is what dave shows in the video. Frequency resolution is extremely low and the noise floor is high, because the Rigol appears to work with the screen buffer only, that's only a couple hundred data points.

In newer firmware versions (04.01 didn't have it, 04.03.SP2 has it), they added a "Mode" option which lets you switch between "Trace" (old behaviour) and "Memory*" which I believe works with the sample memory. Switching to mode, the FFT looks quite a bit better (better frequency resolution and better SNR), although still nowhere near as good as some of the other scopes:



Adjusting the memory depth has no effect though, I'm guessing to always use just a few thousand points. What matters in the sampling speed: the highest resolution you get (Hz/Div) is 1/1000th of the sampling speed.
 

Offline voltsandjolts

  • Supporter
  • ****
  • Posts: 1113
  • Country: gb
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #41 on: January 28, 2016, 09:04:05 pm »
Haha, my Agilent 6000 series FFT is no better than the Rigol. Well, maybe a little less noisy but... jeez.
 

Offline artag

  • Frequent Contributor
  • **
  • Posts: 594
  • Country: gb
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #42 on: January 29, 2016, 02:08:27 am »
Haha, my Agilent 6000 series FFT is no better than the Rigol. Well, maybe a little less noisy but... jeez.

Try turning the precision mode on (in the utility menu). It slows it down a lot, but the FFT is much improved.
 

Offline uncle_bob

  • Supporter
  • ****
  • Posts: 2441
  • Country: us
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #43 on: January 29, 2016, 02:37:13 am »
Hi

If you take a look at one of the other videos on the Rigol scopes, Dave goes into the master oscillator that drives the sampler on the scope. What he finds (pre fix) is really ugly. In the case on the video they fix it to improve a jitter problem with the scope.

To some extent an FFT is measuring the clock signal as much as the test signal. A "dirty" clock will indeed limit your FFT floor and muck up the resolution of your bins. To get a good FFT, you need a good clock into the ADC. Yes, you also need the other stuff discussed at length in other posts here. The net result will still only be as good as the "weakest link" in the signal processing chain.

Bob
 

Offline blueskull

  • Supporter
  • ****
  • Posts: 14744
  • Country: cn
  • BA7LKP
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #44 on: January 29, 2016, 05:06:25 am »
I don't really think FFT makes sense these days. Cheap true SA for <$1k, or even $400 are everywhere, using crappy SDR chips you can even get one for <$100.

A properly designed lab grade USB SA, such as a Signal Hound, is just $900, which I got a new one for only $600, probably cheaper than a single license option on a mid range scope.

After all, you get only 8 bits from your ADC if you are using a scope, and that resembles only 50dB of SNR. A modern SA can get you much higher SNR.

The only reason not to use a true SA is probably because you do not have 50 Ohm output IMHO. Whenever I can get a 50 Ohm output with proper signal level, I will not use FFT.
 

Offline Muxr

  • Super Contributor
  • ***
  • Posts: 1347
  • Country: us
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #45 on: January 29, 2016, 05:34:11 am »
I don't really think FFT makes sense these days. Cheap true SA for <$1k, or even $400 are everywhere, using crappy SDR chips you can even get one for <$100.

A properly designed lab grade USB SA, such as a Signal Hound, is just $900, which I got a new one for only $600, probably cheaper than a single license option on a mid range scope.

After all, you get only 8 bits from your ADC if you are using a scope, and that resembles only 50dB of SNR. A modern SA can get you much higher SNR.

The only reason not to use a true SA is probably because you do not have 50 Ohm output IMHO. Whenever I can get a 50 Ohm output with proper signal level, I will not use FFT.
Folks who break a bank to get a $400-$800 scope aren't going to be able to afford an SA, the scope already is a stretch and it's a far more important instrument in general electronics outside of RF/Radio work. Having a usable FFT even if limited in a pinch is nice. R&S and Gw-Instek demonstrate that it's possible to implement a half decent FFT in a scope.

This is why I think FFT in entry level scopes can be a deal breaker between different scopes, I will probably be tempted to recommend R&S and Gw-Instek to friends.

Even for me personally, the cost isn't necessarily the issue, I would be tempted to get a Signal Hound but it's not supported by my OS. And I don't need an SA enough to justify making bench space for a standalone unit which usually doesn't even cover audio frequencies I sometimes dabble in.
« Last Edit: January 29, 2016, 05:37:14 am by Muxr »
 

Offline AlessandroAU

  • Regular Contributor
  • *
  • Posts: 168
  • Country: au
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #46 on: January 29, 2016, 02:23:41 pm »
I just want to point out something that may not be apparent. Everyone loves to complain about the 8 bit limitation of scope ADC's and how it limits you to 50dB of dynamic range.

This is plainly NOT TRUE. For any FFT you have an increase in the "base" SNR based on the number of samples you take, it's called processing gain. With the rigol's 24M memory you can easily push 120dB... if the internal clock wasn't so shit.

http://www.analog.com/media/en/training-seminars/tutorials/MT-001.pdf
 

Offline uncle_bob

  • Supporter
  • ****
  • Posts: 2441
  • Country: us
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #47 on: January 29, 2016, 06:08:08 pm »
I just want to point out something that may not be apparent. Everyone loves to complain about the 8 bit limitation of scope ADC's and how it limits you to 50dB of dynamic range.

This is plainly NOT TRUE. For any FFT you have an increase in the "base" SNR based on the number of samples you take, it's called processing gain. With the rigol's 24M memory you can easily push 120dB... if the internal clock wasn't so shit.

http://www.analog.com/media/en/training-seminars/tutorials/MT-001.pdf

Hi

The bit gain in the decimation process is highly dependent on a number of factors. It works well for a small signal buried in a bunch of random noise. GPS receivers do amazing things with one or two bits on the input. It may not work quite as well for a large signal into the scope that has relatively little noise. That's one of the reasons "IF based" post narrowband filter DSP approaches often needs to add a dither source to keep things sane. 

Bob
 

Offline nfmax

  • Super Contributor
  • ***
  • Posts: 1235
  • Country: gb
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #48 on: January 29, 2016, 09:17:51 pm »
Sorry if this is a dumb question - I'm not a DSP-guy.

Why do these scopes need this huge number of FFT points?   Basically, if you have 1024 horizontal pixels on the screen, shouldn't a 1024-point FFT (+/- some slack) be good enough?  Couldn't they mix the selected window down to  DC (like a direct conversion receiver), filter, maybe resample  and then a 1024-FFT over DC to span-width?  Software mixing/filtering/resampling/1k-FFT should be much cheaper than a direct 128k-point FFT, shouldn't it?

Yes, you can do this, and in a rather cleverer technique called 'zoom FFT' processing. What Dave didn't mention, and indeed almost nobody does, is that the FFT mode in these scopes is not just an FFT, it also includes a power calculation, so that the trace on the screen is a power spectrum. They all do this because the output from an FFT is actually complex, and scopes have difficulty displaying complex waveforms. The power in a complex waveform is the sum of squares of the real and imaginary components, so we can easily show that. However, if you don't apply the power calculation you can do additional processing before display.

Suppose we make a 4096 point FFT. The output of this comprises 2048 frequency bins plus a DC bin (for a real input). You can think of each of these bins as the output of a narrow band filter, with an amplitude and phase response that depends on the window used, and a centre frequency which is a multiple of the basic frequency resolution of the 4096 point FFT, i.e. 1 over the time duration of the 4096 points. The real and imaginary components of the FFT bin output give you the signal that gets through this filter, down converted to DC, like the I and Q outputs of a 'zero IF' receiver. If we want a finer frequency resolution, we have to acquire more points, over a longer time, so that the bins are more closely spaced, and make a longer FFT.

However, if don't want to see all the frequency range, an alternative is to make a series of short FFTs of successive 4096 sample sections of the signal, and look at the sucessive complex outputs of the bins in just the frequency range we are interested in. This is another signal - complex valued this time - which varies over time and represents how the output signal from the filter corresponding to the FFT bin changed over time. For example, 8 successive 4096 point FFTs gives, for each frequency bin, a time sequence of 8 complex numbers. Here's the clever bit: for each bin we are interested in, we make another, 8-point FFT of the complex bin output. Because the data is complex, this gives us 8 frequency mini-bins in place of the original big bin. We apply the power calculation on the output of these FFTs and display the result. You can do this to as many of the original FFT bins as you want, but because of the extra overhead in the short FFTs it's only really worth doing if you are interested in just a small part of the frequency range. But it does let you show a low-resolution, full bandwidth power spectrum (based on the successive 4096 point FFTs) which is updated rapidly, at the same time as the higher resolution, zoomed section, updated less often. Hence the name of the technique.

We were using this approach in the '80s on a 68000 micro in a dynamic signal analyser for machinery vibration, but I believe it was first used in sonar processing some years earlier. Since memory is cheap and processors are fast, it seems to have fallen by the wayside, but heck, it still works!

Max
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 2576
  • Country: gb
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #49 on: January 29, 2016, 09:40:03 pm »
It was interesting to see the various scopes in FFT mode but I was left a bit disappointed because I don't think enough time was spent on each one to see how much you can exploit FFT mode and make some decent measurements that would be difficult on a conventional swept spectrum analyser.

For example, I still use an old Tek TDS2012 DSO quite regularly and I use the FFT mode on it quite regularly too. It only has a 2K FFT but you can still do a lot with a 2K FFT if you exploit the sample rate and alias limitations of the scope and turn them to your advantage. Obviously, you have to be wary of alias issues but a skilled operator can spot these and know to ignore them or manage/remove them.

Here's a few things I can do with the creaky old 100MHz TDS2012 in FFT mode.

I can look at an unmodulated  FM signal from an old 2m Yaesu radio (FT290R 145MHz) and zoom in to a narrow span of a few kHz or less and look at the close in phase noise of the radio's carrier with a decent refresh rate.

I can also watch the spectrum in 'close to real time' when speaking into the microphone and watching the scope display the (<20kHz) BW of the approx 145MHz FM test signal.

I can also watch the FT290R transmission as it slowly drifts in frequency by a few Hz per second as it warms up from a cold start. Obviously, the scope has to be warmed up and stable beforehand.

I can look at a ~145MHz cw test signal being on/off keyed at a 1Hz rate (once a second) and zoom right in and see the modulation sidebands at 1Hz, 3Hz, 5Hz spacings from the carrier etc. Obviously, the scope will be slow at this but try doing this on a typical/conventional spectrum analyser. Not many will be able to do it.

It's probably at its most useful when looking at classic SSB, AM or FM modulation in FFT mode. It can display a span of a few kHz really fast/fluidly. This old scope doesn't have much in the way of spurious free dynamic range but it's going to be better than the typical IMD distortion seen on an AM or SSB signal from a ham or CB radio. You can also do Bessel nulling on NBFM to measure FM deviation with it and the display refresh will be quite fluid.

It would be interesting to see if the Rigol or the other scopes can be configured to do similar. My scope is very basic and old so I suspect the other scopes will be able to do similar tricks (but with even better results)
« Last Edit: January 29, 2016, 10:32:37 pm by G0HZU »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf