Author Topic: EEVblog #845 - Oscilloscope FFT Comparison  (Read 57317 times)

0 Members and 1 Guest are viewing this topic.

Offline rs20

  • Super Contributor
  • ***
  • Posts: 2317
  • Country: au
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #50 on: January 30, 2016, 10:42:20 pm »
With any algorithm it's "garbage in, garbage out" principle. FFT from 8-bit ADC in the scope will be bad, no matter how clever you try to analyse the data (unless they invent some oracle chip that will extract the additional information using heaven microwave link).

Is that really true? If you have a one million point FFT on 8-bit data, any given point in your output spectrum will have noise reduce by a factor of sqrt(1M). Put another way, if you add twice as much 8-bit data, you are effectively averaging over twice as many cycles.

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

Precisely!

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?

The computational cost of FFT is only O(N log N); it's not some sort of horrendous polynomial monster or anything like that. So to presume that "just doing direction conversion" on millions of points is easier that doing a 16k FFT is, I suspect, simply wrong.
 

Offline KE5FX

  • Super Contributor
  • ***
  • Posts: 1872
  • Country: us
    • KE5FX.COM
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #51 on: January 30, 2016, 10:57:23 pm »
Is that really true? If you have a one million point FFT on 8-bit data, any given point in your output spectrum will have noise reduce by a factor of sqrt(1M). Put another way, if you add twice as much 8-bit data, you are effectively averaging over twice as many cycles.

Of course it's not true, but that doesn't keep people from mindlessly repeating it every time this subject comes up.  A personal pet peeve.

Wait'll they learn how a sigma-delta converter works.   ;D
 

Offline uncle_bob

  • Supporter
  • ****
  • Posts: 2441
  • Country: us
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #52 on: January 31, 2016, 06:27:49 pm »


Wait'll they learn how a sigma-delta converter works.   ;D

Hi

Then throw something cross correlation into the mix .... That's one example of how you can (unless you are careful) take things to the point that the numbers you get actually are *better* than the reality of the signals involved.

Bob
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 26682
  • Country: nl
    • NCT Developments
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #53 on: January 31, 2016, 10:41:39 pm »
Most of us hobby players are often interested in using FFT becuase we cant afford a proper SA for the bench.  And sometimes FFT is at least a chance to get some sort of data in the frequency domain that is better than nothing.
FFT is most certainly not a poor man's spectrum analyser! A spectrum analyser typically sweeps the signal so it is less effective in capturing glitches. With an oscilloscope you can capture a worse case trace and use FFT to get it's frequency contents. Also spectrum analysers typically start from 9kHz to several MHz. An oscilloscope can be useful to do analysis in lower frequency bands as well.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline uncle_bob

  • Supporter
  • ****
  • Posts: 2441
  • Country: us
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #54 on: February 01, 2016, 12:38:45 am »
FFT is most certainly not a poor man's spectrum analyser! A spectrum analyser typically sweeps the signal so it is less effective in capturing glitches. With an oscilloscope you can capture a worse case trace and use FFT to get it's frequency contents. Also spectrum analysers typically start from 9kHz to several MHz. An oscilloscope can be useful to do analysis in lower frequency bands as well.

Hi

So in that context. If I pay nothing for FFT, does it matter that it sucks?

As usual, I'm babbling ...

I've been in a *lot* of design reviews / spec reviews where the concept of "marketing check box" comes up. Translated into common English that's a spec with no details and no apparent strong need behind it. The assumption is that the customer needs to have it present, but has no real need for it. As an engineer, I find this whole idea crazy (to say the least). The next step is to put the feature in with as low cost as you possibly can. If there is a problem, it will be fixed later. That part is driven by the common assumption that any problem can be fixed with a firmware update.

So here we are on the *other* end of that conversation. We are the customers. We bought the scope or MSO. It had FFT listed as a spec. Was that up there with bandwidth on a check list for any of us? I doubt it. It most certainly was not for me. Was it the first (or second or tenth) thing I checked on a new scope? Nope. For me at least, yes, it's a marketing check box. It's not something I expect to work well. If I need a spectrum, I grab one of .... errr ... several .... very large, very heavy, very used, HP analyzers.

So, did the marketing guys get it right? In my case they did. Does that make the whole idea of a "not a spec spec" any less crazy? I'm not sure it does. It does indeed validate the business sense of doing it that way.

Bob
 

Offline mikerj

  • Super Contributor
  • ***
  • Posts: 3231
  • Country: gb
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #55 on: February 01, 2016, 11:11:01 am »
FFT is most certainly not a poor man's spectrum analyser! A spectrum analyser typically sweeps the signal so it is less effective in capturing glitches.

Update rate is an advantage, but in many other areas it is a poor man's spectrum analyser.  For starters there is no ability to capture only signals of interest and filter out unwanted signals which may be at a much higher level, which a spectrum analyser does.  A spectrum analyser gives you greater dynamic range (the noise floor on the cheaper scopes is much higher than a proper SA).

With an oscilloscope you can capture a worse case trace and use FFT to get it's frequency contents. Also spectrum analysers typically start from 9kHz to several MHz. An oscilloscope can be useful to do analysis in lower frequency bands as well.

OTOH even low end spectrum analysers go up to 3GHz or more, well outside the range of any reasonably priced DSO.  At the high end (e.g. 50GHz or more) then you have little choice but to use a spectrum analyser.
 

Offline Mechatrommer

  • Super Contributor
  • ***
  • Posts: 11519
  • Country: my
  • reassessing directives...
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #56 on: February 01, 2016, 05:54:02 pm »
Actually, the Rigol isn't quite as horrible as it appears in Dave's video, once you update it.
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.
here's the review link, and the linked original youtube by Rigol's Application Engineer himself... i didnt know they have a very fluent english speaking engineer there...
http://www.arbenelux.com/en/improved-fft-on-rigol-msods1000-series/
https://www.youtube.com/watch?v=dD-3ni7tYM4&feature=youtu.be

Adjusting the memory depth has no effect though...
there is. i just dick around with manual memory depth (after changing Mode from Trace to Memory), there are difference. i think i like 12KPts more than larger memory. now Rigol gave the power  of manual memory length selection for FFT to us. attached pics is i tried dicking around to get what Dave's got in the OP review... i tried 2 memory depth, 12K and 120K... to get 5KHz/div, you need to dick around with timebase (see attached). excuse my poor FG jittering 1MHz unspecified distortion level. i set it at seomwhere 2.5Vpp 50ohm terminated coax (i forgot to reset the DSO to 1X probe, at 10X setting, it read as 25Vpp, just ignore it or divide by 10). although with this high depth memory i'm suspecting or seems to be FFT aliasing when we set to max freq/div (namely 250KHz/div) as shown in 3rd and 4th picture. 500KHz harmonics at 1MHz fundamental? weird... but when set to Dave's setting, it looks ok.. maybe he can re-evaluate with his FM generator and see if the DS1Z with 4.3.SP2 update (SP1 according to the Rigol's engineer) may catch up with the rest...

let me enlarge the 12KPts memory FFT that match up with Dave's FFT test range (1MHz center, 5KHz/div, 20dB/div) and its accompanying aliasing artifact @ 250KHz/div...



« Last Edit: February 01, 2016, 05:59:34 pm by Mechatrommer »
Nature: Evolution and the Illusion of Randomness (Stephen L. Talbott): Its now indisputable that... organisms “expertise” contextualizes its genome, and its nonsense to say that these powers are under the control of the genome being contextualized - Barbara McClintock
 

Offline Mechatrommer

  • Super Contributor
  • ***
  • Posts: 11519
  • Country: my
  • reassessing directives...
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #57 on: February 01, 2016, 06:29:26 pm »
and to encounter this video from gwInstek... i've made some more capture at the same 1KHz square signal using DS1054Z 12KPts FFT. i believe the video below is superseded (aka obsolete) by the latest Rigol's FW. except the FFT respond time... FWIW... edit: ... "so what are you waiting for? ... update now!"

« Last Edit: February 01, 2016, 06:54:21 pm by Mechatrommer »
Nature: Evolution and the Illusion of Randomness (Stephen L. Talbott): Its now indisputable that... organisms “expertise” contextualizes its genome, and its nonsense to say that these powers are under the control of the genome being contextualized - Barbara McClintock
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 26682
  • Country: nl
    • NCT Developments
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #58 on: February 01, 2016, 06:33:50 pm »
FFT is most certainly not a poor man's spectrum analyser! A spectrum analyser typically sweeps the signal so it is less effective in capturing glitches.

Update rate is an advantage, but in many other areas it is a poor man's spectrum analyser.  For starters there is no ability to capture only signals of interest and filter out unwanted signals which may be at a much higher level, which a spectrum analyser does.  A spectrum analyser gives you greater dynamic range (the noise floor on the cheaper scopes is much higher than a proper SA).

With an oscilloscope you can capture a worse case trace and use FFT to get it's frequency contents. Also spectrum analysers typically start from 9kHz to several MHz. An oscilloscope can be useful to do analysis in lower frequency bands as well.

OTOH even low end spectrum analysers go up to 3GHz or more, well outside the range of any reasonably priced DSO.  At the high end (e.g. 50GHz or more) then you have little choice but to use a spectrum analyser.
What I'm trying to say is that both a spectrum analyser and FFT in an oscilloscope are equally useful tools each with their own strong and weak points.
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14016
  • Country: de
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #59 on: February 01, 2016, 07:49:19 pm »
In the lower frequency range, FFT is a rather good alternative to a classical SA. Some of the more modern SA actually also use FFT instead of the classical logarithmic converter. The somewhat limiting factor with a typical scope is the relatively low resolution of the converter and thus a limited dynamic range. With sufficient long averaging the dynamic range is not that bad - as all frequencies are samples simultaneously there is plenty of time compared to a slow scanning classical SA.

So in many cases the FFT is a cheap alternative / extension for a SA for lower frequency signals. Even if you have a SA, scanning gets very slow at low frequency / low filter bandwidth. If you don't have an SA, the FFT function can be a substitute in many cases - sometimes worse than the SA and sometimes better.

The test dave did, was also just one aspect of the FFT - using high frequency resolution, which is something classical SAs are not very good at. Another point to compare would be the noise level / dynamic range. Some of the scopes did quite well, especially when using averaging or large number of points. Also the linearity of the input / ADC might be important: a poor input could add harmonics or IM products that are not there. Here averaging / oversampling does not help.
 

Offline max666

  • Frequent Contributor
  • **
  • Posts: 367
  • Country: at
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #60 on: February 01, 2016, 08:24:03 pm »
and to encounter this video from gwInstek... i've made some more capture at the same 1KHz square signal using DS1054Z 12KPts FFT. i believe the video below is superseded (aka obsolete) by the latest Rigol's FW. except the FFT respond time... FWIW... edit: ... "so what are you waiting for? ... update now!"

I have been thinking about updating ... the firmware that is ;)
But I've read there are bugs introduced by the 00.04.03 SP2 firmware, so I'm not sure if I should update from 00.04.02 SP4?

Edit: Oh and I too would like to see Dave show the FFT again with updated firmware  :D
« Last Edit: February 01, 2016, 08:26:01 pm by max666 »
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 3007
  • Country: gb
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #61 on: February 01, 2016, 11:27:38 pm »
Here's my old Tek TDS2012 scope displaying a bandpass sampled signal at about 70MHz using FFT mode. The input signal is being on/off keyed once a second and you can see the modulation sidebands at 1Hz, 3Hz and 5Hz (etc) offsets.

So this old scope can display the bandwidth detail of this very narrow signal in a 25Hz span at 70MHz. Not many conventional analysers can do this. Obviously, the signal has to be narrowband or the display will fill with alias terms but because this signal is quite narrow it can be bandpassed sampled in FFT mode at a very low sample rate.

It would be interesting to see how easily other scopes can be configured to do this. My Agilent E4406A signal analyser can display this signal very well but none of my other analysers can.
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 26682
  • Country: nl
    • NCT Developments
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #62 on: February 02, 2016, 12:34:05 am »
Any scope can do that.. (if it has FFT). For fun show the time domain signal...
Gw Instek GDS2204E:
« Last Edit: February 02, 2016, 10:52:09 am by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Offline Mechatrommer

  • Super Contributor
  • ***
  • Posts: 11519
  • Country: my
  • reassessing directives...
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #63 on: February 02, 2016, 03:09:54 am »
But I've read there are bugs introduced by the 00.04.03 SP2 firmware
what bug? i havent brick any, except slower GUI respond than the older version, but if you rotate knob quickly, it will reflect on the scope, just slower GUI. the older version also got bugs, so it makes very little difference knowing the fixed some esp this FFT stuff...
Nature: Evolution and the Illusion of Randomness (Stephen L. Talbott): Its now indisputable that... organisms “expertise” contextualizes its genome, and its nonsense to say that these powers are under the control of the genome being contextualized - Barbara McClintock
 

Offline nixfu

  • Supporter
  • ****
  • Posts: 346
  • Country: us
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #64 on: February 02, 2016, 03:39:44 pm »
FFT is most certainly not a poor man's spectrum analyser! A spectrum analyser typically sweeps the signal so it is less effective in capturing glitches. With an oscilloscope you can capture a worse case trace and use FFT to get it's frequency  contents. Also spectrum analysers typically start from 9kHz to several MHz. An oscilloscope can be useful to do analysis in lower frequency bands as well.

A common technique for using FFT as a pseudo-SA is to use a broad spectrum  noise generator which basically outputs noise signals all across the spectrum at once, instead of sweeping. 

You can then use this to see the frequency response/cutoff points of filters, RLC circuits, etc. using the scope's FFT.     The same broad  noise signal technique is often used by people who have an SA, but don't have a tracking signal generator.
« Last Edit: February 02, 2016, 03:44:12 pm by nixfu »
 
The following users thanked this post: maelh

Offline max666

  • Frequent Contributor
  • **
  • Posts: 367
  • Country: at
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #65 on: February 02, 2016, 09:20:10 pm »
But I've read there are bugs introduced by the 00.04.03 SP2 firmware
what bug? i havent brick any, except slower GUI respond than the older version, but if you rotate knob quickly, it will reflect on the scope, just slower GUI. the older version also got bugs, so it makes very little difference knowing the fixed some esp this FFT stuff...

I was reading this:
upgrade New Rigol DS1054Z oscilloscope to firmware 00.04.03.SP2?

alsetalokin4017 didn't go into detail, he just said "The SP2 firmware did fix a couple of issues that I personally found annoying, but introduced a couple more annoying issues. For my purposes the first point outweighs the second point."

I guess if it's only the "Measurements Fail with Math On" bug, then I'm going to update.
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 3007
  • Country: gb
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #66 on: February 02, 2016, 11:17:28 pm »
Here's what the E4406A signal analyser shows for the 1Hz OOK signal at 70MHz. This is a plot I did some time ago.

In the second image I've used the Agilent 89600 SW to take data from the E4406A and then measure the 99% occupied BW of the 1Hz OOK signal and this should be about 38Hz in theory. This is also an old plot and it looks like it was done at 10.7MHz and not 70MHz.

I tried using my old HP54540C in FFT mode to look at this but it was a bit of a struggle to get a decent display. I've not had this scope long and I've only used it a few times but the FFT mode isn't as good as I expected. It can do up to a 32K FFT which is much better than the little Tek TDS2012 but I much prefer to use the Tek in FFT mode as it is much easier to drive.
« Last Edit: February 02, 2016, 11:20:26 pm by G0HZU »
 

Offline rs20

  • Super Contributor
  • ***
  • Posts: 2317
  • Country: au
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #67 on: February 03, 2016, 05:07:51 am »
Apropos to nothing; I find it funny that oscilloscope vendors and users refer to it as "FFT", i.e., Fast Fourier Transform. FFT is just the fastest implementation of the DFT, the end user shouldn't care if a powerful FPGA running naive DFT was used instead.

Not a criticism per se, just something I find kinda funny -- the way that the optimized solution has become synonymous with the original problem.
 

Offline Mechatrommer

  • Super Contributor
  • ***
  • Posts: 11519
  • Country: my
  • reassessing directives...
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #68 on: February 03, 2016, 06:04:11 am »
"Measurements Fail with Math On" bug, then I'm going to update.
i've tried to reproduce the "statistic measurement freeze failure during math on" bug, and "zoomed time activated, change persistent and brick the scope" bugs on SP2, they are not there anymore. whats a bit annoying is the slower respond time, but thats not a big deal for me as quick scroll still be reflected in the GUI albeit the slowleness... and i found trigger failure weirdnessness bug probing CAL signal https://www.eevblog.com/forum/testgear/rigol-ds1054z-unable-to-trigger-cal-signal-at-lt-0-6v but afaik its not happening on the other external signals, and i'm not sure if this is specific to SP2, this probably also happening in the older FW.. imho this high resolution FFT in SP2 is far outweigh any older (or probably newer) bugs, just stay away from them, i havent encounter anything serious so far, but there is another thread to report such bugs... ymmv.

« Last Edit: February 03, 2016, 06:11:53 am by Mechatrommer »
Nature: Evolution and the Illusion of Randomness (Stephen L. Talbott): Its now indisputable that... organisms “expertise” contextualizes its genome, and its nonsense to say that these powers are under the control of the genome being contextualized - Barbara McClintock
 

Offline alsetalokin4017

  • Super Contributor
  • ***
  • Posts: 2055
  • Country: us
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #69 on: February 03, 2016, 04:26:55 pm »
"Measurements Fail with Math On" bug, then I'm going to update.
i've tried to reproduce the "statistic measurement freeze failure during math on" bug, and "zoomed time activated, change persistent and brick the scope" bugs on SP2, they are not there anymore. whats a bit annoying is the slower respond time, but thats not a big deal for me as quick scroll still be reflected in the GUI albeit the slowleness... and i found trigger failure weirdnessness bug probing CAL signal https://www.eevblog.com/forum/testgear/rigol-ds1054z-unable-to-trigger-cal-signal-at-lt-0-6v but afaik its not happening on the other external signals, and i'm not sure if this is specific to SP2, this probably also happening in the older FW.. imho this high resolution FFT in SP2 is far outweigh any older (or probably newer) bugs, just stay away from them, i havent encounter anything serious so far, but there is another thread to report such bugs... ymmv.
When I reported the Measurements Fail bug to Rigol USA, they were immediately able to see it on their test scopes with the SP2 firmware, and agreed that it was a bug. The problem with reproducing this bug is that it happens after a totally random time interval, sometimes it will hit within seconds and sometimes it takes 30 minutes or more. It is definitely there though. I don't know why you can't see it on your scope. Lots of other people have also seen this bug with the new firmware. The workaround is to keep the Math trace set up, but turned OFF, when you are using measurements, and only turn the Math ON for a brief time when you need to see what it's doing. It still might bite you though, since sometimes it only takes a few seconds for the bug to hit.

As far as the Freeze bug goes: when I tested both my old scope running Boot Version 0.0.1.2 and the new scope running Boot Version 0.0.1.3, both running the latest SP2 firmware, the old scope still was able to freeze, the new firmware made no difference at all for this bug. I had them both for three days before I returned the "freezable"  one to Rigol, and I tested them side-by-side (actually one on top of the other) on my bench. The new firmware apparently does _not_ fix the freezing issue on scopes that have Boot Version 0.0.1.2, as far as I can tell.  The workaround for those who have this bug is to be very careful about using Persistence and Zoom at the same time.

The SP2 firmware also introduces the "Pluses" spelling error and sometimes the "pluses" counters and edge counters miscount the actual number of pulses on screen.
No workaround here--- just suck it up and ignore the spelling error, and count your pulses by eye to see if the counter is off-by-one.

And of course it makes the scope very slow to respond to some controls, like especially the channel vertical position controls. So use "safecracker mode" on your fingers and turn the knobs slowly.... very slowly.... and wait for the scope to catch up.

But your problems with trigger level are something else. As I posted at the time it was being discussed, my scope has no problem triggering on the Calibrate signal down to smaller voltage levels.

The FFT does seem to work better now though.
The easiest person to fool is yourself. -- Richard Feynman
 

Offline Mechatrommer

  • Super Contributor
  • ***
  • Posts: 11519
  • Country: my
  • reassessing directives...
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #70 on: February 04, 2016, 05:01:53 am »
So use "safecracker mode" on your fingers and turn the knobs slowly.... very slowly.... and wait for the scope to catch up.
you dont have to slow turn until you see something on the screen... if you turn quickly, the next screen update, the big offset you made on the knob will reflects, sometime it just goes off the grid. eventually the scope did catch up from behind, its just the GUI "update pooling" will take sometime before its visible on the screen. i just need to learn the right knob turn speed from experience, to get the increment i want on the next screen update... except if i want a very precise Voffset then i will go as you said "safecracker mode"?
Nature: Evolution and the Illusion of Randomness (Stephen L. Talbott): Its now indisputable that... organisms “expertise” contextualizes its genome, and its nonsense to say that these powers are under the control of the genome being contextualized - Barbara McClintock
 

Offline markone

  • Frequent Contributor
  • **
  • Posts: 660
  • Country: it
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #71 on: February 05, 2016, 07:00:39 pm »
Sorry for the OT,

but take a look at this beauty, Dave's FFT reference signal (1Mhz, FM 5Khz, Dev 500Hz) under italian ELAD FDM-S2 16bit-122.88MSa/s sampler analysis, until 60Mhz and 6Mhz span blows out the windows everything showed here in terms of sensitivity, resolution, dynamic and update rate (almost real time) at mere 430 Euro plus taxes.

Not a DSO, but if you need high resolution FFT within its spec, coupled with a DC block and a fixed attenuator and / or active probe for High-Z it does a great job.


 

Offline AlessandroAU

  • Regular Contributor
  • *
  • Posts: 168
  • Country: au
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #72 on: February 06, 2016, 12:00:11 am »
He is the FFT from an AD7760 eval board.

2.5MSPS, 24bit
 

Offline rs20

  • Super Contributor
  • ***
  • Posts: 2317
  • Country: au
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #73 on: February 06, 2016, 12:12:46 am »
He is the FFT from an AD7760 eval board.

2.5MSPS, 24bit

Now try it with the OP's 1 MHz carrier  :P

Also, did you do the FFT on your PC? How many points? How long did it take? Doing an FFT is trivial, after all, if you have GB of memory and no particular time limit...
 

Offline hpw

  • Frequent Contributor
  • **
  • Posts: 345
  • Country: 00
Re: EEVblog #845 - Oscilloscope FFT Comparison
« Reply #74 on: February 07, 2016, 12:20:54 pm »

If some is interested about the dynamic range on 12...40 bit

Hp
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf