Oscilloscope for fft functions

[Electric2000]

What is the max fft point of rigol ds1054z?

[tautech]

As this is an older thread without mention of new models, SDS1202X-E should be mentioned with 1 Mpts FFT.

[ruairi]

Tautech,

Perhaps you are not aware how much of a turnoff your relentless Siglent sales pitch is?  It completely undermines the brand IMO and I have ruled out Siglent products because of it.  The question above your post was about the FFT points on a Rigol, nothing to do with Siglent.

Being diligent in answering Siglent related queries or correcting misinformation is good business, what you do is not.

Ruairi

[tautech]

Tautech,

Perhaps you are not aware how much of a turnoff your relentless Siglent sales pitch is?  It completely undermines the brand IMO and I have ruled out Siglent products because of it.  The question above your post was about the FFT points on a Rigol, nothing to do with Siglent.

Being diligent in answering Siglent related queries or correcting misinformation is good business, what you do is not.

Ruairi

Ruairi, I try to make posts that are informative and factual to the best of my ability.

If I was to reply directly to E2000's post I would have quoted it specifically, instead I checked all posts in this thread and found 'up to date' info missing and simply added it, it was only that E2000's post brought this thread to my attention.
Each and every viewer/reader/member will take my input in a different way and if you take some time to check my postings they cover all manner of subjects, that I also hope is reflected in my signature.

FYI of the several 100's of Siglents units I have sold in the few years I've been a distributor only a very few are to EEVblog members so my efforts here can hardly be called self interest.

[hendorog]

Tautech is enthusiastic to be sure, but I don't think your comment is fair.

The first thing I thought of when I saw "Oscilloscope for FFT" in my unread posts list, was that Siglent FFT picture as I know it has very nice FFT for the price.
I came to the thread to point it out and saw he had beat me to it.

[ruairi]

Tautech,

We'll have to agree to disagree on your approach.  For me Daniel at Keysight to Rich at R&S are an example of what a company rep can bring to the forum, they always seem to arrive when needed but don't wedge their products in to every conversation. That predisposes me to their brands and I have gear from each company, your posts have the opposite effect for me.

Hendorog - it's a personal opinion and I'm probably in the minority, I'm happy that you find Tautech's posts helpful.

Apologies for the thread derail.

Cheers,
Ruairi

[nctnico]

GW Instek has a new MDO-2000E series which is geared towards FFT / frequency domain analysis.
http://www.gwinstek.com/en-global/products/Oscilloscopes/Digital_Storage_Oscilloscopes/MDO-2000E_Series

[MrW0lf]

What is the max fft point of rigol ds1054z?

16k

[Fungus]

What is the max fft point of rigol ds1054z?

16k

Unless you own an Ethernet cable and a PC, in which case the FFT size can be the entire memory of the 'scope.

For serious FFT a dedicated spectrum analyzer will always beat a DSO for high frequencies and a $200 Analog Discovery will usually beat one at lower frequencies (it has a 14-bit ADC).

And now we pass you to your regularly scheduled Picoscope fans...

[MrW0lf]

And now we pass you to your regularly scheduled Picoscope fans...

BTW did you ever consider why I like Picoscopes? Heres why:
If there is no feature (like graphing phase between two signals) I can just take couple cups of coffee, and dream up some math formula that will settle the matter:
https://www.picotech.com/support/topic31641.html

Green trace is phase(A,B). Work still in progress but shows certain promise that might be related to bode plots However would not hurt if they finally implement anti-alias in analytical mode

[Fungus]

BTW did you ever consider why I like Picoscopes?

Not really, no. 

[b_force]

What is good and what is poor? 

In 90% of the cases I see people using the wrong type of time window for what they are looking for and blame the equipment.
It also depends on the bandwith you're looking for.

[nctnico]

For serious FFT a dedicated spectrum analyzer will always beat a DSO for high frequencies

No because a DSO shows the time and frequency domain together. Sometimes it is very usefull to see which part of a signal (time domain) influences which part of the frequency domain.

[Wuerstchenhund]

Unless you own an Ethernet cable and a PC, in which case the FFT size can be the entire memory of the 'scope.

No, because the longest sequence a DS1054z can capture is 24Mpts (which is still quite a lot) so that's longest sequence you can use as a basis for FFT.

Of course you could stitch together multiple acquisitions but then you'd introduce errors.

Forget it. Somehow I read "..entire memory of the computer" instead of "...scope"    Mea culpa!

For serious FFT a dedicated spectrum analyzer will always beat a DSO for high frequencies

No, not always. A decent lab-grade SA offers better dynamic range and overall RF performance, however capturing high BW signals is generally easier with a scope (which can use it's full analog BW) while the RBWs of a swept SA are a lot more limited (usually between 1MHz and 10MHz, with very few offering more than 10MHz) unless you use a newer high-end model.

For serious FFT a dedicated spectrum analyzer will always beat a DSO for high frequencies

No because a DSO shows the time and frequency domain together. Sometimes it is very usefull to see which part of a signal (time domain) influences which part of the frequency domain.

True, but many better SAs can show the time domain as well, often even time and frequency domains together.

[Fungus]

Unless you own an Ethernet cable and a PC, in which case the FFT size can be the entire memory of the 'scope.

No, because the longest sequence a DS1054z can capture is 24Mpts (which is still quite a lot) so that's longest sequence you can use as a basis for FFT.

And 24Mpts is ... what percentage of the memory of the 'scope? 

[Karlo_Moharic]

If you want to analyze signals in frequency domain and you are on budget , try software defined radio (sdr) on ebay. They come in all kinds of price ranges  and with little effort they can make a usable signal analayzer.

[shteii01]

Tek DSA601/602...

[Mr. Scram]

Even though it's probably not going to be it king of the hill, it's good to note that there is an application to do FFT externally with the DS1054Z. That way you can record with the oscilloscope, but do the heaving lifting with your computer. Also, newer firmware versions apparently have improved FFT.

http://hackaday.com/2015/09/22/a-better-spectrum-analyzer-for-your-rigol-scope/

[Wuerstchenhund]

Tek DSA601/602...

Seriously? 

http://w140.com/tekwiki/wiki/DSA600

"FFT: complex, forward and reverse, max. 16384 points; windows: Rectangular, Bartlett, Hanning, Hamming, Blackman, Blackman-Harris..."

[Kalvin]

If you want to analyze signals in frequency domain and you are on budget , try software defined radio (sdr) on ebay. They come in all kinds of price ranges  and with little effort they can make a usable signal analayzer.

HackRF (and clones) is quite inexpensive device going from 1MHz to 6GHz, and there are some FFT analyzer software available. For example this https://github.com/pavsa/hackrf-spectrum-analyzer is build for the spectrum analysis purposes. Even using a cheap SDR dongle can be used up to 1.7GHz. There are some SDR dongles which can go down to very low frequencies using so called direct sampling mode: http://www.rtl-sdr.com/rtl-sdr-direct-sampling-mode/. However, the dynamic range will be around 48dB using these less expensive alternatives. So, depending on your budget there are some options how to proceed.

As others have already pointed out, using a suitable PC software with the oscilloscope for analyzing the captured waveform will also be very handy. So, you do not have to select the oscilloscope per FFT, just make sure that you can get the samples out to PC.

[David Hess]

Tek DSA601/602...

Seriously? 

http://w140.com/tekwiki/wiki/DSA600

"FFT: complex, forward and reverse, max. 16384 points; windows: Rectangular, Bartlett, Hanning, Hamming, Blackman, Blackman-Harris..."

Tekwiki neglects to mention it but the 11K series also supports waveform math functions like differentiation and FFT magnitude *and* phase which together may be used to do vector network analysis using a step signal source.  Averaging and filtering may be applied to the FFT instead of the time domain waveform.  Math also allows the FFT magnitude display to be corrected for ENBW.  The 11K series are the oldest DSOs that I know of which support these things.

At some point I may still pick up one of these DSOs for the above but I would prefer a better option short of a new DSO that costs more than my car.

[MikeP]

This is interesting thread. I asked a question about FFT in the RTB2000-thread. Unfortunately, the answer still is not present.

 Please explain - why I see so big difference in results? What settings need to be used? Why there is dependence from the THD level? In other words - the dependence from the settings decreases at a large THD.

 Now I use the rejector filter. These good old days!

 The RTB2000 is the great tool. But I can't use it.

 Thanks.
 Mike.

[nctnico]

FFT works best in sample mode. Averaging and high-res will filter the signal and peak detect adds extra noise. You also need to be aware of aliasing products at low samplerates especially when the FFT length is short so you can't have high samplerates (yes, 128kpts is short).

[Wuerstchenhund]

Tek DSA601/602...

Seriously? 

http://w140.com/tekwiki/wiki/DSA600

"FFT: complex, forward and reverse, max. 16384 points; windows: Rectangular, Bartlett, Hanning, Hamming, Blackman, Blackman-Harris..."

Tekwiki neglects to mention it but the 11K series also supports waveform math functions like differentiation and FFT magnitude *and* phase which together may be used to do vector network analysis using a step signal source.  Averaging and filtering may be applied to the FFT instead of the time domain waveform.  Math also allows the FFT magnitude display to be corrected for ENBW.  The 11K series are the oldest DSOs that I know of which support these things.

They aren't. The LeCroy 9400A (which came out in 1985) supports these when soft option WP02 is installed, which allows advanced FFT up to 25kpts. The LeCroy 7200a modular scope (1989) should, too, although I don't know what the max FFT size is.

At some point I may still pick up one of these DSOs for the above but I would prefer a better option short of a new DSO that costs more than my car.

If you want to do some TDR stuff then there are quite a few low-cost options available. The LeCroy 9300 Series from the '90s supports the above features with the (now free) software options WP02 and WP03, while offering FFT up to 1Mpts. There are various models of the 9300 Series with different sample rates and bandwidths, most of which can be had for a few hundred dollars in working condition (although these days you can find the successor LC Series for similar money, with faster processing and FFT up to 8M).

Also, there are TDR options for the HP 54120A/B/D sampling scope mainframes, one up to 20Ghz and one up to 50Ghz (scope and TDR mainframe are called 54120T). The scope frames are cheap, however the TDR heads are not but with some patience you can find working examples of scope + head for <£1200.

[David Hess]

FFT works best in sample mode. Averaging and high-res will filter the signal and peak detect adds extra noise. You also need to be aware of aliasing products at low samplerates especially when the FFT length is short so you can't have high samplerates (yes, 128kpts is short).

Unfortunately many DSOs do what you describe and apply averaging to the waveform before the FFT when what is desired is to average the FFT output.