Oscilloscope for fft functions

[freebil]

Does hantek dso5102p has better ftt than rigol ds1054z?

[Wuerstchenhund]

Important note: in FFT mode you use it like a spectrum analyzer, meaning you will have a GUI that allows you to center freq, span, Res BW etc and the tome domain parameter like sampling rate etc are updated accordingly. Very convenient to use!

That is nothing new, really. Even my old Agilent Infiniium 8000 comes with an 'SA' like interface (standard), and it has been a paid-for option (SPECTRUM) on LeCroy scopes since 2006. Like it is a paid-for option on your RTO (K18).

Also, you may not care, but then RTO is one of the few scopes that retains the phase information of the FFT calculation. Many scopes throw the phase information away and only offer you the magnitude.

Again, that is nothing new. Pretty much every better high end scope has been doing phase for almost 20yrs now. My old 200Mhz WaveRunner LT could do that, and that scope came out in 1998. And again, my old Infiniium 8000 can do that, as pretty much every high-end scope I worked with that was made after 2000.

The display representation looks nice, but I can't see anything really new here.

the combination of having a digital 16 bit MSO, a 4 channel scope and a SA in one box for cross domain analysis is a huge advantage.

You don't have a 16bit scope. You have an 8bit scope (8bit ADCs) with a 16bit software option for which R&S charges pretty much an arm and a leg, with all the associated limitations. It's not the same as a true 16bit scope, or even a true 12bit scope.

You beat me to the punch rfbroadband, I saw this thread yesterday and thought "I should link to info on the RTO to show what a non-toy fft looks like." I wish other vendors took the function half as seriously as R&S.

They do. If you want to see some really sophisticated FFT then have a look at the Keysight Infiniium S.

[rsjsouza]

the combination of having a digital 16 bit MSO, a 4 channel scope and a SA in one box for cross domain analysis is a huge advantage.

You don't have a 16bit scope. You have an 8bit scope (8bit ADCs) with a 16bit software option for which R&S charges pretty much an arm and a leg, with all the associated limitations. It's not the same as a true 16bit scope, or even a true 12bit scope.

I think he meant 16 digital inputs...

[rfbroadband]

the FFT option is free of charge with every RTO scope. The option K18 is an advanced Spectrum analysis option. The spectrum analysis is thus free in every standard config.

I always find it difficult to review a product without having used it myself.

I have used and owned various Tektronix, LeCroy, Agilent (Keysight) scopes over many years. When R&S did come out with the their first scope years ago, I have to admit I, did not even look at it. Once I got my hands on one and evaluated it I purchased one, immediately. At some point I gave a demo to a Keysight sales person and his silence (nope we can't do that, nope we can't do this) was priceless. He then arranged for Keysight to send a scope R&D/marketing guy from Colorado to visit my lab for feedback on their scopes. If I remember correctly, 80% of the > 1.5 hour session was spent on FFT and frequency domain measurements. He took lots of notes.

Some technical marketing guy at R&S did his homework. It is the first scope where I use the FFT function on a regular basis, because a majority of the built scope FFT functions are indeed toy implementations.

In summary, there a many fine scopes out there in the market. The FFT, SA analysis capability including measurement speed, dynamic range, FFT gating,  noise floor etc.the RTO is outstanding. If someone is looking for a new scope for a work environment (not as a hobbyist due to cost) and wants frequency domain capability, I seriously suggest you at least evaluate the RTO.

[rfbroadband]

rsjsouza,
thank you. Certainly I meant 16 bits digital inputs for the MSO option.

The HD option (that claims 16bit) is only a software option and it is not delivering 16bits. The HD option reduces the noise floor by ~9.5 dB, thus 1.5 ENOB. It is a useful option, but the marketing for the HD option is very misleading and I communicated that very clearly! to R&S.

[Wuerstchenhund]

I have used and owned various Tektronix, LeCroy, Agilent (Keysight) scopes over many years. When R&S did come out with the their first scope years ago, I have to admit I, did not even look at it. Once I got my hands on one and evaluated it I purchased one, immediately. At some point I gave a demo to a Keysight sales person and his silence (nope we can't do that, nope we can't do this) was priceless. He then arranged for Keysight to send a scope R&D/marketing guy from Colorado to visit my lab for feedback on their scopes. If I remember correctly, 80% of the > 1.5 hour session was spent on FFT and frequency domain measurements. He took lots of notes.

That's why the best way to evaluate a scope is to request a loaner for say a week so you can spend time with the scope yourself. Relying on that the sales droid knows how to operate and demo advanced functions can be pretty hit and miss, especially if it's a more sophisticated scope.

Some technical marketing guy at R&S did his homework. It is the first scope where I use the FFT function on a regular basis, because a majority of the built scope FFT functions are indeed toy implementations.

Glad you like the RTO but none of the FFT implementations of other scopes in that class are "toy" implementations, and haven't been so for many years. If you as you said had LeCroy and Agilent/Keysight scopes before then you should have been aware that phase information is not thrown away, or that an SA like interface isn't something new, either.

BTW, what's the max FFT size the RTO1000 can do? The R&S spec sheet doesn't seem to mention.

In summary, there a many fine scopes out there in the market. The FFT, SA analysis capability including measurement speed, dynamic range, FFT gating,  noise floor etc.the RTO is outstanding. If someone is looking for a new scope for a work environment (not as a hobbyist due to cost) and wants frequency domain capability, I seriously suggest you at least evaluate the RTO.

This, but I'd strongly suggest to also look at the Keysight Infiniium S. It's FFT is really good (and it can do up to 16 at the same time, vs only 4 with the RTO), and you get a true 10bit ADC on top of it.

[rfbroadband]

sorry, I don't recall the max. FFT size.

I looked at the Keysight Infiniium S for a few days a year ago. It has better jitter performance than the RTO, but the improved jitter was not that important to me.

[Wuerstchenhund]

sorry, I don't recall the max. FFT size.

Doesn't the scope show how much memory it is using for FFT?

I looked at the Keysight Infiniium S for a few days a year ago. It has better jitter performance than the RTO, but the improved jitter was not that important to me.

It's not just the lower jitter, it's better SFDR and SNR, faster ASICs, FFT with very large sample sizes and so on. Especially for FFT, there's currently no better scope in the 1 to 8 GHz class.

[pascal_sweden]

If you want to make your bench look professional, just get both an oscilloscope and a separate spectrum analyzer. The more test equipment, the better and the more impressive it will look. If your bench table is not big enough, go to Ikea and buy a bigger table. Mount an anti-static mat on top of the table to turn it in a work space

Hereby a quick calculation, to proof that it doesn't take 4000-6000 EUR to get good FFT at home:

Siglent SSA3021X: 1500 EUR (supports up to 3.2 GHz after Werewolf mode)
Rigol DS2072A: 750 EUR (supports up to 300 MHz after Riglol)

Conclusion: For 2250 EUR you are set and have a 2 GS/s oscilloscope, a 3.2 GHz spectrum analyzer.
I didn't count the Ikea table, as I assume your bench space is big enough for both devices =)

You have plenty of budget left, so you could as well throw in a Siglent SDG2042X signal generator which got extremely good reviews! It supports up to 120 MHz sine waves.

[Wuerstchenhund]

If you want to make your bench look professional, just get both an oscilloscope and a separate spectrum analyzer. The more test equipment, the better and the more impressive it will look. If your bench table is not big enough, go to Ikea and buy a bigger table. Mount an anti-static mat on top of the table to turn it in a work space

Great idea! And hey, why not randomly place some wires, old PCBs and a few screwdrivers on there as well?

It will look yuuuge, it will look beautiful, really, you'll see! 

Hereby a quick calculation, to proof that it doesn't take 4000-6000 EUR to get good FFT at home:

Siglent SSA3021X: 1500 EUR (supports up to 3.2 GHz after Werewolf mode)
Rigol DS2072A: 750 EUR (supports up to 300 MHz after Riglol)

Conclusion: For 2250 EUR you are set and have a 2 GS/s oscilloscope, a 3.2 GHz spectrum analyzer.

 

I know the fact that someone talks about making a bench "look professional" should have been a clear indicator that there's very little real-world experience involved ('nuff said), and therefore its very likely futile to ask but:

Please explain how a cheap B-brand scope with piss-poor FFT (2kpts?) and a cheap B-brand SA with a lowly max RBW of only 1MHz (according to the spec sheet) is an adequate replacement for a scope with proper FFT, including how this setup can be used to assess low frequency spectra from DC to 9kHz, or modern wideband signals with BW of multiple tens of MHz, both things which are a very common use for FFT in a good scope.

 

[pascal_sweden]

According to the specs (and verified by actual test users):
10 Hz Minimum Resolution Bandwidth (RBW)

[Wuerstchenhund]

According to the specs (and verified by actual test users):
10 Hz Minimum Resolution Bandwidth (RBW)

Well done for stating data that is completely irrelevant

I was talking about the max RBW, which for the SSA3000X isn't 10Hz, it's 1MHz. And since you apparently don't understand why that is necessary, the RBW pretty much tells you how big a slice the SA can capture in one go. 1MHz, which btw is pretty poor even compared to many 30yr old boat anchor SAs, is absolutely insufficient to look at a modern wideband signal.

I have to say it's truly mind boggling how you can always come up with some silly "alternative", usually involving some poor-spec B-brand kit, while being clearly completely oblivious to what the specs actually mean, how this stuff works and where its limitations are, or how it's actually used, and then conclude that this is an appropriate replacement for what was initially discussed. 

[jjoonathan]

If you want to make your bench look professional, just get both an oscilloscope and a separate spectrum analyzer. The more test equipment, the better and the more impressive it will look. If your bench table is not big enough, go to Ikea and buy a bigger table. Mount an anti-static mat on top of the table to turn it in a work space

Great idea! And hey, why not randomly place some wires, old PCBs and a few screwdrivers on there as well?

It will look yuuuge, it will look beautiful, really, you'll see! 

Hereby a quick calculation, to proof that it doesn't take 4000-6000 EUR to get good FFT at home:

Siglent SSA3021X: 1500 EUR (supports up to 3.2 GHz after Werewolf mode)
Rigol DS2072A: 750 EUR (supports up to 300 MHz after Riglol)

Conclusion: For 2250 EUR you are set and have a 2 GS/s oscilloscope, a 3.2 GHz spectrum analyzer.

 

I know the fact that someone talks about making a bench "look professional" should have been a clear indicator that there's very little real-world experience involved ('nuff said), and therefore its very likely futile to ask but:

Please explain how a cheap B-brand scope with piss-poor FFT (2kpts?) and a cheap B-brand SA with a lowly max RBW of only 1MHz (according to the spec sheet) is an adequate replacement for a scope with proper FFT, including how this setup can be used to assess low frequency spectra from DC to 9kHz, or modern wideband signals with BW of multiple tens of MHz, both things which are a very common use for FFT in a good scope.

 

Also, fast events that require a frequency-domain digital phosphor display to catch. You could add a SDR and an upconverter to round out the B team, calibration is sure to give you hours of fun!

[pascal_sweden]

Wuerstchenhund, you seem a very bright person with years of experience in this industry area.
I am sure you are too busy with your current work situation, but if you would start making YouTube movies, you would be very successful, and a lot of people out here would get a lot smarter with your help

[Wuerstchenhund]

Wuerstchenhund, you seem a very bright person with years of experience in this industry area.
I am sure you are too busy with your current work situation, but if you would start making YouTube movies, you would be very successful, and a lot of people out here would get a lot smarter with your help

There are certainly a few things I think I do quite well but unfortunately being in front of a camera isn't one of them. I'm sure if I made YT movies then they'd make Uwe Boll's movies look like works of art in comparison

Also, living in a country where houses and the rooms in them are generally already on the 'compact' side, my personal workshop is currently pretty cramped so I wouldn't really know where to record something (we'll probably move to some place where I shall find more spacious facilities, but that could take well another year or so).

I'm also not really a video guy, and a lot of stuff can be explaned very well even in the format of a forum, supported by images (i.e. screenshots) where required.

[David Hess]

Also, you may not care, but then RTO is one of the few scopes that retains the phase information of the FFT calculation. Many scopes throw the phase information away and only offer you the magnitude. As a result you can use the RTO for large dynamic range bode plot measurements plotting mag and phase response over freq.

Yay! \o/   (Amazingly enough, Dave's forum doesn't have the right emoticon.)

Now what is the lowest end oscilloscope which returns phase information with its FFT instead of throwing it out?

[David Hess]

Also, you may not care, but then RTO is one of the few scopes that retains the phase information of the FFT calculation. Many scopes throw the phase information away and only offer you the magnitude.

Again, that is nothing new. Pretty much every better high end scope has been doing phase for almost 20yrs now. My old 200Mhz WaveRunner LT could do that, and that scope came out in 1998. And again, my old Infiniium 8000 can do that, as pretty much every high-end scope I worked with that was made after 2000.

The oldest DSOs which I know of which provide magnitude and phase from FFTs are the Tektronix TDS520 and TDS540 (with option 2F) released in 1992 and the DSA601 and DSA602 (11k series) released in 1990.  Unfortunately Tektronix limited this feature to DSOs with bandwidths of 500 MHz and higher which I assume was for purposes of market segmentation

I am always on the lookout for other examples which may have lower bandwidth but must be less expensive.

[Wuerstchenhund]

Now what is the lowest end oscilloscope which returns phase information with its FFT instead of throwing it out?

Difficult. I *believe* (don't have access to one at the moment to check) the LeCroy WaveSurfer 3000 can do phase, and this starts at 2ch 200MHz and $3k.

With Keysight, none of the DSOX variants do phase so the "cheapest' one is the Infiniium DSO9064A (600MHz 4ch for $16k).

With R&S, I think FFT phase starts with the RTE, of which the cheapest variant (RTE1022 200MHz 2ch) starts at roughly $6k.

With Tek, the cheapest scope that offers phase seems to the the MDO4000 which for 4ch 200Mhz starts at $7k.

I doubt any Rigol, Siglent (aside from the SDS3000 maybe) or GW Instek scope does phase.

The oldest DSOs which I know of which provide magnitude and phase from FFTs are the Tektronix TDS520 and TDS540 (with option 2F) released in 1992 and the DSA601 and DSA602 (11k series) released in 1990.  Unfortunately Tektronix limited this feature to DSOs with bandwidths of 500 MHz and higher which I assume was for purposes of market segmentation

I am always on the lookout for other examples which may have lower bandwidth but must be less expensive.

I'm pretty sure the old LeCroy 9300 Series (which if I remember right starts at 200MHz) could do phase, at least optional. I wonder what the FFT capabilities are for the 7200/7400 and 9400 Series from the '80s.

There were a few other scopes (i.e. from HP) back then that did offer FFT but I can't remember if they could do phase or not.

[The Electrician]

attached a bode plot measured with the RTO and compared against network analyzer measurements E5061B

How did you create these plots?  I notice the horizontal scale is logarithmic; did you need the K18 option to do this?

[borjam]

I'm pretty sure the old LeCroy 9300 Series (which if I remember right starts at 200MHz) could do phase, at least optional. I wonder what the FFT capabilities are for the 7200/7400 and 9400 Series from the '80s.

From the 9400 manual

I imagine that Lecroy's approach at the time was: "let's display that FFT in every conceivable (and doable) way". Indeed it can be painfully slow if you select too many points (up to 20 seconds if I remember correctly) but it just does it.

[rf-loop]

According to the specs (and verified by actual test users):
10 Hz Minimum Resolution Bandwidth (RBW)

Well done for stating data that is completely irrelevant

I was talking about the max RBW, which for the SSA3000X isn't 10Hz, it's 1MHz. And since you apparently don't understand why that is necessary, the RBW pretty much tells you how big a slice the SA can capture in one go. 1MHz, which btw is pretty poor even compared to many 30yr old boat anchor SAs, is absolutely insufficient to look at a modern wideband signal.

I have to say it's truly mind boggling how you can always come up with some silly "alternative", usually involving some poor-spec B-brand kit, while being clearly completely oblivious to what the specs actually mean, how this stuff works and where its limitations are, or how it's actually used, and then conclude that this is an appropriate replacement for what was initially discussed. 

Who need and who do not need over 1MHz RBW in normal sweeping SA.
Many peoples still survive with example HP8560E and HP8590E series, example, And yes, they are old and yes, they are not even near state of art in any thing.

I have used tens of years spectrum analyzers and I have never need over 3MHz but this is in my use. . Who need more, he buy what he need. There is not universal machine what is good in all things including price. For <2kEur and talking NEW equipments there is not many...  Buyer need brain. He need know what he need and what he do not need and if he do not know what he need but still want buy, why he need buy any.it can ask why. Just for fun or decorate his work room. Most of people who Need SA also know what he need. So he do not need buy universal all in one all perfect machine.  Oh..yes...I know these peoples who never know what they perhaps need tomorrow or next year or after ten year. I remember one people who was, and is still, in this trap. He still take pictures using film camera because next year come more features and some have this and that feature but it is too expensive but perhaps next year this is available also in reasonable price class. I believe he stay infinite without digital camera but all time he read reviews and compares and follow what new is coming and what all perhaps need...

Sometimes, not always, also things can do with skills and using clever methods for jump over some limitations in equipments.

Some may buy 30kEur measurement system when he is doing his telescope main mirror and some use 0.1 eur razor blade and candle. Both get same result in final. Other use brain and skills and do himself, second delegates thinking to machine but then he have head ache when he thing if machine result can trust.

Btw, there is undocumented 3MHz RBW in SSA...   Simple, if you need something what equipment do not have, then buy equipment what meet your needs.  So extremely simple. If I need spectrum and SSA specifications etc meet my needs that's it. If I do not need 100MHz wide realtime analyzer for 6GHz then why need tell that SA noname is poor nad crap when it do not have it.

[Wuerstchenhund]

Who need and who do not need over 1MHz RBW in normal sweeping SA.
Many peoples still survive with example HP8560E and HP8590E series, example, And yes, they are old and yes, they are not even near state of art in any thing.

Many people survive with less than a $1 per day. That doesn't suggest that this is worth emulating.

And I'm really not sure why you bring up these old boat anchors into a discussion about FFT in a scope.

I have used tens of years spectrum analyzers and I have never need over 3MHz but this is in my use.

Great. I regularly need more than 3MHz BW, and while I have an SA that goes up to 20MHz RBW (Anritsu MS8609A) I regularly use FFT on my 8yr old scopes to look at wideband signals up to 160MHz.

For <2kEur and talking NEW equipments there is not many

Yes, in that class aren't many new scopes with good FFT (pretty much only the GW Instek GDS-1000B and GDS-2000E), and none which do phase.

...  Buyer need brain. He need know what he need and what he do not need and if he do not know what he need but still want buy, why he need buy any.it can ask why. Just for fun or decorate his work room. Most of people who Need SA also know what he need. So he do not need buy universal all in one all perfect machine.  Oh..yes...I know these peoples who never know what they perhaps need tomorrow or next year or after ten year. I remember one people who was, and is still, in this trap. He still take pictures using film camera because next year come more features and some have this and that feature but it is too expensive but perhaps next year this is available also in reasonable price class. I believe he stay infinite without digital camera but all time he read reviews and compares and follow what new is coming and what all perhaps need...

Sometimes, not always, also things can do with skills and using clever methods for jump over some limitations in equipments.

Some may buy 30kEur measurement system when he is doing his telescope main mirror and some use 0.1 eur razor blade and candle. Both get same result in final. Other use brain and skills and do himself, second delegates thinking to machine but then he have head ache when he thing if machine result can trust.

Not saying you're wrong but this isn't really relevant to the discussion. How about we focus on the topic?

And I'd like to see how you "work around" the lack of proper bandwidth in your instrumentation 

Btw, there is undocumented 3MHz RBW in SSA...   Simple, if you need something what equipment do not have, then buy equipment what meet your needs.  So extremely simple. If I need spectrum and SSA specifications etc meet my needs that's it. If I do not need 100MHz wide realtime analyzer for 6GHz then why need tell that SA noname is poor nad crap when it do not have it.

So if they have 3MHz RBW which works properly why is this even undocumented? 

But the real point which you so clearly missed was not if the SSA-X is 'crap' or not (I leave this for others to decide),  the point was that the SSA-X was suggested as a perfect replacement for a scope with proper FFT. which it clearly isn't, because it's pretty useless for analyzing modern wideband signals due to it's very low max RBW.

[Wuerstchenhund]

From the 9400 manual

Thanks!   

So even the old 9400 could do phase. That's pretty impressive for a scope from the mid-'80s.

I imagine that Lecroy's approach at the time was: "let's display that FFT in every conceivable (and doable) way". Indeed it can be painfully slow if you select too many points (up to 20 seconds if I remember correctly) but it just does it.

Well, at least you can get the data. Considering the performance of processors back then it's no surprise that it's slow.

[borjam]

So even the old 9400 could do phase. That's pretty impressive for a scope from the mid-'80s.

I imagine that Lecroy's approach at the time was: "let's display that FFT in every conceivable (and doable) way". Indeed it can be painfully slow if you select too many points (up to 20 seconds if I remember correctly) but it just does it.

Well, at least you can get the data. Considering the performance of processors back then it's no surprise that it's slow.

Oh I'm not criticizing it, quite the contrary! It runs on an ancient Motorola 68000.

It reflects the mentality of a long gone era I think. The minimal FFT present nowadays in many oscilloscopes is just a checkbox feature. The people from Lecroy, on the other side, probably thought something like "let's sqeeze every possible bit ouf of this function". Probably their background in the Physics background helped as well.

[David Hess]

Now what is the lowest end oscilloscope which returns phase information with its FFT instead of throwing it out?

Difficult. I *believe* (don't have access to one at the moment to check) the LeCroy WaveSurfer 3000 can do phase, and this starts at 2ch 200MHz and $3k.

With Keysight, none of the DSOX variants do phase so the "cheapest' one is the Infiniium DSO9064A (600MHz 4ch for $16k).

With R&S, I think FFT phase starts with the RTE, of which the cheapest variant (RTE1022 200MHz 2ch) starts at roughly $6k.

With Tek, the cheapest scope that offers phase seems to the the MDO4000 which for 4ch 200Mhz starts at $7k.

I doubt any Rigol, Siglent (aside from the SDS3000 maybe) or GW Instek scope does phase.

The oldest DSOs which I know of which provide magnitude and phase from FFTs are the Tektronix TDS520 and TDS540 (with option 2F) released in 1992 and the DSA601 and DSA602 (11k series) released in 1990.  Unfortunately Tektronix limited this feature to DSOs with bandwidths of 500 MHz and higher which I assume was for purposes of market segmentation

I am always on the lookout for other examples which may have lower bandwidth but must be less expensive.

I'm pretty sure the old LeCroy 9300 Series (which if I remember right starts at 200MHz) could do phase, at least optional. I wonder what the FFT capabilities are for the 7200/7400 and 9400 Series from the '80s.

There were a few other scopes (i.e. from HP) back then that did offer FFT but I can't remember if they could do phase or not.

I will keep these in mind next time I go search for this feature in a new or old DSO, thanks.

While not quite a substitute, the low performance but low cost USB oscilloscopes from Syscomp Electronic Design support network analysis using their built in function generators.  This is another thing which would make me consider buying a new DSO but alas, none support even this.

http://www.syscompdesign.com/Instruments_ep_42.html