Rigol DHO804 Test and Compare Thread

[2N3055]

I'm trying to figure out if my cheapo 150Mhz probes will keep me below Nyquist if I enable all 4 channels with 250Mhz bandwidth enabled. I just poked the probe into the BNC of my SigGen to see what I got.

This is my rise time with the crocodile clip:

This is my rise time with the little spring accessory:

There's quite a big difference between the two!

Conclusions...

Crocodile clip... yes.
Spring... not so much.

First of all thank you for recent info. I did not find all of that in manual, thank you for that.

Also I like the probe test. When in doubt, measure.
Also if you remember previous topics about probes, by Perfoma01 and others, similar results to that.
If you have low enough source impedance, probe will have much higher BW than specced.

As I said, in order to be sure it won't alias with all 4ch , scope should be run in 100 MHz configuration.
With 1 or 2 ch it should not alias much (2ch) or at all (1ch).

Aliasing will be most visible as unstable edges or large overshoots on square wave , but you will generally see it more as spurious tones in FFT. They fold back, like in the mirror..

By using manual control of sampling on a scope we can simulate effects of aliasing so we can see how they look like.
Let me explain with images: same 20 MHz squarewave with 1ns edges (I deliberately stayed from faster edges, this should be similar to Arduino UNO I/O pins ).
It was sampled at 1.25 GS/s, then at 500MS/s and 250 MS/s. Sorry my scope cannot do 625 and 312.5 and this not supposed to be exactly the same, but to illustrate the effect so people can recognize it.. To simulate BW, I enable 200 MHz BW limit but on your scope it will be more like 250 Mhz so aliasing might be easier to show.

On 1.25 GS/s it  looks pretty much perfect. on two lower sampling rates it goes down the hill. Make note that repetition rate is only 20MHz and that edges are no very fast. 1ns edges or faster are pretty much guaranteed today in anything digital., unless you are working only on legacy TTL logic and retro computing...
 
Use this as baseline for comparison. Hope it helps with your evaluation.

[ebastler]

By using manual control of sampling on a scope we can simulate effects of aliasing so we can see how they look like. [...]
On 1.25 GS/s it  looks pretty much perfect. on two lower sampling rates it goes down the hill. [...]

Thanks for those screenshots! But they got me wondering: Is it really aliasing we see there, i.e. the presence of image frequencies, or just the effect of approximating the square wave with a limited number of harmonics? Would these look any different if you had a suitable analog low-pass at the input which blocks all frequency components above the Nyquist frequency?

Edit: For comparison, could you keep the high sample rate but add a digital low pass operation at Nyqist? Thanks!

[Performa01]

By using manual control of sampling on a scope we can simulate effects of aliasing so we can see how they look like. [...]
On 1.25 GS/s it  looks pretty much perfect. on two lower sampling rates it goes down the hill. [...]

Thanks for those screenshots! But they got me wondering: Is it really aliasing we see there, i.e. the presence of image frequencies, or just the effect of approximating the square wave with a limited number of harmonics? Would these look any different if you had a suitable analog low-pass at the input which blocks all frequency components above the Nyquist frequency?

Edit: For comparison, could you keep the high sample rate but add a digital low pass operation at Nyqist? Thanks!

The answer to all these questions as well as the requested screenshots (and many more) can be found in the document at the end of the first posting in the following thread.

https://www.eevblog.com/forum/testgear/sds2000x-plus-bandwidth-aliasing-application-note/msg3919544/#msg3919544

[ebastler]

The answer to all these questions as well as the requested screenshots (and many more) can be found in the document at the end of the first posting in the following thread.

https://www.eevblog.com/forum/testgear/sds2000x-plus-bandwidth-aliasing-application-note/msg3919544/#msg3919544

Thank you! That document deserves time to work through properly...

Upon quickly skimming it, I could not find screenshots which compare the effect of capturing a square wave with aliasing (due to a limited sampling rate 2F) with the effect of limited bandwidth (high sampling rate, but lowpass at cutoff frequency F). Could you point me to the right figures please?

Edit: Ok, I think I got it in principle. The "wiggles" on 2N3055's screenshots can be explained by bandwith limitations, while the jitter is the effect of aliasing, right? If there are direct comparisons of the two effects in your document, I would still appreciate a pointer.

And if there is an explanation how aliasing translates into the jitter effect, please direct me to that one too -- I can't seem to get my head around that effect...

Edit Edit: Ah, saw your new response only after I had hit "save". Reading now, thanks!

[Performa01]

The answer to all these questions as well as the requested screenshots (and many more) can be found in the document at the end of the first posting in the following thread.

https://www.eevblog.com/forum/testgear/sds2000x-plus-bandwidth-aliasing-application-note/msg3919544/#msg3919544

Thank you! That document deserves time to work through properly...

Upon quickly skimming it, I could not find screenshots which compare the effect of capturing a square wave with aliasing (due to a limited sampling rate 2F) with the effect of limited bandwidth (high sampling rate, but lowpass at cutoff frequency F). Could you point me to the right figures please?

• Aliasing: chapter "Pulse & Square Wave Test", figure 22 ff.
• Limited bandwidth: chapter "How much bandwidth do we need?", figure 5 ff.

[asmi]

And I could name few Rigolites(TM) (fans of Rigol ) that were dropping in any discussion and when Siglent was mentioned they would simply just smash the party.. 

I don't remember a single Siglent thread where that would be the case - but maybe I haven't noticed because I admit I don't have time to read them all. But ANY Rigol thread is always swarming full of Siglent shills. What business to they have to be there at all? When I read such threads, I want to get an objective information, not endless shilling and holy wars.

For instance whenever you mention that Siglent has always running segments (triggered capture memory) so you can always stop scope and go back if you wanted to, they would start attacking how that is stupid and useless. Because their favourite brand does not have it. Or when Siglent released 12 bit scope then it was all the rage how 12 bit is not important and that is just a elitistic bulshit to push expensive scopes. Then suddenly apparently 12 bit is all important. By same people. While now and then, 12 bit, like anything else, is important if you have use for it. If not you could be equally well served by 8 bit scope.. etc etc. And Rigol is here and reads it all..

Again, didn't see that. But I do see Siglent shills preaching how 12 bit is not important at all

And then you have users like Performa01.

Do you mean shills like Performa01, who don't disclaim their affiliation and try to pass up their shilling as "unbiased" opinion? Sorry, but I despise such folks. At least tautech has guts to openly admit his affiliation, and I respect him for that.

Based on my experience. So I will preferentially suggest something that Siglent has that I tried and it is proven to work well.

Having biases is OK as long as they are known by people you talk too, so that they can keep it in mind when deciding if they should take your advice on a purchase or not. Because things one doesn't say are just as important (and telling) as things one does say.
And to me as someone fairly unbiased (I have exactly one Rigol and one Siglent TM, so they are even in my eyes ), you do come across as being excessively negaitive towards these Rigol scopes, and constantly comparing them to devices from significantly higher price bracket, which is not helpful at all for potential buyers - we all know that Lexus is better than Toyota, but if you've only got a budget for a barebones lowest trim of Corolla, reading comparisons between Toyota Corolla and Lexus ES are not going to provide you with anything useful.

[Mechatrommer]

And I could name few Rigolites(TM) (fans of Rigol ) that were dropping in any discussion and when Siglent was mentioned they would simply just smash the party.. 

I don't remember a single Siglent thread where that would be the case - but maybe I haven't noticed because I admit I don't have time to read them all. But ANY Rigol thread is always swarming full of Siglent shills. What business to they have to be there at all? When I read such threads, I want to get an objective information, not endless shilling and holy wars.

look at the bright side! they are trying to help $400 scope to push boundary of $4000 scope i really wish them success.

[asmi]

look at the bright side! they are trying to help $400 scope to push boundary of $4000 scope i really wish them success.

I don't know - to me in sounds like a guy who comes into topic about Toyota Corolla and goes on and on about his new Lexus ES. Not helpful nor interesting. If $400 scope will get all features of $4000 one, it won't be a $400 scope anymore. Everything has a price.

[2N3055]

What business to they have to be there at all?

Dude, it is a comparison thread...  With Siglent scope.... 

[2N3055]

look at the bright side! they are trying to help $400 scope to push boundary of $4000 scope i really wish them success.

I don't know - to me in sounds like a guy who comes into topic about Toyota Corolla and goes on and on about his new Lexus ES. Not helpful nor interesting. If $400 scope will get all features of $4000 one, it won't be a $400 scope anymore. Everything has a price.

Go back and read.. If made a critique of anything it was bugs and stuff that needs to fixed..
Nobody said they should add features..
Actually, in this very topic we documented all the good stuff that DHO800 has and is not very well documented or obvious.
And it is actually quite good feature wise, just needs a good debug..

So your complaints are unfounded. If you go and read you will find out I have Rigol DM3068, DP831, DG1064Z and used to have DS1074Z.. Also have Micsig, Keysight, Picoscope..

[asmi]

Dude, it is a comparison thread... 

Dude, I don't care. I want to hear opinions of normal users, not shills. They don't add anything useful to conversation.

[asmi]

Go back and read.. If made a critique of anything it was bugs and stuff that needs to fixed..
Nobody said they should add features..
Actually, in this very topic we documented all the good stuff that DHO800 has and is not very well documented or obvious.
And it is actually quite good feature wise, just needs a good debug..

But the only way to find this out is to spend ungodly amount of time going through flame wars and shilling, so that they can find 10-15 posts which are actually useful and informative. Nobody has time for that.

So your complaints are unfounded.

Yes they are and you know it. Don't even predend otherwise.

[2N3055]

Go back and read.. If made a critique of anything it was bugs and stuff that needs to fixed..
Nobody said they should add features..
Actually, in this very topic we documented all the good stuff that DHO800 has and is not very well documented or obvious.
And it is actually quite good feature wise, just needs a good debug..

But the only way to find this out is to spend ungodly amount of time going through flame wars and shilling, so that they can find 10-15 posts which are actually useful and informative. Nobody has time for that.

So your complaints are unfounded.

Yes they are and you know it. Don't even predend otherwise.

You are wrong. I responded to trolling and shilling. Realtime scope that has nominal 200MHz+ BW and samples at 156,25 MSPS/s is crap and you know it (DHO900). I had argument about that.
I also had argument here about facts.

Yet you would want me not to react when somebody writes complete nonsense.
Why didn't you react and point out inaccuracies? And nonsensical things?

But you come after me because I dare to argue about facts..?
Your post alone just proves exactly my point...

[Martin72]

But the only way to find this out is to spend ungodly amount of time going through flame wars and shilling, so that they can find 10-15 posts which are actually useful and informative. Nobody has time for that.

I think it would be good if people here would finally get back to the subject and not to themselves and other people.
That fills pages here and then complains about the fact that you can hardly find the useful posts in it...Something bizarre.

[asmi]

You are wrong. I responded to trolling and shilling. Realtime scope that has nominal 200MHz+ BW and samples at 156,25 MSPS/s is crap and you know it (DHO900). I had argument about that.
I also had argument here about facts.

This thread is about DHO804 - it's right there in the title! What the hell are you doing here talking about different device? Why is it even here?

Yet you would want me not to react when somebody writes complete nonsense.
Why didn't you react and point out inaccuracies? And nonsensical things?

But you come after me because I dare to argue about facts..?
Your post alone just proves exactly my point...

Your posts has already proven to me everything I needed to know. Now they add exactly zero new information. So please save us all from this noise and let's see if we can increase S/R ratio of this thread.

[asmi]

I think it would be good if people here would finally get back to the subject and not to themselves and other people.
That fills pages here and then complains about the fact that you can hardly find the useful posts in it...Something bizarre.

I only have 4 or 5 posts here, other 22 pages are not mine, and yet their SNR is so low that even the best SA won't be able to make out of all the noise.

[Mechatrommer]

Yet you would want me not to react when somebody writes complete nonsense.

yet when people react to you with informations/facts, you call them rude?

[nctnico]

But the only way to find this out is to spend ungodly amount of time going through flame wars and shilling, so that they can find 10-15 posts which are actually useful and informative. Nobody has time for that.

I think it would be good if people here would finally get back to the subject and not to themselves and other people.
That fills pages here and then complains about the fact that you can hardly find the useful posts in it...Something bizarre.

It could also be due to lack of real content  Perhaps it is better to collect a lot of data + measurements + images, write text in a text editor (go over it a few times) and once you are done, create 1 or more big posts with all your findings bundled together. That is how I typically approach posting equipment reviews / bigger projects. And these threads tend to stay very on-topic. Maybe consider starting a new thread.

[Fungus]

You are wrong. I responded to trolling and shilling. Realtime scope that has nominal 200MHz+ BW and samples at 156,25 MSPS/s is crap and you know it (DHO900). I had argument about that.

Yes, but it DOESN'T sample at that rate, not when you use it sensibly.

Going into every single thread and telling people that it does that isn't helping anybody.

It especially doesn't help people who just arrived and don't yet know who the EEVBLOG "personalities" are.

[Fungus]

I only have 4 or 5 posts here, other 22 pages are not mine, and yet their SNR is so low that even the best SA won't be able to make out of all the noise.

Not even 12 bits is enough to pick out the signal.

[ebastler]

You are wrong. I responded to trolling and shilling. Realtime scope that has nominal 200MHz+ BW and samples at 156,25 MSPS/s is crap and you know it (DHO900). I had argument about that.

Yes, but it DOESN'T sample at that rate, not when you use it sensibly.

Apparently that sensible use can only be expected from the more advanced users of more expensive oscilloscopes. Tautech explained this with reference to the SDS2000X plus series once: 

Most buyers of this class of scope (low mid range) will understand the tradeoff of sampling rate vs frequency (aliasing) and select sufficient a sampling rate for their frequencies of interest or be quite aware to use just one channel on each ADC.

Granted, the SDS2000X plus is an edge case (500 MHz bandwidth at 1 GSa/s), while Rigol is pushing things much further with the DHO924. Rigol should at the very least state clearly in the specs that using the LA costs you half the sampling rate in the analog channels. Very unusual to my knowledge, since in other DSOs the ADC is the bottleneck, while in the DHO800/900 it's apparently the FPGA's capacity to handle digital data streams.

[iMo]

That FPGA could do happily 200-300MHz LA sampling with the 16(32) bits width. The issue/bottleneck is somewhere between the FPGA and the shared external memory (shared with o'scope channels) attached to the FPGA, imho. A dirty hack could be to use the internal bram for the LA, but with much smaller data amount sampled (like XX kBytes), none sampling rate decrease needed with the LA "on" then..
My 15y old openbench-logic-sniffer ($50) can do 200MHz sampling w/ 16bit (24kB bram).

[ebastler]

That FPGA could do happily 200-300MHz LA sampling with the 16(32) bits width.

It even handles 625 MSa/s LA sampling in the DHO900. But then only a total of another 625 MSa/s is available for 1..4 analog channels.

I agree that we don't know which exact part of the digital section constitutes the bottleneck. It might well be the external RAM. There are those two extra RAMs which are only populated on the DHO900 series mainboard, which could be either for LA or AWG use -- I don't think that has been determined?

[tautech]

Most buyers of this class of scope (low mid range) will understand the tradeoff of sampling rate vs frequency (aliasing) and select sufficient a sampling rate for their frequencies of interest or be quite aware to use just one channel on each ADC.

Granted, the SDS2000X plus is an edge case (500 MHz bandwidth at 1 GSa/s), while Rigol is pushing things much further with the DHO924. Rigol should at the very least state clearly in the specs that using the LA costs you half the sampling rate in the analog channels. Very unusual to my knowledge, since in other DSOs the ADC is the bottleneck, while in the DHO800/900 it's apparently the FPGA's capacity to handle digital data streams.

Bold is incorrect.

SDS2000X Plus while a 500 MHz design uses two 2GSa/s ADCs however when 3 channels are activated automatic BW limiters are engaged to ensure Nyquist remains met with sufficient sampling rates.

However SDS2000X Plus can support 2 channels at 500 MHz BW and its -3dB point is ~600 MHz IIRC. Some say 580 others 620 MHz.
This is described in the datasheet:
Sample rate (Max.) 2 GSa/s (interleaving mode)，1 GSa/s (non-interleaving mode)
Where in interleaved mode 4x the rated BW is supported with 2 GSa/s to comfortably meet Nyquist.

[iMo]

That FPGA could do happily 200-300MHz LA sampling with the 16(32) bits width.

It even handles 625 MSa/s LA sampling in the DHO900. But then only a total of another 625 MSa/s is available for 1..4 analog channels.

I agree that we don't know which exact part of the digital section constitutes the bottleneck. It might well be the external RAM. There are those two extra RAMs which are only populated on the DHO900 series mainboard, which could be either for LA or AWG use -- I don't think that has been determined?

Thus it seems the max r/w bandwidth into those external drams is those ~625Mwords/sec (or the twice when interleaved r/w). That is somehow in sync with the ~667MHz max clock of the Zynq FPGA there (assuming the lowest grade FPGA).

The other DSOs I saw have got usually 4x drams where you can share the r/w bandwidth even better.