Siglent SDS2000X Plus

[tautech]

Is here someone from Siglent who could elaborate first hand?

Angus, you are already discussing this with some of Siglent's excellent beta testers.
I can't say who they are. 

[gf]

2: trigger depends on interpolation in some way.

On the one hand, the trigger needs to be calculated in real-time, during acquisition (at least coarsely). OTOH I would not rule out that the fractional sample position still gets fine-adjusted in the postprocessing, honoring the interpolation type then.

I find it interesting, though, that the trigger point appears to be approx. linearly interpolated between the neighbor points in the 2nd image (labeled sinc.png), while it does not lie on the straight line beteen the neighbor points in the x.png image. Subjectively I'd rather expected the opposite.

[2N3055]

Digital trigger engine is before sample memory just after ADC and full HW samplerate.

math and automatic measurements still need complete data without gaps. These data are constructed in a secondary screen buffer, using the interpolation method selected by the user.

I took a measurement with dots. Went to history and selected a frame. The scope remained in stop-mode all the time. So there is no data being updated. The ONLY parameter I changed was x/sinc.


1: dots are a direct representation of the sample values.

2: trigger depends on interpolation in some way. I agree that that won't be a good idea, but if it should happen, the raw data should only move in time.  Remember the scope already sampled; they should not move at all if triggering was done before sampling.

3:if interpolation method has its impact on mathematics and measurements only, the raw data in a frame should not move, because everything that happens with them is, well, visualization.

I think something is unlikely / does not make sense: why would the raw dots of a single frame change their position on screen when the scope is frozen and only that frame is shown? I totally agree with the explanations and assuumptions, but.... something is unlogical for me.

Is here someone from Siglent who could elaborate first hand? I hope I did not introduce myself as too dumb to answer


edit: I modified to hopefully better describe my point

Those two images don't show same sample records. Just look at first 4 points from the left and you will see they don't have same vertical distribution. What happened there I don't know. But those screens are not showing same sample buffer.
A suggestion: set trigger level at exactly 0V. That makes it easier to verify trigger point visually. You get 3 clear points..

1. dots are physical sample point. Correct.
2. Trigger is fully digital. It happens on sampled data. Internal timebase resolution is much higher than 500ps from sampling. ONce trigger event is detected, trigger engine estimates where exactly the signal passed through trigger point and saves that data with that sample buffer. It saves points as they are and save estimated horizontal microposition data. Display engine use it to align everything nicely. Nothing gets changed from original points in sample buffer.
3. I tried on SDS6000 and nothing moves in dot mode. Not a dot. I cannot try 2000X+, I don't have one, sorry. They should function in a very similar fashion, they are based on a same platform.

[Angus]

Nothing gets changed from original points in sample buffer.

Now that I learned that once dot-mode is active we only see raw sample data right from ADC output on screen (and some additional parallel processing for calculations under the hood) that is clear to me.

Even more confusing to see the screen change. ?

SDS6k and SDS2k+ are based on same platform you say? You mean analog frontend and general processing flow? I studied the manual and noticed that at least the firmware seems to be slightly different. e.g. they have located signal averaging in aquisition (which I'd prefer to the hidden math because it is more common), memory configuration looks slightly different.
So could it be that firmware plays a role?

If some sds2k+ owner would be so kind to verify what I encoutered? Is it exclusive to my unit?

Not that this behaviour plays a major role in every day use. It is still surprising and I'd like to understand if I did not get it or if there is -something-.

[2N3055]

Nothing gets changed from original points in sample buffer.

Now that I learned that once dot-mode is active we only see raw sample data right from ADC output on screen (and some additional parallel processing for calculations under the hood) that is clear to me.

Even more confusing to see the screen change. ?

SDS6k and SDS2k+ are based on same platform you say? You mean analog frontend and general processing flow? I studied the manual and noticed that at least the firmware seems to be slightly different. e.g. they have located signal averaging in aquisition (which I'd prefer to the hidden math because it is more common), memory configuration looks slightly different.
So could it be that firmware plays a role?

If some sds2k+ owner would be so kind to verify what I encoutered? Is it exclusive to my unit?

Not that this behaviour plays a major role in every day use. It is still surprising and I'd like to understand if I did not get it or if there is -something-.

No not the analog frontend. And they do have different firmware and hardware details even in digital parts. But general sampling/triggering architecture, how samples flow through the system..

As I said, those two screens you have shown in last post do not show same acquisition buffer. Points are different.
On my 6000, I take single acquisition, set it in a dot mode and if I flip X/sinc nothing happens.
Same as if I go to history buffers and pick one from there...

[rf-loop]

@Angus

Did you note what @2N3055 write also in  second to last message as also in last now (I think we write around same time).

I recommend that before more hassle and before go any more forward you do one very simple thing.

We need to rule out things that are now uncertain about what is actually in the pictures presented.
Like the 2N3055 in his second to last message, I think  the images do not display same ⁽¹⁾ individual "horizontal sweep" aka acquisition or capturte when we talk digital scopes.
⁽¹⁾(I have analyzed these every dots all movements and least it somehow seems impossible they are from one same single shot)

(is it possible that accidentally when you have operated with history buffer then frame (segment) number have not changed, it may happen accidentally easy and your image do not display this segment number.)

So, please, do new test.
8 bit mode.

Signal and tdiv etc as previously in your last two images.
Display Mode Dots and  Interpolation setup what ever but so that you know it.
Take just one Single shot.  (Trigger mode Single.) Everything after this point handle just this one single shot. Inside oscilloscope memory is just only this one.

Just after this  single shot you see there some dots (this rising edge as also in previously)
Do not go to History (and even if you go there is nothing more), do not adjust anything. 

Before adjust or change anything take 1st screen image and tell what was Interpolation, x or Sinc. (just same kind of image as was your two images)

Then do not touch anything but swap only just interpolation.

Take 2nd image and tell what is now interpolation.

Now then you have 2 images where we can see these few dots.

I am really amazed if we can see that the dots have moved.

If these dots really move then we may need start further investigations about what a hell is going on.

[Angus]

I'll try that out as soon as I can. But first I hope that my investigation and questions do not sound harsh or someone is under the impression that I blame my instrument... If so - not intended.

Then do not touch anything but swap only just interpolation.

That is what I very carefully tried (twice) during the test for the above post. But I chose a frame by history and not single shot. So, yes, I'll carefully follow your post. And also set the trigger to 0 V.

Like the 2N3055 in his second to last message, I think  the images do not display same (1) individual "horizontal sweep" aka acquisition or capturte when we talk digital scopes.

Exactly what made me place a post after what you have explained to me about the capture process. The capture -should- not change, but picture looks like.

I am really amazed if we can see that the dots have moved.

I don't know what I would favour.    And how to react if they should be pinned to the screen.

thanks a lot

[2N3055]

I don't know what I would favour.    And how to react if they should be pinned to the screen.

thanks a lot

Well, best solution is that it was a user error  and all works perfectly....

And as RF said, if you can reproduce this, exact steps should be written down and forwarded to Siglent so they can look into it.

P.S. Did I miss it or you didn't say? What Firmware are you running ? That is important.

[rf-loop]

I'll try that out as soon as I can. But first I hope that my investigation and questions do not sound harsh or someone is under the impression that I blame my instrument... If so - not intended.

I think these are always interesting and welcome, and then you can explain or find a more specific principle of operation of the device, which is usually always a good thing to understand when applying the device to study signal phenomena, for example. The only way to successfully test and explore more demanding things is to know the features of the device and the details of how it works. 

But also it is good if result is that there pops up some new unknown bug. Only known bug can fix.

It is extremely unfortunate that the time is over when the manufacturers produced manuals that dealt with the functions of the device in great detail and with quite in-depth explanations down to the last detail, including the basics of the theory. In the old days, both HP and Tektronix, among others, produced such massive manuals and reference guides for many of their devices.
Those times are over. This is unfortunate, especially as equipment has become increasingly versatile and often more complex in its operation. Now, users often have to try and find out how this might work. Sometimes it can lead to long iterative experiments.

[Angus]

I did as told:

In addition I enabled color grading for increased visibility.

Set the scope to single shot trigger which stopped after the shot, as expected. I verified that in History there is only one single shot shown. 1/1.
After checking the history button I did an other shot just for sanity.
I would have placed a bet after being said

I am really amazed if we can see that the dots have moved.

but they do. Don't they?

Second idea was to enable sequence mode for a sequence of 2 and see if there is any correlation.

At first glance I would say that "sequence_2" is similar to "sequence_2_sinc", but "sequence_2_x" is not correlated much.

I do not want to place speculations, but it is strange. Especially strange that you guys cannot reproduce this effect. ;-)

Any ideas for further experimentation for the coming weekend?

[2N3055]

Yeah something is weird..

What is firmware version?

I cannot check myself, I don't have SDS2000X+

[Angus]

What is firmware version?

Ohh shame I forgot to give that information. I suppose it is the lastest firmware.


Software Version 1.3.9R6
FPGA Version 2021-07-16
CPLD: 03
HW version 02-04

[blurpy]

I can reproduce this effect. Same firmware version.

[blurpy]

Also, here is how it looks with the images overlaid.

Switching between x and sinc looks very consistent in the way the dots jump between the 2 positions. So seems to be repositioning based on the same data.

[Performa01]

There is no need to determine the exact trigger point during acquisition. We have a bunch of sample data and the index of the first sample after the trigger point.

Only when the data are to be displayed on the screen and placed into the secondary buffer for measurements and math, we also need the correct trigger time offset, which can be anywhere between zero and the full sample interval.

Quite obviously, the selected interpolation method is used by the trigger position finding. Who would have thought that we get such a lot of flexibility? But then again, it's absolutely necessary to use the same method for the trigger positioning and the data set used for measurements, since some measurements are related to the trigger position, like Delay and T@M.

[nctnico]

I did as told:

In addition I enabled color grading for increased visibility.

Set the scope to single shot trigger which stopped after the shot, as expected. I verified that in History there is only one single shot shown. 1/1.
After checking the history button I did an other shot just for sanity.
I would have placed a bet after being said

I am really amazed if we can see that the dots have moved.

but they do. Don't they?

Ofcourse they move! The trigger interpolator will move the dots so that the trigger point in time matches the trigger level (voltage). The reason for the dots moving is because the sampling frequency isn't an exact multiple of the signal source's frequency so the signal gets sampled at different points in the waveform.  If you run a longer sequence (say 20 segments) and overlay those, you'll see that the points will follow the shape of the signal.

The dots will only stay in a fixed position IF the sample frequency is locked to the signal source's frequency. In that case the oscilloscope will sample the signal at the exact same points in time and thus the trigger interpolator will position the sample points at the same place on the screen.

[Angus]

I tried on SDS6000 and nothing moves in dot mode. Not a dot. I cannot try 2000X+, I don't have one, sorry. They should function in a very similar fashion, they are based on a same platform.

On my 6000, I take single acquisition, set it in a dot mode and if I flip X/sinc nothing happens.
Same as if I go to history buffers and pick one from there...

You can capture using example dots mode... stop scope and after then turn Sinc or x  and switch between these and you can see trigger position stay perfectly enough in right place between real sample dots independent of which one post processed interpolation you select.
I am not sure what interpolation trigger engine is using but it must be in trigger engine because it need work always, independent of our setting for display waveform. Somehow I do not believe it change this interpolation method depending what user have selected for post process interpolation.

Dots mode. there is neither interpolation nor reconstruction, hence we always see the true signal - as long as the trigger data is valid, i.e. not disturbed because of aliasing.

Ofcourse they move! The trigger interpolator will move the dots so that the trigger point in time matches the trigger level (voltage).

Not so obvious for me because of:

Digital trigger engine is before sample memory just after ADC and full HW samplerate.

Thinking about all I learned  during this discussion I come to the conclusion that, yes SDS2000+ obvioulsy behaves like it does and SDS6000 slightly different. It is in the post processing. Maybe they changed the dot mode to a pure 'sample-scope-mode' (which was somewhat implied for all scopes as I understand the above posts) on purpose in order to avoid problems with measured eyes? But, that is speculation.... First thing that came to mind was to ask 2n3055 to have a deeper look at what can be done and happen with the eyes on his sds6k.

This discussion was really interessting and I got to know the community of this forum. Thank you! User rf-loop mentioned the lack of in-depth documentation these days.... Well that is true but you rarely miss it and if I had to choose between documentation and price-tag I have to admit I'd favor price-tag :-(

[nctnico]

Ofcourse they move! The trigger interpolator will move the dots so that the trigger point in time matches the trigger level (voltage).

Not so obvious for me because of:

Digital trigger engine is before sample memory just after ADC and full HW samplerate.

The trigger engine only gives a course indication about when a trigger occured. The trigger time interpolator provides high resolution time information on when the trigger happened precisely. The information from the trigger time interpolator is used by the display engine to position the waveform on screen.

IOW: the trigger engine determines when the DSO stops sampling, the trigger time interpolator determines where the signal goes on screen (horizontal position).

[2N3055]

IOW: the trigger engine determines when the DSO stops sampling, the trigger time interpolator determines where the signal goes on screen (horizontal position).

Yes, pretty much. Trigger interpolator adjusts microposition between two samples, to get better resolution than sample interval (1/sample rate).

[rf-loop]

I did as told:

In addition I enabled color grading for increased visibility.

Set the scope to single shot trigger which stopped after the shot, as expected. I verified that in History there is only one single shot shown. 1/1.
After checking the history button I did an other shot just for sanity.
I would have placed a bet after being said

I am really amazed if we can see that the dots have moved.

but they do. Don't they?

Ofcourse they move! The trigger interpolator will move the dots so that the trigger point in time matches the trigger level (voltage). The reason for the dots moving is because the sampling frequency isn't an exact multiple of the signal source's frequency so the signal gets sampled at different points in the waveform.  If you run a longer sequence (say 20 segments) and overlay those, you'll see that the points will follow the shape of the signal.

But the case was already looked from the data allready captured and oscilloscope is STOP. And in DOTS display mode, then just change the post processing interpolation method (keep only the ADC sample points on the display) and data was changed (dots positions moved)  (and again, it's already one shot in memory and the measurement range in stop mode).) Only if you ignore this fact, the explanation is ok as a simple basic principle.

When I have tested with another Siglent oscilloscope in the same family, these points do not move, they remain as as written in stone. They don’t move, and they shouldn’t move. Do you want to adjust and change the raw data after collecting it.  Somewhere we call this "manipulated truth." No more messing up the sum of mistakes and truth.

[nctnico]

If I interpret his posting right, Angus shows two captures from the history buffer which are overlayed:


This could hind towards why you see the dots moving in his other screendumps; the scope may show a different capture from the history buffer after changing the parameters. Even though the dots moved, they have moved in a way that follows the shape of the signal. It is not a X or Y shift (which would indicate a signal positioning error).

[Angus]

This could hind towards why you see the dots moving in his other screendumps; the scope may show a different capture from the history buffer after changing the parameters.

I think not really. Originall the scope did a single trigger and no sequence.
Sequence was added after that, just to figure out -if- the sope would toggle between these shots.

But the case was already looked from the data allready captured and oscilloscope is STOP. And in DOTS display mode, then just change the post processing interpolation method (keep only the ADC sample points on the display) and data was changed (dots positions moved)  (and again, it's already one shot in memory and the measurement range in stop mode).

That is the point. Thank you for pointing that out I was not sure if I was able to communicate properly.

If no one could imagine a setting that leads to such an effect I doubt that a solution can be found. I am happy to discuss here, but the thread is about different stuff I think.

[blurpy]

I did some THD measurements of the audio spectrum on the SDG2000X some months ago, which you can see here: https://www.eevblog.com/forum/testgear/the-siglent-sdg2042x-thread/msg3606484/#msg3606484

I thought it would be interesting to see how the internal AWG of the SDS2000X+ stacks up, so I repeated the experiment, with a QA402 this time.

The numbers:


In a graph:



It's not too happy below 25Hz, 1Vrms seems to be the worst amplitude, and 20kHz does much better than the other frequencies - even better than the SDG2000X. But with the exception of 20kHz it's doing much worse than the SDG2000X, unsurprisingly. More comparable to the output of the Owon HDS272S.

Adding some example screenshots below as well.

[nctnico]

This could hind towards why you see the dots moving in his other screendumps; the scope may show a different capture from the history buffer after changing the parameters.

I think not really. Originall the scope did a single trigger and no sequence.
Sequence was added after that, just to figure out -if- the sope would toggle between these shots.

But what was the acquisition like before you made the new single shot acquisition? I doubt the history memory is fully cleared before doing a new acquisition so even in single shot mode, the information from the previous acquisition is still there.

What would be an interesting test is to let the scope run with a signal for a while so you are sure the entire history buffer is filled what acquisitions of that signal. Then do a single shot of a signal that is different (lower amplitude for example) and do your settings change test. It will be interesting to see what pops up on the screen.

[Performa01]

Regarding the dots mode issue:

There is no need to do any further tests on this level.

It is already confirmed, that the SDS2000X Plus behaves differently to e.g. an SDS6000. The latter doesn't do any re-positioning of the samples, no matter if x or sin(x)/x is selected. Still the measurements appear correct, even the ones related to the trigger position.

There might be a bug in the SDS2000X Plus trigger handling. Also the huge difference in waveform update speed seems implausible.

I'm in the process of investigating the issue right now, this requires lots of measurements, including waveform update speed under various conditions and settings, including automatic measurments, for both the SDS2000X Plus and the SDS6000. All this will take quite some time.