Siglent SDS3000X HD and upgraded SDS1000X HD

[dEdt]

The SDS800X HD and SDS1000X HD scopes are using the logic probes of type SLA1016.
The SDS2000X HD and SDS3000X HD scopes are using type SPL2016.
While the Data Sheet of the SPL2016 can easily be found, the one for SLA1016 not.
The exceptionally friendly staff of Batronix has provided me with a probe Data Sheet document which I like to share with you, where you can find both probes side by side on page 13 (document numbering page 12).

[KungFuJosh]

As the specification indicates, this is the maximum achievable rate, which is a sweet spot under very specific conditions.
This is the case with every scope; it is not an anomaly.
It is also obvious that this rate is achieved in dot mode because it is faster than vector mode.
So, at the moment, I do not know exactly where the problem lies.

Why is dot mode obvious? I've never seen that stated anywhere. Also, for comparison, dot mode is not required for the Magnova's higher rate. Have you tested the maximum update rate with fast/low memory on a 100MHz sine wave with the SDS' aux out?

[Martin72]

Hi,

Why is dot mode obvious?

Raw, direct sampling data; with "Vector", interpolation and connection are still required.

I tested the update rate very early on with different input frequencies, dot and vector mode, Sinx/x and Lin interpolation, and wanted to do it again “properly.”
Properly in the sense of 2 channels active, because then you should be able to achieve the specified maximum rates.
Don't ask me why it's the other way around with this model than everywhere else.

[KungFuJosh]

Hi,

Why is dot mode obvious?

Raw, direct sampling data; with "Vector", interpolation and connection are still required.

I'm still curious what the real update max is with vector mode. I find it hard to believe that it's only 25k. Test conditions obviously have to be optimal- low memory, high frequency, etc.

[Oldman]

The sds3000xhd is advertised as max. 200.000Wfrm/s.
But this is only achieved when you use DOT-mode and not vector-mode.
Plus you must activate e.g. channel 1 and 2 (you can then hide one channel if you want).
If you do not do this you do not get over max 25.000Wfrm/s.

Where are you getting this "information" from? It sounds like nonsense. There's nothing in the datasheet specifying any of that. It simply says 200k in Normal mode.

I had asked  a number of questions, mainly about the possible speed reductions when you activate measurements, math, fft statistics etc.
In about a week I got a full report (about 23 pages) from Siglent, with pictures, screenshots, measurements, explanations etc. This service in itself is very impressing from Siglent and deserves both thumbs up from me.
Besides this, I of course tested all this myself. The acquisition speed I measured by connecting a counter to the trigger out.

This is a part of the Siglent report:
(3) We must configure the memory depth so that the memory is just full after the contents of the whole display screen are written into the memory. In this way, the oscilloscope can immediately re-trigger and write the next waveform without delay, so as to reach the maximum number of waveforms per second. At present, the optimal configuration is 1k pts, and the sampling rate of oscilloscope is 2Gsa/s.
(4) Turn on the point display mode of the oscilloscope.

Max. acquisition speed they achieved in their test (and mine) is 215000 Wfrm/s. When you are in this situation and switch to Dots-mode the speed immediately drops to 25000Wfrm/s. Dot mode you can find in menu Display / Menu and then choose at type not Vector but Dots (see picture).

Like Martin72 says, this dot/vector mode setting is not uncommon. I myself see this also with other scopes. The R&S MXO do not have this issue for instance.

[tautech]

New firmware for SDS3000X HD models.

Version： V1.0.4.2
~132 MB
https://int.siglent.com/upload_file/zip/firmware/Oscilloscope/SDS3000X-HD_1.0.4.2_EN.zip

Release notes
1. Decode: Added SPMI, Spacewire, USB2.0, CANXL
2. Math:
a) Added operators: tan, atan, envelope, delay, FFT phase
b) FFT: added windows Blackman-Harris and Gaussian
c) Formular editor: supported measure item as the component of a formula
d) Supported Trend and Track as math trace
3. Measure：
a) Added item: Bit Rate, [RD-KHWT-202507041900], UpperLower
b) Supported to limit the display digits of measure [RD-KHWT-202507111920]
c) Supported to display up to 24 items at the same time
4. Analysis:
a) Added Double Pulse Test
b) Added Serial Test
5. Probe: Supported SDP6150A, SDP6150D
6. Roll: Supported Math [RD-KHWT-202506161860] and track cursors [RD-KHWT-202507111919] in Stop mode
7. Display: Supported custom colors for XY trace [RD-KHWT-202411261502]
8. Webserver: supported inverted screenshot (according to the Image style setting in the Save menu); added virtual front panel
9. Save/recall: Supported full screen mode when preview an image; supported to save multi channels data as a single file; supported Auto Save;
10. Utility：Added English help; Supported password for screen saver
11. Fixed several bugs:
a) [RD-KHWT-202505261802] The instrument becomes slower after long time run
b) [RD-KHWT-202506231876] Webserver: some control issue in full-screen mode
c) [RD-KHWT-202504101726] Cannot save a Memory trace to file
d) [RD-KHWT-202502071589] OPC command response error

[Oldman]

Hi,

Why is dot mode obvious?

Raw, direct sampling data; with "Vector", interpolation and connection are still required.

I tested the update rate very early on with different input frequencies, dot and vector mode, Sinx/x and Lin interpolation, and wanted to do it again “properly.”
Properly in the sense of 2 channels active, because then you should be able to achieve the specified maximum rates.
Don't ask me why it's the other way around with this model than everywhere else.

To get the max. acquisition speed in the SDS3000XHD series you need to set the timebase to 50ns/div, normal Acquisition, fast Acq mode , auto memory Mgmt (it will choose 2MSa/s) , max memory size (it will choose 1kpts)  and use a 10MHz signal (that is enough 100MHZ is not necessary). Normal triggering mode, Dots-mode and if you use channel1 also activate channel2.
You can tune the input frequency to give the scope,  after finishing a sweep and after the re-trigger time (spec.as 1.5us) , immediately the oppertunity to start the next sweep. This way you can get not 214kWfrm/s but may be 215kWfrm/s, but this small increase is not that important .

The R&H MXO has 2 settings: In my test to see if Dot mode has influence on speed I have choosen in the Settings/Display mode /Signal Waveform style 'Dot' and also in tha Acquisition menu/Interpolation 'Sample and hold'. Both did not decrease or increase the max. acquisition speed I had at that moment (4,37MWfrm/s).
Of course a R&S MXO is a lot more expensive than the SDS 3000XHD and , yes, the price/performance ratio becomes exponentially worse if you choose scopes in the higher segment. But my solution is I just do not buy the Jaguar X-type 2-seater, but instead a VW golf. This way I can take also a wife and child with me on my vacation and they never bother me about having an expensive oscilloscope........

 I am still curious who actually has tested the Siglent 1000XHD and  2000XHD and what settings were used. I understand from Martin72 he has checked this in the past. Anyone else?

I just installed the new FW Tautech informed us about. I could find Martin72'  sweet spot now at 30kWfrms  (vector mode) when fiddling around with timebase and input frequency.

[KAKUL]

Hi
SDS1000X HD:
I did a quick test and the average on one channel was waveforms/s 123,000  and in 4-channel mode with FFT and filtering enabled the average waveforms/s 30,000.

[Oldman]

and this was in vector mode , not dot mode ?

[Oldman]

I got conformation of Siglent Europe that the 120000 and 100000 Wfrm/s for the sds1000xhd and sds2000xhd are only achieved in Dots-mode and if channel 1 is used, also channel 2 must then be active. So this is the same as for the sds3000xhd (200000 Wfrm/s in Dot-mode and channels 1&2 or 3&4 active).
They also said ' In Vector mode, or when only one channel is enabled, the displayed update rate will indeed be lower, although the internal acquisition rate remains high.'
I am not entirely sure what they mean by that. I will test if rare glitches are still captured in Vector mode as good as they are in Dot-mode. 

[Martin72]

You should let go of this dot vs. vector mode thing...
In addition, the differences only exist in a rather narrow time range; below that, they are the same.
Finally, almost every scope model now has trigger options that can detect a glitch, even though they are not exactly “blessed” with massive update rates.

[Oldman]

You are right of course, only at a specific timebase setting and a matching signal frequency you have this max. Wfrm/s. Every other setting used will show it will drop more and more. But even so, as anyone else on this forum, I am interested in the  engeneering  and achievements  in oscilloscope technology, especially the digital stuff, and find out 'who is winning'

I started to use scopes when they were build with steel and concrete and vacuum tubes with only a tiny circular CRT. In those times we could not even imagine what we now can do in even the smallest scope. Back then you needed a special trolly to move them because they were too big and heavy to just pick them up. Ahh...those were the days when electronic engineers were still men.....

And to get back to Wfrm/s and glitches: if you don't know if there are small, random and rare glitches (and thus do not know how to accurately set up one of the a special trigger modes) , but you need to check: you should use the 'fast segmentation-mode' or 'ultra acquire-mode' or whatever manufacturers call them. In this case the regular dead time is avoided and you always are only left with the re-trigger time, between each waveform-recording, no matter what timebase you have. 
Most oscilloscopes have retrigger times from 0.8 to 1.5us, but only R&S has 0.021us in the MXO series.

[2N3055]

Most oscilloscopes have retrigger times from 0.8 to 1.5us, but only R&S has 0.021us in the MXO series.

And then you switch on any simple measurements and it drops to 100 of triggers per seconds..
That is also a record too..

[Oldman]

Yes , sadly you are correct: when the scope is in run and depending on timebase and signal. But if you use fast segmentation- or ultra acquire-mode etc. these time stamps between Waveform recordings is still 0.021 to 1.5us (depending on the scope) and no matter what timebase and measurements you activated. When you play back the recording, all waveform measurements are calculated afterwards during playback for each waveform. So these in these modes you miss out on as little glitches etc. as possible and still see your actual measurement values.
As everything, it is a compromise.

[2N3055]

Yes , sadly you are correct: when the scope is in run and depending on timebase and signal. But if you use fast segmentation- or ultra acquire-mode etc. these time stamps between Waveform recordings is still 0.021 to 1.5us (depending on the scope) and no matter what timebase and measurements you activated. When you play back the recording, all waveform measurements are calculated afterwards during playback for each waveform. So these in these modes you miss out on as little glitches etc. as possible and still see your actual measurement values.
As everything, it is a compromise.

Yes but I use measurements in normal running mode and that is primary use case for me.
So super fast segmented mode in what is basically a digitizer mode is something I do more rarely.

Also it is definitely not "no matter what timebase you activated". Retrigger time cannot be faster than sweep time + blind time.
If you set scope to 10 ms/div and have 10 div on screen, even if you have scope with 0 blind time, you cannot have more than 10 triggers per second. And blind time will be very dependent on memory depth. Scope will need more time to process 500MS than 1kS.

Catching out little glitches?
It depends on what is frequency of repetition.
And what information you want to extract.

If you need 100% POI (probability of intercept, detection)  in say 100ms period, then your best friend is long capture with long memory. In which case you can have other channels connected to some system references to try to correlate where glitch comes from. You find glitches inside that capture with search.

On better scopes (some of which are still cheaper than MXO series) you have Signal/Wave scan (or some other trade name) that will capture, search, and if detected stop and show, or if no detection, keep recapturing until it finds something. In which case it offers you to copy that detection definition into trigger and then let it run and have 100% detection via trigger.

You can also run scope in simple persistence mode for longer time until you capture something, then by looking into waveform setup a trigger or zone trigger to trigger on problems and also capture them 100% of time.

Once you have a trigger definition, you can then use segmented mode to ensure shortest possible retrigger time, but  funny enough, that is not really necessary then, because if your glitches are so close, you will see them in plain sight while looking at the screen in bog standard mode.. Very reliably, even on SDS800xHD.

There is some merit to fast retriggering in normal trigger mode when using scope in a standard fashion, and that is that in that case CRT emulation is better and scope shows more "natural" view on fast changing signals that resembles CRT scope.
I also find that pleasing to look at, but not really that important, once you are showing more than some amount of triggers per second, it gets good enough..

[tautech]

Yes , sadly you are correct: when the scope is in run and depending on timebase and signal. But if you use fast segmentation- or ultra acquire-mode etc. these time stamps between Waveform recordings is still 0.021 to 1.5us (depending on the scope) and no matter what timebase and measurements you activated. When you play back the recording, all waveform measurements are calculated afterwards during playback for each waveform. So these in these modes you miss out on as little glitches etc. as possible and still see your actual measurement values.
As everything, it is a compromise.

Yes but I use measurements in normal running mode and that is primary use case for me.
So super fast segmented mode in what is basically a digitizer mode is something I do more rarely.

Also it is definitely not "no matter what timebase you activated". Retrigger time cannot be faster than sweep time + blind time.
If you set scope to 10 ms/div and have 10 div on screen, even if you have scope with 0 blind time, you cannot have more than 10 triggers per second. And blind time will be very dependent on memory depth. Scope will need more time to process 500MS than 1kS.

Catching out little glitches?
It depends on what is frequency of repetition.
And what information you want to extract.

If you need 100% POI (probability of intercept, detection)  in say 100ms period, then your best friend is long capture with long memory. In which case you can have other channels connected to some system references to try to correlate where glitch comes from. You find glitches inside that capture with search.

On better scopes (some of which are still cheaper than MXO series) you have Signal/Wave scan (or some other trade name) that will capture, search, and if detected stop and show, or if no detection, keep recapturing until it finds something. In which case it offers you to copy that detection definition into trigger and then let it run and have 100% detection via trigger.

You can also run scope in simple persistence mode for longer time until you capture something, then by looking into waveform setup a trigger or zone trigger to trigger on problems and also capture them 100% of time.

Once you have a trigger definition, you can then use segmented mode to ensure shortest possible retrigger time, but  funny enough, that is not really necessary then, because if your glitches are so close, you will see them in plain sight while looking at the screen in bog standard mode.. Very reliably, even on SDS800xHD.

There is some merit to fast retriggering in normal trigger mode when using scope in a standard fashion, and that is that in that case CRT emulation is better and scope shows more "natural" view on fast changing signals that resembles CRT scope.
I also find that pleasing to look at, but not really that important, once you are showing more than some amount of triggers per second, it gets good enough..

Yup all this and to add what is of most importance for waveform analysis.

We are now working with DSO's and they require a different mindset to what was used in hollow glass oscilloscopes, namely embracing of the Storage capability in Digital Storage Oscilloscope where once a capture is made to then use the tools a DSO provides us with to find abnormalities of interest with the most powerful tool we have, the trigger !

Things have moved on some since I did this Search example with SDS1104X-E but it indicates one of the processes you can use:
https://www.eevblog.com/forum/testgear/siglent-sds1204x-e-released-for-domestic-markets-in-china/msg1370717/#msg1370717

[Oldman]

2N3055 is right of course. I forgot to add the timebase V/div*10 to add to the 0.8-1.5us retrigger time.
And indeed you need more memory if you have a long timebase and want to see narrow events, because you want the sample time as high as possible and with a long timebase , a lot of samples will be recorded.

Besides the re-triggertime I have seen another blind time in digital scopes. If I look at the trigger out signal I saw a gap in the tigger out signal of 500us to a couple of ms every 25-60 Hz (depending on the scope) . Scope manufacturers told me this is the moment where the scope gathers all the recorded information together to, at last , process it and show this on the screen.

[tautech]

Today watching a Youtoob vid from a well known member currently taking leave from EEVblog, he bagged  the Measure feature in Siglent DSO's where he obviously hadn't spent any time learning its capabilities or was too stubborn to attempt to. 

After pulling out a SDS1204X HD and having a good look I still can't see what he was on about. 

So let's take a look...
Hit the Measure button or select it from the top of display dropdown menu.



We can see the measurement as Simple or Advanced and the channel selection to make measurements on and Type allows us access to the type of measurement we need to see and click to add/select and click again to remove.
Each already selected type is clearly highlighted.





If we select Advanced the measurement type management is slightly different yet in some ways more powerful in that we can select types for different channels should others be active. Just clicking the + has the type and channel selection pop up.



Still wondering what he found showstopping about this ?

[Ulrich.G]

Today watching a Youtoob vid from a well known member currently taking leave from EEVblog, he bagged  the Measure feature in Siglent DSO's where he obviously hadn't spent any time learning its capabilities or was too stubborn to attempt to. 

Can you pls. share a link to his video? To understand what he was fighting with :-)

[2N3055]

Today watching a Youtoob vid from a well known member currently taking leave from EEVblog, he bagged  the Measure feature in Siglent DSO's where he obviously hadn't spent any time learning its capabilities or was too stubborn to attempt to. 

Can you pls. share a link to his video? To understand what he was fighting with :-)

Joe simply didn't know Siglent has Quick measure mode also...
He got send a scope from UNI-T and he is testing it and comparing it to other scopes he used..
And testing is going ... UNI-T.. what can I say.
Waiting for his patience to run out.

[mawyatt]

Today watching a Youtoob vid from a well known member currently taking leave from EEVblog, he bagged  the Measure feature in Siglent DSO's where he obviously hadn't spent any time learning its capabilities or was too stubborn to attempt to. 

Can you pls. share a link to his video? To understand what he was fighting with :-)

Joe simply didn't know Siglent has Quick measure mode also...
He got send a scope from UNI-T and he is testing it and comparing it to other scopes he used..
And testing is going ... UNI-T.. what can I say.
Waiting for his patience to run out.

Why would anyone knowledgable in TE consider an expensive DSO from UNI-T, what does UNI-T know about scopes??

Expensive higher performance DSOs, besides the usual upper tier folks, are seeing Siglent and Rigol moving in that direction. Both of which have a long history of DSO development, and learning from LeCroy and Keysight respectively when they supplied the lower end DSOs to these upper tier TE OEMs (believe Siglent still does for LeCroy).

Is Keysight using UNI-T for their lower end DSOs? Our understanding is they parted ways with Rigol sometime ago.

Best

[Grandchuck]

I don't think Keysight is using Uni-T.

A search about relabeling produced "UNI-T has shifted toward high-performance, original designs using their own proprietary ASIC chips (like the MSO5000HD series), making them less of an "OEM-for-hire" and more of a direct competitor to brands like Rigol and Siglent"

[2N3055]

I don't think Keysight is using Uni-T.

A search about relabeling produced "UNI-T has shifted toward high-performance, original designs using their own proprietary ASIC chips (like the MSO5000HD series), making them less of an "OEM-for-hire" and more of a direct competitor to brands like Rigol and Siglent"

UNI-T is unusual company. They did not develop that scope (or any of the latest 12 bit scope) from internal knowledge or tradition because there is none. They probably hired development team(s) or bought IP for whole series from someone.  In China there are dozens of scope manufacturers that make some very advanced platforms, but they do not export to West.
There are also state universities that do research in enabling technologies.
If UNI-T uses ASIC, it was not developed by them internally. They might be using Rigol chips for what we know. Or some chip maker got assigned to create scope chips, so they have domestic source.

Regardless, UNI-T is not really competing with Rigol , and even less with Siglent as serious scope manufacturer. They lack internal development process continuity. And I am talking about system design and software. That is why Keysight, R&S, Tektronix, LeCroy still hold in the high end. The milion of engineer-years in making scopes. Some Chinese manufacturers closed gap in certain segments (and are even leading there). But high end scope costing 25000€ and up is only as good as software capability and it's quality.
UNI-T has long way to go. If they are serious and start doing it the right way. If they keep doing it same way as before, they will never get there. And I am not saying underlying hardware is not capable being good scope. But hardware itself is only a promise something is possible...
I won't even mention support.

[H.O]

Joe simply didn't know Siglent has Quick measure mode also...
He got send a scope from UNI-T and he is testing it and comparing it to other scopes he used..
And testing is going ... UNI-T.. what can I say.
Waiting for his patience to run out.

I haven't seen all videos in the series but I did watch the first and according to that he was not "sent a scope from Uni-T". He bought the thing with his own money. Not a demo unit for review, not a gift, not even a discount.

[2N3055]

Joe simply didn't know Siglent has Quick measure mode also...
He got send a scope from UNI-T and he is testing it and comparing it to other scopes he used..
And testing is going ... UNI-T.. what can I say.
Waiting for his patience to run out.

I haven't seen all videos in the series but I did watch the first and according to that he was not "sent a scope from Uni-T". He bought the thing with his own money. Not a demo unit for review, not a gift, not even a discount.

I didn't watch any of the videos in detail. Thank you for correcting me.  I went and checked and sure he does say he bought it.
I also looked around new videos quickly. I didn't watch in detail, because I am really not interested in that scope.
I am surprised he would buy something this expensive without trying it first.
That is going to go nuclear quick.. he is already in "get your shit together" mode.