Siglent SDS1000X and (MSO) SDS1000X+ series oscilloscopes

[rf-loop]

Before I bought Siglent SDS1102X oscilloscope, I analyzed what I would get for the money.
When I turned on and performed few sessions of  measurements I came across unpleasant surprises.
I am almost inclined to return the oscilloscope. I suppose the cause is not the malfunction of this particular unit, but a design error, or rather firmware.

Check short report: https://drive.google.com/open?id=0ByQgomgYIGXpY1pjSFhUN0hRMUU

Will somebody take me away from this step?

Michal

I have not now enough time to analyze your "findings".
But one first note.
Your signal have quite fast edge.
Now when you turn 2 channels on, maximum samplerate is 500MSa/s.
It means there is one sample every 2ns.
As can see your image where is 2ns/div with 1GSa/s (one channel on) and then right side image 500MSa/s (2 channel on) it is natural that there is high aliasing. In this case aliasing can see as "corners wobbling". This is NOT at all related to manufacturer. it is pure math and physics. Even Siglent can not break base fundamentals.  It is well known that this scope BW is highly over 100MHz. If it is rehjected to 100MHz and after then BW shape drops steep you can not see this effect, because in this case ADC can not see this fast (rising) edge, if edge rise is limited to over 3ns you can see very small amount of corners aliasing. (Alisiang components are this waveform harmonics) 
Now situation change if you shut off Sin(x)/x interpolation and also do not use linear interpolation. Just use dots. Now you can see corners do not "wobble". But, you see more or less dots but perhaps not coninuous line because it have limited wfm/s speed. Turning example 1s persistence may help more nice visibility.
But overall this signal breaks oscilloscope specification (rise time).
If you take oscilloscope what analog front end before ADC is not so wide you can not see this effect because front end limits signal risetime.   Of course if there is more fast samplerate this effect also disappear. As can see when you use 1GSa/s.
It is very important to know signals fundamentals and also oscilloscopes fundamentals.

Then you tell it goes to "roll mode".  Yes, it is default.  But pushing one button you can go to normal sweep mode. And doing trigger setup  you can adjust more things. Example if you set 500ms/div and NOT roll mode. You get 7 second long capture but of course samplerate is only 2MSa/s. (so in practice max sinewave frequency around or under 700kHz and otheer waveforms so that its harmonics stay (enough) below Nyquist frequency (sample frequency / 2).
And also it need note that in this case you need wait before see anything on the screen (of course).

External trigger do not at all have same trigger performance than main channels. EXT trigger use analog side pathway trigger system what have been very normal in digital oscilloscopes also in main channels.
Siglent main channels use full digital side trigger system. But EXT trigger do not. And it can see - lot of more  time jitter and not perfect signal fine adjustment to trigger time position. Main channels digital trigger system outperforms most old digital scopes conventional trigger method just hands down.
Also specifications tell that EXT trigger performance is highly reduced. <100ps trigger jitter is specified only for CH1 and CH2.  It reads clearly in data sheet.

Here some examples and about basic fundamentals you can find useful readings from many sources, also from Keysight, Tektronix and R&S application notes etc.


Here only linear interpolation and dots. Sin(x)/x make lot of more bad things when we go too near f Nyquist.
It is normal  that experienced user also know that Sinc interpolation is not always best for not sinetype signals.


Sine, Linear, Dots. Persistence on.  In this image input signal trise is adjusted so that there is  ~1.25 sample for trise (10%-90%)
Note: As can see,  using linear interpolation and persistence there can estimate also true risetime. (red marks) (as long as oscilloscope analog BW do not reject risetime.)

One "problem" specially for unexperienced (I do not mean length of experience but quality of experience) users in example this Siglent model is that analog front end frequency bandwidth is bit too wide, specially if think lower than 1GSa/s samperates in use. Avoid problems with this, need know how to do. But, peoples want "high Bandwidth..."

Here example one individual SDS1102X

As can see f Nyquist (500MSa/s) is not attenuated even 6dB.  It is clear that fast risetime to input leads lot of aliasing and in this case one special form of aliasing: "corners wobbling".

Know your equipment for better measurements is rule number one.

[sp2iqw]

Thank you for both replies.

I have missed some information. The firmware is the most current 1.12.13R5, FPGA Version 16.8.2, HW Version 3-3. The PC runs Win XP SP3, Intel Z77A MB/i3 CPU, 8GB RAM, SSD. Obviously MS.Net is installed. Otherwise the LAN connection will no function. The weak point is OS but it has nothing common with stand-alone oscilloscope operation.

Yes, I have tried the input impedance of 50ohm. That "jitter" is seen on the waveform "corners" with the second channel turned on with 50ohm internal termination. In fact it is not a jitter but the way a signal is processed inside SDS1102X with only A/D converter.

As per memory depth. With 1Gs/s and 10ns/div one ADC can take only 140 samples per window so we have either 140 or 70 per channel respectively for 1 and 2 channels.

The additional information I have uploaded here: https://drive.google.com/open?id=0ByQgomgYIGXpWEZtQ1N0aG1aNjA

I do not want to break the physics. I know that tested signals (AHC CMOS logic gates) edges exceeds the SDS1102X capability so I do not expect more that can be sampled and displayed after postprocessing.

Any physics limits doesn't explain multiply edges effects which sometimes occurs during normal observation (problem #4 of my previous report). With known and trusted signals I know where I am. But in real terms, when I will see similar display on unknown signal, what to do?, how to interpret a signal?...

Michal

[rf-loop]

About your previous "case 4"

I have tested it with quite same settings you have done.
What I do not know are your signals. What they are exactly and how they are connected and so on.

Here in attached image.
Signals.
10MHz square CH1 as seen its rising edge.

Trigger source Ext
800mVp-p pure sine.

Both signals have same frequency reference, Datum Rubidium oscillator.

Persistence infinite and follow time 3 hours. No any single trigger fails. Also when I adjust trigger level it works rock solid. But my signal are also rock solid and signals to oscilloscope so that they stay clean. (btw, if you use slow scope, it perhaps just do not see these jumps due to extremely high blind time. Turn Siglent for slow acquisition mode  then it works like conventional slow DSO.

This 5ns jumping in your image do not come from oscilloscope trigger system (Exept if your scope is broken. And sorry but I do not believe this.)
Problem is perhaps outside of scope, perhaps somehow in your signals.


----

Then you talk about "smoothing" and that this have some problem. What problem. You still think these corners wobbling is problem in scope. No, they are not. It is just pure math. It works just normally what I can see in your images with this kind of signal what have too high frequency components for this sample speed.
You can inspect it more. Let scope run some time with this 10MHz signal, example over ten seconds using 2ns or 5ns/div. Then stop scope. Go to History. There is now 80k last waveforms in waveform history buffer memory. Select playback interval time example 300ms. Then playback. Look carefully what happen there in signal corners. Change Sin(x)/x off and on. Also playback with dots only.





10MHz square to CH1 and 10MHz 800Vp-p sine (both signals have same Datum Rb as reference clock) to Ext trigger input. Infinite persistence and 3 hour continuous run. As can see no any single jump.



And then in your next document. You have used old  GDS-820C. Screen show that two channels and one channel on, both have 2.5GSa/s. Of course you do not see difference - in practice and in theory.
But, try this "Equal time" mode with GW so that signal is not fully repetitive. THis GW is NOT real time oscilloscope at all with this setting.
This GW scope real time samplerate is max 100MSa/s (I do not know how it is if 1 or 2 channel in use). When you use it in repetitive mode it need sample same signal many many times before it have collected amount of samples what are "Equal" with 2.5Gsa/s. Response for signal changes is very very slow.  But, with repetitive sampling mode (aka Equal time mode) there is one advantage. (If signal is continuous)  It can solve fast edges without aliasing. (but only if signal is repetitive)
With enough fast real time scope you can do (nearly) same using dots mode because there is not interpolation between samples. But because fast wfm/s update rate in can plot so much dots that it looks nearly like continuous line. Slow scope can not do it.  With 5ns/div and one channel use and dots mode this Siglent model can do over 35000 wfm/s.  TFT refresh speed is roughly around 25 times/s. So every single screen may have in this case 1400 waveforms stacked overlay on the screen. This is why dots density is still high. But it do not produce corners wobbling due to aliasing.
If I take my very  old HP 100MHz scope what have 10MSa/s samplerate with same kind of signal, it can produce this same when I wait enough time.
Fast and ultra fast (only analog front end risetime is limit)  risetime measurement without aliasing can do even with one sample in second. If have enough time to wait result.

SDS1000X as many other scopes today are real time scopes and many do not have repetitive mode (Equal time mode) at all anymore.  This kind of scope, if it is fast enough and if signal trigger frequency is enough high (in your case 10MHz) can do nearly same when turn Sinc off and use dots mode and iif need also persistence on. Because there is no sync between scope sampling clock and signal frequency, sampling is random time and it produce mostly quite nice trace.

In my image can clearly see that there is some small amount of trigger jitter. Ext trigger is conventional analog-comparator  type trigger. (what most DSO's have used but more and more modern DPO's not)

Main channels trigger system in SDS1000X is very different and lot of more accurate. It is not made as "old school" DSO trigger. (just like this Ext trig and just what is also in all channels in this your GW). 
Main channels trigger in SDS1000X is fully digital side trigger system with much more accurate timing and less trigger jitter (100ps max). 

It is good practice to first study and after then thing what is the cause and what is effect - causality. If do not know enough how things works how can claim they work wrong.

This is one you can read carefully.
http://cdn.teledynelecroy.com/files/whitepapers/wp_interpolation_102203.pdf
Read it enough times for fully understanding every word.
After then come and justify that SDS1kX works wrongly in this your example case - if you still think so.

 

[zorromen]

Hello, SDS1102x was updated as it should be on 1.1.2.13R5, I stitched it twice, before it was 1.1.2.6, and everything worked there that I need, I work in conjunction with the SDG2042x generator and manage them through VISA commands. So on the new firmware the phase detection team ("MEAD?" And "MEAD? PHA") stopped working, although everything worked before the firmware, or I'm stupid or problem firmware, as other teams are working off. The question may be to return to the old firmware or it will destroy the oscilloscope?

[tautech]

New FW
V1.1.1.2.15
1.9 Mb

Changelog
1. Increase font size for Measurement
2. Optimize self calibration for Channels
3. Decode can be enabled while timebase bigger than 20ms/div
4. Fixed some bug in decode function.

[vpv]

New FW
V1.1.1.2.15
1.9 Mb

Changelog
1. Increase font size for Measurement
2. Optimize self calibration for Channels
3. Decode can be enabled while timebase bigger than 20ms/div
4. Fixed some bug in decode function.

Hello!
Once I made this update and my first channel disappeared.
I did Self calibration, but first channel could not be seen again.
Restart oscilloscope -  first channel not visible again.
I did the second time Self calibration and finally first channel appeared! But a second channel was much more stable than the first channel.
Then I made a third time Self calibration and now everything is fine. 

[GeorgeOfTheJungle]

3. Decode can be enabled while timebase bigger than 20ms/div
4. Fixed some bug in decode function.[/i]

Hi @tautech,

Sounds like great news! Does that fix the decoder bug? Any idea at what sweep does it shut down the decoder now?

[tautech]

New FW
V1.1.1.2.15
1.9 Mb

Changelog
1. Increase font size for Measurement
2. Optimize self calibration for Channels
3. Decode can be enabled while timebase bigger than 20ms/div
4. Fixed some bug in decode function.

Hello!
Once I made this update and my first channel disappeared.
I did Self calibration, but first channel could not be seen again.
Restart oscilloscope -  first channel not visible again.
I did the second time Self calibration and finally first channel appeared! But a second channel was much more stable than the first channel.
Then I made a third time Self calibration and now everything is fine. 

Thanks.
You're the first to have reported the outcome of the update.....interesting, I've got an X+ (MSO) unit here and I'll check for similar unexpected results. Good to know all ended well. 
BTW, did you read the update instructions and comply exactly with them ?

Very important!
It need to update twice to update the machine from 1.1.1.2.6(or below this version) to 1.1.1.2.13R3(or above).The first time is from 1.1.1.2.6 to 1.1.1.2.13R3,the second time is from 1.1.1.2.13R3 to 1.1.1.2.13R3 itself.
 Very important!
Because 1.1.1.2.15 optimized self calibration for Channels, it need to do self calibration once if the machine is updated from 1.1.1.2.13R5(or below) to 1.1.1.2.15(or above).

3. Decode can be enabled while timebase bigger than 20ms/div
4. Fixed some bug in decode function.[/i]

Hi @tautech,

Sounds like great news! Does that fix the decoder bug? Any idea at what sweep does it shut down the decoder now?

Not yet checked George, maybe vpv can look at what timebase setting can be done in Decode mode before it drops the Decode UI.
It had to do with Roll mode settings and while in Decode at slow timebase settings it should not enter Roll mode.

[vpv]

Hello, tautech!
I always update the same day to the new version comes out.
Before this update (1.1.1.2.15), I've been using version 1.1.1.2.13R5 and I've complied with all the instructions that were given.

I want to ask about another thing, it has no relation with the new update.
When I measure the same signal, only when I move the beam up or down and the testimony of "Min=" vary. Is this normal?
I attach the video:
https://youtu.be/rJCneswlRDg

Not shown in the video, but when I picked up the beam oscilloscope 4V, then Min = 0.00mV





Hello,  George!
I do not use now decoding function, but now I tried it and it worked to 20ms.
When I switched to 50ms, decoding button went dark and when I pressed the button again, then I wrote the screen "Function can not be used."
If that was the case before, then this problem is not fixed with the latest update(1.1.1.2.15).

[tautech]

I want to ask about another thing, it has no relation with the new update.
When I measure the same signal, only when I move the beam up or down and the testimony of "Min=" vary. Is this normal?

So you look at the probe Cal with is ~3V @ 1KHz referenced to 0V.
We see the Min above 0V and I ask if your scope had been fully warmed up and a Self Cal performed for accurate measurements to be taken ?
If you try the same again after scope warm, Self Cal and with the waveform at a larger amplitude I think the measurement will be more accurate. Please try this and report findings.


Yes vpv, the Decode is as you describe and in the checks I did today seems much more stable. Packets were decoded without the need for the Decode trigger suite with only a need for the correct edge trigger to be selected.


One more thing about the new FW, I can confirm that immediately after the update and reboot the Ch 1 trace is missing until the Self Cal is performed. It reappeared immediately on starting the Self Cal and remained there after.
Unlike vpv I did not have to perform a # of Self Cals.

[vpv]

My oscilloscope is working for more than two hours. I think that is already hot enough     and I did again Self calibration.
The first two pictures are again from the built-in oscilloscope generator ( ~3V @ 1KHz).





The next two pictures are from an external generator 6V / 20kHz.







I do not have to account of whether the oscilloscope measured accurately or generator is accurate.
The question is why the readings vary the position of the beam, whether it is up or down.
Why should there be a change in the readings from the position of the beam?

P.S. Sorry for my terrible English!

[GeorgeOfTheJungle]

Hello,  George!
I do not use now decoding function, but now I tried it and it worked to 20ms.
When I switched to 50ms, decoding button went dark and when I pressed the button again, then I wrote the screen "Function can not be used."
If that was the case before, then this problem is not fixed with the latest update(1.1.1.2.15).

Hi vpv, and thank you very much!

So if this -which is so clearly stated- simply isn't true "3. Decode can be enabled while timebase bigger than 20ms/div" why would "4. Fixed some bug in decode function" such a loosely worded statement be?

Well done Siglent, well done.

[GeorgeOfTheJungle]

Yes vpv, the Decode is as you describe and in the checks I did today seems much more stable. Packets were decoded without the need for the Decode trigger suite with only a need for the correct edge trigger to be selected.

Yes @tatutech all is fine and dandy except that 1) this has nothing to do with the decoder bug and 2) you surely only checked again @9600 baud and 3) vpv's checks show that what you said WRT the max decoder sweep speed is not true.

Great!

[pascal_sweden]

Update instructions  (Very important ! )

It need to Update twice to Update the machine from 1.1.1.2.6 (or below this version) to 1.1.1.2.13R5 (or above).The first time is from 1.1.1.2.6 to 1.1.1.2.13R5, the second time is from 1.1.1.2.13R5 to 1.1.1.2.13R5 itself.

This seems complete bullshit! Why does Siglent come up with such a confusing process in the first place?

Why can't they make a single pass FW update?

Even if they can't make a single pass FW update, why don't use a different version for the intermediate FW update and the final FW update? At least, that would make sense!

Please pass this on to Siglent Engineering, and ask them if they think it makes sense?

[tautech]

Yes vpv, the Decode is as you describe and in the checks I did today seems much more stable. Packets were decoded without the need for the Decode trigger suite with only a need for the correct edge trigger to be selected.

Yes @tatutech all is fine and dandy except that 1) this has nothing to do with the decoder bug and 2) you surely only checked again @9600 baud and 3) vpv's checks show that what you said WRT the max decoder sweep speed is not true.

Great!

I only have one source of protocols to check FW improvements with George, sorry.

What I said WRT max speed is only what what was stated in the Changelog:
It states: 3. Decode can be enabled while timebase bigger than 20ms/div but it is NOT the case. I do not know why this has been stated but not enabled as I have been told the engineers were working on this issue.
For this to happen the Roll mode needs be disabled when in Decode mode.


vpv, I will look more into your issue, I do not doubt your measurements or the accuracy of the signals, yes they are not the issue. I did ask if you could repeat the All Measure with a larger displayed waveform, say 500mV/div.
When I have some time I will check a unit I have here.

Update instructions  (Very important ! )

It need to Update twice to Update the machine from 1.1.1.2.6 (or below this version) to 1.1.1.2.13R5 (or above).The first time is from 1.1.1.2.6 to 1.1.1.2.13R5, the second time is from 1.1.1.2.13R5 to 1.1.1.2.13R5 itself.

This seems complete bullshit! Why does Siglent come up with such a confusing process in the first place?

Why can't they make a single pass FW update?

Even if they can't make a single pass FW update, why don't use a different version for the intermediate FW update and the final FW update? At least, that would make sense!

Please pass this on to Siglent Engineering, and ask them if they think it makes sense?

It makes perfect sense to me even if you don't understand it.
A few posts back rf-loop said: do not ask why, just follow the instructions (or similar).
We do not need to know how a tool is made only how to use it.

When an update is unpacked from the rar download there are clear instructions how to run the update, the only confusion comes from those not keeping up to date with the new FW. If you do, the double update is only needed to be run once.
In the past some FW packages have contained a cfg file to run before an update.....this is more confusing for users, it is much simpler to run an update twice to enable improvements the engineers have added.
I will guess that all units that did need the double update have been long sold, those that I had were sold last year.

[tautech]

My oscilloscope is working for more than two hours. I think that is already hot enough     and I did again Self calibration.
The first two pictures are again from the built-in oscilloscope generator ( ~3V @ 1KHz)............

(with cold scope).....
Yes there seems some small bug in All Measure Min after a quick check. If you select only Min and/or others from the the Measure menu then values measured are within the accuracy specs of the scope.

Measurements are always more accurate when the displayed waveform is large on the display.

[vpv]

tautech, you're right!
I am only surprised, why there are different values of measurements at different position of the beam on the screen.
In this case I do not care whether the measurement is true, and why there is a change in a different position of the beam on the screen.

[zorromen]

It is interesting to do in future firmware so that you can change the color of the signals of channels 1 and 2, green for example, since the violet does not look very interesting.

[tautech]

It is interesting to do in future firmware so that you can change the color of the signals of channels 1 and 2, green for example, since the violet does not look very interesting.

Similar is a suggestion I've recently forwarded to engineers: to make trace colours user definable.

[zorromen]

What maximum voltage on the probe 1:1 withstands the oscilloscope at 1mv / div? The body is written with 400 Vpk, is it at any sensitivity and at the probe 1:1 or 1:10?
The User Manual shows a certain Quick-Cal mode. I do not have it, it's the latest firmware.

[tautech]

What maximum voltage on the probe 1:1 withstands the oscilloscope at 1mv / div?

To subject any scope to high voltages at low V/div settings qualifies as gross misuse of a scope. 

Always use probes @ 10:1 settings to give additional safety against overloading inputs.
Use 1:1 only on occasions where higher sensitivity is required.

The body is written with 400 Vpk, is it at any sensitivity and at the probe 1:1 or 1:10?

1:1 or direct BNC cable connection @ 1M input settings only.
50 inputs must be restricted to 5V max

The User Manual shows a certain Quick-Cal mode. I do not have it, it's the latest firmware.

No it is not present in 1.1.1.2.15 FW.
Quick Cal is available in SDS2000 and 2000X models and one can only imagine that mention of it in the manual is for future inclusion of it but AFAIK accuracy drift of these units is not such a problem to need this feature enabled.

[zorromen]

It does not matter which maximum voltages are acceptable for different sensitivities. You can apply 300 Vpk to the BNC cable connection and the oscilloscope will stand, or not and on what sensitivity it will not survive? Or at 1:10 you can measure up to 4000Vpk? Where it all is written, how you wrote this gross violation of the region, which area, I do not understand. The instruction does not give a complete answer to this question.

[tautech]

It does not matter which maximum voltages are acceptable for different sensitivities. You can apply 300 Vpk to the BNC cable connection and the oscilloscope will stand, or not and on what sensitivity it will not survive?

There is no mention of operational conditions that aren't survivable but from my days using analogue scopes one always took care not to apply a voltage that would take the displayed waveform beyond the limits of the display.
DSO's are very different in this respect.
The maximum voltages that can be applied are the sum of the probe and input capabilities.

Or at 1:10 you can measure up to 4000Vpk? Where it all is written, how you wrote this gross violation of the region, which area, I do not understand. The instruction does not give a complete answer to this question.

The datasheet and manual give the maximums and condition that apply.
The leaflet that comes with each probe give the specifications for 1:1 and 10:1 operation. IIRC 650V DC is the max @ 10:1 BUT this voltage diminishes with frequency as outlined in the frequency derating graph.

If you intend to connect to elevated voltages of some frequency I suggest you get 100:1 or 1000:1 probes to retain user and scope safety.
If the probe attenuation is sufficient to keep the DSO input voltages below the maximum input voltage then you can measure most anything.

Can I ask what you intend to measure that the maximums are of concern ?

[zorromen]

I'm asking the question more specifically, it is more understandable for me to operate with the voltage that came to the input of the oscilloscope itself, and no matter what the probe. For example, what voltage +/- DC can withstand an oscilloscope at 10V / Div, and 1 mV / Div? And no matter how they got there, maybe by accident, I should understand how reliable my oscilloscope is. So the question about the 50 Om input, as I understand the limitation of <5Vrms is determined by the heating of the 50 Om resistor itself?

[tautech]

I'm asking the question more specifically, it is more understandable for me to operate with the voltage that came to the input of the oscilloscope itself, and no matter what the probe.

The probe always matters, it's division ratio sets what V the scope input will see.

For example, what voltage +/- DC can withstand an oscilloscope at 10V / Div, and 1 mV / Div? And no matter how they got there, maybe by accident, I should understand how reliable my oscilloscope is.

Just as reliable/robust as any other DSO.
Consider that a 10:1 probe can not be safely used at a voltage that will apply the max 400 input voltage then the only scenario that might apply more is a 1:1 probe and to do so exceeds the V rating at a 1:1 probe setting which is just 300V unlike 10:1 which is 600V and both derate with frequency. Study the probe leaflet. 

So the question about the 50 Ohm input, as I understand the limitation of <5Vrms is determined by the heating of the 50 Ohm resistor itself?

Simply yes.