Is this the Siglent SDS1000x-e compensation issue or else?

[Bob Sava]

Same signal shown at different vertical resolutions - once I hear a click of the relay shape of the wave changes (screenshot in purple)

Another issue I see (yellow screenshots) when hooked up to calibration signal I see little lines traveling along the path of the wave (5s persistence is on to show the trail as the 'stick' line moving along)

[Fungus]

Looks like you got one of the units with missing capacitors.

See: https://www.eevblog.com/forum/testgear/siglent-sds1202x-e-compensation-issue/

[StillTrying]

The 3 purple traces look OK to me, probably just the input overloading on the last one, because the trace is coming back from 26 divisions away.

[tautech]

Bob, you need give us better info:
Probe attenuation
Input attenuation
Both channels.....do they show the same symptom
Swap probes and does what you see go to the other channel ? (Dud probe, I've only seen 1 in 6 years)

Further notes;
The attenuation issue is best displayed after compensating probes carefully @ 10x and 1V/div and then using Fine V/div adjust through 1.18-1.22V/div over the input range relay step.
(Before and after step screenshots)

Do the probes show very similar waveforms on 1x and 10x ?

[rf-loop]

Bob, you need give us better info:
Probe attenuation
Input attenuation
Both channels.....do they show the same symptom
Swap probes and does what you see go to the other channel ? (Dud probe, I've only seen 1 in 6 years)

Further notes;
The attenuation issue is best displayed after compensating probes carefully @ 10x and 1V/div and then using Fine V/div adjust through 1.18-1.22V/div over the input range relay step.
(Before and after step screenshots)

Do the probes show very similar waveforms on 1x and 10x ?

But one is sure,

Looks like you got one of the units with missing capacitors.

what ever jokes and falses Mr Rigol tells about Siglent, due to lack of enough  knowledge and experience,  this have nothing to do with well known compensation issue in some SDS1202X-E's (lack of reactance matching capacitors in some manufacturing lot).

But what this is. It is fact that in SDS1104X-E Voltage band 1. have typically slightly different frequency response (-3dB corner bit lower freg) what bands 2. and 3. (Band 1.: 500µV/div - 118mV/div (BNC input level) And Band 2.: 120mV/div - 1.18V/div, Band 3. 1.20V/div - 10V/Div.  But still well inside specifications.

Also it looks that if signal is same in all OP 3. images first 2 are ok (bit ringing edge corner due to impedance mismatch or if probe is used then it is bit ringing. (Also 100MHz model do not come with better P215 probes) This we do not know because OP did not tell anything about signal and how the signal is connected to an oscilloscope - perhaps peoples have "magical eye or crystal ball" but I do not have.

3th Image square wave corner is bit lazy. (also if we look freq response it drops bit faster in this 1. Voltage range than 2. range.  I can see this using roughly around same kind of signal and coaxial with good Tek 50 ohm feed thru termination.  (exept that termination is not good at all because  terminator without scope input reactance compensation)
Also of course this can see if do sweep with sinewave and look fresponse shape from 0 to >200MHz using V band 1. and 2.


Let's take a microscope and go to beach.   Look from afar - you see the beach - you see perfect shoreline. Take the microscope and try find exactly where is this shoreline  - shoreline disappear and the beach disappears. What a mess.

[Bob Sava]

Square wave is from Fy6600, 10kHz, 2.5v p-p, 8ns raise via BNC cable terminated with 50ohm on a tee. 

I see what you say about zooming in, however, and although slower rise falls within 10% of signal edge, I was surprised hat signal changes shape that much when changing vertical resolution.

[Performa01]

Sorry, but none of the replies so far (except #2) gives the right answer.

It is certainly not a probe compensation issue, since it's a direct coax connection and I'm glad to hear that it's even properly terminated at the scope input.

But it has also nothing to do with the frequency response, which is within 1dB up to 130MHz for all sensitivities from 500uV/div - 1V/div for the SDS1104X-E. To illustrate that, here's the respective graph from my review:


SDS1104X-E_BW_Graph

The reason for the different edges is because the input of the scope is severely overdriven and that in an asymmetrical way on top of that. For folks familiar with analog circuit design the hint might be sufficient that the clipping caused by the overdrive is asymmetric for the low frequency content of the signal, but still symmetric for high frequencies. Depending on the severity of the overdrive, all kinds of funny effects can arise when the two very differently distorted signal portions are combined again at the output of the split path input buffer.

Bob, your scope is fine. Just use it the way it's intended. Don't overdrive it more than a few divisions in either direction and be particularly cautious with asymmetric overdriving. There is no danger to damage the scope, but the waveforms and measurements might become just plain inaccurate.

You can try some exercises just to get a feeling for the effect. Try different amplitudes, hence different degrees of overdriving the input and watch the effect on the waveform. Try AC-coupling the scope input instead of DC-coupling and see how the picture changes.

And always remember: never trust the waveform display or measurements on any scope if the input signal is not entirely within the visible screen area. A little overdrive is usually fine, on the high sensitivity ranges like 5mV/div it can even be massive without consequences, but in general you cannot rely on that and have to make sure in advance that the scope will still show correct results whenever you think you absolutely need to overdrive it.

[Bob Sava]

I think I understand what you're saying - would it help if scope had more than 8 bits of resolution?  For completeness sake here are screenshots of the same signal at different portions of the screen (same vertical scale):

[rf-loop]

Sorry, but none of the replies so far (except #2) gives the right answer.


But it has also nothing to do with the frequency response, which is within 1dB up to 130MHz for all sensitivities from 500uV/div - 1V/div for the SDS1104X-E. To illustrate that, here's the respective graph from my review:


The reason for the different edges is because the input of the scope is severely overdriven and that in an asymmetrical way on top of that.

At this time you are perfectly right. Why I told anything about frequency response. Because I have this experience but....then big but. Time ago I measured frequency response using 50mV/div and then 200mV/div and this just because long time ago I have used it due to many reasons, including that with this and 6div "standard" height on the screen this old generator flatness was precicely checked for extremely good flatness. And this level I use many times just from my muscle memory even when generator is now very different. Then I measured also using 200mV/div.
What I find late yesterday...  generator what I used for this have been bad! It  have some reason flatness problem so that flatness is totally bad with some level settins.  So my previous result about more than small diffewrence between V band 1. and 2. response have been totally wrong. Oh my god what mess...  (I need do also many other things again and even some equipments adjustments what I have done need check or do again... shit /&¤&%/¤&%#¤...)

Good in this case is that I find this problem in my one generator now and not later.
-------------------

For OP. This is not 8 bit question.  This is analog front end (before ADC) question and its linearity and range.

As Performa01 told: "The reason for the different edges is because the input of the scope is severely overdriven and that in an asymmetrical way on top of that."

And as you can see in your last first image, this asymmetric overdrive is most bad and start hit limit what really destroy your sinal top because bottom overdrive negative side.

Here is example what are differences in pulse rising edge shape between scope Voltage band 1. and 2.
Of course this is one individual scope, just as out from box, not myself fine adjusted (and can not see any reason for do it).
Reference line is V Band 2.  and running trace is V band 1. (change between 118 and 120mV) 
Reference line scaled for 118mV/div.
Signal from HP programmable pulse generator, signal risetime adjusted for 1.0 - <1.2ns. RG223 and Tektronix 2GHz rated / 50ohm feed thru terminator.

[0xfede]

Hello Bob Sava,

I have to say that is a very nice beahviour for an input stage that overdriven. (Disclaimer: I do not have any Siglent DSO and I'm not connected with them).

I think I understand what you're saying - would it help if scope had more than 8 bits of resolution?  For completeness sake here are screenshots of the same signal at different portions of the screen (same vertical scale):

If you sample with an higher Voltage/Div setting (within the analog limits of the input stage), stop the acquisition and then vertically zoom in the detail an higher resolution will help (when the acquisition is stopped the zoom is performed in the digital domain).

Best,
0xfede

[Performa01]

Bob, as others have pointed out, this has nothing to do with ADC resolution (bits). It is related to the analog (split path) input buffer exclusively and would be exactly the same even if this was an analog scope (with the same frontend).

Older scopes (and/or low bandwidth ones) might not show the same effect, for at least two reasons:

• There might be just one wideband input buffer/amplifier, working from DC to  e.g. 30MHz (but I've even seen such designs that claimed to work up to 300MHz!), where the signal distortion due to overdriving would be about the same for all frequencies. The drawback of these designs is the poor DC accuracy and stability.
  
• If it is a discrete transistor design, supply rails usually will be considerably higher, hence providing more headroom for overdrive. (Even more so for tube designs of course)
  

[Bob Sava]

I understand that at high gain we are in 'garbage in - garbage out' scenario.  Increasing ADC resolution would show more detail at lesser (analog) gain - would it not?

[tautech]

I understand that at high gain we are in 'garbage in - garbage out' scenario.  Increasing DAC resolution should allow me to see the signal in more detail at lesser (anlog) gain - would it not?

I don't get what you're trying to show us or prove Bob. 

So you want to examine the rising edge of a 10Hz square wave.....well it's been explained that you're doing it using the wrong methodology by using amplitude when you should have the entire amplitude on the display and be using the timebase instead.

Old saying....there's more than one way to cook a goose !

[Performa01]

Yes, quite obviously higher ADC resolution would show more amplitude detail. 0xfede has already mentioned it.

But since a general purpose oscilloscope is not a precision instrument, higher resolution is just not required for tasks like this. 8 bits are enough to analyze the imperfections of a real square wave. Just look at the example given by rf-loop. Even without zooming we can immediately see the difference in signal quality between a professional HP pulse generator, connected to the scope with a high quality cable and through terminator compared to a casual setup where a cheap AWG is used together with a cable and termination of unknown quality.

[StillTrying]

Increasing ADC resolution would show more detail at lesser (analog) gain - would it not?

You can sort of simulate what doubling the ADC resolution(1 extra bit) looks like on screen, by STOPping a full screen height waveform and then halving the Y sensitivity of the STOPped waveform. Sines and curves probably show the effect best, you get a thinner slightly smoother trace as expected, but on a 400-500 pixel screen I don't think there's a massive amount in it.