Siglent SDS1104X-E anomaly on what should be a simple task.

[bdunham7]

I recently hauled out a project I've been planning and that is making a replacement LED display for the Fluke 8842A.  That's another story--in short a drop-in replacement is impractical--but I ran into what I think is a shortcoming of the Siglent scope.  I haven't tried to reproduce it with other test signals, so I'll just show you what I have.  What I'm doing is taking the segment and digit scan signals from the VFD drivers of the 8842A and using them to drive a set of 7-segment blue LED displays.  The display has 8 'digits' and 11 'segments' total, so 8 lines x 11 lines.  I'm just looking at 3 digits and the 7 segment lines for now.  The VFD drivers put out a +30V signal on both the segment and digit lines and when both are on, the segment for that digit lights.  This obviously won't work for an LED, so I'm using the segment scan lines to drive the segments of common cathode displays in parallel and the digit scan lines to drive a darlington array which then grounds each common cathode in sequence.  I'm using dropping resistors at this point, but I'll need to use a lower voltage supply for the system to work.  I needed to know the maximum voltage drop across the darlington array, which I decided to determine by connecting a scope between ground and the common cathode of one digit.  The specs say about 1 volt, but I was considering using a 5 volt supply, so I needed to know fairly closely.

I connected Mr. Siglent with one probe to the incoming digit line and triggered off of that, then the other probe to the common cathode connection.  The signal is noisy when the array transistor is off because of cross-feeds from other digits, but when on it seems to go down near zero.  So I started changing the V/div setting and at 2V/div, it indicated a half a division above ground.  I changed nothing except I twisted the V knob 1 click, the scope went to 1V/div and I heard a relay click--but the signal in question was now 4 division above ground. I rechecked everything, thinking I had some strange ground loop or capacitive coupling or bad probe (the probe was the Siglent PP510 and it is falling apart), so I tried another probe, a different channel, moving the grounds.  I also tried both 1X and 10X, nothing changed and the problem was repeatable.

So, I tried some other scopes.  The problem didn't occur on my Tek TPS2024 and even my ancient Scopemeter gave me the correct result.  Well, both of those are isolated, so perhaps still a ground?  Now this is not a fast signal, the scan lines stay on for ~1 millisecond, which makes it difficult to even see the signal with an analog CRO.  But, I still have a Tek 2221, 60MHz 20MSa/s, and it is common ground just like the Siglent or any other scope.  Even that worked properly. 

I think I know what the issue is now, thanks to a comment David Hess made a while back, and I'm sure Siglent will claim it is user error.  However, what I'm doing is not an unusual use of a scope and if this is a limitation, it would be good to know about it. 

Photos are a rough sketch of what I'm doing (on an actual envelope), the Siglent at 4 different V/div settings and then the Scopemeter, Tek TPS and Tek 2221 on just the final screen.

[masterx81]

You are overdriving the input amplifier.
A good read:
https://www.evaluationengineering.com/instrumentation/oscilloscopes/article/13002369/overcoming-overdrive-recovery
Every scope have it's characteristics.
Common "problem" on dso scopes.

[tautech]

You are overdriving the input amplifier.
A good read:
https://www.evaluationengineering.com/instrumentation/oscilloscopes/article/13002369/overcoming-overdrive-recovery
Every scope have it's characteristics.
Common "problem" on dso scopes.

Certainly overdriven in the last screenshot and the giveaway is the long vertical tails in this screenshot:



For the first screenshot the complete waveform is entirely on the display so another explanation is needed and the possibilities are; the unmatched probes, channel crosstalk, probe capacitive loading, circuit PSU not up to the task or inadequate decoupling yet the blip on the wider pulse at minimum amplitude is likely to have any impact at all as it seems well removed from any threshold levels.



It is my view the scope is correctly displaying something and further investigation should discover what yet if it has no effect on the operation of the DUT why bother.

[bdunham7]

For the first screenshot the complete waveform is entirely on the display so another explanation is needed and the possibilities are; the unmatched probes, channel crosstalk, probe capacitive loading, circuit PSU not up to the task or inadequate decoupling yet the blip on the wider pulse at minimum amplitude is likely to have any impact at all as it seems well removed from any threshold levels.
It is my view the scope is correctly displaying something and further investigation should discover what yet if it has no effect on the operation of the DUT why bother.

Sorry if that wasn't clear, but the first screenshot is not a problem at all--nothing is wrong.  I just put the arrow there to show the area that I was trying to measure.  The messy waveform is just a side effect of how I set it up to measure that specific voltage drop.  The only time the scope is properly connected is when the green trace is high and the driver is on, the rest of the time it is bleed-through into a disconnected high impedance circuit.  So the entire waveform during the time the green trace is low is just garbage.  Of course that garbage is what is overdriving the scope, so I can't ignore it.

Certainly overdriven in the last screenshot and the giveaway is the long vertical tails in this screenshot:

You can't see it in a still photo, but all of those bits hanging by those tails were dancing up and down like marionettes (puppets on strings).  I actually don't see a lot of evidence that the waveform was badly distorted and the area being measured has the appropriate shape--flat in the middle and low at the extremes--but it is offset upwards by 3 volts.  So an entire millisecond goes by and there's no sign of any recovery.

[bdunham7]

You are overdriving the input amplifier.
Every scope have it's characteristics.
Common "problem" on dso scopes.

Yes I am, and while every scope has those characteristics, almost none seem to specify them.  The only reference I could find that did was a statement from someone at Cleverscope stating that their recovery time was 1 sample, but no further specifics.  The way I amplified my area of interest is not at all an uncommon way to do it, thus my initial confusion.  I expected it to work because I've done things the same way with other scopes. As I showed, the other DSOs either didn't overload at that same level or they exhibited much faster recovery times, if that is exactly what is going on here. 

So according to Tektronix, they don't specify a recover time from an overdrive situation, they say that if you are concerned about it, you have to characterize the scope yourself.  So it seems that the main characteristics would be 1) at what level the overdrive condition occurs and 2) how long it takes to recover.  It seems to me that 'more than a millisecond' is a pretty long recovery time for a 200MHz scope.

[Fungus]

Yes I am, and while every scope has those characteristics, almost none seem to specify them.

You're operating the chip outside it's maximum rating. What happens next is unknown even to the chip manufacturers. 

[bdunham7]

You're operating the chip outside it's maximum rating. What happens next is unknown even to the chip manufacturers. 

It says "400V_pk right on the front of the scope.  I surely haven't exceeded that!

Anyway, this isn't a complaint post that needs a resolution.  It is an issue any scope user should be aware of and this one caught me off guard because the characteristics of this scope seem a bit different than all the others I tried.  Since expanding a particular area in the way I did is a useful measurement tool, I would think it would be worthwhile to determine what the limits are of this scope in using that technique.

[tautech]

The specification you are exceeding is:

Offset Range (Probe 1X)
500uV~118mV: ±2V
120mV~1.18V: ±20V
1.2V~10V: ±200V

Datasheet P8 Vertical System
https://siglentna.com/wp-content/uploads/dlm_uploads/2020/02/SDS1000X-E_DataSheet_DS0101E-E04C.pdf

[bdunham7]

The specification you are exceeding is:

Offset Range (Probe 1X)
500uV~118mV: ±2V
120mV~1.18V: ±20V
1.2V~10V: ±200V

Datasheet P8 Vertical System
https://siglentna.com/wp-content/uploads/dlm_uploads/2020/02/SDS1000X-E_DataSheet_DS0101E-E04C.pdf

That appears to just be the maximum setting for the vertical position knob in various ranges--which I confirmed by trying it.  I was interested in the 118/120 thresholds and those are the points that I hear a relay click.  So I set up a test signal to see if and when overdriving appeared to happen.  I used a 5 volt rectangular pulse of 100 milliseconds at 5 Hz with a superimposed 100mV 1ms 100Hz rectangular pulse.  The goal would be examine the bottom trailing edge to see the step down and subsequent pulses in as much detail as possible.  I found overdriving and recovery artifacts even within the offset specs you posted, although they became much more severe as the scope shifted down from 120mV/div to 118mV/div, as you might expect.

The test signal:



Expanded to 10 division, 6 of which are on-screen.  You can see 8 or 9 milliseconds of distortion already.



More distortion at 200mV/div, which according to your specs would be good for +20V offset.



Now at the minimum 120mV/div before the range switches and the relay clicks.  15 ms of ringing.



Fine adjustment one click down to 118mV/div, relay clicks, fairly extreme ringing for ~50 ms.



Reducing the top of the large signal to 1.5V to stay within the offset spec.  Obviously still 'ringing', at an absurdly low frequency.



And now back up just over the threshold at 2.1 volts.  It appears to matter a lot how much it is overdriven.  But clearly I can't closely examine my falling edge this way with this scope.



And here is the Tektronix expanded all the way to 20mV/div with a 5-volt top--that is 250 divisions of overdrive and it still recovers quickly enough so that I can see the step down and subsequent pulses just as they should be, without a hint of overshoot, ringing or distortion.



This makes it pretty clear that you need to keep the signal on-screen with the Siglent scope to avoid distortion--and the 'recovery' takes a very long time as scopes go.  I'm not sure what the technical reason is for this, but it is a limitation to know and understand, I guess. 

[tautech]

I'm not sure what the technical reason is for this, but it is a limitation to know and understand, I guess.

Know, understand and work with.
AC input coupling sometimes can help. 

[bdunham7]

AC input coupling sometimes can help. 

Definitely not in this case.

[bdunham7]

Nothing better than a Jim Williams application note on point:

https://www.analog.com/media/en/technical-documentation/application-notes/an120f.pdf

Appendix B is most pertinent, but the whole note is of interest. 

[tautech]

AC input coupling sometimes can help. 

Definitely not in this case.

It should.
The total waveform amplitude if using AC input coupling will be shifted with 50% above and below the channel 0V marker which will allow you to offset the signal further before exceeding the offset limitation.
The attenuation of the probe used however can put you into a different offset spec but some quick mental sums can disclose if that might be a problem.

[bdunham7]

It should.
The total waveform amplitude if using AC input coupling will be shifted with 50% above and below the channel 0V marker which will allow you to offset the signal further before exceeding the offset limitation.
The attenuation of the probe used however can put you into a different offset spec but some quick mental sums can disclose if that might be a problem.

I have a signal that never goes to zero, and I'm trying to measure the absolute difference between zero and one particular part of it.  I don't see how AC coupling could ever accomplish that.

And I see the forum has struck again with the photos not showing properly, at least for me.  What gives?

[tautech]

It should.
The total waveform amplitude if using AC input coupling will be shifted with 50% above and below the channel 0V marker which will allow you to offset the signal further before exceeding the offset limitation.
The attenuation of the probe used however can put you into a different offset spec but some quick mental sums can disclose if that might be a problem.

I have a signal that never goes to zero, and I'm trying to measure the absolute difference between zero and one particular part of it.  I don't see how AC coupling could ever accomplish that.

With Measurements or Cursors.

And I see the forum has struck again with the photos not showing properly, at least for me.  What gives?

Forum bug. Insert pics the old way with links to your uploaded thumbnails inserted between [ img] flags.
Use only the 2nd attachment (more attachments) to also get around the attachment bug.

[bdunham7]

With Measurements or Cursors.

Still not seeing that it is possible.  Let's try a simple example.  I have a triangle wave of about 5 V_rms riding on an approximately 10VDC signal.  I don't know either value exactly.  What I want to do is measure the voltage relative to ground of the signal at the lowest point of the triangle wave.  How can AC coupling manage that?

[tautech]

With Measurements or Cursors.

Still not seeing that it is possible.  Let's try a simple example.  I have a triangle wave of about 5 V_rms riding on an approximately 10VDC signal.  I don't know either value exactly.  What I want to do is measure the voltage relative to ground of the signal at the lowest point of the triangle wave.  How can AC coupling manage that?

Later I can arrange similar within the limits of an AWG and grab some screenshot examples. Poke me if I forget.

[kcbrown]

And I see the forum has struck again with the photos not showing properly, at least for me.  What gives?

I've never seen it work properly.

What you have to do is to attach them as you did, then post the message, then examine the result in your browser (call the result of this the "view" of the message), and then edit it in another window or tab (right click on "modify" and tell the browser to open in another window or another tab, at your discretion).  Expand the images in the view of the message, and for each one:

1.  right click on the expanded image and tell it to copy the image location,
2.  paste it in the appropriate location (i.e., wherever in the text you want the image to appear) in the edit box, then
3.  select the resulting pasted URL in the edit box (via click and drag while holding the mouse button down) and
4.  click on the "image" icon, which will enclose the URL in "image" tags.  The image icon is the leftmost icon in the second row of icons that appears right above the editing box.  If you hover your mouse pointer over the image icon, it should say "Insert Image".

Then save the message.

With any luck, the above makes some sense.

[bdunham7]

And I see the forum has struck again with the photos not showing properly, at least for me.  What gives?

I've never seen it work properly.

What you have to do is to attach them as you did, then post the message, then examine the result in your browser (call the result of this the "view" of the message), and then edit it in another window or tab (right click on "modify" and tell the browser to open in another window or another tab, at your discretion).  Expand the images in the view of the message, and for each one:

1.  right click on the expanded image and tell it to copy the image location,
2.  paste it in the appropriate location (i.e., wherever in the text you want the image to appear) in the edit box, then
3.  select the resulting pasted URL in the edit box (via click and drag while holding the mouse button down) and
4.  click on the "image" icon, which will enclose the URL in "image" tags.  The image icon is the leftmost icon in the second row of icons that appears right above the editing box.  If you hover your mouse pointer over the image icon, it should say "Insert Image".

Then save the message.

With any luck, the above makes some sense.

Thanks!  Got 'er done.  That's a lot of extra work...

[kcbrown]

Thanks!  Got 'er done.  That's a lot of extra work...

Yeah, it's really annoying.  I don't understand why the site operators can't fix it or get it fixed.  It's not like it's an obscure feature that hardly anyone would use.  It's a very common thing for people to want to do here.

[tautech]

With Measurements or Cursors.

Still not seeing that it is possible.  Let's try a simple example.  I have a triangle wave of about 5 V_rms riding on an approximately 10VDC signal.  I don't know either value exactly.  What I want to do is measure the voltage relative to ground of the signal at the lowest point of the triangle wave.  How can AC coupling manage that?

Simple examples displaying the usefulness of AC coupling.
Waveform details deliberately not stated as the displayed info is sufficient to understand what it is. DC coupling tells that story.
Channels 2 and 4 show the same waveform and the offset value when DC coupled shows how much the waveform is above 0V. Measurements deliberately omitted as we have graticules.


AC coupling removes that offset and allows us to examine parts of the waveform to the limits of the channel offset specification. Channel tabs and menus tell the story. 
Any part of a waveform can certainly be measured with cursors.

[bdunham7]

I'm listening, but not convinced yet.  That looks a bit messy and I think you may be overdriving your scope as well in that example, even with the AC coupling.  Offset range and amplifier dynamic range are different and you are driving it 190 divisions off the top of the screen.

Humor me--try the exact test signal I was using.  I used the SDG2042X, with 0-5V volt rectangular pulse of 100 milliseconds at 5 Hz on CHQ and a 100mV 1ms 100Hz rectangular pulse on CH2 and then combined them in the output menu.  I want to pretend that small pulse only exists on the trailing edge of the big one--the rest are just for comparison--and I'd like to see if you can use your technique to accurately measure and perhaps cleanly display that pulse as accurately as the Tek scope does in my last picture.

[StillTrying]

Offset range and amplifier dynamic range are different

Yep, everyone is confusing the Y offset voltage with the pk-pk dynamic range of the variable gain amp.

There's some triangular waveforms distorting because of the amount off screen waveform here.
https://www.eevblog.com/forum/testgear/siglent-sds1000x-series-oscilloscopes/msg1208426/#msg1208426

[bdunham7]

The interesting thing--I see it mentioned in the other thread as well--is this is mostly a low-frequency phenomenon.  I say 'mostly' because it causes a residual offset at higher frequencies too.

I tried a different setup--a +/- 5 volt square wave with superimposed 10kHz 100mV pulses.  I set the channel to 200mV/div with a +5V offset so that the lower portion of the wave should be right on the center grid line.  Set it to 1 Hz, captured the trailing edge and then some of the flat section so that you can see that -5V section of the square wave does indeed eventually line up with the center grid.  But the trailing edge shows many milliseconds of recovery from the overdrive.  Then I increased the frequency to 1kHz, now  you can see the bottoms of that are relatively flat, but offset upwards by 200mV.  AC coupling made little difference.  At 200Hz, you see a bit of recovery 'bounce', looking sort of like the one Tautech posted.

[py-bb]

You posted a link to this from one of my threads and I agree your situation is terrible, the pictures show a stark contrast.

What do you think of a Rigol DS1052E?

I'm after a digital one to capture things that don't repeat and I'm not asking for recommendations as such, I'm asking "is it afflicted by the same deal breaking problem?"


Or was yours somehow defective? The probes perhaps?


Thanks