Strange signal shift on Siglent SDS1202X-E

[kvd]

I wonder about this issue with my Siglent SDS1202X-E scope.

I am feeding a square wave from my function generator into channel 1, and into the external trigger input. The signal is split at the output of the function generator with a BNC splitter.

The scope is triggered from the external trigger input (EXT/5).

Now a strange phenomenon occurs. The signal trace from channel 2 is shifted left when channel 1 is displayed. This is shown in the screen dumps. The signal trace of channel 2 without channel 1 visible is saved into a reference channel, to show the displacement. In the third screen dump note that the tracking cursors do not exactly follow the trace of signal 2.

The acquisition mode is normal. I found out that this phenomenon does not occur when the acquisition mode is averaging (4).

What is causing this effect? Note that the signal on channel 1 has an amplitude in exces of 150V, whereas the signal on channel 2 is less than 0.5V. Is this a shielding effect on the analog front-end of the scope, is it a firmware bug, is it a phenomenon caused by my way of measuring, or something else?

Koen

[jjoonathan]

An offset of 2ns is suspiciously close to 1 / 500MS/s, which I believe is the sample rate with two channels enabled (since 1GS/s is the sample rate with 1 channel enabled).

[tautech]

An offset of 2ns is suspiciously close to 1 / 500MS/s, which I believe is the sample rate with two channels enabled (since 1GS/s is the sample rate with 1 channel enabled).

Yes, quite possibly so.
Although kvd is using AC coupling and triggering so we can't see the trigger point to see any timing correlations. Why is unclear. 

Also in the last shot there's only 70 data pts so just one minor division error might explain it. IMO @ 10ns/div propagation delays is unlikely to be the cause.

[kvd]

Both of these sound plausible explanations. Thanks!

[tautech]

Both of these sound plausible explanations. Thanks!

For interests sake are the results the same using DC coupling for both inputs and the trigger ?

Are offsets such so you need to use AC coupling ?

[Keysight DanielBogdanoff]

This is a guess, but i suspect it might be a trigger hysteresis thing.

The trigger point of a scope when it's doing an edge trigger is actually more of a trigger range - the signal has to hit two comparators in the right order for the trigger to happen. As a result, the signal can sometimes show up slightly shifted on the screen. Then, scope makers shift the waveform back in software so that it doesn't look funny. You can check if your scope does this by switching from a rising edge trigger to a rising-or-falling edge trigger. The alternating edges make it where the scope can't get away with shifting the waveforms, so it'll appear that the signal crosses the trigger point slightly shifted.

I suspect that for this scope, when channel 1 is not displayed but the scope is triggering on that channel, there's no shift. But, when you turn on channel 1 the hysteresis adjustment kicks in. I'd bet that the timing relationships between the two signals isn't actually changing.

Just a guess, as it's not our scope, but this would be what I would look into. You could try putting the scope into rising-or-falling edge trigger mode and see if you get the shift.

[kvd]

An offset of 2ns is suspiciously close to 1 / 500MS/s, which I believe is the sample rate with two channels enabled (since 1GS/s is the sample rate with 1 channel enabled).

Yes, quite possibly so.
Although kvd is using AC coupling and triggering so we can't see the trigger point to see any timing correlations. Why is unclear. 

Also in the last shot there's only 70 data pts so just one minor division error might explain it. IMO @ 10ns/div propagation delays is unlikely to be the cause.

Results are the same with DC coupling for both channel 2 and the trigger. Also with more datapoints the same effect shows.

[kvd]

An offset of 2ns is suspiciously close to 1 / 500MS/s, which I believe is the sample rate with two channels enabled (since 1GS/s is the sample rate with 1 channel enabled).

Even with a slower timebase the effect is more or less the same. The shift in waveforms is about 10ns for channel 2 without and with channel 1

[kvd]

This is a guess, but i suspect it might be a trigger hysteresis thing.

The trigger point of a scope when it's doing an edge trigger is actually more of a trigger range - the signal has to hit two comparators in the right order for the trigger to happen. As a result, the signal can sometimes show up slightly shifted on the screen. Then, scope makers shift the waveform back in software so that it doesn't look funny. You can check if your scope does this by switching from a rising edge trigger to a rising-or-falling edge trigger. The alternating edges make it where the scope can't get away with shifting the waveforms, so it'll appear that the signal crosses the trigger point slightly shifted.

I suspect that for this scope, when channel 1 is not displayed but the scope is triggering on that channel, there's no shift. But, when you turn on channel 1 the hysteresis adjustment kicks in. I'd bet that the timing relationships between the two signals isn't actually changing.

Just a guess, as it's not our scope, but this would be what I would look into. You could try putting the scope into rising-or-falling edge trigger mode and see if you get the shift.

Results are the same with triggering on rising-or-falling edge

[kvd]

I did some more tests:

the effect does not occur with channel 1, and either displaying channel 2 or not displaying channel 2 

[nctnico]

I'd say it is a rounding error somewhere.

[StillTrying]

Have you tried all 3 probes on the same signal.

Tried finding the EXT/5 signal's center position by adjusting the trigger level up and down.

Some of the displayed X trigger positions don't make much sense to me, on the last one SDS00019.png (and others) the X trigger marker is nowhere near the CH1/EXTtrig signal's rising edge.