Waveform shift small vertical scale, large amplitude

[Mike Fikes]

I'm trying to gain an understanding of the Rigol DS1054Z's behavior where the waveform unexpectedly shifts when at a small vertical scale and the amplitude is larger than a certain amount.

In concrete terms, I am feeding the scope a square wave where the top is at 0 V, but the bottom varies from around -1 V to -3 V.

If I have the vertical scale set to 500 mV per division or more the top of the waveform stays around 0 V as expected. But if I go to 200 mV per division or below, if I then increase the amplitude beyond about 2 V, the top of the waveform starts to shift down.

I'd just like an understanding of what is going on inside this scope that might be causing this. Am I saturating the measurement range at smaller scales, causing the scope to hit its upper (and lower) ADC ranges, leaving it with no choice but to shift the waveform to display what it can measure?

To help explain what I'm seeing here's a 1-minute demonstration https://youtu.be/IHBjYxnT3KY

[ledtester]

I would trust what the scope is telling you.

Whatever you are doing to your signal generator it is no longer producing a square wave with a top output voltage of 0 volts.

What signal generator are you using? What parameter(s) of the signal are you changing?

Just to make sure... CH1 is DC coupled, right?

[Mike Fikes]

The signal generator is a Rigol DG1022Z. I'm varying the bottom voltage of the square wave. (On that signal generator this is the "LoLevel" parameter for the square wave.)

Yes, CH1 on the Rigol is DC coupled.

FWIW, I'm unable to reproduce this when the same signal is fed into a Tek TDS 5104. The Tek maintains the top of the signal displayed at 0 V, even at vertical scales of 200 mV per division and much lower...

[ledtester]

It's possible that it's a firmware issue with the DS1054z. I don't have a DG1022, but I can try to replicate it with a different signal gen.

[magic]

If I have the vertical scale set to 500 mV per division or more the top of the waveform stays around 0 V as expected. But if I go to 200 mV per division or below

Obviously something to do with the scope, not the generator. The generator doesn't know what the vertical scale is.

It might be some weird problem with the input amplifier too. Perhaps the -3V peak causes it to saturate to the negative rail at sufficiently high gains and then it goes crazy? What if you reduce frequency to 100Hz?

[Mike Fikes]

Yes, slowing things down to 100 Hz provides more insight into the behavior. I'm attaching screenshots for amplitudes of 500 mv, 1000 mv, 1500 mv, 2000 mv, and 2500 mv. (I forgot to add the 50 ohm terminator—you still get the same curve sequence just at a factor of 2 difference if you add it.)

Also interestingly, if you try other frequencies (high enough to make things look "square"), you can easily get the top level rising significantly before it falls.

This particular scope has 00.04.04.SP4 installed. If you have access to one of these scopes, it's interesting to play with this to see what it does.