Thanks for pointing out that this is probably an unreasonable expectation of performance.
I will experiment with the DS1054Z further and make sure it doesn't glitch with most of the signal on-screen.
So the rapid increase in error at the 500mV and 750mV vertical-scale transitions is probably the scope increasing the input amplifier gain and increasing the effective overdrive.
And similarly moving the signal further off the screen at low vertical-scale is increasing the overdrive.
The DS1054Z seems to have rather substantial overdrive recovery time once it starts ringing.
Ringing seems to last around 15-milliseconds or until the (real) signal level goes back down.
So that could explain why it seems worse at lower frequency.
Sweeping the frequency at constant-voltage shows the ring stays the same, but is truncated.
Decreasing voltage at constant-frequency decreases ring amplitude until it suddenly vanishes, but ring time-period remains basically constant.
With a 1V (±0.5V) square-wave (substantially off-screen) toggling from 76mV/div to 75mV/div, the error is ~75mV when the signal is positioned at the top of the screen and ~300mV when the signal is positioned near the bottom (signal more off-screen) so it's just still on screen after it jumps.
This is in 1x-mode, so 750mV transition equivalent in 10x-mode.
But no ringing or other signal shape change, just attenuation.
And then it actually has an increase in signal level at the 51mV/div to 50mV/div scale transition.
Only a few mV of unexpected change with a 500mV square-wave entirely on-screen.
Keeping on-screen seems like it might also help with the unexpected level jumping I was experiencing.
So overdrive on the DS1054Z also seems to cause a voltage-scale error, even without any waveform shape change.
I tried to trigger this sort of glitch on the TDS-210 again, this time trying to maximize overdrive and it seemed to have no perceptible glitches at all. It seems practically immune from this sort of issue.
But maybe it is the unusual one and not the DS1054Z?
After experimenting with the TDS-210 a bunch more I think I have finally induced it to overdrive glitch.
With a -5V to +2V 100KHz square wave I can get it to attenuate and turn slightly sawtooth.
It has a transition at 200mV/div (1x-mode) where the waveform suddenly becomes attenuated and changes shape.
But still no ringing at all.
It shows a change in rounding of the rising edge with no attenuation toggling from 204mV/div to 200mV/div at 10Hz, 100Hz, 1kHz, 10kHz.
Then the same rounding edge runs out of pulse length and starts attenuating at 100kHz and makes the (slighly curved out) sawtooth top.
Maybe it's switching in some sort of filter to prevent ringing and that rounds the edge?
Changing the vertical-offset on the TDS-210 never seems to have any effect on waveform shape.
EDIT:
With different waveform settings, I did get the TDS-210 to ramp the square-wave top slope when adjusting the vertical-offset.
Kinda like the DS1054Z, but more changing slope rather than changing curvature.
So changing the vertical-offset on the TDS-210 _can_ change the waveform shape in overdrive.
With these settings and ramping the frequency I can see that there is a flat angled top separate from the rising-edge curve.
So the TDS-210 definitely does have overdrive issues, it's just a bit more subtle about it with the lack of ringing than the DS1054Z.With the keywords posted I found another thread:
https://www.eevblog.com/forum/testgear/ds2072-and-sds8102-hw-fail-sw-bug-or-what-the-hell/Quote from thread:
The first thing I learned from the engineer who taught my how to use an analog scope was:
"If the signal is not fully contained within the graticule, you can't trust that it's being displayed correctly."
The same still holds true for DSOs - and the input errors can possibly affect the measurements. Some scopes handle it better than others - doesn't matter about the price of the scope.
Sort of related:
https://www.eevblog.com/forum/reviews/rigol-ds4014-strange-waveform/They also mention a TDS-210 not being affected, but they fixed it on the rigol by changing probes.
What is your signal source? I would start looking at impedance mismatch. Your frequencies are kind of low, but who knows.
The internal 1-kHz 3V square wave and a 50-ohm function generator both with and without a 50-ohm terminator connected (didn't seem to make much difference, just cut amplitude about in half).
But then I am using a scope probe and not just coax.
I have a DS1074Z-S on which I can also create nasties in the signal by overdriving it, so as an answer to the OP, I don't think your scope is defective.
It is good to know this is probably expected behavior.
Thanks for the thread link:
https://www.eevblog.com/forum/testgear/ds1074z-zoom-bug/Overcoming Overdrive Recovery on High-Speed Digital Storage Oscilloscopes
http://cp.literature.agilent.com/litweb/pdf/5989-0068EN.pdfComponent and Measurement Advances Ensure 16-Bit DAC Settling Time
by Jim Williams
http://cds.linear.com/docs/en/application-note/an74f.pdfEvaluating Oscilloscope Overdrive Performance
Why do most oscilloscopes have so much trouble recovering from overdrive? The answer to this question requires some study of the three basic oscilloscope types' vertical paths.