DS1102E unexpected behaviour: changing vertical position distorts waveform

[Xevel]

Hi,
I have a RIGOL DS1102E that has served me reasonably well for a year or so. Today, while visualizing a simple waveform (PWM output from a microcontroller through a series of resistors, 3.9KHz, ratio between 30 and 60%, low level of around 800mV and a high level of of around 3.4V), I ran across something I would not expect: changing the vertical position of the waveform on the screen changes the waveform when I'm in 200mV/DIV resolution or lower. The waveform distorts and shifts, so much that when the vertical position is at its upper limit, +2.0V, the signal appears with a lowest point of -1.7V (which is simply impossible). The plateaus also show a slope that depends on the vertical position (much more pronounced when moving the position up than when moving it down...)

Higher resolutions (500mV/DIV and higher) do not show the unexpected behavior, and moving the position does not affect the waveform, it stays a nice square between its expected value. Just like I would expect from a scope.

As a control, I tested again with the calibration test signal of the scope, with the same behaviour on all ranges (works OK on 500mV and more, not on 200mV and below).


DC coupled, Y-T, triggered auto, original probe 1x, CH1, nothing on CH2 and it's turned off. No math operation, no BW limit, no digital filter, no cursors, no measures, nothing fancy at all. Ground lead is well connected to GND.

Firmware version 00.04.00

Any idea what is happening? Is there an option somewhere that I could have activated that would make the scope act like this? Is it possible the scope is damaged?
Is it normal and have I accumulated too many all-nighters to understand what is going on (I'm doubting my own sanity at this point)?
Thanks

[Xevel]

It just occured to test on CH2. The result is the same.

I noticed that there is a relay clicking when changing from the higher ranges (OK) to the lower ranges (Not OK)... a damaged component on that sub-system maybe?

EDIT : I also tested with AC coupling, the result is the same.

[skipjackrc4]

Does the scope have an internal calibration?  That would be worth running.

[Xevel]

There is indeed a self-cal option.
I just ran it (the scope should be ON for more than 30min, which it has been, and I disconnected everything from the input as indicated) and the result is the same, no improvement.

[han]

any picture?

[Xevel]

https://www.dropbox.com/sh/rsrcsjqozn3hjrk/AAAVh3L4Y1e8itSuR1Y2I4gqa

The probe is on the test signal used to adjust probe compensation. Here it's a square with 3V Vpp, from 0V to 3V.

There is one pic at 500mV/DIV, that shows the signal as expected, then the rest of the pics are at 200mV/DIV with various vertical positions, from -880mV to +1000mV (cf file name).

EDIT : DC coupling. With AC coupling we can have a look at the top of the "square" wave, and the distortion is different, more like an overcompensated probe. The amplitude of the distortion is also variable depending on the vertical position.

[jahonen]

You are overdriving the scope input amplifier. Pretty much every scope will exhibit the same behavior when input is overdriven enough. So what you see is perfectly normal. Waveform should be observed when trying to overdrive, and volts/div adjusted only as far as the waveform remains undistorted.

Regards,
Janne

[Xevel]

Well, now that you put a name on it, it's much easier to get info on it, thanks a lot
Huge thanks to all of you. I still have much to learn on  my test equipment it seems.