DSO112A displays -100mV DC with shorted input

[gf1]

Yesterday my DSO112A arrived and I'm quite surprised by the capabilities of this little gadget

I noticed a strange behaviour, though, which I cannot explain.

• DSO is running on battery power (in order to rule out any influence from USB power supply) and rests on a table, test cable is not yet connected to the input
• Calibrate offset again (just to be sure)
• Timebase does not matter - it happens with slower (say 10ms/div) and faster timebases (10µs/div)
• Switch to 200mv/div or 100mV/div, AC coupled (AC vs DC coupling does not make a difference, though)
• Everything fine, the DSO shows a horizontal trace at 0V with a little bit of noise
• Connect the MCX-clip cable to the input and let the cable hang down.
• Still everything fine, the cables pick up several millivolts of AC hum and noise from the environment, but the average is still at 0V
• Now I short the two alligator clips of the MCX-clip cable (and still let the cable hang down)
• The noise almost disappears, however the horizontal trace moves down to about -100mV DC

I have no clue where this DC reading comes from, since there is no DC applied to the input (but the input is shorted). The loop at the end of the cable may still pick up some noise of course, but this would be AC, not DC. And even if DC were applied to the input, the AC-coupling would block it (AC coupling has proven to work in other scenarios I have tested).

Interestingly, if I pick-up the DSO and hold hold it in my hand, and touch the alligator clips with my other hand, then the unexpected DC offset disappears and the trace moves again to 0V. The DC offset also disappears if I don't short the alligator clips directly, but connect them to a low-Ohm resistor instead.

EDIT: Touching only the shorted alligator clips alone, without touching the case of the DSO as well, does not let the unexpected offset disappear.

Has anybody an idea what's going on here?

Thanks,
gf1

[gf1]

Added some photos.

[jye1]

This is strange. I can't find explanation yet.

[gf1]

Zooming in vertically to 20mV/div (and moving up the trace) reveals also a small periodic AC waveform added to the -100mV DC offset, which does not look like noise:

20mV/div and 1us/div reveal also an AC component

Consequently I tried to measure this signal at the AOUT test point (ADC input) with a faster 1GSps scope, and got indeed a periodic 48.9 MHz signal (so the AC waveform on the DSO112A display is obviously an alias of the real signal, since the ADC clock seems to be only 5MHz):

Signal on AOUT test point (ADC input)

So I guess that an oscillation in the analog front end causes this strange effect.