Hi Attila, thank you for the reply.
The connection is very stupid:
1+ (orange) with W1 (yellow) and 2+ (blue)
1- (white-orange) with W1-GND (black) and 2- (white-blue).
In the past it was working fine, just a few days ago I noticed this strange behavior... With the same connection, and setting W1 for 1kHz 1Vpp, I clearly see CH2 is lower than CH1, while they are the same at 100kHz.
I cannot explain this behavior at HW level... What could be damaged this way?
I only worked with max 5V signals, therefore I would exclude any AD2 abuse, unless sever ESD occurred (I have grounded mat and wrist...).
BTW I noticed that if I simply swap the terminals of WaveGen1 (W1 to 1- and 2-, its GND to 1+ and 2+), the curves are as they should be... Verified with scope too.
Walter