Yep, had the same issue with one of mine, just started doing it when I got it out of the drawer one day. Annoyingly it's the one that I did some hacking on (shortened the leads to get rid of some peaking) so no chance of a RMA, even if it were within warranty.
I did a bit of a bodge to improve it by figuring out how the offset circuit worked and padding one of the reference resistors. Can't remember much more detail than that but very annoying to have to do this - given others have the same problem would no longer recommend these even given the cheap price.
Interesting ! Did you find that the micro added in the anti-offset into that 3rd and final op-amp ?
I was trying to add on to the Dave-Cad schematic but the one that I was just working on appears to be slightly different. For instance, the main differential amplifiers are in a single chip. Then the 3rd op-amp appears to be a dual OPMI labeled TI amp. AND, I cannot seem to trace the 50 Ohm output to that 3rd op-amp. Not in the short amount of time I spent on it anyway.
Answer to the question, how much offset... It varies. On this latest probe, changing the 500X to 50X does not appear to change the absolute output offset. The probe is still useful though. Just have to offset the scope and try to remember where zero is if I am measuring DC.
It also can be worse and maybe even better at different times when letting it sit. But it is always enough to be annoying.
I think that it is offset enough where the processor cannot quite put it right again.
Dave, is your HVP70 probe the one with 100X that you showed in your video ? That is a nice compromise. Hopefully they have gotten the CMR fixed as promised.
boB