A few days ago, I thought that I parked my scope at the end of the night, but apparently it was still tracking. When I came home that evening, the scope was on the side of the pier - pushing into it!!! Sh### !!! Considering that the scope was upright when I left in the morning, it was probably on that pier for a couple of hours :-(
On the next night of imaging I noticed that suddenly both axis seemed to have backlash: it took PHD2 often several corrections until the scope finally moved. In DEC, it happened whenever the direction changed (afterwards, the corrections worked immediately as the gears were now connected). In RA it mostly happened in one direction (probably when PHD tried to move the scope against the siderail tracking) - it worked in the other direction (the gears were already connected from tracking).
I did the most obvious thing: remesh the gears. I was very careful with the RA gear - wanted to make sure that I didn't switch the gears as I would have needed to redo my PEC. But nothing changed. The next day, I did it again, this time tightening the mesh screws more then normal - still the same effect. I thought that the gears were mainly too tight and loosened them again - no change.
The good news is that the mount tracks REALLY well (especially with PEC on) and that I'm currently using the FLI ML16070 camera which has pretty large pixels resulting in a 1.6 pixel / arcsec scale! So, in most cases, the images are still OK.
I posted on the ap-gto mailing list about this.