Hey Guys, I now made several fingerjoints, the problem is: The corners are not straight, they are always a bit above 90° so that they don’t match exactly. You can see some pictures here: MPCNC Primo Schneewittchen - Oldenburg, Germany - #21 by Tokoloshe
I know that the bindings are not sanded down, but in the top view you can see that there is a lot of air, like half a mm above and below the finger. Could it be that the machine is too fast so it has got too much momentum? I cut the pieces at 1200mm/min with 2.2mm depth.
The wood being thicker than the pockets are deep is not a problem either (it is a problem, but I know why. The wood was sold as 6mm thick, it is factually 6.4mm thick, wich I didn’t know when making the model.).
Any tips are appreciated, I love fingerjoints and want them to be beautiful. 
(I know, there will always be the “round” corners, even with “Freifräsen” in Estlcam.
Is your z axis perpendicular to the xy plane?
The cuts are straight in the Z direction.
I made a quick and dirty sketch. It looks sth. like the one on the right from above (not as much deviation, to be honest, but you can still see it slightly), it should look like the left one:
Looks like something a finishing pass could help with, if you’re not already.
Argh, my picture is not good. the ones that are not straight are smaller than the others, not bigger. I drew it wrong. I am sorry for the confusion. So no, a finishing pass would not help. I’ll try it a little slower tomorrow or in the next few days and will tell you how that went.
If you switch between conventional/climb cuts, it might cut outside instead of inside. But a finishing pass may help anyway, because the first cut is farther away.
Thought the same - in static mode, it will probably be. I saw such deviations when my old steel/PU belts approached their lifetime limits and at the same time I found out that my burly core was anything but solid. Plus some play in the bearings settings. Primo parts are ~ 70% printed now.
In other words: How does your (Philipp) router mount react to an enhanced earthquake/meteor impact test ™?