Hi Stephan,
Does it not even come within 50 µm of correct focus?
If it can't even do that, you might want to check that the grub screws on the flexible coupler is as tight as humanly possible :-). The EAF shaft has an indentation, so it is not hard to align one of the grub screw to the flat part of the EAF shaft and then tightening the second will secure it solidly. I don't know if the shaft on tour OTA focuser has a flat indentation; mine is shiny stainless steel and I had to really tighten the grub screws on that end. You can also check by putting a mark on the fine focus knob and see if it always return back to the same spot after moving back and forth by a few hundred EAF steps.
With my 450 mm focal length Petzval (an FSQ-85 with a 1.01x flattener). I have set my Auto Focus Settings step size to 20. And ASIAIR usually gets within one or two steps of correct focus. Much narrower than that, and ASIAIR will be trying to focus on noise.
It is just that I am more critical than the ZWO people. My focus scale is 4.2µm per EAF step. With the price you pay to get a Takahashi spot size, you might as well try to focus as well as possible. So, for critical stuff, I do my own V curve, offline, and then direct the EAF to go to what I compute as minimum.
I don't know how much you consider something to be out of focus too (different people have different tolerances).
I took this last night when trying to experiment with a stack of an OSC narrow band filter (a Radian Quad) with a cheap LRGB filter to select between the narrow bands on a monochrome camera (ASI2600MM). Since I was not seriously doing any imaging (I was just checking for the halos around a bright star, in this case of Xi Cyg) I just let ASIAIR do the auto-focusing last night.
So ASIAIR is at least capable of this degree of focus (single 180s exposure at gain 100, with a stack of a cheap Antlia L filter and the Radian Quad):
http://www.w7ay.net/site/Images/XiCyg.JPG
It is probably 10 µm off true focus. The image covers a full APS-C frame.
Chen