alaluji I noticed that when doing AF every hour, it showed different final focus values.
Is this so or is there something I should change in some parameter?
Not much you can do about it. There is a bug.
The normal way to autofocus is to use a V-curve (hyperbola curve fit). ZWO uses a U-curve (parabola curve fit). But that is not the primary bug, though.
The good thing about the V and U curves is that they are made of multiple data points (different EAF values), and because of that, the curves are quite accurate even though each point's measurement is noisy. The minima of the curve is the focus point.
However, what ZWO does is to run a second pass (this is the bug). On the second pass, it is using individual star sizes to discover the minima, instead of using the curves' minima.
As a result, once the focus reaches the diffraction zone (i.e., inside the critical focus zone -- the CFZ), where the curve flattens out, it is very prone to star size estimation error. It will stop looking once it finds the start size value starts to increase. Because of this, it will stop early in the CFZ, instead of going to the curve minima.
This noisy reading also causes the autofocus function to stop searching for a minima at different times -- what you are seeing. You can see this "early" abort by looking at the red dot relative to the U curve. You will see that almost all of the time, the red dot stops before the minima of the curve.
This is not always a problem per se, since you are often placed inside the CFZ already. However, when temperature causes the OTA length to change, you can be knocked out of the CFZ almost immediately.
ASIAIR's autofocus will achieve focus for the casual user, but not for someone why wants a tack sharp image of the stars.
As a result of this, I use my own V-curve program (I wrote it for macOS), or a Bahtinov mask when it is not too cold outside.
Chen
