So few days ago, I decided to check the performance of the Ekos Autofocus algorithm, and so I pointed the scope at Kochab and started the autofocus routing with the CCD (QSIwsg) binned 4x4. After a couple of minutes, the autofocus completed, and I then attached the Bahtinov mask, and captured the following:



So it appers the autofocus routine nailed the optimal focus pretty closely as you can see from the diffraction spikes.

Here is also the "V-Curve" for the autofocus run:



Now there is still a lot to improve on autofocusing routine as it can get dead locked on occasions. This was done with RoboFocus which is one of the few devices that "just works" under Linux.