Before describing my problem let says that I have corrected 2 days ago the polar alignment and have now 14" polar alignment error on both axis.
New test last this night.
I selected as target Theta Aquilae close to the celestial equator to get a better pulse movement in RA.
Unfortunately, there is no more improvement. With 5000 as pulse the calibration process moves just a fraction of pixels in RA (on display) and/or does some erratic moves forward/backward in both X and Y. The spot doesn't come back to the origin before starting DEC process.
In autoguiding session the RMS error on RA is between 0,25" to 0,5". Two months ago, calibration process worked differently, my RMS error on RA was between 0,10" to 0,25".
I am not sure that the guiding correction is working while autoguiding.
I tested PHD2 using INDI server for mount and camera autoguider. The calibration always failed on "not enough move on East/West axle".
I am not sure this is a INDI problem, but the only change in 2 months is an update of INDI/Ekos/Kstars and my setup didn't change.
As the movements of the mount on both axles and tracking looks fine I don't understand what the problem is.
Another issue is that the mount parking position is lost between 2 sessions. But this is another problem.