First night under the stars.
I started out with the robofocus on the ttys0-port and the EQMOD and ST8 on USB0 and USB1. It seemed to work fine for 2 hours, but suddenly I received the robofocus timeouts again while I was playing with the autofocusing process. The SBIG camera also went out. Changing the robofocus on an serial-to-usb converter and rebooting the whole system did not reproduce this error again. I remember last time the SBIG camera also giving weird errors, so it's likely this robofocus-problem must have something to do with a conflict with the camera, or the communication between the two devices
I tried the autofocus-process a few times between the L-filter and a photometric V-filter. The L-filter focuses around 9600 ticks. The V-filter focuses around 9700 ticks. The RGB filters are parfocal with the L-filter but I rarely use them. When I'm in the neighborhood of the focus, the autofocus works perfectly. He does a few in- and out-'s and then finds the best focus. (Tolerance: 1% and steps=5 ticks). This gave the best focus for me. However, when I then change to the other filter, the system cannot recognize the out-of-focus donuts as stars. Stalling the focus process and hence the whole routine.
This said, because I live in an urban environment with heavy lightpollution, I do mainly timeseries of variable stars.
Under windows and CCD-commander, I used the L-filter, 20s exposure at 3x3bin to get as many stars as possible on the image for doing platesolving. I also focused using the L-filter. Then I did the timeseries on the V-flter on 2x2 bin. CCDcommander automatically refocusses a certain amount of ticks when changing the filter. It would be nice if Ekos could do the same. For timeseries of faint variable stars, I still use the L-filter. A pragmatical solution for using Ekos is thus to only use the V-filter in my workflow.
I found that I cannot save the 3x3 preference for doing platesolving in Ekos. However, when I played a bit with it last night, he found every time a solution with the V-filter at 1x1. But experience tells me this is not enough in areas with not much of stars and under hazy sky.
I normally don't use the guiding-option, my sub exposures are only 30s long and experience with CCDcommander shows me that using the automatic calibration of the autoguiding, it fails more than it works. I however tried it with Kstars, but the application suddenly crashed completely. It's not that important for me, since I don't use autoguiding.
At the end of the night, I could with these settings successfully do an automatic and robotic sequence of images.
The only thing I remarked is that when the mount is parked, and when I click "unpark" in the startup sequence, he doesn't unpark. There is still a bug.
However, I had to play a bit with the settings and return to the scope for a few times. I got a few times a hard crash of Kstars, god knows why. But the most important things is that I have together my workflow and it seems to work if you don't ask too much and don't click too much around. I'll test further the next opportunities under the sky.