There appears to be a time and location conflict in the indi_lx200_10micron driver when the mount is updated from its own GPS (MGPBox), resulting in a range of bad behavior including failed meridian flips, bad pointing, and horrific tracking when the mount and INDI and KStars fight over precise geographic coordinates and time.
I build a mount model in separate software (ModelWizzard 4) that leads to incredibly precise pointing without KStars and 2-axis tracking using built-in encoders. But when doing production imaging in KStars I get horrible arcMINUTE-scale star trails as the mount and software fight over the time & location (see example images attached below).
The time mismatch between GPS/10Micron and Kstars frequently leads to
failed meridian flips
as the mount reaches a safety stop before KStars realizes it's time to flip.
Another symptom is that the reported geographic site position in the INDI driver doesn't retain arcseconds in geographic coordinates (for consistency with LX200?) but the GPS-reported coordinates in the mount keypad does. This was
reported by fenriques in this thread.
It may be that the arcseconds numbers are just not being reported in the GUI, but the rest of the symptoms suggest that the actual geographic coordinates are being *truncated* in the driver, resulting in the DEC thrashing I see in my subs (unless I just ignore the mount coordinates).
As Andrew points out in this thread
, I can get pretty good results by just ignoring all the time & location info in the mount and GPS. To do this I have to turn off the MGPBox syncing in the mount hand controller set the geographic coordinates by hand, and set "KStars updates mount" in the Ekos settings. But I get nearly PERFECT results in TheSkyX by just trusting the mount to do its own thing and not trying to second-guess it in software.
This is a tough issue to diagnose because it involves the mount computer as well as the INDI driver and KStars/Ekos. I can save logs but not sure which logs are involved. Maybe I need to collect a mount log from the 10micron computer and report it to the vendor? But pretty sure this is cross-talk between INDI and 10Micron so hard to know what info to provide to whom.
Attached are two unguided 180-second subs of M74. Polar alignment is identical in these two unguided subs (well under 1 arc minute PA error). The first (with coordinates overlaid) shows INDI and the mount fighting over the time and location. The second shows good 2-axis tracking with all GPS info disabled. It gets even better (eccentricity under 0.3) when I use other software like TheSkyX that just leaves the mount alone to do its own thing. Everything I report here is for KStars 3.5.0 (regular ppa, not nightly) on ubuntu 20.04.1 running on R Pi 4.
This post is following up on three other related threads which I suspect are due to the same issue:
indilib.org/forum/general/8328-3-5-0-flip-failure-10micron.html
indilib.org/forum/general/7928-why-is-th...y-mount.html?start=0
indilib.org/forum/general/7928-why-is-th...y-mount.html?start=0