This is related to my earlier post indilib.org/forum/ekos/7010-ekos-enters-...g.html?start=0#54165

The problem reoccurred in a more serious form with a different mount and a completely different installation on a different computer. This is now with my Atlas EQMOD mount and the most recent Kstars nightly installation from yesterday.

www.dropbox.com/sh/z7gtg2gx0xqwd7c/AAC7j...uppy9aH3Y2zP8Ta?dl=0

The videos show two problems with the meridian flip: FIrst, it wants to start before the meridian is passed (first video) and then in the second video it fails to park again and goes into a new endless loop of flip, realignment, capturing... Alignment using the scheduler also failed repeatedly (reminiscent of the problem Magnus had a week ago).

The log is also included in there, the problem starts pretty much at the beginning. The latter parts of the log record my imaging of M51 after manually slewing to it after it had passed the meridian and just running a regular capture sequence, without using the scheduler. Ignore that, that seems to have worked fine.

I think I found and fixed the issue, can you git pull and try again?

Pulled and compiling right now, but it may be a few days until I can test in the field, as the weather is changing again and it will be cloudy for the next few nights.

Best

Jo

PS: Compiled and ran first test on the simulator. So far so good. Flip completed OK in the simulator.

Unfortunately, that did not fix the problem. I have not been able perform a productive run of the scheduler on M51. The mount will home in onto the correct position, solve, refine the position, but once the final position has been reached, a flip command is issued, the mount changes pierside and everything starts again. That is with the git pulled source build after Jasem's changes (see 2 posts up). No more problems after parking, though.

Log is attached:

I then purged Kstars, reverted back to the stable Kstars (April 25, pulled from repository) and that worked fine. The scheduler performed as expected and the imaging run was uneventful and completed normally. Also no problems after parking. Dropbox link to this more extensive and IMO normal and uneventful log file here ( www.dropbox.com/s/ma658dbmmjdj084/log_22-06-59.txt?dl=0 ).

Looking at your local time of 21:43 I would expect a meridian flip to be possible for M51 at about that time.

However the mount reports that the pier side is already East, looking West. So I don't see why one was needed and indeed the mount reports that the flip failed because the pier side has not changed. It started East and finished East.

Looking at the HA it is -0.443 so it looks as if the mount did the flip early. Is this expected?

I can't see the mount, where was the OTA? On the East or West side of the pier?

Have you set the mount so it does the flip early?

That looks like the problem, I set the simulator to flip an hour early and it immediately tries to flip, but the flip fails because it's already flipped. It should do nothing.

The mount was set to flip if HA > 1.5 degrees, so approximately 6 min after passing the meridian. That would have been around 2215 h local time.
I was standing next to it, watching it flip back and forth several times until I finally stopped it, purged Kstars and reinstalled the stable version from April 25. After that it performed flawlessly. That is all contained in the longer log in the Dropbox, which I think is largely uninformative since behavior was normal.
M51 was passing the meridian around 2210 h local time, so about half an hour later than the time point you are mentioning. Plenty of time to align on the Western pier, but it was moving back and forth every time immediately after alignment had finished. In either direction.

How do you set it to flip an hour early? I only see the option to flip if HA > x degrees or y hours. I see no option to set the flip at HA < hourly angle.

Please have a look a the videos I provided the link for in my first post of this thread, the first one illustrates the same problem that is persisting.

BTW, I had tested the same installation in the simulator previously and it did NOT show that problem. I let it run through a meridian flip with the same settings and that completed normally. So I am not sure why it would behave differently when connected to a real mount vs a simulated one.

But it happened with 2 different mounts (iOptron and EQMOD) on two different computers (Pi4 and Zotac miniPC, respectively). And when going back to the older stable installation the problem is gone (which is probably also the reason why nobody else is reporting this).

I've pushed what should fix this. I was able to reproduce the problem by setting the simulator to flip early and when that happened the problem described happened, the time to flip was reported as -11 hrs or so, the pier side as East and the hour angle as a small negative value.

I've allowed for the early flips, the pier side is still East, the Hour angle negative but the time to flip is now over 12 hrs indicating that the next flip will be in 12 hours which is what is expected. As a result there is no flip attempt.

The code is in the master but will need an overnight build before it will be in the bleeding edge build.

Interestingly this problem and the other problem have different root causes, this one an early pier side change and the other one a failed park and an incorrect pier side report from the mount. The symptoms look similar but that doesn't mean that the same fix is needed.

I still need the slew tests with logs I described in the other thread.

BTW, with a round world when is overnight in the context of builds?

Which Kstars build do you need for the slew logs? You want the logs for the iOptron driver, correct? I am using the CEM60 driver, I assume that is the same as the ieqpro you are requesting. I need to check that when fire up that mount.

Please explain one thing for me, please: How do you set the mount to flip an hour early? Can this be done inadvertently and could I have introduced that problem unknowingly? Where is that setting?
It makes no sense running elaborate tests if the problem is that a simple setting in Ekos is wrong. I want to check that that is not the case.

Thanks, Jo

The simulator has a flip position setting in the simulator tab of the simulator driver. It's normally the last entry on the simulator tab.
For a real driver there could be many ways. Some mounts can be set the flip early - or late. A mount that started with an hour angle error which was corrected with a sync could cause this, so coud a time zone or DST setting error. Having a different time in the mount and Ekos could have an effect.The 'elaborate' tests have nothing to do with this second problem - and bear in mind that the 30 minutes or so it will take you to run these test is far less than the time it will take me to analyse the results and work out how to implement a solution.

As I said the two problems have diferent causes and need different solutions.

I just want to make sure that I understand the parameters to the extent my meager cognitive abilities allow.

But thanks for explaining. DST is definitely a possibility, since I did not observe this over winter. I will hint for that setting.

It may affect my iOptron mount, since there I must route the connection to the Pi4 through the handset. And I definitely did NOT change the DST setting there recently.

Thanks for that important pointer!

Jo

PS: will try to get you the logs ASAP, but may be weekend as in Wouter’s case.