Hi All,

Sorry for the lengthy post but I couldn't resist. I was so impressed with Kstars and Ekos I had to write about it

First of all many thanks for the great job you’ve done with Kstars and Ekos. I didn’t expect it to work so well. I managed to get all my devices working during the first session and almost got to serious imaging. There were some small problems on the way though.

Kstars crashes a lot when zooming and pointing map to found targets. However it might be “fixed” by Jasem here:
lists.kde.org/?l=kstars-devel&m=138807351307331&w=2
Just tried that and similar fix applied also to asteroidscomponent.cpp solves the zoom problem.

Ekos is a modal window. On my small screen it almost completely covers the kstars map. When I try to minimize the ekos window, kstars also gets minimized, so I can’t use the map. That is pretty annoying

It happened a few times that I closed INDI configuration panel and I wasn’t able to reopen it from Ekos. I’m not really sure when it happens. Also when closing Ekos and Kstars, indiserver is not stopped automatically. One needs to remember about that or subsequent attempt to run INDI in ekos will fail (to connect to devices).

Guiding didn’t work for me. I wasn’t able to go through calibration. The error message I got was that the mount didn’t go back to the initial position in DEC axis. I remember I had pretty big backlash in DEC axis. PHD was able to cope with that. That is a pity that ekos can’t (so I have to finally fix it). Anyway, there is one question I didn’t find answer for when reading the: do I need to align the guider with the mount? I mean that the edges of the image are aligned with star motion when moving mount in RA/DEC.

After attempting to guide and calibrate the guider, EQMOD stopped working. I wasn’t able to slew the scope to new target. Restarting INDI usually helped, but it happened once that neither restarting INDI nor restarting kstars helped. I even tried disconnecting EQDIR. Only laptop restart helped resolving the problem.

Plate solving worked well. However there is one annoying feature. Every time I start EQMOD I have to fill in the scope and guider aperture and focal length. I wasn’t able to save these values between restarts.

Autofocus didn’t work at first but the camera was totally out of focus. After coarse manual focus I used auto-focus and it worked pretty well (didn’t try to cross check with mask though).

CCD (capture) module wasn’t used a lot. I only took preview pictures because of the missing guiding. There was one annoying bit though. I had 2 cameras defined: main and guide. Every time I started ekos, guide cam appeared in the CCD module and Align module. This was tricky as plate solving didn’t work at first. Only after a moment I realized that I was using guide cam instead of main cam for Align.

To sum up, I was very happy to go that far during first full session. I wasn't able to image under Linux before because of lack of Atik drivers. Now the drivers are there and kstars with ekos seem to be pretty good for astrophotography. There are some problems, but not that major.

Cheers,
milosz

Thanks for the valuable feedback! Ekos can only improve with user feedback and yours is highly useful!

I applied the fix to asteroidscomponent.cpp now as well, so hopefully no more annoying crashes

Actually, Ekos is a modeless window, and I can minimize it without affecting KStars at all. A modal window is one where you have to close it in order to continue using the application. At any rate, I added an improvement and made it a top-level window, so it will appears now in the taskbar as well.

I couldn't reproduce this issue, INDI server is always shutdown when KStars is closed no matter what. Can you share some of the details of your OS/Desktop environment?

Well, Ekos expects RA/DEC to return in a certain number of steps. I increased the "turn back time" slightly, so it will try to go back to original position for a few more times before it gives up. But if you fix the DEC backlash, then it is even better!

I'm not sure if I understand your question correctly, but Ekos works with the guide camera oriented arbitrary with respect to the RA/DEC axis.

It would be great if you can enable file logging. Go to options tab, click enable debug, then select "To Log File", then check all the checkboxes next to "Logging". The log file will be verbose and useful to diagnose any issues.

You only need to fill those values once. Then go to options, and click Save Configuration. From KStars Ekos options, you can select when to load the configuration, select "On Connection" so they are loaded whenever you connect to EQMod.

Fixed. Now it always defaults to the primary CCD in Capture, Focus, and Alignment modules.

Thanks for the report again and hopefully you can manage to get Ekos to fits your requirements!

It appears in the task bar, but I'm not able to minimize ekos and leave the kstars maximized. Maybe this is the desktop problem. I'm using LXDE on Debian sid.

LXDE on Debian sid. I have 2 virtual desktops.

There is a slight chance to have clear skies tonight, so I will try increasing the number of iterations (docs site says it's possible). As for the second question, you understood correctly. Some guiding programs require camera to be oriented in line with the mount axes. Good to hear I don't have to worry about that.

I will try that. I have verbose output turned on and some logs showed up in the window. However after the guide attempt the logging was silent - no logs

I pressed 'save' and assumed the values will be restored on the next connection. I will check the settings.

It's a big relief to have option to use Linux for imaging. So far I was bound to Windows (XP) which misbehaved badly.

However on Windows I had one feature that I can't find in Ekos. It's implemented in astrotortilla. I couldn't get AT to work, so I took some code from there and crafted my own solution. I was using the image from the previous session to point the telescope in the exactly same location. Do you have a plan to have something like this available. The procedure is pretty simple and all pieces are already in place:
1. take fits from previous session
2. plate solve the picture and save the coordinates as reference
3. slew telescope to the reference
4. take picture after slew
5. plate solve and compare to the reference
6. go to 3 until the current position is within X arcsec (say 10) from reference (with my HEQ5 it's close to impossible to get exactly identical position)

Cheers,
milosz

Yup, I bet it is most likely a desktop problem.

Increasing the number of iterations _might_ help, but the default 5 is more than sufficient under most situations. I have HEQ5 and I use 3 iterations actually.

If logs stop, most likely cause is a driver crash. It gets restarted automatically, but you'd have to issue Disconnect/Connect again.

I'll update to docs to reflect Ekos options setting.

With EQMOd, there is no need for this as after you're done with your session, you can go to the align tab and click "Write File". this will save the alignment points already figured out from previous syncs (you need more than 3 for N-Star to work properly). The next time you start EQMod, it will load the alignment points and in principle, the GOTO should be accurate.

It seems I have to live with that, at least for now

It didn't help I played a bit with different ways of guiding. Neither guide port nor eqmod pulse guide worked. Increasing guide rate in eqmod helped a bit, but calibration still failed in DEC axis. I checked worm gear backlash and I can't see it. So most likely there is some backlash on spur gears. I'm not sure how to get rid of it without belt-mod.

BTW, what are your settings for guiding? Maybe I'm missing something obvious? Are you using eqmod or guide port on camera? Today guiding failed on PHD as well because of gale we had. But at least phd calibration passed (sort of).

I wasn't able to reproduce it again \o/ There is something weird with turning on sidereal tracking in eqmod. Pressing 'Track' button doesn't always work. Probably I'm missing some step. Today in the last attempt it just worked. I pressed 'Track' and than 'Sidereal' and it did the trick. I suspect my previous problems with guiding and slewing after guide were related to this. Today I was watching the logs and I got 'guiding not allowed without tracking' error.

I'm writing about different thing. I assume you refer to alignment. I don't have permanent setup and I do alignment every time I start the session (starting from deleting the points from last session). But I meant something different. When I image the same object for couple of nights I would like to have the same frame every time. The easiest way to do it is to take one of the frames from last session, plate solve it and slew to the solution. But this isn't always accurate, so it requires 2 - 3 iterations. With permanent setup 1st slew might be accurate enough. But still there is no solve-image-from-previous-session feature in Ekos at the moment.

Anyway, I'm slowly progressing

Cheers,
milosz

I am using the ST4 on Lodestar connected to the mount. In Ekos, I just select Lodestar as the one issuing guiding corrections. The only thing I change from default settings is pulse value, I make it 1500 ms otherwise I might get "drift too short" error. Can you notice visually that it cannot get back to its original position?

EDIT: I faced calibration failures under two circumstances:
1. Mount is low on power.
2. Mount is somewhat mechanically restricted (once a cable twisted and froze the DEC axis).

But in your condition, it might be worm/spur gears or more serious issues. Try now with EQMOD in sidereal tracking.

Actually, pressing Track will have no effect, it just sets the mount whenever _future_ coordinates are sent, then it is suppose to "track" them. You got it right, the sidereal button needs to be switched on if not it's already. I think on first power on and connection, it is set.

I implemented it today, you can check it from KStars GIT, or wait a day or two until the PPA packages are built. I only tested this under simulation though, it's raining today. There is a new button "Load & Slew" that does the magic.

I don't think the mount is damaged. I'm able to calibrate with PHD under Windows. So this is surely software problem, not entirely hardware. Today I managed to pass the calibration and the guiding was decent (I will post some pictures after I stack them). The trick was to increase the pulse width, guide speed (rate) and match it with exposure time. It seems that if the exposure is shorter than the pulse width, mount ignores some pulses and calibration fails.

I noticed problem with eqmod pulse guide. When guiding through the ST4 port, everything works fine. When guiding through eqmod, the directions are reversed. So instead of guiting back to initial position, Ekos is pushing the mount away from initial position.

I also tried PHD1 on Linux. There seem to be some bug either in PHD or in eqmod driver. The log was showing pulses of 0ms width. I suspected the same problem with ekos, but I wasn't able to reproduce it after switch to ST4 and back to eqmod pulse guide.

To sum up, there is surely some mechanical problem with the mount, but it's not something terrible. Tuning guide settings did the trick of passing calibration. The guide didn't work great though (around +/- 3arcsec). There was almost no wind and I noticed the guide sometimes passes too big corrections. Anyway, it sort of works now and just needs fine tuning.

When I powered the mount today, the 'sidereal' button was pressed and 'track' button wasn't. Pressing track didn't have any effect. Pressing 'sidereal' again (although it was pressed) turned the tracking on. Looks like a bug in eqmod.

Yay, great! I will try to build the HEAD version during the weekend.

Cheers,
milosz

Did you perform a meridian flip between the calibration and the guiding process ?.
I'm not an expert in autoguiding, but maybe it's due to the 'longer pulses' you set in the calibration?. 1500ms seems a bit large, but of course it depends on the mechanic performance of the mount. Are these errors afecting both RA & DEC or only one axis?

The not-slewing bug seems to be due to the locale settings as discussed here: www.indilib.org/forum/general/226-eq6-do...espond-properly.html . geehalel has found a possible solution.

Minaya

No I didn't. I know that it would be pretty natural to have this effect after the flip. I performed calibration on both sides of meridian and the effect was the same. It happens during calibration, not during guiding.

It was on both axes. The long pulse was due to the fact I wasn't able to pass through calibration without it. Anyway, fine tuning guiding is a minor problem comparing to other issues here.

Hmm, I think I don't suffer from it. My locale is en_GB. I was referring to the fact that although the 'sidereal' button was pressed, mount wasn't actually tracking (just after connecting to eqmod). It works similar way to Windows, but on Windows when the button is pressed, the mount is tracking.

milosz

3 arcsec is not great, I usually get less than 1 arcsec for a long time. However, I also noted the guiding sometimes sends big corrections, it has to do with the centroid algorithm that keeps track of the star. By default it is set to "Fast" which is from experience less susceptible to noise than "Smart" threshold algorithm; however, I might try to try to throttle it but this needs more testing when the weather clears up.

I never used the EQMod guiding, I will try to test this soon to find out the issue.

Yup, I can confirm that. I informed Jean-Luc so hopefully it will get addressed soon.

Have you tried to defocus slightly the guiding scope?. Maybe it helps the centroid algorithm.

Or maybe you can play with 'smoothing' of the guiding signal. In PHD it often helps with fine-tuned mounts.

I'm using an OAG and it's already slightly defocused. I'm going to check tuning the PID settings.