I am working from a Bortle 4 sky area and when using the focus module, sometimes it appears to work properly (verified with Bahtinov mask) and other times it ends up reporting that it has focused but I end up with donut stars. I would love to be able to use this but I just don't have confidence that it will give me a good result. Has anyone else seen this? For example, last night I started imaging with manual focus and then about 1.5 hrs in, I used the autofocus module and it worked fine. A few hrs later, I used it again, the stars looked OK and went back to bed. Turns out all my images from that point on were no good. Help?

Thanks

Please take a look at the advice in indilib.org/forum/general/6167-for-those...-issues.html?start=0
and let us know what you tried and how it goes.

I experienced a similar thing last night. I was using a Triad Ultra filter and as a result my galaxy target (M101) was dim. My initial focus module solve on Regulus (based off a bahtinov mask starting position) was good. When I went to solve on target M101 it didn't auto-find and I couldn't provide - a bright enough target. I solved this by moving off target to a bright nearby star and focusing on it.

Later the "focus after 60 minutes" triggered and that killed stars completely when trying to do that. The log isn't clear as to the problem, so I have upped my ZRam to 4 G in case it was related to my 90M fits files.
Otherwise it might be related to recent bleeding edge code that appears to indentify my Z7 as having a size other than 8256 x 5504 (I lost track of what it thought) it should be. It seems to want to auto-populate the wrong value after interrogating the camera. I'll report on this separately if I can find something that suggests is not just me (again).

[EDIT: I will try the suggestions that Hy made just before I posted to see if that helps in my case as well]

If your stars end up donut shaped, you are using a mirror and what you are seeing is the central obstruction. The focuser cannot distinguish between a star and the ring resulting from the donut. Refractors do not have that problem.

Been there myself.

The only way to prevent that is to reduce the initial step size so that the focuser is not moving out or in so far that the star degrades to a donut.

Once that happens, you can only move it back manually. If the focuser gets confused because of clouds and moves too far out of the focal plane, you are cooked.

As far as I know, there is no fix that recognizes the donut shape of the star and adjust focus accordingly to move back to a point.

I am using a DSLR so it is not the central obstruction. It could be stepping too far out and then getting messed up as you say. What initial step size would you suggest?

The central obstruction comes from the secondary mirror, it nothing to do with the imaging camera

I would use an initial step size that never results in a donut. I. E. Moving the star out of focus, but not so far that it degrades into a ring.

You need to determine that empirically fir your setup.

I may be wrong, but I think the j dain is referring to using it with a DSLR and DSLR lens combination, no mirrors involved.

Also limiting the maximum travel will help prevent going too far off and getting donuts, but you'll have to get close to focus first.

Using the Polynomial algorithm may also contribute to the issue, as it makes larger moves than the linear algorithm.
You might try using Linear, if you're not already doing that.

When I am using a DSLR lens, or a refractor for that matter, I am getting pancakes, not donuts, i.e. no hole in the middle. The donuts are diagnostic of reflector telescopes with the secondary mirror creating the central hole.

I see that with my RC8 and I know exactly what jdain is talking about. It can make automated imaging with regular refocusing with a reflector prone to failure. Any wisp of a cirrus cloud at the wrong time can ruin imaging for the rest of the night.

If the reflector is too far out of focus and the central hole appears, the focus module will lock on to the rim of the ring, not to the ring itself as it is looking for a Gaussian distribution to determine HFR. The ring would create a well, not a peak and the algorithm would not know how to handle that. Ironically, when moving towards actual focus, once the hole disappears during the focus movements HFR will then paradoxically increase and the focuser will conclude that it is moving away from optimal focus and reverse direction again.

Currently there is no way to recover from that. The only way to avoid that is to reduce the initial steps the focuser moves away from its current position so that no hole appears, then move back in until optimal focus is reached.

I have not yet tried that with Hy's linear algorithm to see how that handles this problem, but I suspect step size will also be the critical parameter there.

Jo

I tried all of the notes above in the link provided and it still seems to get into an endless in 500 pts and out 500 pts and never achieves focus. Someone else mentioned limiting the amount of travel but I'm not sure what setting he is referring to or the value I should use. Any other suggestions? Thanks

Please attach or point us to a log file in verbose mode with at least focus checked and perhaps indi and capture too.
indilib.org/support/logs-submission.html