×

INDI Library v2.0.6 is Released (02 Feb 2024)

Bi-monthly release with minor bug fixes and improvements

Autofocus how exact?

  • Posts: 116
  • Thank you received: 2
Hi all,

I use Ekos autofocus with a Sesto sento focusser. But i can't get an precise autofocus. stars are not sharp when autofocus completes.
Does anybody use the same equipment as i and can get everything sharp?

Regards,

Joshua
3 years 11 months ago #51875

Please Log in or Create an account to join the conversation.

  • Posts: 398
  • Thank you received: 117

Replied by Doug S on topic Autofocus how exact?

I have a different motorized focuser, but setup is pretty similar for all motorized focusers. In your case, I recommend a quick review of the following:
www.cloudynights.com/topic/606001-initia...otic-focusing-motor/
Once you know your scope/focuser paired sensitivity, and can get near the critical focus zone, then (hardware problems aside), it's just a matter of setting the step size for the autofocus routine big enough to exceed seeing (which will allow the curve to be defined), but with steps small enough as to not jump over your CFZ. If you haven't calculated the sensitivity math, I recommend just playing with your step size. Run once to get close, and then run again with a smaller step size until you hit the convergence sweet spot in the CFZ. Step size too small causes seeing/algorithm convergence confusion. Step size too big causes a jump over the CFZ. An educated guess from your post is that your step size is too big. Once you're in the CFZ, adjusting the step size smaller is futile...your equipment can't focus any better. Knowing your telescope/focuser sensitivity can help resolve that question. Good luck....
The following user(s) said Thank You: Richard Beck
3 years 11 months ago #51884

Please Log in or Create an account to join the conversation.

  • Posts: 643
  • Thank you received: 62

Replied by Magnus Larsson on topic Autofocus how exact?

Hi!

Focus is a very very interesting topic... there are quite a few of us struggling with it, at least I am! :)

But: step size - what do you mean? I have a MyFocuser Pro, I can choose Full or Half. But no "step size". I can choos how many ticks it will go in the first move, and I can set the max travel (so as not to travel out of CFZ). But what do you mean by setting step size? What did I miss?

(I actually think that my focusing would be faster, but maybe not more accurate, if I could set a step size to say "3 ticks")

Magnus
3 years 11 months ago #51891

Please Log in or Create an account to join the conversation.

  • Posts: 398
  • Thank you received: 117

Replied by Doug S on topic Autofocus how exact?

I'm referring to the step size as defined in the "mechanics" subtab of the focus tab (lower left). In that sub-tab, you can set the step size, max steps, settle time, max travel, and backlash size. By constraining the initial step size and max step size, the algorithms have a "work space". When I was first exploring my EAF focuser, I would widen the parameters to get close, then run again with a smaller constraint so as not to allow the algorithm to overshoot the CFZ. This is why knowing the size of your CFZ, and matching it with your telescope's focuser and motorized step size is important. But, not doing the calc isn't terminal. You can just play with the step size and max step size until you witness the algorithm's curve behaving. Good seeing helps, and bad seeing can be pretty confusing to convergence.

Of course, if you set these too low, you may not be able to find the right starting place, or keep up with large temp changes. If you set too high, you'll jump over the CFZ. You're right...a bit tricky, but once you get the feel for it, it's easy enough. I think the trick is to run the autofocus manually iteratively, until you get the right feel for step size and max steps.

Cheers, Doug
The following user(s) said Thank You: Magnus Larsson
Last edit: 3 years 11 months ago by Doug S. Reason: Additional clarification
3 years 11 months ago #51895

Please Log in or Create an account to join the conversation.

  • Posts: 116
  • Thank you received: 2

Replied by Joshua R on topic Autofocus how exact?

Thanx i wil look in to that.
3 years 11 months ago #51896

Please Log in or Create an account to join the conversation.

  • Posts: 278
  • Thank you received: 17

Replied by S on topic Autofocus how exact?

Thanks for the link to the CN thread, very informative! There are some V curves posted on that thread, e.g.


This looks so much nicer than what I am getting (azi eaf). Does your v-curves look like this also?
3 years 11 months ago #51899

Please Log in or Create an account to join the conversation.

  • Posts: 398
  • Thank you received: 117

Replied by Doug S on topic Autofocus how exact?

No, I don't see that kind of curve. That curve looks to me to allow a pretty large step size (or HFR wouldn't go from 6 to 1.2). My curves are typically constrained by step size limits to be more shallow (at the low end) and noiser. I do experience seeing confusion, but it comes with the territory as my scope is F2.2 (so the CFZ is only 12 microns!). When I do get confusion, I just run the autofocus again. Most times, it only takes one or two runs to converge.
Last edit: 3 years 11 months ago by Doug S.
3 years 11 months ago #51900

Please Log in or Create an account to join the conversation.

  • Posts: 116
  • Thank you received: 2

Replied by Joshua R on topic Autofocus how exact?

As i understand as long as HFR is to high then you run the autofocusser again with smaller steps. is that correct?
3 years 11 months ago #51917

Please Log in or Create an account to join the conversation.

  • Posts: 398
  • Thank you received: 117

Replied by Doug S on topic Autofocus how exact?

Yes, you can "eye-ball" it via a zoomed image, or watch the HFR for a minimum and sanity check against what you think the seeing is. If in doubt, run the autofocuser again with smaller steps. If the seeing confusion is too bad, it won't converge. If reasonably stable, it likely will. Expect the position to be different between runs, but not dramatically so. If in doubt, run again. At some point, confidence will come in the answer.

FWIW, if you don't know your CFZ size, here's a simple way to find it:
CFZ=(roughly) 2.44*F^2 (where F is focal ratio). The answer is in microns. As an example, an F2.2 scope has a 12 micron CFZ.
Next, take the focuser motor steps/revolution, and pair it with the mechanical telescope focuser thread pitch.

As an example, the Celestron HD/RASA manual focuser has a thread pitch of 0.75mm/rev. The EAF motorized focuser has 5760 motor steps/rev. Thus, 0.75mm/5760 = 0.13(microns) per motor step. To be in the 12 micron CFZ, my RASA must be within 12um/0.13 um/step = +/- 46 steps. I like a step size near ~20 steps (~5 chances for the algorithm to hit a count in the zone), and I allow a max size of a few hundred. A smaller step doesn't really gain much (assuming decent seeing); you just want to be able to settle in the zone. Too big risks jumping the CFZ.
Last edit: 3 years 11 months ago by Doug S.
3 years 11 months ago #51922

Please Log in or Create an account to join the conversation.

  • Posts: 1957
  • Thank you received: 420
I typically use the Linear focus algorithm with the following configuration for my ASI EAF and Esprit 80ED:
Settings tab:
- Full Field
- Annulus: 25% and 80%

Process tab:
- Detection: SEP
- Algorithm: Linear
- Tolerance: 5%

Mechanics:
- Initial step size: 15

and all the rest default. And I use 1x1 binning and exposure times of 0.5 sec. I typically get HFR values of 1.3 and better depending on the seeing.

EDIT: very important: you *need* to manually focus to something acceptable (say, HFR < 2) before using the auto-focus functionality otherwise it will take too long to settle. Also, SEP doesn't handle BIG stars or donuts very well.
The following user(s) said Thank You: Keith Cobby
Last edit: 3 years 11 months ago by Wouter van Reeven.
3 years 11 months ago #51923

Please Log in or Create an account to join the conversation.

Time to create page: 0.536 seconds