Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Autofocus in SGP rarely works

  • Please log in to reply
16 replies to this topic

#1 jcschultz01

jcschultz01

    Lift Off

  • -----
  • topic starter
  • Posts: 21
  • Joined: 09 Aug 2018
  • Loc: Twin Cities, MN

Posted 21 September 2020 - 07:14 AM

Hi

I am using a Pegasus Astro Focus Cube V2 on a Celestron Edge HD C9.25 and the SGP autofocus routine.

 

I had to increase the  focus step size to 500 in order to get any kind of "U" or "V" shaped curve for 9 data points.

I have manually focussed the telescope so it is close to focus and disabled smart focus for the EDGE.

The other major options are in the attached screen shot. 

 

With this large step size I get a "U" shaped cuver but the quality is rarely over 80% for the full 9 points and the auto-focus fails after 3 attempts.  

 

I have attached and SGP log file too if that helps. 

 

Is my binning at 2x2 still too low? I noticed some comments in the log fiile "AF frame was too large... downsample = 0.75...". 

Can I turn down the "quality" needed for an autofocus position?  A curve with 70% quality looks OK to me but the autofocus routine apparently wants > 90%????

 

I have tried various numbers of data points, step size,  min star diameter,...   but nothing gives me reliable autofocus so I am imaging with autofocus turned off.  

 

Thanks for any suggestions.

 

Attached Thumbnails

  • 2020-09-21 AutofocusSettings.jpg

Attached Files



#2 rhart426

rhart426

    Vostok 1

  • -----
  • Posts: 149
  • Joined: 16 Sep 2017

Posted 21 September 2020 - 07:45 AM

A 'U' shaped curve implies that your step size is too large; it should be more of a concave 'v' shape.  Preliminary question; did you meticulously follow the auto-focus setup procedure in the help files?  It is easy to overlook steps and/or kinda just blow through it, and 9 out of 10 times that I help with these issues following the procedure to the letter fixes the issues.  In particular, taking the time to manually calculate the correct step size is critical.


  • Raginar likes this

#3 lucam

lucam

    Viking 1

  • -----
  • Posts: 740
  • Joined: 09 Oct 2017
  • Loc: Upstate NY, USA

Posted 21 September 2020 - 08:01 AM

U or V shape depends on how close to the minimum of the focus curve one gets. The focus curve is a hyperbola so if the ratio of largest to smallest HFR is 2-3, you should resolve the U. If on the other hand with your 9 steps the ratio of max to min HFR is closer to 5, it will look like a V. 

 

You can change the acceptance value for the quadratic fit. There is a variable quadratic_fit_min_quality that can be changed by the user in the sequencer GUI. See the attached image. It's possible that with the aperture and focal length of the SCT you may have too much bounce from seeing and the curve goes up and down instead of just looking like a smooth hyperbole. For starters, try spreading your points more until you have a max to min HFR ratio of about 5. Once you manage to get that to work well, you can try to fine tune and reduce the spread to sample the bottom of the curve better. 

 

Screenshot 2020-09-21 08.52.55.png



#4 kathyastro

kathyastro

    Vanguard

  • *****
  • Posts: 2,262
  • Joined: 23 Dec 2016
  • Loc: Nova Scotia

Posted 21 September 2020 - 08:05 AM

I had lots of trouble getting a good autofocus until I followed the documented setup procedure.  It tells you the correct step size for your gear, so you don't have to guess.  Now I get excellent autofocus.  (I also did a focuser transplant, which helped.)


  • Dynan likes this

#5 StephenW

StephenW

    Apollo

  • -----
  • Posts: 1,207
  • Joined: 04 Feb 2005
  • Loc: San Jose, CA

Posted 21 September 2020 - 09:15 AM

Looking at your focus curve:

 

- You have backlash:  Have you set backlash compensation?

- Your step size is too small, causing a "flat" curve"

 

As Kathy says, follow the SGP instructions on how to set everything up and it should work fine.

 

sgp_focus.jpg

 


  • lucam likes this

#6 bacon

bacon

    Sputnik

  • -----
  • Posts: 38
  • Joined: 05 Jul 2019
  • Loc: Denver, CO

Posted 21 September 2020 - 01:38 PM

I use the Pegasus on an Edge8 and I agree that backlash compensation is essential.  I use 1000 steps judging that it requires at least 1000 to move the focus knob an appreciable amount.  And of course double check that you're building the curve with CCW focus movements.  Just 7 points and 300 step size work well for me, though the gearing on a 9.25 is likely different.  Also, filter offsets (if your set is not parfocal) go a long way to getting a proper curve on the first try.


  • Raginar likes this

#7 KNak

KNak

    Explorer 1

  • *****
  • Posts: 92
  • Joined: 30 Aug 2011
  • Loc: Ontario, Canada

Posted 21 September 2020 - 04:56 PM

I use the Pegasus on an Edge8 and I agree that backlash compensation is essential.  I use 1000 steps judging that it requires at least 1000 to move the focus knob an appreciable amount.  And of course double check that you're building the curve with CCW focus movements.  Just 7 points and 300 step size work well for me, though the gearing on a 9.25 is likely different.  Also, filter offsets (if your set is not parfocal) go a long way to getting a proper curve on the first try.

 

Very similar to me on my C11 XLT:  7 steps at 350 step size, backlash compensation = 2000 (always CCW) motion.  Works well for me.  My ratio of HFR from maximum to minimum on the curve is about 2-3.


  • Raginar likes this

#8 Coconuts

Coconuts

    Viking 1

  • *****
  • Posts: 585
  • Joined: 23 Sep 2012

Posted 21 September 2020 - 05:25 PM

For me, a timely topic!  I am just coming up to speed on SGP, and my focuser is the Optec Leo, which is a very nicely design and built piece of gear.  Its three fine-pitch leadscrews, gear driven by a microstepped motor, provide a very high resolution of 0.08 um, far below that of more typical Crayford and rack and pinion designs.  The factory measured backlash is 5 um.  I will dutifully enter the values in SGP to the best of my ability, but it would be great to hear from anyone who has already worked out the best values to use with the Optec Leo.  Since f-ratio matters a lot in focusing, my astrographs are f/2.8 and f/4.0. 

 

All the best,

 

Kevin


Edited by Coconuts, 21 September 2020 - 05:26 PM.


#9 DMRandall

DMRandall

    Vostok 1

  • -----
  • Posts: 105
  • Joined: 25 Dec 2008

Posted 21 September 2020 - 07:40 PM

I'd forgotten how I calculated the step size.  Thanks for the reminder, I need to do this with my other scope now too!



#10 jcschultz01

jcschultz01

    Lift Off

  • -----
  • topic starter
  • Posts: 21
  • Joined: 09 Aug 2018
  • Loc: Twin Cities, MN

Posted 21 September 2020 - 10:08 PM

Thanks for the suggestions everyone.

 

So I followed the procedure in the manual (RTFM) and came up with a step size of 875 steps!! for autofocus (I had been using 200 to 500). I ran this along with backlash compensation of 30 (this might still be too low).

 

SGP found focus in one pass (9 points) with a focus range of 1000 to 8000.  

(Pegasus Astro - Focus Cube V2 with Celestron EDGE HD C9.25).

 

I am still concerned about the HFR but don't know if this is an issue with the smoke from the fires (fire.airnow.gov) or there is something not quite right with the Celestron.

 

Howver I can now autofocus!!

 

 

Attached Thumbnails

  • 2020-09-21 Autofocus stesize 875.jpg

  • StephenW, kathyastro and lucam like this

#11 imtl

imtl

    Viking 1

  • *****
  • Moderators
  • Posts: 968
  • Joined: 07 Jun 2016

Posted 21 September 2020 - 10:23 PM

Now to fine tune and improve it. 

 

So the title of this thread should be changed to "Learning how to get autofocus in SGP to work on Edge 9.25"" instead of the autofocus in SGP rarely works....


  • GrandadCast and RossW like this

#12 StephenW

StephenW

    Apollo

  • -----
  • Posts: 1,207
  • Joined: 04 Feb 2005
  • Loc: San Jose, CA

Posted 21 September 2020 - 11:24 PM

Definitely looking a lot better! :)

 

>I ran this along with backlash compensation of 30 (this might still be too low)

 

From the graph I posted earlier from your log file, it took about 2 steps to eat up the backlash, and that's when you were using a step size of 500 - so I'd try something in the 1000 range.

 

And other than running out of focuser movement(!), and the time it takes to move, you can't actually have "too big" a backlash compensation setting.



#13 Stelios

Stelios

    Voyager 1

  • *****
  • Moderators
  • Posts: 10,221
  • Joined: 04 Oct 2003
  • Loc: West Hills, CA

Posted 22 September 2020 - 01:46 AM

I would use a backlash of at least 2x step size. So I would use at least 1,500 to 2,000 (if indeed 875 is the proper step size). 

 

I am still concerned about the HFR but don't know if this is an issue with the smoke from the fires (fire.airnow.gov) or there is something not quite right with the Celestron.

 

 

I have seen HFR's with my 8" Edge from 1.5 to over 3, depending on seeing. 

 

Howver I can now autofocus!!

 

Great, if only you could now change the thread title...


  • imtl likes this

#14 kathyastro

kathyastro

    Vanguard

  • *****
  • Posts: 2,262
  • Joined: 23 Dec 2016
  • Loc: Nova Scotia

Posted 22 September 2020 - 07:53 AM

Much better!

 

The "wing" on the right end of the graph indicates that you still need a little bit more backlash compensation.



#15 lucam

lucam

    Viking 1

  • -----
  • Posts: 740
  • Joined: 09 Oct 2017
  • Loc: Upstate NY, USA

Posted 22 September 2020 - 08:04 AM

Now that you have a good V curve, you may want to start reducing your step size to get more resolution around the HFR minimum. The ratio of max to min HFR in the focus curve you posted is about 4, you can bring that down to 3 or so and have more points to really hug the curvature of the U curve.

 

As others have said, bump up your backlash compensation - you cannot have too much backlash compensation so there is very little downside to having more rather than less. Also, make sure that the focuser driver does not apply additional backlash compensation outside of SGP..  



#16 jcschultz01

jcschultz01

    Lift Off

  • -----
  • topic starter
  • Posts: 21
  • Joined: 09 Aug 2018
  • Loc: Twin Cities, MN

Posted 22 September 2020 - 07:24 PM

I am bumping up the backlash compensation tonight.  Not sure I want to jump to a backlash 2x the step size but I'll keep that in mind as I increase backlash.  Increased from 30 to 300 at the moment.  I'll check the curves tonight.

 

And just a a point of reference talking to inexperienced people (like me), I now appreciate the comment as well as the work to plot the graph  from my data that someone posted above but at the time I didn't have the foggiest what to do about the comment. 

 

And if another  comment was serious on changing the title of this thread to "Learning how to get autofocus in SGP to work on Edge 9.25with a Pegasus v2" I would be happy to change the title but I don't know how. 



#17 StephenW

StephenW

    Apollo

  • -----
  • Posts: 1,207
  • Joined: 04 Feb 2005
  • Loc: San Jose, CA

Posted 22 September 2020 - 07:48 PM

>Not sure I want to jump to a backlash 2x the step size but I'll keep that in mind as I increase backlash.  Increased from 30 to 300 at the moment

 

Per the earlier log you posted (and the graph extracted from it), 300 isn't enough.  You need to use 2x your step size.

 

And again, you generally can't have "too much" backlash compensation


Edited by StephenW, 22 September 2020 - 07:49 PM.



CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics