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

Does Dithering Throw Off PHD2 w/ Predictive Algo?

  • Please log in to reply
13 replies to this topic

#1 SilverLitz

SilverLitz

    Viking 1

  • -----
  • topic starter
  • Posts: 742
  • Joined: 17 Feb 2018
  • Loc: Louisville, KY

Posted 29 May 2020 - 10:16 PM

I finally got a clear night after more than a month, and I decided to mount my ES ED102CF w/ Hotech FF and my ASI183mm-Pro.  This is the 1st time I have tried the ED102 in over a year.  I am waiting on a M48/M42 adapter that was order from AMZ a week ago and has not shown up, so I could not try my EdgeHD 925 w/ COAG.  With the small ASI183 and 714mm, M101 frames up very nice, so I wanted to see how the ASI183 does at 0.7"/px and f/7.  I am exposing Lum at 120s and RGB at 240s.

 

When I started, the ED102CF was guiding the best it ever has on my G11, with the 1st 10mins pegged at 0.5-0.6" total RMS.  But then on a dither, DEC got way out of whack and it seemed that the predictive algo was showing the effect of this 1 off, causing several minutes of periodic big dips/rises.  After 8-10 minutes of these problems, and the error dips transitioning to error rises, it settled down back into the 0.5-0.6" total RMS range.  By the way, I dither with every frame.

 

Is this normal?

 

Is there a way to have PHD2 ignor the deviations around dither?



#2 imtl

imtl

    Messenger

  • *****
  • Posts: 480
  • Joined: 07 Jun 2016

Posted 29 May 2020 - 11:22 PM

Several things might be working against you here.

 

1. You got some backlash and PHD2 is issuing bigger and bigger pulses to try and get the DEC back centered after dithering. Eventually causing the mount to over react and go into oscillations.

 

2. Your DEC aggressiveness is set too high and its causing an over-correction. Same goes for really low MinMo. 

 

3. Combination of the two.

 

I have seen this happening with my mount and that is what was causing my problem.


  • SilverLitz likes this

#3 44maurer

44maurer

    Viking 1

  • *****
  • Posts: 875
  • Joined: 12 Jun 2016
  • Loc: Los Angeles, CA

Posted 30 May 2020 - 12:58 AM

8-10 mins to settle? There is some other problem. I settle back down Within a few seconds. 
 

I don’t have a clue what it could be, but when it dithers, does it lose the guidestar and it takes that long to find it again?



#4 Kevin Ross

Kevin Ross

    Viking 1

  • *****
  • Posts: 922
  • Joined: 28 Mar 2018
  • Loc: Traverse City, MI

Posted 30 May 2020 - 01:12 AM

Sounds like a lot of Dec backlash.

 

BTW, the Predictive PEC algorithm only applies to the RA axis, not the Dec axis. So if you're seeing problems in Dec, I don't think it's because of the PPEC algorithm.


  • Stelios, SilverLitz, Stricnine and 1 other like this

#5 ChrisWhite

ChrisWhite

    Fly Me to the Moon

  • *****
  • Posts: 5,820
  • Joined: 28 Feb 2015
  • Loc: Colchester, VT

Posted 30 May 2020 - 05:09 AM

Kevin beat me to it. The PPEC algorithm is only working with RA. 

 

Now it is a good question you pose though.  Slewing resets the "learning" in the algorithm, which is a MAJOR bummer because the best focus routine I have used is when the software slews to a single star to run AF.   On the Cem60 I do field focus generally, although I have been abandoning the PPEC algorithm of late and going back to good ole hystersis. 

 

I think that PPEC really works well when several factors are also working well.  Seeing needs to be really good.  With bad seeing, PPEC seems to get confused.  NO wind.  If you have wind, PPEC can't seem to handle that very well.  Some nights, for some reason PPEC seems to just crush it.  I get guiding below 0.4"/px and life is good.  Other nights I fight it.  It's possible that your dither action is causing problems, but more likely your balance is off somewhere or you have backlash to deal with.  If you are using a different RA algorithm does DEC suffer when you dither? 

 

In any event, I do like PPEC but whether I guide at 0.4"/px or 0.6"px makes no difference to my stars.... so I think I'll probably just stick with Hystersis and adjust min/mo based on seeing like I did before PPEC and like I do with my Mach 1. 


  • SilverLitz and OldManSky like this

#6 SilverLitz

SilverLitz

    Viking 1

  • -----
  • topic starter
  • Posts: 742
  • Joined: 17 Feb 2018
  • Loc: Louisville, KY

Posted 30 May 2020 - 07:13 AM

8-10 mins to settle? There is some other problem. I settle back down Within a few seconds. 
 

I don’t have a clue what it could be, but when it dithers, does it lose the guidestar and it takes that long to find it again?

I was not clear enough with my wording.  No, it did not take 8-10 min to find the star, it never lost the star.  Instead of the guiding measurement being in the +/- 0.5" range, it deflected to ~-3" and then quickly recovered (though a "U" not "V"), but I saw this type of error recurring over the next 8-10mins, though lower in amplitude and eventually on the + side.  Then it settled back into the nice tight range, and actually tighter for longer than I have been used to.  I was within 30min of the meridian, and M101 is high in the sky.



#7 SilverLitz

SilverLitz

    Viking 1

  • -----
  • topic starter
  • Posts: 742
  • Joined: 17 Feb 2018
  • Loc: Louisville, KY

Posted 30 May 2020 - 07:21 AM

Several things might be working against you here.

 

1. You got some backlash and PHD2 is issuing bigger and bigger pulses to try and get the DEC back centered after dithering. Eventually causing the mount to over react and go into oscillations.

 

2. Your DEC aggressiveness is set too high and its causing an over-correction. Same goes for really low MinMo. 

 

3. Combination of the two.

 

I have seen this happening with my mount and that is what was causing my problem.

You may be right.  I ran Guiding Assistant prior for the 1st time in a long time, as I was using a different scope.  GA's 4 min run looked VERY good, with 2" RMS total without any guiding, DEC was ~0.4", drifting very little, and RA had peak-peak of 6.7". 

 

But GA suggested lower MinMo than I was previously using, and my DEC aggressiveness was 100.  GA  also suggested a little longer time periods than previously as well.  GA measured DEC backlash at ~2500ms, but did not give a warning.  I implimented GA's suggestions.



#8 Der_Pit

Der_Pit

    Apollo

  • *****
  • Posts: 1,478
  • Joined: 07 Jul 2018
  • Loc: La Palma

Posted 30 May 2020 - 07:25 AM

How well are you balanced?  And where were you observing (position of the CWB)?  Could be that the big DEC jump started throwing RA away from the flange of the worm and it started wiggling between two states.  I had seen similar on my GP-DX, and had started preloading it with a constant pull.  


Edited by Der_Pit, 30 May 2020 - 07:25 AM.

  • SilverLitz likes this

#9 SilverLitz

SilverLitz

    Viking 1

  • -----
  • topic starter
  • Posts: 742
  • Joined: 17 Feb 2018
  • Loc: Louisville, KY

Posted 30 May 2020 - 07:31 AM

How well are you balanced?  And where were you observing (position of the CWB)?  Could be that the big DEC jump started throwing RA away from the flange of the worm and it started wiggling between two states.  I had seen similar on my GP-DX, and had started preloading it with a constant pull.  

I balanced slightly East heavy, using a 1.5# weight velcroed to the counterweight, when scope is on West side of pier, and the extra weight removed after meridian flip.  This behavior was prior to the flip (~30min before meridian) with the extra weight.



#10 SilverLitz

SilverLitz

    Viking 1

  • -----
  • topic starter
  • Posts: 742
  • Joined: 17 Feb 2018
  • Loc: Louisville, KY

Posted 30 May 2020 - 01:25 PM

The problem did NOT re-occur.  I looked at my results this morning, and my 4-hours of post-flip came out good (I went to bed after a couple of frames post-flip).  I only deleted 1-2 subs because of star irregularities, in addition of 2 out 4 subs during my early issues.  I had to cull a few more due to some thin clouds and rotationing into Louisville's light dome.

 

On the bright side, my initial results of the ASI183 at 0.7"/px and f/7 was much better than folks led me to believe; promising.  Other than having to increase my exposure times to 2x more than I was using with the Esprit+FR; pure f/ calculations would led me to longer at 2.8x, but it seems that I have been going well over the "swamping read-noise" as well as keeping to my dyadic exposure intervals in order to limit my dark library.  I expect my better result with the ASI183's small pixels and f/7 has to due with my home's Bortle 5 conditions versus many poster's Bortle 7.


  • 44maurer likes this

#11 AhBok

AhBok

    Mercury-Atlas

  • *****
  • Posts: 2,642
  • Joined: 02 Dec 2010
  • Loc: Lakeland, TN

Posted 30 May 2020 - 05:51 PM

I have wondered about this recently when I noticed that my RA guiding “seemed” to anticipate dithers for frames where I was skipping dithering. It did not destroy my guiding, but seemed to have a small effect that appeared periodic. In the end, I decided it was probably more imagined than real. I use permanent PEC in my mount, along with Predictive PEC and get excellent guiding.
  • SilverLitz and Der_Pit like this

#12 ChrisWhite

ChrisWhite

    Fly Me to the Moon

  • *****
  • Posts: 5,820
  • Joined: 28 Feb 2015
  • Loc: Colchester, VT

Posted 30 May 2020 - 05:54 PM

I have wondered about this recently when I noticed that my RA guiding “seemed” to anticipate dithers for frames where I was skipping dithering. It did not destroy my guiding, but seemed to have a small effect that appeared periodic. In the end, I decided it was probably more imagined than real. I use permanent PEC in my mount, along with Predictive PEC and get excellent guiding.


I have not noticed dithering being factored into the ppec algorithm. I can't say with confidence that it isn't. If it is, that would be a MAJOR oversight by phd devs.
  • SilverLitz likes this

#13 klaussius

klaussius

    Viking 1

  • *****
  • Posts: 521
  • Joined: 02 Apr 2019
  • Loc: Buenos Aires

Posted 01 June 2020 - 09:04 PM

The problem did NOT re-occur.  I looked at my results this morning, and my 4-hours of post-flip came out good (I went to bed after a couple of frames post-flip).  I only deleted 1-2 subs because of star irregularities, in addition of 2 out 4 subs during my early issues.  I had to cull a few more due to some thin clouds and rotationing into Louisville's light dome.

 

Your problem sounds a lot like DEC backlash caused by dither. If you see that problem again, you may want to look into the different dither patterns PHD can be configured to use. Some patterns are specifically designed to avoid causing backlash at each dither (like spiral patterns, for instance).

 

Bear in mind that non-random dither patterns will require that you make sure you complete a whole pattern cycle at least to get decent dithering.


  • SilverLitz likes this

#14 bobzeq25

bobzeq25

    ISS

  • *****
  • Posts: 20,137
  • Joined: 27 Oct 2014

Posted 01 June 2020 - 11:30 PM

Not so minor point.  This is one reason why imagers prefer better mounts, even though it seems like overkill in some situations.  The less guiding you have to do, the better.


Edited by bobzeq25, 01 June 2020 - 11:31 PM.

  • SilverLitz likes this


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