Dec Jumps during Imaging

118 views
Skip to first unread message

mark matzner

unread,
Apr 25, 2016, 3:09:31 PM4/25/16
to Open PHD Guiding

Gents,

Last night’s short session allowed me to capture 8 subs of M51 before the clouds moved in. Subs 1, 2, 4 and 8 were OK.  Each of subs 3, 5, 6 and 7 showed a large southern DEC excursion. 


Noticeable in all of the failed subs is that there is an initial large DEC signal in the direction of the bad behavior, followed by the system’s effort to correct.  In the “Sudden Large Excursions – “Gremlins”” section of your Basic Tutorial, you mention this behavior in passing.  But I’ve not found an explanation for what might be the cause and correction.  My guess is this results from substantial DEC backlash, but who knows?


I see that I had raised this same issue back in September of last year.  But the difference between then and now is that in September there were no initial signals to fail, just correction signals.


As always, you guys do a great job with the product and support.  Any light you can shed on my issue is greatly appreciated.

Thanks,

Mark


12"LX200 Classic (3048mm F.L) on a Mitty Wedge - Dec is balanced and RA is neutral balanced toward zenith

Starizona F/7.5 SCT Corrector mounted inside the Feathertouch focuser  (f7.16, 2182mm FL)

Hutech OAG5 off axis guider

Lodestar guide camera with a nose mounted Scopestuff 0.5x reducer  (f/3.58, 1091mm FL)

Atik 383L+mono camera, Atik EFW2 filter wheel

PHD2, 2.6.1 dev7 for guiding, SGPro v2.5.1.5 for automation 

Location:  ~ 32.54 degrees latitude


Included are the SGPro, PHD2 and Lesvedome log files along with the SGPro Sequence file.  Also a copy of sub #3 to show the problem.  


https://www.dropbox.com/sh/zqdcs659ngbr2kv/AADRENdscEoZhZcv0L9nEfZQa?dl=0

Bruce Waddington

unread,
Apr 25, 2016, 9:01:12 PM4/25/16
to mark matzner, Open PHD Guiding

Hi Mark.  Thanks for looking at the data yourself and sending all the useful information in the first pass.  I think the issue you've identified is caused by your use of a truly huge backlash compensation size - over 15 secs!!!  That's really not going to work.  The backlash comp feature is really intended for mounts that are much better behaved - much smaller backlash to begin with and very predictable behavior.  In your case, the huge backlash comp pulse actually works ok in many cases but then causes these big excursions in other situations.  The backlash comp feature has to assume that the gears can be kept fully engaged (or nearly so) in either direction.  This is a reasonable assumption for better-behaved mounts but not for others.  In your case, I think the normal guiding behavior can result in the Dec gear train entering a dead zone where the gears aren't engaged and nothing really happens with the guide commands.  Then, when there's a direction reversal, the backlash comp pulse may be much bigger than what's needed.

 

If you ran the Guiding Assistant, the recommendation would not have been to try using backlash compensation - it would have been to guide in only one Dec direction.  I think that's what you need to do unless you can make significant mechanical improvements in the mount.  We considered putting an upper limit on what you could specify for backlash comp but didn't do that in the interest of letting users experiment.  But maybe we should revisit that decision or at least put up a warning if a huge amount is specified.  

 

To guide in one direction, just avoid a near-perfect polar alignment and watch how the Dec slowly drifts in one direction (without ruining your subs).  Then specify Dec guiding in the direction that will compensate for the drift.

 

Good luck,

Bruce

 


--
You received this message because you are subscribed to the Google Groups "Open PHD Guiding" group.
To unsubscribe from this group and stop receiving emails from it, send an email to open-phd-guidi...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Bruce Waddington

unread,
Apr 25, 2016, 9:12:10 PM4/25/16
to mark matzner, Open PHD Guiding

One other point I forgot to mention.  The big excursions caused by the backlash comp, as far as I can see, always follow a large dither in Dec.  Considering how much your mount struggles with Dec guiding and direction reversals, I don't think it's advisable to dither in Dec.  Doing so basically interrupts whatever steady-state behavior you've established in Dec and makes it much harder to get things back under control.  If you can't give up Dec dithering, you might try using the spiral-dither feature, which will tend to reduce the number of direction reversals.  Even then, there will be points where you create some instability and will probably lose some subs.

 

Bruce

 


mark matzner

unread,
Apr 26, 2016, 9:26:53 AM4/26/16
to Open PHD Guiding
Bruce,
Thanks for the input.  
I recognize that my old classic has some limited mechanical attributes, so I'll definitely turn off the BL compensation.  I've never had the motivation to try DEC guiding in only one direction, but now seems like a good time.  And in order of preferences, the DEC dither may also go.

I am curious about the spiral dither application that you've written.  Is it compatible with SGPro, and if so, how do I implement it?

I'll follow-up with a report on what did what, assuming we get some skies here in central Texas.

Ciao,
Mark



Andy Galasso

unread,
Apr 26, 2016, 10:57:00 AM4/26/16
to mark matzner, Open PHD Guiding
On Tue, Apr 26, 2016 at 9:26 AM, mark matzner <mdma...@gmail.com> wrote:

I am curious about the spiral dither application that you've written.  Is it compatible with SGPro, and if so, how do I implement it?


Spiral dither and Bruce's dither app are two different things.

The Spiral Dither option is available in the 2.6.1 development snapshot builds and can be enabled in the brain, Global tab:

Inline image 1 

When the dither mode is set to Spiral, then each dither will move the lock position one step along a square spiral path:

Inline image 2
In this way the Dec direction reversals are less frequent than the default, random dither pattern where Dec reversals are quite frequent, for example:

Inline image 4

Bruce's PHD2 dither app is a way to allow imaging apps like MaxImDL that lack PHD2 integration support to work with PHD2.   Since you are using SGPro, which already has excellent PHD2 integration, you do not need to use the dither app, and it does not add any new dithering functionality beyond what is already present in PHD2.

Andy

mark matzner

unread,
Apr 26, 2016, 11:04:53 AM4/26/16
to Open PHD Guiding
Did't see that!
Thanks for the clarification Andy.

Mark

peter wolsley

unread,
Apr 26, 2016, 12:02:12 PM4/26/16
to Open PHD Guiding
Andy and Bruce,
I tried to study Mark's guiding log to see if I could figure out all the guiding actions.  I believe that there is a significant interference issue when using the "fast switch" option and using backlash comp.  IMO whenever the "fast switch" feature acts it does not allow the backlash feature to act.  Here is how I came to this opinion...

Using PHD Log Viewer at 2016-04-24 20:20:13
For the majority of the dither actions in Mark's guide log I noticed that the "fast switch" feature activated and caused PHD to issue a guide pulse that was in keeping with the amount of DEC dither.  The "fast switch" feature activated because PHD has been issuing DEC guide pulses in one direction and then the dither required a significant movement in the opposite direction.  The meant that the mount was asked to move thru it's backlash.  The 2,500ms backlash comp pulse did not happen so the mount never got very far thru it's backlash.  The DEC guide star position slowly drifted as the mount effectively wandered in it's backlash.  Eventually the guide star drifted enough for PHD to detect a small change in direction that had persisted for three cycles.  Because the error was less than 3 times the DEC min move the "fast switch" logic did not act which then allowed the backlash comp 2,500mS pulse to happen.  Because the mount had not truly moved thru it's backlash this 2,500mS backlash pulse pushed the mount way too much which quickly took up any backlash and moved the guide star over 15 arc-seconds in DEC.

IMO I would suggest that the backlash comp feature, when enabled, should always occur when a change in direction occurs.  The "fast switch" modifier should only decide how quickly a correction should occur. 

Based upon Mark's guide log I would suggest that not using "fast switch" would have resulting in better guiding.

Peter

peter wolsley

unread,
Apr 26, 2016, 12:12:17 PM4/26/16
to Open PHD Guiding
Andy and Bruce,
I just realized that this may not actually be an issue with "fast switch".  This may be an issue with dither.  I.E. maybe when dither occurs it does not allow backlash comp to act.  Regardless of whether it's caused by dither or "fast switch" I believe that once backlash comp is used it needs to be used every time the DEC axis is commanded to change direction.

Peter

jcbe...@gmail.com

unread,
Apr 26, 2016, 9:56:46 PM4/26/16
to Open PHD Guiding, mdma...@gmail.com
Hello,

I've been working with a number of challenges trying to establish PHD2 guiding with my Mac using an Orion Atlas Pro AZ/EQ-G mount then opened this thread.  Apologies in advance for hijacking the thread but thought this may be directly related.  I had no idea what could have been creating the images I was getting but it seems I have have been imaging as the program was dithering?  I've uploaded an image to drive, the link is below.  I have also included a Debug log and Guide log.  In the Mac version I'm not seeing these advanced settings in Brain link, are they not part of that build?

Joe



Bruce Waddington

unread,
Apr 26, 2016, 10:01:02 PM4/26/16
to peter wolsley, Open PHD Guiding

I guess I have a different view on this.  The bottom line for me is that simple backlash comp is not going to work with a mount that has this much backlash.  We don't know what the actual bl measurement was in the Guiding Assistant, so that's an open question.  If it was really 15 secs (of time), that's hopeless - the OP needs to guide in one direction.  If it was much lower than that to start, the bl adjustment algorithm may have malfunctioned because of the ceiling being applied by the Dec max-duration.  That's something we will fix.  In the interim, users should make sure the backlash comp amount is <= max-dec-duration.

 

I don't think the fast-switch option has anything to do with the problem.  All it does is tell the algorithm to switch directions more willingly when faced with a large excursion - that would cause the bl compensation to happen more readily, not less.  It's true that dithering doesn't use the bl compensation, but I don't see that as a problem either.  The normal guiding machinery will move the star to the new location regardless of backlash - it may just take longer.  And most users probably use the "fast-recenter" option, which applies large guide pulses to accomplish the dither.  Either way, the backlash is likely to be cleared, and the settling parameters should keep any of this from causing problems.  If you look at the dithers, I think all of them appear to have applied the correct number of Dec guide pulses to get the job done.

 

As I said, I think the OP should either make major improvements to his mount or use uni-directional Dec guiding.  We could probably invent a fancy algorithm that would help his particular mount but that's not the business we're in.  Whatever we do has to apply to a huge range of mounts, most of which we know little about in terms of mechanical behaviors.

 

Cheers,

Bruce

 


From: open-phd...@googlegroups.com [mailto:open-phd...@googlegroups.com] On Behalf Of peter wolsley


Sent: Tuesday, April 26, 2016 9:12 AM
To: Open PHD Guiding

--

Bruce Waddington

unread,
Apr 26, 2016, 10:33:37 PM4/26/16
to jcbe...@gmail.com, Open PHD Guiding, mdma...@gmail.com

Hi Joe.  If you wouldn't mind, would you please re-post using a new thread and subject line.  Mark's problem is related to backlash compensation, a feature you're not using.  

 

Thanks,

Bruce

 


Sent: Tuesday, April 26, 2016 6:57 PM
To: Open PHD Guiding

Cc: mdma...@gmail.com
Subject: Re: [open-phd-guiding] Re: Dec Jumps during Imaging

 

Hello,

--

steve

unread,
Apr 27, 2016, 1:59:53 AM4/27/16
to open-phd...@googlegroups.com
On 27/04/16 04:01, Bruce Waddington wrote:
> In the interim, users
> should make sure the backlash comp amount is <= max-dec-duration.
>

Hi
Is that a default value? I've changed nothing from default. Must I look
for these values?
Cheers,
Steve


Bruce Waddington

unread,
Apr 27, 2016, 10:22:58 AM4/27/16
to steve, open-phd...@googlegroups.com
Backlash compensation is disabled by default. If you run the Guiding
Assistant at some time and measure your mount's Dec backlash, you may see a
recommendation to turn on backlash compensation - that's how it normally
gets enabled. Unless you've done this, you can forget about it. Both of
these parameters are on the algorithm tab of the Advanced Dialog (brain
dialog).

Bruce

-----Original Message-----
From: open-phd...@googlegroups.com
[mailto:open-phd...@googlegroups.com] On Behalf Of steve
Sent: Tuesday, April 26, 2016 11:00 PM
To: open-phd...@googlegroups.com
Subject: Re: [open-phd-guiding] Re: Dec Jumps during Imaging

mark matzner

unread,
Apr 28, 2016, 10:47:41 AM4/28/16
to Open PHD Guiding, stev...@gmail.com

Bruce and Andy and Group,

Here are the results from last night.  Lights were taken in the following order, times and quantities:

      a)    LUM  450s qty 20

      b)    GRN 300s  qty 10

      c)    BLU  300s  qty 10

      d)    RED 300s  qty 10

 

Two things were done before the run.  I turned off the Backlash Compensation and changed the dither to Spiral.

      1.    There no longer appears to be an errant signal at the beginning of a DEC excursion; just the recovery effort as expected. 

      2.    Using Pixinsight’s Blink process, I could see the spiral dither movement from one sub to the next.  And the excursions that occur don’t appear (to me) to be closely tied to the dither operation.

 

The good news is I only felt compelled to throw out one (1) LUM sub, and no GRN subs.  However as the session wore on, I had to discard 4 (of 10) BLU subs and 7 (of 10) RED subs.  So as evening turned to morning and the scope moved from east to west (and temps dropped about 6C), the sub loss rate increased substantially.

 

I had previously tried to reduce the mechanical backlash in the DEC.  As best I can tell, most of the BL is in the initial nylon gear set rather than at the worm and worm gear.  I don’t know how one could go about altering this gear set.

I also struggle with whether or not to guide in only one direction, since out of the first 30 subs I only lost one.  Then,  after that the bottom fell out.

 

A couple of things that come to mind to try are:

      1.    Update the polar alignment.  Could a less than perfect one be aggravating the situation?

      2.    Add weight to the scope nose to increase the DEC bias (probably towards north).

 

Thanks for your thoughts, and I appreciate everyone who has contributed to this discussion.

Mark


bw_msgboard

unread,
Apr 28, 2016, 3:55:58 PM4/28/16
to mark matzner, Open PHD Guiding, stev...@gmail.com

Hi Mark.  If we start with what the data show, we can look at a typical example:

 

 

This is a sudden excursion of 15 arc-sec in Dec that has nothing to do with guiding.  As you said, these happened more frequently as the imaging session went on.  We can natter all we want about backlash compensation or other software strategies, but this is a lost frame, plain and simple.  It’s hosed.   Unless you can find the source of these things and eliminate them, you are going to have to discard many frames.  Looking at your sample image, my first thought was that something had shifted or moved abruptly during the exposure but that’s just an educated guess.  What follows is just my personal opinion, which you are certainly free to discount or ignore.

 

Now that I see more of the context of what you’re trying to do, I’m not surprised you’re having trouble.  If I have the picture right, you’re trying to image at over 2200mm focal length with a Meade Classic on a fork mount.  And you are apparently using a separate guide scope assembly with a relatively long focal length – 1100mm.  Having worked with a Meade SCT (post-classic) for many years, I would predict you’re going to throw away lots of frames – that’s the nature of what you’re trying to do (I had many images like the one you posted).  Frankly, it’s a recipe for frustration unless you really enjoy trouble-shooting and tinkering around with your gear.  Unless you’ve modified the OTA, I think you don’t have a mirror lock – so you can be sure the mirror is going to move a bit as the scope rotates to different parts of the sky.  Even with a mirror lock, it will move to some extent, and your camera will see movements of as little as 5u.  Next, it’s pretty difficult to use a separate guide scope when imaging at >2000mm focal length unless you have very good gear.  I don’t know how you’re getting to 1100mm focal length on the guide scope – whether you’re using a Barlow lens or the guide scope is really that large.  In any case, differential flexure is likely to be a constant battle and, again, it will change as the scope’s pointing position changes.  Long focal-length imaging almost always requires use of an off-axis-guider set-up of some kind.

 

If you can’t upgrade your gear, which I understand not everyone can do, you might consider the following:

  1. Use the Meade 0.63x focal reducer on the scope (but stay away from the 0.3x focal reducer)
  2. Get an off-axis guider
  3. Dither only in RA and don’t dither after every frame.  Your mount is seriously challenged to begin with, so ask it to do as little as possible.
  4. Guide in only one Dec direction
  5. Take lots and lots of exposures of <= 5min and just discard the ones that are bad.  If you retain enough frames, you can still get a decent result.

 

Sorry for the Dr. Doom commentary, but I don’t think I can help you much at this point.  Guiding is not really your problem IMO.

 

Good luck,

Bruce

 


From: open-phd...@googlegroups.com [mailto:open-phd...@googlegroups.com] On Behalf Of mark matzner


Sent: Thursday, April 28, 2016 7:48 AM
To: Open PHD Guiding

Cc: stev...@gmail.com
Subject: Re: [open-phd-guiding] Re: Dec Jumps during Imaging

 

Bruce and Andy and Group,

--

image002.jpg

mark matzner

unread,
Apr 28, 2016, 6:04:38 PM4/28/16
to Open PHD Guiding, mdma...@gmail.com, stev...@gmail.com, bw_m...@earthlink.net
Hi Bruce and thanks for the response.
I know how things can get lost in the shuffle, so just to let you know, I am the original poster.
Per the data on my original post, I am using a Starizona SCT Corrector that yields about f7.16 (FL going into the imaging camera).  I do use an OAG with the Lodestar camera for guiding.  To further reduce the guiding FL, I've added a 0.50x reducer just in front of the Lodestar, cutting its FL to f3.58 (or at least that's what I believe).

So at this point, I will try next the dither in RA only.  Then the guide in one direction.  I'll follow-up on the results once the skies clear.

Cheers,
Mark

bw_msgboard

unread,
Apr 28, 2016, 6:19:20 PM4/28/16
to mark matzner, Open PHD Guiding, stev...@gmail.com

The OAG is a good thing, but I don’t think you’re going to make much headway until you can eliminate these spontaneous shifts in Dec.  Dithering and uni-directional Dec guiding don’t have anything to do with that problem as far as I can see.

 

Good luck,

Bruce

 


From: mark matzner [mailto:mdma...@gmail.com]
Sent: Thursday, April 28, 2016 3:05 PM
To: Open PHD Guiding
Cc: mdma...@gmail.com; stev...@gmail.com; bw_m...@earthlink.net
Subject: Re: [open-phd-guiding] Re: Dec Jumps during Imaging

 

Hi Bruce and thanks for the response.

Reply all
Reply to author
Forward
0 new messages