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

A Theory of Guiding

This topic has been archived. This means that you cannot reply to this topic.
36 replies to this topic

#1 Madratter

Madratter

    Hubble

  • *****
  • topic starter
  • Posts: 13,277
  • Joined: 14 Jan 2013

Posted 22 November 2015 - 12:34 PM

I see many many posts asking about guiding. I also see a lot of advice, some good, some middling, and some very bad. I think part of the reason for this is that people don't have a good understanding of what they are actually trying to achieve!

 

Many of the problems that are attributed to guiding are not actually due to guiding. This is especially true for those who are using guide scopes instead of Off Axis Guiders. There is inevitably some differences in flexure between the guide scope/camera and the main scope/camera. This is called differential flexure and if you have a guide scope your images suffer from it. It may be very small but unless you really know what you are doing, it is much more likely the effect is large. Since beginners usually use a guide scope, their images usually suffer from this to some extent. The way you can tell is to blink through a series of original, unregistered subs. The stars will be moving from sub to sub in the direction of the elongation you are seeing in the stars.

 

Other common sources of elongated stars are collimation problems, and tilt in the imaging system.

 

Unless doing adaptive optics like guiding (Frank the author of Metaguide is a proponent of this), what guiding is good at is eliminating slow movements and drifts. What it is NOT good at is eliminating fast fluctuations. I have explained that before but I'm going to repeat it here.

 

Seeing (or other rapid fluctuations) can temporarily displace a guide star (just like watching the bottom of a pool or pond through ripples on the surface of the water). That displacement is temporary and will change relatively quickly. So say that seeing has displaced the star 1 arc-second from where it is on average. If you have aggressive settings (say 100% aggression), the mount will be told to move to get the star back in the average position. Often times with seeing the star will have already gone back to the center on its own! It is even possible it has moved to the other side of center. Now we have moved the mount unnecessarily.

 

So with no guiding, we were 1 arc-second off and the star has now returned to center. But if we were guiding, the star was 1 arc-second off and now is one arc-second off in the other direction. Even worse, consider if the star now moves back to the original position due to seeing. That plus the one arc-second correction in the other direction and we are now two arc-seconds off leading to a huge correction that can cause even further problems. By guiding on these rapid fluctuations we have made things worse, not better.

 

If you don't understand the above, go back and reread it until you do because understanding it is crucial.

 

There are many causes of slow drift with our astrophotography systems. Among them are polar misalignment, refraction, flexure in the system, periodic errors of the mount, etc. So when guiding, you want to setup the parameters of whatever guiding program you are using so that it detects these slow drifts but ignores fast fluctuations. In order to know HOW fast you need to correct problems, it is good to know how long you can go without guiding and still get round stars. Typically mounts at typical focal lengths can often do 30 seconds. Better mounts with periodic error correction, corrections for some of the other problems like refraction, etc. may be able to go 5 minutes or even longer.

 

So here is what I want you to understand. If you can go 30 seconds without guiding and still get round stars, why on earth do you feel it is necessary to correct problems that are showing up on a second by second basis?

 

Here are some overall recommendations:

1) Slow down you guiding exposures to be multiple seconds. How long you can go will depend on the quality of your mount but even a poor mount can probably go 4 seconds or even longer. This has the side benefit of giving you more guide stars with better signal to noise ratio. Doing this helps average out seeing and other fast fluctuations that you would just be chasing.

 

2) Turn DOWN your aggression settings. These settings typically are used so that only some of the error seen in actually corrected. In the case of declination, Very low values can be used as long as you mount will hold the star steady about some line that does NOT have to be zero on the guide graph. In the case of the RA axis, you may need somewhat higher aggression settings than the declination axis due to periodic error, especially with lower quality mounts. However, typically people still run with more than they need.

 

3) If your guide program has hysteresis settings, consider using them. They factor in the direction that drift has been trending. Again, this is especially helpful with the declination axis.

 

4) The typical advice is to use minimum motion settings close to the RMS error you are having on an axis. I'm going to go against this common advice and recommend trying lower. These settings typically help BECAUSE people have their aggression settings too high, and on their declination axis because they have their hysteresis values too low.

 

Here is an example guide graph following these recommendations. The guide camera exposure time was 4 seconds. Polar alignment error was less than two arc-minutes so drift was reasonably slow. This was a high quality mount so you might not be able to go quite this far. But you might be surprised how far you can take it, especially in declination, even with a lower quality mount, and how it will improve your guiding.

 

GoodGuidingSoulNebula20151103.PNG

 

(Note: On this guide graph - my minimum motion settings were up at levels more commonly recommended.)

 

For what it is worth, I'm located on the East Coast of the U.S.A. That is NOT an area that has good seeing on a frequent basis.


Edited by Madratter, 22 November 2015 - 12:37 PM.


#2 nitegeezer

nitegeezer

    Galactic Ghost

  • *****
  • Posts: 7,691
  • Joined: 27 Nov 2007

Posted 22 November 2015 - 12:46 PM

This just got added to the Best of threads here.  

 

Reading this explanation has helped things click for me.

 

Thank you MR!   :bow:



#3 rigel123

rigel123

    ISS

  • ****-
  • Posts: 24,963
  • Joined: 29 Jun 2009

Posted 22 November 2015 - 01:03 PM

Great information to go by.



#4 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 26,034
  • Joined: 10 Jan 2014

Posted 22 November 2015 - 01:41 PM

Just to add an alternative interpretation of minimum motion here...while I agree at a high level with MR's approach, not everyone has the same sky conditions, and not every approach always works the same for everyone. I have tried the long hysteresis approach myself over about four nights during the last time I was able to image, and I was unable to get the same kind of results a MR here.

 

Quick note about minimum motion. This could be thought of as low pass filtering as MR has stated in the past, however it could also be thought of as guide pulse throttling. With a small minimum motion, your going to generate guide pulses more frequently. IF the hysteresis approach works for you (in my case, I've tried it many times and I am not able to use a hysteresis as large as 50 as MR has done here), then lots and lots of small guide pulses might be find. If the hysteresis approach does not work, then having a small minimum motion means you could get more guide pulses than are likely necessary. The goals are ultimately the same as MR has stated so nicely here:

 

If you can go 30 seconds without guiding and still get round stars, why on earth do you feel it is necessary to correct problems that are showing up on a second by second basis?

 

 

When a long hysteresis results in worse guiding performance than a shorter hysteresis, the other option is to use a larger minimum motion such that small star movements, and preferably most star movements, are ignored by PHD and only those that deviate enough to START affecting your star quality in your subs are actually corrected. I still agree with MR that LOWER aggression should generally be used, even with a larger minimum motion. You still don't want to over-correct, you want to guide. If your corrections are resulting in the star jumping back and forth around the midline in the graph, then your too aggressive.

 

Some hysteresis will usually help, as it will account for the recent trend in general star motion (i.e. drift), but if you have very good PA or just have wild stars, a long hysteresis may not always work. I find myself at 10-15 most of the time in my DEC axis with PA of around 30"-1'. I've tried as high as 50-60 and a small minimum motion like MR recommends here, (on several occasions), and my RMS went from an average of 0.6" to 1.25-1.8" until I dialed hysteresis back and increased minimum motion again. With a larger min. mo. I can often get away with a single guide pulse in DEC every 10-15 guide periods, which with 3-4 second guiding is maybe once a minute. 


Edited by Jon Rista, 22 November 2015 - 01:44 PM.


#5 entilza

entilza

    Soyuz

  • *****
  • Posts: 3,826
  • Joined: 06 Oct 2014

Posted 22 November 2015 - 01:50 PM

Great thread.

 

With 4 seconds of exposure how quickly does this graph settle?  Ie. after first getting ready to guide and you try these settings how long should one wait to see for it to settle before playing with the settings?  (1 minute, 2 minutes?)

 

I also see in your graph DEC corrections on one side only, this seems to be the ideal situation.  However it also appears there is a correction almost ever instance.  Any comments on this?  Should the weight of the DEC axis always be lens heavy as well?

 

Thank You.



#6 nitegeezer

nitegeezer

    Galactic Ghost

  • *****
  • Posts: 7,691
  • Joined: 27 Nov 2007

Posted 22 November 2015 - 02:19 PM

 

I also see in your graph DEC corrections on one side only, this seems to be the ideal situation.  However it also appears there is a correction almost ever instance.  Any comments on this?  

 

I will make a comment here, although please consider the source!!  As far as DEC guiding goes, I don't think it matters if you are getting guide pulses all the time.  If the DEC is way off but flat that will be fine.  Looking at MR's chart he was running at .5" +-.5".  Even if it were up at 2", if it is still at +-.5" that is the important part.  The 2" is not a likely case, but just emphasizes that the variation is the important part.



#7 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 26,034
  • Joined: 10 Jan 2014

Posted 22 November 2015 - 02:26 PM

With 4 second exposures using an AT8RC and OAG on an Orion Atlas (belt modded), I have found it takes about 20-30 seconds for my guiding to settle. That is about 5-6 guide periods. That is in line with when I was guiding the 600/4 with shorter exposures...it would take 5-8 seconds to settle with 1-1.5s exposures. That is with my settling factor at 0.65 rather than 0.25 in BYE, and the minimum cooldown of 1 second. The thing about settling...if it is taking you up to two minutes to settle, then your trying to settle too low. You need to tweak your settling factor for your skies. To settle at 0.25 you need to have a pretty smooth graph with few star motions. If you have a choppy graph, then you should be settling at a higher factor. You can't get better than your average star motion, so it's wasteful to bother trying.


Edited by Jon Rista, 22 November 2015 - 02:28 PM.


#8 entilza

entilza

    Soyuz

  • *****
  • Posts: 3,826
  • Joined: 06 Oct 2014

Posted 22 November 2015 - 02:35 PM

Jon, Regarding settling I see what you mean you are referring to the settling factor in BYE or SGP.  What I meant here was after initially setting up guiding how long does it take to get this graph to smoothen out with 4 second images.  



#9 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 26,034
  • Joined: 10 Jan 2014

Posted 22 November 2015 - 02:42 PM

In my case, and I don't know if this is the case for everyone, when I dither, I don't usually see any significant perturbation in the graph. PHD will pause the graph, dither (offset the lock position), then restore the star to lock position, with the graph paused. It will use a maximum-duration guide pulse to return the star to lock, and I never see that. I may see a slightly higher star movement after that, or if it takes longer than 2 seconds to move the star to the new lock I might see part of that initial movement. Usually, my graph looks the same before, during, and after a guide pulse because PHD isn't actually recording the larger change in position when it dithers (which, fundamentally, just means that PHD changed the lock position for the star by a few arcseconds, then moved the star from it's previous position to the new lock position.) I let it "settle" to make sure that the gears in the mount have truly settled in...but, I rarely actually ever see anything significant in the graph. I don't dither super aggressively these days anymore though, and I also only dither every few short subs (if I'm using 300s subs, I may go two to three subs between dithers), or only once every 20-30 minutes when using long subs. I dither enough to move the star centroids by a few pixels, but that's all. So my gears probably don't need a lot of settling in either...if your dithering very aggressively, instead of 5 guide periods you may want 10 or so (which would be maybe 40-60 seconds to settle.)


Edited by Jon Rista, 22 November 2015 - 03:09 PM.


#10 Madratter

Madratter

    Hubble

  • *****
  • topic starter
  • Posts: 13,277
  • Joined: 14 Jan 2013

Posted 22 November 2015 - 02:57 PM

Great thread.

 

With 4 seconds of exposure how quickly does this graph settle?  Ie. after first getting ready to guide and you try these settings how long should one wait to see for it to settle before playing with the settings?  (1 minute, 2 minutes?)

 

I also see in your graph DEC corrections on one side only, this seems to be the ideal situation.  However it also appears there is a correction almost ever instance.  Any comments on this?  Should the weight of the DEC axis always be lens heavy as well?

 

Thank You.

 

Thanks. The guiding corrections being on one side only are a direct result of my low aggression settings and high hysteresis settings. The fact it is correcting basically every guide exposure is fine because what it is correcting out is the slow drift with these settings.

 

As for whether the balance should be lens heavy, that depends a LOT on your particular system. With my MyT I was best off with a perfectly neutral balance. With my Atlas I was best off front heavy.



#11 Agatheron

Agatheron

    Ranger 4

  • -----
  • Posts: 390
  • Joined: 03 Nov 2010

Posted 22 November 2015 - 04:53 PM

This is excellent, thank you for posting this. I've been one of the culprits showing bad calibration. I have a clear evening ahead, so I'm going to give this a try.



#12 Madratter

Madratter

    Hubble

  • *****
  • topic starter
  • Posts: 13,277
  • Joined: 14 Jan 2013

Posted 22 November 2015 - 07:55 PM

Here is a practical application of the above with a mount that doesn't cost 6000$. This is with my Orion Atlas on a night with wretched seeing.

 

ClearSkyChart.PNG

 

GuidingAtlas20151122.PNG

 

The RA axis isn't guiding quite as well as the Declination Axis here. That is attributable in large part to the fact I'm not running PEC on the Atlas. I could probably bring those numbers down somewhat by either:

1) Training PEC.

2) Bringing the aggressiveness up a bit more.

3) Bringing the exposure time down somewhat.

 

Still, this is not a bad graph and would be alright for guiding my AT8RC at 1190mm. And again, remember just how bad that clear sky chart is for the seeing.



#13 Thirteen

Thirteen

    Gemini

  • *****
  • Posts: 3,192
  • Joined: 12 Jul 2013

Posted 22 November 2015 - 08:11 PM

Granted I've been using an AVX but the disconnect between the RA RMS and the DEC RMS has me sometimes wishing my DEC guiding was worse just to round out the stars.  I commonly get the target graph that shows the straight line across the bullseye due to the "noise" in the RA axis.  I see this pretty regularly when people post guide graphs and I see it in this the one posted above.  I suppose if the total RMS value is within your seeing and image scale limits you may not even see it, but I see elongation due to this.  

 

I'd love to know how people commonly deal with it, or if its just a matter of getting your image scale double or triple the RMS of your guiding?

 

By the way, thanks for the opening of this topic.  It's definitely useful dialogue.


Edited by Thirteen, 22 November 2015 - 08:14 PM.


#14 Madratter

Madratter

    Hubble

  • *****
  • topic starter
  • Posts: 13,277
  • Joined: 14 Jan 2013

Posted 22 November 2015 - 08:20 PM

I've done the math here before, but unless your seeing is better than about 2 arc-seconds, the differences above make only a minor difference in eccentricity values.

 

I could deal with it as above, or as you mentioned by intentionally worsening the guiding in declination. I personally think that is somewhat wrong headed, but it is an approach.



#15 Jon Rista

Jon Rista

    ISS

  • *****
  • Posts: 26,034
  • Joined: 10 Jan 2014

Posted 22 November 2015 - 08:46 PM

If you have very good seeing as MR states, then really the thing to do is not make DEC worse, but make RA better. In my case, on the rare occasions when I get 2" or better seeing, my samples in the PHD target will cluster around the RA axis and barely deviate in DEC. I get mostly a rectangular distribution of centroid samples. In those cases, I could try increasing min. mo. or increase aggression...but the thing is...the star isn't really moving much in DEC. Really, it isn't moving much in RA either....my problem is my mount's PE, and that is primarily what is being corrected. So the solution to the elongation problem is probably for me to program PEC (which I don't do often, although now that I am more permanently set up in my back yard, I may well do just that). With PEC, my ~28" PE should drop, and the distribution of samples in RA should tighten up. 



#16 Madratter

Madratter

    Hubble

  • *****
  • topic starter
  • Posts: 13,277
  • Joined: 14 Jan 2013

Posted 22 November 2015 - 08:50 PM

Here is the first sub of the night at 100%. The stars are nice and tight with no visible elongation. Grant you this is with an image scale of 2.28 arc-seconds. But you would be surprised how little difference it would make if I was using the AT8RC.

 

SH 2-208 GuidingExample.jpg



#17 Thirteen

Thirteen

    Gemini

  • *****
  • Posts: 3,192
  • Joined: 12 Jul 2013

Posted 22 November 2015 - 08:51 PM

If you have very good seeing as MR states, then really the thing to do is not make DEC worse, but make RA better.

 

I wish it were that simple.  My mount is just noisy in RA. 



#18 Phil Hosey

Phil Hosey

    Vanguard

  • *****
  • Posts: 2,335
  • Joined: 29 Oct 2008

Posted 22 November 2015 - 08:55 PM

Thanks Madratter.  I just made some changes to my settings based on your post and it improved my RMS numbers quite a bit.  I'm running about .4" on both axis now which is almost half what it was.  I increased my exposure to 3 seconds (it was 1 second) and reduced my Dec aggressiveness.



#19 Madratter

Madratter

    Hubble

  • *****
  • topic starter
  • Posts: 13,277
  • Joined: 14 Jan 2013

Posted 22 November 2015 - 09:03 PM

 

If you have very good seeing as MR states, then really the thing to do is not make DEC worse, but make RA better.

 

I wish it were that simple.  My mount is just noisy in RA. 

 

 

If your mount is indeed noisy in RA and it isn't just PEC, that is just plain problematic. Again those are the kind of fast changes that don't guide out well. You can try upping the aggressiveness somewhat or lowering the exposure time, but if the problem is truly as you state, it probably won't help much if at all.

 

Thanks Madratter.  I just made some changes to my settings based on your post and it improved my RMS numbers quite a bit.  I'm running about .4" on both axis now which is almost half what it was.  I increased my exposure to 3 seconds (it was 1 second) and reduced my Dec aggressiveness.

 

You're welcome.

 

I do need to mention something here for those who dither. I don't care if the declination axis develops a bias (is not centered on the 0 line). But this can matter when dithering depending on how you have the settings for settling set. I personally have mine set pretty liberally (without bad effect) (2 pixels 0 seconds in SGP). But if you feel you must be more stringent, you might need to up the aggressiveness settings in Declination a bit to eliminate that bias.



#20 17.5Dob

17.5Dob

    Voyager 1

  • *****
  • Posts: 10,267
  • Joined: 21 Mar 2013

Posted 22 November 2015 - 09:29 PM

Here is a practical application of the above with a mount that doesn't cost 6000$. This is with my Orion Atlas on a night with wretched seeing.

 

attachicon.gifClearSkyChart.PNG

 

attachicon.gifGuidingAtlas20151122.PNG

 

The RA axis isn't guiding quite as well as the Declination Axis here. That is attributable in large part to the fact I'm not running PEC on the Atlas. I could probably bring those numbers down somewhat by either:

1) Training PEC.

2) Bringing the aggressiveness up a bit more.

3) Bringing the exposure time down somewhat.

 

Still, this is not a bad graph and would be alright for guiding my AT8RC at 1190mm. And again, remember just how bad that clear sky chart is for the seeing.

Now that I did a complete weight check/ re-balance for my 8" f4, the past two times I've been out, (and both were "ALL NIGHT"), my graphs were better than that, for 16 hrs + using a moderately high aggression and almost no hysteresis. For 20-30min I almost thought I'd targeted a hot pixel for a guide star,it was so flat.

This gives me hope there's room for even further improvements without having to belt mod my Atlas "yet"..


Edited by 17.5Dob, 22 November 2015 - 09:35 PM.


#21 Phil Hosey

Phil Hosey

    Vanguard

  • *****
  • Posts: 2,335
  • Joined: 29 Oct 2008

Posted 22 November 2015 - 09:38 PM

Thanks Madratter.  I just made some changes to my settings based on your post and it improved my RMS numbers quite a bit.  I'm running about .4" on both axis now which is almost half what it was.  I increased my exposure to 3 seconds (it was 1 second) and reduced my Dec aggressiveness.

 

Well, I was enjoying guiding better than I've had it for a while now, then this happened.  Of course this has nothing to do with changing the settings, but it is something I've been struggeling with lately.

Attached Thumbnails

  • PHD2 Error.jpg


#22 entilza

entilza

    Soyuz

  • *****
  • Posts: 3,826
  • Joined: 06 Oct 2014

Posted 22 November 2015 - 10:05 PM

Phil you are using PHD2dev5 i would upgrade to dev7 or just use the base 2.5.0.



#23 qbool

qbool

    Viking 1

  • -----
  • Posts: 624
  • Joined: 27 Mar 2015

Posted 22 November 2015 - 10:13 PM

Thanks Madratter. I usually use a low or no hysteresis with good, but not great, typical guiding (~1.0-1.2" p2p w/image scale of 2.67"). Using a lower minmo+higher hyst does seem to be producing smoother guiding for me... at the moment. Of course, guiding is a moving target, so this probably won't last long: :D

 

guiding.jpg

 



#24 ChrisWhite

ChrisWhite

    Overcast Observatory

  • *****
  • Vendors
  • Posts: 8,864
  • Joined: 28 Feb 2015

Posted 22 November 2015 - 11:07 PM

Awesome thread!  Thanks MR!!!!



#25 CharlesW

CharlesW

    Long time member

  • ***--
  • Posts: 3,659
  • Joined: 01 Nov 2012

Posted 23 November 2015 - 01:12 AM

I believe one thing that often gets overlooked is the quality of focus of the guide camera. We put considerable effort into making sure our imaging cameras are focused to within a few microns but we just eyeball the focus of the guide camera.

 

I would suggest that if your min and max movement settings are correct and have tweaked aggression and are still not happy, you might find a very bright star, throw a Bahtinov on your scope and sharply focus your guide camera. Even if you are happy with your guiding you might try that and see if it improves. You can always defocus. 




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