What Constitutes Acceptable Guiding?

[gnomus]

It's time to come clean and admit that I have no idea what I am doing. 

Although I have been guiding for a few months, I don't really know if my guiding is any good or not.  I only recently found PHD2 Log Viewer.  I then took a bit more time looking for the logs themselves.  Anyhow I attach a screenshot from my last session's guiding (click for a full-screen view).

Am I correct in thinking that this is telling me that my "average" (RMS) guiding accuracy for the session was +/- 0.45 of a pixel in RA and +/- 0.54 of a pixel in Dec?  That doesn't seem too bad to me (if I've understood correctly).  On the other hand I am getting as much of a swing as 1.65 pixels in RA and 2.21 pixels in Dec (if I have understood this graph).  I'm less confident about that degree of accuracy.

Is this good enough or do I need to be getting better guiding than this?  The image itself can be found here, by the way - http://stargazerslounge.com/topic/248909-first-attempt-with-new-ccd-pelican-nebula/

To me the stars seem round enough (but I am just a beginner).  The overall image is a bit softer than some of the better images I have seen here.  I don't know if that is down to guiding, my focussing, the quality of my scope (ED80),  seeing on the night, or poor processing on my part.  

In any event, the main question is am I good enough on the guiding?

Thanks 

Steve

[Thalestris24]

Hi Steve

I guess getting small, round sharp stars is the acid test. You will likely have guiding fluctuations superimposed on periodic error. My own guiding tends to fluctuate between about +/- 4 arcsecs and this seems to largely a consequence of atmospheric conditions since I occasionally get better guiding on dry, clear nights - though they are quite rare here! Also, being in a city there are lots of thermals . Drift aligning has helped my guiding too, as has getting good balance (on my AVX).

Your Pelican image looks pretty good to me - maybe a little soft? I'm sure someone more expert than me can say. Not sure why you're getting the spikes in dec - could be a balance issue, cables dragging? Could also be some flex in the system. Backlash can also cause problems - how is your calibration? Any warnings? Maybe also look at the guiding graph over a shorter timescale. Also have a look see how things behave with guiding off. That will show how much drift you have.

Hope that helps a bit

Louise

[gnomus]

Thank you for responding Louise.  It's reassuring to know I am not a million miles away.  

It could be a balance issue.  I got my new CCD and wheel quite recently.  That is quite a step up in weight and in cabling from what I had before.  I am pretty sure, however, that I have everything balanced out as regards the mount - the scope will hold itself in any position that I put it in with clutches released.  I did get some counterweights for the front of the scope, because otherwise I had to have the telescope sitting way forward in the dovetail (such that about a third of saddle wasn't making contact with anything).  To get perfect balance I have to have the filter wheel hanging out to the East (as the scope points North).

Cabling is another matter.  I have used velcro ties to get everything into one bundle.  I then feed this up the scope and have it hanging over the Saddle knobs (and attached with more velcro ties).  When I was stripping everything down the other day, it occurred to me that it might be a good idea to run all the cables through the hole in the middle of the CGEM mount.  Is that what that hole is for, or would I be making a terrible mistake?

My ST80 is now rigidly affixed to the top of the ED80.  However I need to use an extension tube in the ST80 to get focus with the ZWO guide camera.  There could be some flexure there - if there is I'm not sure what I can do about it.  

My polar alignment might be off.  I set up the mount using a polar scope and an App that gives the position of Polaris on the circle in the polar scope.  When I first started doing this, I would then try drift aligning and found that it was pretty much bang on.  As such, I have tended to rely on the accuracy of my "rough" alignment.  I will pay more attention to this next time - it was certainly easier to get Polaris in the right place when that App was telling me it should be at around 6pm on the circle (as it was in the winter) compared with 11pm now.  I'm thinking of getting that Alignmaster software to help me with polar alignment.

I haven't PEC trained the mount yet, because I've never been 100% confident of my polar alignment - I wouldn't want to make things worse.

That's an excellent tip of turning guiding off to check for drift.  I'll definitely do that next time.

In some ways, I am surprised at how much I know now compared with only a few months ago.  On the other hand, I am becoming increasingly aware of the stuff I don't know.  A few months ago I didn't how much I didn't know.

Thanks for responding.  Does anyone else have any other thoughts?

Steve

[Thalestris24]

Hi again

Sounds like you're pretty well set up. So long as none of the cables are 'pulling' on anything when you're tracking on a target. You should be ok with an extension tube on the ST80 - easy to check by hand if there's any obvious looseness. It's a good idea to make use of the focus lock. I've never bothered with PEC - can't see that there's anything to be gained by it. Your graph would look a lot better if you changed the scale!

Small PA errors aren't usually a problem - you'd see things drift quickly if you had a big PA error and guiding off. Your graph looks fairly symmetrical about the x-axis. It's just those spikes in DEC that look a bit odd (at the scale you're using). It's probably backlash but I'm not sure. Maybe any backlash you might have is something you can tweak with your hand controller (you can with the AVX/Nexstar+)- I keep meaning to do it myself! Certainly with the AVX it's also possible to access the motors and make small adjustments to minimise any backlash of the motor gears (also on my todo list! - I have the YouTube video!).

It's a bit quiet on here at the moment - summer hols, I guess. Hopefully, someone else will chime in.

Cheers

Louise

[gnomus]

I dug out my log for the evening a few days before the one posted above.  It is the same object as before (Pelican) at a similar time of night.  This graph seems overall to be a little better than the first one, although I am still getting rather large (and seemingly random) swings in Dec.  I wonder what could be going on?

I should add that I did not dither on either night.

[Garethr]

Gnomus,

Don't get to fixated on the graph. What are the image like?

Gareth.

[Garethr]

Gnomus,

Sorry just realised you had posted a pic in the opening post. Looks pretty damn good to me, only thing I can think of is you mention perfect balance when you release the clutches. As I understand it most people have a slight inbalance to ensure the gears /cogs are always engaged to prevent any backlash

Gareth

[Thalestris24]

Hi Steve

I just noticed that you are using 'resist switch' for your dec guiding algorithm. It might be better to use 'hysteresis' and, if possible, guide in one direction only. I take it that where you have the dec spikes they don't correspond to losing the guide star?

Louise

[gnomus]

Gnomus,

Sorry just realised you had posted a pic in the opening post. Looks pretty damn good to me, only thing I can think of is you mention perfect balance when you release the clutches. As I understand it most people have a slight inbalance to ensure the gears /cogs are always engaged to prevent any backlash

Gareth

Thanks.  Yes - sorry I should have said that I get everything balanced as best I can.  Once I have done that I set the scope up so that it is to the East of the mount and move the counterweight so that the rig is slightly scope-heavy.  I think that is correct.   

[gnomus]

Hi Steve

I just noticed that you are using 'resist switch' for your dec guiding algorithm. It might be better to use 'hysteresis' and, if possible, guide in one direction only. I take it that where you have the dec spikes they don't correspond to losing the guide star?

Louise

Thanks again Louise.  I have no idea what these terms mean, but am happy to try them.  Would you be able to direct me to somewhere that could explain all of this (I prefer to understand the what and the why)?  I do remember at one time having the Dec guide direction set to North only - and then one night (not either of the nights posted) the Dec fell off the bottom of the graph pretty quickly - I assume due to my PA being quite off that night.  I set it to Auto after that.  

I'm pretty sure that I didn't lose the guide star, because the nights were very good.  I was, however, in the house for most of the time and it is possible that a wisp of cloud passed by (I do have Team Viewer but I am not monitoring yet, only popping out every half hour or so).  Loss of guide star would explain why the spikes seem so random, but would it explain why I seem to be getting spikes in Dec only?

Thanks again for all your assistance with this - it really is much appreciated.

[carastro]

I think people get too hung up on the guide graph, all that really matters is if the stars come out round.  I have been imaging for 5 years or more now and to be honest I don't have much of a clue about the technicalities either, which is why I never switched to PHD2 all those spikes really look quite wrong to me compared to the guide graph you get in PHD1 where I know so long as my lines don't pass above the first line each side of the central line I will be OK.  

PHD1

[gnomus]

Thanks Carastro. Do you happen to know what the scale is on that graph? I see that your RMS value was significantly better than mine.

[Kevstew]

Theres a software update for the mount motor controllers that fixes a dec cogging guiding issue. On my cgem with a nexstar+ hand controller you can view the software level in one of the menus. It will look like this

Hc gem 5.26 5080 (hand controller software)

Mc 6.50 6.50 ( motor controller software)

[ollypenrice]

I think people get too hung up on the guide graph, all that really matters is if the stars come out round.  I have been imaging for 5 years or more now and to be honest I don't have much of a clue about the technicalities either, which is why I never switched to PHD2 all those spikes really look quite wrong to me compared to the guide graph you get in PHD1 where I know so long as my lines don't pass above the first line each side of the central line I will be OK.  

PHD1

That's a good trace, Carole, but round stars are not an adequate test of guiding precision. If you have randomly distributed errors on both axes you'll get round stars but resolution is being lost to what might still be large errors. We found this while initially setting up the Mesu. We had round stars from the start but as we tuned the guiding they became smaller and smaller and, of course, the resolution in the rest of the image improved by the same amount.

The hard information needed to asses guiding is the error in arcseconds at the guider and the number of arcseconds per pixel at the imaging camera. The guiding error in arcseconds must, at the very least, be smaller than the pixel scale at the imaging camera. In order to resolve to the limit of the system the guiding must be comfortably 'sub pixel.'

Since the focal length of the guider and the size of the guide pixels are not likely to be the same as those of the imaging scope and camera we need to use a unit common to them both - and that is arcseconds on the sky. So firstly we just work out the arcseconds per pixel of both guider and imager, then look at the error in arcseconds at the guider. Is this error in arcsecs less than the arcsecs per pixel of the imaging camera? If so, good. If not, you are not resolving to the limit of the imaging rig. Getting the imaging rig error down to well below a pixel is much to be preferred, of course. 

A guide trace on its own is pretty well meaningless. If you want a really nice guide trace bin your guide pixels 4x4. It will make your guiding worse but the trace will look lovely! OAG traces, which are at the same FL as the imaging scope, always look awful compared with those of short FL guidescopes but that's to be expected.

Olly

[glowingturnip]

Is this error in arcsecs less than the arcsecs per pixel of the imaging camera?

Olly

hadn't actually thought about it that way - after over three years of struggling, I'm now finally getting down to about 0.5-0.6" RMS error in guiding, and I'm imaging at 1.42"/pixel.  This makes me very happy 

Gnomus - I suspect your spikes in the Dec guiding are a combination of polar misalignment plus backlash on the Dec axis.  The drifts higher would be drift from the polar misalignment while the Dec is in backlash and not driving the scope, and then the jump lower is when the backlash finally catches and it corrects all in one go.  It does look like quite a high rate of drift though, how far off was the polar align ?  As Louise said, if you turn the guiding output off for a bit then you can just watch the drift, does it disappear off the screen ?

Turning the Dec guiding from auto to either North or South only can help in situations like this, since it will never allow the mount to introduce backlash but will just stay on one side of the gear.  It varies from session to session whether it's N or S though, so you have to experiment - turn Dec guiding off and see which way it drifts, then try N only, if there are corrections are opposite to the drift, then use that, otherwise, try it with S only.  The disadvantage though is if you get a jump in Dec the wrong way for whatever reason then it can take ages to drift back.  Same applies if you dither, it might not be able to adjust to the new dither position.  Better to get a sharp polar align really.

0.7-0.8" RMS error on your second trace is pretty good though, and I do like your Pelican

[Thalestris24]

Thanks again Louise.  I have no idea what these terms mean, but am happy to try them.  Would you be able to direct me to somewhere that could explain all of this (I prefer to understand the what and the why)?  I do remember at one time having the Dec guide direction set to North only - and then one night (not either of the nights posted) the Dec fell off the bottom of the graph pretty quickly - I assume due to my PA being quite off that night.  I set it to Auto after that.  

I'm pretty sure that I didn't lose the guide star, because the nights were very good.  I was, however, in the house for most of the time and it is possible that a wisp of cloud passed by (I do have Team Viewer but I am not monitoring yet, only popping out every half hour or so).  Loss of guide star would explain why the spikes seem so random, but would it explain why I seem to be getting spikes in Dec only?

Thanks again for all your assistance with this - it really is much appreciated.

Hiya

This is worth reading through: http://openphdguiding.org/PHD2-2.4.1i-manual.pdf

Louise

[carastro]

Err, that wasn't my guide graph it was just an example I found to show you what PHD1 looks like.

Carole 

[MattJenko]

If you set to guide on one direction only, what happens when the mount flips, won't it be incorrect, or can PHD handle that case?

[Freddie]

Carole, Olly's comments still apply though. As said, round stars are not a good measure of accurate guiding.

[Thalestris24]

If you set to guide on one direction only, what happens when the mount flips, won't it be incorrect, or can PHD handle that case?

Yes, PHD2 has a box to tick for that!

Louise

[Thalestris24]

If anyone is interested, the latest phd2 development build apparently includes a tool for measuring dec backlash and provides a means for backlash compensation though I expect the latter just duplicates what most handsets can do.

Louise

[gnomus]

...

The hard information needed to asses guiding is the error in arcseconds at the guider and the number of arcseconds per pixel at the imaging camera. The guiding error in arcseconds must, at the very least, be smaller than the pixel scale at the imaging camera. In order to resolve to the limit of the system the guiding must be comfortably 'sub pixel.'

Since the focal length of the guider and the size of the guide pixels are not likely to be the same as those of the imaging scope and camera we need to use a unit common to them both - and that is arcseconds on the sky. So firstly we just work out the arcseconds per pixel of both guider and imager, then look at the error in arcseconds at the guider. Is this error in arcsecs less than the arcsecs per pixel of the imaging camera? If so, good. If not, you are not resolving to the limit of the imaging rig. Getting the imaging rig error down to well below a pixel is much to be preferred, of course. 

....

Thanks to everyone who has replied.  Things are becoming clearer.  Can I check that I have understood this point that Olly makes?......

I entered my guide scope (ST80) and camera (ZWO ASI120) data into PHD2, so the values given for RMS arcsecond deviations are accurate.  For my first trace these were 0.87" in RA and 1.04" in Dec.  Of course these are the figures for my guide scope and camera - the important thing is what this means when using my telescope (ED80 with 0.85 reducer) and CCD (Atik 383L).  I put these into FLO's Astronomy Tools calculator.  This told me that this CCD/Scope combination produces a resolution of 2.18" x 2.18" per pixel.  Since both 0.87" (RA) and 1.04" (Dec) are smaller than 2.18" (imaging resolution), my guiding is adequate (even if it is not very pretty).

Is this correct or is my logic skew-whiff somewhere?

Steve

[ollypenrice]

Thanks to everyone who has replied.  Things are becoming clearer.  Can I check that I have understood this point that Olly makes?......

I entered my guide scope (ST80) and camera (ZWO ASI120) data into PHD2, so the values given for RMS arcsecond deviations are accurate.  For my first trace these were 0.87" in RA and 1.04" in Dec.  Of course these are the figures for my guide scope and camera - the important thing is what this means when using my telescope (ED80 with 0.85 reducer) and CCD (Atik 383L).  I put these into FLO's Astronomy Tools calculator.  This told me that this CCD/Scope combination produces a resolution of 2.18" x 2.18" per pixel.  Since both 0.87" (RA) and 1.04" (Dec) are smaller than 2.18" (imaging resolution), my guiding is adequate (even if it is not very pretty).

Is this correct or is my logic skew-whiff somewhere?

Steve

We've had a family sorrow this week so I'm not really on top of things but in principle I think you're right. Your numbers look nicely sub pixel.

Olly

[gnomus]

We've had a family sorrow this week so I'm not really on top of things but in principle I think you're right. Your numbers look nicely sub pixel.

Olly

I'm sorry to hear that.  Thanks for taking the time to respond to my insignificant, little non-problem.

[gnomus]

Hi Steve

I just noticed that you are using 'resist switch' for your dec guiding algorithm. It might be better to use 'hysteresis' and, if possible, guide in one direction only. I take it that where you have the dec spikes they don't correspond to losing the guide star?

Louise

I ran the mouse over the spikes in PHD Log Viewer.  Mass (m) and SNR stayed pretty constant throughout the graph.  I don't think, therefore, that this was a loss of guide star issue.