How do autoguiders work?

[Rich1980]

From looking around the forum I gather autoguiding is very popular and bordering on essential for DSO photography but I'm wondering how does it work? 

Am I right thinking you point it at your chosen subject and it adjusts to any movement from that subject? 

If that is the case, does a perfect polar alignment matter or will the autoguiding adjust for that too? 

Forgive me too if this is a daft thread as it's all new to me!

[Davey-T]

No daft questions, the intricacies of auto guiding can drive you to drink 

You've got the general idea and perfect polar alignment isn't needed, the longer the focal length of your imaging scope the more demanding guiding becomes on the mount and general equipment.

Dave

[vlaiv]

It works by having second camera / telescope (or attachment to primary scope called OAG - off axis guide) that monitors position of a star (called guide star).

If that position changes by small amount - it adjusts pointing of the telescope to compensate.

These corrections are done in short time period - order of a few seconds.

You still need good polar alignment and all the rest when doing auto guiding as polar alignment error causes more than tracking issues - there is field rotation as well. Polar alignment does not need to be very strict in case of guiding - but it is beneficial to get it right.

[Hallingskies]

NO-ONE knows how autoguiding works.  It is all done by magic.  Just look at the number of threads and opinions on autoguiding.  If there were any science there it would work for everyone, first time, straight out of the box....😃😃🤪

[Stub Mandrel]

Your basic understanding is correct.

Some people let the polar alignment be slightly out, as that tends to mean most corrections are in the same direction, minimising backlash (mostly an issue in DEC as guiding in RA is usually an issue of varying drive speed but not direction).

My own experience is that the better the PA I get, the better the guiding.

The clever bit is that the software measures the relative brightnesses of all the pixels in the star and uses these to average out the centre, this can be done accurately to a  surprisingly small fraction of a pixel. This allows quite small, short-focal length scopes to be used as guiders and still allow tracking well enough for much longer scopes.

 

Software wise I use PHD2, I now get consistent results, with the actual accuracy usually determined by the transparency and seeing. This meant 'tuning' the mount and experimenting with different guiding algorithms to see which worked best with my setup.

I think I can be safe in saying that what works with me may not work as well for someone else as you need to match the way the software works to the hardware  - we are asking the scope to point with an accuracy to shooting a bullet through the earth into a target the size of a football field, so tiny differences in setup mean measurable differnces in how teh rig reacts to guiding.

Good news is that the number of choices is reduced by the ability of guiding software to learn how your rig reacts so you just need to experiment with a few algorithms and they will do tehir own fine tuning.

[Rich1980]

Ah I see, and is field rotation where the framing of your shot would basically rotate due to not being alone right? 

 

Am I right thinking the brighter objects M31, M42, etc would need less guiding as exposure times won't be as long yes? 

 

Thanks too for the responses. 

[ollypenrice]

1 hour ago, Stub Mandrel said:

Your basic understanding is correct.

Some people let the polar alignment be slightly out, as that tends to mean most corrections are in the same direction, minimising backlash (mostly an issue in DEC as guiding in RA is usually an issue of varying drive speed but not direction).

My own experience is that the better the PA I get, the better the guiding.

The clever bit is that the software measures the relative brightnesses of all the pixels in the star and uses these to average out the centre, this can be done accurately to a  surprisingly small fraction of a pixel. This allows quite small, short-focal length scopes to be used as guiders and still allow tracking well enough for much longer scopes.

 

Software wise I use PHD2, I now get consistent results, with the actual accuracy usually determined by the transparency and seeing. This meant 'tuning' the mount and experimenting with different guiding algorithms to see which worked best with my setup.

I think I can be safe in saying that what works with me may not work as well for someone else as you need to match the way the software works to the hardware  - we are asking the scope to point with an accuracy to shooting a bullet through the earth into a target the size of a football field, so tiny differences in setup mean measurable differnces in how teh rig reacts to guiding.

Good news is that the number of choices is reduced by the ability of guiding software to learn how your rig reacts so you just need to experiment with a few algorithms and they will do tehir own fine tuning.

Yes!  This is the brilliant bit which allows you to guide to a fraction of a pixel, contrary to what might seem reasonable. You don't guide on a star, in reality, you guide on the 'centroid' of a star which is far more accurate.

Autoguiding involves an exquisite bit of engineering which can turn a mount's natural error of maybe 30 seconds of arc into an error of maybe 0.5 seconds of arc. And this at a cost of a couple of hundred euros. It sets up a virtuous feedback loop of quite fantastic efficiency. Do it!

Olly

Edit

 

Edited November 3, 2019 by ollypenrice

[vlaiv]

7 minutes ago, Rich1980 said:

Ah I see, and is field rotation where the framing of your shot would basically rotate due to not being alone right? 

 

Am I right thinking the brighter objects M31, M42, etc would need less guiding as exposure times won't be as long yes? 

 

Thanks too for the responses. 

Yes, field rotation is due to RA axis of scope not being ideally parallel to axis of earth rotation. Over the course of the night, although you don't touch scope and camera - resulting shot will be slightly rotated in relation to first one (there will be progression in rotation angle between each subsequent frame). This is not an issue if angle is too small during the course of single exposure. Software that stacks subs can correct for any rotation between subs.

It only becomes a problem if your polar alignment error is big and there is visible rotation of the field during single exposure - stars in corners will star to trail while those in center will still be circular.

There is no such thing as less guiding - you either guide or not. But yes you are right, need for guiding depends on exposure length. If you are using short exposures there is less need for guiding, but that depends on the mount. Some mounts don't even need guiding for exposures up to a minute or so. Some mounts show trailing of stars after dozen of so seconds. Better the mount - longer exposure it can handle without guiding.

Mount performance is complex topic and there are many aspects of mount performance that are related to guiding. If you are guiding - you should not worry much about peak to peak periodic error if your mount is mechanically smooth - guiding is going to take care of that. On the other hand, mount with small periodic error that is "rough" and has error in position that changes abruptly will be hard to guide with good results. 

[ollypenrice]

4 minutes ago, Rich1980 said:

Ah I see, and is field rotation where the framing of your shot would basically rotate due to not being alone right? 

 

Am I right thinking the brighter objects M31, M42, etc would need less guiding as exposure times won't be as long yes? 

 

Thanks too for the responses. 

First point: any deviation from perfect polar alignment will generate field rotation. However, if it amounts to less than half a pixel at the edge of field in the sub length you are using, will you see it? Put it this way, it won't be your biggest problem as an imager!

Second point: if it were true that brighter objects needed shorter subs you'd be right. However, my own M42 used 15 minute subs and my M31 used 30 minute subs. Yes, both involved short subs blended in to capture the brightest details but very few objects can be done entirely in short subs. When I say 'very few' I should say that I have yet to encounter one.

Olly

[tomato]

43 minutes ago, ollypenrice said:

Yes!  This is the brilliant bit which allows you to guide to a fraction of a pixel, contrary to what might seem reasonable. You don't guide on a star, in reality, you guide on the 'centroid' of a star which is far more accurate.

Autoguiding involves an exquisite bit of engineering which can turn a mount's natural error of maybe 30 seconds of arc into an error of maybe 0.5 seconds of arc. And this at a cost of a couple of hundred euros. It sets up a virtuous feedback loop of quite fantastic efficiency. Do it!

Olly

Edit

 

As Steve Richards said in his book ‘Making Every Photon Count’ (which I strongly recommend you read if you want to get into Astrophotography) “sooner or later you WILL want to auto-guide....”

[Rich1980]

4 hours ago, ollypenrice said:

Yes!  This is the brilliant bit which allows you to guide to a fraction of a pixel, contrary to what might seem reasonable. You don't guide on a star, in reality, you guide on the 'centroid' of a star which is far more accurate.

Autoguiding involves an exquisite bit of engineering which can turn a mount's natural error of maybe 30 seconds of arc into an error of maybe 0.5 seconds of arc. And this at a cost of a couple of hundred euros. It sets up a virtuous feedback loop of quite fantastic efficiency. Do it!

Olly

Edit

 

You lost me with the numbers haha but fraction of a pixel sounds impressive. 

And when the time may come to do delve into this, I've a orion 80mm scope I could use? Old laptop laying about, so would just need to buy a guide camera? 

[Rich1980]

4 hours ago, ollypenrice said:

First point: any deviation from perfect polar alignment will generate field rotation. However, if it amounts to less than half a pixel at the edge of field in the sub length you are using, will you see it? Put it this way, it won't be your biggest problem as an imager!

Second point: if it were true that brighter objects needed shorter subs you'd be right. However, my own M42 used 15 minute subs and my M31 used 30 minute subs. Yes, both involved short subs blended in to capture the brightest details but very few objects can be done entirely in short subs. When I say 'very few' I should say that I have yet to encounter one.

Olly

Ah so like shooting bracketed photos to merge afterwards, how do you work out exposure times when working to 30 minutes etc? 

 

I'm gonna need to buy that book for sure! 

[ollypenrice]

8 hours ago, Rich1980 said:

Ah so like shooting bracketed photos to merge afterwards, how do you work out exposure times when working to 30 minutes etc? 

 

I'm gonna need to buy that book for sure! 

I use CCD cameras. The situation is not the same for CMOS chips. I find with one of my cameras that it continues to go deeper up to about 30 minutes so, if I'm chasing faint signal, that's what I use. The other camera seems to stop gaining after about 15 minutes so, again, that's what I use. You actually have considerable flexibility in exposure times doing AP. It is not comparable with daytime photography in this respect.

There is an elegant way of combining short subs with long in Photoshop. http://www.astropix.com/html/j_digit/laymask.html Surprisingly, perhaps, there is hardly ever a need to do so but the famous target for which it is essential is M42. This had 11 seconds, 50 seconds and 15 minutes: https://www.astrobin.com/380941/?nc=user

Regarding the numbers I quoted over guiding, it's very simple. A basic, decent mount like an NEQ6 may have a periodic error of 30 seconds of arc. That just means it will, at any time, be pointing to within 30 seconds of arc of where it should be. If an autoguider brings that down to half a second of arc the mount is now 60x more accurate. That's a lot!

Olly

 

 

Edited November 4, 2019 by ollypenrice

[vlaiv]

8 hours ago, Rich1980 said:

Ah so like shooting bracketed photos to merge afterwards, how do you work out exposure times when working to 30 minutes etc? 

 

I'm gonna need to buy that book for sure! 

Exposure length depends on several factors.

There is a limited amount of signal that sensor can accumulate, and if you "over do it" - it will end up saturated and actual signal in saturated areas is lost. This happens on bright parts of target and star cores in long exposure. For this reason you go with additional short exposures to fill in missing signal - like Olly mentioned.

Apart from those short "filler" exposures, best exposure length is as long as you can. Fewer long subs will give better result over more shorter exposures for the same total integration time. However at some point you get to a place where difference is so small that it is not worth doing longer - any difference that exists will be too small to be perceived. This depends on read noise of the camera and other noise sources. Once read noise is sufficiently small compared to other noise sources - difference in final result becomes small. CCD cameras have larger read noise and thus require longer exposures. CMOS cameras have lower read noise and can use shorter exposures.

One can work out rather easily what is best exposure length for their setup (scope / camera / light pollution levels), but even then you don't need to go with that exact exposure to get good results. Sometimes people choose to use shorter exposures for other reasons - like tracking / guiding precision, bad weather and such. If there is wind gust or some other event that causes sub to be ruined - less data is wasted if sub exposure is short (it's better to discard one 2 minute sub rather than 15 minute one - less time wasted).

[Rich1980]

Quick update, Making Every Photon Count arrived today from the proud sponsors of here, so a bit of bedtime reading for the next few nights. 

Thanks again for the info and advice, really learning a lot on here. It is very much appreciated. 

The orion nebula photo is amazing too Olly.