PHD - Craig finally determined to add Polar Alignment!

[g__day (Matthew)]

Hi all,

Thought I'd mention to folk a rather interesting development on the Yahoo PHD forums. Craig has finally advised us on the 12th Dec that due to overwhelming demand he is plannng to add polar alignment capabilities to PHD:

OK, by far the #1 requested feature for PHD Guiding is some tool to help users polar align their scopes. I've had 3 requests just this week for it! So far, my hopes that this will go away or that someone will do it in OpenPHD haven't come true, so ... let's talk here as a group.

Now, there are several ways in which a computer can help you align your mount:

1) It slews around to a bunch of stars, centers them (or has you help here), figures out (or already knows) exactly what they are, figures out your flex and other misalignment issues. This is the kind of thing that T-point does. In the limit it will tell you "Turn your Az knob 1/8 turn clockwise." Uber-cool, but it needs to know *a lot* about your mount, scope, camera, needs to slew, may need to plate solve, and nee ds more time than I have to code. So, we're not going to be that cool!

2) It could track your error as you're guiding and tell you something about your current error. To know this, it will need to know where in the sky you are pointing. Folks with ASCOM mounts will have this info available, but anyone (like me) guiding with an ST-4 port won't. Now, you could tell PHD where in the sky you are imaging and, with some math I'd need to think about a bit, it should be able to say things like "I think you're aimed a bit high" . While this sounds pretty cool, it's also an after-the-fact kind of thing. It won't be the most efficient way to get you polar aligned.

3) It could have you go through a drift alignment procedure, giving you the info you need to adjust the mount. Drift alignment needs two stars -- one near the meridian to adjust your azimuth and one near the horizon to adjust your altitude (http://www.astropix.com/HTML/I_ASTROP/TRACKED/POLAR.HTM). Once calibrated PHD knows N, S, E, and W of course and if it knew you were on the meridian star it could say "Hey, it's going north -- you should move your mount to the east".

Now, #3 already exists to some degree. If you fire up the graph, turn off guide outputs, and watch the Dec curve, it will tell you if the star is moving N or S. If you know you're on the meridian star, you can use this to adjust your azimuth. Of course, if you're on the horizon star, you can use this to adjust your altitude. What's not there is the "cheat sheet" to help you remember these bits or walk you through them. This could be as easy as a detailed bit in the Help file or it could be some dialog that pops up (has you select meridian vs. horizon star and E or W star for the horizon, shows the current N-S error, and what this means to adjust.

Collective thoughts?

Craig

Everyone seems to like Option 3. I've pointed the man to this site and these algorithms if it helps!

Allen Gilchrist wrote and excellent calculator in 2002 from memory – else try http://canburytech.net/DriftAlign/Spreadsheet.html or download the 1.5MB calculator plus doc and data

http://canburytech.net/DriftAlign/do...rift_align.zip


Calculations required documented on the same site for all http://canburytech.net/DriftAlign/Equations.html

[DavidTrap (David)]

Option 3 works really well. If I didn't have plate-solving and PolarAlignMax at my disposal, that would be my preferred method. A friend uses this method and with an Orion ST80 guidescope, he always gets a suitable star in frame. The whole process can take <10mins for alignment that is satisfactory for imaging. You usually get a good idea of movement on the graph within 20 second.

A cheat sheet that advises which way to turn the mount would be ideal, but I guess the infinite variables between refractors vs newtonians vs Cats, plus or minus diagonals might make it a bit harder to say an upward deflection on the graph equals north or south, etc. Once you've worked out what works for you, you can write your own cheat sheet. A well thought out set of instructions to guide people through this initially would be a great addition. The guides I've read previously on the web give the general gist of what to do, but can be confusing.

DT

[adman (Adam)]

a very helpful thing that could be added to PHD is to give the rate of drift as a number (arcsecs/minute?). That way, once you get to know your own mounts adjustments, you will know how far to adjust (eg number/fraction of turns of the altitude adjustment) without having to go through too many iterations of correction / observation

Adam

[allan gould]

I use Alignmaster (option 1) followed by PhDguiding as option 3. Works perfectly for me.

[Enrique]

I use EqAlign http://sourceforge.net/apps/mediawik...=Main_Page_(en) It uses drift alignment and works very well!!! It shows you clearly what to do.

Regards,

Enrique

[g__day (Matthew)]

The desire most folk express is readuot polar misalignment - but also show it on the screen - put a big red X on where the guidestar should be, so when one changes the mounts position through the adjustment bars - polar alignment is fixed once the star is on the X.

[[1ponders] (Paul)]

Originally Posted by adman

a very helpful thing that could be added to PHD is to give the rate of drift as a number (arcsecs/minute?). That way, once you get to know your own mounts adjustments, you will know how far to adjust (eg number/fraction of turns of the altitude adjustment) without having to go through too many iterations of correction / observation

Adam

This is exactly what K3CCDTools Does Adam. Its brilliant. Makes life so much easier and you can change your graph axis scales as well.

I hope Craig is looking long and hard at how Peter K has his drift explorer in 3 setup. I'm yet to find a simpler method for any gem mount with or without goto ability

[g__day (Matthew)]

Something I would love - would be if someone really mathematically gifted looked at the detailed formulea, to find a way of confirming polar misalignment of both axes whilst tracking a star at almost any elevation or compass point.

I understand why we do two stars in specific locations, but we have powerful computers at our beck and call. I always wondered if the results of solving just one axis at a time might be too simplistic. I would be reassured if the algorithms used take into account elevation and assumed refraction and simultenously modelled polar misalignment in both axes.

I know the model would find it easier and a higher confidence interval if a star 20 degree above the east and one almost straight up (from originally pointing East) are used. But I would be only too delighted to see the software calculating polar misalingment in both axes simultaneously!

Stasticians are used to dealing with this sort of challenge (where two variables interfere with each other to differening degrees - colinearity).

I would have expected the best approach would be measure at Star due East elevated at 20 degrees above the horizon for 5-10 minutes, then raise elevation 10 degrees and repeat again for 5-10 minutes and so forth all the way to 90 degrees, or back from 90 all the way to due west 20 degrees above the horizon. This should give the best curve to fit points to. Then go to a star 45 degrees above the horizon facing die East and show on the screen exactly where that star should be so folk could manually adjust their mounts alignment bolts.

The check high, check low and repeat just feels ... 1900s sort of mathematical prowess. If a PC today can (pardon the foreign financial jargon) caluclate a complex long look forwad kick back derevitative for a advanced synthetic in multiple curriences on several exchanges in a few milli-seconds, surely fiting data to a hyperbolic path with multiple points to determine its offset deltas in two dimensions isn't harder?

[DavidTrap (David)]

PoleAlignMax does something along these lines.

Plate solves an image, slews x degrees in RA, plate solves again then repeats once more.

I guess this removes the issue of periodic error affecting the calculation. Provided the mount really does slew x degrees between shots, the calculation should work.

DT

[AndrewJ]

Gday Dave

I guess the infinite variables between refractors vs newtonians vs Cats, plus or minus diagonals might make it a bit harder to say an upward deflection on the graph equals north or south, etc.

All i do is release the dec lock and "manually" push the diagonal end of my scope "towards the pole". In my LX200gps, this effectively means the corrector "physically" moves "away from the pole".
This effectively simulates a star drifting "towards the pole".
ie Sth for us
I then orientate my camera such that the star "rises" on the screen when i do this.
No need to "think" which way the image is presented then.
If the star is rising on the screen, its drifting Sth and a std cheatsheet tells you which way to adjust the wedge.

Similar logic can be used with a gem.

Andrew

[g__day (Matthew)]

I like the approach PEMPro's Polar Align Wizard takes to determine compass directions and image size.

You simply slew to a star - take a reference image, halt RA for 5 - 10 secs and take another reference image - that gives you East West axes + image scale. Next it slews North and South and compares the image to the first two - and voila you have all your compass barings too!

[White Rabbit]

Align master and option 3 for me as well. Couldnt be easier, or quicker, done and dusted in 10mins.
I remember the bad old days of spending 45 mins drifting and still not getting a great polar alignment.

[g__day (Matthew)]

I'm delighted to say Ray Gralack (author of PEMPro) has joined the Yahoo PHD polar alignment conversation and is graciously offering some advice too!

[RobF (Rob)]

Originally Posted by [1ponders]

This is exactly what K3CCDTools Does Adam. Its brilliant. Makes life so much easier and you can change your graph axis scales as well.

I hope Craig is looking long and hard at how Peter K has his drift explorer in 3 setup. I'm yet to find a simpler method for any gem mount with or without goto ability

+1 for a simple drift number and first derivative. It's just so convenient. I don't use K3 anymore because it doesn't like my SSAG, but a bit of code to monitor Y drift in Maxim gets me aligned very well in a few minutes. I've always wondered by Phd, and even Maxim, didn't include such an incredibly valuable feature.