Confusing Failure to Guide / Polar Alignment

[JSeaman]

A few nights ago I went out and spent about 3 hours just trying to sort polar alignment, I used drift alignment and recorded my steps, the abridged version is:

    Slewed to bet cnc which gave me a -3 meridian offset and +9 declination
    I adjusted and pulled the error to -.25
    Next, I went for omi tau which gave -82 and +9
    I brought this it to <1'

I grabbed a quick screenshot of the error and it was showing -0.64' (attached). I thought I had cracked it. 

Tonight I went out and it seemed all over the place so I ran the guide assistance, it now says I'm somewhere between 15 and 19 arc minutes out on polar alignment, see second screenshot attached

Everything is tight, this is in an observatory on a pier and it hasn't been bumped. Can anyone begin to hypothesise what on earth is going on for me?!

Edited February 20, 2020 by JSeaman

[michael8554]

Hi

When you did your initial PA, how long did you wait ? It can  take over a minute before the Dec error settles down to a true reading.

Also you don't need to get PA down to zero, a 5 arcmin error is easily guided out.

Michael

[JSeaman]

Hi Michael, I was waiting about 5 minutes each time

I was only aiming for <5 but ended up pulling it in really well

After a bit more time playing with this last nightI found that I was regularly getting very poor performance in DEC for a few minutes and then guiding resumed as normal. This meant the guide assistant was seeing the bad start and producing some very scary numbers. Example graph attached, note the first big hump in DEC

My guess is that there is a DEC backlash issue or something I don't understand going on 

[michael8554]

Hmm, interesting. 

Even if you have a lot of Dec Backlash, or conversely Dec Stiction, in the Guide Assistant run the guiding is disabled, so the PA error should be real?

Unless you have your scope balance spot on, leaving it free to wobble in the breeze so to speak? 

A little bit of imbalance in one direction, with the PA set to drift slightly the same way, means the Dec Backlash is taken up, and correction should only be in one direction, with no reversals. 

Unless you Dither, that complicates it. 

But after the initial drift you seem to have good results, so fingers crossed! 

Michael 

[ollypenrice]

I would get the software out of the polar alignment entirely and do it using the classical drift method. There is nothing to go wrong in the old way. In an observatory you're not re-doing it night after night so any increase in time it takes won't matter. And because it is bound to work you may save time. As it is you have one bit of software telling you one thing and another telling you something different. No surprises there!!

Olly (powered by steam.)

[JSeaman]

I think the scope balance is good

I do indeed dither and you're right, it guides just fine after the initial period

I was thinking about ditching PHD and drift aligning manually, thanks for that. The confusing bit is that the guiding seems OK (not brilliant but good enough) after the initial calamity

[alacant]

12 hours ago, JSeaman said:

hypothesise

Hi

If you've built on soil and have had a lot of rain or it's been especially dry for long periods, ground movement? Especially noticeable here in summer when the ground dries to quite a depth.

[JSeaman]

9 minutes ago, alacant said:

Hi

If you've built on soil and have had a lot of rain or it's been especially dry for long periods, ground movement? Especially noticeable here in summer when the ground dries to quite a depth.

Ha that's a good idea given the current climate but it's in cement for a couple of square metres

[JSeaman]

After adjusting the DEC backlash this seems to have gone away, just under 1" for last night which is not bad considering how windy it was!

[newbie alert]

Looking at your images I'd say you was far too soon to adjust and think that you're spot on, if you look at the top image it's says you're 3 pixels out with the red line ascending upwards.. that's pixels and not arc secs, so convert that into arc secs and it's quite abit of error..

When I drift using phd2 I let it settle for at least that time span again and get the error down to 0 pixels...do that on both axis  and you get really good figures, granted it takes some time but because you have a fixed pier it's not like every session, I do that on a tripod and setup/ teardown system..

Also on your pointing prior to running the guiding assistant I have it on DEC 0 and I run it for at least a worm cycle, I know it says 2 mins or at least until the PA errors settle but I get better results running it for far longer

[michael8554]

The top image reports the PA error as - 0.64 arcmins (3 pixels)

His guide setup has a image scale of about 2 arcsecs/pixel.

So 3 pixels equates to a PA error of about 6 arcSECS.

6 arcMINS would be an acceptable PA error, easily guided out, so 6 arcsecs is insanely good.

As shown by the final guiding image which is okay, apart from the Dec stiction at the start.

Michael

[JSeaman]

Thanks for the input and Michael you obviously have a better understanding than me!

Last night I decided to try a drift align, my approach was as follows:

1. I slewed to Bet Cnc and left the EQMod rate at Sidereal

2. I moved in DEC to know which direction was which and monitored the star for 10 minutes

3. There was no movement at all in DEC or RA, I was zoomed in so I could see the 3 pixels the star illuminated and it didn't change

 

4. Next I slewed to Omi tau and checked DEC direction again

5. I saw movement this time but only in RA (see picture) so I didn't correct anything

 

My guiding performance (having made no corrections in either direction) was 1.08" (0.56 px) and I could do 10 miuntes subs. I don't have a good feel for how bad this is but I think I was around 0.6" previously, guide snapshot also attached

[michael8554]

Not sure, are you complaining about sub 1arcsec guiding and 10 minute subs? 

Just for convenience, I'd rotate your guidescope so that RA movement is "horizontal" (think you should Recalibrate after). 

Also HFD 4.97 for your guidescope focus isn't good, try using that reading to get down to 3ish.

Michael 

[JSeaman]

Not complaining, trying to identify how good this is. I found a PHD doc this morning that said 1" and lower is expected for good guiding and a bunch of people saying they have 0.5" and lower, I don't really know what good looks like

I read an article yesterday saying not to waste time rotating the camera because you can just move in DEC to find out which is which (I haven't done it before but it sounded logical)

I'll tweak that guide camera focus again, I can definitely improve that but was following the defocused methodology to allow a greater number of pixels for a star which potentially improves guiding. 

[alacant]

On 21/02/2020 at 10:20, JSeaman said:

after the initial calamity

Hi

Yeah. That quite often happens here, especially early on after sunset. Best to start imaging anyway; it seems to take a while to settle and the graph can look awful. After an hour or so seems to be the time by which to gauge how well you're doing. One factor you can't control are the atmospheric conditions and I think that that that plays the biggest part in all guiding. One night you can be great, the next, all over the place. Nothing equipment wise has changed, the seeing however has.

3 hours ago, JSeaman said:

people saying they have 0.5" and lower

On nights with steady sky -and there' aren't many-  maybe, but remember to look first at the the sky, then the images. Try to avoid looking at the numbers (or the graph!).

Cheers and good luck.

EDIT. EQ6 geared? Try the PPEC algorithm for RA with a fixed period of 120. 

Edited February 28, 2020 by alacant

[JSeaman]

Thanks for that feedback, seeing is definitely a factor! It sounds like 1" isn't terrible then?

I'll look into PEC now as it is indeed geared

[michael8554]

If your 10 minute stars under UK conditions are round, then why worry about the figures? 

PHD2 will cope with any guidecam angle, it's just a case of being more convenient for the user. 

For instance, if you use your guidescope as an aid to centring your target. 

Defocusing the guidecam used to be suggested for PHD1, but the PHD2 developers will pounce on your HFD.

Maybe defocus a little if you have a really short FL scope yielding tiny stars. 

Michael 

[JSeaman]

I have round stars but they are still bloated, I think that's a feature of the camera (314L not +)

Judging by my (limited) experience of O3 and S2 I need to push towards 20 minute subs, I haven't tried one yet but figured I'd try and get my guiding as good as I can first that's all.

[newbie alert]

22 hours ago, michael8554 said:

The top image reports the PA error as - 0.64 arcmins (3 p

His guide setup has a image scale of about 2 arcsecs/pixel.

So 3 pixels equates to a PA error of about 6 arcSECS.

6 arcMINS would be an acceptable PA error, easily guided out, so 6 arcsecs is insanely good.

As shown by the final guiding image which is okay, apart from the Dec stiction at the start.

Michael

The first image shows a error of 3 pixels or .64 arc secs at that time, I'm positive it would be higher a little while later considering the upward trendline 

Can't see where it shows the guidescope image scale, surely if it's showing over 1 arc min how can it have 6 secs of error?

Unless I've totally missed something..

[michael8554]

ST80 guidescope and ZWO 120 gives you the guiding image scale.

JS said he'd waited several minutes, so that upward trend should be steady.

So the PA error isn't going to change, only the drift is getting bigger. 

Where did 1 arc min come from? 

Michael

[newbie alert]

22 hours ago, michael8554 said:

ST80 guidescope and ZWO 120 gives you the guiding image scale.

JS said he'd waited several minutes, so that upward trend should be steady.

So the PA error isn't going to change, only the drift is getting bigger. 

Where did 1 arc min come from? 

Michael

I already said that he was far too soon, the line wouldnt have either a upward or downward trend and would show a 0 or 1 pixel error

His last image shows a 1.08 rms error..

[tooth_dr]

The RMS is one, but the RA peak to peak is over 4.  How much of an impact of this likely to have on the star shapes?

[michael8554]

On 21/02/2020 at 07:24, JSeaman said:

 I was waiting about 5 minutes each time

That's usually long enough for PA to settle down. 

1 hour ago, newbie alert said:

already said that he was far too soon, the line wouldnt have either a upward or downward trend and would show a 0 or 1 pixel error

You're really confusing me. Are you saying that if he waits long enough the PA error will go down to 0 or 1 pixel, without further adjustment ? 

Or he should have run the Guide Assistant for much longer? Might well have moved past that unwanted stiction/ backlash phase. 

Michael 

[newbie alert]

8 hours ago, michael8554 said:

That's usually long enough for PA to settle down. 

You're really confusing me. Are you saying that if he waits long enough the PA error will go down to 0 or 1 pixel, without further adjustment ? 

Or he should have run the Guide Assistant for much longer? Might well have moved past that unwanted stiction/ backlash phase. 

Michael 

Of course not, that would be just stupid, I'm saying as originally said that he was far too soon to make an adjustment, leave it twice as long then adjust.. I wouldn't be happy with a 3 pixel error  and I'm on a tripod..

And yes with the guiding assistant I'd run it at least for the full worm period..

[JSeaman]

Could you explain that, I thought 5 minutes or so would be plenty for polar alignment, why would you wait for the worm cycle to evaluate polar alignment?