Confusing Failure to Guide / Polar Alignment

[michael8554]

Like I said, 3 pixels is 6arcSECS of PA error, you don't need to waste time getting that good, 5arcMINS is good enough and easily guided out, according to the PHD2 developers. 

As would your RA Periodic Error, which could be tens of arcmins. 

Michael 

[JSeaman]

Thanks, I spent yesterday making a second motor for the dome and tonight is due to be clear so I'll try things out then

I currently don't use PEC at all, do you think it's worth investigating?

[JSeaman]

Things are looking better tonight so I guess seeing is a factor as I haven't changed anything (see attached)

I'm currently letting EQMod automatically assess the PEC, will report back shortly!

[alacant]

32 minutes ago, JSeaman said:

EQMod automatically assess the PEC

Hi

Even better we find is the PPEC RA guide mode in PHD2 instead of hysterisis. It should help you lose the PE and -hopefully- the sawtooth with it.

HTH

[JSeaman]

I did indeed read about that and I believe you can run both so I thought I'd start with the PEC. I've run through the calibration process and it's just looping now so I'll see if there is any effect

 

[JSeaman]

So it seems ever so slightly worse, the first picture is pre-PEC, the second pic is post-PEC, no significant change but it certainly doesn't seem better

 

 

[alacant]

16 minutes ago, JSeaman said:

slightly worse

We never got vspec to do anything against the pe either. 

You could try loosening RA until you can feel backlash. That may go some way to lose the spikes.

But hey, if your images are ok, I'd leave it.

Cheers.

 

Edited March 1, 2020 by alacant

[JSeaman]

I tried PPEC just for the sake of it (seeing has gone through the floor) and there is a definite improvement, I just changed the algorithm and kicked it off, is it really that simple?!

 

 

[alacant]

9 hours ago, JSeaman said:

is it really that simple?!

No. It's magic!

 

[JSeaman]

Certainly feels like it, no set up, no parameters to set, just change a drop down and voila 0.2" improvement immediately

[michael8554]

I get the impression that in Guide Assistant, the PA reading changes as the unguided RA goes through its PE cycle. 

So the true PA reading may only occur when the RA line comes back from off the scale, to just crossing the horizontal axis,. 

I raised the question with the PHD2 developers. They said:

"Since the continuous update of the PA estimate is based on a linear fit, it  will normally be very unstable until a sufficient amount of data have been

collected. That's why we force a 2-minute minimum time window in the GA. Beyond that, the apparent Dec drift can be affected by other things: higher

measurement uncertainty with long focal lengths and poor seeing, mechanicalf lexure, wind gusts, etc. " 

Re the EQMOD PEC, a nice curve generated, but it would have been nice to see the actual value of the PE. 

Finally be aware that the PHD2 PPEC is calculated afresh every time you start your guiding for the night, so let it run for about 8 mins to let it get the full PE curve to correct. 

Michael 

Edited March 2, 2020 by michael8554

[JSeaman]

Thanks again Michael, the surprise for me was that the PPEC seemed to have immediate improvement, I was expecting to have to wait ~10 minutes or so

I'm hoping for a few more hours tonight/tomorrow to evaluate further but it certainly looked like an improvement

[alacant]

57 minutes ago, JSeaman said:

looked like an improvement

We found that on a geared eq6, a fixed 122s period -without adjustment- had almost immediate effect. But start imaging immediately anyway. If the first few frames aren't as good as the rest -they usually are- then you can always discard them.

Cheers

[JSeaman]

Well tonight I have very similar results to a few nights ago, 1" of RMS and no real idea why! I guess it's just down to seeing as nothing else is changing but I really thought it was going to do better tonight 

[alacant]

1 hour ago, JSeaman said:

no real idea why

What does the ga recommend? That will give an idea of the seeing.

Edited March 2, 2020 by alacant

[JSeaman]

Just got back after leaving it for an hour and a half to find the guide camera had immediately failed in PHD2, I've seen this happen a number of times since moving to the ZWO 120mm mini, just got things back up and running and it's now at 0.8" so I'll try and run like that for a while, if it breaks again I'll run the guide assistant though.

I'm running out of time as I'm imaging the cone and it's already past the best time to do it so I'm capturing O3 with the moon up (I know you're not supposed to) just so I can get a final image this year

[JSeaman]

Looks like I just about have enough data to form an image so I ran the guide assistant, not happy about DEC: 

[alacant]

9 hours ago, JSeaman said:

not happy about DEC

Hi

Backlash or rotational resistance; the horizonrtal-ish bit at the top-

Can you feel any backlash in DEC? Try to rotate the axis with the clutch tightened. If you can, take up the slack until it's minimal. Otherwise, loosen until you can feel it's just free. The other culprit for DEC sticking is the cone bearing at the base of DEC where the cw shaft slides. Cleaning and tightening the collar finger tight only usually solves any sticking.

Remember it's a cheap, mass produced mount. If you want to go further, you'll probably have to strip it bare, replace bearings and convert to belt drive.

Cheers

[JSeaman]

I did the backlash a week or so ago, there was no noticeable movement but I slackened off the hex keys, felt movement then nipped up per the astrobaby guide and it seemed to go completely. Testin N/S swings using EQMod showed no real latency

I'll have a look at the counterweight bearing next but I seem to be achieving good enough results to keep going as is for now, maybe I'll do the belt drive in summer when no pictures can be taken!

[alacant]

3 minutes ago, JSeaman said:

achieving good enough results

Excellent; imaging rather than 'testing'. I still haven't worked out why there is so much of the latter!

 

[JSeaman]

~0.8" tonight and finished an image

[chrisshillito]

On 02/03/2020 at 08:38, michael8554 said:

Re the EQMOD PEC, a nice curve generated, but it would have been nice to see the actual value of the PE. 

For PE analysis (PE is much more than just a "value") then you would pull the PE capture file that EQMOD produces as part of its autopec into PECPREP. This will give you full analysis or the error signals contributing to the overall PE.