Very weird plate solve problem

[BrendanC]

The other night I wanted to image NGC 7822, which is at RA 00:01:09, Dec 67:25:17

My location is in the UK, near Aylesbury.

When I did my first Goto++ in APT, for framing, it worked fine. I started the plan, it worked fine. When it reached meridian flip, I actually wanted to do something else with the laptop (calibrating PHD2) so I stopped the plan, thinking I could do the calibration, and then restart the plan post-meridian.

However, when I did, the scope slewed, took an image, plate solved it - and told me it was 319,091 pixels out and cancelled the Goto!

I tried to get it to work several times again, without PHD2 (which I think is a red herring here) but every time I got the same error, not with the precise same number of pixels but a similar figure of around 320 thousand-ish.

I tried again tonight, before meridian, and got the same error.

Everything else works fine, all Goto++ via plate solving. No problem with any other object.

Any ideas? It's really oddddddd!

Edited November 5, 2020 by BrendanC

[Padraic M]

Hi Brendan, I'm not an expert but I have had a similar experience with APT recently - not so much after a meridian flip, but doing a Goto++ to a new target close to the meridian. Out of interest, have you calculated the distance in degrees based on your imaging scale? Degrees = 319,091*scale/60/60. I'd guess that your scale is somewhere between 2"/px (180deg) and 4"/px (360deg).

I had my mount slewing almost a full 360 degrees, doing a new plate solve, and then heading off again in the other direction. I could have been there all night! I didn't fix the problem, and as far as I can remember, I just picked a target farther from the meridian for a few minutes. I have Make Automatic Flip set to Off under Session Craft.

I don't think that parking to Home position and then redoing Goto++ made any difference - it made the first slew and then stopped as you said. I'd be interested in how others manage this problem as it doesn't sound like this is 'working as designed'.

[BrendanC]

Very interesting. 

I did wonder whether the pixel value was something derived from my camera's resolution and arcminutes, played around with the figures for a bit but couldn't see anything obvious. I'll play around with the maths a bit more and see if I can pull anything out, esp regarding degrees.

Meanwhile, it does seem to be something to do with the meridian, as you say. I did the same, chose an object further away from it and everything was fine again. But I waited for quite some time before retrying NGC 7822, during which time it must have passed the meridian by quite some time, and it still wasn't having it, and when I tried again it was quite some way before the meridian.

I've posted about this on the APT forum too, so let's see if Ivo picks it up.

[BrendanC]

Ah, here you go - 176.401437 degrees.

Hmmm. Very suspiciously close to 180, don't you think?

Could it be something to do with the scope thinking it's the other way around?

I'm wondering whether this thread on the APT forum could be something to do with it? Although frankly I only just about grasp most of it: https://aptforum.com/phpbb/viewtopic.php?f=10&t=2794

Edited November 6, 2020 by BrendanC

[Padraic M]

All I can suggest is to clear everything back to basics - reboot the mount and software, clear alignment points in EQMod, start from Park and do a 1-star alignment.  Goto++ shouldn't need any more than that to be accurate. I usually do a solve/sync at Polaris and then again on Vega or Caph/Shedar depending on what's visible and it usually works well.

btw Ivo is indisposed at the moment so he mightn't respond at his usual speed.

[andrew s]

When very close to the meridian a small amount of cone error could make the mount point to the wrong side of the meridian compared to its "side of mount".

Not sure if this helps but just a thought.

Regards Andrew 

PS 3 to 4 deg would be a very large cone error though.

Edited November 6, 2020 by andrew s

[BrendanC]

Thanks all.

@Padraic M I did most of those things. How do you clear align points in EQMOD? Or is that done when you exit and restart it? Also, regarding one-star alignment, I generally just polar align, then crack on with plate solving my way around the sky. Do I need to go back to one-star alignment generally, or are you saying this would be a good 'back to basics' way to identify what's going on? Also, I hope Ivo is OK - Jim has responded on the APT forum and I've shared the log file with him so let's see what happens.

@andrew s That's an interesting theory. I must admit, cone error has been on my mind too of late, mainly because I've never checked it but do intend to use SharpCap's Conesharp utility at some point. There is a slight misalignment in before/after meridian flip frames which I assume indicates cone error, but seeing as I was doing fine, I never got around to fixing it. As you say, that would be quite some error though. By the time I'd given up trying to plate solve to NGC 7822, it must have been an hour past the meridian.

 

[Padraic M]

1 hour ago, BrendanC said:

How do you clear align points in EQMOD?

It's done in EQMOD panel, but I suspect it is done with a restart too.

1. Press the Spanner icon for settings
2. In the Alignment/Sync section, clear the alignment data (or, open the notepad and clear each point)

[andrew s]

Unfortunately,  I have no experience of your kit or software but after a significant error I would "home" my mount before proceeding.  Maybe you can do the equivalent,  home, sync etc.

Regards Andrew 

[BrendanC]

Thanks - I'll try and clear the points. Already set the mount to home and park several times!

I think I'll wait for Jim at the APT forum to have a look through the logs and see what transpires...

Thanks for all the suggestions.