PS2 Suddenly Unreliable in Centering?

Ken,
Sorry. No testing tonight. I loaded my mount and told it to slew to Capella – looked over and said – that’s not right. Turns out my DEC drive was dead. I just finished tearing it down and reassembling it and I should be at least temporarily back… Had a big hail storm last week and it tore thru my mount cover. I dried everything off and did not think too much about it. Afterall the mount was working. Well tonight the DEC stepper was frozen solid – undoubted a bit of rust on one of the fine teeth. Once I got to it, I was able to break it free with pliers and now it appears to be working fine. Put a bit of WD40 in, but I will order a new stepper just in case. So, no testing and the next days do not look promising – tonight was the only small window for a while.

rgbtxus, I too find PS2 blazingly fast, usually. For your 2 seconds to solve, are you including the significant setup pause when it first starts before it starts posting the region count? I seem to recall that by itself was several seconds. If it was already fairly close to the target, the region count part would be easily less than 1 second.

It may be that I am getting really fast solves from Astrometry.NET because of 2 factors:

  1. In ansvr config I set the Scale error estimate to 2%. This is really important. Larger numbers will make it run much slower. I also set the time limit to 60 seconds. This combination has never ever failed to solve a legitimate image for me, even in very star poor regions.
  2. My hardware is a cheap quad core Windows 8.1 pc, but it still runs at a respectable 3.2 ghz, so is probably very comparable to what most people are running. More importantly, I think, is that all the code and data tables for Astrometry.NET reside on my SSD drive, which of course gives blazingly fast disk IO.

An additional factor for me that is unusual is that all my initial plate solves follow a slew to a new location which my dome must then also slew to before the plate solve process starts with the first plate solve image. Unfortunately, SGP never seems to pay any attention to where my dome is, so the dome is almost always still moving when the initial plate solve image is taken, which of course makes the solve fail and then there is an additional 1 minute delay waiting for a retry.
For me, when PS2 tries to solve the image of the inside of my dome, it takes at least 2 minutes to conclude it can’t solve that image, then we have a go with Astrometry.NET which fails at 60 seconds. This I will have to double check on. Both solvers should be smart enough to immediately know there are no stars in the image and quit after 1 or 2 seconds.
Another time trait I have experienced with PS2 is that it is very fast if the image is quite close to the hint, say <20 pixels. When It was a lot more, say 50 to 200 pixels it would be very slow. Since my initial slew is usually in the range of 30 to 80 pixels, this often made the first PS2 solve very slow. Astrometry.NET always gives me about the same time, regardless of how close it is.

Bottom line for me is that a complete plate solve process takes around 60 seconds to get closer than 20 pixels to my target. This assumes I have already done a slew (but not center) to the location and waited for the dome to finish rotating into position.

I am still waiting for clear skies to track down my recent PS2 problems a little further.

I experienced failure with the remote astrometry.net on the Win 10 machine.
Seems SGP does not recognize a valid internet connection? The log states this.

[2016-04-15 3:28:21 PM] [DEBUG] [Main Thread] Checking for valid internet connection…
[2016-04-15 3:28:22 PM] [DEBUG] [Main Thread] Error validating internet connection: Unable to connect to the remote server

But, yes the machine IS indeed connected through the ethernet cable. Win 10 Network and Sharing Center states so and I can surf the net fine.

Any suggestion as to why SGP is not recognizing my connection on the Win 10 machine?

I have no idea about the internet connection issue with Win 10. I am running 8.1. All is supposed to work well with Win 10.

I do suggest you install Astrometry.net locally. Very easy to do and absolutely reliable. Internet issue will go away. If you use the parameters I mentioned, it ought to run fast for you. Even if it is not as fast as I see, it is still the most reliable of all the solvers available and everyone’s solver of last resort.

Terry,
Since you are using the remote (online) astrometry.net solver it is quite likely the failure is because the astrometry.net server was down or not responding. I would highly recommend you switch to the local astrometry (ansvr) for blind solve failover.

Thanks, Joel:

Just trying to get this Win 10 machine fully functional with SGP.

Just tried again and SGP log reports the same failure to detect internet connection.
On Win 7 machine SGP log verifies valid connection detected and the remote astrometry solves fine. So, it’s not that astrometry.net site is down. It’s something peculiar to this Win 10 machine.
My ansvr install works fine on both machines BTW. But I’d like to clear up this failure of SGP to detect my connection on the Win 10 machine.

Oh OK. Is your network connection set to “Private network” in the network and sharing center? When I set up my win 10 computer recently I had to change it from public to private.

[quote=“Terry, post:23, topic:3407”]

Thanks, Joel:
It was indeed set to public network. Changed it to private successfully but still get the same message in the SGP log:

Checking for valid internet connection…
[Main Thread] Error validating internet connection: Unable to connect to the remote server

Shot in the dark here…what antivirus program are you running? I recall having some communication issues with AVG installed, but I don’t think it was related to astrometry.net That’s probably not it, but it may be worth disabling your antivirus to see what happens.

Thanks, Joel.
Disabled MBAM. No joy. :frowning:

This does look like a tough nut to crack here Terry. I have not thought of any other things to try. Joel’s suggestion to turn off antivirus was a good one to try, but unfortunately that was not it. I have personally run into many ‘gotchas’ with Windows 10 both personally and with my client base of the database manager that I have developed, market and support for over 30 years. Specifically recently the user’s antivirus program on 2 Windows 10 machines was completely screwing up my products operation, but only on Win 10 not Win 8. Thanks MS. Many other Win 10 problem issues not related to antivirus have cropped up.

I have gone into some detail on this to make the following suggestion. Do I understand correctly that ignoring the issue of SGP trying to plate solve Astrometry.net over the internet and getting an internet connection error, you are able to run plate solves completely successfully with SGP using your local install of Astrometry.net?

So why not forge ahead using the local install? Based on the info in this post, it is my opinion as a software developer that you have a unique internet connection problem here that SGP is simply noticing and reporting. The fact that no other users have reported this problem supports this idea.

“Do I understand correctly that ignoring the issue of SGP trying to plate solve Astrometry.net over the internet and getting an internet connection error, you are able to run plate solves completely successfully with SGP using your local install of Astrometry.net?”

Yes, that is the case. Both PS2 and the local install of astrometry seem working fine.
My recent failures with PS2 have all been on the same field . . . M94. After I understood why astrometry was not being successfully invoked when PS2 failed, and started ansvr to use the local solver, the failover worked.

Why the PS2 failures on the M94 field is still a mystery.
With ansvr working, the failure of SGP to detect a valid internet connection, while puzzling, is of no consequence.

Thanks for the help!

Great, that’s good to hear. Another good example why Astrometry.net is everyone’s solver of last resort. Always works if it gets a legitimate image and you have the camera pixel resolution set correctly.