It has bugged me for a long time that the Camera Simulator draws reasonable star fields, but, as declination gets larger, they get less and less realistic. Today I bit the bullet and spent a couple hours reviewing the math for drawing star fields, and finally, found the original error. I have a pull request outstanding for the fix, it's a one liner in the bit that maps stars to ccd positions. There was one point where it was using the Dec of the frame center rather than Dec of the star in one of the big long calculations for position. I guess I made that error when I originally wrote the math in the simulator, and it never got tracked down and identified.

With this correction in place, simulator now draws correct star fields that solve with astrometry.net, even when the field overlaps the pole.

This became a real issue for me today, i'm trying to revive a program I wrote years ago that does polar alignment calcs, and it depends on correct star fields when pointed at the poles. I didn't get anywhere on the program itself today because I spent far to long poring over long complicated equations trying to find which one had the error.

And I was wondering why astrometry.net sometimes fail without any good reason with the Simulator! I hope this resolves that!

Using the latest PPA, which I think includes this fix, I still can't plate solve successfully around (north) pole area. At parking position it doesn't solve at all and for example if I track the next closest star (not Polaris) in Ursa Minor it gets into loop when doing alignment and ultimately fails:
2016-12-21T12:45:42 Maximum number of iterations reached. Solver failed.
2016-12-21T12:45:42 Target is within 0° 00' 40" degrees of solution coordinates.
2016-12-21T12:45:41 Solution coordinates: RA (17h 26m 54s) DEC ( 86° 34' 20") Telescope Coordinates: RA (17h 26m 51s) DEC ( 86° 34' 23")

South from that it starts to work ok, for the next star in Ursa Minor does solve so it's definitely better than before.