astrosatch Also asiair seems to choose only dimmer stars despite abundance of brighter ones.
ASIAIR will not pick saturated stars, Andrej.
In the past, when users can select their own stars, you often see them selecting saturated stars. You see lots of screen captures on Facebook where the small image at the top left corner of the guide window shows severely flat-topping and ignored by the rest of the Facebook crowd anyway.
Flat-topping causes a lot of harm when trying to detect centroids. And we often see HFD of over 7 or 8. Remember, we are trying to estimate the centroids to at least 1/10 of a pixel (which can already be a significant part of an arc-second; while we all attempt to guide better than 0.5 arc second RMS).
On the opposite end, guiding with really dim stars is also a bad idea because a loss of signal to noise ratio also ruins accurate centroid finding.
So, there is a "goldilocks" region of exposure time and gain, where you will get the maximum number of guidable stars. This is where good dynamic range helps. (I actually tried an ASI178 for guiding, but the frame rate was abysmal within ASIAIR guiding framework, so I gave that idea up).
As to disappearing (and not recovering) stars, it is still a bug that I think ZWO has not yet fixed. At the time they moved from beta to released v1.6, the person I was working with was still in the process of fixing it.
We know that a star that goes away should recover, since we know that can again get many guidable stars when we stop and restart guiding. The same star that disappeared will again reappear when guiding is restarted. So it is not something hard to fix. 15 minutes of coding and a couple of nights of testing should do it. Another thing that they also need to fix is the problem of long term multi-star autoguiding under field rotation -- so be sure to polar align really well. (Dithering may cause the centroids to recalibrate, so perhaps dithering will fix the multi-star field rotation problem too, but I don't have their code to know for sure). At the time I was in communication with them (during v1.6 beta), they were not aware that field rotation can cause confusion among the centroids of multiple stars. So, if you have been autoguiding for 20 minutes, and start seeing stars disappear and not reappearing, it could perhaps be caused by field rotation. In the meantime, try to polar align to better than 10 arc seconds and hope the tripod does not dig deeper into the ground even a millimeter through the course of the night :-).
For what its worth, I have switched back to using IR guiding by using an ASI462MC camera with a 685nm IR-pass filter to get both a more stable set of guide stars, and also to get the Bayer cells of the ASI462 to behave almost like it is a monochrome camera, with all color components have mostly equal spectral efficiency; so less of the Bayer Moiré pattern. The 462 is based on a chip for surveillance cameras, so when there is no visible spectrum light, it behaves quite close to a monochrome IR camera.
Because the cutting off of so much of the spectrum, I usually have fewer stars to guide with (for example, just 9 instead of 12 stars) but the threshold of the ASIAIR multiple stars with my particular mount is only beneficial to between 5 and 6 stars anyway, so I don't feel bad not having all 12 stars.
And IR guiding appears to be more stable with multi-star guiding. I seem to get fewer star losses. But I haven't had many clear nights to play with to say for certain that IR guiding is that much better than full spectrum when used in conjunction with ASIAIR's multi-star centroid algorithm.
So, while waiting for ZWO to fix the bug, and if you already have an IR-sensitive camera at hand, you might try IR guiding.
At least you can try different light pollution filters with a full spectrum camera to try to achieve better signal-to-noise ratio. Or even to use a red or Wratten 29 filter with a monochrome camera to get halfway close to IR guiding. Both these will reduce the flux of most stars, but you may gain better SNR and be less affected by seeing.
Whatever you do, be sure to focus the guide scope really well. It is one sure factor in improving signal to noise ratio (and thus minimize star loss).
Chen
