I determined why ACP can't start the AO at all unless the guide star is right smack centered on the full frame guider image. Dan had the Autoguider Image set to "Flip Horizontally". This caused ACP to pick up the guide stars on the opposite side of the detector and after measuring the SNR, it gave the wrong coordinates to the guider for tracking. The track box was always empty.

After getting past that one, we are now at the same problem being experienced by many other AO users... as soon as the main exposure starts, the AP stops tracking. I did some experimenting and videotaping, and what's actually happening is that the AO is losing a great deal of sensitivity at the moment the main imager starts exposing. You can see this when ACP is set to force the AO to greatly over-expose the guide star.

I will produce a video that will make this crystal clear.