Last night, after much anguish, I finally had a breakthough in fixing my guiding problems. I discovered that corrections with a frequency higher than five seconds led to a situation in which I was basically chasing the seeing.
It didn’t occur to me, because I would expect that if guiding at one or two seconds, not five, especially since I’m guiding with a focal length of 230 mm.
However, guiding integration times of seven to ten seconds yielded pretty perfect stars. Unfortunately I had time for only one 120-second experiment before it clouded up, but the guiding graph was flat, so I’m confident that things should be fine now.
More tests tonight!