I just found excellent explanation on this topic. Probably best article I read this week, and it has formulas
FWHM = (FWHMoptics 2 + FWHMseeing 2 + FWHMtracking 2) 1/2
This adding in quadrature is what I was alluding to above when I said it was similar to noise. Again, the problem with this analysis is that it assumes the errors are Gaussian in nature. That is not necessarily a good assumption.
However, it does provide some insight as to why the tracking error cannot be totally ignored if smaller than the seeing, and why improving the tracking error has diminishing returns.
For those trying to follow the formula above those are squares, not multiplied by 2.
Edited by Madratter, 14 January 2021 - 02:13 PM.