Thanks Terry and Merlin, I really appreciate your time and efforts.
You seem to have identified the issue.
When I calibrated using the O2 line, I got one l/pix reading, but as I refined it after croping the l/pix changed.
I don't understand how the focus could change, unless the spectral image is not flat?
Would the multi-line calibration produce a better result?
I will post an image of etacar taken with the same setup, on the same night.