Thanks Jared. I'm not using temperature compensation. I refocus every 30 or 60 minutes depending on conditions and always refocus on filter change. The issue that I see is difficult to explain so I will come at from another angle.
I would expect, for example, that the reading for the position of the focuser displayed by SGP would be consistent for a specific physical position of the focuser draw tube bit it isn't. A reading of 2500 for example I would have thought would always correspond to the same physical position of the focuser draw tube. It doesn't seem to though as each night the starting position count is smaller than the previous night even though the focuser draw tube is more or less in the same position. Over time the starting position noted in SGP at the commencement of the imaging session gets smaller and smaller.
I suspect that this may be an issue with the way that Robofocus is reporting the step count so I will be exploring that angle as well.