I have done some dimensional studies on some of my parts and was surprised by what I found. ODs were undersized, IDs were also undersized and no correlation to overall size. This does not make any sense. If an ID has extra material on it making it undersized then the OD should also have extra material on it making it over-sized. Essentially an under cure or scan situation. Now if the whole model were shrinking that would make both undersized however then there should be a correlation between the overall size of the part and the amount of shrink. The errors were random relative to component size however they fell in the range of 0.014 inches +/- 0.004 inches. Has anybody else looked at dimensional data? Does Form Labs Have any info that can help in understanding this? Now this could be an artifact of how the beam is offset to take into account the beam diameter. Since the beam diameter is about .012 inches (guess based on smallest feature size) the .014 inches is awful close and could indicate an error in the way the scanning software calculates the offset.
Sounds like galvo calibration issues. Look at a bunch of the threads talking about circles turning into ellipses. Most seem to be over in 1 axis and under in the other, but maybe yours is under in both.