Hmmmm...... must have missed this the first time around.....
A client has a faro arm....... not at all sure which version.
Apparently the USER has as much if not more effect than the actual machine..... We needed a decent model of a part, actually the mounting surface vs the spindle of a wood router spindle motor (old one) that we wanted to use to drive a test setup at medium RPM (a bit under 20k rpm).
We wanted the dimensions accurate to make the holder to support the motor and position the collet where it needed to be to couple to a shaft (via a somewhat compliant coupling). Actually, the client's guys were going to make the model and draw the parts, but we needed to approve.
Anyway, the client's in-house machinist used their Faro arm to get the relationship dimensions.
For whatever reason, the results were no better than 0.002, and not always that good. A few were out far enough that their draftsman and I were very puzzled. I assume their guy messed up, and possibly their unit was not the best one.
We ended up doing better by measuring. Their draftsman, who is good, could not use the data that came off the arm, because it wasn't internally consistent. He and I measured the unit with mics and calipers, etc, and got some very different numbers.
He used our numbers to draw up the support, and the result fit perfectly. Indicating off the shaft of one part to the shaft of the other at two levels showed we were well under 2 thou error, and the coupling could take 6. So we were good.
I don't know what would have happened with the arm's measurements.
I also don't know why they were so bad, but they sure were. So just using the arm won't get you perfection, by any means. You have to do it right, and THEN maybe you get the results mentioned above......