Circles from curves

[[Cl...]]

I have a hexalobe (six sided star) Go-NoGo gage used to check screws. I am scanning 2d curves around the OD and recalling those points into circles. Evaluating them as max inscribed/min circumscribed to get the major/minor diameters. Inside each curve I did a best fit/parameters/alignment (would I tick translate X&Y, or rotate about Z)? Do I then create an alignment from the curve and use it for the circles? Is this the correct way to approach this, or is the best-fit even necessary? My goal here is to optimize accuracy.

[[Is...]]

not sure about the purpose of your curves but just consider best fit adjusts the actual values (no the nominals) then if you extract some dimentions from a best fit curve (i.e. using intersections) the values are not accurate. Not the case with recall Points, however if you plot the curves, what you will see in the graphic will be different than what you see in the data..

[[Cl...]]

So then what/how would I use this option?

We have a worn hexalobe gage used to check screw heads, and
are trying to get the actual major and minor diameters of this gage.

[[Me...]]

If I'm not mistaken, if you're scanning these curves based off a model, and an accurate model to boot, in training we were taught NOT to use Best Fit on Curves. I specifically asked about that in the Curve refresher course I took 2 years ago.

[[Cl...]]

OK, based on that, I'll stay away from that approach.