Perpendicularity of a cylinder and how to evaluate

[[Li...]]

When evaluating perpendicularity of a cylinder to the top and bottom surfaces what evaluations should I use and not use.
I have attached two images of parts we make here. I inspect the cylinder with two circles(top and bottom) and then create a cylinder with "recall feature points. After that I always go to evaluation, filter/outlier, edit, click high pass, and connect segments.
So for the characteristic of perpendicularity what evaluations are to be used if any? What I really want to know is if I evaluate the characteristic by using constraint it always makes my results much less of a deviation. For example, lets say Datum A is a cylinder I created with two top/bottom circles and need to know how square it is to the top surface/plane(Datum C). I run the part and I get .0004" and my tolerance is .0005" If I go to eval. and click constraint and constrain Datum A to Datum C what is actually going on. I normally see a drop in my actual.  Is this wrong or right to do. It would help if I knew for sure what Calypso does to give a result for Perpendicularity and what is being done to those results when Constraint evaluation is used.

Attached are images of just two of hundreds of "bearings" we make. Generally, we hold Perp of the OD to .0005" and Perp of the ID to .0005" to both the top and bottom of the part.

Bearing snapshot.png20220519_092908.jpg60417 set-up.jpg

[[Cl...]]

I would measure two circles, top and bottom on the OD of -A- & two (top & bottom) on the ID of -B-. Then recall both sets of circles into 3d lines.

[[Ma...]]

One note for perpendicularity.
When you are measuring planes then Calypso is evaluating them as rectangles.
Perpendicularity is measured by length of feature, so deviations are bigger closer to corners.

For our purposes we report deviations in axis directions.
Maybe runout would be better value then perpendicularity.

See pic in attachment.