I have two cylinders on a part that need to be concentric within .002". I noticed that on some parts, not all, the concentricity reports out of tolerance at .0039 for example but then when I rotate the part 180 degrees, the concentricity is suddenly .002 or less.
I align the part by measuring a top plane, leveling to this and setting z origin. Then I measure a circle with 9 hits and make this the x and y origin. Im not locking rotation as I really dont think it matters in this case since its a circular part anyway. I then measure the datum as a 3 level cylinder with 9 hits each level, then measure the next cylinder 9 hits 3 levels.
I don't see how rotating it 180 degrees changes anything especially since I'm taking 9 hits at each level on diameters less than 2.5 inches I should be, in my mind, hitting enough spots to capture anything weird going on with the part. What am I doing wrong here? Thanks.
Yes, I did do that and it was .002" for the perp, now I'm leveling and setting xy origin to datum A and setting the top plane to z origin. I also constructed circles from points and rotated the points taken at one of the three circles and created a cylinder from that. and seem to be getting better results.
I don't know enough about ASME, but in ISO world (ISO1101), it cannot be a concentricity (circles aren't coplanar), and it cannot be a coaxiality, because the ratio between length and diameter is too small.
I believe that it should be a localization (TP).
What about trilobe effect if you turn the part and the start point of 180° ?
Reading through the ASME concentricity I don't see why that callout couldn't be used. I do know that ASME 2018 has removed concentricity but this is an older print. Trilobe effect would be a problem with the probe tip, correct? So your saying to rotate the part but also rotate the points with it?
If this were me, I would level my alignment to the top plane before extracting the A-datum cylinder.
Then, I would assess perpendicularity between top plane and the cylinder.
My presumption is that the A daum axis length is too short to effectively use as a datum, and whatever uncertainty in your hits on that cylinder is projecting / compounding exponentially. If i'm right, your perpendicularity will also be all over the place.
If so, Level to the top surface and check both cylinders with profile or total runout to that top surface, and i bet you will get much more consistent results.