Looking for strategies measuring a difficult part.

[[Da...]]

Here's the part: 

Sorry about the extensive editing; new corporate rules about sharing of customer prints.

So, Datum A is a plane created from the flat collars of 4 bushings sitting proud of the surface; 2 of them are .095" wide on each edge, and 2 are .039" wide on each edge, in the Z+ surface.

Datum B is also a plane, created by the .095" wide collar edge on 2 separate bushings on the Y+ surface

Datum C is the diameter of the upper bushing on Datum B.

 

My base alignment, all with a 1mm probe tip:

8 points on datumA, 3 on each of the larger collar flanges, and on on each of the smaller flanges.

6 points on datumB; 3 on each of the collar flanges.

4 points for the circle of datumC

 

My DCC alignment, all with a 3mm probe tip:

14 points on datumA; 4 on each larger flange collar, 3 on each smaller flange collar

8 points datumB; 4 on each flange collar

50 points datumC (profile measurement), "circle in cylinder"

 

Then program measures (with 3mm probe tip), all profile measurements:

Position of bushing below datumC

Perpendicularity of both datumC bushing, as well as the one below

Positions of all the circles of the bushings in datumA

 

My issues:

DCC measurement of datumA has difficulty staying on the tops of the bushing flange collars, probably skewing the datum level.

As a result (or maybe not?), positions of all these holes are out. Things on the B datum are good, though.

 

What I was wondering is some way of centering each planar measurements for DatumA in particular, perhaps using an interim alignment of diagonal holes in the pattern to the base alignment plane? Or, is there some way to not only center each planar surface on each bushing, then consolidate them into a "greater datumA"?

Frankly, I'm betting that this weird datum setup is, has, and never will achieve what the print is asking for, nor would anybody at the customer be able to get an accurate measure either. But, new CMM, and so CMM program is to be made.

No rush on this one; I still have some dubious results from a manual check done by in-process inspector that I can go with.

Bushings pictured below. Yeah, I know....

[[To...]]

Please sign in to view this username.

  I'll address one issue regarding the scan paths on the planes.  I often use a Circle Path strategy for planes like these and use a formula for the location from the circle in the middle.