Profile of a Surface - 4 areas of half sphere

What do you mean by "perpendicular to mid point" ?

In your case, I would create feature sets with scans (left, right, top and bottom), and then dimension the surface profile. I guess you used "findnoms".

Line C is created Perpendicular from axis Datum Cylinder B to Midpoint C1-C2. 

As you can see I have individual requirements, I can try take individual points on each surface to create feature set then report profile of a surface.

Yes, I use Nominal Method = FINDNOMS on Execution tab inside Linear Open Scan.

Thanks for your suggestion.

I think I would call the designer.

In ISO world, A is the tangent plane, on the free side material, using minimax algo (1 translation, 2 rotations)

Then, B is the max inscribed cylinder perp to A (so it's a maxinscribed circle constructed from projected hits of the cylinder on A) (2 translations).

In this case, C cannot be a point, it must be a line to fix the third rotation (so it should be the cone axis 16°, but I don't see how constructing this !!!!!) - Or using B-C instead of C ?

About the difference between values (surface or line), I think it's logical, because both profiles concern a linear scan. Maybe PC-DMIS shouldn't allow to dimension a surface profile on a line ?

Datum C should be the pattern of 3 mid-points.  However, PC-DMIS does not currently allow mid-points to be used as a datum pattern.  A work-around would be to construct a cast sphere from each of your mid-points and then use the sphere's for the datum pattern...

PNT4       =FEAT/POINT,CARTESIAN,NO
            THEO/<60.5,30,-2>,<0,0,1>
            ACTL/<60.5,30,-2>,<0,0,1>
            CONSTR/POINT,MID,PNT2,PNT3
PNT7       =FEAT/POINT,CARTESIAN,NO
            THEO/<29.75,47.754,-2>,<0,0,1>
            ACTL/<29.75,47.754,-2>,<0,0,1>
            CONSTR/POINT,MID,PNT5,PNT6
PNT10      =FEAT/POINT,CARTESIAN,NO
            THEO/<29.75,12.246,-2>,<0,0,1>
            ACTL/<29.75,12.246,-2>,<0,0,1>
            CONSTR/POINT,MID,PNT8,PNT9
DAT_C1     =FEAT/SPHERE,CARTESIAN,OUT,LEAST_SQR,NO
            THEO/<0,0,0>,<0,0,1>,1
            ACTL/<0,0,0>,<0,0,1>,1
            CONSTR/SPHERE,BF,PNT4,,
DAT_C2     =FEAT/SPHERE,CARTESIAN,OUT,LEAST_SQR,NO
            THEO/<0,0,0>,<0,0,1>,1
            ACTL/<0,0,0>,<0,0,1>,1
            CONSTR/SPHERE,BF,PNT7,,
DAT_C3     =FEAT/SPHERE,CARTESIAN,OUT,LEAST_SQR,NO
            THEO/<0,0,0>,<0,0,1>,1
            ACTL/<0,0,0>,<0,0,1>,1
            CONSTR/SPHERE,BF,PNT10,,
            DATDEF/C,FEATURES=DAT_C1,DAT_C2,DAT_C3,,

The pattern will then control the rotation.

When scanning the surfaces, you need to scan multiple cross sections so that you collect data across as much of the surface as possible (within the defined areas).

Something like this...

...rather than a single cross-section...

I think you may still need clarification from the designer though, because of the way they've shown the profile call-outs.  Technically, they could have just had a single profile of a surface and it would mean the same thing.

Datum C should be the pattern of 3 mid-points.  However, PC-DMIS does not currently allow mid-points to be used as a datum pattern.  A work-around would be to construct a cast sphere from each of your mid-points and then use the sphere's for the datum pattern...

PNT4       =FEAT/POINT,CARTESIAN,NO
            THEO/<60.5,30,-2>,<0,0,1>
            ACTL/<60.5,30,-2>,<0,0,1>
            CONSTR/POINT,MID,PNT2,PNT3
PNT7       =FEAT/POINT,CARTESIAN,NO
            THEO/<29.75,47.754,-2>,<0,0,1>
            ACTL/<29.75,47.754,-2>,<0,0,1>
            CONSTR/POINT,MID,PNT5,PNT6
PNT10      =FEAT/POINT,CARTESIAN,NO
            THEO/<29.75,12.246,-2>,<0,0,1>
            ACTL/<29.75,12.246,-2>,<0,0,1>
            CONSTR/POINT,MID,PNT8,PNT9
DAT_C1     =FEAT/SPHERE,CARTESIAN,OUT,LEAST_SQR,NO
            THEO/<0,0,0>,<0,0,1>,1
            ACTL/<0,0,0>,<0,0,1>,1
            CONSTR/SPHERE,BF,PNT4,,
DAT_C2     =FEAT/SPHERE,CARTESIAN,OUT,LEAST_SQR,NO
            THEO/<0,0,0>,<0,0,1>,1
            ACTL/<0,0,0>,<0,0,1>,1
            CONSTR/SPHERE,BF,PNT7,,
DAT_C3     =FEAT/SPHERE,CARTESIAN,OUT,LEAST_SQR,NO
            THEO/<0,0,0>,<0,0,1>,1
            ACTL/<0,0,0>,<0,0,1>,1
            CONSTR/SPHERE,BF,PNT10,,
            DATDEF/C,FEATURES=DAT_C1,DAT_C2,DAT_C3,,

The pattern will then control the rotation.

When scanning the surfaces, you need to scan multiple cross sections so that you collect data across as much of the surface as possible (within the defined areas).

Something like this...

...rather than a single cross-section...

I think you may still need clarification from the designer though, because of the way they've shown the profile call-outs.  Technically, they could have just had a single profile of a surface and it would mean the same thing.

Thanks for your suggestion about how to define Datum C but is kind of specific the way they want it to be defined, just in this case after more than 10 years this new sphere feature design is giving us a little bit of a problem, see datum C definition image showing how different it is now. Before was midpoint C1-C2 at Polar Radio of 36 from Cylinder B now is midpoint C1-C2 at X45.36 from Cylinder B. 

There is a metal sphere that lock rotation of the part, from previous design the contact points was on top of two planes very close where points C1 and C2 are defined so, profile of those two planes was controlled max 0.06. With this new design they want to control 4 contact independent areas of the metal sphere going in a spherical machined feature. I agree with you about having a single Profile of a Surface on each sphere will control what they want but as we all know, Germans tends to overcomplicate engineering

Thanks for your replay