2019 ans spherical radius issues

[Jim Poehler]

Good morning,
I am having some real difficulties trying to understand why 2019 R1 is returning some quite unbelievable numbers on a spherical radius measurement. I have a radius nominal of 5.167 and no matter what measurement construction I try, it returns a value of 5.187 to 5.189, no matter what adjustments are made at the lathe by the machinist. I have tried individual vector points, scans, all return similar values that I know are incorrect. I have even used the sphere center, as taken from the CAD model as the origin of the part and taken sample hits on the CAD surface and called out a PR for those points. The all should be 5.167 but instead they are all over the place, They range from 5.0456 to 5.1404. Any ideas? Has anyone else encountered such an issue with 2019 R1 and radial call outs or should I instead begin investigating the CAD itself? Thanks in advance for any help.

[AndersI]

Sounds like a loose tip or a fracture in the tip shaft...

[Jim Poehler]

Anders, thanks for the suggestion but I see the same issues when I run an offline program without the actual physical probe and cmm, just the cad model.

[UKCMM]

Are you sure it is a true spherical radius and not just a radius blending to a diameter.

[Jim Poehler]

The print calls out a spherical radius and the cad model also returns the correct SR. When centered on the CAD SR with individual points back to the cad surface, the polar radii are all over the place. They should be the same as the overall SR when called out as a sphere.

[Matthew D. Hoedeman]

If you do NOT have 'find noms on execute' marked, when you run offline, all it is doing is using the programmed nominals for all features and constructions. IF your nominals in your program ARE from the cad data, then the CAD data isn't a sphere.

[Jim Poehler]

I am running with find nominals marked. Whether or not the CAD data is a sphere our senior engineer is looking at right now.

[KIRBSTER269]

You check with a O.D. Mic? Maybe that sphere is not very uniform? Go around it several different ways, write down the results, to reassure that the part isn't bad?

[Jim Poehler]

Kirby, I can;t do that cause it is an inside radius. Kind of like a parabolic disc. Actually its part of a camera lens housing.

[Jim Poehler]

For anyone following this thread, apparently what works is a 3d distance from point to point. This gives the old PR info that used to work. Not sure what and when they broke this but PR from the location tab does not work and gives random values from the center of a sphere. 3d point to point give cad accurate information. If anyone wants to see if they can replicate this please do or is it just my copy of 2019 R1 that is broke. I know I used to use PR in the work around method for finding the radius value of a small arc diameter. Just does no work now.

[JEFMAN]

For anyone following this thread, apparently what works is a 3d distance from point to point. This gives the old PR info that used to work. Not sure what and when they broke this but PR from the location tab does not work and gives random values from the center of a sphere. 3d point to point give cad accurate information. If anyone wants to see if they can replicate this please do or is it just my copy of 2019 R1 that is broke. I know I used to use PR in the work around method for finding the radius value of a small arc diameter. Just does no work now.

For me, PR works only in 2D, not in 3D, so I can't see how it could work in 3D on a sphere...
For 3D features, I would ASSIGN/V1=SQRT(DOT(SPH1.HIT[1..SPH1.NUMHITS].XYZ-SPH1.XYZ,SPH1.HIT[1..SPH1.NUMHITS].XYZ-SPH1.XYZ)) which should give an array of radii.

[Jim Poehler]

Thanks Jefman,
Nice to know that. Unfortunately I rarely use assignments but will keep that code in mind if I need to use it again. Not quite sure I understand it though.

[AndersI]

The polar coordinate system is <PA, PR, Z>, PR being the 2D projection down to Z=0 - the higher up on the sphere, the shorter the PR is (zero on top).

I don't think PC-DMIS can do spherical coordinates, even though JEFMAN can :-)

[Jim Poehler]

Thanks Anders, nice to know as well.