Reference points alignment problem

[jvenlet]

Anyone have a lot of experience with PolyWorks + Romer Arm + Part Fixtures? I'm losing my mind, pulling out my hair, banging my head against a wall, etc.

I'm having an issue with aligning to a part holding fixture with the Romer arm on PolyWorks versión 2014 IR18 (build 7338).

We have a part fixture which we have aligned to 3 bushings using the Reference points alignment. At this point, everything seems to be fine. All of our deviations are 0 along the alignment axes. Watching the probe in real time on CAD shows the prone touching exactly where it is expected.

When we go to align on the datum nets that the part will actually sit on, using a reference points alignment (after the first iteration, when you get to fine-tune the points locations), all of the data drops ~6.00 mm. Looking at the bushings locations, they all show -6.00 in the primary axis (Z). I cannot figure out why the data no longer stays on CAD. I have tried linking to the part CAD as reference object, linking to the fixture CAD as reference object, and linking to no reference object. Oddly, the alignment deviations show up within tolerance, even though all of the hits, and the 4-way and 3-way locators show up 6 to 8mm low in Z. It's not even skewing the data, its just straight shifted by about the diameter of the ruby...

I've even re-programmed everything from scratch and still no change.

Also, I have checked the fixture on the CMM. The bushings and the datum nets are NOT 6mm out in relation to each other.

[Jeff]

What diameter tip are you using? It sounds like you vectors are wrong and it is compensating in for the tip in the wrong direction.

[jvenlet]

6mm tip, and that almost makes sense.

Unfortunately, that would mean that PolyWorks is improperly showing my CAD surfaces because the CAD surfaces appear correctly oriented.  The original program was written using the fixture CAD, which would (should) have set the vector correctly for the actual measurement.  Then changing to part CAD (which would be inverted compared to the measurement) the exact same things happens.

I will try inverting those surface elements, when I get a chance.

Thanks  :)