Feature size changes when using pre-alignment vs local alignment

[[Ti...]]

I was reviewing a program by a co-worker today. The sphere diameter was too large so I looked at how it was created. I tried making another identical feature but just using the pre-alignment.  It came in 0.05 mm smaller and right at nominal.

I carefully looked at both 'restore point selections' on the mesh. And I toggled back-and-forth on both and could not see 1 single triangle difference. They looked identical. Is there an odd reason I am seeing this behavior?

 

Thanks.

[[Ti...]]

I'd like to add both spheres have the exact same sigma, residual, deviation, and selected points.  they measure 0.05 difference in diameter and toggling between both selected triangles on the actual feature look 100% identical, no triangles change. Perhaps send to support?

[[Be...]]

Hi,

this sounds strange. Sometimes only one point of a triangle is selected. This leads to the fact that the triangle is not rendered as selected but the selected points are used for fitting. If also the selected points are identical you should send your data to the GOM support for further investigations.

Regards,

Bernd

[[Lu...]]

There is not such thing as "two identical features". in over the years scanning and inspecting parts, i have never had two identical results, there are very very close but still not identical.

First, are you using CAD data to create nominal data and applying meas.principle?

if you are then you have better chances to get a second feature close to the first one.

if you not using CAD then depends where you click on the mesh to create the feature and believe me you will waste your time trying to click in the same spot twice.

Is nearly impossible to look at the  restored point selection data and compare because is thousands of points.

Also an spherical feature is to me the worst shape to scan, it will have noise no matter what, it always deflecting light back to the camera, but you can edit the scans and delete the noise if you want better results.

in the pass i experimented with ceramic gauge blocks and the square ones ware within

+/-.0005 but the spheres the best i got was +/-0.0015 (like i measured 1 inch diameter sphere

and i will get 1.0015").

i hope you take these comments as a healthy way to learn.

[[Ti...]]

I'll send in a support ticket on this...it is very strange behavior.

I'll reiterate what I said before. The sphere was measured using Gaussian 3 sigma against a CAD model in a local alignment. I'm not a fan of having a bunch of a alignments running around so I created an identical feature using the pre-alignment. The pre-alignment and the local alignment are very different. Using the pre-alignment, the diameter check became 0.05mm smalller and was right near nominal. I checked the selected triangles visually on both methods and they looked identical, then the sigma, residual, deviation, and selected points were all the same numbers.

I had 3 people try to find any difference in triangles selected when toggling back from each method and nobody could find one. That's when I investigated the sigma, residual, dev, and selected points and found those had the exact same numbers.