hexagon logo

2022.1SP6 Parallelism issue

I have 2 planes, PL1 is datum A (primary using minimax), and PL2 constructed from 7 basic scan/circle/plane. (least square, but it shouldn't change anything !)
Using geotol, it gives 7,1 µm
using legacy, it gives 35,1 µm
using assignments (ASSIGN/V1=DOT(PL2.HIT[1..PL2.NUMHITS].XYZ-PL1.XYZ,PL1.IJK) , then ASSIGN/V2=MAX(V1)-MIN(V1)), it gives 36,1 µm.
Then, constructing 7 planes from scans and construct a plane from their centroids, it gives about 8 µm (so I'm wondering if geotol uses them to calculate instead of hits ?)

Don Ruggieri , neil.challinor
Parents
  • FCF parallelism (and perpendicularity) has reported much better than the individual hits indicate for years (most PC-DMIS versions). Congratulations, very few people question near-perfect numbers.


    How do you figure? I use CADGRAPH=ON and TEXT=ON and the results always correlate with what is reported. And when using Legacy, the results have been very similar...
  • I stand corrected, the undocumented enhancement is a bit more extensive than I realized. Looks like when using tangental planes, the constructed planes work as described in help files, but when asking software for dimensions, the reported values don't jive. Examples are purely test data, datum A being perfect, while the test plane is intentionally set to deliver a 1mm parallelism (although being a square that can be tilted equally 2 ways the computer should not have been able to construct single calculations). Note the ijk values of constructed plane types and reported parallelisms. (code and report to follow)...
Reply
  • I stand corrected, the undocumented enhancement is a bit more extensive than I realized. Looks like when using tangental planes, the constructed planes work as described in help files, but when asking software for dimensions, the reported values don't jive. Examples are purely test data, datum A being perfect, while the test plane is intentionally set to deliver a 1mm parallelism (although being a square that can be tilted equally 2 ways the computer should not have been able to construct single calculations). Note the ijk values of constructed plane types and reported parallelisms. (code and report to follow)...
Children
No Data