"Default" Math Vs "Legacy"?

good to know! thanks for the clarification.

louisd
#26.1
louisd commented
Yesterday, 06:06 PM
It seems like the "I can't validate its output" is a function of your level of experience and knowledge.
You don't use your gut to validate confidence in a process, you use math.​

Well said louisd and I totally agree.

For the record, we recently went through a validation process with a well known US aerospace OEM. They have their own, internal process for software validation which they perform every couple of years to inform their decision on which metrology software packages they use and whether they update to a new version or not. They fully approved our new math and GD&T capabilities and are in the process of disseminating that information to their in-house CMM programmers and suppliers. Hexagon is also working closely with NIST in an effort to have them verify our "new math" - constrained L2 (which, incidentally, they supported as the default datum math solution when proposed to both ASME and ISO) . Historically, NIST have only offered certification for Least Squares calculations. I believe PTB (Germany) offer Least Squares and Chebyshev certification but, to date, nobody has been able to provide certification for GD&T.​

neil.challinor As far as I know the math for fitting features is what PTB and NIST test. I am only aware of one software package that was actually accepted for GD&T by a 'well known aerospace company' and it wasn't PC-DMIS. Your are right in that this is not something actually tested by organizations such as PTB or NIST but your comments seems to float back and fourth between feature fitting and GD&T which is a bit misleading.

Oak Ridge National Laboratories has their own software

for your viewing pleasure...

https://www.researchgate.net/publication/308021435_A_Constrained_L2_Based_Algorithm_for_Standardized_Planar_Datum_Establishment

KUDOS TO ALL POSTERS!

Really good practical tests and applications: there is a difference between theory and practice!

Thanks NEIL! :Those constructions are not necessary when working with the geometric tolerance command because it performs the constrained L2 calculation automatically in the background when you have DEFAULT datum math and DEFAULT feature math selected. As you are seeing, when working with the geometric tolerance command, most constructions are unnecessary and can actually get in the way.

I thought it was memory leaks or a runaway script infecting legacy with "New Math". I know it changed, and now I now it was intentional. Thats mvaluable info


I like the Calypso flavor . I think the new MIN/MAX with two separate centroids is valuable, I would like to see the diameter (a size feature) re-comped around and origin that was passed through a roundness style filter as a confidence check

with centroid of FORM (Roundess), centroid of MIN /CIR, CENTROID of MAX CIRC


​​​​​​​CHEERS