[2021.2 SP1] GeoTol TP calculation is wrong?

- Fully constrained alignment
- Two holes located in a plane in X
- The plane is the datum for a TP for these two holes
- GeoTol calculates the TP according to the hole positions in the alignment, but displays the coords as if in GeoTols datum alignment?

From what I can gather, the actual distance between the measured features are 29.985 between each other, which if GeoTol calculated this
correctly would be in tolerance.

Parents

NinjaBadger over 2 years ago

Actually, I see from the location dimensions they're both approx 2mm lower than nominal. GeometricTols has shifted them around 1mm up.

I'm hazarding a guess it's doing a LeastSq best fit on them (because of the maths you can get the same number from a situation where you have a one positive and one negative deviation, as if you have two deviations in the same direction) - it's why we have a Vector Least Sq algorithm in the Best Fit alignments.

Out of interest, if you did a Least Sq best fit alignment on the two holes (2D Rotate and Translate in the X workplane) and then output the Location again do you see around a 1mm deviation after the best fit?
Cancel
Vote Up 0 Vote Down

Sign in to reply

Cancel

Reply

NinjaBadger over 2 years ago

Actually, I see from the location dimensions they're both approx 2mm lower than nominal. GeometricTols has shifted them around 1mm up.

I'm hazarding a guess it's doing a LeastSq best fit on them (because of the maths you can get the same number from a situation where you have a one positive and one negative deviation, as if you have two deviations in the same direction) - it's why we have a Vector Least Sq algorithm in the Best Fit alignments.

Out of interest, if you did a Least Sq best fit alignment on the two holes (2D Rotate and Translate in the X workplane) and then output the Location again do you see around a 1mm deviation after the best fit?
Cancel
Vote Up 0 Vote Down

Sign in to reply

Cancel

Children

No Data