Bance wrote: ↑Wed Oct 05, 2022 9:11 am
In terms of contact, there is a well known problem in OCCT that occurs in certain situations, we refer to this as the Co-planar issue.
Ok, I understand. If I apply this to my own experience, it opens up previously unknown possibilities. In my first case, I "subsequently glued" a pad based on the same plane, positive in the initial case and reversed in the "glued" case.
This usually manifests itself where solids are tangentially connected, overlap is the most secure way.
In my second test attempt I had used a symmetrical pad, adding a second pad on the "positive" side did not cause any problems, but on the "reversed" side it did. Now I learned at some point that the solver performs an adjustment against convergent conditions. And this is based on floating point arithmetic. I guess that went just fine on the positive side, while it led to a problem on the reversed side. As a solution, I would suggest not to negotiate the necessary offset for such manipulations to the decimal place, but to take into account a small deduction, which should then guarantee the overlap.