Hi Michael, all of STEP, SAT, and IGES exports go through the same import/export library and the library must be running into some error while it traverses the geometry to export. I'll investigate it to see what's going on.

- Michael

Hi Michael, it seems that the problem has to do with a kind of problematic trim boundary in this particular area here:




One of the trim boundaries in that area has a little edge piece that kind of sticks out and then doubles back on itself, it seems that this is confusing some part of the processing in the IO library. I'll see if it is possible for it to just ignore the error and keep doing the export but in the meantime I've attached a tuned up version of the model where I have untrimmed the surface in that area that had the doubled-back edge in it and retrimmed it to have a clean boundary and this version now seems to export to those formats ok now.

You may want to work on a model like this that has some pretty tiny little sized details in it at either 10 times or 100 times scale and then scale it down at the end when you're done - some commands target a fitting tolerance of 0.001 units in size and when you have little edges and surfaces that are about that size some stuff can get kind of glommed together in awkward ways.

For a couple of steps in the retrimming operation, I scaled the model up by 100 times in size and then did the trim.

- Michael

Michael-

Thanks for the help and the helpful information. I started this particular project at a larger size and was convinced by our CAD department that I should be working to scale. I've been having a lot more problems with little things, here and there, once I started working to scale, but I certainly didn't put the two things together. I'll make sure that I keep that in mind. Thanks so much for the help and insight!

Hi Michael, it's theoretically nice to be working at scale, especially when working with multiple connecting parts that may be imported from separate files with multiple different people working on them and stuff like that. That's probably part of the process that your CAD department is thinking of when giving that advice. The problems from that kind of multi-user process don't really apply so much to a project that's focused on just one single object that only one person is working on though.

The problem with features that are pretty tiny in size (like around 0.001 units or so across basically) is that many functions need to work at a tolerance where they either glue things together that are close enough or keep refining the result until it is accurate enough, and if these things end up thinking that 2 edges are close enough to be glued together for example but gluing them together makes an awkward self-intersecting shape then that can lead to structure problems.

I'm on a kind of long ranging mission to make MoI handle working with little tiny details more robust by having a mechanism that tightens up the tolerance automatically to be some relative fraction of the size of the stuff that you are currently working on. But that's kind of an ongoing work in progress, there are still enough things that work on a 0.001 unit tolerance that it tends to be better to not have entire surfaces being created at that size.

The "tighten tolerance to relative size" type thing is somewhat delicate because it's also not so good for the tolerance to get too tiny as well, if it gets too small things that should get joined together might not get joined, and some operations that keep refining results until they are accurate enough could take way way longer to finish and generate really dense and heavy results.

So anyway if you're working with an object that when modeled at scale would have individual details much smaller than say around 0.05 units in size you will probably be better off doing most of the work on a 10x or even 100x (if details are particularly small) scaled up version and then scaling down at the end. It's just the numeric size that you want to consider, not any specific physical size, like it doesn't matter if you're working in kilometer units or mm units it's that you don't want to be working with numbers that start to have quite a bunch of zeros in them after the decimal point.

- Michael