CAD Forum

Posting this ASM based on a question over at the Swamp. Tangency mate has a tendency to flip if you use just the geometry. For something like this using a sketched line along the bottom of the rod and mating that tangent to the roller will solve the problem here...If I read the question correctly. See attached ASM.
Here's the post over there:
https://r1132100503382-eu1-3dswym.3dexp ... UhC1mlKe9Q

The problem is that toroidal surface on the roller. Can't really mate to it. We have a similar thing on our door rollers that travel on a rail.

Created a circular centerline sketch on the roller surface for mating. Then mated it tangent to the center plane on the rod.

Tried creating another sketch circular centerline on the toroidal surface instead of at its centerline. Then mated that tangent to the rod outer surface. Was a little worried it would flip the mate but it didn't.

Spoke too soon....second method jumps the roller to the other side of the rod if you move it too fast. You can still simulate the motion but you can't move it too fast.

jcapriotti wrote: ↑Thu Aug 19, 2021 10:40 pm Spoke too soon....second method jumps the roller to the other side of the rod if you move it too fast. You can still simulate the motion but you can't move it too fast.

So, I think it'll work if you put a line on the bottom of the rod and mate that sketch to the sketch of the roller. Then it can't jump because it's constrained...I think...Maybe not...But, I think it would be a little more robust?

I think we may be touching on something from the old forum, where they asked to be able to restrain a mate within a sketched line or arc segment. In this case instead of creating a sketched circle, you'd sketch an arc say 90 degrees, and then it would be impossible for the mate to "jump" to the other side because there wouldn't be a sketch segment there to jump to...But, I may be misremembering that discussion, and I haven't had enough coffee yet...

I'm not crazy about mates to lines if I can avoid it, so instead of drawing a line at the bottom I think I would create a plane there instead. Similar to the first solution but more robust I think for cases where the roller groove center is not the same dimension as the rod.

jcapriotti wrote: ↑Fri Aug 20, 2021 9:53 am I'm not crazy about mates to lines if I can avoid it, so instead of drawing a line at the bottom I think I would create a plane there instead. Similar to the first solution but more robust I think for cases where the roller groove center is not the same dimension as the rod.

image.png

The problem with this is that it can also flip direction. Solidworks solvers seem to take the easiest path, that's why moving fast without solving a bunch of intermediate points makes the tangent mate flip. I had an engine model where one of the valve lifters was constantly flipping. I think maybe two points mated to a line or axis would stand a better chance. This flipping business has been around for decades, and understanding that planes and faces have normal vectors, you'd think they'd be able to keep the normals in the same relationship they were in as assigned, but this just seems beyond them. So lots of things keep flipping.

matt wrote: ↑Fri Aug 20, 2021 10:07 am The problem with this is that it can also flip direction. Solidworks solvers seem to take the easiest path, that's why moving fast without solving a bunch of intermediate points makes the tangent mate flip. I had an engine model where one of the valve lifters was constantly flipping. I think maybe two points mated to a line or axis would stand a better chance. This flipping business has been around for decades, and understanding that planes and faces have normal vectors, you'd think they'd be able to keep the normals in the same relationship they were in as assigned, but this just seems beyond them. So lots of things keep flipping.

Does SE handle this better?

matt wrote: ↑Fri Aug 20, 2021 10:07 am The problem with this is that it can also flip direction. Solidworks solvers seem to take the easiest path, that's why moving fast without solving a bunch of intermediate points makes the tangent mate flip. I had an engine model where one of the valve lifters was constantly flipping. I think maybe two points mated to a line or axis would stand a better chance. This flipping business has been around for decades, and understanding that planes and faces have normal vectors, you'd think they'd be able to keep the normals in the same relationship they were in as assigned, but this just seems beyond them. So lots of things keep flipping.

Points are tough here because when it moves the part slides along the rod and circular sketch. Maybe if I had made the roller separate so it could rotate, a point would work. I got lazy and made it all one part.

Solid "Flipping" Works

mattpeneguy wrote: ↑Fri Aug 20, 2021 10:25 am Does SE handle this better?

I don't have the same amount of experience working with assemblies and other things that works tends to flip in edge. I haven't seen the problem in edge, but that doesn't mean it's not there.

matt wrote: ↑Fri Aug 20, 2021 11:10 am I don't have the same amount of experience working with assemblies and other things that works tends to flip in edge. I haven't seen the problem in edge, but that doesn't mean it's not there.

I have not seen it flip anything yet. The big difference between SE mates and SWX mates is: Edge WILL NOT FLIP on purpose. If it ever flips it would be a bug. Even though Parasolid has flipping tendencies, Edge can prevent it MUCH better than SWX. I can't tell you how many times I've felt like beating my head on the desk because of SWX's "automated mate-flipping helpfulness".
Notice there is no way to opt-out. With Edge, if you flip one mate, it will cause errors in all the rest until you manually flip the ones you want. It sounds like more work, but I'd rather do it that way than having to delete and re-create mates to prevent flipping.

@mike miller Do you not get this? I always select No and don't ask again, unfortunately only good for the session.

Make sure this isn't unchecked. They should make it a drop list (Prompt, Change alignment, Don't change alignment)

jcapriotti wrote: ↑Fri Aug 20, 2021 11:42 am @mike miller Do you not get this? I always select No and don't ask again, unfortunately only good for the session.

image.png


Make sure this isn't unchecked. They should make it a drop list (Prompt, Change alignment, Don't change alignment)


image.png

Nope, I don't. Must be I've posted too many snarky remarks about DSS. Actually, yes. Once I found the dropdown halfway down the page and set it to prompt.....


(2021 SP1, BTW)
I do have to wonder....will this actually prevent mates from flipping accidentally in the middle of the night? I'm skeptical.
Just for fun, check out this video. It took me awhile to figure out the problem and then fix it. One mate in a sub-assy had flipped.

Forgive me if I've duplicated, but how about a distance mate from the roller axis to the rod's parallel plane.
I can't paste an image into this reply.

mike miller wrote: ↑Fri Aug 20, 2021 11:45 am
I do have to wonder....will this actually prevent mates from flipping accidentally in the middle of the night? I'm skeptical.

No. Nothing stops SW mates from flipping for no good reason. It just stops solidworks from trying to "fix" your mates for you.