I don't get it...

Nugget

rollers_test.skpI'm attaching the problem (SU-8, SP-3.2) Take away the bar, and the rollers spin just dandy. Take away the rollers, and the bar drops as expected. Leave everything as it is, and the bar gets "clogged" in the rollers. The rollers I expect to have a "surface," and the bar I also expect to have a "surface." Consequently, I make sure in my arrangement that there is no contact between the "surfaces," and if you look closely, you'll see a gap between the bar and all the rollers on both sides. What happens, though, is that these "surfaces" get caught within other surfaces, as if they were all made of adhesives. What am I not getting, here? Thanks in advance.

Nugget

I still don't get it, but it's coming into shape. I simplified the example (which already was a gross simplification of it), and noticed that with just two rollers across from each other (with the gap), the rollers somehow interfaced across the gap. I intuited that my problem might be one of scale, so I redrew it scaled at 20 times. The hinge objects now fit inside the rollers, and don't intersect one another in a side-by-side configuration. The simplified, scaled example "works," but strangely, so I still don't get it. Why are the rollers getting batted about by a completely, ideally smooth surface?

Attaching the latest update.

rollers_test_2.skp

Nugget

Something to do with the bounding box...

Making the cylinder a convex hull makes it completely smooth and then it doesn't turn when brushed by the dropping bar, regardless of how close the rollers are, or even if they are actually intersecting it. Magnetism doesn't help. Urf...

Maybe such mechanical kinds of purely interactions are not possible with this add-on?

chri

hi

Perhaps , it would be a good thinks to try readback collision géometry to understand .

Please , take a look on the next pictures .

first time , you select readback collision géometry for your parts
in a second time , you run the simulation
in third , you will see collisions .

Bye - excuse me for my bad english

chri

readback collision géometry 1.png

readback collision géometry 2.png

Nugget

You have been most gracious in your assistance. Thank you. That was exactly the clue I needed. I need to avoid rotating and scaling, or at least be certain the actual body corresponds to the collision "body". Thanks very much! Here's the working version of the example:
rollers_test_3.skp

chri

Hi

marvelous

if you'd like , you can see this topic
http://sketchucation.com/forums/viewtopic.php?f=61&t=46796

bye

chri