Rolling Contact

matthew_stacy

Alright you clever people.  I've got a roller tangent-mated to the faces of a curved track.  How do I make the roller spin one revolution for every circumference traveled (PI*D per revolution)?  Ultimately the intent is to create a curved rack & pinion.

Insert dummy sketches/surfaces/parts???  What's the best way to attack this task in Onshape?

https://cad.onshape.com/documents/69889dc07c27ca6f06c4b818/w/ce8642eb62a37671e22d3576/e/c13ca2ea499288cd966187cd

 

john_mcclary

Thanks Evan, 

It works pretty much like the sketch i drew earlier. But i stole the first premade chain model i could remember. Because i made them enough times to knew it takes a while. 

Basically, the cylinder is rotating pi*D times for each mm of distance traveled. Which is just a rack and pinion mate.

Now create a track in this case a U shape, but i don't see why it couldn't be a more complicated surface, as long as Onshape can handle it. I know in the past long chain animations fail easy.

Hind sight.. Probably don't need the chain at all. May have over complicated things. But it was something I was familiar enough with that could attach to a slider mate

May try again with my old mining cart model. When I get more free time.

TimRice

I imagine @john_mcclary is warming up his thinking hat for this one 

EvanReese

This makes my brain hurt to think about. I'm dropping a comment to be notified when the clever people show up 

matthew_stacy

More coffee @Evan_Reese!  You can do this.

@TimRice we've got cool software, talented support staff and a great user community.  I'm eager to see what @john_mcclary, @alnis_smidchens, @mahir and so many others come up with.  'Till then I'll twiddle my thumbs.

john_mcclary

I think it would just be a rack and pinion mate that drives the rotation, then have the pin follow a path that is offset around the U-Shape.

tim_hess427

@john_mcclary - how would you have the pin follow the path? 

I played around with this. One issue I ran into was that the pin would rotate the wrong way when it gets to the other side of the "U" shape because the rack and pinion relation relates the rotation to the +/- direction of an axis and NOT the +/- translation along a path. That means it also didn't rotate at the correct speed around the curve.

mahir

I don't think it can be as simple as a rack and pinion because there is no true linear or rotational degree of freedom to work with. R&P only works if you have something like a revolute and a slider mate.

However, I have a thought. How about two tangent mates. One regular one between the pin and the path surface. And one dummy tangent mate between cycloid surface and a dummy cylindrical surface with small radius that is tied to a specific spot on the pin surface. Tricky part would be generating the cycloid. It would have to follow the U path and also not have any sharp points that would trip up the second tangent mate.

john_mcclary

I was thinking a rack and pinion related to a linear dummy object that is the same overall length of the path "laid flat"

then you could make a dummy cylinder part that is tangent to the U that doesn't have rotational freedom (always points north..)

now you may need another dummy part that is flexable (like a chain) that can curve around the U.
Mate the end of the chain to the Dummy cilinder (actually this last link could double as that dummy part)

I can't really get to making a real concept yet, but here is the jist

john_mcclary

You actually may still need a dummy washer between the blue and green part, it needs to stay normal to the world so it doesn't affect the pinion rotation speed, i think 

mahir

I see where you're going. But to get that translation from path length to linear length is pretty messy. I'll work on the cycloid, and you can work on the menagerie of dummy chain links

john_mcclary

you can start in @NeilCooke's old cable carrier document I guess

Energy Chain - Copy | Energy Chain (onshape.com)

mahir

I got reasonably far, but hit a snag when trying to place the second tangent mate. Does tangent not support polynomial extruded surfaces?

https://cad.onshape.com/documents/1511f9c77d7c79b7bcecd27f/w/c55b8a7d4bf01e1c1d289c24/e/b7d978a222720ff457978682

john_mcclary

I did a quick and dirty edit of that cat track file.

proof of concept works.

I'll leave it to you to fine tune all the dimensions and parts...
Energy Chain - Copy | Energy Chain (onshape.com)

mahir

Yeah, that works better than the double tangent. It was buggy at best.

john_mcclary

Yea i was a little surprised when i saw your wavy curve. I wouldn't even know how to sketch that 😉

EvanReese

daaang! this was worth the read. Not gonna lie, I was team cycloid, but John's done it again. I thought of the cable carrier too, but definitely didn't (and still don't totally) understand how to turn that into this. 👏👏👏

john_mcclary

Thanks Evan, 

It works pretty much like the sketch i drew earlier. But i stole the first premade chain model i could remember. Because i made them enough times to knew it takes a while. 

Basically, the cylinder is rotating pi*D times for each mm of distance traveled. Which is just a rack and pinion mate.

Now create a track in this case a U shape, but i don't see why it couldn't be a more complicated surface, as long as Onshape can handle it. I know in the past long chain animations fail easy.

Hind sight.. Probably don't need the chain at all. May have over complicated things. But it was something I was familiar enough with that could attach to a slider mate

May try again with my old mining cart model. When I get more free time.

tim_hess427

I think this all boils down to this: We need some sort of "path" mate. 

This is difficult because all of the mates (except tangent) are limited to straight line motion or rotation and because the path fully reverses on itself. If there was a to just say "make this mate connector follow this path" and then make relations based on distance translated, this would be pretty trivial.

john_mcclary

agreed, still would like a real path mate

matthew_stacy

@john_mcclary, @tim_hess427, @mahir, and everyone else who put in so much thought and effort on this conundrum,

THANK YOU!

Chain / slider / revolute / rack-&-pinion is an interesting approach.  I had limited success with that, but agree that chains seem to be computationally unstable in Onshape.  They WORK ... until they blowup.  I was able to reduce the chain link to a simple sketch (straight line segment with a circle on each end) rather than modeling it as solid or surface.

I also find it interesting that the tangent mate behaves differently on surfaces and solid.  With surfaces, only a single face can be selected.  Whereas with a solid, all adjacent tangent faces are included.

Let's hope for a path mate in a forthcoming OS update!

john_mcclary

Right, using a simple sketch instead of 3d parts for the chain should work just as good
Good Idea