I’ve noticed a lot of teams have used urethane/poly cord flat belts in the past for ball related robots (https://www.mcmaster.com/urethane-belts). Seeing that this might be useful this year, I have a few questions on how to use them:
how do you determine the proper length that the belt needs to be with a given c-c distance? I’d imagine it needs a somewhat significant amount of tension to work.
why does using a crowned roller help with keeping the belt in place? maybe its just me, but that seems counter intuitive.
I can’t answer your first question as I have no experience with polycord, but I can try to answer the second question. Basically when the flat belt starts to wander to one side, tension is created on the other edge of the belt (the edge opposite the direction the belt is wandering) by the crowned roller. This tension pulls the belt back to center. If you look around on YouTube you can find some videos that explain this concept a bit better than I can through Chief Delphi.
I understand that (first capital just made a nice video), I was just more curious on how the length of the belt itself is determined. I’m assuming this isn’t like chain or timing belts, where the length is pretty straightforward, since it doesn’t need to be stretched to work well.
Starting with 9% over nominal, minimum tension and varying based on number of parallel belts you have on the same shaft has always worked for me. Belt is stretchy and forgiving so it hasn’t ever been too much of an exact science for us.
Crowned rollers are pretty magical. You can either print crowned pulleys or create a crown on an existing drum by layering tape. Here’s a solid visual explanation for the concept:
I’m just curious, but is there a more physics based breakdown of this explanation? how does the net tension force (i’m assuming its the tension force in the perpendicular direction that corrects this) point in the direction of correction?
It’s very much a guess and check thing, but I’ve used 5-7% shorter than path length for round belts, and 2-3% shorter than path length for flat belts around an inch wide.
This is only true for flat belts - for round you need a groove or comb or something. But for flat belts - the belt has a natural tendency to ride on that point, it is going faster than the rest of the roller due to the greater diameter. The belt kind of “hugs” the crown with that extra sideways tension seeming to cause it to recenter more easily. That’s my guess at least. In any case, that’s how it works - if you try and make it fall into a groove or something it absolutely won’t work like that.
Here’s a good explanation of how crowned rollers work. I’m still mystified by the physics magic, even though I kind of understand the principal. We used some with flat belting for our cargo intake last year and it worked awesome. Sideways forces on the belt can still push it off though if there is enough, just fyi.
I don’t know the physics for sure either, but could it have to do with the Poisson effect? So as the belt is stretched by the tension, it tends to get narrower. That creates a friction force from the belt at every point not on the center of the crown that points toward the center of the crown. If you take the belt off-center from the crown, the point of maximum strain also moves off center and thus the transverse contraction is stronger on the side of the belt near the crown than the side farther from the crown. The near side of the belt “digs in more” due to this increased frictional force from the transverse strain and the increased normal force from being the side with higher tension. Ultimately, the near side of the belt pulls harder to move itself down to the edge of the crown than the far side, despite the far side being having more surface area. If the edge of the belt slips all the way past the center of the crown, then the belt is going to continue sliding in that direction to a region of lower tension.
That’s a theory, at least.
… And it feels like it’s already being undermined/maybe supported(?) by @BryanLee’s post above me. I’m not sure.