Way back in 2019, I was a sophomore learning design in the off-season. My team was making a crazy intake mounted on this double-jointed arm, and we needed to reverse the wheel directions off of a single motor. Here’s the robot: https://youtu.be/q8uYv1yMzXI
To reverse the direction we used 4 3d printed gears, but our lead design mentor really wanted to do an infinity timing belt at the time(another mentor stepped in and told him he was crazy). However this year I saw Killer Bees with one on their shooter:
Absolutely fantastic and precisely what you would expect from such a top-notch team! But does anyone know how to replicate it? The smooth roller paired with 3mm belt seems critical, but I have no idea of how to find center to center distance calculations or if it’s even possible to do it mathematically.
If anyone happens to know how to pull it off, I’d love to hear how and try to test it out for my team!
We did it on out 2020 robot but used round belting material that you cut to length and then melted the ends back together. So we didn’t have to calculate the length and I’m not sure how one would be calculated. The round belting material worked great since you didn’t have to worry about it rubbing against itself and tearing itself up. This was also on a low torque application not a shooter mechanism. I know this doesn’t help much but it gives you an idea on how to do it.
We did it with round poly cord for our shooter this year. It worked great with no issues! The cut-to-length and weld together nature of poly cord really does make it perfect for this application, but you need to allow that some slippage is likely to occur. For a shooter application, though, that slippage turned out to be minimal, and solved all of our problems with shots bouncing out. Just take a look at the difference between the first few matches at Northern Lights and our shots at North Star (we added the second wheel between events).
First time I’ve seen a toothed belt used in a figure 8 configuration. As the others have mentioned we have used the round poly belt in a figure 8 configuration on our robot. Last time it was to have one motor drive two items in an opposite direction. This year however it was a driving a single item and we did it to get around a frame member. It didn’t matter that it was changing the direction.
To lower slipping and/or reduce the necessary tension you can do two side by side.
The other plus in a single driven item application is that it reduces the potential for slip when you have a small pulley driving a large pulley as you get greater wrap on the small pulley.
Yes you do get some wear where they touch in the middle, but it is very minor, our applications have lasted multiple events with just a slight sign of some ploy dust.
Trial and error with belts that are close to the perimeter needed (wrap a string in the configuration and just measure) is probably the fastest way to find the correct belt assuming you have a decent supply of teeth counts on hand. There may be a calculator out there that takes into account the 3D nature of this configuration.
This is pretty common in my experience, works well on short runs, or longer runs with larger diameter pulleys (as long as the teeth are not close to interlocking). Sure there is some efficiency loss.
We decided to do this for weight savings reasons (lose the weight of 2 gears, 2 bearings, and a shaft). To find the center distance we made an educated guess with a 2D sketch where the belt run lines were tangent to the “other side” of the pitch circle of the pulleys. Obviously this was wrong, so we reprinted the pulleys until it worked, and tested using a TPU 3D printed belt while we waited for the real belt to arrive.
Anyways, we were extremely happy with the performance of the twisted belt, so we want to do an off-season project of making a calculator for the correct center distances of a twisted belt. This will probably happen in the fall, but I’ll make sure to post the findings.
We ran a “roll-ie” separator pin just like 33 did but not sure on how necessary it was. 2771 also ran twisted tooth belts on their shooter with no separator. Whether a separator is necessary probably depends on the particular geometry/application along with how mentally tolerant you are to seeing the belt rub on itself.
That’s awesome, I can’t wait to see the calculator! Do you think you think you’ll end up making it into an onshape featurescript too?
Did you also end up having to modify the inner spacer to a specific diameter to allow for the lack of rubbing? And is it just HDPE on an 8-32 or is it something crazier?
To pull this off and have a strong enough grip around the pulley to transfer energy, I think you would need two idlers to wrap 180 around the pulley. You could also slot them in that configuration to make sure you have perfect tension. It’s probably a net equals or worse though on just belting and gearing unfortunately.
We did this in 2020 when our polycord belts didn’t transfer enough torque. You are actually able to twist the belt to make it so the teeth don’t touch, which only makes it worse for calculating belt length.
You could model the belt, I’ve done this for a couple of belt runs with decent results using this method. You would need to use a 3D sketch for the belt pitch line, not impossible but a little trickier than a 2D belt run.
I could see a setup with one large idler working. The catch is that it needs to be placed correctly enough to give enough wrap.
Also remember: You don’t need to wrap 180 on any given pulley–you need to wrap enough that it doesn’t slip. (The usual number I see when doing calculations is 6 teeth, but I take that with a big grain of salt.)
Most calculators take the tangency points where the straight lines of the belt touch the circle of the pulley, calculate the straight line distance between those points and then add the pulley wrap distance (and then obviously adding a small amount to the C-C distance to create the correct tension, etc). If the pulleys ae far enough apart that the half twist of the belt does not really affect the length of the belt run between the two pulleys, then it seems like a pretty trivial exercise to use the tangency points from the top of one pulley to the bottom of the other as the straight run length (rather than the traditional top to top, bottom to bottom lengths) and then obviously use the wrap lengths between those same tangency points as well.
Originally a single wheeled shooter, we CNC’d small extension mounts for bearings on each side, bolted them to the existing holes for the backboard supports (blue churro’s, you can see them in the picture). 3d printed the pulleys (using TTB inserts!), and cut/welded the polycord to size. Honestly, the hardest part of the whole thing was replacing the existing shaft through the main shooter wheel with a longer one. This is mirrored on both sides, as we were a little worried about slipping.
