rhitee05
10 kW
I wanted to branch off the >2 ratio drive concepts from JennyB's thread to avoid confusion and hopefully attract more interest. The goal of this thread is to discuss ideas for multi-ratio and CVT-like friction drives. I have a couple of ideas and hope others will chime in with comments or their own ideas. I'm hoping we'll be able to come up with a design promising enough to try a prototype!
Motivation:
Several people have been having good luck with their single-speed friction drives, but experience with a variety of drive types has shown that having gears available (electrical or mechanical) is always better. Especially when using RC ESCs, its useful to be able to avoid low-speed, high-current operation. A few two-gear ideas are being tossed around in the thread linked below. If two gears are good, more are better! The more available ratios, the closer more we are able to operate the drive at a sweet spot of efficiency. I think a useful multi-ratio drive has to span a range of gears at least 2:1. For reference, the 9-speed cluster on my bike has a range of 2.7:1.
Original thread link:
http://endless-sphere.com/forums/viewtopic.php?f=28&t=21342&start=0
Here is a quick drawing of the original concept I had:
It's a very simple design. The roller surface is a circular arc around the pivot point (the small circle at upper left). As the whole assembly rotates, the radius of the roller gets smaller and the ratio gets higher. I've drawn it with the motor mounted above and belt drive, but you could also do a direct-drive. An advantage to the belt is it gives you another reduction step so there's more freedom in the choice of roller diameter and motor Kv. A potential issue with this concept is that there will be some difference in speed across the contact patch with the tire. This might not be an issue, or it could cause some additional wear. Distance to the tire does not change so the pressure would remain constant.
To address that problem, I came up with a second concept:
This concept has several discrete ratios rather than being a CVT. There's a section of constant radius for each ratio, so this would avoid the differential speed issue from the CVT. The section widths and spacings are chosen to try and follow the pivot arc so the distance to the tire is about the same. The mount would have to move slightly to absorb the changes is distance as the gears shift. A disadvantage is that the contact point with the tire will rotate from the top slightly toward the side as the drive goes to the highest ratio. I also think the control mechanism would have to be more complex. The CVT can be controlled closed-loop, so the actuator doesn't need to know the position of the pivot, it only needs to know when it hits the min/max stops. For the multi-ratio drive you would need to use sensors to lock the drive in one of the fixed positons.
Construction:
I'm a fan of wood for ease of prototyping and because I don't have access to a machine shop. It would be relatively easy to make a wood roller on a lathe to any arbitrary profile. A hardwood like oak would probably be the most suitable. The finished roller could then be repeatedly dunked in polyurethane to build up a nice, thick coating. The resulting surface should be soft enough to get good grip and not cause much wear on the tire, but hard enough to last for a while. Skateboard, roller skate, etc. wheels are made of polyurethane, after all. When the coating does wear, the roller can be sanded and re-coated. A wood roller would also be lighter and have less inertia than a solid metal roller of similar radius.
My initial concept would probably use a roller with 4" max diameter, 4" long, and based on an 8" radius to the pivot point. Such a roller would require about 2500 RPM to reach 30 MPH. That's probably a bit low to be driven directly by an RC motor, but a slight reduction in the belt drive (less than 2:1) would bring it in-line. Something around ~1.7:1 would work for a 170 Kv motor at 24V. I haven't come up with a way yet to make the ratio self-adjusting, so an RC servo and some simple control electronics would be required.
Motivation:
Several people have been having good luck with their single-speed friction drives, but experience with a variety of drive types has shown that having gears available (electrical or mechanical) is always better. Especially when using RC ESCs, its useful to be able to avoid low-speed, high-current operation. A few two-gear ideas are being tossed around in the thread linked below. If two gears are good, more are better! The more available ratios, the closer more we are able to operate the drive at a sweet spot of efficiency. I think a useful multi-ratio drive has to span a range of gears at least 2:1. For reference, the 9-speed cluster on my bike has a range of 2.7:1.
Original thread link:
http://endless-sphere.com/forums/viewtopic.php?f=28&t=21342&start=0
Here is a quick drawing of the original concept I had:
It's a very simple design. The roller surface is a circular arc around the pivot point (the small circle at upper left). As the whole assembly rotates, the radius of the roller gets smaller and the ratio gets higher. I've drawn it with the motor mounted above and belt drive, but you could also do a direct-drive. An advantage to the belt is it gives you another reduction step so there's more freedom in the choice of roller diameter and motor Kv. A potential issue with this concept is that there will be some difference in speed across the contact patch with the tire. This might not be an issue, or it could cause some additional wear. Distance to the tire does not change so the pressure would remain constant.
To address that problem, I came up with a second concept:
This concept has several discrete ratios rather than being a CVT. There's a section of constant radius for each ratio, so this would avoid the differential speed issue from the CVT. The section widths and spacings are chosen to try and follow the pivot arc so the distance to the tire is about the same. The mount would have to move slightly to absorb the changes is distance as the gears shift. A disadvantage is that the contact point with the tire will rotate from the top slightly toward the side as the drive goes to the highest ratio. I also think the control mechanism would have to be more complex. The CVT can be controlled closed-loop, so the actuator doesn't need to know the position of the pivot, it only needs to know when it hits the min/max stops. For the multi-ratio drive you would need to use sensors to lock the drive in one of the fixed positons.
Construction:
I'm a fan of wood for ease of prototyping and because I don't have access to a machine shop. It would be relatively easy to make a wood roller on a lathe to any arbitrary profile. A hardwood like oak would probably be the most suitable. The finished roller could then be repeatedly dunked in polyurethane to build up a nice, thick coating. The resulting surface should be soft enough to get good grip and not cause much wear on the tire, but hard enough to last for a while. Skateboard, roller skate, etc. wheels are made of polyurethane, after all. When the coating does wear, the roller can be sanded and re-coated. A wood roller would also be lighter and have less inertia than a solid metal roller of similar radius.
My initial concept would probably use a roller with 4" max diameter, 4" long, and based on an 8" radius to the pivot point. Such a roller would require about 2500 RPM to reach 30 MPH. That's probably a bit low to be driven directly by an RC motor, but a slight reduction in the belt drive (less than 2:1) would bring it in-line. Something around ~1.7:1 would work for a 170 Kv motor at 24V. I haven't come up with a way yet to make the ratio self-adjusting, so an RC servo and some simple control electronics would be required.