This year team 708, Hatters Robotics, which is located in MAR built a new, custom drivetrain. This drivetrain was conceptualized on the ride to world championships last year and became an off-season project. We competed with this drivetrain at Team 341's off-season competition, Ramp Riot. With this our alliance made it all the way to Finals and managed to beat some teams that had competed on Einstein this past season. This drivetrain is gear driven, but unlike most gearboxes, the box is a piece 1x3 aluminum tubing and serves as the 2 walls of the robot. This serves as protection for the gears and saves a considerable amount of space for electronics and manipulators. 3 cim motors went directly into the gearbox/drive module. When testing the efficiency of this system, a non-biased mentor of another team found that it operated with 98% efficiency In addition to this, the team also put fans on top of our motors which, even after a competition, made the motors cooler than the room. 8020 with a 1020 profile connected the two modules which allowed to you remove a module by taking out 4 bolts. This made for extreme modularity as we could have the entire drivetrain taken apart in less than 2 minutes which might be incredibly helpful in between competitions in the pits. The entire CAD for Autodesk software is available at http://www.team708.org/drivetrain.html Feel free to ask relevant questions in the chat, provided that it is related to the post.I’m excited to see a design like this out here! I have played around with a gear box concept like this for a while but never got comfortable enough with it to prototype.
http://www.chiefdelphi.com/forums/attachment.php?attachmentid=15702&d=1388683349
Did you disassemble the CIM pinion gears for inspection? I am curious how they held up, they can take a large load when the drive train rocks.
It was great to see you guys at Ramp Riot with the new system. Last time I talked to your team moderator he said you used steel gears for the prototype because you had them on hand, and they made the whole base weigh quite a bit. What weight are you at right now and have you calculated what it would be with aluminum gears?
I love these sleek designs where, from an outside observer, the drive system looks almost too simple. Where are the gearboxes??? 
When we had finished the competition, it did not visibly look as though the gear’s teeth were chewed up, however the team decided to keep the drive base intact so we could have practice matches against ourselves so it couldn’t be inspected extremely well.
According to the CAD, each module had weighed 18.34 lbs with the steel gears, so the base with now board, electronics, or manipulators weighed about 37 pounds and the real version was about that. Just working off of the CAD, each base would weigh 16.31 lbs so the same base drivetrain would be roughly 32.6 lbs instead. Using gears made from aluminum 7075 is something our team is planning on doing so our base weight is lower.
Is that 32.6lbs with all six cims?
Very nice, however, I doubt the 98% efficiency number. Spur gears have a maximum practical efficiency around 99% for a single reduction… how was this tested?
Anyway, super nice drivetrain. Gearbox in tube should be really interesting, and I love how you can remove the siderails if needed. That mitigates part of the problem with gearbox-in-tube.
I think Thomas was missing the meeting we calculated the efficiency. We used a clamp style ammeter and checked the current through the battery cables at full speed with the robot up on blocks. It worked out to right around 2.8 amps per cim and we compared that to the listed 2.7 amps of a cim at free speed. We were all surprised by the results but the improvements were noticeable. We ran the same 6:1 reduction last season with the am14u 3 cim gearbox and the new drive was noticeably quicker.
We have a meeting tonight so I will have Tom take a pic of the cim gears and get a more accurate weight. But the weight savings going from steel to Al gears would be around 5 lbs total.
And Griffin, you know our shop is open if you need to use any of our machines or want a closer look at the drive.
So 98% unloaded. That’s still really good.
How did you get 98% from those numbers?