[DampRobot]

First, I'd like to say that this is shaping up as a really good thread. I haven't seen this level of discussion of design on CD in a while.

I'd like to add one more thing about how frame design effects the live vs. dead axle choice (and visa versa).

It's often in a team's best interest to protect their chain runs behind a frame member. Of course, there are exceptions, but it's generally worth it to put a chain run away from where game pieces, other robots, and debris can get caught up in it.

In a dead axle setup, the chain necessarily needs to be adjacent to a wheel. In many cases, it's even on the outside of the wheel. If teams want to have a frame member between the field and their chain run, it needs to go on the outside of the wheel. Because it's "right there," often teams make the decision to use that frame member to hold the other end of their dead shaft, and to use that shaft as a frame stiffener (as others have mentioned).

With live shafts, it becomes possible to separate where the wheel and the chain are. Sure, you can make them adjacent, but if you do, it's usually easier to go with a dead shaft setup (because it becomes desirable to have an outside frame member, and so on). With the live shaft, you can put the wheels on the outside of the frame, and keep the chain protected on the inside. If you've got that, you've got a drive with many of the aspects of a WCD.

Of course, as Andrew eloquently pointed out, you can do live sheetmetal drives, and WCD like drivetrains with dead axles (I guess I define a WCD kind of narrowly). And, there's nothing keeping you from direct gearing a wheel with a sheetmetal or dead axle setup. Just ask 971.

Perhaps I'm just saying that drivetrains tend to converge on either a WCD setup (sliding bearing blocks, cantilevered hex axles, aluminium box construction, center wheel direct driven off the gearbox) or a sheetmetal drive design (two side sheets with the wheels and chain runs in between, dead axles). Which one you chose really depends on what manufacturing resources you have. And, of course, there's nothing keeping you from modifying one of the more typical approaches into something that fits your team better.