This is an idea I had to make a smaller two speed gearbox. The CIMs drive the ball shifting gears directly. The intent is to add a second stage of reduction with gears or chain in the drive rail. With these gears (11:60 and 18:54) the spread is pretty close (1.82), but I think it’s workable. A modest second reduction stage can get you approximately 8 and 15 fps.
Three CIMs are shown, but my instinct would be to run only two to save weight and avoid brownouts.
The motors are too close together to fit the stubby air cylinder, so a standard one is used.
All of the ball shifter components are stock, except for the pneumatic coupler, which has to be modified to accept the 1/4-28 thread on the air cylinder.
CAD files can be viewed and downloaded here:
Any chance we can see the CAD for this?
I’ve tried this concept a few times (putting the shifter shaft in the first reduction), and the problem I consistently run into is that the gear ratios end up way too close together if the CIMs drive the ball shifter gears. I’ve tried fixing this by having the CIMs drive the ball shifter output and using the shifting gears as pinions for a second stage, but then the gearbox gets HUGE. How have you gotten around these problems?
I’d assume either the shifting spread is going to be 14/11 or they’re cutting their own gears to meet the teeth count. I was experimenting with gears around 14 teeth and either your driven needs to be huge (100+ teeth) or you need to have a secondary. Might be able to get away with 6inch wheels to reduce the driven gear, but he’s still going to have to make custom cim gears to get anything near 2.5 shifting spread.
Edit: Did not realize a description had been posted. Your teeth count does not line up 71 vs 72, is that okay to use?
Vexpro sells a pinion gear that has 11 teeth, but the profile is set up so that it is used as a 12 tooth gear (as far as spacing between shafts is concerned).
One thing that I would double check with this kind of design is that the motors are spaced around the shifting shaft in a way that prevents the pinions from over constraining the shifting gears and binding. With the shifting gears being 54 and 60 teeth (both divisible by 3) I don’t think there would be an issue with the motors being spaced evenly around the gearbox, but it doesn’t look from the render like they are spaced that way.
I’m not so much saying that the two ratios literally wouldn’t work / spin, though if that’s an issue then there’s another problem. I’m basically saying the two speeds / gear ratios on the ballshifter won’t be very different, which really minimizes the benefit of shifting in the first place.
That depends on what you want to use each gear for. I personally think 1.82 is perfect.
What is the thickness of the plates? They look like 1/4". You could probably get away with 1/8" or less depending on how they are mounted to the chassis.
You make a good point about over constraining things. It’s not so much the location of the motors in the gearbox, but the angular alignment of the 11T and 18T gears on all of the motors have to be the same. This is probably not easy to do…
After some more thought, I guess there’s 3 main requirements:
- Spacing of the motors (easy peasy)
- Correct alignment of both pinions on each shaft (hmm, tricky)
- Assembly such that the pinions of each motor are in synch with every other motor (this part would be fun!)
For what it’s worth, the Penguineers used a shifting design with two pinions per motor successfully in 2012. Their solution to these issues was apparently to make 4 gearboxes, one for each wheel. 
I didn’t spend a lot of time looking at the alternatives, but the selection of ball shifter gears is pretty limited. When I found a combination that worked and gave a 1.82:1 spread, I stuck with it. The thought occurred to me that making your own ball shifter gears wouldn’t be too difficult. It may open up some interesting possibilities.
The plates are 0.25". You’re right they could be thinner, but I like the support they provide for the bearings.
I think you could make a fixture with pins to align the pinions and a guide for the broach to ensure the keyways are all clocked the same. Is it worth the effort? Probably not.
I know 971 has run dog shifting gearboxes with two pinions on the motor shaft. I wonder how they solved the problem.
For comparison, the COTS ball shifter has a spread of 2.65:1, though it can be customized to 2.16:1 or 3.68:1. A prevailing school of thought is to have one gear for general use with a high top speed, and a low gear that is much slower designed for pushing indefinitely without tripping circuit breakers. This gearbox’s tight spread would either have a limited top speed or a low gear too fast to accomplish indefinite pushing. If your design objectives are different, that’s fine; I was just curious what your intent here was.
Well to be honest this gearbox was simply a design exercise. I have heard the rule of thumb about gearing low enough to push for 2 minutes without tripping a breaker. But isn’t that usually against the rules? 
I’m not sure why you would want to gear that low.
If it’s a pin, yes. But if it’s a t-bone or both robots are pushing against each other, it’s perfectly legal (or it was in 2014). Also, you may want to switch into a low gear if pushing a heavy game piece (i.e. mobile goals in Diabolical Dynamics) or when kicking off a sprint.