I CADed it just to see how low to the ground a 3 CIM Shifting gearbox could be. I’s not really finished, but I don’t like the design enough to continue working on it. As always, feel free to ask questions or make comments. (even if they’re ridiculous like this gearbox 
)
I would be wary of double loading the CIM pinion like that.
That’s not a cim shaft, that’s a regular 3/8" hex shaft.
You’re really starting to push the envelope for these sorts of gearboxes, but there is a limit and you’ve reached it. Now the weight of your box is becoming almost solely dependent on how well you lighten the plates and gears; each new design has to have a certain minimum of gears in it to work.
Have you tried a different design like with belts?
Team 254 uses dog shifting in their gearboxes. Which means that beyond just double loading the pinions, they are cantilevered at the end of the CIM shaft. In this design, and the others that I have posted recently, the pinions are very close to the base of the shaft. This means that the CIM shafts will take a fraction of the torque seen in 254’s gearboxes.
Specs:
Weight 11.2 lbs including everything shown.
Height 3.3 inches tall. Top of the gearbox is 3.75 inches of the ground if running a 4 in wheel.
Gear ratios are 4.57 to 1 in high and 9.52 to 1 in low. (not really ideal)
What does dog shifting have to do with pinion loading at all? Almost all gearboxes have cantilevered motor pinions, including this one. The only time I’ve seen a fully supported first stage pinion is when a team pressed 3/8 hex sleeves onto their CIMs.
While the distance from the motor and thus bushings matters, it’s still going to be a concern here. It might work, but be careful.
I do like the low profile design a lot, but cleanly integrating three motors into a gearbox without excessive extra weight is indeed quite the packaging challenge - you’ll always compromise something in the design. So you’ll probably always get some nitpick on CD 
I think he was saying that the bending moment induced by the torsion on the pinion gear is less when the gear is closer to the mounting plate. This would decrease the bending moment and the likelihood of bending the CIM shaft. I recall seeing one bent shaft on our 2013 robot (caused by some significant shock loads in the PTO/hanging mechanism). We used the same mounting scheme in 2014 and saw no bent shafts.
The dog shifter design usually spaces out the pinion by one gears face width. The design he has shown can place the pinion almost directly against the mounting plate.
I couldn’t have said it better myself! 
(Obviously, sense I tried and failed lol)
How much ground clearance do you have? Assuming a 4in wheel
IIRC, (I’m not at my computer) it’s got half an inch on a four inch wheel.
Why don’t you like it exactily? I think a flat configuration like this at least from a total robot packing components standpoint is ideal, I wish there was a COTS option with a configuration similar to this.
I am in love with the packaging as well, but there are too many compromises necessary to make it fit like this.
-The gear ratios are close together, and they’re not really adjustable.
-You would have to take a standard 64t gear and cut a ball shifter bore in it.
-It requires two expensive bearings that are 0.75" id 1" od.
-two of the motors go through 1 or 3 stages of gearing, and the other goes through 3 or 5 stages of gearing.
-It’s heavier than my other gearboxes.
It would definitely be cool to build one, and you wouldn’t find a gearbox with a lower COG, but I think my other designs are a lot closer to ideal for my team.
As a side note, those bearings only cost a little bit on Boca Bearings, and you could use a bushing if you wanted to.
However, I see why you don’t like this. Custom gears are tricky to machine without a softjaw or machinable collet IME.
The cheapest I found them for was $25. Not super expensive, but that’s $100 dollars per robot that could be avoided. We also don’t have any experience ordering from Boca. Which is also not a deal-breaker, but it adds to the list of difficulties.
It’s a neat idea, if (as you noted), pushing too far unless you REALLY need that low profile. If it’s that important, perhaps you could put one (or maybe even two) of the CIMs on the wheel side of the gear planes, either ahead or behind the wheel. It would limit the wheel size, but it sounds like you’ve already committed to that anyway. Oh, yeah, you’d have to wire or control the wheel-side CIM backwards.
