RotoShift Gearbox and Function Switcher

[TheMindGarage]

I've made a new gearbox for my next car. This gearbox can either be set to a gear ratio (driving the wheels) or to a function, allowing the drive motors to power something else. In total, there are 4 forward gears, 2 reverse gears and 6 function "ports". The reverse gears weren't intentional - they happened to exist when I added the gearing for the forward gears.

The gearbox consists of a turntable with an off-center gear positioned on it. This design uses two off-center gears on the same axle - an 8t and a 12t. This allows more meshing combinations. The off-center gears are driven by a 24t gear in the middle of the turntable, which is powered by the drive motors. Around the turnable, there are many axles with gears on them (12 in this design). When the turntable is rotated correctly, one of the off-center gears meshes with one of the gears on the outside, turning that axle. Some of the axles are connected with extra gears to form a transmission with different speeds; the unconnected ones will be used for functions.

Here you can see the internal workings of the gearbox. These are the gear ratios (including the 3:5 gearing before the transmission):

Gear 1: 1:2.5

Gear 2: 1:3

Gear 3: 1:4.167

Gear 4: 1:5

Reverse 1: 1:3

Reverse 2: 1:5

Functions 1, 3, 4 and 6: 1:7.5

Functions 2 and 5: 1:5

The gear ratios are rather close, and the reverse ratios are too high, but there is little choice in choosing gears since all of them have to mesh properly with the off-center gears on the turntable. I made all the gear ratios quite high since I'm planning to drive the vehicle with 2 EV3 Large motors, which have tons of torque.

There are two inputs - this is purely because I plan to use 2 EV3 Large motors - one on each side of the gearbox. The shifting input drives the turntable with a 28:8 gear reduction - this could be increased, possibly with a worm gear. Note that it is ESSENTIAL to use a MINDSTORMS motor for shifting, since the shift positions are in strange places and not in order.

This gearbox can handle plenty of torque - the gearing up before it does help. However, when under high load, the turntable can move out of place and make gears grind. This gearbox works best with minimal backlash on the shifting input. Also, some clever programming can make the turntable adjust its position a little bit depending on the amount and direction of the load, countering the forces pushing the gears apart.

The gearbox is very compact for its functionality - comparable in size to a 4-speed sequential gearbox. However, it can only be used with a MINDSTORMS motor for shifting, which will make it useless for most of you guys (unless of course someone develops a version that can be controlled by a PF servo...)

[Vikingbrick]

This looks interesting! You answered most of my concerns regarding this design in your last post.

What could be done to the backlash problem is to use 2 pinions to drive the turntable. Use the interconnecting gears between them to equal out any backlash.

Looking at the small driver gear (that changes between the outputs), is it well supported on both sides of the 24t? In the pictures I can only see it has support on one side..

And I think this could be used without the Mindstorms, for example with a "reverse" clock where the operator set the dials.

[ChrisO]

I was looking for a good multi-output gearbox for my EV3 robot. So I was excited when I found this rotoshift gearbox.

I tried to copy your rotoshift gearbox as closely as possible, but I couldn't see exactly how you build the gear/output changing device ( the thing that rotates in the middle).

So first I tried to built something myself, but I couldn't build anything strong enough and at the same time also fit in the available space.
So in the end I made the whole shifter 2 studs larger and fitted in a custom sturdy gear/function switcher.

The major problem I have now is: The rotating mechanism frequently gets stuck. This happens when the 8t gear of the shifter doesn't interlock nicely with on of the 16t or 20t gears of (gear 1-4).

Then either it needs to turn back and forth a bit to interlock nicely again, or you have to use force to get the mechanism to turn further. Neither methods are good solutions. So I am stuck again...

[TheMindGarage]

Gear meshing isn't this gearbox's strong point since most of the meshing patterns aren't "perfect" meshes (they are either marginally too close together or too far apart, leading to either increased friction or increased backlash). You can try making a version with fewer outputs/gears - the fewer outputs, the better you will be able to place them and the better they will mesh. Building a gearbox like this requires a lot of experimenting to find the best meshing patterns possible.