pic: My Take on a Differential Swerve



After seeing Aren Hill’s differential module, I couldn’t help but design my own impression of it. This design uses modified COTS bevel gears, but would still require a prohibitively large amount of machining for use in season by most teams.

The belts for the first stage would sit 0.5" above the ground.
Free speed for drive is 13.5 fps on a 2.5" wheel.
Steering ratio is 61:1.
Estimated weight is 6.5 lbs.

Questions and comments are welcome.

2 Likes

Here is a link to Aren’s design.
Here is a link to more renders and a Facebook discussion.

I mentioned it on Facebook, but this is a really sweet layout, nicely done. It really illustrates the potential of the differential swerve design to pack a lot of power in a small space.

What’s the advantage of this vs a typical compact swerve mod?

See this thread for a length discussion in 2x775Pro differential swerves

tl;dr - (Conceptually) this allows for both 775Pro motors to supply power to the drive wheels, without the need for additional motors for steering the module

Basically allows you put more power down to the wheel using the same number of motors in a lighter tighter solution, but does have other drawbacks of complex machining and programming.

Best explanation (and a good example) that I have seen is in the first post of this thread by nuclearnerd:
https://www.chiefdelphi.com/forums/showthread.php?t=159384&highlight=differential+swerve

Edit: Sniped by Lavery (not really a big surprise there though). Hmm, didn’t see that post before I posted and have only been on for a few minutes…weird.

How does it rotate the wheel to turn the module?

I see that is says differential speed but are both motors not connected to the same gear? How could one run faster than the other and how does that turn the wheel?

Follow Lil’ Lavery’s link.
They aren’t connected to the same gear.

Edit: Briefly,

but a more complete answer is in that link.

It helped me to imagine the bottom gear being stationary. When you rotate the top gear not only does the wheel rotate, but it spins as the bevel gear walks along the stationary gear, rack and pinion-like.

Then you can see that if the gears rotate the same direction, it locks the wheel from rotating but still spins the module. And if the gears rotate at exactly the same speed opposite of each other it doesn’t spin but the wheel rotates.

Oh ok I thought the bevel gears we’re connecting the ring gears together so I couldn’t imagine how theybcould sounds different speeds

If given a bit of time in I could create a cad model that demonstrates exactly how this works more clearly to those confused, but here is the basis with a bit more detail:

The motors for this system both power the wheel and/or turn the module. If both of the gears are driven at the same speed in opposite directions the module remains in the same position but the wheel spins. If the speed of one motor is greater than or less than the other motor, the module will rotate, while the wheel still spins at some rate (the wheel speed is related to the speed of the upper gear plus the negative speed of the lower gear).

To help with understanding, if you want to simply rotate the module in place without spinning the wheels you must rotate the gears in the same direction rather than opposite, and at the same rate. Any difference in rotation rate between the upper and lower gear will cause the wheel to spin.

For those who like equations:

R=rotation of module
S=speed of wheel
V1=speed of top gear
V2=speed of bottom gear
K1 & k2=scaling coefficients (because these aren’t 1:1 ratios)

R=K1(V1+V2)
S=K2(V1-V2)
The signs are very important in these equations. You will notice for there to be no rotation of the module the speeds of the two gears must be equal in magnitude, but opposite in direction.

If you want the wheel to be rotating while also spinning the module, the speed of the two gears must be different, and must not sum or difference to zero, but the total speed of the wheel is determined by the total difference in speed of the gears.

I probably confused more people than I helped understand, and I probably went way too in depth, but I hope this helps someone understand this ingenious idea.

So how can you ever hope to drive straight because the motors won’t ever turn at exactly the same speed?

Probably lots of encoders, if one motor is abit faster just make it match the other. Also a gyro sensor onboard would help.

How does your tank drive robot drive straight? :wink:

-Aren

This may be the first drive train I’ve seen that specifically needs to account for zip ties left lying around on the field from N matches ago.

(Don’t get me wrong, these are still mind-bogglingly cool)

I doesn’t lol

FWIW a zip tie head locked down 115’s drivetrain one match at Chezy Champs this year. Got trapped between the #35 chain and sprocket and they couldn’t dislodge it in high or low gear. Seriously makes me consider chain in tube again.

First off, this CAD is gorgeous.

One big thing I have discovered in the construction ofnuclearnerd’s design (I have all the parts, just a little fab left), is motors will tend towards matching speed and torque provided the two are close respectively. If one motor is running slightly slower than another the two should equalize.

Another important aspect of this design is integrated dampening. This can take the form of induced friction in the gear train, some active mechanism (see original thread), or the freebie dampening - resistance of the wheel to rotate about the azimuth on the carpet. A module would be very difficult to control with the following parameters, not to say it couldn’t be done:

  1. High speed inputs (direct drive from 775 pro w/ no initial reduction in the form of a versa planetary or similar)
  2. Low friction gear-train
  3. No friction on the wheel azimuth (i.e. module lifted off ground)

Of course the inverse it true - For a module in the real world:

  1. Motors do not spin without friction (expect “cogging” in DC motors)
  2. Have some friction in the gear train
  3. Wheel can be assumed to touch carpet in basic driving case

edit; delete