Advantage to six wheel drive?

[PingPongPerson]

Hi Chief Delphiers
I know there have been a lot of posts about six wheel/eight wheel drives over the last year or two, but I couldn't find the answer to my question: Why would anyone use six wheel drive (especially a WCD) unless there are bumps on the field? It seems to me that four wheel drive accomplishes the same thing that six wheel would.
Thanks,
Michael Groom

[Joe G.]

Turning ability.

In any skid-steer drive with more than two wheels, a phenomenon known as turning scrub arises. Wheels are dragged sideways as the robot rotates, creating significant frictional resistance to turning. This resistance can be large enough to completely prevent the robot from turning at all. Typically, this point is reached if the wheelbase is significantly longer than it is wide, as is the case in a typical, long orientation four wheel drive.

There are a number of ways to reduce turning scrub within a four wheel drive, none ideal. Casters, slick wheels, omniwheels, and unpowered wheels reduce tractive force, and shift the robot's pivot point. Shifting the robot's center of gravity to one end or the other can make it unstable. Wide orientation drives don't work with every robot design. Shifting wheels closer to the center of the robot makes the robot tip-prone.

Drop center 6 wheel drive takes full advantage of a robot's 38"x28" footprint, does not sacrifice traction, and reduces turning scrub. Since only four wheels ever contact the ground at once, the wheelbase at any given time is short and wide. But the presence of 6 wheels, spanning the length of the robot, gives the robot great stability.

For more on the phenomenon of turning scrub, I highly recommend this white paper: http://www.chiefdelphi.com/media/papers/1443

[Michael Blake]

I agree with the 6-wheel WCD advantage...

Is there any advantage in using live-axle vs. dead-axle in a WCD setup?

THANKS!

[Joe G.]

Live axles vs. Dead axles is more a matter of personal choice. There are definite pros and cons to both, and great teams have used both.

Live axles are generally easier to work with. With a hex shaft system, wheels and other components can be switched out extremely quickly, and independently, so you don't have to remove a sprocket to change the wheel. They are generally easier to build in a cantilevered setup. Live axles can be built inside of structural members, and transfer torque, allowing sprockets to be mounted further away from the wheels. This also makes live axles suitable for direct drive. Live axles also make the attachment of encoders easy.

Dead axles are less expensive and time consuming to build initially, since you don't have to broach parts, buy hubs, or deal with keyways. Axles often take the simple form of shoulder screws, further reducing their price. Dead axles can also be squeezed into tight places, since they don't need bearings at the interface between axle and structure. In fact, dead axles can double as standoffs, and serve key structural roles! See the Revolution swerve module for an example of this. Finally, some argue that dead axles allow faster acceleration, since the motor does not have to spin the mass of the axle.

[Michael Blake]

THANK YOU, Joe G... that was _two_ excellent explanations... it's _appreciated_... ;-)

[Ether]

Originally Posted by Joe G.

some argue that dead axles allow faster acceleration, since the motor does not have to spin the mass of the axle.

seriously?

[Andrew Schreiber]

Originally Posted by Ether

seriously?

But the added mass of a sprocket doesn't matter? I'll leave it as an exercise for the reader to find out how wide the shaft must be to have the same moment of inertia as a 22 tooth aluminum sprocket.

[Tristan Lall]

Originally Posted by Andrew Schreiber

But the added mass of a sprocket doesn't matter? I'll leave it as an exercise for the reader to find out how wide the shaft must be to have the same moment of inertia as a 22 tooth aluminum sprocket.

To be fair, there's generally a sprocket on a live axle too....

[Michael Blake]

Originally Posted by Andrew Schreiber

But the added mass of a sprocket doesn't matter? I'll leave it as an exercise for the reader to find out how wide the shaft must be to have the same moment of inertia as a 22 tooth aluminum sprocket.

BUT, isn't the same sprocket you mention required for either approach?

Live-axle Mass = sprocket, shaft, wheel
Dead-axle Mass = sprocket, wheel

In fact, though _both_ use bearings... the live-axle would have at least _two_ friction points... and the dead-axle would have _one_ friction point...

[PAR_WIG1350]

Originally Posted by Michael Blake

Live-axle Mass = sprocket, shaft, wheel
Dead-axle Mass = sprocket, wheel

Live axles also often use hubs for the wheels and sprockets, if not, they probably use heavier sprockets since now a hub must be incorporated into its design. I'm not saying that this has any significant effect on the acceleration, I'm just pointing it out.

Originally Posted by Hawiian Cadder

I think a lot of people think that live axles allow for better reliability, because the forces acting on a dead axle are always in the same orientation, they are more likely to fail due to fatigue, or bend (depending on the application).

But some designs allow for dead axles to rotate and might even facilitate it for the reason you mentioned.

[lemiant]

Most of these reasons are so minor as to not matter. I like dead axles, because all I need is a bolt, no hex shaft, broaching, bearings, hubs or set screws. It is much easier to assemble and repair (unless you guys have some really efficient live axle set-ups I hacmven't seen.)

[s_forbes]

I like this thread, there are some entertaining responses.

Generally I prefer dead axle systems, since they are usually faster and easier to build (for those of us with less tooling) and leave more time to develop interesting robot mechanisms that aren't drive trains. The standard six wheel drive built from a kitbot is also great for the same reason.

[Jared Russell]

In the entire history of FRC, I would be surprised if there has been a single match that was won or lost because one robot didn't have to "overcome" the inertial of a shaft where another one did. We are talking absolutely puny theoretical advantages here.

(But I have seen many, many matches won or lost because a robot did not have a reliable or well designed power transmission solution.)

[Michael Blake]

Originally Posted by PAR_WIG1350

Live axles also often use hubs for the wheels and sprockets, if not, they probably use heavier sprockets since now a hub must be incorporated into its design. I'm not saying that this has any significant effect on the acceleration, I'm just pointing it out.

So, a better representation would be:

Live-axle Mass = sprocket, hub(s), shaft, wheel, mounting screws
Dead-axle Mass = sprocket, wheel, mounting screws

BOTH approaches use bearings... the Live-axle would have at least _two_ friction points... and the Dead-axle would have only _one_ friction point...

We shouldn't underestimate the friction/drag created at each bearing-point on a Live-axel setup... if you insert a shaft into one secured flange-bearing and spin with your hand and then take the same shaft and insert into two secured flange-bearings and spin with your hand, you'll notice it takes more effort to spin the shaft... now times that by six wheels...

ALSO, the mechanics of the Live-axle setup causes more friction/resistance at the bearings because of the angular pressures caused by the shaft pivoting inside the bearings as the pressures on the attached wheel (at far end of the shaft) change...

I KNOW these are _minute_ differences (or, are they?)... but, it's fun to think this through...

[JamesCH95]

I've done a little math (pardon my mixed units):

150lb robot traveling@ 15ft/s has 711 Joules of energy

Six 1-foot axles of 1/2 in OD made from steel rotating @ 14.32 hz (15ft/s w/ 4inOD wheels) has 0.148 Joules of rotational energy.

If those same axles are moving with the robot @ 15ft/s they have 18.8 Joules of energy.

What difference does that make?

Rough numbers here:

Assume 500W power train (i.e. 4 CIMS with some inefficiency and non-peak power output slapped on)
Assume perfect traction

It will take 1.4220s to output a total of 711J, i.e. get a dead axle robot to 15ft/s

It will take 1.4222s to ouput a total of 711.148J, i.e. get a live axle robot to 15ft/s

The amount of rotational energy in the axles at full-speed is utterly trivial. If you need that extra 0.2ms to get to full speed... well... good luck. You could shave 0.03 lbs of static mass from your robot and break even.

I may have used rough numbers, but we're talking multiple orders of magnitude of triviality.

Bottom line: live or dead axle, it does not matter.