Trouble with pneumatic tank drive

[Joe Johnson]

Aspect ratio is all.

Wider is better. Longer is less good.

The reason is that when you turn a WIDE robot the wheels don't have to move normal to their roll direction (i.e. scrubbing) very much, they just roll. When you turn a LONG robot, the wheels practically do nothing BUT scrub along the carpet. It's this scrubbing along with grippiness that kills you (and your battery).

That the reason that you lower the middle wheels; It effectively halves the Width/Length Ratio because either the front two axles are on the ground or the rear two, but not both. So the distance in the length direction halves while the width stays the same. Less scrubbing so you can keep your grippy wheels.

What do to?

Well, if you're 4 wheels in the corners, you need to either get a third axle and lower that center wheel or you need to get rid of some of that grippiness on at least one set of tires (also you might want to read Dr. Joe's Seven Stages of Dealing with a 4WD Robot with Grippy Tires)

Dr. Joe J.

[IKE]

Originally Posted by Joe Johnson

Aspect ratio is all.

Wider is better. Longer is less good.
...snip... Dr. Joe J.

Gulp, I can't believe I am going to do this, but I somewhat disagree with Dr. Joe.....

While aspect ratio will dictate whether or not you can turn, the overall length and weight distribution on the wheels touching dictates how much power it takes to turn at a given turn rate. What is turn rate you ask? It is angle of the turn divided by the time to execute the turn.

For instance, lets compare two scenarios of wide and long for those "back in the old days" of 28x38 dimensions.

Long bot was a 6x6 drop, and on 8" wheels, the point of contacts with the carpet were at least 4 inches from each edge and then of course the center. total length touching was 38-4-4 or 30 inches. If you dropped the center, they you were riding on an effective wheelbase of 15" while the width was typically 1" inside each for frame, and the crown of the tire was inboard another inch each side (2" wide pneumatic tires) for a effective width of 24". This robot turns well. Assum that the CG is right in the middle so all four touching wheels have the same contact. In this scenario, the lateral forces are the same as the longitudinal and for a COF of 1.3, the resultants work out to about 0.91. The longitudinal forces when pivoting act about a 12 inch radius (24/2), where as the laterals act a a 7.5" radius (15/2). To turn 90 degrees in 1 second, the tires end up traveling around a 14" arc, and 90 degrees of that arch would be 22 inches of travel. 22 inches of scrub in 1 second on a 120 lbs robot is about 2400 in*lbs/second or about 268 watts (also 268 J).
For the old wide bot, the wheel base was 28-4-4 or 20 inches, and the width was 38-2-2= 34 inches. This robot will turn well, but to turn the same 90 degrees in 1 second it will have the tires traveling about a 19.7" radius for a total travel of 31 inches which for a 120 lbs robot is 3700 inch*lbs/sec or 420 Watts of power (also since 1 second 420 J)

So to perform the same turning action, the wide bot actually required about 56% more energy to execute the action.

While it will turn just fine, this will cause your motors to get hot. Especially if chasing down balls.

*I did this math back in 2010 after watching 1918 have hot motor problems with their awesome wide bot that year.

[Joe Johnson]

Originally Posted by IKE

Gulp, I can't believe I am going to do this, but I somewhat disagree with Dr. Joe.....

<snip>

It's okay I can take it.

The wider is better statement is just a notional idea not a hard and fast rule.

It is not as sexy to put in all the qualifiers: wider (with respect to length) matters and that it is the groups of tires that actually touch the ground matters and that location of CG matters (though often if you have an arm or other moving subsystem you don't always get control of the exact location of your CG with respect to the tires that are on the ground) and that it CoF for tires is often different for scrubbing vs. the direction of rolling and and and.

Dr. Joe J.

[Wayne TenBrink]

Originally Posted by IKE

*I did this math back in 2010 after watching 1918 have hot motor problems with their awesome wide bot that year.

I thought your analysis sounded familiar to something we experienced back in the day. Besides having only the 4 corner wheels on the ground with that chassis, there was a lot of friction in the chain routing/tensioning. You could have fried an egg on those motors by the time we got thru the playoffs at MSC.

[Al Skierkiewicz]

Another thing to think about with this setup is what actually is taking place while you are scrubbing the high friction tires in a turn. You will subject your bearings and other drive parts to exceptional side loads which in turn raise the friction there as well. With all of of this increase in friction, you will send the motors into a near stall condition which will take you to somewhere north of 100 amps per motor. Depending on your electrical design, the electrical loss could be significant and therefore reduce the available current to each motor. If one side has a much longer wire length than the other side, then turning in one direction may give you half the loss that turning in the other direction might produce.
What Dr. Joe is alluding to is shorter wheelbases will reduce this side friction as you turn. You may get a better handle if you were to draw the force vectors encountered in a turn for each of the wheels. The wheels closest to the center of turn will have the least side force which in turn produces the least (relative) electrical load.
A fully charged and new battery can supply up to 600 amps but not at 12 volts. Typical batteries may only produce 500 amps for the beginning of the match but much less towards the end. Your mileage will vary on final ratios, size of tires and where on the motor curve you have chosen to run.

[GeeTwo]

If you're committed to this layout, and you're using one chain per side as is most common, I suggest putting the CoG significantly towards the axle which is not coupled by chain to the center wheel, and note that you will get your best acceleration away from that axle.

Try to find some way to increase the center drop. As I noted above, recall that the center axle is already a dead axle; don't be shy about using cover plates, putting a bolt through them, and mounting your wheels on bearings to accomplish this.

The team had already bought the nano tube for our air cannon before I became seriously involved. Looking at it later, this seems to me to be a great way to mount mecanum if you can get the right gear ratio, but for anything else it leaves much to be desired.

[Orion.DeYoe]

Can you specify whether this is 4 or 6 wheel drive?

[dradel]

You may want to take a look at your gear ratio as well

[dradel]

Like I said earlier take a look at your gear ratio.

I think there are going to be more than a few teams not getting the performance they should due to using the kit bot gear boxes without changing the ratio. The ratio selected was based on a 6" wheel not 8-8.5 that I have seen teams say they are using.