Arrgggg Stupid Fisher Price Motors

[Adam Y.]

Errr Im taking physics right now and I just realized that there isn't really any force of friction in a hovercraft. Of course now your giving me interesting applications for the using this knowledge.

[Al Skierkiewicz]

Without going into any detail over the technical discussion so far, all holds if you are trying to drive up an incline at 90 degrees. So your calculations are very likely correct. When running on a level surface the power needed to move a robot is reduced with friction being the greatest force preventing movement. The friction comes together as the sum of the drive train friction, (which is minimally affected by robot weight with the use of bearings to transfer the robot weight to wheel shafts) and the wheel to surface friction which is affected by robot weight and floor matierial. All of the forces are still present but change dominance depending on the attitude of the robot. The power required for movement is that which is needed to overcome the sum of the frictions plus a little extra to get the mass in motion. Someone explained it to me as to why a 20,000 lb aircraft can get off the ground with only a 10,000 lb. thrust engine. You are not trying to drive the plane straight up, just to get up enough speed to let lift take over and you are flying.
Remember that many robots were able to design lift mechanisms a few short years ago when the competition demanded it so there must be ways of designing a mechanism to lift 130 lb.
Good Luck All
P.S. This is just an explanation from an electrical guy.

[Paul Copioli]

Al,

I'm assuming your response was regarding my calculations. My last post was only to clear up what many already know. If you design your robot to put out 100lbs of force per wheel and your coeficient of friction between your wheels and the floor is not high enough; then you will be wasting output power, becasue your wheels will spin. I also noticed my 2 posts were confusing if read one after another. The friction I was talking about in my first long post is drivetrain friction (as you mentioned, too) and the friction in my last post was friction between the carpet and wheels which provides tractive force (or the force to propel). Bottom line:

A drag racer can have all the output torque in the World, but it does no good on ice.

-Paul

[Adam Y.]

becasue your wheels will spin.

Spinning wheels=good depending on how you look at it. Last year our robot got into a pushing match with another robot but ending up winning(of course our robot still lost) the pushing match because our wheels would slip while the other teams destroyed there motors because of too much traction.

[Paul Copioli]

I guess it all depends on your point of view. From where I am standing, if you design your drive train to have enough friction at 30 amps per motor (but no more), then you will not fry your motors. You should never be in a situation to have too much traction (or not enough torque).

-Paul

[Al Skierkiewicz]

Paul,
I was suppporting your claims without having gone through the math. As an electrical guy I was just trying to express my way of looking at the premise that would make sense to someone who wasn't following the math. But the friction we are talking about has another side and that being, if you were trying to drive through a different surface, mud for instance, the rolling friction is very high and therefore requires more driving force. (soft tires on carpet would be the same type of rolling friction)
However, limiting to 30 amps on the drive motor would not get you going. I was experimenting with a bare, in my hand FP motor, on a hard regulated supply capable of 25 amps continuous and kept shutting the supply down just trying to get the motor to turn. Instantaneous current is very high on this motor and when in a drive train may actually exceed 100 amps in real world starting conditions.

[Paul Copioli]

Al (and others reading this),

I agree with you 100% with regards to the starting current instantaneously going up to over 100 amps on many motors (FP and Drill included). The point I am trying to make here is to NOT DESIGN your robots pushing torque at a current draw over 30 amps per motor. If you do, and you get in a pushing match, chances are you will trip those annoying little resettable breakers. I have seen over the years (and I'm sure you have too), many teams pick a design torque at max power which is usually at 50 - 60 amps or higher (for FP, drill, and Chiphua) only to see their breakers keep tripping during long pushing matches. I am trying to save some rookies the agony of a trippy robot drive system.

-Paul

[Al Skierkiewicz]

Paul,
No arguement there. We limit our power curves in software and hardware so that we can go "turbo" when needed for a specific task without hitting breaker reset.
AL

[kmcclary]

Originally posted by Al Skierkiewicz
Paul,
No argument there. We limit our power curves in software and hardware so that we can go "turbo" when needed for a specific task without hitting breaker reset.
AL

Hey Al...

What's your method of limiting the power curve in software? What's your sensing and control hardware for this (ex: current sensors of some kind feeding back to RC's digital or analog inputs)?

I'd love to see your setup. Can you describe the architecture and code?

- Keith

[Al Skierkiewicz]

Keith,
We haven't used feedback to limit control at this point. The software just scales back during normal operation and goes to full control when a button on the joystick is pressed. Our "turbo" button is just used on drive control when we want to get somewhere in a hurry. We have used this method for a few years now. When we have used the drill motors for driving with the attached transmission, we lock the speed shifter into one speed only with a couple of tywraps so it can't bump out of engagement.
The gear ratios are chosen for optimum torque at the lower speed and are obviously different for the FP motors or the drill motors.
Al
P.S. Although we have considered using our current sense to control software we think it needs a little more work so that the drivers are not confused by the action of the software. Practice, practice, practice is the key.

[kmcclary]

Originally posted by Al Skierkiewicz
Keith,
We haven't used feedback to limit control at this point. The software just scales back during normal operation and goes to full control when a button on the joystick is pressed. Our "turbo" button is just used on drive control when we want to get somewhere in a hurry. [...] The gear ratios are chosen for optimum torque at the lower speed and are obviously different for the FP motors or the drill motors.
Al

So are you saying you'd do something like this:
- Set up the gear ratio of a drivetrain motor to be (for example) "60A max before tire breakaway"
- Full joystick push now outputs only a 50% duty cycle PWM signal in each direction (30A limit)
- When you hit the "turbo" button, it would then jump to 100% duty cycle in each direction.

If not, then what exactly are you doing?

Originally posted by Al Skierkiewicz P.S. Although we have considered using our current sense to control software we think it needs a little more work so that the drivers are not confused by the action of the software. Practice, practice, practice is the key.

You're right... You have to have a LOT of practice time reserved if your program tries to "autopilot" things like speed ramping, or it'll be immediately compensated for by the driver who doesn't expect it.

Man oh man we had a LOT of trouble last year on my previous team with the driver "overriding" the software filtering. <chuckle> Whatever the programmer tried, the driver would adapt to it... Asking him not to do that didn't matter. He wasn't even conscious of this reaction. Deep down inside he expected a certain behavior from the robot, and by golly his unconscious body's behavior with the controls was going to make it happen that way!

When the driver started gripping the joysticks at their BASE with a thumb and forefinger, and started slamming them full in each direction, we gave up! We tore all the filter code out, and simply asked the driver to "PLEASE don't ask the machine to reverse too quickly or you'll blow the breakers"...

What the heck... Although a little jerky at times, the robot ran just fine, and admittedly the code was shorter.

- Keith

[Al Skierkiewicz]

Keith,
I am not sure exactly what the numbers were. (I'll leave that up to the software guys.) I think it was closer to 70%/100% though. The turbo mode was only intended to be used in low current manuvers like open field running, diving for the goals from dead start etc. I will try to have our driver explain it better.
Al

[kmcclary]

OK...

BTW: Am I correct in my description of your approach, or are you doing it a completely different way? (Set up the drivetrain to be TOO strong, cut the "full scale" PWM level back to compensate until "turbo" button hit, and no feedback sensing whatsoever from the motors?)

- Keith

[patrickrd]

Originally posted by Paul Copioli
Mu DOES NOT depend on weight

That's not entirely accurate, but usually a good simplification to make

If your wheels are spiked or is made of material that digs into the carpet at all, instead of rubbing against it, the F=mu*N approximation becomes less and less accurate. In reality, as you increase N, mu becomes less and less to some asymptote. According to my experiments anyway

[Paul Copioli]

Patrick,

Your experiments must have some other variables coming into play. Mu is a constant of proportionality as defined by experiments. Its purpose is to show the relationship between "relatively flat" objects sliding against one another. When you have spikes or other geometry such that you have many such defineable surfaces in many orientations, Mu no longer holds true. I agree with you that Mu has its limitations, but Mu does not change with respect to weight.

-Paul