Physics of the Flywheel Launcher

VEX Robotics Competition Discussion VRC > Nothing But Net (15/16)
21

That would depend how you arrange them. If you do it in parallel (i.e. all of them touching ball at the same time) then yes - it is 1/3 or 1/4.

However, if you arrange them in serial, i.e first boost stage, second boost stage, then you can do it in whatever increments of delta V you want. I have doubts that anyone would ever need more than two boost stages here. You simply don’t have many extra motors to spare and single stage boosters seem to be working well enough in early prototypes.

It is quest for greater accuracy that will drive all design experimentation.

22

Nice answer, and I agree two stages should be sufficient.

23

Thanks mate!! This will definitely help my new team that I’m mentoring. So thanks!!

24

I don’t know what cody had in mind when he was doing power requirement analysis and asked the above question. But it reminded me of something very important that is obvious to me but may not be obvious for many of the beginners.

Sooner or later you realize that to put more momentum (velocity) into the ball you need to squeeze it more between the rollers. And the more it is squeezed the greater the uncertainty of its exit angle will be.

There could be some successful individual designs that minimize that. But in general, if you take a large sample of the soon to be built NbN launchers you will find this conflict: you either have a powerful launcher or you have an accurate launcher.

Many people will recognize that and say: well we have to make decision what is more important for our game strategy and go with it .

But that is not true. There is a way to achieve both of the goals if you resolve that conflict. If you’ve been reading this this thread you might have a hunch of where it is going.

When asked: why cannot you have both - many will point me to the relationship between squeezing and the uncertainty of the exit angle and there is no way around it.

I believe, the reason for that is that there is a deep preconception that greater velocity must equate to a greater power (transfer) and vice versa.

Once you break free from that notion it will be easy to see that you can build a two stage launcher where first stage will provide most of the power, while the second stage will provide little extra power, but have very predictable exit velocity and angle. In fact you may find that running both stages at the same speed could be an option.

So should you do two stage booster? I don’t know, there could be no cookie-cutter design at this time. You need to experiment, see how it works and change things that don’t meet your goals.

25

I completely disagree, and after thinking it over, the only advantage to a two stage launcher I see is the extra time you get from the second stage. All you really do is increase the ball-engaged-with-roller time, which I feel there’s a better way to do that (still thinking belts).

Also worth noting that a two stage system will engage the ball on the first stage longer, ultimately meaning the first stage gets to apply more energy. This is because the ball will be at speed when it hits the second stage and will pass by it faster than the ball at rest in the first stage.

Here’s the thing. Let’s say that you cram the ball between as many rollers as you please, yes the exit velocity’s direction has error.

Same thing happens when we fire bullets out of guns, that’s why really accurate guns have reasonably long barrels.

The bullet is guided by the barrel, and all that error in velocity goes into friction from rubbing against the barrel, which is lost, but at least the bullet goes where we want it to.

I figure the same is true of our launchers, why not fire it once, as hard as you want, but make sure you have some barrel AFTER the launching bit to kill off as much as that error as you can. The barrel will likely need to be non-trivial sized, like at least a few inches.

As for how much to squeeze the ball, this will almost surely require lots of trial and error.

I’m considering a 2 roller belt system that feeds into a lengthy barrel all on an adjustable pivot. The belts cannot be VEX chain, they need to be curved to loosely match the ball. I’m wondering if 3D printed rubber belts would fly. I would stretch this around two high speed pulleys with a guiding structure behind the belt.

one stage
long ball-engaged-with-roller time
relatively simple
shouldn’t jam if built properly
should be pretty accurate with good control software

Oh and I’d include little bumps on the surface of the belt specially to gain traction on the ball.

I recon I can get at least 5 inches of ball-on-“roller” time with a belt.

26

Would you mind elaborating on why this is the case? It’s not clear to me why the uncertainty of the exit angle goes up with the amount of compression.

27

Think of friction as the wire that transfers the kinetic energy from the rollers to the ball.

If you want to carry a lot of energy, you need a big wire.

Squeezing the ball gives you more friction. As always it’s a balance, you always want just enough friction any more will result in waste but too little and you won’t manage to actually get the energy from point A to B in time.

Is my guess, not a physicist.

28

In case of the gun, expanding hot gases from the charge is the agent that pushes the bullet. The barrel keeps the bullet in line while it is being accelerated along its length.

In case of the launcher the ball is accelerated quickly while it is in contact with rollers. Then barrel keeps ball in line while it oscillates and bounces around. The length of the barrel gives ball the time to convert all that bouncing energy into heat. Ideally, when ball leaves barrel it would have only linear component of the velocity.

The problem with barrel is that it may actually decelerate ball quite a bit. If ball needs to push a lot of air through long barrel - then you lose energy. You may want to either design barrel with some holes (will ball bump into them?), replace it with thin guides, or just increase power output from your boosters.

29

Right so our variables are…

speed of rollers
diameter of rollers
number of rollers
number of roller stages
length between stages
number of motors to use (which amounts to torque on rollers)
barrel (yes/no/how long)
whether to spin the ball up or down
angle of shot

I’d go with…

speed of rollers

TBD

diameter of rollers

Guessing larger is better

number of rollers

Keep it simple, 2

number of roller stages

1

length between stages

null

number of motors to use (which amounts to torque on rollers)

probably 4 or 6 (2-3 on each roller)

barrel (yes/no/how long)

yes, but short IE 4-6"

whether to spin the ball up or down

probably neither

angle of shot

let software figure this out based on distance to target, based on a firing table calibrated with data collected at various battery voltages obtained by hacking a PC power supply with one of my voltage regulators that has a pot to adjust the voltage, all sampled with many shots at roughly the same height above sea as the convention center which sucks because I live in Denver (much higher elevation than everywhere else) so I’d likely have to do this before each competition at the venue.

30

If ball is perfectly uniform and disengages from both rollers at the same time, then it will just oscillate perpendicular to velocity vector while it is traveling to target.

However, if situation is not ideal and it disengages from one of the rollers first, there will be pushback against the other roller while the ball starts its first decompression (it will oscillate a number of times).

Without experimenting with the actual balls for the game I have no idea if the pushback is significant enough to make large angle uncertainty. It all depends on the properties of the ball’s material. Maybe the effect will be minimal and we don’t have to worry about this at all.

I just wanted to put forward some theory so that teams who design their launchers with other balls would not be surprised when the behavior changes with the real ones.

31

Angular momentum of rollers, shafts, gears, motors, etc.???

32

You got it! It might not be as simple as it seemed at the game unveiling. It all depends on the material the balls are made of.

If you are lucky the variations based on all those parameters you listed will be next to nothing. But there is a chance it is going to be very hard to hit the target from the base or anywhere near.

It will be an interesting year in any case.

33

I’m no expert, but as I understand it, gun barrels are made long for three main reasons.

First, the force generated by the propellant acts over a longer distance, therefore the projectile can get up to a higher speed (Force is integrated over longer distance). The higher speed means greater momentum, so the projectile better resists variations in air density, air currents, etc.

Second, longer guns allow the barrel to be rifled, and rifling gives the projectile angular momentum, which helps keep it pointed forward so it can benefit from its aerodynamic shape (instead of the projectile tumbling).

And third, gun marketing people read Freud.

34

I agree, I don’t really think a barrel has much to offer for a ball being shot by flywheels or a catapult. As you said, a barrel is good for rifling, which is not helpful for a perfectly spherical projectile. Furthermore, there are no gaseous propellants pushing the ball. On a VEX robot, I think a barrel will mostly just cause the ball to bounce around and lose momentum unnecessarily before leaving the robot, without a very significant increase in the launcher’s accuracy.

Although I suppose Freud would recommend a barrel to increase your chances of being picked at alliance selection…

35

I obviously understand that the explosion and expanding gas behind the bullet is the main reason behind the barrel, but I also believe that a short barrel will make for more accurate shots in this game.

IDK, something to test I suppose.

36

Question:

How will this barrel be designed? If it is a fully / half closed barrel, how will the internal friction between barrel and balls be taken into consideration? Would this cause further inconsistency?

37

This is a bit off the current topic, but how would you estimate the loss in the angular speed of the launcher wheels after launching a ball? I’m thinking maybe conservation of energy of the wheels-ball system? But that’s assuming we know the exit speed of the ball I think…

38

I think your question is very much on topic.

I’m guessing that for a two wheel launcher, the amount of energy that the electrical system can provide (via the motor) is negligible during the fraction of a second that the ball is in contact with the wheels. So perhaps the only significant energy source the ball could draw from is the angular speed of the wheels, spinning motor mass, shafts, etc. The compression of the ball will convert some of that energy into heat. Make some assumptions about gear inertias, compression losses, etc. and crank through the math. It would be very interesting to see which variables most affect the outcome.

39

I would agree with cody. At this time I think short barrel is the way to go. Unless somebody demonstrates barrelless launcher that is consistently accurate. To avoid air resistance issues I would cut a two 60-90 deg segments out of plastic tube (actually bent plastic sheets) and put them in front of the rollers, since I expect any deviation from the straight trajectory are going to be in that plane.

If you use quad encoders to measure speed before and after ball launch you will know how much the speed has dropped.

I couldn’t have said it better!

For the best efficiency you, probably, want to keep motor speed above 92-95% of the rated speed at all times. See this thread for more info about motor torque-speed-power-efficiency curves.

Flywheels need to be designed in a such way that their moment of inertia is enough to provide ball all required velocity without dropping below those 92-95%. Once again, if it calls for too heavy flywheels you can make them lighter and see if it is still good enough.

40

Using the same mass, you can also increase the angular moment of inertia of your flywheels by placing that mass at a larger radius from the point of rotation. The moment of inertia of a mass around its axis is highly affected by its distance from that axis - in fact, that distance gets squared in the equation, so its radius has a lot of influence.