Launchers with elastics

Has anyone else done any testing with launchers involving elastic power? I’m having issues so far with mine and I am wondering if anyone else has had better luck. So far we’ve only been able to launch the ball a few feet with 36 rubber bands split between the two sides of a channel on slides, and this alone draws almost enough current to kill a breaker while moving the rest of the motors(2.5A-2.7A). In order to try and get to the required distance, we may end up drawing +4A and killing our motors in order to pull enough rubber bands back. This is just our design however, I’m just wondering if there is anyone out there doing better than this.

Maybe show a picture of how you wind up the rubber bands? Slowing/ gearing that down should ensure you don’t stall the motors as you wind up.

As Tabor has already noted, how you gear this will make all the difference. Just be aware that when motors stall (such as at the end of stretching your elastic), they draw a large current and therefore overheat quickly. So you will want to release your elastic the moment you have stretched it as far as you want it to go. If you pause for too long, you will “trip out” the motor for sure.

Also, I would suggest looking into using Vex’s latex tubing to store the energy. Rubber bands are notorious for changing their elastic constants very quickly after just a few uses, etc. The latex tubing probably doesn’t change so much, so I’m thinking the results might be more consistent.

I applaud your interest in developing an elastic system. It seems that everyone is trying to build flywheel shooters right now but I’m interested in knowing if elastic systems might have some sort of advantage over the flywheels. If nothing else, if designed properly, maybe elastic launchers might be less prone to overheating their motors?

Before I saw flywheels on the forums, my first idea for NbN was an elastic launcher. Specifically something like what you are talking about (a crossbow like launcher?). After I did the math I was sorta pushed away from the idea, but I did make a prototype for fun. Here is my experience.

Firstly, if you are trying to shoot cross-court with a fixed 45 degree launcher you need the ball to leave the launcher at roughly 22 ft/s which is about 7 m/s. The balls weigh 0.11 lbs which is roughly a mass of 0.05kg. However, you need to take into account everything that will be moving at launch. In your case, you probably have a little metal “sling” that the ball sits in that is connected to slides. Lets say this is 1 lb, with a mass of 0.45 kg. These combined give us a mass of 0.50 kg. Finding our kinetic energy at launch (0.5*mv^2) gives us about 12.25 Joules of energy.
In order to get our launcher up to our desired speed of 7 m/s we will need to store 12.25 Joules of usable energy in the rubber bands. This plot gives the force/displacement curve of #32 rubber bands as they are loaded to three peak loads.

The key thing to notice is that the energy you store in each rubber band does not equal the usable energy you get out. This is due to hysteresis caused by the generation of heat as the rubber band is stretched. So lets decide we are going to stretch the rubber bands 0.2m form their original length (approximately 8 inches, shown by the black line on the plot). Each rubber band will require 0.95 Joules to stretch it and will give back 0.82 Joules. This means we will need to have about 15 rubber bands and will require 14 Joules to stretch. However, one important thing to note is that when a rubber band is stretched this far, it has a load of 9.3 Newtons. This combines to a total load of 140 Newtons which is about 30 lbs. This can be quite a bit to put onto a normal VEX axle.

After doing all this math I wanted to know if it would really work, so I built it. I made something that resembles a crossbow. My “sling” slid on two slides about 3 holes apart. I ended up needing about 24 rubber bands and had to draw it back around 6 to 8 inches. It felt like it had a draw strength of 50 lbs and when angled at 45 degrees shot over 17 feet reliably. For my prototype, I did not involve any motors at all; it was drawn back by hand. I did not want to put 50 pounds onto the little VEX axles or the pulley. The larger HS axles could probably take it if a short enough one was used.
However, I did some math and this is what I got. Taking my prototype, I backtracked and roughly calculated the energy required to fully draw it back to be around 24 Joules. Now a 393 motor can be ran continuously at a power of 4 Watts. This means that a single motor should be able to draw the launcher back in about 6 seconds, 2 motors in 3 seconds, 3 motors in 2 seconds, etc. You can see that if you want to have a rapid-fire launcher you will need to dedicate quite a few motors to it.
The next problem is the release mechanism. I figured there were two choices. Firstly, you could have a transmission shifter that threw the gear train into neutral when you wanted to launch the ball. The other option would be to make a CARR mechanism. The first has the advantage of letting you change your drawback distance to account for rubber band fatigue as well as change your shot distance at the cost of an extra motor/actuator. The second doesn’t require any extra motors but has a fixed draw distance.

Rubber band fatigue is that last thing I considered. You need to realize that every shot from your launcher will be different. As rubber bands are stretched, they are fatigued. For each sequential draw, that force displacement graph changes. You would have to take this into account somehow in your code to ensure your shots are reliable.

Based on your description, it sounds like you have too much friction in your system or your “sling” is too heavy (32 rubber bands is too many IMO), you aren’t using enough motors, and/or you haven’t incorporated some sort of release mechanism and are instead just letting the rubber bands spin the motors during launch. The latter will eat away at your stored energy like crazy.
Sorry that was so long. I just figured you could save some time by learning from my experiences.

No need to apologize. This is great stuff. Thanks for sharing what you’ve learned.

Would it help at all if, instead of a sling-on-a-slide, you accelerated a mass that “punched” the ball and launched it like serving a volleyball?

I attached photos of the system I was using. It uses a 7:1 geared CARR system powered by two motors. We hooked it up to a DC tester to give constant voltage and enough current, but it was spiking well over enough to shutdown the robot if I was moving while firing. Besides a small torque on the gear caused by an issue in spacing, the only thing restricting movement is friction in the right slide. I’m going to try another system tomorrow and Thursday which may be able to shoot farther at less cost of motors. But I will try to reduce friction and use the math you calculated to try and duplicate your tests.

This video’s from Clean Sweep. However, I’m not sure how accurate it is.

There is also the choo-choo system. I’m not sure how adjustable it is, but it’s a thought:

In the following video, there are details around minute 2:55

That is definitely an idea. When it comes to collisions you have two main types: elastic and inelastic. An elastic collision is when momentum and kinetic energy is conserved (think billiards). For an inelastic collision, the two objects stick together and only momentum is conserved (think car crash). Some of the kinetic energy is lost from the system, often via heat. If you are going to hit a squishy ball with a weight, it will be somewhere between perfectly elastic and perfectly inelastic. However, for simplicity we will imagine that the ball is hard and the two collide perfectly elastically (this is a better case than inelastic for us).

So we can say that both kinetic energy and momentum will be conserved. Momentum is the product of mass and velocity (mv) and kinetic energy is (0.5*mv^2). Before impact our “hammer” will have a mass and velocity. This gives us an initial momentum and kinetic energy. After impact, both the ball and hammer will have momentum and kinetic energy. These values after impact should sum to the values before impact.
If you break out your pen and paper you will notice a few relations. Firstly, the velocity of the ball after impact is directly related to the ratio of the hammer mass to the ball mass. Secondly, as the hammer mass is increased, so is the kinetic energy required to get it moving to a desired velocity. So here is some math.

As mentioned previously, we want the 0.05 kg ball to be moving at 7 m/s after impact. This equates to a kinetic energy of 2.5 Joules and a momentum of 0.35 kg*m/s. Let’s say we have a 1kg hammer. Looking at the energy and momentum equations, we can solve for the velocity needed to transfer 1.2 Joules of energy to the ball to be 7.35 m/s. This corresponds to an initial kinetic energy of 27 Joules which is quite a lot to build up with 4 watt motors.

I whipped up a plot in Matlab that compared the mass of the hammer to its required initial kinetic energy as shown below. The second plot is zoomed about the minima.

It is clear that the kinetic energy is minimized to about 5 Joules when the hammer mass is equal to that of the ball. This means in theory a single motor could get this hammer moving up to speed in under 2 seconds. However, at this mass, the hammer would need to be moving at 14m/s (44ft/s, 30mph) before impact. That would be interesting to say the least. Also note that the speed of the hammer would jump down to half of it’s initial upon impact; you definitely would not want any motors connected to it when it hit the ball. You could reduce this speed and the magnitude of the jump by increasing the mass of the hammer (revisit our initial 1 kg calculation) at a cost of the amount of energy you need to put into the system.

You also need to take into account the amount of momentum in your system. As you increase the mass of the hammer you increase its initial momentum. Too much momentum in the wrong direction can cause your robot to move, axles to bend or break, or even your entire robot to flip over.

Overall it looks like it may be worth investigating further. One should note that the collision was assumed to be inelastic so these figures are very optimistic.

Wow, that’s some very good and insightful information! Do you have any picture of it?

I think one of the issues with this, and probably most systems like this, is the amount of bulk you need to move to shoot the ball. It probably won’t make much different for the motors pulling back, but I could imagine that it will reduce the amount of force shooting the ball. This is something that would really need to be experimented with and constantly improved to find out how much you can optimize this system.

Sorry, but I can’t tell… How does the motor and gearing system disengage from the sliding part? If it’s not disengaging, then that’s a lot of mass to sling around at the same time you’re catapulting the ball. You need some fast way to disengage so that the only things that your rubber bands are accelerating are the ball and a lightweight sliding basket of some sort.

EDIT: Oh, maybe it’s like a choo-choo mechanism? It wasn’t clear to me at first. Sorry. Is a CARR what this system is actually called?

one of my old ideas from tossup is to use rubber bands attached to a gear and the gear has some teeth cut off so the motor isnt connected and all the energy stored is released at once

I’m getting about 1.23 Joules for that. I think you forgot the 0.5 in the K.E. equation. Otherwise, I’m very impressed by your breakdown of this problem. Very cool!

Friction is definitely going to be your enemy, especially with those old-style slides. When I made my prototype I sued the newer slides.
I would first disconnect everything from the launcher including the motors, rubber bands, and the CARR. You should be able to tilt the launcher forward and back and the “sling” should slide easily. If not, adjust the position of the slides until it does.
Secondly, it looks like your CARR has a draw distance of about 4 inches? If that is the case, then by looking at that Caltech plot I linked you would need around double the rubber bands I calculated before. Still though, with 36 rubber bands you should be getting more than a few feet. Are you angling the launcher upwards?
I would say first get something that works by hand, then try to introduce the motorized aspect. Adding motors is only going to reduce your performance, so if it can’t make the shot you want by hand there is no need to waste time putting motors on it before you fix it, right?

Also, here is a little calc for the torque on your motors. A rubber band drawn 0.1 m (4 in) exerts 5 Newtons. 32 of these totals 160 Newtons. Now you are going to apply the most torque to your motors when the CARR arm is pointing to the side across an arm of 0.025 m (1 in). This results in a torque of 4 Nm on the CARR. Your gear ratio reduces this to a torque of 0.57 Nm (5 inlb) which when split across the two motors evenly ends up with 0.24 Nm (2.5 in*lb). According to jpearman’s REV2 motor curves this is well within the same continuous-use range of the 393. However, you report that your motors are drawing 2.7 amps (each?) which would seem to indicate nearly 3 to 4 times as much torque is being loaded on the motors. The only reason I can think for this torque to be there is because of friction. Lithium grease may help some.
Hope that helps.

Thanks, I edited the original post to reflect this correction. I was going a little to fast with my math.

Someone else should weigh in here, and I’m sorry if I’m wrong about this, but I think your definitions of inelastic and elastic collisions are wrong.

First, I thought it was only momentum that is conserved in an inelastic collision, not kinetic energy
Second, I don’t think billiards are a very good example of an elastic collision
Third, just because the objects bounce apart doesn’t make it an elastic collision

You are right, freak flashback to HS physics and I mixed them up. Edited. Thanks!

Elastic collisions happen when there is a perfect transfer of kinetic energy. In reality, this never happens. In billiards, a small amount of kinetic energy is converted into mechanical (sound) during impact. Otherwise, I think this is an excellent example. Here, billiard balls have a relative coefficient of restitution of 0.804 with only rubber band balls and golf balls having higher. But I am open to alternatives suggestions.

True. But in order to be perfectly elastic they need to bounce, see what I mean?
Edit: You could have a collision in which the two objects do not “bounce” but instead travel in the same direction at the same speed. This would still be elastic provided energy was not lost.

Sadly, perfectly elastic collisions just don’t happen. Like I said, hitting the ball with a “hammer” isn’t going to be perfectly elastic, but I am forced to make that assumption. Without any balls to test with, I have no idea to what degree the collision would lose energy. I am sure the ball will absorb some as heat when it compresses, but I have no clue how much or as to what other avenues there may be for energy loss. I did state at the end that this was a very optimistic calculation, and this is a prime reason why. Another reason is that I did most of this math on my Windows calculator in the span of about 5 min, so there is probably a few mistakes. However, I think the conclusion stands that this method of hitting the balls may be worth investigating further.

It was 2.7 amps total for both motors. I will try adding some mass to the paddle and changing that slide. I will also loosen the joints and fix the spacing to try and reduce friction. Thanks guys for all the input!

To reduce your max current draw, you need to reduce the total amount of energy you need to store (energy stored by the rubber bands is roughly linearly related to the max torque on your motors) or change the gear ratio on your CARR to reduce the torque load on the motors.
You can reduce the amount of energy you need to store by reducing friction like you said or decreasing the mass of your paddle. Increasing the mass will make you have to pull the rubber bands farther back (or add more of them) which will put a larger load on your motors and will only make your current draw problem worse.