Additional questions about improving pneumatic catapult

Technical Pneumatics
1

Hello.

As some of you may know, I’ve been doing some research on pneumatics for a few days now and have certainly learned a lot; we’ve spent some time this weekend and we’ve been able to develop a pneumatic system and catapult to be (hopefully) put on the robot later this week. However, we’re still having issues with the ball not going nearly far enough. We want to be able to fix these issues before we try to put anything on the robot, or else it may simply be a waste of our 6 hours out of the bag.

Current state of the system:

Everything is wired and FIRST-legal. We can control the system from the robot (through spikes), and we are using 2 of the old grey solenoids (shown below, from 2006?) which have the blue manual switch on the side. There are 2 small metal tanks on the high pressure side, and 2 sets of 2 small metal tanks on the low pressure side, one set on each solenoid. There are 2 pistons which have a 1.5" diameter (outside), which I’m guessing gives them a 1-1/16" inner bore?

When we fire, the solenoids receive a signal from the spikes and just stay open until we explicitly close them again. This has the pistons open as far as they can, as quickly as they can.

Video from our most recent test:

Testing rig:



Actual robot:


So, there are a few things I want to point out about the video/setup above. First of all, you may notice that the ramp on the catapult is different than the one on the robot. The one currently on the robot is made almost entirely of wood; we are going to use metal so that this can actually support the weight of the ball and the torque of the pistons.

You may also notice that the angle of the ramp on the catapult is much lower than the one on the robot. We started on the catapult with the pistons connected a bit further into the catapult and with the ramp roughly at the same angle as what is currently on the robot. However, that wasn’t able to push the ball significantly. After we raised the top end of the ramp a little bit (increasing the angle) and slid the pistons back a bit against the ramp (so that it is closer to the hinge), we were able to get the shot that you saw in the video above.

Right after the first shot I also did a second shot of the pistons (at 60 psi) without the ball so that the speeds may be compared; clearly, the second shot is faster, which leads us to believe that adding a third piston may help.

That being said, we have some issues to get through before we place this system on the robot:

  1. The ball doesn’t go nearly far enough. If we were to inbound the ball from the human player, we’d have to essentially be right under the truss to make this shot work.
  2. The ball will go an even smaller distance if the ball isn’t being held the way it is in the video.
  3. The entire system vibrates quite a bit, though that may be fixable by more supports and/or braces.
  4. The pistons start off approximately half extended and end fully extended (they have an 8" stroke).
  5. The ball needs to be held in place right now, or else it will fall off the ramp before it shoots.

We have some ideas for how to improve the system as well, and we’d like some comments on them before we go ahead and implement them just because time is running out before the next competition (this upcoming weekend), and we only have 6 hours with the robot; thus, every second counts and we don’t have much time to experiment.

  1. We are considering adding a third piston exactly like the first two, stuck right in the middle. Hopefully, the extra force will increase the acceleration?
  2. We think that if we are able to use the full stroke of the pistons somehow we will do much better, but we’re not quite sure on the specifics on how to do that.
  3. We’ve heard of teams using a latch or something similar to be able to do release all 60 psi inside the cylinders at once instead of having it move as the pressure is building up. However, we have some ideas but we’re not sure on the specifics.
  4. We are considering ordering new high flow solenoids as permitted by the new rules; we’re not entirely sure if there is a bottleneck at the solenoid or within the tubing, however, so we would like to figure that out before spending the money on new solenoids.
  5. We are also planning on reinforcing the wood on the ramp (at least where it is being used to hold up the pistons) with some metal strips.

This all being said, in general the reason I’m posting here is to find advice on how to improve the system beyond simply the barely functional state that it is in now. Thank you for your time,
Hades.

2

There are several things you can do to help speed up your shot. The most dramatic way that we have discovered is to latch the cylinders. We use a simple gate latch to hold the end of our catapult down. While it’s held down, we open the valves, allowing air to flow into the bottom of the cylinders. When you release the latch, there’s already considerable pressure built up behind the pistons. This method also allows easy varying of the shot by choosing different times (in milliseconds) to latch before each shot. If you have any questions on this method, please ask.

3

We also implemented a dual cylinder pneumatic launcher - similar to what you are trying to achieve. There are a few things I would do to yours to maximize your launch:

1- At the back of your launcher arm you need small “armlets” (we called them) that will cradle and hold the ball. This will be the last point of contact as the ball is going through the arc of the launcher and will (drastically) effect the launch trajectory. Your launch appears to be going more up, and needs more forward. The angle and length makes a big difference - even a 1/4 inch changed distances and heights for us.

I had a proto launch video that I cant find , but I will link these two of our completed robot:

http://www.wbng.com/news/local/TigerTronics--249661331.html

You can see the geometry better than I can describe.

2 - You are giving up a good amount of energy by only using half of the stroke on those pistons. You can pull them backward or raise them up and use up more stroke. I don’t think you need a third. Based on that, you need to be mindful of cam-locking the throwing arm.

3 - I agree with the gate latch and pre-pressurize the cylinders in a retracted position and then just release the latch. (It worked for us).

Looking at your construction - You will need to heavily reinforce it. I know it’s a prototype- even what we did in our competition robot was put under immense stress and pulled itself apart and needed reinforcement during the regionals.

Airflow was a bit of an issue as well and we minimized all runs of tubing and extra T’s and such in the system.

4

I know I told you guys to move the pistons closer to the pivot point but have you tried it with them not that close? :stuck_out_tongue: You really should be using the full extension of those pistons, we have the same cylinders as you and initially used the same solenoids so it’s definitely possible. You may also want to decrease the angle between the pistons and the catapult if you want to avoid using a latch. I’ll try to get some pictures of our shooter for you. Our shooter’s resting angle is only several degrees below the horizontal and is ~60 degrees above the horizontal when fully extended and because it’s nearly horizontal when it’s resting the ball sits neatly on top and doesn’t have to be held. We dominated with our shooter during qualifying matches finishing in 4th so it’s a proven design, just requires some tweaking. Also you should avoid dry shooting like you did in that video like the plague. Good luck guys.

5

That’s certainly one thing we were considering. However, the problem is that if an opposing team’s ball was to get into our system then we wouldn’t be able to release it without shooting it, which would be a penalty on us. How were you able to avoid this problem?

Interesting design, congrats on going to nationals and on making the news. The problem that I’d think we’d have with implementing something like that is that you’ve got the pistons pointed horizontally (roughly), instead of roughly vertically as we do. I’m not entirely sure how we could stick to this and still be able to have the pistons start at their low point and extend fully, though I’m certainly open to any suggestions for anything that may have slipped my mind. We considered mounting it similar to the way you do, ie. horizontally, but the way the robot is built right now I’m not exactly sure where we could go about doing that; again, if you have any suggestions I would love to hear them.

Could you perhaps post a bit more information about your gate latch system, and what you’re using to latch/unlatch it?

Gotcha, thanks. If this ends up getting mounted to the robot, we’ll still have approximately 10 pounds to work with, so nearly all of that can go to reinforcements and it almost certainly will; I figure a bit of energy is being lost in vibrations (as seen on the testing rig), so some reinforcement should certainly help.

We will make sure we do that, thanks. Did you find that the solenoids were limiting you after you did this optimization?

6

We have our pistons extended about 1/3 of their 8" stroke. We don’t use a latch on ours…instead, we rely on the geometry of the system to keep the ball from launching until the cylinders build air pressure, and then whammo, they move and launch the ball. The cylinders start out almost parallel to the arms.

I think it works ok, we were finalists…but if you want to build a really good one, find team 842’s web page, where they show how they built theirs. It’s different, and better.

https://sites.google.com/site/falconroboticsfrcteam842/frc-robots/2014-dream

launcher.jpg
launcher.jpg2048×1536 890 KB



launcher.jpg
launcher.jpg2048×1536 890 KB

7

This is a problem, to get a fast reaction time, you need to have cylinders at the bottom of the stroke. So this is a problem if you are not using a latch and pre-pressuring the cylinder prior to launch.

Think of a drink cup, it has a lid, you poke two straws in the lid, one is the input and one is the output. Put your finger over the output and blow on the input. How long do you have to blow before you can’t blow anymore? Now make the cup twice the size, how long? Now eliminate the cup, and hold the other end of the straw. Buy starting the motion with the piston in the middle of the cylinder, you have a large cup between the input and the output. To move the cylinder, you need to pressurize the whole volume to the psi above the load to move. Moving large volumes of air through a straw takes time, reduce the volume reduce the time.

Try to remount the cylinder to use the first half of the stroke and not the last half for speed.

If you go with the latch method, they that volume of air in the cylinder, is basically a low pressure storage tank, which would be beneficial to that system.

8

With our entire robot, our intake arms could raise or lower and allow the ball to be captured, or not captured. Our easy floor pass was raising our arms all the way a short burst backwards driving and the forward and the ball would fall on the floor where it was. In competition we wound up playing forward and generally shooting the ball, or playing defense and shooting over truss and didn’t need the pass often. In my own opinion, the chance of an accidental “catch” was so slight we didn’t worry about it. We would have to actively try and pick up an opposition ball.

Ha - we aren’t - you can chalk that up to news reporter misunderstanding. We are in NY and went to the Troy, NY and Richmond, VA regional competitions. We did make it to semifinals and got to compete against the winning alliance though, and that was great fun. We live in a small area though, and were trying to drum up some interest in the program. I posted that video because it was the final configuration, and had some decent shots of the geometry.

It’s all a trade off and I think either could work. We are more horizontal, and in fact I was trying to have the whole arm go beyond the pivot point as azcalg mentioned, effectively cam-locking it so we wouldn’t need any latch. In the end, we were concerned that we would accidentally fire without a real latch if we got bumped too hard. with your specific robot and mounting points, you probably need to adjust some geometry until you find something that works.

http://www.tractorsupply.com/en/store/national-hardwarereg%3B-v22-gate-latch-zinc

Fairly standard style gate latch. We actually used a small pneumatic piston to retract the gate latch. 1.5 in stroke or so, and 1/2 diameter? It was tough to get it working right though. We ended up angling the latch mount so the whole latch was mounted on an incline so the opening followed the arc of the launch arm. We also mounted a u-bolt to a back cross vrace across teh launcher arm as the piece that was latched.

DONT DO THAT. :slight_smile:

Because it’s curved it had a tendency to let the sides of the launch arms rock back and forth awkwardly. Instead - use a straight bar stock somehow- or we assume that would have been better. I can try and get a picture tomorrow to show what I mean.

The force and sudden stop on this was able to rip through a number of plastic bearings. Don’d underestimate how much reinforcement you need. We were fine for a practice and had to strengthen it at competition because it was slowly dismantling itself.

Also - be safe. Don’t stick your head over it. Our was certainly strong enough to seriously injure someone. We used a nylon safety strap around the whole launcher for carrying on and off field.

Yes, it appears solenoids were limiting us. We wound up using a storage tank on the typical low side behind the solenoid, and another on the front side of the solenoid on the launch tubing. That was plumbed straight in to the bottom of the piston and we did did not use any exhaust side fittings on one of the pistons. This was also pre-charged, and allowed a quicker transfer of air in to the piston. Our launcher would stay upright, so to retract we only plumbed one of the retract side cylinder ports to the solenoid.

We have a few post season ideas to play with and see if some other arrangement would have made for quicker cycling times on the pneumatics.

9

We actually just switched to a pneumatic launcher. Getting the geometry of the launcher right is critical.

Here are some photos and video of our launcher.

We found that moving the pivot on the launcher arm and moving the bottom pivot we could dramatically change the trajectory and force of the launch.

You should try to use more of the stroke. It will allow you to get more acceleration time through the launch.

We also adjust the hard stop at the end of the launch to change the trajectory of the shot.

10

Wow! Such rapid progress!

Is your hard stop adjustable during a match or is it only adjusted in the pits?

11

Only in the pits, and it won’t be adjustable once we dial it in on practice day.

12

I’m on the other side of this–I think there are two ways (at least) to make a pneumatic catapult work and the one my team used is to not retract the cylinders the whole way and instead fill the cylinders ‘chamber’ to 60 psi while the catapult is latched. If you simply hold down the catapult with your hand while these fill and then release, you’ll see how that would work.

The other way is what tr6scott is talking about, it would be to use small diameter cylinders and lots of stroke and large Cv valve, hoping to get enough travel to get speed.

Since you have large diameter cylinders, I think you need to stick with the first plan. You need a latch. The tractor-supply gate latch will work fine, but if you want simplicity and no failure, I recommend you put a small piece of steel on the end of the catapult and hold it down with an electro magnet. The one we use is rated at 180 lbs of pull and uses 5.6 watts.

In stock

$62.19 Each

5698K116

Also, I think your current catapult arm is too heavy. We use PVC. Really. Just clamp it with u-bolts—don’t drill holes in it.

I don’t recommend three cylinders–recovery time will take too long using that much air.

13

So, part of the issue we found was that the ball was rolling off the ramp as it was firing. We fixed that by adding a bit of metal to the end, the ball now follows a much more predictable path.

We’ve ordered high flow solenoids as suggested by team 842 on another of my threads; this should hopefully allow the air to get in a lot faster. The other thing that we did is adjust the angle a bit so that the pistons start at the bottom of their extension and are stopped half way through or so by a rope.

We have, however, come across a new problem. Namely, the cheap hinges that we are using take a heavy beating from the pistons and experience a lot of torque against the point where they’re supposed to be hinging; we’re looking at steel hinges right now, and we plan on hinging the entire ramp instead of just at 2 points. Hopefully this will allow for better power conversion.

After doing these changes, the ball is able to follow a path that’s just barely high enough for it to get over the truss; we hope that the high flow solenoids, which will arrive tomorrow, coupled with a stronger hinge or two, will allow us to fully use the power.

14

I have no particularly good pictures offhand, but here’s something:

early prototype video
picture from before first event, I think

We have two cylinders (I’m not sure of bore size) that are 8-10 inches stroke, I believe. The up and down strokes are powered by separate SMC double solenoids like http://team358.org/files/pneumatic/SMCManualOverride.jpg from kits past, so not particularly high flow. This allows the shot to only vent the topside from atmospheric instead of 60psig. Also, only one downstroke port is connected; the other has no fitting. These push a bar of c-channel whose fulcrum is a bolt. The cylinders point just above the fulcrum, allowing pressure to build before overcoming the weight of ball and bar. The ball is held in place over the bar by the surrounding 80/20 bars. It can make the truss and the high goal from in front of the low goal.

It took a while of playing with dimensions and number of cylinders to get right. There have been several tweaks since these images: a little spacer to elevate the arm at the rear, moving the connection between arm and cylinders a couple holes forward, adding the second valve to not fight pressure on top half of cylinder. Presently, we have 7 accumulators (of at least 3 models, darn space problems) to store air and have an on-board Viair compressor.

Compared to the OP, the shooter is quite close to the ground, so the power is available for a good shooter.

It was finalist at 16-3-0 (11-1-0 through quals) at Traverse City.

Does this help?

Edit: summary of my suggestions: place cylinders more in line with fulcrum, try to use more of stroke, lighten the arm, eliminate flex, keep the ball steady. Use at own risk :slight_smile: