pic: Off-Season Kicker Design



A few details:
-Overall 326:1 gear ratio
-Ratchet pawl system to prevent backdrive
-~200lbs of spring force
-Load and kick time is a combined ~1s

what motor are you using? our kicker this year was a CIM in a 144-1 bane-bots planetary that could generate 200ftlbs of torque and pulled back 275 LBS of surgical tubing. if using a cim, that gear reduction might not be necessary.

Yeah, we’re using 1 CIM. The reason for the huge reduction is in case we want to add more spring force if necessary. The other part of it is that this is also the backstop for our roller, so we can’t afford to have the roller + ball drive the kicker back and have the ball encroach past 3".

When you kick, does it back-drive the motor?

~

  1. where is the sensor?

  2. Are you releasing it with an AM dog clutch? I guarantee it takes too much force to release - more then a small piston can provide.

We’re planning on putting a potentiometer on the shaft for the kicker

  1. Are you releasing it with an AM dog clutch? I guarantee it takes too much force to release - more then a small piston can provide.

Yes, we are. Shouldn’t the ratchet-pawl make it easier to release? And also, we could replace the standard piston used on shifters with a larger piston of the same bore, couldn’t we?

Yes but your talking a lot of force, I would personally move up to a 1.5" bore or 2" bore. If you have the weight and air why not use a 2" bore actuator?

Or you could ask someone how to do the math and get the correct size?

-RC

Yeah, you really need a huge amount of force to release the dog gear. We had trouble this year with our dog gear, even with a 2" bore pneumatic. Actually, we had to significantly angle the faces of our dog gear to reduce the force required to fire it. There might be a better way to address the problem, like using a larger diameter dog gear. I also saw some teams this year use a motor with a lead screw/cam to push out the dog gear, which you could also try.

We used an anti-backdrivable window motor, and attached it to parts of a toughbox (1 stage). We made an assembly that, using a 2" bore piston, separated the gears in the direction of force, so the piston held the two gears together when not kicking. This was after we were unable to release a dog gear, even after greasing it and pulling it by hand. I don’t know if there are any good pictures of it, but if you’re at IRI, you can come to our pit and ask us about it.

In a nutshell: We un-mesh the gearbox by attaching the motor side to a plate that moves to pull the two gears apart.

you really need a huge amount of force to release the dog gear

If you design the spring attachment points so that the spring is almost parallel to the kicking lever when in the armed position, then the holding torque will be very low, even with large kicking energy stored in the spring. Then it is easy to actuate the release mechanism.

I drew a sketch but this particular thread has attachments disabled. Think of an over-center latch.

The disadvantage to this approach is that you cannot vary the kicking energy in software.

~

I was thinking about this, and the way you guys did it inspired another thought for me:

Make a slot in the gearbox where that final gear is so that the gear can slide, and then have a pneumatic cylinder attached on either side of the shaft. When the pneumatic is at one end, the gear meshes. At the other end it disengages and you get a kick. That shouldn’t take nearly as much force.

(The slot shouldn’t be too hard to figure out. Just something following a line at a smaller angle than that of the line tangent to the pitch circle of both gears, right?)

We had a sorta C-shaped bracket, with the shaft supported on both flanges and the motor on one, a pivot point on one end and a pneumatic piston on the other. It was a really big piston, 2" bore.

One other thing to keep in mind is the slight rotation of the output gear that is generated when you unmesh the gears. Since you are not pulling the gears straight out, you are pulling them on an angle, that action will rotate the output gear slightly. Make sure this rotation causes the kicker to go out rather than in. This way, the gears will be easier to unmesh and require a smaller piston (it is not fighting the springs to unmesh).

One final thing to keep in mind is the default state of the pneumatics. We used a double-action cylinder (air pushes both directions) but a single coil valve. We made sure that the default state (when no power is applied to the valve) left the gearbox meshed, so it kept the kicker in the box and avoid accidental firing. We left the pneumatics system full all the time to prevent the gearbox from firing (actually, we shipped the robot to Atlanta with 120psi of air in it’s one air tank, and it still had 80psi in that one tank when we took it out of the crate)

I would actually recommend a pot over an encoder because it’s position is absolute instead of relative. Your programmers will thank you.

The key is to isolate the pot from the shaft. We did this with a short length of 25 chain and a plastic sprocket on the kicker shaft and pot.

Not all encoders are relative. There’s an absolute encoder in the 2010 KoP:

http://www.austriamicrosystems.com/eng/Products/Magnetic-Encoders/Rotary-Encoders/AS5030/AS5030-Downloads/AS5030-Downloads

Why is that the key? Did you try it both ways and it failed when directly connected and worked when chain-driven?

~

The key to not damaging the pot.

We anticipated that the stresses on the shaft would break a potentiometer if it was connected directly to the shaft. We isolated it with a chain so that the rotary motion was transferred, but any other forces were not. We used a standard 270-degree pot.

Although a magnetic encoder would remove any mechanical connection between the sensor and the shaft, so that might be a better solution then a potentiometer.

Isolating the pot from the shaft (via a chain or belt etc) is certainly a solution, but not the only way.

If you use the pot to replace one of the shaft bearings then of course that’s going to put unacceptable loads on the pot. But if you mount the sensor directly to the shaft (through a simple coupling) on the outboard side of one of the end bearings it will not bear the shaft loads.

~

Essentially what we did. We had a gear on the end of the winch shaft. That gear drove an idler shaft that had the pot attached to the end of it. The pot (and the encoder that replaced it) were not bearing any loading from the kicker.

We had a gear on the end of the winch shaft. That gear drove an idler shaft that had the pot attached to the end of it.

Sure, that’s a good solution. But the point was it’s also possible to mount the pot (or encoder) directly to the shaft through a simple in-line coupler on the end of the shaft. With that approach there would be no undue loads on the pot. It’s not the direct connection to the shaft per se that causes the loading problem, it’s if you try to use the pot in place of a load bearing.

~