Monday, June 11, 2012

FRC 2012: Shooter v2

So most important bit of the game this year, getting balls to go into hoops. Simple Right? Fuck no.

The most common and "easiest" shooting mechanism is a one or two wheel baseball pitching machine alike. Either using two wheels or a single wheel with some sort of hood, then just a roller to get the ball close enough for the shooter well to grab onto it and fire. Again sounds easy right, and again the answer would be no.

So what was wrong with version 1? Shooter gearbox was cantilevered out way to far adding to much vibration to the system, there wasn't enough reduction for torque, PID never got implanted fully, feeding was a guessing game, hood wasn't optimal, not even close to enough testing, and the whole assembly was a massive pain the butt to work on. In the end the system did allow use to score a few 2 pointers per match at best, but I wanted 1717 like 3 pointers...

First thing that was done was while the sheet metal hood while a cool concept it was scrapped and replaced with a 1114 two angle lexan hood. This allowed for key and fender shooting and was lighter. Very simple and very effective.  It was also machined out more carefully and less last minute, which turned out to be pretty nice.

Now the angles probably need a bit of tuning for the most optimal shooting performance but that rickty gearbox needed to go before any testing was done. After having to deal with how annoying the first revision was to work on revision two was to have a cantilevered wheel, and the whole assembly was designed so that anything could be removed quickly and without frustration; motor can be replaced in 2 minutes flat. It was designed, as everything should be, in Solidworks and changed a million times before ever reaching a drawing file. The design was pretty simple, using the L-brackets and the bolt holes from one of the stand offs on the old shooter; a new gearbox made up of two plates, 5:1 single stage gear reduction and a key slotted shaft was born.

Also at this time we got our sponsorship with Makerplace going and with access to full size and precise bridgeports/lathes, machining consisted of 4 parts. 2 Plates, 1 Box, and 1 shaft, this ended up taking 11 hours..., working to be faster at doing this is something on my to do list...

Well the easy part was done, a solid gearbox and shooter assembly with a good amount of reduction and a top end measured speed of 4100RPMs, now to give it a close loop control system. There is a EP4 4x optical encoder mounted on the back of the gearbox, using a basic PID loop designed with velocity in mind  [thanks 399!] and swaging some constants values eventually it basically worked. I will spare the technically details of this for another day because even for me it wasn't very fun.

With all of the above, we can score a whole lot better:

But that doesn't mean it is done, everything keeps iterating. The shooter PID needs a setfoward constant or something to help it spin up faster, with 330watts of power behind it it shouldn't take 4secs to spin back up. And the biggest issue left is the feeding. The belts just power the ball into the shooter, the issue is depending on where the ball is sometimes the balls go into the shooter faster then the last one. This of course results in inconstancies [seen in the video], and needs some kind of mechanical ball loading system.

No comments:

Post a Comment