Hello CD,
I am trying to design an offseason shooter/turret like those used in the 2017 season. It seems like most turrets use what is essentially a large sprocket with a hole in the middle to make the whole thing turn. I have two questions:
Where do teams find / what kind of tolerances are required to make such a large sprocket?
Is there a way of making a turret without using a large sprocket?
That is all,
18Sampson
We had access to a metal laser cutter with enough precision to cut a custom 20dp gear for the turret, instead of a sprocket. It worked well, but aluminum gear teeth are fragile, so take care to protect it.
Yes, actually. There’s a couple of possible ways.
Note that both of these methods are friction-dependent, and thus possibly less effective at holding position than a chain/sprocket. As noted you could use a gear instead.
I want to preface my answer up front by saying I have never actually built one of these, just picked the brain of lots of teams, and inspected a few.
I’m not an expert on question 1, but the second question I have looked into a lot. The most effective way of building a turret without custom machining I have seen is what I call the reverse belt method. Basically you take a lazy susan bearing (Amazon Link below) and attach a timing belt to it with the teeth facing out. You can then mount a standard pulley to the end of whatever motor or gearbox you choose, and physically mount the assembly so that the pulley interfaces with the teeth of the belt as if they were two gears meshing. The Photo below is from team 174 in 2012 and shows roughly how the belt looks on the bearing. Cant find a better picture showing a motor from my collection unfortunately.
This is what we did in 2012. You could also do the same thing with chain using these links to fasten it to the turntable bearing. The timing belt method would definitely be smoother.
I think 1533 used a linear servo to aim their turret.
Gears can 3D printed rather easily although for full sized turret gear multiple prints may need to be connected.
Another very simple option is to wrap and attach chain to the turret, using the chain like gear teeth, and spin the whole thing with a sprocket. You won’t get a full 360 rotation with this, but you will get most of it.
195 used the geometry from a standard vexpro gear to create a custom gear on the outside of a lazy susan bearing and then used a wire EDM to create the part. The bearing was not too large and if you are making a turret for the 2017 game, you can settle with a fairly small bearing with an ID just enough to get a ball through with still some wiggle room.
if you want cheap, and not much machining…
in 2009 we used a V belt…the big pulley was not expensive, and we used a hacksaw to remove the center, leaving enough of the spokes to bolt the turret base onto. It is supported with bearings that fit into the edge of the pulley.
In 2012 (rookie year), we only needed about 270 degrees of rotation, so we just put the turret in the “center” position, offset a motor/gearbox with drive sprocket, and wrapped chain around the lot, securing it to the turret on the side opposite the drive sprocket (with screws, IIRC). You could do the same with a timing belt.
Another option we considered was making a groove and running a v-belt in it. As with the Colson idea above, do not trust an encoder based on motor or drive sprocket rotation.
Last season my team used a sprocket on (I can’t remember the name off the top of my head) and a giant lazy susan bearing set something like this. we then proceeded to drill a hole in the outer ring to thread in a bolt that would go between chainlinks so essentially as the motor sprocket pulled the chain the torque would be delivered to the bolt and pull the outer ring of the bearing around as the inner ring stayed mounted in place.
I was on the team at the time this was constructed so here is our experience. Yes, we tried to use a linear servo and it worked for a short period, but we had many issues with the reliability of the servo, eventually breaking completely. I don’t recommend the AndyMark linear servo for turrets, instead I would definitely recommend a reliable motor for turret positioning.
We actually laser cut a sprocket out of 7/64" thick Delrin for our turret for 2017. Worked well for us, but you need access to the right machinery and the right software for creating the sprocket’s tooth pattern.
Actually I’ll challenge that a bit. a ~200 tooth gear has very little curvature from the perspective of each tooth. Rather than mess about with involutes, we modeled our laser-cut turret gear with flat sides at 20 degrees to the pitch tangent (like a rack). Worked just fine.
If you’re using solidworks, you can use any gear profile generated by the toolbox library. It’ll do any DP / tooth count and you can copy the teeth to make a custom part.
As stated earlier by Brainiac33, the linear serve did not last. It was a great idea and initially functioned as designed, but too little attention was paid to specs and it failed due to sizing/application…like other robots on the field hitting us. In the end we just locked the shooter down and used swerve drive as our turret.
However, our swerve drive module design is basically a 360° turret and can easily be sized to anything you want, including the ball bearing design and encoder. (See thread for details https://www.chiefdelphi.com/forums/showthread.php?t=152393)
The gear profile was cut on a waterjet and generated in PTC CREO, though Solidworks also has a profile generator. The ball bearing path was cut on a lathe with a circular cutting tool. The swerve gear and bearing design was robust and failure free for 90+ matches, so for us, its ready for future turret when needed.
All this detail and links to the models should be in the link so I won’t repeat it here.
I believe 971 laser cut delrin for their rotating turret; see the black ring here: http://frc971.org/files/2017-TechnicalDocumentation-971.pdf and here: https://roboticscompetition.news/2017/04/24/team-971-talks-about-their-paintball-inspired-feeder-shooter-serializer-and-their-effective-design-process/
Seems like most teams laser cutting delrin cut it ~3mm and sandwich it for thickess as the heat starts to melt it otherwise.
987 I believe 3d printed theirs out of multiple segments (2?) on a markforged printing nylon to make the gear ring, plus other parts for retention etc - I don’t recall if it was reinforced with fibreglass/kevlar fibre or not.
We actually specifically chose a cycloid tooth profile for the turret gear so that we could router the plates. We laser cut our hood gears, but didn’t have access to laser time on the timeline we needed when we made the turret gears. There was still a stack of three 1/8in plates.
Are there any downsides of the cycloid tooth profile turret gear other than needing to make your own pinion?
Nothing major that I know of (or that would make a difference for FRC turrets). I think that the cycloid profiles are such that you can not interchange gears and pinions (IE a gear designed for a 10T pinion is not the same as a gear designed for a 12T pinion). Involutes do not have this problem (in addition to not having the problem of separate profiles for pinions and gears), and are thus more generally useful as a profile.
I would say that needing to make your own pinion is almost an upside, as both the pinion and gear are able to be made with mill, router, and lathe machine resources. It is much harder to do that with an involute profile.
Here is our early prototype of our very first time attempting a turreted shooter system.
What we did was attach an aluminum plate that was smaller in thickness than the chain used to turn it to the lazy susan turntable.
We bolted the ends of the chain to the sides of the plate. By having the chain ride along the plate we did not need to have it a sprocket on the lazy susan.
Using bolts and nylon rollers as the tensioner to the gearbox you could change the drive sprocket as well as the Versa Planetary gearing to get the ratio/rotation speed you desired. To tension the chain the roller would be moved.
I would recommend to have limit switches just before the end of the travel.
Here is an photo of one of the earlier iterations without the limit switches.
https://www.chiefdelphi.com/media/img/723/723e77663e649a8958d4469c2f694c88_l.jpg
This worked out very well for us since we have very limited manufacturing.
I can ask the team for additional pictures and documentation if you like, let me know…
Aloha!
