The inner “core” of the turret is the rotating portion, and has a very large thin section bearing (like 8" +). The outer race is held stationary and attached to the robot.
In the middle of the “core” is a heavy duty thrust bearing, which keeps the turret pinned down. If you can imagine, the turret assembly can only be pulled out (if you were taking it apart), due to the way the OD and ID of the thin section bearing are held by flanges. This thrust bearing holds things “down” so that the thin section bearing doesn’t have any “bad” forces on it, like trying to wrench out the ID from the OD. All the forces are kept strictly radial and sideways.
It’s a pretty ingenious setup. There’s not really another kind of equivalent COTS bearing that would allow for such a large member to hang off one side like their lift mech.
We had an arm on a turret this year and we used the armabot turret. I was worried about it when we were first building it and contacted them. They said they were more worried about the bolts than the bearings. I knew the bolts could take it so I went with that design.
This proved to be true. As we drove off of hab level 2 in a sandstorm at championships our climber dug in and stopped the bot cold. The arm broke a #25 chain on a 72 tooth sprocket. That force also went through the turret bearing with no ill effects. I’m not saying that 254’s design is not better or stronger, but there is a COTS part that can do this.
I think what Vinny was saying was that while the actual turret itself could rotate multiple times, the actual turret, with the wires, cannot. The example in the video is before 254 has added in the wiring for their arm. Perhaps with the bidirectional energy chain and the wires, they can’t rotate multiple times because the chain ends up wrapping around the turret.
Foreward:
We happy to answer questions about our robot (our kids especially love talking to anyone in the pits). We do not upload our CAD because we feel that incentivizes people to merely blindly copy and develop their own design skills or understand why we did things the way we did.
Another note about our designs, just because we are 254 doesn’t mean it’s perfect, we have only a 6 week season like everyone and sometimes that means “just sending it”. Do we overbuild things and rely on custom billet parts a lot? Yes. Do we overvalue aesthetics? Yes. Should we have built a turret in the first place and made our season feel like a dumpster fire? Perhaps not, though it turned out well in the end. This was by far (a solid order of magnitude) the team’s most ambitious, stressful, and challenging build/competition season in recent memory and I’m not sure we’d make all the same decisions again. That said, here’s some details that you won’t find in our technical binder (which will likely be uploaded sometime in the summer).
Wire Management:
Due to the limitations of the bi-directional IGUS (“Bigus”), the turret could only rotate +135deg and -310deg, the code told the turret which way to spin to point to the target, and sometimes if the drivetrain was quickly rotated the turret would have to whip around to keep pointing whilst not going past the hardstops. The Bigus was constrained above and below by a shelf and cover made of bent 1/16" polycarb that attached with velcro onto the bumper rails and could easily be removed to replace batteries/bumpers and service wiring.
Turret Design:
The bearing design for the turret was done to maximize stiffness and strength. Making a turret that was strong enough to withstand the moment generated from our arm running into a rocket or other robot with the elevator at max travel and drivebase at full speed was the driving spec.
The turret shaft (highlighted in blue) consists of a giant billet part which holds an upper, 6.5" ID, 7.25" OD, 0.375" WD Kaydon X-contact bearing we purchased from Ebay. This bearing outer race was held from below by a counterbore in the fixed 3/8" plate that was bolted to the 1x1x1/8" crossrails that were welded to the drivetrain. The outer race is clamped from above by clamp ring via screws into tapped holes in fixed 3/8" plate. The inner race of the bearing rotates with the turret and is held from below by a clamp ring via screw into tapped holes in the turret shaft, and above by a shoulder on the turret shaft.
Lower on the turret shaft a shoulder exists where a 25 72T vex sprocket with a milled-out center is attached (#10-32 screws clamp and transmit torque). This sprocket is driven by #25H chain from a 22T 1/2" hex sprocket that is the output of the turret gearbox. An SRX Mag Encoder counts rotations 1 gear reduction back from this output due to packaging issues. The gearbox is driven by 1 775pro.
On the bottom of the turret shaft another 1" ID, 2" OD, 0.5625" WD high load radial ball bearing is held from above and below by internal and external retaining clips, respectively. It is held radially in another billet piece that attaches down to the drivetrain weldment’s bellypan via standoffs (this billet piece just barely sits above the breakers on the PDP). This extra bearing greatly increases the moment capacity of the turret, giving it a second bearing point 1.75" away that can transmit load into the bellypan, which is very strong in inline tension.
The claw (V7 in CAD, V3 of what was actually built) is designed to intake both discs and balls from all sides of the robot. The claw in the robot photo at bag’n’tag was bad, it popped balls and couldn’t pick-up discs, not even from the ground. We abandoned floor discs and focused on making a versatile intake that would be great at funneling balls and could still do discs. Thus, we took inspiration from others and redesigned the intake in time for SFR to feature 2 rollers to hold the ball and 4 3" flex wheels on arms that could actuate between a ball funneling state and a disc intaking state.
At SFR and SVR, the intake was massive and game object would sweep outside the bumper perimeter when turreting, requiring the drivers to carefully drive around the rocket and cargo ship. The new intake for champs was much shorter but required a longer (16.5" at SFR/SVR → 21" at Champs). It was also lighter (15.6 → 10.2 lbs) by removing the bottom roller (we determined the bottom roller was unnecessary after seeing others), one of the 2x2 and both 1x1 crossbars. These combined to result in a ~50% reduction in torque, allowing it to be counterbalanced by a lighter gas spring (90 → 40 lbs), accelerate faster, and not cook our wrist 775pro.
Neos were used to power the top roller (2:1 reduction) and side wheels (10.125:1 reduction). 3D printed pulleys via the Markforged were used all over to make deadaxle combined pulleys, 16T pulleys with built in spacer, etc. The Neos were certainly excessive in their power output, but allowed for less gearing than a 775pro would require and the brake mode allowed for the disc to be held with little current draw.
A retro-reflective banner sensor was used to detect when the ball was all the way in and automatically stow the arm and intake.
The arms of the side wheels were actuated by a single pneumatic cylinder. Routed 1/4" aluminum 10DP “gear plates” were attached to the polycarbonate plates to keep the arm from parallelogramming. 1/4" polycarbonate plates were used here and to hold the top roller due to their ability to absorb shock loads without permanently deforming.
The top roller is our standard 1" OD, 1/16" wall roller with surgical tubing stretched over it we’ve used for lots of intakes in the past. We made custom aluminum endplugs (one side counterbored for a 3/8" shoulder bolt, the other with a 3/8" thunderhex CNCd to drive the roller). This roller setup is lighter than using a 1/2" thunderhex with numerous flex or mecanum wheels.
We try to use a belt for the first reduction of 775’s wherever possible because they are quieter and don’t wear down like the 32DP gears spinning at thousands of RPM do. It’s also nice to be able to package the motor in a weird spot out of the way if needed (see our carriage gearbox). This belt run still has some backlash (there’s no tensioner in place).
One more interesting thing that we might reconsider doing differently next time was our decision to make a single stage elevator and double jointed arm, as opposed to something like 3476’s architecture with a 2-stage elevator and linear extension, allowing for purely prismatic DOFs. Having the arm made for weird singularities and dexterous work-space limits, particularly with reaching rocket disc level 2, that drove a lot of the arm and intake geometry. The intake and arm for champs were designed use a massive top level master sketch that coupled all the driving parameters from the scoring and intaking states to determine all the dimensions.
Requirements such as size of ball, reaching a disc with a ball in the way, scoring on all 3 levels, etc, provided plenty of constraints for the sketch.
Yes. You can simply get the pulley profile from a VEX or McMaster pulley by opening it, cut extruding away a flange so you can see the profile, sketching on that surface and using “Convert Entities” to grab the contour, saving that sketch as a Sketch Block and importing that sketch block into your new part, where you can add you own extra thin flange, bore out for 1/2" hex, add nubs to ride on bearing inner races, etc.
Thank you for taking your time and effort for writing up an in-depth analysis on your robot, I really appreciate it.
Although it’ll take some time as I try to digest your design! Thanks!!
Of course not, we planned on having a two-roller ball intake with a disc system mounted above it. The bottom roller of the intake was to be Velcro and pickup the disc from the floor and then the low jaw of the intake would articulate up and handoff the disc from the Velcro roller to the disc system on top. However, we ran out of time and didn’t really build a full intake prototype with all the systems, so just had to send what we had for the robot photo and plan on redesigning for SFR.
