Clamshell Sheet Metal Drivetrain Concept


My team and I have been working on an offseason robot to shoot basketballs for PR events and such, we are hoping to use it at our school’s basketball games. I figured now would probably be a good time to try out new things before the season as we are a relatively new team, so I designed this clamshell style drivetrain(Like 971 in 2016). We have an amazing sheet metal sponsor who we love, Bridge City Steel, and they offered to help us out with this project. They have a laser cutter and a cnc brake with pretty good tolerances, so this project seemed appropriate.

I would love any input/improvements y’all have for this design. Thanks!

We are also struggling a bit trying to get our flywheel prototype to shoot far enough. Any input would be greatly appreciated.


With no information about your shooter it is impossible to make specific suggestions so here are the basics of how they work.

To shoot anything a certain distance requires you impart a specific amount of energy to the object.

When using a flywheel to transfer that energy, the flywheel must contain more energy than the amount you want to deliver to the ball.

The greater the total amount of energy the flywheel is storing the greater amount it can transfer to the ball.

The longer the time the flywheel and ball are in contact the greater amount of energy that can be transferred.

Slip between the flywheel and ball is another important factor. Coefficient of friction and the normal force determine how much slip will occur.

So to make it shoot farther you can.

  1. if slip is not the problem increase the energy stored in the flywheel by increasing it’s rotational inertia. You want to add weight and do so by placing as far out from the axle as possible. You can increase the stored energy by increasing the speed however that can reduce the time to transfer the energy which means you need more energy. Note however you may already be on the edge of slipping and either way you can seen a lack of improvement or maybe even a loss.

  2. increase the time the ball is in contact with the flywheel. This can be done by moving to a curved reaction plate or increasing the difference between the entry angle and exit angle.

  3. spread the transfer of energy over more time by moving to a compound system where flywheel A imparts some of the required energy and then flywheel B imparts the rest of the required energy. This also has the benefit of reducing the likelihood of slip.

  4. if slip is a problem then you can reduce it by increasing the coeffecient of friction between the 2 surfaces and/or increasing the normal force or compression. Note there can be too much compression, instead of the energy transferred being used to accelerate the ball it is used to compress the ball which doesn’t get you distance.

Note nowhere is throwing more power at it mentioned. The power of the motor plays a very small part on how far you can shoot. Motor power and where in the motor’s curve it is operating are what determines how quickly the flywheel can recover the lost energy and be ready to shoot another shot at the same distance.


What advantages do you see with this style of drivetrain over a WCD, for example? Or is the goal of the project just to try something new? It seems like this clamshell style is harder to design/manufacture and less serviceable than an equivalent WCD or parallel plate style drivetrain. Am I missing something?

Also, is there a reason you have the gearboxes in the middle of the chassis with all of the wheels belted to the gearbox, instead of having one of the wheels direct-driven? Seems like it would reduce the part count and make the drivetrain more reliable.

As far as the shooter goes, the advice Mr. V posted is good. Relative to the objects we usually shoot in FRC, basketballs are heavy, hard, and non-durable (it’s easy to rub off the orange covering with a flywheel). I would definitely look into using multiple wheels at different speeds to help accelerate the ball more slowly for a longer period of time.

And if you don’t need to shoot multiple balls in a short period of time, you may want to look into spring-powered catapults or linear punchers. They can transfer large amounts of energy in a short time more easily, at the expense of reload speed.


If you’re shooting basketballs, there’s going to be quite a bit on energy required to launch those things. I would be a little concerned with the way the sides with the wheels are mounted to the rest of the frame. I’d opt for some angle brackets or tubing.

You’ll find good shooters if you look at 2012. Put backspin on it, we learned that the hard way. Look at 67’s 2012 shooter


If you are shooting basketballs, might I recommend going the linear puncher route (a-la 118/971/67/2826/179 in 2014)? May be a little different to aim but there will be way less damage to the ball (and they look more impressive from a PR standpoint :wink: )


I chose this drivetrain style for a few reasons. We have sponsors who can waterjet and laser cut, but unfortunately at the moment we don’t have a sponsor with CNC machining capabilities nor do we have them ourselves, so making the drive rails for a WCD becomes much harder. We considered bearing blocks, but for a PR robot we felt that sheet metal would look flashier as well, which was one of our goals for this robot. Additionally, we haven’t worked much with bent sheet metal beyond gussets before, so it’s good practice for something we will probably see in industry.

As for the direct driven wheel, we are worried about maintenance. As anyone who has worked with the KOP knows, replacing that direct driven center wheel requires removing either the transmission or the outer plate, and as we can’t remove the outer plate in this design, we would have to remove the transmission. This is also an issue as we are using belts. It just makes maintenance much harder. And as for why belts over chains, we have always used belts and chains have given us some grief in the past.


Sorry about that, we are trying a single vertical flywheel with a curved wooden hood. At the moment, we are trying two 4 inch plaction wheels which seem to grip very well for the flywheel. Our power source is two 775pros on a 7:1 versaplanetary because that’s what we had available


I’d definitely try increasing the rotational inertia of your system as the first step, as those wheels don’t have a lot of mass.

One way to add to that inertia is with additional wheels, they don’t need to be able to contact the ball, they just need to add inertia. In fact placing them where you could use a larger diameter wheel while the current system is left intact would be the most bang for your buck and quick to implement.

Keep in mind when rotation is involved all mass is not created equal. 1 oz mounted at the rim of the wheel will add much more rotational inertia that 1 oz mounted near the axle.

Another way is to add weights to the wheel. The weights used to balance automotive tires and wheels are a good option. Since you are in Oregon you can no longer buy lead weights, which is good for the environment of course, and in this case actually good for your application too. The type of stick on weights now available are strip of very heavy rubber with VHB tape on the back that you cut to length.

So head to your local Les Schawab, put on your sad HS student just trying to do a cool science project face and ask them for a couple of feet of it. If you can run a continous strip on the inside of the rim where the ends form a tight butt joint you don’t have to worry about keeping the wheel in balance. If that doesn’t work then carefully cut on the lines to ensure each strip is close to the same weight and carefully place them as equally as possible.


First thing that jumps out at me is “how do you put the wheels on?”.


Initially, the bottom of the side tubes are open, so you just slide them up into the tube, though the cutouts in the belly pan are big enough to allow them to fit through for easy maintenance. Once the wheel is somewhat aligned with the bearing hole, you slide the hex shaft through and tighten two 10-24 bolts into the threaded holes in each end, using a washer to hold the shaft in place. In there somewhere you would add the pulleys and belts but the basic idea is the same, just remove the shaft to remove the wheels/pulleys then the reverse the put them on.


With the way you’ve described assembling the wheel I don’t see how you loop the belts onto the pulleys, especially since you have no finger access. I think having the separate panels forming the drive rail instead of the clamshell improves serviceability significantly. You can still achieve that closed off look by putting a removable top panel in place.

I strongly recommend taking a look at 1114’s public CAD release of their past sheet metal drives. Specifically the one they CAD in the Simbots Solidworks series.


Put the belts onto the pulleys (which are mounted on the wheels) before lifting the wheels into place. We’ve done this with the front/rear wheels on the AM14U series kit chassis when we don’t need to mess with the drive wheel. The belts coming off the drive shaft look short enough that you’ll need to do this practically under the drive shaft to get the inside wheels, but we quite similar with chain and 8" wheels in 2016 STRONGHOLD.


If a belt connects two wheels you can’t load it on both wheels and then insert the wheels through an opening. This works on the AM14U chassis because the inside and outside C channels are separate. It is more difficult on this design because the belly pan and top of the clamshell block access to move the belts into place.

Also, what is your ground clearance? Community service / demo robots generally need higher ground clearance than flat field FRC bots to deal with uneven pavement, doorways, potholes (ok maybe just in MI).


For an off-season project, your team should learn how to use chains properly so that you don’t suffer the grief you did in the past. Many other teams are able to use chains successfully. There will be situations in the future where chains may be a better choice than belts.

It might be a good idea to prototype this with flat plates (or even thin sheets of plywood) held together with aluminum angles, or something like that, to verify that your assembly process is as easy as you think it is. You might find this is more difficult than you think especially if there is tension on the belts. If there are no problems, you can proceed with confidence. If there are problems, you have a chance to design them out before you send your files to your sponsor. Unless your sponsor is particularly generous, you may not be able to get enough iterations on your parts.

Your design sounds like it may have to be assembled in place somewhat like the octanum module design that a team I worked with in the past made. It took two people around an hour to install one of those modules in the chassis. At our first event that year, we watched 148 swap one of their octanum modules in less than 2 minutes. There were some subtle but very significant differences between those two designs.


I’m hearing a lot of great feedback on this concept. I wonder if perhaps a better plan might be to go with a west coast drive. We don’t have a cnc router or the like at our disposal currently, so until we get one of our own/ find a sponsor with these capabilities, maybe use the bearing blocks from VEXpro? What would you all recommend in terms of drivetrain for a team that can’t do machining in aluminum tubing but has fairly solid sheet metal capabilities?


When the side panels are in place this is true, but it looks like there’s a wide open space when the side panels are removed. So you can remove the whole side power train, swap wheels, and return the whole thing. Given that the plates are already at an essentially fixed distance apart, this shouldn’t be that much more work than pulling the churros on the KoP, though you may have to disconnect the power and sensor wires. (I’m assuming that the gearbox attaches to the side module, not the main chassis.)


You may want to consider using the VEX 2 x 1 tubing with gussets riveted on. The tubing does not have to be cut very precisely so a bandsaw or a hacksaw will be good enough. I like their West Coast Drive Bearing Blocks. They work well and are quick and easy to install correctly. The holes that have to be drilled in the tubing are all clearance holes so they don’t need to be very precise either. You can use this off-season project as a trial run for your 2019 competition chassis. If you make it slightly smaller than a typical “legal size”, you can work out all the electrical and control system layout. While you are at it, throw in a compressor and a bunch of solenoids whether you need them or not so you have an idea if they will fit nicely or not.


The Robowranglers have made plenty of WCD-style drivetrains in sheet metal to learn from. Their offseason “X009” project was the basis of every drivetrain they’ve run since 2015, as well as future “X000” offseason projects. Their recent CAD assemblies can be found here. In addition to taking advantage of their sponsor’s resources, sheet metal construction seems to make it easier for them to run a “chain-in-tube” drive and add chassis clearance for obstacles (i.e: angling the front and back of the chassis to expose the outer wheels).

That said, it’s definitely worth considering the VEXpro clamping bearing blocks + pre-drilled tubing route. If you’ve got the workflow down, it’s possible to make a running VersaChassis WCD in a few hours.


If you can spare the room, go with larger diameter shooting wheels.
In 2012 my team used 4) 8" wheels, a 4:1 transmission and 4 old style 775 motors. The new 775 motors are faster and more powerful, so 2 motors should be enough especially with a curved backrest.



As I understand it, this is to shoot actual basketballs, not the much smaller and lighter nerfy balls of 2012. Size 7 basketballs (High schools to NBA) weigh 22 oz and have essentially all of the mass on the 29.5" circumference surface. The 2012 game pieces weighed 11.2 oz and had a circumference of about 25" according to the 2012 game manual, and the mass was distributed pretty evenly throughout the volume. That’s about half the weight (and mass) and less than a quarter the moment of inertia about an axis through the center.