Emergency last-minute hole saw modifications.
Is it necessary?
You’ll save more weight by not having something than by redesigning it. If your team has typically struggled to meet weight requirements, factor that into the criteria you use to make design decisions. For example, with drivetrains consider shifting gearboxes, 3 motor gearboxes, or swerve are all heavier than a single speed 2 motor WCD/KOP drive. Same with adding extra DOF to mechanisms, building to max frame perimeter/height, etc.
Is it efficiently designed?
Slotted extrusion is heavy. Don’t build 80/20 elevators.
Mecanum wheels are heavy. Don’t build mecanum drives.
Use brushless motors. Some quick math on example weight savings:
Intake: a 1 stage versaplanetary + 775pro + talon is .5 lb heavier than a falcon
Drivetrain: 2 CIMs + 2 talons + mag encoder is 4 lbs heavier than 2 falcons (8 lbs for both sides), and 3 mini CIMs/talons is 5.1 lbs heavier per side (10.2 lbs total)
Use smaller screws and thinner material. 1/4-20 bolts very rarely necessary. Unless you’re doing sheet metal, pocketing should be a last resort. The effort to weight savings ratio is very high, and often you’re not getting much structural benefits from thicker, pocketed material than thinner+solid.
If you use 3D printed parts, take advantage of the ability to print complex geometry to simplify what might otherwise need multiple parts. Design the parts smarter and not just thicker, and take advantage of print directions.
Is it effeciently integrated?
A lot of robots I’ve seen have a lack of integration at the top level, leading to extra superstructure where they could have used shared mounting points, or non-ideal mechanism locations requiring a lot of bracketing.
Did you CAD everything?
If you’re using CAD to check weight, make sure all material properties are properly set, and consider what you didn’t CAD in. The one year as a student my team didn’t easily make weight, I found several pounds of bolts and nuts that we hadn’t CADed. Fasteners, belt/chain, wires/connectors, and everything else adds weight, and you should factor in a safety margin when checking CAD.
One golden rule which can be rephrased and sub-ruled in many different ways: “Think about the load paths”. Most of these things are obvious when you’ve enough experience and you take the time to actually think; if you get in the habit, you begin to spot room for improvement like a habit.
Loads don’t evaporate. Newton’s law states every action has an equal and opposite reaction. In more accessible terms: every link passes along the load; the force doesn’t vanish. Think about how every load you put on your robot will be reacted by the rest, including the drivetrain- if you do it ahead of time, you can be smart rather than scabbing tons of gussets on post-facto.
Use big sections: A 2x2, 1/16" wall box tube has about 5.3 times the bending stiffness, and 2.6 times the bending strength, but weighs the same as a 1x1, 1/8" wall box tube. This is why airplanes and racecars are stressed skin monocoques: moving material to the outside helps BULK strength.
Use thickness where you need it: bulk strength isn’t local strength. Take the above advice to the extreme: If your arm is made of .015" wall 8x8" box, sure, it’s really stiff and strong… but the wall will get dented, and maybe even pierced (.015 is aluminum foil territory).
Spread it out: If you can connect something with two bolts 1" apart, or two bolts 5" apart, do the latter, as this will decrease the force seen at the bolted connections by a factor of 5 during (in plane) bending loads. Also, if you’re connecting soft material to hard material, use a washer to distribute the potentially high stress. Using washers and scab plates can let you put thickness exactly where you need it (at the interfaces) rather than where you don’t (in the middle).
Be strong like a tree: If compliance isn’t an issue, a good way to avoid damage from impacts is to make your mechanism flex more. Energy = force times distance; if you can increase the distance your part deforms, you decrease the force your part must withstand from an impact.
Less things to build and break: Reduce your parts count. Bracket directly from one thing to another, or don’t even use a bracket, because fasteners and webbing for fastening to is the overwhelming majority of the weight in an optimized part. The overwhelming majority of structural failures occur not in the middle of a part… but where it meets others. And of course, the fewer links you have in a chain, the less likely it is to break. The best fastener is to not have a fastener.
Get to the point: If you need to connect A to B, connect A to B directly. Don’t make square frames when trapezoids will do. The more linear the load paths, the stronger your part. Parts are always weaker in bending than tension/compression.
Use smooth, properly-sized fasteners: Threaded fasteners will egg out over time, especially if they’re in a loose hole. If you can use pins or rivets to transmit load, these have smooth diameters which are less prone to egging out. This is why welding 6000 series aluminum, which absolutely demolishes the material strength, can still be stronger than a bolted joint: it doesn’t introduce a stress concentrator which will egg out over time. Glues and epoxies are fantastic fasteners (if not for the whole wait-til-it-cures aspect)
If you need to use FEM/FEA to design the part, start over: Designs should be simple. If you can’t visualize and understand the load paths (the magnitude of the loads is a different story), you’re already going down a bad path. Come up with a better design you can actually understand.
I was surprised how much weight we were able to get out of the robot by pocketing the large aluminum gears. We were able to get between 0.25 and 0.35 lbs out of each gear.
We haven’t ventured into 8-32 territory yet (it took years to make 10-32 the default for us!), but this is seriously missed by a lot of teams, especially ones who’ve been around a long time and got stuck in a groove before bumpers became the norm.
Also, if you can’t swing $200 on a rivet tool get this sub-$30 one. It’s been very reliable in our shop even on 3/16" rivets.
Another one I haven’t seen yet: consider your spinny bits too. If you have an axle full of wheels, could that be Versaroller or maybe a plastic pipe with broached end plugs? (We ran PVC pipe for conveyor rollers this year, and it held up fine.)
This image re-inforces Andy’s point. The small gear was providing power to one side of a robot’s drive train, until all of its teeth stripped off. During an elimination match. Guess which alliance was eliminated.
How big was that thing supposed to be
Before they were stripped, the small gear had 14 teeth. So its original pitch diameter was 0.7 inch and its overall diameter was about 0.8 inch.
If youre using the KOP chassis, put holes in the steel drivetrain gears. Also, design with angle bar and channel as well as tube, use 6mm aircraft plywood rather than .25" polycarb in certain areas that won’t experience significant shock loads, and DESIGN YOUR ROBOT IN CAD USING PROPER MATERIALS.
4414 didn’t have a piece of thick wall tube on the robot.
That along with all the weight savings from using falcons (lots of single stage reductions and didn’t use all motor slots) we never had to watch our weight with what was, in my opinion, an extremely chunky robot.
Also spend the extra 2 bucks and buy the west coat product lightened gears. That stuff adds up.
My first four days doing work for AndyMark was Championship 2016. It was peppered with heartbreaking stories of teams with aluminum gears in their drivetrains that just gave up.
There’s a reason the EVO Shifter (which came out for the 2017 season) uses weight-relieved steel gears (example) instead of aluminum.
The 14-tooth gear is particularly prone to failure (this happened all the time in 2016, due to the preference towards large wheels in that season). Aluminum gears are fine in a lot of cases, but not in that particular gearing.
We reduce weight primarily by Swiss cheesing.
However, our real secret is to take the robot outside onto the abrasive asphalt parking lot and do drive practice for a few hours. Not only does the robot lose weight from the tire treads wearing down, but fasteners and other odds and ends fall off. In addition, this tactic provides additional benefits such as reducing the overall robot height.
Hey, I mean, if a mechanism falls off, it counts as weight savings. Right?
Yes. Bonus points if this is achieved via high speed collision with a parked automobile or a dumpster.
*teach a freshman how to use the lathe and save the $2 per gear.
As many have pointed out, if you’re “removing” weight, you’ve already lost. The robot needs to be light from the start. Make mechanisms preform multiple jobs (looking at the dual elevator teams in 2018), cut down on the number and length of frame members, pick 6wd or even 4wd over 8 when it makes sense, etc. Using a PTO is pretty close to the point where it’s basically the same weight to just add more motors (if you have the slots) which is unnerving.
There’s a post somewhere around (in the “very old” category) where someone was talking about how the Apollo spacecraft were de-FODded. They’d be put on shake tables or similar to shake all the loose stuff out of the capsule. Several 5-gallon buckets of FOD later, the capsule was FOD-free (well, hopefully…)
One way my team removes weight every year: “EricH, get your foot off the scale!” (I do a pre-inspection of our robot every year. Often I’ll lean a foot on the scale just a bit to throw it off…)
Funny story on aluminum gears: I once was looking to lose weight AND size from something at work, and took a look at the VEX aluminum gears. Well, I looked at that… and then I decided I’d better do some engineering, and started working through the gear tooth stress analysis. (Quite a fun little math problem, for any enterprising students, BTW–and if you’re really enterprising make sure to do a fatigue run on that as well.) I didn’t like the numbers I got. (Of course, it might have helped if the book I was checking didn’t only cover steel gears…) I checked again… and designed for steel gears. Aluminum just wasn’t going to be strong enough for that particular application.
I tried to suggest removing all brackets and joints on the robot in favour of attaching them via constrain tool. Mechanical apparently doesn’t like it when the CAD team makes jokes.
I don’t know about the other two, but IIRC our KOP drive weighed in at around 20 pounds with all things factored in last year, which is pretty similar in weight to a COTS swerve using SDS-style modules