I said this project was almost ready to go a few months ago, then shelved it and never got around to finishing it… until now! I’m pretty happy with how it turned out, but there are a few things that I think could use improvement.
~75 lbs with Battery, Electronics and all fasteners (but without bumpers)
Dual Wheel design – inspired by 2767
Incorporates a wave washer to account for field variation – inspired by 2767
4 CIM drive power, 4 Bag or 775pro Azimuth power
SRX mag encoders for drive, VP encoders for azimuth (these are custom modified to be standalone)
Ball-shifting gearboxes
True, independent swerve
Module Specifications:
954 Aluminum Bronze Thrust Bushings and washers
2.5" Colson wheels
~1/16" Wave washer for field variation
Azimuth power provided through a VP
VP Steel Bevel gears
Thrust and radial loads completely isolated from drive torque
Drive Gearbox Specifications:
2.78x spread
2.83:1 reduction in high, 16.6/20.5 fps (Adjusted/Free)
7.87:1 reduction in low, 6.0/7.4 fps (Adjusted/Free)
VexPro ball-shifter shaft and gears
COTS shifter piston from McMaster
Lightened BrecoFlex AT5 30 tooth pulley, 16 mm width
Custom 3-D printed shaft coupling and piston mount
3-D printed shell
1/4" plate, fits in a 5.5"x 3" footprint
~5.5 pounds inclusive of all hardware, shell, pulley, encoder and C
Let me know what you guys think! I still have more media to track down, and if anyone is interested I will be finding all of my renders and screenshots of old revisions and also making imgur posts for them.
Very interesting design. I feel you on the “said I would finish it and then shelved it and never got around to finishing it.” That first 90% is always the easiest part even though it usually involves the most challenging work. That last 10% is what really gets everything finalized. I can’t count how many folders I have on my computer with offseason design projects that got 80% done, either because I got bored or side-tracked, or I got most of the way done and then thought of a better way to redo the entire thing.
I’m trying to figure out the weights you have listed. You said the entire thing weighed 75lbs. Now that included battery and electronics, so excluding those it is probably around 50lbs? (13 for battery, 12ish for electronics - probably less but let’s call it that). 50lbs sounds pretty heavy for a drive base.
What is confusing me is you said that a module with all the fasteners etc. weighed 5.5lbs. This is definitely on the low side for module weight, usually anything 6lbs or less is considered light for a swerve module right now. So 4 of those adds up to 22lbs, which means everything else (chassis) weighs 28lbs?
Either you have some material properties wrong, or somehow your chassis is way overbuilt. I think there is something off with the weight estimates you are getting, as nothing about the chassis in that picture makes me think it weighs 28lbs. Might want to double check your materials, etc. or see if you’re picking up any suppressed or hidden components in your mass properties of the assembly.
I’m really impressed with the documentation, renders, and files for everything. Staying organized always helps with having success and makes it easier to iterate on the design!
I’m not sure I love how far apart the two wheels are on each module. It makes the footprint of each module pretty considerable. This in general is one of the drawbacks to the 2-wheel setups, but yours seems to be a bit wider then others (2767 to use your example). Did you have any drive behind the decision of using the double wheel setup other then getting inspiration from Stryke Force?
Also have you looked into how well the wave washer would work for the posi-traction of each module? I’d have to think about it more on whether it will do the same job and act the same as a regular spring like 2767 used. Perhaps someone from Stryke Force can shed some light on if they’ve ever looked into using something like this instead of the regular spring.
Great CAD work! I’m curious to hear what improvements you think could be made like you stated.
Sorry, I should probably have been more specific-- Each drive GEARBOX is 5.5 lbs with all fasteners, 3D prints and gears, bearings, pulleys, CIM etc. Each module is an additional 2.35 lbs and the misc. hardware (azimuth motor and gearbox) is an additional 2 lbs. This brings it to almost 10 lbs per corner. It’s definitely a overestimate I think, and to be honest I probably didn’t put enough time into making sure every item weight was spot on. I would guess the weight IRL would be +/- 5 pounds. If you have suggestions on making it lighter, do tell!
I’m not sure I love how far apart the two wheels are on each module.
It’s about 3/8" wider than 2767’s, and I think there isn’t a meaningful amount to shave off without greatly increasing manufacturing complexity. I went with dual wheel because 1) It’s interesting, and 2) It inherently keeps COG low. In any case, it was an interesting CAD project and let me do some interesting design for the swerve fork.
Great CAD work! I’m curious to hear what improvements you think could be made like you stated.
Thanks! If I thought about it I probably could have made a much more elegant solution for the azimuth encoder, but ¯_(ツ)_/¯
Is there a reason you custom-made the module housing from sheet metal instead of using C channel? Seems like if you can find the right size it’d be significantly stronger and easier to manufacture.
If it were to be built by my team it would be easier to manufacture since it’s built out of 2D plate rather than 3D C-channel. (We have a CNC router in house, so the swerve fork could be CNC’d in about 30 minutes)
It could definitely be made out of C-channel though.
The wheels do look really far apart and I too am not a fan of that. You say it’s only 3/8" more than ours so at this point, it’s time to try it. On a positive note, we have never had a problem turning our duel setup, just the opposite. You should never need a 775pro to turn an azimuth. The biggest negative with duel wheels is the “drunken swerve walk” when maneuvering at super low speeds. A good thing with this design is if this is a problem, redesign the saddle (thing that holds the axle) and bolt it on! Nothing has to change above the saddle. A scary thing about duel wheels is if you hit a “curb” with one wheel, big forces go into the azimuth gearbox. Again, we haven’t had any issues with that yet.
An interesting phenomena that has happened is the wheels have gone from a crowned shape to trapezoidal. Some say it’s from azimuth scrub, some say it’s from hard cornering.
The reason we didn’t use a wave washer is because we wanted a really low spring rate. You want the same amount of down force whether compressed or not. The wave washer has a spring rate somewhere in the stratosphere which greatly reduces the effect you’re trying to achieve.
Unfortunately it looks like this design will never get built (I graduated last year, so this was kinda just for fun).
You should never need a 775pro to turn an azimuth.
I put the 775pro in the CAD since its the biggest motor that would ever be needed, and I wanted to make sure it would clear everything. If it were to actually be built this would probably use a BAG on azimuth.
Emphasis mine. I’m not sure it keeps the COG any lower then other options. Using a standard single wheel offset bevel layout would have just as low COG, while having less parts -> less weight. It can often be more complicated machining wise but if designed correctly it really shouldn’t be too hard to do. There’s also other ways around that if you get creative.
Mike, I know you guys discussed it some in your swerve tech paper, but other then a simpler saddle to machine, what have you found the benefits to be for the dual wheel? Some of the things you brought up above have always been some of my reservations with it. I’m just not sure what benefit it would give me over a single wheel offset bevel configuration, and it seems to have more downside (“drunken swerve walk”, the curb issue, more parts/weight, larger footprint, etc.)
This was my thought when I saw it. I’ve been wondering for simplicity sake if it would be easiest to just have the “base plate” that everything mounts to on a standard module just spring mounted on the chassis, the the entire module just has the suspension you’re looking for. The way 2767 packaged and executed it was very elegant, so I think that’s probably the ideal way to do it, but putting the entire plate on just some basic spring mounts might be simpler.