33 Killer Bees 2019 Offseason Swerve

Below is a link to the CAD files for our 2019 off-season swerve module that we are building for further development.

This module is an evolution of the module we competed with during the 2018 season. That module was an evolution of a concept originally pioneered by Aren Hill, and executed on a competition robot by 2451 PWNAGE. Link to the thread for our 2018 module can be found here.

The things we wanted to focus on for this newest version were as follows:

-Increase wheel base to be as large as possible for increased stability
-Keep height of module below the top of a bumper that is as low as 1" off the ground (so 6" tall)
-Keep good ground clearance to be able to drive over various game elements (i.e. 2019 HAB1 & depot)
-Update both drive and steering motors to NEOs
-Switch to a dead axle setup for the wheel for simpler bearing setup and easier wheel swap
-Support shaft under vertical bevel gear with bearing so it is not cantilevered (biggest issue in 2018)
-Switch to an CTR Mag encoder for the steering instead of the MA3 encoder (had potential ESD issues in 2018)
-Keep as much as enclosed as possible to protect components and prevent jamming

With those items in mind as well as some smaller detailed stuff that I can’t think of off the top of my head, this is what we ended up with. Most of the custom parts are printed on our Markforged printer, with the plates cut out of black nylon on our Omio router. We are currently already in the process of making the parts for one module and have ordered all the necessary parts for four. We will be building up one module first to make sure everything checks out and then building a complete chassis. Some of the basic stats on this version:

-13.37fps free speed (18:80 GT3 SDP Timing Belt 116T; 32:16 20DP VEX Spur Gear; 18:45 1.25 Modulus KHK Bevel Gear; 3" wheel with printed hub and RoughTop Black SBR Tread)
-5.9rps on the rotation of the module (18:72 GT3 SDP Timing Belt 95T; 24:96 GT3 SDP Timing Belt 125T)
-2.07" from floor to bottom of module plate
-5.60" from floor to top of module cover
-2.25" from center of wheel to edge of frame (X & Y)
-7.5" from edge of frame to edge of module plates (X & Y)
-5.48lbs estimate: SolidWorks (does not include frame rails, we expect the weight to be lower as the Markforged parts come out lighter then the CAD estimate).

Below is a link to the GrabCAD site with both 2018 SolidWorks and STEP files. Below are some renders of the module. Feel free to ask any questions and we will be updating this thread throughout the off-season with any developments and lessons learned.

GrabCAD Link






41 Likes

This thing is sick, kudos to you guys on iterating the design and making it look so good. cant wait to see this driving around next year!

3 Likes

Thanks for posting! Still trying to figure out some of the details – how do you sense azimuth?

So there is a 96:24 GT3 belt coming off the module as part of how the rotation of the module is powered. We are picking up the shaft of the 24T pulley and have 2 stages of 16:32 20DP printed gears to make up the 4:1 ratio that the 96:24 belt stage had. This way the CTR Mag encoder is at a 1:1 ratio relative to the module for the azimuth. You can see this in the third render.

The bottom portion of the mag encoder is missing as the model for it has issues saving as a STEP file. We have a custom part we designed to replace the first half of the mag encoder housing so that it mounts with #10-32 screws instead of the little #3-48 screws it comes with.

3 Likes

That housing is beautiful, just downloaded the CAD so that we can have it!

Thanks, now I see it. Very slick.

There will be some lash in the azimuth, I guess.

True. We did something similar to this in 2018 and didn’t seem to have a ton of issues with it. If I could space the plates apart vertically a bit more then we could put a gear profile on the top of the module and pick it up with a 1:1 gear ratio, but we wanted to keep the distance between the plates 2" to easily mount to the frame.

We have some ideas around some other ways we could potentially handle the sensor for the azimuth, but I’ll wait until we have some potential development there before sharing. If it works it would allow us to get rid of those extra shafts and be 0 lash.

Also for those curious, in the SolidWorks model, the Master Sketch part file that was used to plan the design and derive most of the custom parts off of is included in the assembly, so you can look at that to see some of how we arrived where we did. Most of the parts are created such that if we change a dimension in the master sketch, all the parts that are derived off that master sketch part will automatically update. This makes for integration later in the design process much easier.

1 Like

This is amazing!
One question, do you need to have mirrored swerve modules or 4 identical?

I usually oppose swerve for schedule and budgetary reasons, but that model is making me question myself. That thing is beautiful.

1 Like

This is so cool! I was wondering where you got your bevel gears - I see a KHK logo on the 18t gear, but I don’t see a large bevel gear on KHK’s catalog like your 45t gear that interfaces with the wheel. Are those custom?

The 45T starts from KHK, and then the back of the bevel hub will be machined away, the center will have the round bore added, and finally the bolt/dowel pin pattern will be added.

One of the nice details of the design is that the two nylon plates are symmetrical, so you can easily assemble left or right hand versions of the module, no mirrored versions of any parts are required. That said since everything is kind of enclosed in the module and the overall module is symmetrical, it shouldn’t matter much, other then if your electrical team cares about which motor is in which place just for organization etc. Basically with this design, it’s up to you!

1 Like

This is insanely cool! If you don’t mind me asking, where did you guys get your belts and pulleys?

Most of the pulleys are being printed on our Markforged, with the exception of the 18T GT2 pinion pulleys on both of the NEOs for the drive and steering. We sourced these from B&B manufacturing online and will be opening the bore to 8mm and broaching for a 2mm key to interface with the motor shafts.

The belts are all GT3 belts ordered from SDP.

3 Likes

100% in agreement with the above.

Just gonna say this is like a Da Vinci for nerds… pure work of engineering artistry.

1 Like

What is the reasoning to not use an off the shelf pulley from vexpro or something?

For one, VEX doesn’t sell pulleys larger than 60T in the GT2 category. And rather than buying the pulleys from SDP, printing them is pretty easy for us and results in a lighter part.

and if you already have an MF printer they’re cheaper too.

Only with a $5000+ entry to play price tag