Strange Swerve 4.0 Triple Strange 1533

[Brainiac33]

Introducing, Strange Swerve 4.0!

This summer our team designed and created our fourth swerve iteration in preparation for next year's challenge! We have managed to use swerve in the past two game challenges and hope to continue. Strange Swerve 4.0 is lighter, and more robust than our previous 2016 design, weighing in at 5.4lbs per module. For comparison, in our previous modules, our motors alone weighed in at 4.1lbs. Our design is using mainly water cut parts, one of which is a custom 100 tooth gear. We have an encoder on both our rotation and drive motors. The rest is standard vex gears, gearboxes and wheels. Here are some pictures of the module:

http://imgur.com/pVh66Bl.jpg


http://imgur.com/Lw7dvd9.jpg


http://imgur.com/TOxPpLq.jpg


Going a little more technical, our main plate is sandwiched by the 100 tooth gear and a top plate, all of these parts have grooves designed to hold bearings. This allows us to continuously rotate our wheel on the custom bearing. Here are some of the renders from our 3D step models that we will release soon, as we fully intend on using them in 2017.


http://imgur.com/N0Awnoi.jpg


http://imgur.com/k4Fz0fP.jpg


Comments, feedback, and questions are welcome! You can help us brainstorm some improvements for our design if you wish, or we can explain something on our module more in depth.

[frcguy]

Very cool. Is there any way we can get the CAD?

[asid61]

Very cool! There's so many things I love about this module! :D
1. Waterjet everything!
2. lightweight!
3. Compact!
4. Double custom ball bearing!
5. Versawheels!
6. Low profile!

Very good job, and I hope I can see it in action next year.
How is the single reduction VP + 100 tooth gear working out? It looks like you're running something close to 10:1 x 100:36, which ends up around 576 rpm, which is very fast for a swerve module. I would change to a 2-stage VP, but that's about the only way I could see this die. The 100 tooth custom gear seems like extra work when you could use an 84t gear instead, but you seem to have good use of the extra space.

The only thing I can see bad about this in the long run is that it might be difficult to integrate a CIM into it, unless you weld a 3/8" hex onto the output shaft for the bevel gear.
I would love to see the CAD for this, if you're willing to part with it.

[Cothron Theiss]

Looks awesome! How are you making the groove in the gear for the ball bearings?

[ahartnet]

very cool design in a lot of ways. I like the plate being sandwhiched with the grooves for bearings.

I'm sure someone will chime in and tell me I'm wrong - and maybe y'all have tested it and can tell me I'm wrong...but I wouldn't want to use the 775pros for a drive motor. If you get in a pushing match or need to play defense you'll risk burning a motor or at least decreasing its life span. At least with the new VPv2's it could be a quick change out - but I wouldnt risk it personally.

[cad321]

The only thing I can see bad about this in the long run is that it might be difficult to integrate a CIM into it, unless you weld a 3/8" hex onto the output shaft for the bevel gear.

What would prevent them from using a CIM on the current VP using the adapter sold by VEXPro (http://www.vexrobotics.com/vexpro/motion/gearboxes/217-4018.html)?

[asid61]

What would prevent them from using a CIM on the current VP using the adapter sold by VEXPro (http://www.vexrobotics.com/vexpro/motion/gearboxes/217-4018.html)?

That's a really good solution, apart from increasing the height of the module a bit.

[dmpage318]

How is the single reduction VP + 100 tooth gear working out? It looks like you're running something close to 10:1 x 100:36, which ends up around 576 rpm, which is very fast for a swerve module. I would change to a 2-stage VP, but that's about the only way I could see this die.
Both VP gearboxes will likely have a second stage added once we get all the modules together. We just put in single 10:1's for initial testing and you're right, it's EXTREMELY fast.

[Noudvanbrunscho]

Very very nice, reminds me of the module from 696 in 2015.
I have a question about your gears, are they custom made or where did you buy them?

And what is the total weight of the module?

[Cothron Theiss]

Very very nice, reminds me of the module from 696 in 2015.
I have a question about your gears, are they custom made or where did you buy them?

And what is the total weight of the module?


...weighing in at 5.4lbs per module...
...Our design is using mainly water cut parts, one of which is a custom 100 tooth gear. We have an encoder on both our rotation and drive motors. The rest is standard vex gears...


...

[Noudvanbrunscho]

...

Ooh oops, missed that part, went directly for the photo's. Thanks!

[iyportne]

Looks awesome! How are you making the groove in the gear for the ball bearings?

Grooves in all plates made the same way. Mount plates on fixture in lathe and use a 1/4" Dia parting tool to cut out groove.

https://youtu.be/vE3397650W4

[iyportne]

What would prevent them from using a CIM on the current VP using the adapter sold by VEXPro (http://www.vexrobotics.com/vexpro/motion/gearboxes/217-4018.html)?

Nothing...the plan was to make a perfectly scalable module..the power choices and ratios are now all in the versa gearbox, not the mechanical structure. Everything from lightweight and lightning fast to slow and powerful.

I know we'll never win the CD debate about Swerve vs. WCD...but let me just say if you are designing Swerve for pushing matches, you're kinda' missing the whole point anyway.

[iyportne]

...I have a question about your gears, are they custom made or where did you buy them?

Only custom gear is the housing gear. All others are COTS. Note that housing bearing/gear design would be scalable to any turret, turn-table or arm pivot and prevent those frustrating (yet awesome to look at) sheared gearbox shafts.

[marshall]

Well done!

[SenorZ]

Fantastic. We've been toying with swerve all summer (mostly in design) and it didn't turn out this slick.

I think elegant is the term to use (as in an elegant solution).

[Type]

One question for you, why did you decide not to use the VersaPlanetary encoder for the turning motor?

[marshall]

One question for you, why did you decide not to use the VersaPlanetary encoder for the turning motor?

I'm not on the team but it's a 1:1 ratio the way they have it setup now. Using the encoder block would result in a ratio that is not 1:1 and would then make for some complex math and silliness to figure out the angle of the module.

[Cothron Theiss]

Ooh oops, missed that part, went directly for the photo's. Thanks!

Haha, those are some great pictures to look at. The last bit of text was easy to miss.


Grooves in all plates made the same way. Mount plates on fixture in lathe and use a 1/4" Dia parting tool to cut out groove.

https://youtu.be/vE3397650W4

Oh wow, that's exactly what I was wondering. That looks great! What's the actual diameter on the ball bearings? I know in commercial bearings, the inner and outer races are cut into a concave path so that the balls automatically center in the path. What made you choose to do use parting cutter? It'd be easy to hold tight tolerances on the depth with, but it'd leave a flat bottom in the groove. Are you guys seeing any issues with the balls catching or jamming since they can move side to side relative to each other and since you don't have an inner cage?

[nuclearnerd]

Are you guys seeing any issues with the balls catching or jamming since they can move side to side relative to each other and since you don't have an inner cage?

I'd be more worried about the balls brinelling such a soft race. I'd love to hear how this design fares over a large amount of use.

The cage problem (if it is a problem) would be easy to solve. Just make sure every second ball is smaller. The smaller balls will act as spacers, and since they won't contact both races, they can turn in the opposite direction as the larger balls, eliminating rubbing. Bearing balls are available in increments as small as 1 thousandth of an inch. You can even use plastic balls, since they won't be load bearing.

[iyportne]

I'm not on the team but it's a 1:1 ratio the way they have it setup now. Using the encoder block would result in a ratio that is not 1:1 and would then make for some complex math and silliness to figure out the angle of the module.

Right...the code library is based on absolute position. The encoder is a TT Electronics/BI 6127V1A360L.5FS from Digikey, a linear angle position sensor, makes the code robust and the bot powers up knowing the wheel angles. Of course, the mechanical challenge has always been to get the encoder to match the rotation of the swerve module, but they got that down now after a couple of design interations.

[iyportne]

...What's the actual diameter on the ball bearings? I know in commercial bearings, the inner and outer races are cut into a concave path so that the balls automatically center in the path. What made you choose to do use parting cutter?...Are you guys seeing any issues with the balls catching or jamming since they can move side to side relative to each other and since you don't have an inner cage?...

The balls are 0.250" DELRIN bearings. The channel depths are 3/32" leaving a 1/16" float gap between plates. The parting tool is actually 0.250 circular arc so the channel is a concave path with nearly full contact with bearing circumference. Running dry there is no apparent bearing-to-bearing interference. The first prototype was cut with CNC Mill using a ball end mill, but that was very slow and produced too much cutter advance ridging perpendicular to ball path.

[CalTran]

That swerve looks dope. ~22lbs for a swerve is insane.

[Type]

I'm not on the team but it's a 1:1 ratio the way they have it setup now. Using the encoder block would result in a ratio that is not 1:1 and would then make for some complex math and silliness to figure out the angle of the module.

Thank you, I must have missed that.

[Cothron Theiss]

The balls are 0.250" DELRIN bearings. The channel depths are 3/32" leaving a 1/16" float gap between plates. The parting tool is actually 0.250 circular arc so the channel is a concave path with nearly full contact with bearing circumference. Running dry there is no apparent bearing-to-bearing interference. The first prototype was cut with CNC Mill using a ball end mill, but that was very slow and produced too much cutter advance ridging perpendicular to ball path.

Ahh, ok. All of that sounds good. I've never used a parting tool with a convex face.

Also, this may just be the angle, but the teeth on your watercut gear look more shallow than the teeth on the VexPro gear. Is this just a trick of the light, a mistake in the manufacturing, or did you use a slightly different profile from what's on the VexPro gears?

[dmpage318]

Also, this may just be the angle, but the teeth on your watercut gear look more shallow than the teeth on the VexPro gear. Is this just a trick of the light, a mistake in the manufacturing, or did you use a slightly different profile from what's on the VexPro gears?
The teeth are in fact more shallow. They were designed to match the profile of VexPro gears but this is how they turned out from the waterjet. We're not too concerned about it because it won't be under a big load and won't be turning very quickly.

[Brainiac33]

Here is the link for the Strange Swerve CAD MODELS!!!
https://grabcad.com/library/strange-swerve-4-0-1

[frcguy]

Here is the link for the Strange Swerve CAD MODELS!!!
https://grabcad.com/library/strange-swerve-4-0-1

Awesome! Excited to take an in-depth look.

[Lil' Lavery]

And the VersaPlanetaries v2 arrived just in time to swap out all the burned out 775 Pros! ;)

Kidding aside, I'll be intrigued to see how the 775s hold up should this make it to the competition field. Pushing matches aren't the only time a drivetrain could experience high current situations, nor are they always entirely avoidable (especially in games with designated scoring areas and/or chokepoints).

I'll also be interested with regards to how the modules hold up to side loads on the wheels. This is a very aggressively packaged module, and it has custom bearings to boot. I'm wondering how well those bearings will deal with scrub forces on the wheels, and what potential there is for those forces to get passed even further up the module.

Certainly a very ambitious and advantageous take on a swerve module.

[Cothron Theiss]

Just now noticed that it's a 1/4" bolt serving as the axle. That seems a little thin to me, but if it works, that'll be great.

[dmpage318]

Just now noticed that it's a 1/4" bolt serving as the axle. That seems a little thin to me, but if it works, that'll be great.

You can't see it because of the spacers, but the bolt passes through a 1/2" aluminum tube.

[Brainiac33]

We have just finished assembling all four of the modules, and have them mounted to a base chassis. The final weight is 25.2 lb. for all four modules AND our base chassis. Just as a heads-up, we will be giving a presentation on swerve mechanics, design, and programing at the FIRST North Carolina FRC workshop at NCSSM on December 17th so join us if you can. In the meantime, here are some more pictures of our design!

http://i.imgur.com/yhWVOlg.jpg


http://i.imgur.com/9sAH88M.jpg


http://i.imgur.com/wRzWHNK.jpg

[OccamzRazor]

This looks pretty compact and quite a bit easier to repair compared to what you had on the last robot. What was the overall weight of the last system minus the tank treads and whatnot?

What kind of power savings do you expect from the changes?

Can't wait to see this one at the workshop in Durham.

[Bennett548]

Have you had a chance to run this under load?

My concern is that the axial force of the bevel gears has to travel all the way thru the churro standoffs, thru the delrin bearings, and then up the plates holding the bearings. I would imagine the bevel gears will push away from each other quite a bit, which greatly reduces the efficiency and will wear down the gears much faster than normal.

I would suggest taking the plates that carry the horizontal bevel shaft bearing and extend them upwards, screw in a cross plate and put in a bearing that rides on the 3/8" hex shaft output of the versaplanetary. That way the axial force of the bevel gears won't push them apart very much.

Otherwise it is very cool. Put up some videos when you start driving!

[iyportne]

It's alive! Strange Swerve 4.0 made its first public appearance at the 2016 NC FRC Workshops.

https://youtu.be/VbXx9Hj13eI

Base as seen weighs in at 34.6 lb without battery.

[ToddF]

We've been looking for a swerve module buildable with our resources. Our swerve drive project has been on hold until we got access to a CNC machine. With this being waterjet cut, we finally have a viable path to building our own swerve modules. I think we've got our next summer development project. I can't wait to see how this holds up in competition.

[Brainiac33]

During our build process, we tried to stay as organized and precise as possible, we made these layouts at a 1:1 scale with the actual pieces just so we had layouts of all our parts.

http://imgur.com/0g3eYAi.jpg

Here is a google docs link to a much higher-res pdf if thats easier to download and use.

https://drive.google.com/open?id=0B4c0EAYkfeSrQ3hrcTJWa1AyTlE

[Brainiac33]

In order to make swerve even more accessible for any team, we wanted to make our code public and available for anyone to use on their prototypes and designs. Here is the link to our github: https://github.com/ecgrobotics/Strange-Swerve