This looks really cool! Are the CADs for the modules posted somewhere? Also, is there any reason why the radio is mounted underneath the robot?
We will be posting CAD files and Code to github before kickoff so the design is eligible for use during build season if we choose to use it. The only reason the radio is under the robot is because that was the easiest place to fit it. This test chassis is pretty small.
Nice swerve! Is the design based on Aren Hill’s swerve module?
Looking at the Pictures in the beginning I see your encoder for module rotation, but I don’t see any drive encoder. What encoder are you using, and how is it attached to the cim?
Also, what bearing are you using for module rotation? It looks like it isn’t being held in by bolts, but I can’t see the picture well enough to be sure.
I believe they have what looks like a cimcoder from AndyMark. It’s what we used on our 2017 swerve for drive encoders as well. It mounts on the end of a cim, mini cim, or cimile.
looks awesome. Second the bearing question. It looks like a large diameter thin section. Putting all the load through a radial bearing? looks like your axial load path is through bolt heads retaining the bearing and not a lip/flange though guess loads are low enough that doesn’t really matter.
The design is heavily inspired by Aren’s design. Many props to Aren for a great design. Here are a couple other threads talking about our module design.
https://www.chiefdelphi.com/media/photos/45452
https://www.chiefdelphi.com/media/photos/45357
Cory is correct, we are using CIMcoders on the drive motors. We haven’t done it yet, but our plan is to try magic motion profiling with Talon SRX controllers. We aren’t currently use the CIMcoders for teleop driving.
The bearing we are using is a KA035XPO/SA035XPO equivalent purchased from an Alibaba supplier to get the cost within reason.
These are awesome! How well does using a pulley for turning work? How hard was it to machine the pulley on the module?
Cory is correct, we are using CIMcoders on the drive motors. We haven’t done it yet, but our plan is to try magic motion profiling with Talon SRX controllers. We aren’t currently use the CIMcoders for teleop driving.
Thank you and Cory, we will investigate these encoders to see if they will work in our design.
The bearing we are using is a KA035XPO/SA035XPO equivalent purchased from an Alibaba supplier to get the cost within reason.
Thank you very much for sharing this. The Silverthin bearings seem to be much smaller for the diameter than other bearings available, great for packaging the module compactly.
Using belts for the steering of the modules has been working great. We haven’t had them skip a tooth yet. Once the belt is tensioned there is very little backlash.
Machining the big pulley piece isn’t to bad. We start with 4" diameter X 1.25" piece of 6061 stock and it is machined in 2 setups. We have the machining time down to about 40 minutes for the complete part on our little Tormach 440.
To make the teeth easier to machine and not need such a long, small diameter end mill we are cheating a little with the tooth design. We are actually only engaging every 3rd tooth on the belt, which turns out to be fine on such a large pulley.
We are using X contact thin section bearings which are actually designed and rated for thrust loads.
As far as the axial load path. The part that is tricky to see is the wheel mounts overlapping the inner race of the bearing. I agree it isn’t as ideal as a lip or flange. We have beated on this bot quite a bit and so far we haven’t had any issues with the bearings. However, this bot is quite a bit lighter then a normal competition robot.
Nice! It looks like you used a third pulley (actually a bearing?) to tension the belt. How hard was it to get the tension right, and could it be done without the extra pulley?
Also, for the Bag motor, it looked like you had 3:1 in the pulley and you were going to try 21:1 in the VP, for 63:1 total, according to Anand’a suggestion. Is that what you ended up with for the turning motor?
Those are 2 bearings stacked on top of each other to tension the belt. The top and bottom plates have slotted holes where the bolt goes through to adjust the tension. After making the 1st module we adjusted the slots so that at the loosest position they are tensioned perfectly now. I don’t expect the belts to stretch much over time. I think it would be possible to do without the tensioner, you just might not get the tension perfect the first time and need to adjust the part design slightly. This is on my list of potential improvements/simplifications for the mk2 design.
We did go with 21:1 for the VP. We honestly haven’t tried any other ratios so we can’t say it’s the best, but it has been working great. We are running PID onboard the SRXs.
Really smooth, the design looks great and the programming solid. I hope we can see this fielded next year!
One warning though: I recommend Munchskull’s CIMcoder over the Andymark one. I (like several other teams) had very poor experiences with the Andymark ones last year.
Wow that thing looks awesome! And awesome programming to match! YouTube appropriately pulled up a video of 1717 to follow yours.
Fantastic design. This is a great module, simplified and effective. Couple questions:
-
What is the weight of a single module? This has to be towards the lighter end of modules I’ve seen but I’m curious what the final actual weight is.
-
In your code did you do any dampening to the turning input dependent on current translation the robot is doing?
Great job, always great to get projects like this done in the offseason.
Nick
-
The weight of a singe module including the motors is 7.2 pounds.
-
We are not currently doing any dampening of the turning based on translation. We have decreased the overall steering sensitivity by scaling the steering stick’s input by 0.5 factor. On a full size bot I don’t think we would go that low though. This little test chassis is especially sensitive to steering due to its low mass, small wheel base, and small track width.
Here is the link for the CAD of the robot shown in the video.
https://github.com/woolfepr/Swerve-MK1
This doesn’t seem to be a CAD file.
There is a CAD file in that GitHub, here is a picture that should help you locate the zip file. Once you have downloaded the zip file you the need to unzip it and there should be the CAD to the swerve drive. If you have any more problems getting it open contact Patrick Wolfenden or Me.
https://drive.google.com/open?id=0B0lqPoc2XRt0Y3pQMENsVXl4Rk12RWJrRmRoS0Y4T0xtNGVv