paper: Team 2410 Swerve Drive Project

Thread created automatically to discuss a document in CD-Media.

Team 2410 Swerve Drive Project
by: CalTran

This project write up is Team 2410’s project to pursue an off season project of developing a coaxial swerve drive.

Attached is the Team 2410 Project form for Swerve Drive. As well, attached is a Bill of Materials, and a picture of the mostly finished CAD. Read and critique. Mechanical critiques can be directed here:

SwerveDriveProjectForm.doc (78 KB)
BillOfMaterials.xlsx (11.5 KB)

I have heard of “continuous turn” potentiometers, that theoretically can turn indefinitely (with a set range of values that reset every 360 degrees or so). I don’t have any experience with them - I’m hoping somebody will post who has - but these could potentially enable multiple revolutions per pod, as encoders would.

These are called magnetic encoders:

Absolute encoders can be magnetic or optical. All the optical encoders in the KoP are incremental.

There are several types of sensors that can be used

  1. Incremental encoders: give pulses to say they’ve moved some part of a rotation. Quadrature encoders can tell which direction, pulsed encoders give a special pulse every revolution

  2. Absolute encoders: these come in two common varieties, optical and magnetic (as Michael pointed out above). These are the most commonly used in FRC application that require infinite angular sensing

  3. Infinite rotation potentiometers: quite simply these are pots that have no hard stops. They do exist, but they’re usually not very good for this type of application because tend to have a dead-band.

Correct. I forgot about that. What are the performance differences between the two?

An incremental encoder provides pulses that can be counted (in the processor or FPGA) to give elapsed angle or speed. An absolute optical encoder has special markings on the disc (called a Gray code) which encode the disc’s angle.

Last year we used BI 6127V1AL.5FS magnetic sensors. Go to Digikey and search for that part number. Notice the cost. Down load the data sheet for specifics.They out put an analog voltage that corresponds to the angle. The voltage varies from .25 volts to 4.75 volts. With all things considered this works out to about .5 degree resolution. This is more of a c-rio limitation. For higher resolution one would have to go to a serial (SPI) type sensor. CUI has a capacitive absolute angle sensor with 12 bit resolution. Look at the AMT series. Avago also has a serial solution. The Bi sensors worked fine for us. The Bi’s have a bronze bushing so they cant take high rpm’s or mis alignment. There are special couplers to protect sensors from misalignment. You may want to check first choice for an example.

That part number gives me nothing in DigiKey. Could you post a link to the page?

Sorry I left out the " . " Link

3.2 Schedule

If this project is to be used in FRC season then it would be preferred for the project to be complete by January 5th. The ideal completion date would be somewhere mid-December, which would allow for a few weeks of practice to get used to the omnidirectional nature of the drive and non-intuitive control.

Conception of Idea………………11/05/12
Research of Swerve Drives…….11/12/12
Design Team Swerve……………11/24/12
Purchase of Materials……………11/27/12
Fabrication Complete……………12/07/12
Programming Complete…………12/10/12
Practice Time…………………….12/11/12

Is the fabrication complete?

No, sadly we got a bit behind when our teacher went to the Autodesk University conference in Vegas last week and we couldn’t purchase materials. We should have the modules completed on Tuesday, hopefully.

I’m curious how your project is coming along. Any good news?

Well, depends on your definition of good news. We’ve completed a module, and are working on the other 2-3 modules, depending on what material we can scrounge up for our rapidly ending off season. This prototype may continue on into regular season, with an “upgraded” swerve debuting on our robot at Greater Kansas City, game allowing of course. If the game does not allow for it, then we’ll simply finish the prototype, slap it on to a base, and stick our 2012 robot sans-drive on top for a snazzy demo robot.

If your goal is to have an effective scoring design for 2013, it’s probably a bit late in the game to implement swerve.

There is a lot of tuning, debugging, iteration, etc… on the software and controls for it.