2 CIM Swerve Concept

[magnets]

Here's my attempt at designing a swerve drive. Any feedback/criticism/questions is welcome.
The weight per module is a little over 5 lbs without CIMs, with most of the weight coming from the steering motor/gearbox and the aluminum tube.

[CalTran]

Err, what might the benefit of two CIM swerve be over the more conventional one CIM?

[magnets]

Originally Posted by CalTran

Err, what might the benefit of two CIM swerve be over the more conventional one CIM?

I wanted to reduce weight by having only three modules, but I didn't want to only have the power from three CIMs. Also, because I have a lot of torque from two CIMs on each module, I can gear the module so that it is pretty fast, but can still do well in a pushing match.

[BBray_T1296]

Originally Posted by magnets

Also, because I have a lot of torque from two CIMs on each module, I can gear the module so that it is pretty fast, but can still do well in a pushing match.

With almost any wheel, you will run out of traction long before you run out of torque, especially with the wheels you show in the CAD

[Chris is me]

Originally Posted by BBray_T1296

With almost any wheel, you will run out of traction long before you run out of torque, especially with the wheels you show in the CAD

Maybe you'll run out of traction before you hit *stall torque*, but you certainly won't run out of traction before the motor is loaded with more current than what the resetting breakers can handle.

[Nate Laverdure]

Am I understanding correctly that both CIMs drive the small pulley? How will you extend the CIM shaft all the way through the rectangular tube? A shaft coupling?

[magnets]

Originally Posted by Nate Laverdure

Am I understanding correctly that both CIMs drive the small pulley? How will you extend the CIM shaft all the way through the rectangular tube? A shaft coupling?

Yes. The CIM underneath the smaller pulley has an 8mm shaft coupling that connects to the longer shaft that is connected to the little pulley on top of the box. There is a bearing on the top of the box.

The original design was for a 1 CIM swerve, and I haven't made up my mind as to which one is better. I posted the 2 CIM version because it looks cooler.

Originally Posted by BBray_t1296

With almost any wheel, you will run out of traction long before you run out of torque, especially with the wheels you show in the CAD

This is true, but this is geared to go pretty quickly, so there isn't really that much torque. JVN's design calc shows the top speed to be over 13 fps, and the free speed is over 16 fps. The wheel shown here could be swapped for another with a better tread. Also, if I reduce the number of motors to only three, JVN's calculator shows that I need to drop the max speed down to about 7 fps.

[Nate Laverdure]

Originally Posted by magnets

...an 8mm shaft coupling...

The keyway in the CIM output shaft does not go all the way to the end (pdf link). You could mill flat spots in the shaft to provide landing areas for the set screws on your shaft coupling, but I'd expect this to be a point of failure.

[CalTran]

Why might going 13fps be advantageous over 7fps? Have you found it advantageous, excluding 2008, to have a high top speed?

[Ryan Dognaux]

Originally Posted by CalTran

Why might going 13fps be advantageous over 7fps? Have you found it advantageous, excluding 2008, to have a high top speed?

Both 2011 & 2013 are games that come to mind where having a fast drive train came in handy, especially if your strategy involved driving the length of the field. Take a look at the Einstein matches from this past year and you'll see robots screaming across the field. Shaving seconds off your cycle time and fitting in an additional cycle could win you a match in 2013.

[magnets]

Originally Posted by Ether

Here's a link to a PDF giving the equations for inverse kinematics for an N-wheel swerve, where N>2.

Thanks, this is really useful!

Originally Posted by Nate Laverdure

The keyway in the CIM output shaft does not go all the way to the end (pdf link). You could mill flat spots in the shaft to provide landing areas for the set screws on your shaft coupling, but I'd expect this to be a point of failure.

I agree, this is the weakest part of the design. In the original 1 CIM version, the CIM was mounted on the upper plate and there was a hole cut out the bottom part, so no shaft coupling was needed.

A better solution might be to take a large diameter aluminum shaft about 3 inches long, put an 8mm hole in the bottom for the CIM, use a CNC mill to make the pulley part on the bottom, then turn down the rest to a 3/8" diameter.

Originally Posted by CalTran

Why might going 13fps be advantageous over 7fps? Have you found it advantageous, excluding 2008, to have a high top speed?

Yes, we've found that having a really high top speed can make a really good robot if you have a lot of driver practice.

In 2013, if we were blocked from leaving the feeder station and returning to our pyramid, we could really quickly go around/under the pyramid faster than we could shift to our pushing gear and push a defender out of the way.

In 2011, being able to quickly grab a tube and quickly return was extremely important.

In 2003, the ability to go fast/have control of the robot was the whole game. That's why teams like 111 who had really good control of their robot could win so easily.

[Nate Laverdure]

Originally Posted by magnets

A better solution might be to take a large diameter aluminum shaft about 3 inches long, put an 8mm hole in the bottom for the CIM, use a CNC mill to make the pulley part on the bottom, then turn down the rest to a 3/8" diameter.

Alternatively you could try to move both CIMs up to the upper plate and route the belt in a triangle around the large pulley.

[magnets]

Originally Posted by Nate Laverdure

Alternatively you could try to move both CIMs up to the upper plate and route the belt in a triangle around the large pulley.

I tried that, but when I did that, the belt calculator showed that only 2 teeth of the belt would be engaged.

[CalTran]

Team 16, Bomb Squad, ran a pretty effective three wheel swerve for a few years. IIRC, there's documentation on here somewhere about it. Check the white papers.

[Ether]

Here's a link to a PDF giving the equations for inverse kinematics for an N-wheel swerve, where N>2.