1st Attempt at Swerve Drive

[SerpentEagle]

I've been working on this swerve drive train in the past couple evenings, thought I might drop it by here and get some feedback.



Specs:

-6 CIM
-Theoretical ~18 ft/s
-~40 lbs

Link to model: https://grabcad.com/library/crabdrive-01-1

A lot of it isn't really detailed yet, but hopefully it is enough to convey the idea.
Any and all feedback is appreciated, thanks.

[Landonh12]

I think it would be better if you made it a full swerve drive instead of a crab drive. I'd split the CIMs up and use one per wheel so that you can independently drive each one. That would also drop your CIM count to 4 so that you can use CIMs in other places.

[John Weissman]

As a mentor for 1640, we have been doing swerve for 5+ years. If you look on our website team 1640 wiki, you can find out how we build our swerve drive, including programming, cad, and the math behind it.

[TheModMaster8]

Thats a really nice design! slick and slim, if you are going to make another version and are looking for more info here's the one my brother made https://grabcad.com/library/swerve-d...or-frc-robot-1
he has since made a better,smaller, and lighter one. if you need or want any help just ask

[Moskowapplepi]

We have done swerve in the past but what are the 3 motors in the middle of each side for? We did individual modules rather than the entire drive train, in case you wanted to know why I asked

[marshall]

This design looks awesome!

We experimented with a similar design earlier in the season but did not use it this year. The one thing we looked at carefully with this style of design was the strength of the cross members with the CIM motors requiring so much material to be removed. If you are building a drivetrain with 6 CIM motors then you are designing for defense and you can expect to take some hits so you need to design for that.

[SerpentEagle]

Originally Posted by Landonh12

I think it would be better if you made it a full swerve drive instead of a crab drive. I'd split the CIMs up and use one per wheel so that you can independently drive each one. That would also drop your CIM count to 4 so that you can use CIMs in other places.

Yeah I was thinking about that, but my main goal for this design was to create a drivetrain that had smooth, basic omnidirectional movement, yet can still travel as fast and defend as well as a tank drive. 6 CIMs provide exceptional power for that.

Originally Posted by John Weissman

As a mentor for 1640, we have been doing swerve for 5+ years. If you look on our website team 1640 wiki, you can find out how we build our swerve drive, including programming, cad, and the math behind it.

Awesome, Ill take a look. Thanks.

Originally Posted by TheModMaster8

Thats a really nice design! slick and slim, if you are going to make another version and are looking for more info here's the one my brother made https://grabcad.com/library/swerve-d...or-frc-robot-1
he has since made a better,smaller, and lighter one. if you need or want any help just ask

Looks great, I wont hesitate to contact you!

Originally Posted by Moskowapplepi

We have done swerve in the past but what are the 3 motors in the middle of each side for? We did individual modules rather than the entire drive train, in case you wanted to know why I asked

Those are the motors driving each pair of wheels (There are supposed to be belts, its not shown).I did this to increase robot acceleration and pushing force. This drivetrain was oriented towards a 2014 like game, and I noticed that the one main drawback of the swerve designs of that year was robot speed and acceleration; they just couldn't keep up with many tank style drives, so in theory 6 CIMs should be a big upgrade to this.

Originally Posted by marshall

This design looks awesome!

We experimented with a similar design earlier in the season but did not use it this year. The one thing we looked at carefully with this style of design was the strength of the cross members with the CIM motors requiring so much material to be removed. If you are building a drivetrain with 6 CIM motors then you are designing for defense and you can expect to take some hits so you need to design for that.

Yeah, I did kind of laugh at the amount of lightening holes I put in. Ill be sure to do some FEA analysis of the frame later on.

[asid61]

First of all, that looks really nice, great render! Your modules look beefy even if they are a little tall, and your 3x3 siderails look strong as well.

Some possible changes you could make:
-4 cim + 4 minicim, 1 of each on each wheel. This lets you have omnidirectional motion. As you have it right now you can move in only 2 axes, and 4/4 gives you even slightly more power than 6 cims. That being said, I really like the form factor of your CIM gearbox.
-Are those HTD profile or trapezoidal profile timing belt pulleys (looks like the latter)? HTD and GT2/GT3 are much better due to the greatly increased working load if you are using trapezoidal.
-I can't open your assembly that I downloaded from Grabcad. If you want other people to be able to look at your full assembly, you need to use File -> Pack and Go and upload that. That will put all the parts and assembly into a single folder, and update all references. Or, upload a step file (but the former is easier for SW users to deal with). Right now I can only view the file in GrabCAD's viewer.
-Good use of extended inner ring bearings for separating the shaft collar from the outer race of the bearing.
-You may not need a 2-stage VP for rotation. You get a speed of 220rpm for turning each module with just a 10:1 x 84:14 reduction, but it's your call what suits you of course. Changing would result in a 7A draw on the BAG motor normally.
-Double shielded or double sealed bearings are much better than open bearings for your shaft support, simply because stuff doesn't get into to them as easily. Look into using 6800 series metric bearings as well; they can be super handy due to their decreased OD (and resulting lower weight). I don't know if you need to switch for sure though, due to the low turning speed and orientation.
-Why the spacer in between the caster box and the 84t gear?
-Using long screws to clamp on a tube to connect the corners will not be very strong. You should stick an insert inside there to prevent crushing (plastic is ok)
-Your free speed looks like 19fps, is that correct?
-Nice attention to detail relieving the corners for the VPs! I really like the amount of thought you've put into this. The attention to small details really is amazing.
-You may or may not want a bellypan, as the sheer (shear) strength of the frame rails might be enough.

[SerpentEagle]

In response to asid61:

First of all, that looks really nice, great render! Your modules look beefy even if they are a little tall, and your 3x3 siderails look strong as well.

Thanks, the modules are a bit tall. This is for a 2014 like game, so I might find a way to lower them later on.

Some possible changes you could make:
-4 cim + 4 minicim, 1 of each on each wheel. This lets you have omnidirectional motion. As you have it right now you can move in only 2 axes, and 4/4 gives you even slightly more power than 6 cims. That being said, I really like the form factor of your CIM gearbox.

I was thinking of 8 mini cims, but that's even better. I didn't realize how much movement I was sacrificing with the 6 CIM config; I don't really understand how swerve movement works yet. I should probably look into it.

-Are those HTD profile or trapezoidal profile timing belt pulleys (looks like the latter)? HTD and GT2/GT3 are much better due to the greatly increased working load if you are using trapezoidal.

Those are trapezoidal pulleys; I kinda just threw in whatever I found on McMaster Carr. Ill make sure to swap them out in the future.

-I can't open your assembly that I downloaded from Grabcad. If you want other people to be able to look at your full assembly, you need to use File -> Pack and Go and upload that. That will put all the parts and assembly into a single folder, and update all references. Or, upload a step file (but the former is easier for SW users to deal with). Right now I can only view the file in GrabCAD's viewer.

Interesting.. I saved with all parts referenced. Ill make sure to use the pack and go method as well as upload a .stp file.

(Done)

-You may not need a 2-stage VP for rotation. You get a speed of 220rpm for turning each module with just a 10:1 x 84:14 reduction, but it's your call what suits you of course. Changing would result in a 7A draw on the BAG motor normally.

I should have mentioned this, but those are actually single stage (10:1) reductions with an additional stage for versa encoders. I was a little worried about torque, but it seems like it should be fine.

-Double shielded or double sealed bearings are much better than open bearings for your shaft support, simply because stuff doesn't get into to them as easily. Look into using 6800 series metric bearings as well; they can be super handy due to their decreased OD (and resulting lower weight). I don't know if you need to switch for sure though, due to the low turning speed and orientation.

Ill definitely take a look into metric bearings; I kinda got scared away by the idea of metric units

-Why the spacer in between the caster box and the 84t gear?

Partially to hold the caster tube in place (holds down a flange on the tube) and partially to provide clearance for the versa planetary gear and shaft from the top of the module. Really the caster mechanism is in general sort of sketchy; that will be the first thing I will be fixing.

-Using long screws to clamp on a tube to connect the corners will not be very strong. You should stick an insert inside there to prevent crushing (plastic is ok)

I was actually intending to use short bolts on the inner wall of the tube; Do longer bolts offer significant structural benefits?

-Your free speed looks like 19fps, is that correct?

Thats what if was originally, but I forgot that I lowered it to about 15.25 ft/s (~19 ft/s with a 19% speed loss constant)

-Nice attention to detail relieving the corners for the VPs! I really like the amount of thought you've put into this. The attention to small details really is amazing.

Thanks! Thats one of the biggest things I learned this build season. The killer lies in the details.

-You may or may not want a bellypan, as the sheer (shear) strength of the frame rails might be enough.

I'm planning to add one.


Many thanks for your feedback! You should know that you are a huge help to the CD and FRC community.