Swerve Concept
 

Hi I have been working on this design with a mentor on my team, its getting pretty close and thought id see what you guys think.

Some of the gears I realized after endearing were not mated properly. I fixed them so don't worry.

I have a few questions as well, what do you think the best way to hold the gears in place is? also, what do you think of the design overall?

If you are confused just ask, whenever i show this to someone they are usually confused.

Re: Swerve Concept
 

Are you asking how to hold the crown gears in the module in place? or the spur gears between the motors.

When we did our car/crab drive in lunacy and breakaway we used one single brass pillow block to hold both crown gears and support the axles. of course, it looks like the large crown gear shares an axle with the wheel here, where ours had another stage lower down

For the spur gears (depending on live/dead axle) there is hardly any axial load (gravity isn't much) if you can machine groves for E-clips they would work just fine. Shaft collars (especially the thin and light VexPro ones) would work as well

EDIT: looking back at the pictures, PVC or Delrin (or anything really) spacers would hold the spur gears

Re: Swerve Concept
 

What are you using for your bearings between the rotating module and the frame?

Also, if this were to be made, how would you make the .125" sheet metal part? Getting that to line up nicely for a gearbox is easier said than done.

As others have said, the normal gears and their shafts can be held in place with snap rings or spacers. It is also useful to turn down the end of the 1/2" hex shaft to 1/2" round and use a round bearing to keep the shaft from sliding out. The bevel gears will have thrust loads, which are along the axis of the shaft so it's important to be aware of this in your design. Team 1640 uses a thrust bearing probably similar to this one http://www.mcmaster.com/#6655k13/=smu6aw

I've also seen teams get away with using a thrust washer similar to this one
http://www.mcmaster.com/#5906k411/=smu6ry

Where do you plan on purchasing the bevel gears from? Also, where does the wheel go?

Re: Swerve Concept
 

We will have 2 wheels on either side of the bottom gearbox with the bevel gears.

The module will be cut and bent at a laser cut place near our build space.

The bearings for rotation is a big thrust bearing on the part in the upper gearbox, and a beveled bearing(I think thats what its called, mcmastercarr isn't loading for whatever reason ATM). EDIT: its a roller bearing

Im not sure where we got the bebe gears from, we don't meet until the 7th, but ill ask then

Re: Swerve Concept
 

Why do you want wheels on each side? What reduction are you using for the BAG motor to pivot them?

Also what are the flanged plates made of? 0.125" aluminum?

Re: Swerve Concept
 

2 Wheels allow for more traction and less stress on the bevel gears, plus its just helps with overall stability, we are using the new 4in vex wheels

We will probably use .125 in

Re: Swerve Concept
 

This one?

Well keep in mind that, while more wheels will keep the tread pattern from wearing down as quickly, doubling the number of wheels will give only a very minor improvement in traction since, while it is simplified, friction very closely equals N*mu. Having a wheel on each side will significantly increase the torque required for pivoting and it will probably take a minimum of a quarter of a second or so to rotate a module 180 degrees.

Re: Swerve Concept
 

On a deformable surface like carpet, the n mu approximation isn't a valid assumption, he'd pretty much need to test it himself with his wheel setup in order to figure out how friction would be effected. With a rough top tread I would think you would get more friction, but probably not with a smooth vex wheel, but this is just a guess. The increase in turning torque is correct though and if more surface area was beneficial I'd just have a single thicker wheel to take advantage of the low torque required to turn near the axis of rotation.

Also, getting the holes on the two 1/8th inch pieces of aluminium to line up is going to be very challenging, I would recommend making it out of a single piece of extrusion or two flat plates with standoffs.

Re: Swerve Concept
 

Do you know of any quantitative data that teams have collected about this with different wheels/treads? I have been unable to find any.

Re: Swerve Concept
 

Nope. Vex has some CoF's listed, but they say nothing about the testing method or how it's related to surface area. Later this summer I plan on doing a test to determine this with different wheels and to figure out which gives the best traction, but this probably won't be for a while.

Re: Swerve Concept
 

Dangerous statement here. I agree with everything you said EXCEPT the claim of .25 seconds to go 180*. That speed would actually be pretty quick for most swerves people are currently running.

Re: Swerve Concept
 

We will see how accurate our vendor is, we may have to do standoffs, we'll see

As for the wheels, you bring up a good point, the 2 wheels will add a lot of friction. I don't foresee that it will be too much of an issue.

I think the best thing we can do is build one module like it is, then we'll have to tweak it quite a bit. What i like about it is at the form factor, its very small, and its all gear driven with no belts or chains which means less chance of things breaking.

Thinking about it more, It would also be pretty difficult to make this design with a single wheel. I still think we will have enough torque to turn the wheel with adequate speed. There really is only one way to find out..to test it.

Re: Swerve Concept
 

2 wheels doesn't inherently equal more traction. It would result in pretty much the same stress on the bevel gear as well - how do you figure that it would result in less?

Please don't guess.

The edge case I know about that seems to affect pushing force based on wheel width is roughtop tread on 4" wheels. I remember reading data somewhere (but not experimenting myself, something to do in the fall) which showed a negligible difference in traction for 6" and 8" wheels of different widths as well as wedgetop tread. Specifically 4" roughtop tread wheels have noticeably better traction in a 2" wide configuration as opposed to 1" wide.

I have never seen any data on Colson wheels or Vex wheels with regard to width versus traction. Based on purely subjective experience, I think people are somewhat overstating the benefits of a wider wheel in terms of carpet traction. We would have to do testing to be sure though, perhaps in the fall.

Yes, definitely a factor here. Your turning torque is a function of the size of your moment arm when spinning the wheel, so the farther from the center the tread is in contact with the ground the greater the torque needed to turn the module.

Alternately they could leave the bends in, with a bit more clearance away from each other, and use standoffs to actually hold the plates together to ensure alignment. This keeps the rigidity benefits of the bends.

OP: Have you thought about how to mount this to a frame? A popular idea on the west coast is to make a 2x1 frame and mount the gearbox to each side of the 2x1, using the existing frame as a big spacer. Also, what are your gear ratios?

Not at all, you can easily do some math to get an approximation of how much torque it takes to turn a wheel and whether or not your system has enough reduction to do so.

Re: Swerve Concept
 

It would be advisable to go further on the speed side of the motor curve than one would be with the reduction that I used in my calculation, hence a minimum of a quarter of a second. I suppose that half a second would be more reasonable in this set up.

Re: Swerve Concept
 

Some things to think about. Weight of a swerve drive is often very heavy trying to reduce the weight is a major chore.

Just to keep in mind. The closer your gears are to the final reduction of the drive train the more or a moment they will be taking. This can be come a huge problem when you take large amounts of pushing or something unexpected. The moment is directly related to your wheel size.

Also remember your whole robots weight will be carried through the upper gear box this can cause major problems for thin metal flexing and losing gear spacing causing them to skip/bind I like belts for this reason. The competition is not the most ex-stream things that will happen some one might drop it and one of the poor modules will take the load of a potently 150 pound robot falling from 2 feet. I like belts for this reason.

My team has been designing a swerve drive using the same duel wheel concept.
http://imgur.com/a/H6Qcm
The main goal over the old was was to reduce the size/ machining complexity/assembly complexity/points of failure/ and the biggest was increasing mechanical efficiency.

Re: Swerve Concept
 

I'm not sure where you're getting your numbers at all.

The two wheels spaced far apart will take substantially more torque to turn, there is no way they will get BETTER performance than current optimized narrow centered wheel swerves.

Re: Swerve Concept
 

The original idea was to put holes in the 1in flanges pointing up/ down and bolting that into the frame with a sorta of square of 1x1, (ill post a pic later)

I do see how it could twist and bend, I have some ideas...

Im thinking we may stick with 2 wheels for now, I don't think its worth a total redesign of the bottom GB ATM, What id like to do is just get a prototype made to see how it it all goes together and see what has to change.

I might play around with a 1 wheel version if I get the time to. This is actually my first major CAD project, so I am pretty proud of it, but I see a lot of valid points that ill try to implement. I really appreciate the feedback.

PS:

This is the large tapered bearing I was talking about http://www.mcmaster.com/#5709k31/=smy6cy

Re: Swerve Concept
 

Try 3D printing it out to insure that everything works as expected. It is always easier to re design then spend a couple hundred hours of frustration trying to get a bad concept to work.

Re: Swerve Concept
 

The biggest thing is seeing what our vendor's tolerances are and their accuracy, but I do want to 3D print it as well

Re: Swerve Concept
 

In this post, OP said a BAG is used for pivoting

I estimated that around 80in*lbs would be needed to turn that two-wheel module. A BAG motor has a stall torque of 3.5in*lbs, so, while more reduction would be better, I went with 100:1 in my calculation. Free speed of a BAG motor is 14000 rpm, so 140rpm with the reduction and 108rpm under the load. That works out to 0.56 seconds per rotation, or a bit over 0.25 seconds for 180 degrees.

So, while it may take more torque to pivot and while one may want to run the motor further on the speed side of its curve, that is where I got my numbers.

Re: Swerve Concept
 

We will most likely be using a 100:1 ration on a versa planetary gearbox, then that turns a 34t gear which tuns the 60t turning gear, I'm not to good with these gear calculations so yeah.

Re: Swerve Concept
 

Okay, so about half a second or so.

Re: Swerve Concept
 

Based on assumptions you've made of the loading without evidence.

Time to rotate must also factor in the control loop. It can't operate at full speed the entire time.

I'm not trying to pick on you, just pointing out it's dangerous to publicly state something that can be perceived as fact when it's not a proven fact.

Re: Swerve Concept
 

Don't worry, I appreciate your responses. I'll make my assumptions and thought process more clear in the future when I make an estimation.

Re: Swerve Concept
 

Why would you ever rotate a wheel more than 90 degrees at once?

Re: Swerve Concept
 

I was about to ask this.

Re: Swerve Concept
 

You wouldn't, that was just my picturing of the module rotating, not really applicable.

Re: Swerve Concept
 

We will to calibrate the steering, but only then

Re: Swerve Concept
 

For reference, last year we used Bag motors with 90:1 versa planetary gearboxes (10x3x3) for steering with no issues. We had no further reduction after the Versas.

Re: Swerve Concept
 

If you don't want to switch the direction of your drive motor, like team 118 did in 2008.

Re: Swerve Concept
 

FYI, I just tested ours using smartdashboard. It looks like 200 ms for a 90 degree turn, which matches our design calcs pretty closely.

Re: Swerve Concept
 

Mind posting motor, gearing and what the wheel is? Thanks!

Re: Swerve Concept
 

AM0912
100:1 VersaPlanetary
42T:66T (we think... At least that's close) sprockets

Edit: AndyMark 6" performance wheels

Re: Swerve Concept
 

A swerve control algorithm can always run the drive motors one direction and check to see which way to turn. Least distance. Or the algorithm can look for the shortest distance with motor reversal. Both will work. We have done the latter. This does put a stress mechanical and electrically on the module. The first has increased response times. Now say you have a CVT solution for swerve, but the response is asymmetric on the direction of the drive motors. Works really good counter clockwise, terrible clockwise shaft end. Do you accept greater than 90 degree steering solutions for CVT or is the steering response more important? That's what I've asked our team.

Re: Swerve Concept
 

So... optimally, what RPM should the module turn at? What speed do teams that consistently use swerve revolve at?

Re: Swerve Concept
 

See my response above. I measured a 90 degree turn at 200 ms. We had designed for around 1 RPM, and it looks like we are a little faster than that.

That is with the hardware-enabled ramp on the Jaguar and limiting the output to 75% in the software. Those guys can eat a battery up if you let them.

Re: Swerve Concept
 

A lot faster. 90deg/200ms = 75 RPM.

Re: Swerve Concept
 

you are correct. Although, I think he meant 1 RPS.:D

Re: Swerve Concept
 

MOE just did this test last week. Our test rig was a simple 4 wheel cart with the wheels locked in place. We used 4 inch AM Performance Wheels (1 inch wide) with riveted on blue nitrile*. The cart was first weighed down to 127lb total and dragged with a force gauge. It slipped when the gauge read ~108lb. The test was then repeated with more weight (192lb total) and slipped at ~160lb. We took this to mean that the relationship between weight and force needed to slip was relatively linear as the % increase in force needed to pull was about the same. We then added 4 more identical wheels to the same 4 axles and locked them in place effectively creating a 2 inch wide, 4 inch diameter wheel. The cart was pulled a third time and the reading was ~180lb. This shows that (at least for a 4 inch diameter wheel with blue nitrile) wheel width will certainly give you more traction but it is not anywhere close to 1:1. More like 100% increase in width = 13% increase in traction, so it's worth is still subject to some debate.

We opted to go with the force gauge test because we felt it was safer, easier and more accurate than the ramp test (the cart tended to fall forward off the ramp before it slipped)

*not new, slight wear.

Re: Swerve Concept
 

Ha! Yes. Good catch. 1RPM wouldn't do a lot of good in a 2 min match.

1RPS

Re: Swerve Concept
 

Wow... that's less than half the proposed speed for this design. What is the reasoning behind that? I guess there's a threshold beyond which the speed qualifies as unnecessarily fast, but where do you draw the line?