Questions about crab/swerve drive.

[BenX02]

Hey, our team is wanting to have crab drive next year, so I have been coming up with designs. I have a few questions.

1) The design I came up with has six CIMs, one for each wheel module, and one for each side to turn the modules. My only concern with this design is that it will be heavy. Can someone show me some designs or CADs (or just explain your design) for your designs. Don't worry, I won't rip them off, I just want to get insperation.

2) This is going to sound like a dumb question, but I have heard that a water jet/computerized mill is necessary for building a crab drive system, may I ask why? The machine shop our team meets at has one, but in the past they have not let us use it.

3) As a driver, I have been wondering about the different ways you can drive a crab/swerve. I have seen teams do a tank style with two joy sticks, where it acts like tank drive but if you move the sticks sideways your robot moves sideways ect. I have also seen with one stick and a throttle, and a stick with a Z axis. But which would you say is the best?

4) I know you need lots of sensors, but what kind and where might I get them?
Thanks in advance.

[R.C.]

Ok,

1) First of all you can't use 6 CIMS. You are only allowed to use 4. To turn the modules, teams use globes or another type of motor. There are some designs on CD, I have some already drawn up and will post them up when I get back home. **school is not fun**

2)CNC machining or a waterjet makes life easier. You can simply draw up the parts, CAM it and run it. Automated machining just makes life easier. Ask them if you can use or if they can mill parts for you.

3)It is up to you on the joysticks. I prefer using tank drive.

[Doug G]

Originally Posted by BenX02

Hey, our team is wanting to have crab drive next year, so I have been coming up with designs. I have a few questions.

4) I know you need lots of sensors, but what kind and where might I get them?
Thanks in advance.

You need to use pots on the steering, or possibly encoders (startup position can be tricky). You should also consider encoders on each drive wheel. It has been said many times on these forums and is worth repeating... Building the crab modules is just one part of the system, controlling it is often toughest part.

[BenX02]

Originally Posted by rc_cola1323

Ok,

1) First of all you can't use 6 CIMS. You are only allowed to use 4. To turn the modules, teams use globes or another type of motor. There are some designs on CD, I have some already drawn up and will post them up when I get back home. **school is not fun**

2)CNC machining or a waterjet makes life easier. You can simply draw up the parts, CAM it and run it. Automated machining just makes life easier. Ask them if you can use or if they can mill parts for you.

3)It is up to you on the joysticks. I prefer using tank drive.

Dang! I forgot about that rule. I'm stupid.

I think they will probably let us use it, since we actually did good this year.

That's what I would prefer also. But I have a question. Say next years game is one rug, in previous years you usually had to use omni-wheels for tank drive because of friction, how might you get passed that on a crab drive system?

Sorry if I sound like a noob, I am with crab drive.

Originally Posted by Doug G

You need to use pots on the steering, or possibly encoders (startup position can be tricky). You should also consider encoders on each drive wheel. It has been said many times on these forums and is worth repeating... Building the crab modules is just one part of the system, controlling it is often toughest part.

Okay, and all the sensors available on AM or other websites?

[hillale]

Originally Posted by BenX02

2) This is going to sound like a dumb question, but I have heard that a water jet/computerized mill is necessary for building a crab drive system, may I ask why? The machine shop our team meets at has one, but in the past they have not let us use it.

Precision is key when building a bulletproof swerve drive. This cuts down on everything from chain play to drift. Our 2007 swerve (handmade) was down for some reason or another (thrown chains, etc.) every few matches. Our 2008 swerve (waterjetted) made it through 3 regionals and champs without the need for a fix (until IRI when one of the axles wore far enough to lose it's snap ring which was a less than 1 minute fix with our design).

If you want more info, pm Aren_Hill

[BenX02]

Originally Posted by hillale

Precision is key when building a bulletproof swerve drive. This cuts down on everything from chain play to drift. Our 2007 swerve (handmade) was down for some reason or another (thrown chains, etc.) every few matches. Our 2008 swerve (waterjetted) made it through 3 regionals and champs without the need for a fix (until IRI when one of the axles wore far enough to lose it's snap ring which was a less than 1 minute fix with our design).

If you want more info, pm Aren_Hill

I see. Did you machine your wheel hums out of one piece of metal? I'm guessing not, but it's an idea.

[sgreco]

You definitely don't need a CNC to make a swerve drive, if it's going to be coaxial and you're going to use bevel gears, you probably want to find a shop to help you out. My team made a full swerve on a manual Bridgeport mill in 08.
We used a 4-axis CNC this year, but you certainly don't need that to make it good.

For sensors, my team finds hall effect sensors to be the best because they are absolute sensors which saves time centering the modules(the reading is usually more accurate for us than we found with pots and encoders).

Although you are only allowed 4 CIMs, you probably wouldn't want it for steering anyway. They are too fast. I would use the globe motors. My team used the Denso motors and they worked fine as well.

If you have any questions feel free to PM and I can answer any questions you want about mechanical aspects of swerve.

[R.C.]

Here are some pics of some modules and dt's:

http://www.chiefdelphi.com/media/photos/33368
http://www.chiefdelphi.com/media/photos/33336
http://www.chiefdelphi.com/media/photos/33335
http://www.chiefdelphi.com/media/photos/33334
http://www.chiefdelphi.com/media/photos/32573


CAD:
http://www.chiefdelphi.com/media/photos/32520
http://www.chiefdelphi.com/media/photos/32444
http://www.chiefdelphi.com/media/photos/31626
http://www.chiefdelphi.com/media/photos/31078
http://www.chiefdelphi.com/media/photos/29415
http://www.chiefdelphi.com/media/photos/29412
http://www.chiefdelphi.com/media/photos/28568
http://www.chiefdelphi.com/media/photos/25328

These are just some images and stuff on CD. If you contact the creator, they will for sure help you out. There are a ton of iterations. You just have to find one that fits your needs. On sensors, since MadTown will never do Crab (I mean NEVER). I don't know. But someone will post a link soon hopefully

[big1boom]

Might as wel preface this saying that I only have 1 year (this year) experience with crab/swerve

Originally Posted by BenX02

1) The design I came up with has six CIMs, one for each wheel module, and one for each side to turn the modules. My only concern with this design is that it will be heavy. Can someone show me some designs or CADs (or just explain your design) for your designs. Don't worry, I won't rip them off, I just want to get insperation.

http://www.chiefdelphi.com/media/photos/33181
This is our swerve. My brother posted images of his CAD's later in the thread.


This year we went with 1 CIM total, but for a normal year I would recommend going 1 CIM for each side (r/l) The main problem with this setup is that you have to be exact for the bevel gears, otherwise the will not mesh correctly.
We used globe motors to turn our wheel modules. I have seen other teams use FP motors, or window motors.

Originally Posted by BenX02

2) This is going to sound like a dumb question, but I have heard that a water jet/computerized mill is necessary for building a crab drive system, may I ask why? The machine shop our team meets at has one, but in the past they have not let us use it.

It is not necessary to have a computerized mill, although it does make everything easier.

We had our boxes cut out by an old CNC, then we drill pressed all the holes, then had them welded.

Originally Posted by BenX02

3) As a driver, I have been wondering about the different ways you can drive a crab/swerve. I have seen teams do a tank style with two joy sticks, where it acts like tank drive but if you move the sticks sideways your robot moves sideways ect. I have also seen with one stick and a throttle, and a stick with a Z axis. But which would you say is the best?

This year we went with 1 Joystick controlling throttle, one joystick with twist controlling translation (strafing) and rotation.

Originally Posted by BenX02

4) I know you need lots of sensors, but what kind and where might I get them?

I didn't have involvement in sensors other than mounting them, but I know that we mounted a magnet onto a sensor sprocket shaft, then put the sensor right above it, no contact. This sensor then outputs a voltage that corresponds to an angle. According to our team's programmer, he has to round the values otherwise it will crash/slow down the cRIO.

If you have any more questions, please post here, or PM someone. Most people are more than willing to help

[sgreco]

This a is a swerve I drew up during the build season with the specific intention of not needing a CNC machine. However, you do need a rotary table for your mill with this design.



The sprockets for powering are not shown, but they could simply be attached with a keyed connection. The steering sprocket is shown, but the motors for steering aren't because we didin't know what method of actuation we wanted or which motors we would need elsewhere. I would recommend steering the front wheels together and the back wheels together for a little bit of flexibility in steering modes.

The CIMs can also be down into the frame. The only reason they a sticking up is because of a few specific wiring issues that pertained to my team's 2009 bot.

[BenX02]

Originally Posted by big1boom

Might as wel preface this saying that I only have 1 year (this year) experience with crab/swerve


http://www.chiefdelphi.com/media/photos/33181
This is our swerve. My brother posted images of his CAD's later in the thread.


This year we went with 1 CIM total, but for a normal year I would recommend going 1 CIM for each side (r/l) The main problem with this setup is that you have to be exact for the bevel gears, otherwise the will not mesh correctly.
We used globe motors to turn our wheel modules. I have seen other teams use FP motors, or window motors.



It is not necessary to have a computerized mill, although it does make everything easier.

We had our boxes cut out by an old CNC, then we drill pressed all the holes, then had them welded.


This year we went with 1 Joystick controlling throttle, one joystick with twist controlling translation (strafing) and rotation.



I didn't have involvement in sensors other than mounting them, but I know that we mounted a magnet onto a sensor sprocket shaft, then put the sensor right above it, no contact. This sensor then outputs a voltage that corresponds to an angle. According to our team's programmer, he has to round the values otherwise it will crash/slow down the cRIO.

If you have any more questions, please post here, or PM someone. Most people are more than willing to help

Thanks for the help.

You said the CADs later on in the thread were your brothers right? I see those rotating things on the bottom and top of your modules, can you explain to be what those are? I see them on a lot of different designs.

Sorry about all the questions, I have a lot to learn about crab drive.

[big1boom]

The rotating things are Square Lazy Susans from McMaster

We have one on top, and one on bottom of each box, this defines the vertical plane of the wheel with 2 points, making it much stronger. Most swerve modules are supported from top and bottom to increase strength and reliability.

Keep asking questions, I know that I don't mind sharing information that might help someone.

[EricH]

You're all forgetting one thing.

We don't know that a swerve drive will be good for next year's game.

So, by "next year", you really mean "next fall, as a prototype", riiight? You see, crab/swerve drives are pretty complex, especially compared to tank drive. So you want to do at least one before it counts, just to help you do the one that counts (if you do it). Move up your schedule and do it in the fall, then evaluate it at the start of build to see if it'll be good for the game.

Now, for some help.

1 CIM per side is still probably too weak. It'll work, yes, but you really want 2, or 1 per wheel module.

There are a number of setups, with the 2 most common being a coaxial crab and a more standard swerve. 118 uses coax, 148 used coax last year, and some other teams do as well, I think. In this setup, all the drive motors power all the wheels. A second rotation shaft is used to rotate the modules. You could set this up to act as either all 4 steer together or as a 2x2.

The other main option is to put the motor in the drive module. That's a little tricky on the wiring side, but it's a bit simpler mechanically.

[BenX02]

Originally Posted by big1boom

The rotating things are Square Lazy Susans from McMaster

We have one on top, and one on bottom of each box, this defines the vertical plane of the wheel with 2 points, making it much stronger. Most swerve modules are supported from top and bottom to increase strength and reliability.

Keep asking questions, I know that I don't mind sharing information that might help someone.

Ah, thanks. I will look into those.

Thanks, you've been a lot of help. I think most of my questions have been answered, but thanks again.

Originally Posted by EricH

You're all forgetting one thing.

We don't know that a swerve drive will be good for next year's game.

So, by "next year", you really mean "next fall, as a prototype", riiight? You see, crab/swerve drives are pretty complex, especially compared to tank drive. So you want to do at least one before it counts, just to help you do the one that counts (if you do it). Move up your schedule and do it in the fall, then evaluate it at the start of build to see if it'll be good for the game.

Now, for some help.

1 CIM per side is still probably too weak. It'll work, yes, but you really want 2, or 1 per wheel module.

There are a number of setups, with the 2 most common being a coaxial crab and a more standard swerve. 118 uses coax, 148 used coax last year, and some other teams do as well, I think. In this setup, all the drive motors power all the wheels. A second rotation shaft is used to rotate the modules. You could set this up to act as either all 4 steer together or as a 2x2.

The other main option is to put the motor in the drive module. That's a little tricky on the wiring side, but it's a bit simpler mechanically.

Well of course that's what I meant.

And yeah, I was aware that one would be to weak. I was thinking of using six CIMs (before someone reminded me of the rule that you can only use four), one for each module, and two to turn them.

I have been thinking of the wiring, how might you go about doing the wiring for that? I'm trying to figure out a way so that they don't get wrapped around.

Thanks for the help.

[RMS11]

Originally Posted by BenX02

I have been thinking of the wiring, how might you go about doing the wiring for that? I'm trying to figure out a way so that they don't get wrapped around.

Thanks for the help.

If you have the motor on the module then you would probably need to have a mechanical stop so the module could not spin freely and some kind of wire holder that allows for all the wires to stay organized, but im forgettin the name right now. Or if you do coaxle then its not a problem