Questions about crab/swerve drive.

[AdamHeard]

Originally Posted by Steve_Alaniz

Could you give some examples please? Any drive that might stall will have problems but that is pretty much ...any drive. I think you are referring to a transmission problem that might occur under stress but I wouldn't want to put words in your mouth. You are correct that the slick surface allowed for easy turning but our drive actually worked better on carpet, so it was actually a challenge for the drivers to adjust to the regolith.
Our shafts were 1/2 in steel so they were not going to break but we did worry a bit about the transmission so we used the banebot 16:1. I am under the illusion we were pretty bullet proof. At least we plan to build on what we learned this year.

I'm actually referring to the loads the module itself has to deal with, the sideloads on the module can be massive if you turn while moving, or are pushed, or just about any situation really.

Many modules I saw this year made me think, "wow, if that was on carpet, it would just collapse when you turn".

Working isn't good enough, it has to "Work" while turning at 20fps and getting it from another robot, or being pushed sideways by two robots. If your modules can't survive those loads, it'll be a rough season.

[Al Skierkiewicz]

Why not have a wheel encoder tied to a non-driving wheel. Use two at ninety degrees to each other using omni wheels and then you don't have to worry about running wires down inside the modules.
Please keep in mind that bevel gears (those normally used on coaxial drives) are relatively inefficient in use for driving and require some pretty meaty bearings to keep them aligned under load and stress.
Slip rings can be made by using circular, machined conductive surfaces, insulated from the shaft by PVC, polyeurathane or delrin. Brushes can be bought anywhere, even at Ace or TruValue. All you need to do is keep them aligned in the assembly and have some form of attachment. Using brass allows you to solder wires directly to the rings.

[Steve_Alaniz]

Originally Posted by AdamHeard

I'm actually referring to the loads the module itself has to deal with, the sideloads on the module can be massive if you turn while moving, or are pushed, or just about any situation really.

Many modules I saw this year made me think, "wow, if that was on carpet, it would just collapse when you turn".

Working isn't good enough, it has to "Work" while turning at 20fps and getting it from another robot, or being pushed sideways by two robots. If your modules can't survive those loads, it'll be a rough season.

Good point! So I guess you mean the turning bearing on the module is a critical spot. I'd have to agree with you. (A Lazy Susan probably wouldn't cut it. )
Thanks for clearing that up.

Steve

[Steve_Alaniz]

Originally Posted by Al Skierkiewicz

Why not have a wheel encoder tied to a non-driving wheel. Use two at ninety degrees to each other using omni wheels and then you don't have to worry about running wires down inside the modules.
Please keep in mind that bevel gears (those normally used on coaxial drives) are relatively inefficient in use for driving and require some pretty meaty bearings to keep them aligned under load and stress.
Slip rings can be made by using circular, machined conductive surfaces, insulated from the shaft by PVC, polyeurathane or delrin. Brushes can be bought anywhere, even at Ace or TruValue. All you need to do is keep them aligned in the assembly and have some form of attachment. Using brass allows you to solder wires directly to the rings.

Hey Al! Nice to see you... Really? You'd use slip rings? Ok but I would be concerned about the electrical noise they would make... One is bad enough but four? I guess that's why I never tried it. Still Id like to hear your thoughts.

Steve

[Al Skierkiewicz]

Steve,
You asked an engineering question, so I gave you an answer. I didn't say it was a good one. There is no possible way to not have some noise in a slip ring, particularly one that is carrying a lot of current. However, with large brushes or multiple brushes, you should be able to transfer enough current to be useful. Slip rings for sensor use should be fine since the current is fairly low, and the noise should also be low. Please be sure that power supply for sensors are filtered near the sensor.
There are some mercury filled rotary electric connections out there. I am against using them but I do believe the GDC answered (In a Q&A) that they could be used this year. In the hard hits, and misalignments I see on robots during inspection, I believe the chance for a mercury spill is pretty high. For that reason, I advise against their use.

[Vikesrock]

Steve, I don't know if what Adam saw was similar, but I saw at least two crabs this year that would not sustain the axial loads put on them by rapid change in wheel direction or by being pushed across the carpet perpendicular to wheel direction.

I am far from a mechanical expert (electrical and software are more my thing), but I am near positive that the two crabs I saw would fail in an FRC game on carpet. One team was definitely aware of this and the module was designed this way because of the game. I didn't get a chance to talk to the other team at all to see if they knew what issues they could face if we move back to carpet next year.

[sgreco]

Originally Posted by Vikesrock

Steve, I don't know if what Adam saw was similar, but I saw at least two crabs this year that would not sustain the axial loads put on them by rapid change in wheel direction or by being pushed across the carpet perpendicular to wheel direction.

I am far from a mechanical expert (electrical and software are more my thing), but I am near positive that the two crabs I saw would fail in an FRC game on carpet. One team was definitely aware of this and the module was designed this way because of the game. I didn't get a chance to talk to the other team at all to see if they knew what issues they could face if we move back to carpet next year.

I agree with what you are saying about crabs failing. For an example, my team used made a swerve drive that worked very well, but only on Regolith. If we put higher friction wheels on and drove it on a rug, it probably wouldn't be as successful. We were aware when we built it that the system would only work on regolith with lunar wheels, but that's the only place it had to work. You just have to keep in mind what the challenge is and what the objectives are when you are building it.

For those of you asking questions about swerve for the future, I would really look into 111 and 118. That's where my team got a lot of inspiration from.

If you are interested in looking at my team's swerve, here's a link to a brief description and some photos. http://alarmrobotics.wikispaces.com/Swerve+Design+09

[JesseK]

Just wait... next year's field will be the same Glassliner surface but with a sticky tar coating. It's purpose is to simulate the gravity on the surface of ... Jupiter!

[Andrew Schreiber]

Also, not to knock swerve drive systems but there is something to be said about simplicity. I remember several years ago our buddies from the north, 1114, did swerve. Interesting that they havent done it since then. (Not saying they aren't capable of doing it or that it is bad) Perhaps they concluded that a 6 wheel drop center drive offered more bang for their buck. Remember a swerve drive is usually a significantly harder drive train to fabricate.

Always have a contingency plan, if your fancy swerve driving system won't work you better have something you can replace it with because moving should be step one for any robot.

Adam also brings up a good point, just because teams could get away with being sloppy this year with the robustness of their drive trains does NOT mean that they should. I look at several of these robots and shudder knowing how many pieces they would have been in a couple years ago.

[Jared Russell]

Originally Posted by Al Skierkiewicz

Steve,
You asked an engineering question, so I gave you an answer. I didn't say it was a good one. There is no possible way to not have some noise in a slip ring, particularly one that is carrying a lot of current. However, with large brushes or multiple brushes, you should be able to transfer enough current to be useful. Slip rings for sensor use should be fine since the current is fairly low, and the noise should also be low. Please be sure that power supply for sensors are filtered near the sensor.
There are some mercury filled rotary electric connections out there. I am against using them but I do believe the GDC answered (In a Q&A) that they could be used this year. In the hard hits, and misalignments I see on robots during inspection, I believe the chance for a mercury spill is pretty high. For that reason, I advise against their use.

Regarding mercury-filled electrical connectors, my team has some sitting on our shelf (in 2006 we toyed with the idea of a continuously rotating turret). We didn't end up using them in competition, but you should read the product documentation CAREFULLY if you want to. Currently, no Mercotac electrical contact (Mercotac was the only manufacturer of these as of a couple years ago) is rated for more than 30A of continuous current. This all but rules out these devices for a high-traction swerve drive.

One other warning: slip rings (especially home made ones) always add a bit of noise to a signal. Be careful when using analog sensors with slip rings! (I know from experience!)

[artdutra04]

190 built a custom high-power slip ring for their 2008 robot, along with a commercial slip ring inside of it for all of the sensors. Here's a photo of their [nearly completed] slip ring during the build season:

[Alan Anderson]

Originally Posted by Jared341

Currently, no Mercotac electrical contact (Mercotac was the only manufacturer of these as of a couple years ago) is rated for more than 30A of continuous current. This all but rules out these devices for a high-traction swerve drive.

That's 30 amps per contact, right? The one we used on the 2008 TechnoKats robot's continuous-rotation arm has eight contacts. Use four of them for a pair of 30A supply and return lines, and you still have four left over for an encoder if you want.

(The list price of such things might be higher than the single-component cost limit, but many suppliers have an educational discount that they offer to any team who asks for it.)

[Team1710]

To determine the position of the wheels on a swerve drive, would one want an absolute rotary encoder or a relative rotary encoder?

[Aren_Hill]

We've always used the Ma3 abolute analog encoder which kept its orientation so we didnt have to line up the wheels before we turned the bot on

[R.C.]

Originally Posted by Aren_Hill

We've always used the Ma3 abolute analog encoder which kept its orientation so we didnt have to line up the wheels before we turned the bot on

Aren,

A link would be very handy, we kinda wanted to buy some.

Thanks,

-RC