pic: Ratchety Swerve Progress 2

Aren’s also guilty of making a 6-wheel crab drive (which made it to Einstein). Unlikely designs can go very far in the right hands.
3:1 on a 2" wheel should be fine, but it will probably end up more compact if you keep the CIM outside of the module in a coaxial setup.

I listened to Aren, that’s why the module looks so much different than it used to. I focused on making the module to where someone could take the ratchet off and still have a nice swerve module (albeit the gearing is a bit speedy).

Just out of curiosity, where are you sourcing these pulleys? Any data here forward is based on info from SDPSI. Also, I’d highly suggest considering 5mm GT2 belts because the 5mm HTD profile readily available in FRC applications doesn’t perform as well in terms of power transmission and load carrying capability, but for the sake of dimensional analysis sizes are pretty similar because they share the same pitch.

You probably won’t meet the load handling requirements for an FRC drive application with smaller pitch belting options (GT2 3mm for example). There are design documents available from all of the belt manufacturers that explain how to use their products appropriately.

You probably can’t go much smaller than a 12t pulley for the belt pinion on your motor shaft in a 5mm pitch, because of minimum teeth in mesh requirements (6 to meet stated performance from the belt manufacturer). Even if you choose to ignore that requirement you run out of pulley material between the root of the teeth and the bore of the pulley at about 9t.

As far as the largest pulley (Assuming it’s coaxial to the wheel) you can’t go much larger than 28 teeth if you don’t want your belt in contact with the ground.

If you aren’t going coaxial and you’re using a transfer stage to drive your wheel you have the option to get a more reasonable set of gear ratios, since you might as well take advantage of that interface to get a second reduction.

I’m really curious how you’re going to make this work, maybe you’ve come up with a more creative solution than I’m aware of. If you ignore this and put your belt literally in contact with the ground, supporting the weight of the robot, you’re limited to around 32t.

I’d theorize your maximum achievable ratio assuming (near, you need an idler to get a 6t interface on a 12t pulley with such a short run) conformance to manufacturer specifications is probably 12:28, or 2.33:1, which would make you traction limited with a full weight robot on some of the lower CoF wheel options available (1.07 yields about what you’re after in this case). This robot is capable of about 16.88 ft/s.

Throwing out conformance to manufacturer specifications and a willingness to run your belt on the ground, you’re looking at 9:32 or 3.56:1, which would make you traction limited with a full weight robot on pretty much any conventional FRC wheel option in terms of CoF. This robot is capable of around 11 ft/s.

In regard to the wire tensioning system you described; does that mean you have a fixed amount of rotation per module per match? What happens when one runs out in a match? Also, because of the ratchet you’re going to be doing more degrees of rotation in about 50% of your moves because of the unidirectional rotation.

I think you have a really interesting idea here. Looking forward to seeing what solutions you can come up with for these really significant design challenges.

Thanks! It really does come with it’s own very unique design challenges :rolleyes:

As for your comments: Custom machined pulleys (looking into V belts). Drive pulley is 0.5" in diameter (could be smaller if I were to lock it in via a set screw in the CIM’s output shaft), driven pulley is 1.5" in diameter (yes, it is coaxial to the wheel. As far as tread, if I were to build it I would use pebbletop (1.24 CoF). The “wire tensioner” is probably only going to let out enough wire for 1-2 full rotations of the module before having to reset and let the wire spool back in.

If you really want to do this then I have a number of questions:

  1. How is that gear on the top being attached?
  2. What is taking up the axial load on the module? What is taking up the radial load on the module?
  3. What are the thicknesses of the top and side plates? What alloy are they?
  4. What size screws are you using to attach the plates?
  5. What is keeping the top plate from buckling like in a 3 point bend test given that it’s just a (heavily pocketed) flat plate?
  6. How are you transferring torque from the CIM to the wheel? (I know you said belt & pulleys but can you elaborate on sizes and tooth profiles)
  7. If those belts are going on the outside of the module (like it seems), are you worried about another robot hitting them or the pulleys? What about hitting the CIM output shaft?
  8. How are you planning on turning the module? (That wide wheel is very scrub-y)
  9. How are you manufacturing the custom wheel? What kind of tread are you using?
  10. What in the module is doing the ratcheting? How much static and dynamic loading can it take?
  11. If you burn out your CIM, how can you replace it?
  12. How are you measuring wheel speed? I don’t see an encoder but it might be hidden.

I know it’s a lot, but these are just a few of the questions you need to consider when designing a good swerve module

Ohhhhhh boy gimme a second.

1.) Pulley is machined into the top of the plate.
2.) Please elaborate.
3.) Top plate is 1/8" high alloy low carbon steel, side plates are 1/2" 6061 alu.
4.) Side plates are attatched via 6 1/8" bolts.
5.) Adding 0.25" standoffs now that you bring that up.
6.) V belts (0.5" pulley driving 1.5" pulley).
7.) No and no, the frame will protect from that.
8.) PG71 with another pulley reduction
9.) Prolly a lathe. Pebbletop tread.
10.) The wheel has a ratchet machined into it, the module has a powl attatched (look at the bootom left of the module)
11.) Take off the right side plate and the CIM will come off with it
12.) Left side plate has room for an encoder mount :wink:

That good?

Axial load: How does this support the robot weight while being free to rotate? That is, what sort of “thrust bearing” will you use?

Radial load: How does this support the forces of accelerating the robot laterally across the floor without binding?

Also, either I’m misunderstanding, or others are - Is the ratchet only on the wheel axis, or is there also a ratchet on the swerve axis? I thought it was only the wheel, but the questions about wire wrapping seem to indicate that there shall also be a ratchet on the swerve axis.

You know what? From a lot of your responses in this thread, I kinda get the feeling you’re just trolling everyone with this design. Either that, or you don’t actually have quite the grasp of drivetrain design that you think you do. I don’t say that to take potshots from across the internet at someone who’s trying to make innovative designs. I love seeing people try out crazy stuff just to see if it can work. But you seem determined to go forward with some very nonsensical aspects of this design. Your response to Ari really gives me the impression that you’re just ignoring the advice people are giving you, and I don’t know why.

I’d be happy to offer my advice and critiques as well, but not if you’re just trolling us or you’re not willing to listen to the advice of others that have already posted.

Amen brother.

I feel like I’m reading “Final 2011 Drivetrain” all over again.

No, you didn’t. If you are going to continue to ignore our feedback and throw it by the wayside then stop asking for it and wasting our time.

I don’t want to sound discouraging, but I can’t help but think it’d be better for you to funnel your time and creative energy into something that might actually be useable. Thought experiments are cool and all, but if you don’t make the slightest effort of linking them to something practical then they’re not really engineering.

“Innovation” is not just a term for doing something that no one’s done before. A lot of things haven’t been done because they’re genuinely not very good ideas. 1:1 gearing for a CIM on a robot drive is a bad idea. A ratchet on a swerve module is (likely) a bad idea. Making a swerve module bulky and complex because “minimalistic and elegant designs have been done already” (per one of your posts in a previous thread) is a bad idea.

Admittedly, as someone who seriously considered going to graduate school in pure math for some time, I’m sympathetic to playing with ideas that may have no grounding in practical use. But, when you ask for feedback on a design, most people are not going to assume that you’re doing that. If you’re pursuing this purely as a personal aesthetic endeavor, then you have to make that clear (and, in that case, it’s not clear what good the advice of others will be). Asking for feedback on a design when you’re not actually interested in suggestions for making it more simpler or more practical is a waste of time - both yours and others’.

The comments on this thread are pretty discouraging. Instead of helping you make this module work (which is honestly not hard at all) a lot of comments seem to be rehashing old points or just being plain nonconstructive. The entire last page hasn’t been “oh that’s interesting, but I was wondering why you chose xxx over xxx”; it’s been “Oh man this design is so bad, where can I begin?”
The only things I would really say are major issues are the plate thicknesses. Move to 1/4" on the top place and 3/8" or 5/16" on the side plates and you’ll be golden. V-belts on drivetrains are a no-go for me, but some teams do them (like 1640 this year), so as long as you prototype that particular bit first I don’t see a deal-breaking issue. Keep in mind the pitch diameter of the pulley is larger than the actual diameter by a set amount, so a 1.5/0.5" pulley reduction might turn into a 2"/1" reduction depending on the thickness of your belt.
PG71’s are slow AF btw, you may want to move to a VP or swap out the motor on the PG gearbox (but it’s your call). The ratchet means that you need to rotate the module farther than other swerves.
Making complicated things is absolutely fine! It’s practically the offseason already. If I decided to give up every time somebody told me “hey that design is dumb” I would never get anywhere.
I recommend you work on this module on your own, or remove the ratchet when you post it to CD. The actual ratcheting part of the module is tiny but it seems to be treading on a lot of people’s toes, so just keep it off for now and add it back on for the final version.

EDIT: The wire thing sounds like a cool idea (although I still like coaxial more). If you ever get that done post the cAD somewhere; I can think of a few times I would like to use that.

To Asid: Thanks man! I’m working on making a new top plate that doesn’t have the steering pulley machined into it. I like the 0.5" side plates where they are at, since they are bored out on the insides to support the CIM and could probably protect the wheel encoder. I also am working on CADing a VP, but I just needed something for now. When I’m done I’ll post the wire tensioner CAD just for you pal :wink:

To Oblarg and Cothron: Thanks for offering to give advice! Will definitely take you up on that :slight_smile:

To FRCguy: If you honestly feel like you’re being trolled than why do you keep commenting? I listen to everyone’s advice, but it’s ultimately my decision on whether or not I follow through with it, okay? If you would like to be the slightest bit encouraging than be my guest bud.

Providing feedback takes time and effort. I don’t think anyone was trying to imply that you need to take all advice that is given, but it does get a bit silly to share knowledge and feedback if we aren’t sure that you’re receptive to it.

You’re getting free advice from people who are doing you a favor by peer reviewing your design. A lot of these guys have many years of following through from design phase to manufacturing to testing and iteration. Rarely will you find a forum of people willing to give you as much guidance as CD. Use it to learn if you want to.

I am highly receptive to it. Everything I read in threads like this I soak up like a sponge (with this many engineers on one site, it’d be foolish not to). What FRCguy was implying, however; was that I ignore advice that is given and waste everyone’s time, and that is why I said what you quoted.

I’m still taking you seriously on this, so I’m going to try to give some helpful suggestions based on your answers.

  1. You are going to want to find a better way to attach the pulley. If you’re machining the whole pulley and top plate from a solid block, that’s a huge waste of metal and machining time. If you’re welding it on, that takes a good amount of skill, leaves weld beads, and opens you up to variations in the angle. It should be possible to drill a bolt circle into the pulley and attach it with bolts.
  2. Exactly what Gus said. I don’t see what is supporting the weight of the robot and the sideways forces on the module. Basically what I’m asking is how is the module attached to the robot. I don’t see any of the normal methods of attaching it so I want to make sure this is something you’re considering when you’re designing the module. Usually people will model part of the frame with the module to be clear on how it is attached.
  3. Honestly, I’m not sure if that’s going to be strong enough. You may want to add more and/or embigulate them. Also, something you may want to consider is where you want the shear forces on the bolts. The way you have it now, the bolts shear when the module slides sideways. If you orient them sideways like in the Revolution Pro 2, they shear from the robot’s normal force. Not saying which one is better, just that you should consider it.
  4. I agree with Anand about your plate thicknesses. 1/2" plate isn’t going to be significantly stronger than 3/8", and you may be able to go even thinner. 1/8" is going to be too thin, especially pocketed like that and without flanges to add rigidity.
  5. Sounds like a good idea :wink:
  6. That drive ratio is definitely better than the original :rolleyes: I don’t have any experience with V belts (the standard is to use timing belts), but that doesn’t need to keep you from experimenting with them.
  7. If you’re confident in that, then ok. Personally, I would want that protected. Especially if the pulleys are plastic. If you bend the CIM shaft (even a little) you will mis-align your pulleys and your drivetrain will be less efficient, if it works at all. If you shatter a pulley or throw a belt, you’re basically out for the match. That scares me.
  8. Whatever you do, just make sure that the modules can spin very fast (i.e. faster than a normal swerve) and with a lot of torque (also more than a normal swerve). Since you can only spin your wheels in one direction, you will need to move them twice as far (on average) to get to the same point. Also, since you are using ultra-wide wheels with the grippiest tread, you will need a LOT of torque to turn these. I didn’t look into the numbers, but you may want a motor more powerful than that. The 775pro will probably work, but again I didn’t run the numbers.
  9. If you really want to spend machining time on those wheels, then go for it. I wouldn’t, but it’s not my design. Are they being lathed from solid aluminum or something else? If so, that sounds heavy. Also, how are you attaching the tread to the wheel hubs?
  10. I still don’t see what benefit this provides, especially given the difficulty in manufacturing, but you seem to be emphatic about this part of the design. Plenty of people have given their opinions on the matter and I don’t have anything new to add to that, so I’ll leave that there.
  11. You might want to look into making the CIM easier to remove. If that burns out in one match, you want to be able to have a new one in there by your next match. Having to basically disassemble your module to do that doesn’t seem like it would meet that criteria. I’m pretty sure that there’s a better way to do that, and I’m going to let you figure out what it is.
  12. Good thinking. Just make sure it’s protected.

If you’re willing to consider CD’s advice, I am more than willing to keep providing it. Just so long as you’re serious.

Thank you much! I’m taking all of these suggestions and making a checklist of improvements to be made :slight_smile:

I’m curious how you are deciding on the top speed of your swerve. Are you selecting a target speed and then applying the appropriate gearing, or are you letting whatever gearing will fit determine the output speed?

I’m letting whatever gearing will fit determine the output speed. The thought behind that was “sice I’m using belts, I can put the CIM as close to the wheel as possible (0.05”) and make appropriate sized pulleys after that". For future projects, however, gearing is going to be the first priority :slight_smile: