pic: 6502 Octocanum Pod



6502 Octocanum Pod. Possible use in 2018

How do you plan on actuating the module? Can’t quite tell from the image.

The plan is to have the pivot point be the Colson wheel’s axle. I think there will be a standoff about 5 inches from the Colson wheel axle which a pneumatic piston will be attached to.

What gear ratios do you have now and/or are you planning on using?

I know sometimes teams have their mecanum go slower than their colsons. (Not that I’ve had any experience with them)

The Colson wheel has two stages, the 12:72 and the 30:64 (6 fps adjusted) and the mecanum wheel just has 12:54 (72t gear is an idler) for 17.3 fps adjusted.

Can’t help but think it looks a tad heavy/overbuilt. Any reason you need the Colson to be so big? Smaller wheel means smaller gear, and less weight. You could also probably eliminate the outermost plate.

If you really want to get fancy/ambitious, you could try doing the whole thing with belts instead of gears (WCProducts sells HTD CIM pulleys).

I think the reason @18sampson decided to stick with the larger Colson was because it perfectly matches the mecanum wheel in width. Although, now that you mention it, I don’t see why this couldn’t be done with a couple of slimmer (and smaller OD) Colsons. Would there be a noticeable difference in traction when using smaller Colsons?

I was under the impression belts lead to tensioning problems and more backlash for any kind of encoder/Talon SRX control. We were planning on having the encoder on the inside of the frame on the end of the fixed axle going through the Colson.

I’m probably wrong. This could probably be a lot more compact with belts, but I’m not sure if the sacrifice in durability/rigidity is worth it. Currently with the pocketed gears the module comes in at <7 lbs, so I don’t think weight will be an issue.

Edit: here was a prototype plate we did for giggles earlier today (in Lexan): http://imgur.com/a/2mQFX

Edit 2: and here are some more renders (not completely up to date yet, extra idler shaft is removed now): http://imgur.com/a/9uHf0

Am I Crazy (I know I actually am) or are there two CIMs in this module? Might they be MiniCIMs?

In these renders there are two CIMs. It’s supposed to have a mini and a full.

The effect of a smaller wheel diameter on traction is likely very small.

I was under the impression belts lead to tensioning problems and more backlash for any kind of encoder/Talon SRX control. We were planning on having the encoder on the inside of the frame on the end of the fixed axle going through the Colson.

I’m probably wrong. This could probably be a lot more compact with belts, but I’m not sure if the sacrifice in durability/rigidity is worth it. Currently with the pocketed gears the module comes in at <7 lbs, so I don’t think weight will be an issue.

Nah, there aren’t really any specific backlash concerns with belts. There’s not much of a sacrifice in durability, either, so long as you pick the right belts/pulleys (I’ve had a grand total of one belt failure in 10 years of FRC, and that was back in 2008 before any of us on the team really knew what we were doing, drive-wise).

The real drawback is that it’s a bit more of a pain to swap out motors with a belt input c.f. a gear input, but it’s still not that much of a hassle. It might also be helpful to do a 192-style tensioning solution, with multiple motor mounting holes for slightly different distances.

Quite the opposite actually. Belts (unlike chain in most cases) don’t need to be tensioned if the center-center distances are setup properly. And belts should give you much less backlash than gears. You can minimize the backlash even further (I believe) by pre loading your belts by adding around 0.003" to your center-center distances.

I’m probably wrong. This could probably be a lot more compact with belts, but I’m not sure if the sacrifice in durability/rigidity is worth it.

I doubt the gears are actually adding rigidity and durability to your design. Your plates and stand-offs should be doing that job for you. There is slop in the Vex gears anyway (they aren’t a tight press fit onto the shaft).

Edit: here was a prototype plate we did for giggles earlier today (in Lexan): http://imgur.com/a/2mQFX
I have to say, it’s pretty awesome that you guys are already producing this. Can’t wait to play with you again next year :slight_smile: What setup/sponsors are you guys using to prototype this?

Reading all the comments in this thread, I think I’m going to try and make a belt design now. Seems like it could save space and allow for more standoffs and actually increase rigidity. In terms of manufacturing it, we have a small CNC router that we might try and use, but we are also sponsored by a company in our area that makes medical grade robotic arms for laparoscopic surgery. We want our school to invest in a decent CNC, however, so we can make these ourselves. Either way, I think we can easily manufacture these.

Where can I look to find the right pulleys/sprockets to get such large reductions? Are there any guides on designing gearboxes primarily with belts? I don’t want to try something without having researched the proper way to design it.

This is actually a question we are still researching a bit… we have some options, but do you have any recommendations for NC-based precision machining?

So, WCP’s CIM pulleys are 9mm-wide HTD 5mm, which may work fine for the first reduction (the torques on the motor-side are much lower). However, they are 12t, which makes it effectively impossible to keep 6 teeth in mesh if you’re going to run one belt between two motors and one wheel (HTD 5mm belts de-rate under 6 teeth in mesh), so it’s a bit iffy. Another option is to run 3mm GT2 for the first reduction (the 192 gearbox I linked earlier uses this, with 20t pulleys on the CIMs - you can source appropriate pulleys from SDP-SI).

One thing to be careful of is 9mm-wide HTD 5mm belts in the second reduction, for the same reason that you have to be careful with them in ordinary drive design (note that being careful does not necessarily mean “avoiding them,” in general). The rule of thumb that has served me well (admittedly, it is based on my experience with 6WD with direct-driven center wheels - the forces on the belt are likely somewhat higher in an octocanum, though this rule errs on the safe side so it’s probably still fine) is to make sure that, for 9mm belts, the pulley on the same shaft as the wheel has at least 8 teeth for every inch of wheel diameter. Any less, and you should go to 15mm belt. If you have any doubt, 15mm is always a safe option - but don’t be scared away from 9mm belts just because some people have made questionable design choices and had them snap.

Edit: Updated to be a bit more conservative in my advice, after checking the numbers more closely.

You can get a lot of what you need from Vex/WCP, but the place to go for everything is SDP/SI. Check out their belt calculator here. That calculator can list specific part links and has a ton of pulley and belt options. There are even custom options if you are willing to pay for them.
As for a guide, I’m not entirely sure. We moved almost exclusively to belts last year just from knowledge accumulated from CD. But we still don’t do anything too fancy. Here are a couple of basic threads that could get you started if you want:
https://www.chiefdelphi.com/forums/showthread.php?t=153996
https://www.chiefdelphi.com/forums/showthread.php?t=144718 (problem was low rigidity in shaft mount)
https://www.chiefdelphi.com/forums/showthread.php?t=149325
https://www.chiefdelphi.com/forums/showthread.php?t=152763
https://www.youtube.com/channel/UC98W6C_j1yR8SzDi1ws5Jyw/videos (WCP’s series on belts)

The rapid prototyping you are doing on the CNC will probably be enough to teach you a ton. Bump up that lexan thickness to 1/4" and stick it in an actual robot for testing.

This is actually a question we are still researching a bit… we have some options, but do you any recommendations for NC-based precision machining?
I’ll PM you. I don’t have any hot leads or anything, but I do know of some options.

The edit clarifies a question I was about to ask. The 20t pulleys look good (and I presume we’d use 6mm-wide GT2 belt for that first reduction). The only other issue I could see is being able to reduce as much as we need to in such a small size. WCP/Vex doesn’t have anything over 36T… I’m guessing we’d find larger stuff on SDP-SI? What is 192 using with their two 20t CIM pulleys?

I also presume it is ideal to support each pulley in the gearbox on both sides, like with gears. How do belt lengths work? Do you just cut them to a certain size, or is it something which you have to work out before setting up the pulley center distances?

You can find larger GT2 pulleys on WCP, too, though they may still not be big enough for a one-stage reduction from 20t pulleys with a 4’’ wheel, and currently appear to be on backorder. You can likely afford to go a bit smaller than 20t, as well.

Be aware that nothing on SDP-SI is going to have a hex bore, so you’ll have to broach it yourself if you want to use hex.

For HTD5mm, VexPro does have a 60t VersaPulley, but it’s too close to the diameter of the wheel for my liking.

Edit: It appears that the largest standard GT2 3mm pulley is 80 teeth. This puts you around the ratio you’re running currently. So, doing the first reduction with GT2 belts seems certainly possible, but you won’t have much headroom to increase the gear ratio if you find you need to.

I also presume it is ideal to support each pulley in the gearbox on both sides, like with gears. How do belt lengths work? Do you just cut them to a certain size, or is it something which you have to work out before setting up the pulley center distances?

Cantilevering the pulleys (or gears) ought to be fine, to be honest, so long as they’re right next to the plate.

Belts are available in fixed lengths, which means you have to fit your C-C distance to an existing belt. This isn’t really that hard to do, though.

I have shared the idea of using GT2 belts with the team. First words: “that’s what’s on our 3d printer. It’s microscopic, no way”

If running 9mm-wide HTD belts is considered risky for higher-load situations, wouldn’t running GT2 belts be problematic? Say we were using the same tooth setup as in these renders… wouldn’t there be a lot of torque exerted on the GT2 first reduction as well as the 9mm HTD second reduction?

Also for the first reduction, I’m assuming any quarter inch bore GT2 pulleys with set screws will work on the CIM outputs? Tooth count is a different story, but just wondering so I can check out the options.

In the 192 gearbox you linked, how would you determine the amount of teeth that would be meshing when running a belt around three pulleys altogether? Do the same tooth-count rules apply to GT2 (that you listed for HTD 9mm)?

I was actually inspired to make your module with belts (partially at least) using only VP stuff. You’ll need to play with the reductions a bit if you want the cims to face inwards on the module, but VP doesn’t go any less than 18T on their 5mm HTD Pulleys. Only option is to use GT2 belts instead of the 50:12 on the CIM output our to go for a larger than 50:12 reduction to squeeze the CIMS in. Also there are no shafts or standoffs shown but the deadhub colson actually acts as structure (parallel plate construction wooo) and I just didn’t feel like making a bunch of extraneous parts.

Disclaimer: despite having the resources to make this, we (5511) probably won’t do anything like this unless it’s COTS swerve just because of the simplicity and robustness of the 6WD or 8WD tank. YMMV But we like to keep the majority of bearings in our mechanisms :rolleyes:

http://i.imgur.com/OZypDJMl.png

Not compared to the second stage. Because the second stage is outputting a different speed and torque as the first stage, it is also experiencing a different torque. The second stage sees much more torque and less speed than the first stage does. Using a narrower (and thus “weaker”) belt in this application should be okay. You can double check the load ratings if you want to be sure though.

Also for the first reduction, I’m assuming any quarter inch bore GT2 pulleys with set screws will work on the CIM outputs? Tooth count is a different story, but just wondering so I can check out the options.

THe CIM output shaft is 0.315" (8mm) so you would need to enlarge the pulley bore if you want to use the pulleys that Oblarg linked. Personally, I would try to press fit the pulley instead of using a set screw. I think the set screw would work, though it may give you a little play. You would have to experiment. If you expand the pulley bore with a 5/16th drill bit (0.3125") you would get an interference fit that may work for you. I’m not familiar with what an appropriate interference fit would be for this application.

In the 192 gearbox you linked, how would you determine the amount of teeth that would be meshing when running a belt around three pulleys altogether? Do the same tooth-count rules apply to GT2 (that you listed for HTD 9mm)?

I think you would just count the number of teeth based on the number of degrees that the belt is wrapping around the pulley.

How do belt lengths work? Do you just cut them to a certain size, or is it something which you have to work out before setting up the pulley center distances?

If you check out the video link I posted, it goes into detail about how to choose the right belt lengths for a design. Basically, you pick a belt length (120 tooth, 121 tooth, etc.) and then set your distance in your model based on that length.