Of to note, you cannot assemble the last reduction stage you have on your gearbox. You should look into increasing the cutout size in the extrusion to facilitate the assembly of the last driven gear.
So if the piston is retracted, the left gear would be engaged, but the entire shaft will spin? If the right gear is engaged, the left will not because it is to the left?
What the shifter is doing is changing which gear is connected to the shaft and which one is dead axled. When the piston shifts to the left, the left gear is powered by the motor pinions sending power to the shaft. When the right gear is engaged, the motor pinion drives the left gear (but it isn’t engaged, it’s just dead axled over the shaft), which drives another auxiliary gear which ends up driving the right gear. Since the right gear is engaged, the shaft spins.
Both gears on the main shaft are always spinning, the difference is which gear transmits torque to the shaft instead of spinning freely.
We ran a belt-in-tube drivetrain in 2018 and 2019, but with a 2x2" tube. I would suggest putting lots of access holes near the wheels so you can poke around to line things up for assembly.
I suspect the 3d printed pulleys will not hold up unless you have a really lightweight robot and drive carefully. We used vexpro pulleys. For the center wheel, we took two pulleys and faced off one flange on each to get them to fit in the tube. Using aluminum pulley stock to make your own pulley would be better.
The belts in this drivetrain configuration see the same maximum load as belts in the usual WCD configuration. I wouldn’t be concerned about it.
Nice looking design!
Some advice specific to belt in tube:
The biggest difficulty with belt in tube is assembly and maintenance. You will want access holes on the top of the tube above the pulleys. Even still, getting the shafts to line up with the pulleys and bearings requires some finesse. We are planning to move to Thunderhex which will hopefully help. You can check out the CAD for our 2019 drivetrain here. Our access holes have tabs that we bend up slightly to provide a smooth edge for the belt.
Once you have the belts inside the tubes, your entire team needs to be disciplined about keeping debris out. If metal shavings, rivet heads, or other things get in, your belt will get shredded. Make some polycarb covers for the access holes. We let our guard down this year, and a shaft collar ended up inside which shredded a belt. Fortunately this happened during practice at our shop.
I don’t have any experience with 3D-printed drive pulleys, so I can’t say how well they would hold up. The failure mode would likely be rounding of the hex bore. 30 tooth pulleys will be fine. We use 24 tooth, but have not tried it with anything larger that 6 inch wheels. We buy aluminum pulley stock from BB Manufacturing, which we machine and hex broach. You can find our pulley CAD here.
Is this an “issue”?
You could look at the load ratings for the belting and see if anything is under-rated. Heck you might be able to change the further belt to a smaller width to save space/weight.
If you are concerned about the fact that you’ve doubled your failure points (i.e. if you snap the rear belt, both front wheels are out of comission) you could run two lengths of belt- one from rear to center, one from rear to front.
I am always super worried when teams cut that much out of their frame rails. That aluminum was there for a reason! But if you got away with it, maybe I’m being overly conservative.
We only do it on one side, so it doesn’t really affect structural integrity much.
I feel like you’d be better served by putting the cutouts/covers on the underside. Servicing would require a little upskirt, but then the topside has full attachment potential. little things though.
For a chain version I would want to use 25 or 25h chain. What recommendations are there for 25 vs 25h chain? If a chain did break how tough would it be to change with access holes at sprocket locations.
The 254 style double belt/chain run is what I was getting at. I’ve seen regulalrly loaded belts snap in drivetrains before without any other obvious cause for the damage so I think double loading the belt could certainly exacerbate stress on the belt to the point of failure within the timespan of an FRC season more regularly than single loading the belt would.
What you said about doubling points of failure is another issue that I forgot to include in my earlier post.
We ran #25 chain for the past 3 years without encountering any problems.
If you happen to snap a chain you will need to use the chain tool to repair but they will not fit through an access holes
I believe master links are also an option but I have 0 experience with it
IIRC when 971 ran belts they used large pulleys, wide belts, and center wheel gearboxes for precisely these reasons.
You have to download and install it to a pc, and make some assumptions about torque, but Gates had their belt and pulley ratings in their Design IQ software: https://www.gates.com/us/en/knowledge-center/engineering-applications/design-IQ-software.html
IIRC, you need something like the kitbot size pulleys to be really within the Gates recommendations. (42 tooth, 15mm wide, 5mm pitch HTD belt). The only way you can get smaller is with 8mm pitch belt, which is harder to come by.
I’m getting the feeling that a skid steer style drive that is not the KOP should just use chain.
I am working on a chain version and will update the post when done. I am having a few issues with packaging of the last stage my problem is the vex 16t sprockets have a 1/4" thick protrusion on one side. There is a location where an 1/8 smaller would make it easier but I would rather not have to machine cots parts.
What’s worth noting is that with a decent setup (which might just be a hex shaft with a few clever modifications), you could face off the extra space from those sprockets pretty quickly, using either a mill or a lathe. Since the face doesn’t need to be perfectly smooth or precise, you could probably manage a really quick turnaround on them and have a drivetrain’s worth done in a jiffy.