First attempt at an inverted custom gearbox with VexPro ball shifting internals. The design goal for this is a bare minimum number of non-COTS parts, in fact the entire shifting cluster is unmodified. In addition to the side plates, the only machined parts are three different Thunderhex shafts (which could honestly be left alone by tapping the ends and using pan head screws instead of E-clips or undersized bearings). Packaging everything was a chore, hence the 84t 1st stage with idler gears. “Real Life” speeds are 14.5 ft/s high (5.19:1) and 6.1 ft/s low (12.35:1, 48.5 Amps pushing current draw). I used hex bolts to secure the CIM motors so that the motors could be serviced without removing the rear idler support plate. Since the Versablock up front also uses 10-32s, I’ve switched it over to hex bolts as well for tool commonality at the expense of a little weight.
Questions, comments or criticisms welcome, with any luck we’ll get to build and test one well before kickoff.
It’s tough to get the ball shifter shaft to work in this application. I’ve tried to package something similar to this (using the shifter shaft “backwards” and with the shifting gears as the pinions for the next reduction), and I’m never fully satisfied with the end ratios and the size of the whole thing. Could you potentially reduce size by shrinking the second reduction gear and adding idlers to keep the CIMs spaced out?
9.7 lbs with motors and hardware, assuming I remembered to apply correct material properties to everything.
We used the 2-CIM flipped DS gearbox from WCP this year. It had 4 parallel plates as well and we never experienced any binding or misalignment issues.
I am slightly concerned about overconstraining the main drive shaft, but that’s only passing through three Thunderhex bearings which I heard were more forgiving than hex bearings.
It sure would be nice if you could get rid of those idler gears and that extra plate! This adapter from Vexpro might help, it will let you put slightly larger gears on the motor shafts without having to machine your own pinion gears. You would probably have to adjust the gears used on the output (wheel) shaft to get the speeds you want using this approach, but it saves having to make several parts.
Heh, I actually have some of those adapters squirreled away for that very reason. Alas, the issue with the flipped COTS ball shifter setup is that I couldn’t get enough 2nd stage reduction before the driven gears got uncomfortably close to the carpet on 4" colsons.
Quick thought - how risky would it be to cantilever the idler gears if I was able to find a way to fit a retaining ring on the inside of the 3rd rail? That would entirely eliminate the need for the final plate.
Yes, we ran the pocketed 60T and 64T gears this season. They were fantastic, and are still running great after 105 matches and several more hours of driver practice and demos.
We ran a 3-cim inverted dog shifting gearbox. We used the 8mm to 1/2 hex adapter to get the spacing on the first stage that allowed us to fit the dog gears in between the cims. All of the cim shafts and the first reduction were cantilevered off of the first plate and retained with shaft collars or retaining clips. We were able to make the entire gearbox with only two plates reducing complexity and weight. This made for a very compact design. We didn’t have any gearbox issues the entire season.
This thread doesn’t allow you to do attachments so I uploaded a pic of the gearbox to the gallery, it should show up soon
Added pocketed gears, reduced the rear plate thickness to 1/8", standardized to 10-32 screws. Brought the weight down to 8.9 lbs. I really want to find a way to double up on one of the middle plates but I also really want to avoid running chain-in-tube. The pneumatic cylinder might also hit the 4" wheel.
I posted a couple more pictures of how we were able to reduce the number of plates needed to 2 which will help with alignment issues. Also it reduces CNC machining time as well as helps with packaging the gearboxes into the chassis
Holy snack cakes Batman, that’s a lot of custom plates. How may shafts are constrained over 3 points? I see the output, is that it? (shiftier shaft doesn’t really count)
Go look at 1477’s 2013 robot if you can find pictures. Repackaging a ball-shifter shouldn’t use more than 3 plates, the trick is to recreate vexpro’s ball-shifter gearbox layout as in 2 84t spacings but reorient the shifter shaft in order to flip the cims outward. That way you can also use the vexpro published speed- ratio tables and be able to easily change your high and low speeds throughout season.
In terms of custom parts in the gearbox, you should be at 3 plates, however many standoffs you like, and 2 shafts.
Speaking of space reduction, is it insane to want to use a CIM adapter and a 3:1 Versaplanetary stage? Can those be used in drivetrain applications if it’s well within the load rating guide? That buys me a significant first stage reduction, a 1/2 hex shaft, and a diameter reduction for cleaner packaging near the ball shifting cluster.
I’ve never done this, but I don’t think this is unreasonable at all. As long as you can package the longer motor (may be tough with the flipped gearbox like this), that can probably save you some reduction space in the spur stages.
The trade off would be efficiency. Also, 1533 used 775 pros in versaplanetaries on a swerve and had poor results. Iirc the gears heated up and were in a bind. I would not make that trade off, but if you really need the space…
That’s a valid point, and I agree with you that it is not sufficient evidence to draw a conclusion. I just wanted to share that example. Perhaps a Triple Strange team member can shed more light on it.
The failure mode turned out to be simple to detect. The key was trying to keep the planetary gears and their shafts well lubricated. Basically, the planetary gears started to bind…at best loading the 775pro, at worst stalling the motor. That then led to burning out the motor. The problem continued to cascade if we still tried to move bot during the match (which we always did early on) with one locked wheel, the other 3 swerve modules overworked and reached 775pro stall zone taking out or weakening additional motor(s).
After cooling, the gearboxes loosened up, but never 100% free.
We addressed the issue by frequent tear-down, clean and re-lube or replace gearsets, and put in a new set of 775pros when a match failure occured…not an ideal way to run a whole season, but got us through world’s.
In off-season, we replaced 775Pros and 10:1 gear sets with mini-CIM’s (due to space and weight) and 5:1 (because 4:1 were sold out). It was slower, but well suited for drive teams under development…AND…it was rock solid with zero failures even in pushing matches. Gearbox input shaft speeds were less than half, and friction heat load in gearbox reduced.
Another interesting data point on the 775pros…we had motors fail and still free wheeling and others failed rotor locked (usually shattered or melted debris). A free wheeling failure actually did not impact our match because the other 3 swerve motors could drive the bot will little performance loss…but…with a locked-rotor we learned to simply try to limp to the airship and wait for climb. So, if you have a 775pro fail locked, and have it parallel coupled with another 775pro directly or indirectly, and keep trying to power thru the match, you are going to burn up more that one motor. We have the smoking bot and e-stopped match video to prove it.
Lesson learned?..if reliability under any condition is in…775pro is out.
Can’t wait to see the new motors this season, but may just stick with drama free CIMs for one year.
