A custom shifter-shaft PTO shifter. Uses a 3-position cylinder from Bimba. Low speed is 13.5:1, high is 5.88:1 and PTO is 22t sprocket at 5:1. This results in speeds of 16.5fps and 7.2fps with a 4.2" wheel. Weighs 2.16lbs without cims.
The low speed is a tad high, but I can’t get it lower without making a lot of other gears much larger. At the moment it is enough to get away with, and we can use an 11t pinion to bring it down to 6.6fps with a 4.2" wheel.
This uses a custom 1/2" diameter ballshifter shaft and 3/16" balls. However, by shifting on the upper shaft and using a steel shifter shaft, I’m confident that it would make it through a season.
The amount of lathe work this requires is inordinate; two moderately complex parts, one very complex part, and three basic parts (plus 4 spacers).
I will update the GrabCAD partner space with this model soon here: A Fond Farewell to GrabCAD Workbench - GrabCAD Blog
Really liking this design 
Because this is a 3-position piston, does that mean that when the PTO is engaged, the robot won’t be able to move around?
Also, what is the PTO capable of doing? I’m not sure if a 5:1 ratio might be enough to, say lift the robot’s weight off the ground.
Finally, where can I find the ballshifter PTO? For the dog PTO I did, I had to rip the hub off of a .step file from a wcp gearbox.
Yes, the robot can’t move when the PTO is engaged (like it should).
The PTO is going to a 22t sprocket. I would most likely route it to a 40+t sprocket to lift anything (like the robot) or route it to another sprocket and gearbox next to the mechanism. Given that you would be using 4 cims to lift things, a 12:1 gear reduction should be enough for a linear climber.
I had to design the shifter shaft myself. The idea is that the shifting is on the upper shaft, so whenever a gear is engaged the other gears are not. This also enables me to use only two shafts.
The Vex ballshifter shaft requires at least a 34t gear and uses a 3/4" shaft, plus only has two positions. Using a 3-position 1/2" steel shifter shaft enables the use of a tiny 20t ballshifter gear, and reduces the size of the gearbox considerably.
From the title at first I thought this was an ABS gearbox and I was concerned. Looks good though! Generic comment about supporting your CIMs marginally better goes here.
If a PTO left the drivetrain engaged, you would have no choice but to be driving your mechanism and the drivetrain at the same time. This is generally undesirable. Imagine trying to hang on a bar you are forced to drive away from, for example. The tradeoff of a PTO is that you can’t use your drivetrain while it’s engaged (plus the design hurdles).
Also, what is the PTO capable of doing? I’m not sure if a 5:1 ratio might be enough to, say lift the robot’s weight off the ground.
Nothing about a PTO requires the entire gear reduction to take place in the same gearbox as the drivetrain. Rarely is this the case.
This latest round of gearboxes seems to be missing a very critical (IMO) component for the implied WCD setup they’re for: the outer bearing block for the direct-drive wheel. Would adding that bearing block (such as the WCP one) cause issues? It simply requires adding two extra threaded holes around the existing output bearing hole.
I’ve never designed a PTO gearbox, but this one inspired me to think about it more… as I was thinking about it more, I couldn’t come up with how a gearbox with a single shifter shaft / output shaft could be used to get a PTO… After having thought about it a bit more, it seems to me that it requires having 2 shifter shafts OR having the shifting shaft not be the output shaft.
I had a look at your ball shifting shaft and the shaft itself looks fine. I found it interesting that you made the shaft wall 1/8 thick so the balls can’t fully engage, that might be a problem with wear depending on gear material and tolerances but it could very well be totally fine.
What I am very concerned about is bearing load. Its seems you’re using R1212 bearings and according to lily bearings (http://www.lily-bearing.com/ballbearings/r1212-bearing.htm) they only have a dynamic load rating of 111 newtons and static rating even lower. These numbers may be very conservative (or just wrong) but I would recommend doing some math and researching the ratings of the actual bearings you’d use to see if they’d work. If not, look into bushings or bigger/different bearings.
Interesting that you noted that. All of my other gearboxes have a better cim support, but I elected to forgo it in this case due to the lack of demonstrated need for one. A couple of my old gearboxes which were actually manufactured did not have cim supports and functioned ok, and I figure that will hold true for this as well.
I was planning on not using a bearing block on this one and instead directly pressing bearings into the 2x1 and using a double sprocket, however I will add mounting holes for the WCP bearing block just in case. Thanks for the idea!
I am using Boca SR1212-ZZ, which has a stated dynamic load of 94kg (207lb). I figured that would be enough for these, but I can swap them for bushings relatively easily if needed; that would also remove/reduce the need for spacer shims.
Thanks for pointing that out.
The balls should be able to fully engage with the grooves in the gears. I made sure that the balls wouldn’t be able to “get stuck” in the grooves, but not so much that it wouldn’t engage. What do you mean?
Nailed it! That was my thinking exactly; I’ve struggled with this problem in the past.
You can get around disengaging the drive by lifting the wheels off the ground before engaging the PTO(see 254 or 118in reveal video 2013).
Interesting. Personally I feel like that’s a little sketchy, as you need an extra mechanism to lift yourself to synchronize the two, but I guess if it works, it works.
It works 
In the case of both 254’s and 1114’s 2010 robots, the PTO did disengage the wheels. With 254 this is evident when you watch the robot extend its arm - it has to stop driving to unfold the hanging arm. I don’t know where you’re getting this information from. There are serious risks to trying to align something when the PTO is moving you away from your target. You almost never want your wheels engaged while your PTO is doing something else.
In 2013, if 254’s PTO didn’t disengage its wheels, that’s just because the wheels had already left the ground at that point, so the feature wasn’t necessary. This is unique to the 2013 hanging task.
This is correct. In 2010 we disengaged and in 2013 we did not. In 2013 the drive wheels were already off the ground so we had no need for the three position cylinder. When we were hanging, the drive wheels were also spinning back and forth. No big deal.
Be careful with the three position cylinders though. They do not behave exactly the same as two 2-position cylinders stacked back to back. See this post here for details: http://www.chiefdelphi.com/forums/showthread.php?p=1276175#post1276175
Thank you, that’s very helpful. So the holding force in the middle position is quite low then.
In that case, I might just switch to two stacked cylinders.
Yes, it is. It is worth considering. We got away with it in 2010, but it was a little sketchy. Sometimes it wouldn’t shift. We ended up having to shift to the other gear before shifting back to neutral. Not ideal.