During playoffs at the FIRST NC Guilford district competition, we somehow managed to shatter the 1/2" hex shaft female coupler insert connected to the output of the Versaplanetary gearbox that tilts our climber arm. The coupler supports one end of a ~5" long hex shaft (which is supported on the other end by a hex bearing). That shaft carries two sprockets that drive the (larger) arm tilt sprockets in turn. The arm is on a separate shaft that carries the robot weight, this short shaft just drives the tilt sprockets.
The gearbox is driven by a NEO with a 100:1 reduction, with a very light current limit (4A).
The coupler split into three pieces - two halves of the hex socket, and the spline end. It did not split at the setscrew. I don’t have it - one of our other mentors does - but I’ll post pics if I can.
Is this a fluke, or did we mess up by relying on the coupler to carry some of the strain on this shaft? When our robot climbs the arm tries to backdrive the shaft & gearbox as the robot swings. I was under the impression that the low current limit would limit the torque… but maybe when backdriven that’s not the case…
We could change the design to have hex bearings on each end, but I was concerned about the potential for misalignment between the bearing and the shaft coupler. I’m also not convinced it would have kept this from happening.
If it’s a fluke, fine, we have spare couplers, and it’s possible this one could have been damaged from previous use.
Any advice would be welcome
Thanks
Ben Edwards, 6500 GearCats mentor
I have seen similar failures in the past - once on a climber with what sounds like a similar setup, and once on a shooter. Ever since, when we use them I make sure we minimize or eliminate any side loading on the coupler - I suspect that is what causes the failure.
We have never used a Vex universal coupler for anything other than low load application like a shooter. I’m pretty sure it’s a cast piece that is not really made for any kind of shock load.
A NEO through a 100:1 gear box has enough power to twist a hex shaft like a pretzel (don’t ask me how I know) even at a low amperage.
Using the (updated) JVN calculator and a current limit of 4A I only get ~6 ft-lbs of torque at the gearbox output. Surely that’s not enough to do any damage to the coupler or hex shaft.
We could switch the motor out for a NEO 550 or something similar, but we didn’t really want the arm to tilt super fast and our programming team likes the built-in encoder.
I can see side load maybe being a problem, and adding a hex bearing on the shaft end near the gearbox is something we can do if that’s really the problem. If it’s torque, I want to understand how the current limit isn’t keeping that from being a problem.
I’m still struggling to understand the current limit I guess.
DC motor torque is proportional to current. If I have a NEO with full stall torque of 2.478 ft-lbs at 166A stall current, and I limit that NEO to 4A using the smart current limit feature of the Spark Max, it should develop no more than 4/166 * 2.478 = 0.0597 ft-lbs. Multiplied by 100 for the gearbox reduction gives 5.97 ft-lbs.
To me that means that the output torque of the gearbox will not exceed 6 ft-lbs.
If the arm is tilting forward under load, the motor + gearbox puts 6 ft-lbs on the shaft maximum.
If the arm is pushed against a stop, the motor + gearbox puts 6 ft-lbs on the shaft.
These behaviors are how I understand the current limit works.
What I am not 100% sure of, is, if the arm is pushed backwards the motor & gearbox should spin backwards and resist the arm movement with 6 ft-lbs of torque. I’m not 100% sure this is actually how the current limit works. Can anyone verify this? When backdriven, is the torque/current limit still active?
If the current limit doesn’t protect the gearbox + motor against being backdriven, then that’s a critical flaw in how the current limit is recommended. I’ve read many posts about “using the current limit as a clutch.” Slipping clutches limit torque in both directions.
My backup plan is to build a mechanical friction clutch (spring pushing a friction disc against the sprocket), to limit the torque applied to the chain sprockets. I may do that anyway, it’ll be an interesting project.
Separately, I am going to reach out to VEX, to see if they have any torque specifications on the VP universal female output shaft kit. My clearly wrong assumption was that it was at least as strong as the hex shaft.
A 100:1 planetary gearbox does introduce a fair amount of friction which, in my experience, makes them pretty hard to backdrive.
Make sure your motors idle mode is also set to coast if you’re not relying on it to hold arm position with no commanded voltage, these brushless motors have pretty significant holding torque while in brake mode.
I think the slipping clutch analogy you’ve seen is mostly in reference to drivetrains, where current limits are used to avoid wheels breaking traction and slipping on the carpet. This is sort of the opposite of what you described, current limits mostly ensure that you stall the motor at a lower current to avoid either tripping a breaker or damaging the motor due to heat generation.
Don’t trust the current limit to protect anything except your motor
- build a torque limiter/slipping clutch, or use gearboxes that can take 1000s of ft-lbs
Don’t trust the VEX universal coupler to take any significant side load
- support the shaft at both ends with a bearing
Disappointing. We keep trying to make this system more robust and instead finding the next weak point.
Ours broken in the finals for our climb in 2020. It’s just bad design there is not enough material to sustain heavier load, we now use the hex output with a hex coupler.
I think that’s what I’ll do as well, rather than build a torque limiter, as that will be a much easier adaptation. We have 1/2" hex shaft outputs for the VP gearboxes, it’s a relatively easy swap.
It’s not always possible, but whenever you can avoid a cantilevered output shaft from a gearbox, it’s a good idea to avoid a cantilevered output shaft from a gearbox. Let some chonky plate take the side load, and let your gears spin with as little friction, stress, and strain as possible.
The design we have isn’t cantilevered - the shaft is supported on the other end. It’s just that we were using the coupler to support one end of the shaft.
I definitely agree that cantilevering a shaft is a bad idea, in general (WCD wheel shafts excepted).
The original design (this is V2) had us using the plastic VP single stage clamping gearbox, with a single sprocket on a 1/2" shaft output, turning the arm sprocket through a #25 chain.
But the VP clamping mount flexed so much under load that the chain skipped and eventually popped off while we were climbing. So we changed to the current design which has the shaft supported on both ends, with two #35 chains and two sprocket sets.
Putting a bearing on the coupler end, to me, fell afoul of the “3 bearings on a shaft” problem - you have the two hex bearings and then the bearing in the gearbox, something’s out of alignment unless it’s all drilled at once.
We decided to solve the issue by going with the suggestion from running_nightmares - use the 1/2" hex shaft output and a shaft coupler, then support the other end of the shaft. It seems much more robust. I guess the next thing we’ll break is the gearbox itself… oh well, it got us to district champs so if we make worlds that’ll be the next upgrade.
This can be viewed as the shaft only being supported on one side. Imagine the coupler and or the interface between the coupler and the two shafts flexing.
Using a compliant mount for the motor/gearbox that only resists torque would prevent over-constraining the shaft. A second solution would be to use a compliant shaft coupler to allow for some misalignment between the gearbox output shaft and the driven shaft (see below). For some low torque, low speed applications, we have used a piece of rubber tube with a pair of hose clamps. https://www.automationdirect.com/adc/shopping/catalog/motion_control/drive_couplings#sort=undefined%20asc&start=0
A third solution would be to use a rigid mount on the motor and gearbox but to match-drill the mount to the rest of the support structure after connecting the shaft coupler.
It would still be good to use a compliant shaft coupler to accommodate any flex in the structure between where it supports the motor/gearbox and where it supports the shaft. It doesn’t matter whether the shaft flexes under load or the support structure flexes under load, the resulting misalignment will stress your gearbox.