Shaft Collars


#15

This thing looks cool. Which model are you using? Any opinion on how much it tears up the shaft?


#16

We have used shaft collars as part of a primary torsion member in our chassis (2016), and many other less-exciting applications, and did not have issues with them loosening.

A properly sized, assembled, and torqued one or two piece shaft collar should not come loose in nearly any FRC application. The CoF of dry aluminum to aluminum is somewhere between 1.0 and 1.4, so it would take 1.0 to 1.4x the fastener clamping force to start to move the collar (which is how loosening occurs). To put some numbers to this:

VP shaft collars appear to use #4-40 screws](https://www.vexrobotics.com/shaft-collars.html#Drawing). Assuming that the screw is torqued to a conservative 1/2 of proof load, it should be generating around 386/2=193lbf of clamping force. Combined with the CoF this equates to 193lbf to 270lbf of lateral load support capability. This should be more than enough to hold even a drivetrain wheel in place.

Note that the Al-Al CoF goes straight in the toilet (0.3) if grease, oil, or other lubricants are present. **It is important to have a clean, dry, assembly when putting these together! Otherwise your holding capacity is somewhere between ~20 and ~30% of what it should be! ** We love to clean parts with acetone or 90%+IPA (iso-propyl alcohol) before assembly.

If set-screws cannot be avoided, I use cone-point screws on a D-shaft or other flattened shaft.


#17

The Vex collars are not dry aluminum though they have been anodized. The other problem is that with a single clamping bolt, combine that with the fact that they are sized so they will slip easily over the shaft (in other words not properly sized) and the clamping area is very small. The ones from AndyMark on the other hand are sized such that you have to open them up for them to slide easily onto the shaft.

The Vex ones failed our team and I’ve seen it cause failures on other teams and those were cases where the load was quite small. So for me the ones we have in stock are only to be used as spacers going forward.


#18

I’ve had a lot of issues with the standard VexPro shaft collars coming loose, however their “High Strength Clamping Shaft Collars” have generally held up fairly well (we used one to attach our climbing winch cord this year and never had any issues with it coming loose).

Back in the day we used to machine our own shaft collars and used a set screw to press into the shaft itself rather than using a clamping hold. This method was rock solid (assuming you used locktite on the set screw), but damaged the shaft over time making it harder to slide things on and off. These are certainly easy enough to make yourself with a drill press and a tap (you can probably even modify the Vex or AM collars to attach this way).


#19

Almost every VEX shaft collar that has ever come off on either of my past teams have been due to this. Set screw gets stripped, student tries to make it work and doesn’t get it tight enough. Other than that we love using the low profile single screw ones and only opt out for the high strength ones in critical or high stress areas.


#20

We make custom drivetrain shafts every year. We simply buy eight foot lengths of 0.5" 7075 Hex Shaft from a local warehouse (Online Metals)

We use external E Clips (aka Side Mounted External Rings) on the interior of the robot. This allows for a shorter interior shaft that can be easily removed with needle nose pliers in a sometimes crowed robot interior. On the outside end we tap a ¾+ inch 10-24 thread. We than install (with Loctite) a 10-24 fastener and oversize washer for wheel retention.

For cutting E-clip groves we use a Nickole tool holder with a 1.5 mm (0.059”) grooving insert. A 1/2 “ E-Clip is specified for a 0.046” wide grove with a 0.396 inner diameter. You can either use a blade micrometer (or simply trail fits with an Eclip) to obtain the right depth of cut on the first shaft. Subsequent shafts are simply cut by adjusting the coss-slide depth to the same value as the first.

There are several cheap (<$20) 2.0 mm grooving tools out there using MGMN200 inserts on amazon that would work fine; and are “way better than a hacksaw”.

For tapping holes on the lathe, I use a Brown & Sharpe spring tensioned tap guide in the drill chuck.

With decent attention to detail, there is no issue machining all the axles to the correct CAD length +/- 0.001 inch.

The use of standard c-clips isn’t really worth the effort on hex shaft. C-clips (circlips) can only be fitted over a round shaft into their grove with snap ring pliers. I would only use them on an already round shaft. Making replacement hex gearbox output shafts is the one area where we have used them.

Shaft collars might be ok for some game specific mechanism, particularly for a shaft with no real side loading. I simply would never use one for any wheel retention on a drivetrain.


#21

CoF is a great point. Testing with parts on my desk, anodized 6061 to plan 6061 has a CoF of about 0.5. Certainly not as good as unanodized aluminum, by a big margin, but around 2x better than lubricated! I checked my CRC handbook on aluminum CoF and found vaules of lubricated interfaces between 0.2 and 0.3.

Can you please explain to me why a reduced clamping area or initial fit would affect clamping force? If anything the ‘digging in’ or elastic deformation of a smaller contact area would help. I don’t think the initial fit matters provided that the clamping force is resolved by passing through the shaft and not the gap of the shaft collar going to zero.

In 2017 we used rope guides made from round shaft collars that clamped around just the points of a larger hex shaft that supported a significant force when the rope piled up against them. Still, no issues.

Do you have any pictures or descriptions of how they were implemented, what loads they saw, and what you did to ensure that the bolts were properly tightened?

I ask all of these questions because I get really bummed out when I read about a team giving up on something that others have been quite successful with. I want to understand what the differences in implementation are that lead to success, or failure, because we can all learn from those details.


#22

Everything on our competition robot is new (Robo-Rio is an exception) - unless we are waiting for the part. Don’ ask abou GEMs…


#23

Personally, I’ve always had one or two thin shaft collars fail on me each year- the question is not “if”, but “when”. On our drivetrain, we had one single-screw Vex 1/2" hex shaft collar on each of the idler shafts between CIM pinion and driven gear. Effectively this was a 1-stage gearbox with idlers. Worked fine during the season, but one of them fell off during practice. It showed no signs of failure until it happened suddenly. Most other places on the robot used two-piece collars or screws and washers with no failures, except for a poorly-tightened screw falling off a couple times. Ideally, we’ll be switching to circlips next year. My guess is that either grease got into it, or it was stripped out during the install.
We used shaft collars in a remarkably similar way to make flanges for our 2017 climber, on a 1" axle. But 1" shaft collars get far more force than 1/2" hex ones do, with less stripping issues (#1 problem for us), and honestly not that much more force than people put on the 1/2" hex ones 99% of the time. Back-of-the-napkin calcs are looking like 1/4"-20 clamp screws put out a round 5x the force of a 4-40. The main problem with the Vex ones has been stripping the tiny screws, which is why I want to buy some Torx hardware this year to swap them out. Certainly we can try and fix our problems with Vex, thin, 1/2" hex shaft collars, but why bother when we can just use circlips?


#24

I don’t have any pictures available right now but where the Vex shaft collars failed us were in a similar situation as yours. Switching to the Ruland ones from FIRST Choice solved the problem. Note the Ruland ones we used are anodized as well and they were used on a Vex anodized hex shaft.

As I stated in my original post I’ve seen them fail for other teams as well. The most recent example I saw was a team that used them to hold the shafts for the wheels in their roller claw. Their side loading was 1/4 the weight of the cube. I first saw them have a failure at a practice field while we were using un-bag time. I suggested they make a change then and again when I inspected them at their first event. However they did not follow my recommendation. It then failed again in a match. Once they made the change to the Ruland units they did not have any more failures.

As far as the sizing goes, I can’t say I’ve done proper scientific evaluation but my gut feeling is there are two things going on. #1 a lot of the bolt’s clamping force is going to deform the shaft collar. #2 because of that the area where the collar is actually putting force on the shaft is very very small.

The tiny bolt doesn’t help things either because as others have mentioned it is easy to strip them out.

Ruland’s business is shaft collars and couplings and they have perfected them in the over 80 years they have been in that business. So I’ll stick with them when a shaft collar is needed. Ruland does have a thin line is size or space is a concern, again made using the things they learned in over 80 years of specializing in that business.


#25

Okay… that’s some good information about Ruland, but you didn’t really answer any of my questions about the circumstances of your usage nor did you explain why the different fit or smaller clamping area would be an issue.

Put differently - if we can understand why Ruland shaft collars work more effectively (in some cases that you’ve seen) then perhaps we can all make more educated decisions about what to use. How do you rationalize this one team having them fail under a few pounds of lateral load while other teams successfully use them on their drivetrain?


#26

The Ruland collars have bigger screws so they have more clamping force. The 2 piece collars allow you the distribute the clamping force more evenly.

On the Vex ones, we replace the socket head screw with a #4-40x3/8 Torx](https://www.mcmaster.com/#92610a106/=1dd6odx) head screw. They are better about not stripping the head. Once all the clearance is taken up in the split of the collar, you are not getting any more clamping force. I have seen that with the Vex collars.


#27

Those are two very good reasons. I would opine that the 4-40 screw can provide adequate clamping force, but that a larger screw can do so over a wider torque range and thus is generally more robust.

The gap ‘going solid’ is a serious issue. Dimensional capability of the part could easily explain why some teams have no issues and other teams have trouble.


#28

I’ve not heard of people successfully using the Vex collars in high load applications. I’ve seen them fail in lightly loaded conditions though.

The application on our robot that made me make the call that we will not use the Vex ones again was similar to yours, holding discs and the shaft for our climber last year. The rope would push the discs when the rope pilled up or the robot hung at an angle.

From looking at the picture you posted those are definitely not Vex collars.

Like I said I have no scientific proof of why the Vex ones don’t have enough clamping force, I’ve just seen them fail enough that they are off my list of parts that we will use.

Ruland makes shaft collars in house and that is basically their only business so they have perfected them. Vex has them made for them and it does not appear that they have put a lot of engineering and testing into them.


#29

I never said that they were.

FrankJ covered the two likely causes well I think. I’m going to leave it at that.


#30

I would like to point out that AndyMark sells a robust shaft collar.


#31

Over the years I’ve seen several stripped set screws on shaft collars, but one of the best ways to combat this issue is with education.

Teaching students how to properly tighten hardware is a challenge at first when their assumption is ‘tight’ means as far as physically possible. The biggest contributor to stripped set screws that I’ve seen is failure to explain that there is a logic behind selecting the right sized allen key for the job. Selecting the right size becomes much harder for smaller sizes and has led to many, many stripped shaft collars over the years when someone shoves an allen key in until it feels like it fits.

Teach your upperclassmen the importance of teaching less experienced students the value of using the right tools for the job - stripped hardware makes for a stressful 5 minute field timeout.

Will +1 a previous poster who recommended purchasing replacement hardware for collars.


#32

When I was designing things for ships, we a had a design policy of not using fasteners smaller that 3/8" unless absolutely necessary . Partially for corrosion and partially because that is the smallest size not easily stripped with hand tools.


#33

same rule for trains. Hard to do on robots.


#34

Having spent about a year at sea working, I have definitely experienced what Frank is talking about; nothing exposed outside, and little else, is smaller than about 3/8". However, I have to go with Bob for FRC purposes - there are few things on an FRC robot which require 3/8", and I’ve read CD posts from teams who avoid 1/4". An FRC-appropriate mitigating strategy (separate from training) for this might be to only have screwdriver-handled Allen drivers smaller than 5/32 easily available - no T-handles, no bits to be driven with ratchets or power tools.