In the past, my team has tried a variety of ways to get shafted sensors to stick into the ends of our drive and mechanism shafts, but every once in a while we run into a situation where our sensors are slipping, and we attribute that to our methods of attachment.
We’ve tried regular press fits and interference fits, but in high load, high acceleration situations we still see some slippage. We get the same results when cutting slits into the shaft and using a shaft collar to clamp down on the sensor shaft, though we want to try playing around with that a bit more. While we haven’t personally tried it, some of the teams we look up to have told us that they’ve tried using set screws and didn’t get the best results.
I’m curious how everyone else deals with keeping sensors locked onto the shafts they’re intended to be in? We want to avoid using glue or loctite so that we can easily remove the sensor if needed, but have had to resort to them in the past due to lack of options.
EDIT: I should clarify that we never have these problems with mag encoders, mostly potentiometers
Not exactly the answer you are asking for but my team uses the SRX Mag Encoders, so there is no physical connection between the shaft and the encoder part. We drill a small, press fit hole in the end of the shaft and using machine blocks (basically just very precise little plates), we set the encoder and use an adhesive to ensure the encoder holds fit. We just scrap the magnet and shaft each year.
Disclaimer: I have never attempted these myself nor been on a team who has.
From what I understand, there are two popular methods to prevent this sort of slippage for encoder shafts.
One method is to use a rotary encoder like the GreyHill 63 R and then fix it to the shaft via a flexible material such a surgical tubing. This method has the unique advantage that the encoder doesn’t need to be concentric with the shaft it’s connected to because of the flexible nature of the connecting material.
Another method is to use a magnetic encoder like the SRX Mag Encoder, whereby a small magnet is inserted into the end of the shaft, and the encoder counts the rotations of the magnetic poles. In this case there is no physical connection between the shaft and the encoder, but the encoder still needs to be located concentric to the shaft within fractions of an inch.
Anyone who is more knowledgeable, please correct me if I’ve said anything wrong.
Are you sure the slipping is in the mechanical connection between the sensor and the shaft, and not the mechanical connection between your wheel and the floor?
How constrained is your sensor? If it’s very little (a zip tie on the cable only) you may see some error, if it’s too much (rigidly mounting to a solid part) you can also see slippage.
Press fits should work, and flex collar clamps should work even better than press fits.
I’ve done the split shaft with collar thing you’ve mentioned successfully a bunch of times. Applying a bit of thread lock to the mating surface between the sensor shaft and the split shaft bore can help prevent slip if your bore is a little looser than you’d like.
Also a fan of the SRX magnetic encoder; not only doesn’t it slip, you can decide on the fly (through reprogramming if it’s connected to an SRX) whether to use it as a relative or absolute encoder. Bonus: Most of AndyMark’s gearbox output shafts already have the magnet installed.
Have also used the rubber tube coupling to good effect, as well as a 1/4" clamping shaft couplerto attach a potentiometer to a shaft.
Glad to know all of these alternatives have potential issues. We have only tried glue and it works. We can usually retrieve magnets from glue in axle, but if this is a pain; we replace the axle and stock these. Note this works well if you have simple axles, which we do.
Mags don’t have this problem, though we commonly like to pair mags with pots and the pots are what give us issues with slipping.
The issue with this method is that the flexibility of the surgical tubing creates significant backlash within the system, which for some of our sensors makes the data impossible to use.
The mechanisms that give us these issues are more arms and lifts than drivetrains. The one in question that prompted me to post this thread is our single jointed intake arm for our offseason bot. When we accelerate quickly we find the pot slips. The sensor is rigidly mounted to the frame of the robot via a 3D printed mount that locks the sensors pretty well into place.
Definitely a potential solution - hex encoders are going to change the game for FRC. Unfortunately some sensors won’t have that kind of functionality, which is what I’m having trouble with.
I’ve used this this too, and it works in certain applications. In something like an arm or elevator that doesn’t have really high accelerations or switch directions quickly, it is a decent solution. For something like a drivetrain that accelerates and reverses directions a lot, you’ll be introducing some lag into the system. Depending on what kind of controls you’re using, that may or may not be a problem.
The real solution is to use a misalignment coupler from a company like Ruland. These are designed to connect two shafts without being bothered by slight misalignments. I’ve used some that we’re donated before and and they work very well. It might be too fancy of a solution for FRC purposes in general though.
We’ve used 63Rs regularly for our drivetrain for several seasons, and have used potentiometers for a variety of arm and elevator joints. We’ve varied our attachment methods depending on the task at hand.
Slip in our drivetrains has been occassionally observed via maintenance, but never a huge concern for ultimate reliability of the drive encoders. We’ve used a variety of methods for this, but using a 1/4-20" insert into the shaft and a 1/4" shaft coupler is our preferred method when space is available to support such a solution.
For potentiometers, glue is our go-to solution for when slipping in a shaft is an issue. Not the answer the OP wanted to hear, but it works for us. We’ve also been able to unseat the shafts from the glue, when required.
For arms and lifts we typically also have redundant sensors (limit switches) on the end of travel. These can also serve to help “zero” the potentiometer in any situation where sliping and/or backlash concerns have begun to stack-up.
That might be the issue? You want a very secure connection to the potentiometer shaft, but you do not want to overconstrain the sensor with a firm and rigid mount. You ideally want something that prevents the sensor from rotating at all but does allow it to float a little (e.g. a lexan Z bracket).
Echoing this. Our 3D printed encoder mounts frequently have visible mounts sometimes visibly flex during shaft rotations. While this is less-than-ideal, I’d rather the mount take up that stress than the sensor or the connection between sensor and shaft.
Interesting, haven’t thought out it that way. Is this because of possible misalignment adding resistance to rotation on the shaft? I’d be afraid of a little lexan bracket warping with load.
Try to design the mounting bracket so that it only requires elastic deformation rather than plastic to handle the misalignments. Use the properties of the material to your advantage.
As soon as your encoder shaft and mechanism shaft are mounted just off of concentric at all, they will be fighting and will want to shift to be concentric. If the sensor is very rigidly mounted (assuming here the mechanism shaft is already rigid) then the encoder shaft will be more likely to slip in your press fit or set screw setup.
This is why in an industrial setting flexible couplings would 100% be the go-to solution here. It really is the best way, as it allows the sensor to be rigidly mounted while still allowing for concentric misalignment. That said, these generally tend to be more expensive then most teams would like to spend for an FRC sensor application.
The lexan solution Chris suggested is a good one. Allows the encoder to flex the small amount it needs to to stay concentric, but still still keep the encoder in place. Essentially the lexan plate the sensor is mounted to gives you the flexibility the coupler normally would.
I’ve never been a fan of the surgical tubing method for the reasons you pointed out (backlash).
I’ve become a big fan of using the mag encoder whenever possible, or the VEX VP integrated encoder, or other similar options when possible. But yeah, there will be times when you just can’t make these work for what you need. I don’t love putting the encoder directly into the shaft because it often times makes any disassembly harder. The flexible couplers just work, and are designed for this exact application.
With encoders it’s generally fine, what I’m worried about is a potentiometer moving a bit and throwing the position off. We do use mags wherever we can, though on certain mechanisms we prefer to have encoders and pots, in which case we still need to deal with a shaft.
Yup, I’ve referenced that image more times than I can count. Works great when a little bit of sensor wiggle is okay.