Accurate, reproducible coupling

[itsmike]

I'm designing a testing device that requires an accurate and reproducible quick-release shaft coupling. Basically I'm looking for a way to remove and reconnect two shafts such that the position and orientation are the same every time. "Quick" is all relative; I just don't want any loose parts (screws, pins, etc.) when they are disconnected.

My best thought right now is a tapered hole with a flat, and its corresponding male end. Here's a model:

coupler_connected.jpgcoupler_disconnected.jpgcoupler1.jpg

The taper ensures concentricity while the flat orients the rotation. The two parts are held together by a captured nut (shown in gold), similar to how a garden hose is connected. For a sense of scale, the shaft with the external threads is 3/8" diameter.

First of all, I can't quite imagine how to make the female hole (maybe a wire EDM?). And while it works in my head, can anyone see where this design may fall apart?

I also haven't been able to find any commercially available solution for this...could anyone point me in the right direction?

Thanks in advance!

[RWOLFEJR]

Hi mike and welcome to the forum.

How tight does this connection need to be? How concentric? How are you going to tighten it? Pin wrench or maybe hand tighten and a knurl? Basically... What is the thing going to do? If it doesn't necessarily need to be super rigid and it's main purpose is to drive or just align, then I'd suggest for quick and slick connection... Go to your nearest John Deere dealership and ask to look at a quick disconnect drive for one of their mower decks. Or do a search... Couplings.

Far as the shaft and bore goes you could machine pretty much any shape in there you'd like to locate and drive but you should give consideration to the different costs associated with each. Wire edm isn't the cheapest way to go for sure.

Good luck,
Bob

[itsmike]

Thanks Bob!

The nut will be knurled and tightened by hand. The other end of the gray part is a probe that will be in contact with the objects to be tested. It is about 3-5 inches long and may experience about 5lbs of force at the tip. So I'd say it has to be fairly rigid.

As for concentricity and rotation, I don't have specs yet, so I'm going with "as close as possible" for the time being. This is being mounted at the end of a 3D scanning arm, so accuracy is pretty critical.

I re-read my post and probably should have described the end as a tapered D-shape profile.

I realize Wire EDM isn't the most economical way, but I can't imagine any other way to create this profile.

[jboggs]

Mike,
Your answers are going to depend on how you need this joint to perform. What loads/moments will it see? At what speeds? What accuracies are required? How often will the joint be dissassembled and re-assembled again? True, tapers are used all the time in precision machines, but I just recently changed out a 25 year old design on equipment in our plant from a taper joint to a piloted flange because of excessive vibration at high speeds. The cause of the vibration was that the design of the tapered joint was such that "bubba" could easily assemble it with a very slight misalignment, and he often did.

I am assuming that the flat is there because you need some alignment, which implies to me that the joint will be transmitting some torque. How much? How critical is the angular alignment? For what its worth - I DETEST trying to transmit torque thru a flat. It is a recipe for failure in my opinion.

Machining the taper with a flat in it will be tough. Would something with a key like this be feasible?
tapered shaft.JPG

All things considered I would still lean toward one feature for concentricity (a pilot diameter), and a separate feature for angular alignment and torque transmission (like a key or slot).

[itsmike]

jboggs - This is going on the end of a 3D scanning arm, to connect the arm with a variety of tips. So the joint won't be transmitting torques, per se, but it needs to react the torques applied at the other end of the gray part (I'll call it the 'extension'). The extension may have an angled tip off of its end. At the worst case, I would expect the joint to need to react 10 in-lbs of torque, which I think is reasonable for a flat.

I've considered the keyed design you attached, but I think there will be too much slop between the key and slot. Perhaps a tapered key and slot? I'd also prefer that there aren't any loose parts when the two parts are disconnected, so I'd want the key to be connected somehow or integrated in the female end. We have a variety of operators scanning for us, and small loose parts are just asking for trouble.

Maybe a captured tapered pin that screws down where that slot is and interfaces with a divot on the male end?

Thanks for all of the replies so far!

[PinkertonD]

Hi Mike,

I am with Bob, the flat is not a good suggestion for alignment. It would be far too easy for it to be locked up with a tiny degree of rotational error. You need something with a definite "lump" for location.

I would be opting for something with two devices for location. I would have one for the rotational positioning and the other for the longitudinal alignment. Maybe a quarter-turn device.

Something like this where the slot locates it radially, then the rotational locking pulls it onto the taper for alignment.

[itsmike]

PinkertonD - do you know the brand/model of the chuck you attached? That looks promising, though fabricating that isn't trivial either. I imagine you'd need a 4-axis CNC mill.

[PinkertonD]

Sorry Mike, it was just a stock photo for a search on "quick release endmill holder," or something along those lines. Can't remember exactly. Do some research on quick release tooling as there are some tap holders that also may be useable.

That should be pretty easy to produce on conventional equipment, depending on how many you want to make. The taper is straightforward lathe stuff and the helical locking part of the slot need not be all that prefect as it is just pulling the chuck up on to the taper and that will lock it in the same position each time. Could easily be done with an X-Y move. The radial locating key and the taper will define where it will settle each time.

Once you "zeroed" each head, they should be fine from there. Justa guess, though.

[itsmike]

Hey guys, thanks for the your responses so far. The requirements for the joint has changed slightly - I no longer need to keep the same rotation since we have a way to set a zero on the 3D scanner.

Anyways, I've got another design I want to get a reality check on. If you look at the attached screenshots, the black piece (the 'extension') has a 4-40 threaded stud as well as a 1/4-28 left-handed external thread. The red piece has a female 4-40 threaded hole, and the gold piece (a captured nut) has a female 1/4-28 left-handed internal thread.

The idea is that once both right- and left-handed threads are screwed together, any torque applied on the extension (black part) cannot cause it to unscrew, since the part will jam itself on the captured nut.

Does this make sense? Is it overly complex? Somebody smack me if I'm way off base =)

I've also considered a key, but I'm not very familiar with how to make them so I've shyed away from them. If there's a much simpler design, I will most definitely consider it.

Thanks!

assembly1_annotated.jpg
assembly2.jpg
assembly3.jpg
assembly4.jpg

[PinkertonD]

Couple of points, if you use that approach, I would make the outer thread right handed as it is more natural to use.

4-40sounds very flimsy and easy to snap off to me for something that is to be manually changed frequently.

Relying on just the threads for accurate repeatable location is optimistic at best. You need some form of positive alignment such as a 60-deg taper or something.

Finally, you appear to be over-thinking this. Go visit some CMM manufacturers and see what they use for swapping out heads. It has all been done before.

[jboggs]

Ditto.
What he said.