For my past several projects where I needed some sort of light-duty torque transfer, I would use a 3d printed clover coupling like the one shown below from a pump project I did a while ago.

While this coupler works ok, it doesn’t provide for any misalignment between the input and output axes.

I decided I wanted to see if I could make a u-joint inspired coupler in approximately the same form factor as the previous coupler. Primarily, I want it to fit through the same 30mm bearings, since I have quite a few of them on the shelf.

Since I’m basing this design on a u-joint (or universal joint), let’s take a quick look at what they are and how they work. The below image shows an example of a u-joint. The input coupler and output coupler have pinned connections to the central hub.

Those pinned connections allow the couplers to rotate around that pin, relative to the hub. Because the couplers are 90 degrees offset about the rotational axes. If we consider the Output Coupler as fixed, and we call the axis of rotation through it at the Z-axis, then these two pivots allow the Input Coupler to rotate about X and Y while still smoothly transmitting torque about Z.

The rotational degrees of freedom allow the coupler to accommodate a relatively large range of angular misalignment between the input and output.

But I want my coupler to also allow for radial misalignment. U-joints usually handle this by pairing the joints up. The two angular degrees of freedom, combined with the length of the shaft between them, can allow for significant radial misalignment.

One of the main things I wanted to try out for my coupler was to try to get this radial misalignment in a single joint. My thought for how to achieve this was to allow the hub to ‘float’. In typical u-joints, the pin joints between the hub and coupler are locked in all other degrees of freedom, only able to rotate about the pin. But my plan is to free up one translational degree of freedom. If I allow for translation along the rotation axis, I figured that would give me what I’m looking for.

I also want the couplers to be made up of only common parts I usually have on hand (dowels, bearing balls, etc.) and 3d printed parts.

Below is an image from the first test of this concept. Instead of capturing the spokes of the hub within rigid bearings, the hub has four steel dowels radiating outward from the center. Each dowel sits in a 3d printed V-groove that also has a cylindrical profile. A cylinder crossing over another cylinder makes a point contact. So that gives me four points of contact between each coupler and the hub. These four points of contact restrict four degrees of freedom…the two that aren’t restricted, happen to be exactly the two we want!

To sanity check the design, I started out with something I could just hold and turn by hand. I figured that would give me a feel for any binding that occurs due to misalignment. So I designed the little assembly shown below.

I printed all of the plastic bits from Polymaker PETG, and sized the holes in the hub for a light press fit of the dowel pins. For the dowels, I used four of these 5mm x 15mm stainless steel dowels.

And it felt good! I was actually pleasantly surprised. Granted, this was far from well-controlled or quantitative, but it seemed promising!

So next I wanted to take a stab at packaging it in a way that would match a potential real use case. I also decided it was time to set some actual design requirements.