3D Printed Wheel Hub Drive – Part 2

Part 1 is here.

In this article I’ll talk about some improvements I made to the wheel hub drive from the last article and some other designs I have tried since then.

The planetary gear I was using used 5 planets. You can see in Figure 1 that the planets are so close together that they actually touch each other. This was causing extra friction so I tried 4 planets instead, shown in Figure 2. This seemed to work better.

Figure 1: Helical planetary gear with 5 planets
Figure 2: Helical planetary gear with 4 planets and sleeve bearings

In Figure 2 you can also see that I inserted sleeve bearings. I hoped that the sleeve bearings would help decrease the friction caused by the planets spinning around the bolts that attach them to the Y carriage.

I ended up redoing the rear motor mount, shown in Figure 3. Also shown are the 3D printed bearings I used that go around the motor.

Figure 3: New rear motor attachment

To test the assembly’s strength I tied a rope around it and attached the other end to the ball transfer skateboards I made, as shown in Figure 4.

Figure 4: High load test

I stepped on the skateboard and turned the motor on to see if it could pull me (I weigh about 160 pounds). The results were disappointing. It was not able to pull me when I stood on the skateboard with the 3D printed ball transfers. It was able to pull me when I stood on the skateboard that had 5 ball transfers I bought from McMaster, but it took a lot of current to do so (I believe around 30 amps).

The planetary gear seemed to not be causing much friction once I went down to 4 planets and put the sleeve bearings in. I thought that maybe the bolts that go through the planets were being pulled to one side when under a heavy load, causing the planets to tilt a little. But after taking a video of the planets under load it looked like they were not being tilted.

I believe the problem is mainly with the 3D printed bearings shown in Figure 3 and discussed in part 1. I don’t think they can handle the load I’m trying to put them under. Since there is only one ring of BBs, and the motor and hub spin at different rates, the BBs probably do more sliding than rolling, causing friction. I may have to redesign the bearings to have 2 rings of BBs, one for spinning with the motor and one for spinning with the hub. The plastic parts may also flex under high load, causing more friction.

Other Motor Drive Designs

I created a few more designs that I hoped would be able to handle heavy loads. These designs focused on minimizing friction. The first is shown in Figure 5.

Figure 5: Small wheel design

This design drives a smaller wheel. I wanted to see if the motor could provide the torque needed to rotate the large gear I slid around it. I cut out the wood pieces on my CNC. The wheel was printed out of TPU and attached to the gear via 1/8″ steel rods, as shown in Figure 6 and 7. The tire tread was copied from here.

Figure 6: Wheel, gear, and 1/8″ steel rods unassembled

Figure 7: Wheel, gear, and 1/8″ steel rods assembled

I performed the same kind of load test as before, shown in Figure 8.

Figure 8: Small wheel motor drive being set up for testing

Disappointingly, the this motor drive was not able to pull me either, even when I used the skateboard that used the ball transfers I bought from McMaster.

The last design I tried is shown in Figure 9.

Figure 9: Gear train motor drive

This design used a simple gear train to get a gearing ratio of 4:1 between the motor and the longboard wheel. The wheel was attached to the gear using 1/8″ steel rods like the last design, as shown in Figures 10 and 11.

Figure 10: Wheel, gear, 1/8″ steel rods, and other parts unassembled

Figure 11: Wheel and gear assembled using 1/8″ steel rods

This design was tested like the others, as shown in Figure 12.

Figure 12: Testing the gear train design

This design actually worked! It was able to pull me when I put all my weight on both of the skateboards shown in Figure 12. The current it took to pull me is shown in Figures 13 and 14.

Figure 13: Current draw when the gear train design pulled my weight while I stood on the skateboard using my 3D printed ball transfers

Figure 13: Current draw when the gear train design pulled my weight while I stood on the skateboard using the ball transfers I bought from McMaster

You can see that the ball transfers from McMaster used less current. It takes more force to start pulling a load from a dead stop than to keep it going, so I believe that’s why the current is highest in the beginning (and the startup boost may be too high).


I’m going to move forward with the gear train design. I’ll improve the design in several ways. For example, I’ll try to redesign it so that the middle gear is supported on both sides and I’ll probably make the gears thicker.

3D Printed Wheel Hub Drive

Figure 1: 3D models of hub motor

Part 2 is here.

This week I worked on making a 3D printing a geared wheel hub drive for a Keda 190KV brushless motor. I designed it in Designspark Mechanical.

Figure 2: 3 bearings going around the motor and attachment for the rear shaft

First I designed bearings that go around the motor. The motor is an outrunner, so the outside spins when the motor is powered on. 3 of them are pushed onto the motor, as shown above. I also 3D printed 2 spacers to keep the bearings from moving. Also shown below is the attachment I made that screws into the rear of the motor. A 5/16″ square rod is inserted into the attachment.

Figure 3: Assembling the bearings

Shown above is one of bearings. The bearing is made with 4.5mm BBs. The cap (on the right) is glued in. I used Loctite Super Glue Gel Control and Gorrilla Super Glue Gel Control. Both seemed to work very well.

Figure 4: Helical planetary gear

The other end of the square shaft is inserted into the sun gear of a planetary gear, shown above. The gear was modified from this design by emmett on Thingiverse. With his design, you normally print the ring, planets, and sun gears all at once. The parts will be slightly fused together, so you need to break them apart by forcing the sun gear to turn. Instead, I broke the ring gear into 2 pieces so that I could print each part separately and then assemble them.

Figure 5: Y carriage

Shown above is the Y carriage that attaches to the planetary gear. 5 bolts are going through it. The bolts are what will attach the Y carriage to the planet gears.

Figure 6: Y carriage and motor

The motor and bearings go inside of the Y carriage (or hub I guess you’d call it?), as shown below.

Figure 7: Spacers for the bolts that attach the Y carriage to the planet gears

I printed some spacers that go on the bolts, shown above.

Figure 8: Y carriage attached to planetary gear via 5/16″ bolts
Figure 9: Nylon nuts are used to secure Y carriage to planetary gear

Shown above the Y carriage is attached to the planetary gear. I used nylon nuts to secure it. The nuts are screwed down just enough to prevent the bolts from wiggling around as the gear spins, but not enough to cause extra friction.

Figure 10: Motor inserted into Y carriage hub

Shown above, the motor and bearings shown in Figure 1 are inserted into the Y carriage hub.

Figure 11: Cap that screws into the other end of the Y carriage hub

Next, I made a cap that fits over the other end of the Y carriage hub to keep the bearings from coming out.

Figure 12: Mount that front of the motor attaches to

Shown above, I mounted a piece to the front of motor.

Figure 13: Cap that fits over the motor shaft

Finally, I made a cap that fits over the shaft of the motor.

Figure 14: First test of the hub motor

Shown above is my first test of the wheel hub drive. You can see when I zoom in on the planetary gear that the sun gear is spinning faster (4 times faster) than the planet gears and the Y carriage. This is the gearing that I wanted to achieve.

Figure 15: Tire that fits around the Y carriage hub

Since I didn’t need it for testing, I didn’t print the tire that will fit around the Y carriage hub yet, shown above in black.


There are several improvements that I’d like to make to the wheel hub drive. First, the planetary gear was printed with too small of a tolerance. The ring, planet, and sun gears fit together too tightly making them hard to turn. The gears are generated using an OpenSCAD file and the tolerance is one of the parameters. Increasing the tolerance should solve this issue.

Second, I’m going to replace the spacers shown in Figure 7 with metal washers instead. I believe washers will have less friction and better handle the heat generated by the turning.

Third, I believe I only need 1 bearing, not 3 as shown in Figure 2. I would only leave 1 bearing on the end to support Y carriage hub. With only one bearing, there will be less friction. However, I’m worried that if I take the middle bearing out, the Y carriage hub will strain under the weight I plan to put it under. To help, I’m going to try to redesign the Y carriage hub so that I can reinforce it with at least 8 1/4″ steel rods.

Fourth, and finally, I’m going to see if I can make the bearings shown in Figure 2 without using glue. Even though the glue seemed to work very well, I’m worried that it will degrade over time. I tried to redesign them to use snap-fit joints instead, as shown below.

Figure 16: Redesign of bearing to use tabs

However, they didn’t seem to work too well. I may try to make the snap-fit joints work, but right now I’m thinking that I might be able to use zip ties instead.

I will make these improvements and then post the results in part 2. If I’m happy with the results, I will upload the 3D model files to Thingiverse. Stay tuned!

3D Printed Ball Transfer Units

Ball Transfers I 3D printed.

You can download the 3D models on Thingiverse.

You can buy ball transfers from McMaster-Carr for around $3, so a while ago, last year sometime, I bought some. The ones I bought were pretty noisy. They were so noisy that I couldn’t use them for what I was working on. At the time I thought I could try to make my own, but I didn’t want to spend the time to do so. But 3 weeks ago I decided to try to make my own.

Ball transfers have one large sphere that supported by a number of small ball bearings. For my designs, I bought the large spheres from McMaster-Carr. The small ball bearings I used are just 4.5mm BBs.

Designs I Ended Up Not Using

The designs in this section I ended up not using because other designs worked out better.

Design with pockets for each ball bearing.
Design with grooves that ball bearings would move along.
A design where the 1″ sphere rests on the spring.
Shown here are a couple of the ball transfers I bought from McMaster-Carr. I tried 3D printing a cover for them out of TPU to see if the cover would help with the noise, but it didn’t.

I tried some other variations of the designs above. I experimented with making the hole that the bearings and sphere sit in a little bigger, or a little smaller. For the caps that go over the sphere, I experimented with different sizes for the hole that the sphere goes through. I even tried adding grease to the ball bearings and making the hole in the cap small enough so that the grease wouldn’t come out, but the sphere could still role. I ended up not moving forward with that idea because the smaller hole caused too much friction when it rubbed against the sphere.

Parts that didn’t work out for one reason or another.

Designs That Worked Well

Design with no grooves or pockets. The ball bearings cover the entire bottom, or top, depending on if it’s mounted ball up or ball down.
A smaller design my dad suggested that uses several smaller spheres (0.5″) instead of one large one (1″).

As you can see, just filling the hole with ball bearings worked out the best. No grooves or pockets needed.


I ended up looking at several different materials to use for the ball bearings and spheres. I tried out steel, nylon, ptfe (teflon), and silicon nitride.

The white 1″ spheres are ptfe. The spheres off to the side are steel.
The white 0.5″ balls are nylon. The balls off to the right are steel.
The black 0.5″ ball is silicon nitride

I ended up not doing any testing with the silicon nitride balls . They seemed like they would work well, but they’re just too expensive for what I want to use these for. I bought 2 of them just because I was curious.


I wanted to see if the ball transfers I made could handle moving around my own weight. I found some scrap pieces of wood and screwed the ball transfers to them and basically made a skateboard with ball transfer units instead of wheels. I then found a large, flat piece of wood and laid it on the floor, put my skateboard on it, then tried to push myself along. I tried the different designs and different materials and narrowed them down just by how much friction I felt when trying to push and pull the skateboard and by how noisy they were.

Through how they felt, I narrowed down the designs and materials I liked. I found that the 1″ ptfe balls felt like they had less friction and where not as noisy as the 1″ steel balls. The 0.5″ nylon balls seemed to have about the same amount of friction as the 0.5″ steel balls, but the nylon ones were less noisy. I didn’t order any 0.5″ ptfe balls because they’re too expensive, but I imagine they would work better than both steel and nylon.

Eventually, I bought a luggage scale and did some testing.

Shown in the above image is the skateboard I made that has the 1″ sphere design I liked. The spheres I used i this test were made of ptfe. I stood on the skateboard while my wife pulled on the scale until I moved forward. I found that it took about 20 pounds to get me going. After I started moving, it took about 5 pounds to keep me moving.

Shown in the above image is the skateboard that has the 0.5″ sphere design. I used 11 of them, so 44 0.5″ spheres in total. The spheres were made of nylon. It ended up taking about 15 pounds before I would start moving. After I started moving, it took about 5 pounds to keep me moving.

I could only find 2 of the ball transfers I bought last year from McMaster-Carr. With these, it took 15 pounds to start pulling me. But after the initial 15 pounds it was much easier to keep me going than with my designs. These ball transfers are only suppose to handle a max load of 70 pounds, and I weigh 160 pounds, so I’m sure that these would work much better if I used 4 or 5 of them.


I found ptfe to be the best material to use for the 1″ spheres. I’d like to use it for the 0.5″ spheres and the ball bearings too, but at those sizes they’re just too expensive. I found that my design that has smaller and more spheres (44 in total) worked better than my other design that had 5 1″ spheres. Both my designs were much quieter than the ball transfers I bought, but the ones I bought had a lot less friction.

Overall I’m happy with my designs. I think I’ll be able to use them in my future projects. In the future, instead of using the 4.5mm steel bbs, I may try to use 6mm plastic airsoft bbs and see how well they work.