The Fastest Uni Hub on the Planet. Can I finish for Unicon 18?

Justin_LE · July 19, 2016, 2:22am

For those not familiar with some previous antics, I’ve got a long standing dream of building up a unicycle that I can commute on as fast and as comfortably as riding an ebike. This started with trying to add electric assist to a unicycle, first in a penguin giraffe (circa 2006 Emanual Project ) then to much better success using a direct drive hub motor modified for cranks to go through the middle, which culminated here with very good success
Electric Assist Unicycle Build | Endless Sphere DIY EV Forum
Electric Assist Unicycle at Maker Faire, this Sat. and Sunday

The downside with that build was that it wasn’t geared, so you couldn’t take advantage of the motor much except when going up and down steep hills. That lead to a path of seeing if there wasn’t an easy way to use bicycle hub gears to make a shiftable multi-speed unicycle that could gear the pedals to cruise at 30-40 kph where the electric assist really makes a difference.
This started with the 3 speed Sturney Archer unicycle which worked well enough in principle but in practice had too much backslash and risk of chain skipping to be comfortably rideable:

To the then way more elaborate endeavor of using a Nuvinci CVT transmission which some of you may have seen around at Unicon 17 in Montreal

You can read that thread in full to see all of the challenges and difficulties with the build, and you can see how mechanically things got pretty out of hand. External jackshafts transmissions are not a very good solution.

That made my long for the simplicity of a simple hub motor build just wishing that I had a Schlumpf inside it. What I realized quite a while ago though is that with electric assist, in you don’t really need variable speeds, just a single speed that is geared in the 50-70" realm would be fine and then the motor can look after the extra torque requirements for hill climbing and accelerating from a standstill.

Unfortunately the timing didn’t work to get any Schlumpf gearing parts from Florian before the upcoming Unicon, but I was lucky that in my random old ebike parts bin I had a motor with a 135 tooth ring gear along with some 90 tooth and 20 tooth nylon spur gears I had bought some time ago for an experiment. In theory, that could lead to a planetary gear with a 1 + 95/135 or 174% gear ratio, even more than the Schlumpf hubs at 155%.

So the question, could I use this as the basis for a hub motor electric unicycle that has fixed internal gearing so that everything is clean and tidy in the hub, and can I get this built and tested before we fly out to Unicon 18 at the end of this week?

vookash · July 19, 2016, 2:42am

Do it Justin! I missed the opportunity to try Nuvinci powered one, so I hope I can see the next level of cosmic uni

Justin_LE · July 19, 2016, 2:53am

CAD Models

After a solid Sunday afternoon last weekend tied to solidworks this is what I came up with. Fortunately, we’re in the process of producing an ebike hub motor that is compatible with thru-axle forks and hence uses really large diameter ball bearings, so most of the electric motor components I’m able to pull from that and use without much modification.

The design uses a nimbus ISIS unicycle hub with one flange machined right off and the other flange bolted to a planet gear carrier. This planet carrier rotates with the cranks and turns 5 of the 20 tooth planet gears around the fixed sun gear and causes the ring gear to rotate at that 170.4% ratio.

The ring gear is attached to the left side cover plate after some heavy modifications. Meanwhile, the sun gear is attached to the end of the hollow motor axle, which is stationary and doesn’t rotate.

Justin_LE · July 19, 2016, 3:19am

Parts Machining

It looked from the CAD model like everything could just fit just so, and luckily one thing that we have now at our shop is a CNC milling machine which wasn’t available in the previous uni projects. So last week me and some of the staff at Grin went to town.

Here’s the holder for the sun gear just after milling:

This is the adapter piece that holds the ring gear to the left motor side cover plate being turned down

Precision alignment of all these transmission parts I can only presume is quite critical for smooth gear operation, so everything turned on a lathe was centered accurately with a dial indicator to less than 0.001", like the sun gear bore (original part is http://shop.sdp-si.com/catalog/product/?id=KPS1-95J10)

And the isis spindle

This here shows the back side of the planet carrier. All of those threaded holes line up with the speed holes that are in the Nimbus hub flange, and will serve to bolt the hub spindle to the so that pedal torque is transmitted to the planet carrier.

Finally, quite a moment of truth the first time all of the gear components are stacked together and spun for the first time, which happened just this past Thursday, and of course she works! Hold the sun gear, turn the spindle, and the large plate does almost two rotations.

tholub · July 19, 2016, 8:53am

Beautiful! But will the nylon gears be up to the challenge?

Justin_LE · July 19, 2016, 3:57pm

Yeah, they should be up to that. All the ebike hub motors use nylon gears like this and generate similar peak torque levels to what a rider puts on the cranks (~60-100 Nm), and they usually only have 3 planet gears. Here with the load spread over 5 planets I think there’s a really good margin of safety. The main concern would be if any foreign debris got inside the motor which could wreck havoc on the nylon teeth.

It’s interesting that we don’t see any plastic gears in IGH bicycle hubs even though they are ubiquitous in motorized hubs. I think that’s a result of keeping the target diameter small which means much higher forces for the same torque transmission.

Anyways yesterday I finished machining the main axle which holds the motor stator as well as the sun gear. One side has a 42mm diameter ridge to fit in the ball bearing cap on the frame and provide non rotating reference, and inside this is a 22mm x 32mm bearing for the ISIS spindle to turn in.

And here we are drilling the holes for where the sun gear will attach.

I also got around to installing the planet gears with press-fit ball bearings which should stay put on the small shaft without the need for a lockring.

So now all the key mechanical parts are finished! But before it goes together for a first spin there is a bit of electrical wiring work to do first…

LanceB · July 19, 2016, 4:55pm

As always, your projects are amazing! I will look forward to seeing video of this showing its stuff at Unicon!

As an observation, it looks like starting out from a standing stop will be a challenge. I can see how the electric assist will help, but coordinating it with your own physical launch will be interesting. What size wheel do you plan to run with this?

Good luck!

harper · July 19, 2016, 5:05pm

As always, beautiful work Justin. I agree with Tom. I’m concerned about the shear force on those nylon teeth. But, as you say, if the ebike hubs handle comparable torques, it should work on this as well. Another concern is the press fit pins in the aluminum plate planet carrier. I tried that on my first rendition with dowel pins pressed into a steel disc. The pins elongated the holes but the catastrophic failure was one of the planets galling on the dowel pin and seizing. Good luck, my man.

Justin_LE · July 19, 2016, 6:40pm

I will have an answer to this very soon that I think all of you are going to love.

I’ll be lacing it into a 29" initially, since that’s the largest I can easily travel with to Unicon. That will give an effective wheel of 29*1.704 = 49.5 inches, not quite as tall as a Shlumpf 36 but getting pretty close.

In hindsight I regret not inverting the ring and sun gear positions, then I could have had a gear ratio of 1 + 135/95 = 2.42. Put that in a 24" wheel and still have a 58 gear inches. I think it’d be perfect.

Justin_LE · July 19, 2016, 6:50pm

Oh no! That was the one area of biggest concern for me since I don’t have much firsthand experience with press fit pins, and aluminum is way softer than steel. What was the overall radius where the planet gears were installed in your early prototypes?

I’m hoping that the fairly large distance out of my planets plus the spread over 5 of them means that the forces will stay reasonable. Quick calculation (should have done this earlier!) but we’re at a 52mm radius to the pins, the most torque someone could put on the pedals is about 200 Nm, so that would be 200 Nm / 0.052m / 5 = 770 Newtons or about 175 pounds per pin.

Yikes. I might need to keep the cranks short. With 100mm cranks then a 200lb person standing with full force on a pedal would be more like 90 Nm, so about 80 pounds on the pins. Still a lot for a press fit of a 6mm dowel into aluminum., time to go hang some weights on the gears and see if I need to remachine that part out of steel instead!

-Justin

Justin_LE · July 20, 2016, 1:54am

And we are in luck. I pushed an identical 6mm dowel pin into an identically drilled hole in a piece of aluminum ‘L’ stock to see what kinds of forces were needed to bend the planet carrier pins using a press with a digital readout.

At 200lb, the pin still looked pretty solid and didn’t appear to change it’s angle form before the force was applied.

However, when I got up to around 300lb you could see clearly that the press fit hole was ovalizing a bit and the pin was pointing a few degrees downwards even after the force was released

After 1000lb it was pretty bad

Anyways, this is good news since it looks like the press fit pins on the planet carrier will survive a worst case human torque application, and the build can proceed as planned.

vookash · July 20, 2016, 2:16am

Wow Justin… that looks like magic really happening

harper · July 20, 2016, 8:45am

Mine were on a 2.25cm radius so, much smaller than yours. My solution was to go to a cage rather than a plate. Lengthening the pins and pressing a matching plate onto the open side of the pins would greatly increase the integrity of that part.

Justin_LE · July 20, 2016, 11:46pm

Torque Sensing

And so here’s the deal. All along I’ve wanted to have a pedal torque sensor on these electric assist unicycles, but that’s challenging to do as the pedals and cranks are on a rotating reference frame. However, with a planetary gear drive, the sun gear experiences a drive torque on it too, and so we can sense the rider’s pedal effort from the stationary sun gear support and not have to deal with sensors that are rotating.

That’s why the sun gear support is machined that way with the 4 radial arms, so that they flex a tiny amount when there is torque on the cranks and we can install strain gauges on the surface to sense this quite accurately

What we can hopefully do now is instantly and accurately amplify this pedal torque with the motor, say for every 1 Nm that the rider puts on the cranks, the motor will mirror that with 2 Nm. Then riding this effective 49" unicycle with say 100mm cranks should feel and handle almost exactly like riding a 20" unicycle with 125mm cranks, only you’re moving 3 times as fast.

That is at least my hopeful theory. The strain gauge amplifier board has been installed and tuned to produce a scaled voltage response to the pedal torque OK

And now the hub is all finished and it’s just the rest of the unicycle surrounding it that needs to be built up!

Piece_Maker · July 21, 2016, 3:24am

So that hub alone contains ALL the magic? That’s insanely impressive!

Where are the batteries stored? And how does that cable interact with the thing being turned over (IE. do you have to worry about it tangling with your cranks?)

MartinC · July 21, 2016, 6:31am

Waaoow Justin please make it work for U18! I’m not sure I want to use this before my races, but I DO want to try at least to start with it This is just impressive how fast you physically build this project after, I think, a long and intensive phase of design-thinking. It’s like you pre-thought of everything, and so everything becomes feasible and almost simple when you build it. That really gives a feeling of magic!

Justin_LE · July 22, 2016, 1:40pm

Ha ha Martin, yes you would know all too well! This is pretty much a spot on summary, it’s been a design that’s had 3-4 years of incubating in the back of my mind, so the less than 2 weeks it’s taken to build it isn’t telling the full story.

And speaking of story’s I got it in a state I could ride it my 13km commute home on wednesday night to check that all systems were a go:

The mechanics all worked great, and the amount of backlash on the cranks from the gears is so small to be imperceptible. There was one issue that I hadn’t anticipated though. There was enough eccentricity in the gear drive that the strain gauge for sensing torque also had an offset signal that would undulate just from the position of the crank rotation, to the tune of a ± 10 Nm error even without any actual torque being present.

So as a result I had to dampen/average the torque signal in software so that there wouldn’t be pulses in the motor power at each wheel revolution, and that meant that the torque control was a bit laggy. But when I turned off the automatic torque assistance while riding, then it was really clear how much easier it was to ride with the torque control versus without, even with the delayed response.

It then got a bit more of a workout last night at the Vanuni hockey meetup. Tynan had it down within short order

and other fellow Ben who had never ridden a schlumpf before got the hang of it after 5 minutes or so.

However, after this workout I could hear a regular “tick tick tick” sound coming from inside the hub indicating something mechanical rubbing. So everything came apart to see what’s up and it turns out that two of the press fit planet carrier pins were starting to work their way out of the plate, and the fronts of the pins were rubbing and wearing against the the motor stator support.

We’re trying in a mad rush to remachine a new planet carrier out of steel instead of aluminum, and then I’ll be able to spot weld the backs of the pins in place so that they’ll be locked in position for good. A steel carrier will give me (and Harper too I imagine ) a lot more confidence in mechanical drive holding up to the rigors of field testing for 2 weeks at Unicon.

While I’ve got it all apart, I’m also taking the opportunity to redo the strain gauges so that I have two pairs on opposite legs of the sun carrier, and the hope being that running two sets instead of just the one pair I used previously will help to average out offset variation in the torque signal.

I know that the best ride experience will be had if the pedal torque is immediately and accurately amplified by the motor, that would be the ticket to having a geared 29er feel as easy to mount and ride as an ungeared 20 or 24"

-Justin

juggleaddict · July 22, 2016, 3:09pm

… Man that’s a sexy frame. Reminds me of a black sheep fork.

ernst · July 23, 2016, 3:50am

Impressive! Crazy! Scary (40kph)!

Really great work!

Justin_LE · July 23, 2016, 5:00am

Yeah it was actually the dry run prototype that was kicking around from the project with the Nuvinci unicycle, where I needed the shape done this way for stiffness along the chainline to the jackshaft which was located behind the seat tube:

Without linking the pair of curved tubes together though it has quite a lot of flex in it, so it’s not an ideal design as built but I like the curves and elegance and it seemed a fitting frame choice to use with this build.