Electric Assist Unicycle Build

[justin_le]

Hey Guys, not too much time to write much now, but here is a photo collection for one of the projects I finished and have with me in SF for the Maker Faire this weekend, which might be of interest to some on this forum. The goal was an electric ASSIST unicycle made from a direct drive hub motor. A unicycle that you can ride normally, but with the advantage of power going on the uphills and regen on the downhills to save your knees in both cases.

The implementation should be pretty self explanatory from the CAD section view:

Solidworks Model of unicycle hub assembly

U Uni Hub CAD Section.gif (51.97 KiB) Viewed 1980 times

[justin_le]

Side covers started off as 9" x 9" by 1/2" thick aluminum plates from Metal Supermarkets, and are mounted to the rotary table for machining:

Starts with raw metal stock and a rotatory table.

UA Aluminum Plates.jpg (61.27 KiB) Viewed 10411 times

Bulk of metal removed with a large diameter hogging bit:

UB Hogging Out.jpg (70.14 KiB) Viewed 10411 times

And press fit cavity for ball bearing too:

Cavity for 20mm x 32mm thin section ball bearing

UC Ball Bearing.jpg (52.85 KiB) Viewed 10411 times

[justin_le]

Inside corner of side plate rounded with a ball nose end mill:

UD Rounding Inside.jpg (53.05 KiB) Viewed 10502 times

All the spoke holes and mounting holes are drilled out on the spoke flange, which is part of the side cover plate rather than part of the rotor.

Center drilling 18 spoke holes + 9 mounting holes

UE Spoke Holes.jpg (46.21 KiB) Viewed 10502 times

[justin_le]

After this, the inside surface of the side cover has been finished, so it gets taken off the table, flipped over, and put back on for machining the outside surface.

Aluminum Plate after all inside machining complete

UF One Side of Plate Machined.jpg (49.59 KiB) Viewed 10410 times

Start of milling the outside surface

UG Second Side Starting.jpg (57.86 KiB) Viewed 10501 times

The flange is milled down 2.5mm thick, exposing all the spoke holes, and the rest of the cover is faced off just for aesthetics.

UH Surface Facing.jpg (41.74 KiB) Viewed 10410 times

Outside is finished with a 3/8" corner radius bit for a nice rounded edge:

Outside Corner Radius

UI Outside Fillet.jpg (43.78 KiB) Viewed 10410 times

And finally, the side completed plate is parted off with a small 3/16" end mill:

Separating finished side cover from the raw stock

UJ Parting Cover.jpg (63.96 KiB) Viewed 10410 times

[justin_le]

The spindle for the cranks started off as an off the shelf cotterless unicycle hub from UDC:

UK UDC Hub.jpg (37.9 KiB) Viewed 10408 times

One of the flanges was machined right off, but the other flange left in place as a means for coupling the pedal cranks to the motor casing via the right side cover.

UL Uni Spindle Machining.jpg (55.85 KiB) Viewed 10499 times

The spindle diameter in the middle just happened to be exactly 20mm already, which was the size of ball bearing that I had going over it.

UM Spindle Complete.jpg (63.62 KiB) Viewed 10499 times

[justin_le]

A second side plate was machined just like the first, only this one was made to fit a much larger 45mm ID thin section ball bearing to go over the aluminum stator support:

[justin_le]

The stator was taken from a Nine Continent hub motor that was having hall issues. My plans had the stator bored out to a 46mm hole. However, you can see that if this is done, it would completely eliminate the welding on the center tube that holds the two stamped steel plates together. In order to keep things structurally sound, I first cut some small steel tubes and welded them to three of the "speed holes" that are punched out of the support plates in order to hold the halves together after the center was cut out:

NC Stator with holes punched into plates and rings to fill them

UO Stator Rings.jpg (64.76 KiB) Viewed 10406 times

TIG welding the rings in place

UP Tig Welding.jpg (79.29 KiB) Viewed 10497 times

Then the original axle hole was drilled:

Large diameter drill bits to remove original NC axle tube

UQ Stator Boring.jpg (47.1 KiB) Viewed 10406 times

And then bored open to exactly 46mm:

UR Stator ID.jpg (67.18 KiB) Viewed 10406 times

[justin_le]

The next step was the aluminum stator support, which plays the important roll of preventing the stator from rotating by being affixed to the unicycle frame, and of holding the stator centered in the hub even with the spindle on which it rests can rotate.

Starts off with 3" diameter 6061 aluminum from a scrap yard:

Raw round stock chucked in the lathe

US 3 inch Aluminum Stock.jpg (63.01 KiB) Viewed 10403 times

A fair bit of machining to get her down to size:

Lots and lots of metal to remove

UT Turning Al Down.jpg (49.6 KiB) Viewed 10494 times

Close to the final shape:

Various surfaces for ball bearing support, frame bearing clamp, etc.

UU Close to Parting Al.jpg (53.84 KiB) Viewed 10403 times

And the completed unit after parting off. The 6 holes are to bolt to the actual Nine Continent stator and lock it from rotating.

Completed stator support,

UV Aluminum Stator Support.jpg (33.61 KiB) Viewed 10494 times

Need wire slots for phase and hall cables too, so these are milled into the 45mm bearing support:

Milling of slots for wires

UW Wire Slots.jpg (45.11 KiB) Viewed 10401 times

[justin_le]

The metal rotor ring and magnets were also made from a returned Nine Continent motor, in this case one that had suffered water damage.

UW1 Water Damaged Rotor.jpg (55.8 KiB) Viewed 10491 times

The stock Nine Continent rotors are a little heavy, with a 7mm thick steel ring for the side cover bolts, and a pretty thick aluminum casting over top of that for the spoke flange. Since I had the spokes coming off the side plate, none of this was needed and I could shave off a lot of superfluous metal:

NC Rotor being machined.jpg (68.23 KiB) Viewed 10400 times

In the end, I took the steel ring down to about 2.5mm thick, reducing the original rotor weight of 1.78 kg down to just 830 grams:

UW2 Rotor Ring Weight.jpg (61.35 KiB) Viewed 10491 times

[vanilla ice]

Hardcore work there.. put some gyros on it!

[justin_le]

Here is what all the bits and pieces looked like after quite a few late nights in the shop:

Finished parts for electric unicycle build

UX All the Pieces.jpg (80.46 KiB) Viewed 10392 times

Stator is pressed on and bolted to the machined stator support:

UX2 Stator Pieces Bolted, end view.jpg (73.27 KiB) Viewed 10483 times

Spindle is inserted into the right hand side plate, and original spoke flange bolts to a small aluminum spacer ring so that turning the cranks turns the hub in 1:1 ratio:

UX3 Spindle in Side Plate.jpg (46.32 KiB) Viewed 10483 times

Steel rotor ring with the magnets is snapped onto this side cover and then held snugly down with 9 bolts clamping the edge. This is needed to hold it secure against the magnetic forces with the steel stator is inserted. Otherwise the 2.5mm steel ring isn't strong enough to stay round, and the force of attraction causes it to wrap around the stator and deforming into an oval shape.

UX4 Rotor RIng Secured.jpg (45.58 KiB) Viewed 10392 times

Even then, it was a bit tricky to get the stator in:

UX5 Rotor and Stator Inserted.jpg (88.12 KiB) Viewed 10392 times

After that, remaining side cover plate with the 45mm ball bearing is pressed in place over the wires.

[justin_le]

Here's a couple shots of the built up wheel, after lacing into a wide 700c rim:

Edge view of completed electric unicycle hub

UZ Wheel Build Complete.jpg (43.36 KiB) Viewed 10482 times

UZA Side view of 29 inch wheel.jpg (80.52 KiB) Viewed 10391 times

I've got the wheel mounted in a Kris Holm 29" Frame with a big apple tire. The little battery rack at the back was made from angle aluminum and holds a pair of Hobbyking 5Ah LiPoly packs and an infineon 25A controller. Handlebar off the seat is home-made and has the CA, a throttle, lumenator light, and ebrake lever all attached to it:

UZB Complete KH 29 Electric.jpg (108.57 KiB) Viewed 10391 times

[dequinox]

WOW... Thats amazing Justin, quite a post after what seems like a long-time-no-see!
How does it ride so far?

Brilliant design, and I can't wait to see how the stabilizer circuits you are probably about to design for it turn out

[liveforphysics]

So nicely done and so cool! I can't wait to see the videos!

[grindz145]

Ebikes.CA for the win.

It's so great to see people that are THIS passionate

[liveforphysics]

Was the Honda electric unicycle a source of any inspirations for your project?

[Toorbough ULL-Zeveigh]

nah, bet it stems all the way back to wishing for one of these when he was a kid.

B.C. Transit

bc2.gif (14.6 KiB) Viewed 5938 times

be careful or u might get sued.


correct me if i'm wrong, but it looks like ur doing all the machining by hand, no cnc.
i found every step useful & directly applicable to various projects, none of which are a unicycle.
it'll save me a lot of frustration having ur pictoral roadmap for a guide, it's greatly appreciated.

[amberwolf]

What, no panniers? Just kidding--every time I think I'm creative I see stuff like this.

Those mountain passes are gonna be easier without the rest of all that bike weight this time around, right Justin?

[Drunkskunk]

Wow. Thats Amazing!

[dbaker]

Very nice job, Justin! Amazing what you can do with a 9C! It would be so easy to enlarge phase wires with that large bearing!

Dave

[etard]

Super sick man!! Video with burnout please!!

I've gotta make it to one of the Maker Faires sometime, great to have someone like you representing the ES scene. Thanks for sharing.

[nicobie]

That's neat! Can't wait to see the video.

I'd probably kill myself trying to ride one of those.

[bzhwindtalker]

unicycles are cool, ebikes are cool.
electro unicycle is cool²

adam

[justin_le]

correct me if i'm wrong, but it looks like ur doing all the machining by hand, no cnc.

That's correct, there wasn't any CNC used here, just a standard milling machine with a rotary table. I installed a digital readout system on it a few months ago though which was invaluable though, especially for getting all the ball bearing fits on the first pass. The first few side covers I experimented with by turning them on a lathe, but for something of this large in diameter it was a bit tricky and the rotary table with large milling bits was faster and perfectly accurate.

i found every step useful & directly applicable to various projects, none of which are a unicycle.
it'll save me a lot of frustration having ur pictoral roadmap for a guide, it's greatly appreciated.

Well glad it could be of use. Will be curious to see what these various projects up your sleeve are!

Justin

[justin_le]

WOW... Thats amazing Justin, quite a post after what seems like a long-time-no-see!
How does it ride so far?

So far, with just a throttle and an ebrake for regen this one works surprisingly well! You don't go especially fast on a 29" unicycle, my cadence tops out around 20-22kph, so with the electric assist I am only averaging just over 1 Wh/km, and for every amp-hour I use going up hill, about 0.5 Ah goes back into the pack on regen from the downhill. So that little 36V 5Ah LiPoly battery is good for over 150km at this rate. I'm at 70km right now and haven't charged it yet.

The control loop with a thumb throttle controller PWM isn't ideal though, and it will be much better when I finish a custom controller that is motor phase current regulated. That way between the throttle and the brake you can dial in the exact torque on the hub which will stay constant regardless of speed, rather than always adjusting up and down on the throttle as your speed changes in order to maintain the same torque. On a unicycle, you have other things to think about, and so having to hold and focus on the throttle for every change in the road or change in speed is a bit consuming.

Brilliant design, and I can't wait to see how the stabilizer circuits you are probably about to design for it turn out

So, you are right about this too. In the original plan, I thought it would be great to have self balancing circuitry on the electric assist uni - no throttle and you lean and pedal to accelerate or brake. I also thought you could use it as a 'trainer' device, to make it easier to learn, and then gradually turn down the amount of electronic balancing as you get more comfortable riding. That was the idea at least.

The circuitboard was made to fit in the same enclosure box we use for the CA production, running a PIC18F series chip to take the signals from a rate gyro and accelerometer and then use that to drive a 4 quadrant controller:

Inside the self balancing circuitry

Tilt Circuit.jpg (101.04 KiB) Viewed 5825 times

I struggled for several days with the code trying to get this to work, but found that to have both human operated pedals AND a balancing control circuit running at the same time resulted in the two always fighting each other. The harder you pedaled, the more you injected a disturbance that the motor would correct by braking, and if you tried to stop or pedal backwards then the motor would kick in high gear going forwards. It was a really strange experience, like riding through a viscous fluid where the harder you worked the more resistance you faced.

As somebody who rides a unicycle, the only way I could make it work was either to turn the self balancing gain way down, so that I was basically able to overcome the motor and keep it steady with my legs, or I could turn the balancing gain way up, remove the pedals and just rest my feet and let the electronics do everything. As soon as you had pedals, and the motor, and the balancing circuitry, it was chaos.

I did bring a 2nd orange 24" electric unicycle down to the Maker Faire which had the balancing circuitry to show people though and to see if maybe someone else could figure out the right human/electric mix with it.

Mark having a go on the electric uni after the pedals are removed.

Mark on euni.jpg (88.84 KiB) Viewed 5825 times

But again it was most rideable with the pedals removed, and since my intent was electric assist this wasn't the end that I wanted. That's why in the larger 29" uni, I left the balancing purely to the rider, and the motor is purely there as an assist and a brake, and that seems to work much better. At least for now, there is still a lot to explore here. -Justin