Be careful with polycarbonate it has a tendency to stress crack/craze when exposed to some solvents or to UV then you are left with a lot of microcracks along the bend ^^ (already saw it on some things like ticket machine covers and such)
3 Likes
Threadlocker will also absolutely destroy it.
6 Likes
This is good to know. Chances of me getting threadlocker on this were high, and I likely wouldn’t have cared too much. 
2 Likes
I watched that Easy Composites video on making forged carbon parts a while back and have meant to try it. Looked pretty idiot-proof, unlike some other carbon fiber processes. I could have just used a bit of wood for the footpads here, but where’s the fun in that.
Started out designing/printing the molds. These are too big for my printer so they had to be printed in sections; male mold was glued together, female was held together by two threaded rods. Idea being that this would make it easier to get the part out. Printed at 60% infill, so there’s about 1.5kg of filament here.
Molds were sanded lightly, I’m not real concerned about seeing layer lines in the final part as it’ll likely be covered up. I’d probably recommend doing a layer of primer and sanding it down for a better finish, just make sure this doesn’t mess up your tolerance between the male and female sides.
Those holes in the female side are ejection points I can hopefully use to get the part out without destroying the mold.
I go for 3 coats of PVA mold release, followed by a few coats of a spray on wax mold release (J-Wax).
The part is mostly flat, but does vary in thickness from 5-15mm. Given this, I thought it’d be ok to mix it up a little and use a mix of the chopped carbon and woven fibers.
Followed the calculations in the Easy Composites video to work out the amount of carbon vs resin and it all worked out ok. Just enough resin to make sure everything was wet down enough. No photos of packing the mold as gloves were just covered in a mess of resin and chopped carbon. I will say that spacing the woven carbon at regular intervals throughout the chopped carbon did make things a lot easier, it’s a lot easier to wet down than the chopped stuff and kind of held it in place.
Once the mold was packed it was time to clamp. Used two sheets of masonite to help spread the clamping force, would have preferred some metal plates but just too hard find/cut to size.
Be prepared for a lot of resin to drain out!
Popped the male mold off two days later, and this came off rather easily.
But totally destroyed the female side.
Washed it, cut/sanded off the mold flash, drilled the holes (the countersink was in mold, but not the through hole), and we have a completed part.
Pics from above are a mix of the first and second footpad, second footpad is currently curing so hopefully all goes ok as it did for the first. It’s a bit of effort but seems like a great way to make some stiff/light/decent-looking parts.
10 Likes
The forged carbon looks like it came out really well. Shame about the mould not serviving, maybe more release agent. You could try Maguire’s mould wax on the 3d print, build up layers and let each harden up. Some times warming the mould up with a hair dryer can help it to seperate.
Good luck on the build.
1 Like
FWIW I did manage to demold the other footpad without destroying the mold, not that I’ll even use it again.
Speaking of footpads, finally got around to wiring up the sensors. Four FSRs wired into two parallel groups. I’ve not heard great things about this solution, but I guess I’ll find out. Must have done something to one of them the first time around as I found one of the parallel groups to be always closed 
I think everyone knows carbon fiber is conductive. I was curious though so stuck a multimeter to each end of my footpad, and sure enough, it really was quite conductive. Not a big deal, just means most surfaces the battery will come in contact with are conductive, it’s enough to cause a little concern.
Lined the aluminium and carbon surfaces with Tesa tape, and then another layer of 3mm thick neoprene.
Then the battery was stuck down to the polycarbonate base with double sided tape.
And then the enclosure gets sealed up for hopefully the last time.
Battery still charges, footpad sensors still work, and nothing rattles/moves
12 Likes
It’s critical that the foot sensors work properly or you can have your wheel run away on you (without a rider) and crash.
2 Likes
Really nice work with the FlexiBMS!
what is the display? and that other gray box?
also what charge port?
Last thing is you missed a little opportunity to rewrap the cells with clear shrink and use clear shrink on the whole pack so you could see the cells when fuly put together. Ignoring fishpaper and foam its a great idea 
3 Likes
Yeah, I’ve seen the GT ghosting videos 
Interesting thing about this build is that I’m fairly sure it’ll end up naturally balanced. As in it sits level even when powered off, powered on & armed it just kind of sits there too. Not something I want to rely on, maybe just a bit of redundancy. On the other hand, it it was powered on & armed and you tried to step on one of the ends, well you’d probably get a nasty surprise.
2 Likes
my split pack board has that equal weight distribution property. It is annoying trying to mount the board while its facing downhill because it wants to stay nose down and basically starting in a downhill nosedive position is not ideal.
Good for some stuff I am sure, not amazing overall.
2 Likes
The whole FlexiBMS story is a bit sad really. Even now you could probably rework the firmware to use the latest VESC BMS spec CAN comms and it’ll be a great little BMS. Kind of hard to justify this effort given they aren’t being produced. Thanks again @SimosMCmuffin
The other grey box is basically one of Mitch’s balance buddies (Arduino + CAN module) but hacked to talk to a FlexiBMS instead of a VESC. OLED is connected to this. FlexiBMS powers on the Arduino when the charge plug is connected. It’s a lot of stuff just to show cell voltages, but as I didn’t want to connect it up via CAN to the VESC, it’s the only way I could see this info without opening the board up and connecting via USB. More details here.
I don’t even know how to refer to it . On aliexpress, where it came from, it’d called a “0B FGG EGG 2-pin aviation metal connector”. Rated for 10A.
2 Likes
Close to being done.
Grip tape, a little more cable management, and a bunch of VESC’ing are about it.
Remounted the VESC mounting plate, and the VESC. Thermal paste because I really wanted to capitalize on the frame being a 2kg aluminum heat sink.
Following that the motor mounts and motor go on. This was mostly straightforward except for the motor shaft having a 3mm keyway, while the pulley has a 4mm keyway, so ended up making this. It’s not a great solution TBH.
Printed out two boxes to mount/shield electronics. One in the middle is the Bluetooth module; it still seems to connect ok despite being sandwiched between an aluminum plate and the carbon footpad. Box on the right just includes the two step-down resistors needed for the footpad FSRs with some JST connectors.
If I had a mill I would have cut a channel into the rails to run the battery/footpad wires to the rear, but I don’t. Instead, the cables run along the top of the rail which left them a little exposed. This printed guard should hopefully shield them from accidental bumps.
12 Likes
That thing is looking MEAN!
Maybe you could make rail decals that have something like a shark with the teeth being where the belt is.
That custom key conversion is the cherry on this DIY cake hahaha… or is it literally everything else in this masterpiece.
4 Likes
Definitely open to ideas on what to do for rail decals. Want to get some hand in gears warning stickers made up, part ironic/part serious, that said if you stick your finger in the wrong spot at the wrong time it’s going to get messed up 
There are always VESC stickers too, seems to be the thing OneWheelers do.
6 Likes
I love that sicker. Especially if they’re fiber-based and not plastic. (In other words, paper — so you can soak them with epoxy to apply them permanently)(with a piece of white paper under them)
2 Likes
Cool! Be warned, I did the exact same thing with a mismatched set of keyways, and my franken-key sheared in half after a couple weeks. It’s better to use an unmodified key in the small size, and shim the gap in the larger keyway with a piece of small strip or rod stock + use retaining compound.
1 Like
This is one of the things I’m worried about, was pretty tempted to use retaining compound but didn’t cause I’d like to experiment with different motors/pulleys. Guess I’ll find out
I’d like to switch to an 80kv motor, and I think I’ll find some way to get a 4mm keyway cut in the motor shaft for that.
1 Like
As someone who’s machined keyways in motor shafts… and in pulleys…
…I have to advise against the former. It’s better & easier to machine a 3mm keyway in the pulley instead.
You need an arbor press and a 3mm keyway broach.
1 Like
It’s now finished, well finished to the point where endless tweaking can begin. Few pics before it starts to look trashed.
And the first ride (once I got it rideable).
Wanted to get started on VESC v6 for this, even though it’s still in beta. Beta package didn’t include MacOS build, and the Windows build in there didn’t support Bluetooth so ended up compiling the VESC Tool from source. My build seemed to only include the old firmware so ended up compiling the firmware too. Annoying, but it actually wasn’t as bad as expected. After that I just followed the process over here.
I’ve spent a while now messing with the various params of the VESC balance package, and it’s still pretty “meh” to be honest. Just too soft, and real-time data doesn’t show the amps getting anywhere close to the max I’ve set. May jump over to pev.dev to try and enlist some specialist support.
14 Likes
One of the best things about this forum is that it’s up and open on the web, even Google’able. It’s not really the home of DIY OneWheel stuff, but I’ll keep this thread going because it’s better out in the open than buried in some Facebook group chat somewhere. That said there have been several Facebook/Discord/Telegram groups that have been really helpful, in fact, this post is just a summary of several comments I’ve made everywhere.
So a big part of this build was finding out if this kind of drive system is really feasible for this application, and I have some thoughts. The build itself seems mechanically sound, and I love the aesthetics, but the performance is woeful (my Pint is far better). This seemed off given the specs ‘in theory’ are far better; bigger battery, more powerful motor, better ESC.
My initial thoughts were that I just didn’t have the parameters in the balance package (the thing that makes the VESC do OneWheel stuff) tuned well. I tried many well-established working tunes, and variations of them; none of them seemed to work. Logging via the mobile VESC tool showed I was pushing over 80 motor amps on flat grass with the board really struggling.
So maybe it was the motor tune. Reran the motor detection in as many different ways as possible, always coming up with close-ish numbers. Almost have silent HFI going but it struggles with motor tracking under load, but that’s not going to fix the issue anyway. Ran a very unscientific experiment with the belt drive vs the Pint. It’s pretty simple, let both run at the max speed and try to stop the wheel spinning through whatever means. In the case of the Pint I could actually slow down and probably stop the wheel. Zero chance with the belt drive, wheel speed didn’t deviate at all no matter what I threw at it. Was kind of scary actually. However, repeat at low speeds and you get the opposite results. The belt drive definitely has the power, just not where it’s needed.
Motor KV selection and gearing was chosen to give a reasonable top speed, but I believe the actual KV of the motor is a bit higher than the advertised 50kv. Following calcs based on 14 motor poles, VESC reported voltage of 49, ERPM of 18862, and a duty cycle of 95%.
pole pairs * rpm = erpm
rpm = erpm / pole pairs
= 18862 / 7
= 2695
speed = 2695/60 * (22/49) * 0.280 * pi
= 17.74 m/s
= 63.9 km/h = 39mph
rpm @ 100% duty cycle = 2695 / 0.95 = 2837
kv = 2837 / 49
= 58
I’ve ordered a 12t pulley that will half the top speed and increase the low-speed torque. The real issue is that at these low RPMs the motor seems to be operating very inefficiently. I’m fairly pessimistic that even doubling the reduction will get it out of this inefficient range of rpm/torque. It’s worth a shot for a pulley change!
Probably not surprising that when you research the motor characteristic for low-speed torque you find out it looks a lot like what is used in OneWheel hubs. Lots of motor slots, big diameter, lots of copper windings. I still think electric motors are always better off with a reduction, but I definitely no longer think drive reductions can be used to change the torque characteristics like I needed and expected here.
If the pulley doesn’t work out this project could go several ways;
- Leave it as is and move on
- Just buy a SuperFlux/CannonCore/OneWheel hub, and add a split 6s pack up front to make 18s overall. Would need new VESC and BMS.
- Design/build my own brushless motor with more favorable characteristics.
9 Likes