Skike - A eSK8 parts based eBike

Hey guys, since a few seemed interested in this project I will make this thread my build log for the various revisions and ideas I may try

Before I made my first board I always wanted a ebike, but then esk8 happened, since I had some spare parts laying around why not?

This is a friction based ebike, the motor can grips directly on the tire, no gear, no belts, no complications, it’s intended just as a assist motor, if you try to push huge currents it will overheat the motor and accelerate tire wear

This first version is just a proof of concept, the motor mount is made from laser cut 3/16" steel since its really cheap

This is the parts involve so far

  • FSESC 4.12
  • TB6355 230Kv motor
  • 5S3P NCR18650GA battery

I’m running 25 A motor 15 A battery, definitely need to increase the motor current a bit since it can do better on hills, and make it have two power settings


Today I did the first long ride, 32 km ridden and 148 Wh consumed, so 4.68 Wh/km, and that is without regen enabled, with a tire that’s not optimum for motor engagement and with a bike that have the rear hub completely destroyed with lots of friction, 3 Wh/km seems possible, if not lower

Average speed was 20 km/h, for a guy that haven’t done any serious cycling for a few months and is considerably out of shape. Since we have a lot of hills around here my average speed used to be around 10 km/h

Apart from the wires from the throttle switch breaking mid ride and having to use the bare wires to engage the motor and the battery running out at around 29 km, it was great

The final version will have way more discrete motor mount, all electronics and batteries will be inside a water bottle with a quick disconnect, proportional regen using the dead zone of one of the brake levers, a compact oled display on the handle bars and single button operation with various functions and a secondary battery connection for longer rides

For anyone that have parts laying around I really recommend, eSk8 was making me fat :grin:


Cool Eskatsicle. I like the simplicity. Got any closer shots of the motor/mount? What kind of throttle are you using? I still haven’t fully wrapped my head around the wired throttle setup with the VESC. On my eTricyskate, I just strapped my nano remote to the handle bar. Lol



Will post some CAD views so it’s easier to see

The two horizontal plates clamp the 2 bottom tubes of the frame, the notch in the vertical plate is for a nut to go, this way when you put a screw and a washer from the top plate is locks the vertical plate to it

The throttle, now after the switch broke free is literally two wires that I short together with the side of my hand on the handle bars, on the VESC, I have two resistors wired like this:

GND - R1 - ADC1 - R2 - 3.3V

The “switch” is wired between R2, so when I press it, it shorts R2 and the ADC receives 3.3V

My first idea was to also use to my spare GT2B receiver, but that would involve opening my board to re bind it latter

The ADC app is really confusing with a bunch of options, to work like this just use the “Current”, I still have to see if it will regen when I install another switch that shorts R1


Initial thoughts are the tricky thing will be figuring out the right gear for the motor if you are using the bikes existing drive train…

edit: Oh sorry never mind it’s friction based

1 Like

Looks pretty solid. I wanna try this. Interesting throttle setup.

1 Like

Apparently people have been caught cheating in professional bike races with hidden motors and batteries in their seat posts that engage with the cranks

They can catch them with thermal imaging cameras looking at the seat post


So, I had a thought. Do you think dual drive is possible with this setup if you just make the motor mount a little longer and angled a bit? For a little extra friction and power transfer? Also, do you think a smaller motor (or two) like a 5065 Ollin/Racerstar would have any overall advantages? On one hand I was thinking that the narrow stator and motor can might have a small mechanical “gearing” advantage, but it seems you would lose a little bit of contact area between the motor can and the tire. Then, of course, the heat issue, if it’s even an issue. Not sure how many amps you could push before the can/tire slips.

Once I’m done with the design I can send you the code, it will be something like a single press engage the motor a 50% and double press at 100%, I can divide it in more steps if needed, any touch at the brake levers disengage

But for your case I would get a eBike throttle, it’s pretty easy to connect to the VESC

That’s the main goal, to be invisible, for now a white battery held in place with duct tape is not exactly stealth, I got some weird look today

And about your gearing commentary, I always thought that gearing was crucial for efficiency, until I tried the direct drive simulator in my board, running 6355 with a 14:36 gearing and with 36:36 didn’t have that much effect in consumption, of course it was higher, but not a lot, 10% at maximum

Take a look at the Kepler design, unless you are searching for it you can’t say it’s electric

Another design I’ve made was using a 3 stage reduction to run 10S and take advantage of all the rpm range of the motor, but I think the consumption would be worse due to all the losses, the final stage would be with chain directly to the rear wheel and a freewheel, so no regen

That’s a great question, dual drive is definitely possible, but my personal opinion is that if you want a more powerful bike friction drive is not the way to go due to tire wear concerns, Kepler talks a lot about it, really recommend reading all of his topics about it

I have 5055 @ 190Kv and will test that, but my bet is that they will overheat, they have 80% of the size of 6355, so if we consider a 190 Kv 6355, we can run 80% of the current for the same torque, but when we consider the winding resistance, the 5055 have 40 mΩ, and the 6355 18 mΩ, even with the lower current we still have almost 4 times the heat loss on the smaller ones, that have less mass to soak it and less surface area to dissipate

The 6355 got hot today with 30°C ambient, with sun all the time, but nothing that I couldn’t hold my hand, but once I start pushing the current things may get out of contro

But hey, if they work without cooking and increasing the consumption by a lot we can make it stealthier

Another important thing to consider, the bearings that goes in all 50XX motors are ridicule undersized, I got tired of replacing them when using in my board, now imagine that you have the constant tire contact pressing on them


Another ride to a lake near my house, on the way back there is a 1.2 km long 70 m elevation climb, on the top of it the motor was really hot, couldn’t keep my fingers on it for more than 2 s, so I won’t even bother to test 5055 motors, but on the next ride I will bring my IR thermometer

For a hill like that the assist sure helps, but I can’t go faster than I usually go on the tallest gear without assist, as @Gamer43 discovered, the low speed of the VESC is really bad, and in this case the motor is spinning at just 600 rpm, more current may help since it will go faster and move away from this low efficiency zone

I will definitely make some sort of heat sink on the final mount design



So interesting thing, the reason for poor low speed performance, I’ve now discovered, is because of crosstalk in the current sensing section (plus or minus some quirks in the algorithm). Basically what happens is that the current being fed into the motor also shows up to the VESC as the BEMF. So tracking at low speeds will actually get even worse at the same speed, but higher motor current.

Honestly, I’d just ditch the VESC for a Kelly Controller and sin/cos resolver at this point, should provide significantly better performance. (Although for parts bin, not much you can do).

Really cool project by the way.

1 Like

Thanks, unfortunately another controller is out of question for now due to size

What would solve this problem? A redesigned PCB? And if I fit a encoder to it instead of running sensorless, would it help? Or the quirks in the code are enough to make it perform bad or the same?

1 Like

PCB design and part changes.

I have no experience with using encoders with the VESC, but I suspect using a quadrature encoder would significantly improve performance.

1 Like

So I may try that in the future

I can really feel that tracking lag on my scooter xD. Above a certain speed, I suddenly feel the acceleration pick WAAAY up xD. It’s already a rather fast speed, too.

1 Like

I’m running my board as single drive until I can fix one motor, without touching the current you can really feel it, and hills it climbs once you are fast

1 Like

I think it’s a very interesting idea the drive the wheel directly with the motor can. Are you set on the mount, or do you have plans to create a tensioner to keep it in good contact with the tire as it wears down or changes pressure or what have you?

1 Like

Not set yet, but so far I haven’t had the motor slip and the pressure is not that high, I just pushed it lightly against the tire by hand and tightened the screws

As soon as I get the rear hub replaced I want to test how little pressure I can set it o before it slips

The spring idea is a good one, keeps constant pressure without deforming the tire too much if it is not perfectly centered


Would it be better on the motor if a pulley was designed intended for friction drive so you could use the motor properly rather than using the can?


I don’t think it would change much, unless you add gearing before the drive wheel. But then we are already away from simplicity

Okay cool. I honestly don’t know enough about motors to know what abuse the can , can handle. Good to know.
I’m definitely interested in trying something like this out