Cruise Aider - lean swing arm ebike

[michaels]

Hi!
I'm building up a very lean and low profile ebike and I wanted to get in touch and talk to people actually having some idea what they are doing - that's you! I've started from scratch and I can now say, that I can somewhat ride my ebike. Things are in the very VERY early alpha stage, as I am still carrying my laptop to apply throttle to the motor :wink: . However, what is my main concern at the moment is the unstable roller contact of the motor.

I've attached 2 pics of the current status and I'm planning on documenting more and more precisely when it is a bit closer to being useable.

For reference:
All structural parts were designed by myself and 3d printed on my heavily modified creality cr-10S. The motor is a turnigy aerodrive 6374 with 168 KV rating, which I feel is too big (especially too long) and a bit overpowered. I use a VESC-like controller and a 5S 5Ah LiPo battery pack.

Does anyone have some tipps for me regarding the swing arm length, the direction of a torsion spring and the distance of the motor to the tire/seatstay in it's idle position. A spring, I guess needs to be preloaded towards the tire, right? But then how does it keep the distance to the tire if not used? Also I definitely need endstops! Let's see how this project develops..

 

[fechter]

Check out this one that's similar. Kepler spent years developing this one and it's about as good as it gets for a friction drive:

https://endless-sphere.com/forums/viewtopic.php?f=31&t=86961&hilit=kepler

I think you need a spring to keep it against the tire and some kind of cam that lifts it away if you don't want it engaged.

 

[amberwolf]

AFAICR, in the version of Kepler's drive I have (early, maybe v1), it uses the motor's startup kick to flip the arm into the tire, and the motor's powered rotation itself digs it into the tire. When power is released the spring in the arm pulls it away from teh tire.

You'd have to check his thread for how that works.

IIRC AdrianSM also made a similar drive, and there are some others too. Spinningmagnets has some links in the non-hub motor subforum's Sticky INdex thread.

Just do a topic title search for Friction and you'll find a lot of stuff out there.

 

[michaels]

Thanks Fechter for the link. Actually I've read most of the pages of Kepler's and adrian's threads. I really tried to follow the whole development process, but there are 2 main flaws, the length of the threads - it's more than overwhelming and also some topics are better documented than others.
Thanks amberwolf for the hint that the spring is loaded away from the tire. I know the concept and it works for me, but only when there is enough torque on the motor. Thus, running without a load on the wheel it just disengages very quickly. Pulling the brake a bit and keeping a constant "drag torque" the setup works! Also after this discovery I've tested it with me on the bike and it works as well! However, the engagement is pretty inconsistent and I really see the absolute need for strict endstops and a torsion spring. I will develop these things next. The endstops I already integrated in the 3d print and I just need to add the pin and use it, but the spring I have to somehow buy first - without any idea of where to get stuff like that.

 

[fechter]

I've seen some setups that use a lever mounted on the handlebars and something that resembles a brake cable going to the pivot. You can manually engage or disengage while riding. Automatic engagement is a bit trickier. I suppose you could use a standard brake lever (a short one) to assist engagement and use the spring for automatic disengagement. You could also possibly tie the motor power into the same lever so pulling it engages and powers the motor. Letting off the lever kills the motor and disengages.

 

[amberwolf]

I've seen some setups that use a lever mounted on the handlebars and something that resembles a brake cable going to the pivot.

I used a friction shifter and it's cable to pull this mess down onto the wheel, on one of my first friction drive attempts:
https://endless-sphere.com/forums/viewtopic.php?f=2&t=15570&p=231001&hilit=dayglo+avenger#p231004

If the drive itself had been any good, that would probably ahve worked fine. A gripshifter would've worked better, as it would've locked it down better than the friction shifter did, in increments, but even the friction shifter did well enough.

 

[michaels]

In response to adrian's graphical mechanics work done here for the swing arm geometry, I've done the same (with a little bit more detail) in formulae. It is interesting to see the resulting formula for the normal force that acts on the tire - this is the one that actually holds the motor on the tire.

Note first, that my angle alpha is actually towards the center of the motor, so the trends are equal, but the angle itself is different than in the linked analysis above.
The result is as follows: If the pivot point is inline with the connection line between motor- and tire- centerpoints, we have alpha=0. In that case the normal force is infinite, since we divide by 0. To be mathematically correct we are talking about nonzero very small angles and then the normal force tends to infinity. That means that having small alpha is the best way to make the motor stick to the tire. It is also true that with higher normal forces the efficiency drops, since the tire has to be deformed whenever it comes close to the motor. Since my main concern is a stable condition, however, I'm not concerned with that at the moment.
From the formula for the normal force "N" we can also see the influence of "l", the length of the swing arm. Which is essentially not that important. This suggests that having a smaller arm we can reduce weight and problems with bending arms - my assembly is 3d printed and not THAT strong..
Last but not least we see that the coefficient of friction "mu" has to be bigger than some geometric expression depending mainly on the angle alpha again. Notice how the torque is not relevant to the friction condition.

I'm quite happy about things turned out and that we can clearly see what the geometry is all about! This makes the design process much easier, since I now know that the arm length is not a big issue for friction, but I can use a small arm to increase the stiffness of my design.

Further comments much appreciated, maybe someone finds out more details upon this hint.

 

[michaels]

After calculating the normal force I revised the mounting brakets to include end stops for the swing arm. This worked wonders and after tuning the start throttle code a bit I managed to get pretty reliable and continuous engagement and still have the drive disengage when the motor is stopped. I made a small video showing the engagement:
[youtube]https://youtu.be/uHX9qT7y094[/youtube]

 

[michaels]

Things have evolved quite a bit. The swing arm design works really well now and I am using it on a daily basis for my short commute.


It's the perfect kick to make a city commute more fun and enjoyable. Even on very hot summer days going to work means breaking no sweat.

Like with any project, there's still a lot to do: I want to turn a torsion spring for the swing arm, maybe redesign the swing arm to be mostly supported by long screws, insted of relying only on the 3D printed PLA plastic. I'm planning on using a Hall sensor to do a pedal assist system. And last, but not least I want to move the whole setup below the bottom bracket - but this seems to be in the further future.