Kepler Friction drive comes of age.

I am interested in your latest production also John, thanks...
 
Its been a while since I have done any updates on project. Plenty has been happening with a new bike now in the stable using the friction drive.

This is my stealthiest setup yet.
Bike is a 2018 Trek Domane ProjectOne SLR7. Beautiful bike to ride and now even better with a light assist.
All electronics are hidden away inside the frame. This includes the Vesc controller, Arduino controller, and DI2 components.
The battery is a single 150Whr 6S LiPo that is hidden away in the tool bottle with the plug now part of the bottle cage.
I am using a single button throttle together with a cadence sensor. A short push activates the assist in conjunction with the cadence sensor. This give about 150W of assist. You can then hold the button down for a 250W boost whenever you want. Releasing the button returns the bike to pedal assist control and a short press deactivates the drive completely.
Total converted weight of the bike is 9kg and the battery provides 1 hour of full time assist.

Roadbike.jpg

With regards to releasing a new batch of the drives, the good news is that a limited run is with the laser cutters at the moment. A few minor modifications to the last release of drives which I will explain on the sales thread.
 
Kepler Friction Drive.jpg

Please excuse my ignorance but I'm trying to understand the purpose of the elongated hole (over the middle finger tip) on the end of the mounting bracket. Thank you.
 
I'm not familiar with those tiny motors used in friction drives.
What ampere limits would you possibly chose for this motor?
https://flipsky.net/products/6354-190kv-2450w-2
 
I can share the settings that I am using, with this motor:

https://metr.at/m/eNWv

Code:
Maximum Speed	                242km/h
Motor Current Max	        30A
Motor Current Max Brake	        -10A
Battery Current Max	        15A
Battery Current Max Regen	-5A
Absolute Maximum Current	130A
MOSFET Temp Cutoff Start	85C
MOSFET Temp Cutoff End	        100C
Motor Temp Cutoff Start	        85C
Motor Temp Cutoff End	        100C
Maximum Wattage	                15000W
Maximum Braking Wattage	        -15000W
 
gavinc said:
Maximum Speed 242km/h
Motor Current Max 30A
Motor Current Max Brake -10A
Battery Current Max 15A
Battery Current Max Regen -5A
Absolute Maximum Current 130A
MOSFET Temp Cutoff Start 85C
MOSFET Temp Cutoff End 100C
Motor Temp Cutoff Start 85C
Motor Temp Cutoff End 100C
Maximum Wattage 15000W
Maximum Braking Wattage -15000W

Christ that is really fast for a push bike :lol:
Joking apart, how's it going with the motor? How fast can you go?
 
I currently have two different types of friction drive approaches at the moment. The tire climbing under power design as documented in this thread and the constant contact type as documented in this thread http://www.endless-sphere.com/forums/viewtopic.php?f=28&t=96581&sid=b676dc4b5f5d1b33815c1abbff5571f0

Both versions have their pros and cons and as such I have been thinking of how to combine the advantages of both systems into one optimised system.

In summary, the main advantage of the tire climbing design is that it fully disengages from the tire when turned off and as such there is no drag on the bike.
The design's main draw back is that it only works in dry weather (unless you use some sort of grip tape and are happy to deal with a new set of issues)

The main advantage of the constant contact approach is that the drive still operates in the wet (at lower power settings) and gives you the opportunity to use regen.
It's disadvantage is that there is light but noticeable drag on the bike with the drive turned off.

So how to combine the best of both systems?

It was obvious I needed to have a mechanical activation method that would keep the drive permanently engaged when wanted and was quick and easy to operate. Secondly there would need to be changes to the software to facilitate regen.

Here is what I came up with.

A generic remote fork lockout lever was installed and linked to the motor pivot so that activating the lockout fully engaged the drive (to its end stop) and releasing the lockout lever allowed the motor to just clear the tire. However, this had an unforeseen secondary advantage which was that I could now adjust the motor to tire clearance using the barrel adjuster via the remote lockout lever. This also allowed me to also remove the gravity spring. It should be noted that the motor could still self engage and climb the tire under power as per the original design.

The Vesc was then reprogrammed for regen and the Ardunio software modified accordingly.

Now with the drive engagement locked and the drive activated via the Ardunio, any coasting now activates regen with around 200W back into the battery. Any pedal input deactivates the regen and activates light assist of around 150 Watts. The boost button can be pressed any time for a total of 250 Watts. Of course these figures vary slightly depending on speed and associated duty cycle.

The locked position is great for those hilly rides where regen can be used. Also in wet conditions, the drive will still operate up to around 100 Watt assist. leaving the engagement unlocked is perfect for dry flat road riding where coasting freely is advantageous.

This picture shows the modified fork lockout lever mounting.

View attachment 4

Disengaged position.

View attachment 3

Locked engaged position

View attachment 2

20200107_180134.jpg

Stats for a local test ride. The drive was activated for most of the ride with the the drive in the locked position.
Battery has 180 Whr usable. Based on these figures, 30 Whr would be returned to the battery via regen which is significant for a battery of this size. Based on these figure, the addition of regen would have increased assisted range by approx 7km


Screenshot_20200107-175412.png
 
Rode some local hills today that thankfully were not on fire to give the locking mech a good shakedown and really test the regen.

To make it even more interesting, one of the riders was on a new Specialized Creo Expert so I had something to compare my bike against both in range and performance.

I am currently using a 10S2P battery that gives me 180 Whrs as measured by the Vesc. Over the 50km ride that included around 1200 metres of climbing (4000 ft), I managed to regen 60 Whrs. Regen was set at 150W max which was light enough not to greatly effect my descending speed. Assist was used most of the ride making for a really relaxing ride effort wise.

At the end of my ride, I was showing 35% battery remaining with the Creo showing 30% battery remaining.

Below are the Strava details for the ride

Xnip2020-01-10_17-37-10.jpg

It was a great experience to compare my bike with basically the best e-Road bike money can buy. Below is what I observed.

Firstly the Creo is is a beautiful bike and the drive system is super refined. It has the most natural torque sensing system I have ever tried.

Power wise, bikes were pretty similar at moderate speeds however on steep climbs (15% plus) the Creo smashed me. The slow speeds on the really steep stuff work against the Vesc due to the low duty cycle especially on 36V. The Creo could pump out 250W and gear up as needed. Under those conditions with my bike limiting to around 80W. Better then nothing but certainly not my sweet spot.

On milder climbs, (5% to 10%) I could smash the Creo mainly because this was a Euro spec bike. At 27kph (25kph plus the allowable 10%) I could hold 30km plus leaving the Creo well behind.

Noise wise, the Creo drive is certainly not silent. My bike was noticeably quieter under all riding conditions. This was well noted by the other riders in the group.

The Creo is lightest commercial ebike ever made however its hardly a light weight at 12.7kg (28lb) you sure feel it when you pick the bike up. My bike is 9.8kg and basically feels like a normal midrange road bike when you pick it up. That being said, the Creo does roll well un assisted.

So that's basically the comparison. To be honest, the US spec version of the Creo would be a completely different story as really the 25kph speed limits is ridiculous for this type of bike. That being said, it nice to know we can get close to a Creo without taking out a second mortgage.
 
I really like the look of this for low assistance on an old 2006 vintage CF road bike I have.
Would prefer to use one with a 13s1p 48 volt 18650 battery as that's the voltage my other batteries and chargers are.
I'm a bit confused about what motor might be a possible candidate for this or not.
 
Find a 140kv or lower motor. Flipsky have one that will suit that is a 6354 size.

I presume you have 20A cells in that 13S1P pack.

VESC recommends a max of 12S however the components are rated at 60v so it should be ok.
 
I just read as far back as your explanation of the push button throttle and the single crank sensor - I was imagining a similar system (throttle + PAS -> Arduino -> VESC) for mine, but was wondering how to get a thumb throttle on my roadie handlebars. The single button interface is a nice simple solution to that problem.

The only slight issue is if I'm going for 100% UK compliance, the throttle can only power up to something like 5kph with no PAS input, or 25 kph with it, so I'd also need a speed sensor (probably another single hall + magnet like a bike computer speedo) to keep it within the right band. For that kind of timing I'd probably want to use interrupts; have you tried those with your sensors or are you just polling the value repeatedly?
 
Kepler said:
Find a 140kv or lower motor. Flipsky have one that will suit that is a 6354 size.

I presume you have 20A cells in that 13S1P pack.

VESC recommends a max of 12S however the components are rated at 60v so it should be ok.
Thanks.

So the 140KV Motor is actually H6368 which is tight but should still just fit if I've measured correctly.
The H6355/6354 are 160KV/220KV.
So the 140KV H6368 is the right choice? I'm 185cm, 85Kg, and only wanting assistance with pedaling up to about 40kph.

My BB is Shimano 68mm, BC 1.37 x 24 Road with the screw in bearing caps.
From BB centre to tyre is ~80mm.

Haven't made the battery yet but have Samsung 25R 20A cells (as new) that haven't been welded before.
I bought a malectrics spot welder because I want to do three ebike batteries over the next few months and think the 13S1P is probably the easiest one to start with and will do the two 13S3P (from 39 x LG MH1 and 39 x LG MJ1 lightly used cells) after. Only $2.00NZ per cell delivered. The cells check out really well and even on the BT-C3100 discharge test.

For the 13S1P I think I will need to double up on the nickel (0.2 x 8 )?

My plan was to use the Luna 48V Advanced Charger at 1 or 2 Amps and 80 or 90% of full. Just occasionally take to 100% and then use on my other bike to reduce the voltage again.

So when buying from Flipsky I should also get 'Mini FSESC4.20 50A'.
What about the 'Bluetooth Module 2.4G Wireless' For V4?

Thanks

Michael
 
Hi Kepler

I've been a fan of your drive for a long time, and made a fixed version in my bike that I was road testing with just a simple ADC button connected to the VESC for a long time

Now that everything is tested I've implemented a more complex single button control using an AtTiny but the same problem you had is happening, on the bench or with the motor disconnected is runs fine, but as soon as you ride it disengages the motor shortly after starting it

Now I seen you had problems with FOC and a continuous PWM signal, will see if adding a small fluctuation solves it, because after more than a week of debugging, adjusting debounce for the button and messing with the code I'm out of ideas

Cheers
 
It sounds like its more an issue with the settings in the vesc. Make sure you have the max wattage set really high so it is taken out of the picture. Start with low current settings. Say 10A battery and 20A motor. Also make sure your 2 voltage cut offs are set correcty. See if it will stay engaged then. If ok, slowly raise your current limits and set as close to your desired as you can.
 
Thanks for the tips

Did a few more tests using a short lead with the switch really close and all problems solved, so now I have to find a way to remove the noise from the long cable, at least its not another VESC down the drain

If you are interested I documented the build here, still a work in progress

https://forum.esk8.news/t/skike-a-esk8-parts-based-ebike/7518
 
Hi Kepler,
I have not been doing much with with the friction drive I bought from you for a loooong time. well I decided to have another build with it this time with a 29 inch wider tired bike and I made the motor mount locked so it wont move and use the gravity spring.
I also have made the arduino nano throttle interface. it works well with the flipsky 4.12 controller. I am running a 10s greenworks 2.5 amp battery since I have 3 of them. so right now I am in experimental mode with the bike on a stand and it seems to really move with good power. but the test will be when I get it out for a ride. I had first tried this setup with a 5s and my castle controller and it moved the bike okay up to about 20 mph. but that was close to limit on my castle controller. so I have been reading this post over and over and trying to figure out how to make your one button throttle interface. I have not downloaded visuino yet but I think I will try. can you email me your file and the wiring diagram so that I can try to duplicate? I saw your VESC settings and thought I would like to set my VESC the same but with my current throttle I can not figure out how to incorporate the regen either. so if you could tell me how you hooked up for regen also that would be great.
Thanks so much and I am looking forward to giving this another go. Hopefully this time I can post some pics if I get it all to work.
You have really done a great job evolving your drive into such a clean and simple ebike solution!
Steve
 
I knew you would work out a way of getting regen without touching the brakes.
Can I ask whether you will be making the electronics of this new system available to us mere mortals? The mechanical activationlever install should be pretty DIY although maybe some hints via pics on how you achieved this?
I would be very interested in the reprogramed Ardunio/button setup if you are planning on supplying these?

quote=Kepler post_id=1520498 time=1578391033 user_id=14534]
I currently have two different types of friction drive approaches at the moment. The tire climbing under power design as documented in this thread and the constant contact type as documented in this thread http://www.endless-sphere.com/forums/viewtopic.php?f=28&t=96581&sid=b676dc4b5f5d1b33815c1abbff5571f0

Both versions have their pros and cons and as such I have been thinking of how to combine the advantages of both systems into one optimised system.

In summary, the main advantage of the tire climbing design is that it fully disengages from the tire when turned off and as such there is no drag on the bike.
The design's main draw back is that it only works in dry weather (unless you use some sort of grip tape and are happy to deal with a new set of issues)

The main advantage of the constant contact approach is that the drive still operates in the wet (at lower power settings) and gives you the opportunity to use regen.
It's disadvantage is that there is light but noticeable drag on the bike with the drive turned off.

So how to combine the best of both systems?

It was obvious I needed to have a mechanical activation method that would keep the drive permanently engaged when wanted and was quick and easy to operate. Secondly there would need to be changes to the software to facilitate regen.

Here is what I came up with.

A generic remote fork lockout lever was installed and linked to the motor pivot so that activating the lockout fully engaged the drive (to its end stop) and releasing the lockout lever allowed the motor to just clear the tire. However, this had an unforeseen secondary advantage which was that I could now adjust the motor to tire clearance using the barrel adjuster via the remote lockout lever. This also allowed me to also remove the gravity spring. It should be noted that the motor could still self engage and climb the tire under power as per the original design.

The Vesc was then reprogrammed for regen and the Ardunio software modified accordingly.

Now with the drive engagement locked and the drive activated via the Ardunio, any coasting now activates regen with around 200W back into the battery. Any pedal input deactivates the regen and activates light assist of around 150 Watts. The boost button can be pressed any time for a total of 250 Watts. Of course these figures vary slightly depending on speed and associated duty cycle.

The locked position is great for those hilly rides where regen can be used. Also in wet conditions, the drive will still operate up to around 100 Watt assist. leaving the engagement unlocked is perfect for dry flat road riding where coasting freely is advantageous.

This picture shows the modified fork lockout lever mounting.

20200107_175703.jpg

Disengaged position.

20200107_180027.jpg

Locked engaged position

20200107_180043.jpg

20200107_180134.jpg

Stats for a local test ride. The drive was activated for most of the ride with the the drive in the locked position.
Battery has 180 Whr usable. Based on these figures, 30 Whr would be returned to the battery via regen which is significant for a battery of this size. Based on these figure, the addition of regen would have increased assisted range by approx 7km


Screenshot_20200107-175412.png
[/quote]
 
Hi I previously purchased one of the kepler kits, could i get the config file for regen? I managed to rig up a similar setup with a remote dropper lever but couldn't figure out how to get regen to work.
 
You can get the regen to work without reprogramming the Arduino interface but you will need to change the setup within the Vesc. You will need to make changes in the App settings/PPM from "Current No Reverse" to "Current No Reverse With Brake" You will then need to re scale the Arduino output.

In simple terms, the Arduino outputs a signal from zero to 1. No power output is zero, low power output is 0.5 and full power is 1. This suits the "Current No Reverse" option.

If you select "Current No Reverse with Brake", zero from the Arduino is now full brake, 0.5 is zero power, and 1 is full power. So using the same Arduino programming but with the Vesc re scaled, no button push is full regen, short button push is zero power and zero regen, and long push is still full power.

So with the standard Arduino programming, you will miss out on the low power setting that you would normally get with a short button push.

Ideally, the Arduino should be re programmed to output zero with no button push to give full regen, 0.85 with a short button push to give low power, and 1.0 with a long push for full power.

In summary, re programming of the Arduino is not necessary to convert the drive to provide regen however, to have the 2 power output option, a small change to the programming will be needed. Happy to provide the Visunio file to anyone who wants it (PM me) or if you sent me your existing interface, I will re program it for you at no cost except return postage.
 
The mechanical conversion to lock the drive in using a remote fork lockout is really easy to do.

The gravity spring is no longer needed and you just need to add a 4mm flat washer to the motor attachment bolt to lock the cable down.

Lots of cheap options for remote fork lockouts (just google it). Costs under $10.00


5d25bf0c7ac6976ad39020e1-3-large.jpg


I would recommend you also purchase a universal gear selector cable set and cut this down to suit your bike. Something like this https://www.bicyclestore.com.au/parts/cockpit/gear-cable/universal-gear-cable.html

gear_cable_with_housing.jpg





20200110_163101.jpg

20200110_163147.jpg
 
Back
Top