Boxxbike Valkyrie : 12kw crank concentric mid drive with left hand motor drive

[scrambler]

Dropper post turned out perfect, with the full 125mm extended I have perfect leg extension when pedaling, and at the lowest I can mount and dismount easily or cruse with throttle.
That said, even with the dropper post extended, the seating position is still too forward, and even worst when at lowest. So I decided to try and make an adapter out of the part I mentioned before (pic below)



I cut off the top part to just keep the U bottom piece.
I bent it a little, this is not really needed but when straight, to get the saddle horizontal, I was at the limit of the dropper rotation adjustment. With a little bend I have more rotation range.
Note: These bars are wicked hard, it took a big Vise and a very large hammer to bend them. I guess that is good and means this will be a rock solid adapter.
Below you can see the cut and bent part in place on the dropper post, and my “comfy” saddle on it.
This thing gives you several inches or horizontal adjustment, so you should be able to get exactly what you need.




Now the other posture issue is the fact that the reach is a bit small for me, putting me in a quite upright position. I like more of a trekking posture with about 90deg between arm and Torso, which is also better for my back…
So I installed an adjustable Stem, and that gave me the posture I needed. Next to test on the trails…



And the telescopic Kickstand worked a charm. I used the left hole of the tensioner protection plate, re-drilled and threaded at M6, and drill / threaded a new one next to it.




Other articles
Purchase report:
viewtopic.php?f=3&t=94174&start=75#p1675901

Customizations:
viewtopic.php?f=3&t=94174&start=100#p1676907

Torque sensing pedal assist:
viewtopic.php?f=3&t=94174&start=100#p1679413
viewtopic.php?f=3&t=94174&start=100#p1686510

Regenerative braking:
viewtopic.php?f=3&t=94174&start=100#p1690743

Kindernay IGH conversion:
viewtopic.php?f=3&t=94174&start=100#p1709402

[briangv99]

Scrambler/Sn0wchyld, do you what the differences are between the Cyclone coaxial motor and the one Boxxbike uses?

They look almost identical, but Boxxbikes motor seems to be quieter from the videos I've seen, but it's always hard to tell with camera microphones being so different. Could be quieter because Boxxbike use Silixcon FOC controller.

What are your thoughts?

[scrambler]

I have no info on what they use for their motor but a Key difference is that the Frame is used for the Stator.
Basically they use the actual frame as their stator, which helps greatly with the heat dissipation.

[briangv99]

Interesting, I've seen a video where they recommend drilling a hole in the bottom of the frame to let out water that might get in past the axle. Must be quite tightly packed in the bottom bracket area.

[scrambler]

Mine actually came with that hole already made and a small screw to open it.

Look in the first 10 seconds of the video below, you can see in the back at the top right, a bike with the frame motor cover open and the stator as part of the main frame.
https://www.youtube.com/watch?v=XyKVnZGmMis&t=10s

[scrambler]

Testing on the torque sensor is progressing.

Initial wiring below allowed for both the Amplified Load Cell signal and the throttle to be mixed (with diodes) and both activate the motor (the highest one wins). The load Cell signal creates a very smooth acceleration when being pulled on by hand.





Next I completed the sensor mount(s), and it fits nicely inside the bike triangle.
With the derailleur, there is a lateral chain movement of 14mm between highest and lowest gear, using a standard nylon pulley I get 7mm inside the pulley, so I added 7mm of lateral space on the bracket.



For a single speed, which is what I will have once I replace the derailleur by the Kindernay IGH, I can use a smaller wheel with no lateral play.






The 50Kg Load cell has a 2mV/V signal, so being powered at 12V, it should mean 24mV under a 50kg load. It also has a 1.2mV signal at zero load.

Once mounted on the bike, right now it creates a 10 degree chain deviation, and with that, I get about 15mV when standing on one pedal (72kg applied on one pedal).
• This means I won’t overload the sensor.
• This also mean a 25Kg sensor could probably be used with a smaller chain deviation.
The lower the Pulley, the shallower the chain deviation angle, the less downward force for the same pedal pressure (see forces calculation on a previous post)

The board has a zero adjustment which allows to offset the output Voltage at zero load. I should also be able to use that to adjust how much pressure on the pedal starts producing a positive voltage.

The board also has an amplifier adjustment, so I can use that to decide the max throttle voltage it will produce. Given the power range of the bike, I suspect I will want to have the max Sensor signal to be around 50% of the max throttle, actual road tests will tell.

I did a first Road test with the testing harness, and it works great, direct pressure on the pedal activates the motor in proportion. No lag, Instant response, very progressive. This validates the Mechanical solution of capturing pedal pressure for throttle signal!

Now the board I have is hard to adjust because the Amplifier Pot and the Zero Pot interact with each other and can cause some weird output at specific combination. I am trying to see with the vendor if he has any insights on that.
This is important, because setting the zero appears to be key in not having the system be too rough to use.

In order to do more testing, the next step is to make a clean harness and box for the board and converter, as well as a probing connector so I can more easily finesse the output zero and max, and test how it pedals…

This is definitely not as flexible as using the Arduino, as the Arduino would allow programmable adjustments (easier than tweaking pots on a board), as well as creating smart processing of the Sensor signal. But it is a start and easier for me to implement at this time as I have never dealt with the Arduino before.

Pure Torque sensing is great on the trail in hills and aggressive riding where you have immediate response to power actions on the pedals, but not ideal on the flat going at constant speed as it does not provide a smooth output. When pedaling with force at a medium cadence you get bursts of power separated by a brief nothing. For that situation it would be key to be able to mix both cadence and torque so the cadence sensing can provide a baseline, and the torque the incremental power when needed. This will eventually be possible if we can program the controller to enable the throttle during the PAS mode…

Ideally the most powerful implementation would be to inject the Sensor signal into the controller torque input, and use the controller programming to do advanced and smart mixing between cadence and torque like LMX has done on the LMX64. But that would involve significantly more work and cooperation with Boxxbike….

Good first test, but there is more to be done, so to be continued, and I am looking forward to hear from @Sn0wchyld about the Arduino.

Other articles
Purchase report:
viewtopic.php?f=3&t=94174&start=75#p1675901

Customizations:
viewtopic.php?f=3&t=94174&start=100#p1676907

Torque sensing pedal assist:
viewtopic.php?f=3&t=94174&start=100#p1679413
viewtopic.php?f=3&t=94174&start=100#p1686510

Regenerative braking:
viewtopic.php?f=3&t=94174&start=100#p1690743

Kindernay IGH conversion:
viewtopic.php?f=3&t=94174&start=100#p1709402

[sn0wchyld]

nice work mate... mine has been on the backburner... just found out the other week I (along with many workmates) have been made redundant, and that means moving back to AUS at short notice, so might not have much time for this for a while, depends on what happens in the next couple weeks... certainly sometime early next year it looks like i'll have a lot more time on my hands hahah.

for the code im working on - I will be building it to take 2 inputs, 1 from the torque sensor, 1 from the throttle, and doing a basic 'max' comparison to feed to the controller throttle signal (ie no modification needed to controller) so you can combine signals easily. Also will be working to soften off the throttle response if possible, idk about you but imo could do with a little more modulation / progressiveness at high power modes. Code is pretty simple though, and might help with the 'pulse' nature your seeing (basic low pass filter used).

Testing on the torque sensor is progressing.

Initial wiring below allowed for both the Amplified Load Cell signal and the throttle to be mixed (with diodes) and both activate the motor (the highest one wins). The load Cell signal creates a very smooth acceleration when being pulled on by hand.




Next I completed the sensor mount(s), and it fits nicely inside the bike triangle.
With the derailleur, there is a lateral chain movement of 14mm between highest and lowest gear, using a standard nylon pulley I get 7mm inside the pulley, so I added 7mm of lateral space on the bracket.



For a single speed, which is what I will have once I replace the derailleur by the Kindernay IGH, I can use a smaller wheel with no lateral play.






The 50Kg Load cell has a 2mV/V signal, so being powered at 12V, it should mean 24mV under a 50kg load. It also has a 1.2mV signal at zero load.

Once mounted on the bike, right now it creates a 10 degree chain deviation, and with that, I get about 15mV when standing on one pedal (72kg applied on one pedal).
• This means I won’t overload the sensor.
• This also mean a 25Kg sensor could probably be used with a smaller chain deviation.
The lower the Pulley, the shallower the chain deviation angle, the less downward force for the same pedal pressure (see forces calculation on a previous post)

The board has a zero adjustment which allows to offset the output Voltage at zero load. I should also be able to use that to adjust how much pressure on the pedal starts producing a positive voltage.

The board also has an amplifier adjustment, so I can use that to decide the max throttle voltage it will produce. Given the power range of the bike, I suspect I will want to have the max Sensor signal to be around 50% of the max throttle, actual road tests will tell.

I did a first Road test with the testing harness, and it works great, direct pressure on the pedal activates the motor in proportion. No lag, Instant response, very progressive. This validates the Mechanical solution of capturing pedal pressure for throttle signal!

Now the board I have is hard to adjust because the Amplifier Pot and the Zero Pot interact with each other and can cause some weird output at specific combination. I am trying to see with the vendor if he has any insights on that.
This is important, because setting the zero appears to be key in not having the system be too rough to use.

In order to do more testing, the next step is to make a clean harness and box for the board and converter, as well as a probing connector so I can more easily finesse the output zero and max, and test how it pedals…

This is definitely not as flexible as using the Arduino, as the Arduino would allow programmable adjustments (easier than tweaking pots on a board), as well as creating smart processing of the Sensor signal. But it is a start and easier for me to implement at this time as I have never dealt with the Arduino before.

Pure Torque sensing is great on the trail in hills and aggressive riding where you have immediate response to power actions on the pedals, but not ideal on the flat going at constant speed as it does not provide a smooth output. When pedaling with force at a medium cadence you get bursts of power separated by a brief nothing. For that situation it would be key to be able to mix both cadence and torque so the cadence sensing can provide a baseline, and the torque the incremental power when needed. This will eventually be possible if we can program the controller to enable the throttle during the PAS mode…

Ideally the most powerful implementation would be to inject the Sensor signal into the controller torque input, and use the controller programming to do advanced and smart mixing between cadence and torque like LMX has done on the LMX64. But that would involve significantly more work and cooperation with Boxxbike….

Good first test, but there is more to be done, so to be continued, and I am looking forward to hear from @Sn0wchyld about the Arduino.

[scrambler]

just found out the other week I (along with many workmates) have been made redundant, and that means moving back to AUS

Sorry to hear about your work situation. Hopefully it will work out for the best...

I have a quick question for you regarding the existing Cadence sensing PAS mode.
On mine when starting from a dead stop, it takes about 1/4 pedal turn for the assist to kick in which is not bad, I can actually start the bike in the highest gear.
But when you are moving and pedaling constantly, if I stop pedaling for a brief moment, then start pedaling again, it takes about two full pedal turn for the assist to kick back in.

I am wondering if they set it up this way for some odd reason, or if it is a system limitation. It is a bit of a pain if you are in a hill...

I will ask them eventually, but first wanted to check if yours behaves the same first.

Thanks

[scrambler]

To Brian, I have the 14kW version, and 14kW must be peak, as the motor is set to a max of 8kW sustained.

The Motor Chain ratio is 13T to 97T
On the 12kW version (7kW sustained), the rear sprocket is 88T, don't know about the front but probably the same 13T

The 13T sprocket has a clean Gear tooth profile (not the rough pedaling chain sprocket profile), that may contribute to the silence.

EDIT:
Well, it appears I was wrong on the bike power…
BoxxBike use to make the battery as a 12S x 12P using VTC6 cells.
These Cells have a 20A max continuous discharge and 35A peak discharge, which gave the battery a Max possible Continuous power of 10kW and a Max Possible Peak power of 18kW (12 x 3.6V x 12 x 35A). Plenty to support the 12kW or 14 kW advertising.

But it appears that at some point they switched to the Panasonic / Sanyo NCR18650GA Cells (with the same 12S x 12P configuration).

These cells have the benefit of a slightly larger capacity (3.45Ah vs 3.12Ah) and therefore increase the battery capacity from the previous 1600Wh to 1800Wh which is nice.
But these cells have a max continuous discharge of 10A and a peak discharge of 14-15A.
This limits the battery max continuous power draw to 5kW, and the max peak power draw to 7 - 8kW.

So my 14kW Bike is actually a 5kW continuous and 8kW peak bike.

The controller is still able to deliver high Phase Amps (up to 360Amps) so the bike still has the same great torque and is a great bike to ride with a lot of power to spare, but the 14kW is not real …

I am not sure what is the difference between the bike sold as 12kW vs the one sold as 14kW. May be the bike sold as 12kW, is actually limited to 7kW peak…

But I wanted to rectify my previous erroneous statement.

[briangv99]

Great info Scrambler,. Thanks

I believe the sprockets and chain are #219 go kart derived parts, same as LMX chain drive. I ran my LMX with chain drive initially, and while it wasn't bad noise wise, the thrashing chain sound wasn't pleasant.

I think you're right that the concentric drive arrangement, plus the thick gauge alloy used in the frame must dampen the noise quite a bit.

[scrambler]

It looks like Nicolai is bringing a Neematic substitute to market.
https://en.nicolai-bicycles.com/eboxx-ultra/

6kW / 10 kW
Pinion gear box
Gates carbon belt
2kWh battery

Glad to see this category of bike to start emerging as valuable choices

[efMX Trials Electric Freeride]

posted about it here;)
viewtopic.php?f=25&t=56951&p=1675651&hi ... i#p1675651

[scrambler]

I wonder what PAS they have , if it is only cadence sensing, or if they have also added torque sensing.
And if they have added Torque sensing, what type did they use...

[scrambler]

I wonder what PAS they have , if it is only cadence sensing, or if they have also added torque sensing.
And if they have added Torque sensing, what type did they use...

Actually so far they have no PAS planned.

Not sure if it is an oversight, or they think that pedaling a 60kg bike without Pedal assist is fine....

[scrambler]

Previous posts on this project
• Force calculations
viewtopic.php?f=3&t=94174&start=75
• Early tests
viewtopic.php?f=3&t=94174&start=100#p1679413

I finalized the Wiring of the Torque sensor module.
Below is the final wiring of the Harness at the head of the bike, and the Torque sensor electronics at the bottom.




I modeled and 3D printed a small custom box (pictures below) that contains the 48V to 12VDC converter, the AD620 Amplifier board, a fuse, an ON/OFF switch so the sensor electronics don’t stay powered (my 48V source is always on), and the 4 connectors.



It fits very nicely inside the frame Seat Tube support and below the shock, and is barely noticeable.







Usage Review:
As I mentioned in my last post on the subject, the torque sensor is great on rough terrain and incline as it gives you instant response, and real-time proportional adjustment of the assist. This means you can start from a dead stop in the middle of a hill in any gear, and get going as quickly or as slowly as desired. These are impossible (dead stop start uphill in any gear), or very hard to do with the cadence sensor pedal assist without constantly changing assist level, which is impractical and not always quick enough to respond to highly variable terrain or a sudden obstacle.

On flat terrain when cruising at regular speed, the torque sensor alone is not as smooth as the Cadence sensor PAS, because when cruising at leisurely speed on the flat, the pressure on the pedals is intermittent and so the resulting assist is not as regular as with the cadence sensor. That can be improved with low pass filters as sn0wchyld mentioned, but ultimately smart mixing of the Torque sensor and cadence sensor signals is how hybrid assist systems like Bosch or Bafang really shine. To do that properly, it will require plugging in the Torque sensor signal into the Silixcon Controller Torque sensor input, and use the controller programming tools to combine both sensor signals.

EDIT: There is actually a Low pass filter parameter for the throttle in the Silixcon controller, using it you can really smooth the Torque sensor input and for aggressive pedaling even on the flat, it makes it really nice. The Low pass filter also has the benefit of making the Throttle less brutal. You still have to watch out for the front wheel going up at high levels, but overall it does a great job.
Parameter is acc/asc/lpf and a value of 0.03 seems to do the job well

So right now, I configured one set of modes for cruising on streets and smooth trails using the Cadence sensor assist, and another set for more aggressive MTB situations, rough terrain and steep inclines without the cadence sensor, and where the torque sensor provides an incredibly fun, powerful and intuitive assist. With up to 8kW at the tip of your feet, you can pedal you way up anything

Below are my current maps until I have had more time on different type of trails, and until I have installed and tested with the Kindernay VII.

Maps 1 to 4, “Cruising maps”
Cadence sensor Pedal Assist ON at increasing Watt power (250, 500, 750, 1250)

• Throttle is also enabled in these modes with max torque level (trqlvl=1), so I can use the throttle from a dead stop if needed.
Note: Right now Silixcon does not allow a different max power for the throttle and the cadence sensor assist, so at low assist power the throttle it is pretty useless. But they are working on a new “human watts feature” that “May” allow us to adjust the power differently for the cadence sensor and the throttle. If that turns out to be the case we could have a powerful enough throttle even in low PAS maps.

• Because the throttle is enabled, I can also switch on the torque sensor on at the handle bar to mix both, but unless we could set the throttle power to a higher level than the cadence sensor PAS, there is not much benefit from them competing at the same level. (This would allow the torque sensor to provide burst of power above the cadence PAS)

• Note: BoxxBike has improved the response time of the pedal assist. It kicks in even quicker from a dead stop (it was not bad before with ¼ turn), but more importantly restarts quicker when the bike is going (it use to take two pedaling turns before, which could lead to tricky situations when going uphill).

Maps 5 to 9 “MTB Motorcycle maps”
Throttle and Torque Sensor assist at increasing watt power (2000, 3000, 4400, 6000, 8000)
• Torque sensor can be switched ON or OFF at the handlebar depending on the situation.
• Torque level are incremented from 0.7 to 1 to provide a slightly smoother throttle at lower power. I am still experimenting with that…
EDIT: with the low pass filter on the throttle, I may actually leave the Torque at max for all maps...

Next project is the Kindernay VII IGH when it gets to me
To be continued….

Online folder with all the project pictures
https://1drv.ms/u/s!At3vMAQjaOZLlVGR_MO ... i?e=lSc2gK

Other articles
Purchase report:
viewtopic.php?f=3&t=94174&start=75#p1675901

Customizations:
viewtopic.php?f=3&t=94174&start=100#p1676907

Torque sensing pedal assist:
viewtopic.php?f=3&t=94174&start=100#p1679413
viewtopic.php?f=3&t=94174&start=100#p1686510

Regenerative braking:
viewtopic.php?f=3&t=94174&start=100#p1690743

Kindernay IGH conversion:
viewtopic.php?f=3&t=94174&start=100#p1709402

[scrambler]

Engine brake:
The BoxxBike comes with Regenerative braking that activates on the Throttle cutoff, which is perfect for providing Engine brake in Throttle mode.
Coming from the world of big single 4 strokes Motorcycles I enjoy a good engine brake
Controller parameter that activates it in the maps is comlvl (value o to 1)

Unfortunately, because regen takes priority over any power delivery it acts as an engine cutoff. As a result, this type of regen cannot be available in PAS modes. Pedal Assist provides power while the Throttle is off, so if comlvl is on, it disables power coming from PAS when throttle is off. Comlvl is therefore set to zero in PAS maps.

Regen on actual braking action:
Silixcon also has the possibility to activate Regen with an ON/OFF switch, it is called the static brake.
BoxxBike actually provides a 2 pins connector for it at the head, as they use to do Regen that way with a Custom Switch on the rear brake
I understand reliability issues with their Custom Switch motivated the change.

Good News, is that Magura has e-brake levers compatible with their various models, and these provide an excellent lever switch. They exist both as Closer (Normally Open), and Opener (normally closed).
The one we need is the Closer (Normally Open).


You can find them on their webstore. www.magurausa-shop.com
• Part 2700838 MT7E lever blade
• Part 2700837 MT5E lever blade
• Part 2700836 MT4E lever blade

Notes:
• The switch activates at the very beginning of the lever action, before any actual mechanical braking, so you can activate regen alone that way.
• The only drawback is that they only exist as 3 fingers and not one finger…
• Below is a Video on how to change the brake lever on an existing brake
https://www.youtube.com/watch?v=-Qx7qV9ud4M

You could use one lever on each brake to activate regen with either the front or rear brake, if so you would connect the two switches in parallel.

I personally chose to only add a regen switch lever to the Front brake.
Reasons for that choice were:
• The front brake is the most used brake
• If I brake fully, I get powerful front mechanical braking and some rear braking, for a more balanced and stable deceleration.
• To use the front brake you usually release the throttle, which eliminates the risk of engine cutoff while throttle is twisted, which can lead to brutal leap when regen is released.

Controller configuration:
• The Connector for the ON/OFF switch at the head is connected to the GPIO1 Controller port.
In order to activate it, you need to do two things in the Silixcon controller programming.
...o io/IN_sbrake must be set to 17,0 (Digital input on GPIO1)
...o Then to activate the brake in any map, set the map parameter sbrakelvl to 1 for max regen
• Global Regen intensity is set by the parameter: driver/irefr, and by default it is set to -25. For the Sony VTC6 cells used in Some BoxxBike battery, it should be safe to use up to 50 Amps for max regen.
Note: I found out I do not have VTC6 cells, so I may not be able to increase that regen number. Make sure to confirm what you have before making a change there!!
• Static brake Ramp Up and Down are set by brake/rgnslup and brake/rgnsldn low pass filters.

Silixcon low pass filters are explained below
https://silixcon2.atlassian.net/service ... -531453331

If you want a different intensity for the brake lever and the throttle (engine) brake, you can increase the Global irefr parameter, and then use the comlvl or sbrakelvl to reduce either one.

For example IF you have VTC6 cells, to have a brake lever regen twice as powerful as the current engine brake, you could do
• driver/irefr = -50
• comlvl = 0.5
• sbrakelvl = 1

With that you have Strong Regen on brake lever, and medium Regen as Engine brake on throttle

Still waiting for my Kindernay VII...

EDIT:
Regenerative braking update
Here are additional informations on the regen settings for the BoxxBike / Silixcon Controller.
I mentioned the regen max level was set in driver/irefr
In addition to the parameter driver/irefr, there is another limiter which is driver/limiter/ibneg
• driver/irefr is the Phase current limiter for regenerative braking
• driver/limiter/ibneg is the Battery current limiter for charging / Regenerative braking.

For the New battery with the GA cells, ibneg needs to stay at -25 Amps as this is the max charge current acceptable for the battery.

But because Phase voltage can be lower than battery voltage, irefr could go above 25 Amps when the phase Voltage is lower, without creating a battery charge current over 25 Amps.
Battery charge voltage is 50.4 volt, so 25 Amp max charge current means max Charge power of 1260W
A Regenerative braking of 50A phase current and 25V phase voltage (1250W) would create a safe Battery charge current of 24.8A

So I asked Silixcon if ibneg would take priority over irefr and protect the battery in case we would set irefr higher than -25 Amps, like -50Amps

They confirmed that the first limiter triggered will limit the regen current, and so ibneg would indeed protect the battery if irefr is set higher.

Setting irefr to -50Amps for example, allows for more regen at certain speeds, without risking the battery charge current to go above the 25A limit
If at any time the phase regen current was to create a battery charge current that exceeds the limit set by ibneg, ibneg will limit the phase regen current to keep the battery charge current under 25Amps.

Notes :
• irefr is on one microsecond loop, which is faster than ibneg which is on a one millisecond loop. But one millisecond should be fast enough to protect the battery.
• In addition to the ibneg controller limiter protection, the Battery BMS should have its own charging overcurrent protection, but that would have to be checked to be sure of what the setting is

So even if you have the New Lower discharge / charge rate battery (using GA cells instead of VTC6), You should be able to set irefr to a higher value to get more regen at lower speeds, as long as you keep ibneg where it is set by the manufacturer to keep the battery protected.

Then you can eventually set the engine brake to a lower level than the brake lever using the comlvl parameter.
I personally use the settings below now
• driver/limiter/ibneg = -25 (stock)
• driver/irefr = -50
• comlvl = 0.5
• sbrakelvl = 1

Disclaimer: I am sharing what I am told, I am not responsible for you frying your battery

Other articles
Purchase report:
viewtopic.php?f=3&t=94174&start=75#p1675901

Customizations:
viewtopic.php?f=3&t=94174&start=100#p1676907

Torque sensing pedal assist:
viewtopic.php?f=3&t=94174&start=100#p1679413
viewtopic.php?f=3&t=94174&start=100#p1686510

Regenerative braking:
viewtopic.php?f=3&t=94174&start=100#p1690743

Kindernay IGH conversion:
viewtopic.php?f=3&t=94174&start=100#p1709402

[scrambler]

I was wrong on the Bike power previously; it appears the bike is now delivered with a battery that uses different cells with more capacity but less power.

I edited my last post on the subject with the new information
viewtopic.php?f=3&t=94174&p=1679641#p1679641

It is still a fantastic bike with plenty of Torque and power, but the advertised Power numbers of 12kW & 14kW do not reflect the Bike Power.

[scrambler]

I added some new info in the Regenerative braking Post above, allowing to increase the regen power for some conditions, even with the Batteries equipped with the GA cells that have limited Charging current.

[Lwing]

Hi there,
I’ve been following this discussion as it seems to be the only one on boxxbike I can find. This is the only bike I’ve found that ticks most of my boxes.
I’ll admit the ebike tech is out of my comfort zone, the bike mechanics I’m fine with. Pretty impressed with the owners tech savvy to tweak there bikes.
I see the kw aren’t exactly as advertised, I’m okay with that, but the watt hours? Are these actual, close? I need range, no throttle for some of the things I intend to do, I believe I’ve read 1800 wh, I’m 215lbs, am I correct to think 35wh p/km would be realistic? 50% flat gravel, 50% climbing gravel road up mountain.
40kg is max weight in bike I’d want, the only other bike I see is cyclone, coaxial which would need lots of upgrades
Would need a moto rear rim and tire as the blast rock roads I ride destroys mtb tires.

[scrambler]

The new battery (the one I have) is made with Panasonic / Sanyo NCR18650GA Cells and these are 3.6V nominal and 3.45Ah capacity.

The battery is 12S - 12P which means 12 Cells are connected in parallel to create master cells of 3.6V and 41.4Ah, then these 12 master cells are connected in series which makes the final pack 43.2V and 41.4Ah for a total capacity of 1788Wh

Now most ebike manufacturer actually calculate their battery capacity using a Cell voltage of 3.7V. If you do that the pack is 44.4V x 41.4Ah = 1838 Wh

Now calculating range from that is impossible, as on a 5kW (8kW peak) bike, how you use the bike, will widely change the range.
Things that will greatly influence the range are
Using pedal assist (more or less) vs using throttle only
Speed in pedal assist is pretty much limited to 22mph because of the gearing (unless you install a Kindernay as I plan to do ). That means 35 km/h
But you can go faster on throttle, and that would obviously reduce the range.
How steep are the hills...

Using pedal assist only on trails I usually ride happily in mode 2 or 3 on flat (500 - 750W), and only increase to higher mode (1000 - 1500W) for hills. I am only 165 Lb and I never actually run down the battery, but my estimation for a mix of flat and hills in pedal assist is an autonomy of about 70 km, and on flat possibly 110 km.

Boxxbike number on their site are based on the Older 1600Wh battery (the one that had more power), and they seem conservative.
Their say on single track 50-60km, that is to be compared to my estimate of 70 km.
35Wh/km with an 1800Wh battery would be 51 km, all these numbers are in close proximity
They also say that if you are on throttle in difficult terrain that can be halved to 30km

Now if you have hills, one point to consider and that is not reflected in my numbers above, is the fact that the bike has regenerative braking, and when going downhill on long strips, that can actually bring back some significant range (especially if you increase the regen level as I did).

But if you have tested the range of existing ebikes, and you are going to use the BB in pedal assist mostly, you can extrapolate the range using the 1800Wh battery capacity number vs the one of the bike you tested, and that should be a conservative extrapolation because of regen.

[scrambler]

I just did a battery range test, starting with a battery at 49.3V (4.11V / cell), street riding 30 miles at PAS 2 (500W) to 4 (1200W) to maintain around 20mph, with 0 to 180ft elevation and google maps, showing 420ft going up and 420ft going down (no idea what that info on google map really means ...).
After that I still did about 25 miles of local trails (mostly flat, three short climbs) before the battery reached 37.9V (3.2 V/cell).

So, I would say with moderate climbing and 20mph in PAS 2-3 (500W-750W), the battery provides about 57 miles or 90 km of range for me (20Wh/km)

In line with my previous expectations.

[scrambler]

Purpose:
1. Provide a higher max gear ratio so the bike can be comfortably pedaled up to 35 mph instead of the 22mph currently.
With a 13T rear cog and the 34T front sprocket, a 75 rpm pedaling cadence will cover from 8mph to 35mph.
The overall pedaling ratio range goes from 1.06 – 3.09 with the derailleur to 1.26 – 5.41 with the Kindernay and a 13T rear cog
2. Reduced number of gears more appropriate for a bike of this power.
3. Get rid of the derailleur with a more robust and protected system, requiring no maintenance, and allowing to shift gear when stopped.

System used:
• 148 mm dropout / 12mm thru axle Kindernay VII, with 6 bolts rotor mount and 7 bolts swap cage mount (32H)
• Kindernay XIV brake rotor, 7 bolts - 216mm with brake caliper spacer (203-216)
• Kindernay standard torque arm
• Kindernay Onesie Hydraulic Shifter
• Surly Singleator Tensioner for the pedaling chain
• New 3/32 Chain and 3/32 13T sprocket

Feasibility / Design:
After getting the nominal dimensions of the existing Hub assembly from BoxxBike and the K VII IGH from Kindernay, I used Sketchup to draft a Custom carrier to mount the Rotor and Motor Sprocket on the left side of the hub, as well as a custom Torque arm extension to accommodate the 160mm dropout of the Boxxbike vs 148mm of the Kindernay.

The BB 160mm Rear hub assembly is based on a 142mm hub with a custom extension on the left, composed of a custom carrier for rotor and motor sprocket and a spacer. This makes it possible to recreate using the 148mm Kindernay, with an adapter that holds both the brake rotor and the large 97T Motor Sprocket and a torque arm extension playing the role of spacer.
Because the torque arm needs to be extended to the left of the dropout, the best solution was to use the Kindernay XIV brake rotor which uses the 7 bolts over a larger diameter.
This provides clearance for the torque arm extension, as well as two sets of bolts to attach the adapter to the hub, distributing the effort of the motor Sprocket and rotor on both the 7 M4 bolts and the 6 M5 ones.
The torque arm extension is made of the existing torque arm cutoff and pinned to a Spacer to provide locking on the Hub, then of a custom L bracket pinned to the other side of the Spacer and anchoring on the bottom of the left dropout

On the right side of the hub it called for a small spacer, so I decided to use a Wedge lock washer, https://www.mcmaster.com/91812A474/ which also helps prevent the Hyseq from rotating under shifting if axle is not tight enough.
Note: the BB CAD shows a small recess in the right dropout that did not exist on my bike.

Interestingly enough, it places the center of the wheel at exactly the same place as on the provided CAD of the BB hub, and the chain line is very close to the BB nominal chain line.

Note: once I mounted everything together, I realized I had missed a small interference between the larger Kindernay brake rotor and the front Caliper mount.
I filed the edge of the mount a bit, but decided to also move the Brake rotor 1.7mm to the right as the caliper allowed it (I had a local CNC shop shave 1.7mm off the custom carrier).
This provided ample clearance with the caliper front mount, but did make the clearance between the caliper hydraulic connector and the motor chain workable but small.
In the end if I was to redo it, I would only move the Brake rotor 1mm In, to better distribute clearances.
EDIT: Another and probably better solution, would actually be to use the smaller 193mm Kindernay XIV rotor



Below is the 3D CAD assembly



Parts:
Adapter and spacer were modeled in the free to use Onshape online Surface modeler, allowing me to generate a Step model needed by CNC Shops.
Parts were manufactured in Aluminum 7061 T6
Custom Carrier:



Torque Arm Extension:



Miscellaneous Hardware notes:
The Kindernay XIV 7 bolts brake rotor has holes slightly bigger than the M4 screws (they must be using shoulder screws on the XIV hub). There are no 50mm long M4 screws with a small shoulder. So I found 2mm spacers, which once slotted with a Dremel would spring inside perfectly.
https://www.mcmaster.com/94669A002/
Slotted spring pins for the Torque arm extension:
https://www.mcmaster.com/91611A185/
7x M4 x 50mm screws with small washers and mounted with Loctite 243 at 3Nm
6x M5 screws as provide by Kindernay mounted with Loctite 243 at 6Nm
6x M6 screws with Nylstop locknut and washer for motor sprocket at 9Nm

Wheel lacing, truing and tensioning:
The Kindernay wheel lacing process is very well detailed in the EMTB thread below
https://www.emtbforums.com/community/th ... ion.18818/

The BoxxBike original wheel uses plain 12G spokes (2.6mm). The Kindernay hub normally maxes at 13G (2.3mm). Although using high performance 13G spokes would probably be OK, I decided to go for the new Sapim E-Strong 12G/13G butted spokes that actually has a higher strength than their 12G spokes. This means I had to enlarge the Kindernay hub holes a bit (to 3mm). I was careful to enlarge the hole towards the center of the hub, as not to reduce the material on the outside where it counts.

I was initially going to lace the wheel myself, then have it trued and tensioned at my LBS, but there was a 2 weeks delay, so I decided to give it a go myself. Using the swing arm as a truing stand it went actually fairly easily and also allowed me to register the original wheel centering to make sure the new wheel would match.
https://1drv.ms/v/s!At3vMAQjaOZLlX5ZCfl ... B?e=Ktaoul
https://1drv.ms/v/s!At3vMAQjaOZLlX-6Cka ... y?e=NnPNxf

This actually revealed a small discrepancy between the BoxxBike CAD drawing and the actual wheel position.
BB CAD shows a rim centered inside the hub flanges, but in fact the rim is offset to the right by a few mm.
I simply laced the new wheel with the same small offset.
This means that if I was to redo it, I would eliminate the washer on the right side of the Kindernay, and increase the width of the Carrier on the right by the same amount, shifting the hub and rim right, reducing the need to lace asymmetrically.

For spoke tensioning, I bought a tension meter https://smile.amazon.com/dp/B07YQQQC8L? ... ct_details on Amazon, and used the excellent Parktool WTA web app https://www.parktool.com/wta to record and tune the spokes tension.
Note: BB could not provide a tension value as they said they do it by feel, so to decide on a tension value, I tested the existing BB wheel, and I settled on 100Kgf. But doing that I noticed the spoke tension on the BB wheel was extremely uneven from Spoke to spoke, so I did a truing and tensioning pass on the original wheel to clean it up


Assembled Wheel:







Brake caliper:
The Boxxbike Caliper mount has a front mount that is too short, so on my wheel they had used like 6 flat washers to adjust the position of the caliper… the rear mount also has a weird angle which is not great for the screw direction…

The Kindernay 7 bolts rotor is larger than the standard rotor (216mm vs 203mm) and Kindernay provides a spacer that also has an arm for their torque arm (I Cut that off as it is not needed). The spacer was a little too high on both mounts. The good thing was that it allowed me to get rid of most of the washers in the front, but I did have to reduce the thickness at the rear a bit, so the caliper would sit at the right height.

Not having the professional tools normally required to perfectly level the two mounting surfaces, I had to make due with a vise and a file. I remembered why the first thing they taught me in mechanical engineering school was to file a metal surface flat by hand



Pedaling Chain tension:
BoxxBike has a concentric Swing Arm, so there is no variation of the chain path length during the rear wheel travel. That said, you do need a bit of slack in the chain so you can remove it more easily when the wheel has to come off. So you do need a tensioner.
Kindernay has a dual wheel tensioner which I found overkill and too big for my taste, so I bought a Surly Singleator and mounted it pressing up, which makes for a very stealth configuration.







Shifter installation:
There was only one hole available in the under frame cable system, so I slotted the one hole to pass both cables side by side, clipping them with each other.







Below is the Online folder with all the information.
I will upload a revised version of the Custom Carrier Step Models.
https://1drv.ms/u/s!At3vMAQjaOZLlXzVNSG ... h?e=Sti88z

Other articles
Purchase report:
viewtopic.php?f=3&t=94174&start=75#p1675901

Customizations:
viewtopic.php?f=3&t=94174&start=100#p1676907

Torque sensing pedal assist:
viewtopic.php?f=3&t=94174&start=100#p1679413
viewtopic.php?f=3&t=94174&start=100#p1686510

Regenerative braking:
viewtopic.php?f=3&t=94174&start=100#p1690743

Kindernay IGH conversion:
viewtopic.php?f=3&t=94174&start=100#p1709402

[sn0wchyld]

Brilliant work mate... just one question...

How much?

Purpose:
1. Provide a higher max gear ratio so the bike can be comfortably pedaled up to 35 mph instead of the 22mph currently.
With a 13T rear cog and the 34T front sprocket, a 75 rpm pedaling cadence will cover from 8mph to 35mph.
The overall pedaling ratio range goes from 1.06 – 3.09 with the derailleur to 1.26 – 5.41 with the Kindernay and a 13T rear cog
2. Reduced number of gears more appropriate for a bike of this power.
3. Get rid of the derailleur with a more robust and protected system, requiring no maintenance, and allowing to shift gear when stopped.

System used:
• 148 mm dropout / 12mm thru axle Kindernay VII, with 6 bolts rotor mount and 7 bolts swap cage mount (32H)
• Kindernay XIV brake rotor, 7 bolts - 216mm with brake caliper spacer (203-216)
• Kindernay standard torque arm
• Kindernay Onesie Hydraulic Shifter
• Surly Singleator Tensioner for the pedaling chain
• New 3/32 Chain and 3/32 13T sprocket

Feasibility / Design:
After getting the nominal dimensions of the existing Hub assembly from BoxxBike and the K VII IGH from Kindernay, I used Sketchup to draft a Custom carrier to mount the Rotor and Motor Sprocket on the left side of the hub, as well as a custom Torque arm extension to accommodate the 160mm dropout of the Boxxbike vs 148mm of the Kindernay.

The BB 160mm Rear hub assembly is based on a 142mm hub with a custom extension on the left, composed of a custom carrier for rotor and motor sprocket and a spacer. This makes it possible to recreate using the 148mm Kindernay, with an adapter that holds both the brake rotor and the large 97T Motor Sprocket and a torque arm extension playing the role of spacer.
Because the torque arm needs to be extended to the left of the dropout, the best solution was to use the Kindernay XIV brake rotor which uses the 7 bolts over a larger diameter.
This provides clearance for the torque arm extension, as well as two sets of bolts to attach the adapter to the hub, distributing the effort of the motor Sprocket and rotor on both the 7 M4 bolts and the 6 M5 ones.
The torque arm extension is made of the existing torque arm cutoff and pinned to a Spacer to provide locking on the Hub, then of a custom L bracket pinned to the other side of the Spacer and anchoring on the bottom of the left dropout

On the right side of the hub it called for a small spacer, so I decided to use a Wedge lock washer, https://www.mcmaster.com/91812A474/ which also helps prevent the Hyseq from rotating under shifting if axle is not tight enough.
Note: the BB CAD shows a small recess in the right dropout that did not exist on my bike.

Interestingly enough, it places the center of the wheel at exactly the same place as on the provided CAD of the BB hub, and the chain line is very close to the BB nominal chain line.

Note: once I mounted everything together, I realized I had missed a small interference between the larger Kindernay brake rotor and the front Caliper mount.
I filed the edge of the mount a bit, but decided to also move the Brake rotor 1.7mm to the right as the caliper allowed it (I had a local CNC shop shave 1.7mm off the custom carrier).
This provided ample clearance with the caliper front mount, but did make the clearance between the caliper hydraulic connector and the motor chain workable but small.
In the end if I was to redo it, I would only move the Brake rotor 1mm In, to better distribute clearances.



Below is the 3D CAD assembly



Parts:
Adapter and spacer were modeled in the free to use Onshape online Surface modeler, allowing me to generate a Step model needed by CNC Shops.
Parts were manufactured in Aluminum 7061 T6
Custom Carrier:



Torque Arm Extension:



Miscellaneous Hardware notes:
The Kindernay XIV 7 bolts brake rotor has holes slightly bigger than the M4 screws (they must be using shoulder screws on the XIV hub). There are no 50mm long M4 screws with a small shoulder. So I found 2mm spacers, which once slotted with a Dremel would spring inside perfectly.
https://www.mcmaster.com/94669A002/
Slotted spring pins for the Torque arm extension:
https://www.mcmaster.com/91611A185/
7x M4 x 50mm screws with small washers and mounted with Loctite 243 at 3Nm
6x M5 screws as provide by Kindernay mounted with Loctite 243 at 6Nm
6x M6 screws with Nylstop locknut and washer for motor sprocket at 9Nm

Wheel lacing, truing and tensioning:
The Kindernay wheel lacing process is very well detailed in the EMTB thread below
https://www.emtbforums.com/community/th ... ion.18818/

The BoxxBike original wheel uses plain 12G spokes (2.6mm). The Kindernay hub normally maxes at 13G (2.3mm). Although using high performance 13G spokes would probably be OK, I decided to go for the new Sapim E-Strong 12G/13G butted spokes that actually has a higher strength than their 12G spokes. This means I had to enlarge the Kindernay hub holes a bit (to 3mm). I was careful to enlarge the hole towards the center of the hub, as not to reduce the material on the outside where it counts.

I was initially going to lace the wheel myself, then have it trued and tensioned at my LBS, but there was a 2 weeks delay, so I decided to give it a go myself. Using the swing arm as a truing stand it went actually fairly easily and also allowed me to register the original wheel centering to make sure the new wheel would match.
https://1drv.ms/v/s!At3vMAQjaOZLlX5ZCfl ... B?e=Ktaoul
https://1drv.ms/v/s!At3vMAQjaOZLlX-6Cka ... y?e=NnPNxf

This actually revealed a small discrepancy between the BoxxBike CAD drawing and the actual wheel position.
BB CAD shows a rim centered inside the hub flanges, but in fact the rim is offset to the right by a few mm.
I simply laced the new wheel with the same small offset.
This means that if I was to redo it, I would eliminate the washer on the right side of the Kindernay, and increase the width of the Carrier on the right by the same amount, shifting the hub and rim right, reducing the need to lace asymmetrically.

For spoke tensioning, I bought a tension meter https://smile.amazon.com/dp/B07YQQQC8L? ... ct_details on Amazon, and used the excellent Parktool WTA web app https://www.parktool.com/wta to record and tune the spokes tension.
Note: BB could not provide a tension value as they said they do it by feel, so to decide on a tension value, I tested the existing BB wheel, and I settled on 100Kgf. But doing that I noticed the spoke tension on the BB wheel was extremely uneven from Spoke to spoke, so I did a truing and tensioning pass on the original wheel to clean it up


Assembled Wheel:







Brake caliper:
The Boxxbike Caliper mount has a front mount that is too short, so on my wheel they had used like 6 flat washers to adjust the position of the caliper… the rear mount also has a weird angle which is not great for the screw direction…

The Kindernay 7 bolts rotor is larger than the standard rotor (216mm vs 203mm) and Kindernay provides a spacer that also has an arm for their torque arm (I Cut that off as it is not needed). The spacer was a little too high on both mounts. The good thing was that it allowed me to get rid of most of the washers in the front, but I did have to reduce the thickness at the rear a bit, so the caliper would sit at the right height.

Not having the professional tools normally required to perfectly level the two mounting surfaces, I had to make due with a vise and a file. I remembered why the first thing they taught me in mechanical engineering school was to file a metal surface flat by hand



Pedaling Chain tension:
BoxxBike has a concentric Swing Arm, so there is no variation of the chain path length during the rear wheel travel. That said, you do need a bit of slack in the chain so you can remove it more easily when the wheel has to come off. So you do need a tensioner.
Kindernay has a dual wheel tensioner which I found overkill and too big for my taste, so I bought a Surly Singleator and mounted it pressing up, which makes for a very stealth configuration.







Shifter installation:
There was only one hole available in the under frame cable system, so I slotted the one hole to pass both cables side by side, clipping them with each other.







Below is the Online folder with all the information.
I will upload a revised version of the Custom Carrier Step Models.
https://1drv.ms/u/s!At3vMAQjaOZLlXzVNSG ... h?e=Sti88z

[scrambler]

Thanks, it was a fun project.

The Kindernay VII is 1,000 euros.
The CNC parts about $250.

And then all the tiny bits

Not cheap for sure............

[Enric]

Nice project and ebike scrambler.

I really like the boxxbike, I think it could fit my needs.

But I want to be able to share the route with my friends who ride 250 or 300cc thermal enduro bikes.

I'm not worried about power or being slower than them, in fact their routes are closer to trial than enduro, but my concern is range.

I think that with 1800wh of battery I will not be able to finish the routes with them.

And that is why I want to know if it is possible in some easy way to open the battery compartment to replace it with one that I carry in my backpack, or do a bit of DIY to put a connector to which to connect the battery that I carry in my backpack with a wire.

regards