My RC Mid Drive with Single Stage Reduction

[FZBob]

If one was to implement a Torque Sensing PAS, would e-brake switches be required/desired?

My initial thought is that these would not be necessary. My brakes are far stronger than the motor, so I could stop any time, and just stall the motor if it ran away. I would prefer to keep things simple and save the money and effort. However, I have not actually ridden a PAS bike, so I may be missing something. I would like to hear any thoughts/input!

 

[Drum]

Hi FZBOB,
Great build, and cool that you tried different motors and setups until you settled on one that works for you.

Torque sensing Bottom Brackets: You are right.. the torque applied to the chainwheels by the motor will not be measured by the torque sensor in the bottom bracket, as it will be transferred directly to the chainwheel that is driving the back wheel.
The Thun units (and some others) measure the torque applied to the left-hand pedal only, which may mean that there is a slight delay in responding to an increase (or decrease) in pedaling torque. For "normal" riding they work ok, but for technical climbing it may be an issue.
Sempu and Bafang both make double-sided torque sensing bottom brackets, where the chainwheels are not mounted to the right-hand crank arm, but are on a separate spider mounted on a splined sleeve on the right of the bottom bracket. The crank arms both drive a solid spindle, and that spindle drives the output spline through a torque sensor. This reduces the potential delay in sensing a change in pedal torque.
More info in this thread: https://endless-sphere.com/forums/viewtopic.php?f=28&t=94553&start=25&hilit=sempu
Unfortunately, all the commonly available torque sensing bottom brackets are square taper, and I stopped expecting that system to carry my weight in rough going without wearing / breaking many years ago. Sempu do make (have made in the past?) a TS bottom bracket with ISIS crank interface, and offered to make one for me again if I asked for it specifically. I haven't moved on that because other projects have taken precedence, so can't confirm the availability.
I have no direct experience with implementing these units, but my understanding is the torque is measured and an analogue DC Volt output is created. I think you could program a VESC to work with the output voltage range.
The bottom brackets also need a DC input (from memory 5V or 12V), so you would have to provide that as well.

I would love to hear your thoughts if you do implement torque sensing.. I think you would like it.
Definitely test ride one of the modern Trek / Spec / Giant etc e-MTBs if you get a chance, preferably in similar terrain to where you normally ride. You might not appreciate the extra weight, but it would give you a good idea of the torque sensing control. Just be aware that some brands are tuned to bring in assist quickly, others ramp it in more smoothly.. all of this can be programmed, but it does make a difference to the feel of the bike.

 

[Drum]

Re E-brake switches: I would think for your setup they would not be required. If you stop pedaling, the motor assist would drop to zero quickly, and as you say, in the power range you are working in the back brake could handle the extra torque from the motor.
The TSDZ2 mid-motor system with torque sensing control does not have brake sensors.
I have ridden several bikes with those motors and it seems to work well without them.

 

[Grantmac]

CYC is developing a TS BB with ISIS splines and also a freewheeling crank using a sprag clutch.

 

[FZBob]

For testing/research purposes only, I'm thinking about wiring up a basic cruise control switch. It's not torque sensing, but might get me 1/3 of the way there, and is easy (cheap) to try. Might be nice on long, consistent climbs.

I found this diagram at https://goldenmotor.com/SMF/index.php?topic=2943.0



I would replace the pushbutton with a toggle switch, and the trimmer pot with a regular pot with a knob, adjustable while riding.

Thoughts?

I'm wondering if the diode in line with the throttle Hall Sensor output will cause a bit of throttle dead band due to the 0.7V forward voltage. Any thoughts? Or should I just wire it up and try it... Any suggestions on what diodes to buy?

Maybe wire a cap in for a soft start?

Also, any suggestions on where to buy small electronic components? Back in the day, Silicon Valley had R&D Electronics, which sold a wonderful selection of this stuff. Unfortunately, they are long gone...

 

[falcongsr]

Anchor electronics

 

[Drum]

The diode would drop 0.7V right through the range, so if you didn't reprogram the VESC you would probably lose power at full throttle. Presumably you could reprogram the "start voltage" and "full throttle voltage" in the VESC to match the reduced values caused by the diode?
I have read that there are diodes available that only drop 0.3V, but don't know anything else about them. If you could use them, the difference would be less.
You may have mentioned this before, but is your controller set up as "Torque mode" (Current output proportional to throttle voltage) or "Speed mode" (Motor RPM proportional to throttle voltage) ? The response from the Cruise Control circuit would be different in each case.

 

[Kepler]

Personally I think torque sensing and even a throttle is unnecessary on a low powered assist setups like this. When riding single track, reactions need to be instinctive. Pedal and brake are plenty to think about when tracks get fast and technical.

I think you are are on the right track with considering a form of cruise control however, forget about clunky analogue solutions. The Vesc lends itself so well to PWM inputs and as such a custom programmed Arduino style interface is hands down the way to go. And with the block programming solutions now available, even old dogs like me can program like a pro.

If I was using your setup, I would have a single button throttle and a brake sensor either just on the front brake or on both brakes.

The single button would activate 50% throttle (200W) with a one second push and 100% throttle (400W) with a three second push. Any short push would kill the throttle.

The brake sensor would activate in the free play area of the brake lever and zero the throttle when activated. Releasing the brake lever would return the assist to wherever you originally were.

I use this setup on my friction drive mountain bike ( except I go into regen rather then zero throttle) and the assist feels seamless.

Food for thought.

 

[PaulD]

Personally I think torque sensing and even a throttle is unnecessary on a low powered assist setups like this. When riding single track, reactions need to be instinctive. Pedal and brake are plenty to think about when tracks get fast and technical.

If I was using your setup, I would have a single button throttle and a brake sensor either just on the front brake or on both brakes.

The single button would activate 50% throttle (200W) with a one second push and 100% throttle (400W) with a three second push. Any short push would kill the throttle.

The brake sensor would activate in the free play area of the brake lever and zero the throttle when activated. Releasing the brake lever would return the assist to wherever you originally were.

It sounds like you are describing functions that are automatically taken care of by a torque sensor. Perhaps it depends if you're coming from a moto or bicycle background which is more intuitive and natural. With a torque sensor all you have to think about IS pedalling and braking and rarely both at the same time, Rather than fumbling for buttons. And in my opinion, any offroad bike over 200W should have a torque sensor or throttle rather than PAS and ebrakes or any kind of cruise control. Unless we are talking about non-technical riding, like fire roads.

 

[FZBob]

The diode would drop 0.7V right through the range, so if you didn't reprogram the VESC you would probably lose power at full throttle. Presumably you could reprogram the "start voltage" and "full throttle voltage" in the VESC to match the reduced values caused by the diode?
I have read that there are diodes available that only drop 0.3V, but don't know anything else about them. If you could use them, the difference would be less.
You may have mentioned this before, but is your controller set up as "Torque mode" (Current output proportional to throttle voltage) or "Speed mode" (Motor RPM proportional to throttle voltage) ? The response from the Cruise Control circuit would be different in each case.

You are correct, I just checked the VESC and Min & Max input voltage are menu items. So I should be able to dial out the diode drop. By the way, I'm using -70 Expo, and it makes the throttle very easy to modulate at low power. Below 100 watts, it will easily resolve +/-10 watts.

I'm using Torque Mode, and I like it a lot. Very natural. Way back when I ran the cyclone, it locked into an RPM, and never felt right.

 

[Kepler]

It sounds like you are describing functions that are automatically taken care of by a torque sensor. Perhaps it depends if you're coming from a moto or bicycle background which is more intuitive and natural. With a torque sensor all you have to think about IS pedalling and braking and rarely both at the same time, Rather than fumbling for buttons. And in my opinion, any offroad bike over 200W should have a torque sensor or throttle rather than PAS and ebrakes or any kind of cruise control. Unless we are talking about non-technical riding, like fire roads.

You need to take into account the current limiting the Vesc does for you as the duty cycle drops off. With slower cadence, current is reducing also. You can adjust the current to duty cycle effect by adjusting the motor amps to battery amps ratio.

With many hours on technical single track, I know how well this works. You are not fumbling for buttons, a single press and the system is armed. The zero duty cycle means the amps are limited by the Vesc to a low output and holding the brake lever kills assist completely. Release the brake and just start pedaling. Assist starts off nice and gentle and builds up as your cadence increases. Feather the brakes and the assist is completely off.Yes this is basically a speed PAS system but with only 200W, the bike will just feel light rather then assisted. The 400W setting only gets used on really steep climbs but still has the same current limiting interaction with duty cycle

Sure, a torque sensor would be ideal but probably not a practical solution for this build.

You would be surprised how well this setup works. Its not just some simple on off buttons and is far more intuitive then you would expect. No engineering needed and cost-effective to implement.

Don't get me wrong, i am not trying to sell anyone on this system. Just passing on what l think might be a good control alternative for this build.

 

[qwerkus]

Not sure why so many worship torque sensing. Perhaps I only rode cheap models, but have you ever tried to climb a mountain with torque sensing ? You basically have to push the pedals harder and harder in order to have the motor increase it's power as required, which puts stress on body parts that should be relieved by the presence of a motor in the first place (like my knees)... I find a 12mag+ (24 ist even better) PAS is the simplest and most effective way to control an electric assisted bike (where you still want to pedal, without throttle). The only place where torque sensors really shine is start and stop, as the motor responds almost instantaneously. But than again, with 12mags the delay is negligible. On a purely electric bike (moped or motorcycle), the problem would be different. But even there, I'd recommend a throttle over torque sensing.

For the rest: very interesting tests - thanks for sharing. Love the O ring part. Will try myself.

 

[FZBob]

Interesting discussion on control methods. It gives me something to ponder as I'm out climbing hills...

There are some things I like about the hand throttle. First, I need to limit the amount of leg power required due to my knee issues. Without a throttle, Torque Sensing would require me to pedal harder on steep sections, or turn the assist way up. On flat sections and shallow climbs where I don't use any assist, I would need to turn Torque Sensing off. Second, holding the throttle is an incentive to not use it unless needed. I get better battery mileage, and I can use the exercise...

I wonder if it would be possible to set up Torque Sensing so that if my legs are producing between 0 and 100 watts, I get no assist. From 100 to 180 watts of leg power, I get proportional assist (adjustable). From 180 to 200 watts of leg power, the electric motor would ramp to full power. Basically a rising assist curve, so that my knee never sees over 200 watts. This might mimic what I do with the throttle. Possibly use an Arduino?

Speaking of Arduino, I found this schematic for a Flux Capacitor. Just in case anyone needs one...

 

[qwerkus]

Interesting discussion on control methods. It gives me something to ponder as I'm out climbing hills...

There are some things I like about the hand throttle. First, I need to limit the amount of leg power required due to my knee issues. Without a throttle, Torque Sensing would require me to pedal harder on steep sections, or turn the assist way up. On flat sections and shallow climbs where I don't use any assist, I would need to turn Torque Sensing off. Second, holding the throttle is an incentive to not use it unless needed. I get better battery mileage, and I can use the exercise...

I wonder if it would be possible to set up Torque Sensing so that if my legs are producing between 0 and 100 watts, I get no assist. From 100 to 180 watts of leg power, I get proportional assist (adjustable). From 180 to 200 watts of leg power, the electric motor would ramp to full power. Basically a rising assist curve, so that my knee never sees over 200 watts. This might mimic what I do with the throttle. Possibly use an Arduino?

Speaking of Arduino, I found this schematic for a Flux Capacitor. Just in case anyone needs one...


Flux Capacitor c.jpg

No need for that. A properly programmed controller can do it. Check out casainho's open source firmware for the tsdz2. It's a torque sensor based motor, and his team managed to reverse engineer the code. As a result, the motor is much more reactive, which means less knee pain (I can relate to that!). Still, you have to pedal to start the motor, and at least for me, this is not an option on steep climbs anymore. Throttles are illegal in most part of europe, so not really an option for me, but maybe for others. What I don't like about the throttle option is that it weakens your grip on the handlebar. When there are already 2 fingers taken by the brake levers, only the thumb really has a grip on the bar; if you use it to control the throttle, it kind of feels like riding with one hand only.

 

[FZBob]

One of my initial concerns with this build was reliability of the #25 chain. After 750 miles the chain wear finally reached over 1%. I never clean chains, but I do lube them before every ride (Rock & Roll Gold lube).

It seemed like the last 50 miles it was wearing faster, and felt maybe just a bit more draggy than usual. The pictures show that the little 9T sprocket has lived a full life, and the chain is worn. I have seen worse, especially on motorcycle chains, but it was time to replace. I didn't see any signs of failure (breakage). I went for a ride yesterday, and the bike was definitely quieter, and felt a bit peppier (maybe psychological...) Next time, I will replace the chain and small sprocket at 700 miles. I'm happy with that, as it's on par with what I get out of my bike chains. They do operate in a very harsh dusty environment...

I got 10' of HD chain for $50 and used 1/3 of it. The 9T sprocket was $7. So replacement cost is about $24. Again on par with normal bicycle chain maintenance.

By the way, the electricscooterparts chain breaker works really well. The pins are not peened over, and press out easily. I think in the future I may not need a master link.





View attachment 2

 

[fixvid]

I started doing my own version of FZBob light ebike, it will be a Cannondale Scalpel 29" with D5035 motor, 5s LiFePo battery and Turnigy ESC.

My aim is also as light as possible. I have another bike with Shimano M8000 edrive and it's really nice but frames built for this motor are sick heavy ..and it doesn't have a throttle.

I started design a mount around the frame with a 3D printer which i acctually succeded to 'hug' the complex geometry of the frame tube. The first prototype with low infill can be seen in this picture:

I'm still waiting for the motor pulley for more than a month .. i cannot fix the final position of the motor until then. The 89t chainwheel is heavy steel, i managed to cut it down to only 270gr, i will need that light aluminum chainwheel which FZBob has.

The second step was designing a 32700 LiFePo4 enclosure. I adopted a new strategy using no spot-welding and just a little soldering on the positive side. I 3d printed an enclosure with multiple nickel strips on the negative side. I draw 20 amps from it and besides battery getting warm, no connection to the battery was hot.

I would like to thank FZBob for the light ebike ideea, i was thinking for a long time of it, i just didn't know how to do it.

 

[FZBob]

I
I started design a mount around the frame with a 3D printer which i acctually succeded to 'hug' the complex geometry of the frame tube. The first prototype with low infill can be seen in this picture:

I'm still waiting for the motor pulley for more than a month .. i cannot fix the final position of the motor until then.

I like that motor mount a lot. It does not require drilling the frame like mine did. Any chance you could post the weight?

There might be a nice little side business here for you 3D printing mounts. You could have people send you paper patterns of their frame tubes, and build up a library of shapes. (I would consider buying one if I decide to convert my nice bike...)

I did something similar back when I did my cyclone.

It was aluminum and weighed a ton (500g). Also a pain to fab, as my mill is not CNC.

A couple of suggestions. There can be up to 175 Lb of tension on the chain, trying to slide the mount down the tubes and distort the front plate where the motor mounts to. My Cyclone mount did have a tendency to slide down the frame. One thing I did was to glue my mount to the frame with Silicone. However, my frame is anodized, so no chance of damaging the paint. Your paint looks nice, so I would be careful. If you have a sliding issue, you might try thin double sided tape. That might also serve to keep the dust out from between the frame and the mount, and prevent paint damage. I also tried a strap from the mount to the water bottle cage mount, and that worked well without silicone.

You may want to consider making the front mount plate area thicker, and/or using a metal doubler on the front, as that is the area of greatest stress, and may twist. Also, a metal doubler would give a nice seat for the mount screws.

Did you try Electric Scooter Parts for the pulley? I got one a couple of weeks ago in 2 days for about $7. (Part # SPR-2509D).

 

[fixvid]

I thought about the mount sliding and twisting and solved it like this :

The front plate is one area which could have some improvements, but without pulley i still have to delay this test. If it bends I can adjust the design to counter it. Putting a metal L plate would be a lot more work but stronger for sure.

I don't know the weight because it is just a prototype with low infill and I haven't measured it. It's featherweight now. If I bump the infill to 100% where is needed (the front plate) I think it will be still very light. I was thinking that under the frame I need just 2 small half clamps, no reason for 1 big half clamp. So I can play a lot with it to get the maximum strength to weight ratio.

If someone's interested I can share the stl files but they work just for Scalpel 29er frame.

I am from Romania and thought that ordering from electricscooterparts will be harder and more expensive so I choose the Chinese.

My chain Wheel is shitty and it still has 270gr:

It's so thick that I have to mount it on the outside of the crank set. Not an elegant solution but it should work.

Regarding batteries, I still don't know about what should be a good wh rating, but I can tell that the Shimano m8000 with 500wh battery pack goes like 100km easily(mainly on flats).

The m8000 motor is rated 250w.. But I'm not sure about this one. With right gearing, 30t chain wheel, 46 sprocket, it can climb anywhere. I was climbing hills so steep that I could barely hold my balance on feet. For this you have to have a lot of traction on the back and you have to be over the handlebar. I think the motor sometimes pulls 500w but I cannot measure this because everything is enclosed. The battery is 10s so I cannot imagine how it could pull just less than 10A.

Getting back to the rc motor with 5s lifepo4 that's 16v with 20 to 30amp limit that's around 320w to 480w. That means it can match Shimano M8000 power which has enough power for every kind of hill. Buuuut.... Our rc mid drive is way less kilos than the M8000. We're getting somewhere here. 8)

 

[FZBob]

I thought about the mount sliding and twisting and solved it like this :

The front plate is one area which could have some improvements, but without pulley i still have to delay this test. If it bends I can adjust the design to counter it. Putting a metal L plate would be a lot more work but stronger for sure.

I don't know the weight because it is just a prototype with low infill and I haven't measured it. It's featherweight now. If I bump the infill to 100% where is needed (the front plate) I think it will be still very light. I was thinking that under the frame I need just 2 small half clamps, no reason for 1 big half clamp. So I can play a lot with it to get the maximum strength to weight ratio.

My chain Wheel is shitty and it still has 270gr:

It's so thick that I have to mount it on the outside of the crank set. Not an elegant solution but it should work.

Regarding batteries, I still don't know about what should be a good wh rating, but I can tell that the Shimano m8000 with 500wh battery pack goes like 100km easily(mainly on flats).

The m8000 motor is rated 250w.. But I'm not sure about this one. With right gearing, 30t chain wheel, 46 sprocket, it can climb anywhere. I was climbing hills so steep that I could barely hold my balance on feet. For this you have to have a lot of traction on the back and you have to be over the handlebar. I think the motor sometimes pulls 500w but I cannot measure this because everything is enclosed. The battery is 10s so I cannot imagine how it could pull just less than 10A.

Getting back to the rc motor with 5s lifepo4 that's 16v with 20 to 30amp limit that's around 320w to 480w. That means it can match Shimano M8000 power which has enough power for every kind of hill. Buuuut.... Our rc mid drive is way less kilos than the M8000. We're getting somewhere here. 8)

You are lucky the water bottle screw is right under your mount! I agree that should eliminate any slipping or rotating. Nice clean design.

My motor chainwheel also mounts on the outside of the crank set. It is 1/4" (6.3mm) thick. I have a 40t on the inside (middle), and a 22t on the farthest inside for steep climbs. I have a 11-40t casette on the back, and with the motor I'm not fast on steep climbs but I'm not working too hard.

My battery is 4s, 8Ah. It goes flat at about 100 watt hours. I usually ride 1-2 hours, 10-15 miles, and climb 900-1500 feet. I have never ran out of battery before the end of the ride. I typically use around 150 watts on regular climbs, but occasionally see up to 500 watts. I do get a nice workout.

 

[fixvid]

Thankfully batteries are cheap these days.
I remember you wrote that 4s gave you the optimum pedal cadence and you really didn't want to go higher on voltage because of too much rpm at the crank.
I saw that tuning in the Vesc tool is endless and I was just thinking that you can use any kind of battery and just limit the throttle or pwm to what is your max crank cadence.
With this in mind I could use my bricked 500wh Shimano battery on really long tours (2.6kg) or use 18v makita battery which I have a ton(620g). I have even hilti 36v batteries which are on the heavy side 10s 3p 6ah.

Yesterday I stood all day at the side of my bicycle looking at it and thinking what would be a nice design for a box containing the vesc and the batteries. I have the first templates, it looks I found a good compromise. I will post some pictures when I have the first prototype.

 

[FZBob]

Thankfully batteries are cheap these days.
I remember you wrote that 4s gave you the optimum pedal cadence and you really didn't want to go higher on voltage because of too much rpm at the crank.
I saw that tuning in the Vesc tool is endless and I was just thinking that you can use any kind of battery and just limit the throttle or pwm to what is your max crank cadence.
With this in mind I could use my bricked 500wh Shimano battery on really long tours (2.6kg) or use 18v makita battery which I have a ton(620g). I have even hilti 36v batteries which are on the heavy side 10s 3p 6ah.

I think you may be correct about running more than 16V (I'm certainly not a controller expert!). I would recommend being very careful to make sure that max RPM is limited, as uncontrolled spinning pedals could do some damage... I've been whacked a couple of times just on 4S, and it got my attention...

General safety precaution: When working on the bike, if the battery is present, keep all body parts away from the pedal area! (Sorry for stating the obvious, but it's easy to forget and put your face near the motor while poking controller wiring. Guess how I know...)

 

[fixvid]

It's working:

First alpha version, still have to polish a lot of things, but it's working ..

First problem: i have to change the chainwheel to a nicely machined aluminium one, current one is heavy and i couldn't drill holes to be berfectly on 104mm circle, so it's not centered. Chain sometimes is slack, sometimes is tight.

Second thing which i am very hard thinking of is clutch/freewheel on the crank. When motor spins pedals should not spin. I have my knees busted and i just cannot pedal for long. I still wouldn't use any kind of heavy components, just something light and well tought for a bicycle. My bottom bracket is BB30 and im not sure someone invented one yet.

It shouldn't be rocket science to put a freewheel between the spindle and spider. Or better yet, just take a spider replace it with a freewheel and add the necessary mounting for 2 chainrings BCD 64 and BCD104. I cannot imagine it would be so hard to achieve this at a reasonable weight.

This is a powermeter, but the ideea is the same:

... and this should take care of shin wacking pedals.

 

[tentman]

Great work - watching with interest. It would certainly be great if someone can crack the freewheel problem for this sort of build

 

[fixvid]

Well, Rocky Mountain faced this issue a year ago and solved it:

That's a standard RaceFace crank with a special spider which have a freewheel incorporated.

 

[FZBob]

It's working:

First alpha version, still have to polish a lot of things, but it's working ..

First problem: i have to change the chainwheel to a nicely machined aluminium one, current one is heavy and i couldn't drill holes to be berfectly on 104mm circle, so it's not centered. Chain sometimes is slack, sometimes is tight.

Second thing which i am very hard thinking of is clutch/freewheel on the crank. When motor spins pedals should not spin. I have my knees busted and i just cannot pedal for long. I still wouldn't use any kind of heavy components, just something light and well tought for a bicycle. My bottom bracket is BB30 and im not sure someone invented one yet.

It shouldn't be rocket science to put a freewheel between the spindle and spider. Or better yet, just take a spider replace it with a freewheel and add the necessary mounting for 2 chainrings BCD 64 and BCD104. I cannot imagine it would be so hard to achieve this at a reasonable weight.

This is a powermeter, but the ideea is the same:

... and this should take care of shin wacking pedals.

Wow, that was quick!

My chains always seem to end up with tight and loose spots. I used to obsess over it. Now I'm used to it. Doesn't seem to cause me a big problem.

Before you spend a lot of time putting a freewheel inside the chainring, you might want to try riding with your feet off the pedals to see if the system has enough power without pedaling. I tried it briefly. See the second bike pic in my first post, leaning against the bike rack. For me, the motor seemed weak without me providing some assist by pedaling. However, I'm a big guy...