My attempt at a lightweight EMTB with torque based PAS

HrKlev

1 kW
Joined
Dec 6, 2019
Messages
328
So Im redoing this thread, too messy in last attempt....

Ive lately starting to get interested in single trail riding, and have 2 bikes equipped with tsdz2 motors. One fatbike for the easy rides, and one full suspension Liteville for going up the steep stuff, and trying to get back down without breaking stuff. I mostly use the FS bike in the summer. I like pedaling (well, sort of. I need the exercise, and I get lazy if I have a throttle), so I enjoy the torque sensored lightweight tsdz2. That is, until they started breaking. The spindle broke on the fatbike, and the tsdz2 on the FS bike let out the magic snoke at 982km. There's been some other small bugs with the tsdz2's also, enough for me that I want to move on to something else. Ive learned so much from this page, so I want to make a build thread, in case someone else is looking for something like this.

The easy solution is to buy a commercial emtb, PaulID's excellent kit, or wait and see how the lightest.bike kit will be, but knowing myself, I will never stop customize the bike, so Id rather put something together myself.

Build goals:
- Mid drive
- lightweight (sub 20kg on the FS bike)
- torque sensored
- 6-700w peak power

First problem is finding a solution for the torque sensor. The power from the motor must not affect the sensor, and I need a freewheel. Luckily, the CYC X1 kit has this, and parts are readily available in their webshop. So I got one for the FS bike.

I will use a cycle analyst to process the torque sensor signals and display information.
 
Its been a little back and forth with motors on this bike, but when the tsdz2 burned, I needed to put something together on the FS bike again. I already had a torque sensor and freewheel from the CYC X1, a cycle analyst, and a MY1018 brushed motor. So I converted a bracket to fix the motor to the frame.

IMG_31082020_101731_(1000_x_750_pixel).jpg

Here's the torque sensor and freewheel. I used an old chainring from a cyclone kit to make an adapter from whatever BCD the freewheel is to 104bcd, and attached a 32T and 44T chainring to it

IMG_04082020_231349_(750_x_1000_pixel).jpgIMG_04082020_210508_(1000_x_750_pixel).jpg

The 44T was later replaced with a 48T. After a lot of tweaking with the cycle analyst the bike behaves quite good now. But, the MY1018 is very loud, and not very efficient when it is not in the peak powerband... The tsdz2 got me a lot longer on the same battery. Doing drops and jumps without fearing that the spindle would break was awesome, though! The added ground clearance is also a very welcoming change.

IMG_31082020_101740_(1000_x_750_pixel).jpg

The bike now weights 20.2 kg, so not too bad. The MY1018 must go, though. I have decided to go with a small, geared hub motor that will be mounted as a mid drive, and a baserunner controller. The baserunner has a unique(?) feature to simulate an electric freewheel for use in the clutchless GMAC. I think this will work very nice to keep the motor running and keeping tension in the driveline when not pedaling, so I get instant reaction without the jerk that happens when the motor catches up with the driveline. I have also ordered a Bafang G310 (2.5kg and a AKM/cute XS75 (1.4kg). If the XS75 is powerful enough, I will use it for the FS bike, and the Bafang for the fatbike. Because of reasons, the completition of orders and shipping takes forever, though, and in the meantime Im grinding away on the MY1018.
 
Got some parts this week! Together with the baserunner, I got a bafang g310 and a aikema 75sx. The "big" one is the bafang... the aikema is seriously small!

IMG_12092020_210120_(1000_x_750_pixel).jpg

Decided to try the aikema on the FS bike. Its only 1.4kg, so if it has enough power it will be a nice, lightweight setup. Took a 18 sprocket from an old 8spd cassette I had laying around and bolted it to the side cover, and mocked up some motor mounts. I might have made it a little stronger than I had to, but after everything is welded, the motor and the mount was 2060 grams, so not too bad.

IMG_12092020_210133_(1000_x_750_pixel).jpg

IMG_12092020_210234_(1000_x_750_pixel).jpg

Taped up the extra wires and battery temporarily and went for a test run

IMG_12092020_210257_(1000_x_750_pixel).jpg

Unfortunately one of the hall sensors are either not working, or the wiring in the aikema plug is different from the grin plug, so I had to run it sensorless. I didnt have enough time to play around with it to get a smooth start, so it came alive with a clunk everytime I stopped. Once it was up to speed, it felt pretty awesome! Thanks to the virtual freewheel feature on the baserunner, the response was insane compared to what I am used to from the tsdz2. It was also dead silent, and seemed to pull 6-700 watts without any problem. Just need to figure out the halls now, to get the standing starts dialed in. Then I can do some testing to see if it can take the beating. Bike weights in at 19.6kg now, including pedals and everything.
 
So I cant get the halls figured out. Starting to think its the controller that is the problem. Its the blue hall that is not responding, and if I plug in the brand new bafang motor, I get the exact same error. When I was messing around with trying to make the sensorless startup smoother, the last hall sensor suddenly started working and it was sooooo smooth! Then after a few minutes it was gone againg, and back to the cogging starts.... So its running sensorless atm. Any tips are welcome!

Anyway, I installed a temp sensor in the aikema motor. Turns out it has a speedo hall located like the bafang motir, so I just cut it out and soldered in an NTC resistor. Not pretty, but it works!

IMG_16092020_205656_(1000_x_750_pixel).jpg
IMG_16092020_205629_(1000_x_750_pixel).jpg
IMG_16092020_205538_(1000_x_750_pixel).jpg

Took it out for a small test and it turns out it runs a bit hotter than I was hoping at 5-600w continous. I stopped it at 90 degrees C while it was still climbing. So after reading Grins oil cooling experiment, I went out and poured 35ml of ATF into one of the holes for the brake disc. And voila, it could shed some of that heat. To my suprise the unloaded current went from 1.1A to 0.9A. The motor was a lot hotter to the touch, and internal temp was lower, like I was hoping. I did a quick test and went for a ~300m climb I used to do with the tsdz2. At max 500w assistance, the temperature seemed to stabilize at 80C, so it should take a few hundred watts more for a while if I feel like it. I dont know how much heat a geared hub motor can take, but I think this temperature is acceptable. Ive set the limit to start at 85C, and cut-off at 95C.

The decent was much more fun now that I dont constantly think about the tsdz2 spindle breaking after a drop....

IMG_16092020_205715_(1000_x_750_pixel).jpg

Got myself a 3d printer that I plan to make a battery box with, and some covers. Just need to learn how to donit.... Also definately need to get the halls to work. Might need to seal the motor a bit better. And some paint and clean-up in the wiring mess, and I think I'll be happy with it for a while.
 
Wow, nice lightweight setup! Should be great fun on the singletrack descentes :)
Good bikes those German Litevilles, I did owned some Nicolai bikes, great craftmenship as well ...
 
Thanks! Yes, Im very happy with the Liteville. Its my first "proper" full suspension, so I dont have much to compare it with, though. Im quite amazed how much gnarly stuff these shocks can take up. Its all rocks around here, so just need to be brave enough 😁

Today I found a 10K resistor and wired up a 5V to the halls line this diagram (think I found it on ES, so thanks to anyone who made it):
NktFBwJ.jpg

As suspected, everything checks out and all hall sensors works. So now I just need to wait for the customer service at Grin to answer..... sigh..... I want to cut off the connector on the controller and solder on another one in case that is the problem, but I guess I just have to wait in case its a warranty claim.
 
Super interesting setup! Really impressive.

Do you have any idea if you are running the AKM motor at peak efficiency RPM wise or not?
 
Thanks! Im not really sure how how the effiency curve looks like, but it is rated for 201rpm, so I guess it should enjoy life around there... It is spinning 264 rpm unloaded on full charge, and I am currently running a 19-42t sprocket-chainring setup. I like to keep around 85-90rpm when I ascent on fireroads, etc (tecnical climbs will be more variable of course), so its spinning 180-190 rpm when I am at max power. Im running it at max 550w atm, and it keeps up with my normal cadence as long as its not below 30-40% battery, then power starts to drop, and I have to lower my cadence to get full support. I have a 38t and 36t chainring on their way that I will test out, to see if it can maintain full support in a good cadence for me. Then I will try out field weakening, to get full support on the small, tecnical steps where you have to pedal like crazy to build momentum.

Im quite impressed with this little motor. SO quiet. Just hope it is reliable. Ill order a set of gears, just in case.
 
It seems that you are reasonably close to peak efficiency, then :)

Your setup is honestly very inspiring - might try to replicate you at some point.

Definitely think field weakening is a good option for that use-case!
 
Thats a cool build, havent seen it before! I need at least 500w for 15-20 minutes more or less continiously to keep up with the normal "250w" storebought ebikes in the ascends, though, so I think it would fall a bit too short. The motor Im using is only ~400g heavier (torquesensored BB, motor mount and battery is a lot heavier, though). Ill take the motor in the lathe an remove the spoke flanges and generally see if I can shave some weight when I have time to finalize the build this winter.

Im using 30A phase current, that was the lowest it would let me choose. Phasesuite says it equals 43NM. I guess that is current going into the motor, so I would need to rule out efficiency losses...? With the current 2.1 gear ratio it would be 90NM minus effiency losses at the crank. Commercial ebikes are around 70-85NM I think, and it feels about the same. I have bought a spare set of nylon gears that I want to have in hand before I try any more :)
 
For some motors, metal gears are available.

The nylon gears get soft in the heat long before the stator has any problems.

A hubsink-like device might be really useful here as well :)
 
It was hard enough finding nylon gears for this one, so I dont think anyone makes metal for it. I think its a quite new design on this motor, it has been a q75 motor that looks similar on the market for many years, but that was sensorless. This one has sensors, and a two-stage planetary reduction with helical first stage and a inrunner motor, just like the bafang g31x motors. And its really silent, like the new bafangs. So I'd rather not put noisy metal gears in it ;) Im not sure how much sure how much heat the gears can take, but I have programmed the CA to start throttling at 85C and turn iff at 95C. I made it up to 85C on the top of a 500m ascent with a 2.5X torque assist and max 550w, but it kept it below 90C, so no problem I think. I usually dont ride with that much assist, it was just for testing purpose.

I have some plans for cooling, just need some time to learn how to 3d print :)
 
Nice to hear! Aldo really interested in the life og the nylon gears. There seems to be conflicting information about whether or not the oil destroys them.

3D printning is not hard, but CAD is!
 
Fantastic work there and I'd love to see a video of it in operation and some more close up photos. Having those CYC parts available sure helps. What is the drag like when pedalling with the motor off, does the freewheel totally disconnect the drive system? I'm looking for something like that to extend my range on flat ground or if the battery dies. I'm also over the TSDZ2 and keen to see what Bikee delivers with their Lightest kit. I've been planning a project just like this in my head in the meantime so was nodding in agreement while reading. Really good to see someone putting it all together. :thumb:
 
famichiki said:
Fantastic work there and I'd love to see a video of it in operation and some more close up photos. Having those CYC parts available sure helps. What is the drag like when pedalling with the motor off, does the freewheel totally disconnect the drive system? I'm looking for something like that to extend my range on flat ground or if the battery dies. I'm also over the TSDZ2 and keen to see what Bikee delivers with their Lightest kit. I've been planning a project just like this in my head in the meantime so was nodding in agreement while reading. Really good to see someone putting it all together. :thumb:

Oops, sorry, I must have missed your post. Thanks! There's very little drag when the motor is off. You have to rotate the notor chain and the outer case of the motor, but there's not much resistance. There's no seal other than the bearings (which surprisingly is not leaking much ATF at all, even when the bike is on the side. Theres some moist close to the bearings, but not enough that it has been dripping). You can see the clutch mechanism inside the motor in the picture above where it is disassembled. With the electric freewheel enabled, it is using around 15-20W to keep the motor and chain spinning, so I can just set assistance to zero to get absolute no extra drag at all. Not that I feel much of a difference.

I have confirmed with the customer support at Grin that one of the hall sensor outputs in the baserunner is blown, so I need to send it back. However, I just received the Phaserunner for my other bike, so I will be using it on this bike until the baserunner is back. Looking forward to be able to use the hall sensors and get rid of the cogging from standing starts.
 
HrKlev said:
With the electric freewheel enabled, it is using around 15-20W to keep the motor and chain spinning, so I can just set assistance to zero to get absolute no extra drag at all. Not that I feel much of a difference.

So when you start pedalling it gives you a flat ~20W assist? Or does it vary up to that amount depending on the torque applied?
 
famichiki said:
HrKlev said:
With the electric freewheel enabled, it is using around 15-20W to keep the motor and chain spinning, so I can just set assistance to zero to get absolute no extra drag at all. Not that I feel much of a difference.

So when you start pedalling it gives you a flat ~20W assist? Or does it vary up to that amount depending on the torque applied?

You actually set a fixed phase current value in the phaserunner suite. I think I used 0.8A. It turns out to be around 15-20w battery power. That way the motor and the entire driveline keeps spinning when I stop pedaling. For instance if I must stop pedal for a very brief moment to avoid pedal strikes on the trail, the motor keeps the driveline engaged, and the responce I get when I keep pedaling is very fast. With the tsdz2 it used maybe a second to spin up again, and it was very annoying in technical terrain. Also, when you restart the pedaling, the engagement is super smooth without the "clunk" you get in a typical cyclone-like mid drive system. You can choose stall time, so when you come to a stop, it turns off, but if you briefly loose traction with the rear wheel in a decent, it spins up again and is ready to go when needed. I really like it a lot!
 
It sounds a bit like a walk mode that is always running, is that the general idea?
Do you feel any motor pull or overrun when you stop pedalling, or will it eventually coast to a stop?
Is there just the single stall time setting that times out after you stop pedalling, or does it require the speed to drop to zero? What do you have your stall time set at?

I'm considering adding something like this to the TDSZ2 but wasn't sure whether to make it a constant level or highly limit one of the dynamic modes. I've also thought about adding a "reserve tank" cut-out that saves enough battery to ride without drag for a while.
 
famichiki said:
It sounds a bit like a walk mode that is always running, is that the general idea?
Do you feel any motor pull or overrun when you stop pedalling, or will it eventually coast to a stop?
Is there just the single stall time setting that times out after you stop pedalling, or does it require the speed to drop to zero? What do you have your stall time set at?

I'm considering adding something like this to the TDSZ2 but wasn't sure whether to make it a constant level or highly limit one of the dynamic modes. I've also thought about adding a "reserve tank" cut-out that saves enough battery to ride without drag for a while.

The idea is that it is just powerful enough to keep the driveline spinning. If I lift the wheel it slows down slowly, because it then also drives the wheel. I dont notice it while coasting, it will still come to a stop, and the force is not enough that I notice it stalling. Originally it was made for hubmotors to be able to pedal the bike with zero drag, but it has an added benefit on mid drive systems that it keeps the driveline engaged, and the system kicks in once you start pedaling, instead of first having to wait for the motor to "catch up". It keeps the motor spinning until it stalls out, with or without pedaling.

I think 250ms is max, and that is what Im using. Sometimes I lock my wheel during decents, but most of the time it is short enough that it slowly spins up again.
 
How's this going, did you get the hall sensors sorted out yet? After looking over the thread again I realised that you lock the axles in the brackets and the the hub motor spins as it normally would in a wheel, is that correct? Did you give any consideration to fixing the motor in position and spinning the axle instead?
 
Ive been away for work for some time now, so not much happening.... Winter is coming, so I did sell my tsdz2 equipped fatbike before I left, and bought another one for a similar build, so that will be the next focus when I come home. And some 3d printing for the battery boxes, to finish off both bikes.

You are correct that I clamp the axles and spin the whole hub. These tiny hub motors has inrunner motors and dual reduction, so there's no simple way to convert them. I dont see the point either, actually. Its much easier this way, and if I need more cooling when summer returns, it will be more effective with the case spinning.

The baserunner was defective so I will return it and get it fixed or replaced. I bought a phaserunner for the fatbike, and it works perfect with the same motor :)
 
Had some time to start with the fatbike yesterday. I have decided to use the aikema motor on this bike too, Im really happy with it so far. So I will have to order another one before spring arrives and the Liteville bike will be my main bike again...

I did not remember to buy an adapter or a spare cyclone chainring, so I had to make the 104bcd freewheel spider by hand. That sucked! Got it within 0.3mm wobble, so it will do for now, but I will get another one water cut or something...

IMG_30102020_194351_(1000_x_750_pixel).jpg

Because of the huge drop-out on the fatbike I have decided to use the inner chainring for the motor to not have everything stick so far out on one side. The chainstays limits the size on the motor chainring to 32. No problem, just need to use a 16t on the motor.

IMG_30102020_194403_(1000_x_750_pixel).jpg

Had to tack it up on the bench this time because of carbon frame. Not sure if it likes welding splatter...

IMG_30102020_194420_(1000_x_750_pixel).jpg


IMG_30102020_194507_(1000_x_750_pixel).jpg

I tried to use some smaller diamater materials to save some weight this time, but it seemed a little to flimsy. So I added a crossbar. The part that rests against the frame also needed to have a big surface to hopefully keep the carbon from cracking. So after it was finished, it weighed ~400g, pretty much the same as my first one for the Liteville bike 😜

IMG_30102020_194528_(1000_x_750_pixel).jpg

Ran out of time, so painted it and called it a day. Hopefully I will get to test it next weekend.

Other news on the project is I have started to learn 360 fusion. So fun with 3d printing! Been through a couple versions of the battery box already, but I think the latest version is the one I will stick to. Will print it in a different color of course, the white color is just because thats what we had available that moment.

2 of these 3 pieces are using rivnuts in the frame, and will keep the box from rotating or sliding on the frame:
IMG_31102020_075350_(1000_x_750_pixel).jpg

Battery us quite snug in the box, no movement in there.
IMG_31102020_075409_(1000_x_750_pixel).jpg

Then the whole thing is secured with 3 huge tie wraps. I did use screws and clamps around the tube on a couple earlier revisions, but it was too bulky for my taste. Using tie wraps saved a lot of space. And I like tie wraps 😜IMG_31102020_075438_(1000_x_750_pixel).jpg

I will also print something to hide all the extra wire clutter, so I think it will be quite OK after some tinkering.
 
Back
Top