My attempt at a lightweight EMTB with torque based PAS

Vbruun said:

And the motors still hold Up fine?

Click to expand...

Yes, still doing fine! Had a ride with my friend who bought the tsdz2 fatbike, and the tsdz2 is really loud compared to the aikeema motor, which surprised me.

Have been learning how to use fusion 360 and the 3d printer, and its quite fun! Trying to piece together something that doesnt look like a trainwreck this time... I have just been using leftover pieces of PLA in different colors to print prototypes, and I think Im quite close to the final version now, so I will print everything in the same color soon!

I will take the fatty on the first ride tomorrow in the currebt state, to see how she does. My weight said 15kg before I touched it, and it now says 21kg, including pedals and everything the brands dont include. So Ive added 6 kg including everything. Now so bad. I still have heavy tubes in the tires, which will be removed when I put on the winter tires, so should be quite close to 20kg then.

I'll include pictures with some colorful prototype bits

How's the torque compare to the TSDZ2, strong enough to get up steep inclines without overheating? And how do you find the torque sensing in comparison? Any delays or motor overrun issues?

Did you just use a standard bicycle chain for the drive chain? Any other problems now that you are not running a tensioner pulley, you mention the virtual freewheel helped.

Do you have any plans to print a cover for the motor or a chain guard?

Have you considered torque sensing via chain slack like the Lightest setup?
It would allow you to drop one chainring and the TS bottom bracket. Although if the chainline lined up correctly you could drop the second chainring immediately along with a significant amount of chain.
Any reason you didn't run the motor inside the triangle first either build?

Cheers,
Grant

famichiki said:

How's the torque compare to the TSDZ2, strong enough to get up steep inclines without overheating? And how do you find the torque sensing in comparison? Any delays or motor overrun issues?

Did you just use a standard bicycle chain for the drive chain? Any other problems now that you are not running a tensioner pulley, you mention the virtual freewheel helped.

Do you have any plans to print a cover for the motor or a chain guard?

Click to expand...

I was running the tsdz2 with max 550W, and I'm running this one with the same parameters. I would say its about the same torque. Difficult to compare. I was running the tsdz2 without temperature control, so I think I overheated it a few times. The case was too hot to hold my hand on. This one with the ATF oil I think is a little better, but I have temperature control now, so it ramps down before anythin bad happens. I usually put in about half the power from my legs, and I usually run it at very technical climbs, then down the same path. It doesnt overheat this way. On flat road it takes about 10 minutes with constant 550w before it ramps down. I think 400w constant would be fine.

The delay on the tsdz2 is partly the reason I went this way. With the virtual freewheel keeping the drivetrain engaged, there is very little delay from coasting to full power compared to the tsdz2 which needs to spin up the motor and gears first. There is some slight overrun in certain cases that I am trying to tune out, but not bad at all. I want to tune it to be as responsive as possible, on the technical stuff, so that is the reason for the slight overruns. Still trying to find the perfect settings on the CA. Good thing is I can try out new settings when I am out on the trails, and dont need to wait to get back to a computer to flash new settings. Tuning it a bit softer for normal road/fireroad, etc is very easy.

The motor chain is just a normal 10speed chain, so I only have ot carry one kind of spare links with me. Its doing fine so far.

Grantmac said:

Have you considered torque sensing via chain slack like the Lightest setup?
It would allow you to drop one chainring and the TS bottom bracket. Although if the chainline lined up correctly you could drop the second chainring immediately along with a significant amount of chain.
Any reason you didn't run the motor inside the triangle first either build?

Cheers,
Grant

Click to expand...

Yes, I absolutely did consider the chain slack sensor. Rocky mountain bikes (I think) also has it this way. I could not find a sensor easily available, though, so I went this way. Did not do too much digging to be honest, when I found this BB torque sensor with freewheel I just bought it. Do you now here I could buy a sensor like that?

I wanted the battery inside the triangle, and the weight as low as possible, so that is why the motor ended up there. Its a quite safe position, even though it looks a bit exposed. The tsdz2 survived almost 1000km hanging below the BB, so I would be surprised if it gets damaged hanging the way it does now

Oh, and the final versions (for the time being at least) of the mounts are in the printer. It was a bit more PITA to get the printer to work with PETG filament, but the benefits are that it is stronger and weatherproof, compared to the normal PLA-stuff.

I think you could use a cable to the throttle sensor box from a Sur-ron or even just mount a throttle position sensor from a car.
Then you'd just use the output as a throttle.

I like the idea of being able to do torque sensing using simple components and low tech controllers. I might play with that myself.

Grantmac said:

I think you could use a cable to the throttle sensor box from a Sur-ron or even just mount a throttle position sensor from a car.
Then you'd just use the output as a throttle.

I like the idea of being able to do torque sensing using simple components and low tech controllers. I might play with that myself.

Click to expand...

Hmmm.... Do you mean a spring loaded pulley that pulls on a wire that goes to a throttle box?


Got the bike more or less finished. Need to re-print the CA mount, but I am quite happy with the design. It lowers the CA a lot compared to the stock mount, whuch I dont like that much... Managed to get an over the bar experience down a rock garden and smashed my helmet. The bike landed on the rear derailleur, so I guess that is better than a frame damage. A lot of straightening of the hanger and the gears are perfect again. Need to get a spare hanger just in case. Never owned a SRAM system before, but I must say I like it a lot!

Its quite ready for the winter now, studded tires are waiting in the garage for when snow arrives. Will need to build another battery and fix all the casings for the Liteville bike before spring arrives, and the narrow-tire season begins

That's precisely what I'm talking about. Dead simple and easy to start from a standstill.
Plus you could rig a cable throttle from the grip and run it to the same box, very analog solution using cheap components.

You would have to drive the chain directly though, I don't think it would work with a primary and secondary reduction.

Grantmac said:

That's precisely what I'm talking about. Dead simple and easy to start from a standstill.
Plus you could rig a cable throttle from the grip and run it to the same box, very analog solution using cheap components.

You would have to drive the chain directly though, I don't think it would work with a primary and secondary reduction.

Click to expand...

Its an interesting thought, maybe for another build! I would definately have added a cycle analyst and a cadence sensor as well, to get the tunability. Pure torque sensing doesnt work that great when cadence gets higher, you need the rpm sensor to calculate the power output of your legs.

So I re-printed a cycle analyst mount as close to the stem as possible, then made some wrap-tie mount around the stem. Think it turned out OK, but it is quite a big unit with a lot of wiring this thing.... Tried to clean it up a bit, but its not easy to get it perfect without a lot more job than I am willing to put into it now.

Other things since last update:

- Added thumb throttle. Getting back up in the saddle after a stop in a hill with low traction is way easier this way.

- Installed 26x4.8 spiked tires for floatation. Set up tubeless for running with 4-5 psi. The winter tires were so heavy that there was niw weight loss compared to the summer tires + tubes. Works great, though.

- Added cheap front light. Works OK together with powerful head light.

I think enough ATF has seeped out so the heat is no longer transported to the outer shell. Went into temperature protection today in -6C, but the motor was barely melting the snow that was thrown on it. So I will top it up. Might have a look at some bearings with better seals and using some kind of sealant between inner race and axle and outer race and shell. No big deal. Other than that I actually think the bike is finished, just how I wanted it. I have never liked skiing, so this is a nice way to get some exercise now during winter.

Looks good! I really like the black printed battery and controller boxes. I need to print some for my bike. What material did you use? And which orientation did you print them? How do they handle the cold?

Thanks! I used PETG. Great stuff that prints almost as easy as PLA when you have heat and retraction settings dialed in. Stronger than PLA and a lot easier than ABS. And doesnt break down out in the elements like PLA. Holding up great so far, but I've only had it for a few months, so not much of a longterm review.... The 3 huge tie wraps takes almost all of the stress, so I really dont think it will be failing any time soon anyway.

My print bed is 30x30x40 cm, and the box is something like 34cm, so I had no choise but to print it vertically. So I designed it to be printed like that with the lid on the small end. Might have done it different if I had a bigger printer... I think I will build a triangle pack for the FS bike that fits in the bottom of the frame, so it will be a new design. Never thought designing and printing stuff would be so much fun, should have bought a 3d printer a long time ago

Just a suggestion that instead of a top of stem mount i find it better to get the display out of harms way yet easy to read and access is to mount it beside the stem. A plate that holds the CA with a 1 1/8" hole that fits over the steerer tube and takes the place of a spacer is what I do to solve the problem. Should be easy enough to make one with a 3d printer, I used some aluminum plate.

Hmmmm, that might be a great idea. At least move it a few cm to the side so the huge "box" where the wires come out does not need to sit on top of the stem. Might have to re-do some cabling now

After a loooong winter, days are getti g longer an warmer, so I need to get the Liteville bike up and running again. I got a brand new baserunner (warranty because one of the hall inputs were blown), a Bafang G370 I decided I want to try, and a new battery to build.

First off the motor. The Aikema motor was very easy, just use the torque washers and weld up some mounts. On the Bafang unit, the axle flat on the drive side does not allow a sprocket on the inside, and on the non-drive side the cable goes into the drop-out because the axle flats are so close to the motor.



I decided to re-use the Aikema mounts that I made last year, so I will have to make some spacers. I will take the time to disassemble and clean/paint everything up this time, so they look a bit rough right now for test-fitting

Drive side was pretty easy. Used the two torque plates and a spacer between them to get the offset correct, and welded them together.



Non drive side, I made a spacer in the lathe with the ID of the axke flats, cut out one side for the cable, and welded on a tab for the drop-out, and cut it to size with a grinder. These things look like crap, but I think they will work just fine for the power-level I am using.



And it fits

I had to tear down the Bafang to rip out the speed hall sensor and replace it with a 10K NTC resistor to get temperature readings. I took some pictures, so I will just dump them here for reference. I did have a Bafang G31x here and some spare gears for the Aikema 75sx to compare. It looks like it uses the same gears as the G31x, even the same tooth count (18-30), so it should be able to handle about the same peak torque. The first stage reduction looks about the same as on the Aikema motor, but the second stage is a lot sturdier on the Bafangs than on the Aikema. So given the nylon is more or less the same quality, the Bafangs should hold do very well in this application, as the Aikema is still doing fine in the fatbike, with now up to 800w peak power.

This will never be a hub motor now.... needed some space to get the full adjustment without hitting the frame.



Bolted a sprocket to the motor and mounted it. A spare 36V battery and the baserunner hanging from the bike, and I hooked up the compiter for a quick test, and it seems to spin alright.



First test-print of the controller box. I decided to use the same layout as the fatbike, so this box will house the controller, excess wiring, charger plug and also a power button. It will hang on the seat tube.



As for the battery, I have bought my first ever brand new cells this time. Went with the 21700 samsung 50e cells. Im trying to make a clean build, so I have printed holders for a 10s3p battery, with recess for nickel and wiring, and space for a bluetooth BMS. If it works out like I have planned, it will be smaller than the 10s4p battery on the fatbike, and more capacity. Just need to weld it and I guess I will find out if it works

How much weight increase do you think you'll have from the Bafang motor?

The scale says 1680gr with sprockets, nuts and the torque spacers I made. Havent checked if the 1440gr that is claimed for the Aikema is true, but I will check it next time it is out of the bike. I might put it in the lathe and shave away the spoke mounts (or whatever it is called) some day.

So probably a couple hundred grams increase.

Based on their specified weight that's not much difference. Under 2kg plus mount is really impressive.
The Bafang should be good for quite a bit more power as well.

Threw some things on the scale yesterday for fun, seems quite decent so far. At least for using no fancy materials

Battery (540wh) with box, mounts and bms: 2678g (plus nickel and a little wiring)
Baserunner controller with box: 370g
Motor mount: 460g
Motor: 1680g

Dont know what the extra weight of the torquesensor is, or the weight og the Cycle Analyst. Seems like bike will be close to 20kg when assembled.

Also got to do some painting, and printed batterybox yesterday. Just need to weld the nickel, and some assembly and soldering on the bike, and it should be ready for some testing.

Finally got around to build my battery yesterday. I forgot one passage for a serial connection in my print which annoyed me, but there was two 8x0.20mm nickel strips in serial at that point anyway, so should be more than enough for a battery thay will never see more than 25A peak anyway. Parallell connections are 7x0.12mm. A little overkill here and there, but better safe than sorry. BMS was attached to padding with adhesive on both sides, so it shouldnt move.

I thougth I was really smart when I made the prints, to make sure the main cables would not need to cross, and were as short as possible. Somehow it was mirrored somewhere on the way from brain to cad, so it ended up with the longest cables possible, and three crosses Luckily it was enough space. The outer perimeters of the cell holders fit snug in the battery box, so there is no movement when mounted to the bike.




Also made a new mount for the CA. It now sits a lot lower an more protected. I can flip the bike upside down for quick fix out on the trail without smashing the CA (if it is flat, of course), which is nice.

So, except for some maintenance I need to do, I think I finished this one today! Mounted the battery, soldered the throttle and AUX plug that was hanging from the CA, and finished off some bits and pieces. With some field weakening it hits 120rpm cadense with full support, so I think the gearing should be decent. Im am axited to see how much heat it can handle compared to the Aikema. Might have to fill it with ATF, but it is only running in grease right now. Finished weight ended at 20.4kg, so pretty close to what I was aiming for. Im having a ton of fun with the fatbike at the time, but I am really looking forward to summer, and skinny tires season!

So, not too impressed with the Bafang G370 motor right now. Nylon gears gave up after a few trips. I did have some violent juddering before I got to tune it out, but still.... Running at 35A phase amps, which should translate into 43NM according to phase suite., which is less than what the Aikema motor is running. I did order another Aikema motor so I can have both bikes running. At the same time I started looking for spare gears for the Bafang, but no luck. Only for g310 and 311 to be found.

So, could not resist trying to find out how to draw and print helical gears! After a few test prints with different pressure angles and helix degrees in PLA, I think I am pretty close. The motor ran quite smooth at 250w on these for a while, until I turned it up to 500w, and they gave up after a few pulls. I have a roll of nylon filament coming my way, so I am quite curious to test how strong printed nylon gears are!

Got the nylon filament and surprisingly got a decent mileage out of the motor. After 250km it started to develop the same sound as the stock hears, and found damage to one tooth on every gear. It did last longer than the stock ones, though. Dont know if I might have missed something when I put the mitir together, or if ut is somethung wrong with this specific unit. But it was never as smooth or silent as the Aikema unit even when the stock gears was in there.

So I decided to try something else, the aikema 100h, or cute q100h, or some other name that it is selling under. I was sceptical of the axle cable exit, but it seems like it will be well protected with the cover cap, or I might fabricate something in addition. I added another 300g or so over the g370, but was hoping this would be "big" enough to not need ATF cooling. I still got some tuning left on the baserunner, but so far it seems very promising. Max temp was around 65C on a hot summer day and up to 700w peaks (replaced stock speed hall sensor with ntc 10k temp sensor). Needed to machine down the flanges to make it fit.