Giant Trance Belt drive with MESC @ few kW

[mxlemming]

I normally hang out in the motor technology section, but a number of people have been hankering to see my messy Giant trance.

This thing is more of a test rig for the MESC, though I commute on it frequently, and go blatting around the countryside for fun...

Without further ado... The current state of the bike - Mid June 2021:


The battery is 13s3p Samsung 30Q. Capable of 45A discharge rate at ~48V, it's good for 2 and a bit kW. Currently using a Daly 13s40A BMS, the previous 2 caught fire, because they were crap.


The drive is an 8080 70kV motor from McMaster, and is driven by a 15:96 belt reduction, giving a top speed of mid 30s mph.




The bracket is cut from 8mm plate, cnc'd by a local machinist friend. I had 3 made... 1 on my bike, one on a friends bike and one... waiting to go on my bro's bike.

I run the whole lot from the MESC ESC I have been developing, which is good for a lot more current than the motor can take - the motor is roasting hot, with the ESC not even warm to the touch after pushing ~70A phase for 10 mins.

The lever is vaguelly interesting, incorporating a throttle and a brake:
It works on the principle of hall sensors and magnets in the lever blades, which are read by the MCU on a 50Hz basis, converted to a PWM signal and fed at 50Hz as an RC PPM.

The screen is an SH1106, pretty basic, but great for things like this.




I designed the PCB and had it built by JLCPCB, then soldered the parts myself.Rock bottom STM32F030 K6.

The battery was spot welded. Don't think I would bother again, think I would solder next time. I bought one of the cheap ebay spot welders, and it was a piece of shit. Pulses way too short to do anything, even with a rack of car batteries, and the MCU but out and reset during the battery voltage drop. Absolute junk. So I castrated it's brain, and added an XMC1302 breakout board and an NCP5106 I'd made to control the PWM...




The whole lot now lives in a 3D printed case:


I went through a whole load ofbelt drive combos, the biggest issue being stretch of the belt under braking causing skipping. Now using HTD8m, but will probably go back to HTD5m soon.

Building a 20s4p pack at the moment... TBC how that goes!

Any tips on a bigger motor that could fit in this... the 8080 motor is good enough, but hardly a wild ride.

The controller is a fully custom board, the MESC, documented over in the motor tech section:
https://endless-sphere.com/forums/viewtopic.php?f=30&t=107672

Had 3 revisions so far, latest one is up to something of the order of 16s safe (have run 20s) and 180A pulse tested, seems completely cold at 70A, and the motors give up or the battery BMS craps out way before I get any thermal response.




That's all folks. Epic fun, been a pain in the neck, but the result is way better than a small hub motor, maybe a big QS205 or MXUS 3k turbo would be better, but the shipping for those is crazy. I have the Voilamart 1.5kW hubmotor kit for comparison, that has a bit more power capacity, but it just has nothing on this for power to weight.

 

[mxlemming]

Few more pics of tools built along the way, programming bits and pieces.




New battery build for a friend... I am also building a 20s4p 21700, TBC what that looks like.

 

[cheapcookie]

Effective prototype, really excited for what this thing becomes! hope you get a full suspension bike as well. check these motors out https://neumotors.com/

 

[thepronghorn]

Really cool. For a bigger motor, maybe something like a 12070? I know they're expensive, but MAD's M30 motor and the T-15 should be pretty powerful.

If you want to go really cheap, you could get a hoverboard motor and adapt it to fit your bike, but it needs a rewind to make good power at a reasonable voltage and fairly extensive mechanical modifications to mount a sprocket or pulley.

For a modest power increase, there are 80100 motors out there.

Cheapcookie's recommendation of a neumotor is good too. Madin88 used an 8057 on his build. There have been some reports of unbalanced rotors on those motors though which is a bit disappointing.
https://endless-sphere.com/forums/viewtopic.php?f=30&t=91144

 

[E-HP]

I normally hang out in the motor technology section, but a number of people have been hankering to see my messy Giant trance.

LOL, it's a thing of beauty. I love seeing non-kit builds, the R&D phase, and the fabrication. Plus the 3D printing intrigues me. Something to play with in the future.

Can't wait to see how it develops.

 

[Grantmac]

Would be very interesting to see if it could run the CYC X1 which seems to be limited by poor controller choices.

Especially in a left hand drive setup.

 

[mxlemming]

Cheers all!

Neumotor 8038 looks ideal. Have requested a price. I'm setting my limit at 300gbp.

12070 also good, but there are serious issues with number of pole pairs - 15pp which makes the erpm crazy high with the reduction. Might be possible.

80100i don't think it's worth the effort for the amount of extra oomph. I want 2x the power at least for the next iteration... Surron kind of power with xc tail bike weight.

I have zero interest in bb mounted solutions I'm afraid Grant. If you're wondering about the controller running that... It probably can, but there's no interface for pedal assist yet qwerkus has one and might be working on it. I've got very little interest in doing controllers for money, but anyone is absolutely welcome to build one from the kicad files.

I've also been considering scooter motors hacked up, following in from the illegal bike success, but they're heavy and many pole pairs. UAV motors are generally more optimised for weight which is good. Any tips on a good hoverboard motor? I'm guessing rewind to go faster right? Not immediately looking for a motor rewind project.

Effective prototype, really excited for what this thing becomes! hope you get a full suspension bike as well. check these motors out https://neumotors.com/

This is a 5 inch travel full suspension bike...

 

[Grantmac]

I'm talking about mounting it in the frame (better for unsprung weight) then driving your existing sprocket setup.

 

[mxlemming]

Bit confused. That's a mid drive pedal assist. Would take quite a lot to shoehorn that setup onto my bike.

Seems to be designed for an 83mm, 100mm or 120mm bb, which is virtually unheard of. Standard is 68mm.

1100usd. Not exactly something to buy and nick a motor out of.

Could you elaborate a bit? I'm confused.

 

[Grantmac]

They will sell individual pieces. You can get just the motor and gearbox which are quality items at a decent price.

 

[mxlemming]

That makes sense. Will look into it.

Full suspension bike though with virtual pivot so mid drive requires tensioners on the belt. Will be complicated...

 

[mxlemming]

Effective prototype, really excited for what this thing becomes! hope you get a full suspension bike as well. check these motors out https://neumotors.com/

Cheapcookie's recommendation of a neumotor is good too. Madin88 used an 8057 on his build. There have been some reports of unbalanced rotors on those motors though which is a bit disappointing.
viewtopic.php?f=30&t=91144

Just read madin88 build thread. Sounds like the neumotor is a serious level up. Would love to get one and see if the MESC stands up to 190A for more than a single pulse.

Contacted neumotor... No luck yet. Anyone know how to extort motors out of them and how much they are?

 

[SlowCo]

Very interesting project. :thumb:
I will follow your progress!

 

[mxlemming]

Have completely failed to get a motor out of Neumotor. Will vaguelly keep trying, biking season will only be a few more months though so I guess there is the whole of winter to try...

I got the HTD5M 25mm wide belt up and running. There have been some easy learnings from this:


1) The HTD8m belt just skips far far less. I am pretty sure this is because of the tooth profile, not the size. The 8M belt I have is a Tecnamic Strongbelt Motus, and the teeth are visibly parabolic. I have similar HTD3M Motus belts for a work project. They never skip.

The 5M-25 belt is a Conti Synchrobelt. It's a piece of shit. Standard profile round teeth, and feels rubbery, where the Tecnamic Motus feels... strong... The quality is just so obviously different.

The take-home from this is: Belt quality matters. Don't buy crappy ebay belts, and don't buy Contitech Synchrobelts. I have now had 2 Conti belts, and

I am going to see if I can get hold of an Optibelt 5MHP (the premium branch of Technamic).

However: The 5M belt has allowed a speed reduction, 20:148 (7.4) instead of 15:95(6.3) , which has improved the pickup, more so than I would expect from the 17% torque increase. I suspect the motor and controller just work better at higher RPM. It is also quieter; the 8M belt was noticeably whirry.

 

[BalorNG]

Greetings!
I have a bigger printer, so here is my setup:

It's a 'long haul' setup, not a 'wolf in sheeps clothing' like you want to build though, so efficiency, not raw power is paramount, and I have (and happy with) 24v, 100ah battery so I can get away with single speed reduction with much higher KV and still get useable range of speeds (up to 31mph or so).

I *do* lack low end torque tho, and I can certainly use some more even with dual motor setup + pedalling, a lone motor can barely get me moving even on flat, and sensorless drone motor on a vesc is useless from standstill, even with HFI enabled (but is pretty light and efficient, 0.032 Ohm, 1.5a noload, 115 Kv, rated at 3kw from 50v), and given that I myself weight quite a bit even disregarding 100+ pounds of bike...

A 8080 like yours looks like businesss - avalable in 170Kv from aliexpress for 150$, and matches my voltage/reduction ratio (about 10:1, currently 9:1 with 20T-180T pulley).

I understand yours has a very different KV noload/resistance data would be useless to me, but what about it's interior? You've mentioned in other thread that bearings are crap and are small, but on photos, unlike with flipsky 6384 motor I'm also considering (currently have 6374), pretty large bearings are visible...

A small note regarding your setup and belt skipping... you sure it is not your 'spider' (namely, a thin and 'holy' brake disk) twisting under cantilevered load and putting the pulley out of alingment with the motor? Adapter plate might also contribute, though it is pretty thick in your case...
I thinking on reprinting my pulley with the spider centered *inside* the pulley to remote this twisting lever...

 

[mxlemming]

Wow that really is quite something.

The 8080 motors are underwhelming but substantially more powerful/torquey than the 63mm ones. It might be enough for you, they easily propel me to 30mph with just 1 of them.

The disc rotor definitely doesn't bend. I've been careful to get the forces aligned to the exact center. This type of rotor is frequently used in 220mm disc brake rotors, front brake on a downhill bike which generates insane amounts of torque... Like 500Nm kind of torque. It looks thin but actually it's far far far from the weak link.

The skipping is mainly crap htd belt. I can see it stretch, and the skipping occurs during braking when the loose side is the one going into mesh, so the round profile starts climbing up onto the next tooth. You can literally watch it happen while wheeling the bike forwards with the motor generating reverse torque. I've got a pile of (5) different types (curvilinear, parabolic profiles...) got an optibelt 5mhp, gates 5htd, bando supertorque s5m and optibelt Omega arriving soon to try out plus the current contitech one. I'm getting the bracket upgraded to be stiffer. I'm also adding a booster bearing to the motor. Will do a quick belt write up soon.

I'm looking at building my own motor and am also trying to get a neumotor and an ant innovation AT12070 with a custom shaft. Neumotor like pulling teeth, they ask relentless questions and never manage to quite understand, I've basically given up. The ant innovation guy (who I'm pretty sure is basically just selling his own brand freerchobby motors) has been really responsive so far so if he successfully gets me a motor with a keyed shaft it could be the holy grail for these bikes.

The bearing in the red 8080 motor measures 21mm od 12mm id from memory so even if it's a good quality one it's only 1.9kN rating with a roughly 2x cantilever, and with say 18Nm on a 15mm radius... 1200N... That bearing is probably overloaded by about 25% before you even start to consider fatigue life.

My friend built a bike up with flipsky 6384 and 14 to 120 reduction ratio, got a bit over 50km/h but it got really hot really fast.

If you weren't in Russia I'd give you one of the Red motors but the postage is probably more than the total bill from AliExpress...

 

[BalorNG]

Wow that really is quite something.

The 8080 motors are underwhelming but substantially more powerful/torquey than the 63mm ones. It might be enough for you, they easily propel me to 30mph with just 1 of them.

The disc rotor definitely doesn't bend. I've been careful to get the forces aligned to the exact center. This type of rotor is frequently used in 220mm disc brake rotors, front brake on a downhill bike which generates insane amounts of torque... Like 500Nm kind of torque. It looks thin but actually it's far far far from the weak link.

The skipping is mainly crap htd belt. I can see it stretch, and the skipping occurs during braking when the loose side is the one going into mesh, so the round profile starts climbing up onto the next tooth. You can literally watch it happen while wheeling the bike forwards with the motor generating reverse torque. I've got a pile of (5) different types (curvilinear, parabolic profiles...) got an optibelt 5mhp, gates 5htd, bando supertorque s5m and optibelt Omega arriving soon to try out plus the current contitech one. I'm getting the bracket upgraded to be stiffer. I'm also adding a booster bearing to the motor. Will do a quick belt write up soon.

I'm looking at building my own motor and am also trying to get a neumotor and an ant innovation AT12070 with a custom shaft. Neumotor like pulling teeth, they ask relentless questions and never manage to quite understand, I've basically given up. The ant innovation guy (who I'm pretty sure is basically just selling his own brand freerchobby motors) has been really responsive so far so if he successfully gets me a motor with a keyed shaft it could be the holy grail for these bikes.

The bearing in the red 8080 motor measures 21mm od 12mm id from memory so even if it's a good quality one it's only 1.9kN rating with a roughly 2x cantilever, and with say 18Nm on a 15mm radius... 1200N... That bearing is probably overloaded by about 25% before you even start to consider fatigue life.

My friend built a bike up with flipsky 6384 and 14 to 120 reduction ratio, got a bit over 50km/h but it got really hot really fast.

If you weren't in Russia I'd give you one of the Red motors but the postage is probably more than the total bill from AliExpress...

A few of my own thoughts: putting the motor on thermal paste and heatsinking the adapter plate might work unless you push really hard. Thermal path to the windings is not great, but it may be considerably better than a mere plate.

My (underwhelming) motor with 45 phase amps gets me to 48 kmh, acceleration is like a literal freight train but I suspect that being a recumbent my aero is a tiny bit better than a typical emtb (not by much if my speeds during pedalling are of any indication, I'm a actually a slower than on my MTB but my rolling resistance must be quite abysmal).

The motor gets hot, but not *too* hot, and my large adapter plate gets pretty evenly hot with the motor.
Bet that would change if I get my phase up to 70-80... afraid to do that w/o a thermal probe...

Adding active cooling will be even better of course, I have an old videocard fan with a copper heatsink and intend to experiment with that - that's way worse than blowing the air thought the actual windings like some did, but you don't need to add air filtering lest your motor fills with crap, plus that's impossible on my motor (no holes). Drone motor is way better in this respect, and it designed to pump quite a bit of air (I feel it).
I'm considering getting an encoder eventually and installing it the way robot people do (adding a magnet to the shaft from 'inside').
https://www2.eecs.berkeley.edu/Pubs/TechRpts/2018/EECS-2018-86.pdf
Vescs now support them...

But has *too much* holes for my comfort to install it somewhere low and close to the wheel...

Regarding 18nm of torque from the motor - well, that's an order of magnitude more than a bit more than 2 nm I'm currently getting so I'll be more than happy with half that
If you are serious, there are ways to get stuff cheaper than that from UK using 'postage intermediaries', you'll just send it to a UK adress and than it is my problem from than on... please PM with price/ways to pay you (I have paypal...) if you do.

 

[BalorNG]

Btw, here is implementation that might work for you: (use google translate, but pictures tell the tale pretty well)
https://electrotransport.ru/ussr/index.php?msg=2001223

 

[mxlemming]

Btw, here is implementation that might work for you: (use google translate, but pictures tell the tale pretty well)
https://electrotransport.ru/ussr/index.php?msg=2001223

I'm really not sure what you're suggesting as an implementation. That thread is about vesc 4.12... I have built my own controller with about 4x the power (100V vs 60V and easily handles twice the current) and written my own firmware and have also ported the hardware to the vesc chip and run the vesc firmware... I prefer my own. Fewer features but same power and efficiency and eats its own face much less.

I'm trying to buy bigger motors. Once I have them I'll be more willing to part with one of the old ones. Cruising my fingers that I get a sensible price and so on soon.

 

[BalorNG]

Btw, here is implementation that might work for you: (use google translate, but pictures tell the tale pretty well)
https://electrotransport.ru/ussr/index.php?msg=2001223

I'm really not sure what you're suggesting as an implementation. That thread is about vesc 4.12... I have built my own controller with about 4x the power (100V vs 60V and easily handles twice the current) and written my own firmware and have also ported the hardware to the vesc chip and run the vesc firmware... I prefer my own. Fewer features but same power and efficiency and eats its own face much less.

I'm trying to buy bigger motors. Once I have them I'll be more willing to part with one of the old ones. Cruising my fingers that I get a sensible price and so on soon.

Yea, that's a thread about VESC, but this particular post about active cooling solution of 6384 motor (so should be very similar for 8080 one, or any motor with holes in it for that matter), open a second 'spoiler' section in the post number 361, for lots of pictures. He has an automatic system using 40mm server fan blowing air though the motor interior, acting on data from the thermal probe in the coils, with an air filter to prevent debris getting inside the motor.

But than lots of heat means loss of efficiency, and like I said I'd much prefer a more effective solution that does not need active cooling that eats MORE watts! (though 10w is hardly a blip in the grand scheme of things when it comes to what YOU want).

Still, there a limit what you can achieve even with active cooling - due to saturation.

 

[mxlemming]

So having given up on the neumotor, I found this (well hummina pinned me to it):

Ant innovation at12070 with 120mmOD and 102mm stator. It's more than I'd ever intended to part with for this project but it should turn this into a complete monster.

This is a pic of them assembling the magnets into my motor, i only just realised how big it is.

They're making me a custom shaft with keyway on that standard motor, I'm going with 62kV so that my 20s pack will get it to something like 4800rpm, then maybe i can do a bit of field weakening...

Allegedly 2kg which is a step up from the 1200g red motor but not crazy. Lighter than the neu 8057 (100mm OD)and the torque should be higher with similar max power.

It might be a step too far, an overreaction to the ok but underwhelming 8080 but i can always strap it to a paraglider chute.

The new battery is approaching completion, 20s4p of p42a cells with ANT BMS for a120A discharge rate at 72V(84 max) which sounds like 10kW...


I'm actually now worried that the torque might start to destroy things.

It's a tossup, which will die first out of frame, wheel, htd belt, pulley, or MESC.

 

[mxlemming]

In other news, I programmed the MESC to have a power limiter so despite this upcoming power, I will be able to run it in some kind of eco legal mode at 250W and get something like 100km range if all goes well...

 

[BalorNG]

That's nearly the size of a typical DD bike motor, but with massive reduction for 8x more torque! Crazy!
You'll certainly need to switch to 8T belt though...

I'm actually now worried that the torque might start to destroy things.

Fortunately, you have a build in torque limiter - give too much and you'll just be flipped over backwards .
No, really, must be a genuine concern, a combination of torque reaction moment plus acceleration close to that of a racing motorcycle but with a much shorter WB and higher CG WILL be a concern. I recall a person was seriously hurt like that when he was unprepared for a massive torque output of an electric motorcycle. You might want to upgrade to one of those 'progressive geometry' rigs that look much closer to my bent than your bike

I think forces should be kept within designed range of bicycle disk brakes (around 100nm(https://enduro-mtb.com/en/behind-the-scenes-disc-brake-group-test/)), and you'll certainly want a reinforcement of the frame (vertical crossbrace between the stays)- I am monitoring a welder's blog and a LOT of frame fail somewhere around that point from braking torque.

Also, with 7a of no-load current, I think your bike will not longer be meaningfully pedalable, but likely you are past caring about that at this point

 

[mxlemming]

Well the good weather just started so I'm back on the ebike.

Had some huge controller improvements over the last month or so, so I'm now running full sensorless and hfi kinda works.

The motor and battery is the big change. 3x the battery capacity and a motor 2x the weight... The improvement is huge.

Initially with bench testing I was worried the motor was no good since it was getting pretty hot spinning with no load but it's turned out to be pretty good. It actually seems to stay cooler under load than free spinning.

The belt skip has gone even with the htd5m belt since i got a new bracket and an extra needle roller bearing.

I'm happy enough with it now that I'm going through the tedious process of removing the safety tape and replacing with brackets. The controller went first into a box and now into a box with brackets attached to the seatpost.

But really, I now just want to ride. There's a whole load of to be explored farm tracks, byways, off road... a kilometer from my house.

The biggest problem with this at12070 motor is the noise. It's really screamy and has a resonance at about 15mph/25kmph (an unfortunate speed right?). Fortunately that resonance disappears when I remove the power limiter...

 

[thepronghorn]

Cool update. How does the sliding part for the extra needle bearing work? Do the slotted holes work well enough for belt tension? Are you running an idler for extra tooth engagement on the motor pulley? What's the thickness of the bracket?