Neu 8057 6kW left hand drive hardtail with vesc and encoder

Joined
Jul 20, 2013
Messages
339
Location
Sydney
Hi all,

This ebike previously had a cyclone 1680w aka cyclone mini installed on it. I was considering getting a cyclone 3000w motor as an upgrade and keeping as much the same as possible. I didn't really like this idea as the small cyclone motor with the metal planet gears was not getting great chain life. Running the motor fast and simply not pedalling helped a bit but I wanted to try a direct chain/belt drive to the rear wheel.

I'd always wanted to build an ebike driven by an rc outrunner but was put off by reports of poor bearing life for ev use and the complexity of the reduction. I stumbled across madin88's sweet build log https://endless-sphere.com/forums/viewtopic.php?t=91144. With a big rear sprocket, a single stage of reduction is possible, even with my 29er wheels. The 5/8" shaft diameter and large beefy bearings look like they will be reliable. With 80mm diameter stator and 57mm stator width, the neu motor is almost exactly double the motor of the cyclone 3000w, which has an 80mm rotor and approx. 28mm stator width.

C6rkof4.jpg


Great copper fill.

A forum member was kind enough to help me out and turn a shaft that would exit the front of the motor and with a 4mm keyway. Then I could attach a 219 sprocket with the adapter made by Lightningrods.

I had similar problems to madin. My motor was not factory balanced and had a ton of vibration. I managed to balance the motor with by adding trial weights to the outside of the can. I don't know if the stock bearings were bad or if the vibration damaged them, but they quickly developed some play, even before I had actually ridden the bike. I replaced them with sealed ones.

http://i.imgur.com/puS4zyM.jpg

WS1jSPU.jpg


For controller I bought a pcb for an 18 fet vesc from a fellow on the Vesc Project forums. This uses 80v rated individual gate drivers rather than the rather fragile 60v rated motor driver IC.

The soldered board with copper bus bars.

qnmL4fB.jpg


i5ppF1L.jpg


I'm using the AS5147 rotary encoder as this is supported by the vesc. I 3d printed a cover to go over the shaft to mount the encoder pcb. Then I super glued the magnet on to the end of the shaft. I found the sensorless start in foc was not good. I;m pretty sure this is because the values of the current shunts are really low and the resistance of the motor is also very low so the controller has to cope with a lot of noise during start. The encoder has 14 bit precision on the angle value (more like 11-12 bit with noise and possibly misaligned magnet) and the motor starts really nicely. It is also possible to do silly stuff like smoothly crawling up hills with the rear wheel barely turning.
<Encoder pics and stuff here>

Here is the finished bike. Battery is 14s8p of saggy old Samsung 25R. No rear brake, I'm just using a left thumb throttle for regen and keeping it simple. I will be on the street 90% of the time and it doesn't rain that much in Sydney. I'm not a fan of the ebrake activated variable regen controlled by the main throttle. For the uncoordinated like myself I feel like it is a recipe for accidental WOT. At the moment the bike is way over geared. A 12t motor sprocket and 91t rear sprocket gives 80kph at 4000 motor rpm. This seems absurd but with the motor's high copper fill, I calculated that resistive losses are slightly lower than my cyclone in 5th gear on the 10sp cassette. 5th gear was the gear I would use for climbing steep hills.

LtXtPXk.jpg


I had chain drops on my first few rides. The usual alignment and tension problems. After sorting those out I was able to crank the phase amps up to 350. Crazy stuff. Despite being geared so high and with the 4s battery pack in the front box I can still wheelie.

I have a 115t laser cut sprocket on the way. This should give about 60kph top speed. I've added field weakening to the vesc foc code so I can keep my top speed while being more efficient for hill climbing. I made it trigger when there is a mismatch between commanded phase current and actual phase current. This means it can kick in below top speed to try and bring motor current closer to the requested current from throttle.
 
Awesome!

I've had my eye on those Neumotors for a while and watched Maddin88 build with much interest. Glad to see another build around here with that motor.

It's a bit disappointing that you both had issues with vibration, shafts, and bearings. I thought one of the biggest upsides to this motor was the better manufacturing and quality control than the average chinese RC motor. At least these issues are fixable and its not some terrible inefficiency baked into the the motor design.Its just a shame to have to do surgery on new stuff.

Looking forward to more reviews as you get some more hours on it.
 
DanGT86 said:
It's a bit disappointing that you both had issues with vibration, shafts, and bearings. I thought one of the biggest upsides to this motor was the better manufacturing and quality control than the average chinese RC motor.

Manual balancing made a huge improvement. After getting the right mass of epoxy putty at the right spot on the outside of the can, most of the vibration is gone. The phone app I was using to measure vibration reported almost a 10x decrease in the amplitude of the vibration.

DanGT86 said:
At least these issues are fixable and its not some terrible inefficiency baked into the the motor design.

Is this a reference to Revolt motors ? :? I was reading through that thread. Early on I thought the motor can't be that bad, surely it is just the hall sensors. But it seems several people with different setups and controllers simply can not get it to run efficiently?

I did about 30km of riding yesterday with FOC and encoder. I've previously been riding on the cyclone and the tangent gearbox by Dave and the low noise was nice on the shared pedestrian paths. It's still an outrunner so not DD silent but very quiet for an RC setup. I'm really liking this plastic kart sprocket.

I did have another chain drop just before I made it home. It was a nasty one that derailed at the front sprocket. I'm pretty sure it was from a master link that I reused too many times. :oops: I've since removed the master links and made a solid chain.

mcuIMKY.jpg


Not a great pic but what happened is the tip of the master link pin sheared off at the groove where the clip sits. I suspect what happened was the clip was loose from reuse and the master link backed out of the bushing.

Unfortunately the chain managed to rip the connector off my encoder pcb. Luckily I have a spare and the expensive sensor IC was fine since it lays flat. The encoders are from aliexpress https://www.aliexpress.com/item/Mag...7181-4c0a-9847-2caa3b2a48f2&priceBeautifyAB=0

VN2tL9x.jpg
 
Hi,
nice to see you got it running!

district9prawn said:
I had similar problems to madin. My motor was not factory balanced and had a ton of vibration. I managed to balance the motor with by adding trial weights to the outside of the can. I don't know if the stock bearings were bad or if the vibration damaged them, but they quickly developed some play, even before I had actually ridden the bike. I replaced them with sealed ones.
Yes the unbalanced bell is really annoying.. Kudus that you could fix it by yourself
If one plans to buy such motor (of the 80xx series) you should always ask for that, but they anyway should not ship out motors with that kind of vibrations. We can only hope they learned.
The other thing is if you remove and reinstall the shaft, it could be that balancing is gone. They told me to not do this.

Regarding the bearings i wonder if they were press fitted on your motor?

On mine they were not, which i think could be a problem as well in terms of the radial play. Bearings with normal clearance CN (as you usually get if you buy a bearing) should be installed with a press fit, otherwise there will always be a slight radial play.
If there is a loose bearing seat as it is the case on my 8057, it would be wise to use bearings with C2 clearance, or even better C2 with ABEC5 grade (which are $$).
I'm using the AS5147 rotary encoder as this is supported by the vesc.
Yeah i have seen this. awesome work!
There was also another guy posting pictures in my thread about using an encoder instead of hall's.

Do you think there is an advantage over hall sensors (together with FOC control)?

If one is using trapezoidal 6-step commutation, it is a fact that hall sensors offer accurate enough position, but with FOC where the magnetic field can be applied in any direction, it could make a lot of sense to have finer position resolution..

Do you know more about that?
I had chain drops on my first few rides. The usual alignment and tension problems. After sorting those out I was able to crank the phase amps up to 350. Crazy stuff. Despite being geared so high and with the 4s battery pack in the front box I can still wheelie.

Good to know you can do wheelies at 350A 8) That makes me believe there is some room above on my conservative 250A setting.
Do you have the 100kv or 75kv motor? Is this a VESC 6 based controller?
btw: with your actual 12/91 redcution yours would be geared even quicker than mine (when including wheel size into the calculation) :)

Regarding the chain drop i think, at least from the pics, the idler or chain tensioner below the motor could be the problem.
it will worsen the clinch on the small sprocket which could lead to chain jump and chain drop.

If possible, use the largest sprockets you can find or which fit. The more teeth you have on the small sprocket, the less force on the chain, the lower the noise, the longer the lifespan etc..
And make sure everything is aligned properly and doesn't bend (too much) under load.
 
madin88 said:
Regarding the bearings i wonder if they were press fitted on your motor? On mine they were not, which i think could be a problem as well in terms of the radial play.

Do you mean your bearings were very loose in the stator? Able to remove with fingers? My bearings were a press fit into the stator. I did notice that one seat was a fair bit tighter than the other, with both the stock bearing and my replacements. I just used a cheap sealed bearing from https://www.thebigbearingstore.com/r10-2rs-r10-zz-radial-ball-bearing-5-8-bore/

There was also another guy posting pictures in my thread about using an encoder instead of hall's.

Do you think there is an advantage over hall sensors (together with FOC control)?

If one is using trapezoidal 6-step commutation, it is a fact that hall sensors offer accurate enough position, but with FOC where the magnetic field can be applied in any direction, it could make a lot of sense to have finer position resolution.

I think most foc controllers out there will be interpolating between the mapped hall values to get a bit more accuracy. And once the motor is spinning nicely they should really be using foc to get the rotor position. I have the sensorless transition set to 1500 erpm which is something like 3 - 4 kph. And I could probably use much less with no issue. So no I don't think it makes a huge difference, unless trying to crawl up a really steep hill or using PID position control. My main reason for using the encoder was the simple mounting. I saw your hall sensor mounts with the tiny magnets. Nice work :) The main problem with the encoder setup is that the high speed SPI connection needs twisted pair over the clock and signal line with cables over 10cm. Otherwise the controller gets overwhelmed with bit errors and it just doesn't work.

I've tried running the motor purely on the encoder, with no transition to sensorless at higher speeds. It works but at high speed sensorless was still more efficient. It was also a bit sensitive to magnet alignment, so for real world use it makes much more sense to transition to sensorless as soon as possible.

btw: with your actual 12/91 redcution yours would be geared even quicker than mine (when including wheel size into the calculation) :)

I've got the 100kv. The controller is vesc6 based, with shunts on each phase, not just two.
 
district9prawn said:
Do you mean your bearings were very loose in the stator? Able to remove with fingers? My bearings were a press fit into the stator. I did notice that one seat was a fair bit tighter than the other, with both the stock bearing and my replacements. I just used a cheap sealed bearing from
On my motor the bearings were secured with some kind of thread locking glue, but after cleaning off the rests both bearings just fell into the stator. There was virtually no radial play, but definitely no press fit.
 
madin88 said:
On my motor the bearings were secured with some kind of thread locking glue, but after cleaning off the rests both bearings just fell into the stator. There was virtually no radial play, but definitely no press fit.

That doesn't sound good if it's barely a slip fit.
 
I did about 50km of riding a couple of days ago. It rained for pretty much the entire ride but it was a blast. The encoder pcb cracked a bit as it printed poorly and did not have enough supports. One of the pcb bolts was too long and was struck by the cranks a few km into my ride so I had to rip it off. The controller has no automatic sensorless transition for invalid encoder but luckily I had the BT module still plugged in and was able to use phone app and set it to sensorless trap. Didn't use foc as I haven't dialled it in for sensorless and it starts rough. It ran quite well in trap mode. Not much motor whine as motor is geared so high and rpm is low.

The motor was blasted with water and the fan design of the rotor mean't water would have been getting sprayed through the motor continuously. I was riding quite fast so I basically had a water cooled motor with no thermal limits. Unfortunately the temporary plastic box for the controller wasn't quite waterproof enough. I got a cut out about 10km from home and it took a few tries to get the motor spinning again but once it was going it was ok until I stopped again and the controller had to make another sensorless start. About 3km from home it would only cog and grind so I pedalled back. Connecting to the pc the duty was reported as 200%. When I opened it up there wasn't really that much water but over some mosfets and some phase voltage sense resistors there was this gross slime (wtF?). After rinsing and chasing the water off with some iso alcohol it was all good.

Also printed a new encoder holder out of polycarbonate. Seems really strong.
0lZHo3X.jpg


Also received the laser cut sprocket and some steel mounting plates. Need to grind bevels on the tooth tips. I'm keen to gear lower and play with my field weakening code on the road.
woWV3Ap.jpg


I welded the steel plates at right angles. I don't really have much experience with welding. I was doing tig and my pack of mild steel fillers said it would weld ok over the mill scale. I got about 2cm and then the whole thing turned to swiss cheese and little volcanos. So ground the welds and scale all off and redid and it was ok. Still ugly but solid. I also welded some small alloy pieces to my bb mount to hopefully reduce flex. And shortened it so the front wheel doesn't hit the controller box if the fork bottoms out.
3dfM9Qe.jpg
 
parts looking nicely done and welding too, at least from the one side you took the photo :)

Could you share more photos of the motor bracket attached to the frame?
Does the controller support regen and if yes do you use it?
 
Got the brackets back on the frame.

9s0PS1R.jpg



madin88 said:
Does the controller support regen and if yes do you use it?

Yes I have regen set on a second throttle. I have it set quite high (no rear disc brake) at 1.5kw and 150a phase to get as much low speed braking torque as possible. At very low speeds I set up the controller to short the windings in order to get smooth and powerful braking almost down to zero speed. I played around a little bit with active braking but it doesn't seem necessary unless you are trying to lock the rotor in a certain position.
 
Ugh I killed the controller in a really dumb way. The battery postive leg on the electrolytic capacitor cut through the thermal pad and touched the grounded heat sink. The plasma from the resulting short managed arc to the 3.3v regulator and killed all the ICs connected to that rail. Unfortunately it looks like some traces were also damaged as the controller was still dead after replacing all the ICs and a freshly programmed mcu as well. It's going to be a real grind putting all the components on to a new board.

zoIVZ28.jpg


OcNSmQP.jpg


Have some good news though. First, the stronger motor mounts have been rock solid. Not a single chain drop since installing them. Second, the big laser cut sprocket works great. It was cut from 4mm sheet so I was worried it would bind without the tooth tips being bevelled. This was not the case and it ran great without any extra work. Top speed is about 63kph and any extra will come from field weakening.

u8fER1H.jpg


B1N5gxJ.jpg


I switched back to the trusty phase runner. Despite the 15uh phase to phase inductance of the motor it runs well. Not having the freewheel at the rear makes a lot of difference and makes tuning much easier. On the 91t sprocket it feels pretty gutless at only 80A phase but it does perk up a bit as speed increases. I've been abusing the poor PR and have been riding around WOT and spending too much time in thermal rollback. Should probably attach it to some led heatsinks. Feels a bit more lively with the 115t which is encouraging. With a proper controller it should be a wheelie monster. :twisted:
 
Bike has been running well but still on the phase runner so power and torque are quite low. Added a 3d printed chain guide for the rear sprocket. Printed out of nylon so hopefully it has a decent life time. Chain now sounds a bit happier when going over bumps and while accelerating through corners, although the chain guide can be a bit noisy while the chain is rattling against it.

3xzJQIk.jpg
 
I'm thinking of something similar with the (I think inrunner) QS motor 2000w mid drive it has 120 stator diameter and 70 magnet height. It looks very high quality. Some guys use the 3000w motor in dirt bike conversions, it can take 15kw easy (11kg), so the 2000w(~7kg) should be good for 8kw minimum. They have 4.7:1 internal gear reduction.

Where do you have freewheels in your motor? Or does the big chain always run whenever your bike is rolling?
 
Tommm said:
I'm thinking of something similar with the (I think inrunner) QS motor 2000w mid drive it has 120 stator diameter and 70 magnet height. It looks very high quality. Some guys use the 3000w motor in dirt bike conversions, it can take 15kw easy (11kg), so the 2000w(~7kg) should be good for 8kw minimum. They have 4.7:1 internal gear reduction.

Those QS midmotors usually come with a belt drive, but they don't have internal gear reduction.
Regarding the stator diameter they refer to the outer diameter and not the airgap (as usually).
I think the air gap dia of the 2000W motor will be more like 70mm..
 
madin88 said:
Tommm said:
I'm thinking of something similar with the (I think inrunner) QS motor 2000w mid drive it has 120 stator diameter and 70 magnet height. It looks very high quality. Some guys use the 3000w motor in dirt bike conversions, it can take 15kw easy (11kg), so the 2000w(~7kg) should be good for 8kw minimum. They have 4.7:1 internal gear reduction.

Those QS midmotors usually come with a belt drive, but they don't have internal gear reduction.
Regarding the stator diameter they refer to the outer diameter and not the airgap (as usually).
I think the air gap dia of the 2000W motor will be more like 70mm..

The one I linked has 4.7:1 reduction, and low (3500 max) rpm, it is enough to have 1.5x reduction by chain or belt after it. I have the same with a cyclone, 4500 max rpm, and 6:1 internal. If you put 1.5x on the rear you can go 60kph on 72v before field weakening. So the reduction is easily enough.
 
Too bad the controller got damaged and you could not fix it.
I wonder if that large heatsink on the backside which looks like it covered bus bars and many other parts would have been even necessaray.. Normally it should be enough if you just cool the FET's.

Are you going the build same controller again? In the forum you shared the link i have seen there is an improved version of this VESC available now 8)
 
madin88 said:
Too bad the controller got damaged and you could not fix it.
I wonder if that large heatsink on the backside which looks like it covered bus bars and many other parts would have been even necessaray.. Normally it should be enough if you just cool the FET's.

Are you going the build same controller again? In the forum you shared the link i have seen there is an improved version of this VESC available now 8)

I'd agree that the heatsink on the fets should be plenty given how much copper is attached to the pcb. This controller was meant for fighting robots so they would have zero airflow unless they use fans. I'm going to order pcbs for the new version. The active components are all the same as well as the DC link ceramics. Screw terminals for phase and battery is a nice improvement and tvs diodes for over current and over voltage faults should shut everything down much faster. It looks like they ended up going with the same 80v common mode current sensors though. If they used isolated current sense then it would be a legit 20s controller since gate drive and fets are 100v rated. That would make it competitive with something like the asi bac4000. Which I've also been drooling over.
 
Trying a 105t sprocket which is the largest 219 off the shelf tooth size I can find. It seems a lot quieter than my laser cut sprocket. Also less runout than the laser sprocket. Not sure why the laser cut sprocket has so much runout but it is a bit annoying to tension.

MevDYcc.jpg


Also ditched the crappy chain roller which was very noisy and wobbly for a nicer polyurethane one.

vMUqxDI.jpg


With these two changes and some chain lube drivetrain noise has decreased a lot. It's still louder than a bafang bbs but really quiet for an rc outrunner build.

I have an action camera on the way so should be able to do some ride videos soon. I also plan on buying the asi 2000 or 4000 controller. I still plan on building the new version of the vesc but now I can take my time with it and do it over the holiday period.

EDIT: ASI 4000 is on the way :D. To better feed the controller I've also got some high c rate lipo on the way. I'm normally not a fan of lipo for safety and reliability but my aging 25R cells heat up really fast above 6kw.
 
Can you go into more detail or show some more pictures of your bottom bracket and mounting plate setup? I had the same idea of mounting a motor in front of the bottom bracket and driving the rear wheel directly on the left side.

Specifically, it looks like you are running an ISIS bottom bracket? Is it the stock BB that came with the bike or did you have to use something wider? And your mounting plates are clamped by the BB cups to your frame? What thickness steel/aluminum did you end up using for the different pieces of your mount?

Thanks for your time.
 
thepronghorn said:
Can you go into more detail or show some more pictures of your bottom bracket and mounting plate setup? I had the same idea of mounting a motor in front of the bottom bracket and driving the rear wheel directly on the left side.

Specifically, it looks like you are running an ISIS bottom bracket? Is it the stock BB that came with the bike or did you have to use something wider? And your mounting plates are clamped by the BB cups to your frame? What thickness steel/aluminum did you end up using for the different pieces of your mount?

Thanks for your time.

I'm using the cyclone freewheel cranks and 148mm isis BB, same as when I was using the cyclone kit running power through the gears. The motor and shaft are most likely too wide for a standard mtb crankset.

The alloy part mounted to the BB and down tube is 5mm 5005. The motor mount is 3mm mild steel.

With the slots cut into the motor and BB mount, it's sort of like a universal motor mount. The motor can be adjusted along the down tube or left and right.
Tj61vJ3.jpg
 
district9prawn said:
Trying a 105t sprocket which is the largest 219 off the shelf tooth size I can find.

Would you mind to let us know the supplier?
EDIT: ASI 4000 is on the way :D. To better feed the controller I've also got some high c rate lipo on the way. I'm normally not a fan of lipo for safety and reliability but my aging 25R cells heat up really fast above 6kw.

cool. looking forward to see how this combi works.
 
madin88 said:
Would you mind to let us know the supplier?

Got it from cometkarts in the US.

ScooterMan101 said:
What is the reason for switching to a ASI 4000 Controller over a custom 18fet VESC ?

Main reason is that the vesc died. Also when I installed the phase runner afterwards I found the braking and low speed performance was a bit smoother. I previously used the vesc on the derailleur mid drive as I found it coped very well with the sudden changes of speed/freewheel engagement. I think its related to the fact the vesc's foc state observer doesn't rely on rotor speed to calculate rotor position. Also I want to do a lot more tinkering with the vesc code and hardware so a nice bulletproof controller on the bike is helpful.
 
Controller finally arrived from ERT after being probed by US customs for two bloody weeks. :evil:

73FEWjH.jpg


I've been busy in the mean time. Moved all the parts over to a 27.5 plus hard tail. A bit better off road, a sturdier frame and most importantly, better high speed stability.

CeQoZFx.jpg


Controller setup was fairly straight forward. I mostly just copied my settings over from the phase runner and adjusted current limits accordingly. For current and speed loop PI settings, I started with the ASI recommended values for a high erpm motor and low inductance motor. These values worked fine with not a single over current fault so far. I plan to stick to sensorless for now.

A common problem with the phase runner (bac 800) is that it can be sensitive to over current faults while tracking a rotor suddenly chainging rpm. I've encountered this in offroad riding under loose or muddy conditions when traction is lost and the rear spins out. I've gone for a couple of offroad rides in loose sandy conditions and haven't triggered a fault from breaking traction yet. Whatever ASI changed it seems to be working.

I got some lipo to play around with for Christmas as well. Combined with this new controller the bike is now scary fast :twisted:
 
Back
Top