Neumotor 8057/75 middrive project, belt drive, adding halls, 10kW+

[madin88]

The plan for the halls is to install them outside by use of 3D printed parts. Going this route makes more sense, because:

- the winding is terminated in delta which means the sensors would belong somewhere into the theeth and not between (from what i know a 30° shift electrically would be needed compared to a star termination)
- having the sensors outside on an adjustable disc menas i can play with the timing to get lowest no-load losses
- strong magnetic fields from the coils caused by large currents will have no influence to the signal quality

As sensors i will use SS413A, and the magnets which will be glued into the rotor are 4x3x1mm N50M.

this is the CAD design:

i printed the rotor yesterday with the 3D printer from my brohter. thats how it looks on the axle:

And i also designed the 108T driven pulley for the belt drive. This will be cnc milled out of Aluminum with grey anodizing.
It wont be cheap, but thats the price for a belt drive, and i don't know how i could get this fabricated in another way (the belt guide flange makes it expensive).

The motor pulley's are on order from Pfeiffer Industries.
One 18T custom made for my 12mm wide belt, and one 17T they had available which is for 21mm wide belt.

 

[madin88]

Good news!

the motor is running fine with the 3d printed hall resolver, attached to a Adaptto controller. Yes, Adaptto!
It's consumption is around 300W when spinning close to 6000rpm, but i will do more measurements on this later..
Tomorrow i will upload some pics.

 

[litespeed]

Awesome!

Tom

 

[madin88]

Here a pic of the resolver. It is 3D printed with ABS so that it doesn't melt at higher temperatures. Don't think that this was that easy.. It took 4-5 redesigns in CAD and prints to make everything fit togehter (especially to get 3x4mm magnets into the slots).
Luckily the printing time was not long due to the small size (took about 15 minutes). For the hall carrier it took only one print.
I soldered a pull up resistor to each sensor as stated in the datasheet, and a small cap to stabilize the 5V supply voltage. This cap may not have been necessary, but i have the meaning you never can have enough caps in the supply lines of any electronics -> the more the better

As you can see the hall signals have properly shape, and the accuracy is not bad. There are small variations in signals length which is probably cuased by not 100% identical distances between the magnets.
But still not bad for 3D printed parts, and as the no load constumption is quite good i can live with that.

 

[madin88]

The staus of this project so far:

Motor: here
Adding Hall Sensors: done
Controller: already on the bike
Belt: here
driver Pulley: on order from Pfeiffer Industries
Dual disc Hub: on order from LMX bikes
17x2,5" tire and tube: here

36h 17x1,4" moped rim: very very hard to find at the moment -> any help for a source is welcome!
Spokes: will be ordered once i have hub and rim
driven pulley: still not sure how to get this done. CNC manufacturing would be very expensive..

 

[macribs]

As you can see the hall signals have properly shape, and the accuracy is not bad. There are small variations in signals length which is probably cuased by not 100% identical distances between the magnets.
But still not bad for 3D printed parts, and as the no load constumption is quite good i can live with that.

Nice work. Looking forward to see this completed. Left side drive?

What impacts can variations in signal length give? I mean there have been reports of people swapping out original hall sensors with better quality ones and a more careful placement that resulted in a motor that run smoother, with less heat and less load on controller due to better timing. Think the variations you see is big enough to impact the result in a negative way? Btw, have you ever measured say a hub motor or any other motor with original halls to see how accurate the wave pattern/signal length are, to use that as a reference point to this motor?

Will you be running Adaptto max-e with this motor, or are you set on using maybe another controller?

Do you think others will benefit from swapping standard halls with a setup like you have here? Timing, heat, etc?

 

[madin88]

Nice work. Looking forward to see this completed. Left side drive?

right side drive. instead of the freewheel so no more pealling (but this bike is anyway registered as moped).

What impacts can variations in signal length give?

It means the controller will drive the phases (push current through them) either a tiny bit to early or to late. It is micro senconds.

think the variations you see is big enough to impact the result in a negative way?

I cannot give a detailed answer as i have no comparison, but the motor was spinning at correct RPM and the no load consumption was not bad.
Aside from vibrations which came from the bell (it is not balanced unfortunately) it did run very smooth and hang nice on the throttle.

Btw, have you ever measured say a hub motor or any other motor with original halls to see how accurate the wave pattern/signal length are, to use that as a reference point to this motor?

I own this oscilloscope only for a few days now and aside from that i just measured signals from a QSV3 hub motor.
The signal had a differnt wave form, but in view of the midline-crossings it was similar.
If you look at the pic i shoot from the osci you see how the signal should look like with NEUTRAL timing.
When the phase voltage goes above the midline (or where the sine curve is crossing the x-axis), the hall signal should go up (5V), and if the voltage goes below the midlind the singal also should go down (0V).

Will you be running Adaptto max-e with this motor, or are you set on using maybe another controller?

Yes, because Max-E and 20s battery is already on the bike. And as it seems to run this motor well, there is no need to look for a differnt controller.

 

[madin88]

Here as an example the position sensors of the D7500 motor:

or QS mid-drive motor:

 

[MotorboaterBen]

Also testing out the Neu 80xx series motors here. I have an 8019 (pictured) paired with an Emsiso emDrive 150, and then an 8057-75 up next for testing. I outfitted the motors with an RLS RM44AC sin/cos encoder - super straightforward to do with a bored hole in the shaft for the magnetic actuator and a 3D printed part to mount the encoder.
Currently playing around with P and I gains, will be setting the motors against each other soon with a torque sensor between (my home-made mini dyno) to get some data.

 

[madin88]

Also testing out the Neu 80xx series motors here. I have an 8019 (pictured) paired with an Emsiso emDrive 150, and then an 8057-75 up next for testing. I outfitted the motors with an RLS RM44AC sin/cos encoder - super straightforward to do with a bored hole in the shaft for the magnetic actuator and a 3D printed part to mount the encoder.
Currently playing around with P and I gains, will be setting the motors against each other soon with a torque sensor between (my home-made mini dyno) to get some data.

That sounds impressive. Do you run a boat with the 8019?
It would be great if you could share some pictures of that encoder, and how you managed to install it.
Looking forward for the dyno results.

At the moment i am waiting for a new shaft with keyway, and the large driven pulley.
I think in about 3-4 weeks i can install everything on the bike and do a test run.

 

[MotorboaterBen]

Not running a boat with the 8019, but will probably through the 8057 on an outboard at some point for fun.....
I did have Neu 4400 on a boat back in college that moved pretty quick: https://www.youtube.com/watch?v=QGvg7ujdlDo

A little more detail on the encoder:

1. I used an RM44AC sin/cos analog encoder from RLS: https://www.rls.si/eng/rm44-up-to-13-bit-encoder-base-unit
2. This is a two part encoder, one part is the actual encoder/reader and the second part is the magnetic actuator, which sits in and rotates with the shaft.
3. I used a precision ream to bore a hole in the back end of the motor shaft to seat the actuator with loctite.
4. I made a 3D printed part to house the encoder and set it the optimal distance from the actuator.
5. And it all works!

Attaching a few pictures here:

 

[madin88]

I did have Neu 4400 on a boat back in college that moved pretty quick: https://www.youtube.com/watch?v=QGvg7ujdlDo

Wow that boat looks quick. Thumbs up for such impressive build!
The datasheet of the encoder reads well and i like how easy it can be added with that small magnet.

Let me know how your setup with emDrive controllers works and if you get similar no-load consumption as i get with Adaptto.
The problem with Adpatto is if you adjust the settings, the numbers jump like 0,53° to 0,99° to 1,3° as for instance with nothing in between.
Or jump from 399µsec to 460µsec, 530µsec for indTiming (hall sensor delay).
Now if a value of 430µsec would be optimal, you are fu*** up.
I see roughly about 300-400W on the Adaptto display at 6000RPM, but it could be lower because the motor was vibrating alot beacause of my shaft conversion (it wasn't centered well so the bell was out of balance).

btw: how does your motor run in terms of vibrations/balancing?

 

[ridethelightning]

this is interesting.

so far iv been happy with hubmotors and 15kw, but have always been interested in mid drives for the same reasons- poor road contact on rough, not so good off road etc.

the main deal breaker has been the noise of a chain drive though.
i dont want a whipper snipper after using nice sinewave /hub.

if the belt drive turns out to be practical and quiet, i might go this route myself, maybe mod my nyx frame, make a new swingarm.

iv seen it said that a belt wont last long at 10+kw, but i suppose it depends on the belt quality and depends on the cost, as i wouldnt mind replacing the belt every few months as a tradeoff.

 

[madin88]

The past days i made some progress. I have receiced the LMX dual disc hub (new version with dual row bearings black anodized), the 17x1,4" moped rim from Prowheel and Sapim spokes from Yojimbos garage (ebay seller). But the best thing is the parts for the belt drive also came in.

For building the wheel i used Sapim strong spokes and Holmes moped nipples. The washers on the spoke head should make it more robust at this area.

Adding a platform for a middrive motor onto a normal tubular downhill bike swingarm was no easy task.
I decided to add a 4mm alumnium plate by using liquid steel epoxy glue and plenty of bolts. The powder coat i grinded off for optimal bonding.

The very nice fabricated pulley made of 7000 series aluminum. It has 108T and suits 12mm wide Gates 8M poly chain carbon 8)

 

[madin88]

iv seen it said that a belt wont last long at 10+kw, but i suppose it depends on the belt quality and depends on the cost, as i wouldnt mind replacing the belt every few months as a tradeoff.

If it lasts long or not depends on the torque and RPM of the smaller driver pulley.
There is a software from gates where you can enter all parameters and it will tell you than if it is within the specs or not.

Regarding the program my setup can do around 13kw continuos at about 5000 RPM (which would be about 830 rpm on the wheel).
But if i enter same 13kW at ONLY 1000rpm on the motor, the max recommended motor power drops below 3kW. Basically this means there is a limit in terms of motor torque when using small pulleys like the 18T i am using.
If accelerating hard or climbing very steep hills i will definitely use it byond the specifications. To make it a bit better for high torque i will install an idler near the pulley so that more teeth grip into the belt.

btw: Zero motorcycles are running same type of belt but only 2mm more width (special size for them).
So 12mm vs 14mm or 10kW from my setup VS ??? how much kW has a Zero?

 

[ridethelightning]

Thats coming along really nicely!
great pics as usual.

I think the 4mm alu plate on the swingarm will make a big difference in rigidity.

btw: Zero motorcycles are running same type of belt but only 2mm more width (special size for them).
So 12mm vs 14mm or 10kW from my setup VS ??? how much kW has a Zero?

Luke or dr bass will know that :wink:

were the parts difficult to source? what was the cost of the belt, pully and hub aprox?

 

[skeetab5780]

Sub’d

Nice looking rear wheel!

 

[madin88]

Thats what the Software from Gates puts out if all parameters entered.
Following calculation is at 14kW input (10kW with so called 1.4 service factor) at 5000RPM on the motor. As you can see it can handle above 14kW (output power on the motor shaft, not input power).

But at only 500 RPM and same 14kW, the power should be limited to about 2kW on the shaft to stay within the specs.

 

[Rube]

Hi Maddin, thanks for posting, quite an inspirational build. Interesting use of adhesive. Couple of questions if you don't mind, did you include a 'fence' on only one side of the pulley to aid machining? No need for a fence on the second side?

Looking at the bolt pattern on the swing arm, I'm wondering if you drilled through each side of the chainstay to a second bottom plate?

 

[ridethelightning]

sorry i missed you actually got that pulley custom fabricated.
did you think of making the other side of the belt guide as a seperate ring that bolts on?
im really looking forward to seing this thing working

 

[madin88]

Hi Maddin, thanks for posting, quite an inspirational build. Interesting use of adhesive. Couple of questions if you don't mind, did you include a 'fence' on only one side of the pulley to aid machining? No need for a fence on the second side?

It is called belt guide flange
yes as this pulley is CNC machined i went this route. Normally such guide flanges are separate parts which will be attached to the pulley after the tooth profile was made.
I don't know if it would need a second flange, but if i spin it by hand now the belt stays on the pulley, and afaik the sprocket from a Zero also has only one flange.

Looking at the bolt pattern on the swing arm, I'm wondering if you drilled through each side of the chainstay to a second bottom plate?

No second plate, but plenty of bolts screwed into the tubes and beside them with plenty of "liquid steel epoxy" everywhere and between.
I have done lot's of work on it, drilling holes, fixing it onto the work bench etc and i cannot see any cracks or similar signs of fatique so far.
The Quick Steel has a strength of about 10-20N / mm² and as the surface area is so large i believe it will be up to the task which will be about 150kg pulling force on the belt leading to 200Nm on the wheel.
Time will tell if it holds.

check out the pics!

Thats coming along really nicely!
great pics as usual.

were the parts difficult to source? what was the cost of the belt, pully and hub aprox?

Thank you!
I can tell you that parts for this kind of belt drive are very difficult to find. I think i had luck that Pfeifer Industries had the pulley i needed in stock and even a taper wall bushing for the 14mm axle with a keyway. Custom fabrication for only one or two pcs would have been really really expensive.

The prices:

18T pulley + Taper wall bushing was about 130USD
12mm 8M GT carbon belt 50USD
LMX Hub + cart sprocket carrier about 220USD
12mm axle for 150mm dropouts 20USD
Rim + Spokes 150USD

Aluminum plates + glue + bolts etc about 50USD

 

[madin88]

 

[Punx0r]

But at only 500 RPM and same 14kW, the power should be limited to about 2kW on the shaft to stay within the specs.

Check how much power your motor actually outputs at 500rpm - it will be a lot less than 14kW!

As a sense check I would take motor stall torque, multiply by driver pulley radius and check how much belt-pull force that will give. You may find the belt is actually within the belt spec

 

[madin88]

Today i installed the motor to the swingarm. I wish there would be a bit more clearance between it and the wheel (so i could add the fender i used before), but anyway i'm quite happy that there was standard belt length availabe that suit my needs.
What i'm missing now is the belt idler and a new motor shaft with keyway.
The shaft i got with the motor is faulty as it has some play in the bearings (motor runs noisy with it and has a lot of vibrations). Additionally it measures only 13,8mm at the sprocket side where it should be 14mm.

I told Neumotor about the problem and they will send me a new shaft with a keyway, but they are kinda slow (i called them already 3 or 4 times and email support almost not present).

The thing with the vibrations (it needs balancing when the shaft gets converted) is another thing which needs to be fixed.

 

[ridethelightning]

[/quote]
The prices:

18T pulley + Taper wall bushing was about 130USD
12mm 8M GT carbon belt 50USD
LMX Hub + cart sprocket carrier about 220USD
12mm axle for 150mm dropouts 20USD
Rim + Spokes 150USD

Aluminum plates + glue + bolts etc about 50USD
[/quote]

large driven pully not included in list?
let me guesse, it takes balls of steel just to think about the cost?...

altogether awesome solution in evolution from hub drive. i have no experience here, what do you think the difference in performance dynamics will be compared to the hub, more in terms of power, with the unsprung mass issue aside?

im thinking better accelleration and hill climbing at low speeds, perhaps less torque at higher speed and maybe less top speed?