high rpm motor and controller

[spkpn]

Hi all,

for my second conversion, I am desperately looking for a BLDC motor and controller combo that has the following properties:

1. motor goes up to 8000RPM
2. torque control
3. 4-8 kW
4. compatible with system voltage in the range of 36, 48 or 60V
5. motor and controller < 2000$

the reason that i need such high rpms is that I want to use the original cvt of the converted scooter.
One possible motor is Joby JM1S, the problem is that I cannot find any sensorless controller that has torque control (speed control somehow defeats the purpose of a CVT imho).

Thanks for any advice.
rgrds,
spkpn

 

[Lebowski]

should be possible.... 22 poles at 9000 rpm means you need 100000 e-rpm capability.

With my controller IC I spun a small RC motor at 50000 erpm while clocking my algorithm
at 40kHz. Clocking the algorithm at 100kHz would make 125 k-erpm, but 50k-erpm at 40kHz
is not the limit (I chickened out because 7000 rpm with a wobbly motor is really scary).

The controller IC has torque throttle so that's OK too. Sensorless start of the motor
is the most difficult, it's an option in my IC and will probably work with the Joby motor
but I cannot guarantee this. In a twist-and-go scooter sensorless push start is not an option
(it has a kickstart maybe ?), glueing in 3 hall sensors doesn't look difficult though.

To use my IC you need to find a 60V 6kW output stage... Or you could buy a cheap
(speed controlled) 60V 6kW controller and swap out the controller chip. Why not
go to 100V, then an 18 FET controller should be enough ?

 

[gwhy!1]

Hi all,

for my second conversion, I am desperately looking for a BLDC motor and controller combo that has the following properties:

1. motor goes up to 8000RPM
2. torque control
3. 4-8 kW
4. compatible with system voltage in the range of 36, 48 or 60V
5. motor and controller < 2000$

the reason that i need such high rpms is that I want to use the original cvt of the converted scooter.
One possible motor is Joby JM1S, the problem is that I cannot find any sensorless controller that has torque control (speed control somehow defeats the purpose of a CVT imho).

Thanks for any advice.
rgrds,
spkpn

A CVT works on the principle of the speed of the motor, I cant quite see why you would need torque control :?

 

[spkpn]

Lebowsky, thanks alot for your input!

22 poles at 9000 rpm means you need 100000 e-rpm capability.

isnt e-rpm=poles*rpm=198000 ? (or is it pole pairs?)

Sensorless start of the motor
is the most difficult, it's an option in my IC and will probably work with the Joby motor
but I cannot guarantee this. In a twist-and-go scooter sensorless push start is not an option
(it has a kickstart maybe ?), glueing in 3 hall sensors doesn't look difficult though.

I thought that it might work since the scooter has a centrifugal clutch, so at rpms below something like 3000 there is only friction of the cvt that has to be overvome. And yes there is also the possibility of the kickstart, funny idea an electric scooter that is turned on with kickstart I like .

With my controller...

could I buy one of your controllers?

PS: stupid question, what is "IC"
PS2: we live in the same city^^

thanks again,
skpkn

 

[spkpn]

also to you, thanks for your remark.

A CVT works on the principle of the speed of the motor, I cant quite see why you would need torque control

I imagine that if I have speed control, then whenever I open the throttle to whatever position indicationg a higher rpm than the momentary rpm of the motor, then the motor will start to spin up with a given torque (maybe depending on rpm). However, it seems difficult to achieve soft acceleration in that setting as it seems that there would always be a hard "jolt" when the acceleration begins. IMHO the percentage the accelerator is applied should map to the force of acceleration, not to a speed at which acceration will decease..


rgrds,
spkpn

 

[Lebowski]

Lebowsky, thanks alot for your input!

22 poles at 9000 rpm means you need 100000 e-rpm capability.

isnt e-rpm=poles*rpm=198000 ? (or is it pole pairs?)

Sensorless start of the motor
is the most difficult, it's an option in my IC and will probably work with the Joby motor
but I cannot guarantee this. In a twist-and-go scooter sensorless push start is not an option
(it has a kickstart maybe ?), glueing in 3 hall sensors doesn't look difficult though.

I thought that it might work since the scooter has a centrifugal clutch, so at rpms below something like 3000 there is only friction of the cvt that has to be overvome. And yes there is also the possibility of the kickstart, funny idea an electric scooter that is turned on with kickstart I like .

With my controller...

could I buy one of your controllers?

PS: stupid question, what is "IC"
PS2: we live in the same city^^

thanks again,
skpkn

I looked at the datasheet of the motor via the link you posted. I counted 22 magnets on the
stator. Assuming they alternate north - south - north etc you'l get 11 sine-waves per rotation,
or 11 e-rotation per mechanical rotation...

Hey, you're also in the Zurich area ? Cool

IC means Integrated Circuit, its basically the brains that measures all motor parameters
and sends the correct signals to the power stage to make the motor turn.

You could buy an off-the-shelf 18 FET controller and we could try a brain transplant ....

 

[spkpn]

You could buy an off-the-shelf 18 FET controller and we could try a brain transplant ....

any suggestions on which one would make a good host for your IC aka "brain"?

Hey, you're also in the Zurich area ? Cool

quartier friesenberg to be specific

 

[Lebowski]

any suggestions on which one would make a good host for your IC aka "brain"?

quartier friesenberg to be specific

I don't know if you have the battery already ? I cannot guarantee it'll work 'cause
I only have experience with the stuff I build myself but, I would chose a 18 FET Lyen
('cause they're all the same except for the hi-power modifications Lyen makes, and
he is the only seller I know where you can get a standard e-bike controller). If the
chosen Lyen controller can deliver the power at the voltage you need than that takes
care of the properly designed and modded output stage. Compared to the output stage
the rest you need for my controller IC is relatively simpel stuff, most or all of it will
already be available inside the Lyen controller (like the 5V and 15V supplies).

I know that area, I used to work in the Binz and walk there during lunchbreak
I'm in Adliswil.

There's a guy somewhere in the area that converts old Vespa's to electric but afaik there's no CVT in those....

 

[gwhy!1]

you can archive very smooth acceleration using a cvt ( no need for a torque based speed controller ) the initial engament goes through a CF clutch and this slips initially from point of engagment for around a additional 1000rpm ( this will be equivalent to a torque controller ) but after this initial engagement you will just need to alter the rpm, the ramps ( on the variator ) and torque spring takes care of everything else once you are moving. When using a electric motor the initial clutch engagement needs to be a quick as possible ( removing the springs works very well )

 

[spkpn]

there seem to be also 24FET LYEN controllers, are those even more suitable for the enterpise?

24 FET 10000Watts Muscular High Current LYEN Controller $299

I don't know if you have the battery already ?

I have a123s for my other conversion, once I figured out the best way in building packs of them, I plan to also use them in this conversion (TBD: 16s3p or 16s2p). I want to run a low voltage setup for savety reasons, this scooter will be used from some of my familiy members none of which I want to see electroshocked

the initial engament goes through a CF clutch and this slips initially from point of engagment for around a additional 1000rpm ( this will be equivalent to a torque controller ) but after this initial engagement you will just need to alter the rpm, the ramps ( on the variator ) and torque spring takes care of everything else once you are moving. When using a electric motor the initial clutch engagement needs to be a quick as possible ( removing the springs works very well )

you sure have a point there...Now I want to try both and see wether there is a noticable difference in torque vs speed mode 8)

again, thanks to both for your valuable help.

 

[spkpn]

checking on the available motor windings again, I might have to go with a 72V set up (24s1p orf 24s2p).

 

[Lebowski]

I used to have a 150cc scooter with a CVT (hell, my car has a proper belt-cvt ).
On the scooter I always had the feeling the motor rpm was dead constant. The cvt
was operated based on the rear-wheel rpm. I thought about it and still think it is
possible to accellate while keeping motor rpm constant and to have the motor torque
determine vehicle speed...

 

[gwhy!1]

I used to have a 150cc scooter with a CVT (hell, my car has a proper belt-cvt ).
On the scooter I always had the feeling the motor rpm was dead constant. The cvt
was operated based on the rear-wheel rpm. I thought about it and still think it is
possible to accellate while keeping motor rpm constant and to have the motor torque
determine vehicle speed...

Yes a cvt adjusts the gearing to the rear wheel speed , but by increasing the rpm of the motor will force the cvt to shift higher up the ramps ( nothing to do with torque ) . If the motor can not supply the power then the gearing (rear wheel speed ) will stay the same. The function of the torque spring at the rear of the cvt is already doing the torque control for the acceleration and regulates the speed of the rear wheel not the the front part of the cvt ( ramps ).

 

[Arlo1]

I used to have a 150cc scooter with a CVT (hell, my car has a proper belt-cvt ).
On the scooter I always had the feeling the motor rpm was dead constant. The cvt
was operated based on the rear-wheel rpm. I thought about it and still think it is
possible to accellate while keeping motor rpm constant and to have the motor torque
determine vehicle speed...

Yes you are correct its the rpm of the rear wheel that determines the clutch shifting and If you set the clutches up right once moving the rpm will stay the the peak of the HP curve till top speed. I think torque throttle will be better for CVT!

 

[gwhy!1]

I used to have a 150cc scooter with a CVT (hell, my car has a proper belt-cvt ).
On the scooter I always had the feeling the motor rpm was dead constant. The cvt
was operated based on the rear-wheel rpm. I thought about it and still think it is
possible to accellate while keeping motor rpm constant and to have the motor torque
determine vehicle speed...

Yes you are correct its the rpm of the rear wheel that determines the clutch shifting and If you set the clutches up right once moving the rpm will stay the the peak of the HP curve till top speed. I think torque throttle will be better for CVT!

Once the clutch is engaged unless the rpm from the motor drops to around 3krpm it will stay engaged, its the roller weights and torque spring that needs to be set to keep a ice motor at is peak power through out the speed range. I am open minded about the whole torque controlled throttle but I still really cant see if it would be of any more use than just a rpm based throttle.

 

[Lebowski]

Once the clutch is engaged unless the rpm from the motor drops to around 3krpm it will stay engaged, its the roller weights and torque spring that needs to be set to keep a ice motor at is peak power through out the speed range. I am open minded about the whole torque controlled throttle but I still really cant see if it would be of any more use than just a rpm based throttle.

Yep, but the clutch and roller weight assembly are at different places.
The clutch is at the motor side and responds to motor rpm.
The roller weights determining the cvt ratio respond to rear-wheel rpm.

If the weights are such that a 10% wheel speed increase changes the gear
ratio by 20% you can actually have the scooter accelerate with the motor rpm dropping...

 

[gwhy!1]

Once the clutch is engaged unless the rpm from the motor drops to around 3krpm it will stay engaged, its the roller weights and torque spring that needs to be set to keep a ice motor at is peak power through out the speed range. I am open minded about the whole torque controlled throttle but I still really cant see if it would be of any more use than just a rpm based throttle.

Yep, but the clutch and roller weight assembly are at different places.
The clutch is at the motor side and responds to motor rpm.
The roller weights determining the cvt ratio respond to rear-wheel rpm.

If the weights are such that a 10% wheel speed increase changes the gear
ratio by 20% you can actually have the scooter accelerate with the motor rpm dropping...

the rollers (ramps) are at the motor end and the clutch is is at the other end of the belt along with the torque spring, the clutch only is needed for putting the ice into its power band for pulling away and for disengaging the engine at standstill its not used at all once moving.

 

[spkpn]

I think this is a good example for a question that is best resolved by experiment...so I will go torque controll and then we can compare our experiece gwhy

 

[Arlo1]

Did you guys see my hybrid scooter build??? My rpm will be higher then yours

 

[Lebowski]

I think this is a good example for a question that is best resolved by experiment...so I will go torque controll and then we can compare our experiece gwhy

If you want to try and convert a Lyen controller to use my controller IC, send me a private message...

 

[Lebowski]

just ran some tests

Assuming you want 100000 erpm using 20 kHz PWM (I'm assuming a Lyen uses 20 kHz PWM)
and 40kHz loopfrequency I tried a scaled test as I dont have a 100000erpm motor. I scaled
the test down to 50000 erpm.

50k-erpm with 10 kHz PWM and 20 kHz loopfrequency on a small RC motor worked fine

 

[Arlo1]

just ran some tests

Assuming you want 100000 erpm using 20 kHz PWM (I'm assuming a Lyen uses 20 kHz PWM)
and 40kHz loopfrequency I tried a scaled test as I dont have a 100000erpm motor. I scaled
the test down to 50000 erpm.

50k-erpm with 10 kHz PWM and 20 kHz loopfrequency on a small RC motor worked fine

I belive it is below 15000 PWM and the max e-rpm is 120,000 for the china controllers. I scoped some things on one last year and I was able to see 65uS on time during PWM (might be higher thats just the most I could see) Marko said he thinks they are 120,000 erpm max!

 

[CrazyJerry]

The CVT might not be a bad idea but keep in mind that slippage will manifest as heat on the drive unit (more-so than the driven unit) and that is a waste of energy. If it's set up right this should be minimal. There are a variety of spring/weight combos for many of the CVTs out there. Depending on the model/make there is also the option of opening the drive end of the CVT up and welding additional weight depending on your needs - just be sure to balance the assembly for longevity. To better capture the torque curve of the diesel that's what I had to do and been using this setup on the diesel bike conversion since 2005
( Changzuki : http://www.diesel-bike.com/Diesel_Bike/Driver_rear.JPG )

If you are looking to automatically keep the electric motor in a tight rpm range throughout the mph range, the CVT can do a pretty good job assisting you. They are not inexpensive unless you can find one used.

~CrazyJerry

 

[Lebowski]

just ran some tests

Assuming you want 100000 erpm using 20 kHz PWM (I'm assuming a Lyen uses 20 kHz PWM)
and 40kHz loopfrequency I tried a scaled test as I dont have a 100000erpm motor. I scaled
the test down to 50000 erpm.

50k-erpm with 10 kHz PWM and 20 kHz loopfrequency on a small RC motor worked fine

I belive it is below 15000 PWM and the max e-rpm is 120,000 for the china controllers. I scoped some things on one last year and I was able to see 65uS on time during PWM (might be higher thats just the most I could see) Marko said he thinks they are 120,000 erpm max!

If you scale up my test I've basically shown that with 100kHz loopfrequency
and 50 kHz PWM it can do 250,000 erpm (and this is sensorless vector control with full sinewave out !).
With a not-wobbly motor and someone braver than me :? the limit will probably be higher.

This was with the little RC motor of the video in my controller IC thread, so with only 5 windings
on one tooth per phase.. Slowing it down by hand I managed to put 120 W in this little motor
with uncooled (and only lukewarm) 4115's on 65 V

 

[Arlo1]

Im with you leblowski. Your ic is awesome but its not worth using on these china boards!