[njloof]

Originally Posted by efan

this controller+motor seem to be a new development, and I dont know of anyone who has used it..but it may be worth checking them out.

http://shop.greenstage.co.nz/product...-motor-package

Anybody actually seen or used this (new?) AC motor/controller combo? I see the Greenstage guys have a good history here, but this seems to be a first for them, and beats out the current HPEVS Curtis/AC50 combo by a good margin.

Thanks to efan for pointing this out!

[Elegancec]

Hello, Im also very interested into this new drive. From the data it would be my favourite, @njloof, thank you for posting in my thread (http://www.diyelectriccar.com/forums...s-74449p2.html) this great news!

[Tesseract]

Originally Posted by njloof

...I see the Greenstage guys have a good history here, but this seems to be a first for them, and beats out the current HPEVS Curtis/AC50 combo by a good margin....

Hmm... interesting find. Always good to know what the potential competition is up to.

Note that this system uses a *trapezoidal* BLDC motor, not a sinusoidal type like, e.g., the Remy HVH250 motor. Without getting too wonky, the switching pattern needed for trapezoidal BLDC makes it impossible to use any type of field-oriented control algorithm - only 6-step "V/f" is allowed. This means poor low speed torque, and high levels of torque ripple, noise and vibration at all speeds. Here is a good overview of the two different types of BLDC motor.

As a result of these severe shortcomings, trapezoidal BLDC is really only suitable for small fans and servos, not traction applications.

[MalcolmB]

Great link Tesseract!

The only trapezoidally controlled brushless motors I've had experience with are little ebike hub motors. I guess the torque ripple explains why they growl so much under load? But I still don't grasp why they produce poor torque at low rpm, though I know from reading reports on Endless Sphere that this is a big limitation.

According to that link: "Sinusoidal commutation results in smoothness of control that is generally unachievable with trapezoidal commutation. However, while it is very effective at low motor speeds, it tends to fall apart at high motor speeds."
Can you put a rough figure on the upper rpm limit for sinusoidal control, or does that depend on the motor design as well?

Malcolm

[Tesseract]

Originally Posted by MalcolmB

Great link Tesseract!

Yeah, I thought so to when I read it. I've got all kinds of very expensive books on this stuff and none of them come close to explaining the basic difference in the two types of BLDC stator construction as well as that document.

Originally Posted by MalcolmB

The only trapezoidally controlled brushless motors I've had experience with are little ebike hub motors. I guess the torque ripple explains why they growl so much under load? But I still don't grasp why they produce poor torque at low rpm, though I know from reading reports on Endless Sphere that this is a big limitation.

Generally speaking, it's because there are only 6 points where the stator windings and rotor magnets are perfectly aligned in the trapezoidal BLDC; at every other rotor angle some of the phase current goes towards uselessly trying to counteract (or add to) the flux of the rotor magnets rather than moving the rotor through space.

Originally Posted by MalcolmB

According to that link: "Sinusoidal commutation results in smoothness of control that is generally unachievable with trapezoidal commutation. However, while it is very effective at low motor speeds, it tends to fall apart at high motor speeds."
Can you put a rough figure on the upper rpm limit for sinusoidal control, or does that depend on the motor design as well?...

The higher the rotor RPM the higher the stator frequency and therefore the higher the AC losses in the motor, thus the design of the stator sets a practical upper limit on the RPM, assuming the rotor can handle the centrifugal* force, that is. However, the higher the RPM the faster the processor in the controller needs to be to calculate the switching state needed to create the correct voltage vector for the rotor's actual position. In other words, at a high enough RPM the rotor will move a significant angular distance in the time it takes the processor to calculate the voltage vector. A graceful fallback is to drop the number of voltage vectors until you ultimately end up at 6-step operation.

* - the oldest argument on the interwebz is that centrifugal force doesn't exist, but everyone knows what it means so let's pretend that it does.

[njloof]

Originally Posted by Elegancec

Hello, Im also very interested into this new drive. From the data it would be my favourite, @njloof, thank you for posting in my thread (http://www.diyelectriccar.com/forums...s-74449p2.html) this great news!

I can't take credit for @efan posting this... But thanks anyway

[njloof]

Originally Posted by Tesseract

Note that this system uses a *trapezoidal* BLDC motor, not a sinusoidal type like, e.g., the Remy HVH250 motor. Without getting too wonky, the switching pattern needed for trapezoidal BLDC makes it impossible to use any type of field-oriented control algorithm - only 6-step "V/f" is allowed. This means poor low speed torque, and high levels of torque ripple, noise and vibration at all speeds.

Great article, and makes sense -- it describes quite well what I see on my brushless e-bike as well

But... why not filter the discrete Hall signal into a sinusoidal curve? Or, if you've got a decent processor available, predict the true position of the rotor? At worst the motor is hooked to a flywheel, and at best to a 1+ ton car; the change in rotational velocity of the rotor should be relatively small during 60 degrees of rotation...

[Tesseract]

Originally Posted by njloof

...
But... why not filter the discrete Hall signal into a sinusoidal curve?...

Firstly, because the phase angle of a waveform is always shifted when you filter it, but secondly, the poor performance of the trapezoidal BLDC is directly the result of its simpler construction. If you want better performance you need to rewind the stator. No two ways about it. And just to make it extra clear, it doesn't matter if you know exactly where the rotor is, there are still only going to be 6 points equally spaced about the stator where you will get maximum torque for a given RMS phase current.

[njloof]

Originally Posted by Tesseract

And just to make it extra clear, it doesn't matter if you know exactly where the rotor is, there are still only going to be 6 points equally spaced about the stator where you will get maximum torque for a given RMS phase current.

Because you can't excite all three windings independently?

[Elegancec]

So I think its clear, the controller can also only operate with this treapezodial brushless motor and with no other AC motors, because of its simpler design.

The torque curve would be interesting.