Paul and Sabrina's Cheap 3 Phase Inverter (AC Controller) with Field Oriented Control

[MPaulHolmes]

Holy cow, I take a break from the internet for a day and I miss out on all sorts of stuff! haha. I modified an encoder that was for my old ACIM, but forgot that it has no index pulse, so it won't work. It's a shame too, because it was not easy doing what I did to get that thing to go on a larger post. The mods were extensive.

I don't really understand what the reluctance flywheel thing is, but it would be awesome to get it working and be inexpensive.

W.R.T. field weakening, here's what I did and it works well on the bench with a flywheel:

The allowable voltage can be thought of as a circle of radius R. Let Vd and Vq be the "direct" and "quadrature" voltages. When (Vd, Vq) lies outside that circle of radius R, you don't have enough voltage to allow Id--> IdRef and Iq--> IqRef. So, you figure out the scale factor to shrink (Vd, Vq) down to land right on the circle of radius R. Then, you multiply IdRef and IqRef by something proportional to that scale factor. That's the "field weakening" part. Then, allow IdRef and iqRef to ramp back to where they really want to be (what I call IdRefRef and IqRefRef).

[P-hack]

@paul, wrt field weakening, the end result is full scale but increasingly narrow spvwm waveforms as rpm increases, no? (no is an acceptable response, just trying to understand the end result here).

@cs Are you familiar with DTC?
Direct torque control - Wikipedia, the free encyclopedia , it sounds a little bit like what you are describing.

[Astro]

Originally Posted by MPaulHolmes

...I don't really understand what the reluctance flywheel thing is, but it would be awesome to get it working and be inexpensive...

Looking at the picture it looks like a Crank angle sensor.
The crank angle sensor is mounted close to the teeth on the cars normal flywheel. These are the teeth that the starter motor normally engages with to turn over the ICE engine. Also called the ring gear.
The sensor outputs alternating pulses of positive and negative voltage. A positive voltage as a tooth approaches the sensor, zero transition as the tooth is aligned with the sensor and then a negative pulse as the tooth leaves the sensor.
This gives an AC signal with a frequency the same as the frequency of the teeth passing the sensor.
So if you know the number of teeth on the ring gear you can work out the flywheels RPM and also the amount of angular change.
To get the absolute angle of the flywheel you would need to provide some sort of index. cts_casemod suggested welding two teeth together to get a longer gap between positive and negative pulses on those teeth. This may be difficult to detect or discriminate from a change in RPM especially at very low RPM.
Maybe a rare earth magnet added to a tooth would give a higher voltage pulse for that tooth and make it stand out from the crowd? So you would still get the same pulse interval as the other teeth so angle change and RPM detection wouldn't be affected and have a voltage threshold trigger for the index detection.

[MPaulHolmes]

Originally Posted by P-hack

@paul, wrt field weakening, the end result is full scale but increasingly narrow spvwm waveforms as rpm increases, no? (no is an acceptable response, just trying to understand the end result here).

Yes!

[freebeard]

To get the absolute angle of the flywheel you would need to provide some sort of index. cts_casemod suggested welding two teeth together to get a longer gap between positive and negative pulses on those teeth. This may be difficult to detect or discriminate from a change in RPM especially at very low RPM.

Not as soon as you engage the starter.





I know, I know. But it's a useful place for an airconditioning pump or a pump for the Coanda nozzles. Assuming you even need a ring gear.

[Astro]

Originally Posted by freebeard

...But it's a useful place for an airconditioning pump or a pump for the Coanda nozzles. Assuming you even need a ring gear.

I would have expected running anything from the ring gear would be rather noisy with its square cut teeth. Of course now that you have mentioned it, it does seem a handy spot. I was out in the garage looking at the crank angle sensor i removed during the de-icing of the EV. I noticed the starter motor and had a thought. Maybe i should mount the starter motor back in the vehicle. It could sit there forever doing nothing as it only engages with the ring gear when energised. Then i would have a backup motor in case of any major failure of the traction setup. Enough to get me off a busy road or back up the hill to my garage.
Of course it wouldn't be wired up to the ignition switch, that would be asking for trouble, but wired to a switch somewhere for emergency use.
I have several times had to resort to using the starter motor on a dead ICE vehicle to get it into a garage or onto a trailer. Without the burden of the non running ICE motor the starter motor would probably do a good job as a rescue motor. That it runs off of the 12V supply would mean it would still work even if say a fuse had blown in the HV traction supply.
I am only suggesting seconds of run time not minutes. The starter motor is very over volted at 12V and does not like running continuously. The magic smoke would escape if ran for too long with a large load.

[cts_casemod]

Astro,
I saw that on the past so perfectly viable. Perhaps you want to make it high tech and add an electronic switch just to avoid the startup 'kick' (should be significant without the ice).
I was going that route but i had to take the starter ring to fit my adapter.

Regarding the noisy setup, its not that noisy, its AC and needs a reluctance sensor transducer to read the signal on the zero crossing point. The amplitude is not constant. Or replace that with a hall sensor, with built in logic.

P-hack,

let me clarify my point is observation of the motor parameters under field weakening operation. Feel free to provide stuff you find regarding that operational range (PFC, Slip, etc)

With that in mind, slip is proportional to torque, but not in the same way as during the constant torque/constant flux region. Look at Paul example using the FOC.

Ask yourself what causes the breakdown torque limit. That will help you understand why those concepts are not linear above nameplate.

[Astro]

Originally Posted by freebeard

...I know, I know. But it's a useful place for an airconditioning pump or a pump for the Coanda nozzles. Assuming you even need a ring gear.

Originally Posted by cts_casemod

...Regarding the noisy setup, its not that noisy, its AC and needs a reluctance sensor transducer to read the signal on the zero crossing point. The amplitude is not constant. Or replace that with a hall sensor, with built in logic...

Sorry i thought freebeard was talking about running an air-conditioner pump (A/C) from the ring gear. It was the mechanical noise of the meshing square cut gears that i thought would be noisy in such a set up.
As for electrical noise affecting the sensor reading the ring gear. In the ICE setup it would have had to be immune to quite a bit of electrical noise from the spark plugs, alternator, starter motor etc. So hopefully it should still work ok in a EV set up.
The best thing about a sensor on the ring gear is that many conversions would already have all the bits required. If the flywheel has been removed then maybe not but if the flywheel is still there then it is a low/no cost method if it works.

[e*clipse]

Just wondering why one would WANT a flywheel on an EV??

It's just added inertia - necessary on an ICE; useless - detrimental - for an electric motor.

If you need a crank angle sensor, there are lots of much lower inertia methods - for example Paul's last setup.

Paul - is there any possibility of ( in a useful way ) connecting your new motor to a much smaller shaft? Something like an exposed back shaft?

- E*clipse

[MPaulHolmes]

There is a shaft on back and front. The dang thing sticks out like 1 foot in both directions. haha. 0.875" near the motor, and it gets smaller. I forgot how much smaller. I'll measure it today. Also, way the heck away from the motor (like 8") the shaft has a right angle cut on it, with the other 3/4" of the circumference of a cross section still circular.