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

[e*clipse]

Sorry, no.

So far I haven't found anything like that for these motors, in all my searches for simple numbers like resistance and inductance.

If anyone finds anything I would love to see it. Maybe even a paper about a buried permanent magnet synchronous motor?? Perhaps with some of our own tests we could fill in the blanks of a more general model?

- E*clipse

[MPaulHolmes]

I've got the 700v worth of 12v SLA batteries for the test. Well, I should probably keep the test to maybe 350v or 400v with 600v igbts. After a few more days I'm going to see if I can hook that thing up and first get the PI loop tuned in a locked rotor condition. I'm sure I'll need help knowing what wire to hook where for the resolver board and the motor. My understanding is, the control is identical to a PMSM except that you change the Id and Iq as rpm goes up. It's not as simple as Id = 0, Iq = throttle. But If I remember, it's just a formula. No big deal! haha. What sort of oil goes into it?

SENSORLESS DONE! Now for debugging... (in other words, sensorless 1% done. LOL)

[thingstodo]

Originally Posted by e*clipse

Ok, now I'm very confused...

Sorry - I do that to everyone!

Perhaps you're talking about an induction motor? They do all sorts of things that I honestly don't understand.

The stator or stationary part of the MGR should be similar to an induction motor - coils with inductance and resistance the adds up (vectors) to impedance? The rotor is magnets so the fields are not induced .. simpler and no weird magnetic coupling ...

The motor that I'm using (the MGR motor) has very similar specifications to the newer (2010) Prius motors. The Prius motor is a slightly larger motor (60kW vs 50kW) with the same number of poles and very similar windings (number of wraps, wire guage, etc) The rotor is nearly identical, just scaled to be slightly larger. I'm not planning to rewind the motor, just use it as it was designed. That's why I'm stuck with the high voltage.

OK - got it. I was wondering where the 4 pairs of poles and such came from ... my mistake ... assumptions!

You can see there's huge amount of BEMF @ 13,000 RPM - about 475V.

So you only get current from applied voltage minus BEMF, right?

(BTW - how do you relate RMS line-neutral to the Bus voltage?? )

For AC, peak to peak is your bus voltage. One phase (or 2) connected to + and one phase (or 2) connected to -. RMS is peak / 1.414

Say you're supplying 650V, the voltage difference would only be 175V. According to ORNL's locked rotor tests, the amount of current required to get 35Nm of torque ( the motor's aximum output @ 13,0000 RPM ) is about 40A. I know it's not this simple, but just using Ohm's law says the maximum impedance for 40A would be 4.375Ohms.

Yes. 4.375 ohms at what phase angle? https://en.wikipedia.org/wiki/LC_circuit

It gets messy

Now here is where I'm really confused - is there an increase in impedance -because of the increased frequency - and/or a decrease in effective voltage that limits the maximum speed? (assuming they won't physically explode) Also, to further complicate things, these motors provide a large % of reluctance torque - greater than 50% at high rpm.

There is an increase in impedance due to frequency, and it also changes phase angle

There is a decrease in the effective voltage due to BEMF

So I guess the answer is .. BOTH

I am not going to take a shot at the magnetic circuit effects - switched reluctance design and effects seem to me to be more of a black art than actual engineering and science.

[e*clipse]

Originally Posted by MPaulHolmes

I've got the 700v worth of 12v SLA batteries for the test. Well, I should probably keep the test to maybe 350v or 400v with 600v igbts. After a few more days I'm going to see if I can hook that thing up and first get the PI loop tuned in a locked rotor condition. I'm sure I'll need help knowing what wire to hook where for the resolver board and the motor. My understanding is, the control is identical to a PMSM except that you change the Id and Iq as rpm goes up. It's not as simple as Id = 0, Iq = throttle. But If I remember, it's just a formula. No big deal! haha. What sort of oil goes into it?

SENSORLESS DONE! Now for debugging... (in other words, sensorless 1% done. LOL)

Oh! Geez! this is awesome!

I'll set things up with a motor so I can (hopefully) keep up with you.
Did I give you the circuit drawings & info w/ that adapter board? I'll try to sort that out. One big thing - it's output is analog - so the interface program w/ the QEI pins will need to change.
Regarding Id and Iq, you can think of it as two curves - Id and Iq varied to get the best current angle. ( I think it's related to rotor speed ). The PM torque decreases along an arc from 0* to 90* electrical. The reluctance torque follows a sine-like curve, peaking at about 45* electrical. The total motor torque is a sum of the two curves. Frankly I'd be turning cartwheels at this point if I could get the motor to simply "go around" using the Id in sensorless mode.

A question that I can handle - LOL - The motor uses ATF.

[MPaulHolmes]

We should tell it to stop using ATF before it gets cancer, ruins its liver, and shoots somebody. haha

EDIT: I should have said sensorless induction done. But I read a couple papers about sensorless permanent magnet, and it wasn't worse. It was the same principles, just with a couple different things. Sensorless hybrid switched reluctance/pmsm is probably just a tweak of permanent maget.

EDIT2: I think you did give me that stuff, but the room is taken over by our house guests (1 mom and 5 daughters) haha. I'll have to use a snow shovel to get to the motor.

[e*clipse]

Originally Posted by MPaulHolmes

We should tell it to stop using ATF before it gets cancer, ruins its liver, and shoots somebody. haha

EDIT: I should have said sensorless induction done. But I read a couple papers about sensorless permanent magnet, and it wasn't worse. It was the same principles, just with a couple different things. Sensorless hybrid switched reluctance/pmsm is probably just a tweak of permanent maget.

EDIT2: I think you did give me that stuff, but the room is taken over by our house guests (1 mom and 5 daughters) haha. I'll have to use a snow shovel to get to the motor.

To me, controlling any form of synchronous motor (BLDC, SR, etc )would be easier than induction because you don't have to deal with slip. However, getting an induction motor to run seems pretty easy - just plug it in LOL! Controlling and induction motor, OTOH is still voodoo!

I've been shoveling out my workspace of mountains of old notes, etc. Some are actually important - like how FOC works. That is a pile by itself. While I'm digging I'll attempt to find the resolver driver stuff. I'm going to need it soon myself anyways....

I did finally find a plug so I can connect the ICD2 to the 5 inline pins on the driver board. It needs an adapter, because the old ICD2's used an ethernet type plug.

[e*clipse]

Originally Posted by thingstodo

Sorry - I do that to everyone!

No worries! It's making me think. LOL!

The stator or stationary part of the MGR should be similar to an induction motor - coils with inductance and resistance the adds up (vectors) to impedance? The rotor is magnets so the fields are not induced .. simpler and no weird magnetic coupling ...

That's how I understand it. Physically, the stators are identical.

OK - got it. I was wondering where the 4 pairs of poles and such came from ... my mistake ... assumptions!

So you only get current from applied voltage minus BEMF, right?

Yes - that's what demands the high voltages.

For AC, peak to peak is your bus voltage. One phase (or 2) connected to + and one phase (or 2) connected to -. RMS is peak / 1.414

Ok,actually this question applies to any 3 phase motor: The ORNL test said that the BEMF was averaged line to Neutral for the three phases. Assuming a wye connected motor (the MGR is wye connected) how do you find the neutral? I've looked for a common connection point, and can't find it, even when I took the motor apart. Also, if it is line to neutral, isn't that just the Vdrop of one phase??
Now I'm REALLY confusing myself, because this obviously won't work. I mean, at 13,000 rpm, they say the line-to-neutral RMS voltage is 475V. But - the motor runs with 650VDC applied at the bus. If a bus voltage of 650V is used, then 650/1.414 = 459.7Vrms I mean 459.7Vrms < 475Vrms.

**Edit** Is this where field weakening comes into play, effectively reducing the field strength of the rotor field and thus reducing the BEMF?

Yes. 4.375 ohms at what phase angle? https://en.wikipedia.org/wiki/LC_circuit

It gets messy

Good point - and I'm confused already without throwing a capacitor into the mix. It does show the value of a low impedance bus capacitor for the motor.

There is an increase in impedance due to frequency, and it also changes phase angle

There is a decrease in the effective voltage due to BEMF

So I guess the answer is .. BOTH

I am not going to take a shot at the magnetic circuit effects - switched reluctance design and effects seem to me to be more of a black art than actual engineering and science.

Believe it or not, I'm actually starting to get a handle on this - I really appreciate your help! I guess if I can start asking dumb questions, at least I have that level of understading.

[thingstodo]

Originally Posted by e*clipse

Ok,actually this question applies to any 3 phase motor: The ORNL test said that the BEMF was averaged line to Neutral for the three phases. Assuming a wye connected motor (the MGR is wye connected) how do you find the neutral? I've looked for a common connection point, and can't find it, even when I took the motor apart. Also, if it is line to neutral, isn't that just the Vdrop of one phase??

L - N can be a theoretical measurement. If your three phase transformer is Y connected, then there really is a neutral point and you can measure to it. If ORNL used an isolation transformer between their controller and the motor (not a normal method, but I've seen it used) then a 460V controller can boost up the output voltage to say 600V. If the transformer is Delta-Y, then the output voltage could be measured at the transformer terminals, Line to neutral.

Each phase is measured line to neutral, so it's the measurement of each .. I guess. Line to line measurements is OK when you have sine waves. When you use pulsed output from a controller, the waveform does NOT look like a sine wave until you put the motor/load on the output terminals. But the line to neutral doesn't really look much better ...

[e*clipse]

Ok - that's interesting - thanks!

I looked closer at the ORNL Prius test motor, and there is an exposed neutral wire. Also, the diagram of the testing apparatus shows the inverter directly feeding the motor - no isolation transformer. From the inverter to the motor, 4 lines are shown, indicating either 3 phases and a neutral or 3 phases and ground.

Unfortunately, my motor doesn't have an easily accessable neutral. I cannot find one on the exposed side of the stator windings. Perhaps there is one on the other side; right now that's hidden by the case. I can probably take it apart further to investigate this - it would be nice to have access to the neutral for testing.

Here is some interesting stuff regarding their test apperatus:

ORNL’s dynamometer test cell and Opal-RT-based speed and current feedback controller were adapted to provide the torque needed at each reference speed. Thus, as the applied torque from the dynamometer was varied manually, the controller regulated the torque producing current appropriately. The current controller consists of two standard proportional-integral (PI) controllers for the direct and quadrature currents, id and iq , respectively. These d-q components are obtained by applying the d-q transformation to the three-phase currents which have a fixed reference for steady state operation. The transformation converts the three-phase currents into two-phase vectors, which have a reference that rotates with the rotor. Therefore, precise rotor position feedback is used during this transformation.

The steady state torque equation for the salient PM machine is expressed by:
Tl = np * ( Ld - Lq ) * i d * i q + np * K * iq , (1)
where
np is the number of pole pairs,
Ld is the d-axis inductance,
Lq is the q-axis inductance, and
K is the back-emf and torque-current factor.

The total torque given by Eq. (1) consists of two torque terms which are reluctance torque and PM torque, respectively. PM torque is produced only by the current component along the q-axis. If current is applied which results with a negative component along the d-axis, positive reluctance torque is developed since the difference, Ld -Lq , in the first term is negative and all remaining variables and constants are positive in the motoring region. In theory, there is an infinite amount of d-q current combinations that will satisfy a particular operation condition. There is an optimal d-q current combination in which the motor efficiency is maximized for each particular torque. It is difficult to determine the optimal current trajectories for the entire torque-speed range, as complex factors such as effects of saturation and harmonics must be considered. Therefore, the DAQ was used to monitor the system efficiency to ensure that the controller is operating optimally.

[thingstodo]

Update for Sep 17

It seems that I owe Paul an apology ...

But first!

foc test 15 showed a slip in the encoder coupler. No useful information.

foc test 16 - run-rotor-test does not complete. The motor is kinda bouncing around. After shutting down power to the high voltage, I cycle power to the controller and play with 1 and 2 for motor control just a bit.

foc test 17 - the encoder cable is run over the motor and may be picking up noise. 5081 is as high as the number gets during run-rotor-test, but it is a successful test. But I think I figured out Paul's algorithmn. After it gets too close to the max rpm, it goes into regen to slow down (jerking the motor) then tries the next settings.

Sorry Paul - I thought this was 'hunting' .. it was anything but!

The encoder cable is not the worst cable I could have selected - cat5 cable. But it is not shielded, it is untwisted for several inches where it enters the encoder, the control board, and the voltage divider board. I think having the encoder cable run over and beside the motor has the encoder signal very noisy and the controller cannot tell what speed the motor is really turning. I think that this affects the rotor test and how smooth the control is.

foc test 18 - the encoder cable is run as far from the motor as possible, the failed cable pin is re-tightened. The results are higher - they top out at 7136. Selecting index 35 with 25 amps going to the motor speeds up the motor acceleration. The regen tops out over 52 amps during the run-rotor-test

Run the motor with 1 and 2 a bit. Check regen - works OK but then reverses the motor

I'll post the video this weekend, but I would call the testing tonight a qualified success!

I guess I need to purchase a quick replacement potentiometer (monday), and a longer-term hall effect throttle test bench replacement.

The command to log currents sounds very interesting ..

EDIT: Add video links

foc 15 https://youtu.be/yuZTMYdZfUk
likely foc 17? https://youtu.be/t7AjgZB0WSQ
foc 18 https://youtu.be/LxKoBbXoY2M