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

[Astro]

How about just having a controller per wheel. Then adding a simple (or complex) throttle pre-conditioner circuit. The pre-conditioner would take input from speed sensors on each wheel as well as the main throttle input.
It could generate 4 throttle signals, one for each of the controllers.
The pre-conditioner could be a simple traction control type set up for regular cars and something much more complex for racing situations.
It may also be able to compensate for controller/motor failures.
Switch between 4 wheel drive and 2 wheel drive.
If there is regeneration available it may also be able to use that to give even more control.
With the addition of some accelerometers and gyroscopes it could be heading into the realm of stability control.
Also for reliability, using a throttle pre-conditioner may mean that if it ever failed it could be bypassed and the throttle directly feed the 4 throttles. You wouldn't have all the nice traction/stability control type features but at least you could drive home.
It would also mean standard controllers for whatever type of motor could be used.

Just a thought.

[e*clipse]

Originally Posted by cts_casemod

The prius motor is a hybrid between a switched reluctance and a PMSM. How are you going to do the angle estimations?

exactly!

Please correct me if I'm wrong, but I've only heard of "angle estimators" in regards to induction machines and "sensorless" machines.

I am going to use the resolver that comes with the motor to sense the motor's location and speed. I've made a proto circuit to drive the resolver and filter the resolver's sin and cos outputs for an AtoD converter. The proto works; the board design is currently being made.

According to Toyota's engineer who designed the new generation motors and ORNL, the phase relationship is fixed between the PM and reluctance torque in the Toyota motors. Basically the PM motor produces peak torque at a current angle of 0 degrees. The peak reluctance torque occurs at 45 degrees. The peak combined torque occurs at 35 degrees, but this differs a bit according to conditions like motor speed, etc. This corroborates with some information I've found about synchronous motors in general.

This is the big reason I'm interested in this controller. It's FOC is critical for achieving the full range of RPM and maximum torque with this motor. It's not just hype.

[P-hack]

Originally Posted by e*clipse

It's FOC is critical for achieving the full range of RPM and maximum torque with this motor. It's not just hype.

Unlikely. If you use a SVPWM switching scheme you are going to get pretty much max torque/rpm as it maximizes phase to phase voltage (which is what maximizes RPM/torque). It would be "hype" to think that FOC is "critical" for max rpm, especially in a synchronous machine. Do you have a link?

[cts_casemod]

Originally Posted by e*clipse

exactly!

Please correct me if I'm wrong, but I've only heard of "angle estimators" in regards to induction machines and "sensorless" machines.

I am going to use the resolver that comes with the motor to sense the motor's location and speed. I've made a proto circuit to drive the resolver and filter the resolver's sin and cos outputs for an AtoD converter. The proto works; the board design is currently being made.

According to Toyota's engineer who designed the new generation motors and ORNL, the phase relationship is fixed between the PM and reluctance torque in the Toyota motors. Basically the PM motor produces peak torque at a current angle of 0 degrees. The peak reluctance torque occurs at 45 degrees. The peak combined torque occurs at 35 degrees, but this differs a bit according to conditions like motor speed, etc. This corroborates with some information I've found about synchronous motors in general.

This is the big reason I'm interested in this controller. It's FOC is critical for achieving the full range of RPM and maximum torque with this motor. It's not just hype.

Sure. Angle estimations based on the operational characteristics as that motor generates two types of torque. So you need to calculate the optimum commutation angle, where the maximum torque per amp is generated

Rather than elaborate, on that engineer description, have a look at this video from 16:00 onwords

https://www.youtube.com/watch?v=bZwLFpXhFbI

The bottom end is, if you treat the prius as a PMBLDC you will not be able to get the full performance from it, so it does require some extra work, to which FOC on its own wont solve.

[cts_casemod]

Originally Posted by P-hack

Unlikely. If you use a SVPWM switching scheme you are going to get pretty much max torque/rpm as it maximizes phase to phase voltage (which is what maximizes RPM/torque). It would be "hype" to think that FOC is "critical" for max rpm, especially in a synchronous machine. Do you have a link?

Not only SVPWM, I seen many slip control drives having a resolution of 1Hz or worse. The bottom end is FOC generally uses better hardware and therefore performance of such drives is superior. More yet, there has been a great research on FOC and other vector techniques over the last decade or so and it is generally quite easy to take advantage of the work, for example, sensor-less operation.

But if we compare like by like (same hardware, SVPWM, 3rd harmonic, same waveform resolution, CPU and code refinements) the situation changes, on induction machines, this is.


As to e*clipse description on the Prius motor, the standard FOC libraries have nothing to do with high speed operation, but rather torque. You can test bench the motor and it runs fine, but on the road the torque produced will not be satisfactory since the motor is no longer operating on the optimum point.

FOC, itself, will not be the solution for this problem either, hence my question on #1320 on what algorithm was being used to make sure the commutation happens on the right angle depending on the operational characteristics

Of course this doesn't apply directly to induction. Slip control only works for Asynchronous motors.

(and I re-wrote this paragraph as I assumed we were still talking about induction)

[e*clipse]

Originally Posted by P-hack

Unlikely. If you use a SVPWM switching scheme you are going to get pretty much max torque/rpm as it maximizes phase to phase voltage (which is what maximizes RPM/torque). It would be "hype" to think that FOC is "critical" for max rpm, especially in a synchronous machine. Do you have a link?

You know what, I feel like this is deja vu all over again. I've already posted the links, etc for what I'm talking about. Go look at previous posts. Look up the links or my previous posts about this. Then you can argue with Munihero Kamiya or the folks over at ORNL about this. I'm moving on.

Since you know so much, you can probably find your own links.

[e*clipse]

Originally Posted by cts_casemod

Sure. Angle estimations based on the operational characteristics as that motor generates two types of torque. So you need to calculate the optimum commutation angle, where the maximum torque per amp is generated

Rather than elaborate, on that engineer description, have a look at this video from 16:00 onwords

The bottom end is, if you treat the prius as a PMBLDC you will not be able to get the full performance from it, so it does require some extra work, to which FOC on its own wont solve.

Thanks for the link cts_casemod!

Hmmm, ok - then I guess I mis-communicated. I understood angle estimators to be necessary when you didn't have an accurate way of measuring the motor angle (such as sensorless control) or were using an asynchronous machine.

I am using an extremely accurate position sensor, coupled with current sensors on at least two phases, bus voltage sensors, and motor temperature sensors. Motor speed will be derived from the position sensor. So, the bottom end is I thought I was using all this information to accurately calculate an angle - not "estimate" it.

That's actually an extremely helpful link - I was going through all the equations in my motor control books trying to come up with the equation he gives at 21:20 - this saves me a LOT of time. The problem is I'm still trying to calculate the currents in the first place. They are functions of the motor resistance, inductance, and frequency. The ORNL tests on the 2010 Prius give resistance values, but they don't seem to provide any inductance numbers. One VERY interesting thing he points out in the video is that the optimum current angle is also related to the magnitude of the CURRENT. Jeeeze! This makes backing out the current from those ORNL output plots even more difficult. Since the current is related to the frequency, it could be controlled by either the PWM pulse width or even the switching frequency. Fortunately, all the ORNL tests are conducted with a switching frequency of 5kHz. Changing that number can dramatically change results. At this point, I'm trying to back out some reasonable inductance numbers, which are heavily reliant on switching frequency.

For P-hak, the description from about 12:30 onward describes the motor. At 19:00 he points out that the optimum torque angle is NOT 90 degrees, agreeing with what I said in my previous post. He specifically says it's not a simple "quadrature relationship" at 19:30 - this means straight SVPWM won't be enough to maximise the output of the motor.

The common assumption is that a PMBLDC motor can be pushed to 2X the base speed. In the case of the Prius motor, that's only 2*3000 RPM, or 6000 RPM. However, since the Prius motor is an IPMSM, using FOC WITH THE RIGHT ALGORITHMS ORNL was able to push this motor to over 13,000 RPM.

So, no, I am NOT treating the motor like a PMBLDC motor; that should have been apparent in all my previous posts. Notice he quotes ORNL in the video, just like I did.

So, no I'm not going to be using FOC with fixed parameters, I will be using equations that take into account the motor's operating conditions, including it's speed.

[e*clipse]

I think it's a good idea.

I'm going to try an analog circuit first; hopefully it will be more reliable than my code. It won't be as smart, but hopefully it can be more reliable.

If I can figure out a 100% (or damn close) reliable program to do this, I will be tempted to give it a shot.

Either way, there will be a big red, well tested "bypass" button on my steering wheel. Maybe a big red "oh ****" button to cut all power would be good thing as well.

- E*clipse

Originally Posted by Astro

How about just having a controller per wheel. Then adding a simple (or complex) throttle pre-conditioner circuit. The pre-conditioner would take input from speed sensors on each wheel as well as the main throttle input.
It could generate 4 throttle signals, one for each of the controllers.
The pre-conditioner could be a simple traction control type set up for regular cars and something much more complex for racing situations.
It may also be able to compensate for controller/motor failures.
Switch between 4 wheel drive and 2 wheel drive.
If there is regeneration available it may also be able to use that to give even more control.
With the addition of some accelerometers and gyroscopes it could be heading into the realm of stability control.
Also for reliability, using a throttle pre-conditioner may mean that if it ever failed it could be bypassed and the throttle directly feed the 4 throttles. You wouldn't have all the nice traction/stability control type features but at least you could drive home.
It would also mean standard controllers for whatever type of motor could be used.

Just a thought.

[cts_casemod]

Originally Posted by e*clipse

So, no I'm not going to be using FOC with fixed parameters, I will be using equations that take into account the motor's operating conditions, including it's speed.

Ill give it a try mailing Dave and asking if he could supply some of that stuff he shows on the video. I would not be surprised if he already had al algorithm
to use with texas instaspin libraries. That would make your work significantly easier.

[P-hack]

Originally Posted by e*clipse

Hmmm, ok - then I guess I mis-communicated.

You also hijacked an induction motor thread.