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

[e*clipse]

Originally Posted by P-hack

Actually I lied, the output boost diode bounces between roughly 300A and zero, 180 "degrees" out of phase with the bottom switch bouncing between 300A and zero, with a typical 30% inductor ripple anyway (i.e. 166uh here @ 20khz, ~%50 duty cycle).

It isn't a cop-out, it is just puzzling, why the form of a capacitor is dictating and severely limiting the boost circuit design. asthetics?!? do you see divinity in symmetry? Fibonacci's three phase controller? the boost switch/leg needs to be at least 3x larger for "hotrodding" purposes, sorry.

I would seriously like to improve the boost stage's capability. Right now, it would come at the expense of reducing the rest, because of the diameter constraint. I suppose I could increase the diameter and add those boost stage switches. At this point I'm not even sure I **want** a boost stage. However, the system advantages say it's worth at least playing with. If it doesn't work out, there's the opportunity to provide more output by symmetrically arrranging more phase leg switches.

The goal with this design is to provide as high a power/volume and power/weight controller as possible, in a "build it yourself" type format.

Therefore, options not considered were things like direct die switches, etc. I've seriously considered the stuff Toyota did for the Prius controller. I think Toyota did an amazing job with both the controller and the motor.

However, it would be impossible, within the design constraint of making this buildable, to use the direct die switches they used. In addition to that, they used a lot of tricks with the capacitance, cooling, etc. It's a very evolved, elegant design.

That said, I feel extremely happy that this design offers very close power/weight and power/volume numbers. One thing you cannot do when attempting this is leave a lot of cards on the table, like providing 3X headroom for "hotrodding" purposes. You will notice that you don't have anywhere near 3X headroom in the Toyota design. It would require a complete re-design to boost the power output that much. If you want that, I suggest a more normal design using IGBT "bricks" that can be added until you're happy.

Regarding symmetry and the capacitor - yes, there is divinity in symmetry. The capacitor is critical for the volume/weight reduction. The capacitor functions best when loaded as physically symmetrically as possible.

SBE, the capacitor's manufacturer, has a ton of info on their site about this. They strongly emphasize symmetry and a low inductance bus-plate system. This inverter design actually takes their recommendations a step further with complete symmetry. The benefits of symmetry and lower inductance provide lots of opportunity for higher performance. However, this is buried in the subtleties of SiC switches and high switching frequencies.

Thanks for the thought provoking questions; I'll see what I can do to improve the power output of the boost phase.

- E*clipse

[P-hack]

Originally Posted by e*clipse

Just wondering: I mean it seems you make quite a few mistakes yourself when reading my posts.

I said nowhere "two pase motor."

I'm afraid the dual full-bridge idea for svpwm-ing a 2 phase motor is my own insanity.
(if you use 4 leads instead of 3, you can get full reversal of each coil)

edit: studying 2 phase also makes the field oriented control a bit more straightforward, no coordinate transform.

[e*clipse]

Originally Posted by P-hack

I'm afraid the dual full-bridge idea for svpwm-ing a 2 phase motor is my own insanity.
(if you use 4 leads instead of 3, you can get full reversal of each coil)

edit: studying 2 phase also makes the field oriented control a bit more straightforward, no coordinate transform.

That's ok. I do really appreciate your suggestions.

Just as a starter, maybe now is a good time to figure out what a good boost-converter/inverter system would require.

For example, what boost ratio (difference between battery voltage and required bus voltage) is practical?

It seems it would be wise to look at stuff like **obtainable** EV batteries, and stuff like that. For example, the Leaf batteries I'm using have a 360V nominal pack voltage and can put out about 600A max. The pack's capacity is 24kWh. These are ORNL test numbers.

The pack's current can be reduced by increasing the voltage, with the difference between pack voltage and required bus voltage "fixed" by the booster.

The main down-side to adding more cells is cost and weight. If the pack's capacity is adequate for the intended use, then adding more cells will merely add weight and cost.

The catch is that the boost stage and battery pack must also supply enough current to meet the power requirements. For example, my setup will use four motors/controllers with a peak power output of about 75kW each. If I use the standard Leaf pack's numbers, the boost stage of each controller must supply 208A. ( 75kW/360V ) Since there are four motors, the poor battery must supply 4*208A = 833A !

So, in this example, the boost stage requirements are do-able, even with the un-optimized controller design. However, the battery cannot supply 833A. ORNL testing indicates it can supply about 600A. So, in this example I would **have to** either increase the battery pack voltage or add a parallel string to stay within the current limit, reduce the power requirements, or consider a different battery option.

In another example, say 1 motor can put out 150kW @ 650V. Using the Leaf pack, the boost stage would be required to provide 150kW/360V = 416A. In this example, the current design boost stage design would not work because three TO-247 packages can't handle the current. The options at this point are to re-design the controller or increase the battery pack voltage. If the boost stage's limit is 300A, then the system would work with a 500V battery pack. (150kW / 300A ) The 300A is well within the battery's capability, so the setup would work. The main issue would be the added weight/cost of the extra cells.

I'm just putting this out to start some conversation about this. Adding a boost stage definitely offers some opportunities. However, engineering is about compromise, and hopefully we can come up with something reasonable and useful.

- E*clipse

[P-hack]

Yes, lots of compromises, you probably have considered it, but it seems like the liquid cooled volt pack can push 900 amps, and you already have some liquid cooling in your future it seems with these motors (and hotrodding).

But instead of paralleling another pack, I would series it, since those motors prefer high voltage. Another 39 leaf modules will get you to ~650v, considering how many motors you are feeding... It was designed for a ~100hp car, and you are looking for ~400hp, so more batteries is entirely reasonable.

Also it occurs to me that the boost converter approach seems to be limited to hybrids, where you have a engine-generator to power the drive motors when needed. I don't know of any straight EV's that use a boost converter. The hybrid battery mostly fires up the engine and adds some power back on accell from regen. The boost inductor is out of the way of the high power current flow. The boost switches have a very different reality than the inverter switches as well, and have to handle 3x the current on average (with high peak loads and voltages).

I'm currently thinking the boost converter is probably best avoided in an EV, relying on field weakening and other control tricks instead (or motor retermination/reconfiguring, but that won't help your battery hp demands, only more batteries will do that).

[e*clipse]

Originally Posted by P-hack

Yes, lots of compromises, you probably have considered it, but it seems like the liquid cooled volt pack can push 900 amps, and you already have some liquid cooling in your future it seems with these motors (and hotrodding).

But instead of paralleling another pack, I would series it, since those motors prefer high voltage. Another 39 leaf modules will get you to ~650v, considering how many motors you are feeding... It was designed for a ~100hp car, and you are looking for ~400hp, so more batteries is entirely reasonable.

Also it occurs to me that the boost converter approach seems to be limited to hybrids, where you have a engine-generator to power the drive motors when needed. I don't know of any straight EV's that use a boost converter. The hybrid battery mostly fires up the engine and adds some power back on accell from regen. The boost inductor is out of the way of the high power current flow. The boost switches have a very different reality than the inverter switches as well, and have to handle 3x the current on average (with high peak loads and voltages).

I'm currently thinking the boost converter is probably best avoided in an EV, relying on field weakening and other control tricks instead (or motor retermination/reconfiguring, but that won't help your battery hp demands, only more batteries will do that).

Totally agree. Yes, the power output of the pack will have to increase. And yes, I'm adding some pack cooling. In my area, the temperature frequently gets above 100F average. The "tarmac" and parking lot temperatures can see 120F. Under - hood temperatures...

The problem I ran into when I figured a 650V pack was the solution is primarily weight. If I remember correctly, I was facing a 800lb pack, even doing the various pack-lightening tricks I could figure out. This may not seem so bad by lead-acid numbers, but I have a project goal that the car will be lighter than the original.

But, that's just me. I'm curious how others are looking at this issue. Would a controller like this be useful, or just a highly optimized thing for one specific application? Would a mild boost - less than 2X - be useful to others? In this case I'm also trying to design something that other people might want. This is why I'm taking to heart your suggestions about either improving or ditching the boost stage.

- E*clipse

[P-hack]

well I know you have an idea for a solution in search of a problem, but lets focus on what you have to work with then.

the mgr is listed as 50kw, but you mentioned 75kw, assuming you get there with more current and not just revs, then at 360v you should be able to approach 41kw, which is only 114 amps per motor, or 456 amps, a comfortable margin below 600 amps.

Your units also have gearing fixed, which is a problem from a top speed perspective probably, but with all 4 units you should be able to exceed 3000 lbs of force, might even approach 1g off the line if the car is light enough (does your car weigh less than 3000 lbs?), but I don't know how you figured 75kw.

If you go with stock 50kw mgr figures the breakdown speed can be interpolated from the lower voltage (360v leaf pack assumed, might have to charge it fully to account for voltage drop)

50kw * 360 / 650=27kw. 610 wheel rpm * 360 / 650= 337 rpm where the torque falls off. I dont know your wheel size, but if they are 2 feet rolling diameter, that is 24mph. And 75 amp per motor, 300 from the battery.

Even with 650 volts, it will peak before 44mph.

But taking the simple case, 360v, does field weakening in a pmac take extra current? I.e. you have to "cancel out" the existing rotor field somehow to increase the rpm? Or are you able to somehow shift to a more reluctance based torque angle when the bemf rises?

Also, 650/360 is 1.8. The voltage change from wye/delta is 1.732, you should *really* make sure there isn't a trivial way to change it over to delta, because it would be nearly a perfect fit for the leaf battery then. you would be at 50kw * 4 @520 battery amps max (@375v).

[P-hack]

or at least determine if it is wye or delta...

Determine if the motor is WYE or DELTA - Electric motors, generators & controls engineering - Eng-Tips

Disassemble the motor and apply low voltage DC to a pair of leads. Check the magnetic field strength around the inside of the stator with something like a hacksaw blade. You may wish to vary the applied DC voltage until best results are obtained.
Delta: One winding group will be magnetized strongly and the other two groups will be at half strength.
Wye: Two winding groups will be strongly magnetized and one group will be either not magnetized or weakly magnetized.

[e*clipse]

Originally Posted by P-hack

well I know you have an idea for a solution in search of a problem, but lets focus on what you have to work with then.

How about if we try this - a little mutual respect. It will help considerably in keeping the conversation both productive and positive.

As a start, I've done a ton of research on this motor, including reading just about every paper published on it. I also have one in pieces on my workbench, so I have an excellent idea of it's physical details. I also made a spreadsheet that includes every detail of a car/battery/motor system that includes voltage sag, aerodynamic drag, tire drag, etc etc. This spreadsheet has real numbers based on real data, not guesses. This spreadsheet gets extremely close to real-world EV test results.

the mgr is listed as 50kw, but you mentioned 75kw, assuming you get there with more current and not just revs, then at 360v you should be able to approach 41kw, which is only 114 amps per motor, or 456 amps, a comfortable margin below 600 amps.

That's fine - you're well on the way to proving why the last person who tried the MGR was dissappointed in it's performance. He used one MGR, a 360V pack and a motor controller designed for a BLDC motor. He gave up on it when the car couldn't exceed 45mph.

Your units also have gearing fixed, which is a problem from a top speed perspective probably, but with all 4 units you should be able to exceed 3000 lbs of force, might even approach 1g off the line if the car is light enough (does your car weigh less than 3000 lbs?), but I don't know how you figured 75kw.

Yes, the car will weigh less than 3000lbs. No, single speed gearboxes aren't the main problem. Ask Tesla.
I based the 75kW on the fact that current is limited by the cooling requirements. The MGR relies entirely on whatever heat can be removed from the case. Providing better cooling would allow more power output. A real cooling system may allow a 25% increase in torque. I'm merely building in some controller/battery headroom to allow this improvement.

If you go with stock 50kw mgr figures the breakdown speed can be interpolated from the lower voltage (360v leaf pack assumed, might have to charge it fully to account for voltage drop)

50kw * 360 / 650=27kw. 610 wheel rpm * 360 / 650= 337 rpm where the torque falls off. I dont know your wheel size, but if they are 2 feet rolling diameter, that is 24mph. And 75 amp per motor, 300 from the battery.

Again, you're just proving again why it's not a good idea to run with 360V. Also, ORNL tested the Prius motor (a very similar design) at 650VDC bus, 500VDC bus, and 225VDC bus. In short, the results show that you can't simply linearly interpolate that point where the torque falls off. At 650V, that point is 3800 rpm. At 500V, it's 3200 rpm. At 225V, it'sabout 1500 rpm.

In the start of this you stated that "well I know you have an idea for a solution in search of a problem."
Well, you're demonstrating what the problem is.

Even with 650 volts, it will peak before 44mph.

But taking the simple case, 360v, does field weakening in a pmac take extra current? I.e. you have to "cancel out" the existing rotor field somehow to increase the rpm? Or are you able to somehow shift to a more reluctance based torque angle when the bemf rises?

Yes, you can "cancel out" the BEMF and use the reluctance torque using FOC. Quoting ORNL:

Although the maximum PM torque is produced at a current angle of zero electrical degrees and the maximum reluctance torque is produced at 45 electrical degrees, the maximum total torque is produced at about 35 electrical degrees. This optimal current angle varies with many conditions such as current magnitude and speed.

So yes, I'm relying on this fact and the ability of this FOC controller to make it possible.

Also, 650/360 is 1.8. The voltage change from wye/delta is 1.732, you should *really* make sure there isn't a trivial way to change it over to delta, because it would be nearly a perfect fit for the leaf battery then. you would be at 50kw * 4 @520 battery amps max (@375v).

The motor is wound wye. The common connection point for the wye is hard to get to, even for measurement purposes. I think Toyota knew what they were doing when they made these (understatement of the century ), so I'm attempting to match their control methods to obtain optimum performance. I think it would be a mistake to completely change the motor in an attempt to stick with a bus voltage that is far below the one the motor was designed for.

The numbers I posted were just examples of reasonable scenarios. I really don't want to hijack this thread talking about the MGR. Paul had expressed interest in building a motor controller with a built-in boost stage. I'm trying to help out with that and as a starting point figure out how much boost would be reasonable. I've found a combo that works for me; I'd like to know if this would be useful for others. If no one has any interest in this, I'll just build what works for me; it's much easier.

- E*clipse

[P-hack]

I appreciate the wye confirmation, but otherwise it isn't making a lot of sense to me.

The 300kw model S weighs about 5000 lbs, 3000 lbs is arbitrary, right?

millions upon millions of of motors have been switched between wye/delta, because it is a lot simpler than a rewind, you might want to check with a motor shop if you are not going to put in more batteries to get to 650v. It is an incredibly useful thing for others.

But in the mystery that is toyota, the biggest question is the 4x75kw assertion. that exceeds your pack hp capabilities, and it is a question of the coercive strength of the permanent magnets, push them too hard and they get destroyed, regardless of how much cooling you have in the stator. Do you think toyota didn't consider that in their 50wk rating? Or when they limited battery boost power to ~30kw? Nobody is using 300kw of boost converters in an EV, and the battery cooling plan sounds sketchy at best.

Also I never said single speed gearboxes are the main problem, but choosing the best ratio is certainly is a major consideration. I assume you are stuck with the mgr ratio, which is fine, and all the more reason to give delta a serious try, since you might not be able to throw more amps at the motor without destroying the magnets, but you might be able to push the breakdown rpm a lot higher than stock if you do wind up with more batteries AND delta.

[thingstodo]

Update for Nov 22

DC Motor rotation re-test after equipment moved around and re-assembled in my garage

https://youtu.be/K9JOKahTosM

The video format is a bit different. It uses Google Hangouts live to window 3 camera views.

I'm not sure I like it - what do you think?