Getting the right windings...

[weber]

- weber I would debate you on multipolar machines but I have nothing to bring to the party.
Although I did peruse Marathon's Premium Efficiency range.

At random I selected 30Hp size regarding torque /mass correlation.

2-pole 499lbs   286 frame Probably some slots are left empty
4-pole 519lbs   286 frame
6-pole 689lbs   326 frame So same power, but larger frame.

I say torque/mass ratio increases significantly (but not quite linearly) with pole count. You say it shouldn't increase at all. So why haven't you given the respective torques or torque/mass ratios for these motors, so we can see which claim is supported?

[jonescg]

T2, I will actually be running a 640 V nominal DC bus, so I shouldn't have any trouble getting past 4000 rpm before employing field weakening. The upper limit on all electric motorcycles in this series is 700 V DC max, so I aimed for a top-of-charge voltage of ~705 V, which should settle to about 640 V DC under load. By the time the pack is almost spent, it will be closer to 600 V DC. The PM150DZ is happy at voltages up to 720 V DC, so I could possibly go even higher if the rules permitted.

It will be a direct chain drive from the driveshaft to the rear wheel, and I am estimating a 14:41 reduction, so at 3800 rpm I should be going about 155 km/h. This is about as fast as most turns on Australian racetracks get, with the exception of turns 1 and 3 at Phillip Island - they are stupidly scary fast! I should be able to hit 200 km/h down the straights at 5000 rpm while the BEFM is being suppressed, but power is held constant. Hell, even my old GS500 would do 200 km/h, it just took a month to do it

I will be running a cooling system for the motor, and a separate cooling system for the inverter. This is because the motor can get to 90'C without ill effect but the inverter would start to give up at these temperatures. In fact, if they were in the same cooling loop I would effectively be heating the inverter up!

My pre-race procedure might include purging the cooling loops with ice water, only to bring it up to room temperature on the sighting lap

[T2]

-jonescg Yes, good idea to separately cool the inverter.

You haven't revealed your battery pack details. I estimate 200 LiPo cells and climbing. Do you have a BMS ?

While racing how do you propose to keep an eye on that many cells ?

I am wasting my time here. The supplied graphs are all very well but either the controller is limited to continuous current or it is not. You can't flip back and forth between peak torque and continuous if you've been pulling continuous for any length of time. So I'm wondering what does that 60 sec peak torque value mean to the controller ?

I will say that choosing the AFM-140-3 would have ensured that you need never enter the field weakening zone. Although it appears that the controller will have to supply 20% more current to this motor in actuality the battery current will be the same for both motors.

That is until 4400 rpm is reached at which point the controller current will take a slightly deeper decline, to prevent overheating of the -4 motor, as it enters the field weakening zone. I didn't think that field weakening would exact a penalty so thanks to Weber/Coulomb for pointing that out by the way.

The -3 motor on the other hand will continue to 5000 rpm without taking that steeper decline as shown by the continuous torque line for the AFM-140-4.

Yes that continuous torque line is an indicator of the max motor current using the Kt value of 1.81N-m/Amp for the -4 and the less favourable value of 1.36N-m/Amp for the -3.

Best of luck.

[jonescg]

Yeah, the demands I'm placing on this motor are huge, that's why I think I can ignore peak anything and stick with constant ratings. I would anticipate needing an average power of ~60 kW. Even if it calls for peaks of 120 kW on occasion, I will know that the motor will be able to handle it.

On the battery front, I'm going with 5 Ah cells, arranged 168s,3p. 504 LiPo cells all told. I will have a battery monitoring system which alerts me to a low cell, however given the sheer number of cells, it would be easier if I monitored groups of 3 or 7 cells at once. Balance ports will be set up at the top of the pack for each of the 8 x 21s sub-packs. The pack will be a completely self-contained unit which won't put out power unless 12 V is fed into it to close the main contactor. Safety first!

[T2]

I think I can ignore peak anything and stick with constant ratings

Even if it calls for peaks of 120 kW on occasion, I will know that the motor will be able to handle it.


So at 4000 rpm graph shows 180N-m for the AFM-140-4 with a Kt of 1.81N-m/Amp which infers a current limit of 100Amps, a 7C rate for your effectively 15Ah pack.

If we assume 640Vdc bus voltage you can expect 464Vac at the motor terminals max drawing 100 Amps. So multiplying by root3 gives 80.4Kw max, not the 120Kw that you mention unless the drive has some inverse time delay overload setting that I am unaware of.

These are some awkward questions which you may be unable to answer before perusing the controller user manual. I would certainly query your supplier. It might be a good idea to be in contact with Geerant and get his views also.

[T2]

- weberI say torque/mass ratio increases significantly (but not quite linearly) with pole count. You say it shouldn't increase at all. So why haven't you given the respective torques or torque/mass ratios for these motors, so we can see which claim is supported?

4-pole 519lbs   286 frame
6-pole 689lbs   326 frame So same power, but larger frame

Why haven't I ? Well,Err.. because you may well be right. LOL

The example I put up proves your point. Sure torque costs mass,
but if you assume that 6-pole as the norm and of course I would expect you to assume, rightly, that torque is inverse to rpm for constant horsepower. That in mind, the 4-pole should weigh in at 459Lbs, not 519lbs !!

Look, I'm not beaten yet. If the facts refuse to fit the theory. Then clearly the facts are wrong. I am beginning to think that for a whole bunch of business and manufacturing reasons Marathon's motor designs are not following reasonable orthogonal projections. Their catalogue is, if you will, an inconsistent data set.

Let's start again. Assume that for a particular rated torque specification we can determine a stator design within which the pole phase layout is optimised for slot dimensions etc

What would we reasonably expect the result to be if we re-cut those laminations for the same slot ratios but scaled down to accept extra slots for an additional pole pair per phase ?

I believe that we would have a slower machine developing the same torque as before. We haven't increased the magnetic material, we have simply divided the available torque producing magnetic field over a greater number of poles. A zero sum game in other words.

So let me try you a new case - where we want to have increased torque to compensate for the slower speed of rotation. For that case then we have to increase the perimeter of the stator in order to preserve the existing poles at their full size while at the same time accomodating the additional new poles. I think we could reasonably expect the new poles to supply the needed additional torque.

In conclusion it doesn't seem likely that any pole layout is an improvement unless a FEA can prove otherwise.

Let us now take the case where there was a more optimum pole number than 2-pole.
Well, a turbogenerator couldn't use it. It must be direct coupled to the steam turbine to avoid gearing down. For the synchronous speeds of 3000/3600rpm there is no other choice than a 2-pole design.
Consider also, their 20 ft long stators with 4ft diameters already stretching mechanical limits. Is anyone expecting us to believe that those long narrow fields are optimal ? I think not, but the optimal graph for fields of that magnitude are unlikely to contain razor sharp peaks either, in regard to length /diameter ratios, or perhaps 650MW machines would not be possible.   

I am going to open a new topic on this forum re motor design, for which jonescg will no doubt be grateful. Even if he never goes there !!    

[jonescg]

The PM150DZ is rated for a continuous 225 A rms and its peak amps are 300 A rms. By my reckoning, 640 V DC = 426 V rms, so continuous 96 kW, and about 120 kW peak. That's just the inverter. The motor will obviously put less to the rear wheel, but I reckon these numbers are enough to move me along at a fair clip without catching on fire.

Rinehart offer some formulae in their data sheet:

Motor power = sqr(3) * motor voltage * motor current * power factor * efficiency.

Assume 0.768 PF and 0.912 for the motor efficiency, and under my conditions I get:

P(motor) = 1.7 * 426 Vrms * 225 Arms * 0.768 * 0.912 = 114 kW constant.

So until everything is set up in the bike I won't really know what's possible, but the calculations are pretty close.

If we take the inverter's efficiency as 97.5% and the motor as 91.2%,

P(battery) = 114 kW / (0.975 * 0.912) = 130 kW. At 640 VDC, that's about 200 A, possibly more with sag.

200 A from a 15 Ah cell is 13C so yeah, I will be hammering them. But that's why I wanted to know that the constant ratings were my peaks, and anything above that is a bonus. The Turnigy cells are rated for 40C continuous, but I suspect it's more like 20C in reality.

If you like I can send you the pdf of the RMS datasheet.

[weber]

The example I put up proves your point.

There's hope for you yet.

Sure torque costs mass, ...

What the cattledogs and I are trying to get through to you is that torque doesn't need to cost mass. More torque can be had for the same mass by increasing the pole count. I think Woody or Johny tried to tell you that some weeks ago but you apparently missed it, or ignored it.

Look, I'm not beaten yet. If the facts refuse to fit the theory. Then clearly the facts are wrong.

Hee hee. I'm afraid that's religion, not engineering, unless you're Einstein. And since you're not Einstein, it would seem you're in the wrong forum.

I am beginning to think that for a whole bunch of business and manufacturing reasons Marathon's motor designs are not following reasonable orthogonal projections. Their catalogue is, if you will, an inconsistent data set.

So why haven't you looked at some more manufacturers' catalogs instead of going on about what you expect to find according to your theory?

There are plenty of induction motor catalogs on the web. I'll let you find them for yourself so you can't accuse me of hand-picking those that support my claim. In the size range of interest to EVers you will find that, although going from 4-pole to 6-pole gives only a small-to-non-existent increase in specific torque (13% in your example), going from 2-pole to 4-pole typically gives an 80% to 90% increase in specific torque.

Surely you've already noticed that for a given power rating and efficiency class, the 2-pole and the 4-pole are often in the same frame size and differ little in mass. And, as you acknowledge, the 4-pole has twice the torque.

I believe that we would have a slower machine developing the same torque as before. We haven't increased the magnetic material, we have simply divided the available torque producing magnetic field over a greater number of poles. A zero sum game in other words.

Since you seem to prefer theory to facts, here's some theoretical hand-waving as to why increasing pole count (without increasing mass) is not a "zero sum game" as far as torque is concerned.

I take it you accept F = BIL where F is force in newtons, B is flux density in webers per square metre (tesla), I is current in amps (at right angles to the flux), and L is the length in metres of the current carrying conductor (in this case a rotor bar near a stator pole). The saturating value of B is a constant Bsat for the magnetic material. I is limited by rotor heating. L is constant for a constant frame size. So none of these things depend on the number of poles. But the force F is developed at every pole and so these all add to produce the torque.



In a somewhat tortured analogy, imagine a horse-driven capstan with two regular horses 180 degrees apart. Then we replace those two horses with four shire-ponies 90 degrees apart (conserving total mass of horse-flesh) that have the same muscle cross-section per leg and so can apply the same force, but their legs are only half as long so they only do half as many rpm for the same number of steps per minute, and they only consume hay at half the rate. So for the same mass of horse-flesh you get the same total horse-power and the same total rate of hay consumption, but you've doubled the torque.

Now if you genetically-engineer those 4 shire ponies to double their metabolic rate (maybe some hummingbird genes) then the capstan can turn at the same speed it did with the two horses, but with twice the torque and hence twice the horse-power and twice the rate of hay consumption.

Back to motors. Unfortunately, the closer the poles are together the less of that flux density (Bsat) actually makes it across the air-gap to the rotor, which is one reason torque doesn't go up linearly with pole count and why power factor goes down with increasing pole count.

According to this article from Baldor Electric Company, 2007
http://www.reliance.com/mtr/b7100_1.htm,
this leads to 4-pole motors having the optimum tradeoff for the range of torques used by EVs. Only if you get up past a massive 1300 Nm (nominal, not peak) does 6-pole become the optimum. But that's for conventional radial-flux induction motors. I'm wondering if the leakage flux can be much reduced in an axial flux induction motor (AFIM) so that the benefit of high frequency (say 400 Hz) operation can be had without the need for extreme rpm and reduction gears, by going to maybe 8-pole.

I note that super-efficient high-specific-torque ironless Halbach-array permanent magnet motors tend to have 12 or more poles.

The quantity that would be a "zero sum game", i.e. constant irrespective of pole-count, for a constant-mass constant frame-size motor (if it wasn't for leakage-flux and other flux-path geometry considerations mentioned in the above article) is not the torque but the power at a given electrical frequency. This should come as no surprise because it is exactly what happens with a transformer.

I am going to open a new topic on this forum re motor design, for which jonescg will no doubt be grateful. Even if he never goes there !!    

If this 14000 rpm Machines topic is intended to be it, then I note that its title and opening remarks seem to presuppose your theory that I am arguing against.

This is a topic for those who have an interest in high speed electrical machines and the gearing systems designed to accept them.

I'm more interested in high frequency electrical machines which don't need to be high speed and so do not need special gearing systems.

[woody]

The example I put up proves your point.

There's hope for you yet.

Sure torque costs mass, ...

What the cattledogs and I are trying to get through to you is that torque doesn't need to cost mass. More torque can be had for the same mass by increasing the pole count. I think Woody or Johny tried to tell you that some weeks ago but you apparently missed it, or ignored it.

A not very quick automated scan through the 161 Aluminium motors in ABB "Catalogue_Industrial Performance Motors EN 04_2008 RevB.pdf" reveals the top 4 torque monsters per kg are all 280kg+ 4 pole 250 and 280 frame machines with 4.01 to 3.61 Nm/kg of DOL breakdown torque: ABB 3GAA 252 033, 222 034, 252 032, 282 032

5th is ABB 3GAA 223 032 - a 281kg 6 pole 225 frame

6th is a 177kg 4 pole 180 frame with 3.51 Nm/kg (194Nm x 3.2 / 177kg) ABB 3GAA 182 032

7th with 3.46 Nm/kg is a 4 pole 132 frame 132 007 which I have in my garage.

A bit further down (25th) at 2.9 Nm/kg is the best 2 pole 132 frame 131 008 which weber has.

Even further down (28th + 29th) the best 8 poles are at 2.85 Nm/kg, 340kg + 403kg 250 and 280 frame 3GAA 284 031 and 254 031.

So it looks like 4 poles have it wrapped up, especially if you exclude motors over 100kg.

That said, the ACIM industry and catalogues are all focussed on DOL 24x7 operation, whereas EVs are Inverter driven for short times (2 hours/day).

The nominal figures are all rounded to the nearest 5hp and based on DOL 24/7 loads, i.e. not super relevant to EV use.

[T2]

-woody I don't think there is a consistent data set out there.

For starters I am sure that manufacturers haven't made every single motor size shown in their catalogues. Since the same frame size can house a range of stators with different stack lengths I suspect the figures in the catalogues are for guidance only. It is likely that they would get updated whenever a specific machine is manufactured and it is also possible they tweak stack lengths between manufactures. Mass is rarely a problem to OEMs. In our company the motor was moved from the skid to the mounting plate, we were never concerned with weight. But only when the motor didn't fit the plate - which happened several times with SEW, if I remember.

I am told that the days when motors of greater than 10Hp were sitting around on shelves waiting for your order of QTY 3 please, are long gone. Too many variants in each size killed that idea. Here in NA they are now all custom ordered. And you wait.

Which brings me to another subject. The possible disappearance of all pole counts other than 2 or 4. Over the last 25 years many AC motors have been connected to softstarts possibly for automation and greentech reasons. Today it is more likely that plant managers will not order up rewinds for these unusual machines and that their replacements will be with the more standard 4-pole machines run from inverters. Just a thought.

[bga]

Where's ACMotor when you need him?

I remember that Tuarn did a survey of power density and torque for various induction motors a few years back and came to the conclusion that 4-pole offered the best torque density. I recall that another factor in his observation was that 4-pole offers a wider useful speed range on the motor, making it a better choice for single gear ratio systems.

It may be better to look at 6,8,12... pole machines to be either special purpose or throwbacks to that pre-inverter DOL world in the same way that dahlander designs and start-delta switching are now relics, only found in museums and many manufacturing plants.

My observation of the same motors is that power density is slightly better with 2-pole, but this pushes the optimum speeds too high for automotive direct to diff application.

The above makes the assumption that we are talking about ABB etc induction motors, different arrangements like axial flux motors with different (larger) effective rotor diameters are unlikely optimise to the same point. I suspect that the reason that 4-pole induction machines are favored is because this represents the best compromise between coupling (extra N-S transitions), magnetic crowding and losses in the rotor conductors.

This would suggest that in multi-ratio geared applications, 2-pole may be a better choice, but the weight and complexity of a gearbox may negate this.

I note that that gear single ratio is only effective in light vehicles, weighing less than about 2000kg. Heavy vehicles such as trucks have so wide a dynamic range that the motor and drive electronics would be grossly over sized if a single ratio system was attempted. The 18-speed roadranger gearbox found on 500hp trucks bears witness to this.

It should be possible to configure the EVO double stack motors to be series/parallel switched to allow the dynamic range of the battery/inverter to be extended in the same way that it is done on paired DC motors. (with 3 x SPCO contactor or 6 x SPNO [EV200] type)

Another observation is that it is easier to build (95+%) efficient small motors in PM designs than induction. Typically, induction motors only achieve 96% in the 400kW+ sizes.

[weber]

Hi bga,

I agree with your excellent observations. I'd just like to add a couple of additional observations.

The slightly higher power density of 2-pole over the 4-pole and hence the 2-pole's greater suitability with a multi-ratio gearbox only applies when the 2 and 4 pole have the same volts-per-hertz winding ratio, e.g. when they are both wound for 230 V 50 Hz.

It was only because we didn't know we could order lower voltage windings from ABB that we went with the 2-pole for the MX-5. The 2-pole was rated at 22 kW while the same size 4-pole was 18.5 kW.

But now that Woody has shown it is possible to obtain a 115 V 50 Hz winding from ABB we would definitely use the 4-pole in future (or the similar volt-per-hertz 4-pole SEW motor from Tritium) because the 4-pole can then be run at twice the frequency (on the same voltage), bringing it up to the same rpm as the 2-pole and producing nearly twice the power (and then we'd need different batteries to be able to supply that power).

So T2, we see it is the increased frequency, not the increased rpm, that gives us more power from the same size motor. Someone just needs to figure out how this trend can be continued to 6-pole and higher pole-count induction motors.

Re series/parallel switching of double-stacked EVO and other PMAC motors: The only reason it's not possible is because they only provide a 3 wire conection to the 3-phase stator windings. You need access to both ends of each phase, a 6 wire connection. If they provided a 6 wire connection, you wouldn't need double-stacked motors to do this. You could get a similar result by star/delta switching of a single motor.

So there's a thought, jonescg. If EVO could provide all 6 terminals, then it could still be the number-4 winding in star, with field weakening required to get to 5000 rpm, and you'd have the additional option of trying it in delta with no field weakening to 5000 rpm, similar to the number-3 winding. But you'd still need a higher-current VFD to take advantage of the delta configuration.

[Edit: Grammar and clarification]

[bga]

Hi Back,

The slightly higher power density of 2-pole over the 4-pole and hence the 2-pole's greater suitability with a multi-ratio gearbox only applies when the 2 and 4 pole have the same volts-per-hertz winding ratio, e.g. when they are both wound for 230 V 50 Hz.

It was only because we didn't know we could order lower voltage windings from ABB that we went with the 2-pole for the MX-5. The 2-pole was rated at 22 kW while the same size 4-pole was 18.5 kW.

Agreed
It's not easy to get the USA windings in Australia!, but there's always the rewind shop -- or DIY for the adventurous. (after the BMS, this should be a doddle )

In this post I will assume that the motors are wound with windings such that neither the voltage or current are unreasonable. (ignoring this issue)

So T2, we see it is the increased frequency, not the increased rpm, that gives us more power from the same size motor. Someone just needs to figure out how this trend can be continued to 6-pole and higher pole-count induction motors.

This is what ACMotor did when he arrived at a 4-pole option. Perhaps it's time to revisit 6-poles.

Carrying on from my (and your?) motor, I have prepared some data assuming the ABB Frame size M2AA 180 MLA from the "general performance" catalog:
2-pole - 22.0kw @ 2928 RPM, Tn = 072Nm Tmax/Tn=2.8 =202Nm (Pmax=61kw)
4-pole - 18.5kw @ 1465 RPM, Tn = 121Nm Tmax/Tn=3.5 =423Nm (Pmax=65kw)
6-Pole - 15.0kw @ 0968 RPM, Tn = 148Nm Tmax/Tn=3.8 =562Nm (Pmax=57kw)

If we now run these motors at 50, 100, 150 hz respectively to normalise them to 3000 RPM, we get:
2-pole - 22.0kw @ 2928 RPM, Tn = 072Nm Tmax/Tn=2.8 =202Nm (Pmax=61kw) Tmax/Tn is lower
4-pole - 18.5kw @ 2965 RPM, Tn = 121Nm Tmax/Tn=3.5 =423Nm (Pmax=130kw)
6-Pole - 15.0kw @ 2968 RPM, Tn = 148Nm Tmax/Tn=3.8 =562Nm (Pmax=171kw)

I have not made any allowance for rotor slip at TMax or magnetic losses at frequencies other than 50Hz. (The slip is likely to be fairly linear with torque up to near Tmax)

Slip for the motors is quoted as follows:
2-pole 72 rpm at 3000 RPM and Tn (2.4%)
4-pole 35 rpm at 1500 RPM and Tn (2.3%)
6-pole 32 rpm at 1000 RPM and Tn (3.2%)

Slip is a phenomenon of the rotor's interaction with the magnetic field, so is related only to magnetisation and torque, not the field frequency. This makes the 4-pole configuration a lot better than 2-pole at normalised (3000RPM) speed:
2-pole 72 rpm at 3000 RPM and Tn (2.4%)
4-pole 35 rpm at 3000 RPM and Tn (1.2%) - big improvement
6-pole 32 rpm at 3000 RPM and Tn (1.1%) - small additional improvement -

It would appear from the above that there are limits to the performance of the rotor as more induced poles are crowded onto it, resulting in diminished returns for 6-pole.

I see a number of other issues associated with higher pole count implementations:

a) These are industrial motors with laminations that have been optimised for 50Hz operation, so severe a departure from this is likely to result in a a lot of (hysteresis) loss in the field iron, or heating problems.

I am assuming that 150Hz is the limit on the stator, so running the motors at 150Hz produces a result along the lines of:
2-pole - 8928 RPM, Tn=072Nm Pn=66kw, Tmax=202Nm Pmax=183kw
4-pole - 5965 RPM, Tn=121Nm Pn=55kw, Tmax=423Nm Pmax=194kw
6-Pole - 2968 RPM, Tn=148Nm Pn=45kw, Tmax=562Nm Pmax=171kw

b) With the high field frequency (300Hz for the 6-pole), the Windings are going to have to be a lot heavier gauge (or multiple parallel ~ 'multi-filar') to accommodate the V/Hz of the 6-pole. This affects the inductance of the motor, which will be much lower, which affects the switching frequency needed in the controller.

c) It is likely that external inductors will be needed to help regulate the switching current in the motor at the higher switching frequencies.

d) Big power Semiconductors, like IGBTs, have relatively long turn on-off times, so are frequency limited, complicating the design.

My conclusion:
is that 6-pole looks to be feasible as a configuration, but at only lower shaft speeds.

For direct to diff drives, 4-pole IMs look to be the optimum.

Caveat:
The above is based on a popular industrial induction motor and can't be simply extended to other types of motor that have different configurations and/or properties.

[Stiive]

Again, notoriously not reading other peoples post, but just skimming over to get the gist.

My 2 cents based purely on theory (for an Induction Motor) would be:

If all other parameter of the motor remain unchanged (i.e. stator/rotor resistance and inductance, mutual inductance, max flux, current/voltage capability), the motors should have very similar power just at different RPM (only affected by rotational losses of the faster motor)

For instance a 4-pole motor might produce ~115Nm up till 5600RPM to produce ~70kW, whereas the same 2-pole motor would produce 57Nm up until 11,200RPM to produce slightly less than 70kW, due to the extra frictional losses of spinning faster. Have verified this with MATLAB.

Now being able to obtain the same stator resistance and inductance while maintaining the same flux saturation characteristics between the 2 motors seems unlikely; therefore your not really comparing apples to apples.


if you really want, for a small fee, i'll jump on RMxprt and solve this issue for good Haha. That'd be the easiest way of solving this, <10min job.

[Richo]

Or you can forget about the peak and just look at the continuous

2-pole - 22.0kw @ 2928 RPM -> 22kW@3000RPM
4-pole - 18.5kw @ 1465 RPM -> 38kW@3000RPM
6-Pole - 15.0kw @ 0968 RPM -> 45kW@3000RPM

[woody]

I don't think it's valid to double the nominal continuous output when you double the speed, that assumes that doubling the speed doubles the cooling, which isn't right.

I would be interested to know the difference between say my 132-007 and weber's 131-008 - is the frame identical? Are the windings identical? (i.e. does mine just have half the windings reversed to give consequent poles).

[weber]

Thanks BGA. When I wrote this:

Someone just needs to figure out how this trend can be continued to 6-pole and higher pole-count induction motors.

I had assumed that the trend of increasing power at the same rpm with increasing pole-count, did _not_ continue to 6-pole or higher in the case of industrial induction motors (except in sizes that are way too big for cars).

What I meant was that someone needs to figure out how to build a different _kind_ of induction motor where the trend _does_ continue.

But your results suggest that in fact it does continue to 6-pole with industrial motors, although they are a little on the large size, for cars.

Agreed. It's not easy to get the USA windings in Australia!

ABB don't do 115/200 V (delta/star) windings for the USA either (or any other country). It has to be ordered with a voltage code of "X" for "other voltages" and then you specify the voltage you want, and then you wait, and wait ... If it was a 6-pole we'd want 76/133 V (50 Hz).

Carrying on from my (and your?) motor, I have prepared some data assuming the ABB Frame size M2AA 180 MLA from the "general performance" catalog:
2-pole - 22.0kw @ 2928 RPM, Tn = 072Nm Tmax/Tn=2.8 =202Nm (Pmax=61kw)
...

No. Ours is a 22 kW 2-pole in a _132_ frame (95 kg) and has Tmax/Tn=3.8 (model number 3GAA 131 008E).

a) These are industrial motors with laminations that have been optimised for 50Hz operation, so severe a departure from this is likely to result in a a lot of (hysteresis) loss in the field iron, or heating problems.

Hysteresis loss power is (initially) proportional to frequency, but so is output power (nearly), so hysteresis loss will remain a nearly constant percentage. Eddy-current loss power is (initially) proportional to the square of frequency, but fortunately it starts off being a very low percentage loss at 50 Hz. Nevertheless, at some frequency eddy-current loss will begin to dominate. (At higher frequencies both hysteresis and eddy-current losses increase as the 3/2 power of frequency). Hopefully, most motors these days are designed with VFDs in mind, and so use thinner laminations and laminations with higher permeability/conductivity ratios.

I am assuming that 150Hz is the limit on the stator, so running the motors at 150Hz produces a result along the lines of:
2-pole - 8928 RPM, Tn=072Nm Pn=66kw, Tmax=202Nm Pmax=183kw
4-pole - 5965 RPM, Tn=121Nm Pn=55kw, Tmax=423Nm Pmax=194kw
6-Pole - 2968 RPM, Tn=148Nm Pn=45kw, Tmax=562Nm Pmax=171kw

If you had used ABB's high output series of 132-frame 90+ kg motors that includes our 2-pole, you would have found that the 2-pole has slightly higher Pmax than the 4-pole at the same frequency. Unfortunately they don't list a 6-pole in that series.

The main point is that, to a first approximation, they have approximately the _same_ peak power at the same frequency. BTW, we should also look at mass when making these comparisons.

But these motors are only warranted to 4500 rpm, and although I'm prepared to take them to 6000 (with proper balancing), you can't mechanically make use of that power from the 2-pole. So your earlier figures comparing them all at the same rpm were the interesting ones. Its interesting that you might get a 30% increase in peak power by going to the 6-pole, but at what cost in heat I wonder.

It would be good to show efficiency and power factor figures for them, even if they are only the nominal figures.

b) With the high field frequency (300Hz for the 6-pole), the Windings are going to have to be a lot heavier gauge (or multiple parallel ~ 'multi-filar') to accommodate the V/Hz of the 6-pole.

Definitely multi-filar to reduce skin effect and eddy current loss in the copper, as well as making it easier to wind. Our 2-pole is already wound 8-in-hand.

This affects the inductance of the motor, which will be much lower, which affects the switching frequency needed in the controller.

Yes. But I think modern VFDs are up to the challenge. Tritium_James?

c) It is likely that external inductors will be needed to help regulate the switching current in the motor at the higher switching frequencies.

If the PWM frequency goes up proportionally there should be no need for external inductors.

d) Big power Semiconductors, like IGBTs, have relatively long turn on-off times, so are frequency limited, complicating the design.

Yes. But the common 8 kHz PWM will still give 20 pulses per cycle of a 400 Hz sine wave.

My conclusion:
is that 6-pole looks to be feasible as a configuration, but at only lower shaft speeds.

For direct to diff drives, 4-pole IMs look to be the optimum.

Caveat:
The above is based on a popular industrial induction motor and can't be simply extended to other types of motor that have different configurations and/or properties.

I agree, but only because of the likely much greater losses in the industrial 6-pole at 300 Hz versus the industrial 4-pole at 200 Hz.

And I note that 4-pole is also the optimum for when keeping the multi-ratio gearbox.

[Edit: Spelling and grammar]

[weber]

I don't think it's valid to double the nominal continuous output when you double the speed, that assumes that doubling the speed doubles the cooling, which isn't right.

Agreed. Some years back I came up with a 0.7 power rule-of-thumb for that, based on some paper I read. I'd forgotten about it until T2 quoted it recently. i.e. if the frequency and voltage go up by a factor of n then the continuous power goes up by approximately n^0.7.

I would be interested to know the difference between say my 132-007 and weber's 131-008 - is the frame identical?

No. Yours is 132 SMD. Mine is 132 SME. Whatever that means.

Are the windings identical? (i.e. does mine just have half the windings reversed to give consequent poles).

No. A 4-pole consequent winding has gaps between the pole groups for the consequent poles to form in. A 2-pole winding has no such gaps. The average span of the coils on the 4-pole consequent is nearly half that of the 2-pole. And of course we don't know if yours uses consequent poles at all. Fully symmetric windings should produce fewer air-gap harmonics (i.e. give a more nearly sinusoidal flux distribution around the air-gap) and so fewer losses.

What's the nearest thing you can find to a 6-pole counterpart to our 2 motors, Woody?

[Tritium_James]

This affects the inductance of the motor, which will be much lower, which affects the switching frequency needed in the controller.

Yes. But I think modern VFDs are up to the challenge. Tritium_James?

Not really a concern. We drive ironless BLDC motors (which do require inductors to even achieve this inductance) down to as low as 50uH per phase. In reality we can go even lower, but only if you can guarantee it's not going to saturate and drop inductance.

The induction machines we've seen are hundreds of uH (our low-voltage wound SEW) up to almost a mH.

[Richo]

I don't think it's valid to double the nominal continuous output when you double the speed, that assumes that doubling the speed doubles the cooling, which isn't right.

Where is the extra heat coming from?

Increasing the RPM requiring higher voltage and giving more power but current stays the same.
There would only be extra frictional and lamination losses which is only marginally more.

You would only need extra cooling if you were increasing the current to get the extra power.

[woody]

That sounds fair. I was assuming efficiency stayed the same, so losses were proportional, but that's not a great assumption.

[weber]

I don't think it's valid to double the nominal continuous output when you double the speed, that assumes that doubling the speed doubles the cooling, which isn't right.

Where is the extra heat coming from?

The frequency has doubled, so the iron losses (hysteresis and eddy-current) will more than double.

[T2]

So T2, we see it is the increased frequency, not the increased rpm, that gives us more power from the same size motor.

Eurodrive
      2Hp 3600rpm 47lbs
      2hp 1800rpm 47lbs
With the DRE90L2 and DRE90L4 frame sizes resp. (1E2 effcy rating)

Explanation they just reversed one pole to wipe out the consequent pole of the 4-pole as Coulomb has suggested. Therefore no weight change. Any attempt to draw the same torque on the two pole will simply cause overcurrent on that machine.

      3Hp 3600rpm 57lbs
      3hp 1800rpm 63.9lbs

This time they decided to modify the two pole further
by either a) shortening the stack - knowing it was already overfluxed
        or b) by removing its feet ?

Look, I don't know. I give up. Perhaps someone with more standing could get someone from the Elec. Eng Dept of a local University to drop Eurodrive a line. Personally I think in the age of inverters the motor manufacturers have finally realized they are oversupplying the market with too many options, consequently standardised stack lengths to promote flexible manufacture have become the norm. How else to explain 6lbs difference between 7.5Hp and 10Hp but 64lbs between 5Hp and 7.5Hp ?

Look, if I needed a 60Hp with only 50Hp and 75Hp options, I would change the timing belt pulley on the 50Hp and run it faster or just accept the incremental cost of the 75Hp. In my experience most motors do not get ordered through decisions of electrical engineers. Mostly electrical draughtspersons or mech engs who are less knowledgeable of the options that inverters present and are more likely to order the next size up anyway. Electrical engineers do have a role however. They are usually called in when the motor turns out to be undersized and "we have to ship on Friday" Although I expect these types of screw-ups are quite foreign to you folks in Australia.

[coulomb]

I say torque/mass ratio increases significantly (but not quite linearly) with pole count.

I happened to be reading this DIYelectriccar post this morning, where Tesseract says:

"The higher the pole count the more "back iron" is needed in the stator. This is because torque is proportional to flux and the more flux required the more iron area is required to support it without saturating."

Right! Why didn't I think of this before? Iron has a particular flux density (flux per unit area of the pole face) limit before saturation. The higher the torque, the higher the required flux, so the higher the required pole area, so the more iron you need, and that increases the mass of the motor pretty much proportionally (not only the stator iron but also the rotor mass and mass of the case and bearings would increase about in proportion to the area of the pole faces).

So that's the basis of the rule of thumb that says that a motor's size and mass is roughly proportional to the peak torque.

My apologies for those of you realised this long ago, and just didn't write it down, or if you did, I mustn't have read it, or not let it sink in.


Author Tesseract goes on to say:

"This also means that higher pole count motors tend to weigh more for a given power output. "

Industrial motors are usually specified for line frequency, 50 or 60 Hz. So for the same power, as the pole count increases, the speed decreases, so the torque increases.

However, our situation is a little different. We don't want decreasing RPM, we want the same RPM, either to replace the ICE when keeping the multi-speed transmission, or to power a direct-drive vehicle; assuming a constant tyre size and other requirements, the RPM range is the same. So for the same power, changing pole counts *in our application* does not change the torque requirement, it only changes the electrical frequency (higher for higher pole count, to achieve the same RPM range).

So purely from a back-iron perspective, it seems that for our application, pole count should not matter as far as motor size and mass is concerned.

Weber claims that merely increasing the electrical frequency has the capacity to reduce the mass of the motor, just as it decreases the mass of a power transformer. I don't have a feel for why that happens in power transformers. I also don't see how the necessity to support a particular flux value to support a particular torque value to support a particular motor power can be overcome.

So Weber: do you have evidence to support your claim? I suppose there are three questions here:
1) Why do power transformers decrease in mass as the frequency increases? What quantities are delivered per cycle so that power increases as the frequency increases?
2) Does this mechanism transfer directly to motors? Does it only work if the stator is made of ferrite? I've never seen an EV-sized motor stator made of ferrite, so I assume it's not practical. A crack in the stator from a large bump would presumably make the motor worthless, and could possibly cause an accident by locking the rotor.
3) If this mechanism can translate to motors, how does it overcome the necessity for the same total flux as a lower pole count, lower frequency motor?

[ Edit: oops! I didn't read past that message to Major's reply, where he shoots a lot of what I said above down in flames. I'll have to research this a bit more and re-present my objections, if any. ]

[weber]

I say torque/mass ratio increases significantly (but not quite linearly) with pole count.

I happened to be reading this DIYelectriccar post this morning, where Tesseract says:

"The higher the pole count the more "back iron" is needed in the stator. This is because torque is proportional to flux and the more flux required the more iron area is required to support it without saturating."

Right!

No! Not right. I'm sorry, but Tesseract is completely wrong. The truth is exactly the opposite. A 2 pole motor needs the most back iron (the outer part of the stator, connecting the slot-teeth) because half of the total air-gap flux must pass thru each of two halves of the back iron at any given time. With a 4 pole you have only a quarter of the air-gap flux passing thru each of 4 quarters of it at a given time.

So if the stator and rotor diamters were the same, then the 2-pole would need shallower slots than the 4-pole. See any textbook on induction motor design.

Why didn't I think of this before? Iron has a particular flux density (flux per unit area of the pole face) limit before saturation.

Yes. And of course it's not only the pole faces that matter, it's around the whole magnetic circuit. And its exactly the fact that you don't want the outer part of the stator to saturate any more than the teeth do, that means a 2-pole needs this outer part to be thicker.

The higher the torque, the higher the required flux, so the higher the required pole area, so the more iron you need, and that increases the mass of the motor pretty much proportionally (not only the stator iron but also the rotor mass and mass of the case and bearings would increase about in proportion to the area of the pole faces)

This is typical of the kind of mistake people keep making in this discussion. Stating some proportionality rule which is correct when all other things stay the same, and then assuming it still applies when other things (such as pole count) don't stay the same.

What produces the torque? It's due to forces acting at a distance from an axis of rotation. What distance? Essentially the radius of the rotor. Does a 4 pole have double the rotor radius of a 2 pole of the same size and mass. No. It may be slightly larger but nowhere near 2 times. And a 6 pole's rotor certainly doesn't have 3 times the radius of a 2 pole's rotor or there wouldn't be any stator left!

What produces the forces? One way of thinking about it is that they are from
F = BIl where
F is the force in newtons
B is the radial flux density in teslas or webers per square metre
I is the current in the rotor bars in amps
l is the length of the rotor bars in metres

Note that the force at any given pole is proportional to the flux density, not the total flux. And the maximum flux density can be the same no matter how many poles the motor has, as it depends only on the saturation property of the iron (as you mention above). The length and number of rotor bars and the maximum current through them can also be the same for a given size and mass of motor, no matter how many poles.

So, to a first approximation, we can get the same BIl at every pole, and so a motor with more poles will have more torque for a given mass and size.

So that's the basis of the rule of thumb that says that a motor's size and mass is roughly proportional to the peak torque.

There is no such rule of thumb! Unless you fix the number of poles. If there was, you'd need to explain the catalog data that has been posted by T2, Woody and myself that clearly shows that a 4-pole has almost double the torque of a 2-pole of the same size and mass?

Weber claims that merely increasing the electrical frequency has the capacity to reduce the mass of the motor, just as it decreases the mass of a power transformer. I don't have a feel for why that happens in power transformers.

So Weber: do you have evidence to support your claim? I suppose there are three questions here:
1) Why do power transformers decrease in mass as the frequency increases? What quantities are delivered per cycle so that power increases as the frequency increases?

It is because you can transfer a certain amount of energy per cycle as the flux transitions from near-saturated in one direction, to near saturated in the other direction and back again. Energy stored in a magnetic field is
E = 1/2 * mu * omega^2 where
E is energy in joules
mu is permeability in henrys per metre (or newtons per amp squared)
omega is flux in webers

i.e. You have a fixed size "bucket" for the energy. But the faster you transfer buckets, the more power you have, as power is rate of energy transfer.

2) Does this mechanism transfer directly to motors?

Yes. Because you are still converting energy in a manner where, at one stage in the process, it is stored in a magnetic field.

Does it only work if the stator is made of ferrite? I've never seen an EV-sized motor stator made of ferrite, so I assume it's not practical. A crack in the stator from a large bump would presumably make the motor worthless, and could possibly cause an accident by locking the rotor.

Ferrite is only needed for transformers that operate at tens of kilohertz or more. We are only talking about operating EV motors at up to 400 Hz. 400 Hz motors and generators are already used in aircraft. (And why do you suppose that is?) They only require thinner laminations of the iron.

3) If this mechanism can translate to motors, how does it overcome the necessity for the same total flux as a lower pole count, lower frequency motor?

There is no such necessity. It has been demonstrated many times that if you take a 4 pole motor and rewind it for half the voltage and run it at twice the frequency, you get nearly twice the power of the same size and mass 2-pole motor.

The problem is that with the standard radial-flux induction motor design this trend does not continue to a 6-pole motor rewound for one third the voltage and run at 3 times the frequency, except in motor sizes that are too big for your average EV.

The problem is apparently one of pole geometry and lamination direction. When the poles get too close together there is too much flux leakage between them that does not cross the air gap. This requires more copper to get the same air gap flux and results in a very low power factor.