Field Weakening VS Gear Box VS Higher voltage battery

[macribs]

I'm starting to think of a motor like a flywheel: Get the important stuff (mass for a flywheel, copper & magnets for a motor) out at as greater radius as possible. Everything in the middle is useless apart from being supporting structure for the important stuff and should be minimised as far as materials technology allows.

Yes but still not a one-for-all solution. While it is true that a large diameter hub will have more torque then say a RC they will impact the ride and handling in different manners too.

For a road bike, commuter or citybike that hub will work fin. Bolt it on wherever. Hub or center. It might even prove to give you
an even more stable ride due to rotating mass. Might even work out well for a trial bike making bike more stable in low speed?

But if you fancy another style of riding as well or for the most part, lets say DH, trail riding etc then all that rotational mass might very well slow your handling down too much to be a great choice. A more narrow radius motor like a RC motor would shine in that setting.

 

[Punx0r]

I bet there's maybe 3kg of copper and magnets in a decent sized hub motor, not much really even at the wheel rim. If we ever get room-temperature super conductors (or even graphene wire at 10x the conductivity of copper) we could replace the magents with field windings, use an air-core and have an all up active-material weight of frock-all.

 

[Arlo1]

I bet there's maybe 3kg of copper and magnets in a decent sized hub motor, not much really even at the wheel rim. If we ever get room-temperature super conductors (or even graphene wire at 10x the conductivity of copper) we could replace the magents with field windings, use an air-core and have an all up active-material weight of frock-all.

You replace the magnets with field windings and you will be back to a brushed motor!

 

[Punx0r]

Doh. I had induction motors on my mind all day due to work, should have been more specific...

 

[halcyon_m]

I bet there's maybe 3kg of copper and magnets in a decent sized hub motor, not much really even at the wheel rim. If we ever get room-temperature super conductors (or even graphene wire at 10x the conductivity of copper) we could replace the magents with field windings, use an air-core and have an all up active-material weight of frock-all.

You replace the magnets with field windings and you will be back to a brushed motor!

Brushes, in and of themselves, are not horrible. They can be designed to last a while, and particularly for DC brushes, the commutator will be simply a smooth ring that can minimize wear and have minimal current passing through it.

The tone of this conversation, though, is quite interesting and forward thinking. If you consider how a motor makes power, it's torque multiplied by speed. Simple concept, but here, because the speed of the wheel is much less than the 200MPH rotor speeds of high speed motors, we are talking about 1) reducing the rotational speed by going to direct drive and eliminating any gearing 2) maintaining the speed at the air-gap by increasing the air gap diameter by a similar amount as the decrease in rotational speed, and 3) stretching out the active components of a small and powerful motor to essentially become a ring-motor. The idea itself is all quite sound, but, as LiveForPhysics mentioned, the geometries of the active components become thin and long, which may pose manufacturing/availability issues. Secondarily, the primary issue I see is gap maintenance. Electric motors, particularly optimized ones, require a small and uniform mechanical gap, which is particularly difficult for radial gaps considering deformation of the wheel from bumps and such, but axial may be easier, similar to the circuit-board style stator of Boulder Wind.

Lastly, the only other issue I see is the practicality of seals, from weather/debris. When the gap is so large, typically the seal has to be similarly large, which creates a significant cost, a wear item, and drag. One solution to this may be sealing in the entire wheel from weather, using a velodrome style wheel, where the rotating seal can then be brought to a significantly smaller diameter.

So to re-cap, if you want an elegant and light solution to powering a wheel, spread out the active material (copper, iron, magnets) to the outer-most diameter practical so that the speeds are high and therefore doesn't require as much torque/force from the motor to do whatever job is required.

Caveat: this is only practical if the operating speeds and electrical frequencies of the motor are within reasonable limits from a mechanical perspective (things flying apart) and a loss perspective (high frequency losses in steel or magnets). The rotational speeds depend on the strength of the construction, where the electrical frequency losses are due to the selection of magnetic material, like the thinness of the laminations, or the magnet segmentation. A high pole number (many teeth) will keep the torque quite high, and reduce the magnetic material (iron) to carry the magnetic fields, but if the frequencies from rotation become higher than, say, 1000Hz, (400 poles, 2.5 rotations per second, ~14MPH) then the frequency-related losses become something that need to be considered.

 

[Punx0r]

Covers to seal off the wheel from dirt should be a simple solution. It would cut off air cooling, but the motor might be liquid cooled anyway.

As you say, you do end up requiring very stiff materials, but that's a classic mechanical engineering problem and I'm sure similar problems have been solved many times in recent history. Hopefully now that EVs are starting to go mainstream we will see a big upsurge in R&D invested in the area and the technology will move along leaps and bounds. I hope core losses will be one thing that is greatly reduced or eliminated. The rise in additive manufacturing and non-metallic construction materials could well make many possible things.

 

[speedmd]

Do Like the disk wheel and PC board motor ideas very much. Smart magnetic ( magnetic only when you want them to be) materials would be interesting developments also. Lots of promise for improvements with lighter materials being used more commonly in rims / wheel components that could adapt motor components easily.

 

[circuit]

What started like a usual technical thread, suddenly got very interesting. Thanks to halcyon_m and others for that.
I have been doing some DIY controller programming myself and did notice a dependency of RPM from phase angle. I was doing a synchronous virtual-SVM modulation. It was "virtual", because actual waveform values were stored in a table, due to lack of resources for complex math (AVR MCU with power stage controller unit, ATmega64m1).
I was measuring actual sector time period from hall sensors and interpolating next statistical sector change, from that I scaled my table in time. And I scaled it vertically based on throttle input (speed-control mode). Since I was not doing actual math on dq plane, I was just able to advance or retard my output phases. And I did notice a variation in speed and torque.
I would like to ask if this is the same thing as done with proper math?

I don't really 100% understand how the inductance comes in to play during field weakening. I understand it like this: when phase is advanced, energy is being stored in motor's inductance during first half of sector, where controller output is higher than BEMF, and released back to battery on the other half, where BEMF is higher? Is this correct? If so, it requires quite high inductance to do so. I don't know any hobby outrunner motor that could be suitable for that, as these have inductance of around 16μH (example for Rotomax 150cc, which is a very smooth-running motor and best build quality I have seen). But some hubs could be good enough.

Also I think that a shape of BEMF on low induction motors could be a real issue for many controllers, even "pure sine" ones, because waveforms will not match, inductance will be too low to integrate that properly and we will get huge IR losses. Am I correct on this one? If so, to solve this, controller would need to exactly follow the BEMF shape to avoid increased losses. I had an idea to measure and store this shape in my low-resource controller's waveform table and use that for phase outputs. Would this approach be good enough?
And also, in proper SVM control with all the fancy math involved, does it actually take that in to account? I've read several application notes and it looked like they only use sinewaves and no actual BEMF is being matched. Maybe except for some current-mode algorithms, but those are rare to find well documented.

To rephrase all that, could it be done without fancy real-time math?

Thanks in advance.

 

[Arlo1]

Think of inductance like a rubber band. If you have more inductance you can stretch the rubber band further hence storing more energy so to speak.

 

[Punx0r]

Having read some of the large thread on the Colossus motor and its inductance, it seems it's a compromise to make the controller's life easier. Specifically, with low inductance, current rise in the winding is so fast saturation occurs before the end of the PWM pulse, and saturation means reactance of the winding plummets and very high currents flow, encouraging controller failure because the controller doesn't adequately limit the current (and also high resistive losses in the winding). One solution is to increase PWM switching speed to avoid saturation, but that results in higher core losses in the motor, due the non-ideal magnetic material.

Compromises abound!

 

[Punx0r]

Just a thought: If the motor controller could boost as well as buck, would field weakening be unnecessary?

 

[Miles]

Just a thought: If the motor controller could boost as well as buck, would field weakening be unnecessary?

That came up earlier: http://endless-sphere.com/forums/viewtopic.php?p=949528#p949528

 

[Punx0r]

Thanks Miles. I'd failed to recall that discussion.

The answer appears to be that it increases controller losses. I figure field weakening also increases system losses, so one solution might be preferable to the other, but neither is ideal.

We merely need a controller that acts like an ideal current source

 

[Arlo1]

Thanks Miles. I'd failed to recall that discussion.

The answer appears to be that it increases controller losses. I figure field weakening also increases system losses, so one solution might be preferable to the other, but neither is ideal.

We merely need a controller that acts like an ideal current source

The ideal solution is to make the EV run without field weakening most of the time. IE normal driving my CRX runs up to ~110 km/h but if I need to pass a semi on the highway its nice to have that extra speed and torque at the higher speed needed to make the pass.
Your normal day to day driving should not use field weakening just for the rare occasion you need some torque at higher speeds as well as more top speed its there.

 

[bobc]

I had to do some work on a commercial EV controllers' PWM. The issue for us in a commercial development was that if there were an emergency shutdown at high fieldweakening, the diodes in the inverter would jack up the dc link (battery) to well over the intended level resulting in either 1) driven wheels breaking traction or 2) battery fuse death quickly followed by death of the entire inverter due over volts on the power semis. Neither very attractive options.
I had to effective install programmable crowbar protection in the modulator to keep the inverter viable, while avoiding lockup.
That's a nice feature of the AIM drives (e.g. tesla) - you can always just switch off all the gate drives if you need to....

 

[Arlo1]

I had to do some work on a commercial EV controllers' PWM. The issue for us in a commercial development was that if there were an emergency shutdown at high fieldweakening, the diodes in the inverter would jack up the dc link (battery) to well over the intended level resulting in either 1) driven wheels breaking traction or 2) battery fuse death quickly followed by death of the entire inverter due over volts on the power semis. Neither very attractive options.
I had to effective install programmable crowbar protection in the modulator to keep the inverter viable, while avoiding lockup.
That's a nice feature of the AIM drives (e.g. tesla) - you can always just switch off all the gate drives if you need to....

Can you share any numbers?
IE what the no load and loaded top speed was with and without field weakening and what the back emf was?

I ask because at 160km/h I have lost synk with my controller and I used the mechanical brakes to slow down. Top speed is 113-130 with no field weakening so I am sure it could have been a problem but it never was.

 

[Lebowski]

The current would be in the order of the fieldweakening current the controller was providing.

during fieldweakening, the extra voltage on top of the battery voltage is fieldweakening current times motor impedance (the inductance).

So when the controller looses sync and shuts down, the voltage differencs is as above, same for the motor impedance so the current
will be roughly the fieldweakening current (a bit less as the controller would also make over-voltage to generate torque producing phase current).

But remember my reservations about putting field weakening in ? bobc is listing the reasons...

 

[bobc]

Can you share any numbers?

There are several reasons why not, mostly it was ~10 years ago & I've forgot & partly it's my employers company IP..... (kop out I know....)
But I think they were looking for a field weakening range from the system of over 2:1 making the back emf worries a real thing; You've put desat protection on your big inverter so it is maybe possible you could get into this state due to a small component failure somewhere. Obviously there's a world of difference between an inverter you've just made for yourself & one that many thousands of customers have paid a lot of money for, both in the likelihood of something going wrong and how much you'll be liable if it does....

 

[halcyon_m]

Designing a motor with a characteristic current (current when shorted) to the nominal rating allows the controller to just short the motor terminals in a fault scenario. The issue bobc mentioned just goes to show that "converter fault --> all switches open" is not always the best assumption.

Bottom line: if the motor is not designed for field weakening, then the discussion is somewhat irrelevant. I haven't seen any hobby motors designed to have the right magnetic curcuit for it.

Larger EV motors typically have this capability. Induction motors sort of do this naturally, requiring no current to push against permanent magnet flux.

 

[Lebowski]

Bottom line: if the motor is not designed for field weakening, then the discussion is somewhat irrelevant. I haven't seen any hobby motors designed to have the right magnetic curcuit for it.

What do you mean by this 'the right magnetic circuit' ? I've seen a lot of nonsence about this even in recent literature (few years old). As far as I am concerned: field weakening has nothing to do with the permanent magnets (as you can do field weakening with external coils) and the only requirement is enough impedance from the (not even motor-) inductance.

 

[liveforphysics]

Designing a motor with a characteristic current (current when shorted) to the nominal rating allows the controller to just short the motor terminals in a fault scenario. The issue bobc mentioned just goes to show that "converter fault --> all switches open" is not always the best assumption.

Bottom line: if the motor is not designed for field weakening, then the discussion is somewhat irrelevant. I haven't seen any hobby motors designed to have the right magnetic curcuit for it.

Larger EV motors typically have this capability. Induction motors sort of do this naturally, requiring no current to push against permanent magnet flux.

The words "field weakening" came from Sep-ex motors, as you can take off from 0rpm with a high stator field and then decrease stator current in a Sep-ex as RPM builds to reduce the BEMF and continue the power band out further.

Field weakening in a PM motor simply means dumping the current into the winding ahead of the BEMF making it impossible (as BEMF climbs to above pack voltage when the motor is spun above base speed.) You can look I'm at it like timing advance, but to be fair it's slightly more involved.

Pure PM motors "field weaken" great. Reluctance hybrid IPM motors "field weaken" great. Induction is slightly different but also permits going well beyond base speed.