Have you been living under the rock for the last six months or so? If not, you have certainly read or heard something about a new type of electric motor. Designed by Linear Labs Inc, and going by the acronym HET, the motor certainly promises a lot. Huge torque first and foremost, good efficiency over a wide speed range, no need for VFD, redundancy – you name it (pretty much).
Or rather, the marketing promises a lot. But how much can the technology actually deliver? Prepare for a sceptical evaluation.
This post consists of two parts.
First, we take a look on what we actually do know about the motor, together with some comments by yours truly.
Next, we take a deeper look on what could be seen as the most controversial part – the motor’s topology and the corresponding claims of huge torque density.
Note: All pics either by my, or from a presentation of Linear Labs, referenced Dec 12 2019. Guess this constitutes fair use or something.
What we know
The most distinctive characteristic of the HET is obviously its less-than-standard topology. You can see it in the picture, but maybe a little mental exercise can help you wrap your mind around it later.
First, imagine a standard outrunner motor, with surface-PM rotor and distributed winding in the stator. Easy enough.
Next, take an inrunner, and put it inside our outrunner, so that the two stators are now yoke-to-yoke.
Still with me?
Next, let’s change the winding a little. Instead of connecting inrunner slot 1 to inrunner slot 7, for instance, let’s connect it to the outrunner slot 1. Repeat the same for all slots. Now, our winding resembles a toroidal inductor, wound around the back-to-back yokes of our inrunner-outrunner machine.
In other words, the coils are now wound around the toroidal stator yoke, like this:
Finally, let’s take the rotors of two axial flux PM motors, and slap them on the ends of our double-machine, turning it into a four-rotor machine.
Still with us? Great, because that’s it!
This topology has been called a toroidal-flux motor – and rightly so, if you ask me. After all, the air gap surfaces do span a cylindrical toroid.
Variable shift between rotors
To keep things complex, the relative positions of the four HET rotors aren’t fixed. Instead, two of them (the end-ones if I remember correctly) can be slightly rotated with respect to the other two.
What this does is that it ‘short circuits’ some PM flux outside the stator winding, decreasing the flux linkage.
This is field-weakening. True field-weakening, without having to inject a strong current component on the d-axis like normally in SPMs. Which is more like field opposition, really.
Anyways, being able to reduce the flux linkage like this can be a huge plus, greatly reducing copper losses at high speeds.
However, it is not clearly how much of the PM flux still penetrates into the stator iron (without linking the winding), and how much circles purely between the rotors. While this doesn’t matter from the circuit point of view, it does for iron losses.
No VFD needed
I call terminology-BS on this.
HET might not need a separate VFD box to be bought independently, but I’m pretty much 100% certain it does utilize some power transistors of one kind or another to output a time-varying voltage. An inverter (or AC-AC converter, if you connect it to an AC socket instead of battery) in other words.
Which is nothing to look down at – an integrated inverter solution is no small feat in itself (even if it didn’t have all the safety and grid compliance and whatnot features of a mature VFD).
Independent control of coils
This is closely related to the previous point. The available info suggests that most, if not all, coils can be controlled independently.
Meaning, the motor can be driven in more-than-3 phase configuration. This probably means lower phase voltages, requiring less insulation between the coils and stator core.
Linear Labs also claims that this improves reliability. Alright, losing one phase out of say nine is no big deal, at least compared to one out of three for instance.
However, there are more to issues to consider than this. For instance, if there is a short-circuit, be it turn-to-turn, turn-to-ground, or coil-end-to-coil-end, large fault currents will be induced by the PM flux. Importantly, not much can be done about this by the motor controller, apart from stopping the machine altogether. How the HET survives this – how ANY motor survives this – would require much closer analysis.
I can’t remember if Linear Labs claims this or not, but they wouldn’t be the first to do so.
Namely, the term “electric transmission” has come to mean being able to switch the connection of coils on the fly. Specifically, they would all be connected in series at low speeds, and progressively in parallel as the speed increases.
From the electromagnetic point of view, this would be very nice. It would enable much higher low-speed torques and/or higher top speeds to be reached with the same inverter rating, and would improve regeneration efficiency at low speeds.
The problem lies primarily on the inverter / power electronics side; doing the aforementioned thing reliably without losses too large.
“All slots are firing”
A claim I have seen many times, often contrasted with normal motor which allegedly only have 1/3 of slots doing it.
I’ll let the less-than-standard terminology pass – anybody with a brain cell can understand the intent, and I’m not a NEMA/IEC poster boy myself either.
A bigger issue is that the claim is simply incorrect.
Consider a normal SPM with one slot per pole and phase, supplied with sinusoidal currents and operated in the id=0 mode. Every single slot is carrying current and producing torque, every single moment, apart from the zero-length instant where the current crosses zero.
Even if you add more slots per pole and phase, and can’t perfectly synchronize the current in each slot to the airgap flux density, the net product is still quite darn good. (Each slot contributes positive torque a ratio of time equal to the winding distribution factor, usually in the 95 percent range or above.)
The claim is slightly less incorrect for BLDC motors driven with trapezoidal modulation. In this case, 2/3 of coils will be carrying a roughly-constant current at any given time (outside the switching instant + transient), while the third one will be open and not carrying a current. Still, 2/3 is exactly 100% more than 1/3.
Now, finally, let’s look deeper into the topology, and the claims of superb torque density in particular.
Let’s address the smallest elephant in the room first.
Of Lorentz force
One claim that is (or at least was at one point) often repeated is that the “Lorentz force” is parallel to the rotor movement at any point in the coil.
The thing is, iron-cored motors rarely rely on actual Lorentz force! The current flows in the winding, while flux flows in the teeth, making the actual JB product very small, often less than 10 % of the total torque!
Where the Lorentz force does appear is the equivalent airgap model, which directly exposes a surface current sheet to the airgap flux. The equivalent airgap model can give good results (exact, even), but only when the equivalent current is properly defined, which can be a nontrivial task for more esoteric topologies. And still, this does not mean that the actual motor produces any of its torque via Lorentz force.
In all fairness, the claim seems to be correct for the equivalent airgap model of the HET. At least on the first glance, it can be modelled with the textbook airgap models: two of the radial-flux type and two axial ones. And in this case, the entire winding is producing equivalent Lorentz force.
Well, yes. But no.
And yet, I argue that the ‘entire winding produces torque‘ claim is misleading.
Why? Because in the end it doesn’t really matter if the entire coil length contributes to the torque, but how much torque we get.
Let’s take another look at this outrunner-inrunner pic here. The core of the inrunner is drawn with a different shade of grey for a specific reason: to underline the fact that it is there.
Lemme really stress this: we added an inrunner inside an outrunner. We didn’t just add another rotor to an outrunner.
If we had followed the latter option, we would have been forced to reduce the flux density of the outer airgap, to avoid oversaturating the stator yoke. And then we would have been again forced to reduce it even further after the addition of the two axial-flux rotors.
In the end, we would have ended with a motor with the same flux linkage* and rated current**. The entire coils would have been active / contributed to the torque, yet the actual torque output would have been the same***! To add insult to the injury, the entire winding mass could have been larger than before, resulting in increased copper losses****.
* Let’s say the original outrunner had a flux of 2x Wb per coil. This flux splits in two when it enters the stator core, like traversing a Y junction. Each of the toroidal coils only sees onr half of this flux, so x Wb per coil. But again, we have twice as many coils now, resulting in the same 2x Wb per pole and phase.
** Assuming the current density, slot area, filling factor, and cooling are maintained.
*** Again, since flux linkage and current are unchanged. Plain energy balance here.
**** In the original outrunner, coil length per pole and phase is 2 x axial length + 2 x (long) end-winding. In our bastardized HET here, the length is 4 x axial length + 4 x (short) radial not-really-end-winding segments. Which, depending on the dimensions like pole count and diameter, might be more than before.
Now, I understand the above example may seem unfair – because it is. We’ll come back to that later.
My goal is to simply keep my readers from violating one of the Great Rules of Motor Design – the flux must go somewhere. Meaning, you can’t justslap on four times as many rotors around a motor and expect a quadrupled output torque.
Instead, you’ll have to make sure there is actually space for the flux to flow, primarily in the stator yoke.
That all been said, I do see some benefits in the HET topology.
For example, think about short large-diameter motors with a high pole count. These motors are very much ring-shaped, with a decent amount of not-so-actively utilized space in the middle. In this case, the ‘let’s put an inrunner inside it’ approach, i.e. adding moar rotorz and the corresponding volume of stator iron does indeed very much increase the torque from the same space envelope, if not the same motor mass.
The four-rotor topology does what is commonly called flux focusing, all feeding flux to the same stator. Thanks to this, Ferrites and other low-grade magnets can be used.
I’ve saved what is perhaps the most important point last. Let’s say you wanna minimize the losses at low speed, or maximize the torque per Watt of copper loss in other words. For that purpose, you want to maximize the ratio of coil flux (divided by pole count, but let’s assume that number is fixed now) to coil length. To achieve this, you can do several things.
First, you’ll want to maximize the average flux density within the coil. The HET certainly does this, as the coils are wound around the stator yoke. This could easily mean around 1.6 to 1.8 T. By contrast, for a typical radial-flux motor the figure would be somewhat below the airgap flux density, so below 1 T most of the time.
Secondly, the coil shape (and the shape of the iron core inside it) should be as close to circular as possible. Manufacture-wise, this might be difficult, but a rectangle is a decent approximation.
Finally, you’ll want to make the coil (or core, rather) as large as possible. Again, this is simple geometry as the circumference grows linearly and the area quadratically.
Whether or not the HET topology is good in this respect depends especially on the pole count, and the given space envelope.
Consider, for instance, a typical two-pole motor. Pretty much the entire motor volume is already in use, carrying either flux or current. So, switching to the four-rotor topology would be tricky if we want to stay under the same space constraints. And even if we did switch to a toroidal winding, we wouldn’t really gain anything, as our earlier example already demonstrated.
If, however, we again consider the large-diameter high-pole-count motor, things change. The yoke profile is probably quite thin, due to the small pole flux. Just adding the outrunner component (assuming the original motor was an inrunner), would (almost) double the yoke thickness, (almost) double the output torque, while doing relatively little the the coil length. And adding the two axial-flux rotors would even further help in this respect.
Drawbacks and open questions
Now, what are some of the drawbacks here?
Manufacture and mechanics
Oh, boy, does it seem complex. Some issues to solve (that may or may not already have been) include:
- Gearbox for changing the relative rotor shift, with its associated losses, cost, maintenance, and reliability issues
- Flux paths in the teeth seem very much 3-dimensional, which would require sintered materials to be used to limit eddy-current losses.
- The coils either have to be wound around the stator core directly, or the core has to be assembled from segments.
- Cooling issues: can enough cooling air be forced inside the rotors?
- Cooling issues, again. The primary cooling path for more traditional motors is usually conduction through the stator yoke, to the frame, to either air or coolant fluid. For the HET, this path seems almost completely missing.
- Design compromises: for example the slot area on the all four faces of the stator has to be equal. Each of the four “sub-motors” of the HET will perform sub-optimally compared to an independently-optimized motor of the same topology.
As a summary:
- The multi-rotor topology does enable better utilization of the space envelope at high-enough pole counts, resulting in improved torque per envelope-volume.
- This does not seem to do anything to improve the torque-per-mass ratio, apart from slightly reduced winding mass.
- Assuming the cooling performance remains unchanged, the reduced winding losses could then be cashed in by increasing the current density, and thus torque, torque-per-mass, and bringing losses back to the original level.
- Likewise, at high-enough pole numbers, the topology can improve the flux-to-coil length ratio, improving torque per Watt of copper loss.
- Again, this could be translated into better torque-per-mass, via increased current density.
- Ferrites work well with it.
- The topology’s not panacea, has many caveats, and will not 4x the torque output of any arbitrary motor.
- Adjusting the relative angle between the rotors yields wound-field-like field weakening control with little to no extra losses
- The claims surrounding power electronics (integrated inverter, independent control of coils with or without series-to-parallel changes, redundancy) are very impressive on their own.
- Some of the marketing phrases and claims are either vague, misleading, or downright incorrect.
Need help with electric motor design or design software? Let's get in touch - satisfaction guaranteed!