SPM VS IPM Motor - And Field weaken, is it worth it

Ok. But i wanted to clarify: the phase of the voltages is of no importance its about the negative d current which produced a magnetic field in oposite direction of the magnets field and there for lowering the over Flux densety in the air gap which lowers the BEMF voltage.
 
That link to the phasor stuff is an awesome explanation Lebowski, as is your thread on the topic.
Interesting that (for me at least) its 3 years later that i take the time to learn from this stuff.

Lebowski said:
In my particular controller algo it will always try to make the torque producing current (Id ?) equal to what is asked for by the throttle. It will raise controller output voltage to achieve this. When raising motor voltage is not enough to produce the wanted Id (?) it will start adding (ramping up) field weakening current, again with the goal to make Id equal to the throttle wanted current. There is a mechanism in my controller that limits the field weakening current, it can never go above the max set value, and it will automatically not go into the region where more field weakening current slows down the motor.

This last limit is dynamically calculated, I do not use a table. From the motor guys at Microchip I understand that this is not a know (in the literature) algorithm, and that it is common to use tables / graphs.

If I remember correctly Arlo1's car only went around 110kmh with no field weakening, and upto 180 with...

Only thing is though, that if the above description is how your current algorithm operates below base speed for a Nissan leaf motor, then our friend arlo misses out on roughly half of the available torque output compared to adding lots of field weakening even at only 3000rpm.

ie the very substantial level of reluctance torque this motor is designed (and mapped for by the factory) to produce is not being realized?
-There must be some other goings on here.
 
It depends on how the Leaf motor is build. Getting power out of a motor is all about aligning the bemf voltage vector with the current vector (assuming no field weakening). You can show mathematically that the reluctance also follows this, i.e. the reluctance of the motor also results in a backemf like voltage vector. My controller does not distinguish between the two voltage vectors but only deals with the sum of the two. In this case max output power is achieved when the back emf and reluctance voltage vectors are aligned, something which is determined by motor design.
 
Lebowski, not sure if i get u right, but if u mean that u align ur current with the "bemf" formed out of the magnetic inducted voltage and the inducted voltage by the current flowing then u would handle IPMs and SPMs not in the optimal operation point. I know ur implementation is not like industry does and there for its difficult to back calculate any values from your controller to the equations i am used to. I would asume from that ur observer tracks the hole inducted voltage which is influenced by the current flowing and there for changing the base angle of the current controllers?
 
tecnologic said:
Lebowski, not sure if i get u right, but if u mean that u align ur current with the "bemf" formed out of the magnetic inducted voltage and the inducted voltage by the current flowing then u would handle IPMs and SPMs not in the optimal operation point. I know ur implementation is not like industry does and there for its difficult to back calculate any values from your controller to the equations i am used to. I would asume from that ur observer tracks the hole inducted voltage which is influenced by the current flowing and there for changing the base angle of the current controllers?
Well the inducted voltage is used for calculating the timing advance. Switched reluctance in addition creates a second voltage vector due to the position dependent inductance. This vector, combined with the current, is where the mechanical power in a SR motor comes from.
 
Switched reluctance motors work totally different than sync machines, you could compare it with a SynRM but not switched reluctance.

For me this is unclear because reluctance will never produce a voltage it's always the current and that is set by yourself so it's no measure for an optimal operating point.
 
tecnologic said:
madin88 said:
tecnologic said:
The flux path i was discussing there was the one from the V grove rotor not the qs3000w u posted up front. For the qs3000w u are right there is not much reluctance torque possible, it will only come by the magnets saturating the stator iron (5-10% max).
The pictured rotor with the V-groove magnets is QS3000W midmotor! :)
So unfortunately it really doesn't have much potential for additional reluctance torque due to the high magnet coverage, or with other words those very thin steel sections between the magnet poles don't allow it.
I was referring to the pic from emoto that he posted after u. For your motor there u are right.
tecnologic I think you were quoting the other rotor pic/ similar.
So Just to clarify :) and be more specific on the qs ipm
If the surrron has a bad flux path due to small gap between magnets, Quoted as ''pinching'' as in the vid i posted @ 17min .
How is The qs ipm rotor a good design... [as it has a possibly smaller gap.] more pinching for iq Flux , therefor affecting reluctance torque potential .
 

Attachments

  • unnamed.jpg
    unnamed.jpg
    31.4 KB · Views: 3,609
Lets take a look at Tesla Model 3 motor which is called "IPM-SRM" hybrid motor and has a large visible flux path between the magnet poles:

ACTUAL-Rotor-close-up-with-laminates-3.jpg


the motor from Toyota Prius 2010 is similar:

tesla_ipm.png


Can we say that if the flux path between the magnets is poor, the potential for relutance torque is also poor?
And lets assume we have a given motor and an induction meter, can we "measure" it's potential for reluctance torque, or the general potential for field weakening?
Is it just always the differnce between Ld and Lq, and no matter if SPM or IPM or any other rotor, that will tell us the potential (i understand that no inductance difference = poor, big difference = superb)?
 
madin88 said:
Can we say that if the flux path between the magnets is poor, the potential for relutance torque is also poor?
And lets assume we have a given motor and an induction meter, can we "measure" it's potential for reluctance torque, or the general potential for field weakening?
Is it just always the differnce between Ld and Lq, and no matter if SPM or IPM or any other rotor, that will tell us the potential (i understand that no inductance difference = poor, big difference = superb)?

This is my understanding of how it works from going through the literature. But regular SPM should never have a large saliency ratio right?

The other parameter you want to know is magnet flux linkage. You can calculate this from kv and pole pair count. Flux linkage / inductance is characteristic current or short circuit current. Field weakening is optimized when this is equal to rated current. This makes sense as motors with strong magnets relative to stator flux will require large currents to oppose the magnets. If you try this calculation with most of our surface magnet ebike motors, you will obtain very large currents.
 
district9prawn said:
madin88 said:
Can we say that if the flux path between the magnets is poor, the potential for relutance torque is also poor?
And lets assume we have a given motor and an induction meter, can we "measure" it's potential for reluctance torque, or the general potential for field weakening?
Is it just always the differnce between Ld and Lq, and no matter if SPM or IPM or any other rotor, that will tell us the potential (i understand that no inductance difference = poor, big difference = superb)?

This is my understanding of how it works from going through the literature. But regular SPM should never have a large saliency ratio right?

The other parameter you want to know is magnet flux linkage. You can calculate this from kv and pole pair count. Flux linkage / inductance is characteristic current or short circuit current. Field weakening is optimized when this is equal to rated current. This makes sense as motors with strong magnets relative to stator flux will require large currents to oppose the magnets. If you try this calculation with most of our surface magnet ebike motors, you will obtain very large currents.
This is my understanding to as I have done MTPA on 6374s and industrial servos both show at least a potential for reluctance torque and both are more or less SPMs. When I had the time to finish my Motortest Rig I will post a video of MTPA on and off in comparison for a simple SPM inrunner (ACT 86BLF04). But this will take some time till i am home again.
 
Lebowski said:
My take on switched reluctance and how it produces a voltage:
(bear with me if there are any mistakes, it was new year eve after all :? )
slide_1-800x800.jpg
slide_2-800x800.jpg
slide_3-800x800.jpg
Mmh. All equations I know but the switched reluctance distracted me😀. And I am still not sure what you are exactly doing with current but I assume we are doing the same, except that I use more computational effort for the same😀.
 
tecnologic said:
Hi Emoto,

its not that both torques fight each other, its more for what purpose the motor designer designed the motor.

The V-grove arrangement of the magnets is actually a very good one to reduce magnet mass to iron ratio with the side effect of a slightly higher max torque. But this is only the case when the motor is driven the way its ment to. Just by the looks of this rotor i would assume that using a current command of 2/3 iq and 1/3 id ist more efficient that id = 0 and the hole current command on iq.

What makes you think just by the looks of this rotor (QS3000W mid) that 33% Id would be more efficient (will produce more torque at same phase amps?) as 0% Id?
To not leave Emotos question unanswered.

tecnologic said:
U are right maddin, but this is due to false FW handling by those controllers. For rpm/Kv > V battery id must always be != 0. regardless of torque command. Every industry frequency converter i know is handling it like this.

hmm, if the controller does it wrong or right is hard to say IMO.
question yourself:
Would it be better to have a higher potential for a gain in RPM with the downside of regenerative braking if throttle will be realeased (only at excessive speeds where BEMF is higher than battery voltage).
Or would you prefer to have less potential for a gain in RPM without the issue of regenerative braking?

I am talking particulary about Nucular controller which does handle FW like this (it is called "soft weakening shut down").
Adaptto recommends to set a speed limitation based on 120% of no-load speed (the controller will not allow the motor to spin faster than this to prevent damage due to too high BEMF).

Has anyone here experience with ASI (Phaserunner), Sabvoton, VESC or any other "ebike" controller which can do FW?
How is it there?
 
Lebowski said:
It depends on how the Leaf motor is build. Getting power out of a motor is all about aligning the bemf voltage vector with the current vector (assuming no field weakening). You can show mathematically that the reluctance also follows this, i.e. the reluctance of the motor also results in a backemf like voltage vector. My controller does not distinguish between the two voltage vectors but only deals with the sum of the two. In this case max output power is achieved when the back emf and reluctance voltage vectors are aligned, something which is determined by motor design.

It would be cool if your controller brain does it as good as the "big industry" which are using tables to achieve MTPA :D

Is the max RPM your controller can spin a motor to also limited by BEMF VS. Vbatt, and would make it a differnce if Mosfet's or IGBT's are used in therms of that?
I am trying to understand the theory behind what happens in practice.
 
One bit here i meant to get back to you about madin is the 'unwanted regen/blowing fets' thing.
-If the controller is half smart about it then it continues to supply field weakening current even when you are off the throttle to keep the motors generated voltage below battery level so no regen needs to happen unless requested.

Its only under a fault condition where at say double base speed and motor heavily field weakened, the controller ( gone into shutdown or missing a signal etc) even for a microsecond fails to 'keep a lid' on the motors excessive back EMF then there could be dramas.
 
toolman2 said:
One bit here i meant to get back to you about madin is the 'unwanted regen/blowing fets' thing.
-If the controller is half smart about it then it continues to supply field weakening current even when you are off the throttle to keep the motors generated voltage below battery level so no regen needs to happen unless requested.

I believe, if the controller would continue to supply field weakening current when we are off the throttle, those IPM motors would continue to spin due to the reluctance torque. This could turn out to the same drama as if your throttle would get stuck at half.

Or imagine what would happen if you ride down a steep hill without any throttle, and motor BEMF starts to become higher than V batt? Would the controller then start to waste current into the motor for doing nothing?
 
If it lost control under field weakening, not such a big deal as long as the battery fuse does not blow. If you look at my field weakening explanation, the extra voltage c9mes from the field weakening current combined with (motor) inductance. Should the controller lose control, the max current the motor will supply to the battery will be the field weakening current the controller was supplying. Not a problem, except for when the battery fuse blows, as at that point in time the controllers FETs will see the high motor voltage...
 
I believe, if the controller would continue to supply field weakening current when we are off the throttle, those IPM motors would continue to spin due to the reluctance torque. This could turn out to the same drama as if your throttle would get stuck at half.
"field weakening current", or more accurately, Id (flux-creating current) doesn't directly represent torque.

The torque equation for IPM is this one. Note that if you zero Iq (torque-producing current), Torque goes to zero.
IPM torque.png

Should the controller lose control, the max current the motor will supply to the battery will be the field weakening current the controller was supplying. Not a problem, except for when the battery fuse blows, as at that point in time the controllers FETs will see the high motor voltage...

Imagine you are coasting at 180kmh in a car, 170% base speed. If the FW collapses now your motor is generating a voltage 170% of the base voltage.
If you are running a 120Vdc battery, at those rpm your motor can generate 204V. The only thing between motor and battery is the fuse, contactor and the body diodes.
So you have a short from 204V to 120V and its energy is fueled by the car inertia. You can't turn off that current because if lows through body diodes. In a super low inertia scenario the current spike will be brief, but in a car, the energy to stop the car from 70% over base speed to base speed has to be dissipated somehow, somewhere.
I heard about shorting phase A,B and C together, that way the energy to slow down the car is dissipated in the motor and MOSFET/IGBT die. That sounds scary, its like pulling the handbrake at 180kmh (and releasing it is not up to you). At least it prevents the silicon to blow with the overvoltage.
A safe(r) aproach should be using a powerstage capable of dealing with your overspeed voltage (to prevent mosfet top and bottom to shoot-through), and a contactor capable of breaking that amount of current. I wonder how this safety issue in handled in automotive applications.

In a bike these could be minor concerns, I dunno. Light FW will at most produce some extra, uncontrolled current flowing to the battery that will last as long as the rider is above base speed, probably not too long. If its extreme the fuse will blow and thats it.
 
Lebowski said:
If it lost control under field weakening, not such a big deal as long as the battery fuse does not blow.
And as long as the battery doesn't see an overvoltage/overcurrent for too long. If it did it would either damage the battery or trip the BMS - and once again you'd seen an open circuit and very high voltages.
 
@maddin: regarding the 30% id. This is just a rule of thumb from other IPMs i have handled so far. This lead to around 17° of current advance which is quite near to the MTPA operating point at high load. It's not ideal but a current 30/70 should give us more torque that 0/100 in an IPM like that.

Handling FW at more than base speed was one of the points why Tesla used Asynchronous motors, just my assumption.

The handling in a industrial IPM is done the following if rpm is bigger than base speed apply id from the table for MTPA. The iq could be 0 in that operating point.

As Marcos stated the torque equation says iq=0 torque is 0.

For safety converters have 1200V IGBTs or SiC Fets and they have large braking resistor to regulate the DC Voltage. But these are not battery supplies they are grid connected. And the PFC can handle 1200V as well. So my bad battery supply makes this more complicated.
 
Does any body know the base speed and Max speed of a leaf or Prius? How what ratios of base to top speed are IPM driven cars use? I would assume most of them have higher voltage parts as needed for base speed. Does any body know the voltage rating of the leaf inverter?
 
Inductance measured on the revolution (with ebay meter, don't know how hard it is to measure inductance accurately) for QS3000W motor:
Max: 0.078 mH
Min: 0.058 mH

Can any estimations of the reluctance torque potential be done based on this?
 
larsb said:
Inductance measured on the revolution (with ebay meter, don't know how hard it is to measure inductance accurately) for QS3000W motor:
Max: 78 mH
Min: 58 mH

Can any estimations of the reluctance torque potential be done based on this?

Check my equations (and I mean check for mistakes :wink: ) a few posts above. From the equation for Power you can get the torque (keep in mind you have 3 windings, not 1)
 
That's close to nothing..
Given P=Torque*(rotation speed) then you can divide and remove the omega from your formula and max torque would be
T=L*I*I/4

Amplitude of the phase to phase inductance difference L1 is 0.010*1e-3 (which really kills it)
Current at 200A (assumption)

Which gives T=0.1Nm

That shows that L1 inductance value must be a LOT larger to get any decent torque. I think the phase to phase inductance difference i measured cannot be correct to use here?
To get 20Nm at 200A you'd need 2mH instead of 0.010 mH difference in d-axis and q-axis inductance.

Can someone check it?
 
I've experienced hundreds of controller faults from various types of field weakening controllers while deep into field weakening and never had that damage anything. This seems odd because the BEMF if you spun the motor up with a drill to that same RPM you faulted at would be above Vdsmax for the mosfet. In practice I've never seen a controller die from it despite so many faults happening when doing motor controller tuning runs for the purposes of maximizing top end.

I may be missing something Lebowski, but I think you might be able to get another big chunk of low end torque out of the LEAF motor with using a reluctance table with an experimentally derived deviation from dVI relationships that has an offset to leverage pulling on the motor iron sooner.
 
Back
Top