FOC for power and SOC (Stator Oriented Control) for regen?

[John in CR]

After seeing Leboski's post that stator oriented control was more efficient for regen braking, I started wondering how much difference it would make. Then on my latest build on the first test ride only one controller was running the 6 phase motor. While everything was fine running nice and smooth under power albeit at half the torque, when I engaged regen having only 12 stator teeth working on the 10 pole pair motor it was rough and noisy during regen braking. It was very similar to the plug braking of a blown controller with 1 phase having shorted mosfets.

With 2 controllers running regen is smooth and silent, except maybe a hint of noise and roughness at the end of the final rotation before coming to a full stop using regen. The huge difference during forward power and regen braking with only one controller tells me that the difference of SOC during braking could be big, so my questions are:
1. How much more efficient could regen energy recovery be...both from a range extension and a heat generation standpoint?
2. With a controller that has FOC control, how hard would it be to implement SOC for braking?

 

[kenkad]

I assume you are running 2 controllers so each has 3 phases. What is the motor you are using for this,

 

[John in CR]

I assume you are running 2 controllers so each has 3 phases. What is the motor you are using for this,

The same motors I've been running for 8 years...HubMonster and MidMonster, very high efficiency hubmotors that have been out of production for about 6 years. Both are 20 magnet 24 slot motors wound as 2 totally separate 3 phase motors electrically wound on alternating teeth.

They run smooth and silent even with cheapie square wave controllers, so this regen issue noted running only one FOC controller clearly demonstrates that FOC is not a very good way to do regen just as stated by Lebowski. Also note that Justin measured noticeably more heat in the controller and motor for the same torque during regen compared to while powering in forward.

 

[Lebowski]

Simply put, for highest efficiency cos(phi) has to be 1 for the device receiving the power. So during accelleration this means FOC (as the motor receives the power) and during regen it has to be SOC (as the battery receives the power).

The (now open source) Lebowski controller has FOC for powering and SOC for regen...

 

[John in CR]

Simply put, for highest efficiency cos(phi) has to be 1 for the device receiving the power. So during accelleration this means FOC (as the motor receives the power) and during regen it has to be SOC (as the battery receives the power).

The (now open source) Lebowski controller has FOC for powering and SOC for regen...

I understand why it's needed. I just didn't realize how much difference it could make. I assume there still isn't an A-B-C kit for an electronics caveman like me to assemble reliable controllers with your chip, and that no one is building them up for sale.

Any hints for how to get a teen interested? I've got 2 that aren't even into riding their ebikes that we built together, it baffles me.

 

[Lebowski]

Maybe you can convince Justin to talk yo ASI to change the regen mode ?

 

[Steph]

Hi Lebowski,

I went through the presentation you made for your controller but I haven't played with complex electrical representations for like 20 years and I'm struggling with it.
So I tried to come up with a more empirical representation of the situation based on the principles you shared instead.
Could you help confirm / correct my model below?

At a high level, I consider that the motor converts electric power to (driving) or from (regen) magnetic-movement (or whatever it should be called) power.

If I understood correctly the theory you presented for your controller (and I assume this is what your call Stator Oriented Control), the core principle is to align the I vector with the U vector (U ^ I is maximal, which should relate to cos(phi) = 1 in your previous comment).
The I vector is sensed at the motor wires.
The U vector is applied at the same wires (what we drive with the half bridges).
That's for the electric part of maximizing the power at "one end of the motor".

Now, at the "other end" we have magnetic power, which is maximized when the rotor field and the stator fields are in synchronicity through the rotor movement (it should be something about Lorentz force but honestly, I haven't figured the model, I'm rather relying on resonant frequency analogy such as pushing a swing: you need to push just at the right time so the swing goes effortlessly).

In between these two, the power transfer is operated by the coils, and their intrinsic inductance introduces a delay (T = L/R) between the current sensing and the associated field.
The alignment between U(t) and I(t) must be targeted at the magnetic field generation in the coil.

When driving, since the current goes from battery to motor, the sensing is done ahead of alignment point and the electric power maximization target is for I(t).U(t-T)
When regenerating, the current goes from motor to battery, the sensing is done after alignment point and the electric power maximization target is for I(t).U(t+T)

And then... efficient coasting should be for I(t).U(t).


Further on empiric interpretation...
To regen, we first close the loop of the coil to accumulate energy in it, then open it to release it to the battery.
By doing this with a driving configuration (-T instead of +T), we actually open the loop when the energy is not fully present in the coil, so we recover only a partial amount of it in the battery, and then we close the loop when the energy should be at its peak (but it's no longer the case since we already let some of it go), so it goes slamming into the circuitry and the RdsOn of the mosfets, which would explain the slight heating that has been mentioned.

Cheers,

Stéphane