FOC for asynchronous motors

[johu]

As it stands asynchronous motors can only be controlled by the stm32_sine software, not by FOC. The reason being that scalar control works really well and is proven over 10 years (the good is the enemy of the better). The motors run at high power output and high efficiency.

Therefor, despite being offered money, I have not devoted any time towards the "FOC for asynchronous motors" topic. I also do not have a good testing platform as my ACIM conversion only has single channel shaft feedback.

This thread is supposed to start the discussion whether we need this feature, what problems of the scalar implementation could be overcome and what new problems might be introduced. And last but not least if anyone besides me can do this.

[yaroslav]

Long ago I began development of its controller, because I did not know many nuances, have developed the firmware in conjunction with a programmer at all on atmege328 in Arduino hardware, the development was used in a child's machine and the speed I got about 20km per hour on the engine 740вт achronim, firmware was the same scalar, but I came up with automatic torque adjustment on the current shunt, it turned out, the algorithm is a bit similar to vector, but still in the future, I could voimat momentum (autopodium) without using the encoder, the slide there was considered to be indirectly, but I liked the good acceleration dynamics in this firmware, if you remove the current correction on the shunt, the dynamics fell, the power consumption depended on the load, I understand that doing so is not the best way but it worked, and automatically adjusted the torque. But when the load increased, the speed fell and then I came up with the idea of regulating the slip by current, this gave its result and the speed began to stay stable under load. If you could connect an encoder to it, then it would be easier, but unfortunately it lacked timers. When you press the gas pedal, even with a small effort, the current was measured and the program adjusted the PWM filling and this regulated the torque.

[yaroslav]

If we compare it with scalar control, then vector control not only adjusts the phase current , but also the slip itself, depending on the load, this increases the efficiency of the engine, reduces the consumption of the battery in some modes.


This was the entire controller, but it is still working, although I have already reworked entire firmware in it

[arber333]

As it stands asynchronous motors can only be controlled by the stm32_sine software, not by FOC. The reason being that scalar control works really well and is proven over 10 years (the good is the enemy of the better). The motors run at high power output and high efficiency.

I agree J. Slip control seems to be very good and robust control. I dont want to change that,
1. I would however like us (as in community plural) to develop some way of filtering the throttle and regen in the area close to throttle 0%. There i still get very high pulses on regen and i suspect in traction also. But latter is dampened by vehicle mass (2nd order), where with regen the mass is the instigator and i see amplitude rises quickly to uncomfortable levels. So for clutch to survive i limit ammount of regen to 30% AND i use only HALFH of this while driving. So i get maximum of 8kW on the wheels from regen. Anything more it quickly becomes undrivable and can damage the coupler.

2. I would like to see a resolver option for slip control. There are many ACIM motors with resolver out there. It is a robust sensor and i am surprised how easy it is to implement with your inverter.
If you need a practical application to test i can provide an end shaft cap with magnet receptacle and a mount for RLS encoder (with encoder/resolver of course) for your Polo ACIM motor. This resolver PCB is the same as the one i gave you for ABZ encoder. I just need actual dimensions.
Interested?

tnx

A

[johu]

If we compare it with scalar control, then vector control not only adjusts the phase current , but also the slip itself, depending on the load, this increases the efficiency of the engine, reduces the consumption of the battery in some modes.

"Scalar Control" in our context does not mean "Just throw some frequency at the motor" but actually amplitude and slip are controlled precisely and in closed loop with encoder. Current sensors are not in the control loop. So I don't expect increased efficiency by implementing FOC.

2. I would like to see a resolver option for slip control. There are many ACIM motors with resolver out there. It is a robust sensor and i am surprised how easy it is to implement with your inverter.

Next sine firmware will support all feedback types for async motors and in turn will not support sync motors (it doesn't work well anyway).

If you need a practical application to test i can provide an end shaft cap with magnet receptacle and a mount for RLS encoder (with encoder/resolver of course) for your Polo ACIM motor. This resolver PCB is the same as the one i gave you for ABZ encoder. I just need actual dimensions.
Interested?

Not to keen on ripping out the inverter and fitting it with a newer brain board (it still used Rev1) and I might also sell Polo to free up space on the drive way. So thanks, but no thanks

[Jack Bauer]

I know very little about motor control theory. What I do know is that I can stomp posers in their shiny new BMWs all day in Der Panzer thanks to Johannes. I expect the comments on youtube to be hilarious when I start in on the model 3 drive units.

[yaroslav]

Vector control would expand the adjustment range, I think that then it would be possible to control the engine with combined windings, it has an extended range of control, vector control should solve the problem with torque, I can adjust the DA-90S engine to either very fast dynamics, but not expense of economy, or economy mode, but this reduces the dynamics, although you may need to add a couple of dependencies to the settings

[Jack Bauer]

So if the community want this then there are three choices. 1) Club together and pay Johannes , 2)Club together and pay someone else , 3)Do it yourselves.

[yaroslav]

Then I have such a question can someone else besides me from the community would like to test the work of vector control? I would be satisfied with 1 option, or modify the scalar control for a specific motor with a combined winding?

[arber333]

Hi Johannes

I must commend you...
Today I tried the latest Sine firmware 4.76R and lo and behold! I have no more regen shaking and surging. This is so smooth i am thrilled.
I noticed at low end RPM there is still some shaking but only in 2nd gear which i think has some gear rattle. So it is a mechanical in nature not software.

I really only wish for resolver code in ACIM code. Can this be done?

tnx

[slow67]

I really only wish for resolver code in ACIM code. Can this be done!

If you look in Johannes’ post above, he said the next sine firmware release will include all feedback types (Im assuming this means resolver and encoders)

[arber333]

If you look in Johannes’ post above, he said the next sine firmware release will include all feedback types (Im assuming this means resolver and encoders)

True, i wanted to say i have everything i need for ACIM but that...

[yaroslav]

Hi, Johannes. I got my hands on a 3 kW vector controller, decided to spin it, the acceleration dynamics of my go-kart increased, but the current from the battery also pulls higher. At the start, the torque is higher, I decided to find out why, I began to clamp the current on the phase and saw how I pressed the pedal, not only the phase current increases, but also the frequency of sliding when the engine has a jammed shaft on the torque wrench. How do you think it is possible to make such a function in our inverter?

[ChazFisher]

A number of years ago, I was the project engineer for a very large inverter/motor drive for a government program. We were powering a linear induction motor, and needed to apply a lot of force at the start of the motor length, with essentially no movement of the runner. We also had to stop that runner plate at the far end, in a very short distance. So FOC was the way to go. I'm not surprised that FOC generates more torque from a standstill in an EV application. The other advantage I see is that when we go into regeneration, Id remains constant and we just flip the sign on Iq. And let the transforms work out the correct slip. (We did have to program in a smooth transition from positive to negative Iq, but we were able to do it VERY quickly.)

I'm still really slogging my way thru just keeping my little e-motorcycle project moving. Apparently my brain takes longer to re-learn what it's forgotten than it took to learn it the first time. But if I get some success with it, I think my project would benefit from an FOC-based asynchronous motor controller. So I'll consider taking it on in the future.