How to control an ac motor with sinusoidal torque output

[Cdz]

Hello, guys.
In field oriented control of ac machines, the electromagnetic torque is proportional to the q-axis current. We can control instantaneous torque by control the q-axis current. If the torque we want is a constant value,with PI controller, no problem, no steady error. But if we want to have sinusoidal torque, with PI controller, there would be a steady error.

What can we do about that? Any other method?

Thank you!

 

[anorlunda]

q-axis?

I'm thinking phase locked loop (PLL). But first, define the problem better. Conservation of energy says that the mechanical power of the generator and load must balance in the steady state regardless of anything electrical. If they are synchronous, then mechanical torques must also balance. Are you controlling an isolated system or a generator/motor connected to the grid? Perhaps the load is just a resistor.

What is the origin of the sinusoidal variations, the AC machine or the external load?

 

[jim hardy]

We can control instantaneous torque by control the q-axis current. If the torque we want is a constant value,with PI controller, no problem, no steady error. But if we want to have sinusoidal torque, with PI controller, there would be a steady error.

Oh ?
Error is difference between measured and desired in whatever variable it is you're controlling..
Do you measure torque directly, or rotor position, or rotor velocity?

 

[Cdz]

Oh ?
Error is difference between measured and desired in whatever variable it is you're controlling..
Do you measure torque directly, or rotor position, or rotor velocity?

No, I haven't measured that. I did a simulation with simulink..though. I will provide a better description of the question later :)

 

[Cdz]

q-axis?

Control of AC machine in synchronous reference frame, the instantaneous torque is proportional to q-axis current. So we need to control q-axis current if we need to control torque.

What is the origin of the sinusoidal variations, the AC machine or the external load?

We need the machine torque to be sinusoidal.

The motor is connected to lead screw, which converts the motor rotation into translational movement. This lead screw drives a spring-mass like system, with a sinusoidal torque from the motor, the lead screw would apply a sinusoidal force on the mass-spring system. What we want to achieve is resonance of the mass-spring system. And that means we need to apply a sinusoidal force to the spring-mass system with a frequency of the system's natural frequency.

That being said, we need to control the motor's torque to make it sinusoidal. We tried doing that with a sinusoidal q-axis current reference with a PI controller. But according to classic control theory, a PI controller with a sinusoidal reference means steady error.

 

[NascentOxygen]

This lead screw drives a spring-mass like system, with a sinusoidal torque from the motor, the lead screw would apply a sinusoidal force on the mass-spring system. What we want to achieve is resonance of the mass-spring system. And that means we need to apply a sinusoidal force to the spring-mass system with a frequency of the system's natural frequency.

Resonance is a property of the mechanical system, it is exhibited regardless of the waveshape of the exciting signal. If the system is relatively low-loss, all you need to demonstrate the system's high-amplitude sinusoidal swing at resonance is a repetitive pulse input of precisely matching period though of relatively low amplitude. In general, the task definitely should not need that the input be sinusoidal.

So, despite this, are you sure you do want a sinusoidal input?

 

[anorlunda]

Your question still makes no sense. You have an AC synchronous motor with external DC field. Therefore, the motor is forced to rotate at synchronous speed.

The motor is connected to lead screw, which converts the motor rotation into translational movement.

There is a certain torque required to spin that lead screw at the synchronous speed, and that is the torque that the motor must provide. Field voltage, or q axis current can not change that. If you were turning the motor at synchronous speed by hand rather than by electric power, the same torque is required.

Therefore if I interpret your descriptions right, there is no way to influence torque using field. There is no way to control q axis current directly, and even if you did it too can not influence the torque. Your question and descriptions make no sense.

 

[jim hardy]

The motor is connected to lead screw, which converts the motor rotation into translational movement. This lead screw drives a spring-mass like system, with a sinusoidal torque from the motor, the lead screw would apply a sinusoidal force on the mass-spring system. What we want to achieve is resonance of the mass-spring system. And that means we need to apply a sinusoidal force to the spring-mass system with a frequency of the system's natural frequency.

hmmm. A problem well stated is half solved.
I assume you're using field oriented control of an induction motor?
How much throw do you need ? A leadscrew garage door opener could give eight feet if you're not in a hurry.
Leadscrew will completely define position of its follower nut.
Torque from motor will divide between accelerating its rotating parts, overcoming friction, and applying force to the follower nut.

Is leadscrew follower nut connected to the mass or to the spring? That's two very different systems.

I'd look into something made to produce force , like

https://www.h2wtech.com/category/brushless-linear#productInfo1

http://www.beikimco.com/motor-products/VCA-linear-voice-coil-actuator-all

http://acm-magnetics.com/acm/pdfs/15-BasicsOfVCMs.pdf

And i don't understand the statement about steady error.. Phase error ? That's what "lead" (s in numerator) is for .

 

[Baluncore]

What we want to achieve is resonance of the mass-spring system. And that means we need to apply a sinusoidal force to the spring-mass system with a frequency of the system's natural frequency.

To help the spring-mass system oscillate requires that you know the ball nut velocity with direction. How do you get that information ?

To oscillate at the natural frequency you need to apply a torque that is proportional to the velocity of the ball nut. That will overcome the system losses and reinforce the natural frequency.

Once the spring-mass system is oscillating, what will limit the amplitude of the oscillation ?