0
\$\begingroup\$

I will have external torques acting upon an axle to which a motor is connected. I want my motor control system to fairly precisely compensate for most of the torque that the motor imposes on the axle itself due to things like motor frictional losses, induced eddy currents, etc.

In other words I'd like it to appear from outside as if the motor doesn't exist. If one were to spin the axle with their hand the axle should continue spinning at that rate until externally stopped, and you should have to do no more work to stop it than you would if the motor weren't present (ideally).

My motor will need to operate around 10W and -200 to 200RPM. Ultimately this system will be used to modify the stiffness of a mechanical torsional spring without adding damping.

What considerations do I have for motor type selection/design if I want to:

  1. mimimize the needed loss compensation torque
  2. simplify the control system

?

I assume brushless non-geared motors are superior from a friction perspective, however I understand permanent magnet motors have significant eddy current losses and also suffer from "cogging" which may complicate control.

Additionally:

  1. What methods exist for predicting/estimating the magnitude of these torques from commonly available motor specs?
  2. Which motor losses are typically constant and which vary with temperature / velocity / other factors?
\$\endgroup\$
9
  • \$\begingroup\$ Why not just detect and maintain the speed with an encoder? If you want to maintain the speed, measure and maintain the speed, not something else. Why all the focus on extraneous and difficult to measure phenomena? \$\endgroup\$
    – DKNguyen
    Jan 22, 2020 at 4:22
  • \$\begingroup\$ I don't think you will be able to control the motor as you want it, since "If one were to spin the axle with their hand the axle should continue spinning at that rate until externally stopped" is kind of a contradiction, unless you measure really well whatever torque you want to cancel, the system wouldn't be able to differ when you intentionally change the axle speed or when friction and what else does it. \$\endgroup\$
    – jDAQ
    Jan 22, 2020 at 4:26
  • \$\begingroup\$ Try thinking about what the control system should track, is it certain speeds, torques, or angular displacements? And, what types of sensor/sensing are you willing to have? Speed? Absolute encoders? \$\endgroup\$
    – jDAQ
    Jan 22, 2020 at 4:31
  • 1
    \$\begingroup\$ @jDAQ I would limit how hard my system fights to correct the error and track the rate of change of that parameter and if the rate of change is high enough and/or a sufficiently large enough pertubation lasts long enough, to grab it and set it as my new target value. Or just to maintain the current value if the error exceeds a limit. \$\endgroup\$
    – DKNguyen
    Jan 22, 2020 at 4:39
  • \$\begingroup\$ @jDAQ Assuming there is a constant loss torque X, the ctrl system could apply that torque in the direction of rotation as soon as it detects rotation. If the loss torque varies with speed, then the ctrl system would need to measure speed and vary the comprehensation torque based on it. I might be missing something or oversimplifying... \$\endgroup\$
    – davegravy
    Jan 22, 2020 at 4:48

2 Answers 2

2
\$\begingroup\$

The simplest solution is to use a coreless or 'ironless' motor. With good bearings the drag at low speed should be be very low. Brushed coreless motors also have the advantage of very low inertia.

As an example of how free a coreless motor can be, the ironless brushless motor below can be used as a compass! The rotor spins so freely that - with no iron to cause cogging or eddy current drag - when unpowered it aligns itself to the Earth's magnetic field.

enter image description here

The downside of coreless motors is higher resistance for the same Kv, since there is less or no iron to concentrate the magnetic force so more turns of wire are required. This means you may have to use a larger motor to get the power (torque x rpm) that you need.

Unfortunately high power low rpm brushed coreless motors are hard to come by these days. Ironless BLDC motors are more common, or you could try building one yourself.

Ultimately this system will be used to modify the stiffness of a mechanical torsional spring without adding damping.

Motors are designed to rotate continuously, whereas springs have limited travel. For this application you might consider a different type of control such as a 'voice coil' actuator, or perhaps use the motor to drive some mechanism which adjusts the mechanical stiffness of the spring.

\$\endgroup\$
0
\$\begingroup\$

most of the torque that the motor imposes on the axle itself due to things like motor frictional losses, induced eddy currents, etc.

Motor frictional losses and aerodynamic drag (windage) are the only losses that the motor imposes on the motor shaft (axle). All of the other losses convert input electrical power to heat, subtracting from the percent of electrical power that is converted to mechanical power.

What methods exist for predicting/estimating the magnitude of these torques from commonly available motor specs?

Since the losses in question are mechanical losses, that is a mechanical engineering question. Determining bearing friction starts with examining bearing specifications during the course of selecting the bearings. The coefficients of friction of the bearings, the rotor mass and any radial and thrust forces exerted on the bearings by the load are the factors that determine the friction torque.

The aerodynamic drag is determined by the surface texture of the moving parts and the design of any rotor fins or fan that are attached to the rotor to move cooling air. If cooling air is mechanically moved through or over the outer surface of the motor, the design of the airflow pate needs to be examined.

Losses can be determined by testing the motor using a driving dynamometer. Another method is to let the motor coast to a rest from full speed while accurately recording the motor coast-down speed vs. time. The friction plus drag torque can be calculated if the rotor plus stator inertia is known. The inertia can be determined by torsion pendulum or other tests.

Which motor losses are typically constant and which vary with temperature / velocity / other factors?

Friction torque is relatively constant with respect to rotor speed except for the initial "stiction" torque at the transition from standstill to motion. Power loss is therefore directly proportional to speed.

Aerodynamic drag torque is approximately proportional to the square of rotor speed. That makes the power loss approximately proportional to speed squared. For cooling air flow, the back pressure needs to be considered.

The effect of temperature on air density and coefficient of friction must be considered.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.