3
\$\begingroup\$

This question is not to solve a particular problem but rather one of principle to understand the main reason a designer would want to include speed feedback in a motor controlled positioning system. And I'm hoping a veteran motor control expert is able to help answer it.

Consider a cascade controller with displacement being the most outer loop (encoser feedback), followed by a velocity controller (encoder or tach feedback) and the most inner loop a current feedback loop (using shunt resistors, hall devices etc.)

It's clear that current is directly related to torque, and so that helps to overcome torque disturbance as well as speeding the effective electrical time constant. It helps to linearize the transductance of current (command) to torque.

But what does the velocity (speed) loop do to improve performance? Does it further provide linearization? As far as I know there isn't a 'velocity' disturbance to be concerned about.

Typically if you pick the P, I and D gains of the displacement controller in the right proportions (tuning) then you can get by without a speed loop between the position and current controllers.

Cascade control promises tighter control, there are more knobs to turn, but what is the real advantage of having speed in the cascade?

\$\endgroup\$
  • \$\begingroup\$ Velocity feedback is viscous damping. \$\endgroup\$ – Chu Mar 7 '18 at 23:51
2
\$\begingroup\$

Feeding back speed is the same as feeding back the rate of change of distance against time so your question really boils down to why you might use a proportional feedback signal as well as a differential signal.

Simple answer is that when used properly, the speed signal fed back can reduce the rate at which the system homes in on the target position and somewhat reduces the effect of overshoot and hunting. In other words, it’s a stabilising factor in the presence of system inertia.

Maybe do some research on three term controllers or PID controllers. P stands for proportional, I for integral and D for differential. Differential is the same effect as a speed signal in a position control system.


why waste the expense of a speed sensor, extra circuitry when you can just increase the P gain in the position loop PID controller to accomplish the same

Try this out: -

enter image description here

Picture License information

\$\endgroup\$
  • 1
    \$\begingroup\$ Right - speed feedback and its loop gain can provide 'damping' in the position control. But why waste the expense of a speed sensor, extra circuitry when you can just increase the P gain in the position loop PID controller to accomplish the same - an increase in damping? \$\endgroup\$ – docscience Mar 7 '18 at 22:57
  • \$\begingroup\$ @docscience please give a specific example where velocity feedback is implemented if you want a more detailed analysis. \$\endgroup\$ – Andy aka Mar 8 '18 at 0:18
  • 1
    \$\begingroup\$ @docscience - "you can just increase the P gain in the position loop PID controller to accomplish the same" - sorry, but that's just wrong. The differential of a sine is essentially a phase-shifted version of the original, so judicious use of a D term allows control of phase shift - and a classic example of an oscillator is an amplifier with more than unity gain and a phase shift of 180 degrees. So stability cannot (in general) be guaranteed without consideration of phase, and this translates to use of the D term in a system. \$\endgroup\$ – WhatRoughBeast Mar 8 '18 at 0:51
  • \$\begingroup\$ @docscience You rarely have a dedicated speed sensor - generally the speed is just calculated from the position sensor (encoder). The velocity loop doesn't add hardware, only software, in these cases. \$\endgroup\$ – Selvek Mar 8 '18 at 15:22
  • 1
    \$\begingroup\$ @docscience you are no closer to having ACTUAL speed measurement with a tacho generator (for instance) than you are by differentiating the signal. If the aim of the game is accurate positioning then the position sensor MUST be absolutely reliable for containing all the relevant information regards position and speed. A tacho-generator will vary its output with temperature and time. You can get accurate ones of course but these cost. An op-amp differentiator (or bit of code) costs virtually nothing in comparison and inevitably gives a better result. \$\endgroup\$ – Andy aka Mar 8 '18 at 15:49
1
\$\begingroup\$

Many position controllers use trajectories for point to point moves. This gives them better control of speed, acceleration, and "jerk" (rate of change of acceleration).

How this process works is that you generate a "desired speed vs time" curve, based on how far you want to move, as well as what max speed and acceleration you want.

Then you operate the motor in a closed loop velocity mode, where the actual velocity is controlled to match the profile velocity. A weaker position loop is wrapped around the velocity loop to adjust the desired velocity.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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