Velocity vs Torque

I'm fooling around with servos again, and I thought I'd put this one to the group.

Velocity command is +/- 10 VDC, and produces proportionate speed in the direction indicated by polarity.

Torque command does the same as velocity, with respect to torque.

Now the question - why would one be better than the other for certain applications?

My theory is, with velocity, we want a controlled speed and don't care how much, or how little, torque it takes to produce it. With Torque control, we want a specific amount of rotary force, and don't care how fast or slow we move to get it.

Am I right? Wrong? Waaaaay wrong?

TM

Peter Nachtwey may have a paper on servo control on his company's website, and if he chooses to reply, he'll be able to post a much more articulate answer, but consider this. If you only have control of the torque, that is, you can maintain the torque developed by the motor at a constant level, then any torque in excess of that required to overcome friction will be available to accelerate the load. If that is the case, the motor speed will increase without limit, with potentially dangerous consequences.

It depends.

Velocity control is much easier if you use a velocity command. Velocity commands can be generated manually by a joystick or from a PLC as well as a montion controller. However this requires the drive to do the speed regulations which I find is average to very poor. How many people tune their drives? How many can do it well? Autotuning works only when there is an ideal system. When running in current or torque mode the motion controller must do the speed regulation and the output of the controller to the drive is proportional to the torque. This means the motion controller must have just a little more code to handle the speed requulation and it can't be done with a joystick or with a PLC. The upside is that the tuning is done in the motion controller. This also means that the drive unit should be 'dumb' ( cost less ) so it doesn't interfere with the motion controller.

I went to one of those Tom Bullock and Jack Grenard servo conferences once and the manufacturers of the motion controllers agreed that the motion controllers does a better job of doing the speed regulation than the drives do. One of the drive manufacturers even told Jack to stop the recorder for 15 minutes while he admitted that he though the future belonged to "dumb" drives and and smart motion controllers because the control was better when done completely by the motion controller. This guy didn't want to say this on the record. This is a different issue now that the motion controllers and drives are often one.

Back to torque vs velocity. For high performance jobs, ( small DC motors with high acceleration rates using PWM ) I would prefer torque mode because the motion controller knows what is going to happen next and can anticipate with feed forwards. All the small motors used for testing here use torque mode. We do have an AC servo that uses velocity mode. In velocity mode the the drive ( inner loop ) can only respond to what is happening.

For bigger jobs flux vector drives use +/- 10 volts is still the norm. Our controller still needs the drive to do all the vector calculations. However, there are motion controller ( not ours ) that can do the flux vector commutation directly and still only require dump amplifiers.

So what are you trying to do?

I too will be interested in more informed responses than my own on this, but let me throw this out:

When you have two independant controllers in the system (A CNC processor talking to a Servo Amplifier), you will typically want to set up the CNC Processor to use speed control and then set up the amplifier to use torque control. This allows the processor to send a speed oriented control signal to the drive who, in turn, is allowed to do whatever is required to get the job done.

While it certainly is possible to set up each of the two systems in speed mode, it tends to be much smoother using torque mode on the amplifier.

Steve

Good point.

Steve Bailey said:

If that is the case, the motor speed will increase without limit, with potentially dangerous consequences.

Click to expand...

Our controller is used in a lot of hydraulic applications where open loop is often used. Open loop still works in velocity mode pretty much as it does on a hydraulic system. However, as Steve points out, a constant openloop output will cause the system to acclerate until it crashes ( normally ) or spins faster until maximum rate speed is reach where acceleration and frictional torques are equal and there is no more net torque to acclerate faster.

Limiting the torque is often a good idea when doing setup. It minimized potential damage.

In some applications where a object must be grabbed but not squished, torque limiting will limit the grabbing force. The problem with this is that the armature is getting warm ( hot ) while holding torque is applied and energy is being used. ( side note hydraulics don't consume energy will grabbing or pressing. )

One more thing since this was brought up. A while back there was a thread about measuring current to determine torque. I was trying to get some one to see the difference between measureing torque externally ( not fast enough ) and reading the torque from the controller or drive. The controller or drive knows what the torque is at all times because it has a model that predicts these values. This model is constantly being updated with measured values of voltage current and the phase relationship between them. Because the motor controller knows the phase relationships, it can take one reading and scale it to find the peak current, RMS current or torque. This can't be done externally without averaging over a few cycles because externally one doesn't know the phase relationships.
Dick correctly pointed out that this data is best retrieved from the drive or controller. Howerver, this data is obtained with a LOT more work than just measuring the current.