We need to select the best command for a servo drive for CNC application. The options are: speed or torque or pulse-direction command. In this application we are going to use motion controller that can generate the above commands and we wonder which of them will generate the best performance in terms of speed ripple? What factors do we have to consider when selecting the control method?
What is the common solution at CNC application and why?

Thanks

 

This depends far more on your controller and amplifier than the command method. Velocity commands were common before controllers got fast enough to command torque. The problem with velocity command is that it is load dependant and you have to tune both your amplifier and your controller for the load. This makes doing things like real time gain modification very difficult, if not impossible. These days I use velocity command only with amplifiers which have very slow update times on their analog input (slower than 1 kHz). Torque command is the most common method used today, and is good for all but the most demanding applications. For reliable toque command you should have an absolute minimum of a 1 kHz servo loop. I prefer to run at 4.4 kHz, or even 9 kHz. The most
demanding applications use Pulse Width Modulation, or PWM. Technically, they all do, as almost all modern amplifiers take the torque command and convert it to PWM. But when your controller uses PWM, your amp is a dumb one, and this removes one layer of control from your system, reducing latencies. I've used this to control sub picometer placement for semiconductor applications.

Davis Gentry
Senior Applications Engineer
Delta Tau Data Systems
[email protected]

 

On the surface of it, the Speed option would appear to give the least speed ripple. However when you get down to specific implementations, it depends on the details of the implementations. A speed command is ultimately translatted into a torque command in a servo driver. Calling a controller a "speed" controller vs. a "torque" controller is just a matter of moving the artifical dividing line between the controller and the drive. All the pieces need to be there, regardless of where those pieces physically reside. As for step and direction commands, I assume you would be using a stepper controller, perhaps connected to an interface that simulates a stepper motor, but is actually a servo motor. In that case the speed ripple is probably worse because information is lost when you force the control loop to go through a stage that gives discrete pulses for each step. However, if the "steps" are small enough, this ripple may be so small as to be unmeasureable.

Robert Scott
Real-Time Specialties
Embedded Systems Consulting