Decades of effort have gone into methods of improving servomotor control. There are many contributing factors that result in better products. Better feedback devices give the controller chips that manage the voltage and current to the motor winding a higher resolution basis for the management of power between the amplifier and motor. These are critical relationships that form the basis for the torque and speed loop management. The resulting ability of the control system to manage the higher level functions like position and coordination of axes is built on the fundamentals.
A major contribution comes from the recent generation of processor technology for motor control. Since the era of the DSP (digital signal processor), a number of newer processors have made it into the marketplace that provide computation speeds that 50 to 100 times faster at lower costs. The new generation processor are generally multi-threading so they can manage messages into and out of the controller without impacting the motor control functions. This gives rise to more economical networking of drives and the ability to do coordinated motion over a network, like EtherCat and IEEE 1588, without any latency problems. At the same time, network speeds have increased logrithmically from 1 gigabaud to 100 gigabaud. Latency issues are largely a thing of the past making the system architecture relatively transparent.
Advanced mathematical tools are being used more routinely to improve motor control by identifying natural frequencies of the combined load of the motor and the driven components. These tools, filters and Bode plots, provide the control system with additional information to avoid specific speed ranges so that the combined load and motor/drive system do not become unstable.
But in the true mechatronics world, where the “system” objectives are to regulate a defined load behavior, the “real” control needs are still somewhat masked by the control system PID style programming tools. If the load behavior is to draw a circle, we can model the control system requirements as voltage and torque commands and achieve a certain quality of result at a target speed. As we increase the speed requirement, the control system becomes unable to manage the mechanics. The rules of the actuators are not part of the control system’s programming tools.
At least, not until recently. The folks at Agile Planet (check out their website) have developed a next generation controller that is able to apply the mechanical properties of mass, inertia, gravity and the actual size and distance between axes of motion to create optimized motion trajectories that eliminate overshoot and jerk in the load. This is particularly significant in terms of error detection and stall detection. These properties in complex actuators require the control system to be able to differentiate between normal operating conditions and something interfering with the motion. All of this can be done from the current loop, but the controller still has to be able to distinguish between a lightly loaded system and a heavily loaded system, and an actual stall.
The Agile Planet solution is currently in use with a wide variety of applications in robotics from medical applications involving patient contact to handling radioactive materials. The shift in the control philosophy is quite significant and opens the door to truly “intelligent” motion control.
Tell Us What You Think!