Jacob Tal
President
Galil Motion Control Inc.
Rocklin, Calif.
EDITED BY John R. Gyorki
|
A typical centrally controlled motion system requires wiring from each component to the central controller. This requires numerous cables and is usually considered a problem. |
|
An Ethernet-based control system requires only one cable from the host computer to a hub which then locally branches out to the other control axes. This scheme reduces the total cable length. |
|
The E-Series motioncontrol scheme uses one controller for a master and the other identical controllers as slaves. The advantage is the host communicates only with the master and provides commands for all axes. The master also holds all data concerning the axes, and the host can interrogate the master controller about the status of any axis. |
|
The motion of the C and E axes is controlled by the first and second slave controllers, respectively. However, as far as the host is concerned, it deals only with the master controller. The master controller then informs the first slave, at the time t1, about the required motion of C, and to inform slave 2 at t2 about the needed motion of E axis. |
Motion-control-system architectures usually come in one of two types, either central or distributed control. Historically, the choice has been more often central control. In fact, more than 90% of the high-performance motion-control systems now use this architecture. But, the picture is quickly changing, as many new devices are becoming available for easy-to-assemble and moreefficient distributed-control systems.
Central versus distributed control
A typical central-control-system approach contains an embedded motion controller plugged into a PC's ISA or PCI bus. This lets the host computer communicate with the motion controller through the bus, a simple and fast process. But this central-control method has one major drawback. It requires that all control signals be wired to the central controller located in the PC, including the motor, encoder, and limit signals. As the various components may be spread over a large area, the long wires needed reduce reliability, add noise that may degrade the system performance and increase system cost.
Over the last 20 years several distributed motion-control systems have been developed to try to overcome this problem. These include Sercos, Profibus, Canbus, and others. They all solved the wiring problem, but proved to have some serious limitations. Sercos, for example, controls motion well, but does not let other devices work on the network. Furthermore, all the proposed solutions, because of some restriction, require relatively expensive components. This, combined with the need for every axis to have a separate motion controller, increased the system cost significantly. And this explains why, over two decades, distributed motioncontrol systems did not gain wider use.
Ethernet-based DMC
Recent advances in Internet technology, however, have made the Ethernet an attractive candidate for motion-control networks. As more users became familiar with the TCP/IP protocol, it became easier for them to connect various devices to the network. This, in turn, pushed down the cost of system components such as Ethernet cards and hubs, leading to further interest in this design approach.
An Ethernet-based distributed motioncontrol system consists of a host computer, an Ethernet hub, and a number of motion controllers. For example, consider a system which illustrates the use of a two-axis motion controller.
Ethernet-based motion-control systems address most of the concerns of the designer. First, they solve the wiring problems. The controllers can be placed next to the motors and amplifiers, so the wires can be shorter. Furthermore, when optical fibers are used for communication, noise coupling and grounding problems can be totally eliminated.
The proposed solution lets users place a variety of devices on the network, and due to the common TCP/IP protocol, all devices can communicate. The Ethernet-based network also reduces the system cost because of the cheaper hub and the Ethernet card. Moreover, where each controller is a dual-axis card, it reduces the cost per axis.
However, compared to central control, this approach still has some disadvantages. The problem primarily involves communication and motion programming. First, the computer program that controls a distributed system has to be more complex because it must talk to different devices at different times. This is clearly a disadvantage when compared to central control where the host communicates with only one controller.
Secondly, regarding programming in a distributed system, the host computer collects the data from all controllers and synchronizes the motion. This is certainly more difficult than dealing with one motion controller where all the data is stored.
Ethernet Distributed Motion Control
A special motion controller was designed to overcome these problems and provide an optimal solution to a distributed motion-control system. The system hardware elements are very similar to those in a standard Ethernet-controlled system as shown in the diagram, E-Series motion-control system.
The main difference is in the controller operation. One of the controllers is designated as master. This device controls the A/B pair of motors; all the remaining controllers are designated slaves. For the purpose of control and communication, the master controller acts as a virtual multiaxis central controller. The host computer then needs to communicate only with the master controller and gives it the commands for all the axes. The master controller informs the slave controllers about their respective tasks. The master controller also holds all the data concerning the axes. As a result, the host may interrogate the master controller about the status of any axis.
To illustrate the operation of such a system consider an example, represented by the included program. It commands the motion of the A, C, and E axes, synchronized according to the figure, Velocity profiles.