TechArchive

This article was originally written in the period 1995-2000

John Doran, Marketing Manager at Optimised Control discusses the role of networking technology in relation to multi-axis motion control.

Machine designers are, in general, very aware of the advantages of distributed I/O, thanks to the clamour of the burgeoning fieldbus industry. But the role of networking technology in relation to multi-axis motion control is less widely understood, even though it offers potentially more scope for gains.

Firstly, how is multi-axis automation is implemented without a communications network? The underlying motion system typically consists of a powerful master controller and simple servo amplifiers operating as speed controllers using analogue voltage signals. All intelligence resides in the controller, which needs to be a high performance computer with I/O hardware. It must read position feedback, perform all position loop calculations and produce an analogue output for each drive. It must also generate motion profiles and supervise and control all other aspects of the machine such as the user interface, power sequencing and fault handling.

Such centralised controllers normally have a fixed computing capacity and fixed amounts of I/O. This traditional architectural format restricts machine design flexibility, and makes it difficult to optimise the hardware. As a result, the multi-axis architectures that are used have become extremely diverse and highly custom in nature. This makes development cycles long and expensive. And the architecture ? depending on the quality of the design ? can be difficult to modify or adapt to meet the growing demands of todayOs automation buyers in terms of flexibility, reconfigurability, ease of use, interface standardisation and integration with higher-level information technology networks.

What is the motion control industry doing to support new distributed architectures? Broadly speaking, two things. The first is the simple provision of a fieldbus interface ? high speed serial links – on conventional drives for connection of distributed I/O, eliminating one major limitation of ready-to-use hardware. With such an interface available, users can plug-in virtually any amount of modular I/O via a bus, and cut substantial amounts of assembly time and cost from large machine projects.

The second, and more challenging trend, is the use of fieldbus interfaces for connecting with the drives themselves. This makes radical advances in motion control technology possible. Over the last couple of years three or four buses in particular have emerged as contenders for the motion system application: CAN, Interbus-S, Profibus and Sercos. Each offers particular trade offs, outside the scope of this short article, and for the purposes of this illustration, we will use explore examples based on CAN. This is a widely used and supported industrial network, and one with a track record in automation equipment – through its historical links with Germany (it was developed by Intel and Bosch).

Once the decision to network drives is taken, two types of architecture are possible. The first retains a central intelligence and sends fairly low level position and speed type commands over the network to the drives – implementing amaster-slavenetwork configuration. The second exploits the intelligence of the networkable drives for performing complex motion control locally, allowing more autonomouspeer to peerarchitectures.

The key benefits of networkable servo drives are:

• reduced wiring and installation costs, higher reliability
• reductions in hardware costs because the central controller no longer needs an analogue output and encoder input for each axis
• a single man-machine interface as all drives are parameterised via the network
• availability of status and diagnostic information
• drift-free, repeatable performance, with no noise problems
• software-based performance and configuration simplifying the creation ofspecials, sophisticated MMIs and links to higher information network levels
• reduced costs resulting from use of multi-vendor supported standards

If an intelligent drives philosophy is chosen, these gains are complemented by:

• further reductions in hardware costs because the controller only has to co-ordinate intelligent nodes (the extra hardware cost in the drives is marginal since each networkable node already has a CPU)
• truescaleabilityof the system allowing costly drive hardware to be matched exactly to requirements
• highly modular architecture, allowing large machines to be divided into simpler design projects for simplicity, and speed to market

In the latter style of networked motion system, instead of executing the entire operation the centralisedmastercontroller provides inputs (limits, stop, start, etc) to the motion functions in the drives, and monitors these operations while in progress. The master then co-ordinates these operations with complementary operations on other axes, by sending commands to mailboxes at each drive.

Control networks do not make traditional drive technology obsolete. Although networkable drives offer significant advantages, there are some drawbacks. The key limitations are somewhat slower response times, and less control over multi-axis synchronisation, because commands must travel over a network. From my company’s experience in implementing systems, networked motion control tends to score very highly in any situation involving loosely coupled axes, as are often found in printing and packaging machines, or materials handling systems, for example. They do less well, and can be unsuitable, where axes must be very tightly synchronised and interpolated as in some sectors of the machine tool industry.

Two example applications illustrate their benefits. The first machine ? a piece of automation for electronics PCB manufacture ? uses direct digital CAN drives to control movement and positioning systems responding to commands from a central controller. Here, the digital nature of the system means that the controller just has to send position information, eliminating the need for an encoder interface on the host system and analogue signals to the remote drives. This saves wiring and at least (UK pounds)500 in central motion controller costs for the six axes of control in this equipment.

A second example illustrates the more powerful peer to peer architecture. This system replaced a centralised motion controller and drives system on a large packaging machine. The use of intelligent drives cut the cabling needed to interconnect the conveyor feeding, synchronisation and carton erection stations to just power and network – a much more manageable arrangement than the previous power and drive cabling which necessitated clumsy drag-chains. The complexity of the central controller itself was greatly reduced, now requiring just a small processor capable of co-ordinating the movement. By using CAN, the manufacturer was able to move from a proprietary network to an industry-standard one (CAN – available in chip form from several leading semiconductor manufacturers), saving considerable expense and software development time, and greatly improving the robustness and reliability of the network. The peer to peer nature of the motion architecture is also simplifying the company’s plans to make their machines more modular, allowing them to create autonomous standalone elements (feeding and registration stations etc) which may be integrated into systems and assigned to work in groups easily, to configure the machine to work with different packaging materials and sizes, or create application-specific machines without the requirement for complex new software programs. In this case, for the eight axes involved, a distributed architecture saved around (UK pounds)1500 in motion controller hardware costs alone. As many packaging machines can often have 30 axes or more ? the potential for direct hardware savings by using this approach can easily be (UK pounds)5k per machine.

There are several higher layer protocols available for CAN, the two most popular being CANopen and DeviceNet. Given the performance requirements of this system, CANopen was chosen because of its lower overhead.

• Optimised Control
• Tel: 0117 935 4201
• Fax: 0117 935 4122