Les Hunt talks to John Orrells about current trends in fieldbus
implementation and what the future holds with regard to Ethernet
A decade or more ago, the world of manufacturing and process automation
seemed hell-bent on achieving a single standard for fieldbus, despite all
the politics and self-interest that tended to get in the way of this
goal. It never came, of course, and with hindsight we can now see how
difficult it would have been for such a standard to keep pace with
technology. Today, 'interfaces' are the key to successful control systems
integration, with software like OPC (OLE for Process Control) galvanising
the market into action, not fieldbus standardisation. But the lack of a
standard leaves the field open to a core of around five or six major bus
types, and that does mean the user has to make a choice.
According to John Orrells, an acknowledged fieldbus expert and head of
Control Techniques' Software & Systems think-tank, users of automation
and drives products should nowadays be more concerned with what control
strategy they intend to implement, rather than what fieldbus they should
use. A good knowledge of what exactly you intend to control will usually
point you in the direction of the right bus. If we take away the purely
political dimension of what bus system appears to be in favour, or in the
ascendancy at the time, then the major determining factor is whether the
system would benefit from a 'centralised' or 'decentralised' control
strategy, he says.
Centralised control strategies require a network master controller -
typically a PLC, which is entirely responsible for controlling
communications over the network. The 'slaves' out in the plant are
generally dumb devices with no local intelligence, and are therefore
unable to support peer-to-peer communication. But manufacturing industry
is already fast outgrowing this technology.
As plants become more automated, so the number of nodes on the network
increases. Production speeds are getting faster, manufacturing tolerances
are much tighter, on-line inspection is forming part of the process
control strategy - it all places great demands on the network. In these
circumstances, the master controller must have substantial computing
power to handle the data burden in real time - and this doesn't come
cheap. Look at drives, how quickly they have changed in terms of their
level of sophistication. Installed in a PLC-based, centralised control
system, the modern 'intelligent' drive is nothing more than an amplifier;
its intelligence is rendered redundant by the configuration.
The current trend is to distribute the controlling functions around the
plant, making use of local intelligent devices to facilitate highly
dynamic applications over the network. This is decentralised control and
it can be a very powerful alternative to the master-slave configuration,
particularly for fast moving machines and production lines. Peer-to-peer
communications is essential for decentralised networks, and this allows
data transfer to be more focused, thus reducing the computing power
required at each node - ideal for real time applications as all nodes are
effectively running in parallel.
Drives manufacturers tend to support both centralised and decentralised
network topologies, in order to provide the broadest choice for their
customers. Our Profibus, InterBus-S and DeviceNet interfaces cater for
centralised control strategies, while we support the distributed control
model with our 'in-the-drive' UD Series intelligent options module and
CTNet, our own fieldbus, says Mr Orrells, who is quick to point out that
CTNet is not, as its name might suggest, a proprietary system.
With a gateway to all industrial networks and a direct connection to
Profibus-DP, InterBus-S, Modbus Plus, Ethernet and DeviceNet, CTNet is a
drive-orientated network that is particularly suited to re