The idea of putting Web servers on machines in an assembly line would have seemed far-fetched just a few years ago. Now, manufacturers increasingly view this strategy as a mainstream technique for getting better control of production.

The advantages of machine-to-Internet access are well chronicled. Hooking a PLC or machine controller up to a Web server permits sending information about machine status anywhere in the world. Top management as well as suppliers and customers can get real-time updates about what's going on in production. A server can notify a technician via e-mail when a machine needs help.

Some newer PLCs and industrial computers implement Web servers as a plug-in card that is part of the controller chassis. However, this tends to be an option only for higher end controllers. And there are wide variations in the capabilities available on these built-in servers. In many cases, they demand that developers have expertise in Web programming languages such as Java to design Web pages for reporting functions.

It's likely the next generation of controllers will better integrate functions of Web servers within the controller itself. But manufacturers can put Internet connections on their production gear today without investing in new controls. There are thin servers that target such industrial uses. Thin servers typically consist of a Pentium-class processor, nonvolatile memory, and built-in software that handles communication between a machine controller and the network. The thin server connects to a network through a standard Ethernet 10/100Base T port, a modem, or a wireless network connection. It connects to a PLC, process controller, or sensor bus through a standard RS-232/485 serial port or a fieldbus adapter.

One additional advantage of this approach is that thin servers often come with development packages that have machine designers in mind rather than programmers. They use wizards and other techniques so that end users need not be knowledgeable in Web programming to set up Web pages that report on machine status.

The PLC controlling the machine, in many cases, need not know the thin server exists. During setup, the typical approach is to run an application development tool on the thin server that loads driver software specific to the brand of PLC or controller. This driver lets the server monitor the status of various I/O points in the system.

For example, take the case of a thin server working with an Allen-Bradley PLC. A-B controllers use a standard serial protocol called DF1. The server would load a DF1 driver that would let it map between contact numbers and rungs of a ladder program running on the PLC. The server would have this information available for use in whatever Web pages it puts up detailing the status of coils, contacts, or other devices in the system.

A similar scenario would apply if the thin server worked with fieldbuses such as DeviceNet or other widely used sensor networks. Documentation detailing the physical mapping of devices on the bus and their bus addresses would be used to setup the server for monitoring their status. Thin servers can also work with proprietary controllers (with proprietary serial protocols) this way, so long as there is protocol documentation from which a driver can be developed.

Finally, the process of connecting thin servers to an Ethernet is similar to that for networking ordinary PCs. Thin servers are often set to request an IP address automatically. But it is sometimes convenient to give the server a fixed, static address. This allows outside viewers to examine information put up on Web pages that the server generates. But whether the server address is static or dynamic, there is no special expertise on the part of the network administrator to configure the connections.