Thomas Edwards
Senior Technical Advisor
Opto 22
Temecula, Calif.

As PC-based controls become more mainstream, a related technology is drawing ever greater attention — plug-in Ethernet I/O. These rack-mounted modules connect sensors and controls to Ethernet, the de-facto enterprise network standard. More and more companies these days are using Ethernet to share production data or control processes either locally or remotely.

Available and inexpensive off-the-shelf network components such as hubs, routers, and switches make deploying Ethernet I/O systems relatively simple. The Ethernet format works especially well for distributed-control and data-acquisition applications.

Distributed control
A key element of these distributed systems are so-called intelligent I/O modules containing on-board signal conditioners and processors. Intelligent modules boost system performance by relieving main processors or computers of tedious, repetitive tasks such as conversion of raw signals to engineering units. Module functions include latching, thermocouple linearization, counting, min/max values, filtering, and others. Certain functions such as counting and latching require no network communication. This dramatically improves system response and fault tolerance, important metrics for speed-sensitive control applications. Performance of distributed systems tends to be size independent where response times of centrally controlled networks scale roughly with the inverse of node count. This is a big advantage because control or data-acquisition projects usually grow by at least 50% from conception to completion.

An I/O unit typically includes one or more plug-in I/O modules, a mounting rack, and a local Ethernet I/O processor. Modules are often multiplexed to a single processor allowing I/O to be added in small increments without exceeding system capacity. Various plug-in I/O module types can accommodate digital or analog signals and a wide range of ac and dc voltage levels. Still others connect Ethernet to serial devices such as bar-code readers, printers, weigh-scales, operator interfaces, badge-card readers, variable-speed drives, analyzers, and message displays.

Besides having greater flexibility, this plug-in scheme eases maintenance because workers can replace a failed module in minutes without shutting down an entire rack. Still, failures are rare because modules have no moving parts and aren’t vulnerable to cyclic fatigue. Usually one or two spares of each module type is sufficient for most installations which reduces maintenance parts inventories.

Despite having virtually limitless expansion potential, there are practical limits to Ethernet I/O systems. Although bandwidth of standard Ethernet (10 Mbit/sec) is adequate for most distributed systems, other nonsystem devices attached to the network, such as office PCs and servers, can bog it down. This is why bandwidth-intensive control systems usually reside off the main backbone on a dedicated, switched (by router or other gateway device) network segment. Data-acquisition systems, on the other hand, can usually attach directly to an existing network because sampling rates are generally lower. However, when bandwidth is a concern, there’s fast Ethernet (100 Mbit/sec) available and a 1 Gbit/sec version is in the works.

Fewer and shorter wires
Because plug-in I/O systems locate near machinery or other data sources, wire runs are kept short which means less induced noise and lower line losses. Besides boosting performance, the approach can also lower installation costs. Some estimates for industrial wiring range from $700 to $1,500 to run eight pairs of line voltage or control signal wires 100 ft in conduit. Communications cabling is also simplified because node-to-node range is much greater with Ethernet than the more distance-limited serial variety. Connecting far-flung Ethernet I/O units to networks requires only a single, low-cost Category 5 network cable (one cable incorporates four twisted pair). And there are several alternative media types to choose from including fiber optics, coaxial, and wireless, all of which can provide even greater range and better noise immunity than twisted pair — important considerations for the factory floor.

Most modules incorporate optical isolation, transformer isolation, or a combination of the two for separating field power and grounds from the control system logic. This protects PCs and sensitive control hardware from damaging voltages, eliminates troublesome ground loops, and provides high common-mode noise rejection. Plug-in I/O with isolation costs about the same as without it, so opt for isolation when possible.

Making the connections
Field-wiring connections are made either directly to the plug-in Ethernet I/O module, or to a terminal on a mounting base into which the module mounts. Both methods have advantages. Direct connections make wiring accessible and can often shorten installation time. For some installations, however, easy accessibility to connections is a problem. Explosive environments, for example, require equipment that cannot cause ignition of the explosive atmosphere under any failure or operating mode. A requirement for explosive certification is that no spark-producing activity can occur without the use of a tool. Here, wiring must be covered to prevent access or tampering. In these cases, look for plug-in I/O that are suitable for Class I Division II hazardous areas.

Components can usually mount in any attitude, although when controlling large amounts of power, they should be oriented to permit maximum natural convective cooling. As a rule of thumb, when components are touchable, they’re probably not excessively hot. Most plug-in Ethernet I/O components can safely operate to 70°C. And most are hermetically sealed for use in any noncondensing environment to 95% relative humidity, as well.

Software considerations
The majority of Ethernet I/O systems use the ubiquitous TCP/IP protocol for data transport over Ethernet networks. Systems often have on-board diagnostics that are accessible by Web servers. A standard Web browser can be used to configure and test I/O, download customized Web pages, set event/reactions and simple network-management protocol (SNMP) traps, key in report-by-exception, and more. (SNMP is an open-network-management tool for monitoring and controlling TCP/IP networks and the devices connected to them.) Some systems, however, employ older or proprietary protocols for data exchange which can restrict such options.

Another factor is software compatibility. Several vendors offer both a control programming language and runtime environment that interfaces with a plug-in Ethernet I/O system. Here again, the communications interface or driver plays an important role, but equally important is the control software’s ability to use all I/O system functions. Interface drivers and control software that access only a subset of these functions mean lost power and flexibility.

The software requirements for data acquisition may be less stringent, but again, a widely supported and available interface is key. Ethernet I/O systems are especially capable of handling virtually any data-acquisition application, whether it is across a plant or the globe.