Mention "fieldbus," and the entire process-control and factory-automation communities heave a mighty sigh—not of relief, but of sheer irritation, impatience, and confusion. In an effort to establish an international fieldbus standard and with a workable standard probably more than five years away, market forces have compelled two quasistandard bodies to publish, promote, and support their complete—but different—specifications for a fieldbus. These bodies, with the unlikely titles of Interoperable Systems Project (ISP) and World Factory Implementation Protocol (WorldFIP), have each sworn allegiance to the official developing standard and to converging their specifications with whatever the standard finally becomes.

Against this confusing and uncertain background, designers of process-control and factory-automation products must decide how to proceed. Falling into line with ISP or WorldFIP means designing on a foundation of slowly shifting sand. Some level of early obsolescence seems unavoidable.

Certainly for the short term, companies that want their products ultimately to conform face a dichotomy: Which army will march the shortest route to the standard? Each side is assembling an armory of silicon and design tools, field trials are in progress, and the battle is about to commence. And, by the way, shoot the first guy who says maintaining conformance means changing only a bit of software.

"Fieldbus" is a generic term describing a digital, bidirectional, multidrop, serial-bus, communications network to link isolated industrial field devices, such as sensors, actuators, and controllers. Overall, by installing low-cost computing power in each field device, fieldbus will replace centralized control by distributed-control networks. Fieldbus will also improve data integrity and introduce device control, calibration, and diagnostic functions. Other benefits, including lower installation and maintenance costs, result from substantially simpler plant cabling.

Considering the maturity of the process and automation industries, you might expect there to be at least some de facto standard digital network. In reality, however, nothing could be further from the truth: The most standard feature of today's industrial-control systems is still the 4- to 20-mA analog signal that has been in use for decades.

But don't imagine that the absence of a digital standard results from any lack of resolve on the part of officialdom. Since the mid-1980s, the Instrument Society of America (ISA) and the International Electrotechnical Commission (IEC) have made tremendous joint efforts to establish a unified fieldbus standard. But, despite their efforts and the volumes of draft documents in circulation, ISA/IEC have so far published a specification (Ref 1) for only Layer One—the physical layer—of a fieldbus.

A multitude of problems, none of which is technology-based, has dogged ISA/IEC's drive for a complete standard. The main difficulty is achieving consensus among the many heavyweight players in the process-control and factory-automation industries. These system suppliers have jealously protected an immense installed product base and substantial future interests. Equally, some suppliers—with hitherto-proprietary bus schemes—are slow to accept the concept of an "open" standard and the prospect of product interoperability.

Other reasons for the delay relate to the ambitious scope of the proposed standard. Although primarily for communicating between field elements in the process industry, the standard will also apply to plant automation in building, manufacturing, and transportation. The standard will specify all layers of an ISO/OSI 7-layer model for a communication protocol and will include an eighth user layer to ensure full product interoperability.

Interoperability, a key fieldbus objective, permits straightforward substitution of bus devices from different manufacturers without affecting overall system performance. Toward that end, all bus devices will contain a software-function block or blocks and a software-device description appropriate to the device. For example, a sensor will contain an "analog-input" function block, and an actuator will contain an "analog-output" block and maybe a "3-term-control" block. The device description differentiates between similar types of devices, such as high- and low-temperature-range sensors. With function blocks, device descriptions, and "set-point" data from hosts, a sensor and an actuator will work together over the fieldbus, forming a separate control system.

ISP and WorldFIP set up shop

Both ISP and WorldFIP have set up organizations to promote their versions of fieldbus. These organizations offer a wide range of services, including publishing specifications, newsletters, and catalogs; offering sources of silicon and design tools; providing seminars, training programs, and field-trial support; and coordinating product-compliance testing. Most of these services support organization member companies. What a company pays for membership depends on its size and the level of support it requires. ISP's annual charge ranges from $1000 to $75,000 for companies with sales of more than $100 million. WorldFIP's annual charge, which is double in the first year, ranges from $1000 to $12,000 for companies with 1000 or more employees.

ISP's principal subscribers include Fisher-Rosemount, Foxboro, Siemens, and Yokogawa. Behind WorldFIP are Allen-Bradley, Honeywell, Square D, and a strong contingent of French companies, including Cegelec, ElectricitŽ de France, Elf, and Telemecanique. WorldFIP's French influence comes from an earlier French National Standard, NFC 46-600, also known as FIP, from which WorldFIP developed. By comparison, ISP emerged largely from a German National Standard, DIN STD 19245, also known as Process Field Bus (Profibus).

Both FIP and Profibus remain alive and well, although fieldbus purists despise the inability of these buses to be "intrinsically safe." Intrinsic safety, an important option for any future fieldbus system, refers to the power level a device can transmit along the bus—a significant factor when a device is processing flammable material, for example. This safety limit also accommodates another fieldbus option that requires bus cables to carry dc power to field devices. Both FIP and Profibus fail to conform to current intrinsic-safety limits because they are based on high-level signaling and because they require 4-wire cables compared with a single twisted pair for the fieldbus.

If you're not already confused, take a look at Fig 1, which shows the origins of the three main fieldbuses. However, Table 1, a comparison of the salient features of the main fieldbuses and their alternatives, might allay some of the confusion.

Table 1—Fieldbus versions
	ISA/IEC	ISP	WorldFIP	LonWorks	HART
Media	Twisted pair	Twisted pair	Twisted pair	Twisted pair, power line, wireless	Twisted pair
Data rate	32.25 kbps (H1), 1 or 2.5 Mbps (H2)	31.25 kbps (H1), 1 or 2.5 Mbps (H2)	31.25 kbps (H1), 1 or 2.5 Mbps (H2)	30 or 78 kbps, 1.25 Mbps	1200 bps
Line lingth (m)	1900 (H1), 750 (H2)	1900 (H1), 750 (H2)	1900 (H1), 750 (H2)	1200 (38 kbps), 2000 (78 kbps) 500 (1.25 Mbps)	1500
Maximun no. of nodes	256	32/segment, 125/region, 64 regions	256	127/subnet 32,385/domain	15
Data-link layer	No specification	Token-passing, centralized	Centralized	Carrier-sense multiple-access	Dual-master centralized
User layer	No specification	Implemented	Implemented	Not implemented	Implemented
DC power on bus	Yes	Yes	Yes	Planned	Yes

The first silicon that supports ISP or WorldFIP comes in the form of device-communication controllers. Generally, these controllers handle all physical-layer functions and, to various degrees, functions in the data-link layer. In these products, 8-bit register interfaces allow direct connection via an Intel- or Motorola-type bus to a device's host processor.

Among the companies offering such controllers are Fuji Electric and Yokogawa. Both companies' controllers support ISP protocols. For the lower 31.25-kbps data rate (H1), Fuji and Yokogawa offer the Frontier-1 and Find-1 controllers, respectively. For the higher, 1- or 2.5-Mbps data rates (H2), Yokogawa offers the Find-2 chip-set pair.

Another company, Shipstar Associates, claims that its Fchip-1 series of controllers supports both ISP and WorldFIP protocols at either high or low data rates. Fchip-1 uses Actel ASIC technology and later will use AMI ASICs when Shipstar begins manufacturing the devices in large volumes. Shipstar also offers a wide range of support tools.

You can also obtain WorldFIP controllers and a PC-based development kit ($15,000) directly from WorldFIP. Controllers in- clude FullFIP2 from Cegelec and FIPC01/2/3 from Telemecanique.

If entering the main battle is not for you, there are other, more immediate and proven routes to a fieldbus. For example, Echelon's LonWorks control network and Rosemount's Highway Addressable Remote Transducer (HART) protocol represent viable alternatives.

Barry Haaser, Echelon's marketing director, shows scant concern that the company's LonWorks is not wedded to a global fieldbus standard. In fact, he believes it's unlikely that such a standard will ever emerge. If you look at the nature of the industry and its players and the history of FIP and Profibus, you see the same battle continuing and moving to the United States, says Haaser. He believes that it's unlikely that we'll see a clear-cut winner.

Haaser says the most obvious need is for a fieldbus that's available today and, more important, that the fieldbus that emerges is at the sensor-network level of a control hierarchy. LonWorks finds most success at this level, which includes sensors, actuators, and distributed-I/O devices. The product range currently has over 700 users worldwide and includes around 100 OEM products. Echelon also offers a product-interoperability testing service.

LonWorks' major features include the wide range of media it supports (twisted pair, RS-485, fiber, power line, and wireless), its low cost ($10 to $20/node), and its size (2 in2/node). LonWorks nodes use Neuron chip controllers from Motorola and Toshiba, and a comprehensive range of software-development and network-management tools are available.

Rosemount's HART protocol has the virtues of being fully developed, available, and well-supported. HART user groups boast an installed base of more than 500,000 units using this open-bus standard. Key to its appeal is the ability to carry a conventional 4- to 20-mA analog signal with digital data on one twisted pair. HART systems use a Bell 202 1200-bps FSK signal superimposed on the 4- to 20-mA signal. The FSK signal has no dc component, allowing the analog and digital data to coexist without interference. In practice, though, this system limits the digital message-transaction rate to three transactions/sec and multidrops/loop to 15 when you use HART in digital-only arrangements.

Mark Portlock, process-control products manager at Arcom Control Systems, says these limitations are not significant for many data-acquisition and maintenance systems and process-control systems with relatively long response times. He says HART particularly suits systems involving physical metering and control, in which temperature, pressure, or flow changes, for example, take seconds or longer. Factory-automation and process-control applications with response times requiring more than 10 message transactions/sec require the use of a high-speed fieldbus.

Portlock believes that 15 multidrops/loop is generous because, although the number of nodes in an overall system may exceed 15, sensors and actuators tend to be in clusters, which generally occupy different sites of a system. Thus, it's convenient to run separate loops to each cluster.

Portlock estimates a roughly 50:50 split between HART users that retain the 4- to 20-mA function and those wanting full digital multidrop operation. Users employing the mixed-signal arrangement use the digital path for data acquisition, control signals, and maintenance and calibration data—which is significant for ISO 9000-compliant systems.

Portlock reports virtually no pressure from users or potential clients to standardize on a designated fieldbus. However, Arcom is carefully tracking developments because the company sees higher speed and distributed-control performance as important for future systems business. Currently, though, Portlock says, the gray area surrounding the fieldbus makes it too early to predict the direction Arcom will take.

For more information on HART systems, the UK-based HART user's group publishes the "The HART Book," a directory of worldwide sources of HART-supported products. For further publications on HART systems, see Refs 2 and 3.

1. "Fieldbus Standard for use in Industrial Control Systems, Part 2, physical layer specification and service definition," Instrument Society of America, SP50.02, 1992.

2. Bowden, Romilly, "HART Field Communications Protocol—A Technical Description," Rosemount, 1991.

3. "Developing Process Systems using HART," Arcom Control Systems.