Solid, effective, reliable, and well-supported standards are rare. Although users need and want such standards, electronic-design-automation (EDA) tool vendors have been slow to support them. The situation is particularly unfortunate in the area of populated pc boards. A number of standards are in use, but they provide only partial-and ineffective-re-sults, including Gerber, the Electronic Design Interchange Format (EDIF), and the CAD Framework Initiative (CFI). To make up for the lack of standards, the EDA industry has borrowed geometry-oriented standards from mechanical engineering, such as IGES and DXF, for drawings, board outlines, and the like. However, these standards, too, meet only some of the users' requirements.

Now, the Standard for the Exchange of Product Model Data (STEP) is emerging as a powerful, professional standard and a model for similar standards in other areas. Developers within the ISO spent considerable resources developing STEP as a standard for product descriptions, and the end result is impressive (see box, "Who develops STEP?"). STEP encompasses a wide range of products, and it covers the products' entire life cycle.

The first of the standards preceding STEP was Gerber, which takes its name from its developer, a photo-plotter manufacturer. Gerber developed the standard as an internal file format describing pc-board layout. Gerber lists commands for a photo plotter and a drilling machine and includes pure graphics with no "intelligence" whatsoever. Gerber has been around a long time, and it works well. More important, nearly every vendor supports Gerber. Despite its advantages, however, Gerber does not integrate well with the other dominant standards.

The next standard to emerge was EDIF, a file format launched in the mid-80s that enabled users to exchange netlists. Over the years, developers have added more and more "speed stripes," but EDIF still effectively handles only netlists. How well EDIF works depends largely on the skill and experience of its users. EDIF-PCB, which extends EDIF from netlists to layouts, is in the works.

The latest major standards effort, CFI, emerged from a group of tool vendors who wanted to achieve better interoperability among their tools. CFI lets users of Unix workstations perform exchanges in real time of netlists among tools from different vendors. Thus, CFI not only defines the data to be exchanged and the mechanism with which to do it, but also includes specifications for communication, windowing, and process interaction. However, like EDIF, CFI effectively supports only netlists. It also lacks facilities for archiving, for exchanging data off-line, and for exchanging data between tools in distributed and heterogeneous environments. These requirements are critical for users who want to replace one vendor's tool with that of another vendor.

CFI's developers are planning a major rewrite to combat these deficiencies. The rewrite will probably add a file format, enable communication among mixed platforms, and remove some of the communication and windowing specifications, leaving those areas to the various operating systems. Like EDIF, CFI will also eventually expand the data it handles beyond simple netlists.

All these standards make for a messy situation. With the exception of Gerber, none of the standards works well; integration among them is nonexistent; and attempting to use them as a group is hopeless. Mechanical engineering has long grappled with the same problems, albeit on another scale and with a different set of market dynamics. However, the mechanical-engineering industry has matured past the IGES and DXF stage and is now moving on to STEP.

STEP standardizes on two levels. First, it specifies the "information model," that is, the concepts and the terminology an application uses. The underlying information model ensures that the receiver can correctly understand the meaning of the information. Second, STEP specifies the actual exchange mechanisms to transfer the information.

STEP's modular architecture allows application areas to have different information models. Application protocols (APs) define these models, which are the most outwardly visible parts of STEP. The information models are all written in Express, an information-modeling language that STEP defines. Although splitting STEP into APs makes the standard more flexible and manageable, the division raises the question of interoperability among APs. To handle this problem, STEP uses an infrastructure for integrating data from different application areas.

STEP includes two exchange mechanisms, a file format and the Standard Data Access Interface (SDAI), an application-programming interface for on-line communication. The file format works in a way similar to the file formats of EDIF and IGES. SDAI is an on-line communication interface similar to CFI. The mechanisms exchange data that are independent of the mechanisms themselves. Instead, the mechanism making a reference to an Express model defines the data.

STEP lets independent labs test conformance against the standards. These labs can issue certificates of conformance to products that claim to support STEP. For a buyer of a system requiring STEP compliance, life is easy: Ask your vendor to show that certificate. This document gives STEP the teeth to ensure that things work reliably in practice.

Users have given STEP strong backing, and vendors view it as critical to the ability to compete in the market. Given this clear and well-financed backing, CAD system vendors are all supporting STEP. In addition, the mechanical-engineering, electrical-engineering, process, and architec-ture/engineering/construction industries will all employ STEP. The geographical-information-systems industry is also creating standards with an architecture similar to that of STEP and may use Express as the architecture's modeling language.

You can use STEP to exchange files, in portable applications, as a standard database, or as a tool for application development. In Fig 1a, the STEP file format exchanges files, and a user must specify which AP STEP is using. In Fig 1b, STEP uses SDAI as the standardized programming interface to the system database. An application written using SDAI and an AP can then move freely between a system having an SDAI for that AP. In Fig 1c, the content of an AP specifies the content of a database. All applications written using SDAI and that AP can then concurrently access that database. These applications serve as "editors" to an AP-defined database and are, therefore, fully interchangeable. In Fig 1d, the various tools within STEP act as an environment for application development with or without an AP. The advantages include possible use of STEP information models and the availability of standardized tools from several vendors.

A series of APs within STEP concerns printed-circuit assemblies. The first in this series is AP 210, "Printed Circuit Assembly Design and Manufacture." This AP supports netlist, component list, board layout, component placement, and administrative data. The AP does not support component data, except for geometry. AP 210's primary goal is to define the information that the design department passes on to manufacturing. AP 210 also works as a shared database for concurrent engineering. The information that describes the printed-circuit assembly is complete, is free of redundancy, includes precisely defined semantics, and comes in a vendor-independent form. From the ground up, AP 210 incorporates both electrical and mechanical characteristics-a critical requirement because many real-life problems result from a mix of these characteristics. AP 210 is then the ideal platform for realistic analog and digital simulation, EMI/electromagnetic-compatibility analysis, thermal analysis, verification of manufacturability, and archiving. The AP 210's vendor-independent database lets users mix any AP 210-compatible tools.

STEP is also affecting its standards' predecessors. For example, EDIF and CFI developers are rewriting both standards using Express. (VHDL also now uses Express.) However, even after their upgrades, EDIF and CFI remain merely subsets of AP 210.

Thus, we have a situation in which a technically superior standard (STEP) must attempt to overcome the current standards (CFI and EDIF). In that situation, the established standard normally wins. (Remember Betamax vs VHS?) However, STEP is different because it has critical mass outside electronics. It is becoming more commercially important than EDIF and CFI can ever hope to become. As the market situation unfolds, STEP will be the dominant player, and EDIF and CFI will become isolated, unimportant historical artifacts. For many players, it will be smarter to use STEP, the complete, well-written and well-integrated standard with massive market support. Finally, since neither EDIF nor CFI has succeeded in gaining wide use, users can discard them without wasting too much investment.

What happens now depends on the major players within the user community. They should become well-informed of the business potential of aggressive use of standards. Any industry group could follow in the footsteps of the German automobile industry, for example. Those manufacturers created and funded the ProSTEP project to define a suitable AP and to develop the software to translate to and from the CAD systems they used. The manufacturers did not ask the permission of CAD vendors before taking this step. They simply decided to take control of their own data-with or without the help of the CAD vendors-so that they could build the information systems they wanted. Initially, the manufacturers did not even invite the CAD vendors to join them. Later, when the vendors did join the group, they provided support for that AP, the AP 214 Core Data for Automotive Mechanical Design Processes.

Finally, it is understandable if the people behind EDIF and CFI are not enamored of something that has "not invented here" written all over it and that is effectively going to put them out of business. However, this should not stop users from adopting STEP in electronics. Times change, and one technology generation replaces another.

Lars Celander is a research scientist at the Institutet för Verkstadsteknisk Forskning (the Swedish Institute of Production Engineering Research), Mšlndal, Sweden. He has worked at the company for five years, helping Swedish industry use the Standard for the Exchange of Product Model Data. Celander received an MSc in Engineering Physics from Chalmers University of Technology. In his spare time, he enjoys sailing, scuba diving, and watercolor painting.