System Design

Electronic Component Information Exchange

The Pinnacles Group creates a new standard for electronic data books.

by Tom Jeffery

Current information for designing electronic components is highly desirable in today's rapidly advancing business environment. Component suppliers everywhere want to reach the marketplace, while their customers want easy access to distributors' information. The Internet offers a practical way to support rapid information distribution because it is pervasive, global, and accessible to millions of potential subscribers. Electronic information distribution permits data to be revised and broadcast immediately, without the expense and time delay of conventional publishing practices. However, unless the form and content of component data are standardized, customers will be unable to compare value among multiple suppliers. In addition, they may have to re-key data from the selected supplier for their computer-aided design systems.

The Pinnacles Electronic Data Book project Imagine a publications group that, after heroically completing last-minute revisions, finishes documentation just as a product is ready to ship, only to realize that printing, binding, and distribution will cause the data book to be six weeks late. Imagine a data book--printed on paper--created using CAD, automated test systems to compile data, and desktop computers to assemble data. Another company enters the data by hand from this paper data book (using expensive engineering resources) into another CAD system. Finally, imagine a design engineer searching for a part to design into a new system. The designer has a set of requirements and a stack of data books. Finding the required part means searching through the books, determining each company's scheme for representing the data, and trusting that the books are not too far out of date. These scenarios are not hard to imagine. Indeed, this is the way business is done today. Printed data books are necessary because they represent the only accepted interchange format.

Now imagine a different scenario. A component manufacturer updates an electronic data book at the last moment and creates computer-readable data that is accessible on the World Wide Web, eliminating the printing, warehousing, and shipping costs associated with paper data books. In this scenario, the designer uses browsers and search engines to look for component information in electronic data books from multiple manufacturers. In the ultimate scenario, the designer copies the desired data directly into his or her CAD system, eliminating the need for manual input.

This new way of doing business may be closer than you think. A team of electronic component manufacturers, collectively known as the Pinnacles Group, and consisting of Hitachi America Ltd., Intel Corp., National Semiconductor Corp., Philips Semiconductors, and Texas Instruments Corp., has been working on the problem for several years and is now planning to expand its scope and broaden its impact on industry. The essence of this team's work is an electronic data-book standard.

What is an electronic data book? It is a comprehensive collection of information about electronic components, including active and passive discrete devices, medium- and small-scale integrated circuits, cells, cores, materials, and connectors. An electronic data book contains the text and graphics that vendors of these devices typically print and distribute. It can include design information in a structured and machine-readable format. In addition, it can incorporate computer-readable data such as CAD files, behavioral and functional models, audio, and video. In short, an electronic data book includes all of the information and support that an engineer needs to design with these components.

Until now, there hasn't been a standardized description of an electronic data book, nor could information be exchanged consistently between different manufacturers' data books. Without these critical elements, it was impossible to create a tool set that could generate, store, browse, and retrieve the required information.

Figure 1. A standard for defining the structure and interchange of an electronic data book

To address this problem, the Pinnacles Group created, and is continuing to develop, a standard for defining the structure and interchange of an electronic data book. This standard is known as the Pinnacles Component Information Standard (PCIS), outlined in Figure 1. In addition to the development of a standard interchange format, the Pinnacles Group is working to define a standard set of terms or vocabulary to be used among component suppliers to facilitate true browsing and searching capabilities. This is known as the Component Information Dictionary Standard (CIDS).

Version 1.2 of the PCIS has been published by the Pinnacles Group. It is built on a defined set of tags in a standard generalized markup language. These tags identify the information content. The set of tags include the following:

• Named divisions as well as generic divisions.
• Specialized database-like structures for parametric data and a mechanism for reflecting this data into the body of a document.
• A mechanism for including models and modeling data.
• Structures for labels relating to the document itself--as distinguished from its content such as literature number, revision history, company name, and address.

Data books, like many other documents, have divisions that may have titles. In most documents, a division is identified solely by its title. Although PCIS allows this type of generic division, it provides named divisions as well. A given named division may have any title, but it can always be identified by a specific tag. For example, to search simple text data sheets from several companies for pin-out information, you would need to search for titles such as "Pin Description," "Pin Function," or "Pin Out" with the alternative spellings "pinout" and "pin-out." In a Pinnacles electronic data book, however, the division has the tag "Pin Out Information," regardless of an additional title attached by a manufacturer.

Examples of named divisions in PCIS include the following:

• Absolute maximum.
• Memory map.
• Architectural functional.
• Package information description.
• Product characterization.
• Features summary.
• Register sets.
• Instruction set.
• Soldering and mounting.

Information classes supplement named divisions. The standard calls for six classes of information in electronic component documents:

• Product summary.
• Detailed product specification.
• Application.
• Safety and environmental compatibility.
• Support tools.
• Reliability data.

Any division of a PCIS document may contain information from a particular class. This allows users to search for information by two independent criteria. For example, reliability data in an absolute maximum division, or safety and environmental compatibility information may be in a soldering and mounting division.

Structured data Perhaps the most important part of a component data sheet is its parametric information, such as the operating voltage, timing characteristics, and pin-out. Most companies supply similar information but in widely differing formats. This information is usually in human-readable format, but the user also needs it in a machine-readable format for CAD tools.

Although each company presents parametric data differently, analysis has shown that it can be described using a common computer-readable format. In PCIS terminology, this common format is known as "the Source." To allow companies to present data in their own format, PCIS provides a "Reflection" which presents data from the Source in any desired human-readable way. A Reflection contains pointers to the data in the Source, rather than containing the data itself. This way, if the information in the Source is updated, the references in the document text are automatically updated as well. Similarly, if a value appears several times in a data sheet, there is no need to find and change all occurrences; only the value in the Source must be changed. The body of the electronic data book contains text, lists, and tables, just as documents do now. Any of these structures may be Reflections of information within the Source.

Product identification elements Product identification elements must always be unambiguous. The common phrase for this identification is "part number," which, unfortunately, is not specific enough to describe the different levels of product information in current documents. A data sheet contains specifications for a product, but the product identification at the data sheet level is rarely the number used to purchase the product. Instead, a more detailed ordering number specifies speed, package, and other data. PCIS defines three levels of product identification:

Specific product ID specifies a physical product that can be purchased, including all the options needed to order the product. The full specific product ID is rarely used in a data sheet.

Product ID fragment specifies products that are related in some manner, such as packaging or temperature characteristics. A product ID fragment identifies a useful set of products through similar features. A product ID fragment that includes a wildcard is more commonly used than a specific product ID; for example, "the 94xx processor" represents "the 9419, 9420, and 9430 processors."

Generic product ID specifies a particular product, or collection of products, at the level described in a single document: a single data sheet, application note, data book, etc.

In addition, sometimes an alternate product ID is used. This is an alias for the specific product ID. An alias may be useful as an internal warehouse or inventory number, a customer's name for a product, or the informal or "street" name for a product.

A product ID in a PCIS document can have many levels of information. The whole document, a division, a paragraph, or even a warning can be associated with a product or group of products, using any level of product ID. In very simple documents, the only product ID may be for the entire document. However, in many documents, parts of the document discuss one product ID or set of product options, while other parts of the document discuss different ones. By inference, if a data structure is not specifically associated with a product ID, it is assumed to belong to the next higher level of product information.

Models and modeling data In the electronic component industry, computer-readable data associated with components, such as simulation models, are rarely included in printed data sheets. This information is often supplied as a separate document, often by a third-party vendor. In PCIS, modeling information can be either encapsulated in a separate file in the standard generalized markup language, or it can be stored in an external file, with internal references in the document. In either case, the model is stored in its native format, and managed through the standard graphics mark-up language (SGML).

Structures for document labels Two types of document labels are associated with data books: general labels and company-specific labels.

General labels contain information about the data book or data sheet, such as the document title, publication date, and what parts of the information have been changed in the current version. General labels are included within the PCIS scope.

Company-specific labels may include information on who wrote or changed portions of the information, what has been reviewed and by whom, and which process, machine, or corporate division produced various pieces of information. Company-specific labels are not included in the PCIS scope.

The CIDS The Pinnacles component dictionary consists of five types of subdictionaries: quantitative, non-quantitative, technology, representation, and general terms. These subdictionaries have been identified to provide a structure for the different types of terms defined. However, from the user's perspective, the CIDS is a single dictionary.

The CIDS offers many different ways to establish context. All words that are used for computer-readable purposes are called "terms." The set of all possible terms is properly called the vocabulary of the CIDS, which in turn comprises the five proper subdictionaries. These five subdictionaries represent the first possible context for a term. This classification is based on the different uses of terms within component information documents and computer systems. The context establishes the framework for understanding.

After the five dictionary classifications, the more refined context of a term is found within the term's definition. Within a particular dictionary, the definition of a term can have several possible elements to clarify the definition's context. For instance, if the term is quantitative, then units describing the measurement are included in the definition. The CIDS specifies the main content elements for any definition in the five different subdictionaries. Since CIDS is a closed set (that is, all terms used within the dictionary should be defined in the dictionary), complex signature structures can be made by assembling different terms. For instance, a TTL rise time could be expressed as " rise_time.TTL " to form a complex signature. Since rise_time is quantitative, within the CIDS Quantitative subdictionary, the dictionary term rise_time with a term definition should be found for TTL technology and the units would be defined as seconds. Additionally, the dictionary term "TTL" would be defined within the context of other subdictionaries.

Change control Each dictionary includes a means of change control that has been defined for electronic data books. Change control attributes also provide another possible context.

The PCIS V1.2 is now being put into practice by several of Pinnacles' member companies. However, to fully realize the power and benefit of this electronic paradigm, the CIDS must be formalized and the standards incorporated in a piloting and implementation phase throughout the industry.

The Pinnacles group is continuing its efforts, now under the auspices of CFI (formerly CAD Framework Initiative, working jointly with SEMATECH), and is submitting its work to the component industry for review and implementation. The next year is expected to show great strides toward making this a truly useful and powerful standard, possibly changing the very way the industry does business.

For further information on participating in this standardization process, please e-mail CFI at cfi@cfi.org.

Tom Jeffery is the principle technical writer at Hitachi America Ltd. (Brisbane, CA).

To voice an opinion on this or any Integrated System Design article, please e-mail your message to michael@asic.com.

integrated system design  October 1996