Report prepared by
James A. St.
Pierre
Curtis H. Parks
Electricity Division, Electronics and Electrical
Engineering Laboratory
National Institute of Standards and Technology,
Gaithersburg, MD 20899-8114
Ronald Waxman
EDA Standards Consulting
2369 Paddock Lane
Reston, Virginia 20291
Available in PDF for printing from the NIST Virtual Library. Go!
Integral to the electronics industry are product data-exchange standards and specifications that enable the design, manufacture, documentation, procurement, and support of modern electronics. The traditional formats for capturing designs and manufacturing information, engineering drawings and paper specifications, are rapidly being replaced by digital (computer sensible) formats. To be effective and efficient this information must be correct, complete, and unambiguous. Among the ongoing technical challenges in this arena are the development of adequate information models and standards that describe the essential characteristics of electrical and electronic products.
Currently, there are at least five established standards that can be used to transfer data among automated tools for fabricating electronic products. The industry's complaint is that these standards are not fully capable of expressing designs unambiguously and that “harmonization” of these standards is needed to avoid costly waste in design and manufacturing. Harmoni- zation of these product data standards is intended to enable the translation of design information from one format to another with no loss, corruption, or modifi- cation of the design intent.
One factor that will affect the future health of the U.S. electronics industry will be its ability to perform current business functions, electronically, over the Internet. The automation of business practices, such as the search for and brokering of component information, will enable the U.S. electronics industry to compete globally in a timely manner. A key element of internet commerce will be to make component information available via the Internet in formats that can readily be incorporated into computer aided engineering, design, and manufacturing (CAE, CAD, and CAM) tools. The Electronic Commerce of Component Information (ECCI) workshop held at NIST Wednesday – Friday, July 15 – 17, 1998, provided a forum for discussion of issues related to the exchange of technical information for electronic components in support of the global electronic component e-commerce (electronic com merce) marketplace. The goals and objectives of the workshop were to identify technologies that could be developed to remove barriers and to create opportunities for businesses and consumers in the electronics industry by accelerating the development of means to access, exchange, and reuse electronic component information. The workshop brought together industry technology experts, including representatives of technical consortia, and trade associations who were inter ested in participating in defining the scope of activity required to establish an international ECCI environment. Individuals from academic institutions, non-profit research organizations, and government laboratories also participated. This report provides an overview of the workshop as well as some of the planned follow-on activity. A full report, which includes all of the presentations given during the workshop, can be accessed via the www [1]. The on-line report, based on the contributions of the attendees, makes recommen dations for follow-on activity consisting of collaborative efforts to further develop the mechanisms for access to component data and the standards necessary to bring about the desired ECCI environment.
In addition to NIST, the co-sponsors of the workshop included the Institute of Institute of Electrical and Electronics Engineers (IEEE), Computer Society (CS), Design Automation Technical Committee (DATC), and Design Automation Standards Committee (DASC), the Silicon Integration Initiative (Si2), the Virtual Socket Interface Alliance (VSIA), and the University of Maryland's Computer-Aided Life-Cycle Engineering (CALCE) project.
One standard that received a lot of attention during the workshop was the Electronic Component Information Exchange (ECIX) standard being developed by the Si2. The NIST Electricity Division's Electronic Information Technologies (EIT) group has provided support for the ECIX project from its inception. ECIX also has ongoing support from a wide variety of industry participants. The ECIX project [2] is dedicated to designing stan dards for the creation, exchange, and use of electronic component information, including application specific integrated circuit (ASIC) cores or virtual components. The ECIX architecture and standards are intended to be extensible, unambiguous, well documented and are maintained under the direction of Si2. Si2 has requested NIST help to move the ECIX standards into the International Electrotechnical Commission (IEC) arena. ECIX currently consists of three draft standards: Pinnacles Component Information Standard (PCIS), Component Information Dictionary Standard (CIDS), and the Timing Diagram Markup Language (TDML). These ECIX standards are intended to enable engineers and design support personnel to view component datasheet information (PCIS), including viewing and analyzing timing diagram information (TDML) contained in ECIX datasheets, and directly support electronic design automation (EDA) tools and automatic generation of EDA library information such as logic symbols. In addition to traditional datasheet information, PCIS can enable controlled access and transfer of supplier pro vided design files (e.g., simulation models, test benches, and physical data) to the end-customer. A key compo nent of the PCIS datasheet is the identification of the links to on-line dictionaries of terminology and compo nent classifications provided by the CIDS.
The workshop was structured as a series of panel sessions. Panel members presented information regard ing their current activities. A question and answer segment was included as part of each panel to ensure adequate time for feedback from workshop participants. The attendee interaction provided the data for the workshop results. In addition to the panel sessions, time was allocated to allow participants to demonstrate their ECCI oriented products and/or capabilities. Some of the demonstrations focused on existing or pending products, while others focused on standards development activities.
Following Panel 3, a discussion session was held in which all participants had the opportunity to help determine “Where do we go from here?” That session took place on the second afternoon, while most of the attendees were still present.
The problems discussed at the workshop include:
The workshop was intended for those interested in creating, accessing, locating, integrating, manipulating, or maintaining information and/or engineering data (e.g., data sheets, simulation models, timing information, thermal data, geometry, etc.) or electronic components, either physical or virtual, such as:
The following is a partial list of data and functionality required to support an ECCI environment. This workshop was primarily focused on the first element in the list “Data content.”
Panelists:
Jim St. Pierre — NIST (Chair),
Bill Russell — Air Force Research Laboratory,
WPAFB, Jeff Barton — Texas Instruments,
Kim Singer — OrCAD,
Jeff Williams — Information Handling Services
(IHS),
Tony Hilvers — Institute for Interconnecting and
Packaging Electronic Circuits (IPC),
Patrick McCluskey — University of Maryland.
A lively discussion followed the presentations by the panel members. It was generally agreed that there is a need for a set of standards that bridges all aspects of a design. Furthermore, most attendees (especially users of component information) agreed that if all of the compo nent suppliers provided component information in the ECIX format, significant cost savings within their organizations could be achieved. A significant portion of these cost savings would come from the elimination of re-entry of design information. Using the ECIX standards, users would be able to directly load their databases, catalogs, or electronic design automation tools, without human intervention. It is nontrivial for the component suppliers to provide a fully documented ECIX datasheet. Some suggested that a phased ap proach would help the suppliers to migrate to the ECIX standard. There was a lengthy discussion centered on the ECIX utilization and deployment plans. There is a need for functional or behavioral information to be included in (or referenced from) an ECIX datasheet. ECIX is working on adding the ability to encode multiple views of a component. Although ECIX is an important compo nent in solving the problem, it relies on the existence of a wide variety of component information standards that can be referenced or “wrapped” by an ECIX datasheet.
A convergence of harmonized standards was also noted as an important goal. One aspiration is to use existing standards wherever possible. The participants agreed that there is a significant need for education of the potential suppliers and users of these standards if we are to achieve consensus. The community that will ben efit from ECCI must be a part of the solution by con tributing to requirement definition, surveys, education, and development.
A question was raised as to: “How much information are the component suppliers willing to provide?” One of the component suppliers indicated that they are con sidering implementing multiple levels of access to their on-line component information. Following this model, only certain customers would have access to the “latest- and-greatest” component information. Customer access to a component supplier's information would be deter mined on a case-by-case basis.
Another issue concerns the mechanisms available to transport the component information? In other words, should the information be human-readable, computer readable, or both? Because ECIX is migrating to the XML format, it will ultimately support readability by both humans and computers.
Another area that needs to be addressed in more detail relates to system design and analysis. There is a need for methodologies and tools to support system analysis (performance, timing, and fault analysis) for a large system based on the components being proposed for the system. Evaluation of a proposed component's effect on a system, and the system's effect on the pro posed component, is an important aspect of the design process. An effective ECCI methodology must provide users with this system-oriented data as well as the com ponent and sub-system oriented data.
This panel highlighted the urgency of getting the component suppliers to commit to providing data in a standard format, such as ECIX. The attendees recognized that the workshop lacked a critical mass of component suppliers necessary to build consensus. The Si2 representatives indicated that they were already beginning a project to develop this consensus by organizing the user community regarding their requirements for a minimal set of parameters that they require from the suppliers. The next step would be to organize a series of meetings at which the users could present their needs to the suppliers in an effort to build the necessary consensus. NIST committed itself to supporting this effort, as did several of the industry participants.
Panelists:
Curtis Parks — NIST (Chair),
Mealnie Yunk — Si2,
Jean Lebrun — Thomson CSF (CIREP),
Steve Waterbury — NASA,
Takeshi Fuse — Fujitsu (VSIA),
Mitsuru Takahashi — Hitachi (E-CALS).
The focus of this panel was to explore current efforts in the standards domain that have a relation to ECCI, including the represented projects and others. Interoper able standards are critical to the success of component information exchange. In addition standardization promotes stability, interoperability, known interfaces, and the ability to more easily address higher-level system complexity.
While there was no specific general discussion period for this panel, several important discussion points were made throughout the panel presentations. These points are listed here:
Based on the input from the workshop, the NIST Electronic Information Technologies Group (EITG) will work with Si2 to determine if the current dictionary translator software can be extended to handle JEDEC's dictionaries. NIST, Si2, CIREP and ECALS representatives agreed to plan on a follow-up meeting in November or December of 1998 to begin defining the details of an international interoperability testing project.
Most of the participants elected to present information and/or demonstrations of their activities in the area of electronic commerce of component information. On Wednesday evening, July 15, time was provided for these presentations/demonstrations, which included presentations by the following organizations: IBM, ECIX, Viewlogic, NIST, Si2, ORCAD, InfoQuick, and Digital Market, Inc.. The following companies offered to provide demonstrations upon demand: IHS, Questlink, ASC.
Panelists:
Jim St.Pierre — NIST (Chair),
Steve Weitzner — EE Times,
Preet Virk — ViewLogic/Synopsys,
Steve Thompson — NIST,
Betsy Dunphy — IBM,
Allen Hefner — NIST,
Ron Mayer — Digital Markets.
A key point brought out during this panel discussion was that consistency among the information provided by different manufacturers is very important. Since contract manufacturers can deal with hundreds or even thousands of manufacturers and suppliers, the more consistently the standard is used, the more they will benefit. Some participants also noted that it would be very bad if every supplier used a different dictionary. On the other hand, there are good reasons not to force the use of a single dictionary. In fact, it is a necessity in the case of new technology development, because the developer of a new technology will often define new terms to describe the technology. There seemed to be some consensus during the workshop that JEDEC is most qualified to create a good dictionary for memories. Recommended dictionaries for certain types of components, while not required, might be helpful. Another sugggestion made was that, if many different organizations use many different dictionaries, perhaps the information services companies such as IHS and Aspect could play a role in translating dictionaries, or providing a mapping between dictionaries. This would allow users to compare two components from different manufacturers, that reference different dictionaries to define their terminology (e.g., IEC and JEDEC). Widespread acceptance or recognition of a dictionary would be important before industry would adopt it. A lot of people expressed interest in a freely available dictionary.
Component manufacturers currently provide different levels of information to different customers. In general, they are not yet making comprehensive information (e.g., simulation models, thermal data, geometry information) for the majority of their components generally and freely available via the Internet. It is hard to tell who in the industry — component manufactures, distributors, information-service-companies, or component customers — will be the driving-force for the acceptance of these standards. These groups will need to create the electronic component information before other people can use it. However, OEMs will play a critical role by showing a preference for suppliers who utilize these emerging standards.
Currently different organizations, and even groups within an organization, communicate component information inconsistently. One opinion: it would be helpful to have some “key” that could be used to refer to related information in other systems (e.g., EDA, PDM, ERP, MRP). This discussion about a “key” deteriorated into a part-number focused debate; but a few interesting points about part-numbers were revealed. Not all organizations use or communicate part numbers in a consistent way. Even a single part in a single bill-of-materials can have many different useful part numbers. For example, during design engineers refer to a part number that may map only to a component's electrical and physical properties. However, the purchasing department refers to the same component by it's “orderable part number,” which also specifies how it's shipped (in a bag or a box, etc). In the debate that followed, many different concepts were discussed (manufacturer's part number, orderable part number, internal part number, end-customer part number, mythical-universal-registry-of-part-numbers, and a few others). The number of different approaches to this problem only adds to the confusion. Many organizations do indeed use part numbers to communicate component and bill-of-material information even though that may not be the best way to do so. Resolution of the part number naming and equivalence problem is very difficult. One suggestion that had some merit was for providing a “mapping” between the various types of part numbers for a particular part. This approach would not require a single unique “key,” but a key could be assigned to each set of part numbers that are “linked.”
While electrical and mechanical properties really are properties of a component itself, “business properties,” (such as “price,” “lead time,” and “availability”) may be more a property of the relationship between the organizations involved than the part itself. Possibly, these properties may be out of scope for the ECIX standard; attempts to standardize them now could be a long process and may not meet everyone's requirements.
The main suggestion for a follow-on activity for this panel was to pursue mechanisms for improving the ability of end-users to process a bill-of-material. It was generally noted that this a very difficult problem, how ever it would provide cost savings to the industry if it could be solved. NIST and the IPC agreed to have follow-up discussions regarding this problem. In addition, there were discussions between a NIST Manufacturing Extension Program (MEP) representative and some of the workshop participants regarding ways that small and medium sized enterprises could get improved access to component information, and in particular, part substitution information. One proposal for solving this problem was a pay-per-use system for looking up component information. Such a system would minimize the cost issue for small and medium sized enterprises which can not afford to purchase an expensive component database system.
Two discussion sessions were held regarding forward plans. The first, when most of the participants were present, took place on Thursday afternoon. The second, when only a small cadre of participants remained, took place on Friday morning.
The first discussion session was organized around major conclusions by the attendees that the workshop give direction for future work. The second discussion session specifically addressed proposed work that can and should be done by Si2 (ECIX pilot study methodology), by IEC, and by IEEE. The following key points were raised during the Thursday meeting:
As a result of this workshop, a follow-on meeting was held in Tokyo Japan, at the EIAJ headquarters, on December 8, 1998, to discuss interoperability testing be tween the ECIX, CIREP, and ECALS projects. A U.S. proposal for testing was presented at that meeting. The ECALS and CIREP representatives indicated that they needed more time to evaluate the proposal. A schedule was established for submitting comments on the proposal, modifying the proposal, and reaching consensus. A mini-workshop has been tentatively scheduled for March 1999, in Europe, to achieve these goals.