ODB++ spec tapped for CAD-to-CAM data exchange

ODB++ spec tapped for CAD-to-CAM data exchange
SAN MATEO, Calif. — The National Electronics Manufacturing Initiative Inc. (Nemi) is recommending that Valor's ODB++ technology become the starting point for building a single, industry-standard CAD-to-CAM data exchange format, EE Times has learned.

Jim McElroy, Nemi's executive director and chief executive officer, will formally present the recommendation at the Apex Conference in San Diego later this week. At that show, the Institute for Interconnecting and Packaging Electronic Circuits (IPC) will hold a convergence meeting, hoping to recruit industry representatives to work on the new effort.

Although industry participants got to vote on the two approaches, some in the losing camp are crying foul.

"I think it's a sham," said Dino Ditta, president of Router Solutions Inc. and a member of the Nemi convergence technical committee. "The thing was totally tilted toward Valor customers."

The recommendation ends a two-year effort by Nemi to mediate a dispute between those backing IPC's GenCAM format and proponents of Valor's ODB++ as a replacement for archaic Gerber files as the standard for transferring printed-circuit-board design data to test and manufacturing. When Nemi took up the task in late 2000, it established a technical committee and a management committee to explore how the technologies could be united.

McElroy said the technical committee first outlined what was required, from a user's perspective, to make an ideal exchange format between computer-aided design and computer-aided manufacturing data. The committee explored how the two technologies in question could converge to meet those criteria — but that didn't work because the pair varied too greatly, said McElroy.

The technical committee then created two subgroups to explore what technology was easiest to change to fit the user needs. The subgroups' findings were presented to a broad range of companies for a vote. The results, which showed overwhelming support for the XML version of ODB, called ODB++, were posted at the Nemi Web site last October.

In addition to Valor, companies supporting the ODB++ approach included Cadence, Hewlett-Packard, Lucent, Mentor, Nokia and Xerox. Those voting for the GenCAM approach included Agilent, Intel, Delphi-Delco, Motorola, Router Solutions and Sun Microsystems.

At the Apex Conference, Nemi will reveal the final vote: 72 percent for ODB++ to 28 percent for GenCAM.

Standard not finished

McElroy was quick to note that the final vote does not approve ODB++ as it stands today, but merely suggests that ODB++ technology should be used as the starting point to create a standard that does meet the user criteria. For that to happen, he said, several features from GenCAM must be integrated into ODB++, and IPC will head the effort to do so.

"This is in no way a completed standard," said McElroy. "The IPC will now have to do quite a bit of work to make the ODB++ technology meet those user specifications. We are pleased that we have gotten this far, but the standard now needs to be completed, and then the industry needs to proliferate its use."

Dana Korf, director of product engineering in pc-board operations at Sanmina/SCI, was on the project-management committee and will likely become a co-chairman of the standards effort, along with Henry Jurgens, manager of engineering systems at Celestica's corporate technology group.

Korf said that a key factor in the groundswell of support for ODB++ was its use at electronics-manufacturing services companies and printed-wiring-board fabrication operations. "It is the prevalent non-Gerber format," said Korf. "I think a lot of people feared using a brand-new standard, and that is why they didn't pick GenCAM as the starting point [for an industry standard]."

Ironically, though, XML-compliant GenCAM is finally ready for prime time, just as ODB++ is slated for revision.

Even though GenCAM standard bearer IPC will support the ODB++ work, this might not be GenCAM's swan song, said David Bergman, vice president of standards technology and international relations at IPC.

"There was a minority of companies that put a lot of manpower and effort behind developing GenCAM, and they prepped their companies to use the format," Bergman said. "They did a lot of the software development on GenCAM, and it remains to be seen if they will participate in developing the new effort."

Ironically, IPC will announce it is starting the standardization effort for the once-rival format on the same day it is releasing the final version of its long-awaited GenCAM data exchange format, raising the ire of those who spent many years developing GenCAM.

Ditta of Router Solutions said the decision could harm the industry. "It hurts us as a business — ODB++ is a terrible format that only Valor employees can read," he said. "The [Nemi] technical committees showed this to those that voted, but for some reason people chose to ignore or perhaps didn't understand our recommendation and they voted for ODB++ anyway."

Ditta said that his company — along with Intel, Teradyne/GenRAD (which donated an early version of the GenCAM technology to IPC, called GenCAD), Agilent and Motorola — were among the many companies that have labored over GenCAM's development for five years and hoped it would become the standard.

Ditta and other GenCAM backers said they are not throwing in the towel, and have started a consortium of companies that plan to use GenCAM — at least until another format is developed.

"We plan to sit in on the meetings [for the new ODB++ effort], but we won't likely be participating in it," said Ditta. "We have to support whatever the industry adopts, but in reality it is going to take three years to really develop this thing and we think it will take an additional two years to turn it into a formal standard. We need something that works in the meantime."

GenCAM, he said, is ready for mass use even as ODB++ enters the revision process. The latest release of GenCAM can be found at the new consortium's Web site.

Nemi's McElroy said he expects that those who have invested in each of the two formats will continue to use them until the converged effort is completed.

"The reality is the majority of industry still uses these arcane Gerber files, and they add on additional files for other parts of the manufacturing and test process. It is very sloppy and unorganized," said McElroy. "Our hope is that if we can get a majority of the industry to use this one converged standard, it will decrease the time it takes to get a product to market and save the industry a great deal of money. We believe the converged effort will get the industry to this end goal faster."

Hope for convergence

Celestica's Jurgens agreed. Even though "there are a lot of industry members in the supply chain as well as tool providers that put an investment in GenCAM" and have begun using it, he said, "What we hope is that based on the momentum we've built up with the Nemi initiative, we will be able to create a standards organization for the convergence and get broad industry participation. We think long-term that is going to solve a lot of problems."

IPC, for its part, considers GenCAM "finished," said Bergman. "From IPC's standpoint, we don't really need to do any work on GenCAM any longer." He said he hopes that most of the people who voted for a converged spec built on ODB++ will actively participate in developing it.

McElroy said that in the process of building the technical committee for the data exchange project, Nemi also recruited people who said they would participate in the development process.

"We have to take our hats off to Nemi for stepping in," said Randy Allen, Valor's vice president of engineering and customer support. "They were very successful in getting a large number of companies throughout the industry to participate and getting an unprecedented level of input on this topic. It is a very important milestone in addressing the data exchange problem."

Korf of Sanmina/SCI noted that the new group within IPC will concentrate its efforts on adding three features from GenCAM to ODB++ to make it a more complete standard. ODB++ lacks an ability to include basic business information, he said, such as contact information. Nor does it support schematics and test. All three of these features will be transplanted into ODB++ from GenCAM.

Korf expects the first release of the standard to be out by year's end but a fully ironed-out version won't be ready for another two years, he said.

Jurgens called industry participation "crucial," saying, "We would like to get a good, broad participation from all the key players, but we especially need good participation from the OEMs. The value of this is really going to fall into their hands. It will allow them to get products to market faster and allow us, Celestica, to give them better service and not spend time on data issues."