standards
Data Exchange Was Dead. Then a Deal Was Forged.
A decade in, IPC-2581 Consortium members say the pursuit toward widespread adoption of the vendor-neutral standard was well worth the rigorous effort. by Chelsey Drysdale
Few engineers working in electronics manufacturing today predate the first efforts to develop and implement an industry-wide standard for intelligent electronics data transfer.

As early CAD tools were introduced in the late 1960s and early 1970s, IPC launched a vendor-neutral effort to describe electronics design data from schematic through test.

Meanwhile, Gerber Scientific had developed in the 1960s the common generic (read: unintelligent) format, colloquially known as Gerber, to describe information sent to its photoplotters. In the early 1980s, Gerber adopted and adapted the format for broader printed circuit board manufacturing.

Not long afterward, other various standards bodies and private companies began work on their own formats. The names may ring bells among veteran readers: STEP, EDIF, GenCAD, and Valor’s ODB (whose successor, ODB++, is still used today).

IPC-2581 was first released in 2004 and was originally patterned after GenCAM. It is used to transmit information between a PCB designer and a manufacturing or assembly facility.

But acceptance was spotty prior to 2009. That year, Mentor acquired Valor, a watershed moment that spurred EDA companies to support IPC-2581 as an alternative to Valor’s ODB++. Motivated by the goal to ensure key standards would be developed by industry consensus, a group of ECAD companies formed the IPC-2581 Consortium in 2011. Within a year, Fujitsu had built a “proof of concept” board using only data sent in the IPC-2581 format, and the group was off and running.

The early consortium members shared their recollections of those early days with PCD&F in September.

Consortia members in a stacked and seated group photo
Figure 1. Consortia members were recognized by IPC in 2014.

Hemant Shah (IPC-2581 Consortium Chair, then program manager at Cadence): Prior to 2010, most customers used Gerber for data transfer to manufacturing. This met the requirements for most PCBs, as high-density, more complex (e.g., flex, embedded) PCBs were not yet prevalent. Some large companies had started to use and depend on ODB++, a more intelligent data transfer format from a (then) small company named Valor. Valor worked closely with all PCB EDA vendors and manufacturing companies, and, like Gerber, ODB++ was considered EDA vendor-neutral.

headshot of Hemant Shah

Hemant Shah

Gary Ferrari (then IPC Designers Council executive director): In the early years, we were dealing with companies whose workers were primarily chemical experts. They were not as savvy with computers and felt very comfortable with simple software such as Gerber Scientific provided. The Gerber software was very basic and simple to operate. It told its hardware to open a round aperture, and it shined a white light through the hole and thus created a round image on photosensitive Mylar. If you wanted to make a straight line, one opened the aperture and dragged the light beam across the photosensitive Mylar to create the line.

That’s it, very simple. As long as the design was simple, this worked.

headshot of Gary Ferrari

Gary Ferrari

Joe Clark (cofounder, DownStream Technologies): The primary factor was complacency – or maybe a better term is “risk aversion.” Unintelligent Gerbers … worked, so no one saw the need to change, and the same format was often the exchange between the OEM and the CM, with the addition of a parts list.
Keith Felton (then group director, product marketing, Cadence; now product marketing manager, Siemens Digital Industries Software): The need wasn’t that strong for a new standard because a de facto one already existed.
headshot of Keith Felton

Keith Felton

Hemant Shah: Adopting a new data transfer format never stuck because it required an ecosystem of software suppliers, design houses, PCB fabricators, assemblers, and testers. This ecosystem of supporters didn’t exist, and no one tried to get them all together in an organized fashion.

Gary Carter (then CAD manager, Fujitsu; now independent consultant, ThingWeaver Solutions): There was a recession and little industry support for IPC-2581. Not many of its original authors and creators were in attendance when I first joined the IPC committee and actively participated.

Too Proprietary
Rick Almeida (cofounder, DownStream Technologies): Note Gerber is not actually a standard; it is a proprietary format owned by Ucamco. Likewise, the first true “intelligent” data transfer format – ODB++ from Valor – is also a proprietary format. The major difference between Gerber and ODB++ was licensing. [In other words,] a company had to be licensed by Valor to utilize the ODB format. This means there was always the possibility – regardless of how remote – that one’s license could be revoked or not renewed. And what will become of the format if Valor ceases to exist as a company, or at least as an independent company? The marketplace – the ultimate decider – has a long memory of past formats that came and went – HPGL, EDIF, etc. – and was understandably skeptical of ODB++.

Jamie Wise (vice president, Wise Software): Besides being a single file, open format, one of the needed features of this new standard was the additional safeguards for IP.

Terry Hoffman (technical leader, Cisco Systems): My guess is other standards were not really a standard, or they were proprietary, which would make them difficult to adopt or gain traction.

IPC-2581 Comes … and Goes

(Ed.: IPC-2581 was developed in 2004, but it took more than six years for the IPC-2581 Consortium to form and back its implementation.)

Gary Carter: The initial standard was developed over many years and released in 2004. However, the internet bubble had burst; industry was in decline, and few paid attention to its publication. Most of the original authors and volunteers were engaged after publication. Budgets were tight.

Joe Clark: It’s worth stating again Gerbers worked! And there were no external factors driving the need for change: technical or economical.

Technical factors as well affected adoption of intelligent formats by the mainstream market. For example, many companies are loath to send the entire design database to their offshore PCB foundries for fear of intellectual property theft, and because ODB++ did not support an XML version, there was not an easy way for users to send out subsets of the full design database.

An ‘Overnight’ Success

Joe Clark: The consortium … developed very suddenly, and the reason for this was due to a single significant event: the acquisition by Mentor of Valor announced in November 2009.

Hemant Shah: Mentor Graphics saw ODB++ as their IP and reinforced that on all companies that were consuming/producing it. Larger design houses that were using ODB++ but not using Mentor’s EDA tools were perturbed and requested their own EDA vendors come up with an alternative solution.

Keith Felton: [There was] concern that ODB++ would stop being open and a de facto standard.

Rick Almeida: This was a game changer, and those EDA companies – the majority at the time – that were dragging their feet on support for this new format now had a strong reason to support [it]. A major competitor now owned the company that owned the proprietary ODB++ format that most were supporting via a license agreement. What now? How will the technical issues be addressed? What input will be allowed externally from Mentor for new versions of ODB?

In our view, self-preservation played a key role in the sudden creation of the IPC-2581 Consortium and its missionary zeal to make IPC-2581 the format of choice for all. The early members of the consortium were EDA companies, along with a few OEMs, most notably Fujitsu, but companies across the PCB design and manufacturing spectrum quickly came on board.

Gary Carter: At a Cadence Design Systems Technical Advisory Committee meeting, a top 10 list voted on by industry identified IPC-2581 adoption as a priority. We subsequently solicited membership across the entire value chain to join our newly conceived consortium specifically aimed to promote and support this standard. Prior to its formation, there was no proactive industry consortia focused on open-source collaboration for this particular IPC standard.

The Consortium Forms

Hemant Shah: The IPC-2581 Consortium was created in the summer of 2011 based on an industry need to have a nonproprietary standard to hand off PCB build intent to manufacturing.

We started with 12 founding members. (Today over 100 companies support the IPC-2581 Consortium.) The founding member companies believed in having an open, neutral standard not controlled by any one company. Everyone was collaborative. The challenge that the consortium faced was there were no software vendors that produced IPC-2581. Therefore, there were no software vendors that consumed IPC-2581. It was a chicken and egg situation: What comes first?

Gary Carter: The initial consortium meetings were filled with excitement and support for further development and implementation of IPC-2581. It was surprising to discover several companies already had implemented and utilized parts of IPC-2581 through the earliest tools that were available to import/export in IPC-2581’s earliest published format. Little of this had been revealed publicly for a variety of reasons. Ideas for further expansion to fabrication and test soon emerged.

Terry Hoffman: The first few meetings after I got involved in the consortium were a learning experience for me. I had never worked outside my company to develop a standard, and I was getting familiar with the people involved and the standard, and learning the IPC standards processes.

Gary Carter: Clearly there was a vast difference of opinion on industry’s willingness to openly collaborate. It took time to plant those seeds. Selling this to Fujitsu was also a challenge. We needed better tools to automate our design realization flow. We managed to win executive support to forge ahead.

Hemant Shah: The collaborative nature of the members, many of whom competed in the open market with each other, fueled the innovation to break the chicken-and-egg situation to create the first and only open, neutral industry standard for stack-up exchange. The passion and vision of the late Dieter Bergman for IPC-2581 was inspiring for many members who had the honor and privilege to work with him.

Selling the Public
[Ed: The consortium made its first public appeal to design engineers at an open session at PCB West in 2012.]
Jamie Wise: Once we finished verifying the process between CAD and CAM, we were ready to move to the next step in building a board using IPC-2581 data. Chris Shaw of Fujitsu provided the design, using Cadence Allegro to export the design in IPC-2581 format. Wise using VisualCAM prepared the files for manufacturing (design comparisons, panelize, pad removal, tear dropping, DfM analysis). John Dingley of JD Photo Data did the photoplots and worked closely with Phil Wain from CCEE to manufacture the boards. CCEE also ran electrical tests for opens and shorts, and AOI on the innerlayers to prove the boards were fully functional. Once Phil notified us the boards shipped, there was a lot of excitement. Chris (Shaw), Gary (Carter) and Ed (Acheson) couldn’t wait to get their hands on the actual board.
front view of Fujitsu test board
back view of Fujitsu test board
Figure 2. The Fujitsu test board.
Hemant Shah: It was a milestone for the consortium to talk to everyone about the adoption of IPC-2581. Almost everyone we talked to agreed there should be an open, neutral format for PCB design handoff to manufacturing. There was never an argument about using IPC-2581. The first challenge was awareness; the second challenge was to get the software vendors on both sides – the design side and the manufacturing side – to support it.

Terry Hoffman: I remember the PCB West forum that I just happened to drop into and argued against IPC-2581 in the beginning and then was for it at the end.

Gary Carter: It was a packed session with a very interesting panel debating the pros and cons of the IPC-2581 approach. The spirited dialog spurred a great deal of discussion and follow-up engagement with industry players. This brought a new group of supporters to the consortium and firmly established our path forward.

Rick Almeida: There was clearly a lot of interest. Like all new things, acceptance and adoption often start with small steps, and this was one of the first steps.

Off and Running
Gary Carter: Design tool companies were the key to getting out of the starting gate. Cadence, Wise and DownStream were instrumental in making that happen. Once we had the design tools available, IPC-2581A and B were utilized to prove a large part of the standard’s capability to create value across CAD and CAM. This led to the fabrication of the first set of boards.

Even those who opposed this open standard – i.e., those in industry who had proprietary products that competed in this space – were beginning to see their customers would benefit from its openness, and some customers were pushing them to embrace it. Design, design verification, and manufacturing assembly were the first to embrace it. Fabrication and test soon followed.

Hemant Shah: Also, winning the DMDII [Digital Manufacturing and Design Innovation Institute] contract to get government funding for supporting IPC-2581 adoption.

Terry Hoffman: The event that stands out the most with IPC-2581 was the release of revision B. That release was the one that would enable my company to actually use the standard. That is when I started the long process of validating the IPC-2581 data in my very conservative company.

Joe Clark: Many milestones in the development of the IPC-2581 format were important and merit listing. However, for us a personal milestone stands out, and that is when we were able to demonstrate to Dieter Bergman – the true champion of the IPC-2581 format – the ability of several EDA tools to exchange the format.

The Fujitsu Proof of Concept

Gary Carter: This was the culmination of our effort to release Rev. B. It drew the attention of many producers and consumers of product data in the PCB industry that were already involved in development and adoption of IPC standards, as well as those involved in the development and marketing of proprietary CAD/CAM/PDM/PLM solutions. A few others read about it in blogs and social media venues. The battle in the fabrication and test space continued to hamper progress, but there were inroads forged there as well. PLM companies began to grasp the value of a digital product structure. We implemented this at Fujitsu and demonstrated how the standard supported it.

Hemant Shah: The companies involved were elated. Manufacturing companies were surprised that IPC-2581 indeed had everything they needed to build the board without any notes or back and forth with the design house customer. Three manufacturers built a real 12-layer product board from Fujitsu Network Communications.

Rick Almeida: This event was a significant step in removing the veil of doubt that many users still harbored. It was not the end of the work that needed to be done to ensure the adoption of the standard, but the beginning of the awakening of the market to its possibilities.

Jamie Wise: One of the most important milestones was when companies like Fujitsu started using IPC-2581 data exclusively for the whole design to manufacturing process and when Axiom set up its assembly “paperless factory” using IPC-2581 data.

Keith Felton: It showed the supply chain was capable of consuming IPC-2581.

Making a Difference
Terry Hoffman: The IPC-2581 Consortium has made a significant difference over the years. It is interesting how some of the features discussed in the consortium meetings have actually been implemented in other data formats after those discussions. So, the good ideas we came up with over the years have not only benefited IPC-2581, but also other data formats as well. Even so, we have seen IPC-2581 being adopted by many companies because of our diligence, our drive, and listening to people’s ideas and requests.
Chris Shaw of Fujistu opening the box with the first working IPC-2581 PCB

Figure 3. Chris Shaw of Fujistu opening the box with the first working IPC-2581 PCB.

Gary Carter, at right, sits next to Dieter Bergman in a full industry meeting conference room
Figure 4. Gary Carter, at right, sits next to Dieter Bergman at an industry meeting.
Gary Carter: More companies have plans in development or are in testing phases of internal implementation. It’s slow, but it’s happening.

Joe Clark: One can argue we dumbed ourselves down in our knowledge of the manufacturing process starting in the 1980s when we farmed out manufacturing to third parties. Some recent papers identify the adoption of an intelligent design exchange format as one of the keys to improving the new product introduction process, but given the results for GenCAM, which lacked any real support from the EDA market, the IPC-2581 Consortium has without a doubt generated awareness of and helped ensure the success of the IPC-2581 format.

Keith Felton: Adoption today isn’t that high because the pressure to use a digital twin process still isn’t that high. People continue to send Gerbers, and fabricators still waste their time cleaning them up due to competitive pressure. Given that most fabrication happens overseas, companies are wary of creating a single/complete product model to minimize IP theft. Companies that have bothered to optimize their process have achieved significant benefits.

Gary Carter: In many large companies there exists an aging workforce who are using a plethora of proprietary tools, processes, standards, and procedures. They also have established or entrenched “ownership” over the legacy methodology that had been in place for decades. Even if/when they do recognize the value proposition of the digital standards IPC has in hand, they often do not have the money, time or interest to invest in it. Or the people involved don’t have leaders who are willing to assume responsibility. Simply stated, “It’s not my job!” So, they continue doing what they have always done to toss today’s document-centric work over the wall and out the door.

10 Years After

We asked our panel, “If you were told 10 years ago that you would still be at this now, would you have committed to it?” Their responses were overwhelmingly supportive.
Gary Carter: Absolutely yes. This effort was a tremendous challenge, and the cooperative work led me to interact with so many different people and professions throughout our industry. Fujitsu and IPC opened the door and made it possible. Many others in industry joined the initiative. The collaboration was amazing to see and rewarding to do in so many ways. I consider my effort promoting this standard to be one of my greatest professional challenges and achievements. I would not want to have missed the journey. My only wish is for Dieter Bergman to have been here with us to enjoy the fruits of his labor. I miss that man!

Terry Hoffman: It is probably good I did not know in the beginning how long I would be involved. However, I am the type of person who likes to follow through until a task is completed. I obviously believe the standard is beneficial and should be adopted, or I would not have been involved for a decade. The people in the consortium I have grown to know and befriend over the years have made it easy for me to continue to work with the consortium and IPC.

Keith Felton: No standards initiative is ever done. It’s a constant evolution as technology advances driving change, and I think the effort was worthwhile.

Hemant Shah: We all want instant gratification, so a 10-year horizon would have been unimaginable at the start. However, the journey of collaboration and innovation is what keeps us all going. The IPC-2581 Consortium is unique and is proud of several innovations. Some of them worth mentioning: The first data exchange format built from the ground up to support all PCB production sub elements – BoM, fabrication, assembly, test, stencil – in the same file, so eliminating sync issues between those elements; the XML-based format allows for interfacing with, or augmentation from, external systems, such as PLM, ERP, and MES systems; bidirectional stack-up exchange between design house and manufacturing house eliminating late-stage surprises for getting the product out the door; bidirectional DFM exchange between design house and manufacturing house accelerating NPI.

Rick Almeida: Yes! Nothing worth doing is easy, nor garners instantaneous success. And there are setbacks. We are piggybacking on the vision and efforts of Dieter Bergman in this regard, and his example of continued diligence is serving us well. (CD)

Gary Ferrari: I teach designer courses and … I’d say that 70% are still using Gerber.

Jamie Wise: There was virtually no support for IPC-2581 10 years ago. Today we have all major CAD, CAM, and manufacturing and assembly companies using IPC-2581 in their daily processes.

Hemant Shah: IPC-2581 has a huge following. Several companies have standardized on IPC-2581. They are not going back. Manufacturing companies have provided incentives to their customers to provide IPC-2581. The approach that has worked is manufacturers quote a specific price if customers provide IPC-2581, and if they don’t, and provide Gerber-based packages, the price goes up.

Gary Carter: It takes top-down executive sponsorship to create and lead the complex, interdisciplinary cross-functional team required to successfully lead digital transformation. There are lots of moving parts. Training is required. Risk-adverse human nature also plays a major factor here. Overcoming these barriers takes a masterful plan and buy-in from all parties.

Industry as a whole must change. It takes visionaries to transform this epic tale of weaving a stream of digital threads into the complete digital tapestry. That is where the goalpost stands and where the greatest opportunity to introduce disruptive technology lives at this moment. I believe younger engineers across our industry will be the ones who bring in the drive and the talent to get us across the finish line.

What any industry needs are new ideas, and one of the benefits of a standard intelligent design exchange format is it opens the door for small independent companies to innovate and create new tools, and competition is a good thing on many levels! In the past when formats were proprietary, it was nearly impossible to plug into an existing process, and this stifled innovation, a death knell for any industry. So, in addition to all the other benefits of IPC-2581 discussed, it also enables opportunities for new ideas and new tools to be created that can plug into an existing process flow.

Chelsey Drysdale is senior editor at PRINTED CIRCUIT DESIGN & FAB; cdrysdale@upmediagroup.com.