Who's to blame for the long drought of board-level EDA design data exchange? My answer may surprise you.
As I said in my previous post, despite early efforts by industry founders, vendor independent exchange of board-level design data across EDA tools remains elusive. To gain additional insight into the cause of today's state of design exchange (or lack thereof), let us consider the three primary beneficiaries: EDA companies, OEMs, and semiconductor vendors (including their distribution channels).
Board-level EDA companies, especially industry incumbents, have had nothing to gain by embracing comprehensive design data exchange such as that envisioned by the Electronic Design Interchange Format. It is counterintuitive to expect vendors to embrace anything that could allow customers to migrate design data away from their solutions. The situation is a conundrum, because the EDA companies must be committed to supporting design data exchange for any efforts to be successful.
For OEMs looking to leverage expected-good semiconductor reference designs, the task of re-entering a PDF schematic into an EDA tool is nontrivial, prone to error, and tedious, especially for analog circuits. In addition, at today's higher operating frequencies, PCB layout becomes as vital to the design as the schematic itself. Entering data that has already been entered digitally by someone else is counterproductive and borders on absurdity.
The ability to identify reference designs quickly and import them intelligently would provide product development teams with another tool for improving response times (to meet changing market requirements) and for making critical, early-cycle design decisions. Engineers could import multiple reference designs into their EDA environment and augment them with application-specific functionality to support rapid prototyping efforts. Designers could also leverage the BOM-centric aspects of many of today's EDA tools to assess key supplier metrics such as lead time, second sourcing, and pricing.
Semiconductor vendors invest considerable resources into developing reference designs that feature their devices. To state the obvious, the component is useless (and revenue is not achieved) until the device is deployed on a PCB board and consumed by OEMs in volume. With few exceptions, people wait in lines to buy a product, not to buy a microcontroller, A/D convertor, op-amp, or transistor.
To foster adoption of their devices and to expedite customer time-to-volume, semiconductor vendors have been providing more complete reference designs, including data on PCB, embedded software, application software, and (in some cases) mechanical enclosure design. Anything that makes using the reference design and reaching production quicker benefits the semiconductor device company. Devices are often at a virtual parity with competitive offers. A more complete, intelligently deployable reference design could tip the balance in the selection process.
Semiconductor vendors and OEMs have the most to gain from board-level design data exchange, and I believe that they are the reason for the current state of affairs. These entities, despite being some of the largest EDA customers, have accepted the status quo with little more than a collective whimper. Yes, you read that correctly. The EDA companies are not the problem -- it's their apathetic and virtually silent customers.
Semiconductor companies are some of the largest customers of EDA companies. They are primarily consumers of IC domain tools, but they could have considerable influence on EDA vendors from a corporate standpoint. It is curious that the semiconductor vendors have not chosen to wield such influence. Perhaps there is a disconnect between development and marketing. The domains of IC design and board-level design are considerably different, so it may be that there is an awareness gap within management ranks. Suffice it to say that, with its influence, the semiconductor industry could have made considerable headway into enabling sensible design exchange decades ago.
Personally, I'm a big fan of standardized board-level design methodologies and tool flows. Customers have made considerable investments in libraries, design data, user fluency, etc. I'm not advocating a wholesale standardization on a common foundation for data exchange, but I do advocate a way to import fundamental schematic and board designs into the chosen EDA toolset.
The near-term solution is to establish an export/import format for schematics and PCBs that is limited to the point of preserving the EDA company's business interests while supporting intelligent importation of published reference designs. The format may be restricted to a certain number of components, for example, or to a single page. The idea is to enable the reference design author to publish meaningful work that can be used by the community. The format could be restricted to exclude data for downstream domains, such as signal integrity tools, 3D design, arrayed reference designators, or anything beyond the information provided today in PDF and Gerber.
The first step would be to obtain backing for the initiative from the semiconductor industry's management, as well as the management at key OEMs. Next, each EDA company should be queried as to the most appropriate format that could be utilized by its particular database. The results should be consolidated into a workable vendor-neutral data standard. Note that the EDA vendors would not be required to disclose any information about their highly guarded data formats -- only what type of incoming data definition would be most suited for import into their databases.
This framework is likely an oversimplification, but I believe a solution is possible in a compressed timeframe, though not without outspoken support from the semiconductor industry and OEMs. In my next column on this topic, I will provide a vision for what this fundamental though limited design data exchange could enable for the industry. In the meantime, I look forward to your comments.