Looking Beyond PC/104-Plus

An Interview with Embedded Systems Analyst Eric Heikkila

By Geoffrey James

The PC104 family of stackable systems remains a small but highly interesting segment of the embedded systems market. The world’s foremost expert on this market is probably Eric Heikkila, the director for embedded hardware systems at Venture Development Corp., a Boston MA-based market research firm. Heikkila holds a BS in Electrical Engineering with a minor in Economics from Bucknell University where his studies focused on electrical control systems, digital system design, and electromechanical energy conversion & power systems.

Heikkila also a co-author of the 14th edition of “Merchant Computer Boards for Embedded/Real-Time Applications,” the industry’s standard reference on market size, share and forecasting for this segment. In the report (published February of 2007), he estimated that the PC/104 family of stackable modules would reach approximately $264 million in 2007 and would achieve a 9.51 percent Compound Annual Growth Rate (CAGR) from 2005 to 2010, far outpacing many other high tech segments. We asked him about the report and how the PC/104 segment is likely to change in the years ahead.

To meet all of specifications, however, the resulting product would definitely increase the engineering requirement for high-density-interconnect (HDI) technology. This fact invited concerns about higher-complexity design processes and subsequent increases in the learning curve, performance-qualification tests, and production yields—as well as subsequent fears of increased product costs. Plus, the designers would need to work with a vendor that was capable of manufacturing HDI printedcircuit boards (PCBs).

GJ: What has changed in the market since you published your report?

EH: Not a great deal in terms of the overall dynamics of the market. The growing economic uncertainty looks likely to have a negative impact on sales for 2008, flatting out our original forecast somewhat. However, we believe that most of the industries that the PC/104 market serves are reasonably resilient and so we’re sticking to our projections for 2009 and 2010. In the short term, we expect cost pressures to create somewhat more demand for PC/104 with USB, which costs less than implementation using the PCI bus. It won’t have a huge impact on unit shipment, but it will definitely be measurable.

GJ: The PC/104 family comprises about half of the stackable embedded systems market. Since PC/104 represents a standard, why are there still so many alternative architectures?

EH: There are specific applications that require features that PC-104 doesn’t support well. For example, with the EPIC architecture, you have more space on the board, which allows you to package more functionality into the entire system. There are also cases where there’s a need for a larger board to better deal with heat dissipation than is generally available on a PC/104 system. Because of this, we believe that these alternative architectures will continue to exist for some time to come.

GJ: Standardization usually drives market consolidation. That hasn’t happened with PC/104. How come?

EH: PC/104 tends to appeal to niche markets, specifically military/aerospace and industrial systems. All of these applications tend to be small in unit volume and require a significant amount of customization. PC/104 vendors must therefore be able to engage closely with the customer and make changes as necessary to meet customer needs. What’s emerged, then, are a large number of relatively small firms, all specializing in particular application areas.

GJ: Why hasn’t the customization moved up a level of abstraction? You’d think that some of the customization could be accomplished in software rather than hardware.

EH: With PC/104, the “secret sauce” that justifies one vendor over another is typically how that vendor handles I/O. Making changes to the I/O capabilities of a system typically requires making changes at the system level, an activity that always means a certain amount of custom manufacturing. If it were possible to make these kind of changes purely using software, rest assured that somebody would be doing it.

GJ: Do Embedded System on Chips (SoCs) represent a competitive threat to PC/104?

EH: To a certain extent at the very low end. However, most PC/104 systems have I/O requirements that cannot easily be met in an SoC environment. In addition, PC/104 systems generally require customization, which is more difficult to do in a SoC environment without undergoing the expense of designing a new ASIC. Therefore, I think that the impact of SoC on the PC/104 market will be fairly limited.

GJ: Why is USB so important for PC/104?

EH: USB enables faster data transfer rates than the old ISA bus without adding as much extra cost as the PCI bus. While PCI remains faster, USB is more cost-effective, which is why we believe it will be a valuable tool in the PC/104 tool kit.

GJ: Why should cost be important, if there’s so much customization going on?

EH: It’s true that the PC/104 market is not particularly pricesensitive, because much of the expense of a system lies in the specialty work and customization that’s required for most applications. However, while the price of the hardware is not a primary concern, it is still a concern. All other things being identical, the ability to offer a functional system at a lower price than a competitor is definitely going to influence the selection of a PC/104 vendor.

GJ: What’s next, beyond USB?

EH: It’s pretty clear that PC/104 with a PCI-Express bus is the next generational step because it will allow data transfer rates in excess of anything that’s available today. Unfortunately, there’s some controversy surrounding how PCI-Express should be implemented, with some companies doing their own early work that’s incompatible with the work from other firms. However, we expect the standard to eventually settle down and provide a platform that’s likely to remain useful for five to ten years into the future.



GJ: Do you see PC/104 penetrating into additional markets?

EH: Today, PC/104 tends to sell into military/aerospace and industrial segments. We believe that there’s some limited opportunity beyond these niches in segments like transportation (e.g. a controller on a high speed train) and medical devices (e.g. a controller on a portable MRI machine). However, there are some markets, like communications, where PC/104 simply doesn’t provide sufficient bandwidth to be particularly useful. So we expect PC/104 to pretty much remain in its niche markets, although we do expect those niche markets to continue to grow.

GJ: Will there ever be more standardization in the PC/104 segment?

EH: Overall, the trend in embedded systems is towards greater standardization. As the technology evolves, system manufacturers are learning to put more functionality into embedded systems. More functionality crammed into a system means that there’s less need for customization, thereby making it more practical to use a standardized architecture. That being said, PC/104 is likely to resist the pressure to standardize because the applications tend to be limited to very small market niches. In addition, PC/104 tends to be used in environments where requirements tend to be strict and inflexible.

GJ: Wouldn’t there be some benefit resulting from greater standardization, like economies of scale in manufacturing the systems?

EH: With military and aerospace contracts, the emphasis is on getting it right, not saving money or cost savings in the manufacturing arena. That being said, even PC/104 will not be entirely immune to the standardization trend. There’s an overall trend in Military and aerospace purchasing to use of lower-cost commercial off-the-shelf components and systems whenever possible. As a result, pressure may develop to standardize around a smaller number of PC/104 implementations.


Geoffrey James is a frequent contributing writer for Embedded Intel® Solutions magazine.