Make the Move to Modular

Whether it’s the increasing demands on control and visualisation electronics or the evermore- complex embedded computer kernels, development processes are feeling the pressure. Add in globalisation’s effect on price points and shortened innovation cycles, and it’s easy to see why companies must improve the efficiency of their development processes to stay competitive.

One way to break out of the eternal circle of electronics development is by modularising the in-house, customer- specific, embedded computer technology from a single mould. Modularisation separates a monolithic whole into pieces called modules, components, or building blocks. You can reap numerous benefits by going modular:

  • A single module enables many variations of the whole.
  • The more modules are used, the greater the diversity.
  • Leaving out or adding a module creates further variation.
  • Old modules can be replaced to increase the lifetime of a modular platform.
  • Modules reduce maintenance/ replacement costs.

When constructing a modular system, the modules need to be standardised and offer standard interfaces. Company internal standardisation of device, machine, and system interfaces already offers a number of advantages.

However, when the interfaces don’t draw upon the company’s own core competencies, greater advantages are possible when using standardised modules from external sources. For example, development expenses are lower, time-to-market is reduced, and design risks are lower. In addition, such modules are available at competitive prices. OEMs are able to develop series variants cost-effectively and increase the speed of product cycles. Moreover, listing the modules reduces the time and energy spent on documentation.

Examples of successful modularisation can be seen in the platform strategies used throughout the automotive industry (e.g., at Volkswagen). PC technology is another example—it incorporates modularity via expansion boards.

Processor logic, however, is mostly monolithic. This is usually sufficient for applications that need the board’s entire functionality, but not for custom-designed boards that need their own interfaces. For such boards, it makes more sense to modularise the processor and chipset logic to increase an application’s performance spectrum beyond the processor socket without replacing the entire board.

Computer-On-Modules offer the right components for every performapplications ance class. ETX 3.0, COM Express, DIMM-PC, X-board, and E2Brain are examples of standard Computer- On-Modules developed by Kontron that have acquired global acceptance, including PICMG standardisation (COM Express).

The de facto standard for PCIbased solutions is ETX 3.0. This module even covers ISA expansion boards and supports the latest serial technology (e.g., SATA or USB 2.0). The ETXexpress COM Express module and its smaller brother, the microETXexpress, plus the latest nanoETXexpress (Fig. 1), are standards for PCI Express-based solutions or for new PCI designs with advised migration to PCI Express.

These standards have become wellestablished over the last few years, even though the market for complete, tailor-made solutions is bigger in terms of number of units than the market for applications with Computer-On-Modules. Today, around 70% of the market is still occupied by full-custom designs, which means that Computer-On- Modules have captured around 30% of the full-custom design market (the market for merchant boards can’t be included in this breakdown).

The use of modules will, however, significantly increase in the future. That’s because PCI Express, multicore, and the increasing integration of chipsets demand greater development time as well as multiprocessing. Virtualisation of software applications will become the main challenge in coming years.

Market researchers like the Butler Group predict that the virtualisation of IT infrastructures will be the dominant technology. By 2010, hardwarebased virtualisation, para-virtualisation, and virtual operating-system environments will be state-of-the-art— inside and outside the server room. Companies need to prioritise these tasks as strategically important and significantly increase the development of modular, standard hardware to remain competitive.

GRAPHICS: THE FIRST NEW BOTTLENECK

Further bottlenecks will emerge because users will want to implement increasingly complex graphic technology. This trend will arise from the widening spread of HDTV and high-resolution digital cameras, etc., which already makes high-resolution technology affordable for industrial applications.

Another factor influencing this trend is the greater use of graphicgenerating processes throughout industry. These will become affordable once processing power increases, enabling penetration of a broad, mass market.

Continue on the next page

Quality management via costeffective industrial optical systems has become huge. Radiology is also undergoing further automation. These represent just two fields of development for high-resolution graphic technology. A large number of machines and systems will be equipped with maintenance interfaces that, for example, are connected online via a high-resolution camera to a central helpdesk and are able to display in parallel high-end 4D animations from the maintenance manual.

This requires not only high performance that will become affordable in the near future, but also extremely fast graphics. This is why Kontron developed a new standard for embedded high-end graphics that, together with XGI, was published as an open standard this summer—the new “Universal Graphics Module” Standards UGM (Fig. 2).

UGM has been developed for long-term-available high-end PCI graphics. The new UGM standard is an economical option for highend embedded graphics requiring a short time-to-market. This is made possible by the use of interchangeable and scalable standard modules. XGI and Kontron offer the UGM graphics module specification free to third-party vendors.

In the future, the specification and use of the UGM brand will be controlled by the forthcoming UGM interest group, whose independent development of the specification will particularly benefit users. The main target industries for high-end embedded graphics are medical and industrial imaging, gaming and entertainment machines, pointof- sale/point-of-information terminals, commercial outdoor broadcasting, public facilities, and highend residential gateways.

Unlike conventional graphics cards that are plugged in at 90 degrees to the baseboard, UGMs are connected parallel to the baseboard. This saves space and thereby allows for extremely flat and very scalable designs. Even more importantly for users, UGMs offer availability of at least three to five years. Plus, the graphics functions, including drivers, are particularly quick and easy to implement in custom designs.

POWER MANAGEMENT: THE SECOND NEW BOTTLENECK

Coupled with the trend in highend graphics for embedded applications comes the field of power management. But is power management a core competence of OEMs? No. Power management is a solution that can also be bought.

Take, for example, the SMART Battery Management Platform called MARS (Fig. 3). MARS was developed for Computer-On-Module designs, but it can also be used for every other customer-specific design. In addition to management interfaces for intelligent batteries (SMART), the modular reference design offers power features such as an extended input voltage range for, say, industrial applications or in-vehicle PCs, connections for all ATX voltages, and fall-back battery support for online battery changes.

In the first instance, MARS will be offered as a complete evaluation board including the appropriate COMs. The Intellectual Property (IP) in the form of layout and circuit data (e.g., OrCAD) for individual solutions can be inserted into the baseboard layout via “Copy & Paste.”

EVERYTHING’S GOING TO BE MODULAR

“My vision is to offer the market a range of modules that is as comprehensive as possible,” says Dirk Fenstel, CTO of Kontron’s Embedded Modules Division. “This will ensure the transition from the full-custom inhouse development of embedded computer technology to a modular concept and enable developers to integrate the modules seamlessly into their application specific designs, either as additional purchases or, as with MARS, externally available IP.”

“We don’t want to compel developers to use modules. They need to decide for themselves which commercial- off–the-shelf modules they want to use for their dedicated designs. But it is important for developers to know that they can obtain modules or functional components for almost every development task from Kontron,” he continues. “This is the idea of Modulus Vivendi: each to his own module. In addition to the purely technical components, we are also working on other parts of our services that will benefit developers of embedded computer systems.

“For example, we offer baseboard design training that we will also offer as an online stream in the future,” he says. “And the Web-consultation with our chief designer of COMs is another important element that we will soon be offering. Finally, we would like to bring the circle of COM developers closer together in a community where everyone can benefit by sharing their know-how and experience and avoid doing the same work twice.

“Our ‘Boards & More’ service that until now has mainly developed customer specific baseboards will have in the future a significantly extended after sales service for our COM, graphic and other modules,” says Fenstel. “All of these efforts will provide developers with a comprehensive toolkit, enabling them to develop their specific applications extremely quickly, efficiently, and competitively in order to stay one step ahead of the global competition. This is the only way companies will be able to enjoy continued success in the future.”

See associated figure.

Hide comments

Comments

  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
Publish