Skip to : [Content] [Navigation]
 

SYSTEMS DESIGN

Computer-on-Modules: Standardized Processing for Medical Electronics

Many medical electronics manufacturers choose to outsource key components of their devices. However, they should consider a new option for embedded computing systems.

Marc Brown and Matthias Huber

Computer-on-modules, as shown here, provide standardized form and function.
With ever-increasing speed requirements for modern processors and system buses, designing a cpu-based board from the ground up has become a daunting challenge for OEMs. Often, OEMs’ resources would be better served by focusing on their core competencies and product differentiation. Therefore, many medical electronics engineers are realizing that the fastest and most efficient method of bringing their products to the market is by outsourcing the manufacture of a device’s key components.

One of the components purchased most often is the cpu core. These off-the-shelf products offer a validated core with fully tested BIOS (basic input-output system) support, high-speed cpus, modern interfaces, and time-to-market advantages. Based on these considerations, many medical OEMs decide that buying cpu cores from a reliable vendor is the best choice.

But medical OEMs must recognize that selecting from the many choices of off-the-shelf embedded computing products is one of the most critical steps in the design process.

Computer-on-modules (COMs) can be used in a broad range of medical electronic applications where they fit mechanically, economically, and functionally. However, it is essential for an OEM to ensure that its computing supplier provides more than just a computer board. The supplier also must be aware and capable of meeting the extensive and evolving requirements of medical equipment manufacturers.

Specialized Requirements

OEMs have a lot to think about when designing medical electronic equipment. Timing and detailed information are paramount in the world of medical devices. In medical imaging applications such as x-ray, ultrasound, and magnetic resonance imaging (MRI) devices, the more doctors can see while examining a patient, the better equipped they are to help the patient. The need for high-resolution images that can be manipulated in real time has driven medical equipment developers to require better graphics, faster processing, and faster communications capabilities. In emergency medicine and rescue services, getting the right information at the right time is vital, and therefore real-time data conversions are essential.

In many embedded medical applications, size is also a consideration. And, as the medical industry advances, the requirements will likely change again. However, these applications must be able to be updated without starting completely from scratch, so scalability and upgradability must also be built into the design.

Beyond the performance and scalability needs, equipment designed for medical use must meet extended longevity requirements. This means that some applications will be expected to last as long as 10–15 years. As in all embedded applications, time to market is also a concern. Therefore, reducing implementation time, including the time allotted for FDA and other regulatory testing and approvals, is also a key factor. Finally, budgets for these products are limited, so costs need to be optimized whenever possible.

Embedded Computing Products

Designing and building a big, bulky piece of equipment that requires an extended development period, such as a full custom motherboard and enclosure, is not a viable option for medical electronics OEMs. Likewise, engineering, debugging, and validating a new single-board computer for each generation of processor and bus can be both an expensive and a time-consuming undertaking. In fact, a full custom design can take as long as 9–12 months to complete.

There are a number of embedded computing options available, including off-the-shelf motherboards, long-life industrial motherboards, and high-volume-application custom products. However, the COM approach offers some very distinct advantages, including high levels of processing performance and I/O bandwidth in a compact form factor. More significantly, COM products can help medical system designers achieve faster time to market, reduce development cost, and minimize design risk. In addition, they can enable OEMs to simplify future upgrade paths and scalability and increase application longevity.

COM Overview

The COM approach puts an entire powerful and complex computer on a small form-factor module that can be mounted on larger carrier boards containing application-specific I/O and power circuitry. (All generic PC functions, such as graphics, Ethernet, sound, USB and other serial ports, and other system buses, can easily be obtained in an off-the-shelf compact module.) A custom-designed carrier board can complement the COM with any additional functionality that might be required for a specific application. The carrier boards can provide all the interface connectors to attach the system to peripherals such as a hard disk, mouse, and display.

Standardized Form and Function. As the COM approach gains popularity, an open COM standard is needed to provide the advantages of modular, off-the-shelf building blocks. The benefits of standardization include the following:

  • Reduced cost. Mass production can lead to a better price performance ratio.
  • Improved quality. Mass production can lead to higher product quality.
  • Standard architectures. Standardization allows reuse of hardware and requires only software teams to develop new applications.
  • Scalability and flexibility. More module offerings can be applied to the same platform.

The Embedded Technology Extended (ETX) standard was developed to provide an open standard to meet the needs of embedded industrial applications with PCI and ISA. The original ETX standard offered a number of advantages including full PC functionality, minimum engineering and adoption cost, low cost, reliable connectors, slim design, and simple upgradability and scalability.

Modules that conform to the ETX standard are compact (95 × 114 mm, 12 mm thick), highly integrated COMs. Since all ETX modules feature a standardized form factor and a standardized connector layout that carries a specific set of signals, designers can create a single-system baseboard that will accept modules that comply with the current version of the ETX standard and those that will comply with future versions.

ETX modules can include a variety of common personal computer peripheral functions including graphics, parallel, USB and other serial ports, keyboard and mouse, Ethernet, and sound. Integrated drive electronics, peripheral PCI, or ISA buses can be implemented directly on the baseboard rather than on mechanically unwieldy expansion cards. This ability to build a system on a single baseboard using the computer as one plug-in component simplifies packaging, eliminates cabling, and significantly reduces system-level cost.

As technology has evolved, the ETX standard has undergone further development in scalability and performance. The new ETX 3.0 specification offers all of the benefits of the original ETX standard. However, it adds 2× serial advanced technology attachment (SATA) without changing any of the ETX pins, making new modules 100% pin-to-pin compatible with previous versions.

With new interfaces, such as SATA and PCI Express, the module definition needed a complete makeover. Kontron spearheaded the COM Express standard initiative within the PICMG group to develop a manufacturer-agnostic definition of COM. COM Express includes specifications for small-form-factor modules to satisfy the high-performance segments of the embedded industry. ETXexpress modules, which are COM Express products, are compliant with COM Express and allow the application of high-speed COMs for PCI Express bus and PCI Express chip set. The 220-pin high-speed surface-mount connectors for ETXexpress offer many performance capabilities. ETXexpress supports hardware based on current bus systems, such as 32-PCI and LPC (the ISA bus replacement), as well as up to 32 PCI Express configuration-dependent lanes, including PCI Express Graphics. Gigabit Ethernet, USB 2.0, serial ATA, and parallel ATA interfaces are supported as well.

Multicore Technology. The pressure to provide more features often requires trade-offs between performance, space, and power consumption. To further complicate matters, shrinking form factors can eliminate the option to increase the footprint to make room for more power or more performance. The good news is that advances in and increased availability of solid multicore processing platforms offer higher computing performance, reduced chip count, lower bill of materials costs, and reduced power consumption. For the same price and power allowance, a modern COM provides twice the performance than was available two years ago under the same price and wattage constraints. Also typical today in cpus is two engines in one, reducing the floor space usage. Cpus with four computing cores are already available and will likely find their way onto modules soon. For example, Intel’s multicore architecture provides a powerful and power-efficient platform that delivers increased performance-per-watt to provide greater processing per square foot while reducing the associated costs and risks of implementing new technologies.

Depending on the actual task, a dual- core cpu produces up to 200% performance compared with a single-core cpu at same module power usage. Module usage allows manufacturers to use proven-to-run cores. The challenge of getting a multicore processor board to work in a stable fashion is removed from the equation. So, instead of redesigning a board from scratch (6–9 months), a new module can be evaluated with an existing baseboard and released within 1–2 months. With an existing baseboard, the risk is next to zero, because there is always a working system to fall back on.

Because dual-core processors provide two complete execution cores instead of one, each core has an independent interface to the front-side bus in addition to its own cache. This allows an operating system with sufficient resources to handle intensive tasks in parallel, which provides a noticeable improvement to multitasking. Thanks to this parallel concept, which was previously available only on expensive parallel computers, dual-core processors make it possible to distribute tasks to several computer units, thereby duplicating performance. The power dissipation of a dual-core processor is virtually unchanged in comparison with a traditional parallel computer.

Today, multicore processing capability is being integrated into a number of standards-based modular, off-the-shelf form factors. For example, one company offers a COM Express module built around dual-core technology and a high-performance chipset that is designed for particularly performance-hungry applications and that maintains reasonable power consumption. This COM product offers good performance-per-watt and supports various user-friendly operating systems.

Exponential performance growth can be expected from future multithreaded design software with virtualization technology. Threads can take advantage of existing and future processor designs to future-proof the application. If existing software is optimized to be multithreaded, even more dramatic increases in performance can be expected. For many applications that in the past could not be run on parallel computers owing to prohibitively high costs, this technology creates previously unimaginable possibilities. It may be able to implement applications that combine, for example, endoscope visualization with multiple cameras, a simultaneous video conference with a specialist doctor, picture archiving and communication systems access, and patient monitoring into one system. In addition, it could work with various operating systems when required, which would reduce hardware costs.

Although the current generation consists of dual-core processors, multicore processors are the goal as technology shrinks and available real estate increases on dies. And quad processors will only be the beginning. Chip makers will continue to push for greater performance, using a combination of improvements in circuitry and more-advanced manufacturing technologies.

Implementing COMs

Today’s COM products can help embedded-system designers achieve faster time to market and reduce development costs. In addition, they can minimize design risks, simplify future upgrade paths and scalability, and increase application longevity, all of which lead to the potential for increased market share.

Performance, Design Flexibility, and Scalability. COMs offer the advantages of high levels of processing performance and I/O bandwidth in a compact form factor. They also allow flexibility in standard-form-factor boards that require upgradable host functionality.

Scalability is also critical for applications, such as those used in medical equipment, that require the flexibility and speed to meet product certifications and approvals. The COM approach enables system manufacturers to quickly and cost-effectively respond to competitive forces. They can meet new requirements by modifying their existing designs and expanding their product portfolios. This holds particularly true of embedded products that require longevity (5–10-year life cycles) and in which processing performance and I/O capabilities must be kept up-to-date.

Reduced Total Cost of Ownership. By combining a commercial off-the-shelf product with a custom baseboard, COMs allow rapid time to market. Because a product’s core has already been implemented, development costs are reduced. Using a COM implementation, medical OEMs can standardize their system core while also offering the ability to customize functions to meet specific application requirements. As a result, OEMs are able to focus on their core competencies and the unique functions of their systems. The end result is a compact, high-performance system designed in a shortened development time.

Figure 1. (click to enlarge) The test and validation phases and the need for design revision are elimina­ted when using a COM. The standard BIOS, drivers, and other functions have been pretested and qualified. It is ready to run when the module is plugged into the system.

Customization Simplified. Since COM modules contain all the components needed for a bootable host computer, their ease of installation is comparable to that of an off-the-shelf component. Without the need for a full custom motherboard design, system expansion and customization for each product can be implemented on an applicationspecific carrier board. Together, the COM and carrier board deliver the functionality of a single-board computer. As a result, a COM implementation can save design efforts and reduce costs without sacrificing performance or features. In a proven and stable COM product, the test and validation phase and need for design revisions are eliminated. Additionally, the standard BIOS, drivers, and other key components should have been pretested and qualified by the vendor, so they are ready to run out of the box. (Obviously, an OEM should verify and document that the vendor did this pretesting.) Therefore, an embeddedmodule-based semicustom design takes a fraction of the time and effort that a similar a full custom design would require (see Figure 1).

Another benefit is that software for the finished application can be developed in parallel on a standard COM evaluation platform, even while the customized carrier board is under development jointly between the COM vendor and the system designer. This translates to the ability to enter the market as much as three months earlier by using a COM product over a full custom design. An added advantage of designing a customized system around COMs is that the same customized baseboard can be reused even if the system calls for new computing core performance in future generations.

Reliable Operation. For reliable operation and a long lifetime, it is important to keep the operating temperatures of computer boards and their components as low as possible. Depending on the amount of thermal energy the components dissipate and the environmental conditions in which they operate, customized cooling approaches may be needed. And in harsher environments, computer boards must withstand severe shock, vibration, and a range of outdoor conditions. Some modules employ heat-spreader plates to assist with conduction cooling in passive and active cooling products. These plates allow the modules to operate in temperatures ranging from –10°C to 75°C. Sometimes, the metal-oxide–semiconductor field-effect transistors (MOSFETs) can be changed to offer extended temperatures ranging from –15°C to 85°C.

Figure 2. (click to enlarge) A comparison of full custom design and embedded modules.

By using the COM methodology, embedded designers can realize the following benefits (see Figure 2):

  • Stable and proven designs.
  • Shortened time to market.
  • Simplified development, allowing OEMs to focus on core competencies.
  • A seamless migration path to new technologies and longer application life cycles.
  • Reduced design risk through both standards-based technologies and interchangeable components.

Clinical Workstation Application Example. Using the example of a clinical workstation, the application benefits derived from using a COM product become clear. Typically, a clinical workstation comprises an IP65-resistant 19-in. panel PC with a pen-and-touch interface for data entry and control in patient monitoring or in ICU stations. Since the ETX module chosen for this application delivers the computer engine while the baseboard hosts the functions of power conditioning, battery backup (uninterruptible power supply function), and media storage, the design can either use a flash or a hard disk. Also designed to the baseboard are the RFID reader, the touch controller, and a PCI expansion card socket. Finally, the ETX module sockets are also connected to the baseboard. In this case, the workstation has a special feature—interconnects that allow the system to be connected to a range of patient-monitoring systems. The system was also designed without fans for reliability and noise control.

The modular approach that focused on the application-specific core enabled the design to be completed within the short time of 12 weeks. A relatively small team of electrical and systems engineers were able to do the concept design and prototyping in record time. The team consisted of one electrical engineer, one systems engineer, and one mechanical engineer, so the team was relatively small. Twelve weeks is record time in this context since such a project is normally scheduled for 6–9 months in time. The development time for a completely new computing product would traditionally have needed two extra hardware engineers and one firmware programmer to do the work, which would have caused the development time to more than double.

This particular workstation has now gone through four performance upgrades without a single change on the baseboard. So instead of redesigning the entire motherboard, the new ETX module was simply fitted, and then the workstation was ready for the approval process. This changes the effort from a multiengineer, multimonth project to a single-engineer, one-week project. Since the I/O interfaces were not touched, the failure risks of the EN 60601-1 Parts 1, 2, and 4 tests were greatly reduced. It is estimated that the cost of retesting alone had been reduced by 40% owing to its predictability.

Choosing a Partner: Looking beyond the Product

Although performance, reliability, form factor, and longevity are all critically important, there are a number of additional factors that medical OEMs should consider before finalizing their embedded computing product.

The embedded computing provider should have the design expertise to solve any thermal, power, or ruggedization issues. For instance, thermal dissipation varies considerably among ETX modules, and therefore proper heat removal from the heat spreader is an essential consideration for any ETX design. A supplier’s engineers should run the modules through cycles of extreme hot and cold temperatures to ensure that the products remain fully functional. High-power ETX modules require that heat-removal devices, such as heat sinks and fans or heat pipes, be attached to the heat spreader or be thermally coupled to the chassis. This type of design expertise can make or break the success of the implementation. In some cases, applications may have very specialized requirements that cannot be met from a standard or tailored product. Therefore, a supplier’s capability to design customized baseboards, backplanes, or systems is a significant benefit to take into account.

The supplier should have a clear understanding of existing and future medical requirements and should be reliable and concerned about customer and patient care. Knowledge of the regulatory certifications and requirements is essential. Precompliance qualification, along with customer support in preparing the documents for qualification, certification, and coordination with an accredited test lab, should be an integral part of the offering.

Because of the long life expectancy of most medical applications, it is also important to find a provider that is committed to the technology to ensure a lasting partnership that will remain competitive in the market. This should include a solid product road map with a breadth of products, opportunities for upgrades, and the foresight to anticipate future problems and design considerations.

Finally, manufacturing capability should not be overlooked. Key factors to take into account include the supplier’s abilities for product tracking and traceability, revision locking, and control, in addition to monitoring parts and supplies for long-life availability—in some cases, up to 15 years. Since cost reduction is always a consideration, manufacturing flexibility such as the ability to seamlessly move volume production overseas is also a plus.

Conclusion

Medical OEMs face many challenges over the life cycle of their products, from passing certifications and meeting performance and reliability requirements to ensuring that their technology and products keep up with evolving needs. Since medical equipment designers now have many choices of embedded computing products, the make-versus-buy decision is a much easier choice. However, selecting the best product is more than just selecting any off-the-shelf board. To ensure their application can meet the current and potential future needs of the medical market, suppliers must not only offer the right technology. They must also understand the unique requirements of medical computing equipment, provide design expertise and manufacturing excellence, and have a strong technology road map.

Marc Brown is COO of Kontron's North America division as well as director of the global Kontron medical division. He can be contacted at marc.brown@us.kontron.com. Matthias Huber is director of Kontron America 's embedded modules division. He can be contacted via e-mail at matthias.huber@us.kontron.com or via phone at 510-661-2220, ext. 277.

Copyright ©2007 Medical Electronics Manufacturing