[Title will be auto-generated] by RTC Media

The magazine of record for the embedded computing industry

January 2009

www.rtcmagazine.com

Wireless Sensor Nets: From Factory Floor to Mountain Top

Microcontrollers Latch onto the Internet Smoothing the Path between COMs and I/O Fine-Tuning Flash for Embedded Apps An RTC Group Publication

VOL.

A FASTER-TO-MARKET STORY

FORUMS

TUTORIALS

TAP INTO A QUALIFIED GLOBAL ECOSYSTEM WHENEVER YOU NEED IT.

Windows® Embedded helps to accelerate the development of smart, connected devices by giving your team access to a highly qualiﬁed and engaged worldwide community of industry leaders and Windows Embedded partners. Take advantage of training, tutorials, and their collective expertise to deliver next-generation, connected devices—faster. For the full story, go to WindowsEmbedded.com/FasterToMarket

WIRELESS SENSOR NETS: From Factory Floor to Mountain Top

26 Restaurant ordering stations such as this Embedded CE-based device can also communicate with wireless handheld devices used by wait personnel.

34 2.5 Gbit/s device brings together any port, service, channelization and functionality.

TABLEOF CONTENTS

36 Hybrid signal processing AMC module based on TI multicore DSP.

JANUARY 2009

Departments

Technology in Context

system integration

Microcontrollers

Machine to Machine

Bringing Internet Connectivity to 5Editorial 10 Not Just for Tree-Huggers: Green Will 8- and 16-bit Microcontrollers Be a Major Opportunity for Embedded Rodger Richey, Microchip Technology

Insider 6Industry Latest Developments in the Embedded Marketplace

Devices Will Require Standards for Success 26Service-Oriented Olivier Bloch, Microsoft Windows Embedded

Solutions Engineering

Industry Watch

Wireless Sensor Networks

Solid State Storage

Application Protocol Taking the Guesswork out of Flash Form Factor Forum (CAP): Uniting the Best of IP and System Design 30 14 Compact 8Small Where There’s a Bridge, There’s a Way ZigBee & Technology 36Products Newest Embedded Technology Used by Wireless Sensor Networks: Industry Leaders Featured Product Maintenance-Free or Battery-Free? 18 Views & Comment 2.5 Gbit/s Device Brings Together 48News, 34 Were There Bright Spots in 2008? Is Any Port, Service, Channelization There Hope for 2009? and Functionality David E. Culler and Gilman Tolle, Arch Rock

Matt Porter, Embedded Alley and Bill Weinberg, LinuxPundit.com

Cees Links, GreenPeak

Industry Insight Interconnect Wars

Siverge Networks

Interconnect Wars? Let Peace Prevail with Interconnect Standards

Robert A. Burckle, WinSystems and Colin McCracken, ADLINK Technology Digital Subscriptions Avaliable at http://rtcmagazine.com/home/subscribe.php

January 2009

JANUARY 2009 Publisher PRESIDENT John Reardon, johnr@r tcgroup.com EDITORIAL DIRECTOR/ASSOCIATE PUBLISHER Warren Andrews, warrena@r tcgroup.com

Dreaming of embedded super powers?

Editorial EDITOR-IN - CHIEF Tom Williams, tomw@r tcgroup.com CONTRIBUTING EDITORS Colin McCracken and Paul Rosenfeld MANAGING EDITOR Marina Tringali, marinat@r tcgroup.com COPY EDITOR Rochelle Cohn

Art/Production CREATIVE DIRECTOR Jason Van Dorn, jasonv@r tcgroup.com ART DIRECTOR Kirsten Wyatt, kirstenw@r tcgroup.com GRAPHIC DESIGNER Christopher Saucier, chriss@r tcgroup.com DIRECTOR OF WEB DEVELOPMENT Marke Hallowell, markeh@r tcgroup.com WEB DEVELOPER James Wagner, jamesw@r tcgroup.com

Advertising/Web Advertising WESTERN REGIONAL SALES MANAGER Stacy Gandre, stacyg@r tcgroup.com (949) 226 -2024 WESTERN REGIONAL SALES MANAGER Lauren Hintze, laurenh@r tcgroup.com (949) 226 -2014 EASTERN REGIONAL SALES MANAGER Shandi Ricciotti, shandir@r tcgroup.com (949) 573 -7660

These boards don’t kid around.

Super strong

with highperformance ARM cores and Pentium processors.

Super small

with boards 1/4 the size of the 104 footprint. TM

Super cool

with extreme temperature ranges and low-power operation.

Billing Maggie McAuley, maggiem@r tcgroup.com (949) 226 -2024

To Contact RTC magazine: HOME OFFICE The RTC Group, 905 Calle Amanecer, Suite 250, San Clemente, CA 92673 Phone: (949) 226-2000 Fax: (949) 226-2050, www.rtcgroup.com EASTERN SALES OFFICE The RTC Group, 96 Dudley Road, Sudbury, MA 01776 Phone: (978) 443-2402 Fax: (978) 443-4844 Editorial Office Warren Andrews, Editorial Director/Associate Publisher 39 Southport Cove, Bonita, FL 34134 Phone: (239) 992-4537 Fax: (239) 992-2396 Tom Williams, Editor-in-Chief 245-M Mt. Hermon Rd., PMB#F, Scotts Valley, CA 95066 Phone: (831) 335-1509 Fax: (408) 904-7214

www.EmbeddedSys.com Made in the USA Single Board Computers • Microcontrollers • I/O Boards PC/104 • EPIC • StackableUSBTM and more Contact us at 818-244-4600 or online.

Untitled-2 1

January 2009

1/23/09 1:53:02 PM

Published by The RTC Group Copyright 2008, The RTC Group. Printed in the United States. All rights reserved. All related graphics are trademarks of The RTC Group. All other brand and product names are the property of their holders.

JANUARY 2009

EDITORIAL

Not Just for Tree-Huggers: Green Will Be a Major Opportunity for Embedded by Tom Williams, Editor-in-Chief

s we begin the New Year, perhaps a wise resolution is to remember that “green” is not just a term used by environmentalists. It is also the color of money. At this writing, a lot of folks in Tennessee are wallowing in the blessings of “clean coal” in the midst of a disaster that puts the national focus once again on the issues of renewable energy and green technology—two topics that are bound to grow into major markets in the coming days and years. These two areas also represent tremendous market potential for deeply embedded control and measurement. In fact, in all areas of embedded systems, the development of very low-power yet computationally powerful embedded processors is opening up application areas that were unheard of just a short time ago. Developments coming out of Intel, AMD and VIA Technologies have brought performance per watt to such levels that enormous savings can be made in power consumption in such areas as server farms for Internet service and telecommunications systems as well as laptop and desktop systems. For example, the new generation of processors is increasing the performance per watt by almost 6x and bringing real power consumption down into the 2-watt range. But since information and communication systems consume only about 2% of the world’s electricity, that savings, while significant on one level, does not greatly impact the world’s energy consumption or the emission of greenhouse gases. The big opportunity for embedded control is that since we have been able to so dramatically reduce the power consumption of computer systems, why not use those systems to increase energy efficiency and reduce consumption in all areas of our lives? For example, industrial activity consumes almost 50% of the world’s electricity generation. And in factories, 65% of the electricity is consumed by electric motors, which are dumb and inefficient. Variable speed drives (VSDs) and more importantly, intelligent motor controllers (IMCs) can adjust motor speed to the needs of the load, and IMCs can adjust power draw for the required torque, while also greatly reducing the current surges that occur when a motor starts up. In some applications, motors start and stop frequently and this is a serious power sink. If such

intelligent motor control were applied to the world’s factories, they could reduce motor-related power consumption by 60%, which works out to about 13% of the world’s electricity usage today. In just one other example, connecting local power sources such as wind and solar to the grid with grid tie inverters (also dependent on small embedded computers to match phase with the grid and keep track of when power is being drawn off or pushed onto the grid) would not only provide non-polluting renewable energy, but also reduce the loss due to transmission over long distances by making power generation more decentralized. To really work right, however, such technologies require the development of the “Smart Grid,” which provides two-way communication between producer and consumer. Power grid engineers talk about a “system stability problem” that can occur when you decentralize power sources on today’s 1920s-era grid. That can be solved by the Smart Grid and would greatly facilitate the incorporation of decentralized renewable energy resources. The control of wind generators themselves is a feedback/control problem that balances optimal power output against things like dynamic stress on the blades and gears to provide maximum power along with optimal life span for the system—all of which is to be monitored and controlled by small, low-powered microprocessors working with sensors like accelerometers in the blade tips, current sensors, pitch controllers and axis controllers to meet the demands. And we haven’t even included the technology to improve efficiency in buildings, both public and private, transportation systems with GPS, RFID and route-planning to avoid such non-intuitive issues such as left turns, which often entail waiting at intersections with idling engines. The possibilities are endless. It seems that everywhere we look when contemplating a renewal of the national infrastructure, opportunities abound. The encouraging thing is that there seems to be an emerging synergy that can bring together the embedded industry with established industries and environmental concerns by creating business opportunities that include a whole array of green jobs along with the move to renewable clean energy sources. January 2009

IndustryInsider MONTH 2008

Code Base Growth, Safety-Critical Requirements Driving Static Analysis Market Recently published research by VDC Research indicates that embedded software engineers using a static analysis tool are working on projects with significantly larger average numbers of in-house developed lines of software code than developers not using a static analysis tool. VDC considers static code analysis tools to include solutions ranging from Lint-based syntax parsers to standards’ compliance checkers to tools using more formal methods for verification. With the overall embedded code base growing at a rate near 10%, VDC expects that an increasing number of development teams may look toward static analysis tools as a means to ensure code quality and functionality. The difference between the average code bases of these user groups can also serve to benchmark the typical code base sizes needed for a project team/company to be considered a legitimate candidate for a commercial static analysis tool. However, application classes tending to have requirements that are more safety- or security-critical in nature embrace static analysis tools at a higher rate; previously, code base size and complexity alone would normally drive development teams to consider additional testing methodologies.

Number of In-house Developed Lines of Software Code that Will Go into the Final Design (Average of Respondents)

Using a Static Analysis Tool

Not Using a Static Analysis Tool

Using a Static Analysis Tool for Mil/Aero Project

Mean

238,898

130,139

146,125

Median

45,000

30,000

22,500

For example, surveyed static analysis tool users currently working on a project for the military/ aerospace industry (which also accounted for the highest percentage of the total amount spent on embedded software static analysis tools in 2007) reported a median of 22,500 lines of in-house developed code as compared to 45,000 by the overall static analysis tool user population. Although VDC believes that an increasing number of development teams will look toward static analysis tools as a software development best practice, the escalating importance of software quality will generate a growing demand for comprehensive testing solutions that can also provide tight integration with dynamic testing tools as well as with other complementary application lifecycle management tools.

LonWorks Networking Technology Becomes a Series of ISO/IEC Standards

LonMark International, a non-profit, member-driven standards development and industry trade association supporting the LonWorks control-networking technologies, has announced that the voting members of the Joint Technical Committee (JTC 1) of the International Organization for Standardization (ISO) and the International Electrotechnical Commission (IEC) have formally approved LonWorks control-net-

January 2009

working technology as ISO/IEC 14908, Parts 1, 2, 3 and 4. This ratification will enhance the adoption and usage of this technology within the worldwide controls market and specifically in building control. While LonWorks is also used in many other markets, such as outdoor lighting, transportation, utility, process control and home automation, the majority of the over 100 million installed devices have been installed in or around buildings projects. The four parts of the standard set comprise the core protocol; the twisted-pair, free-topol-

ogy-wiring transfer medium; the powerline transfer medium; and the use of the Internet Protocol (IP) as a tunneling transfer medium, respectively. LonMark International plans to host an educational seminar and press conference on ISO/IEC standardization at the upcoming AHR Expo in Chicago on January 27, 2009. Interested parties can find more information about these events and the standards on the LonMark Web site at www.lonmark.org. Additional white papers discussing the ISO/IEC standard and LonWorks will also be avail-

able on the LonMark Web site.

IPv6-Based PhyNet WSN Chosen for Emergency Communication System

The PhyNet wireless sensor network (WSN) by Arch Rock has been selected by HITEC Luxembourg S.A. for use in its Nomadic Satellite Communication System (NoSaCo), developed for natural disasters and other public safety situations. NoSaCo with PhyNet underwent a “dress rehearsal” in May 2008 at the ING europe-marathon Luxembourg, an annual event attended by some 4,000 runners and close to 100,000 spectators. NoSaCo, part of HITEC Luxembourg’s Public Safety Suite, was designed as a rapidly deployable ad hoc, high-performance communication system for use in emergency scenarios where existing infrastructure has been destroyed or rendered unreliable. The system uses various wireless technologies, such as satellite, Wi-Fi and short-range radio, as well as video and voice communication and sensor networks that can monitor critical information on-site and transmit it quickly over the Internet to those in charge of rescue operations. The NoSaCo system was already well into the definition phase when, in the fall of 2007, the development team learned about PhyNet. The PhyNet architecture appealed to the NoSaCo team because it was low power, easy to deploy and based entirely on the ubiquitous Internet Protocol (IP)—and specifically on IPv6, the newest and most scalable version of IP. Furthermore, PhyNet was the first multi-tier WSN to address large, geographically dispersed applications by forming multiple highly resilient IP-based sensor networks that users could manage centrally as

Industry Insider

part of the overall IP infrastructure. The key to this architecture was a new kind of router that formed an internetworking backbone between an individual WSN and its server-hosted applications, allowing information to be monitored miles—or even thousands of miles—away from their control and management functions.

One Stop Systems Purchases Ciprico Storage Product Lines

One Stop Systems has announced that it has purchased the exclusive manufacturing rights to Ciprico’s MediaVault, Talon4 and DiMeda product lines. In addition to assuming the existing product order backlog and inventory, One Stop Systems will fulfill new orders for products from customers and distributors, will manufacture the products, including material procurement, fabrication, assembly and test, and will provide technical support to customers and distributors. The MediaVault products are direct attached storage devices used in commercial applications such as radio and television broadcasting and video imaging. They include a 5-drive desktop RAID array (OSS-4105), often paired with Apple systems, a 10-drive rackmount version (OSS-4210) and a 40-drive RAID array (OSS-4440). All of these are offered in SCSI320 or Fibre Channel interfaces. In addition, an 8 and 16 drive PCI Expressbased RAID array (OSS-5108, OSS-5116) provides even higher bandwidth data storage. The Talon4 meets the stringent standards for shock, vibration, and severe environmental and temperature changes typically found in airborne, shipboard, fixed and mobile ground-based installations. The Talon4’s dual controllers meet the Storage Bridge Bay

(SBB) 2.0 specification. The DiMeda NAS (network attached storage) appliance (OSS-1716) offers performance levels reaching 700 Mbytes/s supporting high-performance file sharing and video streaming applications. The 1716 supports both SATA and SAS drives and offers a greatly enhanced RAID feature set. It is ideally suited for the digital cinema market as well as any application where multiple systems are accessing high-speed data transfers.

Alereon Receives FCC Modular Certification for Worldwide Wireless USB Half-MiniCard

Alereon, a Certified Wireless USB technology company for Ultra Wideband (UWB) WiMedia solutions, has announced it has received FCC modular certification for its AL5708 Worldwide Wireless USB Half-MiniCard. The AL5708, which provides Wireless USB connectivity for laptops and desktop PCs, is the first and only FCC approved Wireless USB embedded notebook solution for use in UWB bandgroups one, three and six. Alereon’s AL5708, which is PCI Express (PCIe) type H2 compliant, is based on the AL5000 family chipset that covers the entire UWB spectrum from 3.1-10.6 GHz. In addition, the AL5708 is the only solution to date that supports bidirectional, isochronous data transfers both in and out allowing manufacturers to support common consumer devices such as USB speakers and Web cams. The AL5708 Half-MiniCard is the fifth Alereon product that has received FCC modular certification for Band Groups one, three and six, further expanding the rapidly growing portfolio of Wireless USB solutions, both on the host and device side, which Alereon

will bring to market in early 2009. Alereon also received PCI Express V1.1 certification from the PCISIG for the AL5350 Worldwide PCIe WiMedia/UWB Baseband and MAC, the base chipset used in the AL5708 Half-MiniCard.

Sysgo and Interface Concept Partner for Unified Systems Solutions

Sysgo has announced a partnership agreement with Interface Concept, signaling new collaboration between major European software and hardware suppliers of embedded systems. Sysgo is a supplier of software solutions for demanding safety and security applications. Interface Concept is a designer, developer and manufacturer of electronic embedded solutions. The new agreement between the two companies mainly consists of mutual support for integrating the two companies’ products, i.e. Sysgo’s Industrial Grade Embedded Linux ELinOS with Interface Concept’s PowerPC, PowerQUICC, Intel processors and DSP single board computers. Interface Concept brings solutions to a large array of applications ranging from industrial to the most severe environments, by offering products in standard, extended, rugged and conductioncooled grades. IC not only supports today’s industry standards but is committed to include new exciting standards such as VPX in its product roadmaps to enable its customers’ next-generation embedded applications. With ELinOS Industrial Grade Linux, Sysgo offers an “outof-the-box” experience, which allows embedded Linux developers to concentrate their full attention on the development of their target applications, instead of spending time on gathering the right pieces they need. ELinOS incorporates

the appropriate tools to help them build the system and boost their project success, including a graphical configuration front-end with built-in integrity validation.

Tail-f Partners with Afore to Accelerate Network Management Solutions

Tail-f Systems, and Afore Solutions, a supplier of technology solutions for the networking industry, have announced a partnership to collaborate on sales and system design capabilities aimed at accelerating time-to-market for their prospective customers. Together the companies enable network equipment providers to bring enhanced XML-based management capabilities to market. Tail-f’s ConfD product provides developers with the tools and infrastructure to develop device-side network management solutions including key management interfaces (CLI, Web UI, SNMP, and NETCONF) and an integrated database. ConfD uses a data-model driven architecture that significantly reduces the need for costly and error-prone integration of multiple configuration protocol stacks. Afore delivers industry standard solutions to equipment and solutions providers. With a team of rich engineering resources, Afore continues to create custom technology solutions that are tailored to the particular requirements of the network and equipment provider. Its technology solutions provide value-added capabilities to accelerate service offerings and optimize, support and enhance new service offerings.

January 2009

SMALL FORM FACTOR FORUM

Where There’s a Bridge, There’s a Way

mbedded designers in all market segments have become enamored with the new ultra-low-power Atom and Nano processors from Intel and VIA, respectively. For years, hardware engineers have been asking processor manufacturers to reduce the power envelope of their notebook/mobile PC processors and chipsets for the embedded market. After all, 100Wplus desktop CPU chips outstripped embedded market requirements nearly 10 years ago. Even modern mobile-focused CPUs such as Eden, Celeron and Geode rarely met the thermal characteristics required for a small form factor, fanless CPU board to operate in a small or sealed enclosure. Newer multicore designs were not a step in the right direction. If the software engineers aren’t fully utilizing one core, why add the power requirements of a second core? Atom and Nano, however, have hit the small form factor sweet spot. For years, 486 and Pentium MMX processors reigned in embedded designs, giving way to Eden, Celeron and Geode. The “heirs apparent” are clear, or are they? While desktop PCs and, to a certain extent, laptop/notebook/ sub-notebook PCs are largely cookie-cutter designs, most embedded applications are unique. I/O requirements, processing throughput, size, thermal design, MTBF and operating systems vary widely. What about solid-state mass storage (SSDs)? Serial ports? Expansion buses such as ISA (shudder) and PCI? The reality is that legacy I/O and buses are still in demand. Reconciling these requirements with the new legacy-free chipsets supporting Atom and Nano is the challenge. Designers now get to add the features they need, a la carte style. To add the ISA bus, there are several options. Bridging from the LPC bus involves a low-cost device from several silicon vendors, with modest feature sacrifices such as DMA, address space and single-cycle 16-bit transfers. To preserve these features, a PCI-to-ISA bridge can be used—at a higher cost. Of course, the BIOS must initialize the bridge selected. Unfortunately, the parallel 32-bit/33 MHz PCI bus is not supported on some low-power chipsets. For applications requiring legacy PC I/O-like serial ports, parallel port, or PS/2 keyboard or mouse, one straightforward approach is to add a super I/O IC to the LPC bus. Some chip vendors offer serial port ICs that attach to the PCI bus. Many applications require general-purpose digital I/O. Southbridge ICs

January 2009

usually have a few spare GPIO pins, some are used internally, and few are brought to an I/O connector. Interrupts on change of state are not available. When using a computer-on-module (COM) rather than a single board computer (SBC), it is critical to confirm that GPIO pins are defined on the carrier board interface, and that the module under consideration connects these signals. Where PATA/IDE devices are used, be they rotating disk drives or solid state storage, there is good news for now anyway. Nearly all low-power chipsets still have this interface. Several new SSD standards and form factors will be available in 2009 using USB, SATA, or other interfaces to enable a smooth transition before PATA goes away entirely. At some point, a stand-alone IDE controller will be required to continue to support those devices, much like the original way disk controllers were designed in. This again, however, leads us back to a PCI bus requirement. Moving up to higher speed bus interfaces, PCI Express has already begun to overtake PCI in the large form factor arena (motherboards and blades for card cages). The small form factor market is now rushing headlong into this space as well. Many alternatives are available for off-the-shelf boards and custom designs. One challenge is that PCI is being replaced by PCI Express x1 (“by one”) lanes. Parallel PCI can be obtained by bridging from PCI Express. While software engineers celebrate the transparency of this bridge, they will be working overtime to help the firmware engineer get the legacy I/O to work. A final gotcha here focuses on bandwidth, of all things. PCI supports 4 devices natively, and up to 32 devices with PCI to PCI bridging. Cramming 4 PCI devices onto a single PCI Express x1 lane does not provide for the rip-roaring transmission of data. What to do about the limited PCI Express lanes becomes the topic of a future column. The challenges are compounded when several of the aforementioned bridges and super I/O devices are used together. Getting these neat low-power chips to work in your application can become a bit of a chore, considering how much has to be added to make them fit your unique requirements! Comments about this topic can be sent to sf3@rtcgroup.com.

Colin McCracken

& Paul Rosenfeld

One Box, Countless Options MAGMA Leads the Industry by Providing Innovative, Hi-Tech Expansion and Connectivity Solutions

Many Facets of Magma’s EBU Products • Classic Expansion Chassis • Virtualization Product Applications • Building Workstations/Servers • Global Flash Memory Arrays • Clustering Servers

Practically any type of PCI Express card can be used in the Magma Expansion chassis ranging from Single Board Computer, audio-processing, video capture, graphics, test & measurement, data acquisition, SCSI, Fibre Channel, SATA and off course Magma’s own Universal Host Adaptor.

www.magma.com A Service Disabled Veteran Owned Small Business

RoHS Compliant

Tel: 1.858.530.2511 ISO9001:2000 Certified

Technology I n C onte x t

Microcontrollers

Bringing Internet Connectivity to 8- and 16-bit Microcontrollers Many embedded devices do not need the power and bandwidth of a 22-bit microprocessor, which is, however, often needed for Internet access. With some design refinements, 8- and 16-bit microcontrollers can be adequate connectivity and data rates for small, dedicated devices.

by Rodger Richey Microchip Technology

ave you ever been at work and realized that you left your home’s heater running on high? Or, remembered that you have a large number of people coming to a summer event and needed to start cooling your home earlier than normal? Both situations could be handled easily if you could access your home’s thermostat through a Web browser. Just enter a URL and make a couple of mouse clicks. But the world of a typical Internet application and device is very different than that of an embedded application running on the 8- or 16-bit microcontrollers (MCUs) typically found in today’s thermostats. And, spanning the huge disparity between those worlds raises a lot of intimidating questions. Does Internet connectivity require the uCLinux or Embedded Linux operating systems and a high-end 32-bit processor? What are the costs of those operating systems and the TCP/IP stack for that class of processor? If I only have experience with 8-, 16- or 32-bit microcontrollers, what is the learning curve to develop with a highend 32-bit microprocessor and operating system?

January 2009

Fortunately, there are many solutions providing cost-effective Internet connectivity for 8- and 16-bit microcontrollers— and even entry-level 32-bit microcontrollers—without the overhead of highend operating systems and a high-end, cache-core 32-bit processor. Embedded, Internet-connected applications previously involved devices such as cellular telephones, personal digital assistants and portable gaming units. These devices use high-end, 32-bit microprocessors with megabytes or gigabytes of memory, and they run on Embedded Linux, WIN CE, or some other full-blown operating system. Deeply embedded applications, such as thermostats, security systems or utility meters, have only kilobytes of memory and, frequently, no operating system. Bringing Internet connectivity to low-cost, 8- and 16-bit MCUs faces two challenges: the TCP/IP stack and the network interface controller.

Taming the TCP/IP Stack

The TCP/IP-stack challenge is resolved by reducing the stack and replacing the high-end devices’ traditional operating systems. A minimal uCLinux operat-

ing system requires about 500 Kbytes to 1 Mbyte of ROM and up to 500 Kbytes of RAM. And, the TCP/IP stack may not be included in those numbers. This far exceeds the capabilities of a typical 8- or 16-bit microcontroller. Eight- and 16-bit MCUs will not be streaming video or serving complex Web pages, or other high-bandwidth data, so they do not need a full-featured operating system and the full TCP/IP stack— elements that consume great amounts of program and data memory and MIPS. The requirements for deeply embedded applications can be met by a lower-bandwidth stack and, in most cases, intermittent Internet connectivity. A thermostat, for example, is accessed by a very small number of users—probably one or two people—and that access is not continuous. This enables the overhead of the TCP/IP stack to be reduced to a minimum part of the application. Figure 1 shows a typical TCP/IP organization. Many of the displayed protocols are not required for an embedded application, such as a thermostat, so they can be eliminated—significantly reducing the stack’s footprint. The minimal require-

Some applications will have additional requirements. FTP might be needed to transfer information between nodes, such as uploading new Web pages or firmware for the microcontroller. SMTP may be required to send emails about status changes; such as a vending machine communicating that its stock is low. If an application is serving Web pages, it requires HTTP. Those protocols are in the application layer, as shown in Figure 1, so they use some of the basic protocols already listed. One of the solutions currently available for addressing issues with the TCP/IP stack and operating system is from Contiki (http://www.sics.se/contiki/). As the company’s Web site states, “Contiki is an open source, highly portable, multi-tasking operating system for networked memory-constrained embedded systems. A typical Contiki configuration is 2 Kbytes of RAM and 40 Kbytes of ROM.” These specifications are ideal for our thermostat example, as 8- and 16-bit microcontrollers can easily accommodate those memory sizes. Another solution is the free TCP/IP stack from Microchip Technology (www. microchip.com/tcpip). This stack was developed exclusively for Microchip’s PIC microcontrollers and has a modular design so that only the required protocols are enabled. A minimal implementation may take only 45 Kbytes of ROM and 2 Kbytes of RAM. Both of these solutions provide the small footprint and RFC-compliant protocols to enable scalable, cost-efficient Internet connectivity for 8- and 16-bit mi-

UDP

ICMP

More...

TCP

Application

Transport ARP

Ethernet MAC/PHY (Silicon)

Figure 1

FTP

HTTP

SMTP

Telnet

SNMP

DHCP

DNS

SNTP

ments for Internet connectivity are: • Low-level physical interface driver – Driver for the Wi-Fi or Ethernet interface controller • IP – Global addressing for packets • ARP – Address resolution • TCP – Basic communication protocol with handshaking • UDP – Basic communication protocol without handshaking • ICMP – Ping, for verifying connection status • DHCP – Automatic network-address configuration • DNS – Mapping of URLs to IP addresses

NBNS

Technology In Context

Addressing Physical

A typical TCP/IP stack incorporates more protocols than are normally needed by small embedded devices.

crocontrollers.

Tailoring the Interface Controller

The remaining challenge in enabling Internet connectivity on 8- and 16-bit microcontrollers is the network interface controller. Most Ethernet interface controllers implement only the MAC and use an external PHY. That is because the MAC is digital logic with a relatively small silicon implementation. The PHY, conversely, is analog, which can require a large silicon footprint and costly testing. Very few semiconductor manufacturers integrate Ethernet into their 8- or 16-bit microcontrollers. Microchip Technology has implemented both a MAC and PHY into its 8-bit PIC18F97J60 family; the company’s first in a series of Ethernet-enabled microcontrollers. These integrated MCUs provide small PCB footprints and are quite capable of running the TCP/IP stack and an application’s code. If a system incorporates an external Ethernet interface controller, two factors determine its suitability for deeply embedded applications: the interface and the required resources. Most of today’s Ethernet interface controllers only have interfaces designed for high-volume applications, such as PDAs, cellular telephones and portable gaming devices. These applications will typically use a 32-bit, parallel PCI or PCMCIA standard that is not used on 8and 16-bit MCUs. The standard is also very difficult to implement in software and, even when successful, the resulting overhead severely inhibits throughput. Furthermore, these standards require 80

or more of a processor’s I/O pins, leaving little room for additional functionality in the host MCU. A solution more friendly for 8- and 16-bit microcontrollers is provided by the ENC28J60 from Microchip Technology and the KSZ8851 from Micrel, both of which use a Serial Peripheral Interface (SPI). Initially, SPI might be thought to limit the data rate, but that fails to consider the real-world needs of deeply embedded applications. Deeply embedded applications do not require high data rates, and their requirements can be easily met with careful design. The bandwidth issue usually involves the speed at which the microcontroller can run the TCP/IP stack. But if TCP/IP communications are optimized, the bandwidth issue goes away. An 8- or 16-bit microcontroller using the ENC28J60 Ethernet interface controller can achieve a data rate of up to 400 Kbytes per second—quite sufficient for well-designed TCP/IP communications. In the example of the thermostat, such a design would implement a small number of Web pages with low-overhead displays (Figure 2). The second factor in determining an external interface’s suitability for 8and 16-bit microcontrollers is resources. There are two elements in determining the resources an interface will require on a microcontroller: MAC implementation and RAM requirement. On the MAC issue, a key question is whether the interface controller will implement the full MAC, or will the microcontroller have to implement some of it? Also, can the distribution of the MAC implementation be done in a January 2009

Zero to FPGA in Minutes

Host MCU with SPI Interface running TCP/IP stack

SPI Interface

Technology In Context

8 Kbyte TX/RX Buffer

MAC

PHY Ethernet Connection

ENC28J60 Stand-alone Ethernet Controller with MAC and PHY

Figure 2

Integrate Customizable FPGA Hardware into Your VME Systems �

�

Upgrade and increase your VME system performance Use NI LabVIEW graphical functions for I/O, processing, and communications Reuse existing HDL code and IP Use COTS FPGA hardware for PXI/ CompactPCI and PCI

>> Learn to add FPGA

customization to your VME system at ni.com/vme 800 453 6202

© 2007 National Instruments Corporation. All rights reserved. LabVIEW, National Instruments, NI, and ni.com are trademarks of National Instruments. Other product and company names listed are trademarks or trade names of their respective companies. 2007-9412-821-101

Untitled-5 1

January 2009

12/17/08 10:44:44 AM

ENC28J60 stand-alone Ethernet interface controller block diagram.

way that reduces the microcontroller’s resource requirements and die size? If the host controller is a high-end 32-bit microprocessor with megabytes of ROM and RAM, a lot of requirements are offloaded. But, that would place a significant demand on the typical 8- and 16-bit microcontroller. A more MCU-friendly interface controller implements the full MAC on-chip. On RAM requirements, the typical Internet approach again must be changed to bring connectivity to the 8- and 16-bit world. Typical interface controllers have no internal RAM buffers available to the host controller, and instead simply stream received packets to the host controller while accepting fully formatted packets from the host controller. Since the internal RAM is not accessible by the application, this places an extreme burden on the typical 8- and 16-bit microcontroller. Adding an on-chip buffer to the interface controller allows the microcontroller to slowly accept data as it processes the packet through the various protocols. It also allows the microcontroller to form packets in the interface controller and send them, when ready. In considering interface controllers, Wi-Fi is the latest buzz in the embedded design community. But what are the considerations for incorporating Wi-Fi technology into an end product? If a product is not being produced in massive volume, the way that cellular phones, PDAs, portable gaming devices and PCs are, then it is very difficult to find a cost-effective Wi-Fi solution. It might be possible to find an inexpensive Wi-Fi USB dongle or similar device, but then the problem becomes finding drivers.

In considering the Wi-Fi alternative, the same earlier questions must be asked. What is the network interface? Does it implement the full MAC? How much of the Wi-Fi device’s RAM is available to the application? Most of these questions have not been answered until recently. G2 Microsystems has released the G2C547 in both module and discrete device format. It is a full, 802.11 b/g solution that provides an SPI interface to the application. ZeroG Wireless also plans to offer Wi-Fi modules with an SPI interface and integrated RAM for the application. Both companies offer drivers enabling easy interface-device integration. Until recently, Internet connectivity was the domain of only high-end applications, such as PCs, servers, cellular telephones, personal digital assistants and portable gaming devices. That connectivity was enabled by full-blown operating systems, TCP/IP stacks and interface controllers designed for larger devices. Now, a whole new world of applications is connecting to the Internet—smaller, deeply embedded devices such as thermostats, utility meters and agricultural/industrialmonitoring equipment. This class of application requires low-overhead, focused solutions. Microchip Technology Chandler, AZ. (480) 792-7200. [www.microchip.com].

solutions en g ineerin g

Wireless sensor networks

Compact Application Protocol (CAP): Uniting the Best of IP and ZigBee The popular ZigBee mesh network protocol has caught on as a means of distributed sensing and control. Mating it with IP networks has until now been problematic. A new protocol definition now brings the two together.

by David E. Culler and Gilman Tolle Arch Rock

n several industries over the past three decades, ubiquitous automation has occurred hand-in-hand with the emergence of industry-standard application protocols. Starting with HART in 1980 for process control and continuing with BACnet and LONtalk for building automation and Common Industrial Protocol (CIP) for manufacturing, application protocols have provided a way to integrate different kinds of devices, potentially from different vendors, into an operational network, and to configure the devices to the task at hand. Historically, such protocols were defined for a specific, widely used communication link—e.g., RS485 for BACnet and DeviceNet for CIP—and they specified the physical medium (layer 1), the link and its addressing (layer 2), and the application data representation, naming and services (layer 7) in a vertically integrated solution. Each protocol defines a distributed object abstraction, where each device is treated as a named object containing a set of attribute-value pairs and possibly supporting a set of commands. An application profile defines the attributes in each type of object (i.e.,

January 2009

ZigBee Application Layer

CAP SNMP HTTP FTP ... UDP

Zig Bee Network Layer IEEE 802.15.4 2003 MAC/PHY Sub-Layer

Figure 1

TCP

...

IP 6LoWPAN

802.15.4 Ethernet WiFi ...

Architecture of ZigBee stack compared with architecture of CAP IP stack.

device) and how to get and set the associated values. The protocol defines how to discover these services, how to bind objects together to form specific relationships, and the formats of all the messages exchanged between them. Invariably, these industrial application protocols evolved to support additional communication links. And to take advantage of technological innovation without redefining the application protocol, they invariably adopted IP. IP’s layered architecture separates the link layer from the application layer by a network layer that provides link-independent host addressing and routing of datagrams, and a transport layer that provides application-independent reliability and service naming.

Starting in 2004, ZigBee followed suit by defining a vertically integrated protocol suite that provided a distributed object abstraction for devices on a new low-power wireless link, IEEE 802.15.4. The broad utility of this link led to the definition of a wide variety of application profiles— including home automation, commercial building automation and, most recently, smart energy—which cut across industry segments. Moreover, the short range of this low-power link meant that multihop routing, not just link connectivity, was required in most applications, and the ZigBee organization defined and redefined a series of network layers. One silver lining in the perennial, often incompatible, evolution of the network protocols is that the application protocol remained somewhat independent of the specifics of the IEEE 802.15.4 link layer and the ZigBee network layer. Its compact encodings do reflect the small packet size and limited bandwidth of this link, so it is suitable for use on this and any more capable link. Finally, in 2007, the link used by ZigBee (and other industry standards) became a full-fledged member of the IP

SOLUTIONS Engineering

family with the IETF’s adoption of the 6LoWPAN adaptation layer, which specifies how to carry IPv6 datagrams in a compact form over 802.15.4. Just as with Ethernet or Wi-Fi, applications can be built for 6LoWPAN networks by simply opening a UDP or TCP socket and sending/receiving messages to/from an IP address and port number. Indeed, the two parties in the communication need not be on the same physical link, because routers take care of getting the packets from one to the other. The Compact Application Protocol (CAP) completes the unification of IP and ZigBee by transforming the Zigbee application profile so that it is defined for UDP transport over any IP link (Figure 1).CAP does everything that ZigBee can do when two devices are in the same 802.15.4 network, but it allows any device on any link to operate in the same manner. Furthermore, the links and routing layers can continue to advance beneath CAP without changes to the application or the application protocol.

CAP in Action

The main concepts in CAP are illustrated with an example application scenario (Figure 2). This scenario reflects a typical residential setting and is easily generalized to a building or commercial or enterprise setting. From a network point of view, the usual home router is connecting an Ethernet LAN and Wi-Fi (802.11) wireless LAN to form a private IP network. It connects to the Internet over a WAN link through a firewall with network address translation, so any of the hosts can access Internet resources but outside hosts cannot access private resources. The one addition to the typical home network is that the router also has a 6LoWPAN (802.15.4) interface. The embedded wireless devices, including the electric meter, dryer with load control device, switches, lamp, thermostat and handheld all have IPv6 addresses (in addition to unique IEEE 802.15.4 MAC addresses) and can send and receive messages using UDP or TCP sockets to and from each other or the other hosts in the home network. The flow of communication may be quite asymmetric, but the ability for general communication is

provided by the IP network layer. In this scenario, the LAN and Wi-Fi are an IPv4 subnet, with the LAN containing a home server and an inverter for solar panels on the roof. The Wi-Fi WLAN is used by the laptop and handheld device, as well as an ambient display, such as the common digital picture frame, which displays Smart Energy status. The home router provides standard DHCP to assign IP addresses for those hosts that are not configured with static addresses and performs IPv6-toIPv4 translation.

cludes a switch, lamp, thermostat, laptop and server. The Smart Energy (SE) profile includes an electric meter, dryer, thermostat, home server and ambient display. The thermostat has two endpoints, one for each profile, while the rest of the embedded devices have a single endpoint. The home server has two services, perhaps from different vendors, each with a single endpoint. The lamp, an on/off light device in the HA profile, serves the OnOff cluster by providing an on/off attribute (the state SE: Energy Service Portal

• Demand Response and Load Control Cluster (server) • Load Control Event Command

SE: Metering Device

• Simple Metering Cluster (sv)

SE: Load Control Device

• Demand Response & Load Control Cluster (client) • Report Event Status Command

HA: Remote Control

Internet

• Instantaneous Demand Attr. • 24 bit Int

Utility Network

Firewall

• Thermostat Cluster (client) • ...

WAN

LAN Inverter

Router

HA: On/Off Light

WiFi

LoWPAN

• On/Off Cluster (server) • Off Command • On Command • OnOff Attribute • Boolean

• OnOff Cluster (cl) • OnOff Attr. (Boolean)

SE: In-Premise Display Device

[1] HA: Thermostat HA: On/Off Switch •OnOff Cluster (client) •OnOff Attribute •Boolean

HA: Remote Control

• Thermostat Cluster (server) • Local Temperature Attr. (16-bit int 100ths°C) • SystemMode • Occupied Cooling Setpoint (16-bit int 100ths°C)

• Simple Metering Cluster (client) • Instantaneous Demand Attr. • Signed 24-bit Integer • Demand Response... Cluster (client) • Report Event Status Command

[2] SE:PCT

• Demand Response & Load Ctrl Cluster (cl) • Report Event Status Command

Figure 2

CAP embedded network supporting Home Automation and Smart Energy Profiles on IP links.

The Compact Application Protocol raises the level of abstraction so the user can view this collection of devices as one or more plug-and-play applications. In CAP, as in the ZigBee Application Layer, services are provided by an endpoint on a physical device. Each endpoint is associated with a particular application profile. The application profile defines a set of device types and for each type a set of clusters, which provides the functionality of the service with a standard interface and data representation. In particular, each cluster provides specific attributes and commands. In this scenario there are two application profiles, both of which naturally span multiple kinds of communication links. The Home Automation (HA) profile in-

of the lamp) and two commands to control that state. This endpoint is bound, via the CAP binding protocol, to a particular on/ off switch device in the HA profile that is the client side of the on/off cluster. It maintains an attribute that is the position of the switch and issues the on and off commands to the lamp endpoint. Thus far, the application is identical to what it would be with ZigBee. The implementation of the CAP library is written using a conventional UDP socket interface rather than a ZigBeespecific API; the addressing that is hidden inside of it uses IP addresses rather than 802.15.4-specific MAC addresses. The power of this small change is manifested when the Wi-Fi-connected laptop is used as a remote control for the lamp. The January 2009

SOLUTIONS Engineering

laptop is also in the HA profile and has an application built on CAP, but in this case it is a Windows application using a CAP library that is implemented using the WinSock API. The laptop is also bound to the on/off light endpoint in the lamp. It can query the lamp’s on/off attribute with a ReadAttribute request and can actuate the lamp with the on and off commands. The home server also has an endpoint in the HA cluster, perhaps using a CAP li-

1 16Untitled-4 January 2009

brary written for the sockets API in Linux or Mac OS X, so it can operate as a client of the Thermostat cluster of the thermostat endpoint in the HA profile, allowing the homeowner to program the thermostat using a convenient GUI. Indeed, another wireless 6LoWPAN switch could be bound to the multimedia application on the home server to act as a remote control for the entertainment center. While these devices happen to be attached to different IP links and offer

1/21/09 8:40:31 AM

different operating system environments, they provide a consistent set of interoperable application profiles through CAP. The system operates as if all the devices were on an 802.15.4 network in ZigBee, but IP addressing and UDP transport are used for the physical communication so that the application is link- and platform-independent. This additional level of generality is even more important with regard to the second Smart Energy application in this network. The electric meter is a metering device in the SE application profile and serves the Simple Metering cluster, which provides an InstantaneousDemand attribute, in addition to other attributes. The ambient display is the client of this cluster, providing a convenient graphical presentation of real-time energy use. But these devices, as well as the thermostat, dryer load control unit and energy services portal, also provide the Demand Response cluster within the SE profile. In times of peak demand, a DR command is sent from the utility to the energy services portal, which utilizes the homeowner preferences to decide how to reduce demand, and issues load control commands to the thermostat, dryer and ambient display, which alerts the homeowner. The DR command may arrive directly through the meter, over the Internet, or via a dedicated out-of-band path, depending on the utility; but the IP networking capability allows it to be routed to where the action is initiated. The extensibility offered by CAP is also worth noting: the grid tie inverter for the solar energy system is also on the network, and, though no such device is currently defined in the SE profile, provisions are made for vendor-specific clusters. In addition, all the resources described here are contained within the private IP network, but select portions could be moved out to remotely hosted services with established security mechanisms, including VPNs and SSL, to protect them. CAP also enables the system installer or homeowner to confirm that a new device is present on the network, determine its capabilities, and configure its relationships with other devices in the network. In addition, CAP includes its own security infrastructure to protect communication between devices.

SOLUTIONS Engineering

After the new device connects to the IP network and gets an IP address, the installer can use a simple handheld configuration tool to discover the device and enumerate its capabilities. With CAP, this tool might be a specialized physical device, and it might also be an application running on a regular PC or laptop. The configuration tool uses the CAP discovery protocol to query devices for information about the application profile, device type and specific clusters supported on each of their endpoints, and then presents this information to the installer in a friendly way. CAP can use broadcast communication in small networks, and in large networks can make use of a secondary CAP discovery server with which each device automatically registers at startup time. Once the installer confirms that the new device is in the network and has the desired functionality, the configuration tool can be used to draw connections with other devices. To establish a connection, the tool sends a CAP binding command to the device, which connects a particular client or server cluster in that device with a matching cluster on another device. Inside each CAP binding is the IP address, UDP port and endpoint ID number of the peer device, which is used to transparently connect the light and the switch, for example, or the meter and the ambient display. Certain high-security devices and applications may also need application-layer authentication and encryption when communicating, analogous to secure HTTP or secure sockets. CAP supports shared encryption keys, which may be pre-assigned or requested from a Trust Center service running on a different IP host.

A second key consideration is the technologies on which the application protocol depends. Choosing an application protocol that depends only on the presence of an IP network increases the variety of devices and applications that can be included in the system, decouples the system design from any particular link technology, and helps to future-proof the upgrade path. For the application domains covered by ZigBee, the CAP architecture

Arch Rock San Francisco, CA. (415) 692-0828. [www.archrock.com].

New Rugged, Reliable Servers for Mission-Critical Applications New! 1RU RES Servers - 1 or 2 AMD® Opteron™ or Intel® Xeon® CPUs (quad-core processors) - Up to 32GB ECC SDRAM - Up to 3 removable HDD - Dual redundant, hot-swappable PSUs - Dual redundant DC power option - 0 – 55º C operation - Vibration – 0.9g from 10 – 2000Hz - Shock – up to 20G @ 20ms - 2RU RES Servers also available

3RU RES Servers - 1 or 2 AMD Opteron or Intel Xeon CPUs (quad-core processors) - Up to 32GB ECC SDRAM - Up to 5 removable HDD - Vibration – 0.9g from 10 – 2000Hz - Shock – up to 25G @ 20ms

For Sun® Solaris™, Linux®, and Microsoft® Windows® environments When it comes to computing in harsh environments, nobody is more at home with its surroundings than Themis Computer. For over a decade, Themis has delivered high performance, high availability computing for the most demanding military, aerospace and communications applications. Themis’ new family of 1RU, 2RU and 3RU Rugged Enterprise Servers™ (RES) includes the latest quad-core processors from AMD and Intel, offering ruggedized systems with the fastest, widest range of performance options and scalability. Themis servers provide far greater reliability, improved life cycle management and substantially lower TCO than other COTS systems solutions.

CAP Considerations

The most important consideration when choosing an application protocol is the industry-specific decisions that have been collected and invested in its design. These decisions provide a shared understanding of how to define and describe devices and their functions. As mentioned above, while BACnet and LONtalk are chiefly focused on building automation and industrial control, the ZigBee work so far has produced common definitions for home automation and smart energy. Choosing CAP preserves this design investment.

presents a unified approach to networked embedded systems development.

For more information on rugged Themis servers or its new T2BC™ blade servers that run Solaris™ applications on the Sun® UltraSPARC® T2® processor, please visit us at www.themis.com or call (510) 252-0870.

Transformational. ©2008. Themis Computer, Themis, the Themis logo, Rugged Enterprise Servers, and T2BC are trademarks or registered trademarks of Themis Computer. All other trademarks are the property of their respective owners.

Untitled-7 1

AM January9/11/08 2009 11:25:38 17

solutions en g ineerin g

Wireless Sensor Networks

Wireless Sensor Networks: Maintenance-Free or Battery-Free? Two recent major technology waves were the cell phone and wireless Internet (Wi-Fi). Now there is a third wireless wave coming: wireless sense and control networks that can connect and control all kinds of equipment in our buildings, homes and businesses. by C ees Links GreenPeak

B wakes up

A and B Communication Cycle

A wakes up

A awake period

B awake period

A and B Communicate

Time

C wakes up

C awake period

B and C Communicate

C goes to sleep

B and C Communication Cycle

B wakes up

B awake period

B goes to sleep

January 2009

B goes to sleep

Node A wants to pass a message to node C

A goes to sleep

oday, we are entering this third wireless wave. Also known as “The Internet of Things,” the third wave utilizes wireless sense and control technology to bridge the gap between the physical world of humans and the virtual world of electronics. Sense and control networks do not enhance human communication. Instead, they allow sensors to interact with actuators, creating a more dynamic world and avoiding error-prone, monotonous and costly human intervention. However, the strength of wireless sensor networks can only be fully achieved when the wiring for both the data communication and the power supply is eliminated. The standard wireless sensor network solutions that are available today only solve a wiring problem in sensor applications: making networks easy to install. However, ultra-low-power wireless network solutions can also address the maintenance problem inherent to networks with a high number of nodes and a limited battery life. For example, a network of 4,000 nodes and a battery life of 10 years, means that on average 1 battery per day needs to be changed. Because the true cost of wireless sensor networks has shifted into the area of

Time

Figure 1

By precisely timing and overlapping awake periods, the mesh network is able to route messages from node to node and on to the ultimate destination.

the maintenance cost, wireless sensor networks that gather industrial or machine data have yet to become a cost-effective solution. The majority of the sensors used are still battery powered; these batteries

require regular changing and or recharging. In addition, reintegrating the downed nodes after battery maintenance, further adds to this onerous labor expense. To avoid these high costs, the industrial sec-

SOLUTIONS Engineering

tor and its applications require self-powered nodes that require very little power to operate for long periods of time. This is generating the rapidly emerging opportunity for ultra-lower-power wireless and energy harvesting. As you can imagine, there are many different types of applications that will be able to benefit from ultra-low-power wireless sensor networks. These include monitoring of temperature, vibrations, humidity, position, tank levels, etc. in industrial plants and manufacturing. They can also be linked to the control and actuation of HVAC systems, storage, robot movements, temperature control, etc. But there are many others that are not so obvious. For example, agricultural applications now benefit from the use of wireless sensor networks when temperature sensors or soil moisture sensors are used for remote monitoring of test fields, vineyards or green houses and to control irrigation and fertilization. For many real-world applications, the third wave of wireless—ultra-low-power wireless sensor networks—will provide many advantages including the cost elimination of hard-wiring, the enhanced flexibility in constricted or dangerous areas, ease of installation, increased safety and reduced maintenance costs of sensor deployments. The challenge of designing wireless sensor applications is not limited to deploying reliable wireless communication. Power management is an even bigger challenge. This should not be a surprise since the real benefit in wireless communication is primarily to avoid the wiring cost, so the data cables as well as the power cable need to be eliminated. The biggest technical challenge for developing ultra-low-power sensor networks is managing the energy consumption without reducing range or functionality, like speed and standards compliance. The resulting elimination of battery replacement will then simplify maintenance and provide a higher level of ease of use and safety.

Ultra-Low-Power Consumption

It is obvious that current consumption—milli-amps—and duty cycling are important in wireless sensor networks.

Energy Harvesting Devices Energy harvesters can usually be divided into two categories: the “tricklers” and the “bursters.”

SolarCells The best-known energy trickler is the solar cell, and certain varieties of solar cells can even pull energy from the limited light that is available indoors. To be usable for sense and control networks it is important that the energy is stored and also that the usage of the energy is controlled. A task such as a data transmission is only started when there is enough energy to finish it, e.g., receive the acknowledgement. Solar cells have a big advantage: they are relatively cheap. They also have a serious drawback: they only work when there is light. That is not very good news, as most networks need to work reliably and function 24/7 independent of lighting conditions.

Peltier Devices Another trickler energy harvester is a peltier element that taps into the energy stream that comes into existence when two sides of one object have a different temperature, such as the wall of a house where the outside temperature is different compared to the inside temperature. Usually a temperature difference of 5 degrees Celsius gives enough usable energy. Unfortunately, the dependability of peltier elements is limited because temperature differences usually cannot be controlled, and compared to solar cells, peltier elements are relatively expensive.

Dynamos The most commonly used energy burster is the dynamo, a sort of reverse step engine, which rather than using energy to create motion, uses motion to create energy. Actually, this form of energy creation is quite crude and, once set into motion, produces an amount of energy that is way beyond the requirements of a sensor network and needs to be burned off. This type of energy harvesting is also relatively expensive.

Piezo Devices Another energy burster is the piezo element, where energy is created from mechanical torsion. Also, here, the amounts of energy are relatively large. Piezo elements are also quite sensitive to wear and tear. Making these piezo elements reliable and durable makes them also quite expensive. Looking at the status of the energy harvesting industry, it is obvious that we still have a long way to go. The cost is still too high to be practical. At this moment no inexpensive and high-volume solutions are in sight, with exception of the solar cell, but many companies are pursuing promising ideas. Solar cells are the only energy harvesters that are producing in high volume and at low cost as required by sense and control networks. But to overcome its “dark shortcomings,” small batteries are still needed to store power when no light is available. The good news is at least we are somewhat reducing mounting reliance on batteries and their heavy metals and toxic chemicals.

However, minimizing current consumption is only part of the solution. Five other essential issues are key to developing lowpower wireless sensor applications, the first of which is low-power Wireless Mesh Routing. One of the most dramatic differences between wireless sensor communication technology and other well-known wireless technologies is the ability of sensor nodes to forward messages from other nodes further down a communication chain. This is called mesh routing or multi-hop networking. Mesh networking is an ef-

fective and reliable solution for spanning large infrastructures beyond the range of what a single wireless link can do. In a low-power mesh network, all the nodes, including the mesh routing nodes, operate in low power. Figure 1 depicts how low-power routing works when Node A wants to send a message to Node C, through Node B. All nodes in the pictures are low-power nodes and are in sleep mode most of the time. The breakthrough lies in synchronizing the sleep/wake-up cycles of the nodes to each other. This means a node wakes January 2009

SOLUTIONS Engineering

C1 B

Div Switch C2

Transceiver

PA1

GP500C

Figure 2

Antenna diversity operation.

up when it can expect a message from a neighbor node. This works through very precise synchronization of the transmitting and receiving nodes. As a result, the routing nodes will also be in a nearly powerless sleeping state most of the time, achieving ultra-low-power operation. The more accurately the wake-up schedule can match the communication expectations, the less power is consumed by unnecessarily long wake-up periods. The idea is to let a node be awake in certain “slots.” A node cycles through a period when it is asleep and a period when it is awake. For nodes to be able to communicate with each other their “awake” slots need to overlap, meaning that they have to be awake at the same time. This requires an innovative approach for synchronizing “time” between nodes. The accuracy of the time synchronization will determine the minimum time a node needs to be awake. The repetition rate of the awake slots is predefined for the network and depends heavily on the required responsiveness of the network. For example, how often does a sensor value need to be read out. Another critical design parameter is peak current. When closely examining the power consumption behavior of electronic circuits, it becomes apparent that what looks like a flat current curve at first glance, actually bears more resemblance to a mountain range with peaks and valleys. When certain functional blocks become active, they cause a peak. When two functional blocks switch on simultaneously, they cause a peak of double height. The secret in reducing the peak power lies in carefully managing when the functions are turned on and off and avoiding double peaks at all times. When an energy source has dried

Untitled-7 1

January 2009

11/10/08 10:01:37 AM

up, the electronics cannot communicate and are dead to any meaningful purpose. They need a way to fail gracefully. Loss of power can be a normal event such as a solar cell at midnight or an exceptional condition such as a depleted battery. In both cases the power problem can be dealt with, provided the application is intelligent enough to detect the upcoming problem before the energy source has completely dried up. During this last breath the device should perform a number of actions to inform its environment of the situation, transmit some critical data and put itself in a state that allows fast recovery when the power is restored. By using a built-in Graceful Power Failure function, the system is able to monitor various types of low-power energy sources including batteries and solar cells. They carefully monitor the state of the power circuits and raise different levels of alarms ranging from an early warning to near-death. These alarms are escalated to other parts of the system such that they all can move into a state that fits the alarm condition. Wireless chips are usually quoted on their power consumption in receive and transmit mode. However, in order to achieve low power, the devices are dutycycled, moving between alternate states of sleeping and being awake. The longer the battery lifetime needs to be, the longer the device will sleep between wake-up periods. Unfortunately, electronic circuits are never really “sleeping.” Although the powered-down circuits are not doing anything meaningful, a small leak current flows through the transistors. The leakage can amount to several 10s of µA. Sleep current is not usually considered an important design factor, but it becomes extremely important when designing a circuit to live for 5 years or more on a battery, sleeping most of its life. Finally, there is the matter of wakeup time. As duty-cycling is achieved by switching off as many parts as possible during the sleep state, all these parts need to be re-activated when returning to the active state. Voltage regulators need to settle, clock oscillators need to start up, digital electronics that were powered down need to be put in a known state. In order to avoid unnecessary energy con-

SOLUTIONS Engineering

Reliability

The reliability of a wireless sensor network is related to the availability or absence of a communication path between two wireless devices. The most important enemy of reliability is wireless interference originating from other users of the same frequency band. The most notable interferers for IEEE 802.15.4-based devices that operate in the 2.4 GHz frequency band are Wi-Fi transceivers. Most interferers will not fully block out an IEEE 802.15.4 device, but will cause some wireless packets to get lost regardless of the network stack operating on top. The industrial standards provide a mechanism that allows packet losses to become evenly spread out over time, even if the number of lost packets will not substantially decrease. To further increase reliability some companies have added antenna diversity to the wireless sensor network communication features. Antenna diversity is based on the use of two antennas to improve the quality and reliability of a wireless link. Often there is not a clear line-of-sight (LOS) between transmitter and receiver, and the signal is reflected along multiple paths before finally being received. Each of these bounces can introduce phase shifts, time delays, attenuations, and even distortions that can destructively interfere with one another at the aperture of the receiving antenna. Antenna diversity is especially effective at mitigating these multipath situations because multiple antennas provide a receiver with several observations of the same signal. Each antenna will experience a different interference environment. Thus, if one antenna is experiencing a deep fade, it is likely that another has a sufficient signal. Collectively, such a system can provide a robust link. Figure 3 illustrates how diversity antennas work. In receive mode, the signals are fed to the diversity switch, which is controlled by a signal in the DSP inside

the transceiver. The DSP controls the switching (by control signal C1) and will select the best signal antenna. In transmit mode, one of the control signals C2 or C3 enables one of the power amplifiers PA1 or PA2. B represents a balun that acts as an impedance transformer. For wireless sensor transceivers, the dominant and probably only real standard is the IEEE 802.15.4 specification. However, there have been efforts to use Bluetooth and Wi-Fi for sensor applications. In all the cases reported, Bluetooth and Wi-Fi were used in a non-standard way, in fact weaving the principles of IEEE 802.15.4 into their native implementation. It is nowadays widely accepted that the IEEE 802.15.4 offers the best basis for wireless sensor applications. Even within the boundaries of standards, technology providers can discover and exploit differentiation opportunities. As an example, GreenPeak has developed Transceiver and Network Stack technology that is compliant to the IEEE 802.15.4 standard but includes additional functionalities that enable its use for ultra-lowpower applications. An ultra-low-power application is defined as an application that is able to live off a coin cell battery or off energy harvested from the environment through a solar cell, a vibration energy harvester or any other environment energy converter. See Sidebar “Energy Harvesting Devices.”

Highscore for

Lowest Power to Highest Performance Embedded Module Family Ultra Low Power conga-CA with Intel® Atom™ Z5xx and Intel® System Controller Hub US15W

Low Power conga-CA945

GreenPeak Technologies. **NEED CITY & STATE (512) 464-1188. [www.greenpeak.com].

with Intel® Atom™ N270 and Intel® 945GSE Chipset

Economy conga-B915 Intel and Intel Atom are trademarks of Intel Corporation in the US and other countries.

sumption, it is very important to make the start-up times as short as possible without compromising current consumption in normal operation. Smart sequencing of such start-up cycles is crucial for ultralow-power operation.

Untitled-5 1

with full feature set & economically priced Intel® Celeron™ processor

Performance conga-B945 up to 2.16 GHz Intel® Core™2 Duo

High End conga-BM45 Intel® Core™2 Duo up to 2.53 GHz, 6 MByte L2 cache and 8 GByte DDR3 RAM, high performance integrated graphics engine

Find more details at : www.congatec.us Tel. 7 6 0 - 6 3 5 - 2 6 0 0 sales-us@ congatec.com

January 2009

11/17/08 3:40:38 PM

INDUSTRY INSIGHT

Interconnect Wars

Interconnect Wars? Let Peace Prevail with Interconnect Standards COMs were developed to solve certain problems with SBCs but have some issues of their own when developing the needed carrier boards for I/O. A combination of standard interconnects may bring a solution for painlessly mating CPU to I/O.

by R obert A. Burckle, WinSystems and Colin McCracken, ADLINK Technology

ize, cost, performance, application software development and debugging tools, wired or wireless networking, power budget, time-to-market, temperature range, expansion flexibility, RoHS compliance, ruggedness, interfacing to other peripherals, cabling, packaging, documentation, quality, technical support, certifications—whew! System designers are faced with numerous large and small decisions that impact the success or failure of their projects. Fortunately for the embedded computing element, there are a variety of options for off-the-shelf, standard single board computers (SBCs) and ComputerOn-Modules (COMs) available in all shapes and sizes. Each has its own technical advantages and limitations. The good news is that an engineering team does not have the challenge of creating a custom computer, but rather they can start with a known-good design and focus on their core competency—the application. Therefore, the question facing designers is, “What architecture and form factor should I choose as the embedded computer for my project?”

January 2009

SBCs, COMs and Small Form Factors

Designers have turned toward industry standards and consequently changed from in-house proprietary designs to commercially available embedded x86compatible products. This shift in design methodology brings well-designed, technologically advanced system components at reasonable prices, complete with device drivers and application software building blocks and tools. The amount of time available to develop an embedded system has decreased significantly with times now as short as 3 to 6 months. The popularity of embedded PCs lies in the ability of a user to buy off-the-shelf hardware and customize it with software. The result is a better product developed in a shorter period of time. Over the last decade many different standard board form factors have emerged in the embedded industry for use in industrial, Mil/COTS, communications, transportation, instrumentation and homeland security applications. Each form factor standard defines its size, mechanical dimensions, input/output capability and expansion capabilities

with other boards, mounting and certain other electrical characteristics. Many of these different form factors are managed by an industry consortium or manufacturing group of interested companies so as to monitor, improve and promulgate open standards worldwide. Some of these groups charge for the standards documentation while others are free. Independently managed groups include PICMG (www.picmg.org), VITA (www. vita.com), SFF-SIG (www.sff-sig.org), PC/104 Consortium (www.pc104.org) and others. Many of the popular small form factors (SFFs) are shown in Figure 1. For a given form factor, a board that includes the processor, memory and peripheral controllers can be classified as either a Single Board Computer (SBC) or Computer-on-Module (COM). A COM is not a single board computer, but a system module “component” that must be plugged into a baseboard or “carrier card.” Historically, standard form factors and expansion interfaces have been unnecessarily intertwined, creating barriers to migrating from SBCs to COMs for example. Both have similar features and functions; however, they differ in how they

INDUSTRY Insight

are powered, how onboard I/O is brought out to the “real world” and how the system is expanded with additional I/O.

PC104 PCI104 PC104+

Applying and Managing Power

SBCs are powered directly from a power supply, whether single voltage DC, wide input range DC, multiple DC supplies with power button control (ATX-style), or other. A power cable connects the power source to the SBC. On the other hand, COMs are powered through many small fine-pitch pins on the same surface-mount connector(s) as the bus and I/O signals that run between the COM and the baseboard. Therefore, a COM cannot run by itself in a system; it must be part of a minimum twoboard solution. The power connector resides somewhere on the baseboard. Enough connector pins must be allocated for power and ground nodes to present low series resistance and inductance for good DC and AC characteristics, to meet bulk current load requirements, and to keep EMI emissions and susceptibility in check. Power management using the Advanced Configuration and Power Interface (ACPI) specification is trickier with COM architectures, as the original standard was developed for motherboards that contain a BIOS, OS, chipset, power connector, RAM, video and disk interfaces all on the same board. Whereas embedded SBCs handle most or all of the issues depending on the intended level of support, COMs that are loosely defined shift the burden from COM supplier to OEM customer to get their custom baseboard working with the “black box” module.

I/O Considerations

With an SBC, I/O is brought off the board through a set of connectors, and cables are attached directly to them. In some cases, these connectors are “PCstyle” allowing a LAN cable, printer, keyboard or USB device to connect directly to the board. In other cases, these connectors are pin-headers, requiring a transition cable with another connector at the end that mounts to an enclosure bulkhead or a pin-header that connects to another PCB. In either case, the SBC is a fixed form factor and engineers must design their enclosures and cable schemes to match that of the SBC. An SBC can

COM Express (ETX Express) “Extended”

3.5 Inch (Half Biscuit)

EPIC EPIC Express

EBX

COM Express (ETX Express)

5.25 Inch (Biscuit)

“Basic”

ETX & XTX Common Embedded Computer Form Factors PC104, PCI104, PC104+ EPIC, EPIC Express EBX

3.55” x 3.77” (90mm x 96mm) 4.528” x 6.496” (115mm x 165mm) 5.75” x 8” (146mm x 203mm)

3.5 Inch 5.25 Inch

5.7” x 4” (146mm x 102mm) 5.75” x 8” (146mm x 203mm)

ETX & XTX COM Express Basic COM Express Extended

4.5” x 3.7” (114mm x 95mm) 5” x 3.7” (125mm x 95mm) 6.1” x 4.33” (155 x 110mm)

Mini ITX Nano ITX Pico ITX Mobile ITX

6.7” x 6.7” (170mm x 170mm) 4.72“ x 4.72” (120mm x 120mm) 3.9” x 2.8” (100mm x 72mm) 3” x 1.8” (75mm x 45mm)

Figure 1

Mobile ITX PICO ITX NANO ITX

MINI ITX

Common embedded computer form factors.

be used in a stand-alone configuration or expanded with off-the-shelf or custom designed I/O boards. With a COM module, I/O is brought to a baseboard that is developed by the OEM or by a third-party design house commissioned by the OEM. This custom baseboard is a size that best fits the application and its enclosure or packaging requirements. This means that both the style and location of the I/O connectors match the mechanical constraints of the application in question. Using a COM card is great for applications where the creation of a custom carrier card is not considered a handicap due to cost or time constraints. However, for most low- and medium-volume applications, SBCs are the better choice, alleviating NRE costs or design resources, risk and time-to-market.

Expansion Bus Considerations

In terms of how to attach additional circuitry to system buses, COMs and SBCs

Figure 2

Adlink’s EBX baseboard for COM Express modules.

are similar, despite physical connector differences (slots versus pin-in-socket stacks versus SMT all-in-one connectors). Popular PC architecture buses include PCI and ISA, with PCI Express poised to replace PCI due to its higher throughput while maintaining software compatibility. ISA may be gradually fading away, with several low-speed interfaces in the modern PC chipsets now available to help migrate legacy peripherals and custom logic. January 2009

INDUSTRY Insight

Figure 3

Diamond Systems’ EPIC carrier board places the ETX COM module underneath.

Figure 4

WinSystems’ EBX form factor baseboard with COMIT and SUMIT expansion connectors.

In the realm of buses, the greatest similarity between SBCs and COMs exists between the PC/104-Plus bus (PCI and ISA) and the ETX module de facto standard (which also offers PCI and ISA). Outside of those two, SBCs and COMs tend to diverge quickly. With the all-inone COM connector approach, supporting newer interfaces like PCI Express, LPC and I2C becomes a simple matter of assigning pins within the chosen connector. Industrial motherboards take an analogous approach by using standard desktop PCI Express, PCI and ISA slots, where expansion cards plug in vertically (at a right angle to the motherboard). This puts pressure on the overall size, but does tend to spread the heat dissipation evenly around a large volume. Stackable SBCs have lagged due to divergent philosophies about what to support going forward, since tiny small form factor boards do not have the space to support everything. Tough choices are made by architects with different application priorities in mind. Furthermore, in the

January 2009

small form factor realm, thermal issues get complicated very quickly.

Meeting in the Middle

Given the expense, time and risk of developing a custom baseboard, many embedded system OEMs prefer to start prototyping with existing standard SBCs. In some cases, enclosures were designed around legacy or EOL single board computers based on the EBX form factor, for example. COMs offer the ability to easily replace obsolete processors and chipsets in order to keep systems in production with minimal impact. A hybrid approach using COMs on standard-form-factor production baseboards avoids the tasks of turning a huge ATX-style reference design into a custom small carrier. The off-the-shelf EBX-size baseboard shown in Figure 2 accepts COM Express modules, forming a hybrid SBC+COM solution intended to drop into existing EBX-based applications. Alternatively it serves as a small form factor baseboard reference design when volumes justify a custom baseboard.

One drawback of existing COMs defined by true standards bodies 3-5 years ago is that the modules are too large to fit on small form factor SBCs without blocking the expansion bus interfaces. The baseboard in Figure 2 supports many features, but has a PCI Express x1 slot for expansion instead of a stackable interface. A second example is the design approach shown in Figure 3, which places the COM on the bottom of the EPIC form factor baseboard so that the top side can have a PC/104-Plus expansion interface without being obstructed by the COM module. This works well for new designs, but could require minor modifications to existing enclosures to drop in. However, with the advent of tiny, ultra-low-power processors like Intel’s Atom and VIA’s Nano CPUs, it is now possible to define a new, smaller COM standard so that the module can fit easily on the top side of the baseboard without bumping into expansion I/O cards whose locations were previously defined by EBX and EPIC form factor specifications. The challenge is met by a new standard called COMIT from the Small Form Factor Special Interest Group (SFF-SIG). “COMIT” stands for Computer-On-Module Interconnect Technology, and can be implemented on modules as small as 62 x 75 mm as shown in Figure 4.

Examining the Atomic Structure

The COMIT module at the left edge of Figure 4 steers clear of the vertical space needed for expansion I/O cards. COMIT takes on the role of a pluggable CPU core, much like a Socket 370 or Socket 478 processor. COMIT is a ground-up new computer-on-module architecture, defined by rugged SBC vendors for rugged SBC applications, that targets the space-efficient, low-power Atom and Nano processors. A rugged connector pair from Samtec has been selected to live up to the quality and reliability demands of military, avionics, transportation and industrial applications. Another new standard from SFFSIG is the SBC expansion interface called “SUMIT,” for Stackable Unified Module Interconnect Technology. COMIT allows newer x86 interfaces like PCI Express, LPC and SPI to go to the baseboard, and

INDUSTRY Insight

a mid-size SBC form factor can add circuitry like Gigabit LAN and legacy serial ports and then pass the rest of the unused resources and shared buses to the SUMIT interface where off-the-shelf and custom I/O cards can be added. The SUMIT-A and SUMIT-B connectors are at the center of the board in Figure 4. Worst case, rounding out the I/O means designing a simple SUMIT card to some standard small dimension, which is much simpler than taking on the power supply, power management and termination of standard I/O that are unintended burdens passed along to the OEM by COM manufacturers. The value of being up and running on day one is “priceless.” COMs can enable SBCs, showing more collaboration than competition between the two approaches. COMIT is designed to work with EBX, EPIC or custom form factors to “right-size” a board and its processing power to an application. It also helps to future-proof the system design due to the “socket” approach and philosophy of the CPU core element. SBCs and COM products coexist peacefully in a growing off-the-shelf board market where OEM design expertise, design time frames and design and product costs drive solutions either to the COM side or the SBC side. The real question is, “How much processing power, bandwidth and expansion capability is needed for a small form factor solution for your embedded application?” Computeron-Module products are typically used for higher volume applications where the designer can benefit from implementing a custom I/O board to minimize or eliminate cabling and reduce assembly costs. A COM module must be mounted on the baseboard for all additional I/O features and connections, just like a large socketed component. However, a custom baseboard design requires an in-house design team or contract design for the development, which means added cost, more time and limited flexibility for new or additional requirements (sometimes known as marketing “improvements”) for the applications. This expense must be amortized over a large volume of boards to be cost-effective. Modern board-to-board interconnects and off-the-shelf (standard prod-

uct) baseboards are making it easier for OEMs to get the benefits of both SBC and COM architectures. Together, COMIT and SUMIT once again make it possible to use off-the-shelf components when volumes are modest, with a path toward custom baseboards later when volumes grow. Information about the new standards can be found at www.sff-sig.org.

WinSystems Arlington, TX. (817) 274-7553. [www.winsystems.com]. ADLINK Technology San Jose, CA. (408) 360-0200. [www.adlinktech.com].

January 2009

system integration

Machine to Machine

Service-Oriented Devices Will Require Standards for Success As a new generation of devices emerges that can provide services by accessing Web-based information as well as the information and functions of other devices, a set of standards is needed to let them interact seamlessly across manufacturers. by O livier Bloch Microsoft Windows Embedded

ervice-Oriented Applications (SOAs) have proven to be successful and are now an accepted part of application development and deployment. It had become simple for an application to display or use up-to-date weather data; the application merely retrieves this information from an online Web service. With SOA proven and widely accepted—even expected—as a basic feature for connected devices, the industry is further expanding on this trend. The next stage is to focus on developing Service-Oriented Devices (SODs), with “devices” in this context referring to any equipment running embedded software. Examples of these devices are everywhere: in cars, ATM machines, GPS receivers, television receivers and recorders, restaurant ordering stations, self-service checkout lanes (Figure 1), and many other places. These devices are rising in ubiquity and taking advantage of technology trends such as: • The prevalence of Internet connectivity at most public and private destinations, including Wi-Fi, WiMax and cellular data services (3G, EDGE, etc.) • The symbiotic relationship between Web services and applications, sharing and consuming information, including industrial standards and development tools and platforms such as Visual Studio and .NET • Connected devices that consume and exchange data from servers and databases • The tremendous growing ubiquity of embedded devices. These devices can do more now than just connect and consume data. For one thing, they can leverage existing Web services. For example, a personal navigation device should be able to provide information like traffic, gas prices and points of interest nearby. All devices should be able to present information

January 2009

from their sensors or user input as services. With that in mind, devices should also broadcast their functions as a service so they can be identified and discovered by any other appropriate device. Picture your house filled with intelligent devices such as heating controllers, light switches and dimmers, security cameras, smoke and water sensors, and set-top boxes all exposing their functions and data as services. These devices can then be discovered by other devices, desktop and laptop computers, and local and/or remote servers. Once everything is interconnected like this, a limitless number of scenarios arise.

A Real Need for Industry Standards Support

However, for these and other scenarios to exist, the industry needs recognized standards that enable service-oriented devices to be discoverable, to work together and to expose information and functions as services. Right now, existing solutions are often proprietary and rarely interoperable, making such solutions expensive and not appealing for consumers or developers. In another example, you’d like to expand an existing home automation system with devices from a different manufacturer, which can be especially important if the original manufacturer has discontinued or no longer supports those device models. In this particular case, it’s doubtful you’d find another manufacturer supporting the same field bus, protocols and interfaces as on the original device or devices. Furthermore, manufacturers aren’t inclined to create, maintain or update a proprietary solution due to expense and lockdown of the product line. Using industry standards offers more options like integrating third-party solutions, easy outsourcing of maintenance and support, and leveraging the skills and experience of a greater number of engineers and integrators. Standards do exist today, and Original Equipment Manufac-

system integration

turers (OEMs) do realize the need and advantages of implementing these existing industry standards. But so far they face big issues when trying to do so on their devices, as they’ve been unable to rely on operating systems to provide implementations for these standards. Thus, the OEMs needed to work on developing or integrating missing stacks, things that are definitively out of their expertise and not part of their core business.

Figure 1

Restaurant ordering stations such as this Embedded CE-based device can also communicate with wireless handheld devices used by wait personnel.

Service-Oriented Standards and Windows Embedded

Device Profile for Web Services (DPWS) is a means by which existing industry standards can enable service-oriented device development. It’s described in a document that can be found at: http://specs.xmlsoap.org/ws/2006/02/devprof/. As a subset of the Web Services standard, DPWS defines a minimal set of implementation constraints to enable secure Web service messaging, discovery, description and eventing on resource-constrained devices. The DPWS implementation in Windows Embedded CE 6 is called Web Services on Devices (WSD). Once a device is “WSD enabled,” it communicates in a standardized way with many classes of devices, such as restaurant stations with the handheld order-taking devices used by the waiters (Figure 2). This allows application developers to write software generically to a device class and maintain compatibility across devices from a wide range of manufacturers. Easy development saves device vendors time working on the network layer for communication and lets them focus on investing in rich applications for their chosen devices. Other industry standards like connectivity technologies such as wired or wireless LAN and Bluetooth, or protocol stacks (HTTP, TCP/IP, etc.) do exist and are used because they’re reliable, secure, well-defined and fully implemented in available operating systems. When developing a Windows Embedded CE or Windows Embedded Standard kernel using Microsoft tools, you build your OS from scratch. Adding a TCP/IP stack and HTTP support into your kernel image is as simple as selecting a component from a catalog. This allows manufacturers to spend more time adding value to the system they are building and less time “reinventing the wheel” by redeveloping or integrating technologies that should already be “in the box.” In some vertical domains for devices, there are currently no

industry standards for operation. Microsoft is working on filling those gaps by promoting and supporting these standards, like the Decentralized Software Services Protocol (DSSP). This simple SOAP-based application protocol defines a lightweight service model with a common understanding of service identity, state and relationships between services. DSSP contains a set of state-oriented message operations that supports structured data retrieval, manipulation and event notification. The protocol provides a flexible foundation for classifying applications as compositions of services interacting in a decentralized environment. DSSP’s functions are an extension of the application model provided by HTTP, and are designed to be used on top of existing HTTP infrastructure. While Microsoft is proposing some standards for areas that can utilize them, its biggest effort is on developing and providing platforms that implement existing industry standards. The technologies and tools supporting these industry standards are all available under the Windows Embedded OS family. Another really interesting technology Microsoft is providing is called .NET Micro Framework, a bootable .NET runtime running directly on top of the hardware and supporting hardware-constrained devices. The .NET Micro Framework fits into 500 Kbytes and can run on non-MMU architectures, enabling hardware access directly from .NET code. This runtime supports communication stacks and other interesting protocols, including a managed implementation of DPWS.

Opportunities for OEMs

The first big opportunity for OEMs using industry standards is leveraging existing implementations and stacks; shopping and integrating bits from different vendors or choosing a platform providing all the features “in the box.” Microsoft’s approach resembled the latter by providing the OEM with fully integrated platforms and tools that enable the OEMs to focus on device value, fast time-to-market and lower total cost of ownership. Another opportunity for OEMs to use industry standards is to build interoperable devices. For example, a picture frame manufacturer might offer a wireless picture frame that accesses Web services like Flickr or Live Spaces in order to provide a competitive feature for customers. A home automation device manufacturer would also rather give its customers the ability to use one of their devices even where the customer’s existing installation was purchased from a competitor. Service-oriented technologies also represent a major opportunity for OEMs. Thinking of “services” in terms of architecture brings advantages for development, maintenance and updates of a system. Previously, an OEM had to completely redeploy a system for an update or to add a new component. Now, the OEM can simply update a service that provides new features or a new implementation without interrupting or halting the rest of the system. This also allows the OEM to accurately scale the system globally, dispatch task execution in a distributed architecture, and leverage the computational power of the existing infrastructure. A good example of this setup is how CCR/DSS technology from Siemens speeds up the Delivery Point Resolver address lookup January 2009

coM express and FPga Boards showcase

Featuring the latest in COM Express and FPGA Boards technology SOM-5761 Intel® Atom™ N270 COM-Express CPU Module

Advantech Phone: (800) 866-6008 Fax: (949) 789-7179

Embedded Intel Atom N270 1.6GHz processor with low power consumption Intel Atom N270+ 945GSE+ ICH7M Intel GMA950, Microsoft DirectX* 9.1 Support multiple display: VGA, LVDS, SDVO, TV out Intel 10/100/1000 Mbps LAN Supports wide range power design: +8.5V ~ +19V Supports 4 PCIe x 1 or 1 PCIe x 4, 4 PCI devices Supports 2 SATA ports, 8 USB 2.0 ports e-mail: ecginfo@advantech.com web: www.advantech.com

SOM-6760 Intel® Atom™ SCH US15W COM-Micro CPU Module Intel Atom processor Z510 1.1GHz, Z530 1.6GHz Intel System Controller Hub US15W Supports DDR2 SODIMM up to 2GB Intel Graphics Media Accelerator 500, DirectX 9Ex H.264, MPEG2, MPEG4, VC1 and WMV9 HW decode Display type: 24-bit LVDS/VGA Intel 10/100 Mbps LAN LPC, 1 PCIe x 1, 3 PCI master 8 x USB 2.0, 1 EIDE, 8-bit GPIO, SDIO

Advantech Phone: (800) 866-6008 Fax: (949) 789-7179

conga-CA945 Intel® ATOM™ N270 1.6GHz+945GSE Shipping Now!

congatec Phone: (760) 635-2600 Fax: (760) 635-2601

User friendly congatec embedded BIOS with API for easy software development SODIMM socket for up to 2GB DDR2 memory Dual independent display supported ACPI+ battery support GbLAN, 2 x SATA, and 3 x PCIe Compact 95mm x 95mm size Driver and BSP support Includes CGUTIL software tool enabling the OEM customer to generate their own specific CMOS BIOS defaults, OEM splash screen, and unique flat panel support e-mail: sales-us@congatec.com web: www.congatec.us

FreeForm FPGA - Maximum Performance, Multiplatform FPGA Boards FreeForm/PCI-104 - Virtex-5 FPGA, 3,000,000 gates, Flash/DDR2-400 memory, Ethernet, LVDS I/O, designed for embedded processing (MicroBlaze), Power PC FreeForm/104 – Spartan-3E FPGA, 500,000 gates, digital I/O, counter/ timers, Opto-22 compatibility FreeForm products include FPGA reference designs, industrial temperature range and lifetime warranty

Connect Tech Inc. Phone: (519) 836-1291 Fax: (519) 836-4878

FPE320 3U VPX Xilinx Virtex-5 FPGA-based Processor Card

Phone: (613) 254-5112 Fax: (613) 599-7777

e-mail: sales@cwcembedded.com web: www.cwcembedded.com

e-mail: sales@connecttech.com web: www.connecttech.com

VPX3-450 3U VPX MPC8640D and Xilinx Virtex-5 FPGA-based Processor Card

Xilinx Virtex-5 SX240T, LX155T/220T/330T or FX130T/200T FPGA FMC (VITA 57) mezzanine site Two banks of 256MB DDR2 SDRAM, 512MB total Two banks of 9MB QDR2 SRAM, 18MB total 3U VPX form factor with .8” Pitch Air and conduction-cooled options

Curtiss-Wright Controls Embedded Computing

e-mail: ecginfo@advantech.com web: www.advantech.com

Xilinx Virtex-5 LX155T or SX95T FPGA 1 GHz Freescale Power Architecture MPC8640D dual-core processor XMC mezzanine site One bank of 256MB DDR2 SDRAM Two banks of 9MB QDR2+ SRAM, 18MB total 3U VPX-REDI with .8” Pitch Air and conduction-cooled options

Curtiss-Wright Controls Embedded Computing Phone: (613) 254-5112 Fax: (613) 599-7777

e-mail: sales@cwcembedded.com web: www.cwcembedded.com

engine at the United States Postal Service for dramatically faster mail processing and delivery. The service-oriented model and the concurrency and coordination runtime proposed in the CCR/DSS toolkit (based on DSSP) helps correct unclear, incomplete or incorrect address information for up to 70 pieces of mail per second.

coM express and FPga Boards showcase XMC-442 Xilinx Virtex-5 FPGA-based XMC mezzanine module Xilinx Virtex-5 SX50T/95T FPGA One bank of 256MB DDR2 SDRAM Two banks of 9MB QDR2+ SRAM, 18MB total Three high-speed serial ports XMC form factor, 4-lane PCI Express Air and conduction-cooled options

Avoid Overwhelming the Final User

OEMs must also face the fact that too many different technologies and too much technology negatively influence how technology can be used as a solution. When you buy a wireless picture frame, you have to configure it in order to get it connected to your home network. After you’ve done that, you need to set up the device for Internet access, and only then can you download the images you want. This multi-step configuration discourages many users who don’t want to spend time reading user’s guides and risk having their simple device not work for any number of reasons that could be easily avoided. Technology should make things simpler, not more complicated. This is where OEMs need to have standards-based technologies that scale from sensors to servers. OEMs understand the marketplace is changing. It’s no longer enough to make an embedded device, but to offer a device with services the users want and can use with the greatest amount of function and flexibility. To illustrate this point, these two scenarios show what these technologies can do. First, imagine someone at home, working on e-mail and watching news on TV. They’re planning on going out soon, and will be taking a city bus. The bus they are taking is able to geo-localize itself and “knows” it will arrive at the bus stop 5 minutes late. The dispatcher system can send a notification to subscribers interested in this particular bus line. The person at home can have that notification show on their calendar, on their TV through the set-top box, with an e-mail and with a page or message on their phone. This scenario shouldn’t involve the user in any configuration at any point. Obviously, all the mentioned devices would be coming from different manufacturers, but still seamlessly communicate with each other. Next, consider an implementer deploying printers that use service-oriented technologies like DPWS to be discoverable on a network. The printers would be able to notify a user that they are available for printing, when, say, a user’s default printer is unavailable. The printer would also notify the right support person for paper jams or paper and toner refills. Moreover, the printers can contact the service contractor directly when a serious technical issue occurs, or even when regular scheduled maintenance is required. The scenarios mentioned above aren’t just dreams or visions— some manufacturers are implementing them today. With the right use of industry standards and appropriate OS support for those standards, these possibilities and many others are within reach. Furthermore, in a fast moving and highly competitive market, device manufacturers need to rely on tools and platforms that maximize both value and differentiation to their service-oriented devices. Microsoft Embedded Redmond, WA. (206) 882-8080. [www.microsoft.com/windows/embedded]

Curtiss-Wright Controls Embedded Computing Phone: (613) 254-5112 Fax: (613) 599-7777

e-mail: sales@cwcembedded.com web: www.cwcembedded.com

Titan-V5e Credit Card Size – 2.125”W x 3.375”L x 0.25”H FPGA - Xilinx® VirtexTM-5 FX70T or SX50T Advanced SoC Processing Architecture Memory – Three Banks DDR2 SDRAM up to 533 MHz Operating System – Linux 2.6 with Real Time Modifications

iVeia, LLC Phone: (410) 858-4560 Fax: (443) 557-1680

e-mail: sales@iveia.com web: www.iveia.com

Kontron ETXexpress – COM Express™ Solutions 100 percent COM Express™ Compliant Processing performance up to 2GHz Intel Core™ 2 Duo Processors Low power VIA and AMD CPU based modules also available Up to 4MB system memory on ETXexpress COMs with dual channel memory support

Kontron Phone: (888) 294-4558 Fax: (858) 677-0890

e-mail: info@us.kontron.com web: www.kontron.com/ETXexpress

Setting the Standard for Digital Signal Processing

Pentek, Inc. Phone: (201) 818-5900 Fax: (201) 818-5904

Model 7153 – PMC/XMC Digital Downconverter with four 200 MHz, 16-bit A/Ds Four 16-bit, 200 MHz A/Ds Two or four DDC channels installed in Virtex-5 FPGA DDC decimation range from 2 to 256 or from 2 to 65536 Independent decimation factors for each channel Clock/sync bus for multiboard synchronization Conduction-cooled version available

e-mail: info@pentek.com web: www.pentek.com/go/rtc7153

industry wat c h

solid state storage

Taking the Guesswork out of Flash System Design Application modeling can help determine whether flash hardware and file systems meet requirements for performance and device longevity by understanding the attributes of flash memory and flash file systems.

by M att Porter, Embedded Alley, and Bill Weinberg LinuxPundit.com

evelopers of intelligent devices use ubiquitous and increasingly inexpensive flash memory in almost every type of application. Designs deploy flash to store code and data in mobile and deeply embedded devices up through SBC and bladebased applications. Flash technology selection usually occurs at the hardware level, with focus on cost per megabyte and data transfer rates. Later, platform developers and integrators must decide how to partition onboard flash, to enable OSs to boot, programs to load and run, and use that flash for data storage. Unfortunately, developers often lack insight into how application read/write behavior impacts flash performance, and vice versa. Designers find themselves forced to make educated guesses about viability of flash hardware and software for layers up the stack. To leverage application modeling, it is important to distinguish between the key terms and technologies. The two basic technologies, NAND and NOR flash, take their names from the operations for saving data to memory cells. Both erase by setting all bits to ‘1’, but differ in changing those bits by performing either NAND or NOR operations on the contents. NOR came first and today represents a more costly option targeting direct, bytewise random access. Here we will focus on optimizing NAND design with application modeling, although some flash file systems (FFSs) mentioned also support NOR. Beyond that decision, OEMs also need to choose from two types of NAND flash: Single Level Cell (SLC) and Multi-Level Cell (MLC). At its introduction, NAND flash boasted good performance and moderate capacity, with each (SLC) flash cell having two detectable states to encode one bit (Figure 1). Chip designers were later able to squeeze more capacity out of the same technology by implementing additional states per cell. Four-state MLC is

January 2009

Reference

0 Vt

SLC: 2 states, 1 bit

MLC: 4 states, 2 bits

Figure 1

00 Vt

SLC and MLC logic states compared.

standard today, with six states and beyond on the horizon. The higher density gives MLC a price/performance edge but exacts a cost in that there are smaller gaps between logic levels, which require sensitive read circuitry and can induce slower access and higher read error rates. Higher density exposes bits to effects from adjacent cells, and the additional complexity can reduce reliable deployment lifetime. The SLC-MLC gap can be bridged with software, using error correction code and write cycle counting. Application modeling can also clarify trade-offs. Additionally, in place of “raw” flash, OEMs can use “Managed NAND”—flash chips encapsulated as closed hardware devices. Managed NAND features built-in ECC, wear leveling and abstracted block interfaces, but limits choice of FFS and costs more than raw flash.

INDUSTRY Watch

NAND

NOR

Page

Byte

pressed images, but some designs need XIP to accelerate boot, application launch or for security. Pagewise access in NAND prohibits XIP. For Linux, NOR-based XIP requires cooperation between Read Cycle Fast Fast execv() and the file system, as well as tools to layout executable Erase/Write Cycle Fast Slow files correctly. Note that not every FFS supports XIP even if it supDensity High Low ports NOR. Moreover, XIP is only really practical in RO file systems, since NOR write cycles stall execution pipelines. Lifetime Write Cycles 1,000-100,000 10,000-1,000,000 Flash file systems can work directly on top of raw flash memRequires ECC Yes No ory chips or build on other subsystems. For example, JFFS2 builds Cost High Low on Linux Memory Technology Device (MTD). UBIFS relies on underlying Unsorted Block Images (UBI). UBI builds on top of Table 1 Key characteristics of NAND and NOR flash memory. MTD devices and includes wear-leveling and volume management. By carefully comparing these charactristics, developers can choose Wear Loop FFS R/W Compressed Journal De-Frag Mount Time from a range of legacy, current and exLeveling Back perimental FFS options, with varying JFFS2 R/W Long      attributes and virtues (Table 2). Random Access

YAFFS2

R/W

UBIFS

R/W



Moderate



Moderate



Flash Design and Deployment

Design and deployment involves more than using the cheapest parts CramFS R Short   and configuring kernel support. OEMs need to consider challenges common SquashFS R Short   to all flash and FFS types. One major UnionFS Reflects attributes and capabilities of underlying member file systems difference from rotating media is that flash is composed of erasable blocks Table 2 Comparing attributes of Linux Flash File Systems. that can endure finite erase cycles before becoming unreliable. Wear-leveling involves techniques for prolonging flash service Flash File Systems life using algorithms that spread erase/write cycles across the For embedded developers with background in “bare metal” device to avoid stress on any particular block. An FFS can impleRTOSs, the presence of any file system can seem superfluous. For ment wear-leveling directly, build on underlying subsystems, or developers deploying Linux or another general-purpose, “rich” can be delivered as a built-in as with Managed NAND and flash application OS, file systems are central to the platform. Flash file disk drives. systems are characterized by their different attributes, which can A journaling FFS stores blocks on raw flash devices or partiinfluence their suitability for various applications. tions in a ring buffer, appending recently used blocks to the tail For example, some file systems like CramFS or SquashFS and writing fresh data at the head. Allocation can be determinisare Read-Only (RO). RO file systems can store boot images, core tic or highly unpredictable. Deterministic wear-leveling depends application images and libraries, and other code and data that 6 define device personality. These RO file systems get overwritten 5 Maximum Erase Count 4 only by a complete rewrite of the partition (for upgrades, FOTA, 3 Mean Erase Count 2 etc.). Another attribute is whether the stored data is compressed. 1 0 Even with falling flash prices and rising capacities, most file sys2e+06 4e+06 6e+06 8e+06 1e+06 1.2e+07 1.4e+07 1.6e+07 0 tems offer compression, with RO file systems offering more agTime (seconds) 1 gressive algorithms. 0 Total Erase Block Moves An FFS needs fault resilience. Battery power can fail and 0 many flash-based devices have unreliable line supplies. Conse-0 quently, most popular R/W flash file systems rely on journaling, -1 2e+06 4e+06 6e+06 8e+06 1e+06 1.2e+07 1.4e+07 1.6e+07 0 logging the start and end of write transactions for fault recovery. Time (seconds) JFFS2, YAFFS2 and UBIFS all feature journaling. In addition, 1 0 high-density flash cells encounter write errors as well as read Total Bad Erase Blocks 0 perturbation. MLC and SLC both require Error Correcting Code -0 (ECC) for data integrity. -1 2e+06 4e+06 6e+06 8e+06 1e+06 1.2e+07 1.4e+07 1.6e+07 0 Execute-in-place (XIO) can be an issue. During boot, Linux Time (seconds) and RTOS loaders copy images from block-based NAND flash to Figure 2 NANDsim display of application modeling data. RAM. Running from RAM speeds execution and supports com-



Number of Erases

Short

Erase Block Moves

R/W

Bad Erase Blocks

FAT

January 2009

INDUSTRY Watch

JFFS2 over MTD

RO Boot Partition

MTD 0

YAFFS2 over MTD

RO RootFS

R/W Data & Applications

SquashFS

JFFS2

MTD 1

MTD 2

RO Boot Partition

MTD 0

RO RootFS

R/W Data & Applications

SquashFS

YAFFS2

MTD 1

MTD 2

FFS over UBI RO Boot Volume

VOL 0

R/W RootFS

R/W Data & Applications

UBIFS or JFFS2

VOL 1

VOL 2

UBI MLC NAND Flash Device

MLC NAND Flash Device

Analysis • Bad block management constrained by partition • Wear-leveling with random block replacement • No MLC read disturbance mitigation • Slow mount time from JFFS2 journal design • Metadata conflict with ECC • Compressed file systems Figure 3

• Bad block management constrained by partition • Non-deterministic wear-leveling • Reserves spares for replacing bad blocks & random block moves • No MLC read error mitigation • Faster mounting than JFFS2 • Out of sync metadata leads to inconsistency • No compression in YAFFS2

Modeling and performance results.

on block-level counters to track erase/write cycles. UBI blocks feature wear counting metadata, allowing precise and deterministic accounting. By contrast, JFFS2 and YAFFS2 have no wear counters and randomly choose among available blocks for wearleveling. JFFS2 and YAFFS2 nonetheless perform adequate wear-leveling, but without guarantees. In addition to wear, fragmentation occurs in file systems as volumes fill and contiguous sectors or blocks become scarce. On rotating media, fragmentation leads to longer access times from multiple seeks to gather data scattered across disk platters, and from diffuse allocation metadata. With no heads to move, solid-state media lessens the impact of fragmentation. Fragmentation does affect write times, however, since time-consuming erasure precedes each block write. Consolidating erasures into a contiguous block definitively speeds up erase/write operations. Journaling file systems like JFFS2, YAFFS2 and UBIFS perform de-fragmentation on the fly. Background garbage collection rewrites portions of files into new blocks and consolidates available blocks. Since MLC and SLC flash both suffer bit failure during read and write, designs need error correction code (ECC) to operate reliably. The number of ECC bits, the type of algorithm and flash page size determine the scope of error detection/correction. For hardware and software implementations, if ECC fails to correct errors while reading, writing or erasing, an FSS can mark and exclude blocks from future operations. Linux MTD manages error generation/detection/correction, freeing file systems from

• Separates wear-leveling from FFS • Global bad block management and wear-leveling • Deterministic algorithm • Volumes replace partitions • UBIFS and JFFS2 run over UBI • Block layer over UBI handles any file system • UBI block management triggered on read/write

January 2009

this chore. Also, UBI uses ECC results from MTD to perform read disturbance mitigation in the background by scheduling bad blocks for scrubbing.

Application Modeling

Poor technology choices directly impact product success and company bottom lines. Failure of flash devices from a bad combination of file systems and partitioning leads to returns and high support costs. Flash design and deployment have been “seat of the pants” operations; designers over-provision or make educated guesses to satisfy application requirements. As with many aspects of open source, Linux FFS support presents OEMs with tough choices and scant guidance for making informed selections. Flash cost and density options, and the need for ECC and wear-leveling, can leave systems designers scratching their heads. Application modeling lets OEMs account for updates, data logging, user preferences and data files, and performance requirements. It builds on knowledge about average hours of daily usage and read/write/erase sequences in typical use cases Instrumented systems could log real-world usage data to test design assumptions, but the mismatch between development and deployment time-tables prohibits gathering comprehensive field data: development cycles measure in months, deployment in years. Gathering usage data during field tests cannot provide reliable data to track lifetime viability. While actual service data does validate design of next-generation products, it misses the impact of ongoing advances in cost/density and reliability.

INDUSTRY Watch

Service Lifetime Emulation

Application modeling can provide input to design by simulating real-world usage, at an accelerated rate. Open Source Linux, its file systems and other platform components enable precise modeling by iterative simulation of long-term deployments. With open source platforms, OEMs can track flash wear, blockby-block, on real or simulated hardware; inject errors into FFS code at test points and black-box entries and compress multi-year operation into days or hours using cheap Linux clusters Embedded Alley uses community-based and in-house tools to simulate and emulate flash device and FFS behavior. One of the more useful tools, NANDsim, resides in the 2.6 kernel tree at /drivers/mtd/nand, providing test and simulation for flash and FFS code without actual hardware. Careful system configuration and application scenarios over NANDsim can emulate bit read and write errors and report wear statistics representative of long-term deployment (Figure 2). Embedded Alley uses a combination of real hardware, NANDsim and other tools to validate flash system viability by modeling global read/write behavior to exercise FFS and flash memory. Read/write rates are scheduled from use cases. The tool allows the injection of bit and block errors and tracking comparable parameters across FFS types. It can collect statistics on bad block handling, block wear and CPU utilization. Operation can be accelerated past 100x wall clock time and it can also run power fail scenarios. Using simulation also limits destruction of prototype flash hardware and evaluates candidate file systems prior to hardware availability. Figure 2 illustrates the result of a simple application modeling run on a MLC NAND device running the UBIFS FFS. The top graph shows the maximum and mean number of erasures performed on erase blocks in the NAND device. The middle graph shows total number of erase block moves that are performed as a result of ECC errors. Finally, the bottom graph shows the total number of bad erase blocks that have been detected on the NAND device. This particular example run was performed on NANDsim that was not configured with bad blocks or bit flips so the results of the latter two graphs are expected to be flat. Using NANDsim and other tools, developers can try flash and FFS configurations for technology choices (NOR, MLC/ SLC, managed NAND, etc.) and for evaluating FFS types and partition sizes. Developers can evaluate layered dependencies on subsystems such as MTD, UBI, etc. (Figure 3), compare device and RAM capacity, page size, sub-page capability, CPU utilization and determine read/write rates of actual hardware. Application modeling is not a turn-key operation. Embedded Alley performs application modeling as part of its consulting practice, helping OEMs and other clients design systems to meet product requirements. The key is choosing a reasonable starting point and scenarios that reflect project realities, not every conceivable possibility and permutation. The configurations in Figure 3 reflect scenarios for building a navigation device or similar equipment. Analysis of simulation data and comparisons of file system attributes raised a series of issues for JFFS2 and YAFFS2, leading to the choice of UBI with either UBIFS or legacy JFFS2 on top.

CO C COT OTS OT OTS TS RAID/SAN/NAS/VME

Phoenix International designs and builds rugged COTS Data Storage Systems that plug and play in any application -- from Multi-Terabyte Fibre Channel RAID and Storage Area Network configurations to plug-in Solid State Disk Drive VME/cPCI Storage Modules.

4FF VT BU XXX QIFOYJOU DPN PS DPOUBDU VT BU t JOGP!QIFOYJOU DPN An AS 9100 / ISO 9001: 2000 Certiﬁed Service Disabled Veteran Owned Small Business

This article provides developers with a general understanding of flash technology and file systems for embedded Linux. Discussion of device types and FFS attributes is presented not as an end, but as a means to highlight tough choices and essential challenges. Successful deployment of flash depends on systems engineers not taking storage technologies for granted. Application modeling lets OEMs prototype and test design and use cases over long service lifetimes by accounting for innate, non-obvious characteristics of flash and file systems. Actual product development and deployment realize benefits from application modeling via BOM cost reduction, by improving profit margins from mitigated deployment risk, and by meeting or exceeding product lifetime requirements.

Untitled-2 1

1/21/09 8:35:49 AM

Embedded Alley [www.embeddedalley.com]. LinuxPundit [www.linuxpundit.com].

January 2009

Featured Product 2.5 Gbit/s Device Brings Together Any Port, Service, Channelization and Functionality A single-chip carrier transport device enables equipment manufacturers to develop converged line cards and systems capable of handling any port, service, channelization and functionality while supporting both TDM and packet-based networks. Pioneering a new category of packet transport devices, these devices will enable the first 10 Gbit/s, fully channelized “any service, any protocol” line cards—resulting in 16 times more bandwidth and density than the high end of channelized line cards that exist today. Griffin devices pioneer a new category of universal physical-layer (PHY) capabilities, enabling multi-service line cards to support an extensive array of rates, protocols and interfaces. This is made possible using Siverge’s patented VLSI core technology and circuitry—the Siverge Matrix centralized-resource shared bit-stream processing engine—in the Griffin family of devices. This technology enables Griffin devices to achieve unprecedented performance and functional integration density by allowing a new type of resource-sharing that facilitates the massive multi-channel, multi-protocol processing of thousands of independent data streams. Griffin devices address all layer 2 protocols, any channelization levels (down to DS0), services, applications and port types in a single family of package-, pin- and software-compatible SoCs. Manufacturers can create comprehensive, converged multi-service line cards and systems or offer many variations of footprint- and software-compatible solutions for specific protocols, services, line rates, interfaces and channelization levels. The Griffin devices dramatically reduce board space requirements, power and complexity, while cutting overall system cost with a single design. The Griffin family consists of four devices, each of which is available in three speed grades. The family will allow carriers and network providers to significantly reduce capital and operating expenditures while optimizing investments in next-generation packet-based networks and ensuring service continuation for all legacy data and TDM services. For system vendors, the Griffin family will enable new applications while greatly reducing development time and investment, bill-of-material costs and time-to-market. Carriers, network providers and system vendors alike will benefit from major reductions in product inventory and maintenance.

January 2009

The flagship member is the SV3640/1/2 universal convergence device, which combines as many as 30 devices into a single highly integrated SoC. The SV3640/1/2 integrates SONET/SDH framing/mapping/multiplexing, T1/E1/J1/T3/E3 framing/ mapping, M13 multiplexing, ATM/IMA controller functionality, HDLC/POS/PPP/ LAPS/EOS/FR/MLPPP/MLFR controller functionality, and both Ethernet-over-PDH and Ethernet-over-SONET/SDH functionality (including GFP over VCAT with or without LCAS). Rounding out the Griffin family are three additional devices that provide more specific functionality and cost-optimized solutions for TDM-only applications (SV3600/1/2), legacy-protocol applications (SV3610/1/2) and “Ethernet-over-Anything” applications (SV3620/1/2). Griffin devices can be used in applications ranging from packet-based transport systems and multi-service switches to ADM/MSPP systems, switches and routers. These devices also simultaneously support 2G, 2.5G, 3G and beyond wireless transport backhauling systems as well as BSCs and RNCs. The Griffin family comes with a comprehensive software package including API and companion GUI application that reduces system integration requirements, and an evaluation system for device testing, evaluating and demonstration. Siverge also provides reference designs and an entire line card and system solution in collaboration with network processor and other silicon partners. Related application notes are available under non-disclosure. Pricing ranges from $200 to $1,200, depending on the device and its speed and feature set, in OEM quantities of 10,000. Siverge Networks, Herzliya Pituah, Israel. +972 9 9526600. [www.siverge.com].

Celebrating

Years

Concept Development, Inc.

A Perfect Fit Your Vision — Our Capabilities

Homeland Security

Computer Systems Design

Medical Equipment

Concept Development, Inc. was established to assist companies in meeting their needs for timely and cost effective product development. We accomplish this by collaborating with the customer’s existing engineering group to effectively create an extension of the customer’s engineering resources. CDI has established itself as a flexible and responsive partner for many companies in a wide range of product technologies for an expanding set of markets.

In-Flight Entertainment

Aerospace Applications

Glass Cockpit Scanning & Analysis

A Complete Development and Support Process Define the Requirements

Engineering and Design

Production Phase

s 5NDERSTAND #USTOMER S .EEDS

s $EFINE THE 2ANGE OF %XPERTISE 2EQUIRED

s 0ROTOTYPES FOR $ESIGN 6ALIDATION

s 0RODUCT 3PECIFICATIONS ARE $ETAILED

s !SSIGN THE 4EAM -EMBERS

s 0RE PRODUCTION 5NITS FOR -ANUFACTURABILITY 6ALIDATION

s $ELIVERY 2EQUIREMENTS ARE )DENTIFIED s 0ROGRAM AND 0RODUCT #OSTS ARE $EFINED s &IXED 0RICE 0ROPOSALS ARE 3UBMITTED

s (ARDWARE &IRMWARE -ECHANICAL !RCHITECTURAL s $ESIGN 2EVIEW 0OINTS TO )NSURE #OMPLIANCE s $ESIGNED FOR 4ESTABILITY

s &ULL 0RODUCTION #APABILITY

Production Support Phase s 0RODUCT ,IFE #YCLE 0LAN s 0RODUCTION 5NIT 7ARRANTIES s ,ONG TERM 0ROGRAM 3UPPORT !CTIVITIES s 0RODUCT $ESIGN 7ARRANTY

s #USTOMER 3UPPLY #HAIN )NTEGRATION

Please Call Us 949.623.8025

1881 Langley Ave. | Irvine, CA 92614 | 949-623.8000 | f 949.623.8055 | www.cdvinc.com

&TECHNOLOGY

Products

Radio-via-Ethernet Board for Streaming Audio without a PC

A ready-to-go answer for integrating an Internet radio solution that will play streaming audio without the need for a PC is based on a hard-coded W5100 Ethernet IC from Wiznet and an ARM7 processor and MP3 decoder. WIZ-iRadio’s mother/daughterboard construction provides all needed connectors and a helpful backlit 128 x 64 pixel graphic LCD for program and song-file information. It even comes with its own infrared remote control. WIZ-iRadio allows access to Web-based radio stations broadcasting an audio stream in MP3 format. The module can connect to servers with SHOUTcast, MMS and HTTP, DHCP or static IP. Running from 5V, WIZ-iRadio needs less than 180 mA. Configuring WIZ-iRadio is simple, thanks to a small Windows utility provided free, which allows you to store a list of your favorite network radio stations settings in a simple text file. Complete C source code software is supplied, so you can adapt the program to your own specific needs, especially if you have ARM environment capabilities. Full documentation is provided, with schematic diagrams and complete instructions for use. WIZ-iRadio is available from U.S. distributor Saelig at $149 each. Saelig, Pittsford, NY. (888) 772-3544. [www.saelig.com].

Hybrid Signal Processing AMC Module Based on TI Multicore DSP

Time-to-Digital Converter Delivers 5-Picosecond Resolution

Precise time-to-digital conversion is vital in a variety of applications that include physics applications and 3D mapping using Lidar/ Radar. Agilent’s new single-hit time-to-digital converter (U1050A-002) has a 5-picosecond timing resolution and a wide measurement range of up to 20 seconds, a ratio of 1:4.109. The U1050A-002 TDC has 13 identical hardware channels. Twelve channels are independent stop inputs; the 13th is the common start. The module can operate in either single-start or multi-start acquisition modes with the timing information on all the independent channels encoded relative to the common channel. Time measurement on the U1050A-002 TDC can be based on either the internal low jitter (under 3 ps rms), high stability (±2 ppm) clock source, or an external 10 MHz reference input. Digitized data is fed directly to the onboard FPGA-based data processing unit. This handles the data and subsequent fast readout with direct memory access (DMA) mode, for increased data throughput to the PC. Pricing for the U1050A-002 multi-start, single-stop time-to-digital converter starts at $16,000.

Featuring a high-performance, multicore Texas Instruments DSP designed specifically for telecommunications infrastructure applications and coupled with a high-density Altera FPGA, a new AMC module provides a flexible, reconfigurable platform for high-end digital signal processing. The F1/ GXAM from BittWare features TI’s C6474 high-performance multicore DSP, enabling SoC solutions in one modular, scalable device with the benefits of a small footprint. The device contains specialized coprocessors and supports several standard interfaces including Serial RapidIO, Gigabit Ethernet, DDR2 and McBSP. Robust software tools are available to aid development including TI’s Code Composer Agilent Technologies, (408) 345-8886. Studio and numerous third-party programs. Santa Clara CA. [www.agilent.com]. The onboard FPGA, an Altera Stratix II GX, was specifically designed for serial I/O-based applications requiring high-density, reconfigurable logic. The FPGA provides up to 15 full-duplex high-performance, multi-gigabit transceivers supporting Serial RapidIO, PCI Express, 10GigE, Gigibit Ethernet and SerialLite II standards. It contains up to 132,540 equivalent LEs, over 6700 Kbits of embedded memory, 252 embedded 18x18 multipliers (63 DSP blocks) and 8 PLLs. BittWare offers a comprehensive software and support package for the F1/GXAM. The BittWorks software tools provide host interface libraries and a wide variety of diagnostic utilities and configuration tools. BittWare’s FINe Bridge provides Gigabit Ethernet on port 0 of the common options region. It also provides 10/100 Ethernet and RS-232 on the AMC front panel. The F1/GXAM also implements the standard Module Management Control Interface (IPMI).For FPGA development, Altera’s Quartus II FPGA design flow tool and SOPC builder system-level design tool are available. BittWare also offers an FPGA Developers Kit, which includes fully validated board-level modules for I/O, communications and memory for BittWare’s Atlantis FrameWork. Implemented on the Stratix II GX, Atlantis facilitates the majority of off-board I/O and provides communications routing and processing. It provides an internal dataflow interconnect fabric that allows these modules to be easily connected, and a control fabric that allows them to be easily coordinated and controlled. Atlantis provides a stable, high-performance signal processing platform that frees the designer to focus on application development rather than reinventing board-level infrastructure. The F1/GXAM will be priced under $4,000 in OEM quantities. BittWare, Concord, NH. (603) 226-0404. [www.bittware.com].

January 2009

Video Distribution System for Mil/Aero Defense IP Nets

The demand for real-time distribution and storage of high-quality video continues to grow. System developers need to apply video distribution to a wide range of aerospace and defense applications such as ship-wide naval distribution, local situational awareness, airborne distribution and simulation systems. Curtiss-Wright Controls Embedded Computing has announced VDS, a new digital video distribution system suite comprised of middleware and a comprehensive selection of video hardware. The VDS digital distribution systems support the capture, compression and distribution of video from a wide variety of sources to any display or recording station available over a local or wide area network (LAN, WAN). JPEG2000, the primary compression algorithm employed in the VDS family, provides an optimal mix of quality, robustness and performance with compression that enables the transmission of multiple video streams over standard Gigabit Ethernet or other suitable network connections. VDS also supports the RTP (real-time protocol) standard used commonly for real-time video and audio distribution requirements. RTP controls the flow of video data to ensure optimized transmission over IP networks. VDS is controlled by a browser-based interface (primarily useful during development) and via an embeddable API. Pricing for VDS starts at $4,500. Curtiss-Wright Controls Embedded Computing, Leesburg, VA. (703) 779-7800. [www.cwcembedded.com].

Frame Grabber for Security Apps Provides 16 Analog Camera Inputs

A 16-Channel PCI Express Frame Grabber for security applications uses low-cost color cameras. The PC_EYE/JPG from American Eltec offers onboard JPEG compression hardware for 16 channels along with an onboard scaling device and a fast PCI Express Bus interface with a bandwidth of 250 Mbytes/s. It records analog color camera data via 16 parallel inputs, all of which contain a separate fast AD converter. With this technology, images can be recorded in real time and stored in the PC’s main memory. There are 16 analog-to-digital converters (ADCs) that up to 16 cameras can be connected to. Each camera has its own ADC (SAA 7113-type) as well as its own color separator. This makes it possible to switch cameras in the digital domain, minimizing multiplex times. Since the analog input is not switched, there is no PLL settling delay, which can take up to several frame times in other multiplexing schemes. The PC_EYE/JPG is intended for security applications where video images from up to 16 color cameras are fed into a PC’s main memory for storage or into the graphics board for display. Data is acquired in compressed or in raw format. An onboard scaling unit can transform full-resolution images to CIF size or smaller for efficient display in split screen applications. All standard color and monochrome video cameras are supported in hardware and software. The video standards PAL, NTSC and SECAM can be used. Single-unit pricing starts at under $1,800. American ELTEC, Las Vegas, NV, (702) 878-4085. [www.americaneltec.com].

High-Power, Fully Programmable and Intelligent Standard Power Supply

A new fully programmable and intelligent high-power AC-DC power supply is capable of delivering up to 4920 watts of high-density output (12W/cu-in). The iVS from Emerson Network Power is specifically designed for industrial, military and medical instrumentation applications that require high power, reliability and maximum flexibility. Fully modular in design, iVS power supplies utilize digitally configurable single and dual output power supply modules available in six power ratings. Spanning 210W to 1500W, they support 24 different output choices ranging from 2V to 60V to enable nearly limitless power configurations. All modules are fully interoperable with existing Emerson configurable platforms and can be quickly configured to match exact application requirements, enabling greater precision and energy efficiency. To maximize control flexibility, microcontrollers are integrated into the case and modules are paired with Emerson Network Power’s I2C GUI-based control software. The software enables designers to configure and reconfigure voltages, current limits and inhibit/ enable settings for individual power supplies while allowing them to conveniently export their final configuration specs for mass application. In addition to its module configuration capabilities, the control software offers OEMs and end-users real-time voltage, current and temperature monitoring to provide at-a-glance performance confirmation for the entire power supply, including the case and all its constituent modules. All iVS power supplies also feature field-upgradable firmware, which extends product lifecycles and offers developers a new dimension of flexibility and interoperability. Emerson Network Power, Tempe, AZ. (800) 759-1107. [www.emersonnetworkpower.com].

January 2009

Products & TECHNOLOGY microSD Card Unites Smart Card and NFC-Compatible Contactless Functions

A new mobile security card offers contactless smart card security in addition to the usual data storage functions. The new security feature is provided by a cryptography controller with an NFCcompatible interface integrated in the microSD card along with the flash memory. The Mobile Security Card CL from Giesecke & Devrient serves as a basis for implementing security and payment functions in cell phones and for many other mobile applications. This addresses the needs of digital identity solution providers. The mobile security cards designed by G&D are suitable for use in mobile phones, smart phones, netbooks and other mobile devices, and even in USB tokens. The only requirement is that such devices must be equipped with a slot for microSD cards. Such cards can provide security functions for all of these devices, as well as PCs and laptops, without the need for an external smart card reader. The latest product in this family of cards, the Mobile Security Card CL, has been enhanced to include a contactless interface in addition to the integrated smart card, based on the international standards ISO 14443 and MIFARE. In order to use the card in contactless mode, the device in which it is inserted must be equipped with an antenna for the contactless communication link. The card contains two additional standardized contacts for this purpose. A battery is not required to operate the contactless interface; the necessary energy is tapped from the radio field generated by the card reader. A secure communication link is established as soon as the antenna comes within a few centimeters of the reading device. All types of terminals automatically recognize the microSD format of the mobile security cards. In conjunction with the integrated user identity, this gives the cards a high added value. This can be usefully exploited to develop new, feature-rich products such as an electronic student pass with incorporated encrypted e-learning content and signature applications for electronic document submission, library passes capable of storing downloaded and copy-protected eBooks, or personal identity cards for bank accounts containing a secure banking application.

6U CompactPCI CPU Board with 1.86 GHz Intel Celeron M 440

A new 6U value line CompactPCI CPU board specially designed for costsensitive industrial applications is equipped with the 1.86 GHz Intel Celeron M processor 440, the Intel 945GM chipset, up to 4 Gbyte 533 MHz DDR2 SO-DIMM and the ICH7-R IO controller hub. The CP6001-V CompactPCI board from Kontron is optimized to provide the best cost/performance ratio. Due to its industrial components and 4 Gigabit-Ethernet interfaces, the CPU board is designed for demanding, powerful industrial automation and IP switching applications. The 4HP single-slot CompactPCI board comes with industryrequired interfaces such as a serial port, VGA graphics interface, 4 Ethernet ports, 2 front USB ports, PMC, HDD, CompactFlash and USB flash disk support. With an integrated graphics accelerator on the Intel 945GM chipset, the Kontron CP6001-V provides high-resolution graphics up to 2048 x 1536 pixel @ 75 Hz, and 2D/3D multimedia quality video. The 945GM chipset also enables balanced memory usage between graphics and system for optimized performance (up to 128 Mbytes of dynamic video memory allocation). With the Intel ICH7-R I/O controller hub the board provides USB 2.0 and Serial ATA300. The PMC site, USB flash, CompactFlash socket and an optional onboard 2.5-inch hard disk can all be used at the same time in a single 4HP slot. Utilizing the modular extensions, all industrial requirements for write-protected operating systems and data storage capabilities are supported. Two Gigabit Ethernet interfaces available on the front panel operate independently of the additional two PICMG 2.16-compliant Ethernet channels, which are routed to the backplane. The CP6001-V is provided with a self-installing board support package for Microsoft Windows XP, Windows XP Embedded and Linux. The onboard temperature and voltage sensors can be monitored; the front panel LEDs, hardware watchdog and timer can be triggered to support application-specific features. Kontron, Poway, CA. (888) 294-4558. [www.kontron.com].

Giesecke & Devrient, Munich, Germany. +49 89 4119-2985. [www.gi-de.com].

HD SDI Matrix Routing Switcher with Remote Capability and Simple Front Panel Operation

Capable of switching SMPTE 292M & SMPTE 259M from any of the four inputs to any of its four outputs, the 2444 from Sensoray allows for remote capability and simple front panel operation. It features a built-in, customizable Web page that allows for remote operation while remaining operating system independent. An automatic reclocking and equalization of the incoming signal is applied to HD SDI signals from as far away as 150m and SD SDI signals from as far as 250m away. The unit is enclosed in a simple desktop case; custom packaging dependent upon the end user's needs can be arranged for OEM quantity orders. The 2444 is available with pricing starting at $226 with OEM pricing and customization available. Sensoray, Tigard, OR. (503) 684-8005. [www.sensoray.com].

January 2009

High Port Density 6U VME Gigabit Switches Offer Managed or Unmanaged Options

The rapidly growing demand for high-performance, high-reliability switching and high port density is at the heart of today’s sophisticated, demanding network-centric applications. In response, GE Fanuc Intelligent Platforms has announced two new 6U VMEbus Gigabit Ethernet switches, the NETernity RM980RC and RM982RC. The goal is to provide engineers with more choice and flexibility as systems continue to rely on more Gigabit Ethernet communications in their architecture. The RM982RC is a configuration-managed switch, featuring GE Fanuc’s embedded OpenWare Lite management software that allows the switch to be monitored and/or controlled either locally or remotely. Management features include support for VLANs, MSTP, trunking, mirroring and limited SNMP. It can be configured with either 24 copper rear Gigabit Ethernet I/O ports or 22 copper rear I/O ports plus two fiber or copper ports routed to the front panel. The unmanaged RM980RC provides a lower-cost option for customers not requiring a managed switch, and is available in three configurations. The entry level unit supports twelve Gigabit Ethernet ports routed to rear I/O. Alternatively, a 24-port version and 22port version are available, the latter with two further copper ports with RJ45 connectors routed through the front panel. Both the RM980RC and RM982RC are RoHS-compliant, and are capable of Layer-2 switching and routing at wire speed. 12-, 22- and 24-port rear transition modules are available. Each switch board occupies only a single chassis slot (the 22- and 24-port rear transition modules require two slots) providing maximum port density per slot, and each is able to facilitate communications both within a chassis as well as to support the external network. The OpenWare Lite switch management environment is available on selected NETernity configuration-managed Layer 2 Ethernet switches, providing a common control and monitoring environment that can enhance customer productivity. Configuration and monitoring functions are accessible from a serial console or via a network. Supported access methods include Telnet, SSH and SNMP. OpenWare Lite is easy to deploy and is Linux-based, allowing faster implementation and easy updates to firmware. A familiar Linux command line interface and remote Telnet user interface support allows users to select how they interact with the switch. It is portable across switch fabrics and processor environments. GE Fanuc Intelligent Platforms, Charlotteville, VA. (800) 368-2738. [www.gefanuc.com].

Conduction-Cooled PrPMC/PrXMC Sports PowerQUICC III

Compute-density is the watchword for a lot of today’s high-end applications. The PowerQUICC processor has had a lot of success as a comms processor, but many want it for its pure performance and system-level features. With that in mind, Extreme Engineering Solutions (X-ES) has introduced the XPedite5201, a PrPMC/XMC based on Freescale Semiconductor’s MPC8548E PowerQUICC III processor. This conduction-cooled card comes standard with PCI-X support with optional PCI Express support via an XMC connector. XPedite5201 is an optimal solution for military system designers that need to include additional processing in an existing system, or want to take advantage of the advantageous performance/watt characteristics of the 8548E. XPedite5201, with the Freescale MPC8548E embedded PowerPC e500 core running at 1.333 GHz, includes up to 4 Gbytes of DDR2-533 SDRAM, up to 4 Gbytes of NAND flash and 256 Mbytes of redundant NOR flash. Rear I/O includes two Gigabit Ethernet ports, GPIO and two RS-232 serial ports. The new XPedite5201 is available today. Pricing starts at $6,000. Extreme Engineering Solutions Middleton, WI. (608) 833-1155. [www.xes-inc.com].

GPIB-to-Serial Board Supports RS-422, RS-485

Serial interfaces like RS-422 and RS-485 can be found all over embedded systems, often side by side with more modern I/O technologies. ICS Electronics makes a GPIB-to-Serial Interface Board for interfacing devices with an RS422 or RS-485 interface to the GPIB bus. Designated the Model 4808, this new board provides an IEEE-488.2-compliant, GPIB-to-serial, data path to the device and includes an RS-232 to RS-485 converter that provides a transparent RS-232 communication path to the serial device. The 4808 is a new type of interface board that combines a GPIB interface and an RS-232 interface to control a serial device with RS-422 or RS-485 differential signals. The Model 4808 is an intelligent IEEE 488.2/GPIB to Serial Interface that adapts any device with RS-422 or RS-485 signals to the GPIB or HP-IB bus. The 4808 provides the serial device with an IEEE-488.2compatible interface that responds to all of the required IEEE-488.2 Common Commands and includes the 488.2 Status Reporting Structure. Prices start at $340. ICS Electronics Pleasanton, CA. (925) 416-1000. [www.icselect.com].

January 2009

Products & TECHNOLOGY I/O Expansion Devices for PCI Express Gen2 Support Multicast

Two new PCI Express (PCIe) Gen2 switching solutions are optimized for computing and embedded applications. The new 32-lane, 8-port and 48-lane, 12-port solutions from Integrated Device Technology (IDT) are based on the company’s switching architecture and technology, and support multicast functionality for increased system resource utilization. The two new devices incorporate advanced features to minimize power consumption, maximize performance-per-watt, and reduce total cost of ownership and thermal design complexity. Both new devices support PCIe multicast functionality, which allows any switch port to simultaneously send identical data to two or more switch ports, increasing system resource utilization by decreasing the previously required hardware and software overhead needed to send copies of data in a looped manner. This extension of the PCIe standard ensures consistency in data and table information among multiple system elements and extends the reach of PCIe into emerging enterprise computing and embedded applications requiring robust data coherency and sharing. The Gen2 PCIe switches are fully compliant with the PCI-SIG PCIe base specification 2.0 and provide customers the option of doubling the throughput bandwidth of existing PCIe lanes to 5 gigabits per second of data transfer for cutting-edge designs, or allowing a 50 percent reduction in the number of PCIe lanes and board traces necessary to support link throughput requirements for a more cost-effective design. Each of the IDT PCIe switching solutions has a dedicated evaluation and development kit for device testing and analysis, and system emulation. Each kit consists of a hardware evaluation board with representative upstream and downstream connectivity, and an IDT-developed, GUI-based software environment that enables the designer to tune system and device configurations to meet system requirements. Moreover, to ensure that each OEM system design is optimized for production and meets its time-to-market objectives, IDT provides customers with extensive, collaborative technical support including system modeling and signal integrity analyses, and schematic and layout review services. The 32-lane, 8-port 89PES32T8G2 device is packaged in a 484-ball ball gate array (BGA), and the 48-lane, 12-port 89HPES48T12G2 is packaged in a 676-ball BGA. The new devices are sampling to qualified customers and range in price from $42.50 to $62.00 for 10,000 units. Integrated Device Technology, San Jose, CA.(408) 284-8200. [www.idt.com].

VXS GPU Platform Targets Sensor Stream Computing

Graphics processing units, or GPUs, now rank as the computing industry’s most powerful, programmable floating-point graphicsrendering engines. With recent architectural advancements, the algorithmic scope to which GPUs can be applied has grown dramatically. For traditional signal processing algorithms like the FFT (Fast Fourier Transform), they provide unprecedented performance, particularly performance per watt. With the Mercury Sensor Stream Computing Platform, embedded stream computing customers can benchmark and evaluate application performance in their choice of GPU environments, and then migrate to a larger deployed solution. The VXS-based Mercury Sensor Stream Computing Platform offers scalability in compute power, performance and thermal management, and allows for much greater, tunable performance for a variety of commercial and defense applications. The Platform leverages a dual dual-core Intel Xeon-based VX6-200 SBC, which offers unprecedented levels of compute performance and a wide selection of I/O interfaces. At the heart of the Platform is the VXS-GSC5200 dual MXM GPU module, which delivers very high bandwidth performance to each GPU from the host, as well as between GPUs. Each MXM GPU module can drive up to 3 display monitors (1 analog and 2 digital). Customer shipments of the Mercury Sensor Stream Computing Platform are planned for early 2009. Mercury Computer Systems, Chelmsford, MA. (978) 256-1300. [www.mc.com].

January 2009

“Starter Lab” Turnkey Line for Training & Entry Level SMT Assembly

Contractors and OEMs who require lowcost precision prototyping or low-volume SMT assembly can now make use of the “Starter Lab” turnkey line from Manncorp, which is geared for production as well as for training applications. It reportedly has also gained acceptance by numerous technical schools and colleges as a teaching aid. The SR2100 stencil printer offers precision screw alignment to boards, X, Y and Theta adjusting knobs and double-sided board processing. Frame capacity is 18” x 22” (47 mm x 570 mm). The singlechamber 850 hot air convection reflow oven adjusts to lead-free temperatures to 260ºC. Included are LCD preheat and reflow-phase time and temperature settings.

PCB assemblers on tight budgets no longer have to scavenge for used, almost outmoded equipment, “when for the same or less money, they can now have a single-sourced SMT line with fully integrated lead-free and fine-pitch equipment.” The placer is the latest upgrade of the company’s ECM-5611S with on-the-fly vision placement and available bottom vision for larger components, such as BGAs and CSPs. It will place a wide range of fine-pitch devices on a 10.6” x 5.9” placement area and holds 24 8 mm tape feeders. A computer with LCD monitor is included. Internet-priced at $27,995, the line includes three space-saving bench-top systems: a precision stencil printer, a computer-controlled pick and place unit and a lead-free reflow oven. Manncorp, Willow Grove, PA. (215) 8301200. [www.manncorp.com].

PCIe Dual Copper Gigabit Ethernet Adapter Includes Bypass Function

A PCI Express dual copper Gigabit Ethernet adapter features a bypass function for increased network security. Based on the Intel 82574L Gigabit Ethernet controller, the new ABN-262 from American Portwell features low power connectivity with advanced performance in a footprint as small as 147.3 (W) x 68.9 (L) mm / 5.8” (W) x 2.71” (L). The main focus of the ABN-262 is to reduce network downtime caused by security appliance hardware failure, which then automatically increases the continuity of the business network. It is targeted as a win-win situation for customers who have been plagued by worms, viruses, Trojan Horses and the like. The ABN-262 is a standard low-profile PCI-E x4 adapter that is easy to integrate into a 1U/2U or 4U server

Small Surface Mount Filtered GPS Low Noise Amplifier

A line of surface mount filtered GPS low noise amplifiers offers multiple gain stages of 26 to 38 dB with more than 35 dB of out-of-band rejection at 1575.42 MHz. This new small surface mount design from Spectrum Microwave exhibits a low noise figure of 1.8 dB, and operates from a 5 to 32 volt supply while only consuming 77 mA. These filtered amplifiers offer a 4 pin design for easy mounting and customer selected 26 dB, 32 dB or 38 dB gain options while still maintaining a low 1.8 dB noise figure from 0° to 70°C. Spectrum also offers these filtered LNAs in connectorized hermetic packages for inline booster applications or other harsh environments. Spectrum Microwave’s high isolation, Filtered LNAs are suitable for a wide range of applications including Avionics, Satellite Navigation, Marine Navigation, or Surveying and Mapping systems. The series 310 Filtered GPS LNAs are priced starting from $300 depending on quantity and configuration. Spectrum Microwave, Philadelphia, PA (888) 553-1531. [www.SpectrumMicrowave.com].

The ABN-262 unit includes a built-in watchdog timer with an easily configurable Normal/Bypass mode and incorporates a software programmable time-out setting and a control to automatically switch to bypass/non-bypass functions for Ethernet ports in the event of a host system “hang” and/or power failure. Built with onboard LED and LED pin-out for LAN status and bypass mode and variable LED location for system integration, the new ABN-262 supports most network operating systems and its low profile form factor will fit in a wide variety of systems platforms. The ABN262 is a solution for customers seeking applications for increased throughput and High Availability (HA) on server-based systems, Intrusion Prevention Servers (IPS), Intrusion Detection Servers (IDS), WAN optimization, security appliances and other missioncritical gateways. American Portwell Technology, Fremont, CA. (877) 278-8899. [www.portwell.com].

USB to Serial Adapters are Software Configurable

Sealevel Systems has announced new additions to its SeaLINK USB to serial product line: the SeaLINK/ PC.SC (one-port) and SeaLINK+2/PC.SC (two-port) embedded USB to serial adapter with a PC bracket, perfect for adding serial ports to any PC using the computer’s internal USB connection. The serial ports are software configurable for RS-232, RS-422, or RS-485, which eliminates the need for an external converter and provides a clean, professional installation. All configuration and electrical interface selections are handled through the driver software, so users never need to shut down the computer or open the system enclosure to set jumpers or dipswitches. Standard operating temperature range for SeaLINK products is 0 to +70°C, and extended temperature range (-40° to +85°C) models are available. Both models include a standard-sized PC bracket and ship with an internal USB cable that is compatible with ATX, Mini-ITX and similar large motherboards with 0.1” USB header connectors. For smaller motherboards, an internal USB cable with a 2 mm header connector (Item# CA383) is available as an option. Both items are available immediately from stock priced at $179 for the SeaLINK/PC.SC and $229 for the SeaLINK+2/PC.SC. Sealevel Systems, Liberty, SC. (864) 843-4343. [www.sealevel.com].

January 2009

Products & TECHNOLOGY High-Density 16-bit D/A PMC Module for Industrial Apps

A standard single-wide 32-bit PMC module provides 32 or 16 channels of 16-bit analog outputs. The TPMC553 from Tews Technologies is suitable for applications in industrial, transportation and aerospace/defense COTS. All signals are accessible through an HD68 SCSI 3 type front I/O connector. The software-selectable output voltage ranges are 0-5V, 0-10V, 0-10.8V, ±5V, ±10V or ±10.8V. The output voltage range can be individually set per channel. The conversion time is typically 10 μs and the DAC outputs are capable to drive a load of 2 kΩ, with a capacitance up to 4000 pF. In addition to individual channel updates, the double buffered DACs allow simultaneous update of all channels for simulation applications. Additionally, a sequencer on the TPMC553 allows the periodic updating of enabled channels with a sequence timer range from 10 μs to 167s. Each TPMC553 is factory calibrated. The calibration information is stored in an onboard serial EEPROM unique to each PMC module. The TPMC553 operates in extended temperature range (-40° to +85°C) standard. Driver software for the module is available for operating systems including VxWorks, Windows, Linux, LynxOS and QNX6.

6U CompactPCI PMC Carrier Card Provides Simple I/O Expansion

Two versions of a new 6U carrier card provide an easy and low-cost solution to expand the use of PMC mezzanine modules in 6U CompactPCI computer systems. The AcPC4620 carrier card from Acromag acts simply as an adapter to route PCI bus signals to and from the PMC module through the CompactPCI card slot edge connector. A PLX Technology PCI6540 bridge device provides a transparent 32-bit 33/66 MHz PCI/PCI bridge for data transactions from the PCI bus (system host) to the PMC site. The card is available in two models for air-cooled (-40° to 85°C range) or conduction-cooled applications. All Acromag PMC modules and those from other vendors are compatible. 3.3V and 5V DC signaling are supported.

TEWS Technologies, Halstenbek, Germany. +40 (0) 4101 4058-0. [www.tews.com].

12-/10-/8-Bit Octal DACs Integrate 10ppm/°C Reference in Tiny Packages

A family of 12-bit, 10-bit and 8-bit digital-toanalog converters (DACs) integrates a precision reference in tiny 4 mm x 3 mm DFN and MSOP packages, making them the smallest octal DACs available on the market today. The small size and internal reference of the LTC2636 from Linear Technology are vital for a variety of industrial and communications applications. Optical networking is one such application that requires multiple DACs in a single compact package. These DACs are suited for driving optical attenuators or setting the current levels for laser diodes. By integrating a 10ppm/°C reference, the LTC2636 offers further space reduction for constrained optical networking circuit boards. In addition to selecting one of three resolution options, designers can also choose between a 2.5V or 4.096V full-scale range, making the LTC2636 a good fit for 3V or 5V systems. The internal reference is bonded out, providing a convenient way to drive the reference inputs for other data converters in the system. Alternatively, an external reference can be used if greater accuracy or a nonstandard voltage range is required. Ordering option provides the choice between powering up the DACs at zero-scale or mid-scale, offering flexibility for designs that cannot be forced to ground when power is first applied. The LTC2636 octal DACs join the previously released LTC2630 and LTC2640 single DACs. These are part of a complete family of octal, quad, dual and single 12-/10-/8-bit DACs with internal references in tiny packages. Pricing begins at $2.85 each, in 1,000-piece quantities. Linear Technology, Milpitas, CA. (408) 432-1900. [www.linear.com].

January 2009

The board accommodates front and rearpanel access to field I/O signals. The air-cooled carrier card has front panel cut-outs providing access to a PMC module’s front I/O connector. Alternatively, all I/O signals can be routed through the carrier card’s rear J3 and J5 connectors. An air-cooled rear transition board, Acromag Model TRANS-C4620, is available to map the field I/O on the PMC module to the rear of the CompactPCI card cage. The conduction-cooled model employs a heat frame with wedge-locks and thermo bars for use in applications where ambient or forced air can’t provide adequate cooling. This unit is designed to meet ANSI/VITA-47 Environmental Class ECC4 standards and is ideal for airborne systems, deployment in battleground equipment, and other situations with advanced thermal management requirements. Field I/O signals are routed through the carrier card’s rear J3 and J5 connectors. The carrier cards start at $750. Acromag, Wixom, MI. (248) 295-0310. [www.acromag.com].

Rugged Digital Data Recorder and Signal Conditioner Allow Wide Range of Deployments

A digital data recorder and signal conditioner are specifically designed for use in harsh environments that are subject to extremes of heat, vibration and shock. The DDR-300 and DSC-300 from GE Fanuc Intelligent Platforms are rugged, compact and lightweight, allowing them to be field deployed in space-constrained and weightconstrained environments such as aircraft and in dangerous testing environments that require the units to be nearby and in conditions that prohibit the use of conventional rackmount units.

For applications in which data from numerous sensors must be processed, a primary requirement is to ‘condition’ the incoming signal—amplifying it, for example—before it can be processed. In many applications, the requirement also exists to capture multiple parallel streams of incoming data at very high speeds for subsequent analysis. Such applications include test environments with NVH (noise, vibration, harshness) and munitions testing, and the DDR-300 and DSC-300 are ideal solutions in these environments. The DDR-300 is a complete, self-contained, ready-to-run field-deployable unit. The rugged 32-channel or 64-channel digital data recorder includes an integral 15” XGA TFT display, USB keyboard and mouse, integrated stereo speakers and support for up to 2.5 Terabytes of high-speed (up to 100 Mbytes/second) 7200 rpm SATA disk storage, which may optionally be implemented in a RAID configuration. It features a 2 GHz Intel Core 2 Duo processor and 2 Gigabytes of system memory. Housed in a magnesium chassis enclosed within a fiber-reinforced polycarbonate housing, the DDR300 offers expansion slots for two full-size and two half-length PCI cards and a combination AC/DC power supply. Optionally available are a UPS battery, removable hard disk storage and rack-mounting. It weighs less than 25 lbs. The rugged DSC-300 8-channel signal conditioning module weighs less than 5 lbs, and is capable of handling either analog or digital signals and either differential or single-ended inputs. Gain is selectable up to x1,000 per channel, with a bandwidth of 200 kHz at gains of up to x100. It features a built-in buffered output monitor together with a reference input for calibration, and supports voltage, bridge and ICP (Integrated Circuit Piezoelectric) inputs. GE Fanuc Intelligent Platforms, Charlotteville, VA. (800) 368-2738. [www.gefanuc.com].

Low-Power 3U cPCI SBC Targets Freescale MPC8572E

A new 3U CompactPCI single board computer is based on Freescale Semiconductor’s MPC8572E PowerQUICC III processor. In addition to the MPC 8572E dual core processor, the XPedite5330 from Extreme Engineering Solutions provides a PMC/XMC site along with Gigabit Ethernet and the traditional PCI bus. To provide a system designer with multiple hardware options, the XPedite5330 is available in a full range of convection- or conduction-cooled configurations. The XPedite5330 operates with two 1 .5 GHz e500 cores based on Power Architecture technology, and supports two separate DDR2-800 ECC SDRAM channels with up to 2 Gbytes per channel as well as up to 4 Gbytes of NAND flash and 256 Mbytes of redundant NOR flash. It also provides two Gigabit Ethernet ports. The J2 connector I/O includes GPIO, I2C, PMC I/O and two RS-232/RS-422/RS-485 serial ports. Operating system support includes board support packages (BSPs) for Wind River VxWorks, QNX Neutrino and Green Hills Integrity as well as Linux 2.6. Extreme Engineering Solutions, Middleton, WI. (608) 833-1155. [www.xes-inc.com].

VITA 46 3U Two-Slot Chassis Offers Quick Start Development Platform

A VITA 46 3U dual-slot conduction-cooled chassis is intended to provide 3U VPX system engineers with an inexpensive desktop platform to jump-start their software or hardware development. The XPand1000 from Extreme Engineering Solutions hosts up to two 3U VPX conductioncooled cards providing fabric interconnect between the two slots, as well as easy access to Gigabit Ethernet and serial port I/O. Powered by an external 12V supply with integrated heatsink fins and built-in fans, the chassis is suitable for desktop or lab bench use. Front and rear acrylic panels provide a showcase for safe viewing of the installed cards, and can be removed for access to the hardware. The design allows customers to immediately begin using new 3U VPX hardware, such as our XPedite5370 and XPedite71 70 single-board computers. The complete XPand1000 development system includes a backplane, I/O breakout module, heat fins, fans and power supply. The system is priced at $2,215. Extreme Engineering Solutions, Middleton, WI. (608) 833-1155. [www.xes-inc.com].

January 2009

Products & TECHNOLOGY 2U 20-inch LCD/Keyboard Has CAT5 KVM Switch

The increasing use of networking in embedded applications is fueling demand for more displayed nodes where users tap into that network. Serving those needs, Neuro Logic Systems offers a ruggedized rackmount 20-inch LCD/sealed keyboard and integrated 16-port KVM switch. The RFT-20-2U-MIL-CAT5 is designed for military and harsh environment use and features a sealed, backlit, full-travel keyboard and sealed 38 mm 3-button trackball. When closed, the lightweight, aluminum alloy RFT-20-2U-MIL-CAT5 stows into a single, 17-inch deep 2U space in a transport case or standard RETMA equipment rack. The high-quality, wide view angle 20-inch LCD is protected by strengthened, anti-reflective glass filters. The LCD native resolution is 1600 x 1200. The integrated 16-port CAT5 KVM switch allows the unit to connect up to 16 servers via a standard CAT5 Ethernet cable instead of using the older, larger and heavier analog cables. All the above features are installed in an aluminum housing designed to meet Military Specifications 461E, 167, 810 and 901D. While the RFT-20-2U-MIL-CAT5 was designed for military use, it is priced at a level that allows it to be used in any harsh environment situation including vibration, heat and cold. Pricing starts at $5,750. Neuro Logic Systems, Camarillo, CA. (805) 389-5435. [www.nlsdisplays.com].

PowerPC 7448 Upgrade Provides 20-30% Increase in Application Performance

The RACE++ Series PowerPC 7448 Multicomputer is a high-performance, drop-in product upgrade that adds significant improvements in processor speed, L2 cache size and application performance over previous RACE++ products built on the PowerPC 7447A processor. The new PowerPC 7448 Multicomputer from Mercury Computer Systems comprises a Mercury MCJ6 motherboard and two faster daughter cards that interface directly with the RACE++ switch fabric. By maintaining commonality with previous RACE++ Series PowerPC products, Mercury provides an easy upgrade path for existing system designs. The Mercury RACE++ Series interconnect fabric represents a robust and proven approach for multicomputer digital signal processing, with products deployed in more than 1,000 programs worldwide. The RACE++ Series PowerPC 7448 Multicomputer allows integrators the option to increase application performance while preserving their investment in the existing system chassis, hardware and software—thus providing savings in development costs. The two daughter cards on the RACE++ Series PowerPC 7448 Multicomputer combine two 1.06 GHz MPC7448 processors with AltiVec technology to address today’s power-sensitive legacy upgrade market. The low-power version of the PowerPC 7448 enables a RACE++ product upgrade that has less power dissipation than the previous generation based on the PowerPC 7447A, minimizing customer requalification costs. Mercury Computer systems, Chelmsford, MA.(866) 627-6951. [www.mc.com].

January 2009

2W DC/DC Converters Boast Compact Size

As mixedvoltage systems become more the exception than the rule, embedded systems are requiring more and more DC/DC converters in their design. Fortunately, converter manufacturers continue to offer more compact and functional products. Along such lines, Martek Power offers two new series of compact DC/DC power converters: the 200VFI and 200WFRS. The 200VFI series is a family of cost-effective 2W single and dual output DC/DC converters with 18 models that operate from input bus voltages of 5V, 12V and 24V; producing output voltage levels of 5V, 12V, 15V, ±5V, ±12V and ±15V. Measuring 1.25 x 0.8 x 0.4 inches, the models feature short circuit protection, 4000VAC input/output isolation and built-in PI filter. The 200VFIs are priced at $11.75 per unit for volume orders. Also new are the 200WFS series of costeffective 2W single and dual output DC/DC converters in an ultra-miniature “gull-wing” SMT package. Thirteen models operate from input bus voltages of 5V, 12V and 24V; producing output voltage levels of 5V, 12V, ±5, ±12V and ±15V for a wide choice. With many standard features such as 1000 VDC input/ output isolation, 2:1 Input range and remote on/off control, the units are priced at $6.50 per unit for volume orders. Martek Power Torrance, CA. (310) 202-8820. [www.martekpower.com].

INTO TECHNOLOGY COMING TO A CITY NEAR YOU rtecc.com

Products & TECHNOLOGY Mini ITX Motherboard with Core 2 Duo and Two GbE LAN Ports

Designed for applications requiring high performance in a small form factor such as POS, mobile kiosks, factory automation, gaming machines, multimedia and digital signage, a new Mini ITX motherboard is powered by the Intel Core 2 Duo processor. The MB-73170 from WIN Enterprises features an Intel Socket 479 Core 2 Duo processor supported by the Intel 965GME and ICH8M express chipsets. The Intel 965GME mobile chipset (GMA X3100) supports dual independent displays such as CRT + LVDS, CRT+DVI. By means of an Intel 82573L Ethernet controller, the board supports two GbE LAN ports and the IEEE 802.3u standard for network connectivity. The onboard ICH8M chipset supports HD audio with mic in, line in and line out. Two DIMM sockets support up to 4 Gbytes of DDR2 system memory (533/667 MHz). For those requiring additional multimedia support, expansion is available through one PCI slot, one Mini PCI slot and a CompactFlash socket. Two Serial ATA interfaces provide up to a 150 Mbyte/s transfer rate. The board features six USB 2.0 ports and six COM ports. MB-73170 is equipped with an onboard TPM 1.2 chip to provide a runtime mechanism for encryption support during transaction, communication and signature phases. TPM 1.2 assists in verification of user authentication and authorization. The Intel 965 Chipset enables broad operating system support that includes Windows XP, XP Embedded and Vista (this includes 64-bit versions). Support is provided for these Linux versions: Fedora 9, Suse 11, Debian/Ubuntu, Gentoo and Redhat; as well as the Unix-like FreeBSD. Price without the processor in OEM quantities is $234.

The MathWorks Introduces Simulink Fixed Point 6

A major upgrade within the Simulink product family from The MathWorks, Simulink Fixed Point 6 delivers capabilities for designing and simulating fixed-point systems and generating optimized code for implementation, while creating more efficient workflows. Simulink Fixed Point 6 supports word lengths up to 128 bits to facilitate more accurate designs. It enables engineers to quickly turn a floating-point design built in Simulink, Stateflow, Embedded Matlab functions, Signal Processing Blockset, and Video and Image Processing Blockset into a fixed-point design based on input data ranges and target hardware characteristics.

Win Enterprises, Andover, MA. (978) 688-2000. [www.win-ent.com].

Handheld RF Analyzer Aids Wireless/Wi-Fi Diagnostics

Convenient, cost-effective RF spectrum analyzers are finding increasing demand with the exponential use of wireless technology and the increased challenges of identifying interference and available frequencies. Created for RF analysis, where bringing out an expensive or bulky bench-size analyzer is not reasonable, the HF60100 fills the need at a price that is affordable. The new model, the Spectran HF60100 from Kaltman Creations, is a handheld RF analyzer that features a frequency range of 1 MHz to 9.4 GHz with a sensitivity of 130 dbm and a bandwidth resolution of 1 KHz. Stability is 0.5 PPM yielding an accuracy of +/- 1 db. Convenience is enhanced with automatic unit conversions and unlimited marker points. The HF60100 is sold as a complete analysis kit that includes: HF60100 (1 MHz - 9.4 GHz) RF analyzer, optional internal pre-amp, an SMA stick antenna, a precision handheld LOG Periodic antenna, AC charger/adaptor, aluminum carrying case, mini-tripod stand, 6-foot USB cable, PC software and documentation. Priced at $4,975 Kaltman Creations, Suwanee, GA. (678) 714-2000. [www.kaltmancreeationsllc.com].

The new version can be used with MathWorks Real-Time Workshop Embedded Coder to generate bit-accurate production code from fixed-point models, ensuring the implemented design performs exactly as it did in simulation. Code can be generated for signals and parameters with word sizes from one to 128 bits, accelerating simulation modes and enabling embedded code deployment. Designers can now develop and implement algorithms for modern digital signal processors (DSPs) and microprocessors, which offer large accumulator and integer word sizes for high degrees of numerical range and precision. Additionally, the advisor feature in Simulink Fixed Point 6 moves design from floating to fixed point using an automated system conversion approach that creates optimized workflows in shorter time spans. This in turn facilitates more accurate and cost-effective design of high end, mass production components, especially in the signal processing segment. Simulink Fixed Point 6 is available immediately for the Microsoft Windows, Solaris, Linux and Macintosh platforms. U.S. list prices start at $1,000. The MathWorks, Natick, MA. (508) 647-7000. [www.mathworks.com].

January 2009

Altera Stratix IV GX-based AMC Supports Comms and Military/Aero Apps

An Advanced Mezzanine Card (AdvancedMC) can be populated with multiple densities of the Altera Stratix IV GX FPGA, from 230K to 530K logic elements (LEs). The state-of-the-art Altera Stratix IV GX FPGA at the heart of the S4-AMC (S4SM) from BittWare provides up to 530K equivalent logic elements (Les), 20 Mbits of RAM and 1,024 embedded multipliers in addition to up to 27 full-duplex, multi-gigabit transceivers—24 of which are high-performance, capable of supporting PCI Express Rev 2.0 and Serial RapidIO Rev 2.0. The S4AM connects the Stratix IV GX transceivers to three ports (1, 2 & 3) in the AMC common options region, and 16 ports (4 - 15, 17 - 20) in the AMC fat pipes region. These 19 ports provide a network data and control switch fabric interface on the AMC connector. Additionally, four of the Stratix IV GX transceivers are made available on the AMC front panel via an InfiniBand-type connector. All AMC clocks are also connected to the Stratix IV GX FPGA to facilitate system synchronization.

BittWare’s ATLANTiS FrameWork provides the Stratix IV GX FPGA on the S4AM with fully validated board-level physical interfaces to I/O, communications and memory, along with DMA engines and resource arbitration. A standard, open data interconnect fabric implemented using Altera’s streaming Avalon interface allows the ATLANTiS modules to be easily connected. Similarly, a standard control fabric using memory-mapped Avalon allows ATLANTiS modules to be easily controlled and coordinated. The ATLANTiS FrameWork frees users of the S4AM from having to reinvent low-level IP for the FPGA, and allows them to focus on the development of their application’s unique processing and I/O requirements. BittWare’s FINe Bridge implements a complete host/control bridge for the S4AM with extensive software support, thus removing that complexity from the data plane, I/O and processing implemented on the Stratix IV GX FPGA. The FINe provides Gigabit Ethernet on port 0 of the common options region, along with 10/100 Ethernet and RS 232 monitor port on the AMC front panel, and is connected to the FPGA via a local control bus. BittWare offers a complete suite of development tools to make developing and debugging applications for the S4AM easy and efficient. These software tools provide host interface libraries, a wide variety of diagnostic utilities and configuration tools and debug tools. This tool set is comprised of BittWare’s DSP21k Toolkit and DSP21k Porting Kit. Pricing starts at $5,000. BittWare, Concord, NH. (603) 226-0404. [www.bittware.com].

E-Frame Open Access Chassis Aids Development for 3U Backplanes

The E-frame from Elma Electronic has been available in heights for 6U backplanes and now accommodates 3U heights. The unit does not have side or back walls around the card cage area, providing easy access for test and development. The E-frame was designed with the high power and cooling requirements for VPX (VITA 46/48) systems. However, any backplane architectures in a 3U height can be used, including VME and CompactPCI. The total height of the portable E-frame is approximately 9U, with a convenient carrying handle on top. With a rugged modular aluminum construction, the E-frame tower can support up to 21 slots at .8” or 17 slots at 1” pitch. There are front accessible test points and monitoring LEDs for +3.3V, +/-5V, +/- 12V, +/- 24V and +/- 48VDC. The E-frame features high-performance cooling with 3 x 150 CFM fans under the card cage. The fans are speed controlled with fan fail indication. A system monitor with remote monitoring via Ethernet capability is optional. The black, powder-coated finish of the E-frame enhances aesthetics. Other features include a Rear A/C PEM (Power Entry Module) with fuses, GND Stud and Front located ESD jacks. The unit also offers full RTM (rear transition module) support. Pricing for the E-frame is under $3,500 depending on volume, backplane architecture and options. Elma Electronic, Fremont, CA. (510) 656-3400. [www.elma.com].

New CPPD Line of Frequency Multipliers in Rugged Housings

A new line of frequency multipliers is designed in a rugged SMA housing for easy in-line assembly and operation in test equipment and general lab use. The Crystek Pocket Passive Doubler (CPPD) series features an input drive level of +10 dBm to +20 dBm with a typical conversion loss of 15 dB for the CPPD-0.85-2.0 and 13 dB for the CPPD-2-4. Isolation (Fo, 3Fo, 4Fo) for both is rated at 33 dB typical. Maximum output power rating for the CPPD line is +27 dBm, with an operating temp of -40° to 85°C and a storage temp of -55° to 100°C. The CPPD-0.85-2 will accept an input frequency from 0.85 to 2.0 GHz and multiply by a factor of two to achieve output frequencies from 1.7 to 4.0 GHz. Likewise, the input for the CPPD-2-4 is 2.0 to 4.0 GHz with an output of 4.0 to 8.0 GHz. The CPPD-2-4 is available from stocking distributors at $31.29 each. For additional pricing details, contact Crystek Corporation. Crystek, Ft. Myers, FL. (239) 561-3311. [www.crystek.com]. January 2009

NEWS, VIEWS &

Comment January 2009

Warren Andrews Associate Publisher

Were There Bright Spots in 2008? Is There Hope for 2009?

appy New Year. Well, let’s at least hope it’s going to be a happy year. We’ve seen some rough times in 2008—record layoffs in Silicon Valley and elsewhere in the computer/ electronics business; major losses and dismal forecasts from semiconductor companies, particularly memory makers; lackluster forecasts from everyone from cell phone makers to communications companies, and shortfalls being reported in the industrial control and automation business. And if that’s not enough, we’re seeing new hostility between Palestine and Israel, the potential eruption of hostility between Israel and Iran, saber rattling by the Russians, a continued threat of domestic terrorism, mayhem in the financial markets, a failed auto industry and escalating domestic bail-outs. As Dirty Harry might say, “go ahead, make my day.”

A Great New Year

However, this is a new year and I think it’s important that we all have a positive outlook—or at least try to put a positive spin on some dismal news. First off, not all news for 2008 was dismal. For example, on the technology front, several developments have set the stage for new product development for future years. Many of the developments below have already started generating the platforms—and revenue—that will be at the heart of the embedded-computer business for years to come. One of the most significant technology boosts for 2008 might well have been Intel’s 45nm hafnium-based technology that has enabled the Atom, which in turn is pushing 32-bit embedded control into smaller places where issues of space, power consumption and heat had been previously prohibitive. We need only look at the explosion of small boards and modules taking advantage of Intel’s Atom family (and other small low-power processors such as Via’s Nano) to see that these developments have put a ma-

January 2009

jor tool in the hands of embedded-computer developers. These new products promise to enable a new generation of embedded processors in a broad base of applications—many currently untouched by the semiconductor industry. And, such processor technology has made possible the nextgeneration netbook—some kind of cross between the intelligent phone/PDA and a notebook computer. While this new family of devices is squarely in the commercial space, look for the transfer of many design ideas, chipsets and other assets from the netbook world to the embedded world. It’s already started. What’s ahead for Intel? CEO Paul Otellini envisions an x86based processor in everything. RTC Editor-in-Chief, Tom Williams, says there is much more for Intel coming as it plans to turn “green”—that is, turn its shareholders green with cash while contributing in a major way to save energy. Where’s the technology future? “I guarantee you that Moore’s Law will not end on my watch,” Otellini says. So expect to see Intel continue process development on a grand scale. Already the company has 32nm feature sizes in the works and on experimental die. That’s about a 30% decrease in feature size, translating to commensurate savings in power dissipation and cost. The industry has long been chasing the perfect memory technology—dense, fast, non-volatile and cheap. As a result, memory technology has seen some significant development from the early days of DRAM, SRAM and PROMs on to flash. However all have been essentially the same. The latest effort from HP, though envisioned long ago, comes in the form of a memory resistor, which it claims to have built. The theory of a “memristor” was postulated back in 1971, but it has taken until now to make a working version. HP’s claim is that it would be more energy efficient than flash or other technologies, and can permanently store information. The prototype was made putting a film of titanium

dioxide between electrodes and applying a charge. As the charge flows through it, the atomic structure actually changes. Not yet in production But it gets better. A new approach, from Numonyx, a joint venture recently formed by Intel and STMicroelectronics, is betting on a technology called phase-change memory, which takes advantage of certain materials’ characteristics to change physical structure with the application of heat. The company claims it is now shipping 128 Mbit chips using the phase-change approach. The company claims it will have dramatic density increases in the coming year. An interesting aside is that the phase-change technology has been attributed to Stan Ovshinsky, a name that might be familiar to older readers. Ovshinsky was the founder of a technology that became known as Ovonics, and he worked primarily with photovoltaic cells but studied and developed many patents on phase shifting from amorphous to crystalline states. In April this year, HVVi Semiconductor announced a new way to do RF Power Transistors that allows higher operating voltages (48V and more), thereby outperforming the dominant RF power transistor standard (Lateral Double-diffused metal Oxide Semiconductor (LDMOS)) that’s been used for the past 15 years in radar systems. Look for newer, more powerful and less expensive RF systems coming down the pike. These are but a very few of the many heady developments of the year from the semiconductor side of the fence. However, they have been enabling technologies and have made possible innovation in other areas. SUMIT and COMIT, for example, are two of the latest standards developed to provide module makers and users more flexibility for more powerful cards. Other developments on the module front include a MiniBlade spec for Solid-State Drives on small boards also by the SFF-SIG. GPS is now everywhere. And while people are quick to point out that the system is a lot older than many think (it first became operational in 1978), it suddenly popped back into the mainstream of activity this year with GPS capability showing up on phones and notebooks in the commercial world. Devices that don’t have true GPS capability have made do with cell-tower triangulation or geolocation based on Wi-Fi hot spots. And it’s all crept onto all variety of boards and modules in the embedded world for any number of unique applications from location monitoring of trucks and cargo to knowing where patients are for medical monitoring. And, lest we forget, we are firmly entrenched in the age of the multicore processor, which has enabled countless new SBCs and powerful computing engines. Only a year or so ago, these megaliths reached the commercial market and already have become the de facto processor technology for virtually all applications. And we’re already seeing the next-generation Atom processor emerge in multicore configurations as the demand is voracious. And if that’s not enough, we’re going to be seeing some real-time software capability for these new processors in coming months.

bleak as forecasters would have it. That’s right, it’s OK to look up, the sky isn’t falling. Though in some cases it’s a bad-news, good-news scenario, there is a good-news component. For example, the SIA (Semiconductor Industry Association) reports that global chip sales have declined. However, after saying that, it adds that the sector will remain the U.S.’s second largest exporter for the year. Further, the largest drop has been with makers of memory chips that have suffered price erosion— buy now while the getting is good. The other good news is that researchers can’t agree on what 2009 will look like. While all predicted a drop in 2009, it ranged from 5.6% to 16%, there was universal agreement that there would be a rebound in 2010. Further, many corporate technology chiefs see some bright spots forthcoming. While they agree that not everything is going to be rosy, areas of technology such as online software, mobile applications and security will likely see a boost in coming months. Also, though overall technology spending may be soft, and technology buyers may not be trotting off to buy the latest and greatest, new generations of embedded computers are likely to hit the spot in some soft areas with room for productivity enhancement. And while we’ve seen the price of crude oil drop to multiyear lows, individuals and companies became critically energy conscious only six months or so ago when oil approached the $150 a barrel level. Look for new energy saving devices from motor control to environmental control in the industrial area this year. The bottom line is that the population’s mindset is changing to be greener and more economical, moving those market areas forward with a concomitant reliance on embedded electronics And finally, the military/aerospace market promises solid growth throughout 2009. I just completed my annual market update and forecast for our sister publication, COTS Journal, and despite the change in administration, all indicators are pointing in a positive direction for the merchant market for embedded computers in the military and aerospace markets. Two thousand eight may not have been the most sterling year on many of our calendars. However, of the many people I spoke with in the embedded-computer business toward the end of the year, all had much the same thing to say, “in spite of what everyone is saying, business has been pretty good. I just hope it stays that way.” Happy New Year.

On the Other Fronts

And while we’ve been focusing on technology development and innovation, not all the business and economic news is as January 2009

with an Application Engineer, or jump to a company's technical page, the goal of Get Connected is to put you in touch with the right resource. Whichever level of service you require for whatever type of technology, Get Connected will help you connect with the companies and products you are searching for.

www.rtcmagazine.com/getconnected

Advertiser Index Get Connected with technology and companies providing solutions now Get Connected is a new resource for further exploration into products, technologies and companies. Whether your goal is to research the latest datasheet from a company, speak directly with an Application Engineer, or jump to a company's technical page, the goal of Get Connected is to put you in touch with the right resource. Whichever level of service you require for whatever type of technology, Get Connected will help you connect with the companies and products you are searching for.

www.rtcmagazine.com/getconnected

Company

Page

Website

ACCES I/O Products......................................................................................... 16.............................................................................................. www.accesio.com Birdstep Technology......................................................................................... 20............................................................................................. www.birdstep.com

Products

End of Article

COM Express and FPGA Boards Showcase......................................................28,29......................................................................................................................

Concept Developement . .................................................................................. 35................................................................................................ www.cdvinc.com congatec.......................................................................................................... 21............................................................................................ www.congatec.com Elma Bustronic Corp......................................................................................... 25.................................................................................... www.elmabustronic.com

Get Connected with companies and

Get Connected

Magma............................................................................................................. 9................................................................................................www.magma.com products featured in this section.

with companies mentioned in this article.

www.rtcmagazine.com/getconnected www.rtcmagazine.com/getconnected Micro/sys, Inc................................................................................................... 4......................................................................................www.embeddedsys.com Microsoft Windows Embedded........................................................................... 2........................................................................... www.microsoft.com/embedded National Instruments........................................................................................ 12....................................................................................................... www.ni.com One Stop Systems............................................................................................ 13................................................................................. www.onestopsystems.com

Get Connected with companies mentioned in this article. www.rtcmagazine.com/getconnected Phoenix International........................................................................................ 33.............................................................................................www.phenxint.com Get Connected with companies and products featured in this section. www.rtcmagazine.com/getconnected Real-Time & Embedded Computing Conference.................................................. 45.................................................................................................. www.rtecc.com

Themis Computer..............................................................................................17............................................................................................... www.themis.com VersaLogic Corporation..................................................................................... 51...........................................................................................www.versalogic.com WinSystems..................................................................................................... 52........................................................................................ www.winsystems.com

RTC (Issn#1092-1524) magazine is published monthly at 905 Calle Amanecer, Ste. 250, San Clemente, CA 92673. Periodical postage paid at San Clemente and at additional mailing offices. POSTMASTER: Send address changes to RTC, 905 Calle Amanecer, Ste. 250, San Clemente, CA 92673.

January 2009

When You Design for the Extreme, Choose Your Partner Wisely. Did we mention “Reliability” is our middle name?

ndustrial equipment needs to perform awlessly, night and day, under even the most extreme conditions. Whether designing for the manufacturing oor, clean room or the eld, you can depend on VersaLogic to deliver the highest quality embedded computer products, from prototyping and design-in, through years of product production. We design our boards for high reliability and long-term availability, then run them through exhaustive quality tests, ensuring that we deliver only the best. And with our world class service and ve year availability guarantee, we’ll always be there when you need us. Whether you need one of our standard products or a version customized to your needs, our skilled technical staff will work with you to meet your exact specications. Contact us to nd out how for more than 30 years we’ve been perfecting the ne art of extra-ordinary support and on-time delivery: One customer at a time.

1.800.824.3163 1.541.485.8575 www.VersaLogic.com/RTC

SBCs and I/O Modules Enable Machine-to-Machine Industrial Connectivity Whether wired or wireless connectivity is required, keep in touch with your industrial application with our SBCs and I/O modules. Small, rugged, and operational over extended temperature ranges, they are perfect for pipeline, transportation, MIL/COTS, utility, instrumentation, and homeland security applications. Wireless Support • GSM/CDMA Cellular • 802.11 a/b/g Networking • ZigBee Wired Support • 10/100/1000 Mbps Ethernet • Dial-up Modem (POTS) • USB 2.0 • RS-232/422/485 Serial I/O Other I/O Expansion • A/D, D/A • Digital • GPS Embedded Platforms Supported • PC/104 – 3.6 x 3.8 inches • EPIC – 4.5 x 6.5 inches • STD Bus – 4.5 x 6.5 inches • EBX – 5.75 x 8.00 inches Enclosures for SBCs and I/O Open-frame Panel PCs Operating Systems Supported • Windows® XPe and Linux • x86-compatible RTOS Extended Temperature Operation

Call 817-274-7553 or Visit www.winsystems.com Ask about our 30-day product evaluation!

Off-the-shelf delivery, knowledgeable technical support, and long-term availability makes these products the right choice for your application. TM

715 Stadium Drive • Arlington, Texas 76011 Phone 817-274-7553 • FAX 817-548-1358 E-mail: info@winsystems.com