RSC # @ www.industrial-embedded.com/rsc RSC # @ www.industrial-embedded.com/rsc www.industrial-embedded.com Volume 1 • Number 1 october 20 05 COMPUTING COLUMNS TECHNOLOGY 7 Foreword Thinking 48 Modern interfaces in light of embedded computer integration A fresh start to getting things done By Andreas Geh, DIGITAL-LOGIC AG By Don Dingee 54 Embedded compute models help contain costs 8 Industrial Europe By Ernest Godsey, MEN Micro  Q & A with Ulrich Gerhmann, CEO, and Norbert Hauser, 57 Product Profiles VP of Marketing, Kontron EMEA HUMAN INTERFACE By Stefan Baginski TECHNOLOGY 10 Market Pulse 80 Converging functionality in embedded industrial control IEEE 802.15.4 and ZigBee By Melissa Jones, Ultimodule By Bonnie Crutcher 84 Using software-configurable processors in biometric 98 The Final Word applications It’s all about choices By Philip Weaver, Stretch, and Fred Palma, A4 Vision By Jerry Gipper 87 Product Profiles SENSORS/CONTROL FEATURES TECHNOLOGY NETWORKING 88 Combining a hardware neural network with a powerful SPECIAL: Standards automotive MCU for powertrain applications 16 Opening gates with TCP-to-CANopen By Dr. Paul Neil, Axeon By Holger Zeltwanger, CAN in Automation APPLICATION 20 Performance, implementation, and applications of 90 Open architecture PAC technology drives undersea remotely Powerlink operated vehicles By Frank Foerster and Bill Seitz, IXXAT By Chris Ward, C&M Group TECHNOLOGY 91 Product Profiles 12 Ultra-wideband communication for low-power wireless STORAGE body area networks TECHNOLOGY By Bart Van Poucke and Bert Gyselinckx, IMEC 94 Software architecture for managing resident flash 26 Building low power into wireless sensor networks using By Robert Krantz, Datalight ZigBee technology 97 Product Profiles By Jon Adams, Freescale Semiconductor 31 Creating value with ZigBee Networks By Chris Herzog, Software Technologies Group EVENTS GSPx 34 Deploying ZigBee in existing industrial automation networks October 24-27 • Santa Clara Convention Center, Santa Clara, CA By Tim Cutler, Cirronet www.gspx.com/GSPX/2005 38 Adding with minimal software changes By John Korsakas, Grid Connect ISA Expo 42 Embracing WLANs in the industrial market October 25-27 • McCormick Place Lakeside Center, Chicago, IL By Mukesh Lulla, TeamF1, and Ron Fredericks, www.isa.org/isaexpo2005 Embedded Components APPLICATION WEB RESOURCES 40 Whiteware industry drives Ethernet bus Subscribe to the magazine or E-letter: By Debra Biela, Contemporary Controls www.opensystems-publishing.com/subscriptions 46 Product Profiles Industry news: Read: www.industrial-embedded.com/news Submit: www.opensystems-publishing.com/news/submit Published by: Submit new products: OpenSystems ©2005 Industrial Embedded Systems www.opensystems-publishing.com/vendors/submissions/np Publishing™  / 2005 Industrial Emdbedded Systems Resource Guide RSC # @ www.industrial-embedded.com/rsc OpenSystems Publishing™ OpenSystems Publishing Advertising/Business office: 30233 Jefferson Avenue St. Clair Shores, MI 48082 A n O p e n S y s t e m s P u b l i c a t i o n Tel: 586-415-6500 n Fax: 586-415-4882 Vice President Marketing & Sales Patrick Hopper Embedded and Test & Analysis Group [email protected] n Embedded Computing Design Senior Account Manager n Embedded Computing Design E-letter Dennis Doyle n Embedded Computing Design Product Directory [email protected] n Account Manager Industrial Embedded Systems Resource Guide Tom Varcie n PXI Technology Review [email protected] Editorial Director Don Dingee Print and Online Marketing Specialist [email protected] Christine Long Editorial Director Jerry Gipper [email protected] Technical Editor Chad Lumsden Advertising/Marketing Coordinator [email protected] Andrea Stabile Associate Editor Jennifer Hesse [email protected] [email protected] European Bureau Chief European Representative Hermann Strass Stefan Baginski [email protected] [email protected] Special Projects Editor Bob Stasonis Account Manager Senior Designer Joann Toth Doug Cordier Senior Web Developer Konrad Witte [email protected] Graphic Specialist David Diomede Business Manager Circulation/Office Manager Phyllis Thompson Karen Layman [email protected]

For reprints call the sales office: 586-415-6500 OpenSystems Publishing™OpenSystems Publishing Editorial/Production office: 16872 E. Avenue of the Fountains, Ste 203, Fountain Hills, AZ 85268 Tel: 480-967-5581 n Fax: 480-837-6466 Website: www.opensystems-publishing.com Publishers John Black, Michael Hopper, Wayne Kristoff Vice President Editorial Rosemary Kristoff Communications Group Editorial Director Joe Pavlat Assistant Managing Editor Anne Fisher Senior Editor (columns) Terri Thorson Advertiser Index Technology Editor Curt Schwaderer Associate Editor Jennifer Hesse European Representative Hermann Strass RSC# Company Advertisement Intel Embedded Roadmaps Military & Aerospace Group 2 Group Editorial Director Chris Ciufo 3 WinSystems Embedded PCs Managing Editor Bonnie Crutcher Assistant Editor Sharon Schnakenburg 5 SMA Embedded Computer Solutions Senior Editor (columns) Terri Thorson 15 Diversified Technology PCI based Industrial European Representative Hermann Strass European Bureau Chief Stefan Baginski Embedded Systems 19 Arcom Control Systems Intel Pentium M 25 VersaLogic EBX, PC/104, EPIC, STD 32 37 Vector Electronics Take a Closer Look 41 KUKA Controls Download Free Demo 44 Grid Connect Intelligent Chips & Modules 5301 GD California MVME products now available 5302 Real Time Automation ASCII Barcode to Rockwell PLCs

99 Arcom Control Systems Single Board Computers, Industrial Embedded Systems is published semi-annually by OpenSystems Publishing Industrial Compact Enclosures LLC., 30233 Jefferson Ave., St. Clair Shores, MI 48082. Subscriptions are free to persons interested in the design or promotion of 100 WinSystems Fanless EBX 733MHz P3 with COM industrial embedded systems. For others inside the US and Canada, subscriptions are $34/year. For 1st class delivery outside the US and Canada, subscriptions are $60/year (advance payment in US funds required). Canada: Publication agreement number 40048627 Return address: WDS, Station A, PO Box 54, Windsor, ON N9A 615 POSTMASTER: Send address changes to Industrial Embedded Systems 16872 E. Avenue of the Fountains, Ste 203, Fountain Hills, AZ 85268

 / 2005 Industrial Emdbedded Systems Resource Guide foreword thinking >> By Don Dingee

New, intelligent technologies are driving the transformation of As a designer, marketer, or manager for a company involved industrial automation. These technologies are changing the way with developing or applying industrial automation products and we live, the way we work, and the way things get done. The services, you need to keep informed of these changes. As we very definition of industry is changing as new applications are looked the available information sources, we found no single developed putting these new technologies to work. publication covered an industrial view of intelligent devices in computing, human interface, networking, sensors/control, and I’ve seen this change first hand when our household recently storage as a combined set of technologies. That is why we’ve shopped for a washer/dryer. Our home has some constraints in started this publication, Industrial Embedded Systems. the laundry space – it’s really more of a laundry hallway with a door to the garage opening inward just clearing the front of the In our coverage of new intelligent technologies for industrial dryer. As we started looking at washer/dryer combos we found automation, our goal is to bring you a fresh perspective that many higher-end models with the features we wanted simply combining elements from our focus categories. We look for wouldn’t fit in the available depth. We had to search harder than highly innovative, intelligent products and services that are we expected to find machines that fit our space. I can’t believe we helping create new best-practices and make significant shifts in have the smallest laundry space in America, but it’s apparently how things get done – in manufacturing, retail, entertainment, the case. transportation – anywhere automation applies. Every issue features technology and application topics along with a new But as we continued to shop, we discovered an interesting and product guide section. Contributed editorial from leading innovative solution – and probably a better performing one. We companies and standards consortiums, along with ongoing settled on appliances from Fisher & Paykel designed and built market analysis and perspectives, will help keep you up to date in New Zealand (where they must have smaller laundry spaces, on what is happening in the industry. too). The Fisher & Paykel GWL11 washer is one of the most advanced, efficient machines on the market today using about In this issue, we’ll feature viewpoints on networking standards one quarter of the energy and much less water and detergent than and applications, including TCP-to-CANopen, ETHERNET a traditional washer. Powerlink, ZigBee, 802.11i and Ethernet/IP. We also have views on embedded computing modules, system-on-chip signal This efficient performance is made possible by a clever and stylish processing, flash storage management, and application stories design centered on a programmable, direct drive brushless DC in open architecture control, transportation, and networking, motor system. Working with its 16 selectable agitation actions including a story on the washer/dryer factory shown in our controlled from the user interface panel, its load sensing system cover photo. determines exactly how much water to use based on the weight of the load, and dynamically balances the tub as the load shifts. It’s this kind of forward thinking we’ll be bringing you in each After completing the washing cycle, it transitions to a 1,000 RPM issue of Industrial Embedded Systems. Please ask your peers spin drying cycle emitting a whirring sound remarkably like a to subscribe at www.industrial-embedded.com so they can join turbine engine, saving further energy by requiring less drying the discussion. since it uses incredible centripetal force to squeeze the moisture from the clothes. We look forward to continuing to serve your information needs on technologies for industrial automation. As always, Our new washer probably has more intelligence than the I enjoy dialogue with our readers and encourage you to e-mail first personal computer I owned (circa 1984, an Intel 8088- me with your thoughts (pro or con), ideas, and suggestions at based machine running MS-DOS at what is now a pedestrian [email protected]. 4.77 MHz). This type of intelligence is showing up in so many places now in devices that we interact with daily, both at home and in the workplace. This sweeping infusion of intelligence is often to a degree beyond that we thought possible just a few years ago, and it’s changing the way all industrial automation systems Don Dingee are conceived, designed, and implemented. Editorial Director

Industrial Emdbedded Systems Resource Guide 2005 /  Industrial

Europe By Stephan Baginski

with Ulrich Gerhmann, CEO, and Norbert Hauser, VP of Marketing, Kontron EMEA

In a recent interview 25 percent in North America, and almost 40 percent in Emerging markets. So we have healthy growth rates and a stable market position. Our annualized growth has been about 40 percent over with Kontron EMEA the past six years, with organic growth of 21 percent (excluding acquisitions and currency exchange). CEO Ulrich Gehrmann Kontron has a strong competitive advantage in cost efficient engineering:

n We have the highest research and development headcount in and VP of Marketing the embedded computing industry with some 728 engineers, about 38 percent of our total staff. Norbert Hauser, Industrial n We target applications with high research and development content. n We reuse engineering know-how by a global CAD/CAE Embedded Systems asked system. n We make use of efficient engineering resources in Eastern Europe. them to discuss their n We have low employee turnover (less that 1.5 percent annually). views on the embedded I must stress that engineering is the key differentiator in embedded computers; Kontron is not different from other vendors, but perhaps a little better. Our organization is extremely flat and computing market. distributed. We outsource many services to other specialized entities. IES: Mr. Gehrmann, Kontron AG is constantly introducing new products. Your recent integration of Dolch Computer is IES: How do you view your competitive environment? only one of many recent news developments from Kontron. GEHRMANN: We closely watch Motorola Embedded What in your view makes Kontron an important enterprise Communications Computing, which acquired Force Computers in the world’s embedded computing markets? recently, and companies like Advantech, RadiSys, GE Fanuc GEHRMANN: It is the size of our revenues and our position Automation, and several smaller players. in the world markets. We operate on three distinctive markets: Our strategy is to avoid mega mergers and expensive acquisitions. EMEA, North America and what we call Emerging. Slightly We prefer organic growth. If we choose to do an acquisition, more than 50 percent of our revenues are from Europe, around 40 we carefully select a target for technology expansion (in effect, percent are from North America, and the remainder comes from AdvancedTCA) or for regional expansion (in effect, China). We new emerging markets such as Russia and China. When we look believe that by 2010 the three largest players should own more at the market dynamics, we grow at about 18 percent in Europe, than 50 percent of the market.

 / 2005 Industrial Emdbedded Systems Resource Guide Industrial Europe We will continue to move Kontron up the value chain, from the PCI Industrial Computers Manufacturers Group (PICMG). modules to systems and up to solutions. Kontron sells embedded Development of the official COM Express specification began computers to OEMs with forward integration into vertical in early 2004. Based on the released COM Express standard, markets. Kontron ETXexpress products now bring advanced embedded computing technology for tomorrow’s applications. IES: And which vertical segments experience strongest growth? IES: Mr. Hauser, can you describe the product further?

GEHRMANN: We see embedded computing growing mostly HAUSER: Kontron ETXexpress products are built around in the areas of communications, transportation, automation, and serial differential signaling technologies. ETXexpress modules utilities. Growth drivers are found in outsourcing trends: incorporate interfaces for PCI Express (4x1, 1x16), Serial ATA, USB 2.0, LVDS, Serial DVO, and much more into a 95x125mm n New technology enables new embedded computing products small form factor embedded computing module. Kontron (in effect, UMTS enables wide band connectivity for trucks). ETXexpress modules provide the highest available performance n Cost advantage driven by number of design wins. on the smallest state-of-the-art embedded modules. Additionally, n Standardization trend toward COTS technology. by following modular concepts for embedded design, ETXexpress with Ulrich Gerhmann, CEO, n The total embedded computing market is around $26 billion, modules safeguard research and development investments and according to Electronic Trends and VDC. We estimate our lower cost of ownership. available market to be around $4 billion with a CAGR of and Norbert Hauser, VP of Marketing, Kontron EMEA 15 percent. The ETXexpress-PM delivers up to 2.13 GHz processor performance and supports up to 2 GB DDR2 memory. For Kontron is well placed and our expectations are high. Market applications that require advanced real-time capabilities, the response and prognosis seem to confirm that. ETXexpress-PM has integrated graphics based on Intel Graphics Media Accelerator 900 architecture and also supports PCI IES: Many technology-driven companies experience Express graphics. unhealthy dependencies on single customers or on single industry. I would like to mention another product family where Kontron is offering one of the broadest portfolios on the market. We have GEHRMANN: That is not the case with Kontron. Our biggest developed modular COTS solutions for communications based customer constitutes around 5.3 percent of revenues, and our on the AdvancedTCA and Advanced MC (AMC) standards. We biggest sector is less than 25 percent of revenues. recently introduced two new AMCs – the AM4300 quad Gigabit IES: We have referred to Embedded Computer Technology Ethernet module and the AM4500 high capacity SATA storage several times now. What does it mean in your company? module. These are particularly attractive to Telecommunications Equipment Manufacturers (TEMs) and service providers as they GEHRMANN: We start with our basic product, which we call the can be hot-swapped and are accessible via the front plate. AMCs embedded brain, and we look at the application. Working closely are the smallest and most compact stand-alone Field Replaceable with systems integrators and OEMs, we create solutions for their Units (FRUs) in the entire telecommunications industry. needs. Then we network it to the infrastructure and the real world, providing interfaces to other systems and to people. IES: It’s good to see standards initiatives resulting in new products, which should help OEMs and systems integrators. In other words: Thank you both for sharing Kontron’s viewpoint with Industrial Embedded Systems. n We combine existing technology to make applications intelligent and to offer diversified computing platforms. To learn more, contact Norbert at: n We address a diversified range of applications, not depending Kontron EMEA on a single vertical market and a few customers. Oskar-von-Miller Str. 1 n We are in an engineering-driven market – 60 percent software 85386 Eching (Munich), Germany and 40 percent hardware content (per value). Tel: +49-8165-77-0 n We serve OEM projects; we do not only offer products. Fax: +49-8165-77-219 n We go for low volumes (100-30k pieces Pa), but we seek high E-mail: [email protected] research and development content. Website: www.kontron.com

IES: Can you give us specifics, maybe one of the new products in your COM technology?

GEHRMANN: In mid-2003, Kontron initiated a movement to bring PCI Express technology into computer-on-module or COMs. Intel and a group of leading embedded computing companies later bolstered this movement. These industry leaders formed a working group in January 2004 under the auspices of

Industrial Emdbedded Systems Resource Guide 2005 /  market pulse: IEEE 802.15.4 and ZigBee

Market growth expected n Together, System-in-Package (SiP) and System on Chip (SoC) to increase 200 percent solutions will drive easier system/product development and In December 2004, the lower the costs of adding this wireless capability to sensor ZigBee Alliance signed networks. off on the 1.0 ZigBee specification. In April Regarding sales prices, industry experts expect them to rapidly 2005, four companies, decline. However, the average pricing for all shipments will Chipcon, CompXs, Ember, and Freescale depend on the ratio of Reduced Function Devices (RFD), used as Semiconductor, received endpoints, to Full Function Devices (FFD), used as coordinators. the first ZigBee-compliant In-Stat indicates it is too early to estimate what that ratio will be platform certificates. Figure 1 As technology and the yet, as it depends on what type of topology will be in demand the alliance have continued to move forward, industry analysts most – Star, Cluster Tree, or Mesh (a full mesh requiring FFDs). from In-Stat have indicated that the benefits of 802.15.4 and ZigBee standardization will soon come to fruition, enabling In-Stat predicts 2005 to be an emerging year for 802.15.4 silicon, manufacturers to rein in many more advantages. Some of these with a few end products just barely beginning to appear. Silicon advantages will result in partnerships between companies, developments for SoC solutions that embed a microcontroller offering complete solutions for lighting, Heating, Ventilation, and have been moving at a quicker pace, with several vendors Air Conditioning (HVAC), Automatic Meter Reading (AMR), expecting to begin sampling such solutions this year. Based upon and industrial control applications, for example. discussions with system providers, the closer silicon vendors can get to a “ZigBee-in-a-box” solution, the greater the opportunities A major advantage is that 802.15.4 Wireless Personal Area will be. SoC and SiP play nicely into a box option. In-Stat also Network (WPAN) technology and ZigBee offer OEMs a menu of predicts that 2006 will become a significant springboard year, multiple silicon sources and ZigBee networking layer suppliers. turning pilots and prototypes into deployments that will move According to Joyce Putscher, In-Stat Research Director and this market into volume significance. Principal Analyst, “Additional advantages include pricing competition and system vendor partnering opportunities.” In-Stat has published a complete report on this market study. The report, 802.15.4 SoC & SiP Surge as ZigBee Faces Residential On an aggressive basis, In-Stat analysts forecast that 802.15.4 Competition (#IN0501836MI), covers the worldwide 802.15.4 nodes/chipsets could grow by a Compound Annual Growth market. It presents conservative and aggressive forecast scenarios Rate (CAGR) of 200 percent from 2004 to 2009, with annual for unit shipments, bill of materials average sales price, and shipments surpassing 150 million units in 2009 (see Figure 1). revenue for chipsets. It also includes profiles of leading vendors.

In-Stat researchers also found the following: To purchase this report, or for more information, contact: In-Stat n This low-cost, low-power, low-data-rate networking Attn: Courtney McEuen technology is receiving attention from many companies 6909 East Greenway Parkway, Suite 250 that are involved in industrial control, home automation, Scottsdale, AZ 85254 and commercial building control, spanning everything from Tel: 480-483-4454 nuclear power plants to hotels. The first residential products E-mail: [email protected] are just barely emerging on the scene. n Industry experts expect that commercial building control OR will capture the lion’s share of the 802.15.4 market, in terms Joyce Putscher, Director Converging Markets and Technologies of node/chipset volumes, but not in terms of design wins. Tel: 480-483-4475 Residential automation and industrial applications will E-mail: [email protected] follow respectively. Website: www.in-stat.com

10 / 2005 Industrial Emdbedded Systems Resource Guide Product Profile Resource Directory

RSC# Company Model RSC# Company Model 4601 PHYTEC America LLC PEAK PCAN Products 7201 Intel Intel® 945G Express Chipset 4602 Broadband Energy Networks Inc. Connector C2k 7301 Intel Intel® Pentium® 4 Processors 4603 Advanced Vehicle Technologies, Inc. AVT-418 Interface with Hyper-Threading Technology ® ® 4701 Grid Connect OEM Solutions 7401 Intel Intel Pentium M Processors on 90nm ™ 4702 Mesa Electronics PC104-PLUS MAC HU 7501 Tundra Semiconductor Tundra Tsi148 5701 AMTELCO AMTELCO XDS 7502 Research Centre Module 1879BA1T ASIC 5801 Innovative Integration M6713 7503 Gage Applied Technologies 12-Bit Digitizer 5802 Innovative Integration SBC6713e 7601 Arcom Pentium M 5901 PHYTEC America LLC phyCORE-LPC2294 7701 Carlo Gavazzi Computing Solutions Rugged ATR Chassis 5902 PHYTEC America LLC phyCORE-MCF5485 7702 Silicon Turnkey Express GP3SSA 5903 PHYTEC America LLC phyCORE-PXA270CE 7703 Vesta Technology, Inc. ProtoStar Design 5904 Saelig Co. Inc. USBwiz USB-file ic 7801 MPL AG Unique Panel PC 6001 Technoland (an iBASE company) MB 740 Mini ITX MB 7802 SoftPLC Corporation SoftPLC 6002 Technoland (an iBASE company) IB880 – PentiumM SBC 7803 Advanced Power Solutions APS Power Supplies 6003 Vesta Technology, Inc. ANTERO CONTROLLER 7901 OceanServer Technology, Inc. BBDC Battery Power 6004 Grid Connect Inc. DIMM EX CPU Module 7902 JRM Consultants Inc XP862 Power PC 6101 Digital Logic smartModule SM915 7903 Ardence, Inc. RTX 6102 Digital Logic smartModule SM855 8701 ACP ThinManager 2.6 6201 Inova ICP-PM, ICP-CM CPU 8702 EasyCODE Software EasyCODE V7.5 6301 Inova savings-box 8703 Trilogy Touch Technology Trilogy Touch LCDs 6401 Versalogic Cobra Pentium 9101 Printed Circuits Corp. PCB/PCBA’s Fab/EMS 6501 Versalogic Gecko 9201 SICK, Inc. UE4100 Remote I/O 6601 Versalogic Jaguar 9202 KineticSystems Company, LLC VXI Controllers 6701 Arcom Viper 9203 Mesa Electronics 4I68 Anything I/O 6801 Digital Logic MSM915 9204 Arcus Technology, Inc. Performax USB 2.0 6802 Microcomputer Systems, Inc. MSI-P402 9301 Entertron Industries 5.7” Touch Screen 6803 AAEON Electronics, Inc. PFM-620S PC/104+ 9302 Entertron Industries Elite-2000 ePLC 6901 Altera Nios II Dev. Kit Cyclone II Edition 9303 Entertron Industries SK1600-RIC ePLC 7001 Microcomputer Systems, Inc. MSI-P600 GPS Card 9304 Entertron Industries Smart-PAK Series 7002 Themis TA64 9701 KineticSystems Company, LLC Loggers/Recorders 7003 Themis TPPC64 9702 Lanner Electronics Inc. NS04-4100 7101 Altera Nios II Dev. Kit Stratix II Edition 9703 OSIsoft OEM Data Historian

Industrial Emdbedded Systems Resource Guide 2005 / 11 Networking: Technology Ultra-wideband communication for low-power wireless body area networks By Bart Van Poucke and Bert Gyselinckx

The ultra-low-power radio is a key The BANs will consist of various small, to allow energy autonomy of the battery- component of the autonomous wireless low-power sensor/actuator nodes with powered devices, making high demands sensor nodes in future Wireless Body Area sufficient computing power, wireless on the radio in the sensor node. The Networks (WBANs). Power consumption capabilities, and integrated antenna. Each average power consumption of this radio requirements of the radio interface node should have enough intelligence must be reduced to below 100µW. This are very severe, targeting an average to carry out its task. This can range stringent requirement cannot be met by power consumption of less than 100µW. from storing and forwarding algorithms today’s low-power short-range radios This article explores the capabilities of to complex non-linear data analysis. such as Bluetooth and ZigBee, which the emerging Ultra-WideBand (UWB) These nodes have radio interfaces to make use of standard narrowband radio technology to meet this challenge, communicate with other sensor nodes communication. An interesting alternative including performance data from a CMOS or with a master device or central node is to use ultra-wideband technology. pulser implementation. worn on the body. The central node communicates with the outside world Ultra-wideband modulation In the near future, microsystem technology by using a standard telecommunication for WBANs will largely contribute to an improved infrastructure such as a wireless local area Ultra-wideband refers to a radio quality of life by enabling people to carry network or a cellular phone network. The modulation technique based on their Body Area Network (BAN). This network can deliver services to the person transmitting very short duration pulses, BAN will be used to gather vital body using the BAN, such as the management often of duration of only nanoseconds information into a central intelligent node, of chronic diseases, medical diagnostics, or less, whereby the occupied bandwidth which in turn will communicate wirelessly home monitoring, biometrics, and sports goes to very large values. In 2001, the with a base station. Figure 1 shows the and fitness tracking. Federal Communications Commission aim of this network to provide the user (FCC) authorized UWB technology for with medical, sports, or entertainment The successful realization of this BAN telecommunication, provided that UWB functions, going along with the needs obviously depends on the ability to signals are within the 3.1 – 10.6 GHz of our society to pay increasingly more extend the capabilities of current devices. range, with a minimum bandwidth of 500 attention to comfort, better healthcare, Among the main technological obstacles MHz, and have a power spectral density and longer autonomy. is the energy consumption of all building below -41.25 dBm/MHz. For short-range blocks. This must be reduced drastically applications, UWB systems in general become more efficient than narrowband solutions in terms of power consumption. Moreover, in a WBAN context, the characteristics of the sensor nodes and their environment point to pulse-based UWB as a favorable communication principle. For instance, the sensor nodes of the WBAN have a very low data rate, allowing the radio to operate in burst mode with minimal duty cycle. In a pulse-based UWB, the transmitter only needs to operate during the pulse transmission, producing a second duty cycle inside the burst. This can reduce the baseline power consumption. In addition, the power budgets in the sensor node and in the master device Figure 1 are very different. The power budget in the sensor node is very tight, whereas

12 / 2005 Industrial Emdbedded Systems Resource Guide Networking: Technology the master has a slightly more relaxed is multiplied with the carrier created by creates a pulse that is modulated in time power budget. This asymmetry can be the oscillator, resulting in an up-converted according to Pulse-Position Modulation exploited by shifting as much complexity triangular pulse at the output. (PPM) and additionally shifted according as possible to the master device. In UWB to the chip value. communication, most of the complexity Signal calibration is in the receiver, allowing the realization Up-conversion of the signal to a desired of an ultra-low-power, lowest-complexity center frequency is necessary in order transmitter in the uplink (such as from to fit the spectrum inside the 3.1 to 10.6 sensor to master). And finally, the very GHz mask specified by the FCC. Besides little hardware complexity of a UWB the center frequency (Fc), the bandwidth transmitter offers the potential for low- also must be calibrated. Two calibration cost and highly integrated solutions. circuits are used to set these two For these reasons, UWB modulation quantities. Center frequency calibration is believed to have strong advantages is done using Phase-Locked Loop (PLL) compared to more traditional narrowband and should be carried out only once per radio communication. burst. The bandwidth can be controlled by calibration of the triangular waveform. It The CMOS 3-5 GHz UWB pulser is the slope of this waveform that finally Interuniversity Microelectronics Center defines the duration of the triangle. Figure 3 (IMEC) has realized an ultra-wideband pulser as the central component of a UWB Overall UWB transmitter architecture Besides the pulse generator and the two Measurement results transmitter. Two issues deserve particular The UWB pulser shown in Figure 3 calibration circuits, the overall UWB attention: has been realized in 0.18 µm CMOS transmitter architecture contains a system logic technology to reduce size and n The design and implementation of controller, a modem, and an antenna as cost. The resulting chip size is 0.6 by the pulse generator, being the most the main building blocks, illustrated in 0.6 mm2. The transmitter is flexible in critical component in terms of power Figure 2. The system controller defines both center frequency and bandwidth. consumption. the parameters of the communication It operates between 3 to 5 GHz and n The calibration of the center frequency according to a specified physical layer. generates pulses from 2 GHz (useful for and bandwidth of the pulses, to be It sets the Pseudo-Noise (PN) sequence positioning applications) to 500 MHz compliant with the FCC specification allotted to the user as well as the bandwidth bandwidth (allowing a multiple bands mask. and center frequency of the pulse. The solution), as pictured in Figure 4. For a predefined PN sequence provided by pulse bandwidth of 1 GHz, the measured Pulse generator the system controller is translated by the active power consumption of the system is Traditional UWB pulse generators use modem into triggering instants for the 2 mW for a pulse repetition rate of 40 MHz. very specific components such as step pulse generator. This chip sequence has This allows energy consumption as low recovery diodes, inductors or special a length equal to the number of pulses as 50 picoJoules per pulse for this antennas. These components, however, per bit. For each chip, the pulse generator are difficult to integrate. Complex repetition rate. packaging techniques add to the power consumption and overall cost of the pulser. An interesting alternative is to realize the short high-frequency UWB signals by gating an oscillator. The oscillator center frequency and the gate duration define, respectively, the signal center frequency and the bandwidth. To obtain smooth pulse shapes and reduce side-lobe power, the output amplitude is modulated by a triangular signal.

The pulse-generating sequence is as follows: a triangular pulse generator and a ring oscillator are activated simultaneously. The triangular waveform has a duration that can be adapted in accordance with the desired bandwidth. A gating circuit activates the ring oscillator when a pulse must be transmitted, avoiding useless power consumption between the pulses. The triangular signal Figure 2

Industrial Emdbedded Systems Resource Guide 2005 / 13 Networking: Technology were then used to calibrate the simulations. equivalent circuit in the case of UWB. To calculate the path loss, the distance They determined the front-end power around the perimeter of the human body overhead, which is independent of was measured using a real measurement the output power level. setup in the 3 to 6 GHz band. As expected, the path loss was found to increase with With these parameters, a model was distance, with a large fading variance. developed that allows comparing UWB The fading distribution around the mean and narrowband solutions in terms of could be modeled using a log-normal power consumption as depicted in Figure 5. distribution with a standard deviation Four low-power chips have been chosen between 5 and 8 dB, depending on the as narrowband competitors for IMEC’s position around the body. The path loss UWB solution, taken for their low-power Figure 4 was modeled with an empirical power capability. Evaluation of the UWB transmitter in a decay law. It turned out that, compared to As a result, three zones can be WBAN context free space, the path loss near the body in distinguished, depending on the channel To evaluate the UWB transmitter in a the GHz range is much higher. realistic WBAN scenario, a model had quality. For good channels, UWB performs to be developed that describes UWB UWB versus narrowband: overall power better due to limited front-end power propagation around the human body. This consumption consumption, such as for a communication body channel model allows comparing In a next step, the overall power between two body positions very close to the UWB solution with state-of-the-art consumption of both UWB and each other, UWB reduces the consumed narrowband implementations. narrowband implementations was power by two orders of magnitude from derived. First, it was calculated how much the best narrowband solutions. This The body channel model power needed to be transmitted through matches the expectations for UWB in low- A Finite-Difference Time-Domain the channel, taking into account noise power short-distance communications. (FDTD) simulator was used to model power, signal-to-noise ratio and channel For average channels, narrowband the UWB propagation around the body. path loss. The latter was calculated using solutions are better as soon as transmit The FDTD simulator shows that, after the BAN channel model. Further, two power dominates. For bad channels, transmission of a pulse, the wave is additional components contributed to the neither of the two implementations can diffracting around the human body rather consumed power: the transmitter antenna achieve reliable communication, thus than passing through it. Measurements efficiency and the power amplifier, or its requiring other solutions.

Figure 5

14 / 2005 Industrial Emdbedded Systems Resource Guide Networking: Technology Conclusion and outlook full UWB spectrum between 3 GHz and Bert Gyselinckx is The realization of wireless BANs relies on 10 GHz. In a future stage, the ultra-low Director, Human++ the availability of ultra-low-power radios. power transmitter will be integrated program at IMEC. By using a pulse-based UWB transmitter, into the sensor nodes of IMEC’s body Human++ is a cross- IMEC has been able to significantly area network. divisional initiative reduce power consumption for short-range aiming at developing communication. The flexible UWB pulser Bart Van Poucke is highly miniaturized has been completely integrated in 0.18 µm activity leader of the autonomous wireless CMOS technology to reduce both size and ultra low power radio nodes for healthcare cost. The operating frequency (between activity at IMEC. He and lifestyle applications. He received 2 and 5 GHz) and signal bandwidth obtained his electrical his M.S. degree in Electrical Engineering (from 2 GHz to 500 MHz) are compliant engineering degree from the Rijksuniversiteit Gent, Belgium, with FCC specifications. An energy at the KIHO, Gent, in 1992 and his M.S. degree in Air consumption as low as 50 picoJoules Belgium in 1996. and Space Electronics from the Ecole per pulse at 40 MHz pulse repetition After some years in the Nationale Superieure de l’Aeronautique rate has been obtained. To estimate the industry, he joined the DESICS division of et de l’Espace, Toulouse, France in 1993. overall power consumption in a realistic IMEC in 1999 to work on Medium Access WBAN scenario, a body channel model Control and cross layer optimization for To learn more, contact Bert at: for UWB propagation was developed. For reducing energy consumption. IMEC short-distance communication, the UWB Kapeldreef 75 To learn more, contact Bart at: transmitter outperforms state-of-the-art B-3001 Leuven IMEC narrowband implementations in terms of Belgium Kapeldreef 75 power consumption, proving the potential Tel: +32 16 28 15 37 B-3001 Leuven of UWB for short-range communication. E-mail: [email protected] Belgium Website: www.imec.be It is believed that implementation of the Tel: +32 16 28 83 51 pulser in a 90 nm CMOS technology will E-mail: [email protected] enable the transmitter to operate in the Website: www.imec.be

RSC #15 @ www.industrial-embedded.com/rsc Industrial Emdbedded Systems Resource Guide 2005 / 15 Networking: Standards

Opening gates with TCP-to-CANopen By Holger Zeltwanger

CAN is a popular industrial protocol, takes place on the (layer 2 manufacturer to trigger a Service Data and TCP is a widely accepted enterprise of the OSI reference model), a bridge is Object (SDO) service. This made protocol. An attractive system necessary. Such a device may translate machine builders dependent on the implementation for industrial control CAN messages into Ethernet messages. gateway’s manufacturer; they could not would bridge these two protocols. This However, for this the higher protocols just swap their gateway for that of another article describes the TCP-to-CANopen (meaning the application layer) must be manufacturer. This is why machine specification, a new initiative from CAN identical. This is not usually the case, and builders demanded to standardize the in Automation (CiA) to standardize thus a gateway is necessary. A gateway can communication protocols. gateway protocols. translate the services of one application layer into another application layer. It is TCP-to-CANopen gateway Machine builders are increasingly facing thus that a gateway connects a CANopen A cooperation between -IDA and the task of connecting a machine control network with EtherCAT, Modbus TCP or CiA resulted in a gateway specification, to the company automation network. . Theoretically, which enables developing and If the machine internal controller is based this works just fine. manufacturing standardized gateways. upon an embedded network, a gateway is This is advantageous since controllers necessary, which connects the company In practice this requires commands, which and configuration tools can access any network to the machine bus system. cause the gateway to initiate specific CANopen device via a set of defined The international user and manufacturer tasks on the “other side.” Example: A commands and responses. Additionally, group CiA therefore standardizes controller that is networked via TCP with CiA defined a protocol based on ASCII the communication services and a CANopen gateway requires information commands, which can access a CANopen protocols, which enables communication about a sensor on the CANopen network. device via any device from a TCP-network between heterogeneous and homogenous Until now machine builder had to use a via the standardized gateway. networks. command proprietary to the gateway Hierarchical networks The industrial automation industry has used several communication systems for years. On the lowest level, simple sensor/ actuator bus systems and backplane bus systems are being used (such as in modular I/O modules). On the intermediate level “embedded” networks control the machine: They connect programmable logic controllers, complex electric and hydraulic drives, I/O modules, and other microcontroller controlled devices including human machine interfaces and data acquisition devices. Factorywide automation networks are located on the top level; they are responsible for the communication between the machines.

Bus systems and networks are connected via devices that can interpret between the communication systems. They are called repeaters, if the networks are not logically divided. If communication Figure 1

16 / 2005 Industrial Emdbedded Systems Resource Guide Networking: Standards The gateway specification (CiA 309) supports among others the following CANopen services: n Access services for SDO: for example, read and write of object dictionary, entries in a CANopen device n Access services for Process Data Objects (PDO): for example, configuring of PDOS in the gateway n CANopen Network Management Services (NMT): for example, starting and stopping of a CANopen device n Device error management services: for example, reading a gateway error n CANopen interface configuration services: for example, initializing of the gateway n Gateway management services: for example, setting of the CANopen Node-ID of the gateway n Controller management services: for example, starting and stopping of programs in the gateway

All services are based on basic services (Figure 1). These include: n Request: requesting a service n Indication: Accepting a request Figure 3 n Response: responding to a request n Confirmation: confirming of a service Amazingly, in some applications CANopen CANopen device profiles. They are even is the superordinated communication compatible with the CANopen application A special command format was defined for system in the hierarchy of networks. layer to a far degree. ModbusTCP (Figure 2); the response also In these applications Ethernet-based provides a defined format. The protocols networks connect extremely fast drives, Using the same device profiles greatly on the CANopen network are equivalent which are subordinated to the CANopen simplifies the implementation of to those of the protocols specified in the network. There are no standardized gateways. In these cases, the gateway CANopen application layer (CiA 301/302 commands on the CANopen side at the does not have to reformat the application or EN 50325-4). moment for this scenario. However, such data. Reformatting can require a protocols would make sense for Ethernet lot ofprocessor resources especially in Powerlink and EtherCAT, since these bit-oriented data. But also for analog two communication systems support the values that are used in differing physical

Figure 2

An example of this is the NMT- ModbusTCP start command for a CANopen device in Figure 3. Figure 4 shows the common protocol structure of generic ASCII commands, and Figure 5 shows the ASCII command for the NMT start command. Figure 4

Industrial Emdbedded Systems Resource Guide 2005 / 17 Networking: Standards cascading, a CANopen (reverse) Address Resolution Protocols (CARP) must be defined still. This protocol will translate CANopen node-IDs of one network into unambiguous CANopen node-IDs of the other network. A similar protocol Figure 5 can also be used to connect Ethernet and CANopen networks. units, the microcontroller may have working on standardizing such multiple- to waste considerable amounts of level networks. Also homogeneous Holger Zeltwanger calculating power. CANopen-based network architectures finished his studies at are being considered. the Fachhochschule Nonhierarchical networks Braunschweig/ The future belongs to the nonhierarchical Already multiple homogeneous CANopen Wolfenbuettel (Germany) networks. Even if it still sounds utopian: networks are being realized with the aid in 1976 with a diploma in Master/Slave systems will be increasingly of application profiles. For example, electronic engineering. replaced by decentralized systems. with a CANopen application profile lift He worked at Siemens as a system However, such modern controllers controllers can be realized with eight programmer for two years, then was a require at least partially nonhierarchical lift units, each up to 254 levels. Since all technical editor for German and American network architecture. In nonhierarchical PDO communication is pre-defined, this magazines for more than 14 years. In 1992 networks the gateway is open to both is a logical network. This logical network he founded the international users’ and directions, meaning both sides can can be distributed into several physically manufacturers’ group CAN in Automation initiate communication functions on separated networks, which are connected (CiA). Since then he has worked as the respective other network. Such to each other via transparent gateways General Manager for CiA. nonhierarchical CAN-based networks are (Figure 6). already being used in automotive vehicles To learn more, contact Holger at: and other vehicles. Multiple CANopen networks can also be CAN in Automation implemented if generic device profiles Am Weichselgarten 26 Gateways may even implement multiple are used. In that case the gateways must DE-91058 Erlangen interfaces to be able to realize meshed be standardized though, since they are Germany network architectures. They are spread in not transparent. They must be able to Tel: +49-9131-690-86-0 applications throughout several networks, initiate the corresponding communication Fax: +49-9131-690-86-79 thus the bus systems or networks in functions on the second CANopen E-mail: [email protected] between must be able to “tunnel” or pass network via defined commands. Website: www.can-cia.org on the messages. The CIA group is already If the CANopen networks are to be

Figure 6

18 / 2005 Industrial Emdbedded Systems Resource Guide RSC #19 @ www.industrial-embedded.com/rsc Networking: Standards

Performance, implementation and applications of ETHERNET Powerlink By Frank Foerster and Bill Seitz

With Ethernet as the world standard for industrial automation. Ethernet is the EPL responds to the users and enterprise networks, industrial network world standard for IT and is rapidly machine OEM’s demands by featuring designers are looking to leverage its appearing in more industrial automation microsecond fast time-deterministic popularity while solving real-time control applications. cycle times that separate each cycle into issues. An emerging standard to address cyclic (isochronous) and non-cyclic the needs of automation and control Many automation users and machine (asynchronous) segments. While this is ETHERNET Powerlink (EPL). This OEMs are demanding faster, more mechanism responds to the microsecond article details the main features of the precise distributed control systems that performance demanded by users and EPL specification, details performance synchronize the ever-increasing number OEMs, it also addresses issues of domain benefits, and discusses an implementation of sensors and actuators on a single separation by isolating data from external of the protocol. network. There is compelling justification non-real-time domains, transferring for why Industrial Ethernet is attractive asynchronous data, from isochronous data By virtue of their adoption in for automation applications: at the machine level. This single feature microcontrollers and the number alone eliminates out of sync data from n Ethernet’s 100 Mbps data rate offers of installed nodes, two worldwide encroaching into the time-deterministic one to two orders of magnitude communications standards have emerged domain. during the past 15 years. As an embedded more bandwidth than most current communications standard, Controller Area field-buses, and with the emergence EPL operates on standard Ethernet Network (CAN or CANbus), originally of , even greater and features connectivity to common developed for distributive control in bandwidth is possible. IP network topologies via gateways as automobiles, is a commonly available, n Internet protocols provide integration illustrated in Figure 1. virtually free on-chip serial interface and data transparency on all offered by more than 25 semiconductor networking levels. Overview of EPL manufacturers. CAN is further supported n Ethernet is a proven cost-effective EPL is the first available real-time Ethernet by CANopen and DeviceNet as the worldwide standard. protocol. EPL, version 2 (EPL V2), extends most popular application protocols in n Silicon devices with onboard Ethernet the basic EPL communication mechanism transportation, medical, building, and are available from multiple vendors. of version 1 by adopting the well-proven

Figure 1

20 / 2005 Industrial Emdbedded Systems Resource Guide Networking: Standards CANopen object dictionary concept and its communications mechanisms. EPL V2 has new features like interoperability, flexibility, and configurability, which are essential to fast and accurate real-time communication. CANopen and EPL V2 Today, CANopen is used worldwide in thousands of applications and devices. This experience now directly benefits the further development of the EPL V2 stack. Working with the ETHERNET Powerlink Figure 3 Standardization Group (EPSG), IXXAT Access to and handling of the Ethernet In the event of access requests to these has created the EPL V2 specification with controller is encapsulated within a separate objects, user-specific call-back functions the focus on extensive integration of the module: the Lower Layer Driver (LLD). are linked to every application object that mechanisms developed for CANopen on To implement EPL V2 on a specific CPU enables an event-controlled notification of EPL V2. or hardware platform, the LLD is the only the application. This mechanism allows a EPL application structure EPL V2 module that must be modified. direct, application-specific reaction on EPL V2 can be implemented as a set of This abstraction between hardware- modifications to the application data Controlled Nodes (CN) and Managing dependent routines and the higher layers caused on the bus side. In addition, Nodes (MN) as illustrated in Figure 2. of the EPL V2 software ensures a high saving and recovery configured data The EPL V2 protocol software degree of scalability and adaptability. is supported. contains all necessary functions for full During development, a configuration file implementation. Linking of the time- in the module is used to optimize EPL V2 Operating system support The EPL V2 stack can be used with or deterministic EPL V2 network to a non- functionality to a given application and to without an operating system. The software real-time legacy Ethernet is achieved adapt the source code accordingly. has an internal scheduler that ensures the through a gateway. EPL V2 and legacy optimal allocation of the available process Ethernet cannot be commingled on a Process and service data objects Two basic mechanisms are provided by time resources to the various stack single Ethernet network. EPL for data communication: Cyclic, functions. By using an operating system, Encapsulation of real-time tasks, real-time critical data like I/O or closed the EPL V2 stack is executed as one task. hardware specific adaptation loop control data are transferred via Only basic operating system functions Referring to Figure 3, the lower layers Process Data Objects (PDOs) in the first like semaphores and tasks are required. of the EPL V2 software are specifically part of the EPL cycle while non-real-time These functions are encapsulated by the designed to follow the EPL cycle critical data like configuration parameters abstraction layer and are easily adapted to schedule and to guarantee fast response are transferred in the asynchronous the operation system. times for real-time events on the EPL period of an EPL cycle. In general, every bus. EPL V2 does not contain a TCP/IP CN participating in real-time processes Specifics of the EPL V2 protocol stack stack, which is required for running the is granted bus access in every cycle by IXXAT now offers the first complete EPL CANopen “SDO” protocol via UDP the MN to send its PDOs with cyclic V2 Protocol Stack according to the EPSG since this feature is typically provided by data. SDO data are transferred in the specification for CN and MN/CN. asynchronous period of an EPL cycle. the operating system. Alternatively, an The IXXAT EPL V2 protocol stacks are There is only one asynchronous transfer external commercial or open source TCP/ offered in a generic version for direct possible in one EPL cycle. IP stack can be used. porting to various target platforms and Object dictionary and programming operating systems as well as a complete interface solution, including a stack version The object dictionary, as conceived dedicated to IXXAT hardware reference in the CANopen specification, is the designs. interface between application and communication. Each object dictionary Provided as hardware-independent code, entry directly allocates a reference to the software can be used on various a variable containing application data. target systems based on different µC Communication services, such as PDOs platforms. Access to and handling of the and SDOs, access these application Ethernet controller are encapsulated by a variables directly. Therefore, only minor separate lower layer driver module with a modifications to an existing CANopen standardized interface. This provides users application are necessary to integrate an with the ability to implement the EPL V2 EPL V2 protocol stack. stack for specific Ethernet controllers and Figure 2 hardware designs.

Industrial Emdbedded Systems Resource Guide 2005 / 21 Networking: Standards The delivery includes the code to run the EPL V2 stack directly on a reference platform, which facilitates a quick start from scratch.

The EPL V2 stack has been developed in close cooperation with leading vendors of EPL-enabled devices like B&R Automation, Lenze, Hirschmann, and others who have collectively delivered more than 60,000 EPL nodes to date. EPL V2 performance The performance criteria associated with time-deterministic Ethernet are as follows: Figure 4 1. Overall jitter precision of the system Response times which makes such routines very specific. 2. Minimum cycle time Another important performance criterion The drawback of this approach is that 3. Response time of the nodes is the response time, which is, in the changing the hardware microcontroller Jitter precision of the EPL V2 cycle case of a CN, the time gap between the platform is problematic due to the Jitter is a term used to describe timing reception of an incoming request and uncertainty of performance levels that can deviation of cyclic events. For example, the transmission of the response. The be achieved on alternative systems. if an event should occur every 500 µs response time of an MN is the time gap An alternative to the microcontroller and it actually occurs after 498 µs in the between the reception of an incoming approach is an FPGA with an integrated best case and 504 µs in the worst case, response from CN 1 and the transmission hub/Ethernet Controller and a hardware the difference is referred to as jitter. In of the request to CN 2. The lower the state machine for package handling (both this example, the jitter is 6 µs. Network response times of a node, the higher the in VHDL), which makes the response jitter refers to the jitter caused by the bandwidth usage of the bus. Since they time much faster and more deterministic. network and its components, including have to handle the low-level package The hardware state machine is completely all connected devices. EPL V2 has a handling (reception/transmission) in independent of factors like interrupt maximum network jitter of less than 1 µs. software, typical microcontrollers are latencies or DMA speed. Providing Therefore, the timing fluctuation between able to achieve response times in the such IP in VHDL language reduces any of the connected devices is always range of 3-10 µs. Low response times on the dependencies on specific processor less than 1 µs. microcontrollers can typically be achieved with highly optimized assembler routines, platforms. The FPGA approach guarantees The jitter precision of EPL V2 is response times of less than 1 µs. solely determined by the MN, which is responsible for correctly sending a Start of Cycle (SoC) message. EPL V2 cycle time Most applications target a cycle time of 500 µs. In some applications, 2500 µs are required, but these generally have a reduced number of nodes. The EPL V2 cycle time is determined by the total of all communications on the bus, which is easy to determine by totaling all Request/ Response packets. Since every individual node has to perform specific tasks like SoC processing, PDO mapping, asynchronous SDO traffic and other application tasks, there are also node-internal processes like Network Management and the EPL Cycle State Machine that must be considered. When considering EPL V2, it is important to calculate cycle times that take into account all the message events on the EPL bus. Figure 5

22 / 2005 Industrial Emdbedded Systems Resource Guide Networking: Standards Miscellaneous performance factors Integrating a hub device into every node is recommended. Hubs introduce certain delays to messages sent on the EPL bus. Especially when building daisy chains, the hub delay could become a significant factor. Usually the hub delay is considered to be 1µs. Optimized hubs reduce the delay to approximately 400 ns. Implementation strategies IXXAT offers a flexible and generic approach for the implementation of EPL V2 nodes allowing the customer to select a microcontroller and architecture with respect to price, cycle time, space, and so on. The IXXAT EPL V2 stack can be implemented on virtually any microcontroller. However, performance factors like jitter, cycle time, and response time have to be evaluated individually. Implementation on a single microprocessor (CN) Figure 6 The implementation of the EPL V2 stack on a single microcontroller is easy to FPGA as communication coprocessor The low level routines stay on the NIOS achieve and has little or no impact on the (CN) II and the higher layer part (such as the hardware design. The important task is The implementation of EPL V2 function- object dictionary or PDO mapping) can the adaptation of the hardware abstraction ality on a FPGA is very powerful and be executed on the host processor. This (LLD) of the stack to the on-chip Ethernet extremely flexible. The FPGA contains an achieves shorter cycle times. The FPGA controller. In this configuration, a highly and an Ethernet controller/ contains the Ethernet controller and hub, optimized, specifically tailored hardware EPL packet handler. which guarantees ultra low response abstraction can provide a good result with times of less than 1 µs. Hub latency is respect to response times. approximately 400 ns. In addition, an Altera NIOS II processor The cycle time depends on the overall core is included in the FPGA. This Please note that this approach is not performance of the processor. According processor core provides the flexibility to really suitable for MNs for the following to benchmarks already made with the execute a scalable amount of software. For reasons: Freescale ColdFire MCF5235, a cycle slower CNs (those with slower I/O), the time of less than 250 µs (such as 1 Rx/Tx whole EPL stack plus application software n SoC is inaccurate since a PLC PDO) for this approach is achievable with can be implemented without a host CPU. application (executed on the host the IXXAT EPL software. For faster CNs, only the EPL stack (or a processor) and the EPL stack is part of it) has to be implemented. IXXAT essentially decoupled. This represents a straight Forward EPL is currently performing an implementation n Interface between application and V2 implementation on a microcontroller in which the customer requires a CN EPL stack is more complex and with an optional external hub with the capable of a 200 µs cycle time. In this unsuitable for handling a high number following features: application, a part of the EPL V2 stack is of different frames. executed on the FPGA NIOS II and the n The NIOS II is incapable of handling n Application code and EPL stack other part of the stack is executed on the all MN tasks and CN packets during executed on the same CPU. host processor. an EPL cycle. n Cycle times according to the performance of the controller and The most important advantage of this FPGA as hardware accelerator (MN/CN) the processing requirements of approach is that the host CPU is isolated When using the FPGA as a hardware the application (for example, EPL from EPL V2 critical real-time processes. accelerator as shown in Figure 7, the processing time on the MCF5235 was application and the EPL V2 stack less than 250 µs). Figure 6 illustrates the host CPU with are executed on the host processor. n Response times >3 µs (depending additional FPGA containing the NIOS II The approach makes use of the FPGA on the Ethernet controller and DMA core. The application code is executed on internal Ethernet controller with hub and architecture of the microcontroller). the host CPU and the EPL stack is executed the packet handler, which ensures ultra- n Hub device is optional. on the FPGA NIOS II. Alternatively, the low response times. Yet this keeps the EPL stack can be divided into two parts. interface between the application and

Industrial Emdbedded Systems Resource Guide 2005 / 23 Networking: Standards Because of its 10 years experience with CANopen, as well as other industrial communication networks, IXXAT was selected by the organizing companies of the EPSG consortium to develop the EPL V2 stack. IXXAT was chosen to ensure the full integration of the complete specification (application layer plus low level real-time aspects) in software. Today more than 10 companies are already using the IXXAT EPL stack. The majority of them are from the motor drive industry.

IXXAT invested a total of eight man- years of software development in EPL V2 and released the first fully featured stack in November 2004. Conclusion The IXXAT EPL V2 stack features all mandatory elements of the EPL Figure 7 V2 specification. IXXAT provides a flexible, scalable, open, and powerful the EPL V2 stack simple and makes the performance data provides a complete solution for EPL V2 systems today. It reaction time of the MN (in other words picture of all tasks that have to be handled provides customers with flexibility in PLC) very fast. by CNs and MNs. A processing time of choosing processor architectures and 280 µs (in the case of the MCF5235) for implementation strategies. The licensing Host CPU with additional FPGA: a complete cycle includes, among other model is a royalty free buyout license tasks, the SoC/SoA processing or the that is the desired model for almost n The application code and the EPL cycle state machine. Accordingly, the 250 all customers. stack are executed on the same CPU. µs cycle time for a CN on the MCF5235 n Cycle times are based on the includes the 40 µs processing of the one performance of the controller and pair of request/response plus 210 µs the processing requirements of overhead. the application (for example, EPL processing time for a CN on the Table 1 is a summary of the important MCF5235 is <250 µs). performance data that IXXAT’s n FPGA contains the Ethernet controller implementation of EPL offers. and the hub, which guarantees ultra- low response times of less than 1 µs. Application layer requirements n Hub latency is approximately 400 ns. When EPL V1 was introduced, attention Frank Foerster was focused on the performance: cycle Performance summary times, jitter precision, and node response. The whole EPL cycle can be thought of Today, the integration of CANopen as a series of events that are related to mechanisms into the EPL V2 specification the actual PDOs on the bus and to the has moved the focus to the application overall timing of an EPL V2 cycle. The layer.

Performance Criteria Time

Freescale ColdFire MCF5235 < 250 µs

Infineon TC1130 32-bit Industrial and Communication Controller < 200 µs Bill Seitz (estimated based on the benchmark test) To learn more, contact Frank or Bill at: NIOS II < 300 µs IXXAT, Inc. Fast NIOS II (estimation) < 200 µs 120 Bedford Center Rd., Suite 102

Processing time for a Request/Response Packet on a CN (MCF5235) 40 µs Bedford NH 03110-5442 Tel: 603-471-0800 Delay of the FPGA integrated hub approx. 400 ns E-mail: [email protected] Response time of FPGA integrated Ethernet Controller/Packet handler: < 1 µs or [email protected] Website: www.ixxat.com Table 1

24 / 2005 Industrial Emdbedded Systems Resource Guide RSC #25 @ www.industrial-embedded.com/rsc Networking: Technology

Building low power into wireless sensor networks using ZigBee technology By Jon Adams

A typical industrial network can have The ZigBee Alliance, an open industry system might be the optimal approach a large number of attached devices, group, identified a number of application from a communications performance presenting a challenge in power spaces, including industrial control, point of view. With a more holistic view, consumption. In this article, the benefits of , consumer electronics, the developer might discover that the ZigBee technology for industrial networks personal health care, PC and peripherals, incremental value that the customized are discussed along with how to address and residential/light commercial control as wireless system provides may be offset the power challenge. Particular focus is targets for ZigBee networking technology, unfavorably by cost, added project given to how to get the most life out of which was recently released to the public. complexity, radio testing and certification, battery powered ZigBee-enabled devices. ZigBee technology takes advantage of the and interoperability challenges. robust and reliable IEEE 802.15.4 low- The industrial, commercial, and residential rate wireless standard, and adds to this A standards-based approach can ensure sensor communities have wanted low mesh networking, sophisticated security interoperability between products, which power, cost-effective wireless sensors features, interoperability, and end-product can be very valuable in a sensor data ever since sensor technology was born. certification. The combination of the mining application as it can broaden the The ability to collect information in a IEEE standard bound with the ZigBee number of sensors that are available to simple yet robust fashion, while avoiding networking approach makes a significant provide data. From a customer’s point of the high cost and reliability issues that the paradigm shift in standardizing wireless view, this can mean OEM independence, traditional backhaul or homerun wiring sensor and control communications. Within which can often make investment in a creates, has the potential to completely the markets defined, the Alliance chose to technology more palatable to the customer change the metrics against which the cost concentrate initially on residential/light since they see that they have multiple, of data collection is measured. With the commercial lighting control and HVAC, competing sources for the technology, recent release of the ZigBee specification and industrial monitoring. providing them price, feature, and service to the public, it is now practical to build competition. With standardized products and deploy wireless sensing devices that A custom-designed proprietary wireless available from a multiplicity of OEMs, are at once capable of operational lifetimes system might at first seem to be the best increased product innovation is important measurable in months to years to decades approach to solving a wireless developer’s for the vendors to maintain differentiation from the batteries that power them as dilemma. With intimate knowledge of the and value in the eyes of their customers. well as meeting the cost, security, and sensor environment and its needs, a custom From the OEM point of view, standards- reliability requirements that the network developer expects. ZigBee technology and other wireless standards There is a wide, nearly unmanageable, range of choice when it comes to wireless technologies available for deployment of wireless sensing systems. There are both licensed and license-exempt technologies, standards-based and proprietary, on the market, all with their own potentials for addressing a developer’s product problem. Figure 1 shows a 5 order-of-magnitude data rate range and a variety of wireless technologies developed to address specific market needs.

Figure 1

26 / 2005 Industrial Emdbedded Systems Resource Guide Networking: Technology based approaches allow them source Then again, there are sensors and more It uses two forms of Phase-Shift Key independence and competition in pricing likely, control devices, that consume (PSK) modulation depending on the and availability, and distance them from significant amounts of power at all times frequency band – PSK has been used by radio function and interoperability issues or are permanently tied to the mains. These NASA for deep-space communications since these items are addressed by the considerations may prove important in the for decades because of its ability to silicon vendor. In the case of ZigBee choice of wireless technology to use. provide robust performance even with technology, there are already multiple very low Signal-to-Noise Ratios (SNR). silicon platform vendors out there with IEEE 802.15.4 primer In the past, PSK was infrequently used highly cost-optimized solutions, and the IEEE 802.15.4, a simple packet data in simple systems due to complexity and natural cost reduction through competition protocol, is now available as silicon-based cost, but the modern silicon systems have and volume production allows an OEM to products from a number of semiconductor resolved this to provide easy access to the benefit from the volume market generated companies, with more vendors on the approach. Figure 2 shows the difference in by their competitors. horizon every day. The quantity of performance between the ZigBee wireless platforms now available means that the technology, which is based on IEEE Picking the right wireless “cost-effective” mantra has become a 802.15.4, and other wireless protocols. depends on the sensor reality, and the breadth of OEMs taking It’s apparent that depending on the needs advantage of this straightforward radio Basic RF channel access is via Carrier of a sensor network, any of these wireless technology in a tremendous variety Sense Multiple Access including Collision methods (as well as other, less standardized of applications presages a long and Avoidance (CSMA-CA), which means ones) might fit the bill. Designers are successful life for products based on the “make sure the channel’s clear before you interested in wireless sensors that are not standard. IEEE 802.15.4 was designed for talk.” By doing this simple step, collisions just low power, but more importantly, sensor/control systems with specification between individual transmitters can be energy efficient, allowing them to operate of the RF channels, physical layer reduced and the overall channel capacity on battery for months to years to decades characteristics and behavior, and a simple improves. There’s also an optional TDMA or potentially through energy scavenging set of MAC primitives when compared operation for completely battery-operated techniques. to either Bluetooth or Wi-Fi technology. networks, which takes advantage of A fundamental advantage of using a regular timing beacons between nodes to There are a few other factors that standards-based technology is that it align communications opportunities. The sometimes need to be considered when is continuously vetted for quality and physical layer employs four simple packet defining the right kind of wireless for robustness, with the members of the IEEE types: a sensor. Number 1 is probably data focused on improving the technology and volume per unit time. Consider an image reducing the overall complexity and cost. n Data sensor that captures one compressed n Acknowledgement 100 Kb image every hour, and that frame In the 2400 MHz band, the standard n Beacon is required to be available at a remote specifies 16 channels, each 5 MHz wide. n MAC Command point within one minute after capture. This means that 100 Kb of information needs to be transferred from the sensor to a network within 60 seconds, about 1.7 Kbps. Add protocol overhead, other network usage, and the occasional retry and it sets the requirement for the overall network at probably something on the order of 3-5 Kbps. Not a tremendous data volume for most wireless systems. However, if that same sensor is capturing 30 frames per second continuously, that’s 3,000 Kbps of raw data, which is way out of the class of many low-rate protocols.

Some sensors are extremely power efficient, where the sensor function itself uses a very small amount of power (think of a simple photodetector like a phototransistor or magnetic reed switch security sensor), or is power-managed in such a way that even though the sensor during operation consumes significant power that the sensor function is shut down and cycled on an as-needed basis. Figure 2

Industrial Emdbedded Systems Resource Guide 2005 / 27 Networking: Technology It supports communications in multiple regionally unlicensed frequency bands, MC13213 of which only the 2400-2483 MHz band Background is available worldwide. Configured for Analog HCS08 CPU Receiver Debug Module maximum battery life, a device using IEEE 802.15.4 has the potential to last RFIC 8-ch Timers 10-bit as long as the shelf life of the batteries Tx/Rx Frequency Flash

ransceiver ADC powering it. Switch Generator Digital Memory Control 2 x SCSI

Digital T Logic In the MAC layer, the standard provides a Analog RAM IIC set of primitives that give strong flexibility Transmitter to the network developer, but it does not Low-Voltage 16-bit explicitly specify a network topology. Buffer RAM Interrupt Timers The standard uses IEEE 64-bit addresses IRQ RAM Keyboard Interrupt COP so that each radio is uniquely identified, Arbiter Arbiter Internal UP to 39 and to reduce overhead when a device Power Voltage Clock GPIO joins a network, this address is replaced Management Regulators Gen with a locally generated 16-bit address, which provides the capability for that local network to have more than 65,000 Figure 3 active nodes. AES 128-bit encryption is a features built in to both devices, there are A sensor/control network must have good built-in feature, absolutely important in a a number of power-down states available reliability, and that’s where ZigBee mesh wireless system. The Acknowledgement that allow the application developer to networking adds to the equation. ZigBee packet ensures that when device A sends trade functionality, latency, and average networks contain three kinds of networked a message to device B, device A will power consumption. device: the Coordinator, the Router, and keep trying until the data is transferred the End Device. Both Coordinators and successfully. So where is the ZigBee technology? Routers establish and maintain routing Real platforms and ZigBee Remember that ZigBee takes advantage of tables defining network connectivity; in wireless technology the IEEE standard’s functionality and adds addition, the Coordinator has the special Figure 3 shows a next-generation IEEE network function, security management, role of establishing the base network 802.15.4 silicon platform, this one from and application profiles. ZigBee in effect topology through assignment of the 16-bit Freescale Semiconductor. A single- sits atop the IEEE protocol, managing it network-specific addresses. End devices package system, the MC13213 (due as well as adding an applications interface do not handle traffic other than their own. later this year) contains all the required to which the developer writes their end In most ZigBee networks the Routers functionality to take full advantage of application. Figure 4 shows the conceptual and Coordinator are contained within ZigBee or 802.15.4 technology. On the left diagram of how ZigBee and IEEE products that are connected to the mains side of the block diagram are the 2.4 GHz technology complement one another. – devices like lighting load controllers, radio circuits, while the right side is the HVAC pumps and air handlers, smoke 8-bit MCU as well as various peripheral alarms and the like, as typical sensor devices that are available for application networks ultimately interface to control development. devices. This allows the End Devices to APPLICATIONS ZigBee spend most of their life asleep, waiting Within the are all the necessary for user input or preprogrammed action mechanisms to manage the RF channel APPLICATION FRAMEWORK like temperature, humidity, or occupancy as well as the protocol’s management of measurement. It’s straightforward to packets. This means that the MCU doesn’t NETWORK SECURITY design devices that minimize their average have to deal with RF channel timing issues FRAMEWORK ZigBee power consumption while remaining at the symbol or bit level, and is free to Alliance “connected” to the network. sleep or do other work while messages MAC LAYER Platform are being received or transmitted. In IEEE Figure 5 is a network diagram that shows addition, the transceiver contains timers PHY LAYER the initial tree routes (red), established and interrupt inputs that can further strictly due to the assigned address scheme, offload other protocol chores from the Application are quickly augmented by mesh routes MCU. The MCU contains functions ZigBee Platform Stack (blue) between the routing nodes. These that a sensor or control device needs, Silicon added routes mean that as propagation including ADCs, timers and I/O bits, and changes in the network due to events performs the high-level management of (doors open, furniture moved, people the RF data modem as well as hosting walk), or interference occurs (bursts from the sensor/control application. Because Figure 4 other radio systems, microwave ovens), of sophisticated power management there’s a secondary path to move the data.

28 / 2005 Industrial Emdbedded Systems Resource Guide Networking: Technology time period will vary depending on the need to transfer data, but for most simple sensors transferring battery voltage and sensor state (on or off for a switch, temperature or otherwise) fits easily into this payload.

Alkaline cells are commonly advertised with a seven-year shelf life nowadays – wouldn’t it be nice if the product came with a pair of batteries already installed and good for more than half a decade? The sensor that Freescale has designed using the IEEE 802.15.4 protocol can hit the kind of average power consumptions that make this practical.

All batteries have a self-discharge characteristic. This is due to internal finite high resistance paths between the battery anode and cathode that causes the cell to very gradually discharge. For a typical Figure 5 alkaline AA cell, that self-discharge current is between 10 and 20μA, meaning Controlling average power functionality does this device have and that every three weeks or so of shelf time consumption is the key how does it interact with the sensor and the cell loses approximately 0.05 percent For a simple security sensor as shown in the network? Figure 7 is a ladder timing of its total capacity. Thus, the battery the schematic of Figure 6, the average diagram that shows how an End Device will lose about 50 percent of its capacity power consumption can be made very communicates with the network, allowing in seven years even if it never leaves the low to allow exceedingly long battery the device to quickly wake up to an original blister pack. The goal for a long- life, if the situation calls for it. In addition event, make its observation, move the battery-life sensor is to have an average to the MCU and the RF data modem, data into the network, receive a network current no more than 10 to 20 percent of there’s a crystal for the RF oscillator and acknowledgement, then go back to sleep the self-discharge rate. about 10 capacitors and resistors needed as necessary. The steps in the process are to finish out the basic circuit. RF design specific to the IEEE 802.15.4 protocol Figure 8 shows the sensor’s battery life requirements are very minimal, as most requirements; there may be sensor when powered from those two AA alkaline of the complexity is contained completely processes that are not reflected here as cells. For a typical security system, it’s within the transceiver, removing the need every sensor is different. However, as long, good to know that the sensor is working for the designer to be an RF expert. What’s as the base timing rules are followed, it’s and has not been tampered with or removed not shown in the schematic is whatever practical to keep the transceiver powered – this sensor is probably expected to circuitry that the sensor might need, but down until necessary, or to shut off the report in on a regular basis to the network, in this example, the sensor is a simple MCU while the transceiver is interacting not just when an event occurs tripping magnetic door switch. with the network. The first network the sensor’s magnetic switch. However, message time, 650 µs, represents a this requires the sensor to go through the Now that the sensor has been connected message with 5 or 6 bytes of data payload, above process on a regular basis; turning and batteries installed, what kind of as does the third network message. This on a radio transmitter and receiver pulls a substantial amount of power over a very short time, but it can add up. The chart shows the check-in interval versus the expected battery life. For a check-in interval 18 seconds or longer, the average current is low enough that the sensor can outlast the batteries’ shelf life. Since many security systems expect a report interval between 15 seconds and 10 minutes, this design greatly exceeds the typical sensor requirement. It’s even practical to consider resizing the batteries to AAA and saving a little volume and mass, where the lifetime will still be longer than the user’s memory Figure 6 of when the batteries were fresh.

Industrial Emdbedded Systems Resource Guide 2005 / 29 Networking: Technology Low power, efficient radio technology Jon Adams is currently frequently on both. Prior to Freescale, Jon that is represented by the IEEE standard Director of Radio spent 17 years as a rocket scientist and was and ZigBee networking technology allow Technology for part of the team involved with developing for an entirely new class of sensors where t h e W i r e l e s s and deploying the Mars Network. During only simple measurement is necessary, Mobile Systems this time, Jon received several awards but until now, there was no cost-effective Group, Freescale including the NASA Achievement Award wireless method to move even that simple Semiconductor. Jon (for years 1986, 1992, and 2000) and the data with reliability and the knowledge has more than 20 JPL Award for Excellence. that the information will get to the network years of experience in due to the persistence of the sensor itself. the high technology industry. His current To learn more, contact Jon at: With these powerful tools, it’s practical focus is the chipset/platform for a new Freescale Semiconductor now to develop entire sensor systems that IEEE protocol called 802.15.4, and the 2100 E. Elliot Rd can last for a very long time on batteries network layer atop that standard called Tempe, AZ 85284 while always providing the reliability ZigBee. Jon is also an expert on wireless Tel: 480-413-3439 that the user and developer have come to communication technologies such as Ultra E-mail: [email protected] expect from wired sensor systems. Wideband and Bluetooth, and has spoken Website: www.freescale.com

Figure 7

ZigBee Security Sensor Battery Life 20 18

16 Any interval in excess of 18 seconds allows sensor to 14 surpass alkaline battery shelf life ears) 12 10 8 Typical Alkaline Shelf Life 6 4 Battery Life (Y 2 0 2 AA Alkaline Cells 5 10 15 20 25 30 35 40 45 50 55 60 Check-In Interval (sec)

Figure 8

30 / 2005 Industrial Emdbedded Systems Resource Guide Networking: Technology

Creating value with ZigBee Networks By Chris Herzog

ZigBee is a wireless networking standard ZigBee defines a set of higher level authenticity of transmitted data across that promises to reshape the way many protocols built on top of the IEEE the entire network. With security built companies create the infrastructures that 802.15.4 networking standard. It provides into ZigBee right from the onset, ZigBee their businesses and processes are built network routing and meshing capabilities, helps to ensure that data gets to its on. It will enable reliable, cost-effective, security, and support for both standardized intended destination and only its intended and low-power wireless monitoring and and proprietary devices through the use destination. control products based on open standards. of ZigBee profiles. Designed for use with This article provides an overview of small processors using battery power, Network topologies and capabilities ZigBee technology and discusses some ZigBee enables sensing and control ZigBee supports several network considerations for use and practical applications that until now were either not topologies including star, cluster, and mesh. applications, including topologies and practical or even possible. By supporting these different topologies, cost drivers. a ZigBee network can accommodate Using frequency ranges of 868 MHz, 902- many different application, installation, ZigBee has been developed by the 928 MHz, and the 2.4 GHz band, ZigBee and deployment circumstances. ZigBee Alliance, a group of companies operates at data rates of 20 Kbit/sec to working together to create a standard to 250 Kbit/sec depending on the band in Among the key attributes of a ZigBee enable wirelessly networked monitoring use. Ranges of 10-100 meters between network is the ability to add and remove and control products based on an open devices are obtainable, and when multiple devices on an ad-hoc basis and the ability standard. devices are joined into a meshed network, to have the network form a mesh of the range can be greatly extended through connections between nodes. The primary goals of ZigBee and the the relaying (or routing in network terms) Unlike networks designed for more static ZigBee Alliance are to address the market of packets from one node to another. need for and benefits of standards-based, configurations, ZigBee allows devices interoperable wireless products in areas ZigBee networks consist of a mixture to join a network very quickly (typically such as industrial process monitoring, of Full Function Devices (FFDs) and <30 ms) and supports very quick wake- home and building automation and Reduced Function Devices (RFDs). In up and data transmission (typically <15 control, consumer electronics, medical addition to sensing or control capabilities, ms). This permits new devices to rapidly monitoring, and similar areas. ZigBee FFDs can function as network routers join a network or quickly wake and send is designed to support these goals allowing them to form a continuous, data allowing battery-powered nodes to through simplicity, long battery life, multi-pathed, and highly reliable network. spend most of their time in a very low mesh networking capabilities, security, RFDs are designed to be simpler, less power sleep state and extend battery life reliability, and low implementation cost. expensive devices focusing on their to multiple years. control or sensing roles. Each RFD The ability to form a meshed network The ZigBee Alliance provides standards communicates to the broader ZigBee allows for multiple paths for data definition, interoperability testing, network by associating itself with an within the network and allows extended certification testing, and branding for FFD, which routes data on its behalf. ZigBee devices to ensure they meet all coverage far beyond the radio range applicable standards and interoperate as Security is an important aspect of ZigBee. of a single device. Multiple paths intended. Wireless networks offer flexibility, provide redundancy and reliability by but often at the cost of reduced data automatically dealing with temporary The building blocks security. ZigBee supports a variety of signal blockages, such as a vehicle The ZigBee specifications are designed security strategies ranging from the being parked in a signal path between to accommodate sensing and control IEEE 802.15.4 security for single hop two nodes, or equipment failures. When networks with a wide variety of devices, transmissions all the way to Advanced signal blockages clear, failed equipment a large number of devices (up to 65,000), Encryption Standard (AES). The AES is replaced, or new nodes are added to the low active duty cycles, small data packets, standard uses a core encryption algorithm network; the network can find these new and very long (multi-year) battery life. to support confidentiality, integrity, and potential routes automatically and use

Industrial Emdbedded Systems Resource Guide 2005 / 31 Networking: Technology them to benefit the overall network. When ZigBee sensor networks ZigBee profiles enable devices to find using a mesh topology, the total area of a A typical ZigBee sensing network and effectively communicate with other ZigBee network can be much larger than involves a set of ZigBee-enabled sensors network devices using standardized the range of a single RF signal. with some number of ZigBee routers and requests, and operations. The support a ZigBee Coordinator (typically a router of profiles is the cornerstone of ZigBee ZigBee supports other network topologies with Coordinator capabilities). In addition interoperability. as well. Installations based primarily on a to the actual sensor or controller interfaces, single, well located router can configure a sensor node requires a ZigBee-capable Using ZigBee effectively themselves as a star network using direct, radio and a small microcontroller as Deploying successful ZigBee networks point-to-point communication. Cluster illustrated in Figure 2. requires some understanding of the type topologies can support groups of devices of communications needed, the power while still permitting communication In some cases, a gateway device may be requirements of the devices, ranges and between devices in other clusters as deployed to support connections into and densities of the radio devices, and the shown in Figure 1. out of the ZigBee network to external costs of deployment and maintenance. networks or devices. ZigBee networks are formed around ZigBee application characteristics a single initial device, the ZigBee Profiles ZigBee is focused on applications with Coordinator. The Coordinator chooses Profiles provide a means to identify and particular characteristics. ZigBee is best a radio channel for the network and locate devices of particular types or that suited for applications requiring periodic controls the overall network topology provide particular services. For example, or intermittent data of relatively modest by specifying key networks parameters a light switch can locate a light load size. Typically, this is 40 bytes or so of affecting the way network formation will controller (or vice versa). Beyond simply data. While this is a relatively small, it occur. locating a device, ZigBee applications is usually more than enough for most can determine detailed characteristics sensing and control applications. Sensed Routers are responsible for the structure of devices or applications that enable a conditions like temperatures or pressures and operation of a ZigBee network. They fine degree of control and a high level of can typically be expressed in 1-4 bytes form the backbone of cluster or mesh useful interaction. and control directives are often similarly networks, support the establishment and sized. maintenance of the routes through the For example, a ZigBee light switch can network, and perform the actual routing not only find and associate with a ZigBee operations (the handing off of packets to lighting controller, it can determine specific neighbors who will then relay whether or not the controller is capable of the packets until they reach their final dimming its load or whether it supports destination). In the event of routing optional capabilities, such as sensing its failure, routers can discover alternate current lighting level or whether the room routes to keep traffic moving and can the lighting controller is located in is also discover more optimal routes as the occupied or not. network grows.

Figure 2

Since many ZigBee sensors are intended to be battery-powered, the use of compatible sensing and control technology is crucial. In these situations, sensors must be “battery friendly.” This includes operating on available battery voltages without excessive current draws that would impact battery life. Additionally, sensors should support a low power consumption mode or even completely shut down to save power when the battery-powered node goes to sleep. Sensors should be able to wake quickly without excessive power consumption or long delays that would impact battery life. Controller applications are subject to similar considerations. Figure 1

32 / 2005 Industrial Emdbedded Systems Resource Guide Networking: Technology One way to work around battery of the individual components but the total are eliminated. In addition, battery- limitations is to leverage existing power deployed cost. In the previous example, powered sensors operating on less from the machine or device being the cost of the sensor is more than doubled than 5V are intrinsically explosive- monitored. Many machines have internal when ZigBee-related components are safe, saving hundreds of dollars in power sources, which can power wireless added, but when compared with the total special sensing nodes and installation sensing nodes with minimal impact. The cost of a wired installation, a ZigBee costs (many ZigBee radios and low power requirement of the typical solution is actually a fraction of the processor combinations can operate ZigBee node provides flexibility in regard installed cost. to input voltage and current, which can n The ad-hoc capability of ZigBee can ease the integration into existing systems Any sensing or control network consists of be used for cost-effective monitoring and devices. more than just the individual sensing and of a fleet of commercial vehicles. By control nodes and must include network supporting ZigBee devices integrated Radio range and density infrastructure such as control panels, logic with a vehicle’s on board diagnostic With ranges of 10-100 meters, radio controllers, and other network support or sensing capabilities (such as ODB- placement must be taken into account devices. ZigBee networks require a II), a plethora of information can be when deploying ZigBee devices in Coordinator and routers in addition to the collected when a vehicle returns to order to create an effective and reliable sensing and control nodes. Depending on the maintenance or rental facility. This network. By locating battery-powered the size and configuration of the network, data can identify vehicles with specific sensors and mains-powered routers in a additional routers may be needed to problems or vehicles whose loads or way that multiple devices are within RF extend the reach of the network to all mileages indicate regular maintenance range, an effective mesh can be created devices. While these infrastructure costs is warranted. The rapid network joining offering more routes, higher reliability, cannot be eliminated, the incremental cost and ad-hoc networking capabilities of and redundancy. of adding nodes to a ZigBee network is ZigBee support collecting information only a fraction of the cost of adding nodes quickly from vehicles as they enter a Due to the ad hoc nature of ZigBee to a wired network. facility with no manual intervention or networks, additional routers can be easily configuration. added to an existing network without Installation costs are not the only shutting down or restarting to provide expenses, so it is important to consider For more information new network paths quickly and easily. A the total life cycle costs for the network. The ZigBee Alliance website, few strategically positioned routers can As an example, wired networks often www..org, provides information greatly improve the network infrastructure require physical reconfiguration, which regarding the ZigBee networks, the of a ZigBee network and can be easily can be a substantial cost, and networks ZigBee protocol, and suppliers of added at any time. of battery-powered sensors will require ZigBee-compatible components, systems, periodic battery replacement. Evaluating software, and engineering services. Costs for wired vs. wireless sensor technologies on both deployment and networks ongoing maintenance costs can help For more information on the IEEE In many locations, typical costs for the present a clearer picture of the overall 802.15.4 networking standard, see the professional installation for a wired cost of a potential solution. IEEE 802.15.4 website at www.ieee802. device including material (such as wire, org/15/pub/TG4.html. conduit, and fittings) and labor can be $10 Practical ZigBee applications per linear foot or higher, with most of that Practical ZigBee applications are those Chris Herzog is the cost in labor. If a $20 sensor requires 100 where there is a clear way to realize a President of Software feet of wiring to install, the effective cost return on the investment of wirelessly Technologies Group, a of deploying that sensor may be $1,020. enabling sensor or control networks, either software engineering Conditions such as wet or explosive through direct cost savings or by enabling services firm with a environments require special wiring solutions that were cost-prohibitive focus on 802.15.4 and components and installation expertise that without this technology. Oftentimes, ZigBee technologies, can dramatically increase costs. this comes from exploiting the unique including software capabilities of ZigBee networks, such as protocol stacks, development tools, While a ZigBee network may save low power requirements or support for ad prototypes, and product development. wiring costs, there is a cost for the radio, hoc network changes. processor, and other components necessary To learn more, contact Chris at: to wirelessly enable a device. The cost n Using the battery-powered capabilities Software Technologies Group for an 802.15.4 radio transceiver, 8-bit of ZigBee networks can simplify 10300 W. Roosevelt Road, Suite 310 microcontroller, support components, dealing with explosive environments, Westchester, IL 60154 battery, and ZigBee software can be less such as the production of ethanol Tel: 708-547-0110 than $25 per device in small quantities. or other fuels. With a wireless E-mail: [email protected] connection between a sensor and Website: www.stg.com When evaluating ZigBee as a potential its control logic, expensive and replacement for wired networks, it’s difficult-to-install explosive-safe important to account not only for the cost connectors, housings, and bulkheads

Industrial Emdbedded Systems Resource Guide 2005 / 33 Networking: Technology

Deploying ZigBee in existing industrial automation networks By Tim Cutler

ZigBee technology has industrial The foundation of automation networks: The cables themselves are unwelcome in networks as one of its primary targets. PLCs, sensors, actuators many factory environments, where they Implementation of ZigBee-enabled Regardless of specific application needs, can represent nuisances or safety hazards. devices is attractive, but in many cases industrial automation networks consist of And cabling runs have severe length coexistence with existing networks is PLCs (programmable logic controllers) limits, leaving networks unable to address required. This article looks at ways to that communicate with remote sensors very remote sensors and actuators in such incorporate ZigBee-enabled devices into to gather data regarding such variables as environments as oil fields and large-scale existing network topologies effectively, pressure, temperature, vibration, sound, plants. gaining the benefits while preserving and strain. When the application includes existing investments. a control function, PLCs act on that data Wireless networking has always by issuing commands that orchestrate been a reaction to these issues, with The ZigBee wireless standard is gaining processes among such actuators as relays, a primary requirement of preserving favor fast in industrial arenas, where motors, solenoids, and valves. An example communications reliability. Because it is prized for its low cost and its mesh is shown in Figure 1. factory and industrial settings pose special networking capabilities. The immediate challenges for wireless transmission, most challenge to realizing ZigBee’s many The success of this automation depends notably in interference and multipath benefits, as with all new technologies, on reliable communications between fading, industrial wireless solutions is implementation without extensive PLCs and sensors/actuators. The employ special techniques to compensate re-engineering. communication links have traditionally for noise and interference. One successful been hard-wired, with data transmitted method has been Frequency Hopping For many early adopters, the answer will over cable. Spread Spectrum (FHSS), which ensures come with gateway and sensor modem reliable data delivery in point-to-point techniques that marry ZigBee with In most cases, communications is and star topologies. Another is mesh existing system components. Products driven by the openly published Modbus networking, which uses automatic routing incorporating these techniques are already protocol, supported by the majority techniques to redirect transmissions that entering the market, providing immediate of PLCs, sensors, and actuators used fail in any transmission path. over-the-air mesh links between existing worldwide. The ubiquity of Modbus and PLCs and remote devices, and preserving the sensors and actuators that support it While these solutions are effective, they the usefulness of existing monitoring represent enormous investment, and an are generally found in pockets of need programs, even as they lay the groundwork entrenchment that makes inroads difficult rather than serving as entire network for embedded ZigBee systems of for alternative technologies. platforms. One reason for this is the the future. proprietary nature of available solutions. The basic need Unlike Modbus in the wired arena, the for wireless industrial wireless market has been D e s p i t e i t s characterized by proprietary solutions a d v a n t a g e s , that don’t interoperate with products from wired industrial competing companies (the 802.11b and networking has Bluetooth standards were created for very w e l l - k n o w n different applications, and have limited drawbacks. Data use in industrial settings). Another reason cabling represents is cost, with solutions of any technology a t r e m e n d o u s typically costing upwards of $100 per labor cost, in both node, sometimes much more. initial runs and in reconfiguration. For these reasons, wireless remains an alternative, rather than mainstream, transport mechanism, flourishing Figure 1

34 / 2005 Industrial Emdbedded Systems Resource Guide Networking: Technology primarily where safety and/or distance n Routers, which talk to each other, equipment have introduced devices that considerations make wiring highly to the coordinator and to end provide an over-the-air network interface undesirable or impractical. devices, and automatically redirect between existing PLCs and the existing communications when a link fails for sensors and actuators with which they Enter ZigBee: mesh networking for mass any reason. communicate, and preserve investments industry n Reduced function end devices, which in existing monitoring programs. These The ZigBee standard fills an interesting communicate with routers and the products fall in three basic categories, niche, specifying standards-based products coordinator, but not with each other; shown in Figure 3. that are inexpensive, consume little power, as they are not responsible for routing, and offer a form of mesh networking. they can sleep when not transmitting There is simply no competing standard, and may be battery powered. and ZigBee aims to fill that void while moving industrial wireless networking The automated, self-healing routing into mainstream connectivity. function of ZigBee is especially prized in industrial and factory environments. If the And it stands a very good chance of link between a router and an end device succeeding; its niche is more like a chasm. or the coordinator is disrupted, the router ZigBee’s extremely low cost should simply retransmits its data to another encourage adoption throughout industrial router with a clear path to the destination. Figure 3 settings, extending benefits to even the n Sensor modems ZigBee sensor modems provide a combination multichannel signal conditioner and ZigBee sensor module. Individual versions of ZigBee sensor modems can support various input types, such as 4-20 mA and full-wave bridge, and can provide 4-20 mA transmitters, digital I/O, and excitation voltages. n Modbus modems ZigBee Modbus modems connect to actuators, and combine the function of a serial output device with a ZigBee module. Modbus commands issued by the PLC are transported over the ZigBee network and relayed to the I/O module. n Modbus gateways ZigBee Modbus gateways provide the ZigBee network takeout point, mapping Figure 2 Modbus addresses to ZigBee addresses and formatting responses from ZigBee simplest sensors. Based in standards, it Joining this capability with very low cost remote devices into the Modbus protocol. will enable engineers to choose the best invites wireless industrial networking on These mappings prevent the Master PLC product for each node while ensuring a large scale; mesh networking addresses from having to be ZigBee-aware, and interoperability with similar products industrial challenges, and low cost radios preserve the functionality of existing from other manufacturers. Its low power extend connections to virtually any PLCs, sensors, actuators, and monitoring consumption allows battery power sensor. programs. The application functions as when mains power is impractical. And Preserving existing investments before, except that wireless links have it offers mesh networking, with devices replaced wired connections. automatically routing communications If ZigBee is to achieve significant market around barriers and doubling as repeaters penetration quickly, it’s not enough for Preserving the protocol for long-range links. it to provide connectivity, no matter The ZigBee Modbus gateway’s role is how advantageous. As many wireless made practical by functions that translate ZigBee delivers mesh networking through installations involve the replacement of Modbus functions to ZigBee and back, three device types, examples of which are wiring in an established environment, with assistance from inherent Modbus- shown in Figure 2. ZigBee cost-effectiveness requires that ZigBee similarities. it interface with existing equipment and n A coordinator, which organizes the support existing network protocols. To The mapping of Modbus commands to network and maintains routing this end, several early providers of ZigBee ZigBee addresses at the ZigBee Modbus tables.

Industrial Emdbedded Systems Resource Guide 2005 / 35 Networking: Technology

Table 1 gateway is straightforward. Modbus and other required values, plus a field for actuators, and PLCs with embedded ZigBee both assign a single address to number of registers to read for inputs or ZigBee radios, delivering cost-effective, each network device, which can be an for ON/OFF values for outputs. To read ready-to-use devices for new ZigBee individual sensor or actuator, or an I/ the analog-to-digital channel 0 input of a networks and simplified expansion of O block of sensors and actuators. The ZigBee sensor modem assigned Modbus existing ZigBee networks. Figure 4 shows Modbus gateway’s controller software address 4, for example, the following a future ZigBee device network with converts each address from Modbus Modbus command would be sent to the embedded radios. conventions to ZigBee conventions, ZigBee Modbus gateway: which is the only operation required. As this happens, ZigBee gateways will 04 04 00 00 00 01 still be key components, providing As Modbus reads and writes input and interfaces to get ZigBee data off the output values using a standard set of The following Modbus command would ZigBee network and onto wired networks registers, the use of standard Modbus be sent to turn digital output 1 ON at a that host applications that make use of the commands requires that ZigBee sensor ZigBee sensor modem: ZigBee data. However, the function of modem I/O be mapped to the standard these gateways will change as applications 11 05 00 01 FF 00 Modbus registers. The Modbus gateway become ZigBee aware, streamlining the translates the Modbus register specified in Using these simple techniques, ZigBee mapping process. the Modbus command to the appropriate devices can easily deliver the promise of Tim Cutler is I/O of the ZigBee sensor modem, with this new wireless networking standard vice president of mapping as shown in Table 1. while protecting investments in existing marketing and equipment. Communication is still All ZigBee sensor modem and Modbus O E M / i n d u s t r i a l Modbus-driven; the only difference is that gateway communication is expressed as sales for Cirronet. mesh wireless has replaced wires, while properly formatted Modbus commands, In addition to avoiding the re-engineering of existing including fields for Modbus address, extensive experience systems. function code and register address, and in executive sales Future ZigBee and marketing roles industrial and an MBA degree, Mr. Cutler holds a automation degree in electrical engineering from the embedded network Georgia Institute of Technology and two T h e Z i g B e e patents for microprocessor-based design. gateways, sensor To learn more, contact Tim at: m o d e m s , a n d Cirronet, Inc. Modbus modems 5375 Oakbrook Parkway discussed so far Norcross, GA 30071 introduce ZigBee Tel: 678-684-2000 benefits to existing E-mail: [email protected] n e t w o r k s . A s Website: www.cirronet.com ZigBee moves into the mainstream, the industry will shift Figure 4 toward sensors,

36 / 2005 Industrial Emdbedded Systems Resource Guide RSC #37 @ www.industrial-embedded.com/rsc Networking: Technology

Adding industrial Ethernet with minimal software changes by John Korsakas

Many industrial products were designed simply retransmitted on the opposite Implicit messaging often uses one of the without a growth path to Ethernet port. Other versions of the XPort support following methods: networking, but most included serial an industrial Ethernet protocol (such as interfaces. The benefits of migrating to EtherNet/IP and Modbus/TCP), allowing n Polling – a one-to-one periodic request industrial Ethernet interfaces in older and existing products supporting the Modbus and response between a client and a newer products are now compelling. This RTU/ASCII serial protocol to upgrade server article discusses the basics of Ethernet/ from an RS-232/485 serial port to a high n Cyclic – clients and slaves send their IP and how to leverage off-the-shelf performance 10/100 Mbps Ethernet port. data at regular intervals components to implement it. Overall, with minor hardware changes and n Change of state – data is sent only very few software changes, the existing when it changes Proliferation of Ethernet networking in product can be upgraded to support an n Strobe and multicast – a one-to- automation systems is on the rise, pushing industrial Ethernet protocol. many periodic data production, the vendors of industrial control products either through an acyclic request and such as barcode readers, pneumatic response or cyclic production of data valves, I/O, and so forth, to jump on the Other popular communication models are networking bandwagon. Client/Server and Producer/Consumer.

To stay competitive in today’s automation The Data Structure of a protocol provides market, vendors should now be adding a scheme for organizing, storing, and intelligent networking to their products. retrieving data within the device. One However, implementing support for specific example is the object model an industrial network protocol such as within CIP (Common Industrial EtherNet/IP could take three to nine Protocol), which is the application layer of months to adequately learn the details DeviceNet, ControlNet, and EtherNet/IP. and confidently write the code. An easier Data is organized into an Object/Instance/ way is needed. Figure 1 Attribute hierarchy. At the highest level are Objects, which group together similar Ethernet expertise not required Industrial network protocol basics types of data such as device identity to start Industrial protocols have a significant information, connection parameters, Enabling technologies for industrial number of differences from standard and application-specific data. An Object networks including software protocol enterprise networking protocols. can then contain one or more Instances, stacks or hardware components are Behind every protocol are three basic for example, a device with eight digital available today. Essentially, all of the components: outputs could contain eight instances of details of the network protocol are already the Digital Output Point Object. At the n Communication Model embedded in these components, and next level are the instance’s Attributes, n Data Structure one can act as a bridge for the data from which are the specific data values of n Services your existing product onto the industrial configuration parameters, application network. The Communication Model defines data, and so on. An example of a different data structure is contained within the One example is the XPort from Grid the way data is exchanged between two Modbus Application Protocol, which is Connect. It is a self-contained Serial- networked nodes. This includes what can used by Modbus RTU/ASCII over a serial to-Ethernet bridge with a 186-based be referred to as explicit messaging and line and Modbus/TCP over Ethernet. processor, RAM, and flash memory implicit messaging. Explicit messages Modbus data is organized into flat arrays packaged into an RJ-45 connector just are typically one-shot request/response of 16-bit Registers, 1-bit Inputs, and 1- about the size of your thumb. The standard messages used for setting configuration bit Coils (outputs). Each Register, Input, XPort supports basic serial tunneling, parameters, and implicit messages and Coil is assigned an address used to where data received on each port is typically transport the process I/O to and from the device. locate the data to be read and/or written.

38 / 2005 Industrial Emdbedded Systems Resource Guide Networking: Technology

The organization of data within CIP and previously shown, the data structure of specifying how often to read or write a Modbus is quite different, but each has its EtherNet/IP is object-based. The CIP particular map of the Modbus data as well own advantages and disadvantages. specification defines numerous standard as assigning relative priorities to each objects that help make CIP devices more mapping. The mapped data is buffered The final component, Services, compatible and interchangeable as well as in a data table within the XPort, which is encompasses the commands and ease the task of programming controllers also read and written by the EtherNet/IP functions that the devices support. Some when installing a CIP network. A very server that handles the communication common services are Read Data, Write common type of Modbus device is an with the EtherNet/IP client. Once the Data, Reset, Open Connection, and Close analog or digital I/O module. Since CIP configuration is complete and tested on Connection. These may sound simple defines standard Analog and Digital I/O one of the XPorts, a configuration file can enough conceptually, but each protocol Objects, a mapping should definitely link be downloaded from it and later uploaded implements many of them differently, the Modbus registers, inputs, and coils to the rest of the XPort-based versions largely due to its underlying Data to these standard objects. For other less of products during manufacturing. The Structure and Communication Models. common Modbus device types, Modbus configuration can also be protected to For example, to read a data item with data would have to be grouped and avoid a problem with customers changing CIP, a Get_Attribute_Single message mapped into a generic application object. the mappings to something that is not must specify the Object ID, Instance ID, compatible with your device. and Attribute ID to locate the desired Packets of EtherNet/IP Process (I/O) data data. The Get_Attribute_Single service are compiled through CIP assemblies, Conclusion can read any single attribute within any which are simply data arrays, formed Adding an industrial network to a product object, and some devices may additionally by “assembling” the data from the CIP can be a significant undertaking, but can implement the Get_Attribute_All service objects. Modbus inputs would get mapped be easier if enabling components are to read all the attributes of an object. CIP into CIP input assemblies and Modbus utilized. This allows an engineering team does not define a service to read only a outputs to CIP output assemblies. to quickly implement the networking subset of attributes. Within Modbus, one interface and spend more time focusing type of service to read data from a register The most common Communication on the application of the network in the specifies a base register address where to Model used in EtherNet/IP is a Producer/ product’s environment. Leveraging the begin reading and the number of registers Consumer model. Devices “produce” their industrial network development tools to read. Different, but analogous, services input data by sending input assemblies to available on the market and investing time are defined for Inputs and Coils. Even the one or many destination nodes at regular up front to build in-house knowledge and simplest functionality can vary greatly cyclic packet intervals. Output data expertise can save your company a great from protocol to protocol. received by devices at cyclic intervals deal of time and money in the long run. is “consumed” into the devices’ output Mapping serial data to Ethernet assemblies. The Communication Model of John Korsakas The conversion of serial data from an Modbus is a Polled or Request/Response has a BSE and existing device into Ethernet packets can method of exchanging data. A Modbus MSE in electrical be trivial depending on the compatibility client regularly sends requests containing engineering from and similarity of the protocols required on output data to a Modbus server, and the the University each port. To transparently tunnel the data server responds with its input data. Note o f M i ch i g a n between the serial port and the Ethernet that these two communication models are where he was the port, a standard XPort or similar product not exactly compatible: A Modbus server lab manager of could easily be used. However, this only sends input data when requested the DeviceNet, assumes that the other Ethernet device(s) while an EtherNet/IP server produces ControlNet, EtherNet/IP, and Modbus can parse and understand the format of input data at regular packet intervals. Conformance Test Laboratories for the data, which may not be the case. If the Therefore any bridge that converts the five years. During that time he also device’s serial port supports the Modbus data between Modbus and EtherNet/IP contributed toward several network, protocol and connection to a Modbus/TCP must account for this difference. communication, and testing standards for Ethernet network is required, luckily the ODVA, ISO, and SEMI. For the last two Modbus application layer is very portable The XPort’s EtherNet/IP-Modbus years John has worked at Grid Connect from its serial format to its Ethernet software allows the product engineer to developing industrial networking software format. A clear one-to-one mapping of map the Modbus registers of their existing for embedded systems. all the data can be quickly handled by an product into EtherNet/IP objects and I/O embedded technology such as an XPort assemblies. The configuration of the To learn more, contact John at: with Modbus Serial and TCP. XPort is done through the Ethernet port via telnet and/or web pages and is stored in Grid Connect, Inc. EtherNet/IP is one of the more complex nonvolatile flash memory. The serial port 1841 Centre Point Circle, Suite 143 industrial Ethernet protocols, and of the XPort acts as a Modbus client that Naperville, IL 60563 unfortunately it does not have a similar continually “scans” the Modbus server in Tel: 630-245-1445 x232 serial counterpart. In this example, the host device. This Modbus client can E-mail: [email protected] consider Modbus-Serial data. As also be tuned for better performance by Website: www.gridconnect.com

Industrial Emdbedded Systems Resource Guide 2005 / 39 Networking: Application

Whiteware industry drives Ethernet bus By Debra Biela

Implementing industrial networks can be What fueled the type of communication number of switches. “By using the a challenge as complexity increases, but network for Scott Technology’s system EISC16-100T, we had more ports per the benefits are often worth the effort. was their requirement to use a bus system switch,” says Baldock. Temperatures This article discusses a medium scale that would dominate the industry. Scott’s had to be between 0° to +60° C, similar application of Ethernet/IP in an appliance knowledge of Ethernet/IP supported this to industrial controllers like PLCs. The factory, which is featured in this month’s requirement, realizing this technology’s switches must operate on 24 VDC power cover photo. capabilities; from reducing downtime because the control system I/O voltage is to ease of use in commissioning. 24 VDC. Data rates must be 100 Mbps Few realize the complexity of “My electricians found it easy to because 100 Mbps is more efficient than manufacturing automated production terminate,” says Kilkelly. “We purchased 10 Mbps to get messages to and from systems for appliances, also known tools to do the job properly and a Fluke the I/O.” internationally as whiteware. Scott test instrument to test the cables.” Technology Limited in New Zealand Kilkelly notes that with Ethernet/IP there Baldock continued by saying the VLAN specializes in automated manufacturing is improved flexibility. “We can add or feature of the EISC16-100T minimized systems for the largest whiteware remove devices on the system quickly and the effects of multi-cast messages on all manufacturers. easily without using additional wiring.” ports. “By definition, VLAN allows for the physical network to be configured Contemporary Controls’ distributor, Deployment of the Ethernet as multiple VLANs. This limits the Mantis Systems Ltd. in New Zealand, communication network was ideal for this broadcast/multicast domains and supplied Ethernet switches to Scott application. By linking all components improves performance.” Technology to be used in the machine’s including an additional Ethernet-based form line (shown in Figure 1) for a Welding Technology Corporation The form line (shown in Figures 2 and 3), whiteware manufacturer. John Kilkelly, welding controller based on a Rockwell which is 25 meters long, is housed Scott Technology’s Senior Controls Automation SLC, Scott Technology inside the building. All electronic Engineer, says the form line decoils simplified the line’s cabling and improved components are mounted in control the sheet metal-forming, bending, diagnostic capability. panels adjacent to it. The 51 Ethernet pressing, and welding the metal into a devices are attached to CAT5 cabling, component used in washing machines and The 8-port EIS8-100T and 16-port with lengths up to 30 meters. The line refrigerators. configurable EISC16-100T switches uses a ControlLogix processor Controller from Contemporary Controls, marketed to form the end product and a Welding under the CTRLink trade Technology Corporation SLC Controller name, were employed in the for component welding functions. The system. The switches met the line’s backbone is the Allen-Bradley guidelines: “Fast installation, PLC chassis with individual slots for with no additional labor,” 1 processor, 4 EtherNet/IP modules, and says Kilkelly. “We panel 1 motion controller. mounted them using the mounting brackets so they would be easy to replace if failed. However, not a switch failed compared to others used from competitors, which kept losing messages.”

Tony Baldock, Mantis Systems Manager, says they were able to reduce the Figure 1 Figure 2

40 / 2005 Industrial Emdbedded Systems Resource Guide Networking: Application are linked via the EISC16- “With modern industrial manufacturing, 100T. Six laptop programming the operator expects to have instant results ports (positioned around the on his display terminal,” says Baldock. machine) are connected via “The fast I/O update times of today’s the EIS8-100T. Shielded RJ- Ethernet control systems mean there can 45 connectors on each end of be minimal delay from a change-of-state the cables connect the switches on the machine to providing the operator and other components. RJ-45 with that information.” Strong support for bulkhead receptacles (female) Ethernet from the industry secured Scott are mounted on operator panels, Technology’s decision to move ahead connecting the laptop to the with the system. control system for fault-finding, and so forth. Debra Biela is the Marketing The PC, located at the line’s Communications end, has RSView from Rockwell Specialist for Automation for data collection C o n t e m p o r a r y to determine how well a Controls. In her component is performing. For capacity, she helps example, batch data is given, to promote the such as number produced and company’s image as a global supplier of Figure 3 number scraped. Also, shift start industrial networking products. and end times are determined Each EtherNet/IP module connects to and a fault code will be logged for To learn more, contact Debra at: the EISC16-100T which attaches to the machine stoppage. Contemporary Controls, Inc. EIS8-100T. There are 10 switches on the 2431 Curtiss two lines. Seventeen I/O blocks connect The line is fully automated with safety- Downers Grove, IL 60515 to the EtherNet/IP module via one switched access doors. Operator Tel: 630-963-7070 ext. 105 EISC16-100T and one EIS8-100T. The panels are strategically located along E-mail: [email protected] 10 variable speed drives are attached via both sides to permit monitoring and Website: www.ccontrols.com the EISC16-100T. The welder controller control as well as manual intervention. and seven electronic operator terminals

RSC #41 @ www.industrial-embedded.com/rsc Industrial Emdbedded Systems Resource Guide 2005 / 41 Networking: Technology

Embracing WLANs in the industrial market By Mukesh Lulla and Ron Fredericks

Industrial networks differ in some key characteristics from their enterprise counterparts, particularly in the area of security. This article discusses the trends toward use of wireless local area networks (WLANs) in an industrial context. It gives a perspective on some of the issues surrounding industrial protocols and focuses on new specifications related to IEEE 802.11 to improve security.

The realization of wireless networking as a mainstream technology is now clear as a result of a myriad of products available today that satisfy the demand driven by consumer and office applications. However, the adoption of this technology by industrial device manufacturers has been limited thanks to some significant challenges and security concerns involved in deploying the technology in industrial environments such as the plant floor.

In spite of this, there is no getting around the fact that significant interest in the industrial use of wireless technologies continues to be driven by its staggering advantages over cabled systems in harsh industrial environments and its use in difficult-to-wire equipment, such as those with moving parts or distributed over large areas. This, coupled with the fact that yesterday’s proprietary, closed control systems are giving way to open network architectures such as Ethernet, we see a WLAN future hurtling toward the Figure 1 industrial world at an incredible speed. technologies geared toward local area The Wi-Fi Alliance industry group A quick WLAN primer network (LAN) use, the most predominant further evangelized 802.11 networking Before exploring the applicability of WLAN technology is 802.11 networking by helping promote interoperability wireless local area networks (WLANs) — named after the IEEE 802.11 standards and certification procedures to the industrial domain, let’s start by committee that defined the standards for required of 802.11 devices to be Wi-Fi introducing some of the standards that it. The IEEE 802.11 specifications define compliant. Unlike some of the earlier are relevant to this technology. While a wireless LAN that operates in the 2.4- wireless LAN technologies that sputtered wireless networks can include wide area 5 GHz range and currently provides data before gaining momentum, adoption of cellular networks (including the newer transmission rates of 1 Mbps – 54 Mbps, 802.11/Wi-Fi standards for home and 3G data/voice variants) and the varied using various modulation techniques. enterprise networks has proceeded at a

42 / 2005 Industrial Emdbedded Systems Resource Guide Networking: Technology breakneck pace. With the tremendous networks for performing batch and Trend toward standards popularity of this technology, cost of process monitoring/control typically in The trend toward open standards for Wi-Fi components has also come down areas of real-time control, which can be industrial networks has been underway for drastically while quality, reliability, and vital to a plant’s operation. Systems on a many years now. Ethernet has emerged security have improved as the alphabet control network may also include SCADA as the de facto hardware standard and the soup of 802.11 standards have evolved. A (System Control and Data Acquisition, adoption of TCP/IP for use in industrial natural outcome of this has been that these Supervisory Control and Data Acquisition, networks has also recently gained technologies are now ripe for uses beyond Security, Control and Data Acquisition) ground. Examples include the standard home/office network setup. and M2M (Man to Machine) interfaces. now offering ProfiNet and Modbus now In some cases, they may include many offering Modbus TCP. Control networks Typical industrial network topology of the same computer systems found in now use real-time extensions unique to To recognize the role of traditional the office and embedded devices such as a particular standard over Ethernet as industrial networks, it is helpful to redundant Ethernet switches, real-time a common hardware transport. Upper visualize the various computer and monitoring systems with special control layers of the network stack typically device systems used in an industrial or panels, and display devices, although the use the ubiquitous TCP/IP or a more manufacturing facility as being divided applications are different from those in an recent real-time control standard called into three types of networks as shown in office network. EtherNet/IP, which uses UDP packets in Figure 1. addition to TCP/IP to add improved real- Device networks are the most diverse and time control. Ethernet is thus emerging as Office networks in an industrial facility the most automated of the three networks, a clear and pervasive standard for control mimic that of any typical office, and include different specialized networks, making it convenient to connect supporting common functions such as embedded and electrical systems such as to office networks as well for improved database services, desktop computing, programmable logic controllers (PLCs), cost efficiencies. e-mail, intranet, Internet, and other -enabled equipment for measuring communication/data services. However, flows, temperatures, vibrations, liquid In device networks, there still exist the office network in this case is also and gas volumes, sensor panels, motion competing technologies including some connected to the rest of the plant for control, pressure, fluid, gas, and heat of those used in control networks. Table 1 such purposes as sales, inventory, and controllers. Several of the common open lists common fieldbus technologies and maintenance operations. fieldbus standards, such as PROFIBUS, networks they are used in, each with Modbus, DeviceNet, and CANopen have its unique cabling requirements and Operations personnel such as plant traditionally been used in industrial device competing open standards or organizations operation engineers, production networks. representing the standards. A significant managers, and batch operators use control Used In Plant Standards Body Protocol(s) Office Devices Control Open Device Vendors EtherNet/IP CIP Ethernet, No Yes Yes Association – ODVA DeviceNET No No Yes Industrial Ethernet EtherNet/IP CIP Ethernet No Yes Yes Association Industrial Automation Open Networking Association EtherNet/IP CIP Ethernet No Yes Yes – IAONA Modbus-IDA Modbus No No Yes

PROFIBUS fieldbus, PROFINET No No Yes PROFIBUS International industrial Ethernet No Yes No

Fieldbus Foundation FOUNDATION fieldbus No No Yes

OPC Foundation - Object linking and embedding for OPC standards and specifications No Yes Yes Process Control

CAN in Automation (CiA) CAN, CANopen, CAN DeviceNet No No Yes

Institute of Electrical and TCP/UPD/IP Ethernet, Yes, Yes, Yes, Electronics Engineers – IEEE 802.11 Yes Yes Yes Table 1

Industrial Emdbedded Systems Resource Guide 2005 / 43 Networking: Technology issue in industrial networks used to be Furthermore, the controlled equipment on patch-management as a security the connectivity between one fieldbus may also be installed near wet, moist, or policy, as enterprise networks have grown network standard to the next and then corrosive materials that could damage used to. The ability to support different back to the control network, requiring physical wires and plugs. generations of 802.11 security and the use of converters and adapters, which networking standards concurrently is a big Ethernet has served to alleviate to a large On device networks, too, WLANs present plus to engender graceful obsolescence extent as well. several unique advantages beyond those and incremental upgrades. in office networks. Remote device Why industrial WLANs? maintenance and monitoring becomes The wireless security problem While not without its practical difficulties, quite convenient if a device is networked While initial concerns about the use of the migration to Ethernet and TCP/IP for wirelessly. This is especially true for WLANs may have been grounded in fact, industrial protocols makes the transition devices that are not easily reached because recent advances in the areas of security to WLANs an almost obvious next step of their physical location in the plant have addressed the most insidious of since it complements the use of TCP/IP. (such as dangerous or harsh operating the problems holding back acceptance Furthermore, 802.11/Wi-Fi WLANs are environments, or restricted areas in the of WLANs in industrial systems. New based on similar principles as Ethernet – in plant). In addition to maintenance, basic standards address many link security fact, they were for the longest time referred wireless connectivity of devices located concerns while still maintaining and, to as Wireless Ethernet. The typical Wi- on moving parts such as in a wind tunnel, in fact, enhancing the mobility and Fi server (an access point or wireless robotic, or motor-controlled system is untethered aspects of a wireless LAN router) can offer a wired (Ethernet) point optimal compared to physically wiring a that are key to its acceptance in industrial of connection to an existing network system that can be problematic, especially networks. and a radio antenna or air interface to in corrosive or harsh environments. In bring wireless devices to bear on the other words, WLANs can be used to The inherent link security problem in a factory’s control or device network using untether difficult connections and allow WLAN stems from its very strength — its radio waves. for measurements at the source without wirelessness. Since wireless technologies passing through various intermediate make use of a shared medium, in There are many good reasons to consider connector devices. Last, but not the least, other words, airwaves, with no well- WLANs for the control network on the WLANs make it easier to come up with defined physical boundary that can be factory floor. In addition to the basic proof-of-concept networked systems in an protected, the link is not as secure against mobility aspects of WLANs that are industrial environment without expensive eavesdropping as a wired one, and it is such an asset to office networks, the prototyping. also difficult to restrict network access control floor also has some unique safety to a set of authorized users. The security issues that WLANs can help address. Moving Wi-Fi to the industrial mechanism proposed in the initial 802.11 For example, WLANs may be used to environment standards was, in a moment of misplaced seek out current operational data and So, with all the advantages that wireless bluster, named Wired Equivalent Privacy manage a controlled shutdown procedure technologies bring to an industrial (WEP) but it turned out to be flawed. remotely and allow the proliferation network, is now the time to embrace Newer industry standards such as of control points such as safety shutoff Wi-Fi WLANs? Challenges specific Wireless Protected Access (WPA) provide valves and other mechanisms much more to the use of WLANs in industrial important fixes for issues that plagued the conveniently than a wired network. applications may go beyond those of legacy WEP and at the same time maintain adapting WLANs to the enterprise. legacy compatibility with WEP encryption For example, extensions for real-time hardware. WPA2 and the recently ratified response on industrial Ethernet networks IEEE 802.11i standard add further have to be in place before adopting Wi-Fi. security enhancements to WPA, including Furthermore, radio interference caused by use of the Advanced Encryption Standard or affecting Wi-Fi networks are important (AES) block cipher in the form of Counter factors to be considered, especially in Mode CBC-MAC Protocol (CCMP) and electrically noisy industrial environments stronger integrity check mechanisms. since noise can degrade the performance Table 2 outlines the significant differences and predictability of a Wi-Fi network that between various generations of Wi-Fi relies on unregulated radio frequency security technologies. bands for its transport. Another key factor to consider is the fast-moving world of In the presence of strong link security, Wi-Fi standards, which, for all its evolving privacy and integrity of network data benefits, may involve constantly changing are maintained, but there still exists the technologies that are anathema to robust, possibility of other attacks that may minimal downtime industrial networks allow an intruder to act as a valid client that favor consistent functionality over connecting to the network, or as an evil frequent upgrade cycles. Lastly, the twin access point masquerading as a valid security of such networks needs to be access point to hijack client credentials (or RSC #44 industrial-strength and should not rely worse). Other means of defending against @ www.industrial-embedded.com/rsc

44 / 2005 Industrial Emdbedded Systems Resource Guide Networking: Technology and more reliable. Driven by its growing Features WEP WPA WPA 2 acceptance, industrial networks that once were segmented into distinct proprietary Cipher RC-4 RC-4 AES pieces can now be collapsed cost- effectively into one without significant 128 bits encryption intrusion on the current physical Key Size 40 bits 64 bits 128 bits infrastructure. Added intelligence can now authentication be stuffed into wireless-enabled devices with secure embedded middleware that Initialization Vector 24-bit 48-bit 48-bit will lead to efficiencies of cost and scale never before thought possible.

Per-Packet Key Concatenated Mixing Function Not Needed Mukesh Lulla is the president and co- Payload Integrity founder of TeamF1, CRC-32 MIChael CCM Check Inc., an embedded software company Header Integrity None MIChael CCM s p e c i a l i z i n g i n Check security software for wireless applications. Replay Attack None IV Sequence IV Sequence TeamF1’s portfolio of wireless products Prevention range from an out-of-the-box secure AP EAP-based middleware package (Air Secure Access EAP-based Key Management Group-shared Or Point) for use in industrial embedded Or Pre-shared Pre-shared applications that require AP functionality, to turnkey packages for wireless switches. Table 2 To learn more, contact Mukesh at: attacks such as MAC filtering, which In many applications, strong user or TeamF1, Inc. inventories valid client MAC addresses device authentication requirements 39159 Paseo Padre Parkway #121 that can connect to the WLAN and denies may necessitate the use of independent Fremont, CA 94538 access to all others, and access control/ authentication mechanisms such as Tel: 510-505-9931 blocking of network nodes are required Kerberos or embedded digital certificates E-mail: [email protected] to reduce the chances of unintentional for cryptographically secure access Website: www.TeamF1.com or casual access to the wireless network. mechanisms. Mechanisms that monitor for rogue access The new factory floor Ron Fredericks is the points are also an integral part of industrial president of Embedded WLAN security. Over the past decade, WLANs have revolutionized networks in the enterprise Components, Inc., Sunnyvale, California. Finally, secure management of devices and at home. Industrial WLANs can Ron is the architect of that form the industrial WLAN and help take this further and allow for a web and consulting LAN infrastructure is a frequently integrated information flow, optimize service offering called overlooked item that can turn out to be production processes, increase automation Cycles of Efficiency the Achilles heel of an industrial network. capabilities, and reduce hazardous (CoE) that helps engineering teams Management capabilities allowing for conditions. Network enabling process and device manufacturers seek out and remotely controlling the configuration, control equipment or devices does not integrate embedded components and tools upgrades, and provisioning of the network need to be an expensive, time-consuming more effectively from their vendors. infrastructure devices in all three types of effort any longer. With concerns about industrial networks need to be upgraded wireless technologies out of the way, To learn more, contact Ron at: to secure versions of protocols such as: factories can lower their production Embedded Components, Inc. costs by freeing themselves from wiring 233-i East Red Oak Drive n SSH (a secure replacement for constraints and taking advantage of Sunnyvale, CA 94086 Telnet) full electrical isolation, convenient Tel: 408-390-1895 n HTTPS (HTTP over secure sockets maintenance, and reduced installation E-mail: using SSL/TLS) by leveraging a cable-free environment. [email protected] n SNMPv3 (a secure version of New mobile device policies can now be Website: SNMP) implemented for maintenance workers www.EmbeddedComponents.com n SFTP/SCP (a secure replacement for and emergency crews that improve plant FTP) to ensure security in a wireless operational safety and costs. The use of environment where perimeter Wi-Fi in the plant does not have to stop at firewalls alone cannot be relied upon making existing networks safer, cheaper, to create private networks.

Industrial Emdbedded Systems Resource Guide 2005 / 45 Networking Networking Wired to Device: CANopen Wired to Device: LONworks™ Networking

PEAK PCAN Products Connector C2k FEATURES: FEATURES: n Various low-cost CAN interfaces (LPT Dongle, USB, LIN) n Multi-protocol networking across multiple media and devices n Enable easy connectivity from embedded target to host-PCs n LONworks™ TP and PLC, ModBus, RS232, RS485, pulse, mesh radio n CAN configuration software: PCAN-Developer, PCAN-Explorer n WAN connections - Ethernet, dial-up, cellular, or broadband n PCANopen Magic configuration tool n Secure XML socket interface with open API and complete SDK n More info: www.phytec.com/peak.htm n Massively scalable Network Operating System network manager

PHYTEC America LLC Broadband Energy Networks Inc. 203 Parfitt Way SW, Suite G100 7000 Terminal Square Bainbridge Island, WA 98110 Upper Darby, PA 19082 Tel: 206-780-9047 • Fax: 206-780-9135 Tel: 610-734-1245 • Fax: 610-734-1263 www.phytec.com www.broadbandenergynetworks.com RSC #4601 @ www.industrial-embedded.com/catalogrsc RSC #4602 @ www.industrial-embedded.com/catalogrsc

Wired to Enterprise: Industrial Ethernet

AVT-418 Interface AVT-418 is an Ethernet interface to in-vehicle computer networks supporting: Controller Area Network (CAN); 3 CAN physical layers: CAN-C, CAN-B, and Single Wire CAN (SWC); LIN as master or slave; J1850 VPW (GM, Chrysler) with high speed and block transfer; J1850 PWM (Ford SCP); K-line communications: ISO9141 and ; and Ford UBP. Supports other proto- cols as well. AVT-418 communicates with the host computer via an Ethernet connection for fast, reliable communications. Product FEATURES: documentation, technical documentation, and a price list are all n All protocols supported; no additional hardware to buy or install available on our website at www.AVT-HQ.com. Call or write for n prompt support. Low cost and readily available n Full support for ISO 15765 CAN messaging; also known as Segmented or Multi-Frame messages n Multiple units can be connected to a single host computer; can be remotely located Advanced Vehicle Technologies, Inc. n Free software from our website to quickly get up and running; easy to connect and easy to use 1509 Manor View Road n Prompt technical support available at [email protected] Davidsonville, MD 21035 Tel: 410-798-4038 • Fax: 410-798-4308 www.AVT-HQ.com For more information, contact: [email protected].

RSC #46023 @ www.industrial-embedded.com/catalogrsc

46 / 2005 Industrial Emdbedded Systems Resource Guide Wired to Enterprise: Industrial Ethernet

Networking OEM Solutions Ethernet/IP, Modbus TCP, ProfiNet, DeviceNet, Profibus - choose the level of integration for your design and cost reduce it as your volumes go up. Grid Connect provides you product, module, or embedded chip designs with complete proven royalty-free soft- ware stacks. Our designs are used in OEM designs that require performance and low cost. Our chips can be used as co-proces- sors or as main processors with your application software added. We provide a complete integrated development kit with example FEATURES: designs and software. To save time and focus on your applica- n Ethernet/IP, Modbus TCP, ProfiNet, Ethernet TCP, e-mail, Web server tion, call Grid Connect for your industrial networking hardware Ethernet protocols and software. n DeviceNet, CANopen, CAN protocols, Profibus, Modbus, and many other serial protocols n Complete RTOS and TCP/IP stack included with development kit, C compiler/linker, JTAG debugger support; kits are low cost Grid Connect n EX SOC chip has 30 MIP processor, 256K SRAM, 2 10/100M Ethernet channels, USB controller, 4 serial ports, SPI, I2C, 2 CAN ports 1841 Centre Point Circle, Suite 143 n XPort Module has EX SOC chip, serial to Ethernet hardware inside Naperville, IL 60563 J45 connector; the XPort can be loaded with protocols Tel: 630-245-1445 • Fax: 630-245-1717 n NET 232, NET485, WI232 are complete products with various www.gridconnect.com software protocols available depending on your requirements For more information, contact: [email protected]

RSC #4701 @ www.industrial-embedded.com/catalogrsc

Networking Wired to Enterprise: Ethernet/IP

PC104-PLUS MAC HUB FEATURES: n Stackable PC/104-Plus Card n 100 BaseT Ethernet MAC with a 4 port switch n All ports auto MDX n Switch support full duplex wire speed connections n Price of the 4I66 in 100's is $133

Mesa Electronics 4175 Lakeside Drive, Suite 100 Richmond, CA 94806 Tel: 510-223-9272 • Fax: 510-223-9585 www.mesanet.com RSC #4702 @ www.industrial-embedded.com/catalogrsc

Industrial Emdbedded Systems Resource Guide 2005 / 47 Computing: Technology Modern interfaces in light of embedded computer integration By Andreas Geh

In the embedded computer level on which they operate: interfaces on components will still be available in the industry, both manufacturers the chip, board, peripheral, and network future, whether the new standards are and users are facing the challenge levels (xAN). For system developers, the embedded-capable and how they can be of selecting the right solution from a chip level is no longer critical because checked during the development process. formidable multitude of bus, interface, and module concepts are widely used in this communication standards. Furthermore, area. Consequently, the following aspects Important new serial standards on board this process becomes even more difficult will concentrate on the interfaces that are level: in times of major technological changes, primarily relevant for design-in (board n PCI moving to PCI Express (PCIe) as is the case with the current transition and peripherals). n ISA to Low Pin-Count Bus (LPC) from parallel to serial standards. n TTL to Digital Video Output (DVO) This article first describes the most In recent years, the computer industry has been transitioning from standards for important interfaces that play a role And on the peripheral level: in most x86-based embedded systems. parallel interfaces to standards for serial After that, it highlights the advantages interfaces. This change has far-reaching n IDE P-ATA to Serial Advanced and disadvantages of current and future consequences for the embedded computer Technology Attachment (SATA) standards for embedded computer industry, which must now determine n LPT, Floppy, PS/2 to Universal applications and points out special how long it will continue to support Serial Bus (USB) design-in considerations. older standards and which peripheral n PC Card to ExpressCard Embedded computers can only fulfill their purpose if the interfaces that they Type of Number of Number of Type Speed Max. length have to provide are efficiently configured. transmission conductors subscribers Making the right decision is a matter of trade-offs between size requirements and flexibility, power-saving capabilities and (16-bit/8 MHz) 16 MBps Parallel performance, long-term availability and ISA (theoretical) Single ended 87 30 cm 5 - 10 future compatibility, as well as marketing ~ 4 - 8 MBps 3.3 V (5 V) considerations and genuine needs. The (actual) embedded computer industry is currently experiencing a radical change that is Serial LPC ~ 4 MBps Single ended 6 - 13 30 cm 5 - 10 creating new perspectives: Consumer- 3.3 V oriented trends have to be tested for suitability for industrial applications. (32-bit/33 MHz) Serial interfaces are replacing Parallel 4 - 8 133 MBps PCI Single ended 70 30 cm (No. of parallel interfaces, and some finished (theoretical) 3.3 V (5 V) interrupts) developments first have to prove their ~70 MBps (actual) usefulness in embedded environments. Decision makers for new projects and 32-bit/2 (4/8) x Parallel products have to do more than just AGP (2/4/8) 66 MHz 1 Single ended examine the processor components, 533 – 3132 MBps which are frequently overemphasized in marketing. They must also be familiar with a multitude of interface standards and then PCIe 250 MBps Serial 4 30 cm 1 carefully select and scale them. Since it is (1 Lane) bidirectional Differential not possible to comprehensively examine the huge number of interfaces, one can categorize the interfaces according to the Table 1

48 / 2005 Industrial Emdbedded Systems Resource Guide Computing: Technology From the angle of embedded computer conductor lengths, as well as the number During the design-in for an embedded solutions, the new standards offer only of subscribers, addresses, and interrupts. system, it is a good idea to use module limited suitability for use. Furthermore, As far as integration is concerned, concepts that employ several bus standards they present system developers with availability of components, suitability and/or scalability between parallel and new challenges. for use under harsh environmental serial buses. LPC and PCI should be conditions, susceptibility to EMC used for the conventional peripheral The assessment of an interface for problems, connector and cable definitions, components and when ISA is required, embedded computers in industrial backwards compatibility, costs of the connection should be made using environments depends on several integration and parts, as well as presence conventional bridge chips or FPGAs. essential characteristics. On the physical of software support are among the crucial Using PCIe is only to be recommended side of the evaluation are such criteria as criteria. Tables 1 and 2 show the most when a high-performance graphics cards data rate (transfer volume in megabytes important interfaces at the board level for is to be used or when other high-speed per second), number of conductors, type x86 embedded computer systems. processes (such as Gigabit, Firewire, and of transmission, maximum and minimum

Type Area of application Connector / cable Advantages Disadvantages Availability

• Simple peripherals • Software support • PC/104; DIMM • Bandwidth Decreasing number • I/O functions can be integrated ISA • Simple plug connectors, • Outdated power of components for • Small bus systems • Proprietary etc. consumption proprietary solutions • Proprietary controllers solutions

• Primarily for legacy support Software support for (SuperI/O controller) Reduced LPC No standardization LPC to ISA compatibility High • BIOS extensions conductors problems • µC interface

• PC/104-Plus • Peripherals • CompactPCI • Bus systems • Wide buses • MiniPCI • Widely in use PCI • LAN / framegrabbers / • Few subscribers in High • PCI slots • Software support graphics controllers / PC parallel • PISA Card, etc. • Proprietary connectors

AGP Only for graphics cards Special AGP slots No other applications High for graphics (2/4/8)

• Fast peripherals PCIe • Increasing • Gbit Ethernet • High bandwidth Stringent development PCIe slot • Little chip support • ATA150 • Few conductors requirements (1 Lane) so far • High-end graphics

Table 2

Industrial Emdbedded Systems Resource Guide 2005 / 49 Computing: Technology graphics) are to run simultaneously in for selecting a video interface depend on (bandwidth) are simultaneously on the addition to Gigabit Ethernet or when an the chipset, BIOS or display, dynamic rise. The short life cycles for LCDs are ExpressCard is planned as a removable or fixed detection, or on physical factors also extremely problematic. media. such as the conductor length, resolution, and timing. For external monitors, CRT (VGA) The peripheral level can be divided into solutions were the standard choice until three areas: The conditions for the display control for now, and TTL or LVDS solutions were modern x86-based systems are changing. used for LCDs. Integration required n User interface The embedded graphics controllers that checking the monitor data, adapting n Storage media have been in use until now are being the video BIOS (timing) and making a n Devices phased out of the market, and new chipsets suitable adapter cable. The advantages and graphic controllers no longer support were the low hardware expense and User interface: video the TTL interface. Adaptations to the trouble-free adaptation of exotic displays. Every modern is BIOS are becoming increasingly complex The disadvantages were the high equipped with fast components in order and more difficult for manufacturers to expense for integrating the software, to master the data transmission tasks. One produce. Moreover, the drivers in the the permanent choice of a display type, category that takes on special significance operating system are also becoming and the problems associated with this in in this respect is the video interface with increasingly complex. Modern displays conjunction with product discontinuation its different standards. Video interfaces are are equipped with automatic detection or with adjustments to the BIOS. closely connected to the application and mechanisms such as DDC/EDID, and the meet different applications’ requirements. requirements for graphic performance In the future, DVI-I will be the standard As shown in Table 3, the decisive factors solution used with external monitors. With LCDs, it will be LVDS or DVI. The examination of the monitor data, the generation of an Extended Display Area of Connector / Type Advantages Disadvantages Availability Identification Data (EDID), and the application cable integration of DDC functionality (I2C bus) will be required. The new technology

• LCDs offers several advantages: It makes the • Outdated Modern with low • Simple plug solution independent of the display • Short graphics resolution connectors, conductor controllers hardware, enables automatic recognition (

• Noise continue to use the analog CRT • DVI • Complex resistant connection. Control of connector display • Defined DVI modern LCD • Shielded control High 2. For systems with external displays, standard monitors cable • Large • High DVI should generally be used. • Up to 12 m connector bandwidth 3. For the use of integrated LCDs in the system, LVDS lends itself well as the interface combined with the integration of DDC with an EDID on

• Simple plug the baseboard or in the display. Direct connectors Direct connection • Twisted pair Coding User interface: input and audio LVDS control of High of LCDs in • Flat ribbon varies widely LCDs Since many systems have a Super I/O chip, systems cable • Up to 8 m which is usually required by the BIOS, the PS/2-interface can be used inexpensively for input/output. When designing new systems with a user interface, USB Table 3

50 / 2005 Industrial Emdbedded Systems Resource Guide Computing: Technology should generally be preferred because Packet Interface (ATAPI) to the ATA it makes sense to use scalable modules this alternative is increasingly replacing standard for connecting CD/DVD drives such as DIGITAL-LOGIC’s SM855 and PS/2. For audio applications, the AC97 using EIDE. Table 4 contains a summary SM915 modules. bus from Intel will remain a frequently of ATA standards. used and simple possibility for employing Removable storage media: codecs for diverse requirements. When comparing ATA and SATA (see PC-Card vs. ExpressCard Table 5), it is important to remember ExpressCard technology is the name given Storage interfaces that transfer rates higher than 33.3 MBps to an add-in card that was developed by The storage interface is normally used for (UDMA2) are rarely implemented with the Personal Computer Memory Card three types of storage: embedded systems, because higher International Association (PCMCIA) and speeds (> ATA-4 or UDMA2) require use will be replacing the PC Card over the next 1. The system/ data memory for the of the special 80-conductor cables. These few years. This plug-in card, which is only operating system and operating data cables are not suitable for embedded half the size of a conventional PC Card, 2. The main memory (not examined applications. uses the scalable, high-bandwidth serial here) because its control is usually PCI Express and USB 2.0 interfaces. The integrated in the chipset When developing embedded systems, it USB 2.0 interface is primarily suited for 3. Removable storage media, such as a is still a good idea to use EIDE/ATA as a technologies such as Bluetooth cards or PC Card, ExpressCard, or USB-based standard interface, because it is anticipated Flash memory cards. The high-speed PCI media that hard drives will be available for a Express variant is predestined for devices long time. Furthermore, the 44-pin EIDE with high transfer rates, such as 1394b- or The EIDE/ATA interface is widely used connectors and cables are very well-suited Gigabit Ethernet cards. As already defined as a solution for mass storage devices. for use in harsh environments (shock/ with the PC Card, the ExpressCard is Enhanced Intelligent Drive Electronics vibration). SATA is only suitable for equipped with hot-plug functionality, (EIDE) defines the interface’s electrical systems with external drives or high data which can be used for the PCI Express properties, and Advanced Technology transmission rates. Nevertheless, using a and the USB 2.0 interface. In addition, a Attachment (ATA) defines the protocol. lockable SATA connector is advisable. In sophisticated power management system In 1998, ANSI added AT Attachment order to be able to support both standards, enables the plug-in card to make efficient use of the limited energy resources of a ATA-1 ATA-2 ATA-3 ATA-4 ATA-5 ATA-6 corresponding energy-dependent device. PIO modes 0, 1, 2 3, 4 — — — — added The fundamental advantages that DMA modes 0, 1, 2 Multiword 1, 2 — — — — added Multiword 0 ExpressCard technology has to offer are UltraDMA — — — 0, 1, 2 3, 4 5 high transfer rates, universal use, and modes favorable form factor. Another plus point Max. transfer 11.1 MBps 16.6 MBps 16.6 MBps 33.3 MBps 66.6 MBps 100 MBps rate for the ExpressCard standard is owed to the 40 40 40 40/ 80 80 80 low cost due to the low number of circuit Cable conductors conductors conductors conductors conductors conductors paths. On the other hand, this technology ANSI NCITS NCITS NCITS standard, X3.221-1994 X3.279-1996 X3.298-1997 requires a new connection architecture 317-1998 340-2000 347-2001 year on the system boards and new plug-in Block SMART, CRC, Features transfers, cards such as the ExpressCard/34 and the — reliability 80-conductor — 48-bit LBA added LBA, drive features cable ExpressCard/54. Although applications identification Ultra for embedded systems are currently still Ultra DMA/66, Known as ATA/IDE ATA/IDE ATA/IDE Ultra DMA/33 DMA/100, ATA/66 lacking, ExpressCard technology offers ATA/100 interesting perspectives in the midterm Table 4 compared to the PC Card (shown in Table 6), which is widely used but has significant EIDE/ATA SATA restrictions in embedded environments. 33 MBps (UDMA 2) 150 MBps Speed Up to 100 MBps (UDMA5) 300 MBps (SATA II) When developing embedded systems, it is 44-pin flat ribbon cable (0 – 40 cm) better to do without the PC Card if possible Cable 80-pin flat ribbon cable (0 – 20 cm) Special SATA cable, 7-pin (up to 1 m) beginning with UDMA3 (66 MBps) and instead use USB 2.0 intensively for removable storage media and as a simple, Signals Parallel; single ended; 5 V Serial; differential; 0.5 V high-speed interface (see Figure 1). • Widely used • Thinner cable • Connector suitable for • Noise resistant Advantages embedded use • High bandwidth The ExpressCard lends itself well when • Flexible cable lengths • Completely compatible to ATA interchangeable high-performance • CF-connection possible • Beginning with SATA II: hot plug peripheral cards become important in the • No connector suitable for embedded use system. It is also possible to use scalable • High number of pins • Long standard cable (>0.45 m) Disadvantages • EMC issues • Only one device / channel modules that support both the USB 2.0 • Still no CD/DVD drives and the PCIe standards without passing on existing solutions, as is the case with Table 5

Industrial Emdbedded Systems Resource Guide 2005 / 51 Computing: Technology

Figure 1

the SM855 and SM915 smartModules mounting for DRAM (for automotive and uses the i915GM chipset that is integrated from DIGITAL-LOGIC. avionics applications). into the video controller and supports up to 256 MB of video memory as well as Implementation example: smartModule The new SM915 smartModule shown DirectX9 compatibility. The video The smartModule is a small Pentium in Figure 2 is based on the Intel 915GM controller uses two 18-bit LCD M module with ideal scalability for all chipset and uses a Pentium M processor connections to control the most important interfaces. This module solution with clock speeds of 0.6 to 2.2 GHz as its common LCD technologies. It reaches allows scalability despite the switch CPU. Equipped with a large, 2 MB L2 on a resolution of up to 2048 x 1536 of technology from parallel to serial chip cache, the 2.2 GHz Pentium M765 pixels with a maximum color depth of interfaces. Furthermore, it fits in between achieves performance levels comparable 256 bits. It is also equipped with two PC/104-Plus connectors. For customers, to a 4.4 GHz Pentium 4 processor. The PS/2 ports for a mouse and keyboard, a these two aspects protect investments over maximum main memory is 1 GB of serial interface, a floppy interface, eight the long term despite the technological DDR-RAM. The CPU and memory are USB-2.0 interfaces, an AC97/HDA- transition and ensure the realization mounted interchangeably, but they are compatible sound interface with DTS- of expandable system solutions of mechanically protected against vibrations 7.1 output, a watchdog, an RTC battery, minimal size. The smartModules feature or shock. and EEPROM support. For connecting a mature, integrated cooling concept additional mass storage media, there are a Another of the smartModule’s strengths and are configured with a completely P-ATA133 and two SATA150 hard drive is its graphics performance. The SM915 encapsulated design including mechanical interfaces available. Furthermore, all required power supplies and progressive PC-Card ExpressCard Speedstep power management functions

132 MBps in theory have been integrated. For connecting to Speed 300 MBps 50 - 70 MBps in reality a TCP/IP-based LAN or to the Internet, the smartModule is equipped with a Memory cards Memory cards 10/100Base-T Ethernet controller. The Peripheral cards Applications Peripheral cards (Gbit-LAN, WLAN, Firewire, module can be expanded via the PCI or (LAN, WLAN, Firewire, field field buses) buses) PCI-Express bus.

Connectors 68-pin (type I, II, III) 26-pin (type 34, 54) The SM855 and SM915 require a supply voltage of 5 V (expandable to modern point-of-lead concepts, 4.5 – 30VDC) 32-bit (PCI), single ended, Signals PCIe 1x + USB 2.0 and run on all conventional operating 3.3 V systems. Equipped with a flash BIOS • Small form factor (34 mm) that has a dual-BIOS option makes it • Noise resistant possible to boot from different media Advantages Widely used • High bandwidth • Two standards in one (USB) such as the hard drive, a floppy disk, • Hot plug USB, or via LAN. The modules operate within a standard temperature range from • Limited suitability for • PCIe required -20 °C to +60 °C. At low clock rates (600 Disadvantages embedded use • Few cards available • EMC issues • Consumer oriented MHz) they are also approved for use in the extended temperature range from Table 6 -40 °C to +85 °C. The smartModules feature

52 / 2005 Industrial Emdbedded Systems Resource Guide Computing: Technology Andreas Geh is key account manager for DIGITAL-LOGIC. Prior to his current position, Andreas founded his own engineering office for industrial electronics concentrating on embedded computer technologies. He instituted the concept of including embedded computer, embedded Linux, and embedded communication as basic technologies for systems. For more than 12 years, Andreas has garnered considerable experience and knowledge in development, system design, and project management for embedded computer projects, mainly in the areas of medicine, telecommunication, and industrial automation. His studies focused on electrical communications technology. Figure 2 To learn more, contract Andreas at: a sophisticated cooling concept: The cover the side of a device’s enclosure or to a DIGITAL-LOGIC AG consists of a milled aluminum block with cooling element. The cover simultaneously Nordstrasse 11/F high resistance and a fixed supporting provides EMC protection. Since all of CH - 4542 Luterbach surface on the PCB and is mounted on the the i855GME/i915GM chipset’s power- Switzerland CPU on a special copper core to ensure saving functions are supported, the Tel: +41-32 681-58-00 ideal heat dissipation. This perfectly flat smartModules typically only consume Fax: +41-32 681-58-01 surface enables an ideal connection to 10 watts of power. E-mail: [email protected] Website: www.digitallogic.com

www.ascii-to-plc.com

RSC #5301 @ www.industrial-embedded.com/rsc RSC #5302 @ www.industrial-embedded.com/rsc Computing: Technology

Embedded compute modules help contain costs By Ernest Godsey

The ability to deploy embedded compute for embedded applications and their the very real possibility of disrupting modules in challenging environmental demanding environmental conditions. the operation of the control system and conditions is especially critical for many Different manufacturers have taken jeopardizing freight or passenger train embedded industrial systems that find different approaches to SOMs, but one service. In addition to these challenges, their way into applications where air of the major differences between SOMs the Le Shuttle train traverses under the conditioning and other amenities are and SBCs is the self-sufficiency of SOMs. water of the English Channel through the often limited or nonexistent. This article As a complete compute engine, most Channel Tunnel or Chunnel. Moreover, examines several applications of System- SOMs can function standalone or they the developers of the Le Shuttle embedded On-Module (SOM) designs that save costs can be mounted on a carrier card where control application faced technical while meeting this challenge. additional application specific I/O can challenges raised by the need for multiple be installed. Moreover, SBCs are almost field buses to control the various aspects It’s a familiar refrain for designers of always very bus-specific. For example, of the Le Shuttle’s cars. embedded industrial systems. The unique VMEbus or CompactPCI SBCs requires requirements of the application practically a full VMEbus or CompactPCI system, Of course, as is typical of many embedded dictate custom hardware to get the job complete with card cage, backplane and projects, the timeframe for delivery of done, but the design team’s management power supply. In contrast, a SOM can be a working system was very short, only thinks the cost of the solution ought to placed on a carrier card, which may only six months. This obviously precluded be in the same ballpark as commodity contain a power supply connection. Or, the possibility of any significant custom desktop PCs. This is a huge problem for the carrier might include a bus interface. development work. As a result, the project designers because industrial systems Thus, a particular SOM can be used design team at the MORS company of are almost never manufactured in high without a bus or with an interface to Marseilles, France, turned to MEN Mikro volume the way consumer products are. practically any bus. for an off-the-shelf compute module As a result, development and design costs and mezzanine cards for the multiple cannot be made up over a high-volume Besides the cost savings that the flexible field buses. manufacturing run. architecture of SOMs provides, they also save development time since they can be The Le Shuttle embedded control system Because the process of developing quickly adapted to special requirements. was designed to collect, analyze and application-specific hardware is both In addition, many SOMs operate over the process all safety data for each car that expensive and fraught with design risks, full industrial temperature range. And the makes up the nine Le Shuttle trains. The both of which can jeopardize the success small factor as well as sturdy construction safety systems in Le Shuttle trains monitor of the overall project, smart designers give SOMs greater stability to withstand and control door closing mechanisms, the oftentimes turn to off-the-shelf compute the shock and vibration of industrial train’s lighting and heating systems, alarm modules and standard platforms. This settings. Some SOMs are available with management routines and other safety- minimizes design risks and design time, a conformal coating to protect them related functionality. Each of the trains while keeping their options open in terms against harsh chemicals and airborne is approximately 2,400 feet long and can of system architecture and packaging. contaminants. contain as many as 28 passenger cars and two engines. Embedding a system-on-a-module Underground, underwater Historically, designers of industrial Embedded control systems for The basic topology of each train’s control embedded systems have turned to open transportation applications can be very system featured local control intelligence bus-based Single Board Computers challenging for a variety of reasons, distributed throughout the train, a network (SBCs) and I/O cards mounted in a primarily environmental and design of multiple field buses that connect all backplane chassis. In newer configurations flexibility. Embedded computers in of the train’s local control units and a SBCs have been equipped with a one or railroad applications, for example, are centralized processing unit to manage more mezzanine cards for I/O. Recently, subject to shock and vibration that can the overall operations of the system at the small form factor system-on-module shake loose components from circuit level of the entire train. To make the train- has emerged as an alternative solution boards or cables from connectors, with wide control system more manageable,

54 / 2005 Industrial Emdbedded Systems Resource Guide Computing: Technology

Tailoring SOM solutions Two German companies recently discovered the value of implementing SOMs in control systems for textile processing equipment that is deployed in every corner of the globe.

Zinser Textilmaschinen GmbH and its sister company, Schlafhorst Winding Systems GmbH, are both part of the Swiss Saurer Textile Group. Although the product design groups for the two companies collaborate closely in order to reduce costs, their machines perform different functions in textile mills. Zinser’s spinning equipment is used early in the process to produce yarn from raw materials. Schlafhorst’s winding systems add strength to the yarn by combining strands and subsequently wrapping the yarn around bobbins so it can be used later to produce cloth. Figure 1 Figure 1 depicts how an Embedded System Module (ESM) commands into the control system. However, the Zinser from MEN Mikro GmbH, parent company of MEN Micro, machines needed higher-resolution graphics at 16 bits per Inc. of the U.S., was used as the heart of the intelligent pixel, while the Schlafhorst systems required only eight Man-Machine Interface (MMI) for both the Zinser spinning bits/pixel. By implementing the graphics controllers for each machines and the Schlafhorst winding systems. Because of type of system in the EMS’s FPGA, the different graphic the versatility of the embedded compute engine, the efforts of resolutions could be readily accommodated by reconfiguring the collaborative development team could be shared between the FPGA. This scheme also helped to minimize the the two different types of systems while still providing two development time needed to implement the graphics decidedly different control systems. subsystem in each system.

The MEN ESM was mounted on different carrier cards The ESM used for both the Zinser spinning systems and for each company’s systems. These carrier cards also the Schlafhorst winding systems featured a 266 MHz included the application-specific I/O and other system- PowerPC 8245 processor running either Linux or the OS- specific functionality. And since MEN’s ESM featured a 9 operating systems. Also located on the SOM was 64 Field Programmable Gate Array (FPGA), the same EMS megabytes (MB) of 133 MHz memory in single outline-dual could easily be modified to serve the two different sets of inline memory modules (SO-DIMM), as well as 32 MB of functional requirements. For the different functionality CompactFlash memory. required for Zinser’s spinning systems or for Schlafhorst’s winding systems, the FPGA was simply reconfigured with MEN’s Embedded System Module featured the level of different IP functional cores. flexibility and versatility that allowed it to serve two different yet similar applications with a minimum of extra design For example, both types of systems required a touch screen effort, reducing costs and speeding the systems from both TFT display monitor that is used by operators to input companies to market faster.

each train is logically divided into units Flexibility was a key requirement for network connecting all of the units in a train. which can be anywhere from one to three implementing the network of field buses To ensure failsafe operations, this bus cars. Each unit in a train includes an needed for Le Shuttle’s control system. was implemented as a dual redundant embedded local processing element. Several different local field buses are used communications channel. The local in European train applications. These network for each unit in a train was based To combat the shock and vibration, the include the MultiVehicle Bus (MVB), on dual redundant Bitbus lines. design team was drawn toward small Train Control Network (TCN), CAN, form factor circuit boards for the compute WorldFIP, Bitbus, and others. After much Interfacing a small form factor local control engines. Eventually, the short timeframe analysis, the development team settled unit to both the dual redundant WorldFIP dictated an off-the-shelf 3U Eurocard on the WorldFIP bus as the basis for the bus and two Bitbus lines was accomplished platform as the team’s selection. with a mezzanine card plugged into

Industrial Emdbedded Systems Resource Guide 2005 / 55 Computing: Technology each of the 3U control units. The mezzanine cards were based Dallas, Texas. Ernest has a BS in Electrical Engineering from on the M-Module standard, which supports 100 megabits per Texas A&M University and a MBA from the University of second throughput in burst mode. As shown in Figure 1, the Wisconsin, and is a registered Professional Engineer in the state M-Module is connected to the control unit via a sturdy 40- or of Texas. He holds patents in areas as diverse as analog signal 60-pin connector to ensure operations despite significant shock conditioning, computer architecture, and electronic packaging. and vibration while the train is in motion. Deploying a mezzanine Ernest also serves on the External Advisory and Development card based on the M-Module specification gave the design team a Council of the Department of Electrical Engineering at Texas number of different interrupt options, as well as Direct Memory A&M University. Access (DMA) and trigger lines. To learn more, contact Ernest at: The embedded balancing act MEN Micro, Inc. From textile processing machines to underwater trains, the PO Box 4160 developers of embedded industrial control systems walk a Lago Vista, TX 78645-4160 tightrope between the limitations of ready-made off-the-shelf Phone: 512-267-8883 solutions and more costly and time-consuming custom-developed E-mail: [email protected] solutions. Fortunately, design teams can take advantage of the Website: www.menmicro.com versatility and flexibility embodied in embedded system modules, mezzanine cards, carrier cards, and other system building blocks to overcome many of their challenges.

Ernest Godsey is President of MEN Micro, Inc., a wholly owned subsidiary of MEN Mikro Elektronik GmbH, a privately held German company. MEN supplies board-level products for embedded systems to OEMs worldwide. For the 13 years prior to his tenure with MEN, Ernest held a variety of senior management positions with Interphase Corporation in

Figure 1

56 / 2005 Industrial Emdbedded Systems Resource Guide Boards: Other module <100cm2

Computing AMTELCO XDS The AMTELCO XDS CompactPCI Infinity Series Boards includes a number of boards to meet your specific application needs. The XDS H.110 CompactPCI Infinity Series MC-3 Multi-Chassis Interconnect Boards allow you to connect 20+ PCchassis with 4800+ ports into a single system, with fully dynamic timeslot switching between all chassis. The XDS H.110 CompactPCI Infinity Series Analog Port Boards include High Density Loop Start Boards (16 and 32 ports), which come with Caller ID resources. The XDS High Density Station Port Boards function as an interface to ana- log telephones. The XDS E & M Boards have configurable ports with two- or four-wire Type I or Type V signaling circuits, radio control circuits, and two or four-wire audio circuits. They also can be used to provide PC enhanced service links to PBXs for two-way DID service for analog radio interfaces and four-wire audio circuits. The XDS H.110 CompactPCI Infinity Series BRI Boards support NI-1, NI-2, X.25, and international/Euro ISDN. The XDS H.110 FEATURES: CompactPCI Infinity Series High Density 512-Port Conference n Connect 20+ PC chassis with 4800+ ports into a single system with Boards are ideal for larger conference applications with enhanced the XDS MC-3 Multi-Chassis Interconnect Boards conferencing and flexible multi-conferencing facilities. Software n XDS MC-3 Multi-Chassis Interconnect Boards provide fully dynamic driver packages and service are included with all AMTELCO time-slot switching between all chassis XDS boards. n The XDS High Density Loop Start Board is available with either 16 or 32 ports, and includes Caller ID resources n The XDS High Density Station Port Boards function as an interface to analog telephones n The XDS E & M Boards have configurable ports, with two- or four- wire Type I or Type V signaling circuits n XDS E & M Boards also have radio control circuits and two- or four- wire audio circuits n Use XDS E & M Boards to provide PC enhanced service links to PBXs for two-way DID services and more n The XDS BRI Boards support NI-1, NI-2, X.25, and international/ Euro ISDN n The XDS High Density 512-Port Conference Boards are ideal for large conference applications n The XDS High Density Conference Boards include enhanced conferencing and flexible multi-conferencing facilities n Contact AMTELCO at 800-356-9224 OR 608-838-4194 for quick and reliable service n For a customized application to your product, e-mail your requirements to: [email protected]

AMTELCO 4800 Curtin Drive McFarland, WI 53558 Tel: 800-356-9224 or 608-838-4194 • Fax: 608-838-8367 www.xds.amtelco.com For more information, contact: [email protected]

RSC #5701 @ www.industrial-embedded.com/catalogrsc

Industrial Emdbedded Systems Resource Guide 2005 / 57 Boards: Other module <100cm2

M6713 M6713 is a powerful and flexible DSP + FPGA board of novel architecture for advanced data capture and real-time control in

Computing PCI systems. Designed around Texas Instruments’ most power- ful floating-point Digital Signal Processor (DSP) for high-speed, high dynamic range signal processing, and Xilinx’s latest Field Programmable Gate Array (FPGA) for unlimited customization of the I/O peripherals as well as hardware-accelerated signal process- ing, the M6713 adds high-speed, low latency IO and deterministic FEATURES: control to PC systems in a COTS (Commercially Off The Shelf) solu- n PCI 64-bit/66 MHz tion for a wide array of advanced real-time control applications. n Two OMNIBUS I/O module sites n Reconfigurable FPGA option up to 1.5 million gates n Supports multiple card I/O synchronization n Extensive software support in source form n Custom logic development supported for FPGA Innovative Integration 2655 Park Center Drive Simi Valley, CA 93065 Tel: 805-520-3300 • Fax: 805-579-1730 www.innovative-dsp.com For more information, contact: [email protected]

RSC #5801 @ www.industrial-embedded.com/catalogrsc

Boards: Other module <100cm2

SBC6713e SBC6713e is a high-performance, flexible, stand-alone DSP board with Ethernet connectivity, loaded with I/O peripherals. Built around the powerful, C-friendly, 300 MHz floating point C6713 DSP, it is a fully open platform with 15+ OMNIBUS available off-the-shelf daughter cards that provide a wide choice of A/D and D/A and also supporting simple EMIF bus interface to custom IO daughter cards. TCP/IP is running on a dedicated DM642 co-processor to preserve the C6713 for user-code and other peripheral controls. FEATURES: n 300 MHz TMS320C6713 DSP (Floating Point) n Two OMNIBUS I/O expansion sites n 10/100 Ethernet, RS232 port n FPDP data port to 200 MB/s n Capable of 100 percent Stand-alone operation n 600K gate Spartan-IIE for user-code (optional) Innovative Integration 2655 Park Center Drive Simi Valley, CA 93065 Tel: 805-520-3300 • Fax: 805-579-1730 www.innovative-dsp.com For more information, contact: [email protected]

RSC #5802 @ www.industrial-embedded.com/catalogrsc

58 / 2005 Industrial Emdbedded Systems Resource Guide Computing Computing Boards: Other module <100cm2 Boards: Other module <100cm2 Computing

phyCORE-LPC2294 phyCORE-MCF5485 FEATURES: FEATURES: n SBC subassembly in submini dimensions (60 x 53mm) n SBC subassembly in submini dimensions (70 x 57mm) n Philips LPC2294 ARM7 RISC microcontroller (TQFP-144) n Freescale ColdFire MCF5485 (200 MHz) in 388-pin BGA package n 1 (to 8) MB external SRAM, 2 (to 16) MB external Flash n 64 (to 128) MB DDR-SDRAM, 32 (to 64) MB Strata-Flash n 10/100 Mbit/s Ethernet, 4x CAN, JTAG, 2x serial interfaces n 4x UART, 2x CAN ports, 2.0 USB, PCI 2.2 bus, Ethernet n Available in Rapid Development Kits (Keil, IAR versions) n Available in Rapid Development Kit with Metrowerks CodeWarrior

PHYTEC America LLC PHYTEC America LLC 203 Parfitt Way SW, Suite G100 203 Parfitt Way SW, Suite G100 Bainbridge Island, WA 98110 Bainbridge Island, WA 98110 Tel: 206-780-9047 • Fax: 206-780-9135 Tel: 206-780-9047 • Fax: 206-780-9135 www.phytec.com www.phytec.com RSC #5901 @ www.industrial-embedded.com/catalogrsc RSC #5902 @ www.industrial-embedded.com/catalogrsc

Computing Computing Boards: Other module <100cm2 Boards: Other module <100cm2

phyCORE-PXA270CE USBwiz USB-file ic FEATURES: FEATURES: n SBC subassembly in submini (57 x 71.5mm) dimensions n Unique USB file-system ic – add USB storage to any product! n Populated with the Intel PXA270CE, PXA255 microprocessors n Adds SD media, thumb drive, HD, WiFi, printer, mouse, etc. n 32 (to 64) MB Flash, 64 (to 128) MB SDRAM n No USB knowledge necessary, plug-and-play – knows TCP/IP, too! n 3x serial, MMC/SD, 100 Mbit/s Ethernet, CAN, JTAG interfaces n As easy to use as our popular ALFAT serial-to-FATfile ic! n USB host/client, GPIO expander, WinCE and Linux kernels n USBwiz-OEM ready-made board with SD slot and USB host ic

PHYTEC America LLC Saelig Co., Inc. 203 Parfitt Way SW, Suite G100 1160-D2 Pitts-Victor Road Bainbridge Island, WA 98110 Pittsford, NY 14534 Tel: 206-780-9047 • Fax: 206-780-9135 Tel: 585-385-1750 • Fax: 585-385-1768 www.phytec.com www.saelig.com RSC #5903 @ www.industrial-embedded.com/catalogrsc RSC #5904 @ www.industrial-embedded.com/catalogrsc

Industrial Emdbedded Systems Resource Guide 2005 / 59 Computing Computing Boards: Other module <100cm2 Boards: Other module <100cm2

Technoland Technoland (an iBASE Company) (an iBASE Company) Computing

MB 740 Mini ITX MB IB880 – PentiumM SBC FEATURES: FEATURES: n Socket A for AMD Geode NX processors, up to 1 GHz speed n Supports Pentium® M / Celeron® M processors Up to 2.0GHz n DDR DIMM x 1, Max. 1 GB, DDR266/333 n Integrated VGA, shared memory, CRT/LVDS/DVI n Onboard 10/100 and Realtek 8110S Gigabit Ethernet n Integrated 10/100 BaseT and Intel® Gigabit LAN n Integrated SiS 741CX CRT VGA, optional LVDS n DDR DIMM x 2, max. 2GB, Watchdog timer, Digital I/O n 2 x SATA, 6 x USB 2.0, 2 x COM, watchdog timer n Type 2 CF socket, Mini PCI slot, PCI slot, 2 x SATA, 4 x USB Technoland Technoland (an iBASE Company) (an iBASE Company) 1050 Stewart Drive 1050 Stewart Drive Sunnyvale, CA 94085 Sunnyvale, CA 94085 Tel: 800-292-4500 • Fax: 408-992-0808 Tel: 800-292-4500 • Fax: 408-992-0808 www.technoland.com www.technoland.com RSC #6001 @ www.industrial-embedded.com/catalogrsc RSC #6002 @ www.industrial-embedded.com/catalogrsc

Computing Computing Boards: Other module <100cm2 Boards: SmartModule

ANTERO CONTROLLER DIMM EX CPU Module FEATURES: FEATURES: n ARM-7 Based SBC n Small 30 MIPS DIMM CPU board with 30 MIPS perfomance n IndustryPack® Compliant n Two Ethernet 10/100M Controllers with PHY, 4 Serial Ports, USB n Extensive I/O: Analog, Digital, Serial, etc. n 1 Mbyte SRAM, 2/4/8 Mbyte Flash, watchdog, status LEDs n CANbus, USB, Ethernet n Lead free, industrial temperature, JTAG debugger, SPI n VERY Low Power n Complete software RTOS included, low cost development kit

Vesta Technology, Inc. Grid Connect Inc. 11049 W. 44th Ave., Ste. 100 1841 Centre Point Circle #143 Wheat Ridge, CO 80033 Naperville, IL 60563 Tel: 303-422-8088 ext 111 • Fax: 303-422-9800 Tel: 630-245-1445 • Fax: 630-245-1717 WWW.VESTATECH.COM www.gridconnect.com RSC #6003 @ www.industrial-embedded.com/catalogrsc RSC #6004 @ www.industrial-embedded.com/catalogrsc

60 / 2005 Industrial Emdbedded Systems Resource Guide Boards: SmartModule

Computing smartModule SM915 The SM915 is the smallest CPU module (115mmx70mm) in its class and supports Intel® Celeron® M up to Pentium® M with 533 MHz FSB. The use of smartModule PCs as optimized, tested compo- nents in the development of new products not only reduces the design effort and the development costs, but can also drastically reduce time-to-market. The genius cooling feature ensures perfect thermal connection to a heat sink. FEATURES: Designed for low power consumption, the SM915 is the solution, where size, CPU, and graphic performance are important, along n Intel® Processor Celeron® M 600 MHz up to Pentium® M 760 with high reliability and long time availability. n 915GM 400/533 MHz FSB, ICH6, 256-1024MB DDR2-DRAM n Extreme Graphic, up to 128 MB (UMA), DirectX 9 compatible, CRT and 2x SDVO / LVDS n 8 x USB V2.0, 100MbitLAN, AC97 SPDif 7.1 sound stereo with eight- channel output and two-channel input DIGITAL-LOGIC AG n SM915Bus including PCI-Express and LPC Nordstrasse 11/F n Thermal management solution for operating temperature of –20 °C to +60 °C (optional –40 °C to +70 °C) CH-4542 Luterbach / Switzerland Tel: +41-32-681-58-00 • Fax: +41-32-681-58-01 www.digitallogic.com For more information, contact: [email protected]

RSC #6101 @ www.industrial-embedded.com/catalogrsc

Boards: SmartModule

smartModule SM855 The SM855 is the smallest CPU module (115mmx70mm) in its class and supports Intel® Celeron® M 600 MHz up to Pentium® M 755. The i855GME offers VGA and DVO/LVDS. The smartBus855/915 definition covers all interfaces, such as PCI and LPC bus, Serial ATA, and PCI Express, supporting the Intel®915GM Alviso chipset. The genius cooling feature ensures perfect thermal connection to a FEATURES: heat sink. Designed for low power consumption, the SM855 is the n Intel® Processor Celeron® M 600 MHz up to Pentium® M 755 solution for mobile, medical, military, automotive, POS/POI, or tele- n 855GME 400 MHz FSB, ICH4, 256-1024MB DDR-RAM com applications, where size, CPU, and graphic performance are important, along with high reliability and long time availability. n Extreme Graphic, 64 MB (UMA), DirectX 9 compatible, CRT and DVO / LVDS n 6 x USB V2.0, LAN, AC97 SPDif 5.1 sound stereo with six-channel output and two-channel input DIGITAL-LOGIC AG n SM855Bus including PCI and LPC Nordstrasse 11/F n Heat spreader and cooling feature for operating temperature of –20 °C to +60 °C (optional –40 °C to +70 °C) CH-4542 Luterbach / Switzerland Tel: +41-32-681-58-00 • Fax: +41-32-681-58-01 www.digitallogic.com For more information, contact: [email protected]

RSC #6102 @ www.industrial-embedded.com/catalogrsc

Industrial Emdbedded Systems Resource Guide 2005 / 61 Boards: CompactPCI

ICP-PM & ICP-CM CPUs Considering the base design for the ‘Samurai’ CPU family was drafted some 3 years ago, the typical five-year cradle-

Computing to-grave product lifecycle has been given an impressive boost, by at least five more years, giving the ultimate in product stability – an absolute novelty in the embedded computing sector. The ICP-PM single board computer combines high performance with low power consumption and is available with the latest 2.0 GHz Intel Pentium M 765 processor based on the 90 nm Dothan kernel. With an impressive 2 MByte L2 cache and higher clock frequency, the power consumption remains the same as that of the 1.6 GHz, 130 nm Banias processors but yields even greater performance. The standard configuration has 512 MB DDR 333 RAM onboard with the option to extend this capacity by a further 1 GB through FEATURES: an SO-DIMM interface. Hence, the Inova ICP-PM sits in a class of n Master/Slave 3U CompactPCI CPUs with Passive or its own as it more than exceeds the specification of its nearest Active Cooling & Software Support for all Popular rivals. Thanks to the SpeedStep technology, these CPUs are per- Operating Systems fectly suited to passive cooling in embedded systems. n Pentium M CPUs: For power-critical applications, two ULV Celeron M platforms are • Frequency Clocking in BIOS (600 MHz to 2.0 GHz) available that host either the • Up to 533 MHz Front-Side Bus 600 MHz or 1 GHz processor. The power consumption for the • Up to 2 MByte L2 Cache entire board remains around the 10 W mark. • Up to 1.5 GByte 333 MHz DDR RAM • Dual Gigabit Ethernet Both 4HP and 8HP variants are possible, with a screened version n Celeron M ULV CPUs for higher operational temperatures (-40°C to +85°C) available as • 600 MHz or 1.0 GHz ULV Processors an option. A universal PCI/PCI bridge permits Master/Slave config- • 5W Power Consumption urations without design alteration – yet another advantage when • Passive Cooling comparing similar competitive products. • Dual The 3U CompactPCI format of these CPUs makes them particularly n Phoenix BIOS suited to applications where compactness, low power consump- n SiS Chipset tion and mechanical stability are a prerequisite. These markets n Choice of Onboard or AGP Radeon Graphics include, typically, mobile transportation, robotics, manufacturing systems, telecommunications and energy and environment. n VGA / DVI / TFT Supported Formats n Up to 2048 x 1536 Pixel Video Resolution These CPUs are available for rugged applications through the usual worldwide Inova sales offices or distribution channels. Prices n Up to 3x USB 2.0 Interfaces start at EUR 990.00 and additional technical information may be n CompactFlash Socket retrieved from the Inova homepage: n Comprehensive & Intelligent Rear I/O Fabric http://www.inova-computers.de/icppm.php n Universal 3.3V / 5.0V Design

Inova Computers, Inc. 18275 North 59th Ave. Suite 152 Glendale, AZ 85308 TEL: (602) 863-0726 • FAX: (602) 863-0796 www.inova-computers.de/icpsysclc.php For more information, contact: [email protected]

RSC #6201 @ www.industrial-embedded.com/catalogrsc

62 / 2005 Industrial Emdbedded Systems Resource Guide Boards: CompactPCI

Computing SAVINGS-BOX – ICP-SYSC-LC By understanding the shifting trends and market dynam- ics, Inova have created a future-oriented, innovative prod- uct that is based on the industry-proved stability of 3U CompactPCI, but pitched at a price typically reserved for industrial PCs. The ICP-SCSC-LC is a complete 3U, 30 HP system with the latest processing technology and three free CompactPCI expansion slots to meet the target application requirements. Performance scalability is ensured by embracing the complete spectrum from Intel’s embedded processor roadmap; from the ultra low voltage 600 MHz Celeron M, right up to the latest and highest performance, 2.0 GHz Pentium M processor with its 533 MHz front-side bus. Power consumption, and hence the choice of passive or active cooling, can therefore be regulated to suit the operational environment. FEATURES: The CPU boards are furnished with a standard CompactFlash n Rugged 3U CompactPCI Design socket for use by modern operating systems such as the real-time – 4-Slot CompactPCI Backplane VxWorks or Embedded XP for maintenance-free, high-availability – MTBF > 200,000 hours applications. This combination delivers yet another competitive – 0°C to +70°C Operational Temp. edge since, during its operational life, those costs typically asso- ciated with regular housekeeping maintenance are saved. The – Designed for Windows XP Embedded use of ‘automotive-grade’ hard disks, qualified for operation at n 600 MHz to 2.0 GHz Celeron M / Pentium M CPU temperatures between -20°C to +70°C, ensure added durability – Up to 2 MByte L2 Cache – particularly for applications in harsh environments such as those – Up to 1.5 GByte 333 MHz DDR RAM found in mobile (transportation) sectors. Where necessary, the sys- – Up to 533 MHz Front-Side Bus tem can be furbished with suitable 24x7, 2.5" hard disks for round- – Choice of onboard or AGP Radeon Graphics the-clock operation. – VGA / DVI / TFT Supported Formats The international acceptance and stability of the 3U – Up to 2048 x 1536 Pixel Video Resolution CompactPCI standard combined with the inherent robustness of 19" technology were the inspiration behind the ‘Savings-Box’. – Up to 3x USB 2.0 Interfaces – Dual Gigabit or Gigabit & Fast Ethernet Interfaces The ICP-SYSC-LC is available for rugged applications through the – CompactFlash Socket usual worldwide Inova sales offices or distribution channels. Prices start at EUR 1,490.0 and additional technical information may be – Comprehensive & Intelligent Rear I/O Fabric retrieved from the Inova homepage: – Universal 3.3V / 5.0V Design The 600 MHz Celeron version is just 4HP wide and is suited for those http://www.inova-computers.de/icpsysclc.php applications where moving parts in the form of a fan or hard disk are not permitted. n 70W AC/DC PSU – 115V / 230V Input

Inova Computers, Inc. 18275 North 59th Ave. Suite 152 Glendale, AZ 85308 TEL: (602) 863-0726 • FAX: (602) 863-0796 www.inova-computers.de/icpsysclc.php For more information, contact: [email protected]

RSC #6301 @ www.industrial-embedded.com/catalogrsc

Industrial Emdbedded Systems Resource Guide 2005 / 63 Boards: EBX

Cobra Pentium® M EBX Single Board Computer The Cobra belongs to a new class of embedded comput- ers offering higher processing performance with lower power

Computing consumption, extreme graphics capability, and enhanced rugged- ization. The advanced design makes it suitable for a wide range of higher-end applications with demanding speed and graph- ics requirements, including gaming devices, information kiosks, telecommunications devices, and advanced security systems. In addition, its PC architecture and small footprint make it an excellent choice for traditional embedded applications. The system is built around the Intel Pentium M or Celeron M proces- sor with SpeedStep technology, and their companion chipset, the 855GME, with integrated Extreme Graphics 2 and audio functions. The processor incorporates a number of power saving features, includ- ing the SpeedStep technology and performance enhancing fea- tures, such as a large 1 MB Level 2 cache. The advanced graphics controller provides exceptionally sharp images, ultra-fast render- FEATURES: ing, and smooth sequencing to produce realistic 3-D graphics and n Pentium M or Celeron M processor for lower power consumption, animation. higher performance n Extreme Graphics 2 video provides very high-speed rendering and The board features two SODIMM sockets, which MPEG-2 support support up to 2 GB of system RAM. These rugged, n low-profile RAM sockets provide more reliable operation in appli- Rugged, low profile SODIMM memory sockets accept up to 1 GB RAM each cations subject to shock and vibration. The Double Data Rate (DDR) RAM interface provides fast memory access and increased n OEM, field-upgradeable, customizable BIOS system throughput. n Dual 10/100Base-T or gigabit Ethernet ports provide high-speed networking capability Like all VersaLogic products, the Cobra is designed and guaranteed n Four USB 2.0 ports support high-speed USB devices for long-term availability. From application design to continuing production, it provides a cost-effective solution with low total cost n Two standard + two RS-422/485/232 configurable ports of ownership. The Cobra is manufactured to the highest quality n Enhanced ESD protection for keyboard, USB, Ethernet, and other standards, and is backed with excellent technical and engineer- I/O with onboard TVS devices ing support. Customization is available in quantities as low as n CompactFlash socket provides onboard static memory with no 100 pieces. moving parts n CPU temperature sensor supports software-controlled cooling options n Watchdog timer controls application run-away conditions n Onboard, 32-line, TTL I/O port reduces need for I/O expansion

VersaLogic Corporation 3888 Stewart Road Eugene, OR 97402 Tel: 541-485-8575 • Fax: 541-485-5712 www.VersaLogic.com For more information, contact: [email protected].

RSC #6401 @ www.industrial-embedded.com/catalogrsc

64 / 2005 Industrial Emdbedded Systems Resource Guide Boards: EPIC

Computing Gecko EPIC Single Board Computer The Gecko is an efficient mid-sized SBC with an excellent balance of outstanding performance, low power con- sumption, and a complete set of onboard I/O features. This 4.5" x 6.5" board was designed specifically to support OEM applica- tions where long-term availability is required. The Gecko features the advanced AMD GX-500 CPU. This highly integrated processor features fanless operation and very low power consumption (one watt!). The GX-500 sets new standards for performance, yet it operates at less than half of the power consumption of compara- ble processors. It provides excellent throughput, such as extremely fast onboard transfers (6 GB per second), fast DDR memory access, and integrated high-performance video. The Gecko complements the GX-500 CPU with a full range of onboard I/O devices to form a complete, highly effective single board computer. Based on the EPIC form factor, the Gecko offers a com- pact, cost-effective solution for developers who require a mid-range FEATURES: platform (smaller than EBX, yet with more features than PC/104). n EPIC form factor; industry-standard, multi-vendor support n Like all VersaLogic products, the Gecko is designed GX-500 processor provides 500 MHz-equivalent performance with lower power draw and fanless operation from day one for high reliability and long-term availabil- ity. From initial application design to five-plus years of n Up to 512 MB Double Data Rate (DDR) RAM; 64-bit data path production, its quality and longevity provide a cost-effective, long- n Analog and LVDS flat-panel outputs for 18 and 24-bit video displays term solution. The Gecko is manufactured to the highest qual- n Onboard sound, 10/100Base-T Ethernet, four-port USB, four COM ity standards and is backed with VersaLogic’s excellent, award- ports (two RS-232 and two RS-422/485), IDE interface winning support. Product customization is available in quantities as n Built-in Transient Voltage Suppressor (TVS) devices provide enhanced low as 100 pieces. ESD protection for onboard I/O n Analog input provides eight 12-bit input channels; sixteen-line TTL I/O port reduces need for I/O expansion n Standard PC-style connectors simplify cabling and lower costs in many applications n CompactFlash socket provides onboard static memory with no moving parts n Watchdog timer provides hardware-level safety control for application run-away conditions n Embedded BIOS with OEM embedded features; field-upgradeable; customization available n NEW Gecko extended temperature version for operation from -40° to +85° C; ideal for space-constrained environments with limited cooling available

VersaLogic Corporation 3888 Stewart Road Eugene, OR 97402 Tel: 541-485-8575 • Fax: 541-485-5712 www.VersaLogic.com For more information, contact: [email protected].

RSC #6501 @ www.industrial-embedded.com/catalogrsc

Industrial Emdbedded Systems Resource Guide 2005 / 65 Boards: PC/104-Plus

Jaguar PC/104-Plus Single Board Computer The Jaguar is designed for projects requiring fast processing, compact size, flexible memory options, high reliabil-

Computing ity, and long product availability. It features Celeron or Pentium III processors from 350-850 MHz, and up to 256 MB of RAM secured in a high-reliability latching socket. Additional features include DiskOnChip socket, 128-256 KB Level 2 cache, watch- dog timer, and Vdc-sensing reset circuit. Standard I/O ports include dual floppy, AT peripherals, keyboard, PS/2 mouse, PCI-based IDE, dual USB, LPT, and two COM ports (one RS-232/422/485 selectable). The high reliability design and construction of this board features Transient Voltage Suppressors (TVS) on user I/O for enhanced ESD resistance, latching I/O connectors, watchdog timer, voltage sensing reset circuit, spread spectrum frequency generation for lower EMI, industrial long-life battery, and self-resetting fuse on the 5 V supply to the keyboard, FEATURES: mouse, and USB ports. The CPU temperature sensor can be used n 850 MHz Pentium III processor for high speed processing to turn on additional fans, create local or remote warnings, or take n Celeron processors for extended temperature, fanless, and low other action through software triggers. power versions n The General Software Embedded BIOS® 2000 offers CD-ROM ATI Rage Mobility video chip provides 3-D rendering, MPEG-2, and and Ethernet boot capability. It also features reconfigurable BIOS flat-panel support defaults for optional battery-free operation. n 32-pin DiskOnChip socket for onboard non-mechanical storage n 10/100Base-T Ethernet Both hardware and software (BIOS) customization are available in 100+ unit quantities. Each board is subjected to Environmental n Dual USB Stress Screening, complete functional testing, and a five-year n TVS devices on user I/O ports for enhanced ESD protection longevity guarantee. n CPU temperature sensor controls system cooling programmatically n Watchdog timer stops runaway applications n Intel 440BX chipset – High-speed, front-side bus n PC/104-Plus interface for expansion using PC/104 or PC/104-Plus add-on modules n OEM-enhanced BIOS with reconfigurable defaults

VersaLogic Corporation 3888 Stewart Road Eugene, OR 97402 Tel: 541-485-8575 • Fax: 541-485-5712 www.VersaLogic.com For more information, contact: [email protected].

RSC #6601 @ www.industrial-embedded.com/catalogrsc

66 / 2005 Industrial Emdbedded Systems Resource Guide Boards: PC/104

Computing VIPER: PXA255 SBC The VIPER is an ultra-low power PC/104 compatible single board computer based on the Intel® 400MHz PXA255 XScale® RISC pro- cessor. The PXA255 is an implementation of the ARM compliant, Intel XScale microarchitecture combined with a comprehensive set of integrated peripherals including, a flat panel graphics controller, DMA controller, interrupt controller, real time clock and multiple serial ports. The VIPER board offers a long list of features making it ideal for power sensitive embedded communications and multimedia appli- cations. The board has been designed to take advantage of the power saving modes of the PXA255 processor and other onboard peripherals to achieve an incredible 1.9W typical power consump- tion. The VIPER also supports a very low power standby mode. The VIPER board includes a TFT/STN flat panel graphics control- ler, onboard soldered SDRAM and resident Flash, 10/100baseTx FEATURES: Ethernet, 5 serial ports, dual USB host controller, USB client, AC97 n Processor: 200MHz or 400MHz Intel® PXA255 XScale audio/codec, CompactFlash interface (CF+), and a standard PC104 n bus expansion connector. The PC/104 format is an industrial form Memory: Up to 64Mbytes of soldered SDRAM, 32Mbytes of Flash and 256Kbytes of battery backed SRAM factor measuring 3.8" x 3.6" (96mm x 91mm). n Video: TFT/STN flat panel support up to 6400 x 480 x 16-bit color.. The VIPER is supported with Development Kits for the lead- Support for analog and LVDS output using adapter modules. ing embedded operating systems including Windows CE .NET, n Network: 10/100baseTx Ethernet controller (SMSC 91C111) embedded Linux and VxWorks 5.5. Arcom also provides support n ™ USB: Dual USB host ports - v1.1 (Philips ISP1160) and a factory fit for RedBoot , a utility based on the eCos RTOS, which serves as USB client option (surface mount link configuration) a simple boot manager and download tool for embedded Linux n Serial ports: 5 serial ports ( RS-232, RS-422 and RS-485) applications. n Expansion: CompactFlash (CF+ for both memory and I/O with hot To speed up the process of system integration, you can purchase swap support) the VIPER ICE enclosure (fitted with a 320x240 wide temperature n I/O: 8 buffered digital inputs / 8 buffered digital outputs (+5V TFT / touchscreen display) or the rugged CYCLOPS display terminal tolerant) (fitted with a high brightness 640x480 TFT / touchscreen display). For cost sensitive high volume applications Arcom offers the VIPER- Lite - an alternative configuration based on the 200MHz PXA255 fitted with 64Mbytes DRAM and 16Mbytes Flash.

Arcom 7500 West 161st Street Overland Park, KS 66085 Tel: 913-549-1000 • Fax: 913-549-1002 www.arcom.com For more information, contact: [email protected]

RSC #6701 @ www.industrial-embedded.com/catalogrsc

Industrial Emdbedded Systems Resource Guide 2005 / 67 Boards: PC/104

MSM915 DIGITAL-LOGIC offers the PC/104-Plus CPU module MSM915 with Intel® processors from Celeron® M up to Pentium® M. The standard

Computing version includes 6xUSB-V2.0, 10/100Base-T with boot from LAN, a watchdog, up to 1024MB of DDR RAM and consumes only 10W to 30W. 2 S-ATA and 1 P-ATA drives can be connected. CPU and RAM are protected against vibrations and shock.

By the Intel 915GM chip set, graphics with 224MB RAM and reso- FEATURES: lutions up to 2048x1536, TV output for HDTV and an AC97/HDA with 7.1 DTS sound are available. n MICROSPACE® MSM915 baseboard with exchangeable smartModule SM915 The MSM915 module is perfectly suited for transportation, tele- n Intel® Processor Celeron® M 600MHz up to Pentium®M 765 with 2.2 communication, medical, measurement, aerospace, or automotive GHz application. n Intel® 915GM with 533 MHz front-side bus, ICH4, 512-1024MB DDR- RAM SODIMM n Extreme graphics, 224MB, DirectX 9 compatible, 1xVGA-CRT, 1xDVI- DIGITAL-LOGIC AG D/LVDS or S-Video-out Nordstrasse 11/F n MS/KB, FD, 1xP-ATA, 2xS-ATA, LPT1, 6xUSB V2.0, LAN, AC97-HDA, CH-4542 Luterbach / Switzerland 7.1 DTS sound Tel: +41-32-681-58-00 • Fax: +41-32-681-58-01 n Optimized thermal concept and cooling feature for operating www.digitallogic.com temperature of –20 °C to +50 °C (optional –40 °C to +50 °C) For more information, contact: [email protected]

RSC #6801 @ www.industrial-embedded.com/catalogrsc

Computing Computing Boards: PC/104 Boards: PC/104

Microcomputer Systems, Inc. AAEON Electronics, Inc.

MSI-P402 PFM-620S PC/104+ FEATURES: FEATURES: n Quadrature decoder card with 32-bit binary up/down counters n Intel Ultra Low Voltage 400 or 650 MHz CPU n Available for 2, 4, 6, or 8 channels of encoder inputs n Supports up to 36-bit TFT/DSTN LCDs n Ideal for machine control and robotic applications n 10/100 Mbps Fast Ethernet LAN prot n Individual software reset for each channel n CompactFlash socket / 2 COM / 2 USB ports n Operates -25°C to 85°C from single +5V; cost $275 to $350 n +5V power input only required

Microcomputer Systems, Inc. AAEON Electronics, Inc. 1814 Ryder Drive 3 Crown Plaza Baton Rouge, LA 70808 Hazlet, NJ 07730-2496 Tel: 225-769-2154 • Fax: 225-769-2155 Tel: 732-203-9300 • Fax: 732-203-9311 www.microcomputersystems.com www.aaeon.com RSC #6802 @ www.industrial-embedded.com/catalogrsc RSC #6803 @ www.industrial-embedded.com/catalogrsc

68 / 2005 Industrial Emdbedded Systems Resource Guide Chips: Processors

Computing Nios II Dev. Kit Cyclone II Edition Altera® Cyclone™ II FPGAs are optimized for low cost, while offering the benefits of first-generation Cyclone devices, including a customer-defined feature set, industry-leading performance, and low power consumption. They also offer greatly increased density and more features, all at significantly lower cost. The Cyclone II family provides flexible hardware solutions that include configu- rable multipliers capable of implementing common functions such as filtering, correlation, transforms, encryption, error correction, and video and image processing. Embedded multipliers provide more speed and logic efficiency compared to logic element (LE)- based multipliers. The Nios® II Development Kit, Cyclone II Edition, offers a variety of intellectual property (IP) cores to speed develop- ment, including three Nios II 32-bit RISC processors. The Nios II processors—fast, standard, and economy—are each opti- mized for a specific price and performance range, allowing design- FEATURES: ers to choose a system configuration that is an exact fit for their n FPGA platform for embedded applications using Altera’s low-cost embedded needs. All three processors use the same instruction set Cyclone II family architecture and are 100% binary code compatible. Nios II proces- n Includes a complete set of development tools, hardware, software, IP sors can be added to a designer’s system using the SOPC Builder cores & reference designs system development tool in the Quartus® II development software. • Cyclone II development board with EP2C35 FPGA & 10/100 This kit includes 12 months of upgrades so you can keep your sys- Ethernet tem up-to-date with the latest enhancements to the Nios II proces- • 16-Mbyte DDR SDRAM, 16-Mbyte flash, 1-Mbyte SRAM, 16- sor, peripherals, and system design tools. Also included is a coupon Mbyte external compact flash that provides a low-cost upgrade path to a perpetual license for the • 32-bit PCI mezzanine connector, JTAG and Mictor debug headers, Quartus II design software, 12 months of updates, and the 81 user I/O (5-V tolerant expansion headers) ModelTechnology™ ModelSim®-Altera software. • IP cores with perpetual use license Visit www.altera.com/nios for more information. – Three Nios II 32-Bit RISC processor cores (fast, standard, economy) – Timer, UART, SPI, JTAG UART, GPIO, DMA, Ethernet interface (MAC/PHY) – On-chip RAM/ROM, SRAM, SDRAM, flash, compact flash interface, serial flash Interface – Tri-state bridge, LCD interface, interface to user logic, custom instruction • Integrated development environment for design & debug, including Quartus II software & SOPC Builder system integration tool • MicroC/OS-II RTOC, lightweight IP TCP/IP stack

Altera Corporation 101 Innovation Drive San Jose, CA 95134 Tel: 408-544-7000 • Fax: 408-544-6403 www.altera.com For more information, contact: [email protected]

RSC #6901 @ www.industrial-embedded.com/catalogrsc

Industrial Emdbedded Systems Resource Guide 2005 / 69 Computing Computing Boards: PC/104 Boards: VMEbus

Microcomputer Systems, Inc. THEMIS Computing

MSI-P600 GPS Card THEMIS TA64 FEATURES: FEATURES: n SiRF GPS Architecture with standard NMEA 0183 protocol. n Themis’ new family of 6U VMEbus AMD processor-based SBCs n Full-duplex serial port for navigation and control. n AMD Turion 64 Mobile processor running at 1.6GHz n Differential GPS capability plus a TTL I/O port. n Memory – Up to 4 GBytes DDR333 SDRAM n Active antenna included with a 5 meter (16.5 ft.) cable. n Low 45 watt power dissipation (without PMC) n Operating temperature range -40° C to 80° C. Cost $275. n Supports 32-bit Solaris 9 & 64-bit Solaris 10, Windows, and Linux OS

Microcomputer Systems, Inc. THEMIS 1814 Ryder Drive 47200 Bayside Parkway Baton Rouge, LA 70808 Fremont, CA 94538 Tel: 225-769-2154 • Fax: 225-769-2155 Tel: 510-252-0870 • Fax: 510-490-5529 www.microcomputersystems.com www.themis.com/prod/hardware/ta64.htm RSC #7001 @ www.industrial-embedded.com/catalogrsc RSC #7002 @ www.industrial-embedded.com/catalogrsc

Boards: VMEbus

TPPC64 The TPPC64™ is Themis’ first 6U VMEbus computer board family based on the IBM® PowerPC® 970FX processor. The PowerPC 970FX processor provides maximum performance for existing 32- bit applications and new 64-bit applications. The TPPC64 is available in single-slot uniprocessor and two-slot, dual Symmetric Multiprocessing configurations. I/O extension and graphics boards added to either single or dual FEATURES: processor configurations occupy additional VMEbus slots. The TPPC64 includes two Gb Ethernet ports and Dual Ultra320 SCSI n IBM PowerPC 970FX processor - 1.8GHz clock rate - in single and channels. I/O expansion is supported via a PCI riser. PMC I/O can be dual processor configurations expanded to 4 slots with 2 different PMC carrier boards. n Up to 4 GB of DDR400 SDRAM memory n Two Gigabit Ethernet ports, two USB ports, two serial ports and one SCSI port on front panel n Carrier board PCI expansion supports up to three additional THEMIS PMC slots n Support for Linux OS 47200 Bayside Parkway n Rugged design for reliability in harsh operating environments—up to Fremont, CA 94538 30G shock Tel: 510-252-0870 • Fax: 510-490-5529 www.themis.com/prod/hardware/tppc64.htm For more information, contact: [email protected]

RSC #7003 @ www.industrial-embedded.com/catalogrsc

70 / 2005 Industrial Emdbedded Systems Resource Guide Chips: Processors

Computing Nios II Dev. Kit Stratix II Edition Altera’s 90-nm Stratix® II devices are the industry’s biggest and fastest FPGAs and feature an innovative logic structure that packs more functionality into less area to dramatically reduce device costs. With internal clock frequency rates up to 500 MHz and typical performance >250 MHz, these devices deliver on average 50% faster performance and more than 2x the logic capacity than first-generation Stratix FPGAs. Stratix II FPGAs deliver 50x higher multiplier bandwidth than single-chip, standalone digital signal processors. The DSP blocks have the flexibility and performance to implement fast, arith- metic-intensive applications such as image processing, wire- less communications, military, broadcast, and medical. Each DSP block has dedicated multi-width multipliers to implement DSP algorithms and functions, including filtering, video and image pro- cessing, correlation, transforms, encryption, and error correction. The Nios II Development Kit, Stratix II Edition, offers a vari- FEATURES: ety of intellectual property (IP) cores to speed develop- n Industry’s highest density and highest performance FPGA platform for ment, including three Nios® II 32-bit RISC processors. The embedded applications Nios II processors—fast, standard, and economy—are each n Includes a complete set of development tools, hardware, software, optimized for a specific price and performance range, allow- IP cores, and reference designs ing designers to choose a system configuration that is an exact • Stratix II development board with EP2S30 FPGA & 10/100 Ethernet fit for their embedded needs. All three processors use the same PHY/MAC instruction set architecture and are 100% binary code com- • 16-Mbyte SDRAM, 16-Mbyte flash, 1-Mbyte SRAM, 16-Mbyte patible. Nios II processors can be added to a designer’s sys- external compact flash tem using the SOPC Builder system development tool in the – JTAG & Mictor debug headers, 81 user I/O (5-V tolerant Quartus® II development software. This kit includes 12 months expansion headers) of upgrades so you can keep your system up-to-date with – IP cores with perpetual use license the latest enhancements to the Nios II processor, peripherals, – Three Nios II 32-Bit RISC Processor cores (fast, standard, and system design tools. Also included is a coupon that pro- economy) vides a low-cost upgrade path to a perpetual license for the – Timer, UART, SPI, JTAG UART, GPIO, DMA, Ethernet Quartus II design software, 12 months of updates, and the interface (MAC/PHY) ™ ® ModelTechnology ModelSim -Altera software. – On-chip RAM/ROM, SRAM, SDRAM, flash, compact flash Visit www.altera.com/nios for more information. interface, serial flash interface – Tri-state bridge, LCD interface, interface-to-user logic, custom instruction – Integrated development environment for design & debug, including Quartus II software & SOPC Builder system integration tool – MicroC/OS-II RTOS, lightweight IP TCP/IP stack

Altera Corporation 101 Innovation Drive San Jose, CA 95134 Tel: 408-544-7000 • Fax: 408-544-6403 www.altera.com For more information, contact: [email protected]

RSC #7101 @ www.industrial-embedded.com/catalogrsc

Industrial Emdbedded Systems Resource Guide 2005 / 71 Chips & Cores: Pentium

Intel® 945G Express Chipset The Intel® 945G Express chipset delivers innovative features for interactive clients and many other embedded computing solutions

Computing requiring enhanced graphics capabilities. The new graphics core, combined with the Intel high-performance, dual-channel memory interface, delivers significant graphics performance over previ- ous Intel platforms. With support for dual-independent display, enhanced modes for widescreen flat panels, and optimized 3D, embedded platforms based on the Intel 945G Express chipset can deliver an intense, realistic visual experience without requiring a separate graphics card. Designed for, and validated with, the Intel® Pentium® 4 processor 551* with Hyper-Threading (HT) Technology† and the Intel® Celeron® D processor 341* – both with Intel® Extended Memory 64 Technology§ – the Intel 945G Express chipset platform provides scalable performance, making it an ideal price/perfor- mance solution for embedded computing applications. FEATURES: n ® Support for dual independent display, enhanced modes for The Intel 945G Express chipset consists of the Intel 82945G widescreen flat panels and optimized 3D to deliver intense visual Graphics and Memory Controller Hub (GMCH) and the experience without need for separate graphics card. ® Intel ICH7/ICH7R I/O Controller Hub (ICH). It delivers outstanding n High-bandwidth interfaces including PCI Express® x16 graphics or system performance through high-bandwidth interfaces such as I/O, PCI Express x1 I/O ports, next-generation Serial ATA (SATA II), ® PCI Express x16 graphics or I/O, PCI Express x1 I/O ports, next- and Hi-Speed USB 2.0 connectivity. generation Serial ATA (SATA II), and Hi-Speed USB 2.0 connectivity. n Scalable to higher performance Intel® Pentium® 4 processor with Additionally it features dual-channel DDR2-667 memory technol- Hyper-Threading (HT) Technology and Intel® Celeron® D processor. ogy, Intel® Graphics Media Accelerator 950, enhanced manage- n Dual-channel DDR2-533/667 allows up to 10.7 GB/second of ability and storage security technologies. bandwidth and 4 GB memory addressability for faster system responsiveness. n Support for 64-bit computing. n † Hyper-Threading Technology requires a computer system with an Intel® Pentium® 4 pro- Facilitates upgrades by allowing different memory sizes to be cessor supporting Hyper-Threading Technology and an HT Technology-enabled chipset, BIOS populated and remain in dual-channel mode. and operating system. See http://www.intel.com/info/hyperthreading for more information. n 2.0 GB/s concurrent bandwidth to maximize throughput between the *Intel® processor numbers are not a measure of performance. core chipset components. See http://www.intel.com/products/processor_number for details. n Six PCI masters provide generous system expansion capability. § Intel® EM64T requires a computer system with a processor, chipset, BIOS, operating n Provides protection and fast access to digital photo, video, and data system, device drivers and applications enabled for Intel EM64T. Processor will not operate content through RAID 0, 1, 5, and 10. (including 32-bit operation) without an Intel EM64T-enabled BIOS. n Enables remote, down-the-wire management of out-of-band See www.intel.com/info/em64t or consult with your system vendor for more information. networked systems, regardless of system state. n Numerous video output options (DVI, dual independent display, component, composite, HDTV, and LVDS) in a single-card solution. n Embedded lifecycle support.

Intel Corporation 2200 Mission College Blvd. Santa Clara, CA 95052 Tel: 800-628-8686 www.intel.com/go/embedded

RSC #7201 @ www.industrial-embedded.com/catalogrsc

72 / 2005 Industrial Emdbedded Systems Resource Guide Chips & Cores: Pentium

Computing Intel® Pentium® 4 Processors with Hyper-Threading Technology Intel® Pentium® 4 processors with Hyper-Threading Technology† (HT Technology), manufactured on 90nm process technology, provide increased system respon- siveness in multi-tasking environments and are ideal for next-generation multithreaded applications. With 800 MHz processor system bus and 1 MB Level 2 Advanced Transfer Cache (ATC), they enable highly differentiated and scalable embedded solutions within the growing communications, interactive client, and industrial automation market segments. The 3.4 GHz Intel Pentium 4 processor 551* with HT Technology simplifies the transition to 64-bit computing for embedded customers developing with Intel® Extended Memory 64 Technology§. It is designed for use with both the Intel® 915GV Express chipset and Intel® 945G Express chipset, support- ing a variety of embedded graphics applications. FEATURES: n State-of-the-art 90nm process technology supports advanced, highly The 3 GHz Intel Pentium 4 processor with HT Technology differentiated embedded products. is validated with the Intel® 875P chipset, providing dual- n 800 MHz processor system bus delivers 6.4 GB of data per second. channel DDR memory – DDR 266/333/400 – with Error Correcting n Validated with the Intel® 875P chipset (FC-µPGA4 package) and Intel® Code (ECC) operation. AGP 8X supports the latest graphics con- ® trollers. 915GV Express and Intel 945G Express chipsets (LGA-775 package). n Performance headroom, robust I/O and scalability help reduce total Platforms based on these processors can help reduce total cost-of- cost-of-ownership. ownership for a new generation of advanced, highly differentiated n Increased system responsiveness in multitasking environments embedded products by providing performance headroom, robust supports next-generation multithreaded applications. I/O, scalability, and quality. In addition, Intel offers validated plat- n Enhancements to Intel NetBurst® microarchitecture provide software form design schematics to rapidly meet unique product application and architectural scalability to other Intel® processors. requirements. Board solutions are also available from third-party n 1 MB Level 2 Advanced Transfer Cache delivers high data throughput providers. channel between Level 2 cache and processor core. n Level 1 Execution Trace Cache includes 16 KB data cache; stores up † Hyper-Threading Technology requires a computer system with an Intel® Pentium® 4 pro- to 12K decoded micro-ops in the order of program execution. cessor supporting Hyper-Threading Technology and an HT Technology-enabledchipset, BIOS n Support for uni-processor designs. and operating system. See http://www.intel.com/info/hyperthreading for more information. n Data integrity and reliability features such as ECC and fault analysis Intel® processor numbers are not a measure of performance. and recovery for both system and Level 2 cache buses. See http://www.intel.com/products/processor_number for details. n Validated design schematics help rapidly meet unique product *§ Intel® EM64T requires a computer system with a processor, chipset, BIOS, operating application requirements. system, device drivers and applications enabled for Intel EM64T. Processor will not operate n Embedded life cycle support. (including 32-bit operation) without an Intel EM64T-enabled BIOS. See www.intel.com/info/em64t or consult with your system vendor for more information.

Intel Corporation 2200 Mission College Blvd. Santa Clara, CA 95052 Tel: 800-628-8686 www.intel.com/go/embedded

RSC #7301 @ www.industrial-embedded.com/catalogrsc

Industrial Emdbedded Systems Resource Guide 2005 / 73 Chips & Cores: Pentium

Intel® Pentium® M Processors on 90nm Process Intel® Pentium® M Processors on 90nm process utilize a new microarchitecture to meet the current and future demands of

Computing high-performance, low-power embedded computing, mak- ing them ideal for medium-to-large enterprise communications, transaction terminal, interactive client, and industrial automation applications. While incorporating advanced processor technology, they remain software-compatible with previous members of the Intel® micro-processor family. Intel Pentium M processors on 90nm process are available in both standard and low-voltage versions, providing a variety of perfor- mance and power options. These Pentium M processors are validated with the Intel® E7501 chipset, Intel® 855GME chipset, Intel® E7520 chipset, Intel® E7320 chipset, and Mobile Intel® 915GM Express chipset. Each chipset, when paired with the Intel Pentium M processor on 90nm process, FEATURES: helps create a unique platform that addresses a variety of cus- n Ideal for medium-to-large enterprise communications, transaction tomer requirements. terminal, interactive client, and industrial automation applications. Specific performance and power options: n Available in both standard and low-voltage versions, providing a variety of performance and power options. Intel® Pentium® M processor 760† at 2.0 GHz core speed and n ® ® ® 533 MHz front-side bus (FSB) speed Validated with the Intel E7501, Intel 855GME, Intel E7520, Intel® E7320, and Mobile Intel® 915GM Express chipsets to address Intel® Pentium® M processor 745† at 1.8 GHz core speed and a variety of customer requirements. 400 MHz FSB speed n Dedicated hardware stack manager employs sophisticated hardware control for improved stack management. Intel® Pentium® M processor Low Voltage 738† at 1.4 GHz core n speed and 400 MHz FSB speed Micro-ops fusion for improved instruction execution. n Advanced branch prediction capability. † Intel® processor numbers are not a measure of performance. See http://www.intel.com/products/processor_number for details. n 2 MB Level 2 Advanced Transfer Cache (ATC) delivers high data throughput channel between the Level 2 cache and the processor core. n Second-generation Streaming SIMD Extensions capability adds 144 new instructions: 128-bit SIMD integer arithmetic and 128-bit SIMD double-precision floating-point operation. n Support for uni-processor designs. n Fully compatible with existing Intel® architecture-based software. n 478 µFC-PGA and 479 µFC-BGA packages. n Embedded life cycle support.

Intel Corporation 2200 Mission College Blvd. Santa Clara, CA 95052 Tel: 800-628-8686 www.intel.com/go/embedded

RSC #7401 @ www.industrial-embedded.com/catalogrsc

74 / 2005 Industrial Emdbedded Systems Resource Guide Chips: Bridges and Hubs

®

Computing Tundra Tsi148™ The Tsi148 is the industry’s leading PCI/X-to-VMEbus bridge and the successful next-generation offering to the Universe™ II, the leading VMEbus bridge for embedded systems customers. Fully-compliant with VMEbus standards, the Tsi148 implements the 2eSST protocol that allows the VMEbus to run at a bandwidth up to 320 Megabytes per second. The Tsi148 is a full-featured master, slave, and system controller that can be used in any VME applica- tion and provides an increase in overall processing capability on FEATURES: legacy backplanes, while transparently enabling the high-perfor- n Industry’s best sustained transfer rate of 305 MB/s in 2eSST - higher mance distributed processing that new applications demand. system bandwidth n Multithreading - capable of a number of simultaneous transactions while optimizing bus utilization n Full-featured master, slave, and system controller - can be used in any VME application n PCI-X local bus supports two loads at 133 MHz - reducing component Tundra Semiconductor count 603 March Road n Small device footprint - saves board space Ottawa, ON K2K2M5 Canada n Proven VME backwards compatibility - preserves legacy investment Tel: 613-592-0714 • Fax: 613-592-1320 www.tundra.com/tsi148 For more information, contact: [email protected]

RSC #750103 @ www.industrial-embedded.com/catalogrsc

Computing Computing Chips: Industrial temp processor core Systems: Industrial ruggedized

Research Centre Module

1879BA1T ASIC 12-Bit Digitizer FEATURES: FEATURES: n 1879BA1T MIL-STD-1553B interface terminal n Precision and speed: 12-bit, 400 MS/s sampling n Integrated interface between processor and MIL-1553 data bus n Up to 4 GB (2 GigaSamples) of on-board acquisition memory n BC, RT, MT, simultaneous RT/MT modes n Signal Averaging & FIR Filtering FPGA technology available n 4Kx16 internal RAM, externally expanded to 64Kx16 n Programming-free GageScope oscilloscope software included n 14 x 14mm 80-pin LQFP package n SDKs for quick and easy development of custom applications

Research Centre Module Gage Applied Technologies 3, Eight March 4th Street Box 166 1 Provost, Suite 200 Moscow, 125190 Russia Lachine, PQ H8S 4H2 Canada Tel: +7-095-152-9698 • Fax: +7-095-152-4661 Tel: 514-633-7447 • Fax: 514-633-0770 www.module.ru www.gage-applied.com RSC #7502 @ www.industrial-embedded.com/catalogrsc RSC #7503 @ www.industrial-embedded.com/catalogrsc

Industrial Emdbedded Systems Resource Guide 2005 / 75 Systems: Industrial Ruggedized

Pentium M - SBC The APOLLO uses the Intel® Pentium® M processor to offer the best combination of high performance computing features with

Computing the lowest power dissipation. This single board computer can be fitted with the Intel® Pentium® M or Intel® Celeron M processors with speed options from 600MHz to 2.1GHz. The combination of Enhanced Intel SpeedStep® Technology and the Intel 855GME/ ICH4 chipset ensures that the board is ideal for compact systems with restricted ventilation and can be used to create very high performance fan-less systems. All these features are incorporated onto an industry standard EBX board with standard connectors for many of the I/O connections. The APOLLO includes a hot-swap CompactFlash (CF+) socket for use with both storage and network cards (Wi-Fi, Bluetooth and modems). The single PCI 2.2 compliant slot can be used to drive (up to) a 3 slot PCI riser card for compact system integration. FEATURES: The APOLLO is fitted with an ATMEL Trusted Platform Module n CPU options: Intel Pentium M (1.1 to 2.1GHz) and Intel Celeron M (TPM) device for use in applications which require a high level of (600MHz to 1.5GHz) software security and tight control of application code execution. n System memory: Up to 1024MB PC2700 DDR DIMM memory The board also includes a tamper detect input (which operates n Ethernet: Dual 10/100baseTx controllers (option for 1000baseT with or without main power applied) and a simple LCD character Gigabit Ethernet using Intel 82541PI controller) display interface for systems without a traditional VGA display. The n Video: Dual video controller - allows for two independent video APOLLO is ideally suited for low power, high density server racks, displays (CRT, LVDS) 1U and 2U systems with fan-less passive cooling as well as rugged, n USB: 6 x USB 2.0 (2 accessed via add-on module) secure computing installation. n FireWire: 2 x IEEE 1394a-2000 FireWire ports (100/200/400Mbps) n Serial ports: 2 x RS-232, 1 x RS-232/IrDA, 1 x RS-232/422/485 n Expansion: Single PCI Rev 2.2 expansion slot (with support for up to 3 PCI master cards via a riser card).

Arcom 7500 West 161st Street Overland Park, KS 66085 Tel: 913-549-1000 • Fax: 913-549-1002 www.arcom.com For more information, contact: [email protected]

RSC #7601 @ www.industrial-embedded.com/catalogrsc

76 / 2005 Industrial Emdbedded Systems Resource Guide Systems: Industrial ruggedized

Computing Rugged ATR Chassis The 714 Series is an aluminum chassis that utilizes the dip brazing fabrication process, which seals the enclosure and aids in its natu- ral convection to conduct heat away from the boards and power supply. Wedge lock guides secure the boards into the rack and provide a thermally conductive path for removing heat. Up to 135 watts can be expected from the power supply while the chassis is in a 50°C ambient environment. The power supply plugs directly into the backplane to eliminate the need for power cabling to FEATURES: the backplane. Accommodates VME64, VME64X, and cPCI back- n 3/4, 1, 1 Long and 1 1/2 size conduction cooled air transport rack planes. Custom backplane configurations are also available. chassis n Meets ARINC 404A/MIL-STD-91403 n Accommodates VME64, VME64X, and CompactPCI backplanes. Custom backplane configurations are also available n Backplane I/O breakout area and direct plug in power supply provides Carlo Gavazzi Computing Solutions a more rugged design by reducing the number of cables n Highest wattage available 10 Mupac Dr. n Thermal Simulation Model available Brockton, MA 02301 Tel: 508-588-6110 • Fax: 508-588-0498 www.gavazzi-computing.com For more information, contact: [email protected]

RSC #7701 @ www.industrial-embedded.com/catalogrsc

Computing Computing Systems: Industrial ruggedized Systems: Industrial ruggedized

GP3SSA 8555/8541 ProtoStar Design FEATURES: FEATURES: n First PowerQUICC Mini- ITX Embedded Computer -only $995 n Total Solution: Concept-Prototype-Manufacturing n Storage and Security Appliance for Freescale PowerQUICC III n Reduce Time-to-Market n 1394, IDE, PCI, dual mini-PCI, PMC, copper Gigabit ethernet n Rugged Design Experience n Off the shelf or custom integration with Manufacturing n “Building Block” Design Library n Open Uboot n Your Market Knowledge + Our Product Experience = SUCCESS!

Silicon Turnkey Express Vesta Technology, Inc. 749 Miner Road 11049 W. 44th Ave., Ste. 100 Highland Heights, OH 44143 Wheat Ridge, CO 80033 Tel: 440-461-4700 • Fax: 440-461-4329 Tel: 303-422-8088 ext 111 • Fax: 303-422-9800 www.SiliconTKx.com WWW.VESTATECH.COM RSC #7702 @ www.industrial-embedded.com/catalogrsc RSC #7703 @ www.industrial-embedded.com/catalogrsc

Industrial Emdbedded Systems Resource Guide 2005 / 77 Computing Computing Systems: Panel computers Systems: Programmable automation controller

MPL AG Computing

Unique Panel PC SoftPLC FEATURES: FEATURES: n Space saving, All-in-One Panel PC with 6.5"-19" display (up to Pentium M) n Embedded control software can deploy to various hardware n All around IP65 / NEMA4 aluminum or stainless steel housing n A-B like ladder logic/addressing; import/convert A-B logic n Silent, fanless, specially designed for rugged environments n User functions in C/Java via toolkit, such as e-mail, data logging n Long-term availability n Supports many I/O and motion devices, including our own offerings n Specific low-power designs n Embedded firewall, Web server, SSH/FTP servers, much more! n High flexibility and expansion capabilities MPL AG SoftPLC Corporation Täfernstr. 20 25603 Red Brangus CH-5405 Dättwil, Switzerland Spicewood, TX 78669 Tel: +41-56-483-34-34 • Fax: +41-56-493-30-20 Tel: 512-264-8390 • Fax: 512-264-8399 www.mpl.ch www.softplc.com RSC #7801 @ www.industrial-embedded.com/catalogrsc RSC #7802 @ www.industrial-embedded.com/catalogrsc

Systems: Components – Power Supplies

APS Power Supplies APS (Advanced Power Solutions), founded in 1982, delivers quality power products that are extremely cost-effective in today’s markets. APS offers a wide range of quality power products featuring: – High efficiency switching topologies – Local sales and engineering support – Far East manufacturing (ISO-9000/2000 compliant) FEATURES: – International safety approvals – Wide range of standard off-the-shelf products n 10-100 watt open frame power supplies with universal AC input and – Excellent custom design engineering services 1-4 outputs ranging from 2.5-48 VDC n 100-250 watt U-Frame power supplies with universal AC input and Please contact us today to discuss your specific requirement. 1-4 outputs ranging from 2.5-48 VDC n 50-800 watt industrially enclosed power supplies with universal AC input and 1-3 outputs ranging from 2.0-56 VDC n 150-600 watt industrial PC power supplies with universal AC input Advanced Power Solutions and ATX V2.2 compliant in standard and low-profile packages 7079 Commerce Circle n 5-100 watt desktop and wallmount power supplies with universal AC Pleasanton, CA 94588 input and 1-3 outputs ranging from 2.5-48 VDC Tel: 925-734-3060 • Fax: 925-460-5498 n 20-400 watt medically approved power supplies with universal AC www.advpower.com input and 1-3 outputs from 2.0-56 VDC For more information, contact: [email protected]

RSC #7803 @ www.industrial-embedded.com/catalogrsc

78 / 2005 Industrial Emdbedded Systems Resource Guide Computing Computing Systems: Components – Power Supplies Software: Linux

JRM Consultants Inc. Computing

BBDC Battery Power XP862 Power PC FEATURES: FEATURES: n Complete battery-based ATX power supply solution - in PC104 n Motorola MPC862 Controller with Linux pre-loaded n OEM drop-in supply for up to two 95 WHr Li-Ion battery packs n Motherboard and stock/custom daughter boards n Controller with integrated charger from AC or DC, wind/solar n Ports: Ethernet, wireless, PCMCIA, PCI, SCSI n Regulated DC output - complete ATX power - 3.3V, 5V, 12V n Digital Signal Processing co-processor n Real-time status reported via Serial / RS 232 to host system n Turnkey - HW, SW, custom manufacturing

OceanServer Technology, Inc. JRM Consultants Inc. 151 Martine Street PO Box 90310 Fall River, MA 02723 Santa Barbara, CA 93190 Tel: 508-678-0550 • Fax: 508-678-0552 Tel: 805-564-3119 • Fax: 805-962-0378 www.ocean-server.com www.jrmconsultants.com RSC #7901 @ www.industrial-embedded.com/catalogrsc RSC #7902 @ www.industrial-embedded.com/catalogrsc

Software: Real-time Operating System

RTX RTX is a high-performance, deterministic, real-time Windows solution that provides unprecedented control, saves developers time, reduces system costs, and speeds time to market. RTX enhances Windows’ universally adopted look/feel with fea- tures that enable real-time determinism, better control, and unmatched dependability. RTX application portability is simplified between various Windows operating systems through a compliant FEATURES: Win32 API. n Control Windows with deterministic memory management Free evaluation version at: https://eval.ardence.com/mes n Reduce system costs by eliminating the need for customized or Ardence is a global leader in developing software solutions that specialized platforms enhance the control, security, dependability, and management of n Increase performance through the smallest footprint in the market Windows. and sub-microsecond latencies n Improve reliability with support of standardized x86 HALs - HALx86 Ardence, Inc. as well as PIC and APIC with ACPI n Speed time to market through intuitive development tools that easily 266 Second Avenue integrate into standard Windows IDE Waltham, MA 2451 Tel: 800-334-8649 • Fax: 781-647-3999 www.ardence.com/embedded For more information, contact: [email protected]

RSC #79023 @ www.industrial-embedded.com/catalogrsc

Industrial Emdbedded Systems Resource Guide 2005 / 79 Human Interface: Technology

Converging functionality in embedded industrial control By Melissa Jones

Control systems and human-machine Newer advanced HMI display systems control, and interpretation of data from interfaces can be made more intelligent now support enhanced communication multi-tiered networks a relatively easy and tailored for the end user through between operators and equipment; they task when compared to the same task in clever use of FPGA technology to are more independent and control an system available only a few years ago. partition functionality, address varying increasing number of machines that I/O needs and customize systems to end interact with each other in an ever-evolving A next generation of industrial products user requirements. This article explores enterprise information network. As these will base itself around adaptable FPGA-based architecture strategies and display systems become more capable of embedded modules. These modules fit an innovative application for a high-end communicating over various networks, into multiple instruments within a line yacht control console. they not only support data exchange, of products designed to meet the local they also support remote monitoring and requirements of a device level network. Throughout our industry, virtually all upgrade, making it possible to introduce The technique of installing separate industrial I/O products and processes are self-managing and self-upgrading systems modules with different capabilities on a becoming smarter and more efficient. for greater efficiency. basic base product platform has been the The market drives this change, and one around for years, and the latest technology response is to reduce system complexity Convergence of the HMI and addition is in the inclusion of FPGA and enhance individual subcomponent industrial control systems technology. The basic architecture uses an capability. Another response is to drive In the last few years, as Supervisory FPGA that divides the system hardware more intelligence to the edge of the system Control And Data Acquisition (SCADA) into selectable combinations of functional or network so it is possible to communicate systems have expanded their influence blocks. Modules can be plugged together more rapidly and directly with specific on the market, the HMI displays have in different combinations to create subcomponents or units. One product increased their baseline capabilities. customized system product geared toward that is rapidly changing in the industrial One of the most notable changes is a specific application. To build a system, control market is Human-Machine- the use of microprocessors instead of an engineer simply selects the required Interface (HMI). HMI technology is on microcontrollers to control hardware, and modules to satisfy the needs of a specific a convergence course with industrial the use of high-end operating systems like application. The next generation of I/O computing. Increasingly, HMI devices Linux and Windows CE instead of small- technology will include controls embedded are using microprocessor-based solutions embedded operating systems. One benefit in FPGA configurable HMI displays. with modern commercial grade operating of this change in processing capability is Separate HMI display controllers and systems instead of microcontrollers with the concurrent addition of sophisticated industrial controllers, such as PLCs, are customized embedded operating systems. networking capabilities. therefore converging as the intelligence At the same time, the intelligence and functionality of both evolve toward a traditionally reserved for a host system is Many HMI displays are now very network single embedded system that delivers even moving out to the domain of distributed capable and easily act as a bridge between more versatile control capabilities. programmable controllers. This transfer the various layers in the network hierarchy. is occurring in Programmable Logic HMI displays are now capable of both It is easy to underestimate the significance Controllers (PLC) or Remote Terminal collecting and controlling local sensor of the relationship between complex Units (RTU). Increasingly, this new data and devices as well as receiving and software and hardware. As the complexity intelligence includes reconfigurable monitoring remote device level sensor data of a control system increases, there is hardware technology. More and more through industrial control networks and the a nonlinear relationship between the embedded industrial systems are replacing almost ubiquitous business level Ethernet amount of software or hardware needed to portions of system software in these TCP/IP. Bridging to Ethernet from existing manage the environment. The easiest way devices with hardware, which executes standards is critical in enabling designers to combat this trend is to develop methods upgradeable algorithms embedded in to connect both new and legacy devices to that abstract both the hardware and FPGA technology. a central control unit. The use of software software into larger and larger blocks of and hardware based industrial protocol fully tested hardware-based subsystems. conversion products make the transport, These highly optimized subsystems can

80 / 2005 Industrial Emdbedded Systems Resource Guide Human Interface: Technology be plugged together to form a baseline part of the design that has the greatest monolithic system. This framework is physical platform that improves the challenge – what some refer to as “the last therefore faster and easier to troubleshoot quality of the design while reducing the 10 percent of the design.” and tune to fit a specific application. time taken to implement the complete Having the ability to more discretely system. The entire point is to simplify the Understanding and modifying a framework define and modify the relationships and foundation design and integration process of component elements is faster and interactions between component elements so the design engineer can focus on the easier to understand and modify than a in a building-block approach greatly helps

anywhere onboard: in the captain’s cabin, the owner’s Sailor, you have the control cabin, in the saloon, on the flying bridge, and on the main bridge. One important feature of the system is to advise the To show a practical application of an industrial embedded passengers and crew, by sound and graphic signals, of any solution that extends far beyond the factory floor, a system malfunction such as excess liquid in the bilges, a manufacturer of luxury yachting control systems is integrating person overboard, or smoke in the engine room or any other the latest HMI displays and smart modular controls into their unsupervised area. line (see Figure 1). The use of such HMI displays greatly enhances Their solution uses a network of intelligent HMI displays safety and accident prevention, in part because specific positioned around the boat to receive data from all the onboard software programs can be adapted and applied to each sensors and systems. The system is simple and intuitive thanks individual system. This enables more detailed analysis and to information displayed with easily identifiable graphics reporting by each system with its respective sensors, and integrated with online help. Through the various displays, crew members know with greater accuracy where and when the crew can monitor and control a wide variety of important a problem occurs. onboard systems. The system also enables recording of anomalous events by The displays support a wide range of functions: engine room each HMI display, much like the black box aboard an aircraft. monitoring to check working status and engine efficiency, This permits more efficient analysis of what caused a problem video cameras, intrusion sensors, fuel and water tank level or breakdown in order to prevent recurrences. For security, sensors, radar, GPS, wind station, depth-finder, electrical data available on each monitor may only be handled through switchboard, bilge pump alarms, and air conditioning and insertion of a password. other controls and sensors. Each HMI display not only supplies all information transmitted by an array of onboard For management and monitoring of water systems, the crew sensors but also permits management of each individual can precisely read real-time values from sensors installed system using a built-in touch screen. An authorized operator and calibrated in various tanks and linked to an independent can switch on the air conditioning, modify lighting conditions system control module. The required minimum and maximum in various ship locations, or set a new waypoint on the plotter levels may be programmed for each tank as an added integrated into the system. An HMI display may be installed safety measure.

Figure 1

Industrial Emdbedded Systems Resource Guide 2005 / 81 Human Interface: Technology in determining project parameters that the timing and sequencing of analog data HMI displays is to include the capability satisfy the requirements of the system. collection, and customizable digital timer/ to support host control. An HMI display counters that support high-level sequenced with host control adds a completely new Although device-level network products collection of digital data from quadrature dimension to the types of tasks that it will continue to handle bridging and decoders, frequency counters, or custom can perform. This type of HMI display control functions, many designers will sight-dependent digital sensors. Figure 1 can now gather data locally, remotely want to enhance their project using shows how the inclusion of FPGA connecting to systemwide intelligent local the equivalent of a chip-level design technology in these new edge devices devices, and then process and display the environment. Instead of purchasing mimics the hardwired functionality of information to the operator. Thanks to individual controllers, a system designer original PLC devices. The difference integration and miniaturization, it is now can select FPGA-based modular hardware is that this hardwired programmable simpler to install and adapt in any control blocks that form a prebuilt solution. These functionality applies at the chip level system, be it a new one or one undergoing modules come preconfigured off the shelf instead of at the board or device level. an upgrade. and are actually optimized for specific applications within the factory. Such The future of enhanced The most intelligent HMI displays modules can support both bridging and intelligent display devices coming into the market today contain networked HMI control systems and let HMI displays enhance the function pre-installed IP designs in FPGA devices. designers complete prototypes quickly of existing I/O modules, PLCs, and These are well- tested, optimized designs and move directly into small and medium other automation devices by providing implemented for a specific application, production volumes using the same a visual representation of machine and they provide a hardware-based means platform. processes. These visual images inform to customize the system. The fundamental the operator about what the machine architecture of the FPGA-based product Increasing endpoint intelligence is doing. Substantially, each individual includes the processor and memory reduces complexity system or instrument appears graphically interfaces, and a set of basic system The number of devices in the typical on the HMI display, leaving little room control blocks. These basic system control control system is increasing at a high for doubt about interpretation. The HMI IP blocks provide a standard foundation rate, and one way of reducing the overall display makes machine control easier to for a second set of functional IP blocks complexity of the system is to transfer interpret and more efficient by enriching that implement an optimized application. intelligence out toward the device the operator experience. A current trend in When downloaded to the HMI as a endpoints. Inside these intelligent devices are small-embedded microcontroller modules that perform local control and enhance the capabilities of the software PLC. FPGA technology provides the means to make this happen. The use of hardware acceleration techniques to replace software algorithms is one obvious method of moving intelligence into the endpoint hardware.

Smarter I/O modules have included FPGA technology for some time, primarily serving as glue, interconnect, and control logic. Recently, however, the trend has become to move algorithms that traditionally have been the domain of software into hardware at the far edges of the control environment. Moving intelligence into the hardware endpoint devices makes them capable of handling many of the functions previously performed by a host controller. Advanced software is also moving down to the far edge of the control system; intelligent endpoint devices will perform much of the real-time embedded control locally, thereby reducing traffic on the local network fieldbus.

The types of intelligence moving into the FPGA on the endpoint devices include Figure 1

82 / 2005 Industrial Emdbedded Systems Resource Guide Human Interface: Technology

Figure 2 monolithic mix of soft IP, the IP design design. These benefits include the ability to the interaction between the combined transforms the system into a controller to fix design bugs in the FPGA hardware, devices in the system. customized to meet the requirements for upgrade a system in the field, or simply a specific application. swap out hardware functionality without Melissa Jones is a redesigning the physical board that founder and the current One example of an application where contains the FPGA. The latest trend when president of Ultimodule, the addition of FPGA technology to an using FPGA technology is to capture the Inc. in Sunnyvale, intelligent HMI display enriches the intelligence that originates in software California. Prior to operator’s experience is in the field of and transfer it as optimized hardware founding Ultimodule, video processing. By adding a video input algorithms to perform the optimized she spent seven years module to an HMI display, the FPGA in task at a higher speed. The embedded in marketing positions the HMI display can perform hardware- intelligence in a control system is migrating for Silicon Valley start-ups focused on accelerated video processing of the in two directions: out to the endpoint System-on-Chip technology and embedded digitized video stream. The combined IP devices and into the HMI Display. This IP. Before that, she spent 14 years with blocks form an image processing engine migration is having the greatest effect Philips Electronics (Semiconductors that implements in hardware most of the on the host controller that is supplying and Consumer Electronics) in sales algorithms required of a real-time vision the intelligence. In some ways, the host and marketing positions in the United system (see Figure 2). These algorithms controller is being split apart and moved States and Europe. Jones holds a BA in include programmable linear and nonlinear into the HMI Display and the control local organizational development from New filters, warping, edge, corner, segment, device and suggests that in the long-term, College of California and an MA in and curve detection, image segmentation, the SCADA network looks more and more organizational development from Norwich and the tracking of an object through a like a peer-to-peer environment and less University. sequence of images. and less like a master/slave environment. To learn more, contact Melissa at: Implementing IP using functional building The overall system is easier to comprehend Ultimodule, Inc. blocks is an excellent approach to system because the implementer constructs it 556 E. Weddell Drive, Suite 2 design because the reconfigurable FPGA from intelligent subsystems. This building Sunnyvale, CA 94089 has greater versatility than equivalent block approach makes troubleshooting Tel: 408-734-2300 components designed into a fixed single- and maintenance tasks easier, as well E-mail: [email protected] board level solution. as raising the overall reliability because Website: www.ultimodule.com the behavior of each subsystem is well- Conclusion known. Since the subsystem is previously The use of FPGA technology provides tested, the most rigorous testing applies numerous benefits to embedded system

Industrial Emdbedded Systems Resource Guide 2005 / 83 Human Interface: Technology

Using software-configurable processors in biometric applications By Philip Weaver and Fred Palma

Software-Configurable Processors (SCPs) are becoming an increasingly popular solution to challenging compute problems, spanning the gap between a general purpose processor and a dedicated digital signal processor. The benefits are especially clear for implementing biometric algorithms, which are complex and adaptive. This article describes an architecture for facial recognition, mapping C++ algorithms onto an SCP.

Humans detect and identify faces in a scene with little or no effort. However, building an automated system to accomplish this task has proved quite challenging due to the computational requirements of the algorithms. There has been a sharp Figure 1 increase in the public’s interest for face recognition systems that are both reliable The core technology of a face recognition they match those already stored in the and non-intrusive. Such systems could system consists of several steps outlined database. Each of these steps has their own make criminal identification, monitoring, in Figure 1. First, a stream of 2D camera set of data processing and computational and security access systems more input is processed into 3D surface data. requirements that challenge typical effective. Unlike any other biometric The 3D surface is then examined further embedded processors. technologies – such as fingerprint or iris to identify and measure facial features. scanning – face recognition systems can These measurements are then distilled Computing the 3D surface function in a real-time environment with into a biometric template. This template A4Vision’s proprietary hardware acquires little to no cooperation from the subject, is then compared with a database of surface data using structured light. A who may not even be aware that the previously enrolled templates. Faces pattern of light is projected onto the scanning is taking place. are recognized based on how closely subject’s face while a camera observes

Figure 2

84 / 2005 Industrial Emdbedded Systems Resource Guide Human Interface: Technology the pattern. The observed pattern is The fastest DSPs available may be able recognition, consider one portion of distorted, and this distortion depends to support the computations, but their the algorithm that computes a 3D facial on the individual geometry of the face. complex instruction sets typically require surface from a 2D video image using The pattern can also be projected using manual optimization of assembly code. structured light. This part of the algorithm wavelengths that are outside the visible The time required for this optimization solves for the x, y, and z coordinates light spectrum, which allows the image to dramatically slows down the development using three equations. The calculation is be acquired without the awareness of the process and must be repeated when the performed using Cramer’s Rule: subject. algorithm changes.

Because the distortion of the structured Software-configurable processor light pattern depends on the individual Stretch software-configurable processors Given the system geometry of the face, the 3D surface can deliver both flexibility for developing of equation: be computed based on the image stream algorithms and computational from the camera. First, image processing performance to them in real time. The SCP techniques are used to reduce noise and consists of a RISC core and configurable identify the projected pattern. Once the logic working together seamlessly as a pattern has been identified, the x, y, and single processor. Its programming model z coordinates of each point on the surface is entirely C and C++ based, thus allowing can be determined using trigonometry for rapid development of new algorithms. and linear algebra. The algorithm initially The configurable logic also allows new computes a cloud of points, and then links custom instructions to replace functions Then the points into a mesh. The reconstruction that otherwise require tens or hundreds of of the 3D surface is illustrated in Figure 2. operations to complete. This logic allows Additional information about the surface ordinary C or C++ code to be accelerated, and texture allows a realistic likeness to without manually optimizing assembly be recreated. Computing more points in code, to speeds that often exceed those the original cloud allows the system to attainable with standard DSPs. produce a more accurate surface, but also requires more calculations. The basis of SCP acceleration is instruction specialization, instruction Facial feature extraction and recognition pipelining, data bandwidth, and data Further analysis of the facial surface parallelism. The SCP compiler tools allows features such as the forehead, create a single instruction from a C or C++ The determinant D eyes, and nose to be extracted. Based function containing multiple operations. on the locations of these features, key The optimizing compiler then schedules is computed using points on the face are identified and the the new instructions together with the expansion by minors: distances between them measured. The standard instructions to minimize stalls in points chosen depend on rigid skeletal the processor. structures that typically do not change over time (as opposed to soft tissues that The performance benefits of the custom change as a person ages or fluctuates instructions are substantial, and are in body weight). These measurements multiplied when the instructions perform are stored as a biometric template in an multiple calculations in parallel. The SCP enrollment database. A face is identified architecture includes 128-bit wide register in real-time by matching the biometric files that can hold multiple data objects to template against this database. facilitate the parallel processing. a b Meeting computational requirements Because the tool chain used to program = ad − bc Developing an embedded system to the SCP requires no assembly language c d perform real-time 3D face recognition programming or logic design, changes to the algorithm can be implemented is challenging. As algorithms continue Representative C code is shown in by changing the C or C++ code that to evolve, development using a high- Listing 1. A 3x4 matrix m holds the implements them. Algorithms can level language such as C is desirable. coefficients of the three equations. The therefore be developed and refined in General purpose processors such as the minors array holds minor determinants a high-level language and accelerated popular RISC-based cores widely used used to compute D, Dx, Dy, and Dz as easily. This ability makes the SCP a unique for embedded systems generally have shown previously. From these values x, y, platform for biometric algorithms. high-quality C and C++ tools well suited and z are determined and placed in the to algorithm development, but are unable vector X. to perform the demanding computational Computing the 3D coordinates To illustrate how the SCP architecture tasks demanded by the algorithms. delivers high performance for face

Industrial Emdbedded Systems Resource Guide 2005 / 85 Human Interface: Technology // Non-accelerated version The combination of data streaming identification. 3D face recognition // Compute partial determinants of and wide data registers provides a high systems can meet this need, but one of the A needed for Cramer’s rule. minors[0][0] = (( m[1][1]*m[2][2]- bandwidth path to the configurable barriers to bringing such systems to larger m[2][1]*m[1][2] )>>31); logic that implements the instructions. markets is the requirement for a flexible minors[1][0] = (( m[2][1]*m[0][2]- Seven specialized instructions have been and powerful embedded architecture. m[0][1]*m[2][2] )>>31); minors[2][0] = (( m[0][1]*m[1][2]- crafted to replace the large number of The software-configurable processor, m[1][1]*m[0][2] )>>31); instructions a compiler would create however, has been demonstrated to minors[0][1] = (( m[1][2]*m[2][0]- to implement the code in Listing 1. For enable this biometric application to move m[2][2]*m[1][0] )>>31); this computationally-intensive set of from systems into minors[1][1] = (( m[2][2]*m[0][0]- m[0][2]*m[2][0] )>>31); operations, the Stretch SCP processor an embedded platform. Furthermore, minors[2][1] = (( m[0][2]*m[1][0]- reduced the number of compute cycles by the SCP will continue to be used to m[1][2]*m[0][0] )>>31); a factor of nine. accelerate the compute intensive section minors[0][2] = (( m[1][0]*m[2][1]- to enable different viewing angles for m[2][0]*m[1][1] )>>31); // Accelerated version higher accuracy in real-life deployments minors[1][2] = (( m[2][0]*m[0][1]- // Load three rows of matrix data m[0][0]*m[2][1] )>>31); WRGET0INIT(0, (void *)&m[0]); as well as new algorithms for improved minors[2][2] = (( m[0][0]*m[1][1]- WRGET1INIT(0, (void *)&m[1]); identification. m[1][0]*m[0][1] )>>31); WRGET2INIT(0, (void *)&m[2]); // Compute D WRAGET0I(&row1, 16); Philip A. Weaver is a D = ((m[0][0]* minors[0][0]+m[1][0]* WRAGET1I(&row2, 16); minors[1][0]+ m[2][0]* minors[2][0])>>26); WRAGET2I(&row3, 16); Member of Technical // Compute 1/D // Compute intermediate values for determinants Staff, Applications invD = (1<>26); EI_COMPUTE_ MINORS _1(row1, row2, technical experience x = ((Dx * invD)>>SCALEFACTOR); row3, &unused, &minorsForDet2_2); in software, hardware, // Compute Dy and y EI_COMPUTE_ MINORS _2(row1, and mechanical systems. He was vice Dy = ( (m[0][3]* minors[0][1]+m[1][3]* row2, row3, &minorsForDet1to3_1, president of Engineering for Luidia, minors[1][1]+ m[2][3]* minors[2][1])>>26); &minorsForDet3_2); y = ((Dy*invD)>>SCALEFACTOR); // Compute determinants Inc., and has held senior engineering // Compute Dz and z EI_COMPUTE_DET(mc0, and management roles at Electronics Dz = ((m[0][3]*minors[0][2]+m[1][3]*mino minorsForDetAndDet1_2, &D); for Imaging, Maxoptix Corporation and rs[1][2]+ m[2][3]*minors[2][2])>>26); EI_COMPUTE_DET(minorsDet1to3_1, Quantum Corporation. z = ((Dz*invD)>>SCALEFACTOR); minorsForDetAndDet1_2, &Dx); // Store coordinate in vector. EI_COMPUTE_DET(minorsForDet1to3_ X[0] = x; 1, minorsForDet2_2, &Dy); To learn more, contact Philip at: X[1] = y; EI_COMPUTE_DET(minorsForDet1to3_ Stretch Inc. X[2] = z; 1, minorsForDet3_2, &Dz); 777 E. Middlefield Road // Invert det for multiply Listing 1 EI_INVERT(D); Mountain View, CA 94043 EI_INPUT_DET123(Dx, Dy, Dz); Tel: 650-864-2707 In Listing 2, the matrix computations have EI_COMPUTE_X(&xyz); E-mail: [email protected] been converted to use custom instructions. // Store result Website: www.stretchinc.com WRPUTINIT(0, X); The custom instructions are written in WRPUTI(xyz, 12); Stretch C. Because Stretch C supports all WRPUTFLUSH(); Fred Palma is Vice of the data types and operators in C, the President of Engineering custom instructions are specified using Listing 2 for A4Vision. He code nearly identical to that shown in has extensive senior This example demonstrates the process Listing 1. The code that calculates x, y, management experience of accelerating one part of the face and z now uses special macros to load in leading development recognition algorithm using the software- 128-bit Wide Registers (WRs), perform efforts for several configurable processor. This process can the various parts of the calculation, and commercial applications be repeated throughout this or any other store the results to the vector X. In the in card base identification, access control, application. The resulting acceleration code listing, the WRGETxINIT() macros and card personalization systems. He was allows computationally-intensive initialize streams of data to be loaded a director of Engineering and Product algorithms to run at high speeds using into the WRs, while the WRAGETxI Management for Diebold Card Systems, only C / C++ software programming each load a row of data from memory and vice president of Information Systems methods. for processing. Custom instructions for Oberthur Card Systems, US. then compute the minors and use them Conclusion To learn more, contact Fred at: to compute the determinants. Custom The ability to deliver a face recognition A4Vision instructions are then used to perform the system unaffected by lighting conditions, 840 W. California Avenue, Suite 200 divide and multiplies that determine x, y, background colors, facial hair, or Sunnyvale, CA 94086 and z. Finally, the WRPUTINIT, WRPUT, makeup goes a long way toward meeting Tel: 408-329-4566 and WRPUTFLUSH instructions write public demands for better monitoring E-mail: [email protected] the resulting vector X to memory. systems, security systems, and criminal Website: www.a4vision.com

86 / 2005 Industrial Emdbedded Systems Resource Guide Software: Interface Application Human Interface Human

ThinManager 2.6 ACP’s ThinManager lets you configure, maintain, upgrade and replace Thin Clients quickly and efficiently. Its intuitive interface provides at-a-glance management of all connected ACP Enabled Thin Clients. ThinManager is also a client enabling technology, so each connected client has the same software, assuring uniformity of operation across a variety of Thin Client models.

FEATURES: n Thin Client hardware that never becomes obsolete n Automatic configuration of new clients n Redundancy and Failover n Server Load Balancing n Support for all major touch screens n Proven installations with all HMI and SCADA software ACP 4080 McGinnis Ferry Road Alpharetta, GA 30005 Tel: 678-990-0945 • Fax: 678-990-0951 www.thinmanager.com For more information, contact: [email protected]

RSC #8701 @ www.industrial-embedded.com/catalogrsc

Human Interface Human Interface Software: Interface Application Format: Touch Screen

EasyCODE V7.5 Trilogy Touch LCDs FEATURES: FEATURES: n EasyCODE makes software development exciting and enjoyable! n Trilogy Touch - desk top and chassis LCD – at the best price n Logical, streamlined programming using structured diagrams n Perfect for project maintenance, reverse-engineering, etc. n Supports all major programming languages (C/C++, Java, etc.) n Try it on one of your projects and see the difference!

EasyCODE Software Trilogy Touch Technology 1628 Old Apex Road 1 New Hampshire Avenue Suite 125 Cary, NC 27513 Portsmouth, NH 03801 Tel: 919-468-7256 • Fax: 919-468-6374 Tel: 800-360-9019 • Fax: 603-929-7411 www.easycode-software.com www.trilogytouch.com RSC #8702 @ www.industrial-embedded.com/catalogrsc RSC #8703 @ www.industrial-embedded.com/catalogrsc

Industrial Emdbedded Systems Resource Guide 2005 / 87 Sensors/Control: Technology

Combining a hardware neural network with a powerful automotive MCU for powertrain applications By Dr. Paul Neil

One solution to adaptive control problems is artificial neural networks. This article discusses an application in automotive control using a system-on- chip implementation.

Artificial Neural Networks (ANNs) are rapidly gaining acceptance within the automotive design community as a suitable solution to the types of complex, non-linear problems typically found in powertrain applications. The recent, high profile deployment of an ANN in the 2005 Aston Martin DB9[1] has boosted interest in the underlying technology. However, there are still some perceived barriers to the widespread adoption of ANN Figure 1 technology. These barriers include: of the SIMD architecture optimized to discrete version of the Vindax processor execute a version of the Self-Organizing core with the Powertrain Starter Kit (PSK) n High performance overhead incurred Map (SOM)[2] neural network. A block from Infineon Technologies. The PSK is in software implementation diagram of the Vindax processor core is a reference design for a complete engine n The corresponding cost of a dedicated shown in Figure 1. During the application control system based on the TriCore Microcontroller Unit (MCU) development phase, Axeon supports its embedded microcontroller architecture. n The lack of a standard set of tools and customers by delivering standard tools The starter kit package comprises of a methods to support the development and methods via the Vindax Development hardware platform capable of controlling of ANN solutions System (VDS). up to a 12 cylinder gasoline engine and a Axeon Ltd. is a pioneer in the use of torque converter based transmission, and Axeon has recently completed the first hardware neural networks within the is shown in Figure 2. The hardware integration of a automotive space. Its patented Vindax is designed to be used in-vehicle, technology overcomes and has multiple sensors and these barriers by actuator interfaces to allow providing hardware direct connection to the majority support for ANN of engines from most vehicle algorithms in manufacturers. The hardware the form of a also provides expansion licensable IP core, possibilities for further input thus offloading signal acquisition interfaces, much of the output driver interfaces, and computation from future processor expansion. the MCU, and An FPGA implementation of enabling complex a 64-element processor array problems to be was developed for this integration. solved using a low- This FPGA also houses the array cost co-processor controller, which sequences the solution. The IP core is a variant Figure 2

88 / 2005 Industrial Emdbedded Systems Resource Guide Sensors/Control: Technology execution of the SOM algorithm on function of a given sensor, allowing with major semiconductor suppliers the processing elements, and transfers the performance of the sensor to be such as Infineon, Axeon is looking to data to and from the TriCore system. monitored over time, with the ability accelerate the deployment of this powerful Mounted on a small daughtercard (shown to switch in a back-up (should the technique. in Figure 3), the FPGA implementation sensor fail), or deletion of the sensor is interfaced to the TriCore external bus. completely, providing a significant Footnotes: [1] “Aston Martin DB9 is High-Tech Axeon also provides a library of low- cost to the automotive OEM. Under the Hood,” Mike Thomas, Ford level software routines for the TriCore, n High performance classification. Communications Network, August 2004. which allow efficient access to the The increasingly stringent regulatory [2] “Self Organizing Maps,” T Kohonen, Springer-Verlag, Germany, 1995. Vindax firmware routines and hardware environment for emissions has resources. In addition to providing easy dramatically increased the required Dr. Paul Neil is the access to mission mode Vindax firmware CPU bandwidth to implement Director of North operations such as classification, adaptive effective misfire detection. Vindax has American automotive classification, training, and map context been used on a wide variety of high sales for Axeon Ltd. switching, the low level interface cylinder- count engines to improve Paul’s background is routines also provide access to Vindax detection performance, and thus meet in embedded processor housekeeping operations. The 64-way emission requirements. and System-on-Chip parallelism offered by this implementation n Change detection. Here, Vindax is (SoC) design, including provides massive bandwidth. For a used to create a model of normal design efforts at complex powertrain problem with 16 system operation. Statistical metrics companies such as Vision Group, ARM, input parameters, in excess of 360,000 can then be applied to the output of Pixelfusion, and Siroyan. Paul holds a classifications per second have been that model to determine whether, BSc in computer science and electronics achieved using this system. In addition, the in mission mode, an anomaly has and a PhD in high-level synthesis. firmware provided for the processor array taken place. supports true application multitasking. To learn more, contact Paul at: This raw performance permits the most In many instances, these techniques Axeon Ltd. challenging powertrain applications are used in combination to improve the Davidson House, Science to be addressed directly without imposing performance, accuracy, and reliability of & Technology Park significant additional load on the TriCore. the neural solution. Aberdeen, Scotland AB22 8GT United Kingdom As noted previously, the ANN paradigm This system can be used to address the Tel: +44-(0)-1224-338383 is now being accepted by the automotive following classes of problems: E-mail: [email protected] design community. By offering an Website: www.axeon.com n Development of virtual sensors. architecture to standardize on, the tools to Here, Vindax models the transfer support that architecture, and by working

Figure 3

Industrial Emdbedded Systems Resource Guide 2005 / 89 Sensors/Control: Appication

Open architecture PAC technology drives undersea remotely operated vehicles By Chris Ward

Remotely Operated Vehicles (ROVs) working with offshore oil rigs C&M often uses PC/104 products with the familiar PC operate in one of the most difficult environments on earth. Dealing architecture. These are industrial, virtually immune to vibration with the massive pressures found 5 kilometers underwater is only and small in size. On the plus side, PC/104 stacks are extremely part of the challenge. Weather is often atrocious and subsea rugged. The cost of the boards is low, so supplying built and tested visibility is regularly only a few inches. Surface waves can throw a stacks as spares is commercially viable. The physical size allows 5-ton machine around like a toy. Currents can exceed 3 knots, and smaller diameter housings, which are more suited to the high operators have to navigate around rigging cables, scaffolding, and pressures of deep water, to be used. There are hundreds of PC/104 seabed debris. Engineers routinely work with 5kV voltages, mV manufacturers – the quality and range of products is superb. The signals, digital and analog electronics, computers and software, PC/104 implementation does have a few drawbacks. Dismantling hydraulics, mechanics, fiber optics, acoustics, electromagnetics, a PC/104 stack needs a lot of care, and ribbon cables have to be X-ray equipment – technically challenging even without the kept short to prevent noise issues. offshore environment. On the surface ship, a control panel of switches and joysticks Tasks for ROVs can be simple or complex. Typical construction connected directly to a controller communicate with a network of work can include positioning large structures that are lowered onto one or more subsea controllers up to 5 kilometers away through the seabed by crane barges. Or with special tooling, the ROVs fiber optic links, bringing video, telemetry, Ethernet, and analog can pick up the ends of pipelines and align them to valves on or digital sonar data back to the operators. the structures, inserting gaskets, nuts and bolts before tightening them to preset torques. They operate valves and other machinery One essential hardware feature for ROV controllers is with incredible dexterity using manipulators controlled from the support for Flash disk, enabling system updates that can be surface. On the same mission they may use brute force to ram a e-mailed out and dropped straight onto the networked system by 20-ton pile into position, creating internal shock loads that can the operators, or linked in via satellite from C&M headquarters quickly destroy poorly designed electronics and hardware. in the UK. Although products may change over time, new drivers can be created in a few hours, if necessary. In the past, divers carried out much of this work, but commercial diving rarely goes below 400 meters. In the hunt for new oil Advantages of open architecture PAC reserves, depths are now 3,000 to 5,000 meters or more, calling Software needs to be compatible with the widest range of for a new breed of ROV. This presents a range of challenges for systems. For control, C&M uses open architecture Programmable electronics in the ROV and shipboard control system. Automation Controller (PAC) software from SoftPLC Corporation. SoftPLC control software allows C&M to quickly Carefully selected technology create applications using ladder programming. It has a range of Designers of equipment for underwater use have to be very careful communications options including RS-232/RS-485 and Ethernet in their selection of technology. Companies like C&M Group Ltd. and a peer-to-peer communications link can be set up to reliably (C&M) have built up a niche market where complete ROV control transfer a large amount of data with a single instruction. SoftPLC systems can be designed, built, and delivered in just a few weeks, also has a toolkit that allows developers to easily tie in their own with a control system adapted to the client’s specific needs. The instructions written in C for time critical situations or unusual industry is highly competitive, and ROVs must work perfectly algorithms. C&M has built a library of modules both for hardware while still being cost-effective and designed and built quickly. interfaces and applications such as notch filters, saving and Vehicles, operator consoles and tooling systems can have a life reading set-up files, and interfaces to intelligent sensors. Most of in excess of 10 years, so reliability and long-term support cannot these routines were written purely to optimize existing SoftPLC be compromised. Systems must be able to accommodate new instructions. It is an extremely versatile language. This alone is not tools and technologies over time, and clients need the ability to enough to justify using this product, but it has two other superb expand systems for unforeseen demands. Regulatory compliance qualities. Because it’s simple, ladder can be read and understood is mandatory and ease-of-use and diagnosis along with multi- by other programmers. Trying to get to grips with someone else’s language support is critical. Onsite technical and spares logistical “C” code is usually a nightmare, but a well-written ladder program support must be provided. can be figured out very quickly.

90 / 2005 Industrial Emdbedded Systems Resource Guide Sensors/Control: Application

The biggest advantage of all over custom code is that SoftPLC blend of reliability, performance, price, and most importantly, programs can be monitored and changed in real time. Final customer satisfaction. development and bug finding can be carried out live while the equipment is being operated. The reduction in development The use of SoftPLC open architecture programmable automation and support time, and thus cost, from this simple technique is controller software and PC/104 hardware provides C&M Group incredible. The client can watch during commissioning and with low-cost and fast development time solutions. C&M is at the provide valuable feedback on what works well and what should leading edge of development but has the reliability of established be changed, and C&M programmers can implement live changes designs. Clients can modify systems or engage with C&M or to a system while in-water function testing is going on without others to do it for them. The resulting control systems are simple, going through the compile and download steps required by long-lived, and supportable with a built-in future upgrade path. other products. Chris Ward is Advanced Technology Director of the C&M Group The SoftPLC software is not hardware dependent, so it can be Ltd. He has worked in the underwater industry for more than 25 ported to a wide variety of hardware, with only I/O drivers needing years developing electronics and software for many of today’s to change. C&M has used SoftPLC on PC/104, VMEbus, and ROVs and underwater tooling systems. Designs have ranged from STDbus legacy equipment without any problems. It brings control 8” inspection ROV control systems through to 1,000-ton salvage developers years of proven code and reliability. For low volume systems. C&M Group is now expanding their flexible control and unique products in a highly competitive market, where future applications into other hazardous area environments such as changes and upgrades will be a way of life, this solution is hard Zone 1 products and equipment for the atomic energy industry. to beat. To learn more, contact Chris at: Or contact: Operator displays are critical, not just for monitoring conditions C&M Group Ltd. SoftPLC Corporation and navigation but for diagnostics and fault handling/backup 5-19 Holland St. 25603 Red Brangus Drive systems. The typical OEM solution is a team of programmers Aberdeen, Scotland AB25 3UJ Spicewood, TX 78669 locked away developing code in Visual Basic or C++. However, United Kingdom Tel: 1-800-SOFTPLC one-off projects can’t justify this expense or the potential for Tel: +44-1224-625928 or 512-264-8390 costly mistakes. Modern PC-based HMI/SCADA products are E-mail: [email protected] E-mail: [email protected] incredibly powerful. With SoftPLC, any HMI/SCADA product Website: www.c-m-group.com Website: www.softplc.com is an option so customer standards can be followed, or the best product can be selected for each project. Sensors/Control Better fit for operator requirements Detection: Electromechanical C&M’s ROV control systems are built using a common procedure. Starting with the client’s specification, reviews are held to work out what the client actually needs. The system structure is created and a first pass with mock up screens is designed and submitted Printed Circuits Corp. for approval. A programmer then sits down with the client and modifies the screens and operations to their preference. None of this affects the underlying design, but the client buys into the product through helping to develop an interface that suits their purposes. Designers then tie the dynamic functions into the SoftPLC by linking readouts or switches to an input or output on SoftPLC through a drop down list of I/O created originally in Excel. The software is then loaded to the newly built hardware, debugged, and tested. PCB/PCBA's Fab/EMS

The real test starts in the field, seeing how the system works on FEATURES: the back of a heaving vessel in a force 5 with high humidity, n PCB design/layout and photoplot services, NPI concept-finish vibration, shock loads, and client operators and management. n EMS: Proto-Production, PTH/SMT, hand and auto assembly HD SMT A programmer and an electronics engineer are sent out with the n SS/DS and high density MLB to 12 layers FR4, 4000 series lam system for commissioning. While the operator works with the n SS/DS/MLB, LPI, SMOBC, Electroless Nickel/Immersion Gold system, the engineer is monitoring and adjusting the system to n Box build, repair and test services; data transfer via Web/EM make it work in the real world – with feedback and language translations supplied by the people who actually use the system Printed Circuits Corp. and diagnostics that cope with the reality of the situation. 4467 Park Drive, Suite E C&M systems are open architecture, and operators can be trained Norcross, GA 30093 not only to use the system but how to modify it. By letting trained Tel: 770-638-8658 • Fax: 770-638-8659 operators change the interface to suit their needs without letting www.pcc-i.com them into the application core, it very quickly becomes their RSC #9101 @ www.industrial-embedded.com/catalogrsc system. The use of open architecture technology gives the best

Industrial Emdbedded Systems Resource Guide 2005 / 91 Sensors/Control Sensors/Control Control: Analog Control: Digital

SICK, Inc. Sensors/Control

UE4100 Remote I/O VXI Controllers FEATURES: FEATURES: n ProfiSAFE Interoperability with .GSD file integration n V151, V152, V154: PowerPC based slot-0 controllers n Field Signal Acquisition to Category 4 in accord. EN954-1 n V153: High-performance Pentium 4-based slot-0 controller n Eight Dual Channel, SIL3 (accord. IEC 61508) Safety Inputs n Two PMC card options for SCSI, IEEE 488, Ethernet, USB, etc. n Network Monitoring and Configuration of SICK Safety Devices n V15x: Convert VME controllers into slot-0 VXI controllers n IP67 Enclosure using Quick Disconnect (M12,Mini) Technology n FOXI: PCI host and controllers via 10 MB/s fiber-optics

SICK, Inc. KineticSystems Company, LLC 6900 W 110 St 900 North State Street Minneapolis, MN 55438 Lockport, IL 60441 Tel: 952-941-6780 • Fax: 952-941-9287 Tel: 815-838-0005 • Fax: 815-838-4424 www.sickusa.com www.kscorp.com RSC #9201 @ www.industrial-embedded.com/catalogrsc RSC #9202 @ www.industrial-embedded.com/catalogrsc

Sensors/Control Sensors/Control Control: Digital Control: Digital

4I68 Anything I/O Performax USB 2.0 FEATURES: FEATURES: n FPGA Based I/O Card n USB 2.0 stepper motion control for Windows n 400K gate Xilinx Spartan3 FPGA n Wide range of models from single axis to multi-axis control n Bus mastering PCI bridge n Integrated stepper controller/driver available n 72 I/O bits, LVDS supported on one connector n Low cost (example: 4 axis control for only $395 single quantity) n Price for the 4I68 in 100's is $153 n Easy-to-use language with sample VB6, VC++, Labview

Mesa Electronics Arcus Technology, Inc. 4175 Lakeside Drive, Suite 100 4160 Technology Drive Richmond, CA 94806 Fremont, CA 94538 Tel: 510-223-9272 • Fax: 510-223-9585 Tel: 510-490-4303 • Fax: 510-405-2073 www.mesanet.com www.arcus-technology.com RSC #9203 @ www.industrial-embedded.com/catalogrsc RSC #9204 @ www.industrial-embedded.com/catalogrsc

92 / 2005 Industrial Emdbedded Systems Resource Guide Sensors/Control Sensors/Control Control: Digital Control: Digital Sensors/Control Sensors/Control

5.7" Touch Screen Elite-2000 ePLC FEATURES: FEATURES: n Eclipse Hitech Touch Screen with five programmable function keys n Up to 64 in/64 out, triac or transistor fused outputs n 5.7" 256 color or 16 shade grey screen with analog touch n Up to 16 analog I/O-voltage, current, millivolt, temperature n Free software with online simulator and cross reference n RS232/RS485 communications ModBus RTU and/or ASCII n 4 Meg program capacity with 512K recipe storage n 2 amp continuous rating/8-10 amp inrush capabilities n Five programmable function keys, 75,000 hour backlight tube life n Seven-year warranty, free programming software with simulator

Entertron Industries Entertron Industries 3857 Orangeport Road 3857 Orangeport Road Gasport, NY 14067 Gasport, NY 14067 Tel: 716-772-7216 • Fax: 716-772-2604 Tel: 716-772-7216 • Fax: 716-772-2604 www.entertron.com www.entertron.com RSC #9301 @ www.industrial-embedded.com/catalogrsc RSC #9302 @ www.industrial-embedded.com/catalogrsc

Sensors/Control Sensors/Control Control: Digital Control: Digital

SK1600-RIC ePLC Smart-PAK Series FEATURES: FEATURES: n Up to 32 in/24 out digital I/O and 4 analog I/O n Micro ePLC with up to 16/12 digital I/O and 8/4 analog I/O n AC or DC inputs/8 amp form C relay outputs; 12 VDC PS n 12 VDC or 24 VDC power, DC inputs/7 amp relay outputs n RS232 and RS485 ports - ModBus RTU and/or ASCII communications n 12-bit analog includes voltage, current, millivolt, temperature n Free programming software with simulator n RS232 and RS485 with ModBus RTU and/or ASCII communications n Five-year warranty; UL recognized optional n Five-year warranty, free Ladder Logic software with simulator

Entertron Industries Entertron Industries 3857 Orangeport Road 3857 Orangeport Road Gasport, NY 14067 Gasport, NY 14067 Tel: 716-772-7216 • Fax: 716-772-2604 Tel: 716-772-7216 • Fax: 716-772-2604 www.entertron.com www.entertron.com RSC #9303 @ www.industrial-embedded.com/catalogrsc RSC #9304 @ www.industrial-embedded.com/catalogrsc

Industrial Emdbedded Systems Resource Guide 2005 / 93 Storage: Technology

Software architecture for managing resident flash By Robert Krantz

Solid-state storage is a key element of today’s industrial control systems. Managing flash presents some unique issues for embedded designers, including write speed, write protection, and life cycle. In this article, an approach for a software management system overseeing flash is given, addressing these points along with tips to optimize performance and reliability.

Recent product innovations and larger flash arrays are placing greater demands on embedded systems, especially in the area of managing flash memory. The previous approach of file systems directly Figure 1 accessing flash is being supplanted by dedicated flash media management Current design requirements call for but requires a desktop driver to exchange software, which offers more effective flash flexibility in the choice of operating data with other systems. disk access along with a host of additional system, which is driven by the file flash-optimization functions including: system’s elevated role. File systems were While flash memory management not needed in most embedded designs of can be performed by the file system, n Multithreading support the past, but they are becoming a necessity the functionality will be limited. The n Wear leveling operations in today’s higher-capacity devices. complexity of ever-larger flash arrays n Improved handling of write errors has driven the development of separate n Background compaction routines The high level of interdependence flash media management software that n Safer method for reclaiming memory between the operating system and file resides between the file system and the space system is due to the specific requirements flash media. Figure 1 shows the system n Portability to all flash types and for how each can work with the other. A architecture and flash media manager. This operating systems file system that comes “bundled” with an dedicated media management solution n A standard block device driver interface operating system is usually insufficient for provides a simple, flexible interface, and to the flash disk embedded designs and must be replaced enables greater optimization of flash with a specialized flash file system. functions with specific devices. This article is a general discussion of the properties of flash media management. Of the many file systems available today, Though the use of standard block device In the case of specific examples, they are each has its own unique attributes that driver calls, the media manager presents drawn from the performance of Datalight can dramatically affect performance, the flash array to the file system as Inc.’s FlashFX Pro. data reliability, and data exchange. This traditional disk media – even if the flash variety of features and functions leads part type is changed. This feature gives Foundational elements to trade offs. For example, most FAT- developers the tremendous advantage of Concerns about embedded development based systems (the default format for file being able to change in the flash media and flash media management are directly systems in most commercial operating part without having to alter other system affected by the selection of key system systems) offer ease of data exchange but code. The use of block device drivers also elements: the operating system, the file are weak on data reliability. Conversely, frees developers to select a file system system, the media manager, and the a file system designed to prevent data based upon design parameters (such as flash part. loss, such as Datalight’s Reliance, offers FAT for compatibility or a transaction- excellent performance and data reliability, based system for reliability).

94 / 2005 Industrial Emdbedded Systems Resource Guide Storage: Technology

Physical flash media, available from many manufacturers, is specified by “part” numbers that indicate overall size, block size, and other specifications of these parts. Flash parts are sorted according to the types, NAND and NOR, which are significantly different.

NOR media appears much like RAM in a system and is read by expedient parallel access. NAND media is read by serial interface, one page at a time and requires rigorous error detection and correction, due to its high occurrence of single-bit Figure 3 errors. These issues are offset by the significantly lower cost of NAND as process is temporarily suspended to allow specify the number of spare sectors to be compared to NOR. immediate processing of the read request. available for use during failures. While The read latency, practically eliminated, is this allocation reduces memory available An optimized solution may include both reduced by three orders of magnitude. for other operations, it extends the media’s types, with a large area of NAND for lifespan. affordable data storage and the faster Write operation safeguards reading, slower writing NOR used for Data written to flash memory does not As sectors continue to fail, write and erase code execution. In such a case, a media overwrite existing data. Instead, it is operations can take longer and longer manager will provide transparent access placed into unused space that has been while the system seeks for usable media. to both types of flash media. cleared for new data. As a flash array is When all spare sectors are expended and generally full of both valid and discarded failures exceed a specified number, the Multithreading for speed data, a prescribed amount of this dedicated entire flash disk is placed into a write- With single-thread processing, each “buffer” space must be established during protected state to ensure that no data process must be completed before another initial programming. is lost. can begin. Multithreading systems can suspend one process to allow performance Progress of the write transaction is Compaction: reclaiming space of another – then the suspended monitored through recording the location When content is discarded, it is immediately marked as obsolete but continues to occupy space on the flash media. The compaction process reclaims this space by identifying a block containing discarded data, copying any valid data to a new location, then erasing the block for reuse in subsequent write operations. In addition to freeing up memory, compaction also helps to combine non-contiguous data – similar to other disk defragmentations.

During compaction, the original valid data is kept safely intact until it is verified as Figure 2 correctly copied to the new location. In case of interruption or error, the original process may be resumed. For example, and state of the new data. When all new data remains as a backup. multithreading enables a PDA to suspend data is successfully written into the flash Compaction can occur in the middle of media playback in order to handle an array, the allocation table is updated and a write operation as seen in Figure 4. incoming phone call. the old data is discarded. If an error or This space-freeing task could have interruption occurs, the original data been accomplished earlier to avoid this Figure 2 shows the read latency in a remains intact with no loss of data. single thread system as the low-priority interruption. maintenance operation must be completed When the flash media management This interruption of the writing process before the high-priority read request can software encounters a write error, it can be avoided by running background be processed. needs to mark the error-generating block compactions during system idle time as unusable. It then attempts to perform as directed by the media management Figure 3 shows the same situation in a the write operation at another location software. Figure 5 shows how to avoid multithreading system. The maintenance within the flash disk. Developers can

Industrial Emdbedded Systems Resource Guide 2005 / 95 Storage: Technology and poorly designed wear leveling algorithms. FlashFX Pro avoids deadlocks by performing compaction operations during idle times so cleared space is available before the write operation occurs.

Wear leveling routines may attempt unneeded compactions that do not create any new free space. These have a negative effect on system performance and power consumption. An ideal solution will allow programmers to limit the number of extra Figure 4 compactions performed. The bottom line Development of an embedded solution involves many complex decisions about highly interdependent elements. However, many of these decisions can be simplified by use of a flash memory management system that can eliminate the complexity of interfacing with the flash media part.

A dedicated flash media manager also integrates several functions that optimize use of the media. These include flash-optimized read, write and erase Figure 5 algorithms, wear leveling routines, bad block management. Advanced solutions delays by background compaction. The drives an algorithm that determines the will also include multithreading support controls for this function in FlashFX optimal location for each write operation. and background compaction. Pro can accelerate apparent write speed without additional battery usage. Areas containing fixed content, such as Ideally, embedded developers will be able binary code, are generally defined as to select flash media management software Wear leveling: maximizing lifespan static regions. If such content remains in independent of the file system in order to A block of flash memory will eventually the same location, then most of the rewrite meet their customer driven requirements fail from the cumulative “wear” of erasing. capacity of that area remains unused. for performance and reliability. Ideally, all blocks on the array will remain functional for about the same amount of For example, a 512K array containing Robert Krantz has time. But if not properly managed, the eight 64K blocks, each capable of served many roles within use and wear can be unbalanced – very 100,000 write operations has a lifespan of Datalight over the past low for some blocks and excessively high 800,000 write operations. However, if left seven years including for others. The overused memory can fail unmoved, a four-block static region on technical support, project early and significantly reduce the lifespan this array could effectively prevent the use management, product of the media. Uneven “wear” across of 400,000 write operations. The media’s management, and currently Director of the memory array is addressed by wear lifespan would be cut in half. Strategies and Services for Datalight leveling functions. Engineering. Robert joined Datalight from Some systems attempt to prevent the a desktop hardware background where he occurrence of static regions by stringing Wear leveling operations should cause spent five years building, supporting, and data throughout the flash disk. Others minimal performance impact while deploying computers for personal and employ various statistical models. accomplishing the following: commercial uses. Alternatively, Datalight products actively n Balancing write operations across all monitor block use statistics, detect low- blocks To learn more, contact Robert at: use areas (static areas), and then swap this n Rotating the location of static content Datalight Inc. data into a high-use area. n Preventing deadlocks 21520 30th Dr. SE Bothell, WA 98021 Wear leveling can also help to prevent a To properly perform wear leveling, a Tel: 425-951-8086 deadlock – that is, a small region (two or media management system will keep E-mail: [email protected] more erase sectors) locked in a pattern of track of how often each block in the array Website: www.datalight.com writing and compaction. Deadlocks are has been used. In FlashFX Pro, this data attributed to high levels of static content

96 / 2005 Industrial Emdbedded Systems Resource Guide Storage Storage Hardware: Data recorder Hardware: NAS Storage Storage

Loggers/Recorders NS04-4100 FEATURES: FEATURES: n From 2 to 64 analog/digital channels n Four-drive bay storage appliance with superior data protection n Programmable signal conditioning (gain, filter, and bridge) n An integrated file sharing and backup/recovery solution n Powerful, easy-to-use VersaDAQ™ GUI software n Seamless data recovery by supporting remote boot n Local data storage via CompactFlash™ module n Supports RAID 0, 1, 5, and access control list n Compact, lightweight, ruggedized, and portable enclosure n Built-in DHCP server and printer server functions

KineticSystems Company, LLC Lanner Electronics Inc. 900 North State Street 925 Canada Court Lockport, IL 60441 City of Industry, CA 91748 Tel: 815-838-0005 • Fax: 815-838-4424 Tel: 626-581-1898 • Fax: 626-581-1897 www.kscorp.com www.lanner-usa.com RSC #9701 @ www.industrial-embedded.com/catalogrsc RSC #9702 @ www.industrial-embedded.com/catalogrsc

Storage Software: Storage Application

OSIsoft

OEM Data Historian FEATURES: n Automatic high-volume data collection for OEM applications n 'ECHO' stores up to 800,000 variables per second n Optimized for use in mobile and remote devices such as PDAs n Ideal for optimization and analysis; rapid data retrieval n Developer support from the data historian experts at OSIsoft

OSIsoft 9045 S. Pheasant Ridge Lane Saline, MI 48176 Tel: 734-944-2484 • Fax: 423-283-4894 www.echohistorian.com RSC #9703 @ www.industrial-embedded.com/catalogrsc

Industrial Emdbedded Systems Resource Guide 2005 / 97 The Final Word The Final Word The Final Word The Final Word The Final Word The Final Word The Final Word The Final Word The Final Word The Final Word The Final Word The Final Word The Final Word The Final Word The Final Word The Final Word The Final Word The Final Word The Final Word The Final Word The Final Word The Final Word The Final Word The Final Word The Final Word The Final Word The Final THEWord The Final Word TheFINAL Final Word The Final Word The FinalWORd Word The Final Word By Jerry Gipper

It’s all about

Electronicchoices systems designers have never had as many choices as With all the choices, many complications and questions arise. they have now in functional modules or boards for their embedded Should you stick to standards based products? Should the electronics needs. Hundreds of suppliers worldwide specialize in standard be open to the public or internal to individual suppliers? commercially available, off-the-shelf products. Combined, they What form factor or interface is best? What I/O should be used offer thousands of products. The well-informed designer can shop and in what circumstances? What should I look for in a reliable around for functional modules and boards in a huge array of form supplier? You need to consider system interconnects, product life factors, performance levels, capabilities and feature options. It cycles, and make/buy decisions. is a dramatic change from the days of build-your-own to today’s vast array of off-the-shelf solutions. Software choices also must be made, some before selecting hardware, some during hardware selection, and some after The same is true for all of the peripherals and software that are choosing the final hardware platforms. Development tools are needed to complete the system. There is an even larger selection also important. The right decisions here can ease the design and of peripherals to gather input, display results, interface with shorten time-to-market. humans and other computer systems, store information, and control devices designed for embedded applications. To top it off, You can see the wide variety of choices when you look through software tools and operating systems are optimized for real-time many of the embedded computing vendor’s websites or catalogs. systems ranging from simple consumer electronics to the most Where once they had a single product line or two, there are now complex control systems. Commercially available functional multiple product lines with many options within each. Most of modules, peripherals, and software make plug-and-play a reality them also offer various levels of customization or design services in many cases. that are essential in designing complicated embedded computing systems. All of this means that more energy can be put into designing the best possible product in a reasonably short period of time. These are all topics and issues that we will cover in upcoming Focus can be on the functionality of the end product and not on issues. We will hear from experts in the fields of industrial spending precious design time in debugging basic elements of a embedded systems who will present their perspectives on the system design. Options in the design can be quickly prototyped, challenges faced and their ideas and recommendations for solving verified, and modified without significantly impacting schedules. the design and product problems. An OEM designer’s role changes to system integration more so Our goal is to help you get through the maze of choices facing than chip level design. industrial embedded system designs. We’re glad you’ve made the Module and peripheral design can be extremely complicated and choice to read Industrial Embedded Systems and look forward to is usually best left in the hands of experts who do these designs serving your information needs. day-in and day-out. No one designs their own disk drive or Jerry Gipper monitor. Rarely would you design your own compute modules Editorial Director and peripherals with the selection available today.

98 / 2005 Industrial Emdbedded Systems Resource Guide RSC #99 @ www.industrial-embedded.com/rsc RSC #100 @ www.industrial-embedded.com/rsc