CompactPCI and AdvancedTCA Systems - OpenSystems Media

RSC# 3 @www.compactpci-systems.com/rsc

PERF TECHRSC# 5 @www.compactpci-systems.com/rsc

RSC# 7 @www.compactpci-systems.com/rsc

EDITOR’S FOREWORDEDITOR’S FOREWORDUptime. Downtime.For decades, designers and customers of embedded computersystems have attempted to establish reliability by estimatingMean Time Between Failures (MTBF). This is almost always acalculation, usually based on the failure rates of individual componentsas established by tables like MIL-HDBK-217. Thesetypes of methods provide an estimate of how long a systemshould operate before failure:■ If the components were used within their design margins■ If the components used were good■ If the circuit assemblies were properly manufacturedA lot of ifs. Most MTBF calculators are largely silent about thingslike software, which is usually the most failure prone element ofany modern, complex embedded computer. And new failure mechanismsare beginning to appear. One good example is the increasingsusceptibility of very small geometry integrated circuits to logicfaults and failures due to radiation sources such as solar neutronsthat practically cannot be shielded against (see the October, 2004CompactPCI and AdvancedTCA Systems Editor’s Foreword). And,MTBF is just a statistical estimate. A system with a 30,000-hourcalculated MTBF may fail after a few hours of operation. MTBF’ssister calculation, Mean Time To Repair (MTTR) also assumes thatthe right replacement part or assembly is available and is good, andthat the individual performing the repair or replacement is skilled inthe process. All in all, a lot of big fat ifs and guess-timates. So, areMTBF and MTTR calculations useful? Almost certainly. Are theyenough? Increasingly, the answer is no.Is there a better way?The telecommunications industry has, for years, used availabilityin addition to MTBF as a better measure of a system’s overall reliabilityand robustness. Numbers like 5-nines (99.999% uptime,By Joe PavlatEditorial DirectorCompactPCI &AdvancedTCAor about 5 minutes of downtime a year) or 6-nines (99.9999%uptime, or about 30 seconds of downtime a year) are often used asmeasures of availability. Availability requires a somewhat differentmindset when compared to MTBF thinking. Highly availablesystems are generally architected in very different ways from traditionalsystems. They usually have multiple, redundant resourcessuch as processors, power supplies, and storage. Specialized hardwareand software combine to detect failures and switch out badresources and subsequently switch in good ones. Downtimes areoften measured in seconds or minutes, not hours or days. Of courseit is almost always desirable to replace failed resources with goodones for continued redundancy, and features like hot swap andsystem management help repair personnel keep the still-runningsystem ready for the next failure. The term 24x7 is being replacedin the communications world by 3600 by forever, which is a bettermeasure of real world requirements. Downtimes need to be measuredin minutes at most, not days.Designers of military electronics should be interested in high availabilityarchitectures. Traditional military systems have achieveda level of reliability by robust packaging and careful componentselection, but usually have simple single-resource architectureswithout the capability of failure tolerance and automatic repair.Additional forces are in play that should cause military electronicsdesigners to take a few chapters from the telecommunicationsequipment design handbook and start to think about availabilityinstead of just MTBF. For example, today many necessary componentsare of commercial grade, including almost all silicon.That’s not necessarily a bad thing. One good aspect of this trendis that complex silicon gets cheaper every year, permitting theduplication of many functions for redundancy. Also, today’s netcentricwarfare environment is largely about information technologyand communications. Many of the lessons about makingthose types of systems highly available have already been learnedin the telecom world, including different methodologies for softwarerobustness than those used in the military systems. Sure,environmental extremes and operating temperature requirementswill often make some military electronics systems specialized,but the underlying architectures and components developed forthe much larger communications marketplace should be consideredwherever possible.Keeping modern military electronics systems operating 3600 byforever will be absolutely necessary in the future as warplannersand warfighters make rapid decisions based on real-time information.Putting the best heads together from both the telecom andmilitary electronics worlds would be a great opportunity to furtherthe state of the art for both and to face common challenges,such as better cooling technologies, for the future.Joe Pavlat, Editorial Director8 / CompactPCI and AdvancedTCA Systems / June 2005

CompactPCI and AdvancedTCA Systems / June 2005 / 9RSC# 9 @www.compactpci-systems.com/rsc

RSC# 11 @www.compactpci-systems.com/rscCompactPCI and AdvancedTCA Systems / June 2005 / 11

2. Global walking: Communicationsand diagnostics among various nodepoints within the system.3. Message service: Messaging betweencomponents within a system.4. Event distribution: Failure ormanagement events, what constitutesan event and when it occurs, whogets notified, and how are theynotified.5. Checkpointing service:Checkpointing at certain phaseswithin a task and if errors occur, rollback and relaunch of the task fromthe checkpoint.From a telecom perspective, some ofthese services are enterprise-centric andtoo simplistic. Developers are now addingmessage-based distribution, system management,and logging capabilities, as wellas security that figure much more importantlyfor telecom. Everything will nowbe backward-compatible from the olderspecifications. So, now a critical mass ofservices has emerged that can be adoptedas a high availability system.Q. What plans does Motorola have withrespect to AIS?A. Motorola is actively developing anapplication interface library and plansto integrate it into our AdvancedTCAapplication enabling platform. Also, weplan to work with third-party platformproviders to implement on their platformsas well. The HA software and AIS productswill be our own unique products.In addition, Motorola plans to port theapplication interface library to other systemsof partners that make a good strategicfit for us. The end benefit from thisactivity is portability for applications thatperform high availability tasks within asystem.Q. What kind of interest are you seeingin using AIS specs?A. Oracle and VERITAS have joined andare seeking a way to incorporate the AISspecifications I mentioned earlier intotheir applications. Motorola and the SAForum are also looking into Java-basedalternatives. Since Java is quickly becominga ubiquitous part of Internet programming,integrating AIS services, either asa Java class or native method, opens uphigh availability services for just aboutevery application connecting with theInternet.and Army mandate high availability withstandard form factor solutions. Many ofthe same attributes of streamlining supplychain coupled with high availabilityin telecom are also attractive for defenseapplications.Significant interest also exists in the highavailability SA Forum product beingdeveloped. We originally estimated onedesign win this year with significantdesign win activity in 2006, but we’rewell ahead of 2005 projections.Motorola ECC is changing to meet theneeds of the telecom industry. Likewise,Motorola ECC headcount seems to alsobe tracking this change. Currently, Johnestimates the ratio of hardware to softwareengineers being 60/40. Softwarehead count could reach the point wherethe 60/40 ratio could flip in favor of softwareengineering over the next few years.Boards and platforms are still an importantpiece of the Motorola solution. However,software capabilities are the key ingredientto serving the needs of capturing newbusiness in the telecom industry.For further information, contact Curt bye-mail at cschwaderer@opensystemspublishing.com.We’re also seeing some activity indefense and aerospace, where the NavyRSC# 13 @www.compactpci-systems.com/rscCompactPCI and AdvancedTCA Systems / June 2005 / 13

TECHNOLOGY IN EUROPETECHNOLOGY IN EUROPEStrong businessBy Hermann StrassCompactPCI &AdvancedTCAEuropean business trendsAlcatel, (France), one of the top five carriersoperating globally, has announcedtheir endorsement of AdvancedTCAat the 3GSM World Congress held inFebruary, 2005 in Cannes, France, as itspreferred architecture for the evolution ofits mobile and fixed network infrastructureplatforms. This comes after a yearof working to define and develop modularcommunications platforms builtupon AdvancedTCA, Carrier GradeLinux, and other standards. Alcatelclaims this to constitute one of the world’sfirst applications of AdvancedTCAnetwork equipment. There are otherclaims such as this (see the July/August2004 CompactPCI and AdvancedTCASystems, Technology Update column).Established on May 31, 1898, Alcatel,with sales of 12.5 billion EURO (approximately$16.25 billion) in 2003 (42 percentin Western Europe) operates in morethan 130 countries.The Alcatel Evolution 9130 BaseStation Controller (BSC), based onAdvancedTCA architecture, is one ofthe elements of a radio network. It offersup to 2,000 channels in one rack. TheseBSCs are the physical link between theswitching center and base stations. Theyprovide control of handovers, frequencyuse, and signal power control for everymobile user.Kontron, (Germany), sees itself as thenumber three supplier of open-standardembedded systems products worldwidebehind Motorola and Advantech.About 50 percent of their business is inEurope and 40 percent in the US. TheirCompound Annual Growth Rate (CAGR)last year in EURO currency was about 20percent (30 percent in US dollars). Withalmost 70 percent in capital resources,their business has a solid foundation. It isinteresting that a major part of their businessis in computing and control equipmentfor slot machines in Las Vegasgaming establishments. Kontron sellsmodules, systems, and applications toOEMs rather than to end users.As with many European companies,Kontron sells into the transportationmarkets. Railways (rolling stock andcontrol equipment) in Portugal, Spain,and Corsica are equipped with a varietyof Kontron supplied control equipment.OEM customers in transportation includeBombardier and Siemens. A typical trainmanagement system may use devicesfrom several product families such as:■ An information panel(12-inch TFT controlled by anE2Brain)■ Communications networks– Wire Train Bus (WTB)– Industrial Ethernet– CAN– PROFIbus■ CompactPCI systems■ Remote I/OsAdvanced Mezzanine CardsDespite the recent collapse of the telecommarket, Kontron envisions a 10 percentCAGR in this market fairly soon. Threetypes of AdvancedTCA boards: CPU,Hub, and Advanced Mezzanine Card(AMC) are currently available to the publicand some others to undisclosed OEMs.Kontron sees a great future in AMCs asmezzanines on AdvancedTCA boards andperhaps an even better future in the formof MicroTCA. One example is AMCs thatplug into a backplane directly rather thanas mezzanines on an AdvancedTCA carrierboard. Figure 1 shows the KontronAT8001 CPU board with two AMC mezzaninesinstalled. The Xeon Nocona basedCPU features PCI Express, Fibre Channel,and Carrier Grade Linux. It is too early tospeculate on this new application sinceFigure 1AMCs are not yet widely available. Theirspecification was ratified January 3,2005 by the PCI Industrial ComputerManufacturers Group (PICMG). AMCsuse card-edge connectors suitable fortelecom office application versus gastightpin-and-socket connectors, whichare required in many industrial applicationsas a protection against aggressivegases, moisture, and other factors.European market analysisEvery year in time for the CeBIT Fair,the European Information TechnologyObservatory (EITO) issues their statisticsand market analysis report derived fromEuropean and US sources. The EITO is aEuropean organization supported by privateand semi-government organizationsthroughout Europe. Their statistics indicatean Information & CommunicationTechnology (ICT) market of about 1,959million EURO (approximately $2,547 million),with Europe ahead (32.2 percent) ofthe US (29.4 percent) and Japan (14.8 percent).The European market breakdownof 594 million EURO (approximately 722million) shows datacom and networkingequipment at 37 million EURO (6.3 percent).Only the computer section is largerat about double this size at 12.4 percent.Overall growth rate in the EU was +3 percentin 2004 (+2.9 percent in the US), andshould be +3.8 percent in 2005 in the EU.Regulatory requirements such as WEEE(see the April 2005 Technology in Europecolumn) or Registration EvaluationAuthorization of Chemicals (REACH),and others account for some of the additionalgrowth rate.SBS Technologies (Germany) continuesto grow in Europe (see the October2004 issue of VMEbus Systems magazine,VMEbus Technology in Europecolumn). During the first nine months offiscal year 2005 (ending March 31, 2005)SBS reported a sales increase of 60 percent(10 percent of which was due to thecurrency exchange rate) for the EuropeanGroup and a five percent increase for theAmericas Group. SBS is expanding theirproduction and office space in Augsburg,14 / CompactPCI and AdvancedTCA Systems / June 2005

RSC# 15 @www.compactpci-systems.com/rscCompactPCI and AdvancedTCA Systems / June 2005 / 15

RSC# 16 @www.compactpci-systems.com/rsc16 / CompactPCI and AdvancedTCA Systems / June 2005

TECHNOLOGY IN EUROPETECHNOLOGY IN EUROPEGermany on company property. SeeFigure 2. The Lord Mayor of Augsburgparticipated in their topping-out ceremonyat the end of April to honor this forwardmovingcompany. Augsburg is an oldbut very innovative city. Together withTrier, it shares the title of being the oldestcity in Central Europe that is in today’sGermany (countries such as France andGermany did not exist in those days) withmore than 2,000 years of recorded history.The physics department of the Universityof Augsburg is world class with somesignificant breakthrough discoveriesannounced this year. The Augsburgregion is perhaps the world’s foremostcenter of excellence in environmentalresearch and development (see the April2005 issue Technology in Europe column).Mozart’s parents and their ancestorscome from Augsburg. Rudolf Dieselinvented the diesel engine at Augsburg.So SBS European headquarters (within astone’s throw from Augsburg University)is embedded in an innovative, environmental,and historical environment.SBS develops advanced products usingAdvancedTCA and MicroTCA technology.SBS has developed an AMC processorboard, to be announced in time for theInternational Supercomm Conference inChicago, June 6 to 9, 2005.Figure 2Hermann Strass is an analyst andconsultant for new technologies,including industrial automation, computerbus architectures, mass storagetechnologies, and industrial networking.He is an active member of severalnational and international standardizationcommittees.For further information, contact Hermannat: hstrass@opensystems-publishing.comRSC# 17 @www.compactpci-systems.com/rscCompactPCI and AdvancedTCA Systems / June 2005 / 17

GUEST EDITORIALMILITARY APPLICATIONCompactPCI in the military:Playing to its strengthsBy David CompstonDavid discusses the advantagesCompactPCI’s inherently I/Oorientedarchitecture yields fornetwork-centric warfare and the needfor smaller, lighter solutions.To understand where CompactPCI fits inthe military scheme of things, and the factorsthat will affect its future, it’s importantto understand ”the nature of the beast.”How is the military different from, forexample, the telecommunications marketplace– and how does this difference affectits adoption of new technologies?The first thing to understand is that theoverriding characteristic of the militarymarketplace is its inherent conservatism.Making technology decisions thatcan literally be a matter of life or death– rather than a bad telephone line connection– makes you somewhat cautious.The ideal military technology is stable,proven, known to be reliable, and widelyaccepted, attributes more highly prizedthan cutting edge performance.It is also true that, given the typical militaryapplication’s complexity, its mannerof deployment, and the nature of how thatdevelopment and deployment are funded,vendors measure project timescales inyears or very often in decades. This forcesattention onto issues such as obsolescencemitigation and long term support, againcausing the military to value technologiesthat have proven longevity. Beyondthis, the military faces the requirement tointegrate with an enormous installed baseof legacy systems. As a result, developmenttends to be evolutionary, rather thanrevolutionary.This approach often meant that, historically,the military struggled to stayabreast of technology developments.The landscape changed, however, withSenator William Perry’s memorandumof June 1994 which, in effect, mandatedthat in the future the US defense industry,which represents a huge proportion of thedefense industry worldwide, should nolonger design and develop its own proprietarysolutions, but should rather takeadvantage of the substantial cost savingsavailable from implementing CommercialOff-the-Shelf (COTS) solutions instead.Although primarily intended as a costsavingmeasure, COTS brought new technologiesto military applications morequickly and, through adherence to industrystandards, delivered the high degreeof interoperability that was fundamentalto the military’s requirements. The COTSapproach has also reduced the time tomarket for new applications.The foregoing may give some insightas to why it is that CompactPCI has notthus far made the progress in the militarymarket that might reasonably havebeen expected, especially given the pervasivenature of PCI technology both onthe desktop, and as enabling technologyfor the majority of boards sold into militaryapplications. Although PCI has beenaround for 10 years, it is still, in militaryeyes, a newcomer by comparison withVMEbus. VMEbus is the bus architectureat the heart of the majority of militarysystems and has been around for a quarterof a century. The history of VMEbus is aremarkable one, not least in the ability ithas consistently demonstrated to embraceand adapt to emerging technologies.But if the advent of COTS opened thedoor to CompactPCI, it is the real changein military thinking that is likely to see itestablishing at least a substantial toehold.The buzz phrase in military circles is network-centricwarfare, and it describes anew paradigm in which military “appliances”are viewed as nodes on a network,with local electronic intelligence at thepoint of deployment. Future battles willbe won by the force that can most quicklygather, analyze, distribute, and act oninformation. That’s nothing new in warfare,of course.An important goal of network-centricwarfare is that it should be technologyintensive,not personnel-intensive.“Sensor to shooter” solutions, for example,capture the idea that a potentialtarget can be identified, acquired, anddealt with in a single, seamless electronicprocess that requires no human intervention.Unmanned vehicles, whetherUnmanned Ground Vehicles (UGVs) orUnmanned Aerial Vehicles (UAVs), arethe next logical step in this direction.Limits to VME 3U implementationsWhile either can be of any size (a UAV, forexample, can range from a hand-launchedunit to one which requires a traditionalrunway) the trend is towards small andlightweight to maximize both deployabilityand mission range. This trend presentssomething of a conundrum to designersof military systems, because it points tothe need for a solution built around the 3Uform factor. VMEbus, which would otherwisebe the natural choice, does not readilylend itself to “small and lightweight.”Designed for high performance, multiprocessorapplications in harsh environments,VME is highly scalable, but its 3U implementationhas three important limitations.The first of these is that VMEbus systemsare power-hungry, and thus generate heatthat has to be dissipated. Second, the full64-bit implementation of VMEbus is onlyavailable in its 6U format. This createsperformance constraints for the smallersystems, which may only use the 16-bitimplementation. Third, and perhaps mostimportantly, VMEbus in its 3U form providesnegligible rear I/O, greatly reducingits flexibility.CompactPCI, on the other hand, providesan architecture that is inherentlyI/O-oriented, with the availability of75 pins across the backplane. Its 32-bitparallel bus offers potentially higherperformance than VMEbus in its 3U form.It is designed to support a maximum of8 slots – compared with VMEbus’s 21 slots18 / CompactPCI and AdvancedTCA Systems / June 2005

RSC# 19 @www.compactpci-systems.com/rsc CompactPCI and AdvancedTCA Systems / June 2005 / 19

GUEST EDITORIALMILITARY APPLICATION– and consumes less power. Typically, as thecomponents are closer to the cooling rails,conduction cooling can be more efficient.Building a UAV around a 3U CompactPCIsolution, such as Radstone’s RT4 Power-Pact application-ready platform, can generatesubstantial savings in weight, size,and power consumption. Figure 1 showsthe RT4. The RT4 measures approximately10 inches by 5 inches by 5 inches,making it less than a sixth of a cubic footin volume, and it weighs less than eightpounds. Yet it delivers the same processingpower as Radstone’s PPC7D processorin a 1/2 ATR chassis, which weighsmore than twice as much and occupies25 percent more space.Lightweight design surveillanceradarPerhaps typical of the military applicationsfor which CompactPCI is well suitedis the development being undertaken byTelephonics Corporation, headquarteredFigure 1in Huntington, NY, for Lockheed Martinin support of the US Coast Guard’sDeepwater program. Targeted at theemerging market that is demanding fullycapable surveillance radar in a lightweight(less than 75 lbs.), compact (twoboxes – 1/2 ATR Short Signal Processorand 3/4 ATR Short Receiver/Transmitter)design, and developed for use in UAVs,the project selected CompactPCI becauseof its combination of higher bandwidth(relative to VMEbus), support of a lightweight3U form factor, and interconnectability (via the PCI backplane). Beyondthis, the inherently open architecture ofCompactPCI gives Telephonics access toa range of products from potential suppliers,together with a powerful road map viaPCI-X and PCI Express. The companybelieves that the industry trend towardsMaximum Radar Processing capabilities,Digital Scan Conversion, and interfacesto programmable gate arrays in the areaof signal processing, together with theincreasing emphasis on remotely operated(unmanned) platforms and smaller,lighter solutions will see CompactPCIcontinuing to gain acceptance in the defensecommunity.CompactPCI in its 3U form offers benefitsfor space-, weight-, and powerconstrainedapplications that are veryattractive to the military system designer.Beyond this, the growing number of conduction-cooledCompactPCI boards isincreasing all the time, and the nature ofthe technology means that it is possible todevelop extremely powerful but relativelyinexpensive solutions: Radstone’s IMP2A(see Figure 2) CompactPCI processor, forexample, has the functionality of a 6Ucard in a 3U space.Although it is “immature” by VMEbusstandards, the military market is reassuredby the comparative longevity ofCompactPCI, noting that it has not onlysurvived but also thrived in an industrythat seems to throw up new bus technologiesevery month.But just as CompactPCI seems to becoming into its own in the military marketplace,questions are arising about itsfuture. The desktop PCI architecture onwhich CompactPCI is based is movingrapidly towards PCI Express with its substantiallyimproved performance. PICMGhas intercepted the concern with its proposalsfor 3U Express (and 6U Express),which will provide a native PCI Expressbackplane in the CompactPCI format andaccommodate legacy CompactPCI cards.RSC# 20 @www.compactpci-systems.com/rscWhile the military may be wary of thetransition to a new backplane, VMEbusis going through a similar transition withthe VITA 46 proposal that is designed toaccommodate upcoming switched fabric20 / CompactPCI and AdvancedTCA Systems / June 2005

Figure 2technologies with a redesigned backplane.In the case of VITA 46, military customershave indicated a willingness to acceptchange if it delivers higher performance– and it seems likely that hybrid solutionswill emerge in the case of both VITA 46and CompactPCI.When CompactPCI was first announced,some commentators believed that it wasconceived as a competitor to VMEbus.That proposition has always seemedunlikely, given that the two technologieshave contrasting strengths and weaknesses– which is why, today, it looks as if theywill coexist in the military market space,with CompactPCI leveraging its strengthsto take advantage of the move towardssmaller, lighter solutions. Manufacturerssuch as Radstone will continue to offerproducts based on both architectures.David Compston graduated from theUniversity of Warwick, England witha degree in Computer Science. He isa board industry veteran, having heldlead positions both in engineering andmarketing for Radstone for more than20 years. David is currently Directorof Marketing for Radstone EmbeddedComputing.For further information, contactDavid at:Radstone Embedded ComputingTove Valley Business ParkTowcester Northants UK NN12 6PFTel: 44 1327 359444Fax: 44 1327 322800E-mail: david.compston@radstone.co.ukWebsite: www.radstone.comRSC# 2103 @www.compactpci-systems.com/rscRSC# 2101 @www.compactpci-systems.com/rscRSC# 2102 @www.compactpci-systems.com/rsc

SPECIALHIGH-END EMBEDDED COMPUTINGBeam-forming to scientific modeling: High-densitycompute platforms offer multiprocessor solutionsBy Ian StalkerThe CompactPCI standard continuesto grow in importancefor high-end applications.There is for example increased demandfor Gigabit Ethernet in a PICMG 2.16configuration. This marks a step towardsthe eventual replacement of parallel bustechnology with switch fabric interfaces,yet it must be stressed that the new generationof switch fabric technology suchas used in AdvancedTCA is still sometime away from maturity and generaldeployment. In this article Ian covers anumber of applications that make use ofthis type of technology, signal analysisbeing one of them.Today, the preponderance of CompactPCIapplications are served by classic generalpurpose Single Board Computers (SBCs)and I/O products, where a single microprocessor,paired with application specificI/O, provides sufficient computingpower to perform the requisite task. Manyindustrial control applications, for example,fall into this category. At the otherend of the performance spectrum resideapplications that are essentially computebound meaning that the system designerswill take advantage of as many MIPs andGFLOPs as their fiscal or power budgetwill permit. Simulation and scientificmodeling are examples where there iscontinual need for greater speed and resolution,and which also frequently requiremultiprocessor solutions.Two classes of compute problems needmultiprocessing solutions. The first classcomprises compute farm applicationswhere multiple channels of data needto be processed but have only a smallor medium requirement for interchangebetween the processors working on theproblem. For these applications Ethernetprovides an ideal transport between theprocessors because it is simple to programwith portable software and is alsocost-effective. Scaling up to large systemsinvolves the relatively straightforwardprocess of packaging multiple boards intoenclosures.The second class of multiprocessor applicationsis that in which multiple processorswork with a shared database on asingle problem. These problems typicallyinvolve large amounts of interprocessorcommunications. One example of thistype of application is digital radio beamforming.In these applications one wouldbenefit from augmenting the interboardI/O with a higher performance, low overheadcommunications technology.A high-density 6U CompactPCI computeplatform based on the PICMG 2.16Packet Switching Backplane (cPSB) standard,such as Curtiss-Wright’s CHAMPAV-IV (CAV4) can be adapted to theseapplications with the addition of one ortwo StarFabric PMC modules to provideup to approximately 1 GBps of interboardI/O while significantly reducing processoroverhead. Figure 1 shows the CAV4.Figure 1PICMG 2.16 plusAs previously mentioned, the CAV4employs the PICMG 2.16 PacketSwitching Backplane standard. In fact,the CAV4 does not have a PCIbus backplaneinterface. The PICMG 2.16 standardwas developed to overcome the inherentlimitation of the PCIbus. With a single,shared, parallel bus capable of 533 MBps(best case, 5-slots), CompactPCI systemswere becoming limited by the throughputof their interconnect. The PICMG 2.16standard introduced the concept of usingEthernet (10/100 or 10/100/1000) as themain data transport mechanism withina system. Using the CompactPCI midplaneJ3 connector, the standard definesnode and fabric slots. Node slots haveone or two Ethernet interfaces. Fabricslots provide the Ethernet switching function.Systems comprise one or two switchcards, and up to 20 nodes, supporting atotal bandwidth of up to 5 GBps.The CAV4 extends the PICMG 2.16 principleeven further. It provides five GigabitEthernet interfaces to the backplane connectors.Each processing node, includingthe 8540 control processor, has an independentEthernet connection to the backplane.Two of these interfaces are on thepins defined by PICMG 2.16.Systems built using the PICMG 2.16Ethernet standard are precursors of thenew era of interprocessor communicationsusing switched fabric technology.Standards such as VITA 41, VITA 46,AdvancedTCA, and CompactPCI Expressare all based on high-speed point-to-pointserial interconnect with switching insteadof buses. While these technologies continueto mature, Ethernet will garnermany design wins for the current generationof systems.Ethernet performanceIn the course of characterizing the performanceof the CAV4, the Ethernet throughputwas measured using the Wind RiverSystems VxWorks real time operatingsystem with the Blaster/Blastee test programsthat are included. These programshave two tunable parameters: transmitmessage size and receiver buffer size. Thebest performance, not surprisingly, wasobtained with the largest message sizes.The test used the standard VxWorks 5.5IPV4 network stack without optimizationsto take advantage of the Discovery III TCPchecksum offload feature. Table 1 showsthe performance obtained using PowerPC22 / CompactPCI and AdvancedTCA Systems / June 2005

SPECIAL7447A processors at different clock ratesusing message sizes of 48 KB.With a total of five Gigabit Ethernet interfaces,the card is capable of well in excessof 300 MBps throughput. A system comprisedof many CAV4s would have dramaticallymore Ethernet communicationsbandwidth than that provided by a singlePCIbus.Power consumptionIt is a well-known phenomenon that thepower consumption of microprocessorsand accompanying system logic has beensteadily rising. Desktop processors fromIntel and AMD now top 100 W. In highperformancemultiprocessor computingapplications, the name of the game howeveris computing density. The questionis, “How much real computing work canbe accomplished within the confines ofstandard enclosures and racking systems,without resorting to prohibitively expensivecooling technologies such as spraycooling or refrigerated air systems?”The latest generation of quad processordesigns is starting to push the envelopeof available cooling in the IEEE 1101.10mechanical standard. A precision airmass-flow measurement test was developedby Curtiss-Wright to qualify andaccurately specify the cooling requirementsfor high-power, air-cooled processorboards. In concert with this program,we have characterized the power consumptionof the Compact CHAMP-AV IVat different processor clock rates and inletair-temperatures. Table 2 shows the powerfor a test scenario designed to stress theprocessors and memory subsystem of thecard, thereby consuming power in excessof the majority of real applications.These tests highlight some of the factorsthat influence power consumption. Fasterclock rates are usually accompanied withthe need to power the processor core athigher voltage, causing relatively largeincreases in power for relatively modestincreases in clock rates between 998 MHzand 1064 MHz. The other factor that isperhaps less well understood is processors’drawing more power when runningat higher silicon die temperatures,illustrating the need for effective thermaldesigns and air-management within theenclosure. Freescale’s power estimatesfor the 7448 processor are not publiclydisclosed, but we expect to see powerreductions at equivalent test conditions.Much of the technology required forapplications with high performance needsin the military market such as radar,sonar, and signal intelligence can beapplied in products aimed at the high-endof the commercial/industrial CompactPCImarket space where performance andpackaging density is valued but extremeruggedization is not.Ian Stalker is the DSP productmanager for Curtiss-Wright ControlsEmbedded Computing. He holds morethan 20 years of experience in theembedded industry and has a degreein Electronic Engineering.For further information, contact Ian at:Curtiss-Wright ControlsEmbedded Computing741-G Miller Drive, SELeesburg, VA 20175Tel: 703-779-7800 • Fax: 703-779-7805E-mail: ian.stalker@curtisswright.comWebsite: www.cwcembedded.comProcessor ClockEthernet Performance665 MHz 62.6 MBps998 MHz 79.1 MBps1064 MHz 79.1 MBpsTable 1Core Voltage 1.0 V 1.0 V 1.1 VCore Frequency 665 MHz 998 MHz 1064 MHzAverage Power@ 25 °C inlet47.6 W 54.7 W 64.9 WAverage Power@ 50 °C inlet50.6 W* 59.2 W 70.9 W*Power measured at 40 °C for this test.Table 2RSC# 23 @www.compactpci-systems.com/rsc

SPECIALHIGH-END EMBEDDED COMPUTINGProcessing challenges of shrinking high-endembedded computing systems to fit into smallunmanned air vehiclesBy Bob KahaneLarge Unmanned Aerial Vehicles(UAVs) such as Global Hawkand Predator have been successfulusing today’s high performanceembedded computing solutions. In thisarticle Bob explains that the challengeis to provide similar processing powerfor much smaller UAVs, many of whichhave less than half the payload weightand one-quarter the volume of thePredator.UAVs offer an ability to perform penetratingsurveillance missions as well as persistentsurveillance with low risk and theability to get in close to better see, hear,and sense the situation of interest. One ofthe best-known UAVs is the Global Hawk,a multimillion-dollar aircraft managed asa theatre/national asset similar in dimensionsto the U2 manned reconnaissanceplatform. The Global Hawk, with its largesize, provides a platform for multispectralsensor suites, including Synthetic ApertureRadar (SAR), Electro-Optic/Infrared (EO/IR), and SIGnals INTelligence (SIGINT)payloads. This UAV has proven its worthin battlefields from Bosnia to Afghanistanand Iraq. This success has led to a surgein proposed UAV missions and designsusing a layered approach, with multipleclasses of UAVs to provide persistent narrowand wide Intelligence, Surveillance,Reconnaissance (ISR) coverage. In supportof this mix, smaller UAVs, such asthe Hunter and the Predator, are mostwidely known and have also proventheir utility with a lesser complement ofsensors. When netted with manned andlarger UAVs, smaller UAVs provide significantsynergy and effectiveness in rapidlyassessing the situation. In addition,smaller UAVs can detect targets of interestwith high accuracy and with targetingquality geolocation.Large UAVs such as the Global Hawkand Predator-B have been successfulusing today’s high-performance embeddedcomputing solutions. The challengeis to provide similar processing powerfor the newest UAVs, which are significantlysmaller, and many of which haveless than half the payload weight and onequarter the volume of the Global Hawk(Table 1).Table 124 / CompactPCI and AdvancedTCA Systems / June 2005

System demandsUAV platforms must have signal processingsystems that can function under difficultenvironmental conditions, coveringhigh humidity, extreme heat and cold,dirty air, high altitude, shock, and vibration.Usually UAV platforms must dealwith some or all of these challenges at thesame time.Despite restrictions on payload size andweight, the new signal processing systemsmust be highly flexible. The cuttingedge of defense imaging technology combinesmultiple types of sensors in a singleplatform, giving field commanders a fullspectrumview of the battlefield. Rapidaccess to multiple types of images fortheir specific situation enables these commandersto make more informed, moreeffective combat decisions. The additionof SIGINT payloads for electronicsupporttype missions, such as RadioFrequency (RF) emissions, are characterizedas to emitter type, class, and angleof arrival, and increase the effectivenessof the imaging sensor. SIGINT payloadsprovide passive, 360-degree surveillanceacross wide RF bands of interest and cancue narrow field-of-view imaging sensorsto rapidly detect activities of interest.The most powerful example of this multisensorapproach is the Integrated SensorSuite (ISS) deployed on the Global Hawk.Raytheon developed this powerful imagingsystem, with signal processing supplied bymulticomputers from Mercury ComputerSystems. SAR imagery enables operatorsto view wide areas of terrain, while high-Doppler resolution radar provides a MovingTarget Indication (MTI) capability that canidentify individual moving vehicles, oreven the recoil motion of artillery tubes.Multisensor imaging capability hasproven to be highly effective. OperationIraqi Freedom employed a single GlobalHawk; it flew just 3 percent of all imagery-collectionsorties, yet it generated55 percent of all the time-sensitive targetspassed to attacking units.These results are driving plans to putmultisensor systems on more of thenewer, smaller UAV platforms, as wellas to expand multisensor capability intohyperspectral imagery and ultrawideband(UWB) radar for penetrating foliage. Tosupport the multisensor approach, signalprocessing systems must be able to generateimagery from a shifting and variableset of sensor inputs. They must be ableto perform a set of diverse functions andinterface to a broad range of sensors.In the past, we have relied on Moore’s Lawto help us out. We could wait a couple ofyears and the technology improvements inthe electronics would have enabled significantsize and power reduction. However,the industry has reached a point whereMoore’s Law still increases absolute performance,but not performance per Watt,per pound, or per cubic foot. Although thenumber of transistors available is increasing,the power consumption is increasingat almost the same rate (see Figure 1).The increased infrastructure to handlethe power distribution and heat extractionincurs a penalty in size and weight.Alternative approaches are needed.System expectationsAs the platforms get smaller, the sensorsystems are driven to greater challengesin meeting the performance requirementswithin smaller envelopes of Size, Weight,and Power (SWaP). In addition, reducedRSC# 25 @www.compactpci-systems.com/rscFigure 1

26 / CompactPCI and AdvancedTCA Systems / June 2005RSC# 26 @www.compactpci-systems.com/rsc

SPECIALHIGH-END EMBEDDED COMPUTINGplatform size is driving developers to dramaticallyreduce the cost of the payloads.What’s needed are attributable platformpayloadsolutions that still exhibit appropriatelevels of reliability to match theforecasted life expectancies of the platform.A first-order approximation of systemflyaway costs is $10,000 per pound. Thisfigure was developed many years ago on amultispectral Electronic Warfare (EW) systemdevelopment involving six major EWcompanies and has survived many technologyshifts. Technology has provided morecapabilities per pound, but this approximationhas still proven to be adequate forrapidly approximating system costs.System requirementsThe first basic requirement is that the signalprocessing systems must be physicallysmall enough to fit into the new platforms.For many platforms, the commonly used6U VME systems are just too big; 1U, 2U,or 3U form factor solutions are needed.In addition, the signal processing systemsmust adhere to industry standards forboard design and interfaces, if systemsdesigners are to benefit from CommercialOff-the-Shelf (COTS) solutions.Squeezing the processing power of 6Uboards into smaller form factors demandsthe creative use of a specialized adjunctprocessor. Adjunct processors are devicessuch as Field Programmable Gate Arrays(FPGAs) and ASICs, dedicated to a specificcomputationally intensive operation.Because adjunct processors execute asingle task, they can do it with extraordinaryspeed and efficiency. Developerscan partition signal processing operationsamong different types of processors formaximum efficiency, getting more donein less space while accepting the tradeoffof somewhat greater system complexity.True multisensor flexibility demands thesignal processing engine have a varietyof I/O options, all with high-bandwidthinterconnects to the processors. Ideally,these I/O options connect directly to theprocessing boards, as well as supportingsome form of an industry-standard mezzaninecard.operate in harsh environments, they mustbe able to withstand shock, vibration, andtemperature extremes.Processing density and efficiencyOne approach for achieving processingdensity and efficiency for signal processingis to leverage adjunct processingengines such as FPGAs as programmableprocessors. For some front-end signal andimage processing functions, FPGAs havedemonstrated a 10- to 20-fold performanceboost over a PowerPC G4 processor.However, some front-end tasks, suchas filter weight computation and mostback-end processing, still perform muchbetter on a PowerPC processor. In fittingthe most processing power in the smallestspace for a given application, the trickis finding not only the optimum balancebetween FPGAs and PowerPCs, but alsodetermining exactly which model of eachchip to choose.Application software can be partitionedso that certain algorithms go onto theFPGA. FPGA-appropriate algorithmsinclude fixed-point computations or non-Since engineers develop functional softwareaccording to an overall project schedule,they need access to adequate developmenttools, including algorithm libraries andI/O device drivers. If adjunct processorsare employed, efficient development toolsmust support them. And lastly, becausethese systems are often called upon toRSC# 27 @www.compactpci-systems.com/rscCompactPCI and AdvancedTCA Systems / June 2005 / 27

SPECIALHIGH-END EMBEDDED COMPUTINGdata-dependent operations. Other parts ofthe application, especially data-dependentoperations, are targeted to the generalpurpose processor, which is easier to programfor those types of algorithms. Thisstyle of application partitioning maximizessystem performance while keepingoverall development time manageable.Solution designSeveral approaches provide more capabilityin smaller physical configurations.The first is the use of dense packaging inphysically small but standard configurations.Another approach is the sharing ofprocessing across multiple sensors. Thistechnique provides multimission processingfrom a common set of processors. Thetechnique supports multiple payloads andprovides time-sliced load-leveled solutionsacross a suite of sensors that operatevirtually simultaneously, providing shortyet effective periods where processingelements can be shared.For physically small but standards baseddesign requirements, 3U CompactPCI isa strong choice. It is a widely acceptedstandard, maximizing configurationflexibility with a wide range of marketavailableproducts. The 3U CompactPCIconnector offers outstanding pin density;the J1 and J2 connectors provide enoughpins to support 32-bit PCI with additionalpins left over for sensor I/O. Because ituses PCI as the system bus, CompactPCIalso delivers compatibility with systemsoftware components. Standards-basedI/O flexibility can be further supportedwith a PCI Mezzanine Card (PMC) interface;PMCs are a common enhancementto 3U CompactPCI systems.Development of multimission computingpayloads provides significant opportunitiesto the platform primes as well as topayload developers in realizing increasedindividual sensor processing resourceswithin the SWaP and cost constraints ofthe platform.Example systemMercury Computer Systems’ MCP3 FCNmodule meets these requirements, deliveringhighly flexible signal processing capabilityin a space-efficient 3U CompactPCIformat. (See Figure 2.) The MCP3 FCNemploys a 1 GHz PowerPC 7447 and aVirtex II Pro P40 FPGA. A Discovery IIbridge chip connects the two processingunits. The three avenues for off-boardcommunications and I/O are via:■ PCI bus on the J1 pins of theCompactPCI connector■ Digital Intermediate Frequency (IF)to the FPGA via a direct connectionfrom a subset of the user-definedJ2 pins■ PMC, which can communicatedirectly with the FPGA or throughthe Discovery II chip to thePowerPCTo develop application components targetedfor the PowerPC processor runningWind River’s VxWorks operatingsystem and using the Tornado operatingenvironment, engineers have access to amature set of Mercury tools, including theScientific Algorithm Library (SAL) withmore than 600 routines optimized for thePowerPC. For those parts of the applicationthat run on the FPGA, developerscan use Mercury’s FPGA Compute NodeDeveloper’s Kit, or FDK. This kit is a collectionof Mercury-developed IntellectualProperty (IP), build files, command linetools, libraries, headers, drivers, boarddescriptors, diagnostics, and consultingsupport, all focused on helping engineersefficiently create reliable FPGA-basedapplications.The MCP3 FCN board is also capable ofdeployment in harsh environments. It isavailable in both air-cooled and conduction-cooledversions and is optionallydelivered in either an IEEE 1101.1 orDRTi chassis. This type of space-efficient3U signal processing solution can bebuilt using powerful COTS components,including Mercury’s MCP3 FCN. It issmall enough to be used in smaller platformssuch as UAVs, and flexible enoughto perform multiple missions and interfaceto a variety of sensors.Figure 2ConclusionThe processing requirements of smallerUAVs can be met today with the carefulallocation of the requirements to the availableCOTS processing/adjunct elementsin smaller, denser yet standard packagingconfigurations. These systems canmeet the environmental challenges andperformance requirements of affordableflyaway costs, appropriate reliability, andperformance to support the multisensorrequirements within the platform’s SWaPconstraints.Bob Kahane is director of theSIGgnals INTtelligence/ElectronicWarfare (SIGINT/EW) segment forMercury Computer Systems’ DefenseElectronics Group. He heads thecompany’s RF Center of Excellence(RFCE) in Reston, VA, which providesfront-end products for the SIGINT,radar, and software radio segments.Before joining Mercury, Bob was atRaytheon’s Intelligence and InformationSystems organization for 18 years.Bob is a graduate of the BrooklynPolytechnic Institute with a BS inapplied mathematics and an electronicengineering minor. He has completedmaster’s studies in business administrationat the American University andin electrical engineering at GeorgeWashington University.For further information, contact Bob at:Mercury Computer Systems, Inc.199 Riverneck RoadChelmsford, MA 01824Tel: 703-673-2720Fax: 703-673-2737E-mail: bkahane@mc.comWebsite: www.mc.com28 / CompactPCI and AdvancedTCA Systems / June 2005

RSC# 29 @www.compactpci-systems.com/rscCompactPCI and AdvancedTCA Systems / June 2005 / 29

SPECIALHIGH-END EMBEDDED COMPUTINGCarrier Grade Linux: The cornerstone oftelecoms’ COTS strategyBy Glenn SeilerThe telecommunications industryis in a state of profoundchange and is now beginningits long rebound from the burst of 2001.The industry is finally starting to seesignificant consumer demand for 24x7accessibility to multimedia-based highbandwidth services. Of course consumerswant these new services for lessthan they used to pay for standard voiceservice. This demand is driving the rapidgrowth of Next Generation Networks(NGNs) and thus provides opportunityfor increased revenues and market share,if the service providers can react to themarket. But this increased opportunityis also causing the emergence of newcompetitors to service these markets.These new competitors do not have thelarge inventories and cost structures thatwere created prior to the burst. Serviceproviders are faced with the challengeof providing new services while continuingto reduce capital and operationexpenses. These service providers mustfind cost-effective methods to drive downcosts and get higher margins in returnfor their services.Convergence and transformationThis change is having a far-reaching impacton the entire supply chain of ISVs andoperating system suppliers, semiconductorand hardware component vendors, andmost importantly the Network EquipmentProviders (NEPs). NEPs can no longerafford to develop entire solutions in-houseand must focus their resources on thedevelopment of new value-added services.A key strategy to drive down and managecosts is to develop systems for new marketsusing common Commercial Off-the-Shelf (COTS) components for softwareand hardware. These COTS componentsare transforming the telecommunicationsindustry just as they did for the enterpriseand IT industries in the 1990s when volume-basedx86 servers running UNIX andlater Linux began replacing single-vendorproprietary hardware and OS solutions.In fact, the convergence of applicationservices and network infrastructure applicationsis one of the key forces drivingtelecom COTS ecosystem growth. For along time the telecom industry has beenusing proprietary hardware solutions withproprietary real-time carrier grade operatingsystems, often built in house, and inhousehigh availability and managementsolutions for their network infrastructures.At the same time they often use commerciallyavailable IT or enterprise-based solutionsfor application services such as BSSand OSS. But now commercially availableCarrier Grade Linux-based systems offerthe real-time support and high reliabilitythe network infrastructure requires. TheseCarrier Grade Linux systems combinedwith advances in commodity processorand hardware technology are drivingCOTS into the previous proprietary networkinfrastructure. By leveraging thebenefits of these COTS components manyNEPs are now developing a universalplatform comprised of COTS componentsthat support both application services andtraditional network infrastructure servicesin a single cost-effective platform. NEPscan now use commercially available commoncomponents to replace much of thein-house R&D development for bothhardware and software, creating significantsavings. This drives the creation ofstandardized hardware solutions such asFigure 1AdvancedTCA packaged with carriergrade operating systems and third partyhigh availability solutions. This trend isshown in Figure 1.In particular, it is the NEPs and their customers,the service providers, who aredriving this trend. They have the mostto gain from a healthy ecosystem ofboth open source and proprietary COTSbuilding block components. The benefitsof using open source platforms such asLinux can help companies build securityrich,flexible, and scalable infrastructures,achieving levels of cost and time efficienciescrucial to accelerating developmentprocesses and speeding time to market.And the reuse of high-volume COTScomponents, which has long been a trendin enterprise, is now something the NEPsand service providers can leverage as morestandardized COTS components designedfor telecom are becoming available.Key driversLet’s look more closely at some of thekey drivers for NEPs who are consideringdeveloping NGN solutions using COTSbuilding blocks. What needs do thesecompanies face that make them look forsolutions that use building blocks fromone or more vendors rather than buildtheir own? The key drivers can be categorizedinto three distinct pressure points:30 / CompactPCI and AdvancedTCA Systems / June 2005

Consumer pressuresService providers must simultaneouslyaddress customers’ demands for increasingapplication complexity while deliveringservices at lower cost. At the sametime advances in technology, especiallythe network, are driving demands for highbandwidth 24x7 services.Business pressuresWe still have the development cycles weinherited from the dot-com era, but notthe financial exuberance. Service providersare expected to accomplish more andin less time, but for less cost than everbefore. R&D budgets and operationalcosts are being slashed, yet at the sametime providers must increase margins.Competitive pressuresTo add to the problems, in today’s confusinglyfast-paced business environment,product differentiation is more importantthan ever. The competitive field is stronger,and providers must get to market faster withbetter functionality than the competition.One way that NEPs are relieving thesepressures is by creating universal platformsthat can be reused to build futureproducts. This is where the use of reliable,commercial grade COTS components canprovide significant cost benefits frombuilding in-house solutions. Open sourcesoftware enables leading edge technologyand best-of-breed commercial solutionsfrom ISVs and HW vendors. In addition,NEPs have access to all the functionalitythey need to develop state-of-the-art NGNsolutions while also providing them withthe control they need over their individualprojects. Using these COTS-based platformsallows the NEPs to focus on theircore value and still differentiate theirproducts and services. These COTS-baseduniversal platforms can then be reused fora multitude of NGN network elements,driving down costs and time to market.Industry organizationsdriving COTSHistorically one of the challenges thatNEPs face in achieving COTS solutions isthe high cost and difficulty of integratingthese COTS building blocks into reusableand interoperable components. In orderfor these COTS components to be trulyreusable and interoperable, standardsmust be created to define the availableservices and the APIs that interface tothose services. Many industry organizationshave formed over the last few yearsto help grow and promote the ecosystemfor COTS components being used in thetelecommunications industry.Open Source DevelopmentLabs (OSDL)Recognized as the center of gravity forLinux, OSDL is dedicated to acceleratingthe use of Linux in all markets. OSDL isa key contributor to the COTS TelecomSolution Stack through sponsoring theCarrier Grade Linux Working Group. Thisis a group of Linux distributors, HW platformproviders, and NEPs that are drivingspecifications for the standardization of aCarrier Grade Linux (CGL). Already inits second release, the CGL specificationis the foundation of most COTS-basedsolution stacks being designed today.PCI Industrial ComputerManufacturers Group (PICMG)PICMG is a consortium of more than 450companies who collaboratively developopen specifications for high performancetelecommunications and industrialcomputing applications. Recently,PICMG announced the developmentof a new series of specifications, calledAdvancedTCA, for next generation telecommunicationsequipment, with a newform factor, and based on switched fabricarchitectures. AdvancedTCA’s successin the market has far exceeded initialexpectations. Nearly every major NEPis planning AdvancedTCA-based NGNsolutions.Service Availability Forum (SAF)The SAF goal is the adoption of openstandards to enable the industry to buildhigh availability network infrastructureproducts, systems, and services. TheSAF is driving interface specificationsto ensure high availability of services. AsNEPs move towards a COTS model, theywill need assurances that these new COTSapplications will have the same level ofservices and availability as their legacyin-house solutions. The SAF is definingkey services and the interfaces betweenthe hardware platform, the operatingsystem, and the High Availability (HA)middleware. The SAF is driving three keyinterface specifications:■ The Hardware Platform InterfaceSpecification (HPI) defines the interfacesbetween the hardware and theoperating system and middleware.■ The Application InterfaceSpecification (AIS) defines interfacesfor how HA middleware services communicatewith each other and with theoperating system. The AIS defines keyservices required for a complete HAsystem, including messaging, clustermembership, check-pointing, eventmonitoring, and frameworks.RSC# 31 @www.compactpci-systems.com/rsc

SPECIALHIGH-END EMBEDDED COMPUTING■ Systems Management Specification(SMS) is a complementary specificationthat acts as an umbrella to tietogether the already existing HPI andAIS specifications. It is an SNMP andWeb-based interface specification thatenables the service event and errorreporting by AIS and HPI.Each of these organizations (OSDL,PICMG, and SAF) are driving standardsin key areas of the COTS solution stack.Figure 2 illustrates the areas where eachstandards effort affects the COTS solutionstack. A significant amount of synergyand cooperation exists among theseindustry groups, for example the OSDLCarrier Grade Linux specification evenspecifies the SAF HPI and AIS interfacesas requirements for a Carrier Grade Linuxoperating system platform. The SAF isworking with PICMG to ensure that theHPI interface is mapped to AdvancedTCA.While the CGL specification is designedto be hardware neutral, SAF is consideringwhether to include specific supportfor AdvancedTCA as well.In the lower two layers of the telecomsolution stack, Carrier Grade Linux combinedwith AdvancedTCA is becomingthe de facto solution for telecommunicationsplatforms and substantially lowerscosts for all types of NGN solutionsranging from Radio Network Controllers(RNCs) and GPRS network elements tosignaling and management servers. In theupper layers of the stack, for service availabilityand application services, there aremore choices of third-party COTS componentsdepending on the type of solutionor application. It is in these layers that theSAF APIs for AIS and HPI are so importantto drive standards to ensure interoperabilityand consistency in the servicesthat are required across a multitude ofsolutions.The role of open source andCarrier Grade LinuxCarrier Grade Linux has a unique rolein the telecom solution stack becauseLinux is typically the only componentthat is based on open source. True, commerciallyavailable middleware productsabove the operating system are based onopen source, such as databases and HAsolutions, but these products do not have adominating position in the COTS solutionstack the way that Carrier Grade Linuxdoes. There are significant benefits forthe NEPs that are driving the adoption ofCarrier Grade Linux into the new COTSbasedarchitectures:■ Lower development costs by usingcommercial CGL rather thandeveloping in-house■ Faster time to market by focusingresources on value-add and usingCOTS components where possible■ Reliability by getting commercialgrade quality with tested and matureCGL distributions■ Leading edge functionality foundwith Linux that is not available inmore proprietary operating systems■ Control of projects and no vendordependency; flexibility to ownthe source and change vendors ifnecessary■ Scalability and flexibility ofCGL that can be reused for multipleNGN solutions■ Long-term viability and road mapfrom a commercial vendorBecause of the unique requirementsof the telecom industry, Carrier GradeLinux was designed from the beginningto include functionality for the telecomindustry that isn’t found in typicalEnterprise Linux distributions. One keyexample of this is the area of real-timetechnology. While the new Linux 2.6kernel has made significant advances inmainstream real-time support, some CGLdistributions include even stronger realtimesupport than what is available frommainstream Linux. Carrier Grade Linuxdistributions are beginning to reach therealm of RTOS and include hard realtimewith such technologies as priorityinheritance and user-space prioritization.But true to the nature of Open Source andLinux, these real-time enhancements arenot forks or fragmentation, but rather formalopen source projects that are optionalmodules or extensions to the standardLinux kernel.Other areas of differentiation betweenCGL and Enterprise Linux are in serviceavailability and high availability.For example there is activity in the opensource community for both the SAF HPIand the AIS specifications. Both havelaunched successful open source projects,OpenHPI and OpenAIS respectively,which are gaining traction and arebeing adopted by Carrier Grade Linuxdistributors and other vendors creatingopen source solutions. Other examples ofopen source projects for high availabilityinclude such projects as redundant networkingand safe disk unmounting in thecase of failover. These are all examples ofopen source projects that can be found insome of the Carrier Grade Linux distributionsavailable today that differentiateCarrier Grade Linux from its enterprisecousin. For the telecom solution stack,Carrier Grade Linux can now be used inboth environments providing even higherlevels of reuse and reduced costs. CarrierGrade Linux is robust enough to supportthe enterprise-based service applicationsand has the reliability to serve the networkinfrastructure.Figure 2The majority of new NGN products beingdesigned by NEPs today are based onsome form of Carrier Grade Linux. Asmany as five different Linux distributorsclaim to have a Carrier Grade Linux32 / CompactPCI and AdvancedTCA Systems / June 2005

RSC# 33 @www.compactpci-systems.com/rscCompactPCI and AdvancedTCA Systems / June 2005 / 33

RSC# 34 @www.compactpci-systems.com/rsc34 / CompactPCI and AdvancedTCA Systems / June 2005

SPECIALHIGH-END EMBEDDED COMPUTINGoffering. As of this writing, three Linuxdistributors have already registered theirproducts on the OSDL website. NEPsnow have many choices for delivering aCarrier Grade Linux operating system.Because Linux is Open Source and theCGL specification is available to anyone,NEPs always have the choice of developingtheir own Carrier Grade Linuxproduct. But remember a strong driver ofCOTS is reducing operational expenses,including the R&D budget. While developingtheir own Linux distribution mayseem attractive initially, nearly everymajor NEP has done the analysis anddetermined that the cost of integrating thevarious technologies included in the CGLspecification, maintaining this code, andupgrading the distribution over timeundermines the benefits of leveraging aCOTS-based solution and far outweighsthe costs of purchasing software and servicesfrom a commercial Linux distributor.Using a commercial Carrier GradeLinux distribution as the cornerstone ofa COTS strategy for new NGN productsenables NEPs to leverage the benefits ofCOTS to reduce costs and speed time tomarket while focusing on their valueaddedservice.Glenn Seiler is Director of ProductMarketing for MontaVista Softwareand is responsible for managingMontaVista’s Carrier Grade Linuxstrategy. In addition to his work atMontaVista, Glenn was also responsiblefor managing HA Clustering solutionsat SCO. Glenn has more than 15 yearsexperience managing UNIX and Linuxoperating systems including previouswork with Texas Instruments, SCO,BSDi, and MontaVista Software.For further information, contact Glenn at:MontaVista Software1237 East Arques AveSunnyvale, CA 94085Tel: 408-328-9200 • Fax: 408-328-3875E-mail: gseiler@mvista.comWebsite: www.mvista.comRSC# 3501 @www.compactpci-systems.com/rscRSC# 3502 @www.compactpci-systems.com/rscCompactPCI and AdvancedTCA Systems / June 2005 / 35

SPECIALHIGH-END EMBEDDED COMPUTINGPCI Express enables high-end embeddedcomputing applicationsBy Jim IsonEmbedded computing has various meanings to each companydesigning products for the “embedded” computingmarket. For every definition of “embedded” there is adefinition of what is considered “high-end.” One thing typicallyremains the same when “high-end” and “embedded” are usedtogether… there is never enough processing power to be considered“high-end” while remaining physically or thermally smallenough to be considered “embedded.” This is especially truewhen companies choose to adopt industry standard architecturesbased on widely available commercial desktop silicon. The formfactor may not fit the application, or the high-end computingpower may be overly constrained by the thermal requirements.As PCI Express continues the imminent replacement of PCIas the host add-in card bus of choice, high-end embeddedcomputing applications are poised to take advantage of thisquantum leap in technology through several industry standards.Many of these standard architectures from the PCI-SIG,PICMG, and VITA organizations are pending release in thecoming quarter. These standards will aid the embedded designerwith a multitude of issues associated with high-performanceembedded computing. In this article, Jim explains new technologicaladvancements of the PCI Express bus that will enhancethe architecture choices of high-end embedded computingdesigners. The standards discussed in this article include PCIExpress Cable, COM Express, and CompactPCI Express withan overview of SHB Express, XMC, and MicroTCA.PCI Express, in the most basic sense, is packetized PCI transmittedserially over several transmission media. The media canbe traces inside a backplane, motherboard, or add-in board, orover a twisted pair cable in many standardized mechanical formfactors. It is ideally suited toward high-speed chip-to-chip,board-to-board, and box-to-box applications. PCI Express usesLow Voltage Differential Signaling (LVDS) to transmit the PCIpackets over, in the most basic form, a four-wire bus running ata clock speed of 2.5 GHz. This four-wire bus is referred to as aPCI Express lane. The lane provides a total available bandwidth of5 Gbps. A single lane between two PCI Express end point devices,along with any of the optional sideband signals for enhancedfeatures, is called a x1 (by one) link. Designers can place severallanes between PCI Express end points in parallel to achievehigher bandwidth links of x1, x4, x8, and x16, yielding a range of5-80 Gbps of total bandwidth. Recent PCI Express press releasesby the PCI-SIG plan on doubling the clock rate of second generationPCI Express (Gen2) to 5 GHz beginning in 2006. That wouldyield data rates of 10-160 Gbps late next year.In addition to the hardware portion of the specification, PCIExpress is inherently backward compatible with PCI in regardsto operating system and application software. This compatibilityallows the application and driver developer to use the same softwaretools used to develop PCI-based software. This is in contrastto the add-in card change from ISA/EISA to PCI that requirednew tools and operating systems.PCI Express CableThe first architecture to aid in high-end embedded applications isa PCI compatible cable expansion/extension capability based onPCI Express. PCI Express Cable is a standard undertaken by thePCI-SIG to transmit the host PCI Express bus over a high-speedcable. This can be done internalto a system enclosure or externalin a box-to-box type application.Using a cable as shown in Figure1, it is possible to extend thePCI Express bus approximatelysix to seven meters from the hostCPU complex without the needfor active equalization to suppressthe inherent noise.Figure 1This particular cable is a x8 PCI Express external cable fromMolex capable of transmitting 40 Gbps of data plus the PCI-SIGdefined sideband signals.Transmitting the host bus over copper cables opens a new world tothe embedded designer. The PCI Express Cable enables a high-endcomputing core in a cooler area of a machine to host embedded I/Osubsystems in remote, thermally constrained areas of the machine.The host and I/O system can be of different form factors suited tothe location or performance each system requires. For example,a high-end, dual Intel Xeon class host system could provide thecomputing power for an operator interface and a high-speed datalink to a high-end embedded I/O subsystem based on MicroTCA,PC/104, 3U CompactPCI Express, or proprietary form factor.A compelling application of PCI Express Cable includes anexpansion system, a set of products that extends the host busof a system an arbitrary distance from the host enclosure to anexpansion enclosure. This approach enables designers to insertmore add-in boards into the system than the host system wasoriginally designed for. A simple example of an expansion systemis using a host interface board, cable, and 19-slot expansionchassis to extend a 4-slot ATX motherboard host system to a 20-slot system. Expanded systems in excess of 100 add-in boardsare likely possible utilizing PCI Express expansion.PCI Express Cable has a unique advantage over other expansionsystems currently on the market. With PCI Express acting as36 / CompactPCI and AdvancedTCA Systems / June 2005

oth the host bus and the cabled expansion protocol, it does notrequire drivers or conversion from the host bus to the expansionprotocol then back again. This eliminates a root cause of someof the throughput latency of the expansion link. PCI Expressoffers a level of software compatibility and performance scalabilityunparalleled in even the most modern generation of cabledexpansion systems currently on the market.Other embedded applications for the PCI Express Cable arefound across virtually all embedded markets. For examplea high-speed docking station link for a high-end handheld orportable device useful in medical services, inventory controlapplications, or commercial laptops could employ PCI ExpressCable. Another architecture a cabled solution could address isa noncontinuous backplane. This could take the form of severalsmall backplanes in a nonconventional configuration, suchas arranged in a circle or around a corner. In more traditionalapplications, an internal cable can replace the riser card of a 1Userver where the add-in cards are mounted perpendicular to themotherboard.COM ExpressAnother important standard is COM Express, which packs powerfulPCI Express computing cores in small form factors for theembedded systems designer. COM Express is a PICMG effortto standardize PCI Express implementations of Computer-On-Module technology. COM Express standardizes two separateform factors and several different pin-outs, offering a choice toembedded developers.Important features of COM Express include:■ Processor architecture independent■ Support for Gen1 and Gen2 PCI Express with twoimpedance controlled connectors■ 125 mm x 95 mm x 18 mm and 155 mm x 110 mm x 18 mmform factors■ Support for up to 32 lanes of PCI Express in severalconfigurations■ Support for hybrid modules with a combination of PCIExpress/PCI pin-outs■ Support for high-speed serial I/O and legacy parallel I/O■ Up to 160 W power budget per moduleThese modules allow embedded system designers to focus theircore competencies on a carrier card that includes only the customI/O functions required of the application. The designer can thenattach the COM Express computing core module to the carriercard to form a customized embedded single board computer. Theform factor and capability of the module proves useful in designinghigh-end handheld devices, custom shape carrier boards, andcustomized I/O carriers. The computing core of the carrier can beeasily scaled to the application or upgraded with a new plug-inmodule, protecting the design from obsolescence.RSC# 3701 @www.compactpci-systems.com/rscRSC# 3702 @www.compactpci-systems.com/rscCompactPCI and AdvancedTCA Systems / June 2005 / 37

38 / CompactPCI and AdvancedTCA RSC# 38 @www.compactpci-systems.com/rscSystems / June 2005

SPECIALHIGH-END EMBEDDED COMPUTINGSeveral PICMG member companies have announced COMExpress modules and road maps. Figure 2 shows PFU Systems’basic form factor COM Express module.readily available CompactPCI chassis and newly designed hybridCompactPCI Express/CompactPCI system backplanes. Theplacement of switches and bridges can include direct backplaneintegration, rear pallet bridges, or slot loaded switch/bridge cards.Hybrid systems with CompactPCI Express and CompactPCIfrom One Stop Systems are now entering the market.SHB ExpressSystem Host Board Express is the passive backplane PICMG 1.3specification. SHB Express defines a new PCI Express host singleboard computer form factor to support the passive backplanePCI/PCI Express market.Features of SHB Express include:Figure 2CompactPCI ExpressFor the embedded systems larger than a handheld device theCompactPCI standard has undergone a transformation toCompactPCI Express. CompactPCI Express is a PICMG standardpending release in the second or early third quarter of 2005.The base standard, named PICMG EXP.0:■ 20 lanes of PCI Express and a PCI-X bus on the card edgeconnector■ A dedicated connector for USB, Ethernet, and Serial ATArouting to the backplane to reduce cables to the SHB host■ Increased power capability to the host board to supporthigher performance processorsEmbedded systems based on SHB Express range from “shoebox”style systems to 1U servers less than 17 inches deep. These formfactors prove useful in embedded machine control, SCADA systems,computer telephony, and military communication applica-■ Improves power delivery to individual CompactPCI Expresscards■ Supports Gen1 and Gen2 PCI Express bandwidth with animproved connector■ Includes provisions for CompactPCI/CompactPCI Expresshybrid systemsBase features of CompactPCI, such as user I/O pins, rear I/O transitionmodules, support for telephony buses, and base mechanicalsremain in the CompactPCI Express standard. This means fora 6U CompactPCI Express card the J3-J5 mechanical attributesremain the same as the base CompactPCI standard.In contrast, the J1 and J2 of CompactPCI are replaced withimproved connectors. Power delivery is achieved using a 7-pinUniversal Power Module (UPM). The UPM is capable of deliveringover 400 W of power to individual cards. The high-speedPCI Express interconnect is achieved with a 3-row AdvancedDifferential Fabric (ADF) connector. Two ADF connectors areused to provide up to 120 Gbps of available PCI Express bandwidthwith to the backplane. A mini enriched 2 mm hard metric(eHM) functions in several capacities depending on the slot inwhich it is used. The eHM is a keyed connector that can providerear I/O in 3U card form factors, power to low power (

SPECIALHIGH-END EMBEDDED COMPUTINGtions. The processing power of such systems currently reachesdual Xeon capability from One Stop Systems and other PICMGmembers.Figure 3XMCSeveral other smallform factor PCIExpress architectureswill prove useful tothe high-end embeddedsystem designer.A joint effort betweenPICMG and VITA isunderway to upgradethe PCI Mezzanine Card(PMC) standard to handlePCI Express as wellas other high-speed fabricsignaling. The base standardis known as PICMGXMC.0 or VITA 42 in therespective organizations.Collectively referred to asXMC, the standard definesa small form factor for processorsand I/O boards thatfollows the exact mechanicalfootprint of the PMC standardwith the addition of a high-speedfabric connector. The boardfootprint remains the same as thePMC card at 74 mm x 149 mmfor a single width card. A sampleXMC is shown in Figure 3.A combination of XMC, PMC, and processor-enabled versionsof these standards delivers benefits to the high-end embeddeddesigner similar to those produced by CompactPCI Express orCOM Express. XMCs are designed to enhance a CompactPCIExpress system by adding functionality to a baseboard that is connectedto an embedded backplane. Designers also utilize XMCsas standalone modules connected to a custom carrier card in anembedded system. Like COM Express, this carrier card can beapplication specific due to functionality or mechanical requirements.At 20 Gbps available bandwidth per XMC connector andwith power consumption ratings from 7 W to 20 W, the XMCstandard gives embedded designers a powerful tool.MicroTCAMicroTCA is a specification under investigation in the PICMGaimed at aiding the embedded designer. This specification is anextension to the PICMG Advanced Mezzanine Card (AMC)standard. AMCs are the mezzanine form factor of choice for theAdvancedTCA specification due to advanced features such as highspeedswitched fabric, hot-swap, and IPMI system managementsupport. AMCs accommodate both processor and I/O functionality.The board area of an AMC is roughly the same as a 3U CompactPCIExpress card but has several choices of interconnect fabric includingPCI Express, RapidIO, or Ethernet. MicroTCA aims to adaptthe AMC mezzanine standard into a standalone, embedded architecturewith a high-speed serial fabric interconnect.ConclusionPCI Express will become a valuable tool for the high-end embeddedsystems designer as the standards begin to release over thenext few months. CompactPCI Express and MicroTCA embeddedbackplane based solutions offer a standard, modular, frontplug form factor design option for high-end processors in smallareas. XMC and COM Express offer mezzanine/carrier form factorsfor flexible baseboard design. PCI Express Cable reopens achapter on cabled serial buses with higher performance than waspossible with RS-232/422/485 or USB. In addition, PCI ExpressCable can be combined with any (or several) of the other formfactors to add an extra dimension to the architecture of the highendembedded system.The rewards of increased performance and flexibility coupledwith the abundance of form factors available in PCI Expresscomes at the cost of some added complexity. With XMC andMicroTCA, manufacturers have the option of choosing serialfabrics other than PCI Express. Compatibility between moduleswith different fabrics must be considered. Also, several formfactors have similar features and size that make for challengingarchitecture choices.Systems manufacturers certainly must accept a more consultativerole in overall system design with many new PCI Expressarchitectures to choose from. The availability of off-the-shelfdevelopment systems that are application-ready, integration servicesbased on standards based building blocks, and fast systemturnaround times become critical factors in choosing a manufacturingpartner.Jim Ison is the product marketing manager for One StopSystems and has more than 10 years’ experience in the busboardmarketplace. Prior to One Stop Systems Jim has heldvarious sales and marketing management positions centered onindustrial and converged communications accounts for ZiatechCorporation and Rittal Corporation. More recently, he hasheld the global positions of CompactPCI product manager anddirector of OEM business development with I-Bus. Jim holds abachelor’s degree in Aeronautical Engineering from CaliforniaState Polytechnic University at San Luis Obispo.For further information, contact Jim at:One Stop Systems2235 Enterprise St. #110Escondido, CA 92029Tel: 760-745-9883 x1647Fax: 760-745-9824E-mail: jison@onestopsystems.comWebsite:www.onestopsystems.com40 / CompactPCI and AdvancedTCA Systems / June 2005

RSC# 41 @www.compactpci-systems.com/rscCompactPCI and AdvancedTCA Systems / June 2005 / 41

APPLICATIONMONITORING AND CONTROLUsing remote upgrades to increase revenue anddecrease costs in wireless base stationsBy David GambaThe 3G rollout is now in full force and will be drivingnew data service requirements that will enable wirelessoperators to stem their ARPU decline. David arguesthat base station designs will need to offer moreflexibility to meet the increasingly technically complexrequirements for delivering next generation data services.Though the 3G rollout has been delayed from original prognostications,the transition to 3G is moving very quickly now. Accordingto iSuppli, 2004 marked the last year that any infrastructure dollarswill be spent installing a 2G network, and 2006 will be the last yearfor any investment (and less than five percent at that) for a 2.5Gnetwork installation. The transition to 3G means associated higherspeed standards can now be supported. The time is ripe for mobileoperators to capitalize on the latest standards and quickly beginoffering higher revenue generating data services to their customers.To this end, these operators can take advantage of a new flexiblebase station technology based on Field Programmable Gate Arrays(FPGAs) that enables in-field upgrades to support the latest industrystandards and handset features. This flexible FPGA technologycan also provide innovative ways to reduce operators’ costs, helpingincrease profitability.Wireless network evolutionAs the Third Generation Partnership Project (3GPP) standardsmove from the 2G and 2.5G based standards of CDMA,GSM, GPRS, and TDMA to the 3G based standards of EDGE,CDMA2000, 1xDO-EV, and W-CMDA, new enriched data servicesand advanced functionality will be available. The enhancedrevenue generating services will include messaging, photo transmission,e-mail, Internet access, motion video transmission, ande-commerce. What’s more, advanced features such as Quality-of-Service (QoS) guarantees and bandwidth-on-demand adjustmentcapabilities will supplement these services, giving rise to morecombinations of revenue generating packages for the wirelessoperators.Reversing the trend of declining user revenueThe new data services and feature offerings will serve as a boonto wireless operators as they can now stabilize and reverse theireroding subscriber ARPU (see Table 1). This is especially importantin regions with very high penetration rates such as WesternEurope (79 percent), Japan (69 percent), and North America(58 percent), where new subscribers will not provide the growthnecessary to drive revenues. In addition, by offering new dataservices and advanced features, wireless operators may be ableto reduce their churn rate (especially in regions dominated byprepaid subscribers who do not have monthly contracts) by offeringunique pricing packages.Addressing technical requirements using remoteupgradesTo enable these advanced data services offerings, the big questionfacing the wireless infrastructure industry is: What is thebest way to deliver flexible, cost-effective solution deploymentsto meet these requirements? Given that the 3GPP standards arestill evolving and that distinct geographical variations will existfor quite some time, wireless base station designs are incorporatingmore programmable technologies such as FPGAs intheir designs. For example, the 3GPP Release 5 added a featurecalled High Speed Downlink Packet Access (HSDPA) as a newUniversal Mobile Telecommunications System (UMTS) requirementin its baseband processing specification for Wideband CodeDivision Multiple Access (W-CDMA). This new feature enablesbase stations to transmit data to the handset units at a peak rate of14.4 Mbps, a sevenfold increase over the previous downlink ratesupported by Release 4 (2 Mbps). This performance increaseenables more advanced data services that will help wirelessoperators raise ARPU. Adding HSDPA to the 3GPP standardrequired upgrading deployed base station units. Operators easilyupgraded some base station designs by implementing an infieldupgrade to the FPGAs on the baseband card to add supportfor the HSPDA feature. Other designs, using inflexible ASICs,required either lengthy redesigns and verification efforts beforerespinning a new ASIC or a complementary device and basebandboard redesign to support this new feature.The HSDPA support issue offers a perfect example of how FPGAdevices speed time-to-market product delivery and enable flexibilityfor field upgrades to support future standard changes oradditions. Turning to FPGAs represents a long overdue shiftaway from using ASIC technology, which does not offer the abilityto future-proof deployments against standards changes or theflexibility to support geographic customization.Average Revenue Per User (ARPU)1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006Voice $82 $66 $54 $45 $39 $35 $33 $32 $31 $30 $29Data - - - - $1 $2 $4 $6 $8 $10 $13Total $82 $66 $54 $45 $40 $37 $37 $38 $39 $40 $42Table 142 / CompactPCI and AdvancedTCA Systems / June 2005

RSC# 43 @www.compactpci-systems.com/rscCompactPCI and AdvancedTCA Systems / June 2005 / 43

44 / CompactPCI and AdvancedTCA Systems / June 2005 RSC# 44 @www.compactpci-systems.com/rsc

APPLICATIONFor base stations implementing these new data service requirements,which support the higher speeds demanded by the downlink(HSDPA) and uplink (High-Speed Uplink Packet Access, orHSUPA, technologies coming in a future 3GPP release) speeds,baseband throughput, and processing power must increase.Processing power must grow to support the additional algorithmicrequirements driven by the data service requirements andby an increasing number of users per base station. These designrequirements dictate the use of components, such as FPGAs, thatcan effectively support single-chip parallel processing operations,as ASICs are not flexible enough for cost-effective deployment.Designers can effectively use FPGAs in the baseband modules ofthe wireless base station to implement the required performancelevels. Using parallel processing techniques enables leveragingdedicated integrated signal processing functional blocks. Thesecapabilities allow for flexible solutions that help reduce chipcount and lower power inside the baseband module.Reducing base station operating costs usingprogrammabilityFPGA programmability can also help significantly lower operatingcosts by offering operators increased power efficiencies duringoff-peak times. Wireless operators need to deploy enough basestations and remote radio antennas to support traffic loads duringpeak usage times. If the operator cannot support a handset user’srequest for a connection when the network is heavily loaded, thenthis service quality issue may drive the user to switch to anotherservice. Thus wireless operators are forced to either sufficientlybuild out their own network to support peak loading times or rentenough usage time from an existing infrastructure to meet theirpeak loading needs. For reference purposes, a typical wirelessbase station network is depicted in Figure 1.Figure 1Not surprisingly, the traffic load on a wireless network decreasessignificantly during the late night and early morning hours. Incertain locations, the traffic load also decreases on the weekendsand holidays as well. This loading imbalance makes it possiblefor wireless operators to balance their network during this time.To successfully implement a power-balancing configuration, thewireless base stations must contain the flexibility to perform thefollowing energy saving sequence:1. Do not accept any new transactions.2. Complete all existing transactions.RSC# 45 @www.compactpci-systems.com/rscCompactPCI and AdvancedTCA Systems / June 2005 / 45

APPLICATIONMONITORING AND CONTROL3. Power down all unused sectors in the base station.4. Maintain status monitoring to receive command statusupdates.5. Power up when minimal load activity session activitylapses.This type of flexibility is likely to be implemented in a stagedapproach in a real network, as one wireless base station afteranother is powered down (or removed from the network in therental model) as the traffic load decreases, until a minimal configurationis reached. As an example, suppose that in the networkshown in Figure 1 that 18 out of the 26 base stations can be powereddown or removed from this network. This leads to a newnetwork with the remaining base stations covering an expandedarea as shown in Figure 2.Figure 2Powering down (or removing) these 18 base stations can providea power savings in the network, helping lower the operator’soperating expenses. To accurately estimate the power savings,one needs to note that the base stations remaining in the networkwill experience an increase in power due to base stations beingremoved from the network. The additional range needed to supportgeographical coverage for handsets formerly supported bybase stations that have been removed from the network requiresextra transmission power. Using an industry standard model presentedin Wireless Communications: Principles and Practices [1](and assuming an environmental factor of 2), the signal powerrequired to travel across a distance d from a transmitter to receiveris described by Equation 1.p ij~ (d ij) 2 (1)Equation 1The significance of this model is that if the transmission distanceis doubled by a factor of two, the power increases by a factorof four. However, this power increase is limited to the radiomodule of the wireless base station, which accounts for roughly30 percent of the wireless base station power consumption. Aquick power savings calculation reveals that the power consumptionof the minimal network shown in Figure 2 will be a littleunder 60 percent of the peak load network shown in Figure 1,which allows for an over 40 percent power savings during theoff-peak times.Summary3G will generate new data service requirements that will enablewireless operators to stem their ARPU decline. Given the constantlyevolving wireless standards and the recognized need forgeographic customization, programmable technologies are rapidlyreplacing traditional ASICs. This trend will continue as basestation designs offer additional flexibility to address more andmore complex technical requirements for delivering next generationdata services. At the same time, base stations must maintainthe versatility to avoid obsolescence or limited deploymentby adapting to standards changes and geographic variations. Inaddition, these programmable technologies also offer operatorsan opportunity for cost savings by using the programmable flexibilityto manage their networks through the service periods.David Gamba is senior marketing manager for the StrategicSolutions Marketing Group at Xilinx. In this role, David isresponsible for outbound marketing for all vertical marketssupported by Xilinx solutions. David joined Xilinx in 2004 andbrings more than eight years of experience in the semiconductorindustry, where he served in a variety of marketing and salesroles including technical sales, product definition, and technicalmarketing. Prior to Xilinx, David held various positions atAeluros, Conexant, and Altera. He holds a bachelor’s degreein electrical engineering from UCLA, a master’s degree inelectrical engineering and computer science from UCBerkeley, and an MBA degree from Stanford University.References[1] Dr. Ted Rappaport, Wireless Communications:Principles and Practices, 1996, Prentice HallFor further information, contact David at:Xilinx2100 Logic Dr. • San Jose, CA 95124Tel: 408-879-6146 • Fax: 408-371-4926E-mail: david.gamba@xilinx.comWebsite: www.xilinx.com46 / CompactPCI and AdvancedTCA Systems / June 2005

RSC# 47 @www.compactpci-systems.com/rscCompactPCI and AdvancedTCA Systems / June 2005 / 47RSC# 47 @www.compactpci-systems.com/rsc

PRODUCT GUIDESBCJust what is a blade, anyway?By J. Eric GulliksenThere is a tremendous amountof ambiguity surrounding theterm “blade,” particularly inthe embedded space. It has become amarketing buzzword used to describea variety of different product types,which has created confusion rather thandifferentiation in the marketplace. Thisarticle presents VDC’s definition of theterm, the logic behind this definition,and the ways in which we differentiatebetween blades and other embeddedboard types.BeginningsVDC first encountered the term bladeduring the course of research for ourfirst report on Switch Fabrics and High-Speed Serial Interconnects, publishedin November of 2001. There appearedto be an intimate association betweenthese interconnect technologies and theterm blade. However, there was no cleardefinition for blades, blade architectures,or blade-based systems and, as the termbecame more widely used, this associationwith fabric technology started to becomeblurred. Definitions for blade foundin various glossaries on the Web couldbe applied to Single Board Computers(SBCs) in general. We began to ask engineeringand marketing professionals inboth the embedded board and enterprisecommunities for their definitions in anattempt to arrive at a consensus. Some ofthe divergent responses we received fromthe embedded industry included:■ Just another, sexier word for board■ A board with some sheet metalwrapped around it■ A single board computer that has beenoptimized as a server■ The combination of a carrier boardand a PMC card■ A PICMG 2.16 SBC■ An expansion card that plugs into amotherboard■ Another name for a 1U pizza boxserverHowever, many of the other responsesincluded an element of commonality, inthat they did cite the use of high-speedserial or switch fabric interconnects as theprimary means of interboard data communication.Discrepancies within thisgroup were primarily related to the use ofshared, parallel buses as additional meansof interboard communication, or to functionaltypes.The Enterprise space showed a greaterdegree of clarity, although most Enterpriseclassblades were of proprietary architecturesor form factors. Here, the datacommunication means between bladeswas limited to high-speed serial or fabricinterconnects, with little differentiationbetween functional types.The VDC definitionKeeping these considerations in mind,VDC developed the following workingdefinition for an embedded componentlevelblade.An embedded component-level blade is acomputer board with the following characteristics:1. It is designed to be inserted, usuallyvertically, in a slot in a card cage orchassis mounted on a rack.2. It connects to a passive backplane,and communicates data to other boardlevel components in the immediatehost system only via a switch fabric orother high-speed serial interconnect.Any shared, multidrop, parallel databus that may be present is local tothe blade and is not carried to thebackplane.We therefore base our definition on thestructure of the interconnect architectureand do not limit it in any way by functionaltype, application, local bus, formfactor, feature set, or any particular highspeedserial or fabric interconnect technologyor topology.VDC presented this definition to severalof the individuals that we had previouslyinterviewed, most of whom agreed thatit made sense and provided much-neededclarity and differentiation betweenblades and boards using parallel busesas an interboard communication means.We then presented the definition verballyat the Bus & Board Conferencein January of 2002, and it has now beenaccepted by a majority of the embeddedboard industry.(Note that certain blade servers, whichinclude 1U, 2U, and 4U devices, do notcomply with provision 1 of this definitionand may communicate with other bladeservers via fiber or cable, without a backplane.We consider these to be systemlevel,not component-level, devices.)SpecificationsOf the various open standards in existenceto date, only AdvancedTCA (PICMG 3.x)comprises a true blade specification.CompactTCA will also be a blade specification.Other standards such as PICMG2.16 and 2.17 have provisions that allow,but do not mandate, blade configurationsunder the VDC definition. A CompactPCIsingle board computer example may helpto clarify this gray area. Note that we willuse the term switch fabric generically toinclude both fabrics and other high-speedserial interconnect means that may not befabrics. In addition, the VDC definitionallows for other functional types (such asI/O and mass storage) to be substitutedfor SBCs, and allows for PICMG 2.16and Ethernet Fabric to be replaced byother specifications and technologies, asappropriate. Thus, 2.16 may be replacedby 2.18 and Ethernet Fabric by RapidIO,for example.”Traditional” CompactPCI SBCs do notinclude switch fabric access. Therefore,these cannot be blades under the VDCdefinition, and there is no ambiguity.PICMG 2.16-compliant SBCs do includeEthernet Fabric access via the P0 connector.In other words, these are fabricenabled.Fabric-enabled SBCs may ormay not be blades, depending on theirconfiguration:■ If these SBCs carry both the sharedPCI bus and the Ethernet Fabric to thebackplane as is permitted by the specification,these are not blades. We callthese nonblade fabric-enabled SBCs.These configurations allow backward48 / CompactPCI and AdvancedTCA Systems / June 2005

compatibility with legacy backplanesand systems.■ If, as is also allowed under thePICMG 2.16 specification, the sharedPCI bus is not carried to the backplane,these SBCs are blades underthe VDC definition.What if a fabric-enabled PICMG 2.16SBC is used with a backplane that doesnot have the capability of connecting toor carrying the PCI bus between boards?Does this make the SBC a blade? No. Inthis case, the board is still a fabric-enabledSBC, but it is being used as a blade.Recent study findings onCompactPCI SBCs andCPU bladesVDC’s newly published report, MerchantComputer Boards for Embedded/RealTime Applications Market IntelligenceProgram, 2004: Volume V: Overview,indicates that, to date and other thanAdvancedTCA, the only standards-basedblades available are of the CompactPCIlocal bus architecture and form factor.Table 1 shows the dollar volume shipmentshares of CompactPCI SBCs, in 2004,Configuration“Traditional” CompactPCIFabric-enabled nonblade SBCsCompactPCI CPU bladessegmented into the three configurationsmentioned earlier.As a whole, shipment shares of fabricenabledSBCs, including blades, areexpected to continually increase. Therelationship between shares of blade andnonblade fabric-enabled configurationsis, however, projected to remain relativelyconstant until the CompactTCA specificationbecomes finalized. Ultimately,shipments of CompactPCI CPU bladesare expected to overtake those of nonbladefabric-enabled SBCs, with thelatter becoming relegated to a transitionarchitecture.J. Eric Gulliksen been with VDCsince October of 1999 and is currentlypractice and project director for theEmbedded Hardware discipline, whichShipment shares by percent of $ volumeTable 143 percent44 percent13 percentincludes Merchant Computer Boards andIntegrated Systems for Embedded andReal-Time Applications. He holds BSEEand MMgS&E degrees from WPI, andan MBA from Clark University. Eric hasbeen awarded 17 US Patents, and hasinternational field experience in22 countries.For further information, contactEric at:Venture Development CorporationOne Apple Hill DriveSte. 206, Box 8190Natick, MA 01760Tel: 508-653-9000E-mail: ericg@vdc-corp.comWebsite: www.vdc-corp.comRSC# 49 @www.compactpci-systems.com/rscCompactPCI and AdvancedTCA Systems / June 2005 / 49

RSC# 50 @www.compactpci-systems.com/rsc50 / CompactPCI and AdvancedTCA Systems / June 2005RSC# 50 @www.compactpci-systems.com/rsc

PRODUCT GUIDESBCProcessorFeaturesProcessorFeaturesCompany name/Model numberPentium IIIPentium MPowerPCXeonAdvancedTCAPICMG 2.16AMC sitesPMC sitesCompany name/Model numberPentium IIIPentium MPowerPCXeonAdvancedTCAPICMG 2.16AMC sitesPMC sitesAcqiriswww.acqiris.comPC502 • •Actiswww.actis-computer.comcSBC-6440 • 2CSBC-6872Ax/144-16 • 2ADLINK Technologywww.adlinktech.comATCA-6890 • • 2cPCI-3500A•cPCI-3700A•cPCI-3840 Series•cPCI-6780•cPCI-6810 • • 2cPCI-6820 • •cPCI-6840•cPCI-6860•NuPRO-900•Advantechwww.advantech.comMIC-3351•MIC-33513U•MIC-3357•MIC-3365•MIC-3366 • • 2MIC-3369 • •MIC-3385 •MIC-3389 • •MIC-3369A • •Aitechwww.rugged.comS950 Space Processor•Artesyn Communicationwww.artesyncp.comBajaPPC 750•Katana 752i • • 2KatanaQp • •PM/PPC-750•PmPPC7447•Axiomtekwww2005.axiomtek.comAXIOMTEK SBC83810•Ballard Technologywww.ballardtech.comOmniBus cPCI•Centralp Automatismeswww.centralp.comKPCI6U-586PMC•CESwww.ces.chConduction Cooled RIOC 4070 • 2MFCC 8442•MFCC 8443•MFCC 8447•RIOC 4065•RIOC 4068•RIOC 4070•RIOS 2476•Cluster Labswww.cluster-labs.comCPU 410•Computer Moduleswww.computermodules.comSC2060/SC2050•SC2210•Concurrent Technologieswww.gocct.com2xPMC, Pentium M, SBC • • 2PP 100/01x • •PP 110/01x • • 2PP 120/01x • • 1PP 120/11x • • 1PP 220/01x • • 1PP 312/012 • • 2PP 312/01x • 2PP 332/02x•PP CP1/P3x • •PP CP2/P3x • • 1PP EMB/P34 • 2PP PSE/P31 • 1PP SC2/P3x • 1Continuous Computingwww.ccpu.comLINUXblade XE20 • • 2Curtiss-Wright Embeddedwww.cwcembedded.comG4C – cPCI SBC • •CompactCore '119•DPMC-106•PPC G4C•SCP/DCP-122•Continued on page 52CompactPCI and AdvancedTCA Systems / June 2005 / 51

PRODUCT GUIDESBCProcessorFeaturesProcessorFeaturesCompany name/Model numberPentium IIIPentium MPowerPCXeonAdvancedTCAPICMG 2.16AMC sitesPMC sitesCompany name/Model numberPentium IIIPentium MPowerPCXeonAdvancedTCAPICMG 2.16AMC sitesPMC sitesDiversified Technologywww.dtims.comATC-4130 • • 2DMD Computerswww.dmd.itDMD I815-C•DNA Enterpriseswww.dna-cs.comVS750•Dynatemwww.dynatem.comCHC•CPC2•EKF-Electronikwww.ekf.deCC2-TANGO•CC7-JAZZ•CC9-SAMBA•CCF-Concert•CD2-BEBOP • • 1ELTEC Electronikwww.eltec.comBAB 750 • 1EUROCOM 248•Eonic B.V.www.eonic.comAtlas3-G4•esdwww.esd-electronics.comCPCI-405•EuroTecHwww.eurotech.itCPU-7630/7631•CPU-7635 • • 2Extreme Engineeringwww.xes-inc.comXPedite1032•Fastwelwww.fastwel.comCPC501•CPC502•GE Fanuc Automationwww.gefanuc.com/embeddedCPCI-7506•PMC721TX•VMICPCI-7505•VMICPCI-7699 • 2VMICPCI-7710 • 1VMICPCI-7715 • 1VMICPCI-7716 • 1VMICPCI-7753 • 1GE Fanuc Automation (continued) www.gefanuc.com/embeddedVMICPCI-7755 • 2VMICPCI-7756 • 2VMICPCI-7757 • 2VMICPCI-7760•VMICPCI-7761 • • 1VMICPCI-7806 • • 2VMIVME-7807 • 1General Dynamicswww.gdcanada.comPC6010 • 1General Micro Systemswww.gms4vme.comC161 Aurora • •C190/191 Atlantis-C •C2000 Millennium •C261 Aurora II •C269 Equinox • •C394 Maverick • 2C50x Web-LC • • 1CX269 • 1Mariner II C158•P60x•GESPACwww.gespac.chPCIPPC-1•PCIPPC-2•PCIPPC-2X•PCIPPC-5 •PCISYS-56AE•PCISYS-58•PCISYS-58X•PCISYS-60•PCISYS-PII/III•I-BUSwww.ibus.comIBC 2600•IBC 2601•IBC 2602•IBC 2801 • • 1IBC 2802 • • 1Inovawww.inova-computers.comICP-(M)PIII•52 / CompactPCI and AdvancedTCA Systems / June 2005

PRODUCT GUIDESBCProcessorFeaturesProcessorFeaturesCompany name/Model numberPentium IIIPentium MPowerPCXeonAdvancedTCAPICMG 2.16AMC sitesPMC sitesCompany name/Model numberPentium IIIPentium MPowerPCXeonAdvancedTCAPICMG 2.16AMC sitesPMC sitesInova (continued)www.inova-computers.comICP-CM/ICP-PM•ICP-PIII•ICP-PM•ICP-PPC•Intelwww.intel.comNetStructure MPCBL0001•NetStructure MPCBL5525 • •NetStructure ZT 4807•NetStructure ZT 5504•NetStructure ZT 5515 • •NetStructure ZT 5524•Interface Amitawww.interface-co.comCPZ-PE09 Series•Kontronwww.kontron.comAM4001•AT8000 • • 1AT8001 • 2CP301 • 1CP302-PM•CP303•CP304 • •CP306•CP320•CP321•CP6000 • • 1CP6010 • 1CP6011 • • 2CP604•CP620-PM •cPCI-DMXS64•cPCI-DT64 • •cPCI-DXS64•cPCI-MXP64GX•cPCI-MXS64 • 1cPCI-MXS64GX •DT64 • • 1EB8245 •ETXexpress-PM•Kontron (continued)www.kontron.comMXs64GX•VisionCompact IA•Maxwellwww.maxwell.comSCS750•MEN Microwww.menmicro.comA12c•D3 • 2D3a•D3b•D3c•EM02•EM04•EM04N•F1N•F6•F7•F7N•F9•Mercury Computer Systemswww.mc.comRACE++ 800 MHz PowerPC 7447•RACE++ AdapDev•RACE++ PowerPC 7410•Microbuswww.microbus.comMAT 1019•Miriacwww.miriac.comCPC45 • •Momentum Computerwww.momenco.comCheetah-Cr•Civet-C • 2Puma•Puma-CR • 2Motorolawww.motorola.com/computersATCA Blade • •ATCA-715/717 • • 4CPCI-680•CPCI-740•CPCI-745 • • 1Continued on page 55CompactPCI and AdvancedTCA Systems / June 2005 / 53

54 / CompactPCI and AdvancedTCA Systems / June 2005RSC# 54 @www.compactpci-systems.com/rsc

PRODUCT GUIDESBCProcessorFeaturesProcessorFeaturesCompany name/Model numberPentium IIIPentium MPowerPCXeonAdvancedTCAPICMG 2.16AMC sitesPMC sitesMotorola (continued)www.motorola.com/computersCPV5370•CPV5375 • 2MCP750•MCP820•MCPN750 • 2MCPN765•MCPN805•MPC8540 PowerQUICC III•PowerCore CPCI-6750•PowerCore CPCI-680•PowerCore CPCI-690 • •PowerCore CPCI-690+•PowerCore CPCI-695•PowerCoreCPCI-6750•PPMC750•PrPMC750•MPLwww.mpl.chIPM6•N.A.T.www.nateurope.comNICE-360•National Instrumentswww.ni.comPXI 8176•PXI-8175•PXI-8175 RT•PXI-8176 RT•NEXCOM Internationalwww.nexcom.comMAXI 6600 • 1MAXI 6750 • • 1One Stop Systemswww.onestopsystems.comMillennium Gold • 1Orion Technologieswww.otisolutions.comCPC7510•PMC7500•Performance Technologieswww.pt.comCPC5505 PICMG 2.16 SBC•ZT 5503 SBC • 1ZT 5504e SBC • • 1Company name/Model numberPentium IIIPentium MPowerPCXeonAdvancedTCAPICMG 2.16AMC sitesPMC sitesPerformance Technologies (continued)www.pt.comZT 5515e SBC•ZT 5524e SBC•ZT 5551 SBC•Portwellwww.portwell.comTANC-5260 • •Prodrivewww.prodrive.nlP3P4403•RadiSys Corpwww.radisys.comProcelerant CE•Radstone Embedded Computingwww.radstone.comCP1A • •CP1A 6U•IMP1A•IMP2A•Sanritz Automationwww.sanritz.co.jpSC2050•SBEwww.sbei.comHW400C/M DKL•SBS Technologieswww.sbs.comC5C•CC7•CE7•CK3 • 2CK3-TM•CK5•CL9-cPCI 3U SBC•CM4 •CP7 •CP9•CR9 •CT7 • 2CT8 • •CT9•RL4•Siemenswww.siemens.comCPCI-CPU076•Continued on page 56CompactPCI and AdvancedTCA Systems / June 2005 / 55

PRODUCT GUIDESBCProcessorFeaturesProcessorFeaturesCompany name/Model numberPentium IIIPentium MPowerPCXeonAdvancedTCAPICMG 2.16AMC sitesPMC sitesCompany name/Model numberPentium IIIPentium MPowerPCXeonAdvancedTCAPICMG 2.16AMC sitesPMC sitesSMAwww.SMAcomputers.comCPU 6.2•CPU 7.2•Smart Modular Technologieswww.smartm.comSMARTengine/603ecPCI • 1SMARTengine/750cPCI-6U•Spectrum Signal Processingwww.spectrumsignal.comPRO-3500•Synergy Microsystemswww.synergymicro.comKGM5•KYMD•Technolandwww.technoland.comTL-SBC 7450•Thaleswww.thalescomputers.comCPU860-MD/MR/MM•PMC860•RA and RC PowerEngine7•VMPC6a•VMPC6c • 1Transtech DSPwww.vmetro.com3CPF1•CR9•Trenton Technologywww.trentontechnology.comCP10 • 1CP16•TriEMSwww.triems.comTRL6227 • 1United Electronic Industrieswww.ueidaq.comPDXI-C-P400/P700•Voiceboardwww.voiceboard.comPMC750•VRose Microsystemswww.vrosemicrosystems.comVRM-CC7-X•VRM-CD1-X • • 1There’s more you can do online!•RSC# 56 @www.compactpci-systems.com/rscRead the latest industry news• Search and sort thousands ofCompactPCI and AdvancedTCA Products• Use the “RSC” numbers in this catalogto receive product information directlyfrom vendors56 / CompactPCI and AdvancedTCA Systems / June 2005

RSC# 57 @www.compactpci-systems.com/rscCompactPCI and AdvancedTCA Systems / June 2005 / 57

58 / CompactPCI and AdvancedTCA Systems / June 2005 RSC# 58 @www.compactpci-systems.com/rsc

NEW PRODUCTSNEW PRODUCTSBy Chad Lumsdenwww.compactpci-systems.com/productsCompactPCI &AdvancedTCAARINCCurtiss-Wright EmbeddedWebsite: www.cwcembedded.comModel: P429A PMC RSC No: 20220ARINC 429 communications controller • CompleteARINC interface • Multiple serial communicationschannels in a single PMC module • AMD85C30 serial communications controller • DDC00429/3282/3182 ARINC 429 solution chipset •Custom FPGA bridges the industrial standard PCIinterface bus to these I/O devices and 4 Mbytes ofFlash • Eight DMA channels, four for serial transmitand four for serial receive channels, to reducethe overhead processing of these serial channelby the processor on the host board • SeparateFPGA provides redundancy logic, interrupt handler,watchdog timer, and an external pulse counter• Suitable for adding serial and ARINC 429interfaces to the host processor board with anavailable industrial standard PMC slotBACKPLANECarlo Gavazzi CSWebsite: www.gavazzi-computing.comModel: CompactPCI Backplanes RSC No: 204402-16 slots standard, 2-21 slots custom, and 6U(10-layer) or 7U (12-layer) format • CT bussing(ECTF H.110) • 5 volt/3.3 volt supported • ATXpower supply connectors (7U format), powerblocks (6U format) • PICMG 2.16 and 2.17 backplanesalso available • Hot swap compatibleRSC 20440▲ELMA BustronicWebsite: www.elmabustronic.comModel: ATCA Backplanes RSC No: 20382Controlled-impedance stripline design • Dualstar, mesh, and replicated mesh configurationsavailable • Slot size of 2, 5, or 14; both 5U and7U heights available; other sizes available •Simulation/characterization studies confirm excellentsignal integrity; unique AdvancedTCA probecard • Custom AdvancedTCA designs • Signalintegrity studiesBACKPLANE: FULL MESHKaparelWebsite: www.kaparel.comModel: AdvancedTCA Backplane RSC No: 2038514-slot fabric interface with full mesh interconnect• Base interface with dual star interconnect• Split power distribution (odd slots on A1/B1,even slots on A2/B2) • Bused IPMI-0 connections(optional configuration allows for radial connections)• Synchronization clock interface on P20 •Metallic test and ring generator buses on J10RSC 20385XILINXWebsite: www.xilinx.comModel: Xilinx ATCA Platform RSC No: 20387Four-channel, four-port (16 MGTs) full mesh fabricinterface • Supports IPMI interface and baseinterface ShMC port • Headers for applicationspecificpersonality module • Fully distributedsystem management architecture • Supportsmanagement firmware running on IBM PowerPCprocessor immersed in Virtex-II Pro FPGA family •Supports Linux-based control plane softwareRSC 20387▲▲BLADES: SERVERDiversified TechnologyWebsite: www.dtims.comModel: Targa-14 RSC No: 2042114-slot AdvancedTCA communication solution forservice providers and high-speed data networkapplications • Single/dual Intel Xeon processorbasednode blades with speeds up to 2.8 GHz and1 MB L2 cache • Separate data/control transferaccomplished via hub switch blades • MontaVistaLinux Carrier Grade Edition (CGE) BSP for CPUnode blades • Redundant management in eitherradial or bussed modes • Dual star and full meshbackplanesBOARD ACCESSORIESZephyr EngineeringWebsite: www.zpci.comModel: ZPCI.246 RSC No: 20545Onboard bridge maintains CompactPCI signalintegrity • Onboard PMC slot for PCI logic analyzer/exerciser• All CompactPCI and user I/Osignals are individually isolatable • Supports PMCRSC 20545▲For further information,enter the product’s RSC# atwww.compactpci-systems.com/rscARINC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59Backplane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59Backplane: Full mesh. . . . . . . . . . . . . . . . . . . . . . . 59Blades: Server. . . . . . . . . . . . . . . . . . . . . . . . . . . . 59Board accessories. . . . . . . . . . . . . . . . . . . . . . . . . 59Carrier board: PMC . . . . . . . . . . . . . . . . . . . . . . . . 60Chips & Cores: Other . . . . . . . . . . . . . . . . . . . . . . 60Connector: Hard metric. . . . . . . . . . . . . . . . . . . . . 60Connector: Mezzanine . . . . . . . . . . . . . . . . . . . . . . 60Development platform. . . . . . . . . . . . . . . . . . . . . . 60DSP resource boards: CompactPCI. . . . . . . . . . . . 60Enclosure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60Enclosure + card rack + power supply. . . . . . . . . . 60Fabrics: Fibre Channel. . . . . . . . . . . . . . . . . . . . . . 61Fabrics: InfiniBand . . . . . . . . . . . . . . . . . . . . . . . . 61Fabrics: Switched Fabric . . . . . . . . . . . . . . . . . . . . 61Front-panel hardware . . . . . . . . . . . . . . . . . . . . . . 61I/O: Analog . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62Mass Storage: IDE. . . . . . . . . . . . . . . . . . . . . . . . . 62Mass storage: Plug-in unit . . . . . . . . . . . . . . . . . . 62Mass storage: RAID . . . . . . . . . . . . . . . . . . . . . . . 62Memory: Flash . . . . . . . . . . . . . . . . . . . . . . . . . . . 62Memory: General purpose. . . . . . . . . . . . . . . . . . . 62Motion control . . . . . . . . . . . . . . . . . . . . . . . . . . . 62Power supply . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62Processor blades . . . . . . . . . . . . . . . . . . . . . . . . . 62Processor: Pentium 4 . . . . . . . . . . . . . . . . . . . . . . 62Processor: Pentium III . . . . . . . . . . . . . . . . . . . . . 64Processor: Pentium M. . . . . . . . . . . . . . . . . . . . . . 64Processor: PowerPC . . . . . . . . . . . . . . . . . . . . . . . 64Processor: Xeon . . . . . . . . . . . . . . . . . . . . . . . . . . 64Prototyping and debugging: Boundary scan . . . . . 64Prototyping and debugging: Bus analyzer . . . . . . . 64Routers/Switches . . . . . . . . . . . . . . . . . . . . . . . . . 64SCSI peripheral . . . . . . . . . . . . . . . . . . . . . . . . . . . 64Shelf and mechanical components . . . . . . . . . . . . 64Software: Development tool . . . . . . . . . . . . . . . . . 64System management . . . . . . . . . . . . . . . . . . . . . . 65Telecom. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65Telephony: VoIP . . . . . . . . . . . . . . . . . . . . . . . . . . 65Thermal management . . . . . . . . . . . . . . . . . . . . . . 65Turnkey system. . . . . . . . . . . . . . . . . . . . . . . . . . . 65Video: Frame grabber . . . . . . . . . . . . . . . . . . . . . . 65Wireless: SDR. . . . . . . . . . . . . . . . . . . . . . . . . . . . 65CompactPCI and AdvancedTCA Systems / June 2005 / 59

NEW PRODUCTSNEW PRODUCTSuser I/O on J3-J5 • Ideal for both CompactPCI andPMC board testing • Test points for all CompactPCIsignals • Test points for all user I/O pins • Powertest points simplify current measurements •Individual indicator LEDs show board power statusat a glance • Rigid frame mates and locks withinjectors on test board • 32-bit and 64-bit configurationsavailable at 66 MHz • Short circuit protectionfor +3.3V, +5V, +12V, and –12V suppliesCARRIER BOARD: PMCTechnoboxWebsite: www.technobox.comModel: 4366 RSC No: 20537Adapts 32- or 64-bit PMC (33 MHz) for use inPCI slot • Designed for optimal signal quality• Support for rear I/O • LEDs convey status ofkey PCI bus signals and power • Accommodatesexternal power • Optional fan assembly for additionalcooling of PMCRSC 20537▲CHIPS & CORES: OTHERPotentia SemiconductorWebsite: www.potentiasemi.comModel: PS-1006 RSC No: 20409Extensive primary side monitoring deliveredthe PS-2406 across the isolation barrier via thePI-Link • -48V inrush control and timed circuitbreaker functionality • Programmable sequencingfor startup, shutdown, and fault conditions• Programmable output overvoltage (OV) andundervoltage (UV) warning and fault threshold •Three general-purpose analog inputs for monitoringtemperature and other parametersCONNECTOR: HARD METRICERNIWebsite: www.erni.comModel: Ermet Connector RSC No: 20395Fully compatible with 2 mm HM equipment • 2 mmHM hardware and accessories • Designed specificallyfor high speed differential • Optimized tracewidth and trace space • Supports speeds beyond5.0 Gbps • 4-pair 25 mm provide 40 differentialpairs/25 mm • 3-pair 25 mm provide 30 differentialpairs/25 mm • 2-pair 25 mm provide 20 differentialpairs/25 mmthe carrier board with wiping action to ensure highreliability • Integrated, high-performance Yamaichideveloped YFlex with B 2 IT interconnect technology• Base substrate is LCP material, which has avery low CTE • Contacts designed for high-speedapplications – very short stub • Supports speedsbeyond 12.5 Gbps • Low Dielectric ConstantInsulation Material: – Connector Housing: 3.10 @6 GHz – YFlex: 2.85 @ 6 GHz • Controlled impedancecontacts 100 Ω+/-10Ω • 200 mating cycles •Operating temperature: -40 °C to +70 °CRSC 20400▲DEVELOPMENT PLATFORMAudioCodesWebsite: www.audiocodes.comModel: TP-12610 SDK RSC No: 212662016 voice/fax LBR channels supporting multiplevoice coders • Optional connectivity to 16 T1/E1/J1 PSTN trunks • Dual redundant Base and Fabric(3.1) interfaces • Fabric and base switch blade •Optional application processor blade • MGCP,MEGACO, SIP, and AudioCodes proprietary API •G.168-2002 compliant echo cancellation • Realtimefax over IP/T.38 • Wide selection of vocodersincluding AMR, EVRC, G.729, G.723, and G.711 •PSTN Signaling: CAS, ISDN PRI, and SS7 layer 2termination • SIGTRAN IUA, M2UA, M3UA overSCTP • Tone detection and generation (MF, DTMF,RFC 2833) • Enhanced voice processing featuresincluding conferencing, voice detectors, andannouncementsDSP RESOURCE BOARDS:COMPACTPCIMercury Computer SystemsWebsite: www.mc.comModel: MCP3 FCN RSC No: 20317A rugged, conduction-cooled 3U CompactPCIdigital signal processing board • PowerPC 7447 @1 GHz • Virtex II Pro P40 FPGA Discovery II SystemController • Conduction cooled with options▲for air-cooled lab development • PMC site withadditional general purpose direct-connected LVDSFPGA I/O • Designed for optimal performancewith Mercury’s dual-channel Analog to DigitalConversion PMCENCLOSUREHybriconWebsite: www.hybricon.comModel: High Power Towers RSC No: 20452Cooling up to 100 watts per slot • Eight-slotCompactPCI or VITA 1.7 VME64x switch fabricbackplanes • High-quality construction in a lightweightportable design • Up to 800 watts of power• Front-access peripheral module with provisionsfor mounting CD-ROM, hard disk, and floppydrives • Custom configurations and integrationservices availableENCLOSURE + CARD RACK +POWER SUPPLYELMA ElectronicWebsite: www.elma.comModel: 12R1 COTS RSC No: 20566A rugged chassis with shielding effectiveness •Weighs 20-25% less than Elma’s previous models• Meets MIL-S-167, MIL-S-810E, MIL-S-461D,and MIL-S-901D standards • Holds 6U x 160 mmor 220 mm cards • Available in 22" and 25" depthsor custom • Wide range of backplanes in up to20 slot sizes is available in VME, VME64x, VXS,CompactPCI, or other architectures • Compliantto the IEEE 1101.10/.11 mechanical specifications• Front-to-rear cooling is achieved through a RearEvacuative Cooling system often using 2 x 470CFM Free Air Blowers • System monitoring LEDsfor DC Voltages, over-temp, and fan fail standard• Includes convenient, separate front access todrives via a removable hinge door allowing spacefor drives (or Kingston carriers) • Large patchpanel for I/O located on the rear of the chassis •2 handles per side for easy lifting • Power suppliesfor up to 350-1400 W • Optional louvered frontpanel to meet the International Protection 53 codeaccording to IEC 60529 for drip requirements • 2Uto 9U heights in horizontal and vertical orientation• Modular design allows models to be developedin 2U and 3U heights • A clear alodine coatingprovides corrosion protection and an aestheticallypleasing finish • Designed to withstand theharsh demands of a military environment • Useshoneycomb filters, braided gasketing, and metalimpregnated sheets to seal off all external seams▲CONNECTOR: MEZZANINEYamaichi ElectronicsWebsite: www.yeu.comModel: CN074 Series Connector RSC No: 20400PICMG AMC.0 Revision 1.0 compliant • GR-1217-CORE compliant • Compression style contacts toRSC 20317RSC 2056660 / CompactPCI and AdvancedTCA Systems / June 2005

• Withstands over 15 g’s of shock and vibration •Various options of rope-coil isolators, air springs,and elastomeric isolators are available • Suitablefor making custom modifications quickly, easily,and cost-effectivelyFABRICS: FIBRE CHANNELSANBlazeTechnologyWebsite: www.sanblaze.comModel: PMC FibreChannel HBA RSC No: 20549Two independent, 2 Gbit Fibre Channel ports •SFP based, supports multi-mode optics and copperoptions • Auto-negotiation for legacy connect(1 or 2 Gbit) • Front and rear panel I/O options;Pim Module available • Software supports switchand loop (private and public) topologies • 64-bit,33/66 MHz PMCFABRICS: INFINIBANDMellanoxWebsite: www.mellanox.comModel: 10 Gb/s 24-Port RSC No: 2038610 Gbps 24-port InfiniBand switch • 24 InfiniBand4x 10 Gbps ports with Double Data Rate (DDR)20 Gbps capability • Full-wire-speed capableswitching core (960 Gbps) • Ports configurableinto 12x uplinks (30 Gbps or 60 Gbps DDR) • Full,open-source Embedded Linux Management Kitavailable • Ideal for VXS VITA 41 or AdvancedTCA3.2 backplane fabric • CPU interface for low-costembedded fabric managementFABRICS: SWITCHED FABRICDSS NetworksWebsite: www.dssnetworks.comModel: Model 8261 Switch RSC No: 20484Fourth-generation BCM5690 switch fabric andBCM5464SRKB quad-port transceivers fromBroadcom • High-performance wire speed on allports – 24 Gb total; up to 32,000,000 frames persecond maximum switching rate • Onboard firmwarefor configuration, management, and healthmonitoring • Cell and packet-based “head-of-line”blocking prevention; 1 MB of onboard memoryfor packet buffering • Extended Ethernet framesizes to 9 KB; fully compliant to IEEE 802.3 specifications,including auto negotiation • OnboardMotorola DSP56F826, 80-MHz RISC/DSP processorfor local management; serial port for consoleCL1 and debug▲TeraChipWebsite: www.tera-chip.comModel: Switch Fabric Solution RSC No: 20401AdvancedTCA compliant 160 Gbps solution •Single-chip based solution with low power consumptionof only 15 W • Scalable up to 320 Gbpsin an AdvancedTCA chassis • Switch card redundancyof 1:1 and 1+1 • Line card protection •Directed end-to-end Flow Control (FC) by slot andCoS • Dynamic load balancing • Dynamic cell size• 8 CoS queuing on ingress and egress with WRR& Strict priorityFRONT-PANEL HARDWAREPhillips ComponentsWebsite: www.phillipscomponents.netModel: VME, CPCI Panels, PMC RSC No: 20460VME panels and hardware • CompactPCI panelsand hardware • PMC bezels • PCI brackets •Ejectors • Card guides • Custom moldingTyco ElectronicsWebsite: www.tycoelectronics.comModel: Guide Hardware RSC No: 20384Configurations for front board and backplane as wellas mid-plane and coplanar applications in the RTM• Vertical and right-angle pins to support right-angleand coplanar board configurations • Guide pins areavailable in short or long lengths to accommodatevarious Tyco Electronics connectorsRSC 20384▲RSC 20401RSC#CompactPCI61 @www.compactpci-systems.com/rscand AdvancedTCA Systems / June 2005 / 61

NEW PRODUCTSNEW PRODUCTSI/O: ANALOGParsecWebsite: www.parsec.co.zaModel: PM488: Dual DAC PMC RSC No: 20533Two Analog Devices AD9772A 150 MSPS 14-bitDAC converters • SNR of 70 dB @ 25 MHz andSFDR of 80 dB @ 5 MHz • 50 Ohm AC coupledanalog outputs with 67.5 MHz reconstructionpass band • 32/64-bit 33/66 MHz 3.3 V PCI interfaceimplemented in Altera Stratix FPGA • Internaldata buffers of 16K samples per channel • Idealfor baseband or IF waveform reconstruction,W-CDMA, radar, and Software Defined RadioMASS STORAGE: IDETechnoboxWebsite: www.technobox.comModel: 4170 RSC No: 20541Provides a single Ultra160 SE/LVD interface • LSI53C1000R controller • Front panel connectivityvia 68-pin VHDCI connector with user-selectableactive termination • Automatic setting of signalingmode, bus width, and clock • Front-panel statusLEDs show bus activity, modes, etc. • Flash-residentboot codeRSC 20541MASS STORAGE: PLUG-IN UNITSMAWebsite: www.SMAcomputers.comModel: CMASS7 RSC No: 20845Mass storage module for the SMA Computers3U CompactPCI CompactMAX CPU7.2 with IntelPentium M processorMASS STORAGE: RAIDSANBlazeTechnologyWebsite: www.sanblaze.comModel: SB-SCSI Raid Blade RSC No: 20472Single or dual SCSI drive options with SCSIUltra320 support • In/out high-density SCSI connectorssupport daisy chaining with auto-termination• 36 GB to 146 GB of storage in a 6U, single-,or dual-slot CompactPCI form factor • Can provideRSC 20472▲▲RAID 0 (striping) and RAID 1 (mirroring) functionality• Hot swappable, IPMI support • Removablehot-swap drive version availableMEMORY: FLASHAitechWebsite: www.rugged.comModel: S990 Memory Module RSC No: 205081 GB NAND Flash memory in four memory bankswith 100,000 write/erase cycles • Hardware EDACcapable of single-bit error correction and multiple-biterror detection • Hardware has automaticpower-off switch to latched-up memory to preventdamages from SEL events • Error detection, correction,and switch-off events are communicatedvia programmable interrupts to the CompactPCIbus • Flash File Driver (FFD) VxWorks softwarepackage to provide a file system with level-wearingfeatures • Low power consumption of lessthan 3 WRSC 20508▲MEMORY: GENERAL PURPOSEVirtium TechnologyWebsite: www.virtium.comModel: 2 GB DDR2 ECC RSC No: 20394VM493T5653-CC/D5/E6 – DDR2 Reg. ECC 1.450"height, 0.150" thickness • VM491T5653-CC/D5/E6– DDR2 Unb. ECC 1.450" height, 0.150" thickness• VM483L5625-B0/B3/CC – DDR Reg. ECC 1.400"height, 0.280" thickness, four-bank module withlow-cost 512 Mbit ICs • VM485L5625-B0/B3/CC– DDR Unb. ECC 1.400" 0.280" thickness, fourbankmodule with low-cost 512 Mbit ICs • Ruggeddesigns and BOM control • ECC/Non-ECC optionsMOTION CONTROLPro-Dex/Oregon Micro SystemsWebsite: www.pro-dex.comModel: CIX RSC No: 20476One to four axis of Servo, Open Loop Stepper,or Closed Loop Stepper axis control options •Standalone with high-speed RS-232 port • 16 bitDAC analog resolution • Configurable PID filterwith feed forward coefficients • Encoder feedbackavailable for stepper axes • Two limits, onehome, and one auxiliary output are standard peraxis • Up to eight user definable I/O, expandableto 144 opto-isolated I/O • Constant velocity linearinterpolation (all axes) • Software for Windows98/NT/2000/XP • Electronic gearing • Circularinterpolation • Linear, Parabolic, Cosine, and customprofilesPOWER SUPPLYC&D TechnologiesWebsite: www.cdpoweronline.comModel: CPCI200A-1 RSC No: 20488Active power correction • Complies with EN61000-3-2 • 90-264VAC input range • 3U x 4HP package• PICMG 2.11 compliant • Low airflow – requiresas little as 200 lfm of airflow • Fault tolerant N+1configuration • Output fault isolationPicorWebsite: www.picorpower.comModel: QPI-6 Active EMI Filter RSC No: 2021214 A rating • 80 VDC (maximum input) • 100 VDCsurge 100 ms • >40 dB CM attenuation at 250 kHz• >80 dB DM attenuation at 250 kHz • -40 °C to+100 °C PCB temperature • Efficiency >99 percentat full load • 1,500 VDC hipot hold off to shieldplane • 1.0" x 1.0" x 0.2" System-in-Package (SiP)• SMT Land Grid Array (LGA)Wolf Industrial SystemsWebsite: www.wolf.caModel: SCAMP Power Panel RSC No: 20492Eight individually protected and monitored 15-amp125 VAC outlets; external alarm port; total currentat 30 amps • Eight individual illuminated circuitbreakers • Blue backlit LCD display of voltage,current, and power levels for total and individualcircuits • Ethernet port for Internet or remotemonitoring and control; serial diagnostic and configurationport • Time-stamped log of AC powerquality • 15 ft. 30-amp FT4 rated cable with twistand lock plug • UL and CSA approvedRSC 20492▲PROCESSOR BLADESDiversified TechnologyWebsite: www.dtims.comModel: ATC5231 RSC No: 20125Dual LV Intel Xeon with speeds up to 2.8 GHz and1 MB L2 Cache • Intel E7520 Chipset • 800 MHzFront Side Bus • Supports one on-board 64-bit/66 MHz 3.3 V PMC cards • Two 10/100/1000 Mbpsauto-negotiating Ethernet controllers for the baseinterface • Two 1000 Mbps Ethernet ports for thefabric interface • One 64-bit/66 MHz PMC sitePROCESSOR: PENTIUM 4ADLINK TechnologyWebsite: www.adlinktech.comModel: NuPRO-850 RSC No: 20557Socket 478 Pentium 4 processor, up to 3.4 GHz •Longevity Intel 875P chipset, 800/533 MHz FSB• Four dual/single-channel DIMM, maximum 4GB DDR RAM, ECC or non-ECC support • AGP 8x62 / CompactPCI and AdvancedTCA Systems / June 2005

RSC# 63 @www.compactpci-systems.com/rscCompactPCI and AdvancedTCA Systems / June 2005 / 63

NEW PRODUCTSNEW PRODUCTShigh-performance graphics • VGA, GbE, USB 2.0,IDE, S-ATA, COM, keyboard, and mouse • ePCI-Xbus, miniPCI expansion slotPROCESSOR: PENTIUM IIICarlo Gavazzi CSWebsite: www.gavazzi-computing.comModel: FabricPac Platform RSC No: 204562.16 Packet Switch-compliant, eight-slot backplanewith CompactPCI, H.110, and IPMB buses •8HP Intel Pentium III Processor SBC • Hard drive,CD, and floppy • Your choice of OS (Windows2000, Windows NT, Linux, or Solaris) • Yourchoice of memory • Layer 2 or 3 switch cardPROCESSOR: PENTIUM MGE Fanuc AutomationWebsite: www.gefanuc.com/embeddedModel: CPCI-7506 RSC No: 203013U CompactPCI Single Board Computer for multipleindustrial automation and commercial applications• Intel Pentium M operating at speeds of1.1 GHz, 1.6 GHz, and 1.8 GHz or Celeron M processorat 1.3 GHz • 400 MHz front-side bus • InternalSVGA controller via Intel’s 855GME chipset • Upto 1 Gbyte DDR SDRAM • Dual Gigabit Ethernetinterfaces • Two USB 2.0 ports • Two optically isolatedserial ports supporting RS-232/422/485, oneSVGA port, and a PS/2 connection • Rear I/O supportincludes one Gigabit Ethernet port, two additionalUSB 2.0 ports, two 16550-compatible TTLlevel serial ports, and a parallel port • Up to 1 GBof onboard CompactFlash or an onboard IDE harddisk drive • Support for one watchdog, two 16-bit,and two 32-bit software programmable timers• Operating system support includes Windows2000, Windows XP, Linux, and VxWorksRSC 20301PROCESSOR: POWERPCArtesyn CommunicationWebsite: www.artesyncp.comModel: KatanaQp RSC No: 20389Single or Dual PowerPC MPC7447A processorsrunning at up to 1.3GHz 2-Way SMP Architecture• AdvancedTCA PICMG 3.1 Node (1000Base-TRSC 20389▲▲Base Fabric + Octal High Speed GbE Fabric) Layer2 or 3 • Ethernet switch option Quad PTMC expansionsites • Redundant System Management Buswith IPM • Controller Up to 2 Gbyte DDR SDRAMw/ ECC in SODIMM package Up to 128 MB LinearFlash • Real-Time Clock with supercap backup •VxWorks and CG Linux support • Quality assuredby over 30 years of experience as well as ISO-9001 and TL-9000 certificationPROCESSOR: XEONADLINK TechnologyWebsite: www.adlinktech.comModel: ATCA-6890 RSC No: 20390One or two Xeon and next-generation Xeon processors,up to 3.67 GHz • Intel E7520 chipset,800 MHz FSB • Four DDR II-400 240-pin DIMMs,16 GB memory maximum • Two PMC slots withPCI-X bus with one supporting Jn4/Pn4 to RTM• Seven GbE data ports: Four fabric Interface andtwo base interface • One 10/100/1000Base-T managementport (front panel) • ATI RageXL video •Two Serial ATA, two Parallel ATA, two Serial, andfour USB 2.0 portsKontronWebsite: www.kontron.comModel: AT8001ATCA processor RSC No: 20392Single slot AdvancedTCA PICMG 3.0/3.1 processorboard • Intel Xeon processor, scalable up to2.8 GHz • Dual AMC, one module support • DualDDR-II DIMM for 8 GB of PC2-3200 registered400 SDRAM • Dual Gigabit Ethernet base interface• Dual Gigabit Ethernet plus Dual Fibre Channel onfabric interface • IPMI v1.5 supportPROTOTYPING AND DEBUGGING:BOUNDARY SCANGÖPELWebsite: www.goepel.comModel: SCANFLEX RSC No: 20808JTAG boundary scan hardware • A complete modularsystem consisting of SCANFLEX BoundaryScan controllers (SFX-Controller) with externalSCANFLEX TAP transceivers (SFX-Transceiver)and parallel controlled SCANFLEX I/O modules(SFX-Module) • Optional analog, digital, andmixed-signal channels can be directly added tothe UUT interface • Support of up to eight parallelindependent TAPs whereby each TAP is individuallyprogrammable in many parameters • 32 I/Olines for event control and 8 auxiliary I/O lines foradditional analog and digital functions • There maybe a 5 m distance between the SFX-Transceiver,typically without the need for a separate powersupply or hardware controllerPROTOTYPING AND DEBUGGING:BUS ANALYZERFulcrum9Website: www.fulcrum9.comModel: Tx/Rx BenchBlade RSC No: 20405AdvancedTCA compatible design using the rightangle male HM-ZD connector for bench testingAdvancedTCA hub and node card • Provides four(4) Transmit and four (4) Receive pairs to test acomplete AdvancedTCA channel • Edge-launchSMAs for superior bandwidth and ease of testcable attachment • Differential impedance of 100 W±5% • SMT pads provided on Receive channels •Cut-outs for cable access to receive pairs and easeof card insertion/removal • Rx channels capable ofutilizing high bandwidth coaxial blocking capacitors• Eliminate dependency on logic card availabilityfor system evaluation • Verify and evaluate designcompliance with AdvancedTCA guidelinesROUTERS/SWITCHESDSS NetworksWebsite: www.dssnetworks.comModel: Model 5468 Switch RSC No: 20480Fourth-generation BCM5388 Layer-2 switch;Intel 82546 dual-port PCI-X MAC host interface• 133/100/66-MHz, 32/64-bit PCI-X bus interface;PMC-Sierra PM8363 quad gigabit SERDEStransceiver • Onboard FPGA for management,control, and routing functions; high-performancewire speed on all ports, 16 Gb total • Up to 16 Mframes per second maximum switching rate;onboard firmware for configuration, management,and monitoring • 1.5 Mb of onboard memory forpacket buffering; Extended Ethernet frame sizesto 9 KB; fully IEEE 802.3-compliant • PCI Rev. 2.2and PCI-X 1.0-compliant; VxWorks 5.5 and Linux2.4.xx driver support; FCC certified (pending)SCSI PERIPHERALRed Rock TechnologiesWebsite: www.RedRockTech.comModel: RRTC-1SFA-LW RSC No: 20468Capacity of up to 96 GB; no additional softwareis required for operation as a SCSI bootable drive• CompactPCI form factor occupying one 6U slot• Ultra Wide SCSI LVD interface available at frontpanel and J5 connectors • Can be configured for8-bit, single-ended, and/or SCSI-2 operation, thussupporting legacy systems • Front panel status andactivity LEDs • Rear Transition Module availableSHELF & MECHANICAL COMPONENTSSchroffWebsite: www.schroff.usModel: AdvancedTCA Systems RSC No: 204172 to 16-slot AdvancedTCA systems ideal for telecomand networking applications • Broad range ofconfiguration options available • Backplanes availablein a variety of topologies including full mesh,dual star, and dual-dual star • Backplanes meethigh-speed requirements of next-generation boards• Low cost, field replaceable fan trays reduce laborand maintenance costs • Removable fan trays provideexceptional cooling up to 200 wattsSOFTWARE: DEVELOPMENT TOOLPerformance TechnologiesWebsite: www.pt.comModel: NexusWare Core RSC No: 20813A full Linux-based operating system • Linux ker-64 / CompactPCI and AdvancedTCA Systems / June 2005

nel is specifically tailored toward embedded applications• Complete suite of development toolsand compilers, including the Eclipse platform •Powerful APIs for all onboard hardware resources• Integrated drivers: no need for external busdrivers • Application development with WindowsXP/2000, Linux, or Solaris OS • Pre-packagedapplications and protocolsSYSTEM MANAGEMENTAdaxWebsite: www.adax.comModel: Adax Signaling Products RSC No: 20553Building blocks for system developers; signalingcommunications controllers and lower layer protocolsoftware • Integrated software and bladesfor application developers; integrated signalingprotocol stacks • Complete signaling gatewaysfor everyone; multipurpose signaling gateways •Signaling nodes – SS7 or IP based STPs, HLR,VLR, SMSCs, databases, and more • Signalinggateways and IP signaling points for SS7 andIP switching, routing, back-haul, and tunneling• Media gateways, media gateway controllers,and softswitches • GPRS and 3G nodes includingSGSNs, GGSNs, MSCs, RNCs, and Node Bs• Simulation, monitoring, and billing systems fortest and measurement applications • Narrowbandsignaling for PSTN, GSM, and GPRS networks– SS7 (64k and 2 Mbs HSL), Frame Relay, LAPB/D/V5, and X.25 • Broadband signaling for 3Gnetworks – ATM AAL2 and AAL5, SSCOP/SSCF,SSSAR/SSTED, IP over ATM, and Frame Relay •SIGTRAN signaling for fixed and mobile networks– SCTP, M2PA, M2UA, M3UA, and SUA with SIPinterworking • SS7/IP interworking providing theability to interconnect all three via T1/E1, OC3/STM-1, and Gigabit EthernetTELECOMSBS TechnologiesWebsite: www.sbs.comModel: TELUM 1000 RSC No: 20429155 Mbps full duplex line speed • PCI Expressinterface; PCIe Rev 1.0 compliant; support for afull-duplex OC-3 Interface • Support for 16,000VCCs; 4 MB local memory • Optional AutomaticProtection Switching (APS) • Passes and managesAAL1, AAL2, and raw cells • Segmentationand reassembly of AAL0, AAL3/4, and AAL5 cells• Traffic management supported: ABR, CBR, UBR,and VBR • Single or APS port versions • SupportsATM Forum UNI 3.1 and TM 4.0 • IntelligentPlatform Management interface; onboard microcontroller-basedsubsystem • Hot-swap compliant• Support available for Linux, VxWorks, andWindows 2000/XPTranstector SystemsWebsite: www.transtector.comModel: ALPU Series RSC No: 21158Auxiliary Lightning Protection Unit • Suitable forprotecting Gigabit Ethernet systems using fastactingsilicon avalanche suppression diodes andreduced capacitance protection circuitry thatpermits high signal bandwidth • More than 12different configurations are available on 2 weeklead times • Clamping performance of SASDswhile maintaining a TIA Cat-5e compliant networkconnection • Deliver protection to IEC 61000-4-5standards • Can protect up to eight Cat-5e pairsas well as several configurations of DC power forPOE applications • Designed to self-sacrifice inthe event of a catastrophic event, taking protectedequipment offline • 5.5 inches high, by 4 incheswide and 3 inches deep • Metal or plastic enclosuresare rated NEMA 3R rainproofTELEPHONY: VOIPNMS CommunicationsWebsite: www.nmscommunications.comModel: MG 7000A RSC No: 20393480 IVR, fax, conferencing, VoIP, 3G 324M videosessions • 16 T1/E1 ports • Call control for CAS,ISDN, and SIPTHERMAL MANAGEMENTRadian HeatsinkWebsite: www.radianheatsinks.comModel: ATCA BGA Heatsinks RSC No: 20383Removable ATCA BGA heatsinks can be installedwith no special board modifications needed •Standard BGA heatsink sizes range from 21 mmto 45 mm footprints • Heatsink heights availablefrom 7.11 mm to 9.8 mm for low-profileCompactPCI, AdvancedTCA, and PC/104 applications• Attachment options compatible withvarious chip heights and package types, includingplastic, ceramic, and metal • Black anodizedplating delivers enhanced performance in harshenvironments and natural convection • All productsprovided pre-assembled, with lightweightaluminum heatsink, selected clip size, and thermalpad optionTURNKEY SYSTEMMotorolaWebsite: www.motorola.com/computersModel: Centellis CO RSC No: 20521Centellis CO 21KX features: 12U/19" CompactPCIframework to deliver 5-nines availability • Faultresilientdesign minimizes hardware induced failures• CompactPCI hot swap capability minimizesmean-time-to-repair • PICMG 2.16-compliantpacket switching backplane • Ethernet switchesand shelf controllers on same board, redundantand hot-swappable • Designed for NEBS Level 3for telecom Central Office (CO) applications •EndurX C0 21KX features: PICMG 2.16-compliantCompactPCI packet switching backplane with19 6U node slots • Dual redundant Carrier GradeLinux/Intel architecture processor-based nodes •Redundant Layer 2 Gigabit Ethernet switches andshelf management controllers • High availabilityframework API for HA-aware applications • Policydrivenevent handling/propagation • Flexible softwareupgradabilityPinnacle Data SystemsWebsite: www.pinnacle.comModel: TS2100 Telco System RSC No: 204132.0 GHz, low voltage, Xeon processor • One- ortwo-way configuration • L2 cache with integrated512 KB • Memory: four DIMM sockets for up to4 GB DDR266, registered • SDRAM, 72-bit, ECC,184-pin 256 MB, 512 MB, 1 GB, 2 GB • Front PanelI/O: One USB 2.0 port • One serial RJ-45 port •Fibre: Two small form-factor pluggable connections• LEDs for status, health, hard drive activity,and Ethernet/FibreChannel connections • SwitchConnections: Backplane 12 x 10/100/1000 MbpsEthernet, Front Egress: 3 x 10/100/1000 MbpsEthernet • InterSwitch Link: 1 x 10/100/1000Mbps Ethernet • Chassis: 13U – 22.75" (577.85mm) by 18.00" (508 mm) (D) by 19.00" (482.60mm) (W) • Four fans cable mgmt filter • Systempower: One power distribution board • Support forup to four –48V DC Power Entry Modules (PEMs)• Backplane: 14-slot full mesh,12 system slots,two switch slotsSBS TechnologiesWebsite: www.sbs.comModel: AVC-cPCI-3003-3U RSC No: 20512Lightweight – less than 11 pounds (4.9 kg) includingmodules • Compact for use in small spaces• Six, 3U CompactPCI slots • CM4 single boardcomputer with 750/755 400-500 MHz processor,two 1 MB L2 cache, 1.6 GB/s • MIL-C-38999Series III connectors • COTS AVC rugged conductioncooled chassisSun MicrosystemsWebsite: www.sun.comModel: Netra CP 2300 RSC No: 20504Part of a comprehensive line of NEBS certifiedsystems, storage, and management/availabilitysoftware from Sun • Solaris OS operating environment• 650 MHz UltraSPARC IIi processor • Upto 2.5 GB memory (512 MB minimum) • A rackmountarchitecture, CompactPCI compliant, andadherence to PICMG standardsRSC 20504▲VIDEO: FRAME GRABBERSBS TechnologiesWebsite: www.sbs.comModel: AVC-cPCI-3009 RSC No: 20562COTS rugged conduction cooled chassis • 6 3UCompactPCI slots • MIL-C-38999 connectors •9.5 pounds with modules • 5.0" (H) x 8.73" (W)x 8.75" (D) • Power supply: 100 watts, single ordual • CR3 Single Board Computers: Intel Celeronprocessor, system controller, or peripheral cardoperation • MIL-STD-1553 interfaces: Two dualredundant channels, independent operation as abus controller, remote terminal, and dual functionbus monitorWIRELESS: SDRInnovative IntegrationWebsite: www.innovative-dsp.comModel: Quixote RSC No: 20448600-MHz TMS320C6416 DSP; 2-6 MGate Virtex-II FPGA • 32 Mbytes SDRAM, 8 Mbytes ZBTSBSRAM; 64/32-bit CompactPCI, 66 MHz, 5 V/3.3 V • AD6645 and AD9764 converters • Complextriggering modes with HW event logging • PMCsite w/Jn4 to FPGA DIO • PICMG 2.17 StarFabriccompliantCompactPCI and AdvancedTCA Systems / June 2005 / 65

CompactPCI®ADVERTISER INFORMATIONPage/RSC#andA dvanced TCAAdvertiser/Product description®Systems33 Adax – Signaling Gateway9 ADLINK – CompactPCI Boards45 AEI – Fast Ethernet & Gigabit Ethernet Cards11 Aitech – COTS50 Alphi – PowerQUICC II Single Board Computer47 Artesyn – Communication Products38 Bustronic – AdvancedTCA Backplanes29 Condor – Interface Solutions5 Conec – AdvancedTCA Connectors41 Curtiss-Wright – Embedded Components and Systems57 Diversified Technology – Single Board Computers16 Elma – Enclosures, Panels, Handles63 Elma – Embedded Packaging Solutions3701 Embedded Connect – Seminar Series27 Embedded Planet – Embedded PowerPC/XScale13 Excalibur – Rugged Systems7 GE Fanuc – Single Board Computers2 General Standards – Data Acquisition Boards34 ICS – Data Acquisition Modules15 Kontron – ATCA/AMC Modular Solutions31 MEN Micro – VMEbus and CompactPCI58 Micrel – Dual-Slot Power Controllers3501 NAT – CompactPCI Carriersand Infrastructure Boards26 One Stop Systems – CompactPCI, PCI, PCI-X, PCI Express,MAX Express2102 Pinnacle – AdvancedTCA Solutions43 Positronic – Zone 1 Power Connectors2101 Positronic – Power Connectors49 Radian – BGA Fansinks3 RadiSys – Com Express20 Red Rock – Mass Storage Modules54 SBE – Networking and Communications I/O Solutions67 SBS – PMC/AdvancedMC12 Schroff – AdvancedMC Mechanical Hardware Kits17 Simon Industries – Conduction Cooled Heat Frames44 SMA – 3U CompactPCI2103 Sundance – SMT3000, SMT300, SMT700823 Sundance – SMT791, SMT787, SMT79525 Sundance – SMT6050, Diamond RTOS, GDD600 & GDD800039 Technobox – Adapters and Tools for PMCs3702 Technobox – PMCs and PIMs3502 Technobox – PMCs19 Vadatech – Board Level Solutions68 VMETRO – Vanguard Bus Analyzers61 Voiceboard – VolP on cPCI6 Winchester – ATCA Power ConnectorOpenSystems PublishingAdvertising/Business office:30233 Jefferson AvenueSt. Clair Shores, MI 48082Tel: 586-415-6500 ■ Fax: 586-415-4882Vice President Marketing & SalesPatrick Hopperphopper@opensystems-publishing.comSenior Account ManagerDennis Doyleddoyle@opensystems-publishing.comAccount ManagerTom Varcietvarcie@opensystems-publishing.comPrint and Online Marketing SpecialistChristine Longclong@opensystems-publishing.comAdvertising/Marketing CoordinatorAndrea Stabileastabile@opensystems-publishing.comEuropean RepresentativeStefan Baginskisbaginski@opensystems-publishing.comAccount ManagerDoug Cordierdcordier@opensystems-publishing.comBusiness ManagerKaren LaymanFor reprints call the sales office: 586-415-650066 / CompactPCI and AdvancedTCA Systems / June 2005

RSC# 67 @www.compactpci-systems.com/rsc

RSC# 68 @www.compactpci-systems.com/rsc