tough hardware easy software.€¦ · tough hardware empowered software... easy software. software...
TRANSCRIPT
T O U G H H A R D WA R E
w w w . c w c e m b e d d e d . c o m EMPOWERED SOFTWARE...
E A S Y S O F T WA R E .
Software shouldn’t be hard. That’s why we back our rugged boards and
subsystems with all the tools you need to easily and rapidly design, integrate
and support your deployed electronics systems. You can depend on our
board support packages, built-in test routines, diagnostic utilities, extensive
software libraries and broad operating system support to reduce design
costs and get your product to market faster. That’s hard evidence that
makes Curtiss-Wright the easy choice.
CHAMP-AV6 CHAMP-FX2
VPX6-185
The CHAMP-AV6, CHAMP-FX2 and VPX6-185 are three of our latest generation of VPX/VPX-REDI boards and include a rich set of COTS Continuum system and support software designed to ease integration and reduce system development time.
Dear Readers,
Major topics in this issue of Boards & Solutionsmagazine are Telecom & Networking, SmallForm Factor Boards, MicroTCA, IndustrialComputing and Embedded Linux.
When Linux was born on August 1, 1991, nobodywas aware that 15 years later the scale of supportedhardware would range from tiny microcontrollersup to large multiprocessor mainframes. Today,even in the automation and machine industry, theuse of Linux has become more widely accepted.Without constraining the freedom of open sourcesoftware, it is still necessary that the user- specificintellectual property be protected against externalaccess and illegitimate manipulation. Read more inthe special feature starting on page 5.
According to the Electronics Trend Publications data, the global PC/104 boards marketamounts to around 120 Million Euros. The largest volumes of PC/104 boards are usedin industrial and in defense systems, 50% and 20% respectively. Customer demand forfunctionality and reliability of PC/104 systems very often require additional efforts fromboards developers and specific engineering solutions for their realization. An article start-ing on page 16 will discuss main solutions that bring more processing power, improve reliability, widen operational temperature range and, therefore, expand the PC/104 productsapplication area.
The main objective of the SCOPE Alliance is to help to develop a powerful ecosystem thatenables the Network Equipment Providers to build Carrier Grade telecommunication systems from components supplied by different vendors that fulfill the needs of their customers. The SCOPE Alliance has already developed profiles for the PICMG 3.0 Advanced Telecommunication Computing Architecture 3.0 specification and for theOSDL Carrier Grade Linux 3.2 specification. Read more in the article starting on page 28.
IP-based Machine-to-Machine (M2M) communication is taking off quickly. The GartnerGroup predicts that by 2007, there will be between 100-160 million machine-to-machine connections worldwide that use wireless mobile phone networks. 880 millionnew M2M-enabled devices will be produced annually by 2010, according to the FocalPoint Group. Given these predictions, embedded designers need to consider all optionsand issues involved in the selection of an IP-enabling M2M solution. Read more in thearticle starting on page 36.
Yours sincerelyJürgen Hübner(Editor)
VIEWPOINT
3 June 2007
For more info visit our website: www.aaeon.com
AEC-6710 Worldwide no.1 Water
Proof Controller!
VIA Mark 533 MHz DC 12-48V input 4 COM / 1 USB / Ethernet
AEC-6860 Entertainment Multimedia Engine Intel Core 2 Duo up to 2.0 GHz Rich Video: VGA /TV-out /DVI / LVDS 4 USB / 4 COM / LAN Supports WLAN / LCD TV / HDTV resolution
AEC-6800 Entry Level,
Compact Embedded Controller AMD GX466 333 MHz Optional CAN bus version
AEC-6920 Rugged Fanless
Embedded Controller Intel Core 2 Duo / Core Duo Processor 2 PCI slots / 1 PCI Express expansion GbE Ethernet / 4 COM / 4 USB Anti-Vibration / Anti-Shock
AAEON Technology Europe B.V. - Science Park 5080 - 5692 EA SON - Netherlands
Tel: + 31 (0)40 2678450 - Fax: +31 (0)40 2678451 [email protected]
... Sometimes a breeze can be very important…
... Sometimes a breeze can be very important…
From our BOXER series:
...but not for us!Fanless Embeded Controller Solutions
IP67
June 2007 4
CONTENTS
Cover PhotoMSC
Native mainline Linux: fit for PAGE 12embedded and real-time system
When Linux was born on August 1,1991, nobody expected that 15 yearslater the scale of supported hardwarewould range from tiny microcon-trollers up to large multiprocessormainframes, or that Linux would be-come a real-time operating system.
PC/104 CPU boards: new challenges, PAGE 16opportunities, and frontiersThis article gives an overview of PC/104 and presents boards that ex-pand the product application area of this long-established standard.
Use of MicroTCA technology PAGE 22in carrier-grade applicationsIn this article the applicability of MicroTCA technology in carrier-grade telecom applications is examined with a signalling gatewayuse.
High signal integrity speeds up PAGE 26MicroTCA systems
Based on the results of detailedexperiments with signal behav-iour on the transmission path,ELMA has optimised its layoutstrategy for MicroTCA back-planes.
The SCOPE Alliance imposes PAGE 28profiles on standards
The main objective of the SCOPEAlliance is to help to develop apowerful ecosystem that enablesthe network equipment providersto build carrier grade telecom mu-nication systems from componentssupplied by different vendors thatfulfill the needs of their customers.
Machine-to-machine communication PAGE 36over IP Networks
Using the internet for M2Mapplications takes much of thepain out of the process of col-lecting and communicatingdata. To succeed, M2M devicesdepend on minimizing the timeand level of engineering re-sources required.
Viewpoint 3
Embedded Linux
Security functions on COM Expressplatforms running under Linux 7
Pre-integrated ATCA – an examplefrom practical experience 10
Native mainline Linux: fit forembedded and real-time systems 12
Small Form Factor Boards
PC/104 CPU boards: new challenges, opportunities, and frontiers 16
Earthquake research satellitedemonstrates PC/104 ruggedness 20
MicroTCA
Use of MicroTCA technology in carrier-grade applications 22
High signal integrity speeds upMicroTCA systems 26
Telecom & Networking
The SCOPE Alliance imposesprofiles on standards 28
IP network servers and carrier graderack-mount servers 32
Updating legacy interfaces – Serial RapidIO compared to EMIF64 switches 34
Industrial Computing
Machine-to-machine communicationover IP Networks 36
Windows Vista ushers in a newera in data acquisition 40
Product News 41
embedded-linux-konferenzembedded-linux-konferenz
Zur Anmeldung nutzen Sie bitte das Online-Formular auf
www.embedded-linux-konferenz.deoder senden Sie eine formlose E-Mail an: [email protected]
An der Table-Top Ausstellung nehmen u.a. folgenden Firmen teil:
>> Acal>> Aonix>> Applied Informatics>> Atlantik Elektronik>> Contec>> Digi International>> emlix
>> emtrion>> Freescale Semiconductor>> Kontron>> MSC>> OSADL>> Pengutronix>> Philosys Software
>> PHYTEC>> SIGMA>> Solid>> SYSGO>> systerra computer>> TQ-Components>> Wind River
Aufgrund der großen Nachfrage findet die embedded-linux-konferenz dieses Jahrin Form einer „Mini-Roadshow“ sowohl in Pforzheim (am 26. Juni) als auch inMünchen (am 28. Juni) statt. Im Fokus stehen „Embedded Linux für dieAutomatisierungstechnik“ und „Embedded Linux für die Kommunikationstechnik“.
Hauptsponsoren Die Veranstaltung richtet sich an
>> Einsteiger und Fortgeschritte>> Entwicklungsingenieure und an das technische Management
Das Programm umfasst:
>> Technische Vorträge>> Hands-on/Demo-Sessions>> Table-Top Ausstellung
Teilnahmegebühr(inklusive Mittagessen und Getränke): EUR 119.-* (*zzgl. MWSt)
!!WICHTIG!!Bitte beachten Sie, dass die Teilnahmerzahl der Hands-on/Demo Sessions auf maximal 20 Personenbeschränkt ist. Die Plätze werden in der Reihenfolge der Anmeldungen vergeben.Wir empfehlen daher eine frühzeitige Anmeldung.
26. Juni, Pforzheim Congress Zentrum28. Juni, München NH-Hotel Dornach
Anmeldung und Begrüßung09:30 - 09:50 Registrierung der Konferenzteilnehmer
KEYNOTE10:00 - 10:30 Open-Source-Software in der Telekommunikations- und Automatisierungsindustrie – Wie passt das
zusammen? Sprecher: Ulrich Doll, Homag Holzbearbeitungssysteme
(Vorstandsvorsitzender des Open Source Automation Development Lab)
Session A - Embedded Linux Trendsin der Kommunikationstechnik und Automatisierungstechnik
10:35 - 11:05 Vortrag A1 - Multiprozessor, Multicore, Virtualisierung Sprecher: Irene Hahner (Kontron)11:35 - 12:05 Vortrag A2 - Linux im Telecom-Markt - Standards, Gremien und Technologien Sprecher: Hans Jürgen Rauscher (Wind River)12:10 - 12:40 Vortrag A3 - Power Architecture and embedded Linux: status-quo and future outlook Sprecher: Freescale13:30 - 14:00 Vortrag A4 - Embedded Security Sprecher: Wolfgang Eisenbarth (MSC)14:05 - 14:35 Vortrag A5 - Embedded-Linux in Liebherr-Mobilkränen Sprecher: Erwin Morath, Liebherr15:05 - 16:05 Vortrag A6/A7 - Embedded Linux in der Automatisierung - riskantes Neuland oder etablierte Technik? Sprecher: Carsten Emde (OSADL) und Hans Rauscher (Wind River)16:25 - 16:55 Vortrag A8 - Projektberichte über Embedded-Linux Implementierungen in unterschiedlichen Marktsegmenten Sprecher: Eckhard Voigt (Philosys Software)17:00 - 17:30 Vortrag A9 - Linux-Implementierungen auf Power Architecture-Basis Sprecher: Freescale
Session B - Hands-on/Demos
10:35 - 12:05 Demo-Session - Multiprocessing/Embedded Linux Durchführung: Kontron10:35 - 12:05 Demo-Session über Computer-On-Module Lösungen in der Linux-Welt Durchführung: MSC10:35 - 12:05 Hands-on: Linux-Software komfortabel und schnell entwickeln Durchführung: Wind River13:30 - 15:00 Demo-Session (1. Wiederholung) - Multiprocessing/Embedded Linux Durchführung: Kontron13:30 - 15:00 Demo-Session (1. Wiederholung) über Computer-On-Module Lösungen in der Linux-Welt Durchführung: MSC13:30 - 15:00 Hands-on: Linux Debugging per Software-Agent und Hardware/JTAG Durchführung: Wind River15:20 - 16:50 Demo-Session (2. Wiederholung) - Multiprocessing/Embedded Linux Durchführung: Kontron15:20 - 16:50 Demo-Session (2. Wiederholung) über Computer-On-Module Lösungen in der Linux-Welt Durchführung: MSC15:20 - 16:50 Hands-on: Fehler schneller finden und beheben mit SystemViewer, ScopeTools und Lab Diagnostics Durchführung: Wind River
EMBEDDED LINUX
7 June 2007
Security functions on COM Expressplatforms running under Linux
� In our connected world, we must not onlykeep our personal data such as financial status,bank accounts, and health records secret, butalso more than ever our enterprise and productinformation. Only through strict securitymechanisms we can maintain today’s technol-ogy advance and protect user-specific intellec-tual property (IP). Many innovative enterprisesrecognise the great value in the protection oftheir special know-how against unauthorisedcopying, modification, or cloning.
Furthermore, it is necessary to guarantee pro-tection of critical processes, for example in theautomation industry, robotics, and medicaltechnology, against unauthorised external ac-cess or cloning. For instance, extensive damagecan occur through unintended modification ofthe recipe of a medical product during its pro-duction. Even though this may be preventedtoday through simple password protection,the hardware and the software are still vulner-able to modification without this being evident.Until recently security functions could only beachieved through costly proprietary solutions.
MSC Vertriebs GmbH has now reacted to theindustry’s demand to protect investments in IP,and has supplied its COM Express platformsbased on the Intel Core 2 Duo and Intel82945GM with extensive security functions toensure confidence in embedded applications.Complete system protection already begins atboard start-up with the first execution of theinitial CPU instruction, and goes all the way upto execution of the application. Changes to thehardware, the BIOS, or to the software can berecognised, and the process stopped, even be-fore booting an unsafe operating system or ap-plication. The COM Express modules operateonly in a signed environment from the OEM.
The basis for the security functions imple-mented on the COM Express platform fromMSC are the specifications defined by the in-dustry consortium Trusted Computing Group(TCG). Intel, Microsoft, AMD, HP, IBM, Sony,Infineon, and Sun are among companies head-ing the TCG. The goal is to enhance the securityof the system in use thanks to special hardwarebuilding blocks and software interfaces. One of
these hardware modules, the technology ofwhich is already used in Smartcards, is built inthe trusted platform module (TPM). As op-posed to a chip card, the TPM designed in ac-cordance with the TCG specification 1.2 isbased on a standardised architecture with openinterfaces directly built on the COM Expressmodule, and hence becomes an inherent func-tion of the hardware itself.
The TPM 1.2 from Infineon used on the COMExpress modules is based on their 16-bit secu-rity controller family and firmware with astandardised security function range . The de-vice contains an RSA encryption unit with a keylength of up to a maximum 2048 bit. RSA is anasymmetric encryption method that uses a keypair consisting of a public key and a private key.All data is encrypted by the OEM using a pri-vate key and can decrypted by each system viaa public key. In addition, a hardware accelera-tor is integrated into the TPM to execute anSHA-1 hash algorithm. The hash function al-lows a compressed key value to be calculatedfrom code segments as a representative image of
By Wolfgang Eisenbarth, MSC
Even in the automation andmachine industry, Linux has
become more widelyaccepted. Without constrain-
ing the freedom of opensource software, it is
necessary to protect user-specific intellectual propertyagainst external access and
illegitimate manipulation.
FREE Subscription to boards & solutions magazineEnsure getting your personal copy of boards & solutions magazine free of charge by completing the online form at:
www.embedded-control-europe.com/bs_magazine
June 2007 8
EMBEDDED LINUX
the module. The encrypted hash can be de-crypted using the public key, so that the mod-ule can be authenticated. A true randomnumber generator (TRNG) serves to generatequalitatively high-order random numbers. Themodule communicates over the low-pin-countbus (LPC).
The hardware-supported TPM functions are already executed during the boot procedure togain secure boot functionality. If a differencefrom the expected values of the BIOS code orhardware occurs, the boot procedure stops andthe system cannot be started. The BIOS mon-itors the master boot record (MBR) of the harddisc, the boot sector, as well as security relevantparts of the operating system and the user soft-ware. In order to assure tight coupling of thesoftware with a specific hardware platform, anunique and fraud-resistant equipment identi-fication is possible with the device master key(DMK).
The security functions implemented by MSCare not only executable on Microsoft operatingsystems Vista and Windows XP, but also on theopen operating system Linux. In recent monthsLinux has been gaining strong momentum asan operating system in embedded systems, es-pecially in 32-bit computers. Surely scalability,
the disclosure of the program source, and in themeantime, the available real-time capabilityhave contributed to this success. Additionallycomes the advantage that no licence costs arethereby incurred with Linux, an interesting argument for many OEM manufacturers.
Linux supports a multitude of hardware com-ponents and well-established network proto-cols. The Linux kernel is freely licensable asopen-source software, and the Linux driverpackages, for example for the TPM, are avail-able from the internet (figure 1). The advantageis that the developers can concentrate on theirproprietary know-how, meaning on their ap-plications. And this user-owned software thatdifferentiates them from the competition iswhat has to be protected. In addition, the com-panies can sign their product special softwarepackages before delivery, so that the customercannot simply exchange them later, or individ-ual modules, which could affect the correctfunctionality.
Linux has also increasingly positioned itself asan operating system in the automation andmachine industry. Activities in these marketsare being promoted and coordinated by theOpen Source Automation Development Lab(OSADL), which was founded approximately
Figure 1. Open source driver implementation (Source: Infineon)
9 June 2007
EMBEDDED LINUX
a year ago. The OSADL is a cooperatively or-ganised institution with the goal to develop,qualify, and to standardise a Linux-based op-erating system environment. At the same timethe user community wants to save on devel-opment efforts for qualifying new systems, andto merge different operating systems on a standardised Linux core, without followingproprietary interests.
Important requirements of the automation in-dustry for Linux are substantially longer prod-uct lifetimes, a significantly reduced failurerate, as well as independence from individualsoftware suppliers. Furthermore computer ap-
plications need specific extensions of the oper-ating system, as for example real-time behaviouror standardised software interfaces. One of theeleven founding organizations of the OSADL isELTEC Elektronik, with whom MSC has re-cently agreed close collaboration. The goal of thecooperation is the configuration of the COMExpress platforms from MSC with ELTEC’s real-time Linux functionality. ELTEC has particularlycomprehensive expertise in the area of real-timeLinux with drivers and frameworks. A few newfunctions are incorporated in the current ver-sion of the Linux kernel, such as real-time pre-empt patches or virtualisation, which improvethe real-time characteristics of Linux.
Figure 3. COM Express module CXE-CD945 with dual-channel memory
Figure 2. COM Express platform CXB-CD945 with Core 2 Duo processor
www.comtel-online.com
Backplanes● 3U CPCI 32bit● 6U CPCI 32/64bit● CT-Bus (H.110)● PICMG 2.16● PICMG 2.20● CPCI Bridges● PXI & PXI Express Backplanes● Custom CPCI & PXI Backplanes
Systems● 1U-4U Horizontal CPCI● 3U-12U Vertical CPCI● CPCI Power Supplies● PXI systems● Custom CPCI & PXI systems
Horizontal CPCI Systems
Vertical CPCI Systems
Leading EdgeBackplanes & Systems
RoHS
Backplanes
Power Backplanes
Pre-integrated ATCA – an examplefrom practical experience
EMBEDDED LINUX
� When Airvana decided to develop the uni-versal access gateway (UAG), the greatest chal-lenge, apart from the technical requirements,was the short implementation time-frame. Although Airvana has many years of experiencein the development of network infrastructuredevices for mobile phone providers, it was clearthat the extensive specifications could only beimplemented if the Airvana development teamconcentrated on the actual application and usedan existing pre-integrated AdvancedTCA system , such as that which RadiSys, Wind Riverand OpenClovis presented to Airvana.
Usually the development team select the hard-ware first (usually at chip, not even at compo-nent level), followed by the operating system, andthen further system software and middleware.This procedure is possible with devices withstraightforward complexity, although still onlypartly advisable. Nowadays, the requirements onthe final product are too great; of course meet-
ing technical as well as qualitative standards is al-ready expected in the first version. The wholething takes place under increasing pressurefrom competitors, which finally leads to veryshort implementation periods for the project.
What functions comparatively badly with simple devices, becomes simply outdated whenit comes to highly complex products such asrepresented by network infrastructure equip-ment. The end customer will not pay anymorefor the effort in time and money to re-inventthe hardware and system software wheel! Usingcomponents that have already been developedand tested is the only reasonable way out: Theopen source software movement around Linuxhas been doing this successfully for years; AdvancedTCA has developed it for commercialoff-the-shelf hardware, with commercial carrier-grade Linux to complete it.
Several factors make the integrated solutionfrom RadiSys (ATCA-Hardware), Wind River(commercial grade Linux) and OpenClovis(middleware) interesting for Airvana. Since acomplete ATCA system is available, Airvana canconcentrate completely on the added-value ofits own application when developing. This
By Hans Jürgen Rauscher, Wind River
Wind River Platform for Network Equipment- Linux Edition
June 2007 10
COM Express platforms
The COM Express module CXB-CD945 fromMSC Vertriebs is available with the power-sav-ing low-voltage processor Intel Core 2 DuoL7400 (Merom). The two CPU cores use a 4MB level-2 cache and provide data transfers at667 MHz over the front-side bus. The thermaldesign power (TDPmax) generates only 17W at1.5 GHz, which is almost half the power con-sumed by the Merom processor T7400. Thanksto Intel 64-bit technology, this high-end mod-ule offers the flexibility to process both 64-bitand 32-bit commands. For applications inwhich low power consumption is of high im-portance, the COM Express Module CXB-CD945 comes with the Intel® Celeron™ Mprocessor ULV (ultra-low voltage) 423. Thedual processor impresses with its low power dis-sipation of just 5.5W (thermal design power) ata clock frequency of 1.06GHz.
The CXB-CD945 module, based on the MobileIntel Chipset 945GM(E), offers a wide selection
of interfaces and expansion capabilities. Apartfrom a PCI Express x16-interface and four PCI-Express x1 interfaces, eight USB 2.0 ports areavailable for communication with peripheralequipment. Through an optional Intel 82573GBit LAN-Controller, a fast 10/100/1000 baseTX-LAN-network connection can be estab-lished. Large data volumes can be stored usingthe two SATA II-connections transfer rates upto 300MB/s. ACPI battery management and awatchdog timer are also implemented on theCOM Express module.
Taking advantage of the dual-channel memo-ry technology, MSC has increased the graphicperformance of its COM Express platformCXE. The main memory on the COM Expressmodule CXE-CD945 can be expanded up to amaximum capacity of 4GB by inserting 2equivalent sized DDR2 SDRAMs in both 200pin SO-DIMM sockets. This makes the boardespecially suitable for high-end industrial ap-plications, such as for use in image processingand medical systems, as well as for game appli-
cations. With the introduction of the CXE-CD945 module, MSC expands its portfolio ofCOM Express components based on the ex-tended form factor of 155mm x 110mm. De-pending on the requirement, the processor onthe board can be selected to be an Intel Core 2Duo L7400 (1.5 GHz), an Intel Core Duo L2400(1.66 GHz), or even the Intel Celeron M 440(1.86 GHz). Other versions are available uponrequest. Thanks to the graphics media acceler-ator 950, which is already integrated in the mo-bile Intel®-chipset 945GM(E), the moduleoffers excellent 3D graphic capabilities.
Besides a PCI 2.3-bus interface, four PCI-Ex-press x1 interfaces are available on the COMExpress module for high-speed data transfer.Altogether eight USB 2.0 ports allow flexibleconnection of various peripherals. Fast Ether-net network connectivity is provided throughan Intel 82573 GBit LAN controller. Storage oflarge data volumes is possible through twoSATA II connections with up to 300MB/s orthrough an enhanced IDE port. �
saves development time, leads to earlier prod-uct availability and increases the marketchances drastically. For AdvancedTCA as such,the component idea was already developedwhen the standards were adopted and imple-mented. Anybody who has already developedwith ATCA knows that the integration of thesystem software, that is, carrier-grade Linux in-cluding the Linux board support package andsupport for the hardware management bymeans of IPMI and HPI, taken on its own, rep-resents a challenge. Heterogeneous CPU archi-tectures and network processors with thenecessary parallel micro-code do not necessar-ily make the situation any easier. The additionalnecessity of using high-availability middle-ware, which guarantees almost 100% uptime atthe application level, brings further complexi-ty into the overall system. The approach to im-plement, integrate, test, maintain and supportthese building blocks by your own product de-velopment team is very likely to fail, due to thesheer mass of functions and possible errors.
Therefore RadiSys offers the Promentum prod-uct series (SYS-6010, ATCA-7010, ATCA-2100),which is needed for network devices such as theuniversal access gateway: extensive scaling,which in different customer environments ineach case enables the optimum in interfaces,band-width and costs. With 10 Gigabit per second, there is sufficiently high bandwidth,thanks to Intel IXP network processors withhardware-supported package processing forflexibility and high data throughput, and lastbut not least, maintenance and system stabili-ty. With “Platform for Network Equipment -Linux Edition” (PNE LE), Wind River presentsa commercial carrier-grade Linux softwareplatform available particularly for deploymentin ATCA devices like the Universal AccessGateway.
PNE-LE is integrated, tested, validated and reg-istered at The Linux Foundation for the CGLrequirements version 3.2 . The recently signed-
off CGL 4.0 specification will be implementedin the next PNE-LE version. Like all Wind RiverLinux software platforms, PNE-LE is based onthe current Linux kernel 2.6 and layers ontocompletely buildable pristine open source code.Included in all Wind River platforms is theWind River workbench development suitebased on Eclipse, which includes extensiveerror seeking mechanisms such as user and ker-nel mode debuggers, SystemViewer andScopeTools, which help in shortening the de-velopment time through optimal support of thesoftware development process.
This begins with hardware bring-up, continueswith the boot loader and kernel module devel-opment and the application programming,and ends with development and system testfunctions. Wind River Product Support andProfessional Services are at the customer’s dis-posal for advice and actual support worldwide.
At the interface between different networks, thenetwork equipment must fulfil an availability ofat least 99.999 percent. On the hardware side,this demand is met by ATCA, on the operatingsystem side by CGL. What is still missing is ahigh-availability middleware which makes fail-safe applications possible. Here, OpenCloviswith the application service platform and IDEoffer the necessary middleware. This makes itpossible, thanks to different services, to syn-chronise application configurations and databetween the different entities in such a way that,in case of failure of an application entity, an-other can continue the service smoothly. Withthe pre-integrated combination, consisting ofAdvancedTCA hardware from RadiSys, carrier-grade Linux from Wind River, and high-avail-ability middleware from OpenClovis, Airvanacan access an executable complete system ofhigh quality. Airvana thus saves, to a not insignificant extent, its own development capacity and above all time, which would haveotherwise been invested in self-integrationwithout real added-value for the final product. �
EMBEDDED LINUX
www.comtel-atca.com
Systems● Proven 245W redundant cooling● All intelligent FRUs● Shelf Manager, Pigeon Point or Motorola● Designed for NEBS, TÜV
Backplanes● NEW split backplane design ● Better high speed performance ● Optimized power distribution ● Cost reduction
RoHS
Leading Edge TechnologyAdvancedTCA® solutions
ATCA CO14/CO16
ATCA CO5
MicroTCA the CUBE
NEW! ATCA CO2
11 June 2007
Product News
� Barric: SBC with open-source Linux OSBarric presents the Balloon board; currently aversion of the Balloon board is built for TobyChurchill, a company developing technologyfor speech impaired disabled people. The "Bal-loon" project is an open source, Xscale based,single board embedded computer using anopen-source Linux operating system and a col-lection of open source tools to enable engineersto develop on the platform.
News ID 214
� MontaVista: ADK 5.0 simplifies embeddedLinux development
MontaVista announces the release of Mon-taVista Application Developer Kit 5.0, an up-date of its embedded Linux developmentplatform. ADK 5.0 has been enhanced and re-designed to reduce the complexity of embeddedLinux application development and analysis sothis larger class of application developers can bemore productive.
News ID 72
Native mainline Linux: fit for embedded and real-time systems
EMBEDDED LINUX
� What is real-time and how is it measured? Asgenerally agreed, it is not an average fast re-sponse that makes a system real-time compliantbut the reliability that a defined response in-terval is not exceeded under whatever circum-stances. A system, for example, that reliablyresponds to external signals within 30 minutesis a real-time system whereas another system,that relies on a timely response within 30 mi-croseconds but fails once in a year, is not.
In order to determine the real-time capabilitiesof a system, an external signal is applied re-peatedly to a controller input such as a digitalinput line or a handshake line of a serial or par-allel communication controller. The analysis re-quires that a few lines of code are inserted intothe beginning of the respective interrupt serv-ice routine and into the initial part of a userspace program waiting for this interrupt. Theselines of codes are written in such a way that apulse is generated at an output device. Thispulse can then be measured along with theinput trigger using an oscilloscope. In addition,the respective interrupt line is also monitored.
The interval between the input trigger and thecorresponding level change of the interrupt lineis called “gate latency” or “controller latency”.The interval between the input trigger and thestart of the interrupt service routine is called“interrupt latency”, and the interval between the
input trigger and the execution of the userspace program is called “overall latency”, “pre-emption latency”, or simply “latency” (figure 2).Normally, a large number of single measure-ments is taken over several hours while gener-ating a realistic system load, and the longestmeasured overall latency is taken as the final re-sult. This is then called “worst-case latency”. Togain a better overview, the data points are bestdisplayed graphically in a latency histogram(figure 3).
Initially, the Linux operating system was adopt-ed primarily as a server platform - probably dueto its support of generally available and inex-pensive hardware in combination with its su-perior network capabilities and run-timestability. At a later stage, support for architec-tures other than x86 was added so the use ofLinux could have been extended immediately toembedded systems. However, embedded sys-tems often require real-time capabilities whichLinux as a server operating system was never in-tended to provide. In fact, in a 2001 survey pub-lished in April 2002 as a Venture DevelopmentCorporation Whitepaper (http://linuxde-vices.com/articles/AT6328992055.html), themost important factor inhibiting Linux adop-tion by respondents planning to use Linux wasthe item “real-time limitations”. At this time, itwas a common understanding that a real-timeoperating system must be conceived from the
very beginning as such and that every singlecomponent must be designed with all aspects ofreal-time execution in mind. Thus, retrofittingreal-time capabilities into the native mainlineLinux was deemed impossible. As a conse-quence, several Linux real-time projects werestarted that used the dual-kernel approach. Thisapproach replaces the Linux kernel by a smallreal-time kernel to which all interrupts are sentand which controls the entire system. One of itstasks is a modified Linux kernel; this enables ex-ecution of Linux user space programs and thusprovides binary Linux compatibility of the en-tire system (figure 4).
Examples of solutions using this approach areRTLinux, RTAI, and Xenomai; they have beenused repeatedly and successfully in recent yearsin a large number of industrial applications.RTLinux, RTAI, and Xenomai offer, similarly tothe RT-Preempt patches, a POSIX compatibleAPI to access real-time related functionality.Xenomai, in addition, comes with a collection ofother APIs (called “skins”). They greatly facili-tate the migration from proprietary real-timeoperating systems. In the foreseeable future, it iswell conceivable that Xenomai will adopt theRT-Preempt approach explained below andprovide additional functionality on top of thereal-time capable mainline Linux kernel. In par-allel, the mainline Linux evolved as well. A firstimportant step was to replace the O(n) sched-
By Carsten Emde, Open Source Automation Development Lab,
and Thomas Gleixner, Linutronix
When Linux was born onAugust 1, 1991, nobody
expected that 15 years later thescale of supported hardware
would range from tiny micro-controllers up to large multi-
processor mainframes, or thatLinux would become a
real-time operating system.
Figure 1. Linux Fedora 7 systemrunning a total of seven virtualsystems simultaneously
June 2007 12
uler by Ingo Molnar's O(1) scheduler. This newscheduler was already merged into the devel-opment Linux 2.5 kernel tree in 2002 and be-came the standard scheduler in Linux 2.6introduced in December 2003. It ensures thatthe time needed to schedule is fixed and deter-ministic irrespective of the number of activetasks - an important prerequisite towards real-time performance. Another important stepwhich was also added to the development ker-nel 2.5 and became readily available in 2.6 wasto enable scheduling during execution of cer-
tain parts of the kernel. This feature is calledkernel pre-emption; it prevents a low-priorityprocess from blocking a high-priority processwhile the former is executing a kernel call. Ini-tially, only small portions of the kernel codewere made interruptible but they have been ex-tended continuously.
Due to the very nature of the kernel tasks, therewill always be a certain part of the kernel codethat must be executed exclusively, and thus re-quires locking against other processes. Such
mutual exclusion is done using so-named mu-texes; but a general problem is, again, that alow-priority process executing a kernel func-tion under locking conditions may block ahigh-priority process that is about to executethe same kernel function. This phenomenon iscalled priority inversion, and one of the solu-tions is to bequeath the priority level to themutex and let it execute under the same prior-ity as the calling process. This special mutexvariant that is called Priority Inheritance Mutex(abbreviated PI Mutex) was implemented foruse in the Linux kernel.
In addition to kernel code, drivers and specif-ically their interrupt service routine may alsocause latencies. In consequence, to further re-duce the overall latency, a mechanism must befound to enable scheduling during execution ofthe interrupt service routines. This may not bemeaningful during its initial part when hard-ware is accessed more frequently, e.g. to ac-knowledge the interrupt and to read and writethe various data and status registers of a par-ticular controller. But it may be very effectiveduring the remaining stages of the interruptservice routine where hardware accesses occurless frequently. This feature is called interruptthreading.
EMBEDDED LINUX
Figure 2. The various elements of the overall or pre-emption latency
June 2007 14
Precise timing in real-time systems requirestimers with a resolution in the range of mi-croseconds. The Linux system clock, however,could only be set to an interval of 1, 4 or 10 mil-liseconds corresponding to 1000, 250 and 100Hz, respectively. Several coding attempts to in-crease the system clock failed, since clock fre-quencies above 1000 Hz did not scale. It was,therefore, decided, to separate the timers need-ed for high-resolution timing purposes fromthe system clock timers and to introduce a newfeature called high-resolution timers (“hrtimer”subsystem). This subsystem now also containsthe feature “dynamic tick”; it allows to com-pletely disable the periodic tick when the systemis idle and is used to safe battery power in mo-bile systems.
Until kernel release 2.6.18, only the O(1) sched-uler and a certain level of kernel pre-emption werepart of the mainline kernel. The other features,such as PI mutexes, high-resolution timers and interrupt threading, were only available when thekernel was patched with the “RT-Preempt” patchdownloaded from Ingo Molnar's repository atRedHat http://people.redhat.com/mingo/real-time-preempt/. The various elements of thispatch described above were developed over thelast couple of years by Ingo Molnar and ThomasGleixner with support from many kernel devel-opers worldwide. There was always hope that theRT-Preempt patch would be integrated intomainline Linux one day. But convincing LinusTorvalds that making Linux a real-time operatingsystem would do no harm, but greatly enhance itsusability, was not an easy task. Fortunately, therewere an increasing number of arguments tomake Linux real-time compliant - not only for industrial automation projects but also for standard server and desktop Linux systems: � Audio and video mixing and recording � Reliable timestamps in financial and
commercial transactions� Voice over IP and streaming video services.
At the occasion of the Ottawa kernel summit inJuly 2006, Linus Torvalds finally acceptedmerging the first part of the RT-Preemptpatches and agreed to gradually enter the other
components into mainline. As of kernel version2.6.21, approximately 60% of the RT-Preemptpatches (complete kernel preemption, PI mu-texes and the high-resolution timers) are al-ready available in mainline Linux. It is expectedthat most of the rest will have followed until theend of 2007.
Linux goes VirtualisationVirtualisation of an entire computer requiresan additional level of privilege (the “hypervi-sor” privilege) to ensure that the virtual ma-chine monitor can control both the system anduser level of a virtual machine and also that thevirtual machine cannot interfere with the ac-tivities of the host system or another virtualmachine. Up to about one year ago, this addi-tional privilege level was realised mainly in soft-ware such as products provided by VMware Inc.(Palo Alto, USA). However, most of the newerx86 processors from Intel and AMD provide therequired hypervisor privilege level in hardwarewhich greatly simplifies the design of the virtualmachine monitor. The related CPU flag iscalled vmx (Virtual Machine eXtension) in Inteland svm (Secure Virtual Machine) in AMDprocessors, respectively.
In October 2006, Avi Kivity posted a Linux char-acter device driver that allows to manage a vir-tual system by means of I/O control commands.In addition, he provided a modified version ofthe qemu emulator where the emulator was re-placed by the appropriate kvm commands(http://sourceforge.net/projects/kvm/). The orig-inal qemu emulator was developed by FabriceBellard (http://fabrice.bellard.free.fr/qemu/).Probably because of its elegant design, excellentcode quality and minimum interference with thekernel, Linus Torvalds decided only two monthsafter its initial submission to merge kvm intomainline Linux, and it was first released in kernelversion 2.6.20 on February 4, 2007.
An example of a Linux system that uses thekvm driver and the qemu emulator to simulta-neously run a total of seven virtual systems isgiven in Figure 1. None of the virtual systemsneeded any modification prior to be installedunder kvm. The performance of the virtual sys-tems is not different from that of the host sys-tem as long as straight CPU code is executed,e.g. during execution of Dhrystone or Whet-stone benchmarks. I/O operations, however,have a reduced performance, since the datamust be transfered twice - first from the virtu-al system via the virtual PCI bus to the host sys-tem and then from the host system through thephysical PCI bus to the peripheral device. Thehard disk transfer rate of the system shown infigure 1, for example, amounts to about 68MByte/s in the physical Fedora system and toabout 36 MByte/s in the virtual Fedora systems.
EMBEDDED LINUX
Figure 3. Example of latency data presenta-tion in a histogram. The vertical green barmarks the worst-case latency.
15 June 2007
EMBEDDED LINUX
In comparison to other virtualisation solutions,kvm has the advantage that it is part of main-line Linux, and will therefore be continuouslydeveloped and adapted to other kernel features.Of special interest is a real-time system runninga kvm virtual system with preserved real-timecapabilities of the host system. This would allowto use a single computer hardware for machinecontrol and graphical user interface - even if thetwo functions require two different operatingsystems. As a consequence, system reliabilitywould increase while costs would decrease.
The advent of open source software makes itpossible for the first time for different and evencompetitive companies to share their efforts anddevelop together base components of the oper-ating system. This common effort, however,
could be more effective if an organisation wereavailable to synchronise these activities. TheOpen Source Automation Development Lab(OSADL) was founded for this purpose(http://www.osadl.org). OSADL member com-panies are active in the field of industrial au-tomation such as machine, machine tool andequipment manufacturers, computer hardwareand software manufacturers and software serv-ice providers. The membership fee is primarilyused to delegate the development of machineand automation software components request-ed by a majority of the OSADL members. Cur-rent projects, among others, are: LinuxRT-Preempt patches, industrial I/O framework,migration tools, upstream submission of Linuxkernel components, making kvm real-time com-pliant, real-time Ethernet drivers for Linux. �
Figure 5. The “single-kernel approach” to realize a real-time Linux systemFigure 4. The “dual-kernel approach” to realize a real-time Linux system
Product News� MontaVista: version 5.0 of Linux and
development platformMontaVista releases Linux ProfessionalEdition 5.0, the latest version of its Linuxand development platform. Edition 5.0 isfirst to include the latest advancements inreal time technology including high res-olution timers and other native Linuxreal-time enhancements lead by Linuxkernel maintainer Ingo Molnar.
News ID 76
PC/104 CPU boards: new challenges,opportunities, and frontiers
SMALL FORM FACTOR BOARDS
� Since their introduction in 1991, systemsbased on the PC/104 standard have been wide-ly adopted for various applications, in trans-portation, in industrial control and in defence.Unmanned aircraft control, on-board vehiclecontrol and navigation systems, personal com-municators, add-on cards for customisedboards – all these make a far from full list of themost popular PC/104 applications. Many system developers choose this form factor dueto advantages such as low weight, compact size(boards just 90 x 96 mm), ease of applicationdevelopment and modification, plus the mechanical reliability of both the connectorsand the whole system.
Having been cutting-edge technology in thepast, PC/104 is now experiencing a new surge ofinterest thanks to the number of advanced,highly integrated boards available from itsmanufacturers, along with innovations bringingnew performance levels. To achieve the highestreliability, flexibility and performance charac-teristics, as well as short time-to-market, devel-opers can use many standard off-the-shelfboards from manufacturers world-wide. ThePC/104 board manufacturer consortium estab-lished 14 years ago unites more than fifty activecompanies specialising in development of various PC/104 standard products. According tothe Electronics Trend Publications data, theglobal PC/104 boards market amounts to
about US$170 million per year. Statistics showthe largest volumes of PC/104 boards are usedin industrial and defence systems, 50% and20% respectively. Functionally PC/104 boardsfall into two classes, practically identical in salesvolume: CPU (or processor) boards and input-output boards.
As the reliability of a whole system depends onthe reliability of its individual components,both the system development and the selectionof components (PCBs, connectors, chassis,power supplies, heatsinks etc.) need great at-tention to details. Customers’ demands forfunctionality and reliability of PC/104 systemsvery often require additional efforts fromboard developers and specific engineering
solutions. Below we will discuss main solutionsthat bring more processing power, improve reliability, widen operational temperature range,and therefore expand the PC/104 applicationarea. Attention is also drawn to some innovations.
Many applications require PC/104 systems tooperate over a wide temperature range ,oftenexceeding the range quoted in the componentmanufacturers ‘specifications. To increase reli-ability and widen operating temperature range,more robust versions of components can bechosen, for example tantalum capacitors insteadof electrolytic. However for some other com-ponents, such as high performance CPUs,chipsets, and memories this option does notexist. Therefore when developing a board for
By Alexander Buravlev, Fastwel
This article gives an overviewof PC/104 and presentsboards that expand the
product application area ofthis long-established standard.
June 2007 16
Figure 1. Frequency and TDP data for different processors
Processors / Chipsets Max.Frequency, Thermal Design Power Frequency to TDPMHz (TDP), W Ratio, MHz/W
Pentium M 760 2000 27 74
Pentium M 745 1800 21 85
Pentium M 1.6 1600 24.5 65
Celeron M 370 1500 21 71
LV Pentium M 738 1400 10 140
LV Pentium M 1.1 1100 12 92
ULV Celeron M 373 1000 5,5 182
855 GM/ICH4M FSB 400 MHz 4.3 + 2.2
915 GM/ICH6M FSB533/400 MHz 6 + 2.3
harsh operating conditions, the critical tasks arethermal effects analysis, designing the board soas to ensure its operability even in worst work-ing conditions, and comprehensive final testingover the whole operating temperature range.
Due to the boards’ small size and the far fromideal conditions for heat dissipation inside theclosed cabinet, special attention should bepaid to thermal stability and functionality, notonly at low operating temperatures, but espe-cially at high ambient temperatures. Forced air-cooling is not always acceptable in thesesystems. Their very compact size means aircooling is not very effective, moreover thedurability of cooling fans is low compared toprocessor boards. For instance, the MTBF of anaverage fan does not exceed 20,000 hours, butthe PC/104 processor board itself should havean MTBF well in excess of 100,000 hours. In theeffort to cope with the low heat dissipation ca-pability of PC/104 systems, many manufactur-ers use CPUs with low heat emission, such asthe Pentium III operating at 300…600 MHzwith reduced power consumption.
However, these processors have relatively lowperformance compared to what is needed formodern applications. Moreover, Intel recentlyannounced the phase-out of ultra-low-voltagePentium III processors, so it is risky to count onwhat may be left in suppliers’ stocks. It is more reasonable to use low-voltage versions of IntelPentium M family processors. These CPUs aremanufactured using special technology andhave almost double the advantage in frequency-to-TDP ratio compared to their congeners. Figure 1 presents frequency and TDP data forstandard processors, their low-voltage versionsfor embedded applications, and chipsets. Rela-tive thermal efficiencies measured in MHz perwatt of low-voltage Pentium M 738 and ultra-low-voltage Celeron M 373 are several timeshigher than for Pentium M 1.6 GHz.
Even though developers select a low thermalpower processor for their application, they stillhave to solve the problem of heat dissipation.Only a few companies in the world offer PC/104Plus processor boards with CPUs 1GHz and up.They use various solutions for heat dissipationsuch as heat pipes, low-profile fans, etc..
An effective conduction cooling solution is implemented on boards made by Fastwel. Thecommon practice of PC/104 processor boardmanufacturers is to locate the CPU on the PCIand ISA connector side. But these connectorsand the expansion board prevent effective heatdissipation from the CPU. Fastwel PC/104Plus processor boards are different. They are de-signed to be either at the top or the bottom inthe stack of PC/104 Plus cards and the CPU it-self is placed on the opposite side to the
PC/104 Plus connector. Thus the problem ofheat sinking is significantly simplified since heatcan be drawn from the CPU and GMCH chipvia a heat conductive plate and transferred tothe PC/104 cabinet. In such cases the contactarea is much bigger and the overall heat trans-mission resistance is much lower than withboards where heat pipes are used to transferheat along the board surface to the sides of thePC/104 enclosure. Provision of a low-heat-resistance thermal bridge between CPU andsystem chassis allows use of the whole PC/104cabinet as a large heatsink. This solution givesadvantages in size, weight, price and overall sys-tem design. Additionally it allows the use for the
most demanding computing applications ofhigh-performance Pentium M processors op-erating at frequencies of up to 2 GHz with a 533MHz front-side bus. For applications withhigh data exchange between CPU, memory andI/O the limitation in the system bus can be cru-cial; that is why the increase of system bus fre-quency from 400 to 533 MHz can give a 30%performance benefit in some cases. Systemsused in transportation or industrial environment are often exposed to permanentvibration loads. In these applications, thePC/104 architecture is itself a very good choicethanks to reliable fastening, the small size andlow weight of the PCBs, as well as the reliabil-
17 June 2007
SMALL FORM FACTOR BOARDS
Otto-Roehm-Str. 69 D-64293 Darmstadt / GermanyPhone: +49-6151-817320 Fax: [email protected]
MicroMod Base-Board
I/O-Board with CAN interface. Available in 5 different versions.
PCAN-PC Card
CAN card for PC Card-slots. Also available as optoisolated version.
PCAN-USB
CAN interface for theUSB-port. Also available as optoisolated version.
PCAN-PCI
CAN interface for PCI-slots. Available as 1- or 2-channel version.
PCAN-Explorer
A universal CAN-monitor with the following features: separate receive and transmission list, symbolic representation (e. g. through multiplexer and signal
integrated data logger, 4-channel linewriter, capability to add plug-ins (e. g. Plotter, Instruments Panel or CANdb-
PCAN-PC/104-Plus
CAN interface for PC/104-Plus systems. Available as 1- or 2-channel version.
ity of the PCI and ISA connectors. However, for many applications therecan never be too much reliability. The simplest way to increase reliabil-ity is soldering normally socketed components. The use of a soldered CPUis a routine technology, whereas soldering DRAM chips leads to parallelcircuitry routing, and requires addition of several conducting PCB layers.Moreover, soldered DRAM chips providing 1 GB of memory take about
15% of the PC/104-Plus board usable surface area. In general, solderingof components considerably complicates design and manufacture, butdrastically improves shock and vibration resistance. For example Fastwel’sCPC1600 withstands vibration levels right up to 5G over a frequencyrange 10 to 500 Hz, whereas SBCs with SODIMM memory usually onlywithstand tests up to 2G. There is another useful side effect of having allthe components soldered. The efficiency of additional damp-proof boardcoating increases, bringing down the risk of short circuits caused not onlyby condensed moisture, but also by salt mist, metallic particles or corrosion.
In the earliest stages PC/104 cards were equipped only with ISA con-nectors for plugging in extension modules. At that time the number “104”corresponded to the number of contacts between the connected modules.In 1997 the PC/104 consortium approved a new PC/104-Plus specifica-tion introducing an additional 120-contact connector for PCI interfaceextension modules. Unlike the PCI bus standard connector having 124contacts, the PC/104-Plus PCI bus does not support 64-bit data transfer.PC/104-Plus compatible systems are designed to support up to four extension cards via this bus.
The maximum theoretical bandwidth of the PCI bus within the PC/104architecture is 132 MB/s, while the real throughput does not exceed 55MB/s. The main applications using PCI bus within the PC/104 segmentare: extension boards with Ethernet controllers, video capture modules,digital signal processing boards and other applications requiring highspeed data exchange rates for the CPU. However for many modern ap-plications the bandwidth capacity of the 32-bit PCI bus is not enough.For example, many graphics cards require 500 MB/s and even more.Video-encoding and recording tasks need more capacity. A single-chan-nel MPEG-2 video compression solution using the Philips Semicon-ductors SAA6752 chip requires ~8 MB/s. Hence, ~5 video channels usethe whole bandwidth capacity of the PCI bus. As another example, the32-bit PCI bus cannot provide enough bandwidth to implement the fullspeed Gigabyte Ethernet interface. The most bandwidth-demanding ap-plications can be realised by means of the PCI Express bus. Being serial,PCI Express has a carrier frequency of 2.5 GHz and can provide up to 2.5Gb/s per one x1 lane with the option to combine lanes in various configurations like x4, x8 and x16. The PCI Express bus is now activelypromoted by a number of companies led by Intel as a universal bus forinterconnection of components within a board as well as modules within
SMALL FORM FACTOR BOARDS
Single Board Computers WithBuilt-in Data AcquisitionMore Performance • More Features
More Choices
Fewer boards in a system result in a more reliable, less costly, easier to assemble application. And a true singleboard solution means single supplier support with no integration surprises.
Diamond Systems delivers the industry’s broadest lineupof single board computers with built-in data acquisition— five rugged SBCs ranging from 100MHz to 2.0GHz withstate of the art peripherals AND analog and digital I/O.
Visit www.diamondsystems.ch for more informationabout how single board CPU plus data acquisition solutionsenable more reliable, lower cost and easier to assembleembedded applications.
www.diamondsystems.ch
Model PROMETHEUS ELEKTRA ATHENA HERCULES POSEIDON
Form Factor PC/104 PC/104 4.2" x 4.5" EBX EPIC
Clock Speed 100MHz 200MHz 400 / 660MHz 800MHz 1.0 / 2.0GHz
Memory 16 / 32MB 128MB 128 / 256MB 256 / 512MB 512MB
Exp. Bus PC/104 PC/104 PC/104 PC/104-Plus PC/104-Plus
USB 2 2 4 (4) 2.0 / (4) 1.1 (4) 2.0
IDE/SATA IDE IDE IDE IDE IDE/SATA
Ethernet 10/100 10/100 10/100 10/100 Gigabit
Serial 4 4 4 4 4
Video 3 3 3
Audio 3 3 3
Analog 16 16-bit, 16 16-bit, 16 16-bit, 32 16-bit, 32 16-bit,Inputs 100KHz, 48 FIFO 100KHz, 512 FIFO, 100KHz, 512 FIFO, 250KHz, 2048 FIFO, 250KHz, 1024 FIFO,
autocalibration autocalibration autocalibration auto autocalibration
Analog Outputs (4) 12-bit (4) 12-bit (4) 12-bit (4) 12-bit (4) 12-bit
Digital I/O 24 24 24 40 24
-40 to +85°C 3 3 3 3 1.0GHz only
Figure 3. CPC1700 PC/104 SBC with PCI Express bus
Figure 2. Carrying capacity per contact ratio for different interfaces
a computing system. Due to its flexibility and scalability, the PCI Express bus will be able to meet system developers’ requirements untilthe middle of the next decade. In addition to the high throughput, theadvantages of PCI Express include: lower signal delay values, improveddata burst transfer protocol, and the option to set processing priorityfor data packages (quality of service).
At the physical level, one PCI Express channel is realised as two pairsof low-voltage differential signal lines at 2.5 GHz. This feature bringssignificant advantages for size-constrained boards like PC/104. Lessrouting means fewer interconnect layers in the PCB and more space foradditional components or interface placements on the board. This iswhy silicon manufacturers and embedded boards developers will switchto PCI Express technology, even for applications that do not requiresuch a high bandwidth bus. In summary PCI Express simplifies wiringand saves PCB space; this in return allows to reduce costs or to addfunctionality. Thus utilising PCI Express lanes available from the ICH(south bridge), the Fastwel CPC1600 has an on-board GEthernet con-troller providing two full-speed GEthernet channels. Figure 2 presentsthe carrying capacity per contact ratio for different interfaces.
PCI Express can provide advantages not only for on-board data trans-fer, but for board-to-board connection as well. As an example, mostmodern Intel server boards do not have PCI slots any more, most expansion boards will work via PCI Express interfaces. From a globalperspective the move from PCI to the PCI Express interface looks likea logical step in the overall shift from parallel interfaces to serial. Thus,USB is replacing LPT, and SATA is superseding EIDE. The PICMG con-sortium developing standards for passive backplane embedded systemshas recently announced several specifications in which the PCI bus isreplaced by one or another serial bus, such as PCI Express, Infiniband,RapidIO, and others. Therefore the appearance of a high-speed serialinterface as a data exchange interface in popular PC/104 standard is natural and just a matter of time. The nearest neighbour to the PC/104standard, the EPIC standard (SBCs manufactured according to EPICstandard can accommodate PC/104 modules through the same ISA andPCI connectors), has recently received the pre-release version of theEPIC-Express specification. In this version, the PCI bus is replaced bythe PCI Express bus with four (10 Gb/s – one bank of connectors) ortwelve (three banks of connectors) PCI-Express lanes.
This solution is quite logical, because the bus used for data exchange(PCI) is replaced by a faster one (PCI-Express), while a low speed ISAbus, traditionally used for simple I/O tasks and for signaling is kept unchanged. This idea has inspired Fastwel developers to createCPC1700, the first PC/104 SBC with the PCI Express bus. Actually, itis a modification of Fastwel’s CPC1600 PC/104-Plus CPU board, inwhich the PCI connector is replaced by a four lane PCI-Express one,which can be set up as four x1 lanes or grouped as 1 x4 lanes. .
Both EPIC-Express and Fastwel’s PC/104 Express computing boards arecompatible with the majority of ISA-based application cards developedby PC/104 consortium member companies and other designers who dotheir proprietary application boards with data transfer through the ISAbus. Thus embedded system developers can now use new compactboards equipped with high performance CPU and high-speed serialdata exchange interfaces. Security video processing (downstreaming,encoding, packetisng and storing), image capturing and recognition insurveillance systems, target capturing and tracking in defence appli-cations are just a few possible applications for such high performancecomputing boards as CPC1600/1700 and boards manufactured according to the EPIC-Express standard. These new products open newhorizons in the development of modern high-performance solutionsfor robotics, security, transportation, avionics and defence systems. �
RUTRONIK · Elektronische Bauelemente GmbH75228 Ispringen/Pforzheimrutronik.com
Display and Control Technology
In our Visual World seminar and workshop series, we will present to you the entire spectrum of flat display technology for all requirements and market segments such as industrial automation as well as medical,gaming, maritime or automotive applications.
Learn more about modern information systems as well as the latest technologies for the use of flat displays in almost any environment.
Our seminars are accompanied by experts from our partners and manufacturers.
Committed to excellence.
Workshops 2007
Visual World.
Venues: FH Pforzheim 26.06.2007Düsseldorf 28.06.2007
Earthquake research satellite demonstrates PC/104 ruggedness
SMALL FORM FACTOR BOARDS
� Scientists feel there is sufficient computerpower available to crunch the complex modelsrepresenting the movements of the Earth’stectonic plates and find exactly where the next“crunch” will come. All that is needed is to col-lect data to produce and update the model con-tinuously.
Since the disaster in Skopje 1963 when thou-sands died, scientific models have improvedgreatly. Skopje University has been compressingrocks and breaking them to observe how theybehave under pressure. Just before the rockbreaks, the cylinders applying pressure releaseslightly, thus extending the breaking process almost indefinitely. Piezoelectric sensors mon-itor the rock, all the time preserving it fromcomplete break-up.
Others have been doing similar experimentsaimed at understanding the process and look-ing for behavioural patterns. At the same time,sensors, computers and software have im-proved rapidly. By the dawn of this century, so-phisticated modelling, software and knowledgeof the earthquake process already allowed sci-entists to move to a new level, and to start col-lecting live data to predict future disasters andwarn those potentially affected in advance. Sev-eral systems have been put in place across theglobe and in space to view and measure tectonicactivities.
Launching and operating a satellite-based sys-tem makes severe demands in terms of vibra-tion, electric, mechanical and thermal stress,limited size, and etc. QuakeSat was an innova-tive and compact Linux-based satellite de-signed to detect earthquakes from space. Itsprimary scientific mission was to detect, record,and downlink extremely low frequency (ELF)magnetic signal data, potentially leading togroundbreaking techniques to predict earth-quake activity. It used the Prometheus inte-grated PCI/104 CPU board from DiamondSystems as its main control system. The satelliteoperated successfully in its orbit 600 milesabove the Earth’s surface for about 3 years, twiceas long as expected.
QuakeSat measured a tiny 4" x 4" x 12". It wasdesigned to fit inside an ultra-miniature satel-lite enclosure mounted in a Russian rocket thathad been decommissioned as a result of the for-mer START talks. This severe size restrictionlimited the choices for off-the-shelf CPUboards. A satellite involves extreme design re-strictions that limit the choices of components.Size, weight, power consumption, and rugged-ness are key concerns. Only PC/1O4 was able tomeet these criteria and provide a suitable com-puting platform for the QuakeSat electronics.QuakeSat serves as an excellent example ofwhat can be achieved using rugged PC/104technology.
Diamond Systems’ Prometheus CPU is basedon the ZF Micro ZFx86 low-power integratedx86 CPU-on-a-chip. The board includesprocessor, memory, Ethernet, serial ports, IDE,and a full analog and digital I/O circuit.Prometheus combines two boards (CPU andI/O) into one to reduce the size and weight ofthe embedded system by 50 percent.Prometheus was chosen for use in QuakeSat be-cause it met the following key criteria:
� miniature size: PC/104 (3.6" x 3.8") was theonly easily expandable form factor to fit insidethe small enclosure;
� integrated design: combining the processorand data acquisition functions on a singleboard reduced the size and weight of the CPUand enabled more electronics to fit in the satellite;
� low power; total power consumption of lessthan 3 watts minimised the drain on the satel-lite’s limited solar-collector system;
� rugged design: Prometheus utilises soldered-on memory for extra ruggedness, and has a test-ed and verified operating temperature range of-40 to +85°C. Temperature and vibration immunity enabled the electronics to survivelaunch conditions and operate in a space envi-ronment with extreme temperature swings. �
By Stephen F. Baginski, Diamond Systems
QuakeSat, a compactLinux-based satellite designed
to detect earthquakes fromspace, used the Prometheus
integrated PC/104 CPU boardfrom Diamond Systems.
Prometheus board with ZF Micro low-powerx86 CPU
June 2007 20
ETXexpress is the step-up module for sustainableproducts. Get started with the starter kit. We lookforward to hearing from you.
Info-hotline:+49(0)[email protected]
If it‘s Embedded, it‘s Kontron.
For future proof productsETXexpress™ is based on COM Express (PICMG).
It provides for embedded systems up to
Intel® 2 Core™ Duo processor technology –
with 100% higher performance at only 15% higher
power consumption and Virtualization Technology VT-x.
ETXexpress™ (module size basic)
PCI-Express and PCI-Express Graphic
Gigabit Ethernet - for high connectivity
SerialATA - for fast drives
COM Express connectors - for highest transfer rates
microETXexpress™ (module size compact)
Based on ETXexpress/COM Express
Available with PCI Express or PCI
Compact size for mobile applications
ETXexpress™ & MORE!For your perfect-fi t Baseboard.
ETXexpress™ Flatpanel SolutionsWe drive your displays!
Copy
right
© 200
5 Ko
ntro
n AG
. All
right
s re
serv
ed. K
ontr
on a
nd th
e Ko
ntro
n lo
go a
nd a
ll ot
her t
rade
mar
ks o
r reg
iste
red
trad
emar
ks a
re th
e pr
oper
ty o
f the
ir re
spec
tive
owne
rs a
nd
are
reco
gniz
ed. R
oHS
com
plia
nt lo
go m
eans
tha
t pr
oduc
t w
ill b
e le
ad-f
ree/
RoHS
com
plia
nt la
test
by
June
30t
h, 2
006.
Rev
. #G0
20eu
10_W
MH ET express
www.etx-express.com
®
Use of MicroTCA technology in carrier-grade applications
MICROTCA
� Network equipment providers (NEPs) oftenconsider µTCA as complementary to Ad-vancedTCA, allowing them to cost-effectivelyaddress the low-end range of core applications,such as VoIP Media Gateway, 3G RNC, NGNmedia server, IMS MRF, signaling gateway, orsoftswitch. In particular, NEPs intend to use thesame AMC modules in both µTCA and ATCAconfigurations, leveraging on the commonali-ty of the two platforms, especially for the net-work interface modules and the processingAMCs. Using the same modules will obvious-ly reduce their development and maintenanceeffort and should reduce the cost of goods. Thesupport of carrier-grade applications meansthat those µTCA systems will have to be high-ly reliable, i.e. all critical functions (backplane,power supply, MCH, fans, etc.) at least will beredundant.
However, some major differences betweenATCA and µTCA platforms make it difficult toreconcile the goals of using the same AMCmodules for both while taking advantage ofeach form-factor. In particular, ATCA bladesrepresent an intermediate level of sub-systembetween module and chassis, which does notexist in µTCA. Thus, the notion of FRU (fieldreplaceable unit) differs: depending on the ap-plication and on the maintenance model, theATCA FRU is either the blade or the AMC. InµTCA, it is the AMC or the chassis itself. Be-
cause the FRU impacts on failover mechanisms,physical replacement and controlled restart,while preserving the system functionality, itsdefinition heavily weighs on the system archi-tecture and the functions expected for theAMCs. Also, AMC modules typically complywith a common specification (AMC.0) and, de-pending on their connectivity, with one (ormore) so-called dot-specification (AMC.1,AMC.2, AMC.3 or AMC.4). When modulescomplying with different dot-specs are mixedtogether in a chassis, there may be certain in-compatibilities on the AMC port usage. InATCA, module interoperability issues are main-ly at a blade level, because the blades somehow“isolate” their modules from other modules inthe chassis. Within a µTCA chassis, all AMCs
are “exposed” to each others, directly or via oneor two MCHs. Consequently, module interop-erability issues are impacting all AMCs. Beyondthe features and specifications of µTCA, let’s seehow one can design a signaling gateway withproducts available today.
The SG is an element of the NGN model,adapting the SS7 protocol layers (used in tra-ditional PSTNs) to an IP transport. Accordinglywith the SIGTRAN architectural framework(defined in IETF RFC 2719), SS7 low layers(MTP2, MTP3 and/or SCCP) are adapted totheir SIGTRAN equivalent (M2PA, M2UA,M3UA, SUA) on the IP network side, and a reliable transport is provided by SCTP (RFC2960). Physically, the SG is separate or inte-
By Marc Orange, Interphase
In this article the applicabilityof MicroTCA technology in carrier-grade telecom
applications is examined witha signalling gateway use.
Figure 1. SG software architecture example
June 2007 22
Figure 2. ATCA solution
grated together with a call agent or a mediagateway. Among several architectures definedby SIGTRAN framework, we’ll focus here on anSG supporting an MTP3 layer, and convertingit to M3UA (MTP3 User Adaptation) onto theIP network. Medium and large SGs have no sin-gle point of failure, and all functions are at least1+1 redundant. A redundant SG is typicallybased on N pairs of SS7 I/O controllers, sup-porting a partial or full MTP2 layer and 2 or
more general purpose processors, controllingthe SS7 controllers and supporting a redundantor distributed MTP3 layer, the nodal inter-working function, SIGTRAN layers and IPlayers. Traffic processing (e.g. load balancingand message routing), and node management(e.g. configuration, high-availability, file man-agement) are also performed on general pur-pose processors. For small configurations,protocol layers, traffic and node management
functions are usually implemented on thesame processors. For medium and large SGconfigurations, those functions are distrib-uted on several processors. Let’s consider a dis-tributed configuration (figure 1) with twoMTP3 processors, each using 1 or 2 SS7 con-trollers (i.e. 2 or 4 SS7 controllers total), andtwo other processors executing redundantlynode management and SIGTRAN and IP stack.Redundancy is achieved by internal synchro-nization between mated processors, and theSS7-to-SIGTRAN interworking function isdistributed within the whole node. InterphaseiSPAN 3639 AMC is the SS7 controller of choicefor ATCA and µTCA systems.
It provides 4 or 8 T1/E1 ports, supports up to31 SS7 links on each port, and runs the fullMTP2 protocol on its Freescale PowerQUICCIII processor. In terms of AMC connectivity, theiSPAN 3639 supports one x1 PCI-Express link,on port 1 (AMC.1 Type P) or Port 4 (AMC.1Type 1) and two GigE links, on ports 0 and 1(AMC.2 Type E2). The MTP2 service is avail-able via an API (iWARE) on both connections.For the IP interfaces, a dual or quad EthernetNIC AMC is appropriate, such as InterphaseSlotOptimizer 364G or Kontron AM 4300.Those cards support four 10/100/1000 Gigabit
MICROTCA
Figure 3. MicroTCA solution (single chassis, PCI-Express links)
• AMC • µTCA • cPCI • VME • PMC • custom
Driven by our motto Innovation in Communication we have
extended our portfolio of solutions for (tele-) communication
core and infrastructure applications by a complete µTCA product
line. N.A.T.‘s product range includes turnkey systems, comprehensive
software support and board level products, such as
NAT-MCH, NAMC-8560-xE1/T1/J1, NAMC-STM1/4
and others. For more information visit our web site at
www.nateurope.com or call us at +49-2241-3989-0
•
complete systemsturnkey and custom
• MCHµTCA Carrier Hub
•AMCTelecomms and Network
N.A.T. Gesellschaft für Netzwerk- und Automatisierungs-Technologie mbH l [email protected] l www.nateurope.com
Ethernet ports and an AMC connectivity of up to x8 PCI- Express links,on ports 4 to 11 (AMC.1 Type 4). For the storage, an AMC hard disk, suchas Kontron AM 4500 provides a 40-60 GB via a SATA interface on port2 (AMC.3). Interphase iNAV 9400 Signaling Gateway is a pre-integrat-ed hardware + software solution that can be delivered in all form-factors.In ATCA, the SG software architecture of figure 1 can be implementedwith a pair (or two) of single board computers supporting 2 AMC mod-ules each. Intel, Motorola, Kontron, Sun, offer such ATCA SBCs. For ex-
ample, Kontron AT 8020 provides with 2 Intel Dual-Core Xeon proces-sors, up to 16GB SDRAM and 1GB flash drive. On its 2 AMC bays, it sup-ports up to x4 PCI-Express links, on ports 4 to 7 (for AMC.1 Type 3), twoGigE links, on port 0 and 1 (for AMC.2 Type E2) and a dual portSAS/SATA on port 2 and 3 (for AMC.3) An example of ATCA SG archi-tecture is represented in figure 2. The interface between the MTP3 proces-sors and the MTP2 AMC modules is logically x1 PCI-Express (via port4) which provides the most efficient and reliable interface.
This uses a PCI switch available on the SBC. On the management and IPinterface blade, the GigE AMCs are controlled via x4 PCI-Express links(ports 4 to 7). The hard disks are controlled via the SATA port 2. All nodeinterworking, management and layer synchronization messages use theGigE ATCA switch fabric. This architecture provides a reliable and pow-erful SG solution, supporting 248 SS7 links on 8 E1s (with 2 iSPAN 3639-4 ports), or 496 SS7 links on 16 E1s (with 4 iSPAN 3639-4 ports). The SGcapacity can be extended by adding SS7 blades, with steps of 124 links or248 links (4 or 8 E1s). Those 4 ATCA blades, together with a pair of switchblades, would fit in a small ATCA chassis.
A more integrated – but less flexible - solution would use a single pair ofblades to run all software, a SATA AMC replacing one MTP2 AMC. Thecapacity of such a 1+1 SG would be 248 links / 8 E1s max.. Alternative-ly, an ATCA switch blade, running the management software andequipped with a GbE AMC and a SATA AMC, would preserve the 2 SS7AMCs per SBC and would scale up easily. MicroTCA implementationsof the same software architecture (figure 1) vary with several factors: con-nectivity offered by the backplane, number and type of AMC modulessupported by the backplane, and switching capabilities supported by theMCH. To measure the system compatibility and continuity betweenATCA and µTCA, we’ll use the same AMC modules. A first approach con-sists in replacing the ATCA SBCs with processor AMC modules(PrAMC) and keeping the same interfaces between modules. Figure 3 rep-resents such a solution based on Schroff MicroTCA development system(backplane 23005-406, supporting 10 AMC single-width/full-height slots,2 MCH and 2 power modules).
The interface between each MTP3 PrAMC and its associated MTP2 AMCis an x1 PCI-Express link (via port 4). This uses a direct fat pipe con-nection in the backplane between adjacent AMC slots. For the IP inter-faces, GigE AMCs are controlled by the PrAMCs via x4 PCI-Express links(ports 4 to 7). Hard disks are controlled by the PrAMCs via a switchedSATA connection, available in each MCH. Port 2 is switched by MCH1,
June 2007 24
MICROTCA
Figure 5. MicroTCA solution (dual chassis and GbE links)Figure 4. MicroTCA solution (dual chassis, PCI-Express)
while port 3 is switched by MCH2. All messages for node interwork-ing, management and layer synchronization use the GbE base fabric onport 0 (switched by MCH1) and port 1 (switched by MCH2). GbE andSATA switching are supported by 2 MCHs, such as the CorEdge Networks CEN-MCH. This solution is highly reliable: all functions,modules and internal communication paths are duplicated (via the twoMCHs).
This architecture provides a cost-effective and powerful SG solution ina single µTCA chassis, with a capacity of up to 248 links / 8 E1s. Usingthe same AMCs, it represents an excellent commonality with ATCA.However, the chassis is full; there isn’t room for spare modules and thesolution can hardly scale up / down. Another approach consists in implementing the system reliability at a chassis level, and building a redundant SG with two µTCA chassis. Figure 4 illustrates such a solution, based on two Motorola Centellis 1000 Series communicationservers. GbE and PCI-Express switches represented on this figure areactually located on the MCHs.
The Centellis 1000 Series can support 10 single-width/full-heightAMCs, 1 MCH and 1 or 2 power modules. In this example, the inter-face between each MTP3 PrAMC and its two associated MTP2 AMCsis an x1 PCI-Express link (via port 4). This uses an extended PCI- Express fabric on the MCH, which supports PCI-Express switchingwith all AMCs, in 1 to 4 separate domains. The GigE AMCs are controlled by the PrAMCs via x4 PCI-Express links (ports 4 to 7), alsoprovided by a switch on the MCH.
Hard disks are controlled by the PrAMCs via port 2 and direct SATAconnections between adjacent slots (daisy chained on all AMC slotsthroughout the chassis). All messages for node interworking, man-agement and layer synchronization go through the GbE base fabric onthe MCH via port 0. The base fabrics of the two chassis are connect-ed together via two front GbE expansion ports on the MCHs. With 4spare slots in each chassis, 2 sets of SS7 modules can be added, thusdoubling the SG capacity.
This architecture demonstrates the capability of the µTCA standard tosupport powerful and carrier-grade applications, not only for small, butalso medium and large configurations. With 2 chassis inter connectedtogether, the solution can scale easily from 248 links / 8 E1s up to 992links / 32 E1s. An alternative approach consists in using the GbE connectivity of the MTP2 modules instead of PCI-Express. Figure 5represents a dual-chassis architecture with MTP3 processors control-ling their MTP2 AMCs via the GbE base fabric on port 0. The rest ofthe architecture is unchanged. The use of GbE interface between mod-ules improves the flexibility of the architecture. Indeed, GbE fabrics arecommonly supported on µTCA backplanes and MCHs. Based on sin-gle or dual star topologies, they serve all AMC slots. Consequently, thelocation of the modules in the chassis does not matter. Moreover, theGbE interface is a balanced interface, even though a module can logically control another module. By opposition, PCI-Express, with separate domains, and one root complex per domain, imposes the location of the controlling and the controlled modules.
This signaling gateway use-case confirms that MicroTCA can providea functional environment for powerful and highly reliable telecom ap-plications, with products available today on the market. As MicroTCAtechnology is rapidly improving, it is recommended to check with thesuppliers for functionality and/or further developments before order-ing the products referenced here. It demonstrates that MicroTCA is notlimited to small capacity equipments, but it is also an effective solutionfor medium capacities. Thus, it proves to be a reasonable complementto ATCA as open technology for infrastructure systems. � http://www.cct.co.uk
All Trademarks acknowledged
Email: [email protected]: (+44) 1206 752626
Mar07/UK
�
�
�
choice of Intel Core Duo processor orIntel Core 2 Duo processor
®
®
™
™
up to 4 Gbytes DDR2 SDRAMextended temperature operation available
VME/VXS
CompactPCI®
�
�
�
choice of Intel Core Duo processor,Intel Core 2 Duo processor ordual Dual-Core Intel Xeon processors
®
® ®
™
™®
up to 8 Gbytes DDR2 ECC SDRAMextended temperature operation available
AdvancedMC™
�
�
�
�
up to 2.0 GHz Dual-Core Intel Xeonprocessor LV
® ®
up to 16 Gbytes DDR2 ECC SDRAMdouble-width/full-height sizefor AdvancedTCA or MicroTCA® ™
Dual-Core Processors
on AMC, cPCI and VME
Dual-Core Processors
on AMC, cPCI and VME
High signal integrity speeds up MicroTCA systems
MICROTCA
� The implementation of real-time applicationsrequires a powerful infrastructure for the datatransfer which has to distinguish itself by largetransmission bandwidths, at least possible delays, and minimum risks of loss. High datatransfer rates are required mainly by moderntelecommunication applications. This field extends increasingly towards sophisticatedmultimedia applications and demanding in-dustrial fields of application, such as machineand process control. In particular the MicroT-CA standard contributes decisively towards this development, thanks to its high degree ofmodularity. In this connection, signal integri-ty is of great importance. As we will see in thefollowing, certain reflections on 3 GHz systemshave to be made that take into consideration thesignal behaviour of up to 15 GHz!
The following contribution is meant to speci-fy the “natural enemies” of signal integrity andto sensitise to the fact that with parsimony, youeasily back the wrong horse in this particularfield. Or, put into positive words: it’s worth theeffort to look for the right partner. After all, thecomplete information chain that even includesperipherals is decisive for the quality and im-munity to interference and fault tolerance.Our considerations are based on serial datatransmission. The standardisation committeesat PICMG and VITA realised at an early stagethat for high-speed applications, you have todismiss classical bus structures and switchover to point-to-point connections with dif-ferential pairs. To avoid the formation of long
sequences of logical zeros and ones on seriallytransmitted data packets – which complicateclocking and data regeneration – the 8B/10Bencoding is successful in such applications(see text frame).
MicroTCA, the latest development of the Ad-vancedTCA standards which promises partic-ular success, offers the possibility to transmitdata by Gigabit Ethernet, PCI-Express, SerialRapid I/O, Serial Attached SCSI or AdvancedSwitching. At first, the question arises whymake highly crucial reflections on signal in-tegrity. After all, it’s a XAUI transmission bymeans of an NRZ signal with a data rate of“only” 3.125 Gbps (XAUI = X Attachment UnitInterface, an interface used in 10 Gigabit- Ethernet; NRZ = “Non-Return-to-Zero” signal,a signal that does not return to zero at regularintervals). It is a fact, however, that the edgesteepness of the signals determines the band-width. Thus, at a basic frequency of 3.125 GHz,the signal integrity for a bandwidth of at least15.625 GHz has to be guaranteed. This is due tothe high-frequency, low-level harmonic waves,strictly speaking to the 5th harmonic wave ac-cording to Fourier. With MicroTCA, AdvancedMezzanine Cards are defined as smallest mod-ular assembly. For cost-reduction reasons, thistechnology uses contacts via connectors on thebackplane and gold pads directly on the AMCboard. Due to their small mechanical shape, 170contacts have to be realized via these connectorson a width of only 73.5 mm. To investigate theperformance of different printed conductor
geometries, Elma Trenew has carried out fundamental metrological examinations. Forthis purpose, a MicroTCA test backplane withappropriate high-quality test boards was de-veloped. It served to measure the physicalproperties by means of a TDR (time-domain reflectometer) and VNA (vector network Ana-lyzer). These measurements assisted in derivingdecisive findings on important influencingquantities that affect signal integrity. The meas-urements showed that the Inter Symbol Inter-ference (ISI) effect decisively influences signalintegrity. The ISI effect describes how pulse-
By Oliver Schroth, ELMA Trenew Electronic
Based on the results ofdetailed experiments with
signal behaviour on thetransmission path, ELMA hasoptimised its layout strategy
for MicroTCA backplanes.
The blu! box is a new MicroTCA shelf with optimised backplane for demanding practical tests.
June 2007 26
8B/10B encoding
When using 8-bit words, you will get 256(28) possible bit combinations, 10-bit se-quences however, result in 1024 possible bitcombinations. For the formation of 8-bitwords, the 8B/10B code only uses such bitcombinations that provide multiple statechanges. These are words which contain amaximum sequence of 5 zeros or 5 ones ina row. Thus it is possible to safely generatea clock signal that is required for synchro-nisation from a coded signal with an ade-quate number of state changes. Thanks tothese advantages, the 8B/10B encoding isused in Gigabit Ethernet, 10-Gigabit Eth-ernet, with Fibre-Channel, StarFabric andgenerally with ATM (Asynchronous Trans-fer Mode) interfaces.
27 June 2007
MICROTCA
shaped signals are broadened compared to theoutput signal through dispersion, skin effect, dielectric losses and reflections. In addition, theinductive stub effects (with THT connectors) orbone effects of up to 1.2 nH (that occur whenconnecting SMT connectors to the inner layersof the backplane) contribute to the ISI effect.Since the pulses can be broadened through intersymbol interference over several periods, this ef-fect limits the transmission bandwith that canbe actually used. Each conductor made of copper has different line capacities and induc-tivities – depending on the layout geometry, dielectric insulation and frequency.
Because of this, pulse-shaped signals are broad-ened. Such broadening increases with the phys-ical length of the transmission line. This resultsin different propagation rates and thus intransit-time distortions for each individualcontactor which can be seen in the eye diagramas jitter. For each differential pair, they have tobe adjusted to each other to a few picoseconds(which corresponds to less than 1 mm ofphysical path). Especially for broad-band sig-nals, an optimisation has to be obtained by cor-rectly selecting the basic material to be used andby adopting a clever layout strategy. This hap-pens in several iterative steps since the individ-ual parameters have an influence on eachother. This requires the backplane design engi-neer to have well-founded experience so that heis able to find a carefully selected compromisefor which all factors of influence are evaluatedand in combinazion minimised.
With increasing frequency, the importance ofthe DC conductor decreases, while the effectstaking place in the electric field become moreand more dominant. In the case in hand, themeasurements also showed that the data trans-mission does not take place in the conductor,but mainly in the field. Thus, we do not have toobserve the conductors and connector pins, butthe field in which the data transmission takesplace as a whole. In addition to this, it is neces-sary to reduce the influencing quantities of dis-persion, skin effect, dielectrical losses, stub andbone effects by means of appropriate connectorand layout geometries and thus to positively in-fluence signal integrity, always while taking intoconsideration economic solutions using theleast expensive basic material. But it’s not onlyone interference factor. Beside the factors de-scribed above, another important item has to bementioned: cross-talk between differentialpairs. Ideally, neighboring pairs are separatedfrom each other and shielded by ground planesin which case only few “aggressors” could in-fluence the individual pair. In practical opera-tion, the high number of layers of the boardnecessary for doing so, and thus the high costs involved and the manufacturing difficultiesspeak against it. Plus, the stub effect becomes
more and more important with an increasingboard thickness. The experiments also showedthat in the backplane, the signal is considerablydegraded due to attenuation and losses. X-talk(crosstalk) on the other hand dominates in theconnector. For this reason, the relevant pa-rameter for x-talk measurements is the NEXT(near ended crosstalk) parameter. Oddlyenough, FEXT (far ended crosstalk) benefitsfrom the attenuation on the board and can beneglected provided that this influence is known.
What does this mean in practice?To sum up, the following can be said: It waspossible to optimise the layout strategy for Mi-croTCA backplanes. That way the number oflayers in the board can be reduced and in spiteof that, all requirements of the x-talk limits ac-cording to the MicroTCA specification can atleast be satisfied or even surpassed. Thus, ELMAsucceeded for example in transmitting 6.25 Gi-gabits via one of their AdvancedTCA backplanesin cooperation with their partner Enigma, usingan ingenious encoding. There still were sufficientspare capacities so that an even higher data trans-fer rate is realistic.
The measured parameters can also contributeto simulation programs. Thus, it will be possi-ble in future to investigate the complete pack-age of connectors plus board in HSPICEsimulations regarding the requirements onsignal integrity. This decisively reduces the de-velopment process and supports users in re-ducing the time-to-market. ELMA supportspotential users of MicroTCA technology withtheir blu! box, a ready-to-run MicroTCA shelffor form factor testing and intense board tests.The fully enclosed 19" subrack is 5 U high and200 mm deep. The tested basic system is fullyredundant thanks to 2 x MCH, two powermodules for - 48 V / - 60 V and two coolingunits. The speed of their high-performance fansis PWM controlled. For this purpose, and forthe exact detection of the thermal situation inthe test structure, up to 20 temperature sensorscan be positioned in the shelf.
The system can be enlarged with maximally 12AMC, 8 “full height” and 4 “half height” in accordance with the form factor. All modulesare hot swap components and IPMI- controlled.The inserted backplane benefits already fromthe results of the detailed series of measure-ments described above. It has a 20-layer structure and guarantees optimum signal integrity. Plus, it supports all field replaceableunit functions. High speed routing is effectedvia dual star technology, the FCI connectorsused are developed especially for high-speeddata transfer. Two options are available fortelecommunication and non-telecommunica-tion applications. �
MEN Mikro Elektronik GmbH,Neuwieder Str. 5-7, D-90411 Nürnberg,Tel. 0911-99335-0, www.men.derp
f W
erb
eag
entu
r |
ww
w.r
pf.
de
■ F17 – Core® 2 Duo T7400, 2.16 GHz■ F15 – Core® Duo T2500, 2 GHz■ F14 – Pentium® M 760/Celeron® M
373, 1..2 GHz■ Side Cards – UART, Multimedia,
USB, FireWire etc.
■ Compatible 3U Intel® familywith scalable computing performance
■ Long-term availabilitydue to easy system adjustment
■ For harsh industrial environmentsand mobile applications
Intel®Pentium®M upto CoreTM(2) DuoCompactPCI®/Express
28June 2007
Figure 1. The SCOPE referencearchitecture for carrier gradeplatforms
The main objective of theSCOPE Alliance is to help to
develop a powerful ecosystemthat enables the network
equipment providers to buildcarrier grade telecom munica-
tion systems from compo-nents supplied by different
vendors that fulfill the needsof their customers.
By Timo Jokiaho and Louise Moser, SCOPE Alliance
� In the telecommunication industry, the net-work equipment providers develop and use car-rier grade platforms within their networkelements that must provide high availability, re-liability, serviceability, scalability and perform-ance. Increasingly, carrier grade platforms areimplemented as commercial off-the-shelf(COTS) hardware and software componentsand free open source software (FOSS) compo-nents that are based on industry standardsand/or open specifications.
The SCOPE Alliance is an association of net-work equipment providers that aim to enableand encourage an ecosystem of suppliers ofsuch hardware and software components inorder to promote high-quality telecommuni-cation platforms, reduce development time andcosts, and satisfy the needs of their customers.When different suppliers provide differentcomponents of a telecommunication platform,those components must be compatible andmust interoperate. Furthermore, the compo-nents must be interchangeable so that the net-work equipment provider is not beholden toany one supplier or group of suppliers.
To promote interoperability and interchange-ability of such components, various standardsorganizations have developed specificationsfor those components. Those standards bodies,and the areas they address, include: PCI In-
dustrial Computer Manufacturers Group(PICMG) for physical hardware; Service Avail-ability Forum (SA Forum) for high availabilitymiddleware interfaces and Open Software De-velopment Laboratory (OSDL) for carriergrade Linux. However, those standards areoften appropriate and useful for applicationsother than telecommunications, and they havegrown to encompass many specifications, re-quirements and features that are not appropri-ate or necessary for carrier grade platforms. Asa result, the potential suppliers of the carriergrade platform components do not knowwhere to focus their efforts.
The SCOPE Alliance is an association of net-work equipment providers that review industrystandards and open specifications to determinetheir relevance and applicability to telecom-munication systems. The objective of SCOPE isto identify subsets of existing standards andspecifications (called profiles) that are relevantto the carrier grade base platforms of thetelecommunication industry, as well as to iden-tify gaps in those standards and specifications.The SCOPE Alliance is organized as a programof the IEEE Industry Standards and Technolo-gy Organization (IEEE-ISTO). The sponsormembers of SCOPE are major network equip-ment providers, namely: Alcatel-Lucent; Erics-son; Huawei Technologies; Motorola; NECEurope; Nokia and Siemens Networks.
In addition, the SCOPE Alliance currently has12 contributor members, which are componentsuppliers. The member companies of theSCOPE Alliance contribute experts that reviewthe standards and generate a profile for each ofthe standards and identify gaps in the standards.SCOPE does not itself define standards, and itdoes not provide specifications to fill the gapsthat it has found in existing standards.
The SCOPE reference architecture, shown infigure 1, is divided into the following mainareas: 1) operations and maintenance, whichenable the user to operate and control the un-derlying platform and its applications; 2) toolswhich facilitate the design of complex systems,
TELECOM & NETWORKING
The SCOPE Alliance imposes profiles on standards
Figure 2. The carrier grade base platform ofthe SCOPE reference architecture
TELECOM & NETWORKING
specification of configuration files, interactionsbetween the user and tools, etc. 3) applicationservices which include database managementsystems, storage management systems, appli-cation execution environments, etc. 4) carriergrade base platform which is the core of theSCOPE reference architecture.
The SCOPE Alliance is focused on the controland service plane, to the left of the blackdashed line in figure 1, which is vital for inter-operability between components, rather thanon the user plane, where differentiation exists.The initial focus of SCOPE is on the carriergrade base platform and tools, shown withinthe red dashed line in figure 1. The carrier gradebase platform, shown in more detail in figure 2,includes: hardware, such as blades, shelves andcabinets; operating systems, particularly carri-er grade Linux as well as middleware, particu-larly high availability middleware. The carriergrade base platform also includes specializedprocessing engines, such as digital signal proces-sors, network processing units, fabric routingengines, etc.
Standards and open specifications for variousaspects of the carrier grade base platform al-ready exist, three of which are shown in red in
figure 2. The SCOPE Alliance seeks to selectsubsets of existing standards (called standardsprofiles) that are relevant to telecommunicationsystems. SCOPE also aims to identify gaps inthose standards as they apply to telecommuni-cation systems. Few industry standards fortools for the development, testing and deploy-ment of telecommunication systems currentlyexist, even though such tools standards couldreduce the difficulties of combining togethercomponents from different vendors into a car-rier grade system. SCOPE is encouraging thedevelopment of tools standards, and plans todescribe requirements (called content pro-files) related to functions and properties of thetools. The tools standards in which SCOPE isinterested include: 1) UML model-driven architecture of the object management groupfor design of complex systems; 2) XML formatsspecified by the world wide web consortium forspecification of configuration files; 3) Eclipseguidelines for interactions between tools andtheir users. 4) Unix tools for system develop-ment. The SCOPE Alliance has already issuedprofiles for the PICMG Advanced telecommu-nication computing architecture (ATCA) standard and for the OSDL carrier grade Linux(CGL) standard. High-level descriptions ofthose profiles are included below. The
descriptions below are necessarily abbreviatedand must not be regarded as definitive. For acomplete and authoritative definition of theSCOPE Profiles, see the SCOPE web site(http://www.scope-alliance.org).
The SCOPE profile defines the aspects of a par-ticular standard that are required for telecom-munication systems, and the aspects that areoptional. SCOPE distinguishes between twokinds of profiles: 1) standards profile, which isa subset of an existing specification that is ap-propriate for carrier grade systems; 2) contentprofile, which is a description of requirementsrelated to functions and properties related to aparticular aspect of carrier grade systems. ASCOPE profile can also identify gaps in the particular standard. The SCOPE Alliance hasdeveloped a first version of the profile for thePICMG 3.0 hardware standard, which is alsoknown as the AdvancedTCA or ATCA standardfor telecommunication systems. The SCOPEprofile for ATCA addresses eight PICMG specifications, including the ATCA base speci-fication, Ethernet/Fiber Channel, AdvancedMezzanine Card (AMC), PCI Express and Advanced Switching, Gigabit and 10 GigabitEthernet, AMC Storage, Intelligent PlatformManagement Interface (IPMI) and Intelligent
Platform Management Bus (IPMB). For inter-connections and cabling, some of the require-ments of the ATCA profile are GE dual starAMC interconnections and GE or PCI ExpressAMC fabric interfaces, Ethernet fabric inter-faces 1/2/4 with aggregation or 10Gbit/s perslot, redundant radial or bussed IPMB, manda-tory front cable access and rear access when arear transition module is used.
For shelves and power, the ATCA profile requires 14 or 16 slots per shelf, maximum200 W power dissipation per blade, shelf mustbe able to power a full population of bladesconsuming 200 W, redundant power feed of48 VDC with -60 VDC option, and extensiverequirements on air flow, cooling and noise. For
management and serviceability, the ATCA pro-file requires a redundant shelf manager withmodular software, redundant IPMB and updateinterface, hot swap of blades and in-service re-placement of fans and filters. The ATCA profilealso requires compliance with certain regulatorydirectives. Five of the 25 requirements of theSCOPE ATCA Profile are shown in table 1.
The SCOPE Alliance has also developed profiles(v1.0 & v1.1) for the OSDL carrier grade Linux(CGL) 3.2 operating system specification. SCOPEhas prioritized the requirements of that specifi-cation based on the needs of the telecommuni-cation market for service and control applications.In developing the Linux profile, SCOPE consid-ered support for an open blade environment such
as that defined by PICMG (ATCA). The SCOPEprofile for Linux characterizes the require-ments of the CGL 3.2 specification as manda-tory, desirable, already available, not needed anddid not prioritize. Of the 247 requirements ofCGL 3.2, 108 are determined to be mandatory,24 desirable, 78 not needed, 27 already available,and 10 not prioritized. Moreover, several of therequirements are determined to be incomplete,imprecise, or subject to multiple interpretations.
The SCOPE Linux profile addresses the sevenareas addressed by CGL 3.2, namely, availabil-ity, clustering, hardware, performance, securi-ty, serviceability and standards, as outlinedbelow. Availability refers to such aspects of car-rier grade Linux as robust mutexes, NFS clientprotection across server failures, parallel userinitialization during startup, iSCSI error han-dling support, Ethernet link bonding andwatchdog timers. Clustering encompasses thecluster availability framework, communica-tion service, software upgrades, checkpointservice, diagnostics, event service, failure han-dling, lock service, monitoring, membershipservice, storage systems and file systems. Hard-ware addresses configuration support, hotswap, shelf management, iSCSI, IPMI, disklesssystems and remote boot.
Performance encompasses low latency sched-uling, round robin scheduling, POSIX real-timefeatures, hyper-threading support, and variousrequirements related to timers, threads, mu-texes, priorities and interrupts. Security in-cludes various security mechanisms such as adynamic kernel security module, process con-tainment, buffer overflow protection, accesscontrol, authentication, password integritychecking, log analyses, IPsec for IPv4 and IPv6,PKI and SSL/TLS support for applications,memory limits, and file system, process and execution quotas. Serviceability includes suchtopics as panic, dump, debug, snapshots andprofiling at the kernel level and in user space,diagnostics related to temperature, fans, powermedia, networks, CPUs and memory monitor-ing, and software upgrade and rollback.
The standards area covers Linux standard basecompliance and POSIX compliance, as well assupport for CIM, WEBM, ACPI, PCI Express,iSCSI, MPI, NPF, SCTP, IPv6, IPSec, SNMP,IPMI and the various SA Forum specifications.For each of these seven areas, an example prioritized requirement for the SCOPE LinuxProfile is shown in table 2. The SCOPE Allianceis currently developing a profile for the Appli-
June 2007 30
TELECOM & NETWORKING
Table 1. Some of the requirements of the SCOPE ATCA profile
Table 2. A few of the requirements of the SCOPE Linux profile
FREE Subscription to boards & solutions magazineEnsure getting your personal copy of boards & solutions magazine free of charge by completing the online form at:
www.embedded-control-europe.com/bs_magazine
cation Interface Specification (AIS) high availability middleware interface specificationfrom the SA Forum. The AIS includes an avail-ability management framework and a numberof services, namely the cluster membership, notification, information management model,event, log, checkpoint, message and lock serv-ices. It also defines several redundancy models,including no redundancy, N+M, 2N, N-way andN-way active, as well as other features and options related to high availability. SCOPE iscurrently considering which of these aspects ofthe AIS are most appropriate and useful for carrier grade systems, and thus for inclusion inthe SCOPE AIS profile.
SCOPE is also working on content profiles fortools that are used to develop, test and deploycarrier grade telecommunication systems.SCOPE expects to describe requirements relat-ed to functions and properties of such tools.They are scoping various standards for toolssuch as the UML-based model-driven archi-tecture for designing complex systems, XML forspecifying configuration files, the Eclipse guide-lines for interactions between tools and theirusers, and various Unix system developmenttools. Such content profiles for tools can reducethe difficulties of integrating software compo-nents from different suppliers into carriergrade systems.
The SCOPE Alliance has already developed pro-files for the PICMG 3.0 advanced telecommu-nication computing architecture 3.0specification and for the OSDL carrier gradeLinux 3.2 specification. SCOPE is currently de-veloping a profile for the SA Forum applicationinterface specification, as well as profiles fortools that enable the development, testing anddeployment of carrier grade systems. TheSCOPE Alliance encourages contributions to,and comments on, the profiles that it develops,and encourages companies to build and supplycomponents for carrier grade telecommunica-tion systems based on the SCOPE profiles andthe corresponding standards. �
31 June 2007
TELECOM & NETWORKING
ChicagoJune 18-21, 2007McCormick Place
Booth #2875
Come see us at:
US Toll Free: 1-800-356-9602 Voice: +1-608-831-5500www.emersonembeddedcomputing.com
Expansion capability and wide control makes KAT4000 the right solution
KAT4000, an intelligent Field Replaceable Unit (FRU), is designed as a single slot Advanced Telecom Computing Architecture (AdvancedTCA®, ATCA™)carrier made to:
Order – Up to four Advanced Mezzanine Cards (AMC) expansion modules; AMC connections include: Up to two GbE to ports 0 and 1 in Common Options Region and Ports 4–7 to Fat Pipes Switch Work – Flexible Modular Fat Pipe Switch (FPS) module, GbE and 10GigE/1GbE Span – Ethernet switch for the AMC Common Options Region Last – Quality assured by more than 30 years of design experience and a TL-9000 and ISO 9001:2000 certified quality management system (FM 26789)Extend – Optional features include: Freescale® MPC8548 PowerQUICC III™
running at up to 1.3GHz; up to 2GB DDR2 memory; and up to 1GB of NAND Flash
Set your expectations high. Choose the global technology leader. To learn more about KAT4000 and why it’s the best solution for you, visit our Website.
Emer
son.
Con
side
r It S
olve
d is
a tr
adem
ark
and
Busi
ness
-C
ritic
al C
onti
nuit
y, E
mer
son
Net
wor
k Po
wer
and
the
Emer
son
Net
wor
k Po
wer
logo
are
trad
emar
ks a
nd s
ervi
ce
mar
ks o
f Em
erso
n El
ectr
ic C
o. ©
2007
Em
erso
n El
ectr
ic C
o.
AC Power Systems
Connectivity
DC Power Systems
Embedded Computing
Embedded Power
Integrated Cabinet Solutions
Outside Plant
Power Switching & Controls
Precision Cooling
Services
Site Monitoring
Surge & Signal Protection
Just another reason why Emerson Network Power is the global leader in enabling Business-Critical Continuity ™.
Product News� ELMA: MicroTCA shelf for form factor
and card testingELMA announced blu!box, an immediatelyoperative MicroTCA shelf for form factor andcard testing. The 19’ subrack has a height of5 U and a depth of 200 mm. The tested sys-tem platform is fully redundant due to 2MCHs, two power modules for 48 V / 60 Vand two cooling units. It uses high-perfor-mance fans the speed of which is PWM con-trolled. For this purpose, and for the exactdetection of the thermal situation in the testsetup, up to 20 temperature sensors may beplaced in the shelf.
News ID 228
� PICMG: AMC.2 specification adds Ethernet fabric
The PICMG consortium announces that ithas released a subsidiary specification for theAdvanced Mezzanine Card. PICMG AMC.2defines the implementation of 1 and 10Gbps Ethernet links on AMC.0 modules andcarriers. The intended implementation prac-tice will normally include at least one link ofEthernet.
News ID 87
� Schroff promotes AMC modules, fillersand carriers
Due to new technologies and optimisedmanufacturing processes Schroff is able to re-duce prices for AdvancedMC modules andfillers. Three different AdvancedMC modulewidths, 3 HP, 4 HP and 6 HP at two differentheights each, are available. The relativelynew 4 HP size is classified as Mid-size and en-ables for instance larger assembly heights withMicroTCA power modules and the use ofsmall cooling components.
News ID 149
IP network servers and carrier graderack-mount servers
TELECOM & NETWORKING
� We are seeing a new class of servers that arebridging the gap between enterprise rack-mount servers and rugged carrier grade prod-ucts such as AdvancedTCA and CompactPCI.These products are IP network rack-mountservers and carrier grade rack mount servers.
What makes them special? They have specialfeatures such as rugged mechanical design,available DC power supplies and shorter 20inchform factor for telco standard two-pole rack systems. Also these products are available on extended, 3 to 5 year, lifecycle for long term deployment in the infrastructure. These serverssolve a number of telecom or IP networkingproblems. Each server is designed to meet thespecific needs of telecom equipment manufac-turers and independent software vendors thatsupply fixed and mobile operators. The serveris designed for telecommunications centraloffices or network data centres and deployed ina rack-mounted, space-saving configuration.The server is modular and based on industrystandards, rather than proprietary in-house de-signs, saving telecom equipment manufactur-ers and service providers time and money,improving return on investment and protectionof software continuity. Long life support satis-fies the requirement for lengthy developmentand testing cycles followed by longer deploy-ment windows on stable platforms. The IP net-work servers are application-optimized
platforms in a network appliance type chassiswith high I/O throughput capabilities andoften the option for front or rear network interface connections. This class of servers areoptimized for IP network applications. They arewell suited to a range of uses in the telecom andother market segments. For the telecom marketthey can be a lower-cost alternative where arugged design and high reliability is needed butthere is no requirement for Network Equip-ment-Building System compliance. Many datacentre managers are looking at server consoli-dation and lowering costs associated with heatdissipation and power conversion. Theseservers, equipped to run off DC power and the latest multi-coreprocessor technology,may be the ideal solution.
By choosing Intelprocessor-based-carriergrade servers serviceproviders can quicklydeploy NEBS certifiedstandards-based prod-ucts to reduce time tomarket and be safe inthe knowledge that theyare purchasing high per-formance products witha dependable and reli-able extended life cycle
roadmap. CG-RMS are server products de-signed specifically for use in telecom infra-structure networks where the service providermust conform to country requirements for highavailability and reliability, including the abilityto survive dry or wet harsh winter environ-ments, or in hot, dusty, high humidity harshsummer environment, with an air conditioningfailure for up to four days. And continuing tooperate without error even if neighbouringequipment is electrically unstable in terms ofemissions. It must also be electrically veryquiet and well behaved and survive if neigh-bouring equipment catches fire and burns. Itshould to function during an earthquake,
By Jeff Hewlett, Intel
This article presents a newclass of servers that bridge thegap between enterprise rack-
mount servers and ruggedcarrier grade products such asAdvancedTCA and Compact-
PCI.
June 2007 32
Comparison of different servers
33 June 2007
TELECOM & NETWORKING
unless it is crushed. It must not fail in such away that might cause injury to anyone andwon’t damage anything even if it catches fire. Allthese requirements must be independentlytested and verified.
Finally form, fit, and function which must beeffectively, reliably, and cost-consciously pro-vided can be considered. This is a legal re-quirement and is known as NEBS in the USAand ETSI (European Telecommunication Stan-dards Institute) in Europe.
What are the main features of CG-RMS thatmake them suitable for use in telecom infra-structure networks? Carrier grade servers arecertified for compliance with NEBS/ETSI stan-dards. NEBS stands for “Network Equipment-Building System”. ETSI stands for “Europeantelecommunication standards institute”. Theyare terms commonly used to refer to a family ofrequirements documents that apply to telecom-munications equipment located in a telecomcentral office .This is a very tough and expen-sive test and includes harsh environments, se-vere vibrations and fire. The server is expectedto keep running except in the event of fire. Re-liability of the telephone system is considered anational security issue, is demanded by con-sumers and makes good business sense. There-fore, NEBS/ETSI testing is taken very seriouslyby governments and service providers as well asmanufacturers developing equipment used forthe telecommunications network. Carrier gradeservers have a shallow depth 20″ form factor.Most central office facilities are equipped withtwo post rack systems which require that the in-stalled equipment should be a 20″ form factor.
A telco uses system alarms, a set of threeLEDs, to help monitor equipment and identi-fy the cause of physical system problems with-in the central office. There are a de-factostandard three levels of alarms: minor, major,and critical. The following is a summary ofsome of the hardware components monitored:� fan (failure, speed)� memory (single and multi bit errors, ECC
errors)� processor (thermal trips and internal
errors)� temperature (baseboard and processor
temperature)� voltage (standby, baseboard, processors)� power supplies (presence, redundancy,
temperature)
Carrier grade servers can be configured with either DC or AC power supplies. In a telecomenvironment it is common to use DC power,most telephone systems operate from 48V DC.DC power is more efficient as it removes the re-quirement for AC/DC conversion thus reducingpower consumption and heat dissipation.
Carrier grade servers have longer commerciallife cycle than enterprise market IA processor-based servers, typically 3-5 years, sometimes thein-service life can be 15 years or more. Long lifecycles are needed because qualification proce-dures in a telecom environment are often 12-18months and once the expensive qualification iscompleted service providers need a stable plat-form to enable continued uninterrupted supplyand consequently lower cost. Telecom serversneed comprehensive remote management.Both runtime and pre-boot LAN-based man-agement is preferred and must be capable of“Headless” operation. Standard telecom serversutilize SNMP and management protocol and
typically integrate with HP OpenView or otherglobal framework management systems. Alsothey will incorporate CIM when appropriate.SNMP is simpler and uses UDP the most pop-ular protocol. TMN uses connection-orientedcommon management Information Protocol.TMN is a large, complete system which is moreexpensive to run but emerges where the focus ison reliability.
Integration is usually obtained via a Q adaptorfunction. The QAF translates between the q ref-erence point, which is implemented as an OSImanagement protocol stack, and the m refer-ence point, which is implemented as an inter-
25 Years of MIL-STD-1553 Service!EXPERIENCE
www.ddc-web.com
DDC Electronik GmbH (Germany) Tel: +49-89-15 00 12-11DDC U.K. LTD. (England) Tel: +44 (1635) 811140
DDC Electronique (France) Tel: +33-1-41-16-3424
34June 2007
By Wayne Cullen, Tundra Semiconductor
� EMIF64 is a proprietary interface developedby Texas Instruments, which has served the in-dustry well for many years. However it is nowbeing used in applications for which it wasnever intended, such as DSP to DSP links. TheEMIF standard represents a mature and stableparallel external memory interface, which hasproven useful for many applications. Howeverits capability is limited to master-only, which re-quires expensive CPU interrupt service routinesin order to transfer data into the device fromother masters in the system. A large softwareoverhead can also be required to support theEMIF interface, depending on the size and fre-quency of the data transfers. This is demon-strated in figure 1, which shows a traditionalEMIF application example, in the form oftransferring results from a CR ASIC to DSP bythe CPU Interrupt + EDMA Method.
By choosing a state-of-the-art series interfacesuch as Serial RapidIO, a number of general ad-vantages are gained: RapidIO supports 1.25, 2.5and 3.125Gbaud rates per link, and up to eight4x links or sixteen 1x links. It is deterministic andlow latency, and provides a non-blocking switch-ing fabric. RapidIO features configurable CPUInterrupt control and support for error man-agement, as well as support for congestion con-trol via performance monitoring statistics. It alsoprovides CRC handling for hardware error re-covery. The inclusion of hardware-terminatedendpoints results in a low software overhead. Ra-pidIO also requires only low-level configuration
and functional support, while offering a highlyabstracted message-passing API. It also has theadvantage that the CPU overhead is not de-pendent on the size of the data transferred, forexample with small control messages.
Figure 2 shows the same application as figure 1when implemented using Serial RapidIO. Thespecific benefits of taking this approach in pref-erence to EMIF64 can be summarised underthe following headings. The limitations ofEMIF64 include the fact that it is not an openstandard interface, and has its bandwidth lim-ited to 8Gb/s half duplex. It is also not a scala-ble solution. Serial RapidIO on the other handhas an open standard interface, scalable band-width up to 20Gb/s and scaleable architecture.
EMIF64 is a lossy system, which does not storeand forward nor does it offer data prioritiza-tion. There is also non-deterministic latencyacross a switch. Serial RapidIO is a lossless sys-tem with guaranteed packet delivery that has
four priority levels available.Deterministic latency across theswitch is featured. When anEMIF64 interface is used, FPGAdesign and verification re-sources are required. Theamount of test bench effort en-tailed should not be underesti-mated either, and finallycontinual product support is re-quired. With Serial RapidIOhowever, no silicon design effortis required, and the solution isless costly to implement due tothe higher relative IO require-ments of EMIF64. At the sametime PCB complexity is reduced- a single 64-bit EMIF interface
requires approx 97 pins, meaning that an eightport switch requires 776 interface pins alone -giving a consequent reduction in cost. Serial Ra-pidIO provides CRC handling for hardware-based error recovery, whereas there is no errordetection/correction available with EMIF64.The latter also offers no status or acknowl-edgement feedback, while Serial RapidIO pro-vides error management and reportingfunctions. Furthermore the wide parallel in-terfaces consume more PCB space than the se-
TELECOM & NETWORKING
Updating legacy interfaces – Serial RapidIO compared to EMIF64 switches
net management protocol stack. The mostcritical task of the QAF is to translate betweenTMN’s information model, which uses the OSIguidelines for the definition of managed objects, and the internet’s structure of man-agement information. TMN/SNMP Q-Adaptorenables seamless integration of legacy SNMP elements into the TMN environment, so that aTMN manager using CMIP can manage re-sources handled by SNMP agents. In order tointegrate the heterogeneous management
functions efficiently, the distributed networkmanagement architecture such as TINA(Telecommunications Information NetworkingArchitecture) maybe another choice. An inte-grated network management architecture forTINA manager and TMN and SNMP agent,where the TINA-based EML (element man-agement layer) subnetwork management op-erations are mapped into the relatedmanagement operations of the TMN/GDMO-based or SNMP-based NE agent.
HP, IBM and Intel have a breadth of experiencein designing servers for telecom environmentsand are now on the third generation of CG-RMS high performance telecom platforms.These servers are designed to enable ecosystemsuppliers deliver compelling telecom solutions,based on a wide choice of operating systemsand middleware. All telecom platforms featurerobust server management, high availability,and an extended life cycle for superior customerinvestment protection. �
Figure 1. Block diagram showing the transfer of results from aCR ASIC to DSP by the CPU Interrupt + EDMA method
35 June 2007
TELECOM & NETWORKING
rial alternative. One area where the two solu-tions are more or less equal is that of power re-quirements. The endpoint power isapproximately the same when configured withequivalent bandwidth settings. EMIF has a half-duplex bandwidth of 8Gb/s when operating in64-bit mode at 133MHz. Serial RapidIO has afull-duplex bandwidth of 4Gb/s when in x4mode at 1.25Gb/s. Switch power is also ap-proximately the same, although this does de-pend on how the FPGA is implemented, andwhich features are included.
Figure 3 shows a block diagram of the TundraTsi578 Serial RapidIO switch, which is a 80Gb/sfull-duplex Serial RapidIO switch that is openstandard compliant and also compliant to theSerial RapidIO interconnect specification V1.3,the most recent version. A highly scalable so-lution for mesh, fabric and aggregated systems,the Tsi578 provides designers and architectswith configuration options to match the preciseI/O bandwidth needs of a wide range of net-working, wireless and video infrastructure ap-plications. It is configurable with up to eight 4xlinks or up to sixteen 1x links, and each 4x linkis decomposable into two 1x links. The switchsupports 1.25, 2.5 and 3.125Gbaud rates, witheach port configurable to 1.25, 2.5 or 3.125Gb/s.The ports are completely independent and theswitch supports mixed speed and width con-figurations.
Ease of use features include ‘hot swap’ - live insertion and extraction of field replaceableunits. In terms of general performance, theswitch features low latency due to packet cut-through, and offers full duplex operation withline rate termination and a non-blockingswitch fabric, and prevents head-of-line block-
ing. It also features integrated programmableXAUI SerDes. The TSi578 is implemented in0.13µm CMOS technology and provided in a27mm x 27mm, 675-ball FCBGA package. Thethird-generation Tsi578 switch uses innovativeswitching fabric management to increase thedata throughput of next-generation communi-cation infrastructure platforms, including ATCAand MicroTCA applications. These platformsbenefit in particular from the ability of theswitch to route packets to over 64,000 endpoints,independent unicast and multicast routingmechanisms, and error management extensions.
Multicast routing support enables simultaneousinterconnection of Serial RapidIO-enabledprocessors and peripheral devices at an aggre-gate bandwidth of 80Gb/s. In addition the ex-tensive non-blocking switching fabricmanagement features include fabric monitor-ing to supervise and manage traffic flow, errormanagement to provide the fabric controllerwith proactive issue notification, a program-mable buffer depth for guaranteed bandwidthand independent unicast and multicast routingmechanisms. Traffic shaping is enabled bygood visibility of throughput, enhanced per-formance monitoring statistics, and advancedscheduling algorithms. The port flexibility ofthe Tsi578, combined with the use of very low-power, high-speed SerDes means that powerconsumption can be readily optimised.
To help simplify signal path routing, the switchalso supports I/O lane swapping.The device re-quires 1.2V and 3.3V power rails and is ratedfor industrial and commercial temperatureoperation. The switch also supports the ACJ-TAG version of the IEEE 1149.6 JTAG standardfor PCB testing of high-speed interconnects. �
Figure 2. Block diagram showing the transfer of results from CR ASIC to DSP by Direct IO usingRapidIO
Email: [email protected]
Infoline: 0 81 53 / 93 08 - 333
Modular – das passt!
Tiny Box TB5200 Mini-Industrie-Terminal mit Linux
• Sehr kleine Abmessungen 125x105x30mm³
• Robustes Aluminiumgehäuse• Leistungsstarker 32-bit Embedded
PowerPC Prozessor mit 400MHz• Integrierter Grafi k-Controller bis SXGA,
Parallelnutzung zweier Displays mit unterschiedlichen Inhalten möglich
• Erweiterter Temperaturbereich • Passive Kühlung
• CAN, RS232, USB, Ethernet, CRT, TFT,
Touch, Backlight, CompactFlash
„Ready-to-use“ Komplettgerät mit geballter Leistung auf kleinstem Raum.
36June 2007
Figure 1. Internet protocol stack for modem communication
Using the internet for M2Mapplications takes much of thepain out of the process of col-
lecting and communicatingdata. To succeed, M2M de-vices depend on minimizing
the time and level ofengineering resources re-
quired.
By Alan Singer, Connect One
� The internet is a loosely organized, interna-tional collaboration of autonomous, intercon-nected networks that use a range of internetprotocols for host-to-host, peer-to-peer, andpeer-to-host communication. Internet con-nectivity requires conformity with internetstandards by service providers who furnish ac-cess to the internet, hardware manufacturersand software developers. Because the internet isbased on open standards (as opposed to pro-prietary protocols), it offers the opportunity ofa ubiquitous, low-cost medium for communi-cation. TCP/IP is the basic communication pro-tocol that enables any company to develop aproduct that can interoperate and “talk” withproducts developed by other companies usingInternet protocols. It has turned the Internetinto the global communication backbone andhas become the ideal medium for networkingM2M applications.
Machine-to-machine (M2M) communicationconnects non-PC devices to the internet andother networks based on the internet Protocol.M2M is the fundamental technology of Post-PC, third-generation computing technology. Itprovides remote access over the internet to non-PC devices such as point-of-sale terminals, fleetmanagement or automatic vehicle locationterminals, home appliances, data loggers,telemedicine devices, industrial controllers,utility meters, building controls, vending ma-chines, SCADA devices, and much more. Theseand many other devices—powered by embed-ded microcontrollers—number in the billions
and are deployed throughout the world. Ofcourse, not every device needs to communicate.But where there is a real business need, such asimproving productivity or lowering the cost ofservice, there’s justification to seek the most ef-ficient or cost-effective means of communica-tion. Using the internet for M2M applicationstakes much of the pain out of the process of col-lecting and communicating data. Fleet man-agement systems track the inventory, route,tolls, location, mileage, engine RPMs, gas con-sumption, brake wear, tyre pressure, stops,hours driven, etc. The process is done auto-matically, freeing the driver from extensivepaper work and the company from hours oftabulation, review and reconciliation of records.Twenty years ago, trucking and taxi fleets usedradio frequencies in the VHF or UHF bands forcommunications—a relatively expensive andanalog system. Today, most fleet managementterminals use IP-based cellular digital net-works such as GPRS and CDMA2000 for com-munication, which are much easier to integratedata into an enterprise’s IT system.
Device communication needs to be rock solidover a medium that is dynamic and unpre-dictable. The internet is a rapidly evolving andoften unstable network environment. Methodsof connectivity and communication and secu-rity protocols all change as technology ad-vances. As new protocols are introduced, adevice’s communication application needs to beupdateable since most industrial devices have along in-field product life. Remote updateabili-
ty of the communication tasks is very impor-tant since many devices are located in hard-to-access locations or are mobile. The internetoffers the possibility to remotely update the de-vice application in an unattended manner.Deployed devices can benefit from being retro-fitted with IP connectivity. Elevators or vendingmachines installed more than 20 years ago, forinstance, were not designed with any expecta-tion of connectivity. To bring them online re-quires a solution that works with older,stand-alone technology. And new devices needto be designed so that their application can be remotely updated and will not need retrofittingas the communication protocols advance.
Obviously, enabling an existing or new designto include IP connectivity requires that the so-lution fits within limited resources. To succeed,M2M devices depend on minimizing the timeand level of engineering resources required. Thedeveloper has tight fiscal budgets to meet. And,of course, the device, particularly older, stand-alone technologies, has limited computing re-sources (both processing power and memory).As in the desktop world, connectivity comes ina variety of formats. Similarly, depending on thetechnology available in the specific region,connectivity might be achieved by dial-up,wireless modem, 802.11a/b/g wireless LAN, or10 and 10/100BaseT Ethernet LAN. A globalnetwork of devices could quite conceivably in-volve a mixture of technologies based on thetechnologies prevalent in specific areas of theworld where the various devices are located.
INDUSTRIAL COMPUTING
Machine-to-machine communicationover IP Networks
37 June 2007
INDUSTRIAL COMPUTING
Software libraries and shareware have long beenavailable for developers who wish to add theTCP/IP protocol stack to their application as a“do-it-yourself” solution. Processor manufac-turers today are bundling TCP/IP stacks withtheir silicon. Wireless modem manufacturersalso bundle support for TCP/IP in theirmodems. All these solutions require that youhave some internet programming expertise. Onthe other hand, IP Controllers and deviceservers enable customers to use their currentapplication and hardware with minimum or noredesign and reprogramming. IP Controllers dothis cost-effectively, efficiently and offer maximum flexibility.
The internet offers many methods to send andreceive data, images, audio, video, and otherfiles. The Open Systems Interconnection Ref-erence Model classifies communication proto-cols, applications and physical media intoseven layers, built one on top of the other. Themodel helps in understanding networks and indeveloping products that can communicatewith each other.
It should be noted that “TCP/IP” technicallyrefers to the lowest layers of the internet proto-col stack. Above the physical layer used by thecommunication device and the data link layerused to connect the physical device is the net-work layer of the TCP/IP stack. At this layer wehave the internet protocol stack, a “connection-less” protocol that does not provide a continu-ing connection between the end points that arecommunicating. Therefore it is an unreliablemethod of delivering data from one device toanother. Here data is packaged in a unit called apacket, which contains complete addressing in-formation. Each sender and receiver of infor-mation (known as a host) on the internet has anaddress known as an “IP address” that unique-ly identifies it from all other hosts on the inter-net. Each packet has an IP header that containsthese addresses for identifying the sending andreceiving hosts. Above this is the transport layer,where one can open TCP and UDP sockets be-tween two points. On the presentation layer arethe protocols that define ways of sending and re-ceiving data according to the IETF (Internet En-gineering Task Force) standards known asRFCs. These standards include the FTP filetransfer protocol, e-mail protocols such asSMTP and POP3, web protocols such as HTTP,and command protocols such as TELNET.
Let’s take a look at the five main alternatives forInternet-enabling a device:� buy or develop a TCP/IP software stack, or
download shareware from the Internet andintegrate the protocol stack into the host ap-plication;
� buy a microcontroller that includes a bun-dled TCP/IP stack and runs the application;
� buy a wireless modem with an embeddedTCP/IP stack;
� buy an IP controller chip that offloads the internet tasks from the MCU;
� buy a device server to IP-enable a deployed device.
Most likely, an M2M device developer will wantto add the IP stack to the application, especial-ly if the application uses a 32-bit processor. Thisoption requires the developer’s organization tohave or to subcontract embedded internetprogramming expertise in order to avoid timeand cost overrun and to ensure reliability inevery network. If the user has some control overthe network, if connections are predictable, and
if only one or a few TCP or UDP sockets areused, this is a relatively easy task. If, however, thedevice will be used on multiple networkswhere connectivity is unpredictable and notcontrollable, then the task is not so simple. Anembedded design must be ruggedized to ac-count for many different implementations ofthe internet protocol standards or RFCs (Re-quest for Comments) on servers located in dis-parate networks, ISPs, wireless operators andtelecoms operators. On an embedded hardwarelevel, if the design does not use a 32-bit proces-sor, the host CPU may be underpowered to runthe low-level TCP/IP protocols and upperlayer internet protocols, security protocols (if
June 2007
June 2007 38
INDUSTRIAL COMPUTING
needed), and the application simultaneously ifthere is a lot of data to send or if the processoris slow. It may be necessary to add additionalmemory to store the internet protocols and addmore buffers if the device does not have ade-quate memory resources and if more than onesocket will be used.
Taking these steps may require a total hardwareredesign of the product that needs to be IP- enabled. It is not recommended to mingle theapplication with the internet connectivity tasks,because the dynamic nature of the internet requires frequent updating of the internet protocols and configuration parameters, whilethe application tends to be stable and does notoften require updating. In any case, a method-ology must be implemented for safe remote updating of the Internet connectivity applica-tion. Using an off-the-shelf IP software stack ordeveloping one’s own IP stack may require a lotof integration, customization, modification,testing, and fine-tuning. This is especiallytricky if upper layer internet protocols or internet security protocols are required. Intaking this route, developers must maintain theIP stack or have a maintenance contract withthe software vendor. There is no guarantee thata stack that works in the lab will function reliably in every network around the world. Ifinternet connectivity is not the developers’ expertise, this is a very risky undertaking. Buy-ing a software stack is the least risky of thesechoices, but developers again need to maintainthe protocols to provide the fastest support totheir customers.
Several manufacturers offer microcontrollersthat include internet protocols and are designedto run the application simultaneously. Bundledsolutions may be adequate for some MCUs andinternet connectivity tasks, but mingling the in-ternet protocols and application may degradeCPU performance, depending on the micro-controller’s processing power and speed, theamount of data transmitted, the protocolsused, and if they are fully implemented. Becauseof limitations on internal chip memory andavailable processing power, on-chip internetprotocols are often scaled down. It’s importantto check that the customer’s desired protocolsare supported and are RFC-compliant. Since
most MCUs have limited internal memory, itmay be necessary to increase the target board’smemory. Finally, because an MCU vendor’s expertise is in silicon, not in internet commu-nication, the amount and quality of support forthe IP stack may be limited. All this adds up toadditional cost in terms of hardware and development time.
Most major manufacturers of wireless modemstoday offer modems with some implementationof the internet protocol stack. These modemsstill require developers to add internet com-mands and configuration parameters to theirapplications. Users must carefully evaluate thecapability of the modem to perform the re-quired IP communication tasks. Recent modelsof IP-enabled modems from some manufac-turers offer a variety of upper layer protocols,while low-end models offer limited or noTCP/IP functionality. Some wireless modemsenable developers to write their application inJava on the modem processor. This may be fineif the application is a new one, but if the appli-cation already exists, it may not make sense torewrite the application on the modem. Also, ifthe developer does not have internet program-ming expertise, this is a time-consuming andinefficient choice. Most wireless modems havea limited amount of memory and therefore in-clude a minimal implementation of the IP stackand may not be able to adequately buffer thedata if there is a lot of data to send. Most wire-less modems do not offer the possibility for amodem-attached device to be remotely accessedfrom outside the network. Because wirelessmodems operate on a private IP network op-erated by a wireless service provider, themodems receive a dynamic IP address that is as-signed internally by the wireless network’sgateway server. This dynamic IP address cannotbe accessed from outside the network becausethe IP address is not known outside the
What are my limiting factors?
There are a number of issues associated withchoosing an IP-enabling solution:
Hardware design time Material costCost for development toolsApplication modification or developmenttimeSoftware development or acquisition costLicensing fees or royalties (if applicable)Selecting the right protocol for your applicationInternet programming competenceMingling the internet communication taskswith the host applicationTime-to-marketProtocol maintenance
There also are a number of risks:
Technological risk: will the solution performreliably under most conditions?Marketing risk: will the solution come tomarket on time?Financial risk: will the solution come in onbudget?
Figure 2. Internet protocol stack for LAN and WiFi communication
Figure 3. iChipBlock diagram
INDUSTRIAL COMPUTING
network. Furthermore, access to the modem is restricted because mostnetworks have a firewall to keep out unauthorized users. This is a prob-lem that the user must deal with, since it is outside the realm of themodem manufacturer.
Just as there are controller chips that offload certain functions from a hostprocessor, such as VGA, DMA, and interrupt controllers, an IP controlleroffloads internet communication tasks from the host processor. The roleof the IP controller is to mediate the connection between the host and theinternet via the physical medium. An IP controller enables developers tomodify their existing design with minimal changes to the hardware andminimal or no change to the application. They can use the existingprocessor, memory, and application, and just add a few commands to setthe internet configuration parameters and to activate internet commu-nications.
An IP controller offloads internet connectivity tasks from a host proces-sor, enabling it to exclusively and efficiently run the device application.When offloading the internet connectivity, customer can use their cur-rent application, operating system, and remain focused on their area ofexpertise, which is the device itself. Offloading eliminates the possibili-ty of having to update the operating system, CPU, and memory in casethe customer wishes to add new internet functionality or protocol sup-port to the application in the future, which also eliminates the need fora major rewrite of the application. An IP Controller also offers the pos-sibility to remotely access devices that have dynamic IP addresses.
An alternative to buying an IP-enabled wireless modem is to use a stan-dard wireless modem with a wireless internet adapter, which provides theIP connectivity. In this case, the customer can choose the most suitablewireless modem from the many modem manufacturers. This solution ispractical when customers don’t want to change their current wirelessmodem or application, or if the functionality available on IP-enabledwireless modems does not meet their requirements. Since this solutionis more expensive than an IP-enabled wireless modem, the internetadapter must offer higher functionality. In the case of IP-enabling de-ployed devices via dial-up, wired or wireless LAN access, device serversexist for these different communication media. They generally include achoice of internet protocols and security protocols. An important con-sideration is to make sure that the device server can work with commu-nication devices from a wide range of manufacturers.
IP networks provide many benefits to users of M2M applications. Amongthe benefits are lower costs of infrastructure, operations, communicationsand support, and higher throughput, reliability, versatility, and security.Because the internet is a complex, dynamic, and inconsistently imple-mented medium that is constantly evolving, an IP connectivity solutionmust be as dynamic as the internet. It must be adaptable, simple to useand maintain, yet sophisticated enough to deal with the internet’s in-herent inconsistency. The challenge to embedded designers is to devel-op a solution cost-effectively and in a timely manner. Market researchindicates that M2M will be an active segment in the embedded market.According to the Wireless Data Research Group, between 2004 and 2008,the market for hardware, software, professional and wireless network serv-ices for machine-to-machine communications will have grown at a 27%compound annual growth rate, from $9.3 billion to $31 billion. TheGartner Group predicts that by 2007, there will be between 100-160 mil-lion machine-to-machine connections worldwide that use wireless mo-bile phone networks. 880 million new M2M-enabled devices will beproduced annually by 2010, according to the Focal Point Group. Giventhese predictions, embedded designers need to consider all options andissues involved in the selection of an IP-enabling M2M solution. As themarket is taking off quickly, the winners in today’s market will be thosecompanies that are first to widely deploy reliable IP solutions. �
#01PERFORMANCE
IMAGE ADVERTISEMENT
HIGH PERFORMANCE - ACTIVITY WHICH LEADS TO AN EXTRAORDINARY RESULT. MISSION SUPPORTED BY ABOVE AVERAGE DEDICATION, AMBITION AND MENTAL STRENGTH EXCEEDING THE GOAL.
// High performance within all disciplines
Plastic injection moulding
MPS
Backplanes
19" subrack
POLYRACK Electronic-Aufbausysteme GmbH Steinbeisstraße 4 // 75334 Straubenhardt // Germany Phone +49.(0)7082.7919.0 // Fax +49.(0)7082.7919.330www.polyrack.com
POLYRACK Tech-Group offers holistic and high performance solutions - especially custom requirements of Electronic packag-ing, Microcomputer packaging systems, Backplanes and Thermoplastic components.
Cases FreeTec
Windows Vista ushers in a new era in data acquisition
INDUSTRIAL COMPUTING
� Windows Vista is the new operating systemfrom Microsoft and came as Windows biggestlaunch after Windows XP. Following years ofanticipated release, Windows Vista rolled offearly this year and offered major improvementsin areas of performance, security, and visualpresentation. Windows Vista, codenamed“Longhorn” during its early development,comes as an opportunity for engineers and sci-entists to solve critical technical problems andachieve greater, faster data processing not pos-sible on earlier Windows releases. According toanalysts, Windows Vista is expected to beadopted more quickly and broadly than anyprevious Windows release, eventually creatinga big pool of potential customers who look forsoftware applications taking full advantage oftheir optimised systems.
Windows Vista offers several new featureswhich set it apart from other Windows releas-es. A major enhancement is Windows Vista’simprovement in PC security. Windows Vista in-corporates the User Account Control (UAC) —a new user-privilege model designed to preventvirus attacks and illegal software applicationsfrom halting or taking control of the system,initiating installation, or changing firewall set-tings. Launched in both 32-bit and 64-bit edi-tions, Windows Vista enables the user to choosethe edition depending on their computer hard-ware. For applications intended to run on
computers with more than 4GB of RAM, Win-dows Vista 64-bit edition supports memory-in-tensive applications including vision acquisitionand large-scale data processing that are not pos-sible on 32-bit operating systems. In WindowsXP, programming is done using the Win32 API.In Windows Vista, the .NET Framework 3.0(formerly known as WinFX) interface deliversa straightforward way to program applicationson Vista territory. Since programmers havebeen integrating their software applicationsusing .NET Framework-based code for severalyears now, they can conveniently integrate anddeploy these applications using Windows Vistawhile keeping their existing hardware.
Expecting that 32-bit applications will be exe-cuted on 64-bit versions of Windows operatingsystems, Microsoft included an abstractionlayer called WOW64 (Windows-on-Windows64-bit) on all 64-bit releases. WOW64 is a light-weight translation layer subsystem and featuresidentical interfaces on all 64-bit versions ofWindows. The primary purpose of WOW64 isto create a 32-bit environment that provides therequired interfaces for executing 32-bit Win-dows applications in a 64-bit system withoutmodifications. Since WOW64 is included on all64-bit versions of Windows, users may enjoy asmooth transition from a 32-bit environmentto Windows XP and Windows Server 2003 for64-bit extended systems. However, even with
WOW64, 32-bit programs executed on Win-dows Vista x64 Edition cannot take advantageof the larger 64-bit address spaces or wider64-bit registers on 64-bit processors.
On a 32-bit Windows system, the maximummemory available is 4GB which is shared be-tween the operating system, memory-residentdrivers, applications, and application data.Therefore, the maximum memory available tospecific applications running on a 32-bit OS is2GB or up to 3GB with a special operating sys-tem and compiling options. On a 64-bit system,the WOW64 infrastructure could allocate asmuch as 4GB of RAM to a 32-bit Windows ap-plication without addressing memory space forthe operating system and drivers. This capabil-ity extends a solid 4GB memory allocation forresource-hungry applications and posts a significant increase in computing efficiency.Adlink has released three 32-bit software libraries that run on the 64-bit edition ofWindows Vista. These include the PCIS-DASK,D2K-DASK, and WD-DASK.
As these packages conform with the WOW64interface, integration with a 64-bit OS comesautomatically and without any need for addi-tional user configuration. Software drivers thatrun in the kernel must also be 64-bit. Howev-er, users are faced with several conditions to obtain the full benefits of Windows Vista x64
By Sam Ku, Adlink Technology
According to analysts,Windows Vista is expected to
be adopted more quickly and broadly than any previous
Windows release. ThereforeAdlink is providing for
Windows Vista both 32-bitand 64-bit versions of kernel
drivers for its full range ofdata acquisition cards.
June 2007 40
Edition. Theoretically, an application runningon Windows Vista x64 Edition should improveperformance because of access to larger avail-able memory even when the application is run-ning via the WOW64 layer. It is important tomention that Windows Vista x64 Edition is notnecessarily faster than its 32-bit counterpart. Insome cases, x64 may even perform slower be-cause of the unrelated OS overhead of largerpointer. In actuality, 32-bit emulated applica-tions running through the WOW64 layer couldnot address more memory than they could in a32-bit system. The time before Windows Vistax64 Edition gains widespread adoption is alsoan open issue.
Adlink is committed to delivering completesoftware support for its line of DAQ productswith varying form factors that include PCI Ex-press, PCI, CompactPCI, and PXI. In general,Adlink not only provides drivers for main-stream OS such as Windows and Linux, but alsooffers support for third-party applicationssuch as LabVIEW and MATLAB. Adlink alsoprovides ActiveX control sets. For WindowsVista, Adlink is providing both 32-bit and 64-bit versions of kernel drivers for its full range ofDAQ cards. As earlier stated, PCIS-DASK,D2K-DASK, and WD-DASK also support Win-dows XP and Windows Server 2003 for 64-bit
extended systems, including AMD64 and Intelx86-64 system architectures. (PCIS-DASK isWindows kernel driver APIs for Adlink NuDAQPCI/cPCI Series. D2K-DASK is Windows ker-nel driver APIs for Adlink DAQ-2000 andPXI-2000 Series. WD-DASK is Windows kerneldriver APIs for Adlink PCI/PXI-9820 high-speed digitiser.) All DASK kernel drivers/li-braries provide a common set of ApplicationProgramming Interface allowing easy access toall the functions of Adlink products. In addi-tion, these advances in application developmentenable the users to perform detailed operationsand experience superior performance and reliability from their data acquisition systems.Microsoft requires a digital signature on alldrivers starting from Windows Vista x64 Edition.
By requiring signed drivers, Microsoft aims toreduce failures and improve stability by settingthe spotlight on vendor responsibility with re-gard to driver bugs and glitches. For this reason,all Adlink 64-bit DASK drivers are digitally-signed for Windows Vista x64 Edition. Usingthe WOW64 subsystem, Adlink has also verifiedon Windows Vista x64 Edition its completerange of ActiveX control sets including PCIS-OCX, D2K-OCX, and WD-OCX, as well asDAQBench — a 32-bit ActiveX control formeasurement and SCADA/HMI. �
41 June 2007
INDUSTRIAL COMPUTING
smartModule855the smallest CPUModule in its Class
_ Intel® Celeron M 600MHz up to Intel® Pentium® M 745 1.8GHz
_ i855GME 400MHz FSB, ICH4, 128-1024MB DDR-RAM
_ Extreme Graphic, 64MB, DirectX 9 compatible, CRT and DVO / LVDS
_ 6 x USB V2.0, LAN, AC97 SPDif 5.1 sound stereo with six-channel output and two channel input
_ SM855 Bus including PCI and LPC
_ Heat spreader and passive and active cooling feature
_ Operating temperature -25°C to +60°C (option: -40°C to +70°C)
DIGITAL-LOGIC offers a large variety of Embedded Computer in PC/104, EPIC, EBX, 3.5", smartModule and other form factors.
Further information:
www.digitallogic.com
Product News
� IBASE: IPC motherboard supports Core 2Duo processors
IBASE announces the MB886, an Intel 945GExpress chipset based industrial motherboardwith high-bandwidth interfaces such as dual-channel DDR2 and 1066/800MHz system bus.Supporting the latest Intel Core2 Duo proces-sors, the MB886 motherboard is ideal for use inmedical, industrial automation, gaming systemsand other demanding applications.
News ID 104
� ADLINK: 1U industrial computer chassis for network security appliances
ADLINK introduces the RK-136, a 1U rack-mount industrial chassis designed for use withhalf-size industrial cards and providing a 5.25"bay for a slim CD-ROM drive, a 3.5" bay for aslim FDD, and two internal 3.5" drive bays. Thecooling system incorporates three 4cm ballbearing fans-two at the front and one at therear-generating a combined ventilationthroughout of 21 CFM. For a convenience, twoLEDs are located on the front panel to indicatedisk activity and system power.
News ID 10
� Amplicon: 3U 19-inch rackmount industrial computers
Amplicon has enhanced the versatility of itsImpact range of industrial computers with thenew Impact-R 3010 series. Comprising 3U 19-inch rackmount industrial computers theseries is designed to provide an economical orspace advantage where 2U designs provide insufficient expansion requirements and 4Udesigns require more than the available rackspace.
News ID 10631
� AXIOMTEK: 15-inch industrial panel computer for POI and HMI
AXIOMTEK launches a 15-inch, fanless oper-ation, rugged-design panel PC featuring IntelCeleron M 1GHz/0KB processor. ThePANEL2155-810-FL1GE is an industrial gradepanel computer and has excellent graphic performance, DualView function, and multipleI/O. It is a cost-effective, full-function ideal forPoint of Information and Human Machine Interface applications.
News ID 10675
� Schroff : ATCA and TCA systems pass vibration and shock tests
As some users demand shock, vibration andearth quake tests the AdvancedTCA and Mi-croTCA systems from Schroff were tested in anindependent laboratory in Munich in accor-dance with the standards IEC60068-2-6 (vi-bration tests), IEC60068-2-27 (shock resistancetests) and IEC60068-2-57 (swing tests with timelaps processes) and passed
News ID 10679
� Artila: ARM9-based Linux-ready box computer
Artila Electronics announces the new ARM9-based, Linux-ready industrial box computer, theMatrix-520. The fanless ARM9 RISC CPU andstrong metal case design make Matrix-520ideal for the industrial applications needing areliable front-end controller. The Matrix-520,powered by a 180MHz ATMEL ARM9 RISCCPU, comes with onboard 32MB SDRAM and16MB Flash memory which has Linux 2.6kernel, busybox utility and boa web server per-installed
News ID 205
� E.E.P.D.: COM Express with Core 2 Duoprocessors
E.E.P.D. announces that the PROFIVE P65COM Express, the latest member of the COMExpress family, is now available. The P65 CPUmodule is powered by the latest Intel technol-ogy, including the Intel Core 2 Duo Processorsand the Intel GM965 Express chipset. The IntelGM965 Express Chipset integrates graphicfunctions to support analog VGA, serial DVOdisplay interface with up to 8 GB of memorysupport.
News ID 229
� ADLINK: 4-axis stepping and servo motioncontrol card
ADLINK announces the PCI-8154, a 4-axispulse train modularized motion control card forcontrolling steppers, servos, and linear motors.The PCI-8154 delivers high-frequency pulserates of up to 6.55 MHz and features a hardware-controlled emergency stop to terminate move-ment in case of system malfunction, a softwaresecurity protection to prevent illegal copying ofcustom software, and a card index switch to sup-port multiple PCI-8154 cards in a single system.
News ID 202
� Diamond: PC/104+ 8-channel serial communications module
Diamond Systems announces Emerald MM-8Plus, a 8-channel multi-protocol serial com-munications PC/104-Plus module that dependssolely on the PCI bus for its processor interface.The module offers improved performance overtraditional PC/104 (ISA bus only) solutions.
News ID 61
� VIA: Pico-ITX form factor reference designVIA Technologies announces the VIA VT6047Pico-ITX form factor reference design, a smallfull-featured x86 mainboard designed for ultracompact embedded PC systems and appli-ances.The VIA VT6047 Pico-ITX mainboardwas designed to be powered by one of VIA’s en-ergy efficient processor platforms, such as theVIA C7 or fanless VIA Eden.
News ID 143
� AAEON: PC/104 expansion cardAAEON announces a new PC/104 expansioncard PFM-T800. The PFM-T800 was designedto serve the customers who have the need forPC/104 interface expansion. PFM-T800 featuresa PC/104 form factor that satisfies market de-mands. In addition, the PFM-T800 has threemain functions of PCI to ISA, USB and COMport expansions.
News ID 227
� Interface Concept: 6U cPCI/VME full Giga Ethernet switches
Interface Concept has announced theComEth4020a, a line of 6U cPCI/VME full GigaEthernet Switches, being PICMG 2.16 and VITA31.1 compliant, for embedded applications.ComEth4020a is built on the latest generation ofGigabit switch engine and PHY transceiver.
News ID 68
� ACAL: ETX baseboard with 2.0 GHz Core Duo processor
Newly available from ACAL Microsystems is theAmpro XTX830, a plug compatible solutionETX baseboard which uses the Intel 2.0 GHzCore Duo processor (‘Yonah’), up to 2MB level2 cache and takes DDR2 SODIMM RAM up to2GB. The module also contains (6) USB 2.0ports, IDE as well as Serial ATA interfaces tocover both legacy and high-speed disk drives.
News ID 187
� Embest: ARM SBCs based on SamsungS3C2440A
Embest has released two Single Board Com-puters, both of which are built around theS3C2440A ARM920T microcontroller and arewell suited for portable and mobile products.The SBC2440-I has two Ethernet ports, sup-porting a variety of other onboard peripheralssuch as 1Mbyte Nor Flash, 64Mbyte NandFlash, 64Mbyte SDRAM, USB Host and USBDevice, RS232 and TTL, IDE interface, batterybacked RTC, LEDs, reset buttons.
News ID 178
� IBASE: mini barebone system for 3.5-inch SBCs
IBASE Technology announces the CSB100mini barebone system series supporting its 3.5-inch single board computer products that aredesigned for a variety of fanless and low powerapplications such as digital signage, POS andpanel PC. The CSB100-881 supports the IB8813.5-inch SBC featuring the Intel PentiumM orCeleron M processor.
News ID 10647
� MSC and ELTEC: real-time Linux for COMExpress modules
MSC and ELTEC have agreed to a close collab-oration to support real-time Linux /OSADL forCOM Express products. The COM Expressmodules from MSC will be equipped withELTEC’s real-time Linux extensions, which aremainly driven by the OSADL (Open Source Au-tomation Development Lab)-Consortium.MSC offers comprehensive security functionsfor their modular COM Express platformbased on the Intel Core Duo and Intel82945GM from system start to execution of theapplication code, which can also be used withLinux.
News ID 220
Rittal: compact cooling package for MicroTCA systems
With its new CCP Compact Cooling Package,Rittal provides increased security in the climatemanagement of MicroTCA systems - with asystem availability of at least 99.999%. Bythese means, slide-in electronic componentscan be cooled down completely by a powerfulair/water heat exchanger (including a fan unit)with a cooling output of up to 1,200 watts.
News ID 36
June 2007 42
PRODUCT NEWS
FREE Subscription to boards & solutions magazineEnsure getting your personal copy of boards & solutions magazine free of charge by completing the online form at:
www.embedded-control-europe.com/bs_magazine
� Schroff : ATCA, AMC and MicroTCAbrochure
Schroff has compiled a detailed, 48 page brochureabout the three new standards: AdvancedTCA,AdvancedMC and MicroTCA. Users who requiresystems with high transfer volumes and absolutereliability, for instance for telecom applications orcomplex automation tasks, will find all importantinformation at a glance.
News ID 82
� Curtiss-Wright standardizes power management features
Curtiss-Wright has launched an initiative tostandardize Power Management features on itsembedded COTS SBC and DSP engine products.These new features will provide system design-ers with sophisticated tools to control power dis-sipation and consumption and ease thetechnology insertion of next generation solutionsinto existing systems. The Power Managementfeatures are supported through a new Curtiss-Wright developed common software API and in-clude Power Disconnect, CPU power throttling, Peripheral Component power throttling, andPower Surge Protection.
News ID 137
� AAEON: medical stations with 17’ or 19’ TFT
AAEON introduces two new Medical Stations,the ONYX-175/195, which are based on theIntel Core 2 Duo processor which deliver a per-formance improvement of more than 100% ascompared to systems running traditional sin-gle-core processors. With dual core computingengines, ONYX can simultaneously execute twocomputing tasks.
News ID 232
� DSM: Core 2 Duo based NanoServerDSM Computer's newly developed 96M1558embedded system from the "E6" NanoServerfamily has been specially designed for high-endapplications. The industry-standard serverbased on the Intel Core 2 Duo (T7400) proces-sor has a 2.16 clock rate. The two cores can usethe 667 MHz front-side bus to access theshared system memory with a maximum ca-pacity of 4 GB. To allow the large data volumesto be transferred, if required, the NanoServerprovides either two PCI slots or, as an alterna-tive, a PCI and a PCI Express x1 expansion slotfor short cards.
News ID 130
� Kontron: Core 2 Duo based COM with optimized graphics features
Kontron announces its latest addition to thefamily of ETXexpresss Computer-On-Modules - the Kontron ETXexpress-MC. Builtaround the Intels Cores 2 Duo processor (up to2.2GHz) and Intel'ss 965GM Express chipset theKontron ETXexpress-MC will offer optimizedgraphics features and will be the most robustand power-efficient COM Express solution.
News ID 225
� GE Fanuc: 6U CompactPCI SBCs with Core2Duo processor
GE Fanuc has announced the CT11 , CP11 andCR11 6U CompactPCI CPU boards with inte-grated dual core Intel processing technology:versions are available with the Core2 Duoprocessor, the CoreDuo processor and theCeleron M processor. The new family supportsboth a 32- and 64-bit processor architecture,and is designed to meet the exacting require-ments of commercial, telecommunications andmilitary/aerospace organizations, with specificapplicability to simulation/training andtest/measurement.
News ID 75
43 June 2007
PRODUCT NEWS
Industrial Computers & SystemsVia Calamandrei, 91 52100 AREZZO - ITALYPhone: +39 0575 26979 Fax: +39 0575 [email protected] www.seco.it
SECO and M706R: small and fine
M706RDiabolically Perfect!
• Intel® CM & PM• Chipset Intel® 855GME• 6 USB 2.0 • Ethernet 10/100 and 10/100/1000 • DVI and VGA • 4 Serials • Parallel • Compact Flash Slot • IDE• Audio• PS/2 Mouse & Keyboard • PC/104 and PC/104+• ATX Connector
� Kontron: 19-inch/4U industrial server withfour Xeon cores
Kontron announces a 19-inch/4U IndustrialSilent Server, equipped with two cost-saving, en-ergy efficient Dual-Core Intel Xeon processors(5000 series / LGA711 socket), a redundantpower supply unit, a hot swappable RAID 0/1/5subsystem and an optional IPMI system hardwaremanagement module.
News ID 138
� ADLINK: 14-slot 3U portable PXI chassiswith 15" touch panel
ADLINK releases the PXIS-2690P, a 14-Slot,3U portable PXI chassis fully compliant withthe PXI specification, rev. 2.2. The PXIS-2690P is designed for military, aero-defense,field testing, and in-vehicle testing applicationsby providing a touch panel 15’ LCD display,keyboard, touchpad, DVD combo drive, and500W power supply in an aluminum alloyframe.
News ID 190
� DSM: ATX IPC with Core 2 Duo processorDSM announces that the Infinity "96M1594"industrial computer available as ATX version isnow also available with Intel Core 2 Duoprocessor. The two cores use a 1066 MHz front-side bus to access the single/dual-channel mainmemory (DDR2 667 SDRAMs) with 4 GB asmaximum memory capacity.
News ID 122
� Avalue: full line of industrial motherboards for various applications
Avalue has developed a full line of industrialmotherboards, including Mini ITX, ITX, MicroATX and ATX form factors for various applications. Avalue now introduce two mod-els: the RX945G and the EMB-9655. TheRX945G industrial mainboard supports IntelCore 2 Duo and is suitable for high-definitionapplications.
News ID 160
� MEN: migration solution for CompactPCIExpress systems
MEN offers an uncomplicated migration solution from 3U CompactPCI to Compact -PCI-Express. In order to guarantee a smoothmigration with existing ‘old’ hardware and software as well as for configuration of hybridsystems MEN in cooperation with Schroffhave developed their own concept: The advan-tages as to data rate, compatibility and costs ofCompactPCI Express compared with the ‘old’CompactPCI are all combined here.
News ID 219
� Siemens A&D: IPCs with Intel Core 2 Duo processors
Siemens A&D has equipped its compact andrugged industrial PCs Simatic Box PC 627Band Simatic Panel PC 677B with energy-sav-ing Intel Core 2 Duo processors. Demandingcomputing and visualization tasks can behandled simultaneously with the new PCs.They are designed for harsh industrial use atambient temperatures of up to 55 degreesCelsius.
News ID 173
� MPL: Industrial PC with soldered ECC RAMand CPU
MPL enlarges its PIP product family by intro-ducing the PIP11, a Pentium M 1.8GHz solutionwith 2MB internal L2 cache. PIP products dif-fer from PCs by PCB design and construction,the EMI/RFI protection on-board and withinthe housing, the selection of long-term availablecomponents as well as other features.
News ID 165
� apra-norm: new catalogue for 19’ components
apra-norm Elektromechanik presents the newcatalogue 19’electronic packaging with morethan 90 pages. It includes the wide product rangeof 19’ mounting systems from subracks, mount-ing frames, individual configurations, plug-insubunits, VME BUS systems, compact PCI tosystem design and customised applications.
News ID 31
� Curtiss-Wright: rugged 6U VPX/VPX-REDIcarrier card
Curtiss-Wright has introduced the VPX6-215ExpressReach module, a rugged 6U VPX (VITA46) and VPX-REDI (VITA-48) carrier card thatfeatures dual PMC/XMC expansion sites withPCI Express connectivity. The VPX6-215 ExpressReach speeds and eases the expansion ofVPX and VPX-REDI systems through the addition of user-specified PMC and/or XMCmezzanine modules.
News ID 89
� Green Hills partners with Curtiss-WrightControls
Green Hills has announced that it has teamedwith Curtiss-Wright, who will offer developerscomplete embedded hardware/software solu-tions based on the Green Hills INTEGRITYRTOS and the MULTI development tool suitefor high-performance modular computing andintegrated subsystem applications in the de-fence and aerospace markets.
News ID 155
June 2007 44
PRODUCT NEWS
EditorJürgen Hübnerphone +49(0)8092-2477413fax +49(0)[email protected]
Contributing EditorTony [email protected]
For Reader Inquiries and Address Changesplease contact:[email protected]
Sales & Marketing DirectorManfred Blumoserphone +49(0)8092-2477411fax +49(0)[email protected]
Claudia [email protected]
Christiane [email protected]
Sales Office - UK and USA, Benelux, Scandi-naviaBlue Sky CommunicationsMalcolm CameronKent Innovation CentreThanet Reach Business ParkMillennium WayBroadstairs Kent CT10 2QQ UKphone +44 (0)1843 609357fax +44 (0)1843 [email protected]
Sales Office - AsiaJean Cheng, [email protected] Hung, [email protected] Media Information & Services7F-3, No. 26, Sec. 2, Ming-Quan East Rd.Taipei 104 Taiwanphone +886 2 2563 1186
Head Office
ICC Media GmbHRauwagnerstr. 585560 Ebersberg / Germany
Editorial Office UK36a Blackacre RoadTheydon BoisEssex, CM16 7LU
Copyright© All rights reserved. No part of this publication may be reproduced or transmitted inany form or by any means without the prior express written permission of ICC Media.
Although we make every effort to present up-to- date, accurateinformation, boards&solutions will not be responsible for anyerrors or omissions or for any results obtained from the use ofsuch information. The magazine will not be liable for any losscaused by the reliance on information obtained on this site. Fur-thermore, boards&solutions does not warrant the accuracy or completeness of the information, text, graphics in this maga-zine. The opinions expressed in the articles are those of the au-thors and not necessarily the opinions of the publisher.
� GE Fanuc: 3U CompactPCI SBC in fiveruggedization levels
GE Fanuc announces the IMPCC2 3U Com-pactPCI single board computer and multi-function PMC carrier card ‘ the latest memberof the PowerPact 3 family. Available in five air-and conduction-cooled ruggedization levels, theIMPCC2 is characterized by an SBC power dis-sipation of only five watts.
News ID 99
� AAEON: compact board with VIA C7/ Edenseries processors
AAEON announces a new compact board -PCM-8120 - that can easily adapt to diverse ap-plications in the competitive industrialembedded market. In addition to the VIA C7processor from 1GHz to 2GHz, this model canbe equipped with an Eden V4 Bus (400MHz to1.5GHz) processor. Moreover, 240-pin DDRIIDIMM system memory is up to 1GB.
News ID 10649
� Acrosser: PCI-Express half size CPU cardAcrosser introduces the AR-B1842, a PICMG1.3 PCI-Express half size CPU Card. The AR-B1842 is based on the Intel Pentium-M proces-sor. Features such as DDRII SDRAMSO-DIMM memory, Slim Cooling Fan, TV-out,programmable watchdog, AC’97 Stereo Audio,Serial ATA, and CF+ Type II are included.
News ID 83
� Schroff: switch mode regulator power supply with 48 V output voltage
Schroff has extended the ecopower switchmode regulator range of power supplies.The AC/DC switch mode regulator ecopow-er SEC148-200 with 56 V output for batterycharging, particularly designed for Datacomand Telecommunications applications, isnow also availabe with an output voltage of48 V.
News ID 207
� BittWare: Stratix II FPGA-based AMCBittWare has announced the release and avail-ability of the GX-AMC (GXAM), the compa-ny’s inaugural FPGA-only board based onAltera’s line of high-density, high-end FPGAs.At the heart of the GXAM is the AlteraStratix II GX FPGA which has been specifi-cally designed for serial I/O-based applica-tions providing up to 20 full-duplexhigh-performance, multi-gigabit transceiverssupporting PCI Express, XAUI, Gigibit Ethernet, Serial RapidIO, SerialLite II andother standards.
News ID 69
� Nexcom: fanless embedded system for factory automation
Nexcom introduces the Nexcom 3200 FanlessEmbedded System which has been specificallydesigned for use within factory environments.With has small form factor and features that in-clude memory back-up, RS-422/485 interfaceswith auto-flow control and ±15 KV ESD pro-tected and isolation.
News ID 208
� Telelogic: new requirements definition & management solution
Telelogic announces a new requirements defi-nition and management solution that is opti-mized for the needs of software developmentprojects. Called Telelogic DOORS Fastrak, thenew Web-based solution provides require-ments definition and management for collab-orative development projects with shortdelivery timelines.
News ID 10667
� Avalue: dual core solutions for multimediaapplications
Avalue introduces its latest solutions for mul-timedia, high-speed data processing and trans-fer, the ESM-945GM and the ECM-945GM.The combination of Intel 945GM with en-hanced graphic engine and Core Duo proces-sor gives the 945GM series the performance tohandle versatile applications and brings a qui-eter operating environment to the desktop.
News ID 9
� Green Hills and Objective Interface collaborate for SDR optimization
Green Hills and Objective Interface Systemshave announced that they have collaborated onan innovative technique that drastically reducesthe memory footprint of the software used tobuild software defined radios (SDRs). This re-duction in resource usage translates into sig-nificant cost and footprint savings in radioproduction.
News ID 90
� Curtiss-Wright: PowerPC VME SBC withXMC and CompactFlash
Curtiss-Wright announces its Power Architec-ture-based rugged VME64x 6U single boardcomputer, the SVME/DMV-184. Available inboth air-cooled and conduction-cooled con-figurations, the 184 combines its Freescale8641 PowerPC processor’s dual 1.0 GHz coreswith a comprehensive suite of military andaerospace I/O capabilities and a broad range ofsoftware support.
News ID 147
Stollmann: ISDN access via USB for embedded systems
Stollmann’s new TA+USBA gives embedded de-vices with USB interfaces access to the ISDNnetwork without special device drivers. In-creasingly, embedded devices are beingequipped with USB rather than serial interfaces.The software applications in use, on the otherhand, will often continue to depend on trans-mitting serial data.
News ID 171
� Siemens A&D: industrial LCD monitors Siemens Automation and Drives presents an in-dustrial LCD monitors that can be used to sup-plement any Simatic PC from a distance of upto 30 meters. This enables users to distribute thedevices over a wider radius in the plant ac-cording to requirements.
News ID 41
� MEN: CompactPCI server blade with DualCore Xeon
MEN Mikro Elektronik presents the new singleboard computer D7, a CompactPCI bladeserver in double Eurocard format. The D7 isequipped with one or two Intel Xeon dual coreprocessors with a frequency of 1.66 GHz andIntel’s server-chipset E7520. It can be used bothas system or peripheral slot board for modernapplications in telecommunications.
News ID 50
� Nexcom: full product catalogue now available
Nexcom announces availability of its full prod-uct catalogue. The 174-page brochure containsdetailed product specifications, marketing ma-terial and ordering information for all of Nex-com’s range of Industrial Computer Solutionsincluding Fanless Embedded Computer Sys-tems, Single Board Computers, Passive Back-planes, Embedded CPU Boards, NetworkAppliances and Blade Servers.
News ID 193
� Advansus: VIA C7/CN700 Mini-ITX motherboard
Advansus announces the first in a series of VIAC7 Mini-ITX motherboards. VCN700-LIC10features a fanless solution integrated with anonboard ultra low power VIA C7 1.0 GHzprocessor and a CN700 digital media chipset.Along with high speed performance and richfunctionality, VCN700-LIC10 provides a cost-effective solution for Point-of-Service, KIOSK,thin client, digital signage, and medical applications.
News ID 123
45 June 2007
PRODUCT NEWS
� IEI: Core 2 Quad-based ATX motherboardIEI Technology introduces IMBA-X9654, an in-dustrial workstation motherboard providingexpansion capabilities with three PCI slots, onePCIe x16 slot, and three PCIe x4 slots by a PLXswitch chip.The IMBA-X9654 industrial moth-erboard with ATX form factor and Intel Q965chipset offers space for four LAN cards.
News ID 2
� Green Hills software selected for advanced missile system
Green Hills Software has announced that SaabBofors Dynamics has selected the Green HillsSoftware INTEGRITY RTOS, networking mid-dleware and Ada-MULTI integrated develop-ment environment for the next phase of itsRBS15 project. The RBS15 is a next generationfire-and-forget, multi-purpose surface-to-surface missile system.
News ID 51
� Kontron: fanless rail PC with two coresKontron announces the release of the KontronThinkIO-Duo, the latest model in its fan-less PCrange. The compact Kontron ThinkIO-Duo(70mm deep / 35mm slot level) is equipped withan Intel Core Duo U2500 processor with a clockrate of 2 x 1.2 GHz and 512MB or 1GB of RAM.Along with high computing performance andinternal Flash (512MB/2GB), this passively-cooled PC, also offers a customizable Compact-Flash socket for data storage or backups/updates.
News ID 30
� ELMA : high-performance TFT panel displays
ELMA is enhancing its competence in the in-dustrial PC area with a new line of high-per-formance TFT panel displays. This marks afurther step in the direction of customer-friendly single-source, single-responsibilityprocurement of visualisation and operatorcontrol elements. Tailored solutions can be pro-vided for nearly every imaginable application,combined with the advantages of economicalmass production.
News ID 39
� Advansus: Q965 MicroATX motherboardfor gaming
Advansus introduces the iQ965-IGM MicroATXmotherboard, powered by Intel Q965 Expresschipset and Intel Core2 Duo processor up to
1066 MHz. Accommodating up to 8 GB of DualChannel DDR2 800 SDRAM, iQ965-IGM offersa powerful platform to system developers.
News ID 200
� Concurrent: AMC with two hard disk drivesor two CompactFlash sites
Concurrent Technologies announces an Ad-vancedMC mass storage module supporting ei-ther two 2.5’ Serial ATA hard disk drives or twoCompactFlashdrives - Raid 0 and Raid 1 sup-port is also available. The AM 600/20x is de-signed to add user definable mass storage drivesinto a double-width full-size or mid-size AMCslot in a system, e.g. MicroTCA, where func-tionality, power, space and cost are key elements
News ID 166
� LynuxWorks to provide RTOS for Galileonavigation system
LynuxWorks announces that it has been se-lected to provide the real-time operating systemfor the European Space Agency’s Galileo proj-ect. As part of the core element of the Galileosystem, LynxOS-178 will be used in the criticalGalileo Mission Segments (GMS), which lies atthe heart of the system.
News ID 98
� One Stop Systems: PCI Express x4 Expresscard
One Stop Systems announces the release of itsPCI Express (PCIe) x4 Express Card, enablinglaptops to operate with high-speed expansioncapabilities at 2.5 Gbps to a x4 downstream de-vice including an expansion chassis or storagesystem. The PCIe x4 cable downlink connects toall One Stop Systems’ expansion chassis for additional add-in board capacity.
News ID 174
� DIGITAL-LOGIC: scalable, fanless computersystem for multimedia
DIGITAL-LOGIC offers a silent and scalablecomputer system for high performance appli-cations such as video streaming servers andmultimedia computers. The MPCX50 is basedon the Intel Core Duo L2400 processor withpassive cooling. It provides features such as a2x1.6GHz clock rate, an Intel 945GM chip set,a 2,048K L2 cache, a modular memory designof up to 2GB, a maximum 1Terabyte hard diskcapacity, and expandability via PCI Express.
News ID 10633
June 2007 46
PRODUCT NEWS
Adv
erti
sers
Inde
x
COMPANY PAGE
AAEON 3
Advanet 29
Advantech 13/15
ATCA Summit 47
COMMELL 14
Comtel 9/11
Concurrent Technologies 25
Curtiss Wright Controls 2
Data Device Corporation 33
Diamond Systems 18
DIGITAL-LOGIC 41
Linux Konferenz 5/6
EMERSON 31
Fastwel 43
Harting 37
Kontron 21/48
MEN Mikroelektronik 27
N.A.T. 23
Peak Systems 17
POLYRACK 39
Rutronik 19
Schroff 24
SECO 43
Syslogic 39
TechNexion 8
TQ Components 35More information about each news is available on www.Embedded-Control-Europe.com/bs_magazine
You just have to type in the “News ID”. —
Critical Technical Information,Product Announcements, Leading Vendors …
The Meeting Point betweenUsers and Vendors in theATCA/MicroTCA Industry
Learn how to develop systems using the latestprocessors, interconnects, backplanes, softwareand hardware standards.
Topics: Get the latest on AdvancedTCA, MicroTCA,AdvancedMC, Service Availability Forum, marketresearch, interoperability, rugged systems, power,cooling, shelf management and applications.
Who Should Attend: Communications andembedded systems designers, test engineers,solution providers, consultants, telecom hardwareand software specialists, venture capitalists,analysts, marketing and PR personnel, engineersand engineering managers.
www.EuroATCAsummit.com
September 18-19, 2007Disney Newport Bay Club HotelParis, France
If it‘s Embedded, it‘s Kontron.
AM4010 Intel® Core™2 Duo processor 1.5 GHz
or Core™ Duo processor 1.2 GHz or 1.66 GHz
Up to 2 GByte soldered memory DDR2 400
Up to 4 GByte Flash
1x GbE, 1x COM, 1x USB
Allows GbE or PCI-Express as Fabric I/O
Copy
right
© 20
07 K
ontr
on A
G. A
ll rig
hts
rese
rved
. Kon
tron
and
the
Kon
tron
logo
and
all
othe
r tra
dem
arks
or r
egis
tere
d tr
adem
arks
are
the
pro
pert
y of
the
ir re
spec
tive
ow
ners
and
are
reco
gniz
ed. R
ev. #
B001
eu02
_WM
H
AM4301 4x Gigabit Ethernet (GbE)
via RJ45 connectors
Intel® 82571EB Gigabit Controller
AM4520 Up to 73 GByte capacity
10,000 rpm, 4.1 ms average seek time
AM4010 – designed for Performance, Capacity, Flexibility
Kontron AdvancedMC modules give
you the most in design versatility
and customization for both your
AdvancedTCA and MicroTCA-based
projects.
From world class processor modules
to storage and I/O modules, Kontron
makes it easy to choose the right
modules for the best system con-
fi gurations.
Incredibly modular, Kontron AMC
modules are the smallest fi eld-
replaceable units that meet all the
requirements for carrier grade relia-
bility, scalability, and manageability.
®
MEMBEROF THEYEAR2006
info - hotline: +49(0)800 7253756w w w.kontron .com [email protected]
new