The magazine of record for the embedded computing industry
INTELLIGENT CONTROL Keeps
Battery-Powered Devices Going and Going ATCA: Not Just for Telecomm Anymore Standards Move Wireless Networks to Mainstream Middleware Pulls Together Complex Platforms
An RTCâ€ˆGroup Publication
GE Fanuc Intelligent Platforms
AXISLite. For those who think that seeing is believing. Power, productivity, portability. And now, free to try. Since it launched, AXIS – the powerful yet easy to use multiprocessor software development environment for VME, VXS and VPX from GE Fanuc Intelligent Platforms – has been at the heart of leading edge applications that are being deployed much faster. But we know that there are plenty more organizations out there who could benefit from its sophistication, its flexibility and its ease of use – but who we haven’t yet convinced. That’s why you can now download AXISLite for free from our web site – so you can see for yourself how it lives up to our claims.
Being free, AXISLite doesn’t do everything AXISView does– it’s a scaled down version. Even so, we believe there will be enough there for you to see what you’ve been missing. If state of the art multiprocessing and DSP systems are what you’re developing, and performance and portability are important to you, you’ll see that nothing else comes close to what AXIS can do. www.gefanucembedded.com/axis
© 2008 GE Fanuc Intelligent Platforms, Inc. All rights reserved.
INTELLIGENT CONTROL Keeps Battery-Powered Devices Going and Going
12 ATCA Breaks Out of Telecom
38 XMC/PMCs with x4 sRapidIO Ports and VPX Intelligent I/O Carrier
46 Multi-Slot PCI Express-to-PCI Extension Systems
Editorial Broadband Everywhere? Don’t Hold Your Breath
Technology in Context
ATCA Breaks Out of Telecom
Integrating with Middleware
ATCA and RapidIO Meet Demanding Semiconductor Applications
Ian Shearer, Mercury Computer Systems
Hardware and 28Prevalidated Middleware Platforms Speed System Integration Jim Lawrence, Enea and Sven Freudenfeld, Kontron
9 Solutions Engineering No Processor Is an Island: Developing Multiple Processor 34 Wireless Sensors Systems with the “New” CORBA Featured Products 38 Standards Will Fuel the Spread of 18 Wireless Network Technologies & Technology Newest Embedded Technology Used by 46Products Industry Watch Industry Leaders Industry Insider Latest Developments in the Embedded Marketplace
Joe Jacob, Objective Interface Systems
Niek Van Dierdonck, GreenPeak
News, Views and Comment Big Changes for 2008: New Technology on Nervous Economic Footing
Mobile Power Management
Mobile and Portable Power Management Systems 24 Optimizing
Event Logging Enables RealTime Systems Analysis 42RTOS John Carbone, Express Logic
Kim Rowe, RoweBots Research
Digital Subscriptions Avaliable at http://rtcmagazine.com/home/subscribe.php
C OTS Your product
cPCI PMC VME PCI ATX choice of OS
Publisher PRESIDENT John Reardon, johnr@r tcgroup.com EDITORIAL DIRECTOR/ASSOCIATE PUBLISHER Warren Andrews, warrena@r tcgroup.com
Editorial EDITOR-IN - CHIEF Tom Williams, tomw@r tcgroup.com MANAGING EDITOR Marina Tringali, marinat@r tcgroup.com
Standard & custom designs
COPY EDITOR Rochelle Cohn
Extended temp. conduction or convection cooled SBCs
Fast, flexible, reliable
CREATIVE DIRECTOR Jason Van Dorn, jasonv@r tcgroup.com ART DIRECTOR Kirsten Wyatt, kirstenw@r tcgroup.com GRAPHIC DESIGNER Christopher Saucier, chriss@r tcgroup.com DIRECTOR OF WEB DEVELOPMENT Marke Hallowell, markeh@r tcgroup.com WEB DEVELOPER Brian Hubbell, brianh@r tcgroup.com
ISO 9001:2000 Certified 321-452-1670 Visit us: www.OTIsolutions.com Untitled-1 1
11/8/07 9:42:51 AM
Advertising/Web Advertising WESTERN REGIONAL ADVERTISING MANAGER Stacy Gandre, stacyg@r tcgroup.com (949) 226 -2024 WESTERN REGIONAL ADVERTISING MANAGER Lauren Hintze, laurenh@r tcgroup.com (949) 226 -2014 EASTERN REGIONAL ADVERTISING MANAGER Nancy Vanderslice, nancy v@r tcgroup.com (978) 443 -2402
The magazine of record for the embedded computing industry
INSIDE SALES MANAGER Carrie Bowers, carrieb@r tcgroup.com (949) 226 -2029
Billing The magazine of record for the embedded computing industry
Mag gie McAuley, mag giem@r tcgroup.com (949) 226 -2024
To Contact RTC magazine:
Free Online www.rtcmagazine.com Spotlighting the Trends and Breakthroughs in the Design, Development and Technology of Embedded Computers. Search Archived Editions along with the Latest News in the Embedded Community. www.rtcmagazine.com An RTCâ€ˆGroup Publication
HOME OFFICE The RTC Group, 905 Calle Amanecer, Suite 250, San Clemente, CA 92673 Phone: (949) 226-2000 Fax: (949) 226-2050, www.rtcgroup.com EASTERN SALES OFFICE The RTC Group, 96 Dudley Road, Sudbury, MA 01776 Phone: (978) 443-2402 Fax: (978) 443-4844 Editorial Office Warren Andrews, Editorial Director/Associate Publisher 39 Southport Cove, Bonita, FL 34134 Phone: (239) 992-4537 Fax: (239) 992-2396 Tom Williams, Editor-in-Chief 245-M Mt. Hermon Rd., PMB#F, Scotts Valley, CA 95066 Phone: (831) 335-1509 Fax: (408) 904-7214
Published by The RTC Group Copyright 2007, The RTC Group. Printed in the United States. All rights reserved. All related graphics are trademarks of The RTC Group. All other brand and product names are the property of their holders.
Broadband Everywhere? Don’t Hold Your Breath by Tom Williams, Editor-in-Chief
“But if everyone communicates by telephone, we’ll have to wire the whole country!” “But if the public travels by airplane, we’ll have to build airports everywhere!”
he reservations above that were expressed about two earlier world-changing technologies. Interestingly, we haven’t seen that kind of reaction to the spread of the Internet. Instead, we are witnessing the kind of hype expected with a new, disruptive technology along with huge expectations accompanied by often unrealistic assumptions. We needn’t deal with the issue of hype. That’s the usual background noise and the only way to cope with it is to develop the proper set of filters. Just as one example, I recently saw a truck in the area belonging to the local cable monopoly. We won’t name the company, but the sign on the side of the truck read, “[Company Name] Triple Play.” Ahem. Those of us in this industry know what the expression “Triple Play” really means. It means data, voice and video over broadband Internet Protocol. The cable company happens to deliver telephone service, cable TV service and Internet access. But that’s NOT what we mean by Triple Play. Now, IMS adds mobile service to the expected IP-based multi/media world. Universal IMS is even further removed from everyday reality. The latter two phenomena are the result of the potential of a universal broadband Internet. This is complicated by the fact that in some isolated pockets that potential has actually come close to realization. And, inexorably, it will continue to spread and grow. But it’s not here yet. There are, of course, lots of great examples of the potential from mostly urban areas about how great full broadband service is—and they are quite believable. But get outside those centers and things look quite a bit different. Out here in California’s Santa Cruz Mountains—just “over the hill” from Silicon Valley—those living outside the centers of Santa Cruz or Monterey, that is, in the hills, have limited options. I, for instance, live on an idyllic ranch in the hills, but the population is not dense enough for the county’s cable monopoly to bother with supplying its “Triple Play.” Most folks make do with satellite TV, which is not bad. For Internet connection, however, there are basically three options: dial-up, satellite or a form of
line-of-sight wireless that can deliver from 1 to 4 Mbit/s bandwidth. We do have phones (and yes, we have flush toilets), but “wiring the country” for broadband Internet is a different matter altogether. And it will take longer and it will go by much different routes. Unlike the telephone service, which (at least until recently) was directed by a single, monolithic entity and whose early expansion partially followed the spread of the population, the Internet is very decentralized, allowing huge service providers to exist in markets that make sense to them while smaller ISPs can fit into more local niches that may or may not be attractive to the big players. In the end, they all connect to the same “cloud” and one day we all may have access to the same range of services and speeds. This, by the way, is why open standards like ATCA, AMC and the SAF-based middleware interfaces make such sense. They enable service providers to get into the game at a small scale and are also attractive to larger ISPs and TEMs, who can begin adding value at a higher level and begin competing sooner in their respective markets. While that may all be just peachy, it is still a long way until outlying areas come in to the broadband fold on a reliable basis. We hear about “fiber to the home,” but seldom of “fiber to the farm.” I just spent a morning with a guy from my ISP who was trying to get me a better signal on my 900 MHz wireless link. It turns out that due to the trees in the path, he was getting a stronger signal from the reflection off the mountain than when he tried to aim directly at the access antenna. We finally decided we’re going to have to move the antenna to the roof of a neighbor’s house and upgrade service. That’s just for basic Internet—we haven’t even considered the idea of IPTV. While wireless connectivity is definitely a vital part of an overall IP-networked world, I think the bandwidth demands of IPTV, let alone full IMS service, will far outstrip what can be done with wireless. Sound difficult? “But to have IMS, we’d have to lay fiber all over the country!”
One-Stop Shopping Bring Us Your List Everything you need for your CompactPCI® Applications Performance Technologies’ CompactPCI® products and solutions are the industry’s most capable and complete open-standards offerings for telecom, aerospace and defense, and commercial applications. We create innovative and robust solutions that enable you to get your products to market faster. Our integrated solutions include: Advanced Managed Platforms™ : These flexible, scalable platforms are ideal for a wide range of compute-intensive applications. Coupled with exceptional support services, we have what you need to quickly and cost-effectively develop and deploy your system. High-Performance Blades: Single and dual-core SBCs, Ethernet switches, intelligent shelf management, voice processors, serial and T1/E1 communications I/O, and RAID storage blades.
Platforms. Blades. Software. Support. Many solutions. One vendor.
NexusWare ® : Our CGL 3.2-registered and POSIX-compliant Linux distribution includes a full operating system, development tools, protocols, HA middleware, and SA Forum-compliant APIs. Best of all, NexusWare's total integration with our blade and platform solutions means you have shorter development cycles. Contact Performance Technologies today and bring us your product development shopping list.
IndustryInsider FEBRUARY 2008
Test and Measurement Industry Trends toward Software-Defined Instrumentation and Use of Multicore
in high-end applications in particular for government, aerospace and transportation areas. Mr. Ulrich Gehrmann, Management Board Chairman of Kontron AG, stated that the acquisition would reinforce Kontron AG in its core business, and would provide Kontron AG with a significant footprint in France. Mr. Gehrmann added that it would compensate Kontron for the sales volume relinquished with the disposal of its mobile computer business in America—this business unit generated sales of USD 25 million in 2006 and was sold in August 2007 to Crane Co. based in Stamford, CT.
Test engineers in industries ranging from aerospace and defense to consumer electronics are facing the challenge of testing increasingly complicated designs with shrinking timelines and budgets. To address these issues, engineers and scientists are incorporating new test and measurement technologies that are capable of meeting complex design requirements without raising costs. National has identified five trends it anticipates will significantly influence the test and measurement industry over the next three years. To continue realizing performance gains without increased clock rates, processor manufacturers are developing processors with multiple cores on a single chip. With multicore processors, test engineers can develop automated test applications capable of achieving the highest possible throughput through parallel processing. One issue facing test engineers is that test instrumentation is not updated as rapidly as the deConnected with technology and vices being tested. TheGet functionality of these complex devices is being defined by the software companies providing solutions now embedded in them, such as most smart phones, which gives design engineers the ability to add Get Connected is a new resource for further exploration features faster than ever before.technologies This is increasingly challenging into products, and companies. Whether your for goal many test engineers because most stand-alone instruments often lack the measurement capabilities of the most recent stanis to research the latest datasheet from a company, speak directly Catalytic Merges with Application or jump to a firmware company's technical page, be the developed and embedded in dards due towith theanfixed userEngineer, interface and that must Celoxica ESL, Changes goal ofengineers Get Connected is to put to youainsoftware-defined touch with the right resource. them. Thus, test are turning approach to instrumentation, which Name to Agility Design Whichever level of service you require for whatever type of technology, gives them the to quickly and user interfaces to meet specific Getability Connected will helpcustomize you connect their with theequipment companies and products Solutions application needs integrate testing directly into the design process. you areand searching for. Following its merger with www.rtcmagazine.com/getconnected Another area experiencing rapid expansion in the test industry is the increase in system-level Celoxica’s ESL business, Catatools for field-programmable gate arrays (FPGAs). More manufacturers are including FPGAs on lytic has announced the company has become Agility Design Solumodular instruments and giving engineers the access in software to reprogram them according to tions Inc. The name change retheir requirements. For example, test engineers can embed a custom algorithm into the device to flects the expansion in company perform in-line processing inside the FPGA or emulate part of the system that requires a real-time size, product offering, geographic response. New system-level tools are emerging that provide engineers with the ability to rapidly Get Connected with technology and companies providing solutions now reach and company vision resultconfigure FPGAs without writing low-level VHDL code. recent merger. Get Connected is a new resource for further exploration into products, technologies and companies. Whether ing yourfrom goal isthe to research the latest Test engineers are also facing new challenges as the use of RF and wireless applications speeds theisdevelopdatasheet from a company, speak directly with an Application Engineer, or jump to a company's technical page, the goalAgility of Get Connected to put you is expanding. RF inand traditionally have level been very specialized fields, type butofthe industry touchwireless with the right resource. Whichever of service you require for whatever technology, ment of signal processing algoConnected help you connect with the and productsinto you are searching for. is experiencing Get a trend where will wireless capability is companies being integrated more products. Soon, rithms offering complete soluwww.rtcmagazine.com/getconnected RF instrumentation could become as ubiquitous as general-purpose instruments such as digital tions for algorithm acceleration, prototyping and implementation multimeters. This growth in adoption requires test engineers to learn wireless protocols and keep in both software and hardware. pace with the rapid introduction of new standards. The solutions include Agility’s As semiconductor devices become more complex, the process of testing each part completely unique software technologies for with a traditional vector-based methodology is increasingly difficult. Complex systems-on-a-chip Matlab to C and C to FPGA syn(SoCs) and systems-in-a-package (SiPs) require a system-level functional test more closely related thesis and a rich portfolio of synto testing components placed on a printed circuit board than a typical chip test, but they still rethesizable algorithmic functions quire the high speeds demanded in production test for the semiconductor industry. The strategy and FPGA hardware platforms. of testing a device by emulating actual real-world signals provides a better method of functional Agility completes the solutions test for these types of high-speed systems. with services delivered by a team of expert users to help customers meet deadlines and delivery Get Connected companies and Connected Kontron to Acquirewith Thales agreement.GetThales Computers the Thales Group. The considered requirements. The new Web site products featured in this section. with companies mentioned in this article. SA will turn over more than €20 acquisition would be subject to Computers is www.agilityds.com. Corporate www.rtcmagazine.com/getconnected www.rtcmagazine.com/getconnected million in its 2007 financial year. contract finalization, to the inKontron is currently in disheadquarters are located in Palo The company commands strength formation and consultation of cussions with Thales and has Alto, California. Work Councils of Thales SA and made an offer to acquire the Thales Computers SA, as well French Thales Computers SA. as to French relevant Authorities The company is 100% owned by Get Connected with companies mentioned in this article. www.rtcmagazine.com/getconnected Get Connected with companies and products featured in this section.
End of Article
ANSI Upholds Reaccreditation of VITA/ VSO
VITA, the trade association dedicated to fostering American National Standards Institute (ANSI) accredited, open system architectures in critical embedded system applications, was notified by ANSI that the latest appeal filed by Motorola on the decision by ANSI to reaccredit the procedures of VITA/VSO has been dismissed. Motorola originally appealed to the ANSI Executive Standards Council (ExSC) Panel, who denied the appeal and affirmed the original decision made by the ExSC to reaccredit the procedures of VITA/VSO to include ex-ante policy. A second appeal was made per the ANSI appeals process to the ANSI Appeals Board. The appeals statement and supporting documentation that was submitted by Motorola in connection with this appeal, together with the appeals material originally before the ANSI Executive Standards Council, was provided to the members of the ANSI Appeals Board via letter ballot. The letter ballot was issued in accordance with clause 11 Appeals process of the ANSI Appeals Board Operating Procedures in order for the members to determine “whether the appellant has established a prima facie case that the decision appealed from was clearly erroneous.” The ANSI Appeals Board Panel decided, based on the record before it, that the appeals statement and record did not establish such a prima facie case. Accordingly, the ANSI Appeals Board dismissed the appeal without an appeals hearing. This decision completes the appeals processes available at ANSI.
Zigbee Smart Energy Profile for Efficient Metering and Management
ZigBee has announced that it has completed development of its ZigBee Smart Energy public application profile. ZigBee Smart Energy offers utility companies a global open standard for implementing secure, easy-to-use wireless home area networks for managing energy. The profile also offers product manufacturers access to a burgeoning green marketplace by establishing a standards-based technology for new products designed to enhance energy management and efficiency by consumers everywhere. ZigBee Smart Energy enables wireless communication between utility companies and common household devices such as smart thermostats and appliances. It improves energy efficiency by allowing consumers to choose interoperable products from different manufacturers giving them the means to manage their energy consumption more precisely using automation and near real-time information. It also helps utility companies implement new advanced metering and demand response programs to drive greater energy management and efficiency, while responding to changing government requirements. A number of Alliance members are currently building products that will be certified by the Alliance to support ZigBee Smart Energy. ZigBee Smart Energy offers innovative electric, gas and water utilities support for advanced metering, demand response, load control, pricing and customer messaging programs. It provides communication and control for devices such as in-home displays, programmable communicating thermostats, water heaters, lighting, smart appliances, plug-in hybrid electric vehicles, plus energy service portals and energy management systems.
Over Four Billion Embedded Systems Shipped in 2006—Still Resisting Commercial RTOS
Recently published research by Venture Development Corporation (VDC) concludes that over 4 billion embedded systems/devices were shipped worldwide in 2006. According to VDC’s “2007 Embedded Systems Market Statistics” report, significant growth in the number of embedded shipments is expected to continue over the coming years. Furthermore, VDC estimates that embedded systems using no formal operating system (with no software on the device that is considered to be an operating system by the project team) or in-house developed operating systems as their primary operating system, represented the majority of total embedded system shipments in 2006. Through 2009, VDC expects the number of embedded devices shipping with a commercial and/or open source operating system to grow at a faster rate than shipments of devices with an in-house/proprietary operating system or with no formal operating system. The trend toward the use of formal third-party operating systems within today’s embedded systems projects is driving this transition. However, VDC believes that migrations in operating system selection will impact total embedded unit shipments less visibly in the shorter term, as the number of products shipping in any given year will always be heavily represented by designs from years past.
Wireless IP Companies Announce Joint Marketing Agreement for WiMAX
Two producers of highly optimized, power-efficient semiconductor IP for WiMAX and emerging wireless standards will offer a joint solution combining baseband processor firmware and MAC layer plus protocol stacks as a complete wireless technology package for system-on-chip developers Coresonic AB, a provider of baseband processor technology for next-generation multimode wireless modems, has announced a joint marketing agreement with SySDSoft, Inc., a provider of embedded software for the wireless broadband market, to promote and market each other’s complementary technologies as complete packaged solutions to mobile device developers and manufacturers. SySDSoft designs baseband and RF/analog circuits for the growing wireless broadband market, with a product portfolio covering a variety of technologies such as WiMAX, Wi-Fi, Bluetooth and wireless USB. Its IP cores can also be implemented in the latest generation of mobile WiMAX IEEE 802.16-e/WiBro wave 2-compliant MAC for mobile devices. The initial focus of this collaborative effort is on producing a WiMax demonstrator, and the companies intend to add Wi-Fi and Bluetooth support to meet customer needs.
E The magazine of record for the embedded computing industry
THE 2-for-1 EDITOR’S CHOICE AWARD 2007
very year RTC offers our advertisers an extra opportunity to feature their products and achievements in our Annual 2-for-1 promotion. This year we’ve decided to highlight one of these submissions to present our first ever “Two-for-One” Editor’s Choice Award. This award goes to the company that demonstrates outstanding technical achievement as presented in their Products spotlight.
May I have the envelope please?
This year’s iPod Nano® Award goes to: Acromag for their inclusion of PMC-VLX/VSX Virtex-5 FPGA I/O and AMX-A30 HighSpeed A/D with Virtex-4 FPGA Thank you to all our advertisers for their continued support of RTC.
PENTXM2 and PENTXM4. Up to Four Cores in a Single Blade. Looking for the best in high speed server class performance for embedded applications? Look no further than the PENTXM2 and PENTXM4 family of manageable single board computers (SBCs)—only from Thales Computers. It delivers the sizzling computing power your applications require while meeting the demands of thermally constrained environments.
Desire real peace of mind? The extended lifecycles associated with the Intel Xeon processor and E7520 chipset, along with Thales’ experienced, long-term support, is your guarantee of outstanding availability and a fully protected investment. PENTXM2 and PENTXM4 VME Blades. Once again Thales sets the standard in real-time COTS system performance.
Want power? These Thales VME blade servers are ready for your most bandwidth intensive applications. At the heart of the system is a single or a twin 1.67GHz Dual-Core Intel Xeon processor combined with an Intel E7520 server class Memory Controller Hub (MCH). That translates approximately up to four times the performance of previous single-core products. Need flexibility? With a full range of standard hardware via triple USB2.0, dual SATA-150, dual PMC slots and PCI-Express available on the XMC slot or the rear P0 interface, you’ll achieve even greater performance and flexibility. EFI Open Standard Firmware enables the card to boot Linux 2.6, VxWorks, LynxOS, Microsoft Windows, QNX Neutrino and Red Hat Linux operating systems.
Dual Head Graphics XMC
For more information please contact: Laurent Richard Tel: +33 (0)4 98 16 34 00 e-mail: firstname.lastname@example.org
www.thalescomputers.com Untitled-2 1
ATCA and RapidIO Meet Demanding Semiconductor Applications The necessity for highly available platforms supporting dense and complex processing in the semiconductor industry has led to the adoption of Advanced Telecom Computing Architecture (ATCA) equipment for top-end control applications. by Ian Shearer Mercury Computer Systems
exploration er your goal eak directly al page, the resource. chnology, and products
ATCA Breaks Out of Telecom
leading semiconductor company re- real-time performance. Semiconductor cently had a situation in which de- manufacturing equipment has throughmands placed on its equipment were put as a key performance criteria, so “real increasing, and the existing bus-based time” becomes more aggressive with evcomputing system was running out of ery new generation of equipment. While bandwidth and introducing unacceptable this level of speed and accuracy is defilatency. The company required a fabric- nitely at the top end of the performance based solution that was cost-effective and requirements, the approach described scalable, yet still supported an existing here applies—albeit in a smaller scale—to proprietary I/O format. Advanced Mez- less demanding systems. zanine Cards (AMCs), standard ATCA panies providing solutions now carrier blades and a RapidIO fabric met Complex Application Makes Big ration into products, technologies and companies. Whether your goal is to research the latest the demanding, low-latency control-loop Demands lication Engineer, or jump to a company's technical page, the goal of Get Connected is to put you requirements this industrial equipment As with any multiprocessor applicaice you require for whatever type ofoftechnology, While tion, moving data around the system is as ies and productsapplication. you are searching for. these standards were not specifically designed for this market, important as the processing. The compathey are capable of meeting requirements ny’s compute platform used a common for a broad range of applications. bus structure to pass data between proThe application is basically an ex- cessors, and between processors and I/O tremely complex control loop. Rather than modules. Over recent equipment generaa simple three-term proportional-inte- tions, the amount of data to be processed gral-derivative (PID) controller, achieving and the processor clock speed increased, nanometer control accuracy requires nu- but the bus capacity remained limited. merous terms and algorithms that neces- Data movement soaked up a greater prositate using multiple processors to achieve portion of the available time for each operation and the bus became a bottleneck to achieving required performance. The Get Connected next-generation compute platform therewith companies mentioned in this article. www.rtcmagazine.com/getconnected fore had a number of enhanced require-
End of Article
February 2008 Get Connected with companies mentioned in this article. www.rtcmagazine.com/getconnected
ments including deterministic latency, support for proprietary I/O, flexibility and cost-effectiveness. It also needed to be on a clear technology path into the future. Control-loop applications have hard, deterministic real-time constraints, and violating the input-to-output latency results in machine failure and unacceptable performance. The results are down time. Achieving a low mean latency is insufficient; the peak system-level latency must also be constrained. Computing the right drive current to apply to an actuator is of little value if the current is not applied at the right time; think of hitting the brakes really hard. There was also a need to support legacy and proprietary I/O. The old computing system was linked to other parts of the equipment by a set of proprietary interfaces. There were so many such connections that replacing them with standardsbased interfaces was unrealistic. While it was highly desirable to adopt a standard platform for the compute platform, that platform had to be capable of easily supporting custom I/O modules. Moving to a new design platform should allow easy implementation of dif-
Technology InContext ferent options. In this instance, the overall semiconductor equipment is produced in a number of versions, with different levels of performance and various options. Also, the equipment includes a large number of controllers with varying levels of complexity. Using a platform that is scalable in terms of performance, processor count and I/O configuration supports that complex product mix. Point solutions—one-off solutions with no roadmap—are often expedient but generally delay issues until some future date. The old bus-based processing architecture had been used for many years very successfully. Unless a similar architecture was used again, maybe just repackaged to make use of new standards, there would be significant disruption in adopting a new platform. It did not make sense to make such a change without knowing the new solution would last for some time and support adoption of new technology as it becomes available. Finally, even in top-end applications driven by innovation and performance, cost is an important factor in selecting a compute platform. In a cyclic industry such as semiconductor equipment, managing production costs can determine a company’s ability to survive.
A switched fabric was the obvious next step from a bus architecture providing a number of performance and usability benefits. RapidIO is the current embedded fabric of choice, being an open standard, providing high bandwidth, and being widely supported by processor and FPGA manufacturers. Most significantly, RapidIO implements point-to-point assured delivery in hardware, with the result that latency is both low and highly deterministic. The closest rival for a fabric interconnect, embedded Ethernet, has a high software overhead and poor determinism. While deterministic low latency was imperative for the target application, RapidIO offers other features that provide important system-level benefits. Using a bus-based communication requires close coupling between software and hardware. Moving data between I/O nodes and multiple processors over a shared bus requires a packetized TDM implementation, which means all internal processing needs to ad-
AMCs FPGA Compute Node FPGA Compute Node FPGA Compute Node FPGA Compute Node
AMCs PowerPC Node PowerPC Node PowerPC Node PowerPC Node
4 Serial RapidIO Switch Fabric
DSP Node DSP Node
I/O Module Control
1 GE Switch
AdvancedTCA is a telecomm standard built around switched fabrics and supports RapidIO through the PICMG 3.5 and AMC.4 standards and is conducive for many other applications including industrial control.
here to a rigid timeline to hit bus availability slots. RapidIO endpoints simply send out data packets when the data is ready, and the fabric takes care of interleaving packets, message queues, etc. This decoupling of software and hardware significantly eases software design, allowing flexibility in the application structure. Another benefit of RapidIO is its ability to handle different levels of traffic priority. This allowed high-priority (deterministic data) and low-priority (system management) packets to share the same fabric with minimal latency impact. In fact, benchmarking activities carried out during system design showed that up to 60 Mbytes/s of low-priority traffic can be moved across critical areas of the fabric with less than 10% latency impact on critical data.
AdvancedTCA is a telecomm standard, so why use it for an industrial control application? Well, for starters, it is built around switched fabrics and supports RapidIO through the PICMG 3.5 and AMC.4 standards (Figure 1). Because
ATCA is an established standard, there are numerous manufacturers of processors, I/O modules, chassis, carriers, etc. providing a strong competitive landscape. In addition, the AMC standards allow for fine granularity in system definition and scaling. For example, it is easier to add a single processor by plugging in an AMC rather than two-to-four processors on a full blade. So ATCA fits well for a “building-block” approach to platform implementation. Hardware is only part of the story for a processing platform. Any respectable processor card vendor offers drivers for the supported interfaces. In this case middleware was developed to isolate the application from the underlying technology, reducing application development time and making future technology transitions easier. ATCA also offers system-level benefits that add value. The base interface (Ethernet) provides a simple method of booting the complete system, with it being common practice to use ftp to boot from an external server. The IPMI infrastructure supports a range of configuFebruary 2008
Technology InContext a proprietary high-speed bus for timecritical data—the RapidIO fabric supports these multiple traffic flows in a deterministic manner.
Mercury’s Ensemble2 product line includes control (PowerPC and PowerQUICC) and DSP processors, FPGA-based AMCs, standard carrier blades, a RapidIO switch blade and various sizes of ATCA chassis.
ration management and machine health monitoring capabilities that can prove extremely useful in the large and complex systems typical of semiconductor equipment. Being designed for the high-volume telecomm industry, the standard also supports attractive cost metrics, particularly as adoption accelerates.
The new compute platform is based on Mercury Computer Systems’ Ensemble2 product line, which includes control (PowerPC and PowerQUICC) and DSP processors, FPGA-based AMCs (Figure 2), standard carrier blades, a RapidIO switch blade and various sizes of ATCA chassis. By using AMC carrier blades with onboard RapidIO switching, the resulting system configuration is a compute
platform that meets the customer’s deterministic low-latency requirements in a physically small, high-density package. A custom-interface AMC translates between RapidIO and the proprietary protocol used throughout the rest of the system. This uses standard serial RapidIO endpoint IP, implemented in an FPGA, to interface to the rest of the compute platform and existing customer IP in the same FPGA to support the proprietary interface. The system is scalable in both processing and I/O by selecting the appropriate AMCs, while the RapidIO fabric provides the necessary flexibility to enable this to be done without compromising performance. Whereas the old processing system used two buses—a standard bus for control and monitoring alongside
In any industry, value must increase over time. In a processing platform that means either reducing cost or increasing performance, or both. MicroTCA offers an obvious route for cost reduction, reducing the switching and support infrastructure costs for a RapidIO platform. While the equivalent system provides lower bandwidth (being dual star rather than mesh architecture) the target application is not bandwidth-constrained. Since fewer switch “hops” are required to traverse the system, MicroTCA actually offers lower hardware latency, giving a small benefit to this critical system characteristic. Cost reduction is assisted by moving to multicore processors. Application developers must take care to make use of the benefits of such a processor, but in this case middleware—a hardware abstraction layer—was provided to isolate the user from the platform, easing the adoption of new technology. While ATCA was designed for the telecom market, it is a very capable standard that is appropriate for a much broader range of applications. Switched fabrics are replacing bus-based architectures where bandwidth and/or latency are important. By having the fabric firmly embedded at the heart of the standard, ATCA is well placed for such applications. The system management capabilities provided by the IPMI infrastructure add value wherever there is more than a minimal level of complexity. RapidIO, as a lean, low-latency fabric specifically designed for embedded applications, is gaining traction in embedded control, so it is not surprising that RapidIO and ATCA together are adopted in this space. MicroTCA, with its lower infrastructure overhead and lower cost base, will broaden the applicability of such fabric-based control platforms to less demanding and more cost-sensitive applications. Mercury Computer Systems Chelmsford, MA. (978) 256-1300. [www.mc.com].
Which Way do You Want Your 10Gb Ethernet?
2500MB/sec 10G b 250MB/se
Software Stack Conventional NIC Technology
Silicon Stack Critical I/O XGE
Silicon Stack Technology from Critical I/O. 10Gb Ethernet at Wire Speed. [Problem] You’re expecting 10Gb Ethernet to deliver a whole lot more performance to your embedded system. But what if you invest in it and get no gain at all? The performance of nearly all existing 1Gb applications are limited by the software overhead associated with the TCP/IP protocol stack. This bottleneck is in the software stack, not the network hardware. So, simply upgrading to 10Gb pipes will not improve your system’s performance. [Solution] Unlike conventional Ethernet interfaces or processor-based “offload” products, Critical I/O’s Silicon Stack technology eliminates this inherent bottleneck by offloading protocol processing to silicon; thereby achieving sustained line-rate performance, microsecond latency, and rock-solid deterministic behavior. And, Silicon Stack is 100% compliant with Ethernet standards, allowing you to leverage existing applications and hardware.
XGE Silicon Stack Ethernet vs. Software-based Stack
Software Stack 10Gb
40 varies with protocol
1Gb Throughput max sustained rate in MBytes/sec Host Overhead
Determinism typical variation Reliability
Horrible ± 200 μsec Poor when under heavy load
Very Low 12 μsec
Rock Solid ± 1 μsec Excellent under all load conditions, no dropped data
Standards Will Fuel the Spread of Wireless Network Technologies Wireless products and technology for sensing and control applications have become a reality, and the widespread adoption of wireless technology is only a matter of time. For that to happen, product integrators require technology standards to provide product interoperability, a large body of knowledge and development sources, second sourcing and flexibility.
by Niek Van Dierdonck GreenPeak
exploration er your goal eak directly al page, the resource. chnology, and products
he average home user has fifty light Application switches in the home; a facility manEmbedded Software ager receives a daily status update of all 10,000 lights in the large office buildNetwork Stack ing; an industrial plant operator receives Embedded Software an alarm that the mains power has failed and that the uninterruptible power supply Wireless Transceiver has commanded all heavy machinery to Hardware “Chip” switch to a safe state. All three examples are typical sensor and actuator cases: the home switch trigFigure 1 Basic architecture of a panies providing solutions now gers the home light, the office building wireless sensor device. ration into products, technologies and companies. Whether your goal is to research the latest luminaries deliver a remote status report, lication Engineer, or jump to a company's technical page, the goal of Get Connected is to put you thewhatever factory power supply light. Of course reliability is important ice you require for typeuninterruptible of technology, the state of the machines. too, but an occasional second pressing of ies and products(UPS) you are changes searching for. In all three applications, devices commu- the switch to make the light turn on won’t nicate over a network. In all three applica- create a critical situation. tions, wireless communication can offer The situation is quite different in the a huge cost saving opportunity given the office building. The facility manager usuhigh installation costs of cable networks. ally guarantees a minimum service level Under the hood, however, the three to the building owner or occupant and applications are very different. In the relies on automation systems to perform home application the main drivers are maintenance tasks. Failing systems causlow cost and low power for the wireless ing a hiccup in the maintenance schedule switches and low cost for the wireless can have huge financial impact. Worse, an intruder switching off all lights of a large office building at 6 PM on a dark winter Get Connected day can cause panic and casualties: nothwith companies mentioned in this article. www.rtcmagazine.com/getconnected ing less than a terrorist attack. So reli-
End of Article
February 2008 Get Connected with companies mentioned in this article. www.rtcmagazine.com/getconnected
ability is essential in commercial building applications. Because commercial applications often have hundreds of devices running on batteries that will eventually fail and need costly labor effort to replace, low power is more critical in commercial applications than in the home. Industrial automation is probably even higher up on the reliability scale, even a slight glitch in a safety application might cause fatalities. On the other hand, industrial automation is usually less cost-sensitive than home and commercial building applications. These are only a few of the parameters that define the diversity of sensor applications. Other parameters include latency of the communication, the number of nodes in a network, the complexity it takes to install, commission and maintain a network, etc. It should come as no surprise that for such a diverse application space a “one-size-fits-all” strategy just does not work—not for the technology, nor for the wireless standards that specify how wireless technology works. Standardization organizations have understood that scoping is required to answer the vast diversity in requirements.
A C R O M A G
F P G A
I / O
S O L U T I O N S
Simply Powerful. Powerfully Simple.
High-performance FPGA I/O designed for faster and easier development. Only Acromag gives you a reconfigurable FPGA wrapped with just the features you need for fast and easy implementation. These PMC modules provide a high-speed I/O interface with plenty of memory for efficient data handling. • Virtex-5 FPGAs (VLX155T/110T/85T or VSX95T) are optimized for logic or DSP • Two banks of high speed 32M x 16-bit DDR2 SDRAM handle high volumes of data • Two banks of 256K x 32-bit dual-port SRAM optimize DMA transfer to the bus • PCI-X 133MHz 64-bit interface ensures fast data throughput • Engineering Design Kit speeds development of custom FPGA applications.
Plug-in I/O cards extend the I/O capabilities
Affordable FPGA-based I/O for all your projects
For details on the fastest and easiest way to develop FPGA I/O visit
AXM I/O extension modules expand interface support for a wide variety of analog and digital I/O signals.
Acromag’s full line of industry pack and PMC modules feature Xilinx FPGAs at a range of price and performance points. Why settle for generic I/O when you can costeffectively design just what you need?
AcromagEmbedded.com, or call 1-877-295-7085
All trademarks are the property of Acromag, Inc. Xilinx and Virtex are trademarks of Xilinx, Inc. and used with permission.
877-295-7085 or 248-295-0310
The Wireless Transceiver chip Application
ZigBee Pro Commercial
ISA-100 Wireless HART Industrial
A view of the most prominent wireless sensor network stack standards.
2.4GHz / 868MHz / 915MHz
20 kbps up to 250kpbs
Typical average power consumption
Up to 65536
Up to 8 nodes
The main parameters of IEEE 802.15.4 compared to Bluetooth.
Some technology providers have usually taken one or two fields of specialization in their quest to be excellent in a few key areas rather than try to do it all equally well, but not well enough for every individual application. Integrators, OEM companies and users of wireless technology are further away from the technology and generally are better aware of their own key requirements than of the key requirements of completely different applications. Therefore many people are often bewildered by the emerging number of seemingly competing standards.
Hardware, Software and Chips… Oh My!
The vital forces behind standardization are: interoperability across brands, second sourcing availability, competition between technology providers to drive prices down, compliance with global regulations and the opportunity to tap into a large body of knowledge. But there is more. Some technology components are so expensive to develop that they can only generate an economic return through very high volumes. And when volumes need to be large, the presence of a global market is paramount. Standards are an excellent
vehicle to generate global awareness and to prepare for such a global market rampup. The basic architecture of a wireless sensor system consists of three layers, as depicted in Figure 1. The lowest block in the architecture, the wireless transceiver, is required to translate digital information (the bits and bytes) into a wireless electromagnetic signal that has the right format to be broadcast by the antenna at the transmitter-side and be reconstructed at the receiver end. In previous generations of wireless technology, you either had a transmitter for transmission only, or a receiver that was only capable of reception. Nowadays, technology has shifted to combined reception and transmission devices as it enables powerful concepts that improve reliability and performance. A straightforward example is the acknowledgement principle: when the receiver successfully receives a message, it sends an acknowledgement to the original transmitter in order to confirm correct reception. Without this principle, the transmitter has no way of knowing whether the message ever arrived. Consequently, transmit- or receive-only technology is considered unreliable and obsolete.
Chip manufacturers need high volume sales to generate meaningful return; high volumes require global markets; and for a global market to take off, technology history has shown that the existence of a standard is essential. This was true for Wi-Fi (wireless Internet), technically termed IEEE 802.11 (a/b/g/n/…). Bluetooth chips are based on a standard defined in the IEEE 802.15.1 specification. For sensor networks, the IEEE 802.15.4 (a/b) standard was set up in 2003. The fact that all three mentioned technologies were standardized under the wings of the same organization, the IEEE, proves that they were conceived for different purposes and not to compete with each other. Indeed, Wi-Fi was conceived as an alternative to wired Ethernet PC communication: high data rate networks with a base station at the center and PCs nearby (i.e., a star-network topology). In order to achieve the application requirements Wi-Fi consumes a fair amount of power—usually sourced from a laptop battery—and data rates degrade quickly when the distance to the base station increases. Bluetooth was conceived with the mobile phone as the center of the universe: it connects the phone to an earpiece, to a GPS device and to a laptop. The Bluetooth data rate of 1 Mbit/s is large enough to carry voice, but is at least one order of magnitude smaller than that of Wi-Fi. In return, the power consumption is lower, most often sourced from a mobile phone battery. In general, the communication range is also smaller than that of Wi-Fi, which is perfectly compatible with the applications as the phone is usually in the vicinity of the earpiece, the laptop and the GPS device. Sensor applications have totally different requirements. Power consumption is probably the most apparent difference: sensors often have to work for years on a coin cell battery or on energy harvested from the environment through a solar panel or a vibration harvester. The battery cannot be recharged like a laptop or a phone battery. Other sensor-specific application requirements are related to automatic network organization, reliability, communication range, the large number of nodes to be supported in a single net-
work, etc. In return, a lower data rate is generally acceptable because most sensors generate fairly small amounts of data and not even continuously. For wireless sensor transceivers, the dominant standard and probably only real standard is the IEEE 802.15.4 specification. The first version was ratified in 2003, with an update in 2006. Several vendors offer transceiver chips. Some of them are a minimal implementation of the standard. Others offer add-ons that are useful in some application segments, such as GreenPeakâ€™s own GP-2000 transceiver, which has many power reducing features targeted toward coin-cell and battery-less applications. There have been efforts to use Bluetooth and Wi-Fi for sensor applications. In those cases, Bluetooth and Wi-Fi were used in a non-standard way, weaving the principles of IEEE 802.15.4 in their native implementation. Today it is widely accepted that the IEEE 802.15.4 offers the best basis for wireless sensor applications. Table 1 compares the main parameters of IEEE 802.15.4 and Bluetooth. Besides the IEEE 802.15.4 standard, a number of technology suppliers have chosen to build a proprietary transceiver. The main motivation seems to be a reduction of the complexity and thus a potential lower cost point. It remains to be seen if a proprietary solution will ever reach sufficient volumes to actually reach that theoretically lower cost point. Additionally, reducing the complexity automatically goes hand in hand with sacrificing performance and thus limiting the applicability.
work node or even a whole branch of the network. In response, the network stack needs to re-organize the communication routes through the network by establishing new links in order to provide uninterrupted connectivity to all parts of the network. The other responsibility of the network is to ensure that messages can travel from a source node to a destination node in a reliable and efficient way. Efficiency here
means that latency requirementsâ€”that is, the travel time of a messageâ€”should be met and that bottlenecks in the routing of messages need to be avoided. The broad application space has widely varying requirements and thus calls for flexibility in the communication technology. Hardware alone cannot offer this flexibility. The network stack comes to the rescue here, because a large part of it is generally implemented
The Network Stack
In essence the network stack has two responsibilities. First, it forms and maintains the network. An important consideration in wireless network stack design is the ability to cope with the constantly varying quality of the wireless links between nodes. For example, in a building automation application, people moving around with their mobile devices have a formidable effect on the link quality, because when a person stands in between two nodes, the link quality will reduce drastically. So the network stack needs to take into account that links can disappear at any moment, possibly isolating a netFebruary 2008
Support for wireless mesh routing
Ability to cope with very large networks
Built-in security features
Comparison of the major features of ZigBee, ZigBeePRO, ISA-100 and Wireless HART.
in software. And software, as compared to hardware, does not have as high an up-front investment cost, meaning that a software investment can live with lower volumes than hardware and still lead to a healthy return. The consequence of these economics is that today we see several Network Stacks standardized, some of them in progress, others already completed. All the current standards build on top of the
IEEE 802.15.4 specification. In other words, these standards assume an IEEE 802.15.4 foundation and sit on top of it (Figure 2).
The Impact of the Zigbee Alliance
The ZigBee Alliance is an independent standardization organization that is driven by a large group of technology providers and OEM companies. The most
recent milestone the alliance achieved at the end of 2007 was to finalize the specification of two Network Stacks: the ZigBee Network Stack and the ZigBee PRO Network Stack. In essence, ZigBee PRO is a superset of ZigBee, adding functionalities related to the ability to scale up the network size and to better cope with wireless interference from other technologies. From a usage point of view, the ZigBee Network Stack is very suitable for residential â€œhomeâ€? applications, where home networks typically contain from tens to a few hundreds of devices. The ZigBee PRO features make it especially suitable for larger applications, very often in the commercial building space. The drawback of ZigBee PRO versus ZigBee is that the extra features require a larger program memory size, which automatically translates into higher cost. In the extremely cost-sensitive consumer market, every extra cost limits the likelihood of adoption. However, thanks to the ever-decreasing cost of silicon, we predict that in the short term the cost difference between ZigBee and ZigBee PRO
4 channels of uncompressed video via USB | Model 2255
* Little or no latency * Simultaneous capture from 4 composite video sources * Total capture rate of 60 frames/sec from all channels for NTSC 2 channels at 30 frames/sec each, 4 channels at 15 frames/sec each * Full frame rate capture on all channels in monochrome or scaled down modes * Multiple output formats and resolutions * Powered through USB * Easy to use API; multiple units supported by the driver. * Linux and Windows SDK
1 20Untitled-1 February 2008
Unlimited, live technical support, forever. Evaluations are now available. For more info email: email@example.com
2/7/08 2:07:40 PM
will be negligible and that most applications will adopt ZigBee PRO. Although the ZigBee Alliance does not explicitly rule out industrial applications, a number of large industrial automation companies have identified the need for extra features, which are not on ZigBee’s top priority list. The two most important “Industrial” features are deterministic latency and deterministic reliability. Latency is the time a message needs to travel from the source to the destination. If the source is a PLC and the destination is a machine, it is easy to see why tight control over latency is important. That is why the standards that explicitly target industrial automation exploit the IEEE 802.15.4 feature called Guaranteed Time Slots to offer latency determinism. In different words, the IEEE 802.15.4 has a feature that allows better control over when a message will arrive. Guaranteed Time Slots are not exploited by ZigBee. The second most visible add-on in industrial automation standards is related to reliability. Reliability is related to the availability or absence of a communication path between two wireless devices. The most important enemy of reliability is wireless interference coming from other users of the same frequency band. The most notable interferers for IEEE 802.15.4-based devices that operate in the 2.4 GHz frequency band are Wi-Fi transceivers. Most interferers will not fully block out an IEEE 802.15.4 device, but will cause some wireless packets to get lost, regardless of the network stack operating on top of it. The industrial standards provide a mechanism that allows packet losses to become evenly spread out over time, even if the number of lost packets does not substantially decrease. The result can be called deterministic reliability.
ISA-100 and Wireless HART
ISA-100 and Wireless HART are the two driving industrial wireless automation standards. ISA-100 is the brainchild of the Instrumentation, Systems and Automation Society (ISA), a non-profit technical society for focusing on industrial automation. The ISA-100 is expected to deliver a standard specification in the course of 2008-2009.
Wireless HART is not a full industrial sensor protocol but an add-on to the old but very popular HART industrial (wired) bus standard for industrial automation. In essence, Wireless HART provides an alternative to the wired message transmission protocol of HART. As ISA-100 and Wireless HART are fundamentally solving the same problems, they have recently joined hands in an effort to examine whether both standards can be merged into one. In a first version they will most likely not be interoperable and will require a network bridge to interface. A follow up version might define a common language. The advantages of the industrial standards are not totally meaningless in commercial building automation, but probably not essential to it either. At the same time the industrial standard features add substantial cost, which residential and commercial application are not likely to accept as these markets are typically much more cost-sensitive than industrial applications. Table 2 lists some of the features of the standards discussed.
Proprietary Wireless Technology
As in all fields of technology, there are proprietary wireless sensor technologies. We define proprietary as a technology that is dominated by a single company. Proprietary does not mean that the specification is not open, because sometimes it is. But a single company still controls the direction of the technology, effectively leading to a monopoly. Proprietary standards have often been designed around a single or limited set of applications. In practice, a proprietary technology can develop much faster than a technology standard because there is no need to reach consensus among different companies. Quite often the proprietary standard can be technically superior to standards when used within their limited set of target applications. Conversely, it is uncommon that a proprietary technology is able to address the broader space of applications that a standard addresses. The two most notable proprietary technologies in wireless sensor communication are Zensys’ Z-Wave and Cornis’ Wavenis. Z-Wave is targeted at residen-
tial automation, as exemplified by the support of a maximum of 237 nodes. This number is sufficient for homes, but is not suitable for larger commercial installations such as hotels and office buildings. Wavenis has generated traction in Automatic Meter Reading applications, and is currently being marketed for other applications as well.
Even within the boundaries of standards, technology providers discover differentiation opportunities. As an example, GreenPeak has provided Transceiver and Network Stack technology compliant to the IEEE 802.15.4 standard and with additional functionalities for ultra-low-power applications. An ultra-low-power application is an application that is able to live off a coin-cell battery or off energy harvested from the environment through a solar cell, a vibration energy harvester or any other environment energy converter. Another evolution that is likely to appear soon in standards is low power routing (LPR). In an LPR network, battery powered devices are able to receive messages from nearby devices and forward these further down a longer communication chain. Standards offer this functionality only for mains powered devices, because a device is required to be in a continuous listening state, consuming a significant amount of power. LPR adds a time synchronization mechanism to the network, allowing devices to wake up simultaneously to initiate communication, avoiding the need to be always on. GreenPeak Zele, Belgium. +32 52 45 87 20. [www.greenpeak.com].
INDUSTRY INSIG H T
OPTIMIZING Mobile and Portable Power
Mobile Power Management
Today’s mobile and handheld systems continue offering more features and subsystems and are demanding more sophisticated power management to extend battery life and offer convenience to the user. Modular software and microprocessor control add features and flexibility for mobile power management. by K im Rowe RoweBots Research
ower management systems have always ranged in size for portable applications from the human-managed Apollo systems to cell phones and BlackBerry devices we see today. The next generation of mobile and portable devices—including portable computers, tools and robots along with cell phones, video games and even hybrid cars—is being designed with ever-increasing demands for longer operation, longer battery life, higher performance output, less heating and greater energy density. Mobile applications are at various phases of maturity and offer a range of challenges depending upon their maturity. Cell phones are mature and their issues are well understood. They can operate in a standby mode that has a small draw with components powered down and automatically started as required. Off-line chargers (chargers that use 230/120 AC power at 50/60 Hz) work as detachable accessories for these and other applications. Other well understood applications include: portable tools, portable video game players, medical devices and many other applications. Newer and less understood applications are generally from the fields of service robotics, field robotics, telecommunication systems and hybrid vehicles. We’ll be looking at some of the newer areas of
energy or mobile power system management with particular attention paid to mobile robotic applications.
System Features and Size
The two key variables in the area of mobile power management systems are: system size and level of isolation. Graphing applications on the grid in Figure 1 offer instant insight into the demands of the management system and the complexity of the system. The trends in these applications are readily apparent. All systems appearing in the left half of the figure are generally driven by primary or secondary cells (rechargeable batteries) and have some local logic to control power management—not a full communication system. Recharging is always done off-line as a secondary activity, and power management complexity is limited to powering down into a quiet mode or powering down subsystems that are an integral part of the electronics. The systems on the right half of the figure have networks and internally distributed power control among the subsystems. Complete subsystems are managed over the network, and subcomponents of the subsystems are managed locally. Most of these systems share a single battery source. All systems on the top half of the figure have some type of accompanying
charging system, typically either solar, hydrocarbon based, mechanical or nuclear. Sometimes sensor arrays run with primary cells (not rechargeable) and are completely disposable devices. All systems on the bottom half of the figure live in friendlier environments and don’t need separate fueled charging systems. They rely upon off-line charging in the case of systems with rechargeable or secondary cells and on battery replacement of primary cells.
Common Power Management Features
Additional features of these systems are software driven and are best categorized into deterministic functions or those functions which are repeatable, remote control functions, remote reporting functions, self-correcting functions, supply sharing functions, safety features and diagnostics. Deterministic functions include the generation of a sequence of events and the recognition of events. The generation of events includes features like soft start, start sequencing of supplies and restart on error. Recognizing events includes features like failure prediction, advanced power-down conditions, data logging and how many restarts or retry events are used before reporting errors and shutting down. February 2008
Harsh Self Organizing Sensors
Video Service Toys
Friendly Figure 1
Mobile systems are very diverse in their application and their operating conditions, often resulting in vastly different requirements for power management.
Remote control functions allow a supply to be remotely controlled and monitored. This includes coordinating the action of multiple supplies, setting voltage and current limits and shutdown sequencing. Remote reporting functions include reporting variations in current, voltage and temperature as well as calculated values like efficiency, power and power factor. Self-correcting functions include temperature compensation of the voltage reference, calibrated output values, temperature-driven current limits and current-driven feedback selection (hysteresis limits). Generally they also include charging features like battery charging temperature and voltage monitoring and battery charging management. Supply sharing features are those features that allow a supply to be shared between two pieces of electronics. This can reduce weight and components if the features never work simultaneously. This would include turning each application on and off as well as managing the power supply to feature mapping to have overall control. Safety features that are generally included are a watchdog timer, brown-out control, low voltage detection and an oscillator fault for the control processor. Other safety features would include careful monitoring of battery temperature, voltage and current with automatic shutdown.
Diagnostics are a main feature that may or may not be included depending upon the application. Big networked applications require this while smaller applications do not.
Today, all of these architectures look very similar with the exception of various hybrid vehicles. This basic architecture is shown in Figure 2. The charging subsystem is either integral in the case of harsh environments, or detachable in the case of more friendly environments. Friendly environments are powered off-line and harsh environments are powered with a variety of different sources that can travel with them. By raising the bus voltage under careful control, the charging system can reverse the flow of energy from the batteries and recharge them. This is a complex and demanding task that depends upon the type of batteries, discharge levels, cell voltages, charge current, charge voltage and a variety of other factors. For that reason, charging subsystems of this nature are always run by a microcontroller. The demanding communications requirements and sequencing also require a real-time operating system to manage complexity. The battery subsystem is generally quite simple. In some cases, special pro-
tection must be included to avoid degassing or venting of batteries and voltage surges (regeneration for example). Large choppers are required in these cases to reduce voltage and protect the batteries against damage. Sometimes a secondary cell is included as a backup. This allows the bias supplies to be run from a secondary source and provides greater reliability through redundancy and guaranteed system control. The power buses can be redundant or not and can offer separate buses for bias supplies or not. There is a great deal of flexibility here depending upon the application. The communications bus or buses can use a range of protocols and wiring. Common options include: CAN, I2C, TCP/IP, UDP/IP, asynchronous serial I/O, SPI, USB and others for larger systems. Smaller systems use direct control of submodule supplies via the microcontroller. Larger systems achieve a greater degree of reliability by providing distributed operation and not directly coupling all the power modules. Overall, one microcontroller is required to run one power module due to sensing and other limitations. Multiple power modules require multiple microcontrollers—each module with one controller and each with a separate communication capability. Multiple master operation is a good idea to avoid a single point of failure.
Real-Time OS Software Architecture
In a power management system, each component needs a microcontroller and each microcontroller may or may not use a real-time operating system (RTOS) to provide the different software features, run the PID loop to control the power supply and provide communication facilities. An RTOS has a few major advantages that make it superior for developing a power management product, particularly if you are developing multiple management systems for different applications. The main advantages are: • modular approach, which eliminates retesting and rework • independent off-the-shelf communication servers • off-the-shelf timer support • integrated development tools • synchronization ability
LEADERSHIP BUILT ON OVER 40 YEARS OF INNOVATION
Four decades of quality manufacturing, design innovation, and process control has earned us the time-honored trust and conﬁdence of a global network of customers.
DDC IS YOUR DATA NETWORKING SOLUTION • MIL-STD-1553 COTS Boards and Components • Fibre Channel • Gigabit Ethernet • High Speed 1760 • Enhanced Bit Rate 1553 (EBR) • ARINC 429
Toll Free: 1-800-DDC-5757
Power Management System Architecture Power Buses
DSP Libraries I/O Servers
DSPnano Services DSPexec
Rechargeable Battery System Module 2
Charge Engine or Line
A power management system should be modular with buses for communication and power transfer among modules.
• nested interrupt capability • resource management • standards-based to eliminate training With this RTOS approach, threads can be developed to provide specific features. For example, all diagnostics could be provided in one thread and only used for those designs that have the space and need for these diagnostics. The overall architecture of DSPnano, which provides these features, is shown in Figure 3. The other main advantage of the thread-based approach is that it supports nested interrupts without special coding and allows most of the functionality to be moved from the interrupt service routines into the threads where resources can be more easily controlled. The net result is a better response time of the system. In such a system, the main piece of the control is done by a PID loop implemented in a thread that monitors voltage and maybe current in some applications and computes new PWM parameters. Depending on the power supply, this thread will control the configuration of the PWM outputs to drive
Hardware H/W & Interrupt Management
the various transistors and/or gate drivers in the design. Different conditional compilation or completely different selectable threads can be used to provide a broad set of configurations to the power supply designer as off-the-shelf modules. Various state machines, self-correcting features, advanced power down and soft start features could be included with this thread to provide a diverse set of features. The limitations of such a system are small, allowing complete reconfiguration by the designer to meet a broad spectrum of power supply topologies and features discussed herein. Another selectable thread will be the I/O module. A broad set of off-the-shelf I/O is available, and by using an RTOSbased design the user could easily select between UART, SPI, TCP/IP, Ethernet, USB, CAN and I2C without changes to their application program. All could be done with minimal resource requirements and standard I/O interfaces. Another additional and optional thread could be an LCD display. Standard high-level C calls could be used to open the device, write to the display and close
Architecture of a softwarebased power management system.
the device. Status updates could be provided. Additional features could provide touch-sensitive operation and menu systems for user interaction. A full range of communication is possible with this supply and other supplies. Startup sequencing becomes a matter of timely communication between the various supplies using one of the standard I/O servers or another mechanism that the designer chooses. Other modular thread-based features include data logging, predictive maintenance and failure, power factor correction, and power and efficiency calculations. Of course, all related information can be easily communicated to other supplies and overall controllers in the system—one of the biggest benefits of an integrated power management system. Additionally, diagnostics and safety features can be added both within modules and in separate threads. This allows the designer to easily configure supplies to make a range of features available easily with full line pricing. Because features are almost entirely software based, simply adding memory (depending on the feature) and changing a configuration is enough to extend the design and provide greater benefits to the customers. Today, advanced fully digital power management systems with tiny real-time operating systems like DSPnano offer the most modular, lean product developmentbased approach to portable power system design and maintenance. Modular features eliminate testing and fit with full line supply development with software-based features rather than hardware-based features. RoweBots Research (519) 208-0189. Kitchener, ON, Canada. [www.rowebots.com].
All the cards...
…for all the solutions Bringing P.A. Semi to PMC, Extreme Engineering Solutions introduces the XPedite8000, the highestperformance, lowest-power PrPMC (Processor PMC) solution available today. With the P.A. Semi PA6T-1682 integrated platform processor, the XPedite8000 offers: • Dual 2.0-GHz Power Architecture processor cores. • Two independent DDR2 SDRAM channels for maximum bandwidth. • Two front-panel Gigabit Ethernet ports. • Two PTMC-compliant P14 Gigabit Ethernet ports. • PCI /PCI-X PMC interface operating at up to 133 MHz. • Linux, VxWorks, and QNX support. For customers looking for one vendor to provide the complete system solution, X-ES provides full component selection, operating system support and integration services.
©2006 Extreme Engineering Solutions firstname.lastname@example.org
©2007 Extreme Engineering Solutions
Integrating with Middleware
Integrating complex building blocks for reliable, high-availability network services can be a daunting and costly process. The higher up the integration scale you can start with a prevalidated platform, the sooner you can begin adding unique value and get to market.
by J im Lawrence, Enea and Sven Freudenfeld, Kontron
exploration er your goal eak directly al page, the resource. chnology, and products
Prevalidated Hardware and Middleware Platforms Speed System Integration
elecommunications applications User Applications have been growing steadily with no sign of decline. Network services, SA Forums AIS such as IP-TV, social networking and 4G presence-enabled services, continue to Standards Based COTS drive growth, setting the foundation for a System Management Building Blocks broad spectrum of content delivery platforms. For instance, as social networking SA Forum HPI begins to converge with communications, it will transition from a fun pastime to a Operating Systems Shelf Mgr valuable business tool providing functionVirtualization panies providing solutions now ality such as advanced customer relationration into products, technologies and companies. Whether your goal is to research the latest I/O App ship management (CRM) capabilities. lication Engineer, or jump to a company's technical page, the goal of Get ConnectedCMM is to put you Blades Blades Whiletypethese types of services may ice you require for whatever of technology, Intelligent Platform Management Bus (IPMB) luxuries now, they will soon ies and productsseem you arelike searching for. -Chassis FRUsbecome an integral part of our lives both personally and professionally. This means Figure 1 Integrating the complex that the demand to deliver content and building blocks in todayâ€™s provide services will grow very rapidly, platforms requires placing heavy demands on the commusophisticated and complex nications infrastructure, while requiring standards to glue them all significant scalability along with unintogether. terrupted service availability. Competition is intensifying as network equipment manufacturers (TEMs) must keep up with providers (NEPs) and telecom equipment time-to-market demands, quality of experience (QoE) expectations and increasing complexity of the network, while focusing Get Connected on differentiating their applications. with companies mentioned in this article. www.rtcmagazine.com/getconnected Building a distributed, highly avail-
End of Article
February 2008 Get Connected with companies mentioned in this article. www.rtcmagazine.com/getconnected
able and reliable system to deliver these services is a complex and often daunting task, particularly since back-end design is increasing in its complexity. Designing the entire system in house is no longer a realistic use of resources nor is it a cost-effective option. Instead, developers are looking to a commercial off-the-shelf (COTS) approach that is driven by standards in order to accelerate and take some of the risk out of the development cycle and ultimately meet delivery schedules. By using COTS building blocks from the hardware computing platform up to the operating system (OS), High-Availability (HA) middleware and certain protocol components, NEPs and TEMs are given the fundamental elements to create a carrier-grade platform. The benefits of a carrier-based platform with a true open architecture foundation are realized in the form of highly differentiated products that are scalable. This frees up valuable engineering resources that can then be used to design applications that add value to and reduce the time-to-market of more innovative services. Integrating all of the complex building blocks is essential and can provide a
Wicked Fast. Pure Reliable. Datalight FlashFX Pro & Reliance: Flash File System Platform
Datalight more than 4 times faster than JFFS2
2500 2000 1500
More than twice as fast as YAFFS2
Datalight more than 20 times faster than JFFS2
More than 10 times faster than YAFFS2
Now for Linux
Reliance + FlashFX Pro
Reliance + FlashFX Pro
How did we do it?
Measured at the lowest level in our system, we see speeds up to 400% faster than JFFS2. Due to the architecture of our platform, the Linux cache helps us achieve performance in excess of the raw speed of the flash chip when measured at the application interface level. All data was gathered using Linux kernel verion 2.6.19 on a Logic PD MX31 Lite with Freescale MX31 processor and 56MB STMicroelectronics (ST512W3A) NAND flash. As with any performance analysis, results on your system may vary.
Want to know more?
Visit www.Datalight.com/WickedFast for product details and specifics on test environment.
Gotta try it for yourself?
Contact us for a FREE evaluation version and experience a more responsive, reliable product for yourself. Visit www.Datalight.com/contactus or phone 1.800.221.6630
Mount Time Datalight more than 5 times faster
Reliance + FlashFX Pro
More than twice as fast as YAFFS2
Time in seconds
Pre-ported verions also available for Microsoft WindowsCE, Wind River VxWorks, and Mentor Graphics Nucleaus PLUS. Porting kits available for use with virtually any 32-bit operating system. All Datalight Software Development Kits include source code and access to our renowned Technical Support. Visit us at: Embedded World 26-26 February 2008 Booth 11-125
ESC Silicon Valley April 14-18, 2008 Booth 5031
High Availbility Middleware
System Management HW Discovery, Resource Mgmt, Event/Alarm, Management I/F
Embedded Management Provisioning, Monitoring, Accounting, Thresholds
High Availability Framework Fault Mgmt, Checkpointing, Hot Upgrade API Change Mgmt
Core Services - OSAL, Messaging, Debug, Log, Monitor
Operating System Hardware Platform
Overview of high-availability middleware.
number of unique technical challenges. As a result, straightforward integration management that has been validated and tested is rapidly becoming a necessity. The SCOPE Alliance has defined a reference architecture for a generic Carrier-Grade Base Platform (CGBP). This architecture, which includes hardware, operating system, operations and maintenance functions and tools, also specifies middleware as a fundamental component for service availability. As CGBP building blocks become commoditized, the industry cooperates in many initiatives to specify and implement an open architecture. In addition, SCOPE creates profiles for The Service Availability Forum (SA Forum), the main organization active in the middleware standardization effort. The SCOPE Alliance has also published the ATCA profile, which provides guidance for a common platform to create carrier-grade platforms that fulfill the needs of NEPs and their customers, the service providers.
The ATCA Building Block
The advent of AdvancedTCA (ATCA), the first standardized hardware platform to meet carrier-class requirements, provides the hardware building blocks and flexibility to integrate complex high-performance systems from off-theshelf components. Processing capabilities and available bandwidth increase with multicore processors while maintaining a
smaller footprint and lower power performance than were achievable in past rackmount configurations. Manufacturers who take advantage of the latest multicore processors in these COTS form-factors will be able to build faster, more scalable systems without upgrading the framework or increasing floor space. Combining ATCA blades with Advanced Mezzanine processor Cards on a carrier-grade, standardbased platform allows network management to take place entirely on one ATCA slot on the ATCA switch blade, relieving the bandwidth from the fabric and maximizing the footprint of the overall system. Delivering reliable high-performance solutions that scale with the demands of the market is quite promising with such advancements. Selecting the appropriate hardware to support a given set of communications protocols and applications is just the beginning of the engineering workload associated with launching a new carrier-class platform. Along with the robust, highly intelligent, high-availability and reliable hardware components provided by ATCA also comes a degree of complexity in the details of virtually every facet of the system. Besides the standards-based COTS system management building blocks, there are a number of other elements that must all work together seamlessly (Figure 1). System design engineers must also integrate the associated OS and in some instances the Board Support Package (BSP)
with the associated supporting drivers for the components on the board or system, and develop middleware to integrate the hardware with the application reliably. The management capabilities for all the hardware, fabrics, software and system components are quite sophisticated, and experts knowledgeable in the complex standards are required in order to pull all the building blocks together into a cohesive system. Robust operating systems are necessary to maintain dependable systems in high-availability environments, allowing for continued service with an interface to the user base that allows the specifics of the hardware to remain transparent.
The Daunting Task of Integration
While the benefits of using the ATCA standard are many, it still requires a certain level of integration effort that can take from 6 to 12 months to make sure all the building blocks work seamlessly together. In addition, integrating the hardware platform can require a great deal of support in the form of program management, functional experts, quality assurance, tools and deployment support, all of which adds up to a tremendous amount of precious personnel, time and money resources. To begin with, integration efforts are on different levels starting from interoperability on the hardware level when using multiple sources for the system components. There are also the considerations of thermal, mechanical, fabric connectivity and IPMI interoperability. This first integration task can become quite complex. Having all the tools to perform this task is already a significant investment not to mention the engineering time to perform that validation and integration. When integrating multi-sourced standard components, further challenges arise when it comes down to identifying which â€œvendorâ€? is at fault when problems occur. The next level of integration requires that the preferred OS is working and supported on the desired blades and might require an additional validation effort. Validating the manageability within the system can be a major undertaking. Even by using standards-based components, the system management (middleware), HPI
and shelf management all need to be validated as a cohesive management unit. Even if the components are designed based on standards or a recipe, every vendor may have a different method of implementing it. For a product to be successful, it needs to be a complete solution with hardware, middleware, OS, etc. Integrating all these elements is a yearâ€™s worth of intense work, which can be a time-consuming and costly task for a systems provider. The following outlines an example of the cost associated with resources and lost revenue due to incremental time-to-market in a real-world network application developed in house:
From the initial procurement phase (which involves component selection, procurement and learning curve) to carrier-class integration and validation of the hardware platform, to deployment support (including debug and component upgrade), the incremental time-to-market can add up to over 700 days. The lost revenue due to this delay can add up to a loss of $1 Million for every month not in the market, which totals to an astounding cost of nearly $24
Million. Within this, the portion associated with just developing the custom middleware to meet the requirements can total up to more than $500,000. Whereas, the build and validate portion can add almost $250,000.
The Emergence of Middleware
Even given the difficult, detailed and time-consuming nature of pulling the pieces of the platform together, embedded system companies should not be discouraged from developing ATCA-based carrier-class systems. In fact, the rapid middleware ecosystem growth provides new opportunities for realizing fully integrated carrier-grade base platforms. The SCOPE Alliance and SA Forumâ€™s specifications and guidance to TEMs is beginning to gain recognition for the portability, interoperability and increased innovation they enable. Standards-based middleware provides TEMs with off-the-shelf highavailability software to complement its carrier-grade equipment (Figure 2). Frequently there is a lapse between the availability of the hardware and the
date that it is possible to deploy applications due to the schedule cost of the backend software development. This gap can be filled with middleware platforms that provide chassis management functions, interprocess communications and services that are scalable from deeply embedded to large, complex systems.
Partnership with a Viable Platform Integrator Is Key
TEMs can realize significant time-tomarket, reduced risk advantages by partnering with proven hardware and middleware experts that can provide integrated, validated and tested platforms. When choosing a viable platform integration partner, developers should make sure the system is clearly defined in terms of well-defined hardware and middleware with the operating system. The complexity of the undertaking requires purposedriven integration. One must be aware of program management and risk mitigation capabilities along with a clear assessment of the amount of resources, including the
DIAMOND SYSTEMS IS DATA ACQUISITION! When you need data acquisition for your embedded application, we offer many patented, high performance solutions. Choose from either our industry leading single board computers with integrated data acquisition, or our stand-alone PC/104 I/O modules.
NEW! ATHENA II
0# !NALOG )/ -ODULE 0ATENTED AUTO AUTOCALIBRATION FOR MAXIMUM ACCURACY s BIT ANALOG INPUTS AT +(Z s BIT ANALOG OUTPUTS AT +(Z s DIGITAL )/ LINES s
-(Z OR -(Z COMPACT FORM FACTOR 3"# s )NTEGRATED DATA ACQUISITION WITH AUTOCALIBRATION s /N BOARD -"