May 2011
www.rtcmagazine.com
"
The magazine of record for the embedded computing industry
Energy Harvesting Brings in THE JUICE
Devices Link Up through the Cloud An RTC Group Publication
Hypervisors Blaze the Path to Multicore Diversity
"
New Specs Fuel System Design Options
DESIGNED WITH YOUR APPLICATION IN MIND.
Embedded Solutions Built Military-Tough. At Extreme Engineering, you will find products that are as tough as the applications they go into. From boards to integrated systems, our embedded solutions are rugged and reliable—ensuring your application is a success, no matter how extreme the conditions. Extreme solutions for extreme conditions. That’s the Extreme way.
Extreme Engineering Solutions 608.833.1155 www.xes-inc.com
Energy Harvesting Brings in “THE JUICE”
46 COM Express Module Boasts Rugged RS-DMM Memory Module
47 VPX Boards Pair Virtex-5 FPGA with PCIe Interface
TABLEOF CONTENTS
48 Instrument Simultaneously Measures and Records Temperature, Velocity and Pressure
VOLUME 20, ISSUE 5
Departments
Technology in Context
TECHNOLOGY DEPLOYED
Sources of Low Power: Energy Harvesting
New System Specifications
Energy Harvesting Applications 5Editorial 14 The Network that Binds: Pulling the Are Everywhere Home into the World of Digital Services Tony Armstrong, Linear Technology
6
Industry Insider Latest Developments in the Embedded Marketplace
Form Factor Forum 10Small Shootout on the Oak Trail & Technology 44Products Newest Embedded Technology Used by Industry Leaders
EDITOR’S REPORT New Network Technologies Enter the Home Smart Grid Meets the Digital Home 12The Tom Williams
TECHNOLOGY CONNECTED Devices and the Cloud
20 Storing Device Data in the Cloud Kurt Hochanadel, Eurotech
TECHNOLOGY IN SYSTEMS Hypervisors, RTOSs and Multicore
24
Embedded Virtualization on x86: A Technical Look at Approaches and Solutions
Legacy Modular 32Modernizing Systems Design with CompactPCI Serial Barbara Schmitz, MEN Mikro Elektronik
Memory Spec Raises 36Rugged the Bar for Rugged Modular Computing
Markus Friese, Lippert Embedded Computers
Industry watch MicroTCA in Networks
Systems for the Evolving Wireless Infrastructure 40MicroTCA Tony Romero, Performance Technologies
Timo Kühn, Real-Time Systems
Virtualization Meets 28 Embedded Real-Time Needs in Multi-OS Systems Christophe Grujon, TenAsys
Digital Subscriptions Avaliable at http://rtcmagazine.com/home/subscribe.php RTC MAGAZINE MAY 2011
3
MAY 2011
Conduction Cooled VME Solid State Disk Phoenix International’s VC1-250-SSD Conduction Cooled Serial ATA (SATA) based Solid State Disk VME blade delivers high capacity, high performance data storage for military, and y, aerospace p industrial applications requiring rugged, extreme emee envi eenvironmental i ron ronmen me tal and secure mass data storage.
Publisher PRESIDENT John Reardon, johnr@rtcgroup.com
Editorial
LLow ow w Ope OOperational Op per pperational e r aatio era tio ti ioo nal al TTemperature Te Temp em emp mperat mp pper e rraatu er ature tur tu urre re --4 -40 -40° 40° C
t H NT PQFSBUJPOBM TIPDL t H SNT )[ SBOEPN WJCSBUJPO tt -PX QPXFS DPOTVNQUJPO -PX QPXFS DPOTVNQUJPO -PX QPXFS DPOTVNQUJPO t 0QFSBUJPOBM 5FNQFSBUVSF ÂĄ UP ÂĄ$ t #VSTU EBUB USBOTGFS SBUF UP .# TFD t 4VTUBJOFE EBUB USBOTGFS UP .# TFD t *OEJWJEVBM QPJOU UP QPJOU EFWJDF DPOOFDUJWJUZ t *OUFHSBUFE 4-$ /"/% 'MBTI t .FFUT NJMJUBSZ BOE *3*( EFDMBTTJGJDBUJPO TUBOEBSET FFUT NJMJUBSZ BOE *3*( EFDMBTTJGJDBUJPO TUBOEBSET E E
High Operational Hi Temperature +85° C
Operational Altitude to 80,000 feet
Art/Production CREATIVE DIRECTOR Jason Van Dorn, jasonv@rtcgroup.com ART DIRECTOR Kirsten Wyatt, kirstenw@rtcgroup.com GRAPHIC DESIGNER Christopher Saucier, chriss@rtcgroup.com GRAPHIC DESIGNER Maream Milik, mareamm@rtcgroup.com LEAD WEB DEVELOPER Hari Nayar, harin@rtcgroup.com
Advertising/Web Advertising
'PS PVS FOUJSF MJOF PG TUPSBHF QSPEVDUT XXX QIFOYJOU DPN t 714ďšş283ďšş4800 An ISO 9001: 2000 CertiďŹ ed Service Disabled Veteran Owned Small Business
Untitled-6 1
EDITOR-IN-CHIEF Tom Williams, tomw@rtcgroup.com CONTRIBUTING EDITORS Colin McCracken and Paul Rosenfeld MANAGING EDITOR Marina Tringali, marinat@rtcgroup.com COPY EDITOR Rochelle Cohn
10/16/09 11:43:57 AM
The magazine of record for the embedded computing industry
The magazine of record for the embedded computing industry
WESTERN REGIONAL ADVERTISING MANAGER Stacy Mannik, stacym@rtcgroup.com (949) 226-2024 EASTERN REGIONAL ADVERTISING MANAGER Shandi Ricciotti, shandir@rtcgroup.com (949) 573-7660 ASIAN REGIONAL ADVERTISING MANAGER Jessica Marinescu, jessicam@rtcgroup.com (+852) 2548 5100
Billing Cindy Muir, cmuir@rtcgroup.com (949) 226-2021
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 RTC Group Publication
4
MAY 2011 RTC MAGAZINE
To Contact RTC magazine: HOME OFFICE The RTC Group, 905 Calle Amanecer, Suite 250, San Clemente, CA 92673 Phone: (949) 226-2000 Fax: (949) 226-2050, www.rtcgroup.com Editorial Office 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 2010, 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.
EDITORIAL MAY 2011
Tom Williams, Editor-in-Chief
The Network that Binds:
Pulling the Home into the World of Digital Services
T
he week I am writing this—the third week in April—has been the subject of an attempt by some to declare a “digital detox week” where we avoid the use of cell phones, texting, social networking, blogging and the like. Fat chance. Embedded intelligence and connectivity have succeeded in binding most of the advanced world into a solid net of digitally linked entities that is only going to get more densely entwined. And most of us seem to love it. This observation might seem out of the scope for an embedded systems publication, but there are lines that are starting to blur between what have been considered “industrial” and “commercial” systems and applications and what has traditionally been thought of as “consumer” electronics. For example, we decided to not cover the design of cell phones, TVs, audio/video receivers and the like and their wired and wireless connection within the home. However, we have long been following developments in machine-to-machine, or M2M systems. M2M systems are made up of intelligent nodes that communicate autonomously with each other to accomplish some application, such as fleet monitoring, in which freight and destination data are automatically collected and used by a scheduling and routing system to make the most efficient use of a transportation system like a truck fleet. Often added to these elements are monitoring systems on engines, bearings, fuel consumption data and the like to additionally optimize maintenance operations. Fleet management also incorporates GPS data for obvious reasons. Similar M2M networks underlie the maintenance of large commercial buildings controlling lighting, HVAC systems, security and overall energy usage. Now very similar approaches are being taken to expand such networked control systems into the home, which represents a truly enormous market. What was once considered “home networking” was properly thought of as a consumer space since it mostly involved control of entertainment systems—and additionally, the home computers—connected via a wired and wireless network inside the home. Increasingly, this network connects to the outside world and the Internet via a set top box. OK, but it is not ending there. The idea of a “universal remote,” a device that lets you control all your consumer electronics by eliminating the truck load of individual remotes, is going to evolve into a home control center, because the kind of network that exists within the home is going to evolve to more closely resemble the kind of network used to control commercial
buildings and industrial facilities. Gradually, it is expanding to also include home security like surveillance cameras, smoke detectors and intelligent door locks that can be monitored either from within the home or remotely over a smart phone app. Communication with security service companies, police and fire services is a natural addition. The next step involves energy usage and monitoring and will be closely associated with the build-out of the Smart Grid. The ability to monitor the energy consumption of individual appliances including the charging of electric vehicles and the ability to schedule major appliances to run at times of lower electricity costs will be realized by leveraging the intelligence that will be built into such devices. Users will be more able to control the costs of their energy usage thanks to the intelligence built into their home networking systems as well as the capabilities that will be supplied by the Smart Grid such as automatic metering. We can envision the development of home control devices with displays that will be used to integrate the data and interaction with the much more pervasive home network, a network that is no longer just a “consumer product.” Rather, this network will be a bridge to a host of automated services provided by companies that are now active in these areas in more traditionally industrial settings but which will expand their offerings to the individual home using very similar technology. At the same time, devices like cell phones that were once relatively simple communication devices have evolved into smart phones and tablets that have the ability to interact remotely with both home and commercial networks as well as the Internet. Where the mix of apps on a phone or tablet may once have been tilted heavily toward entertainment and social networking, it will increasingly include means of interaction with a host of M2M entities, both in the home and those involved with the user’s work. Since this mix of private and job/commerce-related interaction will take place over the same media infrastructure, it will be increasingly difficult to separate “consumer” from “industrial” applications. The same user may use his or her smart phone or tablet to check on whether the kids unlocked the door and are home after school as well as to check on the status of a shipment or the maintenance schedule of a fleet of trucks. Both sets of applications will be supported by essentially the same technology and software infrastructure, differing mainly only at the application level. The market potential for this sort of broad connected world is enormous. RTC MAGAZINE MAY 2011
5
INDUSTRY
INSIDER MAY 2011 Power.org Publishes Power Architecture 32-bit Application Binary Interface Supplement Power.org, the organization that promotes and develops standards for Power Architecture technology, announced the availability of the Power Architecture 32-bit Application Binary Interface Supplement that is current with Power ISA 2.05 (Power Architecture 32-bit ABI Supplement 1.0). The supplement, collaboratively prepared by Code Sourcery (now part of Mentor Graphics), Eager Consulting, Freescale Semiconductor, Green Hills Software, IBM and Wind River, provides detailed documentation on the current state of the 32-bit Power Architecture processor-specific ELF ABI as implemented for the Linux Operating System and embedded environments. The new document, which includes every unique update generated by interested parties of the ABI, was published under the GNU Free Document License (Version 1.3). The original 32-bit PowerPC ELF ABI document, “System V Application Binary Interface: PowerPC Processor Supplement,” which was published in September 1995, was created cooperatively by IBM and SunSoft using UNIX System Laboratories copyrighted material with permission. Since then, the advancement of Power Architecture hardware, improvements in security, and the maturation of the generic ELF ABI specification, necessitated numerous changes to the specification. The information in the new ABI document was largely derived from the 64-bit Power ELF ABI, E500 ABI, EABI, TLS ABI and Secure-PLT ABI. By publishing the supplement under the GNU Free Document License, future loss of revision authority is prevented. Fragmentation caused by changes made since the original ABI documentation was written, has been corrected or documented as incompatible by certain implementations. Fragmentation had occurred in the register reservation/usage overlap (small data areas vs. TLS ABI), complex number passing conventions (GPRs vs. Passed as Structs), and overlap in relocation number reservations and type duplications.
Wind River Expands Mobile Expertise with New Android Development Center
Wind River has set up an engineering team in Stockholm, Sweden focused on mobile technologies. Wind River’s increasing research and development investments include a concentrated effort to grow its Android expertise for a wider range of Androidbased devices including tablets, media phones and other device classes. In the past year, Wind River has achieved a significant
6
MAY 2011 RTC MAGAZINE
acceleration in the number of mobile solution design wins and triple-digit growth in its mobile business segment. Development efforts from the new team include advanced user experience capabilities and Android enablement for broader device categories. In total, Wind River currently has over 20 development centers covering not only mobile but also a wide range of technology specialties. Wind River’s development centers span the globe with locations in Austria, Canada, Germany, France, Israel
and Romania and across the United States and Asia including China, Japan and South Korea. Over the last two years, Wind River’s engineering headcount has grown by nearly 30 percent. Additionally, Jerry Ashford recently joined Wind River as vice president and general manager of mobile solutions. Ashford leads the mobile solutions business worldwide, reporting to Michael Krutz, vice president of worldwide solutions and services at Wind River. Ashford’s responsibilities encompass all mobile operations including business planning, growth strategies development and global team management. Ashford brings over 20 years of experience building and leading successful teams around the world with particular emphasis on emerging regions such as Latin America, Eastern Europe, Middle East and Africa, and Asia Pacific with extensive experience in the China market. Prior to joining Wind River, Ashford was vice president of the Emerging Markets Software Line of Business at Sun Microsystems and also previously worked at Motorola.
Intel and Apple Debut New Thunderbolt Serial I/O—Well Beyond USB 3.0
For some time, Intel has been working on a high-speed multiprotocol serial interconnect it has called LightPeak due to its implementation on optical media. Just recently, Apple announced an implementation of what had been LightPeak but implemented as an electrical interface on its latest model of MacBook and renamed Thunderbolt. The biggest news about Thunderbolt is its speed, which clocks at 10 Gbit/s in both directions. Thunderbolt does not introduce any new protocols, but rather like LightPeak is able to
carry multiple protocols interleaved in its data stream. These include USB, SATA and DVI as well as high-definition video signals over DisplayPort. Both data and video signals can be sent simultaneously over dual 10 Gbit/s channels. The Thunderbolt controller chip interfaces to the Intel peripheral controller hub (PCH) via a x4 PCIe channel and a DisplayPort interface. There are differing views on whether or how the new Thunderbolt technology will affect the acceptance of USB 3.0, which is considerably slower at 400 Mbit/s, but which is already finding wide acceptance in the market. In addition, while USB is an open specification, Thunderbolt with its proprietary controller chip may involve costs such as royalties for its use. Still, the combination of a PCIe controller with Thunderbolt, which uses PCIe as an awning for transferring different data protocols, could mean universal highspeed transfer for many of these existing technologies. In addition, Intel reportedly still has plans for the optical implementation, which would significantly extend the distance of such a high-speed transfer technology.
OpenSAF Announces Release 4.1 of High Availability Middleware
The OpenSAF, an open source community, has announced the public availability of Release 4.1 of its HA middleware platform. OpenSAF Release 4.1 implements all major functions of the Service Availability Forum (SA Forum) Application Interface Specifications (AIS) and includes important enhancements from OpenSAF Release 4.0 including general usability, overload and performance enhancements, and also implements new features such as TCP/IP as alternative
transport protocol and software upgrade rollback. Release 4.1 offers new technical enhancements to Release 4.0, including: • TCP/IP added as an alternative transport protocol for middleware inter-process communication, in addition to the default protocol TIPC. • Software Management now supports rollback, a method to recover to the system state effective at the beginning of a
software upgrade without downtime. • Seamless integration with AM4J (JSR319)-compliant Java Application Servers, enabled with contributed AM4J Java agent that maps AMF Java API to the AM4J API. • Data capacity for the Information Model Management (IMM) service has been improved (tested with 400K objects with an average size of 300 bytes). • I MM now supports schema up-
grade, a feature that allows backward-compatible changes to the class definition in a running system without disturbance.
Hard Drive Market Consolidates as Solid State Storage Gains Ground
These days you can readily buy a desktop PC with a main hard drive of 1 terabyte or better. The prices for high-capacity disk drives have fallen through the
floor. This, among other things, is leading to a consolidation of hard drive manufacturers. Seagate and Samsung have announced a broad agreement that the two companies will “combine” their HDD businesses. Seagate will pay Samsung $1.38 billion, half cash and half newly issued shares of Seagate stock, which will result in Samsung owning 9.6% of Seagate, earning Samsung a nomination for a position on Seagate’s board. In addition, the companies
783.05 (0.63%) This data is as of May 10, 2011. To follow the RTEC10 Index in real time, visit www.rtcmagazine.com. COMPANY -
PRICE
CHANGE
52-WEEK HIGH
52-WEEK LOW
MARKET CAP
Adlink Technology
2.00
0.000%
2.02
1.98
$240.56M
Advantech
3.35
0.419%
3.35
3.30
$1,730.60M
Concurrent Computer -
Elma Electronic Enea Interphase Corporation
-
6.07
-0.979%
6.15
6.07
$56.02M
489.90
0.000%
489.90
483.06
$111.94M
7.58
-2.070%
7.79
7.39
$133.89M
4.45
-0.670%
4.49
4.45
$30.65M
Kontron
12.20
-0.600%
12.34
12.15
$679.07M
Mercury Computer Systems
19.18
0.524%
19.29
19.10
$578.83M
Performance Technologies
2.25
0.000%
2.25
2.25
$25.01M
PLX Technology
3.37
0.898%
3.37
3.35
$150.02M
RadiSys Corporation
9.09
0.888%
9.10
9.00
$220.96M
RTEC10 involves time sensitive information and currency conversions to determine the current value. All values converted to USD. Please note that these values are subject to certain delays and inaccuracies. Do not use for buying or selling of securities. RTEC10 is sponsored by VDC research
Market Intelligence & Strategy Consulting for the Embedded Community Complimentary Embedded Market Data Available at: www.vdcresearch.com RTC MAGAZINE MAY 2011
7
INDUSTRY INSIDER
entered into sourcing agreements, with Samsung supplying NAND to Seagate for SSDs and Hybrid HDDs, and Seagate supplying HDDs to Samsung for PCs, notebooks and consumer electronics. The companies will jointly develop “enterprise storage solutions.” The deal also involves patent cross licensing. The acquisition of Samsung’s HDD business by Seagate comes close on the heels of the pending merger of Hitachi’s HDD business with Western Digital, leaving—with Toshiba— at most three serious suppliers of hard drives The deal between Seagate and Samsung seems to portend a coming shift in mass storage technology. Seagate will take the manufacture of HDDs off Samsung’s hands while lining up a supply of NAND technology for the future—a future that will increasingly be dominated by solid state storage. Seagate looks like it is positioning itself to migrate from the world’s largest hard drive manufacturer to eventually be a major player in solid state storage—which is surely coming. Even the tiny 1.8-inch hard drives will eventually succumb to SSDs, whose continual price drops and capacity increases are being driven by enterprise systems.
Microsemi to Acquire AML Communications
Microsemi has announced that it has signed a definitive agreement to acquire AML Communications for $2.50 per share in an all-cash transaction. The total transaction value would be approximately $28 million net of AML Communications’ projected cash balance at closing. The transaction is subject to customary closing conditions, including the approval of AML Communications’ shareholders, and is expected to close around the end of June 2011. AML Communications had previously entered into a definitive merger agreement with Anaren, Inc. on February 14, 2011 whereby Anaren would ac-
8
MAY 2011 RTC MAGAZINE
quire AML for $2.15 per share in an all-cash transaction. On April 5, 2011, AML Communications’ Board of Directors determined that Microsemi’s proposal to acquire AML represented a superior offer to its shareholders. Headquartered in Camarillo, CA, AML Communications is a provider of microwave amplifiers and subsystems for defense electronics applications. The Company has a broad product portfolio and is a key supplier to major defense programs with Raytheon, Lockheed Martin, Northrop Grumman, L-3 Communications, BAE and others.
Multi-Time Programmable Non-Volatile Memory in 40nm Logic CMOS
Using nonvolatile memory that is on-chip and can be reprogrammed, (SoC) designers can achieve significantly lower costs (70% less), higher performance (24X increase), and improved integration by replacing external serial EEPROM and NOR flash in high-volume mobile and consumer applications. Implemented in standard CMOS with no additional process steps or wafer process adders, the Itera technology from Kilopass provides up to 1 Mbit of storage capacity and 1024 cycles of reprogrammability in applications such as time stamp, key revocation, firmware updates and trimming adjustments. Kilopass is expanding its product line to address a much broader range of applications for embedded non-volatile memory with its patented and proven 2T antifuse technology. With Itera, the company breaks new ground in bringing highly reliable, lowcost embedded logic MTP NVM to advanced process geometries used in SoCs for today’s highvolume consumer and industrial applications. New products to be released in the months ahead will continue to expand the horizons of embedded NVM, removing long-standing challenges to NVM integration across an increasing range of markets, applications
and SoC designs. The current implementation in 40nm process technology has a roadmap to take it to 28nm—with attendant increases in capacity and/or reprogrammability. SoC designs that currently require external MTP NVM to store code or configuration data that change over the life of the product can now use Itera to boost performance and reduce the billof-material costs and space of the final design. Itera enables a typical design to achieve a twentyfour fold increase in performance over Serial Peripheral Interface (SPI) Flash and EEPROM solutions, and realize a cost savings averaging about $6 million for a chip that achieves a 10-million unit per year run rate.
EVENT CALENDAR June 5-10, 2011 Design Automation Conference San Diego, CA www.dac.com
June 6-8, 2011 Embedded Systems Conference Chicago, IL www.eetimes.com/chicago
June 21, 2011 MILESTONE Conference Military Electronics Design Baltimore, MD www.milestone2011.com
June 23, 2011 MILESTONE Conference Military Electronics Design Nashua, NH www.milestone2011.com
June 28-30, 2011 Mobile Computing Summit Burlingame, CA mobilecomputingsummit.com
August 9, 2011 Real-Time & Embedded Computing Conference Denver, CO www.rtecc.com
August 11, 2011 Real-Time & Embedded Computing Conference Salt Lake City, UT www.rtecc.com
August 23, 2011 Real-Time & Embedded Computing Conference Irvine, CA www.rtecc.com
August 23, 2011 Real-Time & Embedded Computing Conference San Diego, CA www.rtecc.com
August 23, 2011 MEDS-Medical Electronic Device Solutions Conference San Diego, CA www.medsmag.com If your company produces any type of industry event, you can get your event listed by contacting: sallyb@rtcgroup.com.
This is a FREE industry-wide listing.
SMALL FORM FACTOR
FORUM
Colin McCracken & Paul Rosenfeld
Shootout on the Oak Trail
H
ollywood has done and overdone its share of old west shootout films, from Clint’s flicks to the Chisholm Trail to the star-studded big budget blockbuster Silverado Trail. Setting its sights on the modest embedded market, Intel’s market machine has entered Oak Trail for an SFF shootout. Incumbent sheriff ARM with its posse of ASIC and standard chip manufacturers lie in low power wait states for the would-be x86 contender to mosey into town. We reported in this column several years ago that x86 and RISC processors inhabit distinct territories, with customers squarely in one camp or the other, since the price/performance/ power profiles didn’t overlap. With shrinking transistors, both sides are closing in on each other, x86 from above and ARM from below, while competing for the coveted smart phone and tablet PC frontier. 2011 promises to bring an epic duel due to multicore ARM Cortex-based chips with integrated GPUs. Of course, some chips will be “gone with the wind” as the consumer market churns, while others just might loiter around for the minimum 5-7 year lifecycles the embedded market requires. The Intel Developer Forum (IDF) in China was chosen as the primary launch event for the “Oak Trail” platform. You know that you are at an IDF keynote when you see the Intel speaker talk about smart phones while being recorded by dozens of ARMpowered smart phones in the audience, and you wonder whether that is as close as Intel will ever get to being inside a smart phone. 10” tablet PCs, however, are still well within the mainstream Windows 7 + x86 platform as proven by Intel’s tablet design wins. Oak Trail, available first in tablets this summer and subsequently for embedded lifecycles as well, has accomplished some notable milestones as the latest Z-series Atom platform. At 3.0 watts thermal design power (TDP), the “Lincroft” code name processors—1.5 GHz Atom Z670 and 1.2 GHz Atom Z650— have reached a new low TDP among Gigahertz x86 mobile and ultra-mobile processor families with integrated graphics. The companion chip / chipset called the SM35 Express chipset (“Whitney Point”) is rated for 0.75 watt TDP. Systems should
10
MAY 2011 RTC MAGAZINE
be designed to cool the 3.75 watt platform TDP, although the typical tablet use case may draw no more than 1 watt, pushing battery life up to cover an 8-10 hour shift for medical and warehouse workers, for example. This is well below the 5W Queens Bay platform (Atom E620-E680 series), although depending on I/O and expansion requirements, the Tunnel Creek (E6xx) processor can run without the Topcliff I/O hub to get down to three and a half watts. Although the power consumption is still high by single-chip RISC standards, Intel has done something with the SM35 Express chipset that deserves high honors. The unprecedented low 0.75W TDP directly results from self-restraint on the I/O side: SM35 Express contains only USB client ports, the LPC Bus, SPI, I2C, SDIO ports and a SATA port (LVDS comes from the processor). We’ve been so conditioned to super-sized PCIe lanes, USB 2.0 and 3.0 ports, Gigabit Ethernet MACs and SATA gen 2 and 3 assault rifle expansion interfaces and the expensive FPGAs with SERDES PHYs for bridging down to low-speed peripherals, that we’ve lost sight of an easier, greener weapon. Medical tablets, inventory logistics computers, and even remote self-powered sensors can attach A/D converters, 3-axis accelerometers, temperature sensors, actuators, transducers, serial ports, GPS receivers and other low-speed peripherals using appropriate low-speed buses. This is the way that ARM-based ASICs and SoCs have done it for years, and Intel finally got it right. Unfortunately, this low-power platform might not attract most of the off-the-shelf computer board vendors who won’t appreciate the value of these low-speed expansion interfaces. And we’re still trying to explain the reason for “Express” in the chipset name. Other x86 manufacturers are digging deeply into their tool chests, largely around graphics processing such as AMD’s Fusion, VIA’s Nano and DMP Electronics’ Vortex86 families, to find their embedded niches. But the shootout on the Oak Trail features Intel against the ARM posse—NVIDIA, TI, Qualcomm, Freescale, Broadcom, Marvell and others. Gentlemen, take your 20 paces…
editor’s report New Network Technologies Enter the Home
The Smart Grid Meets the Digital Home As the Smart Grid is built out with its security and intelligent networking, opportunities arise both in commercial and industrial facilities and now in the home for leveraging local networking to better manage energy as well as services for connected devices. by Tom Williams, Editor-in-Chief
T
he proliferation of wireless networks in homes, offices and commercial buildings may at first appear to be about more easily setting up computer networks that can link together in a local network that is then also connected to the Internet. The potential, however, is much more vast. We have seen wired and wireless networks at work in commercial buildings for managing energy efficiency and HVAC systems, lighting and security. Wireless sensor networks are in use in many fields from agriculture and environmental monitoring to industrial applications of many sorts. Now a huge market potential appears ready to open up for homes that will connect with the coming Smart Grid in what is becoming known as the “Internet of Things.” The Internet of things is simply the connection of everyday objects that are networked for monitoring, control and interaction. The existence of very low-cost microcontrollers and a whole world connected via IP networks makes the connection of anything electrical—from sprinkler system controls to children’s toys—fairly straightforward. The kinds of connections that will link the home with the grid via the all-pervasive Internet will be aimed at some very specific developments that are driving this connection. The first of these is the build out of the Smart Grid itself. The Smart Grid is both an energy
12
MAY 2011 RTC MAGAZINE
distribution system and an intelligent data network. Its main purpose is a more reliable and efficient distribution system and one that is much more robust and resistant to catastrophic failure than the 70-yearold system that is now in place. One of the requirements for the Smart Grid to achieve its potential is for users, commercial, industrial and residential, to be able to manage their energy consumption, and this means that they must be able to interoperate with the Smart Grid. To this end, the U.S. National Institute of Standards and Technology (NIST) has worked with the Zigbee Alliance to sanction the Zigbee Smart Energy Profile 2.0 (SEP 2.0) as a standard for connecting the home area network (HAN) with the Internet and with the Smart Grid. While based on Zigbee, SEP 2.0 is designed to be independent of any MAC/ PHY so it can interact with a variety of networks including the HAN and the broadband Internet. From the home or from other locations, this protocol can be used to connect a wide variety of relatively low-speed devices over wired and wireless connections, enabling the potential expansion of access to the home to different products and services, many of which have not even been thought of yet. In the home, we see the convergence of the connection of traditional “consumer” products such as stereos, TVs, PCs and laptops
with more “industrial” devices like heating systems, utility meters and other appliances along with the Internet of Things, which can include exercise machines, security cameras, lighting systems, smoke detectors and alarms and telemedicine devices. Many, but not all, of these things parallel what has become a wide practice in commercial buildings. Tying all these seemingly disparate devices together is the venerable IP network. The ubiquity of IP networking in Wi-Fi, Ethernet, Internet and now Zigbee has made it the de facto center of connectivity between subnet domains and the connection to the wider world. It has put in place an infrastructure that can be expanded and adapted to even more application areas and domains of connectivity. Several other factors driving the expansion of the home network and its connectivity to the wider world are the push for telemedicine to help keep down the rising costs of health care, and the sort of feedback loop that comes from an infrastructure that is coming into place. That is the fact that the mere existence of broadband connectivity into the home will motivate service providers to leverage that presence to offer more goods and services. Do you want to control your sprinkler system from a web page? Perhaps a knowledgeable service provider can analyze your situation (location, climate, etc.) and design a watering program to suit your needs, upload it and charge a small monthly fee for monitoring and maintenance. This is but one of myriad possible examples. Given that the market and entrepreneurial spirit will come up with many ways to use such connectivity, getting a useable networking infrastructure into the home has its own set of requirements and these must start with where we are today. That is currently an audio/visual/data network that is mostly wireless and that connects entertainment systems and PCs to a central hub such as a router or set top box and then out to the Internet. In the home of the future (even if it doesn’t include a flying car), predicts Adam Lapede, senior director for Atheros, the home network will include other subnets, specifically one
editor’s report
to manage energy and one for monitoring and control of other devices. All three will have different requirements but will be linked together by an IP network using the IPv6 protocol. All three will also present opportunities for expanded applications and business models down the road. Since the home environment is not nearly as predictable or back-configurable as many commercial facilities, there will always be a need for a mix of wired and wireless connectivity to allow for such things as various construction materials or devices housed in metal enclosures that can block wireless signals. At the same time it is not always practical to run cables through a house—or an apartment building for that matter. Such buildings already have plenty of installed wiring in the form of their electrical lines. To take advantage of this, the HomePlug Powerline Alliance has developed a standard for powerline communication called HomePlug Green PHY, which is also compliant with the IEEE 1901 draft standard for powerline networks. The Green PHY specification is a subset of the HomePlug audio/visual (AV) specification, which is designed for higher bandwidth needs in the home. Atheros is also now offering a Green PHY emulation platform for the development of “Internet of Things” products that will operate within the home network environment (Figure 1). There can also be things like Wi-Fi to HomePlug gateways and all these segments are united under IPv6, which will tie together subnets with different requirements and capabilities (Figure 2). The first of these is the energy management subnet, which includes government-mandated higher levels of security. It is from here that the big energy consumers in the household—major appliances, electric vehicle chargers, HVAC systems, etc., can be monitored and controlled. For example, when time-of-day electricity pricing is implemented on the Smart Grid, major appliances should be able to detect when rates change and turn on to take advantage of the savings. Such appliances should also be available for both secure remote diagnostics and be able to
send messages, alerts, etc., to the home owner. Here HomePlug Green PHY will be sufficient as it will in the third subnet, the monitor and control subnet. Monitoring and control of small devices, “Things,” has a large potential in terms of the variety of devices and the combinations of applications in which they can participate. These include telemedicine devices, some of which might also or alternatively be connected to the higher bandwidth audio/visual/data subnet running the HomePlug AV. They will also include cameras for security or to identify users of other connected devices such as the garage door or the front door lock, intrusion and fire alarms, connections to sprinkler and pool systems and the list goes on. Important for both the OEMs offering this infrastructure and its users is not so much what gets communicated from what devices. That is a matter for the appliance marketers and their ilk. What is important at this level is that it be a seamless, plug and play network environment where the user plugs in a device with the same confidence that its data will be available as there is that electricity will flow through the plug. Audio/Visual/Data
Figure 1 The PL-14 HPGP emulation development kit includes standards-based wired and wireless technologies to enable scalable IP infrastructures for smart grid, smart home, security, building automation, remote health and wellness monitoring, and other machine-to-machine (M2M) applications. Atheros, San Jose, CA. (408) 773-5200. [www.atheros.com]. HomePlug Powerline Alliance [www.homeplug.org]. Energy Management Subnet
Charging Station
Energy Management GW
Utility Backhaul
PLC HP-GP • 3G/LTE • 900 MHz wireless
Monitor and Control Subnet Smoke Fitness Detector PLC HP-GP Health
Security Cameras Light Switches
BB-WAN Home GW (BB/AP)
Figure 2 The Internet of Things in the home will involve different subnets with different requirements in a mixed wired/wireless environment. The HomePlug Av and HomePlug Green PHY specifications have been developed to tie together these domains via IPv6.
RTC MAGAZINE MAY 2011
13
Technology in
context
Sources of Low Power: Energy Harvesting
14
MAY 2011 RTC MAGAZINE
technology in context
Energy Harvesting Applications Are Everywhere The development of extremely low-power devices for wireless sensor networks along with advances in IC efficiency for energy harvesting have made possible a wide array of potential applications. by Tony Armstrong, Linear Technology
T
he concept of energy harvesting has been around for over a decade; however, the implementation of ambient energy-powered systems in the real-world environment has been cumbersome, complex and costly. Nevertheless, examples of markets where an energy harvesting approach has been used successfully include transportation infrastructure, wireless medical devices, tire pressure sensing and building automation. Specifically in the case of building automation systems, such things as occupancy sensors, thermostats and even light switches have eliminated the power or control wiring normally associated with their installation and used localized energy harvesting systems instead. A key application of energy harvesting systems is radio sensors in building automation systems. A wireless network utilizing an energy harvesting technique can link any number of sensors together in a building to reduce HVAC and electricity costs by adjusting the temperature or turning off lights to non-essential areas when the building or rooms within are unoccupied. Furthermore, the cost of energy harvesting electronics is often lower than running supply wires, or the routine maintenance required to replace batteries, so there is clearly an economic gain to be had by adopting a harvested power technique. Nevertheless, many of the advantages of a wireless sensor network disappear if
Free Energy Source
Energy Harvester/ Manager
Sensors, A/D, ÂľController
Wireless Transmitter/ Receiver
Figure 1 The main blocks of a typical energy harvesting system or wireless sensor node.
each node requires its own external power source. Despite the fact that ongoing developments in power management have enabled electronic circuits to operate longer for a given power supply, this has its limitations, and power energy harvesting provides a complementary approach. Thus, energy harvesting is a means of powering wireless sensor nodes by converting local ambient energy into useable electrical energy. Ambient energy sources are all around us and they can be converted into an electrical energy by using a suitable transducer, such as a thermoelectric generator (TEG) for temperature differential, a piezoelectric element for vibration, a photovoltaic cell for sunlight (or indoor lighting) and even galvanic energy from moisture. These so called “free� energy sources can be used to autonomously power electronic components and systems. With entire wireless sensor nodes now capable of operating at microwatt average power levels, it is feasible to power them from non-traditional sources. This has led
to energy harvesting, which provides the power to charge, supplement or replace batteries in systems where battery use is inconvenient, impractical, expensive or dangerous. It can also eliminate the need for wires to carry power or to transmit data. In addition, otherwise wasted energy from industrial processes, solar panels, or internal combustion engines, can be harvested for useful purposes.
Characterization of Energy Harvesting Applications
A typical energy harvesting configuration or wireless sensor node (WSN) is comprised of four blocks, as illustrated in Figure 1. These are: 1) an ambient energy source, 2) a transducer element and a power conversion circuit to power downstream electronics, 3) a sensing component that links the node to the physical world and a computing component consisting of a microprocessor or microcontroller that processes measurement data and stores them in memory, and 4) a communication component consisting of a short range RTC MAGAZINE MAY 2011
15
technology in context radio for wireless communication with neighboring nodes and the outside world. Once the electrical energy has been produced, it can then be converted by an
energy harvesting circuit and modified into a suitable form to power the downstream electronics. Thus, a microprocessor can wake up a sensor to take a reading or
Elements within the WSN Power Supply (or battery)
Factors affecting power consumption
Sensors
Physical to electrical signal conversion Complexity of supporting components Signal sampling Signal conditioning
ADC
Sampling rate Aliasing Dither
Microprocessor
Core operating frequencies Operating voltages Power proportional to process & computational load Ambient temperature Application code Peripheral utilization
Radio
Modulation scheme Data rate Transmission range Operational duty cycle
Discharge rate Battery dimensions Supply voltages Type of electrode material used DC/DC Efficiency
TABLE 1 Factors affecting power consumption of a wireless sensor node.
16
Untitled-5 1
MAY 2011 RTC MAGAZINE
measurement, which can then be manipulated by an analog-to-digital converter for transmission via an ultra-low-power wireless transceiver. Several factors affect the power consumption characteristics of an energy harvesting system of wireless sensor node. These are outlined in Table 1. Of course, the energy provided by the energy harvesting source depends on how long the source is in operation. Therefore, the primary metric for comparison of scavenged sources is power density, not energy density. Energy harvesting is generally subject to low, variable and unpredictable levels of available power, so a hybrid structure that interfaces to the harvester and a secondary power reservoir is often used. The harvester, because of its unlimited energy supply and deficiency in power, is the energy source of the system. The secondary power reservoir, either a battery or a capacitor, yields higher output power but stores less energy, supplying power when required but otherwise regularly receiving charge from the harvester. Thus, in situations when there is no ambient energy from which to harvest power, such as darkness in the case of a
4/6/11 8:55:09 AM
technology in context photovoltaic cell, the secondary power reservoir must be used to power the WSN. Of course, from a system designer’s perspective, this adds a further degree of complexity since they must now take into consideration how much energy must be stored in the secondary reservoir to compensate for the lack of an ambient energy source. Just how much will be required depends on several factors. These will include: • the length of time the ambient energy source is absent • the duty cycle of the WSN, i.e., the frequency with which a data reading and transmission has to be made • the size and type of a secondary reservoir (capacitor, supercap or battery) • whether enough ambient energy is available to act as both the primary energy source and have sufficient energy left over to charge up a secondary reservoir when it is not available for some specified period State-of-the-art and off-the-shelf energy harvesting technologies, for example in vibration energy harvesting and indoor photovoltaics, yield power levels in the
Untitled-13 1
order of milliwatts under typical operating conditions. While such power levels may appear restrictive, the operation of harvesting elements over a number of years can mean that the technologies are broadly comparable to long-life primary batteries, both in terms of energy provision and the cost per energy unit provided. Furthermore, systems incorporating energy harvesting will typically be capable of recharging after depletion, something that systems powered by primary batteries cannot do. As already discussed, ambient energy sources include light, heat differentials, vibrating beams, transmitted RF signals, or just about any other source that can produce an electrical charge through a transducer. Table 2 illustrates the amount of energy that can be produced from different energy sources. Successfully designing a completely self-contained wireless sensor system requires readily available power-saving microcontrollers and transducers that consume minimal electrical energy from low-energy environments. Fortunately low-cost and low-power sensors and mi-
crocontrollers have been available for a couple of years or so; however, it is only recently that ultra-low-power transceivers have become commercially available. Nevertheless, the laggard in this chain has been the energy harvester. Existing implementations of the energy harvester block shown in Figure 1 typically consist of low performing discrete configurations, usually comprising 30 components or more. Such designs have low conversion efficiency and high quiescent currents. Both of these deficiencies result in compromised performance in an end system. The low conversion efficiency will increase the amount of time required to power up a system, which in turn increases the time interval between taking a sensor reading and transmitting this data. A high quiescent current limits how low the output of the energy-harvesting source can be, since it must first overcome the current level needed for its own operation before it can supply any excess power to the output. A new generation of energy harvesting ICs brings a new level of performance that was just not possible with discrete implementations. Examples of these are the
17
4:26:15 PM RTC MAGAZINE 3/31/11 MAY 2011
technology in context
GREAT PRICES ON DAQ
LOWEST COST 12-BIT USB-1208 Series • 8 analog inputs
Energy Source
Typical Energy Level Produced
Typical Application
Small solar panels
100s of mW/cm2 (Direct Sunlight)
Handheld electronic devices
Small solar panels
100s of µW/cm (Indirect Sunlight)
Remote wireless sensors
2
Seebeck devices (which convert heat 10s of µW/cm2 (Body heat) energy into electrical energy) Seebeck devices continued
10s of mW/cm2 (Furnace exhaust stack)
Remote wireless actuators
Piezoelectric devices (which produce energy by either compression or deflection of the device)
100s of µW/cm2
Handheld electronic devices or remote wireless actuators
RF energy from an antenna
100s of pW/cm2
Remote wireless sensors
TABLE 2
• 1.2 kS/s to 1 MS/s sampling
Energy sources and the amount of energy they can produce.
• Up to 4 analog outputs
Pros
• 16 digital I/O, counter
from $129
TEMPERATURE
USB-TEMP and TC Series
Cons
Thermal
Able to continuously supply DC power.
Temperature differentials can be difficult to implement in enclosed environments like airplanes.
Piezoelectric
Does not need to be manually charged. The Aircraft provides plenty of vibrational force.
Each element has to be manually tuned to the aircraft’s vibration frequency.
• Measure thermocouples, RTDs, thermistors, or voltage
TABLE 3
• 24-bit resolution
Comparison of piezo and thermal harvesting methods as applied to aircraft.
LTC3109, LTC3588-1 and LTC3105 from Linear Technology. As a result, they have become a catalyst for growth for the makers of energy harvesting systems because they can harvest energy from very low levels. This level of performance, coupled with the cost-effective price points of the transducers, microcontrollers, sensors and transceivers, has led to an increase in market acceptance. This is one of the reasons why there is a significant amount of attention being given to such systems in a variety of applications around the globe.
• 8 channels
from $329
HIGH-ACCURACY USB-2416 Series
• Measure thermocouples or voltage • 24-bit A/D, 1 kS/s • Up to 4 analog outputs, digital I/O, counters
from $1099 Included Software and Comprehensive Drivers, Free Tech Support, Unmatched Warranties
(800) 234 -4232 mccdaq.com ©2011 Measurement Computing Corporation 10 Commerce Way, Norton, MA 02766 info@mccdaq.com
18
Remote wireless sensors
Untitled-15 1
MAY 2011 RTC MAGAZINE
4/6/11 12:35:36 PM
A Real-Life Example: Aircraft Health Monitoring
The structural fatigue of today’s large fleet of aircraft is a real-world issue since it can lead to catastrophic consequences if ignored. Currently, airframes are monitored through more inspections, through improved structural analysis and tracking methods, and by incorporating new and innovative ideas for assessing structural integrity. This is sometimes referred to as “health monitoring of aircraft.” This process incorporates sensors, artificial intelligence and advanced analytical techniques to produce real-time and continual health assessment. Acoustic emission detection is a well-established method of locating and
monitoring crack development in metal structures. It can be readily applied for the diagnosis of damage in composite aircraft structures. A clear requirement is a level form of “go,” “no go” indications of structural integrity or immediate maintenance actions. The technology comprises low profile detection sensors using piezoelectric wafers encapsulated in polymer film and optical sensors. Sensors are bonded to the structure’s surface and enable acoustic events from the loaded structure to be located by triangulation. Instrumentation is then used to capture and parameterize the sensor data in a form suitable for lowbandwidth storage and transmission. Thus, although wireless sensor modules are often embedded in various airplane sections for structural analysis, wings or fuselage for example, powering them can be cumbersome. Therefore, these sensor modules are more convenient and efficient when powered wirelessly, or even self-powered. In an aircraft environment there are a number of “free” energy sources available to power such sensors. Two obvious methods that could readily be utilized are thermal energy harvesting and/or piezoelectric energy harvesting. In the case of thermal harvesting, there can be some difficulty regarding an air-
technology in context plane. The best opportunity for capturing temperature differential in an aircraft is between the aircraft “skin” on the inside of the cabin and the internal cabin temperature. Since the Seebeck effect is the underlying thermodynamic phenomenon that converts thermal heat to electric power, the main equation to take into consideration is: P = ηQ Where P is electrical power, Q is heat and η is efficiency. Larger TEGs that use more heat, Q, will produce more power, P. Similarly, the use of twice as many power converters will naturally produce twice the power given that they can capture twice the heat. Larger TEGs are created by putting more P-N junction in series; however, while this created more millivolts per delta T (mV/ dT), it also increases the series resistance of the TEG. This increased series resistance limits the power available to the load. Therefore, depending on the application requirements, it is sometimes better to use smaller TEGs in parallel rather than using a larger TEG. Regardless of the configuration, TEGs are commercially available from a number of suppliers. Piezoelectricity, on the other hand, can be generated by applying stress to an element, which in turn creates an electric potential. The piezoelectric effect is reversible in that materials exhibiting the direct piezoelectric effect, the production of an electric potential when stress is applied, also exhibit the reverse piezoelectric effect or the production of stress and/ or strain when an electric field is applied. In order to optimize a piezo transducer, one needs to characterize its source for vibration frequency and displacement. Once these levels have been determined, a piezo manufacturer can design a piezo that is mechanically tuned to the specific vibration frequency and size it to provide the necessary amount of power. The vibration in the piezo material activates the direct piezo effect, which results in the accumulation of charge on the output capacitance of the device. This is usually pretty small so the AC open circuit voltage is high—on the order of 200 volts in many cases. Since the amount of charge generated from each deflection is relatively small, it is necessary to full-wave rectify this AC signal and accumulate the cycle-by-cycle charge on an input capacitor.
With respect to the energy source choice, there are trade-offs between thermal and piezoelectric sources. Nevertheless, regardless of which method is selected, both are viable and practical solutions and can be readily deployed with current technology. Table 3 summarizes the pros and cons between these two methods. System designers and systems planners have to prioritize the need of their power management from the onset in
Untitled-3 1
order to ensure efficient designs and successful long term deployments. Fortunately, there is now a growing number of energy harvesting power management ICs from the leading high-performance analog IC manufacturers to greatly simplify this task. Linear Technology Milpitas, CA. (408) 432-1900. [www.linear.com].
19
9:49:01 AM RTC MAGAZINE 2/16/11 MAY 2011
Technology
connected Devices and the Cloud
Storing Device Data in the Cloud IT operations have been increasingly benefiting from the economy and flexibility of cloud computing. Now the rapidly multiplying number of embedded devices in all manner of applications can take advantage of making their data available in the cloud as well. by Kurt Hochanadel, Eurotech
I
The Device Cloud
Traditionally, data from embedded devices is either downloaded manually or through some proprietary means within
20
MAY 2011 RTC MAGAZINE
Worldwide IT Spending* by Consumption Model 2009, 2014 ($B) 600
Worldwide IT Product and Public IT Cloud Services Spending ($ billion)
n the past few years, the business world has adopted a new buzzword— cloud computing. Although many industries are embracing cloud computing, the embedded device market has yet to find a standard, cost-effective way to connect distributed devices to the cloud. It can take weeks, months and even years to plan, procure and deploy IT infrastructure to connect embedded devices to the network and capture valuable data. According to International Data Corporation, worldwide revenue from public IT cloud services exceeded $16 billion in 2009 and is forecast to reach $55.5 billion in 2014. This rapid growth rate is five times the projected rate for traditional IT products (Figure 1). Embedded devices have gone from being isolated tools to being a part of everyday life as they connect data over various ubiquitous networks. Networked devices are becoming commonplace among consumers and in industries ranging from healthcare to transportation, giving way to the arrival of the embedded cloud. The embedded cloud enables embedded developers to deploy services on robust, reliable and flexible communication infrastructure into the business enterprise.
Public IT Cloud Services Traditional IT Products
500 $460B
400
300
$360B CAGR 27% 5%
200
100
12% 4%
0
Figure 1
$16B 2009
$55B 2014
According to International Data Corporation, cloud offerings will gain significantly on traditional IT products by 2014.
a dedicated infrastructure. The data is downloaded at specified intervals, such as each night, or once per month as in utility bills. Many times the data is never used, or perhaps a user will refer to the data after a problem has occurred. Machine-to-machine communication solutions can be divided into four fundamental building blocks as shown in Figure 2. First, organizations must select and integrate sensors and edge device hardware while considering factors such as comput-
ing platform, interface, power supply and housing options. Next, developing the device firmware includes the operating system, drivers, networking, application and a user interface. Finally, the integrator must figure out how to connect the system to the business application via a dedicated communication infrastructure, which includes a connectivity broker, infrastructure, application, database and user interface. Device-to-cloud solutions, on the
technology connected
other hand, can simplify the process and greatly reduce the time-to-market. By connecting devices to business logic through the cloud, users can deliver valuable data between distributed devices and vital business applications using standard practices and commonly used methodologies. In many cases, using a cloud service saves time, money and increases efficiency by effectively acting on data in real time utilizing sound business logic. However, developers must have the correct building blocks in place in order to get to market with a cloud solution and leverage the benefits offered. The Device Cloud is a term coined by Eurotech (Figure 3) to describe how organizations can bring data from device to business application with an integrated solution to turn bits of data into valuable and actionable information. Compared to the conventional approach shown in Figure 2, the Device Cloud in Figure 3 takes care of the entire communication infrastructure piece of the puzzle instantly. The Device Cloud stores the device data in the cloud, providing organizations access to that data in real time through standard methodologies while also archiving the data in a secure redundant manner for easy use with analytic applications. The Device Cloud is also scalable, secure and many times more cost-effective than traditional infrastructure. As cloud computing becomes more mainstream, organizations are evaluating whether a cloud solution can be an efficient alternative to traditional computing networks.
The Benefits of Storing Data in the Cloud
Consumers and businesses alike have realized the benefits of storing data in the cloud. Most individuals are using the cloud in some capacity—e-mail, social networking, photo sharing sites, online banking, and thousands of other everyday uses. Storing device data, however, is more complicated than these consumer services
Conventional Approach
Internet Hypervisor Business Application Sensors & Device Hardware
Device Firmware/ Application
Communication Infrastructure
Business Application Integration
Time-to-Market: 8-24 Months Initial Investment: $$$$$$
Figure 2 In the past it could take years to connect embedded devices to the network and capture data.
Eurotech Approach: Everyware Device Cloud
Competitive Advantage Sensors & Device Hardware
Device Firmware/ Application
Cloud-based Device Data Management and Delivery
Business Application Integration
Time-to-Market: 2-6 Months Initial Investment: $$
Figure 3 Today, ready-made solutions like Eurotech’s Everyware Device Cloud greatly reduce deployment time for IT departments.
but can reap even more value. When device data is stored in the cloud, users can access that data anytime and anywhere through secure online portals. Device data becomes available not only to one specific location or one manager, but to corporate hubs or any executive looking at data from an overall business perspective. Users can increase efficiency with the cloud by allowing employees more flex-
ibility in accessing important data, software programs and management functions. For instance, a manager who uses a cloud-based device solution can work on tasks regardless of location, rather than needing to be at the office. Storing device data from a retail warehouse in the cloud, for example, makes shipping and receiving data available to corporate headquarters in real time— allowing for an immediate asset manageRTC MAGAZINE MAY 2011
21
technology connected
Publisher (Source)
Subscriber (Sink)
Broker
sub(topic) pub(topic, data) pub(topic, data)
Figure 4 The publish/subscribe communication model.
ment system. Suppose a hot new toy was shipping in time for the holiday season. In the past, retailers might receive daily tallies of inventory. With cloud computing, they could use RFID scanners to see an accurate picture of inventory at any given moment and redirect inventory, manage production and operations on timely data. The cloud provides operational efficiencies that did not exist in the embedded device world before. For instance, in the healthcare industry, if a patient had a catastrophic event and was using a home health device connected to the cloud, their doctor could receive an alert in real time, and life saving systems could be deployed.
Building Blocks for a Device-toCloud Solution
If an organization decides to collect and store data from distributed devices in the cloud, there are important considerations during the product development cycle that can make connecting to the cloud simpler. Of course hardware and operating system are important building blocks on the device, but in the interest of space we’ll discuss an ideal and efficient interconnect protocol and application framework for a device-to-cloud solution. A variety of protocols connect embedded devices, such as HTTP or SOAP. Some of the drawbacks associated with these options include thousands of bytes sent as a header for the message, rigid formats and point-to-point communication. HTTP and SOAP provide access to a single embedded device, and some network topologies cannot handle the bandwidth required as it builds up with hundreds of embedded devices. Making efficient
22
MAY 2011 RTC MAGAZINE
use of bandwidth is also a cost-saving measure since some networks charge by the byte of data and it makes little sense to send messages with several thousand bytes of overhead. IBM and Eurotech’s Arlen Nipper developed the Message Queuing Telemetry Transport (MQTT) protocol 10 years ago. As described at www.mqtt.org, the MQTT protocol enables a publish-and-subscribe messaging model in an extremely lightweight way. It is useful for connections with remote locations where a small code footprint is required and/or network bandwidth is at a premium. MQTT provides publish-and-subscribe messaging that enables devices to send and receive alerts and data when significant events occur (Figure 4). With a single publisher and many subscribers, engineers can send information from a single point to many other devices or listeners interested in receiving the information. One can liken this concept to the social networking tool Twitter. A single person posts information, and many subscribers view the information simultaneously. Embedded devices can utilize the MQTT protocol to collect data from multiple devices while using limited bandwidth and providing the information to many subscribers. As a result, the system is relatively simple to set up and provides a useful network interconnect protocol for device-to-cloud solutions. In addition to a lightweight and flexible interconnect, device-to-cloud solutions need a device application framework built so that connecting to the cloud and adding the business logic is the simplest part of the final product development.
In a device-to-cloud solution, the application framework should include all of the functions and utilities vital to making a product platform useful for the end application as well as for the long term management of the product as it matures and evolves. In most cases this application framework includes functionality such as: • Device management (OS updates, application updates, configuration management) • Application abstraction (running multiple applications or plug-ins in the same environment) • Application log management (Dynamic log level, Log file rotates) • Application messaging infrastructure • Application event infrastructure • Application module management (Start/Stop/Unload/Load) • Runtime statistics and meta data • Configuration management • Version management Fortunately, the open source community has provided an application framework that performs these functions and more. The specification to this framework has been provided by the OSGi Alliance. The original OSGi specification was completed about 10 years ago as the set-top box industry was struggling to find a common framework to run multiple applications on a single box while securely and reliably managing these applications within the embedded device environment. The OSGi framework can run on any Java platform, and OSGi technology adopters benefit from improved time-tomarket and reduced development costs because OSGi technology provides for the integration of pre-built and pre-tested component subsystems.
Choosing a Cloud Service
Developers can connect to the cloud over their own IT infrastructure, they can interface with public cloud computing companies, or they can work with a third party to deploy a hybrid solution of private and public cloud capabilities to connect a device. Developers should choose a type of cloudbased service depending on the end application and business logic that will be added to the solution, as well as their own business processes and in-house IT skill sets.
technology connected
The ideal device cloud solution processes data coming from any device through a cloud computing solution that offers high availability, scalability and complete data security. By leveraging end-to-end solutions now becoming available in the embedded industry, customers can deploy their embedded devices and start receiving data immediately without the need to create provision and maintain a costly IT infrastructure.
A Device Cloud Use Case
The Device Cloud has been implemented in a distributed retail application for a company that has 50,000 machines at retail locations worldwide. In the past few years, it has become challenging and unreliable for the company to use phone lines to collect data from their machines at retail locations. The cost to maintain a phone line and modem connection at retail locations continues to escalate, and uptime and availability decreases while equipment failures continually require technicians to visit retail sites and download data
manually—a very costly, slow and labor-intensive practice. The original plan was to contract with multiple vendors to develop a new embedded system device to transmit and receive data with a wireless connection. The device development was projected to take just less than a year. Instead of building a new device, the client chose Eurotech’s Everyware Device Cloud, allowing them to access an integrated end-to-end platform consisting of hardware, software, cloud computing, data management and application integration. By working with an outside vendor, the client was able to focus on their core competencies instead of spending the time to research and develop their own wireless communications system in-house. By accessing, storing and transmitting the data through the cloud, the client did not have to expand its already-full data center. Now, any executive or retail manager can access data with a simple web connection. The client estimates they will receive a positive return on investment for the new data collection system in just one
year based solely on the cost savings from eliminating the modems and phone lines. One of the largest benefits of the new cloud system is that the client’s machines can output status errors in real time. As soon as the machine encounters an error, the client can diagnose the problem remotely and sometimes fix it without sending a technician. When a technician is necessary, they know what to expect and can bring the appropriate tools and replacement parts, making the most efficient use of resources. The client says, “We will see significant operational cost savings with the new system since we’ve calculated that truck rolls can cost between $200 and $300 per hour, depending on where the machines are located. Being able to diagnose a problem with the machines remotely, or make only one trip with the right equipment, saves time and money.” Eurotech Columbia, MD. (301) 490-4007. [www.eurotech.com].
VersaLogic Tiger, Strength & Performance! Low power Intel Atom Z5xx processor on a PC/104-Plus form factor VersaLogic’s Tiger is a compact single board computer on a rugged 3.6” x 4.5” PC/104-Plus form factor. Featuring the low power Intel® Atom™ Z5xx (Menlow XL) processor, Tiger packs powerful 1.6 GHz performance backed by legendary VersaLogic quality. Available in both commercial (0º to +60ºC) and industrial (-40º to +85ºC ) temperature versions! Q Q Q Q Q Q Q Q
Intel Atom Z5xx processor up to 1.6 GHz Low power, 6W (typical) High-performance video and HD audio Gigabit Ethernet Up to 2 GB DDR2 RAM PCI & ISA expansion Fanless operation Industrial temp. (-40º to +85ºC) version
Add VersaLogic’s long-term (5+ year) product availability guarantee and customization options and feel the power of the Tiger! With more than 30 years experience delivering extraordinary support and on-time delivery, VersaLogic has perfected the art of service, one customer at a time. Experience it for yourself. Call 800-824-3163 for more information!
Recipient of the VDC Platinum Vendor Award for five years running!
800-824-3163 | 541-485-8575 | www.VersaLogic.com / tiger
Untitled-5 1
23
2:32:20 PM RTC MAGAZINE 5/9/11 MAY 2011
ploration your goal k directly age, the source. ology, d products
technology in
systems
Hypervisors, RTOSs and Multicore
Embedded Virtualization on x86: A Technical Look at Approaches and Solutions Multicore CPUs are causing virtualization techniques to be reevaluated and tuned to fit the needs of real-time embedded systems. For this there are options and tradeoffs, and a careful examination of the issues is needed to make the right decisions. by Timo Kühn, Real-Time Systems
W
ith virtualization becoming an increasingly widely used techApplication Application nology in embedded systems, it is helpful to take a look at some of the “virtualization-specific” terminology and the special aspects of virtualization when used in industrial and other embedded Guest OS Guest OS systems. A hypervisor is software utilizing Virtual Virtual virtualization to run one or multiple opHardware Hardware nies providing solutions now erating systems on one single computer. Other Drivers ion into products, technologies and companies. Whether is to research the latest Applications Hypervisor Instead of running directly onyour thegoal real ation Engineer, or jump to a company's technical page, the goal of Get Connected is to put you hardware, these so-called “guest operatyou require for whatever type of technology, systems” now execute in environand productsing you are searching for. Host Operating System ments provided by the hypervisor, called “virtual machines.” Physical Hardware A virtual machine may be part of the actual computer it is running on, but Figure 1 it may also be an environment completely Type II Hypervisor (host based). different from the real hardware. As hypervisors create and administer these alization solutions, there are still at least system,” as shown in Figure 1. From a virtual machines, a hypervisor may also two differentiating factors: the method of user’s point of view, the computer starts be called a “virtual machine manager” virtualization used and the classification up normally with its regular operating or VMM. While this is true for all virtu- of virtualization into Type I and Type II system, then launches one or multiple virhypervisors. tual machines in which further operating A “Type II hypervisor” executes systems (the guest operating systems) may Get Connected as a program within a regular operat- be started. The host operating system exwith companies mentioned in this article. ing system, the so-called “host operating ecutes directly on the hardware, not in a www.rtcmagazine.com/getconnected
End of Article
24
MAY 2011 RTC MAGAZINE
Get Connected with companies mentioned in this article.
tech in systems
virtual machine and does not need to be modified. This type of virtualization is therefore perfectly suited for home and desktop PCs. Still, the most common virtualization solutions are of Type I, also called “bare metal hypervisors.” A Type I hypervisor executes directly on the hardware and therefore does not require any host operating system (Figure 2). In the domain of server-virtualization and in virtualization solutions for embedded systems alike, there are some very good reasons for not having a host operating system. First of all a host operating system always presents a security risk, and second, the computing power consumed by the host OS reduces the guest operating system’s performance. The categorization into Type I and Type II hypervisors frequently gets criticized as not being specific enough, and as a matter of fact, some virtualization solutions don’t fit easily into either one of these two categories. Especially since there are hypervisors showing attributes of both Type I and Type II hypervisors, one has to look very closely to correctly assess and categorize a specific virtualization solution.
Types of Virtualization
One way to virtualize an unmodified operating system is through simulation. In this case a full computer is simulated in software. A simulator has to be able to interpret machine code and process this code in software like a real processor would. This includes all instructions, not only a subset of privileged instructions. This approach is very computing intense and therefore very slow. The great benefit of this solution is the possibility to execute and debug an operating system that was designed for certain hardware on completely different processor architectures. In paravirtualization, the code of the guest operating system itself is modified to meet above requirements. This could mean that in the simplest case, the inconvenient parts of the code are simply worked around. Wherever this is not possible, the instruction or code in question is replaced by a call into the hypervisor. The result is that at runtime the modified guest operating system code will lead to a
Application
Application
Application
Application
Guest OS
Guest OS
Guest OS
Guest OS
Virtual Hardware
Virtual Hardware
Virtual Hardware
Virtual Hardware
Hypervisor Physical Hardware Figure 2 Type I Hypervisor (bare metal).
switch into the hypervisor. There the original intention of the guest operating system is analyzed and then the instruction is virtualized. Paravirtualization always implies that the guest operating system is modified, and for this the source code of the OS, or at least part of it, has to be available. In case the source code is not available there is, of course, still a chance to then binary patch critical code. Through reverse engineering it has to be detected where and in which files the machine code that has to be modified is located. This code is then replaced by modified code, either by patching the original files or during runtime after the files have already been loaded. Still, binary patches are critical as reverse engineering never guarantees that all critical code sections really have been patched, and updates of operating system files may cause a maintenance nightmare. Independent of these technical aspects, usually vendors of operating systems like Microsoft do not allow their code to be modified and / or reverse engineered, causing legal uncertainties. If an operating system is to be virtualized without any modifications, full virtualization has to be used. The guest operating system executes most of its code on the real hardware processor(s) and then only if the operating system is trying to execute code that might have system wide impact, does the hypervisor step in. In other words, if the guest operating system
wants to execute machine instructions that are potentially dangerous either for the hypervisor itself or for other guest operating systems, the hypervisor catches those instructions and then virtualizes them according to the properties of the virtual machine. This can be achieved, for example, by executing the guest operating system on a privilege level below the hypervisor’s privilege level. Critical instructions are then configured to require the higher privilege level, therefore causing an exception or exit when trying to be executed by the guest operating system without the necessary rights. The hypervisor then gets to process this exception and virtualizes these instructions without the knowledge of the guest operating system. Before hardware supported virtualization technology (Intel VT-x, AMD-v) was available, the only possibility to fully virtualize an operating system on the x86 processors was the so-called “ring shift.” In this case the “kernel mode” of the guest operating system was pushed down at least one privilege level (also called “RING”). After this, all machine instructions that are only allowed in kernel mode, lead to an exception. All exceptions end up in the hypervisor and are then processed before the guest operating system is allowed to continue its normal operation without even knowing it was virtualized. Some disadvantages of the ring shift approach is that not all instructions that need to be handled by the hypervisor actually lead to RTC MAGAZINE MAY 2011
25
an exception, and many instructions that don’t need the hypervisors attention do cause exceptions. Hardware-based Virtualization Technology (VT) is a processor feature providing a number of advantages to virtualization software. From a technical point of view, VT is a new processor mode along with a few new instructions needed to use this functionality. A hypervisor can configure the execution of a guest operating system in such a way that certain machine instructions lead to a switch into the hypervisor. This is called a “VM exit.” The hypervisor defines which instructions and events should cause an exit. Handling of the exits, i.e., providing the virtualization itself, still has to be provided by the VMM software. This means that VT itself does not provide virtualization but it provides some advantages to the hypervisor, making virtualization easier, more secure and faster if used correctly. These advantages are very valuable for full virtualization, but whether VT is used for paravirtualization or full virtualization or for both is up to the hypervisor. Using VT therefore does not necessarily mean that operating systems are fully virtualized.
Security and Safety
Virtualization adds security or safety. This is something that can be read quite frequently but usually no explanation comes with this statement. What kind of security or safety? What has to be secured or what has to be safe? In the domain of desktop and server virtualization, these questions usually can be answered rather quickly. If multiple general purpose operating systems are run, it simply has to be guaranteed that they cannot hurt each other or the hypervisor itself. In an embedded or especially in an industrial environment, this has to be looked into in more detail. Usually so-called real-time operating systems (RTOSs) providing certain functionalities at top priority are in use. An RTOS is usually a closed system with very few interfaces to the outside world and usually cannot be changed or modified by the end user of a system, therefore presenting a very limited security risk. Desktop operating systems, also known as general purpose operating sys-
26
MAY 2011 RTC MAGAZINE
Context
Tech In Systems
Guest Virtualization Computing Time Overhead
Guest 1
Guest 2
Hypervisor
Launch Virtual Machine
Figure 3
Resume Trap, Virtual Machine Exit, Exception Virtualization Software
Time
Distribution of computing time might be unpredictable for guest OS.
tems (GPOSs), however, are a completely different story. The operating systems themselves are not necessarily unsafe or not secure, but end users may transfer data, install updates or third party software and device drivers, and very often access to local networks or even the Internet is possible, therefore presenting a great risk factor. For the above reasons it becomes obvious that it makes sense to protect the real-time operating system from general purpose operating systems if they are running on the same physical machine. This also means that a safe Type II hypervisor could only be built on a closed and safe host operating system. Going the other route, utilizing a GPOS like Windows as a host operating system unfortunately cannot provide a safe system as the GPOS in this case runs directly on the hardware with no possible way of control. Errors in the GPOS such as bugs in device drivers or malware can crash the system or at least cause some unpredictable behavior, potentially also affecting the real-time operating systems running in parallel. This also means that in case of a targeted attack, access to the RTOS cannot be blocked, which could result in a system failure or even industrial espionage. Ever since the Stuxnet computer worm attacking industrial controllers was in the news,
it became obvious that security aspects are important issues to consider when choosing a hypervisor solution. With a Type I hypervisor this is much easier. A Type I hypervisor does not come with the functionality of a full operating system and for that reason already is not as vulnerable. Still, the type of hypervisor alone is not the only factor when it comes to security. Just as important is the way virtualization is applied. If for instance Microsoft Windows is running alongside an RTOS that needs to be protected, then the execution of Windows including all its device drivers and applications needs to be monitored very closely to keep the RTOS safe. Protecting a guest operating system is therefore achieved by keeping a close eye on others. In order to virtualize a guest operating system in such a way that it cannot impair another guest and to prevent its instructions from having a system wide impact, paravirtualization is not the way to go. Only when using full virtualization can a hypervisor completely control an operating system, minimizing risk. Still, the hypervisor has to actively provide protection, security and safety. Simply running multiple operating systems does not say anything about safety or security, even if all of them are fully virtualized. Simply using hardware-based VT does not guar-
tech in systems
antee safety or security. How safe a technology utilizing VT really is also depends on how VT is used and if all possibilities for monitoring and control are in use.
Real Time and Virtualization
If real-time operating systems need to be used on virtualization, then the access to real hardware is indispensible. The physical hardware with a Type II hypervisor is controlled by the host operating system. This means that either the RTOS itself has to be extended to serve as a host and therefore it has to perform its real-time tasks and provide virtualization at the same time, or the RTOS runs as a guest and the host operating system assigns part of the hardware to the RTOS. For this the host operating system usually has to either come from the vendor of the virtualization solution or, if it is a GPOS, it has to be modified extensively. In the last case, where the host operating system either has to be newly compiled or even binarily patched, the solution is usually not called a hypervisor but a “real-time extension.” In the case of a Type I hypervisor, a host operating system is not present and therefore not an issue. If an RTOS is to be used as a guest on a Type I hypervisor, the type of virtualization used is very critical when trying to preserve the behavior of the real-time operating system. One of the main reasons for using an RTOS is the fact that it is deterministic. If full virtualization were to be used to run an unmodified RTOS, the determinism would be greatly affected, influencing latencies and jitter. One of the reasons for losing determinism is because developers of device drivers and applications cannot know which code and which events will lead to an exit or switch into the hypervisor and how long this will take. Figure 4 illustrates the concept of a hypervisor virtualizing two guest operating systems on the same CPU. If an RTOS had to execute fully virtualized and provide determinism at the same time, it would have to be specifically developed for a given virtual machine. Using paravirtualization instead is much more suitable for real-time operating systems. Because the hypervisor only gets to execute code if explicitly called from
the guest operating system, the behavior remains transparent and deterministic. This type of “cooperative” virtualization, by directly calling the hypervisor, also reduces virtualization overhead as the reason for the switch into the hypervisor does not have to be interpreted first. Besides full virtualization and paravirtualization there is of course also the possibility to have a mix of the two implemented in a hypervisor, showing more or
Untitled-1 1
less of the effects described above. And again, if hardware-based VT has been used in the design, it not may or may not affect the real-time behavior, depending on how it is being used. Real-Time Systems Ravensburg, Germany. +49 (0) 751 359 558 – 0. [www.real-time-systems.com].
27
9:23:49 AM RTC MAGAZINE 4/12/11 MAY 2011
ploration your goal k directly age, the source. ology, d products
technology in
systems
Hypervisors, RTOSs and Multicore
Embedded Virtualization Meets Real-Time Needs in Multi-OS Systems The advent of multicore processors has accelerated the use of virtualization to run general purpose OSs on the same die with RTOSs and to run other OS environments over multiple cores. Just what approach you take to virtualization can make a big difference in the results. by Christophe Grujon, TenAsys
E
mbedded systems often employ multiple computing platforms with RTOS Application HMI each running different operating Mem Mem Application systems. Typically, there is a real-time operating system (RTOS) on one platform to RTOS perform time-dependent processing, and a Hyper Hyper Windows general purpose operating system (GPOS) -thread -thread Processor on another platform running other appliVMM cations such as a human-machine interface (HMI). The complexity and cost of nies providing solutions nowsuch multiplatform systems implementing ion into products, companies. Whether goal is to research the latest can technologies be avoidedandthrough the use ofyour embedVirtual I/Os ation Engineer, or jump to a company's technical page, the goal of Get Connected is to put you ded virtualization. you require for whatever type of technology, Real-time and products you are searching for.systems need complete control of their environment in order to provide deterministic performance. UnReal I/Os Real I/Os like general purpose applications that are human or batch driven, real-time systems need to respond to events at prescribed Figure 1 times. In addition, since the majority of the events occur externally to the comEmbedded Virtualization enables combining HMI and Control System onto one system. puter platform and are serviced through the platform’s I/O interface, the RTOS must have direct access and control of the relevant I/Os and associated tasks’ sched- I/O task can be serviced at the precisely uling priorities. In systems that combine prescribed moment. Every undetermined an RTOS and a GPOS on the same plat- delay in processing a time-sensitive event Get Connected form, the RTOS must have priority over translates into an inaccuracy for the realwith companies mentioned in this article. the GPOS in order to ensure that each time system. Therefore, the system must www.rtcmagazine.com/getconnected
End of Article
28
MAY 2011 RTC MAGAZINE
Get Connected with companies mentioned in this article.
tech in systems
BEFORE (Two Platforms) HMI System ...................... ...................... BIOS
Top of Memory
...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... Apps 1 GPOS
FFFF FFFFH
Top of Memory
0000 0000H
• partition interrupts and I/O address space so that I/Os are always serviced by the appropriate OS, without affecting the other OS
AFTER (One Platforms) Control System ...................... ...................... BIOS
FFFF FFFFH
...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... Apps 2 RTOS
0000 0000H
Combined System ...................... ...................... BIOS
FFFF FFFFH
Top of Memory ...................... ...................... VMM(R) ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... Apps 2 ...................... ...................... ...................... ...................... ...................... RTOS ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... ...................... Load ...................... ...................... ...................... ...................... Off-set ...................... Apps 1 0000 0000H GPOS
Figure 2 How a GPOS and an RTOS are mapped into memory when virtualized onto one platform.
not only address the fundamental realtime applications requirement but make sure that the GPOS running alongside the RTOS does not affect the RTOS in any way while executing its own applications satisfactorily. Embedded virtualization does this by partitioning the platform’s hardware resources (CPU, memory, I/O and interrupts) for dedicated service by either the RTOS or the GPOS, depending on the application. This partitioning avoids the unpredictable response that comes from virtualizing hardware under the control of a virtual machine manager (VMM) that is totally disassociated from the guest RTOS. Embedded virtualization differs substantially from the virtualization used by servers, which is designed to maximize usage of the platform and do so by providing a complete set of virtual devices to each of the hosted OSs, with no concern for hard real-time service or direct access to specific I/Os (Figure 1). In addition to enabling the consolidation of two-platform systems into one,
embedded virtualization also opens the door to OEMs who want to improve their control applications by adding advanced HMIs without incurring the expense of adding another computer box, and without having to modify the real-time application code to interface with it.
Embedded Virtualization – What Are the Choices?
Over the past 20 years, two different approaches to embedded virtualization have been developed. They can be broadly divided into two categories: systems based on paravirtualization and ones based on virtual machine managers (VMMs). Both types of solutions provide the following core functions: • assigning scheduling priority to the RTOS • partition system memory such that the RTOS and the GPOS can run side-by-side without corrupting each other • partition I/Os so that the RTOS has complete control over the I/Os that it needs to access
Paravirtualization solutions have been in use the longest. They use software techniques to modify the OSs in order to allow them to work side-by-side without affecting each other or compromising the real-time responsiveness of the system. Implementations provide varying degrees of platform partitioning and have typically been limited to running two OSs at a time on a platform. Some implementations have evolved to the point where the GPOS doesn’t require any modification and the latest version of the GPOS is readily supported. This is a real plus when the object of coupling the GPOS to the RTOS is to be able to make use of the latest software. Over the years some of these implementations have been optimized to provide the best performance for a particular combination of OSs. The downside to this is that each implementation is specific to the particular combination of OSs, and it is an impractical approach to providing a generic virtualization solution to support multiple OS combinations. In VMM-based virtualization, the VMM manages the platform and provides an environment in which the virtualized OSs run. Not all VMMs are designed to support embedded virtualization, however, with the ability to run real-time applications on an RTOS. Early VMM implementations employed paravirtualization techniques to provide a virtual platform. More recently, hardware-assisted virtualization features like VT (supported by Intel processors) have become available and do away with some of the software complexity by providing hardware assistance that is built into the processor. By using the hardware virtualization support, a VMM can be constructed to function without knowledge of the guest operating system. As a result, the VMM can support any OS that is targeted for that platform. Intel processor features like VT-x (a subset of the VT features) ensure that any memory address RTC MAGAZINE MAY 2011
29
Tech In Systems
Embedded Virtualization, GOBSnet and Multicore Processor Partitioned Memory APP.1a
APP.1b
OS
APP.2
APP.3
APP.4
OS
OS
OS
CPU 2
CPU 3
CPU 4
t Sne GOB CPU 1
Allocated I/O’s
Figure 3 Embedded virtualization techniques can partition a platform to provide AMP like characteristics and global object network allows processes from partitioned and distributed applications across the CPUs in the multicore processor to communicate with each other.
that is issued by a guest OS is automatically mapped to the appropriate address location in physical memory (Figure 2). Likewise, the hardware assisted virtualization feature called VT-d automatically maps I/O memory accesses for bus-master DMA devices, enabling native I/O drivers that are part of the guest RTOS application to be used without modification in the virtualized environment. The hardware assist features substantially reduce the complexity of a VMM and make embedded virtualization a more viable solution.
Embedded Virtualization in Multicore Systems
The advent of multicore processors has created a flurry of interest among OEMs of multiplatform systems, suggesting how they can consolidate their computing resources to run multiple OSs on a single platform, allocating separate CPU cores to specific guest OSs. Having multiple discrete cores available is a very powerful feature because it enables the segregation of the operation of each guest OS. This contributes to the operational integrity of each environment as well as guaranteeing
30
MAY 2011 RTC MAGAZINE
computing capacity for each OS. The same virtualization technologies that have been used to host multiple OSs on a single processor can be applied to multicore processor configurations, with the additional ability to allocate dedicated CPU core(s) to each OS. The multicore chip’s memory address space, I/Os and interrupts can be partitioned, with each partition assigned to one of the multiple installed OSs. This provides each OS kernel with its own dedicated processor platform (Figure 3). Applications can be allocated to dedicated CPU resources, and I/Os are serviced by the very CPU that is dedicated to run the corresponding application tasks. Other application threads running on different CPUs are not interrupted unnecessarily, thus optimizing the overall performance of the system.
Extending the Functionality of Embedded Virtualization
Distributing applications across a multicore CPU using embedded virtualization is straightforward, as long as the application can be contained on one core. There are applications however that would
benefit from being distributed across several processor cores by allowing application processes to interact across cores as if they were running on the same core. One way of doing this is by allowing communication-based programming objects such as mailboxes, semaphores and common memory blocks, to be accessed across OS and processor boundaries, essentially making those programming objects global. This opens the way for large applications that were resource-limited on uni-processor platforms to be partitioned and loaded on separate processors of a multicore processor without requiring fundamental changes to the application code. The implementation of global objects necessitates the implementation of a “network,” much like any network, by which processes can discover, set up and manage the inter-process communications. At TenAsys, the implementation called GOBSnet (Global Object Network) consists of the same basic building blocks that exist in its INtime OS, with the addition of a GOBS manager to keep track of the interOS references. Use of GOBSnet doesn’t require any rewrite of application code and is easily implemented with the support of a few initiation and location discovery commands. Processes requiring services of processes on another processor will find them automatically. Figure 3 shows how applications can be assigned to processor cores, and can span across cores—communicating via the global object network.
Searching for the Right Solution from an Application Perspective
Matching the right technology to a specific application can be daunting. There are a multitude of factors that need to be considered. Drawing out a decision flow chart is a possible endeavor but would be so complex to follow that it would be impractical—there would be most likely one combination that isn’t addressed. A better way is to highlight key questions that need to be answered and look for answers that are compatible with the requirements of the application. For example, consider the case of an OEM who is developing a new product with a GPOS HMI and real-time control system that are to be combined on one platform.
tech in systems
Sometimes the real-time core of an application is where a company’s true intellectual property (its “secret sauce”) lies. The software may have taken years to develop, optimize and perhaps certify. It would be extremely difficult to re-engineer it. In this case, using an embedded VMM solution to host legacy real-time software is the easiest solution. The VMM should allow the native application to run on the platform alongside a GPOS such as Windows without modification. Questions to ask when evaluating a VMM: 1. Does it support embedded virtualization? 2. Will it support upgraded versions of the GPOS? Long-life support of the OEM’s application could be an issue. 3. Does it meet the overall performance requirements of the application? VMMs do impact performance even when run on processors that support hardware assisted virtualization. 4. How much does it cost? Note that a VMM is a third operating software component (in addition to the OSs themselves) that adds costs to the overall solution.
who are looking to cost-reduce and simplify their product platforms by combining multiple operating environments using embedded virtualization techniques. It pays to look closely at available options and understand the tradeoffs. TenAsys Beaverton, OR. (503) 748-4720. [www.tenasys.com].
High-performance real-time applications may have a problem running in a VMM environment. The OEM system developer will have to test the application to see if it meets the necessary performance and allows headroom for future enhancements. Paravirtualized solutions generally perform better for a given processor and will probably give better price/performance results. Things to check when investigating a paravirtualized solution: 1. Does it support embedded virtualization? 2. What RTOS and GPOS combos does it support? 3. Will it support upgraded versions of the GPOS? Long life support of the application could be an issue. 4. What multicore support is there? By now it is obvious that not all embedded virtualization solutions are alike. There are many options available to OEMs Untitled-2 1
31
9:35:53 AM RTC MAGAZINE 5/9/11 MAY 2011
technology deployed New System Specifications
Modernizing Legacy Modular Systems Design with CompactPCI Serial
connections: up to 8 PCI Express, SATA, USB and Ethernet interfaces without a bridge (Figure 1).
Evolving CompactPCI standards
Rugged embedded computing applications in industrial automation, instrumentation, telecommunications, transportation systems and real-time data acquisition systems have benefitted from the reliable CompactPCI bus interface since the mid 1990s. However, application requirements and technological advances in the last 15 years have forced the components and systems needed, and thus the standards used, to evolve. The path leading from CompactA new serial specification for CompactPCI adds a host of PCI to CompactPCI Serial developed out of these adjustments to system design. fast serial backplane interfaces via an additional connector exploration The parallel bus architecture of the r your goal while maintaining backward compatibility with legacy original CompactPCI standard (PICMG eak directly 2.0) has recently become a limiting factor page, the CompactPCI systems and modules. when expanded input/output (I/O) perforresource. hnology, mance is required. The first step in the evond products lution of the basic CompactPCI standard by Barbara Schmitz, MEN Mikro Elektronik aimed to supplement the original standard by remaining true to the mechanical parameters (PICMG 2.0), while still providing expanded opportunities for serial communications. ecause CompactPCI is one of the most successful modular system standards, it’s no surprise that the standard has been continually adapted to newer specifications as technology panies providing solutions nowrequirements become more advanced. The latest and application ation into products, technologies and companies. Serial Whether (PICMG your goal is to research the latest specification, CompactPCI CPCI-S.0), ratified in cation Engineer, or jump to a company's technical page, the goal of Get Connected is to put you March 2011, allows new systems to be built using time-proven ce you require for whatever type of technology, CompactPCI technology that incorporates the performance ades and products you are searching for. vantages of serial communications to meet all the requirements Figure 1 of modern modular systems without the increased costs of other Integral to CompactPCI Serial’s performance alternative standards. advantages is the new high-speed connector, available CompactPCI Serial is a recently approved PCI Industrial in different formats to fit specific design parameters. Computer Manufacturers Group (PICMG) standard for 3U and 6U boards in IEC 1101-compatible 19” systems. Like the CompactPCI PlusIO standard, ratified in January 2010, which provides This resulted in CompactPCI PlusIO (PICMG 2.30). Coma migration path for legacy CompactPCI systems to incorporate pactPCI PlusIO allows parallel and serial connections in system serial communication, CompactPCI Serial ensures the reliability designs so designers can take advantage of new serial technology, of the CompactPCI bus interface well into the future. while still using existing legacy technology in the same system. While the dimensions of the backplanes, front panels and It essentially protects the investments companies have made in handles in addition to the well-proven hot plug mechanics of their CompactPCI systems by bringing the older architecture in Get Connected CompactPCI Serial are the same as for CompactPCI, a major line with current application demands. with companies mentioned in this article. new design element is the connector. With a considerably higher This interim step between the parallel backplane and full serialwww.rtcmagazine.com/getconnected signal density and faster transmission frequencies of up to 12 only functionality allowed system designers to save money and time Gbit/s, the new connector supports only modern point-to-point by using the original CompactPCI boards and other devices that were still operational. CompactPCI Serial addresses entirely new systems being built to provide the greater bandwidth and higher data transfer Get Connected with companies mentioned in this article. rates required by increasingly demanding applications. www.rtcmagazine.com/getconnected
B
End of Article
32
MONTH MAY 2011 2011RTCRTC MAGAZINE MAGAZINE
Technology deployed
Figure 2 Each CPU board can be plugged into every peripheral slot because each slot is identical.
Enter CompactPCI Serial
Because its roots lie in a technology with broad market applicability, CompactPCI Serial covers a wide range of applications across several embedded markets. Not only did all technical options need to be considered in the standard’s development, but these features also had to remain cost-effective, one of the primary benefits of the original CompactPCI technology. Important to note is that CompactPCI Serial remains backward compatible to CompactPCI PlusIO and CompactPCI, so it can interface with older systems, too. The result is a standard that is as suited for a simple industry PC as it is for a complex control system up to redundant systems and highly complex computer clusters. Existing solutions from the consumer market can be adapted very easily to CompactPCI PlusIO and CompactPCI Serial. For example, an ordinary 2.5” or 3.5” hard disk can be fitted to a CompactPCI Serial plug-in board and linked into the system via SATA without any protocol adaptation. Similarly, a USB flash memory drive or a WiFi stick can be mounted on a board and linked to the processor board via the USB bus present on every slot. The possibilities are enormous and the user can benefit greatly from the existing PC market. And, CompactPCI Serial can be applied in harsh as well as safety-critical environments in addition to these many sub-markets and all areas with electronics in mobile applications, especially on rail and road, in ships and avionics, and research and development.
Expanding the Ecosystem
Emphasis in the development of this new architecture has been placed on retaining the CompactPCI ecosystem that requires considerations for not only hardware but also for additional factors, such as the knowledge and experience of system designers. The standard has been deliberately developed around the familiar 19-inch technology. Mechanically speaking, this allows both “old” parallel boards and new serial boards to operate side by side in the same chassis. Although VITA 46 VPX is also based on the industry-standard 19-inch platform and incorporates high-speed interfaces, the complexity of its implementation may preclude its use in
Figure 3 New SBCs built on CompactPCI Serial provide a host of connection options.
some environments, especially in cost-sensitive industrial applications. The alternative of using the more cost-effective CompactPCI Serial technology is now a reality. Its attributes, such as the high-speed capabilities available through the new connector, set the stage for additional uses of this technology into applications where CompactPCI would traditionally have failed. But the connector does more than just provide CompactPCI with the benefit of faster transmission. It saves significant costs as a component itself as well as provides increased robustness to enhance system reliability, enabling CompactPCI Serial to be used in more harsh applications than this platform previously allowed. Additional cost-savings are seen in system integration, since the architecture does not require a separate switch fabric infrastructure, greatly simplifying the implementation of CompactPCI Serial.
Examining the Inner Workings
Any system built needs to have its parameters defined for the specific application, and a designer using CompactPCI Serial will face a similar set of questions. Considerations need to be made to determine which serial interconnect is the best to use, such as Ethernet for star and mesh architectures, SATA for RAID systems, USB for wireless connection, PCI Express for super-fast graphics and so on. The simple star topology used in CompactPCI Serial eliminates the need for bridges or switches in a system with up to nine slots, because the system slot supports up to eight peripheral slots. In principle all peripheral slots are identical. Only two are connected via extra wide PCI Express links (Fat Pipes), which can be used for high-end graphics extensions, for example (Figure 2). The 19” racks used by CompactPCI Serial are COTS, as are the PSUs and fans. The backplanes are standard components in most RTC MAGAZINE MAY 2011
33
technology deployed
cases and have two to nine slots or are offered as hybrid backplanes (CompactPCI Serial – CompactPCI PlusIO – CompactPCI). System slot boards come with the latest Intel architecture or, in some cases, with PowerPC. Mezzanine cards offer the flexibility for Ethernet mesh systems. Peripheral slot boards provide I/O variety: hard disk shuttles, XMC and PCI Express MiniCard carriers, USB, fiber optic and audio interfaces, Gigabit Ethernet interfaces
and switches. The recently released G20 CompactPCI Serial SBC (Figure 4) from MEN Micro, for example, offers a multitude of serial interfaces. Standard front I/O includes two PCIedriven Gigabit Ethernet and two USB 2.0 interfaces as well as two DisplayPorts that can be used as an HDMI or DVI connection via an external adapter. A total of eight PCI Express links in the front and back of the board enable fast communication.
For user-specific applications, the rear I/O also provides eight USB ports, six SATA ports, a Display or HDMI port as well as a PEG x8 port and five PCI Express x1 links (Figure 3). With CompactPCI Serial, the possibilities are endless. Systems, boards and backplanes are readily available for both the CompactPCI PlusIO and CompactPCI Serial specifications. In addition to the SBCs, this includes configurations such as 8-slot hybrid backplanes with three CompactPCI peripheral slots, a CompactPCI PlusIO system slot and four CompactPCI Serial peripheral slots. This maximum backplane configuration for both standards enables designers to use PlusIO and Serial in a wide number of applications. The purpose of CompactPCI Serial is not merely to define a new standard, but to offer a better solution for every application— today and in the future. The value of the standard will be determined by how CompactPCI Serial is implemented in useful products for the real applications. By evolving the CompactPCI standard and utilizing existing, dependable mechanics, CompactPCI Serial becomes a cost-effective, timesaving and modern technique for various rugged, mobile and industrial applications. By utilizing the increased functionality of CompactPCI Serial, system designers familiar with 3U and 6U CompactPCI boards can continue to improve their systems without the difficulties of adapting systems to a different standard. This includes applications as specific as image and data management and recording in surveillance systems, camera control systems, different possibilities for integration of wireless communication, audio data processing and computer simulation or a computer cluster in industrial quality control to name a few. As the migration path from CompactPCI to CompactPCI PlusIO to CompactPCI Serial indicates, the CompactPCI bus architecture will continue to evolve with the times to provide the necessary technological updates for embedded system designs well into the future. MEN Mikro Elektronik Ambler, PA. (215) 542-9575. [www.menmicro.com].
34
Untitled-3 1
MAY 2011 RTC MAGAZINE
5/11/11 10:29:06 AM
technology deployed New System Specifications
Rugged Memory Spec Raises the Bar for Rugged Modular Computing Securing memory modules against shock and vibration on small modules has been a challenge often involving epoxy and adhesives. Now a new standard specifies a proven method to secure memory modules to COM modules. by Markus Friese, Lippert Embedded Computers
T
he popularity of COM Express has positioned the modular computing standard for penetration into more demanding markets such as intelligent transportation, military ground vehicles, aircraft and avionics, outdoor mobile communications, imaging, facial recognition as well as shipboard and marine computing. Modules are widely available with mid-range to high-performance embedded processors and chipsets. For a variety of reasons including costs and inventory management, the vast majority of COM Express modules are designed with consumer SODIMM RAM sockets. As far as rugged application suitability is concerned, the use of SODIMMs has been a drawback compared to the many low-end to mid-range PC/104 SBCs that utilize soldered memory. The ruggedness of RAM chips soldered directly on a host SBC or computer-onmodule (COM) is undeniable, as established by numerous shock and vibration test reports and many hundreds of rugged application deployments over the years. However, system OEMs in these markets who want access to the latest dual core processors have few choices other than ad hoc ways of retaining commercial-grade RAM sticks into their sockets.
Brute Force
The most popular practice for securing SODIMMs against high shock and vibration is applying epoxy everywhere the module touches its socket. Compounds such as RTV silicone adhesives and sealants and epoxy encapsulants are well established for potting
36
MONTH MAY 2011 2011RTCRTC MAGAZINE MAGAZINE
components and attaching cables to header connectors, so holding card-edge boards into sockets with epoxy seems reasonable. Breaking the epoxy seal is not an issue as long as the SODIMM manufacturer is lenient when accepting warranty returns. Other methods in use range from clips to brackets to clamps to screws to thermally attaching to the same heat spreader that cools the dual-core CPU. Of course, exposing system memory to tens of watts of heat doesn’t improve the long-term reliability of the RAM. Concerned about contaminants and corrosion 10-20 years after deployment, some system OEMs have even asked their embedded computer suppliers for evidence that the gold plating won’t wear out due to micro-etching of gold fingers against socket pins.
The Pursuit of Ruggedness
System manufacturers targeting harsh environments with their high-performance applications should have the freedom to select COM Express without fancy fastening, fixturing and finger-crossing. Enter RS-DIMM—a new multi-vendor standard from the Small Form Factor Special Interest Group (SFFSIG). The primary goal in creating this standard was to address the need for reliable memory by using a proven board-to-board connector pair methodology with the additional support of mounting holes (Figure 1). As early as 10 years ago, the need for rugged RAM drove some PC/104 CPU manufacturers to design memory modules when large processors and chipsets prevented soldered RAM chips from fitting on board. But these were sole-sourced proprietary RAM modules. RS-DIMM solves this problem two ways— with multiple manufacturers out of the gate, and with a neutral and impartial standards organization responsible for the longterm maintenance, support and revision control of the governing specification. Swissbit is already in production with the module, and Virtium is prepared to begin production shortly. Having two or more sources is usually important to reduce supply risks. The RS-DIMM Specification defines a small 67.5 mm x 38 mm module that stacks 7.36 mm above the CPU board. Ruggedness is achieved by using a 240-pin Samtec BTH/BSH connector pair on the memory module and the CPU board, along with two mounting holes. It fits better on a number of small form factor CPU boards than the wider 72 mm SODIMM modules. Finally, the pin definition for RS-DIMM closely aligns with the SODIMM pin definition, making it straightforward to migrate an existing CPU board to RS-DIMM (Figure 2). Aside from signal integrity, the salient feature of the Samtec connector pair is the gas-tight connection resulting from adequate vertical wiping during installation. Gold-plated card edge fingers snapping into a consumer socket don’t come close to the
Technology deployed
same level of air-free contact assurance.
Performance with Data Integrity
The first commercially available RS-DIMM module was shown by Swissbit (Figure 1) during the Embedded World 2011
RS-DIMM Module from Swissbit. 67,50 + 0,10
-
2x
ø3
,20 +
- 0,1 0
-
5,50 + 0,05
0,23 + 0,05
-
38 + 0,10
2,80
1,20 2,80 +- 0,05
Padsize 0,28 + 0,05 Pitch 0,50 + 0,05
-
62,30 65,10 + 0,05
-
0,05 02 +ø 1,
A Delivery Vehicle
Figure 1
2x
RS-DIMM uses DDR3 technology and is specified for both unbuffered and registered implementations. DDR3 is the state-of-the-art high-performance RAM technology for two of Intel’s three embedded roadmaps: the high-end Core family (Core i7), and now the mid-range roadmap featuring Atom D425 and D525 “Luna Pier Refresh” dual core Atom processors. Last year, the mid-range N270, N450, D410 and D510 processors still featured DDR2 memory, but D425 and D525 processors are replacing them for new designs. The only remaining Intel embedded roadmap using DDR2 memory is the Atom Z-series and its successor Atom E-series processors. These designs are typically provided in small form factor modules with soldered RAM, such as the tiny CoreExpress modules. Consequently, the need for a rugged DDR2 module is rapidly vanishing from the market. Memory sizes up to 4 Gbyte are currently supported on RSDIMM, with optional ECC (error correction circuitry) supported using either 9-chip or 18-chip designs. ECC provides an extra level of data protection for high-reliability applications. Intel has integrated ECC support into Core i7 memory controllers, but not into Atom memory controllers. It’s not hard to see the advantages of a mated connector pair over consumer card-edge-socket schemes. But the proper engineering verification approach is to select a test methodology along with limits that are meaningful and suitably representative of the rugged applications. For 15 years, the stackable processor and I/O community has used 12 Grms vibration and 50 Grms shock in both directions along each axis. This is often referred to as a 6-axis test. ANSI/ VITA 47-2005 (R2007) provides the necessary framework. According to the VITA organization, this specification defines environmental, design and construction, safety and quality requirements for commercial-off-the-shelf (COTS) plug-in units (cards, modules, etc.) intended for mobile applications. COTS plug-in units are widely used in commercial, military, ground, aerospace and mobile applications. Certification of COTS plugin units, by supplying vendors, to this standard will facilitate the cost-effective integration of these items in larger systems. VITA 47 V3 and OS2 Tests were performed to establish the ruggedness of the memory module. To ensure the integrity of the tests, an EPIC form factor SBC was used for the testing to eliminate any chance of COM Express connectors and mounting affecting the test results. During the V3 tests, three boards passed the random vibration profiles. The RS-DIMM modules were tested over vibration frequencies from 5 to 2000 Hertz. In the OS2 half sine wave shock tests, the DVI display adapter cover (shroud) detached from one of the carrier boards, but that unit still functioned properly. Both boards passed this six-axis shock testing (three orthogonal directions, positive and negative excursions). In addition, four shocks were performed in the form of a bench drop shock test, and the units under test (UUT) passed.
Figure 2 RS-DMM module diagram.
show in Germany. The first COM Express module to support this new RAM standard with ECC was demonstrated during the same show (Figure 3). Lippert’s Toucan-QM57 is a high-end COM Express revision 2.0 Type 2 module with an Intel Core i7 processor. The module is specifically built for applications exposed to rugged environments. This is underscored by the integration of the highly rugged RS-DIMM memory module. The memory module is fastened by screws. The raw computation performance of the processor and DDR3 RAM lends itself to image processing, video encoding, communications and other demanding tasks. The COM Express module offers 4 Gbytes of soldered RAM and another 4 Gbytes of RAM by way of RS-DIMM card, which optimizes board space and ruggedness. With the 1.06 GHz dual-core Core i7-620UE processor, the module operates over -40° to +85°C for military, energy and other RTC MAGAZINE MAY 2011
37
technology deployed
Figure 3 The Toucan-QM57 supports both soldered and RS-DIMM memory (right edge).
outdoor environments. The 2.53 GHz dual-core Core i7-610E processor model is rated to +60°C for avionics, transportation, marine, imaging, communications and some automation applications. For these applications, additional embedded features add to the rugged reliability. Lippert Enhanced Management Technology (LEMT) provides condition monitoring and supervision capabilities with a convenient API for popular operating systems.
A Graceful Landing
Three off-the-shelf production CPU boards on their heatspreaders were subjected to stringent VITA 47 testing. Apart from a loose screw and a DVI connector cover, which were considered to be non-critical, the RS-DIMM modules all performed brilliantly. RS-DIMM has been proven to withstand extreme shock up
to 50 Grms and vibration up to 12 Grms along three orthogonal axes. These amplitudes are well established in these target markets over the years, as is the VITA 47 test methodology. Due to the use of board-to-board mated connectors rather than commercialgrade card edge gold fingers, RS-DIMM allows even COM Express modules with high-performance Intel Core i7 processors to be ruggedized for the stringent requirements of outdoor fixed and mobile applications. Standards from three trade groups—VITA, PICMG and SFF-SIG—were tapped to generate this best-in-class modular solution. RS-DIMM allows modular computing to attain a great level of rugged reliability without epoxy, brackets, clips or other support. LiPPERT Embedded Computers Mannheim, Germany. +49 621 4 32 14-0. [www.lippertembedded.com]. Swissbit Bronschhofen, Switzerland. +41.71.913.03.03. [www.swissbit.com]. Virtium Technology Santa Margarita, CA (949) 888-2444. [www.virtium.com].
Marvell Armada300 System On a Module The CSB1724, designed, developed and manufactured by Cogent Computer Systems, Inc., is a high performance, low-power, ARMADA 300 (Sheeva ARMv5TE) based System on a Module (SOM). The CSB1724 provides a small, powerful and flexible engine for Embedded Linux based Gigabit networking and storage applications. y y y y y y y y y
1.6Ghz 88F6282 Sheeva ARMv5TE Core 16KByte I/D Caches; 256KByte L2 Cache 512MByte DDR2-800 and 512MByte SLC NAND Two PCIe x1, Two SATA Gen2 and Two USB 2.0 Two 10/100/1000 Twisted Pair Copper Ports On-Chip Crypto and Security Engines with XOR Dual SATA Gen 2 and Dual 480Mbit USB 2.0 Ports 4-Bit SD/MMC, 2-wire TTL UART, I2C, SPI and I2S 3W typ., 4W Max, <10mw Power Down
Now available at www.digikey.com The CSB1724 is manufactured in-house on our IPC-610 Certified, lead-free capable surface mount line. All products carry a 1-year warranty and are available in commercial and industrial temperature versions. Cogent offers standard and custom carrier boards, royalty free licensing options and more.
&2*(17 "ALWAYS COMPLETE"
38
Untitled-8 1
MAY 2011 RTC MAGAZINE
&RJHQW &RPSXWHU 6\VWHPV ,QF ,QGXVWULDO 'ULYH 6PLWKILHOG 5, WHO ID[ ZHE ZZZ FRJFRPS FRP 4/28/11 11:15:23 AM
Solid-State Drives & Industrial Box PC Showcase Vortex Compact Network Storage
Featuring the latest in Solid-State Drives & Industrial Box PC technologies
Rugged high performance Network File Server Four 1GbE ports AES-256 Encryption capabilities Transfer throughput >100MB/s Supports CIFS, NFS, HTTP and FTP protocols Compact design 4.84” H x 4.88” W x 12.62” D http://www.cwcelectronicsystems.com/ cns_storage.html
Curtiss-Wright Controls Electronic Systems Phone: (937) 252-5601 Fax: (937) 252-1349
Vortex Data Transport System
TRRUST-Stor™ Solid State Disk
DZUS form factor Four 1 GbE ports NFS, CIFS, HTTP, FTP, PXE and SNMP protocols Weighs < 10 lbs Data encryption and key management Supports 3 Removable Memory Cartridges of 100 GB capacity http://www.cwcelectronicsystems.com/ vortex_dts.html
Curtiss-Wright Controls Electronic Systems Phone: (937) 252-5601 Fax: (937) 252-1349
E-mail: systeminfo@curtisswright.com Web: www.cwcelectronicsystems.com
With encryption, unparalleled ruggedization and blazing fast erase, Microsemi’s TRRUST-Stor™ is the first solid state drive designed for secure, high integrity data storage in military applications. TRRUST-Stor delivers reliability, performance, and security unmatched by current commercial SSD offerings. Features: 50 or 100 GB densities Hardware-implemented AES-256 encryption Multiple key management techniques Full disk erase in under 5 seconds
Microsemi Corporation (formerly White Electronic Designs) Phone: (602) 437-1520 or (888) SSD-4-MIL
Embedded Modem Modules, the Half-Inch Modems
Phone: (408) 383-9006 Fax: (408) 383-9007
E-mail: info@whiteedc.com Web: www.whiteedc.com/trrust-stor.html
Bargain Panel PCs 12V-Powered Touch-panel PC/ controllers. CUTOUCH CT1720™ - Basic-programmable, 82 x I/O, 8 x A/D, 6 PWM, for quick HMI. CUWIN3100/5200 7 - 10” color TFT ARM-based WinCE – backlit, Ethernet, from $299. CUPC-P70/220 7 – 15” Atom/Geode-based panel PCs for XPEmbedded/XPpro from $649. www.saelig.com
Serial TTL interface -40C to +85C operating temperature Compact size: 1” x 1” x 0.2” up to 56K bps data rate 14.4K bps fax, voice AT command DTMF, ring and Caller ID detection Transferable FCC68, CS03, CTR21 telecom certifications Global safety: c/UL, IEC60950-1, IEC60601-1 (Medical) approved CE marking
Radicom Research, Inc.
E-mail: systeminfo@curtisswright.com Web: www.cwcelectronicsystems.com
Saelig Company, Inc. E-mail: sales@radi.com Web: www.radi.com
Phone: (888) 772-3544 Fax: (585) 385-1768
E-mail: sales@saelig.com Web: saelig.com/category/IPPP.htm
INDUSTRY
WATCH
MicroTCA in Networks
MicroTCA Systems for the Evolving Wireless Infrastructure As wireless broadband spreads into less densely populated areas, MicroTCA can offer better economy and flexibility than ATCA solutions. by Tony Romero, Performance Technologies
T
he 4G wireless revolution is making significant market advances with major deployments worldwide. On the heels of this explosive growth in new service deployments is the opportunity for equipment manufacturers to develop smaller Long Term Evolutions (LTE) or WiMAX solutions for specific applications, such as deploying in developing nations, rural deployments, public safety, smart power grids, military, prisons, universities and more. Savvy manufacturers and service providers who get to market quickly with a fully functional and scalable system will have the competitive advantage. Using complete, pre-integrated LTE Evolved Packet Core and WiMAX ASN gateway solutions, equipment manufacturers can get to market more quickly and incur minimal development costs. Based on MicroTCA platforms, they can deliver solutions that are carrier-class, cost-effective, scalable, flexible and standards-based with the ability to be custom branded. With the demand for wireless broadband for mobile users, the LTE and WiMAX standards groups established several objectives intended to support bandwidth intensive applications such as video and online games, and to reliably deliver services such as interactive location-based services (LBS). The major objectives in-
40
MAY 2011 RTC MAGAZINE
MME
S1 X2
eNodeB
S11
S-GW & P-GW
HSS
Evolved Packet Core (EPC)
Xpress SIP Service Delivery Platform Element Management System (EMS)
IMS
Figure 1 LTE Network.
clude increased data rates, greater flexibility and simplified network architecture. The target for true 4G is data throughputs of 1 Gbit/s. Real-world data rates will vary based on numerous factors. For LTE, peak data rates support 100 Mbit/s for down-
loads and 50 Mbit/s for uploads. LTE Advanced, scheduled for release in 2011, will support 1 Gbit/s. For WiMAX, the mobile 802.16e standard delivers data rates as high as 40 Mbit/s, while the upcoming 802.16m standard is targeting 1 Gbit/s.
INDUSTRY WATCH
Service providers deploying networks in different regions around the world are looking for greater flexibility of spectrum usage. The radio spectrum for LTE ranges from 1.25 MHz to 20 MHz, and for WiMAX the spectrum ranges from 2 GHz to 66 GHz, the most common being 3.5 GHz. To simplify the network architecture, it is designed as a pure IP-based architecture, eliminating legacy protocols, such as GPRS, and providing simpler compatibility to IMS networks and interoperability between WiMAX and LTE. The three main categories of equipment that LTE and WiMAX bring to the wireless network are: 1) the User Equipment (UE) or Customer Premise Equipment (CPE), 2) the Radio Access equipment also called Base Stations or enodeB (for LTE), and 3) the Core Gateways, which provide multiple control and data services. Other objectives include reduced latency for establishing connections and transmission, plug and play operation and lower operating costs. There is a push to achieve seamless mobility. For LTE this includes support for different radio-access technologies such as GSM, UMTS and CDMA2000.
Applications Where MicroTCA Makes Sense
It is important to note that when describing gateways, one size does not fit all. Many of the early deployments have been targeted to large cities where the service providers can maximize revenue per deployment. ATCA systems with large banks of processors can provide the scale for these large deployments. But as service providers start to target the large numbers of mid-size cities, smaller towns and rural locations, the CAPEX rules change, and MicroTCA becomes a more cost-effective option. In addition, there are other smaller applications that open the door to good revenue-generating opportunities. In the examples below, it is imperative that the gateway be designed as a highly reliable system architecture that can be dropped into locations that may be less
R1 R3 R8
R6
AAA
CSN/Home Agent
ASN Gateway
Connectivity Service Network (CSN)
Connectivity Service Network (ASN) Base Stations
Xpress SIP Service Delivery Platform IMS Element Management System (EMS)
Figure 2 WiMAX Network.
LTE MME Solution [ Northbound to EMS ]
MME Application NAS (24.301)
Custom MIB (s)
GTP-C v1 S102/ A21 (29.277) Gn (29.060)
LTE EPC Component MIB Shelf Management MIB and Web-based Portal
DIAMETER
GTP-C v2 S10 (29.274) S11 (29.274)
DNS
S1-AP Sgs-AP SBc-AP
S3 (29.274) DDDS (36,413) (29,118) (29,168) Sv (29.280) (29.303)
S6a (29,272) S13 (29,272)
S101 (29.276)
SCTP
UDP
TCP
IP NexusWare Linux
Blade Manager Processor Platform manager Processor
Processor 1U PT-AMP5071 MicroTCA Platform (MCH, Power, Cooling)
Greenfield LTE deployment LTE deployment with overlay legacy UTRAN network
LTE deployment with overlay R8 UTRAN network LTE deployment with overlay CDMA network
Figure 3 Example MME Solution.
than hospitable. They must be easily serviceable, remotely manageable, and must be scalable to add more processing power when the subscriber count increases. 1U
MicroTCA-based systems with support for up to six AMC processors running carriergrade Linux, and gateway application software with full standby failover, meet these RTC MAGAZINE MAY 2011
41
INDUSTRY WATCH
LTE SGW and PGW Solution
[ Northbound to EMS ]
SGW/PGW Application RMPv6
GTP-C v1
S5 (29.275)
Custom MIB (s)
S2a (29.275)
Gn (29.080)
GTP-C v2
DIAMETER
S5 (29.274) S11 (29.274)
Gx (29.2120 SG (29.061) Rf (32.299) Ro (32.296)
S4 (29.274)
Gxc (29.212) S6b (29.273)
LTE EPC Component MIB
UDP
SCTP
Forwarding GTP-Uv1 (29.281) UDP
Shelf Management
GRE (S103)
IP
IP NexusWare Linux Processor
Platform Manager
1U PT-AMP5071 MicroTCA Platform (MCH, Power, Cooling)
Greenfield LTE deployment LTE deployment with overlay legacy UTRAN network
LTE deployment with overlay R8 UTRAN network LTE deployment with overlay CDMA network
Figure 4 Example SGW &amp; PGW Solution.
requirements. In fact, MicroTCA systems are already deployed as WiMAX gateways and are on trial for LTE EPC test environments. AMCs offer a right-sized granularity for these types of deployments, where the population count is not as dense as in large metropolitan cities. Developing Nations: Developing regions around the world are rapidly deploying mobile data services, including areas that have never had communications infrastructure in the past. Strong demand by consumers and competitive conditions by service providers are accelerating frequency spectrum auctions. Brazil’s telecom regulator, ANATEL, has recently designated the 2.5 GHZ band to support nationwide deployment of a neutral wireless broadband. Because of its neutrality, it can support either LTE or WIMAX deployments. The intent of these governments is to facilitate rapid economic development via broadband Internet access. Mobile data services in Latin America are expected to grow at a CAGR of 31 percent rate from 2010 to 2015, according to Pyramid Research. Rural Deployments: The American Recovery and Reinvestment Act (ARRA) of 2009, included $2.5 billion to increase the availability of broadband in rural areas of the U.S. This opens the doors for tier-2 and tier-3 service providers to install wireless networks and run a profitable business.
42
MAY 2011 RTC MAGAZINE
Similar to developing nations, there are remote regions where the subscriber count is small to moderate—anywhere from tens to hundreds of users—and these communities have never had broadband service in the past. Setting up and maintaining wired broadband such as DSL or cable is costprohibitive when service providers consider the infrastructure investments in setting up copper or fiber lines to the “last-mile” and setting up the backhaul network from the ISP to the Internet, or “middle-mile.” And since the per capita income in rural households tends to be lower than urban incomes, wireless base stations and gateways provide a cost-effective alternative. Smart Power Grids: Another part of ARRA includes a $3.4 billion funding for 100 smart power grid projects. Each country has different definitions for smart grids (SGs). SGs provide a more efficient means to deliver power from suppliers to consumers with reduced costs, energy saving policies (such as time-of-usage) and higher reliability. Either WiMAX or LTE can be used to interconnect the utilities. Besides standard networking, these systems can integrate control systems and security. Public Safety: In 2007, the Federal Communications Commission adopted the 700 MHz Band for public safety services. The objective is to establish a nationwide broadband communications network that
provides interoperable services for state and local public safety users. Specifically, the public safety bands are 763-775 MHz and 793-805 MHz, with commercial allocation in between. Light weight but carrier-grade systems provide the means to move these systems into locations that require emergency services, and once again MicroTCA systems provide the high availability and right size for the level of subscribers on this network.
System Level Considerations
When one considers the main functions of an LTE or WiMAX network, they typically think of the base station and the core gateways. Figure 1 shows an LTE network and Figure 2 shows a WiMAX network. However, the Element Management System (EMS), which monitors and manages these systems, is also a key component. And lastly, since LTE is a completely IP-centric technology, there will be a large push for Voice-over-LTE, or VoLTE, which will accelerate the demand for IMS (IP Multimedia Subsystems) with SIPbased services. MicroTCA-based systems are very well suited for all these services. Utilizing a common MicroTCA and Linuxbased processing architecture across all these functions reduces the need to develop on and support disparate platforms. It also reduces time-to-market, and avoids much of the cost to stock spare equipment. Gateways can be comprised of the MicroTCA platform, x86-based processor AMCs, and Carrier Grade Linux. Along with third party WiMAX ASN or LTE gateway software, a complete, fully-functional gateway can be developed. Base stations can be developed comprised of the MicroTCA platform, including support for Serial RapidIO or PCI-e options, along with x86-offs. These systems include GPS-based synchronization timing modules complying with Telcordia’s Spectrum 3 requirements, and they support oven-controlled oscillators for reliable hold-over in the absence of a reference clock. There are third-party AMCs on the market that provide the radio-frequency, baseband and control elements integrated on a single AMC card for Layer 1, 2 and 3 LTE-based and WiMAX-based base station solutions. Integral to both LTE and WiMAX,
INDUSTRY WATCH
an Element Management System (EMS) consists of the systems and applications related to managing one or more physical devices that make up a system. It allows these nodes or network elements to be managed in a unified way using one management system, rather than in a distributed, more cumbersome manner. Known as FCAPS, the key functionalities include managing faults, configurations, accounting, performance and security. In addition, Session Initiation Protocol (SIP)-based servers deliver cost-effective, feature-rich, enhanced next-generation network (NGN) services. Based on a pure IP implementation, new service offerings can be quickly developed and readily deployed, networkwide, on IMS-enabled, converged VoIP and TDM/IP networks.
Details on the LTE EPC Gateway Solution
Evolved Packet Core (EPC) is a set of functionalities that comprises the core of an LTE network. The eNodeBs will access these functions to determine how to behave when a user equipment (UE), such as a mobile phone, initiates a service request of the network. The goal of LTE from the EPC perspective is to simplify and better organize the services and functions of the network to reduced CAPEX and OPEX as well as provide higher bandwidth, higher spectral efficiency and lower latency than existing 2G/3G technologies. The three major components to LTE’s EPC are the mobility management entity (MME), serving gateway (SGW) and packet data network gateway (PGW). Since LTE is completely a packet-based network, all communications to and from the EPC will be native IP packet-based and leverage existing commodity infrastructure and simplify configuration and operations. There are well defined interface specifications for communicating between EPC components, thereby reducing interoperability issues between multiple vendor products. The MME’s main function (Figure 3) is to manage subscriber session control plane functionality, which uses the S1-C (C is for control plane) interface to communicate through the eNodeB to the UE. Authentication, authorization, ciphering and security key management are all dealt with by the MME. The S6 interface is used to commu-
Data
Data
Control
Data
Data
Standby
MME
MME
MME
Standby
MME
MME
Figure 5 Configuring the SGW/PGW Solution and the MME functions in MicroTCA Platforms.
nicate with the home subscriber server (HSS) database to manage authorization and accounting. Activating and deactivating as well as assigning an SGW to a UE, tracking the UE, and transitioning roaming handovers is within the MME’s responsibility in the EPC. The MME uses the S11 interface to inform the SGW of the session details. MME provides for intercept of signaling. To be compatible with 2G/3G networks, the MME provides control plane management with SGSN service from those architectures via the S3 interface. External interfaces supported by this solution include S1-MME, S11, S6a, S10, S13, S101, S102, S3 and SGs. The SGW’s main function (Figure 4) is to receive and route all UE packet data and serve as a mobility anchor for UEs transitioning between eNodeBs. The SGW uses the S5 interface to route data packets to a PGW within the same core network and will adhere to the S8 interface specifications when routing to a different network’s PGW such as in the case of a roaming UE. The S1-U (U is for User plane) interface is used between eNodeB and SGW to carry the user traffic and is subject to latency restrictions. Intercept traffic capture and replication is a function within the SGW. To be compatible with 2G/3G networks, the SGW provides user plane access with SGSN service from those architectures via the S4 interface. The PGW’s main function (Figure 4) is to route data packets from the SGW to external services such as the Internet, IP multimedia systems (IMS), or Public Switch Telephone Network (PSTN). Between the SGW and PGW are S5 for inter-network and S8 for other network traffic. Acting as the mobility anchor for other non-LTEbased networks is a key role for the PGW. The PGW performs packet filtering, policy enforcement and interception, charging support and packet screening. The feature set supported by the
SGW/PGW solution includes online and offline charging, static and dynamic QoS, default and dedicated bearers, and SGi interface based on local DHCP, RADIUS and DIAMETER. External interfaces supported include S11, Transparent SGi, Non-transparent SGi, S5, Gn, Gx, Gy/Ro, Gz/Rf and S4. The solution can scale to support up to one million subscribers. Figure 5 shows how the processors are configured into the MicroTCA platform to provide fully functional LTE EPC gateway systems with standby redundancy. A maximum configuration with four data processors and a standby can deliver an aggregate throughput of 3 Gbit/s with current generation AMC processors. With average per-subscriber data rates of 0.5 Mbit/s, this system can support up to 6,000 active subscribers. The modular architecture allows scaling of data processors to support smaller numbers of active subscribers, thus minimizing the cost. Scaling down also allows the integration of MME functions into the same system. Equipment manufacturers who are looking for turn-key LTE EPC or WiMAX ASN gateway applications to get to market quickly with low development costs can leverage these solutions. They are developed to be comprehensive, ready-for-market solutions that are right-sized and cost-effective for numerous wireless broadband applications. Comparatively, the scale of AdvancedTCA-based solutions is too large and they are too expensive to compete in many of these markets. MicroTCA systems cover the gamut of LTE and WiMAX applications from the base stations, gateways, Element Management and IMS. Performance Technologies Rochester, NY. (585) 256-0200. [www.pt.com].
RTC MAGAZINE MAY 2011
43
products &
TECHNOLOGY FEATURED PRODUCTS Energy Harvesting Power Conversion ICs Point to Easier, More Efficient Designs
With the necessary expertise in short supply around the globe, it has been difficult to design effective energy harvesting systems. The primary hurdle has been the power management aspects associated with remote wireless sensing. Now, with the introduction of three dedicated ICs from Linear Technology, that is about to change. These devices can extract energy from almost any source of light, heat or mechanical vibration. Furthermore, with their comprehensive feature set and ease of design, they greatly simplify the hard-to-do power conversion design aspects of an energy harvesting chain. The LTC3109 is a highly integrated DC/DC converter and power manager. It can harvest and manage surplus energy from extremely low input voltage sources such as thermoelectric generators (TEG), thermopiles and even small solar cells. Its auto-polarity topology allows it to operate from input sources as low as 30 mV, regardless of polarity. The circuit in Figure 1 uses two compact step-up transformers to boost the input voltage source to the LTC3109, which then provides a power management solution for wireless sensing and data acquisition. It can harvest small temperature differences and generate system power instead of using traditional battery power. The LTC3109 converts the low voltage source and manages the energy between multiple outputs. The AC voltage produced on the secondary winding of each transformer is boosted and rectified using an external charge pump capacitor and the rectifiers internal to the LTC3109. This rectifier circuit feeds current into the VAUX pin, providing charge to the external VAUX capacitor and then the other outputs. The internal 2.2V low-dropout regulator (LDO) can support a low power processor or other low power ICs. The LDO is powered by the higher value of either VAUX or VOUT. This enables it to become active as soon as VAUX has charged to 2.3V, while the VOUT storage capacitor is still charging. In the event of a step load on the LDO output, current can come from the main
44
MAY 2011 RTC MAGAZINE
VOUT capacitor if VAUX drops below VOUT. The LDO is capable of providing 3 mA of output current. The VSTORE capacitor may be a very large value (thousands of microfarads or even Farads), to provide holdup at times when the input power may be lost. Once Power-up has been completed, the Main, Backup and switched outputs are all available. If the input power fails, operation can still continue by operating off the VSTORE capacitor. The LTC3588-1 is a complete energy harvesting solution optimized for low energy sources, including piezoelectric transducers. Piezoelectric devices produce energy by either compression or by deflection of the device. These piezoelectric elements can produce hundreds of uW/cm2 depending on their size and construction (Figure 2). It should be noted that the piezoelectric effect is reversible in that materials exhibiting the direct piezoelectric effect (the production of an electric potential when stress is applied) also exhibit the reverse piezoelectric effect (the production of stress and/or strain i.e., deflection when a voltage is applied). The LTC3588-1 operates from an input voltage range of 2.7V to 20V, making it suitable for a wide array of piezoelectric transducers, as well other high output impedance energy sources. Its high efficiency buck DC/ DC converter delivers up to 100 mA of continuous output current or even higher pulsed loads. Its output can be programmed to one of four (1.8V, 2.5V, 3.3V or 3.6V) fixed voltages to power a wireless transmitter or sensor. Quiescent current is 950 nA with the output in regulation (at no load), maximizing overall efficiency. The LTC3588-1 is designed to interface directly with a piezoelectric or alternative high impedance AC power source, rectify a voltage wave-
form and store harvested energy in an external storage capacitor while dissipating any excess power via an internal shunt regulator. An ultralow quiescent current (450 nA) undervoltage lockout (ULVO) mode with a 1-1.4V hysteresis window enables charge to accumulate on the storage capacitor until the buck converter can efficiently transfer a portion of the stored charge to the output.
PRODUCTS & TECHNOLOGY
than required to maintain LDO regulation, charge is transferred from the AUX output to the VOUT output. If VAUX falls too low, current is redirected to the AUX output instead of being used to charge the VOUT output. Once VOUT rises above VAUX, an internal switch is enabled to connect the two outputs together. If VIN is greater than the voltage on the driven output (VOUT or VAUX), or the driven output is less than 1.2V, the synchronous rectifiers are disabled and operate in critical conduction mode enabling regulation even when VIN>VOUT. Get Connected with technology and nowgreater When the output voltage is companies greater thanproviding the inputsolutions voltage and than 1.2V, the synchronous rectifier is enabled. In theforN-chanGet Connected is athis newmode, resource further exploration products, technologies Whether your goal nel MOSFET between SW andintoGND is enabled untiland thecompanies. inductor current is to research the latest datasheet from a the company, speak directly reaches the peak current limit. Once current limit is reached, N-chanwithand an Application Engineer, or jump tobetween a company's nel MOSFET turns off the P-channel MOSFET SWtechnical and the page, the goal of Get Connected is to put you in touch with the right resource. driven output is enabled. This switch remains on until the inductor current Whichever level of service you require for whatever type of technology, drops below the valley limit and the cycle is repeated. VOUT Getcurrent Connected will help you connect with the When companies and products reaches the regulation the N-for. and P-channel MOSFETs connected youpoint, are searching to the SW pin are disabled and the converter enters sleep. www.rtcmagazine.com/getconnected In order to power microcontrollers and external sensors an integrated LDO provides a regulated 6 mA rail. The LDO is powered from the AUX output allowing the LDO to attain regulation while the main output is still charging. The LDO output voltage can be either a fixed 2.2V or adjusted via resistor divider. Getpower Connected withcircuit technology prov The integrated maximum point control allowsand thecompanies user to set the optimal input voltage operating point for resource a given for power source. Get Connected is a new further exploration into pro datasheetregulates from a company, speak directly with current an Application Engine The MPPC circuit dynamically the average inductor in touch the right resource. Whichever of service you requir to prevent the input voltage fromwith dropping below the MPPClevel threshold. Get Connected will help you connect with the companies and produc When VIN is greater than the MPPC voltage, the inductor current is increased until VIN www.rtcmagazine.com/getconnected is pulled down to the MPPC set point. If VIN is less than the MPPC voltage, the inductor current is reduced until VIN rises to the MPPC set point. In applications such as photovoltaic conversion, the input power source may be absent for long periods of time. To prevent discharge of the outputs in such cases, the LTC3105 incorporates an undervoltage lockout (UVLO) ,which forces the converter into shutdown mode if the input voltage falls below 90 mV (typical). In shutdown, the switch connecting AUX and VOUT is enabled and the LDO is placed into reverse-blocking mode, and the current into VOUT is reduced to 4 μA typical. Reverse current through the LDO is limited to 1 μA in shutdown to minimize discharging of the output. Get Connected with companies and LiPPERT Embedded Computers, Germany. products featured in thisMannheim, section.
Ad Index
The third device, the LTC3105, is an ultra-low-voltage step-up converter and LDO specifically designed to simplify the task of harvesting and managing energy from low voltage, high impedance alternative power sources such as photovoltaic cells, TEGs and fuel cells. Its synchronous step-up design starts up from input voltages as low as 250 mV, making it suitable for harvesting energy from even the smallest photovoltaic cells in less than ideal lighting conditions. Its wide input voltage range of 0.2V to 5V makes it attractive for a wide array of applications. An integrated maximum power point controller (MPPC) enables the LTC3105 to extract the maximum available power that the source is capable of providing. Without MPPC, the power converter can only produce a fraction of the source’s theoretical maximum capability. Peak current limits are automatically adjusted to maximize power conversion efficiency while Burst Mode operation reduces quiescent current to only 22 uA, minimizing the drain from the energy storage element. The ultra-low-Iq LDO is capable of directly powering popular low-power microcontrollers or sensor circuitry. The circuit shown in Figure 3 uses the LTC3105 to charge a singlecell Li-Ion battery from a single photovoltaic cell. This circuit enables the battery to continually charge when the solar source is available and in turn the battery can power the application from the stored energy when the solar power is no longer available. The LTC3105 provides the capability to start with voltages as low as 250mV. During start up the AUX output initially is charged with the synchronous rectifiers disabled. Once VAUX has reached approximately 1.4V, the converter leaves start-up mode and enters normal operation. Maximum power point control is not enabled during start up; however, the currents are internally limited to sufficiently low levels to allow start up from weak input sources. While the converter is in start-up mode, the internal switch between AUX and VOUT remains disabled and the LDO is disabled. When either VIN or VAUX is greater than 1.4V, the converter will enter normal operation. The converter continues charging the AUX output until the LDO output enters regulation. Once the LDO output is in regulation, the converter begins charging the VOUT pin. VAUX is maintained at a level sufficient to ensure the LDO remains in regulation. If VAUX becomes higher
Products
www.rtcmagazine.com/getconnected +49 621 4 32 14-0.
[www.lippertembedded.com]
Get Connected with companies and products featured in this section. www.rtcmagazine.com/getconnected
RTCRTC MAGAZINE MAGAZINEMONTH MAY 2011
45
PRODUCTS & TECHNOLOGY
High Density PXI RF Switches in Two Frequency Ranges
A new set of PXI RF switches supports up to 10 off SP4T RF switches in a single module and is available in two different versions based on a common switch design. The 40-755 multiplexer modules from Pickering Interfaces come in two versions. The high-density version occupies just one slot of a 3U PXI chassis and uses a high-density multi-way connector that is suitable for switching frequencies to 500 MHz. The higher frequency version uses SMB connectors and is suited for switching signals to 1.8 GHz. The 40-755 is the highest density SP4T switch available in the standard PXI format and is suitable for use in both commercial and military ATE systems. It is a suitable replacement for use in older VXI-based military ATE applications that are being replaced and/or upgraded by PXI/LXI solutions where great numbers of SP4T are used as standard RF switch sub-assemblies. The two models use a switch design based around high-quality electromechanical relays, and in addition to being supported in any PXI-compliant chassis, they can be supported in Pickering Interfaces Modular LXI Chassis. Pickering Interfaces, Grants Pass, OR. (541) 471-0700. [www.pickeringtest.com].
COM Express Module Boasts Rugged RS-DMM Memory Module
A new COM Express module boasts one of the first implementations of the rugged RS-DMM memory specification. The Toucan-QM57 from Lippert Embedded Computers is a COM Express revision 2.0 Type 2 module with an Intel Core i7 processor. The module is specifically built for applications exposed to rugged environments and hence the integration of the highly rugged RS-DIMM memory module. The memory module is fastened by screws. The raw computation performance of the processor and DDR3 RAM lends itself to image processing, video encoding, communications and other demanding tasks. The COM Express module offers 4 Gbyte of soldered RAM and another 4 Gbyte of RAM by way of RS-DIMM card, which optimizes board space and ruggedness. The module also supports interfaces including dual channel LVDS, 2 DisplayPorts, Gigabit LAN and eight USB 2.0 host ports. In addition, there are four SATA ports with RAID support and one PATA interface plus six x1 PCI Express lanes and one x16 lane. With 1.06 GHz dual-core Core i7-620UE processor, the module operates over -40° to +85°C. The 2.53 GHz dual-core Core i7-610E processor model is rated to +60°C for avionics, transportation, marine, imaging, communications and some automation applications. For these applications, additional embedded features add to rugged reliability. Lippert’s Enhanced Management Technology (LEMT) provides condition monitoring and supervision capabilities with an API for popular operating systems. In addition, there is a flag for the reason for a system restart (normal shutdown, watchdog, power loss, system crash), 1024 bytes of Flash memory for system manufacturers’ use, 128 bit encryption key, board and the ability to store the lowest, highest and current environmental temperatures. Finally, the input power current draw is measured and stored, allowing the OEM’s software to monitor its own health by calculating the module power consumption under various operating conditions. LiPPERT Embedded Computers, Mannheim, Germany. +49 621 4 32 14-0. [www.lippertembedded.com].
Isolated PCI Express Serial Communication Family with Speeds up to 3 Mbit/s
A new family of isolated PCI Express serial communication cards features a selection of 4 or 2 ports of isolated, software selectable, RS-232, RS-422 and RS-485 serial protocols. 2 kV isolation is provided between channels and between the PC and bus lines on ALL signals. The cards from Acces I/O Products feature a x1 lane PCI Express connector, which can be used in any available x1, x2, x4, x8, x12, or x16 PCI Express expansion slot. The PCIe-ICM product line has been designed for use in retail, hospitality, automation, gaming, shipboard and defense industries along with applications such as point of sale systems and kiosk design. The PCIe-ICM cards were designed using type 16C950 UARTS and use 128-byte transmit/ receive FIFO buffers to decrease CPU loading and protect against lost data in multitasking systems. New systems can continue to use legacy serial peripherals, yet benefit from the use of the high-performance PCI Express bus. The cards are fully software compatible with current PCI 16550 type UART applications and allow for users to maintain backward compatibility. The PCIe-ICM Series is especially useful in applications where high common-mode external voltages are present. Isolation is required to guard electronics from transient voltage spikes and offers greater common-mode noise rejection in electrically noisy surroundings containing industrial machinery and inductive loads. In addition to protecting industrial applications from accidental contact with high external voltages, the isolation provided eliminates troublesome ground loops. A complete driver support package is provided including an easy-to-use Windows terminal program for testing your COM ports. This simplifies the verification of proper COM port operation. The cards install as standard COM ports in all operating systems including DOS, Linux (including Mac OS X) and Windows 2000/XP/2003/Vista/7. ACCES I/O Products, San Diego, CA. (858) 550-9559. [www.accesio.com].
46
MONTH 2011 2011 RTCRTC MAGAZINE MAGAZINE MAY
PRODUCTS & TECHNOLOGY
4U Server for Transportation – Passenger Infotainment
A new rackmount server is specifically designed to handle passenger infotainment applications used in trains. With long-term availability, the KISS 4U KTC-5520 Transportation application-ready standard platform combines up to two Intel Xeon 5600 series processors with convenient remote management functionality. The KISS 4U KTC-5520 Transportation is also compliant to the following rail standards: EN 50155, EN 50121 and EN 60068. Passenger infotainment applications such as video on demand (VoD) are enhanced by the high simultaneous data processing capacity of the Transportation server. Fitted with up to eight processor cores and 16 threads, it is able to distribute several hundred HD-quality video streams. A high performance density also makes it suitable for consolidating applications that were previously deployed at separate locations. For example, passenger information, digital signage or Internet-on-train, can now be consolidated in a single server system to save space and costs. The Kontron KISS 4U KTC-5520 Transportation also comes with remote-management functionality. An integrated management processor (IMP) supports VGA/2D, BMC and KVM/VM over IP (iKVM) to enable real-time virtualization of the keyboard, monitor and mouse (KVM) plus virtual media from any external PC. With an intelligent platform-management interface (IPMI) over LAN, which complies with IPMI 2.0, the server board also offers an interface independent of operating systems and platforms, which may be used to monitor system temperature, voltage and fan status. And the integrated out-of-band management allows the system administrator access anytime, regardless of whether the system is turned on or not. In addition to which there are three hot-swap chassis fans accessible from the front to simplify system maintenance.
Ad Index
Kontron, Poway, CA. (888) 294-4558. [www.kontron.com].
PICMG 1.3 System Host Board Based on Xeon E31200 Family
A full-size PICMG 1.3 System Host Board provides high performance and flexible PCI Express expansion. Applications include Factory Automation, Gaming, Kiosk, Medical and Military. The ROBO-8110VG2AR from American Portwell is based on the latest Intel Xeon processor E3-1200 family, which offers 32nm Hi-K process technology with energy efficient architecture. The ROBO-8110VG2AR adopts dual channel DDR3 long DIMMs up to 16 Gbyte and ECC support on the new Intel Xeon E3-1275 and E3-1225 processors. The PCI Express 2.0 available on the new Intel Xeon processors provides flexible x16, x8 or x4 lanes for versatile applications. These latest Intel Xeon processors integrate an enhanced graphics engine with 3D performance for a broad range of embedded applications and support optimized Intel Turbo Boost 2.0 Technology and Intel Hyper-Threading, which provides higher performance and increases processing efficiency ROBO-8110VG2AR is based on the Intel C206 chipset, integrates dual Intel Gigabit Ethernet controllers (one of which can support Intel Active Management Technology 7.0), and features four SATA ports (two ports at 6 Gbit/s and two ports at 3 Gbit/s), which support RAID 0, 1, 5, 10 mode. Legacy device support such as serial ports (one RS-232 and one RS-232/422/485 selectable) for traditional factory automation applications is also provided by ROBO-8110VG2AR. The chipset also offers versatile display interfaces that can fulfill the market's interest in multiple displays built in on one board. ROBO8110VG2AR supports dual integrated displays via DVI-I (DVI-D + VGA) and HDMI. American Portwell, Fremont, CA. (510) 403-3399. [www.portwell.com].
Get Connected with technology and companies providing solutions now
Get Connected is a new resource for further exploration VPX Boards Pair Virtex-5 FPGA with PCIe Interface
into products, technologies and companies. Whether your goal A new series of 3U VPX FPGA boards features a configurable is to research the latest datasheet from a company, speak directly Xilinx Virtex-5 FPGA enhanced with multiple high-speed memory buf- page, the with an Application Engineer, or jump to a company's technical fers and a high-throughput PCIe interface. Field I/O signals interface to theresource. goal of Get Connected is to put you in touch with the right FPGA via the rear P2Whichever connector and/or withyou optional mezzanine pluglevel of service requirefront for whatever type of technology, willflexible help yousignal connectprocessor with the companies in I/O modules resultGet in aConnected powerful and card. and products are searching for. of logic-optimized FPGAs to Three models you provide a choice www.rtcmagazine.com/getconnected match the performance requirements. Cards can be ordered with a Xilinx VLX85T, VLX110T, or VLX155T FPGA featuring up to 155,000 logic cells and 128 DSP48E slices. Each model is available in a format designed for use in aircooled or conduction-cooled systems suitable for -40° to Get Connected with technology and companies prov 85°C operation. 64 I/O lines Get are accessible through the Connected is a new resource for further exploration into pro datasheet from a company, speak directly with an Application Engine rear (P2) connector. A series in touch with the right resource. Whichever level of service you requir of AXM extension modules Get Connected will help you connect with the companies and produc are available to provide addiwww.rtcmagazine.com/getconnected tional front-end A/D, RS-485, CMOS, or LVDS I/O channels through a mezzanine connector on the front of the board. Acromag’s Engineering Design Kit provides utilities to help users develop custom programs, load VHDL into the FPGA and to establish DMA transfers between the FPGA and the CPU. The kit includes a compiled FPGA file and example VHDL code provided as selectable blocks with examples for the local bus interface, read/writes and change-of-state interrupts to the PCI bus. A JTAG interface allows users to perform onboard VHDL simulation. Further analysis is supported with Get Connected with companies and a ChipScope Pro interface. real-time and open source applications, products featured in For this section. Acromag offers C libraries for VxWorks, Linux and other operating syswww.rtcmagazine.com/getconnected tems. The boards start at $7,100 with several options for FPGA logic capacity and conduction-cooled extended temperature operation.
Products
Acromag, Wixom, MI. (248) 295-0310. [www.acromag.com]. Get Connected with companies and products featured in this section. www.rtcmagazine.com/getconnected
RTCRTC MAGAZINE MAGAZINEMONTH MAY 2011
47
PRODUCTS & TECHNOLOGY
Extended Temperature and High-Density Isolated Serial Communication PMC Module
A high-density isolated serial communication controller is packaged as a conduction-cooled single-width 32-bit PMC module suitable for applications in transportation, COTS, communications and process control. The TPMC377 from Tews Technologies can operate with 3.3V and 5.0V PCI I/O signaling voltage. It provides 4 channels of RS-232/RS-422/RS-485 selectable serial connectivity with P14 I/O. Each of the serial channels is isolated from the system and against each other by a digital isolator and an onboard integrated DC/DC converter. The cannels can be individually programmed to operate as RS-232, RS-422 or RS-485 full/half duplex interfaces. In addition, programmable termination is provided for the RS-422/RS-485 interfaces. After power-up, all serial I/O lines are in a high impedance state for critical applications. Each RS-232 channel supports RxD, TxD, RTS, CTS and GND. RS-422 and RS-485 full duplex support a four wire interface (RX+, RX-, TX+, TX-) plus ground (GND). RS485 half duplex supports a two wire interface (DX+, DX-) plus ground (GND). All channels generate interrupts on PCI interrupt INTA. For fast interrupt source detection the UART provides a special Global Interrupt Source Register. Each serial channel of the PMC module has separate 64 byte receive and transmit FIFOs to significantly reduce the processing overhead required to provide data to and from the transmitters and receivers. The FIFO trigger levels are programmable, and the baud rate is individually selectable up to 921.6 Kbit/s for RS-232 channels and 5.5296 Mbit/s for RS422 channels. The UART offers readable FIFO levels. The TPMC377 offers an operating temperature range of -40° to +85°C. Software support for major operating systems such as Windows, Linux, LynxOS, VxWorks, Integrity and QNX is available. TEWS Technologies, Reno, NV (775) 850-5830. [www.tews.com].
1U Rackmount Network Server Provides Functionality and Flexibility
A new modular 1U rackmount communication appliance can be a solution for customers by providing network security solutions to medium-high businesses who want to acquire functionality and make a seamless transition to new technologies such as 10G Ethernet. The CAR-4010 from American Portwell is based on the new Intel C206 chipset-based platform, which offers a wide range of CPU support from the latest Xeon processor E3-1200 family to the Core i3-2120 processor. It also supports the latest dual-channel DDR3 1333/1066 memory platform, which supports UDIMM Non-ECC or UDIMM ECC up to 8 Gbyte; up to two 3.5˝ or 2.5˝ SATA HDD or SSD; and up to four GbE SFP + four GbE RJ45 or eight GbE RJ45 onboard ports utilizing latest PCI-E Gen. 2.0 technology via Intel 82580DB Gigabit Ethernet Controller with two software-controlled bypass segments (Fail-Open or Fail-Close). Expansion capabilities include one PCI-E x8 proprietary interface and one PCI-E x16 standard interface for a modular bay for Portwell’s ABN & NIP network interface card product family; fiber and copper port connections including dual-port 10G readiness with the following Intel 10 Gigabit Ethernet Controllers (Intel 82598EB, Intel 82599ES with SFP+ Interface and Intel 82599EB 10Gbase-T Copper Interface); 2x16 character LCD display module options that include EZIO-300 (with 4 buttons), EZIO-G400 (128x32 graphics display module with 4 buttons), or EZIO-G500 (128x64 graphics display module with 7 buttons) and VGA pin-header available for software/application development. Moreover, it implements an IPMI 2.0 feature on board, providing more powerful monitoring and management. American Portwell, Fremont, CA. (510) 403-3399. [www.portwell.com].
48
MONTH 2011 2011 RTCRTC MAGAZINE MAGAZINE MAY
Instrument Simultaneously Measures and Records Temperature, Velocity and Pressure
A new thermal analysis system precisely and simultaneously measures the temperatures of solid materials and the surrounding air, as well as tracking air velocity and air pressure at multiple points to comprehensively profile heat sinks, components and PCBs. The iQ-200 from Advanced Thermal Solutions simultaneously captures data from up to 12 J-type thermocouples, 16 air temperature/velocity sensors and 4 differential pressure sensors. The thermocouples provide surface area temperature measurements on heat sinks, components, housing parts and other locations to track heat flow or detect hot spots. Temperature data is recorded from -40° to 750°C. Air temperature and velocity are measured by up to 16 low profile ATS candlestick sensors, which can be placed throughout a system. Air temperature is tracked from 20° to 65°C and air velocity is measured from 0 to 6 m/s (1200 ft/min). The differential transducers capture pressure drop data along circuit cards, assemblies and orifice plates. Pressure measurements are taken from 0 to 0.15 psi (0 to 1,034 Pa). The iQ-200 system comes preloaded with user-friendly iSTAGE application software, which manages the incoming data from multiple sensing devices, and provides graphic presentation on monitors and documents. The iQ-200 connects via USB to any conventional PC for convenient data management, storage and sharing. The iQ-200 can be factory modified at ATS to measure higher airflows, up to 50 M/s (10,000 ft/min) and air temperatures to 85°C. List price is $24,500, including 12 J-type thermocouples, 16 candlestick sensors and 4 differential pressure sensors Advanced Thermal Solutions, Norwood, MA. (781) 769-2800. [www.qats.com].
Support Across The Board. From Design to Delivery Now, you can have it all.™ Faster and easier than ever before. Our commitment to customer service is backed by an extensive product offering combined with our supply chain and design chain services – which can swiftly be tailored to meet your exact needs. We have dedicated employees who have the experience to provide the highest level of customer service with accuracy and efficiency. All of our technical experts are factory certified on the latest technologies, providing you the expertise to move projects forward with speed and confidence. Avnet offers the best of both worlds: extensive product and supply chain knowledge, and specialized technical skill which translates into faster time to market – and the peace of mind that comes from working with the industry’s best. Avnet is ranked Best-In-Class* for well-informed sales reps, knowledgeable application engineers and our design engineering services – proof that we consistently deliver: > Industry recognized product expertise > Specialized technical skills Ready. Set. Go to Market.™ Visit the Avnet Design Resource Center™ at: www.em.avnet.com/drc
1 800 332 8638 www.avnetexpress.com *As rated by Hearst Electronics Group: The Engineer & Supplier Interface Study, 2009. ©Avnet, Inc. 2011. All rights reserved. AVNET is a registered trademark of Avnet, Inc.
Follow us on Twitter!
www.twitter.com/avnetdesignwire
with an Application Engineer, or jump to a company's technical page, the goal of Get Connected is to put you in touch with the right resource. Whichever level of service you require for whatever type of technology, Get Connected will help you connect with the companies and products you are searching for.
www.rtcmagazine.com/getconnected
Advertiser Index Get Connected with technology and companies providing solutions now Get Connected is a new resource for further exploration into products, technologies and companies. Whether your goal is to research the latest datasheet from a company, speak directly with an Application Engineer, or jump to a company's technical page, the goal of Get Connected is to put you in touch with the right resource. Whichever level of service you require for whatever type of technology, Get Connected will help you connect with the companies and products you are searching for.
www.rtcmagazine.com/getconnected
Company
Page
Website
AMD.......................................................... 52.........................www.amd.com/embedded
Company
Page
Website
ISI Nallatech Inc......................................... 19................................... www.nallatech.com
End of Article Products Logic Supply, Inc........................................ 17................................ www.logicsupply.com American Portwell Technology, Inc............... 9......................................www.portwell.com Get Connected with companies and Avalue Technology...................................... 16................................ www.avalue-usa.com products featured in this section. www.rtcmagazine.com/getconnected
Get Connected Measurement Computing............................ 18.....................................www.mccdaq.com with companies mentioned in this article. www.rtcmagazine.com/getconnected
Avnet Electronics Marketing........................ 49............................. www.avnetexpress.com
Mentor Graphics......................................... 51............................ www.edatechforum.com
Get Connected with companies mentioned in this article.
Get Connected with companies and products featured in this section.
Cogent....................................................... 38................................... www.cogcomp.com www.rtcmagazine.com/getconnected
www.rtcmagazine.com/getconnected One Stop Systems...................................... 11.........................www.onestopsystems.com
Design Automation Conference - DAC......... 35............................................www.dac.com
Phoenix International................................... 4..................................... www.phenxint.com
ELMA Electronic Inc................................... 31................................www.acttechnico.com
Solid-State Drives & Industrial Box PC Showcase........................................................39
Extreme Engineering Solutions, Inc.............. 2....................................... www.xes-inc.com
Innovative Integration.................................. 27........................... www.innovative-dsp.com
VersaLogic Corporation.............................. 23..................................www.versalogic.com
WDL Systems............................................. 34................................www.wdlsystems.com
Scan this QR Code with your smartphone and instantly SUBSCRIBE to receive RTC magazine! Go to www.rtcmagazine.com to download a QR reader. RTC (Issn#1092-1524) magazine is published monthly at 905 Calle Amanecer, Ste. 250, San Clemente, CA 92673. Periodical postage paid at San Clemente and at additional mailing offices. POSTMASTER: Send address changes to RTC, 905 Calle Amanecer, Ste. 250, San Clemente, CA 92673.
50
MAY 2011 RTC MAGAZINE