Spring 2010 • Vol. 19, No. 1

Pentek, Inc.One Park Way, Upper Saddle River, NJ 07458Tel: (201) 818-5900 • Fax: (201) 818-5904email: pipeline@pentek.comhttp://www.pentek.com

© 2010 Pentek, Inc. Newsletter Editor: TNT Resources, LLC.Trademarks are properties of their respective owners.Specifications are subject to change without notice.

A quarterly publication for engineering system design and applications.

● Product Focus:Product Focus:Product Focus:Product Focus:Product Focus:14 New 3U VPX boards!

● Product Focus: Product Focus: Product Focus: Product Focus: Product Focus:RTS 2711 Recording Instrument

Free Technical Resources● Sign up for Virtex-6 CobaltTM

product updates● Sign up for VPX product

updates

● 20102010201020102010 edition of the High-SpeedSwitched Serial Fabrics Hand-book

● 20102010201020102010 edition of the SDR Hand-book

● 20102010201020102010 edition of the FPGAs forSoftware Radio SystemsHandbook

● 20102010201020102010 edition of the High-SpeedA/Ds Handbook

In This Issue● As evolutionary enhancementsto the venerable VMEbus, bothVXS and VPX deliver significantimprovements in data bandwidth,connectivity, power distribution,and cooling. More in the featurearticle.

Rodger Hosking, Pentek VicePresident and Cofounder

As evolutionary enhancements to thevenerable VMEbus, both VXS and VPXdeliver significant improvements in data

bandwidth, connectivity, power distribution,and cooling. When VME was first introduced, itsshared bus backplane interboard transfer ratesof 30 or 40 MBytes/sec was more than adequatefor most applications. As requirements grew,VME acquired new interfaces such as VSB,RACEway, RACE++, VME64, VME320, and2eSST, thereby ensuring a healthy communityof suppliers and a new stream of products.

Well into its third decade of widespreaddeployment, VME adopted the new VXS gigabitserial interface, clearly representing the mostsignificant leap in backplane data transfer ratesthroughout its entire history.

Because VXS delivered such a dramaticimprovement in embedded system performance,the use of gigabit serial technology was extendedto create VPX. The OpenVPX initiative followedshortly thereafter as risk-averse governmentagencies, with concerns about the longevityand maintainability of new technology andarchitecture, mandated the need for industry-wide standards. The hallmark of any successfulstandard is that it continues to evolve with tech-nology, and none offers a better example thanVME’s evolution to VXS and VPX.

VPX and VXS: System-level Development Strategies

VXS: High-Bandwidth Fabric for VMEMotorola’s VME Renaissance announcement

in 2003 unveiled the new VXS initiative, officiallydesignated VITA 41 by the VSO (VITA StandardsOrganization). It defines the implementation ofgigabit serial technology for VME in a logicallylayered specification, while carefully preservingthe legacy VME form factors and bus operations.At the top layer, VITA 41 defines the connectors,pin designations, dimensions and mechanicalstructures for cards and backplanes—completelyfree of any mandates for specific slot intercon-nection strategies, protocols or fabrics.

VXS defines two types of cards: the PayloadCard and the Switch Card. Both utilize the samemechanical outline as a standard 6U VME card.The payload card fits a new gigabit serial con-nector (MultiGig RT2) between the existing P1and P2 connectors, designated as P0. Mostlegacy VME backplanes provide clearance forthe new P0 connector, thus allowing insertionof VXS cards for backwards compatibility eventhough the VXS interface is not engaged.

VXS backplanes have one mating MultiGig RT2connector for each payload card slot. VITA 41does not dictate any specific backplane topol-ogy and leaves plenty of flexibility for variousarchitectures. The simplest configuration is theVXS switchless mesh backplane shown in

Figure 1. Here, three VXScards are joined in a ring con-figuration, with each cardconnected to the other twothrough two 4X links hard-wired in copper as part of thebackplane design. At a serialbit rate of 3.125 GHz, thissystem supports simultaneousdata transfers among the threecards in both directions at anaggregate interboard rate of7.5 GBytes/sec. ➤

Figure 1. VXS offers significant boost in data transfer ratesbetween boards

“Because they eliminate bottle-necks in highperformanceembedded sys-tems, gigabit

serial links arereplacing par-

allel buseson boardsand backplanes”

2

VPX and VXS: System-level Development Strategies

Figure 2. Pentek Model 4207 VXS I/O Processorwith dual PMC/XMC sites.

VPX Extensions: TremendousSwitched-Fabric I/O Capacity

With each new generation of powerful,high-performance embedded solutions—including processors with higher clockrates and wider buses; data converter prod-ucts with higher sampling rates; and FPGAs,RISCs, and DSPs offering incredible compu-tational rates—an equally capable backplanesolution was needed to keep pace withthe data transfer rates and prevent systembottlenecks. By extending the use of gigabitserial links already proven under VXS, theembedded community created the VPX ini-tiative, which was formally defined underVITA 46. As a migration from the earlierVME and VXS standards, VPX shares thesame outline as 3U and 6U cards and sup-ports XMC mezzanine modules definedunder the VITA 42 standard.

While VXS allows only one MultiGigRTS connector on a 6U card, VPX extendsthat number to three for a 3U card and toseven for a 6U card. As a result, VPX pay-load cards support a much higher trafficbandwidth than VXS, with eight to 24 giga-bit serial 4X ports compared to only twowith VXS.

Like the VXS specification, the VITA46.0 VPX base specification does not de-fine backplane topologies or specificgigabit serial fabrics or protocols. As withVXS, implementations of each fabric pro-tocol are defined as sub specifications, or“dot specs.”

As industry started using VPX, a newextension emerged to deal with severeenvironmental requirements. The VITA 48REDI (Ruggedized Enhanced Design Imple-mentation) defines specific mechanicaldesigns for enhanced thermal managementusing forced air, conduction cooling, andliquid cooling methods. It also defines pro-tective metal covers for the cards to satisfynew requirements for simplified field servic-ing in deployed military applications.

OpenVPXThe OpenVPX organization was formed

in January 2009 to promote industry-widestandards and long-term availability of VPXtechnology across the industry. The origi-

nal VPX specification was being used, butbecause it permitted such a wide range ofarchitectures, VPX systems tended to beunique, vendor-specific implementations.

The mission of OpenVPX was to enhancethe original VPX standard by adding a setof well-defined system architectures,nomenclature and conventions to enableinteroperability among vendors. Consistingof key vendors in the embedded-systemcommunity, all eager to convince govern-ment and military customers that VPX wassuitable for current and future systems, thegroup made fast progress and turned overthe completed specification to the VSO inOctober 2009 for standardization under VITA65. In February 2010, the specification wasratified by VSO, while ANSI approval isexpected sometime later in 2010.

The OpenVPX Language forStandardization

OpenVPX defined new nomenclature forsystems to describe the gigabit serial linksin terms of the number of lanes and theirfunction. The term “pipe” is used to definethe number of bidirectional differentialserial pairs that are grouped together toform a logical data channel. Pipe sizesrange from one lane (1X) called an “ultra-thin pipe” or UTP, up to 32 lanes(32X) called an “octal fat pipe” orOFP. The popular 4X link is calleda “fat pipe” or FP.

OpenVPX also categorized thedifferent kinds of traffic carriedthough the pipes as “planes”. Thefive planes defined are the utility,management, control, data andexpansion planes.

In order to define architecturalcharacteristics of systems, several“profiles” were defined. A slotprofile specifies the pipes andplanes found on the backplaneconnectors of each slot. Themodule profile specifies thepipes, planes, fabrics and proto-cols implemented on each card.The backplane profile defineshow the slots are connected toeach other by pipes. And finally,the development chassis profileincludes the backplane profile

and defines the dimensions, power supply,and cooling method.

VPX and VXS ProductsDozens of embedded computer hard-

ware vendors have developed numerousVXS and VPX products. These includebackplanes; complete card cages andchassis; A/D and D/A converters; softwareradio cards; XMC carriers; single boardcomputers; DSP, and RISC array processors;FPGA cards and memory cards.

VXS fully supports PMC mezzaninemodules and the gigabit serial extensionfor PMC modules known as XMC. Both 3Uand 6U VPX modules are also compatiblewith the gigabit XMC interface. The enor-mous base of existing PMC and XMC I/Omodules offers integrators many choicesfor their VXS and VPX application solu-tions. Integrators can immediately takeadvantage of a rich variety of offerings inthe market.

As an example, Figure 2 shows thePentek Model 4207 VXS PowerPC I/O pro-cessor with dual XMC sites and a wealth ofcritical interfaces for high-performance em-bedded systems. At its heart is an on-boardfabric-transparent gigabit serial crossbarswitch. This switch highlights the vital ➤

3

VPX and VXS: System-level Development Strategies

Figure 3. 16-Channel Beamformer System with four Pentek Model 5353 3U VPX boards.

role of serial technology for interconnect-ing resources within the board and to otherboards across the backplane.

VPX ApplicationsThe latest embedded system designs

show a definite shift towards serial fabric-based system architectures. These use bothPCIe and SRIO, primarily to improve board-to-board data transfer rates that handle highersignal bandwidths, more powerful FPGAsand processors, and faster peripherals.

VPX cards in 3U and 6U sizes can supportone or two XMC modules, respectively.Native PCIe and SRIO gigabit serial interfaceson these XMCs are often directly compat-ible with processors, other devices on thecarrier board, and also with VPX backplanecontrol and data planes. Other XMC protocolslike Xilinx Aurora are ideal for raw high-speeddata links, often directly connectable toVPX data and expansion planes.

Because they deliver substantial perfor-mance benefits, the latest FPGAs and gigabitserial fabrics increasingly dominate embed-ded system designs. Although thesetechnologies are prevalent in both VXSand VPX platforms, VPX offers a clear

advantage, not only because it offers manymore links, but because it also simultaneouslyaccommodates multiple protocols.

For example, GigE can handle systemmanagement, while PCIe can be used forcommand and control. Serial RapidIO cansupport high-speed data transfer betweenprocessors, while Aurora can enableFPGAs to communicate raw data across theexpansion plane.

Beamforming Systems forImproved Performance

Beamforming applications use an arrayof antennas to improve directionality of recep-tion and signal quality. The signal arrivaldelay at each antenna is based on the pathdistance from the source. The beamform-ing process adjusts the gain and phase ofeach antenna signal to cancel the delaydifferences for signals arriving from a par-ticular direction.

Aligned signals are summed together toproduce high signal-to-noise reception inthe chosen direction. By adjusting gain andphase in each path, the antenna is elec-tronically “steered” without the need formoving mechanical structures.

Examples of applications that use beam-forming include Direction Finding, wherea beamformed antenna can be steered tolocate the arrival angle of a signal. Two ormore arrays can be used to triangulate theexact location of the source. This is extremelyimportant for signal intelligence and counterterrorism efforts.

Radar Receiver applications use bothone- and two-dimensional antenna designsfor phased-array and SAR (Synthetic Aper-ture Radar) systems. Electronic steeringof the array dramatically improves theangular agility, range and target resolutionof airborne antennas.

Missile detection and countermeasureapplications use beamforming to improvetracking an object for early detection andimproved responsiveness. And lastly, beam-forming allows spatial frequency sharingby commercial mobile phone carriersbecause it divides the cell into severalbeamformed sectors.

Figure 3 shows a 16-channel Beamformerusing four Pentek Model 5353 3U VPX boards.This digital beamforming system takesadvantage of the expansion plane for cascad-ing beamformed data between cards. ➤

X4

PCIe

X4

PCIe

X4

PCIe

X4

PCIe

VPX Slot 2

VPX Slot 1

CPU

Model 5353

VPX Slot 4

Model 5353

VPX Slot 3

Model 5353

VPX Slot 5

Model 5353

FP A

FP B

FP C

FP D

FP A

FP B

FP C

FP D

FP A

FP B

FP C

FP D

FP A

FP B

FP C

FP D

200 MHz

16-bit A/D

200 MHz

16-bit A/D

200 MHz

16-bit A/D

200 MHz

16-bit A/D

CH B

RF In

CH A

RF In

CH D

RF In

CH C

RF In

AURORA

BEAMFORM �

DDC 3

DDC 2

DDC 1

DDC 4

PCIe

4X

200 MHz

16-bit A/D

200 MHz

16-bit A/D

200 MHz

16-bit A/D

200 MHz

16-bit A/D

CH B

RF In

CH A

RF In

CH D

RF In

CH C

RF In

AURORA

BEAMFORM �

DDC 3

DDC 2

DDC 1

DDC 4

PCIe

4X

200 MHz

16-bit A/D

200 MHz

16-bit A/D

200 MHz

16-bit A/D

200 MHz

16-bit A/D

CH B

RF In

CH A

RF In

CH D

RF In

CH C

RF In

AURORA

BEAMFORM �

DDC 3

DDC 2

DDC 1

DDC 4

PCIe

4X

200 MHz

16-bit A/D

200 MHz

16-bit A/D

200 MHz

16-bit A/D

200 MHz

16-bit A/D

CH B

RF In

CH A

RF In

CH D

RF In

CH C

RF In

AURORA

BEAMFORM �

DDC 3

DDC 2

DDC 1

DDC 4

PCIe

4X

4

Special Bundle

Package Offers You

a $2,000 Savings!

Ends 6/30/10

VPX and VXS: System-level Development StrategiesEach of the four modules digitizes four IFsignals from four antennas in the array.Four DDCs (digital downconverters) trans-late the IF signals to baseband and performbeamforming phase and gain adjustments.

A summation engine accepts a propa-gated sum from a previous card, adds thefour channels from the local card and thengenerates a new sum signal for delivery tothe sum input of the next card in the chain.The summation paths use Aurora 4X giga-bit serial links for the expansion planeconnections across the backplane.

VPX and VXS SummaryWhile VPX systems can deliver aggre-

gate transfer rates much higher than VXS,a vast majority of system requirements can

be fully satisfied by the tremendous boostin rates that VXS offers over the legacy VME.Investments made by board vendors andsystem integrators in VXS hardware, inter-faces, middleware, software and applicationswill translate easily into VPX, when systemneeds dictate.

With so many VPX-compatible productsavailable today, and with the new OpenVPXVITA 65 standardization, system integratorscan feel confident selecting VPX architec-tures for high-performance embeddedapplications.

Further evidence that the industry hasembraced VPX are two new extensions.The VITA 66 Fibre Optic Interconnectspecification for VPX defines a family ofFibre-optic interconnects that allows VPX

connectors J2 though J6 to be replacedwith optical connectors. The VITA 67Coaxial Connector specification for VPXdefines a shielded coaxial analog RFconnector using the same mechanicaldimensions as VITA 66. DRS Signal Solu-tions is leading the VITA 67 initiativeand Pentek is actively teaming with DRSon specification acceptance and productdevelopment.

VPX dramatically improves embeddedsystem performance and can achieve ratespreviously unattainable with earlier tech-nology. For all of these very tangibleconsiderations, VXS and VPX will dominateas the preferred architectures for futurehigh-performance commercial and militaryembedded systems. ❑

Featured Products and Special Offers

Model 4207 VME/VXS Processorwith two PMC/XMC Sites

■ Product Information: Model 4207

Model 71620 Software Radio XMCwith Virtex-6 FPGA

■ Product Information: Model 71620

Model 7158 Software Radio XMCfor Very Wideband Signals

■ Product Information: Model 7158

Model 7751D 512-ChannelSoftware Radio PCI Express Board

■ Product Information: Model 7741D

Model 7190 MultifrequencyClock Synthesizer PMC Module

■ Product Information: Model 7190

Model RTS 2721 PortableReal-Time Recording Instrument

■ Product Information: Model RTS 2721

Special Bundle

Package Offers You

a $5,000 Savings!

Ends 7/31/10

5

3U VPX Boards

Pentek Announces a Series ofSoftware Radio 3U VPX Boards

Features■ 3U VPX with built-in support for PCIe■ Four 200 MHz, 16-bit A/Ds■ Two Xilinx Virtex-5 FPGAs■ 1.5 GB DDR2 SDRAM■ LVPECL clock/sync bus for multiboard

synchronization■ Commercial and ruggedized versions■ 32 pairs of LVDS connections to the

Virtex-5 FPGAs for custom I/O■ Product Information: Model 5350

Features■ 3U VPX with built-in support for PCIe■ 256 channels of DDCs in four banks■ Independent tuning for each channel■ DDC decimation from 128 to 1024■ User-programmable 18-bit FIR

filter coefficients■ Four 200 MHz, 16-bit A/Ds■ LVPECL clock/sync bus for multiboard

synchronization■ Product Information: Model 5351

Model 5350: Quad 200 MHz, 16-bit A/D Model 5351: 256-Channel DDC, four 200 MHz A/Ds

14 New Products, designed for communications, telemetry, and signalintelligence applications, deliver extreme performance

Features■ 3U VPX with built-in support for PCIe■ Two 400 MHz, 14-bit A/Ds■ One DUC (Upconverter)■ Two 800 MHz, 16-bit D/As■ Independent A/D and D/A clocks■ Two Xilinx Virtex-5 FPGAs■ 1 GB DDR2 SDRAM■ LVPECL clock/sync bus for

multiboard synchronization■ Product Information: Model 5356

Model 5356: Dual 400 MHz A/D and 800 MHz D/A

Features■ 3U VPX with built-in support for PCIe■ Two 500 MHz, 12-bit A/Ds■ One DUC (Upconverter)■ Two 800 MHz, 16-bit D/As■ Independent A/D and D/A clocks■ Two Xilinx Virtex-5 FPGAs■ 1 GB DDR2 SDRAM■ LVPECL clock/sync bus for

multiboard synchronization■ Product Information: Model 5358

Model 5358: Dual 500 MHz A/D and 800 MHz D/A

Many more on next page ➤➤➤➤➤ ➤➤➤➤➤ ➤➤➤➤➤

Model 5390 Multifrequency Clock Synthesizer

Features■ 3U VPX with built-in support for PCIe■ Simultaneous synthesis of up to five

different clocks■ Eight clock ouputs on SMC connectors■ All clocks phase locked to input

reference■ Very low phase noise and jitter■ Control and status via the VPX

backplane■ Product Information: Model 5390

Model 5308: Front Panel PCI Express x8 Adapter

Features■ Front Panel x8 PCI Express

connection to host PC■ 3U VPX form factor provides a compact

rugged platform■ Cascade mode provides connection

to an additional VPX system■ Fabric-transparent crossbar switch

bridges numerous gigabit serialinterfaces on the board with no latency

■ USB interface for switch programming■ Product Information: Model 5308

6

3U VPX Boards

Pentek Announces a Series ofSoftware Radio 3U VPX Boards

Model 5341: Dual Multiband Transceiver

Features■ 3U VPX with built-in support for PCIe■ Two 125 MHz, 14-bit A/Ds■ Four-channel DDC (Downconverter)■ One DUC (Upconverter)■ Two 500 MHz, 16-bit D/As■ Xilinx Virtex-II Pro FPGA■ 512 MB DDR SDRAM■ LVDS clock/sync bus for multiboard

synchronization■ Product Information: Model 5341

Model 5342-428 with 4 Multiband DDCs and Interpolator

Features■ 3U VPX with built-in support for PCIe■ Four 125 MHz, 14-bit A/Ds■ Four multiband DDCs and one

interpolation filter■ Decimation range from 2 to 65,536■ Interpolation range from 2 to 32,768■ User-programmable FIR filter coefficients■ LVDS clock/sync bus for multiboard

synchronization■ Product Information: Model 5342-428

Features■ 3U VPX with built-in support for PCIe■ Two 105 MHz, 14-bit A/Ds■ 16 channels of multiband DDCs■ User-configurable Xilinx Virtex-II FPGA■ 250 MHz input bandwdith■ 5 kHz to 10 MHz output bandwidth

for 100 MHz sampling frequency■ LVDS clock/sync bus for multiboard

synchronization■ Product Information: Model 5331

Model 5331: 16-Channel Multiband Receiver

14 New Products designed for communications, telemetry, and signalintelligence applications, deliver extreme performance

Model 5342: Multichannel Transceiver

Features■ 3U VPX with built-in support for PCIe■ Four 125 MHz, 14-bit A/Ds■ One DUC (Upconverter)■ One 500 MHz, 16-bit D/A■ Two Xilinx Virtex-4 FPGAs■ 768 MB DDR2 SDRAM■ LVDS clock/sync bus for multiboard

synchronization■ Optional factory-installed DDC cores■ Product Information: Model 5342

Also available: Model 5341-420 with 4 Wideband DDCs Also available: Model 5341-430 with 256-Channel DDC■ Product Information: Model 5341-420 ■ Product Information: Model 5341-430

Features■ 3U VPX with built-in support for PCIe■ Four 200 MHz, 16-bit A/Ds■ Two or four channels of DDCs■ DDC decimation from 2 to 256■ Independent 32-bit DDC tuning■ Power meters, threshold detectors

and beamforming functions■ LVPECL clock/sync bus for

multiboard synchronization■ Product Information: Model 5353

Model 5353: Quad DDC, 200 MHz A/Ds, BeamformerModel 5352: 32-Channel DDC, four 200 MHz A/Ds

Features■ 3U VPX with built-in support for PCIe■ 32 channels of DDCs in four banks■ Independent tuning for each channel■ DDC decimation from 16 to 8192■ User-programmable 18-bit FIR filter

coefficients■ Power meters and threshold detectors■ LVPECL clock/sync bus for multiboard

synchronization■ Product Information: Model 5352

7

Model RTS 2711

Dual-channel 500 Megasamples-per-secondReal-Time Data Recorder Instrument

Features■ Complete dual channel recording system■ 19 inch 4U industrial rack-mount PC

server chassis■ Complete high performance

Windows® XP workstation■ Two 12-bit 500 MHz A/Ds, retaining

8-bit samples for full-rate recording

■ Real-time aggregate sustainedrecording rates up to 1 GB/sec

■ 4 terabytes of storage to NTFS RAIDdisk array

■ Hot-swap SATA drives■ RAID levels 0, 1, and 5■ Windows SystemFlow® Recording

software■ Signal Viewer analysis tool includes virtual

oscilloscope and spectrum analyzer■ File headers include time stamping

and recording parameters■ Digitizes two analog inputs each at

sampling rates to 500 megasamplesper second

■ Ideal for communications, radar,wireless, SIGINT, telecom and satcom

General InformationThe Pentek RTS 2711 is a turnkey record-

ing instrument that allows the user to recordand analyze two high-bandwidth signals. TheRTS 2711 provides sustained, aggregaterecording rates of up to 1 GB/sec forming apowerful dual-channel 4U rack-mountrecording system.

The front end of the RTS 2711 consistsof two Pentek Model 7858 PCIe modulesequipped with 500 MHz 12-bit A/D con-verters. The RTS 2711 retains the eightmost significant bits of each A/D sample torecord two signals at 500 megasamples persecond.

A total of 4 TB of RAID storage is provided,allowing sustained 2 TB recordings at500 megasamples per second simultaneouslyon each of two channels for over one hour.

SystemFlow SoftwareIncluded with this instrument is Pentek’s

SystemFlow Recording Software. A softwareAPI allows users to integrate control of the RTSinstrument into larger system applications.

The RTS 2711 features a Windows-basedGUI (graphical user interface) that providesa simple means to configure and controlthe instrument. Custom configurationscan be stored as profiles and later retrieved

for easy selection of preconfigured settingswith a single click.

SystemFlow also includes signal viewingand analysis tools that allow the user tomonitor the signal prior to, during, andafter a recording session. These tools includea virtual oscilloscope and a virtual spectrumanalyzer.

Built on a Windows XP Professionalworkstation, users can install post-processingand analysis tools to operate on the recordeddata. The RTS 2711 records data to the

native NTFS file system, providing immedi-ate access to the recorded data.

Recorded data can be offloaded from theRTS 2711 via gigabit Ethernet or USB 2.0ports. Additionally, data can be copied todisk using the 8X double-layer DVD +R/RWdrive on the front panel.

Flexible ArchitecturePentek’s RTS 2711 provides a flexible

architecture that can be easily customizedto meet user needs. Multiple RAID levels,including 0, 1, and 5, provide a choice forthe required level of redundancy. The totaldrive capacity is 4 TB using 16 drives whichare organized as two 8-drive, 2-TB arrays,one for each A/D channel.

Channels can easily be added to a record-ing system by adding more RTS 2711 chassis.The SystemFlow software provides the capa-bility of configuring and controlling multipleRTS 2711’s, enabling scalable channelcount and drive capacity.

SystemFlow Signal ViewerAdvanced signal analysis capabilities

include automatic calculators for signalamplitude and frequency, second and thirdharmonic components, THD (total harmonicdistortion) and SINAD (signal to noise anddistortion).

For more information please go to:Model RTS 2711. ❑