CompactPCI shows signs of a military build-up

CompactPCI shows signs of a military build-up

Already a leader in the telecommunications industry, CompactPCI designers turn their attention toward the military with new rugged specifications and improved performance, while PMC leads the way among mezzanine cards

By John McHale

Leaders of the VME single-board computer industry once claimed with confidence that the upstart CompactPCI was no serious competition for their VMEbus. They made this claim with such conviction because as recently as a year or two ago, they believed VME presented the embedded design community with a clear path to fast backplanes that moved data at 320 megabytes per second, and even faster. CompactPCI, meanwhile, would be stuck forever by design at 132 megabytes per second, they said.

Well, things change. CompactPCI experts are finding ways to speed up their backplane databus to compete even with the super-fast VMEbus versions of the future. At the same time, members of the VME and CompactPCI communities apparently are starting to cool their longstanding hostility, and in some cases are even starting to work together to improve their products. Yet looming on the horizon may be a new threat to CompactPCI, and it does not come from VME. Some experts are claiming that fast serial databus technology is set to replace CompactPCI on the desktop. Such a development, experts say, has the potential to undercut fundamental support for Compact-PCI and leave that technology like cut flowers in a vase — it will live for a time, but ultimately will wither and die.

At any rate, embedded systems designers stand to benefit from a growing selection of high-performance single-board computers and backplane databuses.

Proponents of CompactPCI are setting high performance marks to keep pace with VME320 while members of the VME community are joining hands with CompactPCI designers to conduction and convection cool CompactPCI single-board computers.

CompactPCI supporters at the PCI Industrial Computer Manufacturing Group (PICMG) in Wakefield, Mass., are waiting to approve a new CompactPCI backplane databus specification with a performance of 66 MHz. This bus, proponents say, will be able to move data at about 500 megabytes per second.

"The spec should be finished" before the end of this year, and should be more than a match for VME320, says Richard Somes, technical director of the OEM business segment at Compaq in Shrewsbury, Mass., and vice president for technology at PICMG. Somes could not say when he expects integrated circuit manufacturers to release interface chipsets for 66 MHz CompactPCI.

The new 66 MHz CompactPCI boards will have only four peripheral slots, which will make it easier for designers to meet the clock speeds necessary for 66 MHz, Somes explains.

VME320

Last year VME designers were heralding the coming of VME320 backplane as an answer to the challenge of CompactPCI. VME320 is expected to enable VMEbus systems to at least quadruple their performance and remain backward compatible with older VME equipment.

It is based on a star configuration designed by Drew Berding at Arizona Digital in Scottsdale, Ariz., and experts at Bustronics Corp. in Fremont, Calif. The VME320 approach essentially makes the backplane into a simple lumped capacitance instead of transmission line. This enables the rise times and fall times of the signals in the backplane to flow through the switching threshold without hanging or changing direction. This results in a fast signal and a fast protocol.

Silicon for VME320 should be out by the beginning of next year, says Ray Alderman, executive director of the VME International Trade Association (VITA) in Scottsdale, Ariz.

Experts at the PCI special interest group (PCI SIG) in Hillsborough, Ore., are also working on a message-based interrupt system and signaling protocol on CompactPCI that should handle some the interrupt concerns, Somes says.

Meanwhile back in the heart of VME country, CompactPCI experts are hard at work to extend the VITA standards for convection- and conduction-cooling from VME to CompactPCI.

The standard for convection-cooled focuses on 3U, 6U, and 9U Eurocard CompactPCI boards and is 98 percent finished with completion expected by the end of this year, Alderman says. The conduction-cooled standard is just getting started and will be finished toward the end of next year, he adds.

"The war [between CompactPCI and VME] is over," claims Doug Patterson, product manager for military systems at SBS Embedded Computers in Warrenton, Va., and writer of the CompactPCI conduction-cooled standard.

The VITA membership, which includes companies that produce CompactPCI, wanted to cool the heated debate between supporters of the two technologies, Alderman says.

SBS engineers already design conduction- and convection-cooled CompactPCI with their CR6 and CR7 boards. The CR6 will be flying as part of a test system on a military aircraft. SBS officials declined to comment on specifics of the project.

The CR6 rugged 6U board is conduction cooled with an extended temperature range of -40 to 85 degrees Celsius. Its designers have increased that board`s resistance to the effects of shock and vibration by using stiffener bars, wedge locks, and conformal coating. The board works with Intel Pentium and AMD K6 processors, while the CR7 works with the Pentium II.

Lingering rivalry

Although the membership of PCIMG and VITA may be better friends these days, some industry leaders are still taking up sides in the debate. Some say the application should determine the right board form factor, while others insist CompactPCI simply is not a wise selection.

The choice between CompactPCI and VME depends on the application, insists Mark Wade, western regional sales manager for VMIC in Huntsville, Ala. Programs that are workstation-heavy and need high-speed PCI look to CompactPCI, while fire control applications with heavy analog-to-digital activity lean toward VME, he explains.

The military has a long history with VME and will be slow to change, says Andy West, software product marketing manager for the PowerPC group at Radstone Technology in Towcester, England. "Our military customers" continue to demand VME, he adds.

"I have heard a lot about CompactPCI, but not about any real contracts being won," says Gorky Chin, vice president of advanced technology at Vista Controls Corp. of Santa Clarita, Calif. However, if Vista ever were to do a 3U form factor it would be in CompactPCI because 3U VME never took off, Chin says.

Vista provides the integrated mission management computers (IMMC) on the Global Hawk unmanned aerial vehicle from Teledyne Ryan Aeronautical in San Diego. The IMMC, a derivative of Vista`s Flight Control Navigation Unit, performs the aircraft`s flight controls and navigation functions. There are two IMMCs on each Global Hawk.

The IMMC hosts the Global Hawk`s flight control and vehicle management software and communicates with other vehicle subsystems via a MIL-STD-1553B data link, Ethernet, RS-422 and RS-232 serial data links, ARINC-429 interface, and a variety of analog and discrete I/O to perform flight control and management of the vehicle subsystems.

Patterson says he agrees with Chin on the lack of I/O on the 3U VME backplane as the reason for its small growth. The 3U CompactPCI has plenty of pins, which translates into plenty of I/O, he says. The 3U CompactPCI will see strong interest in programs with small-form-factor requirements and embedded applications, Patterson adds.

SBS currently flies 3U CompactPCI on the NASA Pegasus launch vehicle designed at Orbital Sciences Corp. in Dulles, Va.

Officials at the Motorola Computer Group in Tempe, Ariz., are looking to revive their Prologue line of 3U Compact-PCI and tweak it for today`s market, says Jerry Gipper, director of business development and planning at Motorola Computer Group.

Officials at DY 4 in Kanata, Ontario agree with their competitors at Vista Controls and Radstone in their opposition to CompactPCI in the military.

"We`re seeing VME holding up," says Duncan Young, director of marketing at DY 4. There is some concern among DY 4 customers that CompactPCI may not be a long-term technology, Young adds.

PCI will eventually be phased out of the desktop by serial technology, he explains. "I see it muddying the water more than CompactPCI," Young cautions.

"More and more people at trade shows ask about CompactPCI," says Ron Marcus, director of marketing at Synergy Microsystems in San Diego. CompactPCI brings a greater bandwidth to the table, he says.

CompactPCI will become a target for program managers with COTS in mind, says Stan Skronski, president and chief of engineering at Synergy Microsystems. It will have the performance with the 66 MHz specification and the access to desktop technology, he adds.

More and more customers are also asking Synergy engineers to extend the definition of commercial-off-the-shelf (COTS), Skronski says. In other words make the shelf larger and offer more products with extended temperature for rugged COTS, he explains. Hence, the inevitable standardization of convection- and conduction-cooled CompactPCI.

Custom form factors

Experts at Vista Controls have added "a neat little side business" in custom form factors for small embedded applications in avionics and vetronics applications, Chin says.

"They want our original off-the-shelf design, but the 6U form factor is too large for their particular application," Chin explains, adding that his colleagues have designed a custom form factor of their VME controller for the U.S. Marine Corps Advanced Amphibious Assault Vehicle.

Radstone engineers are also seeing a growing demand for custom form factor designs. Some Radstone customers, for example, want the COTS VME and PowerPC solutions in a SEM-E form factor for some avionics applications, Radstone`s West says.

SEM-E is a 5-by-5-inch form factor in three-quarter air transport rack chassis, Chin explains. "Other form factors we have requests for include 3 by 3 inches and 4 by 4 inches," he says.

Radstone recently won a contract from Harris Corp. in Melbourne, Fla., to supply a rugged PowerPC 603e processor and PMC 1553 module on a large form factor into the U.S. Army`s Multiple Launch Rocket System (MLRS) upgrade program.

"We extended the board to 220 millimeters," which enabled "us to keep a PMC site on the board," West explains.

The MLRS is a mobile rocket artillery system, deploying 12 surface-to-surface rockets in two six-rocket pods. A computerized fire control system integrates the vehicle and rocket launching operations.

The present program will use Radstone`s products as a part of an upgrade to the computerized fire control system. The new system will enable the use of advanced mission software with enhanced capability to handle complex munitions.

Next-generation I/O

While memory capacity and processor performance have grown exponentially over the last five years, I/O speeds have not, says Joe Pavlat, president of PICMG and director of strategic planning at Motorola in Monterey, Calif.

"The next-generation architecture for enterprise-level computers must move massive amounts of data around a large, distributed array of servers," Pavlat says. PCI will not be able to evolve to meet the demands of the new architecture, he predicts.

"The bottleneck in a high-performance system has changed," Pavlat explains. "It isn`t the processor`s ability to crunch numbers any more; it is the ability of a system to ship data around in an efficient manner. Not only is the PCIbus starting to look a bit slow, but any parallel bus has a hard time increasing speed, number of nodes, and distance between nodes, all at the same time."

Two different groups in the industry are attempting to solve this problem using "switch fabric" I/O, Pavlat says. "Intel leads Dell Computer Group, Hitachi Limited, NEC Corp., Siemens Information Communication Network, and Sun Microsystems in a consortium developing a version called Next Generation I/O, while a consortium composed of IBM, Compaq, Hewlett Packard, Adaptec, and 3Com are developing a competing architecture called Future I/O," he says.

"In the switched fabric I/O concept, high-speed point-to-point links are made between processor main memory and I/O subsystems such as network controllers or storage arrays," Pavlat explains.

The concept was first introduced at IBM decades ago in the original 360 architecture, Pavlat says.

"The links are very scalable in the sense that they can connect chips to chips, boards to boards, processors to I/O subsystems, chassis to chassis, and racks to racks," he explains. "Because they only connect two points together at any given instant, they have several advantages over conventional parallel buses."

Both designs use copper interconnects for distances of a few meters and fiber for distances as far as a few hundred meters. But Future I/O is a combination parallel/ serial architecture, transmitting a byte per clock tick instead of a bit. The initial Future I/O data rate is a 1 gigabyte per second in each direction simultaneously, Pavlat says.

However the groups differ over philosophy, Pavlat explains. The Next Generation I/O consortium is more of an autocracy ruled by Intel, and Future I/O is more of a democracy led by IBM, Compaq, and Hewlett Packard, he explains.

Intel officials believe they can concentrate on the hardware and somebody else will figure out the software, Pavlat says. "They may be right," he says. Intel does not have a strong history with serial buses, he cautions. Their 1394 bus never really took off, Pavlat adds.

The Future I/O consortium wants to solve all the issues in detail and "carefully think out electrical, hardware, and pack-aging, interconnect, and software protocol issues," Pavlat says. "They are encouraging wide participation and involvement on the part of the industry."

However, they are about a year behind the Next Generation I/O group, planning specification late in 1999, prototypes in 2000, and production in 2001.

Future I/O is proprietary and will fail miserably, VITA`s Alderman predicts. "It will be the next fart heard round the world," he adds.

There will be no compatibility, Alderman says. Each vendor — Compaq, IBM, and Hewlett Packard — will have a different proprietary version, which will eliminate the idea of a standard, he says.

Intel already has prototype Next Generation I/O hardware and successfully demonstrated it at the Intel Developer Forum earlier this year, Alderman says.

"The [prototype`s] architecture consisted of host control adapters (HCA) communicating with target control adapters (TCA) through a serial crossbar switch fabric at 2.5 gigabits per on each link, or about 300 megabytes per second," Alderman explains. The gigabit links can be point-to-point connections directly between HCAs and TCAs, or through the crossbar switches.

"Next Generation I/O is targeted at very high-performance, multiprocessor servers," Alderman says. "These large machines will require a high level of redundancy for their target applications."

Next Generation I/O and Future I/O are a classic tortoise and the hare case study, Pavlat observes. Future I/O is the tortoise trying to solve all the problems and Intel is the hare getting in to the market quickly, hoping to make money fast, and dominate, he says.

Mezzanine boards

PMC is becoming the mezzanine of choice for military designers because it is popular on the desktop — and the desktop drives the technology in a COTS world, says Robert Perez, sales and marketing manager at SBS GreenSping Modular I/O in Menlo Park, Calif.

IndustryPack is still a strong seller, but it does not have the PCI connection to the desktop that PMC and PC-MIP has.

PC-MIP is the newest mezzanine standard on the market and was designed at SBS GreenSping, Motorola Computer Group, and Micro Electronik Neuremburg (MEN) GmbH of Germany. It is small enough that six modules may mount a 6U VME, 6U CompactPCI, or full-size desktop PCI slot board, including room for processors, bridge chips, and trace routing.

Another reason for PMC`s popularity is the growth of 100BaseTX Ethernet, which is too fast for IndustryPack, Perez explains.

Most CPUs have PMC expansion sites and will incorporate a PC-MIP site as well, Perez says. The military is already looking at PC-MIP and the advantages of its small footprint, Perez says.

Engineers at the Raytheon Computer Products Directorate in Marlborough, Mass., have developed the first plan to use PC-MIP in a U.S. military application — the U.S. Air Force Airborne Laser program.

A PC-MIP carrier board that Raytheon engineers designed would upgrade the IndustryPack devices now in the Airborne Laser program. The new board is to use PC-MIP with MIL-STD 1553B databus technology. Like PMC and IndustryPack, PC-MIP will eventually be conduction-cooled, VITA`s Alderman claims.

SBS GreenSping will also offer 1553 versions of PMC and PC-MIP. The PMC-1553 provides two dual-redundant MIL-STD-1553B bus terminals in one wide PMC form factor. Multiprotocol support of 1553B as a bus controller, remote terminal, or bus monitor is part of the package. It supports front panel I/O via two high- density 15-pin (video-style) connectors.

The P1-1553 PC-MIP module provides a dual-redundant MIL-STD-1553B bus terminal in one wide Type I PC-MIP module. The multiprotocol support of 1553B is identical to PMC-1553. However, a remote-terminal-only-version is also available.

DY 4`s Young says he still likes IndustryPack for its small form factor. "I still see a lot of interest in IndustryPack because of its simple I/O and because it has little in the way of detailed design," he explains.

DY 4 engineers announced earlier this year a high-speed Fibre Channel network interface PMC card that offers connectivity with real-time performance by delivering high bandwidth and low overhead communication in harsh environment, real-time, mission-critical system area networks.

Known as the PMC-642, the new card is a dual-port, 1.0625-gigabit-per-second PMC Fibre Channel card that provides an interface between compute elements and Fibre Channel networks

Vista Controls engineers are also offering a Fibre Channel PMC interface with their Model PFC PMC card. The new card achieves a sustained transfer rate of 102.7 megabytes per second, and supports 1-gigabit-per-second rates over copper and fibre optic media.

The Model PFC also enables dual- redundant connections in point-to-point applications. The card is available in laboratory, rugged air, rugged conduction, and full mil-temp conduction versions.

Motorola Computer Group engineers offer the PPMC 750, a PMC card that uses a MPC 750 PowerPC microprocessor with Motorola`s PowerPlus II architecture.

Motorola officials claim the new device simplifies product development and helps speed time to market by enabling OEMs to consentrate on designing and debugging high-speed memory, cache interfaces and 350Mhz microprocessor logic.

SBS GreenSpring Modular I/O also offers PMC-Video, a high-performance graphics adapter solution for PMC carrier boards. It supports a variety of SVGA/CRT monitors and applications in single display configurations, and features the Silicon Motion LynxEM710 accelerator and 4 megabytes of high performance on-chip SGRAM for high resolution graphics displays.

"PMC-Video is an advanced performance graphics accelerator designed for any embedded application that requires high resolution graphics," says Chris Thompson, general manager of SBS GreenSpring Modular I/O. "The module`s basic design—which includes only the neccessary features required to make it an ideal CRT application interface—enables us to keep the price point lower than offered by many competitors."

Single-board players

Dynatem engineers in Mission Viejo, Calif., are supporting 366 MHz on their Pentium II DRC1, which has been available for about a year and has previously been available with Pentium II versions running at 266 MHz.

The low-power characteristics of the Intel Mobile Module combined with the rugged design enable use of the DRC1 in harsh environments. Current versions of the DRC1 support a temperature range of -40 to 85 C. Recently experts tested the DRC1 for compliance with MIL-STD-901D in a system with an isolated chassis. The DRC1 has also been upgraded to support up to 128 megabytes of surface-mount SDRAM.

Experts at General Micro Systems of Rancho Cucamonga, Calif., announced their release last month of the Mini DDrive II, which they claim it is the first VMEbus mass storage card to feature two IDE hard drives and an LS120 super floppy.

The new card enables integrators to add mirrored hard disks and super floppy mass storage to their VMEbus systems by eliminating the need to bolt the drives to the chassis and provide cumbersome cabling. The Mini DDrive II also enhances security by making it practical for operators to remove the hard drives from the system on a daily basis.

"With the Mini DDrive, bolt-on mass storage is a thing of the past for VMEbus systems," says Ben Sharfi president at General Micro. "Now operators can install and remove hard drives the same way they would a ZIP drive on a PC. And field service personnel can perform system upgrades without having to dismantle the system."

Designers can configure the Mini DDrive II, which occupies one 6U VMEbus slot, with either two IDE drives and a standard 3.5-inch floppy, or one IDE drive and an LS120 Super Floppy. Each 2.5-inch IDE drive provides a capacity of as much as 9.2 gigabytes and is available with factory pre-installed operating systems. The rugged 2.5-inch drives, each 10 millimeters high, can withstand shock and vibration as strong as 100 Gs, General Micro Systems officials claim.

Engineers at Omnibyte in West Chicago, Ill., are offering the Galaxy PowerPC+ VME single-board computer, which they designed to use the latest generation of powerful PowerPC 603, 604, or 740 processors, running at speeds as fast as 350 MHz. As much as 256 megabytes of standard or synchronous x32, or x36 DRAM options are also available.

The Galaxy options include a high-speed Mercury Raceway Interface using the PXB (CY7C966) PCI to a Raceway interface, PCI to SCSI PMC interface module using a Symbios 53C895, and a VxWorks Board Support Package to enable application development using the Wind River Systems "Tornado" multi-hosted development environment.

Cetia experts in Burlington, Mass., continue to offer the ruggedizer, a board-level heat sink designed to lower the overall operating temperature of the board, while adding mechanical stiffening and protection.

No one really offers a better rugged solution than Cetia`s ruggedizer, VITA`s Alderman says.

The PMC-video graphics adapter solution for PMC carrier boards from SBS GreenSpring modular I/O has a basic design which lowers its price for CRT applications.

DY 4 provides COTS VME solution for vetronics on U.S. Army Grizzly

Engineers at DY 4 Systems in Kanata, Ontario, are the sole commercial-off-the-shelf (COTS) VME manufacturer for the vetronics package in the U.S. Army`s new Grizzly combat-engineering vehicle from United Defense L.P. in York, Pa.

The Grizzly Vetronics System (GVS), with an integrated, open-system architecture, gives each vehicle`s two-person crew the capability to drive the Grizzly as well as control the mine clearing blade, power driven arm, and self-defense weapons. The GVS also provides interface, status reporting, built-in test, and system/subsystem control functions and will keep the Grizzly going in the event of non-catastrophic damage.

The GVS`s primary and secondary controllers use two DY 4 DMV-977 chassis, each with two DY 4 circuit boards — a DMV-177 general purpose processor mated with a MAX-657 mezzanine module and a DMV-783 graphics board. The DMV-177 features the high performance and low power consumption of a 100 MHz PowerPC 603e microprocessor with 64 megabytes of DRAM with EDAC and 16 plus 2 megabytes of Flash EPROM. The DMV-783 is a dual-output graphics card with sensor and video overlay characteristics.

While the primary controller handles central processing in the GVS, the secondary controller acts like a 1553 remote terminal in normal mode and is capable of serving as a 1553 bus controller in the event of failure in the primary controller. A DY 4 MAX-657 mezzanine card provides interface to the 1553B databus and digitized speech functions.

DY 4`s chassis with processors and graphics boards also provide the Army with a common vetronics core across platforms such as Bradley, Crusader, Amphibious Armored Assault Vehicle (AAAV), M1A2 Abrams, and Heavy Assault Bridge (HAB), DY 4 officials say.

The Army`s Grizzly is designed to counter minefields, barbed wire entanglements, tank ditches, and other fortifications, while still keeping pace with forces on the battlefield. The Grizzly is currently in engineering and development for the Army Tank Automotive Command in Warren, Mich., and is scheduled to be fielded in 2004. — J.M.

The vetronics system of the U.S. Army`s new Grizzley comabt-engineering vehicle uses a rugged single-board computer solution from DY4 Systems.

VMIC provides VME boards for U.S. Navy Aegis cruisers

Engineers at the Litton Corp. Guidance and Control Division in Woodland Hills, Calif., needed a rugged VME solution for data acquisition and workstation electronics to be used on U.S. Navy Ticonderoga-class (CG-47) Aegis cruisers.

So they turned to experts at VMIC in Huntsville, Ala., to provide products that aid Litton in upgrading old systems with modern digital control for propulsion, engine plant, electrical plant, damage, and fuel control.

In addition, VMIC products are being part of the Operator Station Units as part of Litton`s new Integrated Bridge Systems (IBS), which provides real-time monitoring and display of the ship`s operations.

VMIC VME single-board computers used in the program were the VMIVME-7591 and the VMIVME-7696.

The Litton application was not dependent on I/O like a fire control application, and could have used CompactPCI, says Mark Wade western regional sales manager. Litton chose VME because it was what they were used to, he adds.

The VMIVME-7591 uses one Pentium with speeds of 233 MHz (with MMX) or an AMD-K6 processor with speeds up to 300 MHz. Features include an external (L2) cache — 512 kilobytes of synchronous-burst SRAM, 256 megabyte SDRAM using 144-pin SODIMM, and on-board Fast Ethernet controller supporting 10BaseT and 100BaseTX .

The VMIVME-7696 is a Pentium II processor-based VMEbus CPU running at 450 MHz. Board features include 256 megabyte SDRAM using 144-pin SODIMM, on-board Fast Ethernet controller supporting 10BaseT and 100BaseTX interfaces, and PCI fast/wide SCSI-2 with VMEbus P2 I/O.

Special VMIVME-7696 features for embedded applications include eight to 48 megabyte bootable-flash on secondary IDE, three programmable 32-bit timers, and 128 kilobytes of battery-backed SRAM.

The $11 million contract also includes equipment that Litton will use for a land-based engineering system, which will provide for operator training and system engineering development. Litton is expected to award an additional contract for spare equipment in support of this program, VMIC officials say. — J.M.

TheVMIVME-7591, pictured above, will be part of an Aegis combat ship upgrade to difitize propulsion control.

PC/104 ideal for small embedded systems

PC/104 continues to be the embedded solution of choice for military and commercial designers who are not married to a backplane requirement, says Rick Lehrbaum, vice president of strategic development at Ampro Computers in San Jose, Calif., and cofounder of the company. Ampro engineers first introduced the PC/104 specification in 1986.

"Venture Development Corp. [in Natick, Mass.,] projected a 30 percent compound annual growth rate for the PC/104 market over the next three years," Lehrbaum says.

The PC/104 form factor can also be used as an extension for VME boards similar to what mezzanine cards do, Lehrbaum adds.

Compared "with the tradition of industrial backplane buses, PC/104 and PC/104-expandable single-board computers may well be considered an out-of-the-box approach," Lehrbaum says. "Whereas backplane approaches were mostly limited to use in rack-mount industrial computers, the non-backplane PC/104 approach offers a practical alternative to full-custom system designs in space and cost constrained embedded computer applications.

"Thanks to PC/104, there are now two main approaches to using commercial-off-the-shelf computer boards in embedded applications," continues Lehrbaum. First is using backplane buses with multiple slot cards plugged into them. The second approach involves stacking or sandwiching non-backplane computer modules.

While PC/104 is different from backplane-based 3U CompactPCI and VME, the designers of those systems complain that PC/104 does not have the ruggedness of the Eurocard form factor.

Lehrbaum admits that he does not have any military specifications for his products, but points out they still seem to survive in extreme temperatures.

He is referring to the recent Highly Accelerated Life Testing (HALT) of Ampro`s Little Board/P5x as a part if its engineering qualification. Experts at Qualmark Corp. in Denver, Colo., performed testing at an accelerated reliability test center, using a standardized procedure.

The Little Board/P5x operated flawlessly outside of its published standard and extended operating environment, Ampro officials claim.

The temperature lower operating limit and lower destruct limit were found to be -100 degrees Celsius or better, Ampro officials say. The temperature upper operating Limit was 140 C and the temperature upper destruct limit was greater than 150 C.

The Little Board/P5x contains all the functions of a fully configured PC-compatible system. Included are the latest disk, LCD/CRT display, fast Ethernet, and UltraSCSI controllers, plus four FIFO-buffered serial ports, an enhanced capabilities parallel port, PS/2 keyboard/mouse interfaces, USB, IrDA, CompactFlash socket, and laptop-style advanced power management.

Ampro`s Little Board P5/x is currently flying on the Boeing 777 cockpit in the maintenance access terminal display designed by engineers at Rockwell Collins in Cedar Rapids, Iowa.

Engineers at Parvus in Salt Lake City, Utah, offer the PC/104 Quad Motherboard II, which provides a platform for mounting as many as four PC/104 cards in a space less than 1.3 inches high. The PC/104 Quad Motherboard II is for low-height applications, yet requires a space as small as 10.5 by 8.25 inches.

The board consists of four PC/104 16-bit headers with mounting holes and four sets of 0.6-inch standoffs to mount four PC/104 form factor cards. DC power flows to the motherboard through either a 5-pin screw clamp terminal or a 4-pin power connector.

JUMPtec Industrial Computers engineers in San Diego offer the MOPS/586 133MHz PC/104 form factor board. The device is a CPU module with 4 megabytes of DRAM on board. With a SO-DIMM memory module, as used in notebooks, the DRAM may be expanded to a total of 36 megabytes. — J.M.

Radstone and Synergy speed up VME with P-zero connector

Experts at Radstone Technology in Towcester, England, and Synergy Microsystems in San Diego are speeding up the data path between system boards by connecting the bridges between boards with the P0 connector. This connector sits between the P1 and P2 connectors provided by the VME64x form factor.

The Synergy Microsystems device is called the P0-PCI, which provides the benefits of a PCI connection between boards in a VME system. Boards connect with Synergy`s PØ-PCI minibackplane overlay on the backside of the VME backplane. Synergy offers it on the company`s VGM5, VGR5, and VSS4 CPU boards.

There are two advantages of going through the P0 connector instead of the P2 connector, says Ron Marcus, director of marketing at Synergy Microsystems. First is the avoidance of crosstalk with VMEbus traffic on the P2, and second is freeing up pins on the P2 that could be used for I/O on a PMC module.

Radstone uses the P0CC1 carrier card, which is connected via its P0 connector to a PCI infrastructure consisting of a single-board computer fitted with a PMCPCI and a P-PCI backplane.

This is an important development for commercial-off-the-shelf equipment designers who want to maintain a VME legacy product, says Andy West, software product marketing manager for the PowerPC group at Radstone. Designers can achieve greater bandwidth without upgrading their whole system, he explains.

The Radstone single-board computer with PMCPCI plugs into the master slot of the P0 range of P0 PCI backplanes. The P0CC1s then plugs into this PCI backplane system. The P0CC1 and PMCPC1 cards employ Digital 21150 PCI-to-PCI bridges to create secondary PCI busses, which support concurrent transactions with their primary PCI busses.

The bridge chips also enable buffering of PCI data and addresses for posted memory write operations. The primary bus on the PMCPCI card is that of the single-board computer. The secondary bus of the PMCPCI forms the P0 backplane PCI bus that connects to the primary busses of the P0CC1 cards. The secondary bus on the P0CC1 connects to the PMC sites.

The P0 feature is standard on the conduction-cooled board, but it can be used on all Radstone boards, West says. — J.M.