By John RHEA
Defense and aerospace systems designers must use as many commercial off-the-shelf components as possible to keep costs at a minimum, yet many of these devices were never built for the demanding environments of the military and pose their own set of test and measurement challenges
"Your test capability has to be better than what you're trying to test or you can't test it." This is the way Orion Wood, aerospace and defense marketing program manager at Agilent Technologies in Santa Clara, Calif., sums up testing commercial off-the-shelf (COTS) electronic parts for the rugged applications of the military services.
As a spinoff of Hewlett-Packard, Agilent inherited the commercial and the military test equipment businesses of that company and, as a result, has been able to apply the supporting technologies across the board.
Also, as is typical throughout the test equipment industry, Agilent's leaders have seen the dynamic telecommunications market drive their business. Telecommunications pressed its demand for ever more complex electronics and held out the promise for huge potential sales volumes.
However, this situation is changing, Wood points out. The telecommunications industry is maturing as the economy softens, which drives Agilent leaders to look for ways to leverage their commercial product line to respond to military needs.
Testing military electronics has evolved in recent years from the big racks of special-purpose equipment of the past to increasing use of off-the-shelf test equipment incorporated in multiple systems, Wood notes. An example of the former is the U.S. Air Force depot-level testing of the avionics in the F-16 fighter aircraft.
"We've seen the military desire for COTS, so we provide COTS test equipment," Wood says. "It's the same as the commercial realm. The military is trying to do all it can to take the cost out of the system, and we're trying to do the same."
The challenge he faces, which is common throughout the testing industry, is deciding what commercial test equipment he can use as-is and what he must modify for rugged military applications. The lines of demarcation are not entirely clear. For example, there are separate common testing needs for military and commercial satellites. The same is true of radar.
Just as the horizontal lines blur, so do the vertical lines. Testers must cover the entire spectrum of electronics from chips to boards to complete systems. This results in different responsibilities for different levels of vendors.
Multi-level test procedures
Semiconductor firms usually test their own components or outsource the work to specialized third-party test houses. The end users, either the system prime contractors or the military services that will deploy the systems, have their own acceptance test procedures.
In-between are board-level electronic products — Wood cites the example of a transmit-receive module for an airborne radar — that pose what he calls "a tremendous need for analysis of electronics." At the input end this is typically done with signal generators and at the output end, by spectrum analyzers.
As the signals become more complex, the testing equipment must become correspondingly more sophisticated, and that means understanding the analog and digital signals at ever-greater levels of resolution. All the signals going through the air are, by definition, analog. For testing, however, systems must convert these analog signals to digital data streams as early in the testing process as possible to achieve the desired resolution and to reduce testing time and costs.
Greater analog and digital integration sought
Wood says he is looking for a "synthetic instrument" that would incorporate superior analog-to-digital and digital-to-analog conversion so his experts could use the same instrument to test components in both directions. The trend, he says, is toward greater integration of the testing function.
Michael Gauthier, president of ICS Radiation Technologies Inc., a firm in Downey, Calif., that specializes in radiation testing components for space applications, also says he sees a commonality between testing COTS and ruggedized parts. "It doesn't make any difference if it's a space or commercial part, whether it's a standard op amp [operational amplifier] or the latest rad-hard version," he says. "Test methods haven't changed in 25 years, but the types of parts have. COTS-type parts have been used for space and the military for 25 years. We just didn't call them that then."
What has changed is the industry's approach to the test process. Now, instead of maintaining complete in-house facilities, more companies are going to outside vendors like ICS. Portable equipment has become available that enable testing companies to take their test gear on site. That way the customers can rent it by the hour. The end product is test results on CD ROMs.
Gauthier, a veteran of the NASA Apollo and other space programs, maintains that his firm can test systems more quickly, accurately, and inexpensively than the in-house facilities of the prime contractors. His company, in fact, will even rent time at government laboratories to get the job done, he says.
Designers cannot upscreen components destined for spacecraft because these devices all must be from the same wafer, he maintains. ICS experts will typically test five to ten out of a batch of 500 parts, mostly to levels of about 10 to 100 kilorads of total-dose radiation. Chip fabs must package these devices first because they cannot test bare chips. Curiously, components to be used in the International Space Station only must be tested to two to ten kilorads, but that is because the spacecraft is manned and has to be shielded sufficiently to protect the human inhabitants.
The ICS customer base also spans the specialized integrated circuit manufacturers — hybrid companies that buy chips and package them — as well as aerospace subcontractors without their own in-house test capability. "People pay you lots of money in this business to destroy their fancy toys," Gauthier quips.
Engineers at Excalibur Systems Inc. in Elmont, N.Y., buy specialized testing to make their board-level MIL-STD 1553 products competitive with what the major prime contractors can produce in-house. "If it walks or crawls on 1553, we do it," says Mark Blisko, president of the company.
For example, he points out, PCI interface cards do not come in military grades, so the users put them into gate arrays to achieve extended-temperature ranges. "The customer wants COTS pricing with military performance," Blisko says. He maintains that new plastic parts are even better than ceramics. Plastic parts, he says, dissipate heat better than ceramics, which improves reliability and lifespan.
Testing, like the boards themselves, represents a classical make/buy decision involving overhead structure and time to implementation. Excalibur engineers are comfortable farming out the specialized tests, Blisko says, noting that his test specialists can do a 25-to-30-year test involving shake and bake in seven to eight months. He expects to use these procedures in the next-generation 1553 that committees working under the auspices of the Society of Automotive Engineers are currently defining.
At Mercury Computer Systems in Chelmsford, Mass., Eric D'Agostino, principal mechanical engineer, makes the distinction between highly accelerated life testing, or HALT, and highly accelerated stress screening, or HASS.
HALT and HASS
Company experts do both for their line of scalable embedded multicomputing systems, which go into military and commercial applications. Mercury already has a diagnostic capability through its medical imaging line, which includes magnetic resonance imaging, computed tomography, and positron emission tomography.
Test experts perform HALT when products are fresh out of engineering and tested "until something breaks," D'Agostino says. When the tests reach what he considers a reasonable limit they generate a HASS profile that can run on each product that goes out the door. HASS thus represents an ongoing series of tests of the rugged products.
While test personnel can simulate some of the tests through computer-aided design, D'Agostino maintains that testing is necessary to validate board-level products. Furthermore, Mercury has only one test chamber and sometimes has to use third-party testing firms.
"The big plus of in-house [testing] is the cycle time," D'Agostino says. "You don't have to get in somebody's cue. That adds a level of complexity. The third party won't know your product, so you need your own people."
Set against that, however, is the added cost of test equipment, yet this need not be burdensome. Mercury engineers do HALT and HASS only on the ruggedized products. D'Agostino says he regards the process as an extra "safety net" for products that start out as commercial and then must meet industrial and military specifications. "The military customers are happy to pay for the safety net," he says.
At nearby Sky Computers Inc., also in Chelmsford, Mass., Edmond Hennessy, vice president for worldwide marketing and sales, says companies like his have to leverage the components of outside suppliers with their own in-house HALT and HASS, but that any increased infrastructure cost will have to be reflected in the manufacturing cost and therefore the selling price.
How much? Nobody in the industry wants to talk about this, but Hennessy says it is "maybe in the range of five percent" of product costs for the additional tests needed to verify that the COTS products will do the job. The problem is the suppliers' failure to do what he considers valid tests, Hennessy says. Instead, they use "reference designs" or "test by similarity," which Hennessy dismisses as "almost a marketing ploy."
In the case of Sky, which has focused on non-rugged applications, increased testing requirements have resulted from the company's efforts to span commercial and military market segments with its new Xtreme line of multiprocessors intended to cover both with one family.
Meeting the testing requirements for this line, in part, are the test facilities of Sky's parent company, Analogic Corp. It is a different story, however, for other companies, whose experts have no access to facilities such as these. Hennessy sees no easy way around this problem, and suggests his competitors must deal with test issues on a case-by-case basis.
Ballard Technology in Everett, Wash., specializes in the hardware and software interfaces that enable users to test avionics equipment with personal computers. Kevin Christian, customer services manager, says that these testing devices enable companies to use COTS devices where in the past they would have had to pay the premium for military-qualified equivalents.
For example, in the 1553 world Christian expects Compact PCI and PCMCIA cards in COTS versions can slash costs by a factor of 10 — from $30,000 to $3,000 — in non-flight-critical airborne applications. "We burn up a lot of nine-dollar cards in the process," he says, and the U.S. Air Force is already using PCMCIA cards in maintenance programs for radios.
Ballard's line of interfaces cover the ARINC protocols as well as 1553, all the prevalent operating systems and such old platforms as VME. The idea is to offer a family of devices that replace the military-qualified interfaces of the past and cover the spectrum of testing from research and development up to acceptance tests of deliverable rugged systems.
Using one instrument to cover a broad spectrum of test requirements is a strong idea, says Sandy MacMullen, chief scientist at Los-Angeles based Technology Service Corp. (TSC). He claims his firm can handle jobs ranging from a few milliwatts to multi-megawatt radar transmitters.
Measuring the performance of microwave components and systems is always a tough job, he says, because typical laboratory equipment does not provide sufficient accuracy. To fill this gap, TSC developed a special-purpose instrument called the microwave component analyzer that covers the 2-18 GHz frequency band.
This instrument measures the stability of microwave amplifiers by measuring the signal-to-noise ratio of continuous wave (CW) signals and intra/inter pulse noise and transients of pulsed waveforms. Other measurements include insertion phase and gain, and clutter attenuation of microwave components. The analyzer consists of a Pentium III personal computer with plug-in radio frequency (RF) and digital signal processing modules developed by the company.
The U.S. Navy paid for the system originally in 1998 under a small business innovative research contract from the Naval Surface Warfare Center, Crane, Ind., MacMullen says. The goal was to measure the stability of an upgraded version of the AN/SPS-48E radar transmitter from the ITT Industries Gilfillan Division in Van Nuys, Calif., for use in the Ticonderoga-class Aegis cruisers (CG 47).
These ships' radar transmitters had been based on vacuum tube technology. The upgrade was to replace a 60-kilowatt crossed-field amplifier stage with a solid-state pulsed amplifier. Generating 70 to 80 kilowatts, the solid-state amplifier combined the power from 160 550-watt modules into more efficient radial combiners. As is common in switching from tube to solid-state technology, the result was to increase system reliability, maintainability, and availability.
The measurement task was to determine the stability of the solid-state transmitter on a pulse-to-pulse basis quickly and accurately. At the outset of the program, Gilfillan engineers established a stability goal of 65 decibels in a radar subpulse bandwidth of 333 kilohertz. After completing the engineering model of the amplifier, TSC and Gilfillan teamed to measure the stability of the transmitter, a parameter that is fundamental to radar target detection in a high-clutter environment.
The TSC analyzer measured the stability of the transmitter signal as it passed through each stage of the amplifier chain, using the transmitter's input signal as a reference. The analyzer programmed the amplitude and frequency of the input drive signal using an external frequency synthesizer, and collected the transmitter RF output pulses through a directional coupler.
The stability of the solid-state transmitter surpassed Gilfillan's design goal by more than 10 decibels, according to MacMullen, and measurement accuracy was 0.2 decibels. He claims this approach improves the accuracy of these measurements by a factor of 10 and reduces measurement time by a factor of more than 100.
Now the company is trying to sell the instrument, for about $60,000, to "anybody who makes transmitters," MacMullen says.
The net influence of COTS on military and aerospace testing is to raise the bar on the analytical capabilities needed if the system producers are to capitalize on the inherent COTS advantages of lower price, state-of-the-art performance, and assured availability of the components.
The cost of this increased requirement eventually has to be borne by the prime contractors and is reflected in their price to the end user. The primes, in turn face, the classical make/buy decision on who will do the testing. The dominant figures in the industry can do much of this in-house, but a host of specialized firms is springing up to handle the most complex tasks and support the smaller firms that find compelling reasons to outsource this work.
Government users outsource independent operational test and evaluation
ALEXANDRIA, Va. — A new industry has arisen in recent years to conduct the operational test and evaluation of military and aerospace systems that government agencies do not have the expertise to do in-house and that the prime contractors cannot do themselves because it would constitute a conflict of interest.
One of these independent operational test and evaluation firms is Sentel Corp., headquartered in Alexandria, Va., with offices at user sites in Annapolis, Md., Dahlgren, Va., and Eglin Air Force Base, Fla. Jim Garrett, president of the firm, says companies in this industry offer a certain synergy by being able to "act like an operator" and conduct operational tests across related functions.
Sentel last month was selected by the Federal Aviation Administration (FAA) to support the FAA's William J. Hughes Technical Center, Atlantic City, N.J., across the board in such areas as radar, navigation, weather forecasting, communications, and air traffic control. These functions are similar, according to Garrett, and outsourcing them gives FAA an independent appraisal of its hardware and software contractors.
This is a competitive business, and Sentel had to displace the existing contractor, EER Systems, Lorton, Va., for a contract initially good for one year but containing options for eight additional years that could bring the total value up to $35 million. Sentel, like other companies in this industry, provide the same services for the military. —J.R.
Software source code analysis parallels hardware tests for COTS
COLUMBIA, Md. — In parallel with the testing of the commercial off-the-shelf (COTS) hardware for rugged systems, somebody must analyze the source code of the software to migrate that software to the COTS components, says Mike Smith, vice president for marketing at McCabe & Associates in Columbia, Md.
The purpose of this function, he says, is to quantify what he calls the "goodness of testing" to satisfy the U.S. Department of Defense. This is particularly important in such software-intensive applications as the Aegis shipboard radar, Tomahawk cruise missiles, and the Theater High Altitude Air Defense system — better known as THAAD.
These are among the projects tackled in recent years by McCabe. Smith says his experts use such tools as graphics and text, all working off the same code, to determine the "good, the bad, and the ugly."
The company's customers are typically the system integrators, and McCabe delivers a software license, usually in perpetuity. A standard configuration of quality assurance runs $40,00 to $50,000 for the source license.
The idea is to get everybody on the same page, so the software qualification spans integration testing through development testing, and involves the prime contractors and the government customers. In the case of the Aegis program, for example, the same tools were developed for prime contractor Lockheed Martin in Moorestown, N.J., and the Naval Surface Warfare Center, Dahlgren, Va. — J.R.
Who's who in test and measurement
Aero Nav Laboratories Inc.
College Point, N.Y.
http://www.aeronavlabs.com/
Agilent Technologies
Burlington, Mass.
http://www.agilent.com
Ballard Technology Inc.
Everett, Wash.
http://www.ballardtech.com/
Bell Technologies Testing Division
Orlando, Fla.
http://www.belltechinc.com/
CPU Technology Inc.
Pleasanton Calif.
http://www.cputech.com
Dynamic Instruments Inc.
San Diego, Calif.
http://www.dynamicinst.com/
Excalibur Systems Inc.
Elmont, N.Y.
http://www.mil-1553.com/
Fluke Corp.
Everett, Wash.
http://www.fluke.com/
GenRad Inc.
Westford, Mass.
http://www.genrad.com/
Geotest Marvin Test Systems Inc.
Irvine, Calif.
http://www.geotestinc.com/
ICS Radiation Technologies Inc.
Downey, Calif.
http://www.icsrad.com
IFR Systems Inc.
Wichita, Kan.
http://www.ifrinternational.com/
ISE Labs Inc.
San Jose, Calif.
http://www.iselabs.com
Integra Technologies LLC (formerly Lucent)
Wichita, Kan.
http://www.cetc.com/body.html
ManTech Test Systems Inc.
Chantilly, Va.
http://www.mantechtestsystems.com/homepage.htm
McCabe & Associates
Columbia, Md.
http://www.mccabe.com/main.htm
National Instruments
Austin, Texas
http://www.ni.com/
Perkin Elmer Optoelectronics
Santa Clara, Calif.
http://opto.perkinelmer.com/index.asp
Pikes Peak Test Labs Inc.
Colorado Springs, Colo.
http://www.pptli.com
Retlif Testing Laboratories
Ronkonkoma, N.Y.
http://www.retlif.com/
Sentel Corp.
Alexandria, Va.
http://www.sentel.com/sentel/index.html
Spectral Dynamics Inc.
San Jose, Calif.
http://www.sd-corp.com/
Technology Service Corp.
Los Angeles, Calif.
http://www.tsc.com
Tektronix Inc.
Beaverton, Ore.
http://www.tek.com/
Wyle Laboratories
El Segundo, Calif.
http://www.wylelabs.com/ad.html
