Overcoming the Legacy Equipment Replacement Blues

Investing in custom test systems is seldom a popular decision although it may be the best choice among a number of unattractive alternatives. Nevertheless, custom military/aerospace test systems come with a large amount of risk and expense. The exact requirements may be changing as the equipment to be tested is developed.

Custom spare parts must be provisioned together with the basic systems, which increases costs. A big investment must be made to develop, test, and verify the software test program sets (TPSs) that actually perform the automated testing. And, after a number of years, the likely obsolescence of some commercial off-the-shelf (COTS) test instrumentation may threaten the useful life of the entire test system.

Dealing with COTS equipment obsolescence generally is termed legacy instrument replacement. Legacy hardware issues have attracted a great deal of attention, especially from the personnel tasked with purchasing and maintaining critical test systems as well as from the test system suppliers. Among the many factors complicating legacy instrument replacement is the investment already made in the related TPSs, often several man-years of effort and many millions of dollars.

It's as though you needed to replace a broken part on an old car that you used for daily transportation. The part is no longer available, but you have invested $20,000 for a custom paint job and leather upholstery. You might decide to use a rebuilt part, explore other sources of parts, or modify a part from some other car. If the car ran well except for the one broken part, chances are that you wouldn't buy a different car and again pay for custom paint and upholstery.

This is the situation in which military/aerospace equipment maintenance depots find themselves. A substantial investment in TPSs has extended over several years, TPS and hardware problems have been corrected, and the systems generally work well. During the time the systems have been in use, the equipment they are being used to test often has been upgraded or refurbished to extend its life. The end result is aging but working test systems, a new requirement for them to remain in use much longer than originally planned, and a legacy hardware issue that must be addressed.

There are several approaches being used today. Perhaps the simplest solution is to replace the obsolete equipment with a direct form, fit, and function (FFF) equivalent. Some test and measurement instrumentation companies make this kind of product, duplicating the performance of the legacy instrument even down to known eccentricities and undocumented features. With an FFF solution, you simply replace the old instrument and carry on as before. No changes are required to the TPSs.

Many COTS instrument models never were used in large enough numbers for production of FFF equivalents to be economical. In these cases, it may be possible to emulate the older instrument through a combination of new hardware and special software.

A few companies specialize in this kind of TPS migration, distinguishing between this term and translation because a level of performance verification is anticipated. As with FFF hardware, some legacy emulation efforts are claimed to be an FFF solution.

Instrument manufacturers provide command sets compatible with legacy TPSs. This may be a simple solution in specific cases but is not as straightforward as it appears.

A modern instrument indeed may perform a superset of the functions associated with the legacy instrument. However, will a 1-??s-wide pulse generated by both have the same noise level? Will similar frequency square waves have the same rise and fall times? Will it take the identical amount of time for a command to be executed?

Unfortunately, some TPSs work correctly because of these kinds of secondary instrument features. When a new command-compatible instrument replaces the original instrument, the TPSs may not work as well or may even create errors. Unless a replacement instrument can be guaranteed to exactly duplicate the original, additional time and money may be required to verify that the TPS still operates as intended. Nevertheless, TPS translation is yet another approach that can be effective.

In the future, synthetic instruments (SIs) may minimize the effect of instrument obsolescence. With a system based on SIs, generic modules are configured and reconfigured by software to perform the required instrument functions. However, SIs are relatively new, few if any deployed test systems are wholly SI-based, and experience will tell whether this approach lives up to its claim.

FFF Instruments

Geotest-Marvin Test Systems makes a range of direct replacement pulse, waveform, and function generators. A typical unit is shown in Figure 1. Several models of obsolete Wavetek, Tektronix, and Hewlett-Packard generators can be replaced with these 100% compatible instruments.

David Manor, vice president of engineering, explained, “The designs are based on newer technologies and can operate faster or offer higher performance than the legacy instruments. Because it's an FFF replacement, however, these features are not exposed to the user.

“It's important that the new instruments can emulate deficiencies or undocumented features of the legacy equipment. These anomalies are mostly found in the host controller interface. All of our FFF replacement instruments use an FPGA-based architecture, which gives us the flexibility to easily update functionality in the factory or in the field,” he said.

The Pulse Master Series of pulse/arbitrary generators from Tabor Electronics also can emulate several obsolete HP, Fluke, LeCroy, and Tabor generators. Although the Tabor name may not be familiar to some readers, the company has extensive test and measurement instrument experience, having designed and brand-labeled pulse generators sold for many years by leading instrument companies. Tabor products are marketed in the United States by EADS North America Defense Test & Services.

Obsolete HP and Agilent Technologies microwave signal generators can be emulated by the Giga-tronics 2500A Series. In addition, ASCOR, part of Giga-tronics, provides FFF VXI modules that replace obsolete switching products from Tektronix, Racal Instruments, VXI Technology, Datron, and Interface Technology. A few FFF VXI modules not related to switching, such as the Model 3000-4730 12-Channel D/A Module that replaces the Tektronix VX4730, also are available.

Products developed as part of ASCOR's Tektronix VXI replacement card program have identical types and locations of front-panel connectors, the same signal on the same pin, equivalent or improved performance, hardware compatible with the original Tektronix cards, and a plug-and-play driver that handles a mixture of ASCOR and Tektronix modules.

An FFF instrument is the only drop-in replacement that does not require extensive performance verification. The instrument manufacturer has done that work as part of the development, precisely because the product will be used in legacy replacement applications. The bottom line is less risk, lower cost, and shorter downtime.

Compatible Instruments

For Teradyne's M9-Series Digital Test Systems, the company's modern Di-Series Digital Test Instruments is a direct replacement. It can emulate obsolete instruments but also includes new capabilities such as per-channel programmable low-voltage differential signaling (LVDS) transitions. A TPS used with L-Series Test Platform instrumentation is compatible at the source-code level and requires conversion via the TPS Converter Studio tool. The Di-Series instrument driver supports legacy M9 binary code or translated L-Series source code as well as new TPSs through a native C# interface for system integrators and a C-code interface for TPS developers.

The Di-Series and the Ai-760 Analog Test Instrument offer increased density and flexibility. A Di-Series module can be configured as independent virtual instruments and has per-pin capabilities. While some of these features may be used for legacy instrument emulation, the Di-Series also provides room for growth of TPS to address new requirements.

The Ai-760 maintains the earlier Ai-710 parallel test capability but really is aimed at system integration. It replaces multiple conventional instruments such as a 6.5-digit DMM, a timer/counter, an Arb, a digitizer, and a VXI oscilloscope. Clearly, higher density is not necessary in an FFF replacement instrument, but it would be useful were a test system to be redeveloped to address further requirements.

Teradyne's motivation in developing the Di and Ai Series was to support the company's own current as well as legacy test systems. Nevertheless, both the Di and Ai Series are considered core system instrumentation products and were designed to replace other manufacturer's legacy instruments such as digital word generators.

Agilent makes a wide range of products that can replace legacy instruments, usually older HP or Agilent models. The company offers software that translates command sets so that programs written to run on an older model will still run on a new one. Of course, that doesn't guarantee that any undocumented features will be supported, just intentional ones.

A new product can be compatible with a TPS but is not necessarily an FFF replacement for the instrument originally controlled by the TPS. In some cases, making such a replacement might cause no problems at all. Comprehensive TPS validation is needed when a compatible instrument is introduced into an existing test system unless that instrument is FFF guaranteed.

TPS Translation/Migration

According to EADS Business Development Manager Mike Rutledge, “Translation can be a low-risk process given a disciplined approach coupled with a few critical software tools. If the legacy system was well engineered, then the replacement modern instrument can support the required TPS functionality in a similar vein. In addition to preserving previous investments, modern instrumentation and technologies can provide a basis for planned technology insertion. This ensures that customers have the capability to migrate to future workloads without the need to reinvent the wheel.”

EADS prefers the term migration rather than translation because some amount of validation is anticipated. Mr. Rutledge commented that translations are never 100% perfect because too many variables are involved. On the other hand, they can approach 90% with the remainder of the task handled by sound engineering practices in a structured environment. The company has been doing this type of work for more than 10 years.

Geotest also is involved in translating TPSs, but primarily those that use digital subsystems. Mr. Manor explained that the host controller interface plays a less significant role, making translation practical. The critical issue is having a set of tools that can accurately translate test vectors and sequences to the target digital subsystem. In Mr. Manor's experience, translating or porting digital test vectors to a new digital subsystem requires less effort and has less risk than emulating a host controller interface with its undocumented features.

WinSoft® makes a command processor that can emulate a range of instruments. The processor is presented with the unmodified TPS, and it then drives new instruments to emulate the performance of the legacy equipment. Up to 13 instruments can be emulated by one WinSoft Instrumentation System Emulator (WISE™) box, a number set by GPIB limitations. Agilent has worked with WinSoft on WISE projects to refresh system life for many customers using old HP equipment (Figure 2).

Ehud Shany, WinSoft's president/CEO, said that several of his company's customers in the military/aerospace markets had not had success with TPS translation. “The time and money involved with the translation are about 50 times more than with the WISE approach. And, the translation cannot be done in those cases where the source code is not available or if the ATE is classified. In addition, revalidation of the software can be costly, as much as the translation process itself.”

SI Systems

A DoD initiative created a working group for RF SIs, initially known as the Synthetic Instruments Working Group (SIWG). This group completed specifications for synthetic DACs, up-converters, down-converters, and ADCs. The IVI Foundation has taken over with the intention to create IVI classes for these modules as well as switching, signal processing, and software-algorithm or signal-conditioning SI blocks.

SIs attracted a lot of attention after their successful integration in the Agile Rapid Global Combat Support (ARGCS) technology demonstration exercise. In this case, a common core of SI hardware and software replaced legacy RF and microwave test equipment from older test systems. The project proved that SIs could replace legacy instruments and be interoperable. The same SI configuration could perform equivalent measurements that once ran on four types of testers using different hardware.

Test systems based on SIs make sense in many ways. They save space, weight, power, and money and have an upgrade path for future requirements. On the other hand, there are limits to what is practical. For example, a group of functional modules with similar frequency ranges typically is packaged together and programmed to perform a number of measurement jobs. It might be less advantageous to combine precision DC instrumentation with a high-frequency digitizer and RF generator.

Within well-defined functional limits, SI-based test solutions are becoming available from several suppliers. A good example of an SI-based test system that from the outside appears to be a single instrument is Teradyne's Bi-41x Bus Test Instrument. Using the same hardware, it can handle RS/IEA-, ARINC-, and MIL-STD-1553-style buses. These types of serial buses are considered relatively low speed but otherwise quite distinct and, in the past, required separate hardware instruments.

The Bi-41x also is compatible with variations on standard buses intentionally or unintentionally designed into a weapons system. Without the flexibility of an SI-based solution, past test systems have required additional test-specific instruments to address nonstandard buses.

Despite the attractiveness of SI systems, Charles Greenberg, EADS senior product marketing manager and an active participant in the SIWG for two years, cautioned, “The unique capabilities of a legacy instrument may not be duplicated by a generic, purely synthetic approach. Care must be taken to match the measurement modes very closely to avoid changes to the TPS necessary for the replacement SI.

“We sometimes take a hybrid synthetic approach to legacy replacement by assembling COTS synthetic building blocks with custom signal conditioning circuits,” he continued. “In this way we can create the combinations needed to reproduce the required legacy signals and measurements.”

Giga-tronics has combined its signal generation, up/down conversion, and switching expertise in flexible and reconfigurable SI systems. On the other hand, much of the IP developed as instrument firmware in stand-alone products now must be provided in the SI control software. This makes system development more difficult and the resulting software more complex.

Agilent's Application Note 1465-24, Using Synthetic Instruments in Your Test System, covers many aspects of SI-based systems. Comparisons are made between the SI approach and the VXI, PXI, or GPIB architectures. Very broadly, SI-based systems shift the emphasis from hardware to software (Figure 3).

In particular, software component interchangeability is an area that SI vendors must address through new software tools. Software module reuse is as important as hardware reconfigurability in reducing SI system development cost.

Agilent Product Manager Dan Pleasant highlighted the software standardization that SIs support: “An SI module by itself performs fewer functions than a classical test instrument. This makes it easier to write standard IVI drivers that cover all of a module's functionality. It follows that it also should be easier to write application software using those drivers that works well with different underlying test hardware.”

Against this advantage, Mr. Pleasant noted the increased complexity of SI control software. When SI modules are used, a system designer must acquire and embed the necessary measurement science previously resident in separate instruments. NIST traceability will be difficult to ensure because no single manufacturer has control. And, with no front panel, an SI-based test system can be more difficult to debug.

A further implication of reconfigurable SI-based system development is the need for a thorough review of calibration and self-test capabilities. Geotest's Mr. Manor commented on replacing a legacy instrument with one based on SIs. “For complete compatibility, it also is necessary to accurately emulate the legacy instrument's self-test function to ensure that the system's self-test functions still execute correctly with the replacement SI. Using an SI provides flexibility, but the implementation/verification/certification effort can require significant time and effort.”

Conclusion

A test depot manager's main responsibility is described succinctly: He must keep the test systems running. Problems arise when an obsolete instrument has to be replaced. If replacement coincides with a TPS upgrade, perhaps the manager's available options are greater. Some amount of TPS revision might be accommodated given that the software is being changed anyway.

If only the legacy instrument is to be replaced, an FFF solution appears to attract the least risk and perhaps is the lowest cost solution. It certainly is the easiest form of replacement assuming that a suitable FFF instrument exists. If one does not, the choice widens to include compatible instruments and instrument emulation.

WinSoft's WISE emulator is claimed to be an FFF solution based on instruction translation and new, compatible hardware. Mechanically, a 1U-high emulator must be housed together with the new instrument, but this may be only a small inconvenience. The WISE solution requires no TPS changes so it generally is preferred to an approach that does involve TPS modifications and verification.

Similar emulation solutions from EADS, Teradyne, and Geotest also preserve the TPS investment, which often is the primary objective. Depending on the instruments involved, some degree of TPS change may be required.

Longer term, test system rationalization via a common set of SI modules and software components may occur. Certainly, SI-based test systems are becoming more prevalent as are specific types of flexible instruments internally based on SI-style architectures.

WinSoft's Ehud Shany distinguished between SI-based systems designed to minimize future obsolescence problems and stand-alone replacement instruments that address current obsolescence. Teradyne's Peter Hansen, a product line manager, emphasized the major impact SI-based systems can have on life-cycle costs. Because the cost of adding instruments to a test system often is underestimated, the flexibility of an SI-based system can be very important.

To assist initial test system instrument specification as well as later legacy instrument replacement, a standard way to describe instrument capabilities using the Instrument Description XML format is being developed. EADS's Chris Gorringe, co-chair of the ATML working group, explained, “ATML allows the instrument performance specification and capability to be captured in a common and reusable manner. This means that users can compare and contrast different instruments, both for requirement evaluation and as a legacy replacement exercise. Obviously, the real benefit will come when all instrument specification sheets are expressed online in this format.”