During the last few years, test engineers have seen a steadily increasing stream of information concerning synthetic test. The history and meaning of this term as well as a number of claims about its advantages have been made in a variety of articles. If you are a test engineer or a test engineering manager who has experience with high-volume commercial and perhaps cPCI/PXI-based test system architectures, you probably associate synthetic test with virtual instruments. If you are in the military-aerospace (mil-aero) market and have been involved with various government contracts, you probably identify synthetic test with NxTest and ARGCS (Agile Reconfigurable Global Combat Support). In all cases, it is difficult to find an example that takes into consideration the most practical aspects of evaluating and selecting a microwave synthetic test environment for a test application characterized by high-volume requirements as well as high-measurement performance requirements. This article will describe the selection process that leads a customer to choose a synthetic implementation for testing transmit-receive (T-R) modules utilized in a phased array radar system.
Compared to virtually any commercial segment, mil-aero testing does not traditionally deal with high-volume requirements over a relatively short period of time. Consider the case where a mil-aero company successfully breaks into the business of building T-R modules for phased-array radars. When this happens, the company in question typically moves from building one or a few units-per-radar to building hundreds or even thousands of units-per-radar, depending upon the radar application. Imagine being the test engineer working with a team of mixed-signal microwave component engineers who have spent the last few years designing and prototyping T-R modules. After multiple proposals, uncounted number of meetings, numerous design changes, and the accumulation of literally thousands of hours of test data on a few prototypes, the engineering team has won the business of building several tens to hundreds of radars – each requiring at least two orders of magnitude more modules than the total number of prototypes built to-date. In fact, the first manufacturing contract (block or tranche) may very well take about the same time to complete as the entire development period. And don’t forget about the added requirement of needing to build and test extremely well-matched and high-performance T-R modules to the tune of a thousand-times the number built during the engineering development and proposal stage.
The first production test plans are drawn and an analysis is made to correctly size and select the type of test equipment and number of stations needed to meet the production requirements. In support of the analysis, test time benchmarks are derived based upon the engineering test experience with prototypes. At the end of this process, the findings indicate that each module will require about four hours of test time, if the same approach that has been utilized during the engineering development phase is also applied to production. Considering the quantity of modules to be tested and the length of the contract, what are the financial and schedule impacts that these findings could have and what can be done to find a realistic, yet profitable solution?
DEVICE UNDER TEST
Before answering the question above, let’s find out a bit more about the DUT. A T-R module is basically a miniature transmitter-receiver (transceiver) that needs to amplify and transmit phase- and amplitude-controlled wideband pulsed signals as well as to receive them. The transmit-receive function requires fast switching and high isolation. All stages, especially the I/P ports, have to be very well matched over the entire frequency band of operation. The T-R circuitry needs to be as efficient as possible (especially on the transmitter side) to reduce the power requirement as well as to generate as little heat as possible within the smallest possible form factor. This is very important, because a certain number of modules will have to fit in a relatively small space. Heat and reliability (besides continuous and consistent performance) become a primary concern. Other parameters, such as harmonics, compression point/output dynamic range, third order intercept (TOI), receiver noise figure (for best sensitivity), and duty cycle are also very important.
Consequently, to ensure the quality and compliance of the DUT, all the parameters mentioned above need to be carefully measured, optimized and validated during development, and quickly verified during production. Additionally, since these modules are frequency agile, can be electronically steered through a programmable phase shifter, and have variable output/input power, all the characteristic parameters need to be verified under a large variety of conditions and states. All of this leads to a measurement space defined as the number of measurement points in a multi-dimensional measurement volume that truly requires high-speed and high-measurement accuracy to achieve low uncertainty and allow testing to be conducted efficiently and economically. The test environment also has to be extremely stable in order for all measurements to exhibit good correlation, both for the same module and among all modules, since the quality of an array will depend equally on the individual quality of each T-R module.
PRODUCTION TEST PLAN
When testing in an engineering mode, not only all parameters and characteristics of the T-R module prototypes need to be tested extensively but standard performance parameters are also methodically exceeded to be able to provide both performance (how much of an out of spec condition the module must endure) and manufacturing (how much of a variance due to components and production activities) margins. In this case, tests are bound to be repeated extensively, as the necessary modifications and rework are carried out, until the required performance and manufacturing parameters can be guaranteed. Consequently, during the engineering phase, test activities are almost as intense as they are in manufacturing, but they are more focused toward repeating tests on a few modules than performing tests once for each of many modules. During this phase, many companies start defining the test environment needed in production in order to avoid duplication of efforts and also to provide a seamless transition of the DUT test environment from engineering to production.
When testing in production, speed and accuracy are key. In order to be cost efficient, the production environment requires as few test stations as possible (with a minimum number of operators), with the highest throughput possible, while maintaining the highest possible yield. In this situation, two additional factors are of great importance:
• A system-level calibration that is fully consistent and traceable to recognized standards (e.g., NIST). This ensures accuracy, low uncertainty, and consistent measurement results.
• Consistency between the quality of the test environment in production and the test environment used during development. This contributes to high yield and highly correlated measurement results.
TRADITIONAL VS. SYNTHETIC
There are five key time elements in the stimulus-measurement test process: DUT setup, including warm-up; stimulus setup; response setup, calibration and actual measurement.
In a traditional system, a central computer or controller will have to coordinate all the steps of the test process. The calibration of the individual instruments has to be factored with the calibration of interconnecting cables, connectors, etc., which increases the measurement uncertainty from test station to test station. Also, one must establish some level of synchronization among the various instruments to ensure proper signal injection, capture, and analysis. For T-R modules, the DUT requires power, control, communication, monitoring, and needs to be synchronized with the rest of the equipment as well.
In a synthetic system, the stimulus, response, and DUT interface are all tightly integrated and synchronized. The calibration is performed at the system level, up to the DUT interface. Consequently, sources of uncertainty are greatly reduced. The whole system operates as one homogeneous environment and not as a collection of several items.
FIVE VS. ONE MEASUREMENT CHANNEL
Considering the measurements required by T-R modules, five traditional instruments are needed: spectrum analyzer, network analyzer, noise figure analyzer, modulation analyzer, and power meter.
In a synthetic system, however, all necessary measurements are made through a single response channel, often utilizing common digitized data. One does not need to switch among various instruments in order to make measurements, and there is only one measurement channel vs. several measurement channels to be considered.
Measurement time should only be limited by the performance and latency/settling characteristics of the DUT. Once this becomes the case, the test environment is optimized for speed.
For traditional equipment, measurement time depends on each of the instruments being used as well as the switching of measurement and data paths, synchronization, and central control/sequencing at the system level. Measurements in themselves cannot be optimized since they are rigidly associated with individual instruments and cannot take advantage of common, sampled/digitized data blocks.
For synthetic systems, measurements can be optimized by sequencing the various computations to take advantage of common data blocks, which optimizes the measurement space. There is no switching of instruments and measurement/data paths and the environment is inherently synchronized, including the DUT. Control/sequencing takes maximum advantage of the capabilities of all system resources involved. Again, the environment is homogeneous and it allows for continuous computing, signal processing and conditioning operations.
Looking at various benchmark tests run on T-R modules, and considering test campaigns with an average of twenty different test cases, traditional rack-and-stack test systems require approximately three to four hours of test time. A highly optimized synthetic test environment requires only four to six minutes of test time, where one of these minutes is allocated to DUT warm-up time. Even considering benchmarking approximations and the time it would take to cycle DUTs in the production test environment, using a synthetic test environment can increase test speed by 30 times.
When comparing the costs of these two types of systems, a hybrid synthetic test environment, when first procured, is typically within 10 percent of the cost of a traditional rack-and-stack system. Considering the test time saved over a period of time, the hybrid synthetic test environment has superior cost characteristics. If we also take into account the fact that fewer test stations with a smaller footprint and lower power requirement are needed, the economic advantage is undeniable.
FUTURE PROOF AS A SUPPORTING ELEMENT
An additional item that needs to be carefully examined in the selection process is the degree to which both systems can manage obsolescence, adapt to new requirements, and be reconfigured to support multiple products and product lines over extended periods of time.
Traditional rack-and-stack systems manage obsolescence with a “fork lift” approach, i.e., they are replaced. However, if one type of instrument is no longer made, even replacement becomes impossible and test procedure and test execution will have to be modified to accommodate a different instrument. As far as adapting to new requirements and reconfiguring a traditional solution for different products/product lines, this is certainly possible as long as the new requirements fall within the performance envelope of the associated instruments. As an example, if a higher frequency of operation is needed for some or all measurements, all instruments involved must be able to support the extended frequency. Those instruments that do not will have to be replaced, even if just one measurement is involved at the higher frequency.
Synthetic systems, on the other hand, are particularly flexible when it comes to managing obsolescence and changing requirements. Since synthetic test solutions are measurement-based, as opposed to instrument-based, they are impervious to instrument obsolescence. When new measurements are needed, they can simply be coded within the synthetic test environment (as long as the performance envelope of the overall system supports them). If a new requirement translates into a measurement that falls outside of the existing envelope, all that is needed is to replace or add one or more modules whose characteristics – for example, higher frequency of operation – expand the system’s performance envelope to include the new measurement.
PRACTICAL CASE
Once the selection of the optimal test environment type is made, there is still need to implement/customize the environment for the particular application. While it is possible for the end-user to purchase or develop various modules and software, and then attempt to build a synthetic system, there are ready-made solutions that greatly simplify this process. One such solution is Aeroflex’s SMART^E (synthetic multifunction adaptable reconfigurable test environment). In this particular example, the module test environment (with operating frequency up to 40GHz) was selected as the base environment for this test application (see Figure 1).
The criteria that led to this customer selection were based not only on performance and cost of ownership/operation, but also on greatly reduced start-up costs. These included activities associated with the integration (not simple interfacing) of the DUT in the test environment and the conversion of the test plan into a series of executable test programs/sequences.
Some of the tests conducted on the T-R modules are: Pout vs. Pin; total absorbed power; harmonics; noise figure; pulse measurement; recovery time; s-parameters; spurious; and TOI (third order intercept). Figure 2 depicts a screen capture of Pout vs. Pin as well as the associated tabulated data. Figure 3 depicts a pulse-measurement screen capture and some of the associated tabulated data. Figure 4 depicts a spectral/spurious measurement and some of the associated tabulated data. Cumulatively, all the tests listed above were automatically performed in a matter of minutes. Besides test reporting, SMART^E has also extensive test log and debug capabilities. Figure 5 depicts a screen capture of log/debug activities. For production test, batch sequences can also be used to speed up test activities, while still providing pass/fail criteria and results. Figure 6 illustrates such capabilities.
The SMART^E 5100 T-R module test environment’s first large-scale adopters are two major mil-aero companies in Europe. It is being utilized by these companies for T-R module testing associated with synthetic aperture radars both for air defense and satellite systems.
Click here for the illustrations: Figure 1, Figure 2, Figure 3, Figure 4, Figure 5, Figure 6 |
