Background

Trends in military product design have been toward modular, solid-state RF microwave architectures, taking advantage of major improvements in solid-state RF component design. For example, radar architectures are now based on using thousands of solid-state modules packaged in manageable assemblies of up to 30 modules each. This shift in design architecture has precipitated demand for a flexible high-speed, high-quality RF microwave test system.

In 1999, Raytheon embarked on a venture of developing such a system. Besides high throughput and versatility, the goals of the project included logistic goals that would address lowering life cycle costs, technical goals that would achieve high performance, and an architecture that would enable it to maintain technical excellence.

Logistical Goals

The main focus was to develop a system that would permit lowering life cycle costs. The goal translated into a common platform that could be used in a broad spectrum of applications. If achieved this would enable:

Technical Goals

Architecture was a major consideration at the core of the technical goals. The objective was to have a modular system based on industry standards from both a hardware and software viewpoint and minimize dependence upon proprietary designs.

RF Capabilities

The first criterion was measurement speed. Experience with heretofore “high-speed” test systems achieved test times of approximately a half-hour for solid-state assemblies. Classical rack and stack RF test systems needed hours to test complex receiver-exciter assemblies. Goals were to reduce these times by factors from 3 to 10.

Nearly as important as speed was RF measurement performance. A fast system with marginal performance would not have a wide range of applications. This new system had to have measurement performance capability similar to that of typical commercial test instrumentation. For it to be able to do the job, the system needed to have an extensive measurement suite. In the RF world, where cable losses and other transmission line issues can be serious problems, an instrument with good measurement capability at its front panel is only half the solution. Being able to easily extend the measurement all the way to the DUT was a prime consideration. Hence, having flexible calibration options was high on the list of priorities.

Microwave Synthetic Instrument (TRM1000C)

The Aeroflex TRM1000C is designed to provide reconfigurable, high-speed production test equipment for evaluating a variety of different microwave devices such as amplifiers, transmit and receive (T/R) modules, frequency translation devices, receivers, local oscillators, and phase shifters. It can also perform tests on integrated subassemblies of RF components, as well as on full up-systems filled with any combination of active RF, multiport devices.

The basic architecture of the TRM1000C is consistent with the basic architecture described in this book, a CCC cascade, enhanced with compound signal conditioners. The compound conditioners consist of a stimulus up-converter and a response down-converter. Time multiplexing is used to expand the system to multiple inputs and outputs. A calibration and verification system allows for loopback ordinates and application of metrology standards. Figure 5-1 outlines this high-level functionality.

The TRM1000C is designed to dramatically improve module test times and reduce measurement errors introduced by the operator, test hardware, and DUT interface. It is ideally suited for production test applications where throughput and flexibility are paramount. Through its synthetic design, the TRM1000C’s nonspecific RF hardware can be software configured to run specific RF and microwave production tests. The TRM1000C hardware architecture is based on advanced synthetic instrument concepts; the TRM1000C does the same measurements as several distinct microwave test instruments including a pulsed power meter, a frequency counter, multiple sources, a spectrum analyzer, a vector network analyzer, a noise figure meter, and a pattern generator. The full measurement suite is as follows in Table 5-1:

A standard TRM1000C includes complex stimulus generation including pulsed modulation, AM, FM, phase modulation (PM), and a fast response measurement channel. Since the system is synthetic in design and thereby easily reconfigurable, the system can be reused for different programs and applications, therefore maximizing return on investment (ROI).

If you look carefully at the block diagram in Figure 5-1, you can see that the internal design of the TRM1000C follows the standard synthetic measurement system CCC architecture principles for both stimulus and response. A compound up-converter in the stimulus, and a compound down-converter in the response-side orients the TRM1000C architecture toward the generation and analysis of bandpass signals, as is appropriate to its RF measurement mission. A calibration matrix interconnects stimulus and response with the DUT, providing stimulus-response closure. This eliminates many redundancies (for example: duplicated channels, stimulus-side detectors, response-side sources) that would otherwise be necessary in order to maintain calibration.

Supplemental Resources

Practical design realities limited the scope of what could be achieved with a purely synthetic measurement system. For testing the more complex RF products (receiver-exciter elements, frequency translation devices, and so forth), additional nonsynthetic resources were required. These instruments do not need to be inside the high-speed loops of the synthetic instrument and therefore do not interfere with measurement speeds. The supplemental test instrument resources include three RF sources, an oscilloscope, digital multimeter (DMM), and power meter (for troubleshooting purposes). These instruments, the RF switch matrix, and DUT interface were incorporated into Raytheon designed 3rd bay to complement the TRM1000C.

Product Test Adapter Solutions

Some of the products to be tested on the RFMTS were known to be large assemblies. For this reason, time was spent on developing calibration schemes that could be extended out several levels and on providing a rugged, high-performance interface. This approach has permitted using the system on virtually any size RF product. It also accommodates other relatively unique items that support the test, such as pneumatics, hydraulics, liquid cooling, and so on.

Primary Calibration

The TRM1000C utilizes the modular, line replaceable unit (LRU) methodology as part of the system. These LRUs are calibrated at a standard metrology calibration lab and become an integral part of the system. The majority of the LRUs are commercially available NIST traceable standards. The production floor can have spares available to remove and replace, minimizing system downtime when performing periodic maintenance. The LRUs also eliminate the need for external, on-site support equipment; therefore, the user does not need to bring external equipment up to the system.

A list of primary calibrated LRUs is as follows:

Operational Calibration

This is an application-specific procedure that transfers the measurement standards from the system’s calibration LRUs to the system itself. This calibration can handle a number of different multiport devices and any number of DUT interfaces. The DUT interfaces can be as simple as NIST traceable coaxial connectors (3.5 mm), or the interface can be more complex: non-NIST traceable coaxial connectors (for example, GPPO), standard and nonstandard wave guide, or even direct wafer probes. A multitier calibration technique is used to extend the calibration reference plane out to the DUT (de-embed).

1. Ease of use by TPS developers and test system maintainers.

2. Sufficient depth and flexibility to accommodate dealing with very complex, digitally-controlled RF products.

3. The different levels of software employed by the TRM1000C synthetic instrument.

Test Program Set Developer Interface

The objective here was to provide a simplistic means of implementing moderately complex tests, without demanding that all test program set (TPS) developers maintainers be experts in C programming. The solution was to develop C-based test procedures that would be graphical in nature, and then utilize a test executive that would readily permit stringing these test procedures together. This approach requires a few C experts, but permits test engineers to be productive with minimal software specific training.

The test procedure concept simply translates typical RF measurement scenarios in a graphical user interface screen or “panel” where the required parameters can be entered. The TPS designer enters the associated parameters via the panel for each test the designer develops. Test procedures have been developed for all of the measurement types.

A primary objective of the test procedure approach is to maximize reuse and provide cost effective TPSs in a timely manner. The approach has proved to be very effective. Test engineers are able to concentrate on the technical aspects of the DUT and the test scenarios. The pure software designers are doing what they do best—developing the C-based procedures in support of the test designers. In the event of complex tests, the LabWindows/CVI environment provides a very flexible environment for developing custom procedures for unique or even further speed enhancement if required.

TRM1000C Software

The TRM1000C currently uses a scripting language called JavaScript (ECMAScript). Scripts are used to define the logic, control, processing and storage of the measurement and resultant data. Any number of scripts may be loaded at any time and sequenced via a test executive or incorporated as part of a test procedure and then called by an executive. Through commanding of the scripts, the TRM1000C may change from one measurement personality (i.e., vector network analysis) to another (noise figure).

There are two categories of scripts that can be designed: low level and high level. Low-level scripts are interpreted and run one step at a time. A low-level script is not optimized for speed, but is best for situations where the number of firmware states is unknown. Such a situation may arise when conditional branching is utilized. An example of this is when a measurement result is used to decide whether another measurement needs to be made.

High-level scripts are unrolled and loaded into a state table within the processor. The processor can then execute the state table without any interaction with the script. This method is optimized for speed, but requires a known number of states.

After the measurement is performed, the requested data is stored in a local data file (slot zero controller). Since the scripting capability allows for complex, multidimensional data sets to be collected and the speed of the system allows for considerable data to be collected quickly, these data files can be somewhat large. To minimize the data file size and to allow for easy access, the TRM1000C currently utilizes an open file format called hierarchical data format (HDF). Measurement results can then be either routed back through the host driver link as part of the script or the HDF file can be retrieved remotely by the host. Any number of different test executives can be used with the TRM1000C. Test executive software running on the host PC performs all data presentation and report generation activities.