These basic guidelines help in the selection of the test instruments and software needed to assemble an effective RF and microwave automated-test-equipment (ATE) system.
These basic guidelines help in the selection of the test instruments and software needed to assemble an effective RF and microwave automated test equipment (ATE) system.
DENIS GABLE Director of Frequency Synthesizer Engineering Wide Band Systems, Inc., 389 Franklin Ave., P. O. Box 289, Rockaway, NJ 07866; (973) 586-6500, FAX: (973) 627-9190, e-mail: firstname.lastname@example.org, Internet: www.widebandsystems.com.
Automated test equipment (ATE) systems with RF measurement capabilities are used throughout commercial and military facilities when there is a need to perform a large number of repetitive tasks. In commercial applications, for example, RF ATE systems are commonly employed for screening the performance of production semiconductors and integrated circuits (ICs) used in wireless communications devices and infrastructure equipment. In military applications, specially designed RF ATE systems can complete the many tests necessary for qualifying the performance of tactical radios, radar receivers, and electronic warfare (EW) systems. In either case, assembling an effective RF ATE system is a matter of understanding the capabilities and limitations of different function blocks in the system and how to make those blocks operate seamlessly under software control.
Constructing an RF/microwave ATE system (Fig. 1) involves addressing a number of key requirements, determined by the type of device under test (DUT). Although RF/microwave measurements can vary widely from commercial and military applications, most ATE systems share a list of requirements, including
The choices of hardware instruments comprising an ATE system will depend on how these requirements are met. For example, the types of RF tests required for commercial wireless devices can be considerably different that the RF and microwave tests performed on military electronics equipment. Tests for commercial wireless devices are generally established by the requirements of a particular standard, such as the various versions of IEEE 802.11 for wireless local area networks (WLANs) or IEEE 802.16 for emerging wireless metropolitan area networks (WMANs) including WiMAX systems. Tests for military equipment, on the other hand, can involve specially designed waveforms and specific frequencies. An example is the set of advanced waveforms used for testing tactical military software-defined radios (SDRs) such as those developed for the Joint Tactical Radio System (JTRS) program.
Selection of a signal source or sources for an RF/microwave ATE system is dictated by the intended test applications, with a wide range of suppliers and products currently available. Test signal sources vary widely in price and performance, depending upon the frequency range, modulation options, spectral purity, and other performance factors. As a starting point, a signal source or sources for an ATE system should exceed the operating range of the DUT, such as the 2-to-18 GHz span of many radar systems. In some cases, it may be desirable to exceed the DUT's frequency range by at least a factor of 2 or higher, in order to generate and test DUT performance with second- or thirdharmonic signals.
An ATE test signal source should provide the resolution between frequency steps required to emulate the actual application for the DUT, such as switching among communications channels in a tactical radio. Frequency resolution ranging from 1 Hz to 1 MHz is common in modern RF/microwave signal sources, with even finer resolution possible in some digital synthesizers. The test signal source should also provide adequate dynamic range through sufficient maximum output power and through switchable attenuators. Such attenuators should be well designed, since they will impact the overall amplitude accuracy of the signal generator and, thus, the possible amplitude accuracy of the ATE system. The output power flatness and amplitude accuracy of a test signal source for ATE applications should be in the range of 1 dB to ensure minimal contributions to amplitude measurement uncertainty. Inadequate amplitude accuracy can lead to sufficient measurement uncertainty to fail DUTs during production testing that might have passed during those measurements given a more accurate test system.
The spectral purity (phase noise, spurious, and harmonics) of an ATE test signal source should exceed that of the DUT to accurately determine the noise floor of the DUT, especially for active devices. Phase noise, for example, is generally specified as the amount of noise below the carrier level when normalized to a 1-Hz bandwidth at some offset from the carrier. The noise, of course, is spread across the full operating range of the synthesizer, so the 1-Hz convention simplifies the comparison of noise levels from one instrument to another. The frequency and amplitude switching speed of a signal generator for ATE applications should be sufficient to support expected levels of test throughput for a given DUT. Finally, the test signal source should provide the modulation formats required for fully testing the target DUT. If not available internally in an instrument, the signal generator should provide a modulation input port so that the required modulation can be supplied by an external source, such as a pulse generator or arbitrary waveform generator.
At one time, many of the RF/microwave signal sources found in ATE systems were known as free-running signal generators or swept-frequency oscillators. These were instruments based on YIG oscillators or voltage-controlled oscillators (VCOs) capable of wide frequency tuning ranges but lacking in high frequency stability. In modern ATE systems, most signal sources are now frequency synthesizers that can duplicate the tuning ranges and switching speed of free-running sources, but with much improved frequency stability and spectral purity.
ATE architects have a wide range of commercial frequency synthesizers to choose from, with some key differences in performance. The most common class of instrumentation frequency synthesizers are based on indirect synthesis techniques usually employing a phased-lock loop (PLL) and a reference oscillator to stabilize the phase/ frequency of a tunable wideband oscillator, such as a YIG or VCO, by synchronizing its phase with that of the reference oscillator. The approach can provide a stable output when using a low-noise crystal oscillator, such as a temperature-controlled crystal oscillator (TCXO) or oven-controlled crystal oscillator (OCXO), as the reference source. Unfortunately, the tuning speed of this synthesis approach is limited by the time needed for the PLL to acquire and lock when tuning to a new frequency and is generally in the range of milliseconds.
When speed is an issue, a second class of frequency synthesizers employs direct-synthesis techniques to produce output signals often with nanosecond switching speeds between frequencies and/or amplitude levels. Such synthesizers often employ direct analog approaches, in which a low-noise reference source is multiplied, divided, and mixed to create a comb of signals from which a desired signal is selected or "tuned" by means of switched filters. They may also make use of direct-digital- synthesis (DDS) signal generation. A digital phase step word is accumulated, where after, the instantaneous phase is mapped to a sinusoidal amplitude to be converted to the analog domain by use of a digital to analog converter (DAC).The internal bit resolution of these devices is especially critical to the spectral performance of a DDS generated signal. Limited resolution results in high spurious levels and a need for subsequent filtering or other techniques to avoid undesired spurious signals. A direct-analog synthesizer can provide extremely high levels of performance in terms of signal phase noise and switching speed, but is also associated with high cost due to the large component count in terms of required filters, switches, multipliers, mixers, and other RF and microwave components. A DDS-based signal generator offers improved resolution when used in a direct-analog-synthesis approach, but still relies on the best possible reference or clock performance for good spectral purity.
In assembling an ATE system, concerns in specifying the test signal source or sources often have as much to do with mechanical constraints as with electrical performance. Especially in ATE systems aimed at military test requirements, such as radar testing, where multiple-emitter environments must be recreated in the test laboratory or on the production line, the size and power consumption of a synthesizer may be key considerations.
Conventional rack-mount frequency synthesizers may simply occupy too many rack spaces and consume too much power to support a practical multi-emitter test setup. In such cases, one solution involves the use of a compact, modular frequency synthesizer. As an example, a line of modular fast-switching frequency synthesizers from Wide Band Systems (www.widebandsystems.com) includes units that cover 2 to 18 GHz in a modular housing measuring just 6.50 x 6.25 x 1.05 inches and consuming only 19 W of DC power. Because of their small size, a number of these miniature synthesizers (Fig. 2) can be fit into the space normally reserved for a single conventional rack-mount frequency synthesizer, with multiple units still consuming less power than a single conventional-sized rack-mount synthesizer.
Of particular advantage in multiple- emitter ATE systems, the Wide Band Systems modular frequency synthesizers offer single, dual, or switched outputs with +10 dBm output power, ±1.5 dB output-power flatness over temperature and frequency, and 1.25- dB amplitude balance and 5-deg. phase balance between dual or switched outputs. They provide as much as 22-b frequency control for frequency resolution as fine as 5 kHz. For ATE systems requiring fast amplitude and frequency switching for increased test throughput, the modular frequency synthesizers boast maximum switching speed of 5 μs, with typical switching speed of 3 μs.
Of course, modulation is a key part of an ATE system??s capabilities and any test signal source specified for ATE use must deliver the required modulation formats. Modulation can be supplied internally--by a built in modulation source within the test signal generator--or externally, using a separate pulse generator, function generator, or arbitrary waveform generator as a modulation source. In either case, given the growing complexity of both commercial and military modulation formats, the modulation source should provide compatibility with waveform creation software such as MATLAB from The MathWorks (www.mathworks.com). Using an external modulation source based on digital signal generation, waveforms can be defined by means of a text editor or graphical waveform creation program to provide code inputs for the digital source, which then produces modulation waveforms by means of a high-resolution, high-speed digital- to-analog converter (DAC). These waveforms are then fed to the modulation input of the microwave signal generator or generators to produce high-frequency modulated waveforms for the ATE system and DUT.
The other side of generating test signals in an RF ATE system is detection and analysis. While ATE stimulus equipment can be as simple as a voltage supply, analysis equipment can range from programmable digital multimeters (DMMs) to broadband CW and peak power meters, spectrum analyzers, digital sampling oscilloscopes, and microwave vector network analyzers (VNAs). Again, the choice of analysis instrument or instruments depends on the test requirements for the DUT.
Again, test requirements for commercial DUTs can differ greatly from those of military DUTs. At one time, commercial testing involved simple analysis of frequency range, amplitude characteristics (such as output power and output power flatness), and scattering (S) parameters at higher frequencies using test signals with basic modulation formats such as AM and FM. With the rise of commercial wireless standards, however, the performance requirements of a DUT are considerably more demanding, requiring high linearity and wide dynamic range under more complex modulation conditions, typically involving digital modulation formats such as binary-phase-shiftkeying (BPSK), quadrature amplitude modulation (QAM), and quadrature phase-shift keying (QPSK).
For analyzing the performance of a DUT using simpler modulation formats, traditional analysis tools such as CW power meters and swept-frequency spectrum analyzers can provide the necessary capabilities for detecting and displaying information about the impact of the DUT on the test waveforms. But with more complex digitally modulated waveforms as used in modern wireless communications systems, the relatively narrow analog resolution-bandwidth (RBW) filters found in traditional spectrum analyzers lack the instantaneous bandwidth to pass and display the amount of spectrum occupied by digitally modulated signals. For that reason, a number of test equipment manufacturers now offer swept-frequency spectrum analyzers with digital resolution-bandwidth filters, or have even developed signal analyzers that digitize input signals in the manner of a Fast Fourier Transform (FFT) analyzer, to perform signal analysis in the digital realm.
This latter type of instrument essentially consists of a wideband receiver front end that feeds a high-speed, high-resolution analog-to-digital converter (ADC) to convert analog input waveforms into digital data. In the digital realm, digital-signal-processing (DSP) technology can be applied to the captured digital data to implement digital filters, perform inverse Fast Fourier Transforms, and extract information about the amplitude, frequency, phase, and time-domain characteristics of waveforms passing through the DUT.
Traditional spectrum analyzers are specified in terms of frequency range, frequency accuracy, frequency resolution, amplitude (dynamic) range, and amplitude accuracy. The resolution-bandwidth filters provide a range from very narrow (for differentiating signals close together) to very wide (for passing spectrally wide signals, such as modulated waveforms). The dynamic range of a signal analyzer is usually determined as the range between the smallest detectable signals at the analyzer??s displayed average noise level (DANL) and the largest detectable signals as determined by the instruments third-order intercept (TOI) ..
For measurements of some waveforms with extremely wideband spectral content, such as pulsed radar signals, the resolution-bandwidth filters of analog and digital signal analyzers may be inadequate in width to pass all of a test signal??s spectral content for analysis. To capture signals with wideband modulation or spectrally rich pulsed waveforms, spectrum analyzers can be used in zero-span mode to pass as much of the spectral content as possible to the analyzer??s digitizer section. In some cases, the use of a vector signal analyzer (VSA) can provide a wide instantaneous in-phase and quadrature (I/Q) bandwidth that may provide the spectral width to capture even broadband pulsed radar waveforms.
In addition to spectrum and signal analyzers, analysis equipment for RF ATE systems includes CW and peak power meters, VNAs, and high-speed digital sampling oscilloscopes (DSOs). As with the signal analyzers, these instruments should be specified on the basis of the test requirements for the DUT in terms of frequency range, amplitude dynamic range, and measurement speed (throughput). For example, measurements on radar equipment and components require an ATE system capable of performing characterization with pulsed signals. These types of measurements evaluate different signal pulse characteristics, such as pulse width, rise time, fall time, pulse repetition frequency (PRF), and peak and average power.
An average reading power meter can derive values for peak power from measurements of average power, but it can not determine peak power directly. A peak power meter is better suited for that purpose, equipped with a power sensor for the frequency and dynamic ranges of interest. Pulse signal measurements can also be performed with a microwave spectrum analyzer used in zero-span mode, with adequate instantaneous measurement bandwidth to pass the full spectral content of the pulsed signals. Newer spectrum analyzers are designed with digital resolutionbandwidth filters that provide near-ideal response shapes and permit analysis of pulsed (time-domain) signals without the group-delay distortion typical of analog filters.
For time-domain analysis of high-frequency and high-speed signals, commercial DSOs are available with measurement bandwidths to 100 GHz and sampling rates to 40 GSamples/s. Such instruments can be used to perform time-domainreflectometry (TDR) to find impedance mismatches and reflections with TDR resolution as fine as 20 ps.
An important task for an ATE system integrator is selection of the best power supplies for a particular application. As with specifying the other instrumentation pieces of an ATE system, the choices of programmable power supplies involves a complex set of options and a number of tradeoffs. In any case, a properly chosen ATE power supply should provide adequate power and voltage levels for the present application, with enough capacity for future applications.
Two of the more common programmable power-supply formats are standard GPIB-equipped supplies and VXI-based card-type supplies.1 Traditional GPIB programmable power supplies are controlled directly by an ATE system??s host computer or the auxiliary GPIB port of associated test equipment, such as a VNA. They are available as traditional single-box solutions and as modular devices that fit into a rack-mountable chassis. VXItype power supplies fit within dual-slot C-size VXI cards and include single-, double-, or triple-output configurations. In contrast to traditional GPIB programmable power supplies, VXI power supplies are light in weight and can readily integrate into VXI-based ATE architectures. Since they communicate with other VXI instruments directly over the VXI backplane, they are capable of high-speed data throughput compared to GPIB supplies.
Perhaps as important as supplying power is routing signals into and out of a DUT. This key function is usually performed by a programmable switch matrix which, as with the other ATE system components, can be specified in a variety of formats including as a GPIB rack-mount unit or a VXI card. Usually specified as an m x n matrix with a given number of input and output ports, respectively, switch matrices for ATE systems should also be carefully selected to meet the needs for current testing as well as future requirements in terms of adequate numbers of input and output ports, frequency range, switching speed, and power-handling capabilities.
Configuring an RF/microwave ATE system at one time automatically involved the use of the industry- standard IEEE-488 or generalpurpose instrumentation bus (GPIB) interface. ATE system developers now have a wide range of instrument control standards currently at their disposal, with many newer instruments now incorporating multiple control buses, such as GPIB, Universal Serial Bus (USB), and Ethernet Local Area Network LAN. As a further boost to the of instrumentation over LAN, the LAN eXtension for Instrumentation (LXI) standards promote automatic compatibility of instruments with LXI compliance over Ethernet LANs. LXI standards are promoted by the industry group, the LXI Consortium.2
In some cases, the speed of a dedicated modular bus approach, such as VXI, PXI, or PXI Express, with its high-speed backplane, may provide greater data throughput, software flexibility, and lower latency than a system based on older control and connectivity, such as GPIB. Among others, the US Department of Defense (DoD) has encouraged the use of a synthetic instrument (SI) approach in its ATE systems as a means of cutting costs and extending ATE system operating lifetimes. The approach is based on using basic instrument function blocks, such as waveform generators and digitizers, which are then defined as different instruments, such as voltmeters or spectrum analyzers, depending upon the needs of a particular DUT and set of measurements.
Given the large number of instrument control standards, a hybrid ATE system approach allows the system designer to combine multiple instrument buses and add newer standards without sacrificing system integrity. The hybrid approach can help meet new test requirements while extending the life of the ATE system. In developing a hybrid ATE system with multiple buses, the software framework employed is vital to ensuring successful integration and ease of flexibility to adjust to changing instrumentation buses.
Some system developers find graphical test development software, such as Agilent VEE from Agilent Technologies or LabVIEW from National Instruments, easier to use than textbased programming languages such as C++ or Visual BASIC. When cost is an issue, existing code should be reused as much as possible.
System developers at National Instruments3 recommend a layered architecture to separate hardware from software in an ATE system. This simplifies the integration of multiple platforms and eases maintenance. The architecture starts with the device input/ output (I/O) layer which contains the individual instruments and their buses, such as LAN/LXI, USB, and VXI. The next layer is called the computing layer, which includes the embedded and remote controllers used to control modular instrumentation and connect to different buses. Following this, the measurement and control services layer, with hardware and instrument drivers that link hardware to software. Then, the application layer contains individual test programs, such as power spectral measurements for a pulsed radar component. Finally, the system management layer provides a framework to organize and control test programs, manage data, generate reports, and separate operating controls for multiple users.