Pulsed signals are widespread in radar and other electronic warfare (EW) applications, and they must be accurately measured for manufacturing, design of countermeasures, and threat assessment. Pulse measurements are an especially challenging area for signal analysis due to a combination of factors. Fortunately, many of the improving signal processing and analog-digital conversion technologies behind the generation of complex pulse environments also enable new techniques for effective pulse analysis.
In the past, basic pulse measurements generally were made with swept spectrum analyzers. The intermediate frequency (IF) bandwidth or resolution bandwidth (RBW) of the spectrum analyzer was generally narrower than the effective bandwidth of the pulse, so the spectrum analyzer was used to measure the resulting pulse spectrum. The pulse spectrum could then be used to measure basic signal characteristics such as pulse repetition rate or interval (PRI), duty cycle, power, etc. Spectrum analyzers were also used in more traditional ways to make out-of-band measurements such as spurious and harmonics of pulsed signals.
Though indirect and slightly clumsy, the pulse spectrum approach was adequate for simple pulses and signal environments containing only a single pulse train, and where frequency agility was low or could be inhibited. Modern systems use much more complex pulses, and many signals or signal environments include different pulses (along with other signals) from one or multiple emitters, as shown in the real-time spectrum measurement of Figure 1. The combination of complex signals and detailed measurement requirements means that pulse measurements must now be made using digital signal processing (DSP) techniques on digitally sampled signals.
Choosing RF/Microwave Hardware and Software
A critical first step is to choose the main measurement hardware platform. Rapid increases in signal analyzer bandwidths and improved resolution in digital oscilloscopes are constantly changing the tradeoffs that affect pulse measurements. Two different RF/microwave hardware measurement platforms are generally used for this purpose: signal analyzers with a wideband digital IF, and oscilloscopes or digitizers with a sampling rate high enough to directly handle microwave RF/microwave signals at baseband.
The two hardware front end approaches are conceptually similar for most pulse measurements. In both cases, the output of the RF/microwave front end (including subsequent processing) is a stream or data file of I/Q samples of the signal or signal environment. The principal architectural difference is the location of the analog-to-digital conversion (ADC) operations and the type of processing used to focus analysis on the frequency band of interest. Signal analyzers use a fundamental or harmonic analog mixing process and analog filters to convert RF or microwave signals to an IF section where ADC operations are performed. Oscilloscopes (and other time domain samplers such as modular digitizers) sample the RF or microwave signals directly in a baseband fashion, and subsequent downconversion and band-limiting is performed by DSP.
While signal analyzers and oscilloscopes can make many of the same measurements, the best choice in a hardware front end is often dominated by two performance requirements: bandwidth and dynamic range. The high-speed ADCs in RF/microwave-capable oscilloscopes provide extremely wide bandwidth and good phase linearity. In contrast, the slower ADCs and bandwidth filters of the signal analyzers provide higher dynamic range. Where their bandwidth — now as wide as 1 GHz — is sufficient, they have a greater ability to detect and measure small signals, or to handle both large and small signals at the same time.
One practical advantage of the signal analyzer as a measurement platform is that it can support seamless switching among swept, vector, and real-time measurements in a single instrument. By using smart external mixers, this single instrument — via a single user interface — can provide these capabilities over wide bandwidths and up to 90-GHz operating frequencies.
Once a stream of wideband sampled signal data is available, a variety of software solutions are available to meet different analysis needs. Two major types of software are generally used. Built-in software and installable measurement applications have been available for oscilloscopes for some time, and their analysis is focused primarily on pulse timing parameters and time domain measurements. Built-in applications extend pulse analysis to the frequency and time domains in signal analyzers with wideband capability.
Vector signal analysis (VSA) software is the second type of software applicable to pulse analysis. VSA software can be used with many RF/microwave front ends, including signal analyzers, oscilloscopes, and modular digitizers. VSA software performs time domain analysis, but is particularly useful when frequency domain analysis and demodulation (or modulation quality analysis) is needed. VSA software captures multiple pulses and extensive measurement of pulses one at a time.
Real-time spectrum analysis (RTSA) is also useful in pulsed signal environments. RTSA was originally implemented as a separate analyzer type, because the wide bandwidth of RF/microwave pulse analysis required dedicated RTSA hardware. Fortunately, recent improvements in processing power have made this a practical measurement application to add to general-purpose signal analyzers, at initial purchase or as an upgrade. RTSA involves gap-free processing of signal samples, or at least minimizing gaps so that analysis will not miss even very infrequent events. RTSA can be useful for finding elusive signals, and can also be important for triggering pulse analysis.
Combining these pulse measurement solutions can be especially powerful in meeting certain measurement challenges. For example, RTSA can be a uniquely effective tool for generating acquisition triggers for subsequent measurements made by VSA software or pulse measurement applications.
Pulse Analysis Measurement
The process of pulse analysis is often described in terms of three principal steps: triggering, signal acquisition, and measurement or analysis (Figure 2). Triggering can be understood as a general process of time alignment for acquisition of pulse data, since the signals under test are time-varying. The time alignment may involve an explicit trigger from an external source, or it may be generated in one of several ways by the acquisition hardware itself. For regularly repeating signals, the required time alignment may also be a simple matter of choosing a suitable measurement interval via a time gating function.
Acquisition can be as short as a single frame, or a lengthy recording that is intended for post-processing. The recording can be continuous or segmented, with some unnecessary data discarded to improve effective memory length. The bandwidth of the signal acquisition can be focused on the spectrum occupied by a single pulse or a wider signal environment or band, which includes many different ones, and may contain other signals as well.
Measurement can be single frame, or post-processing with analysis that can establish triggering or some form of time alignment or reference to the measurement. In the case of signal capture or recording using VSA software, the center frequency and span of measurement may be altered after time capture.
In understanding the pulse measurement process, the first step above may involve additional complexity: triggering may be derived from some later measurement/analysis processes such as an RTSA frequency mask trigger (FMT). This can make the complete measurement process somewhat recursive.
Functional Blocks of Pulse Measurement
The steps in the process previously described may be performed individually by separate devices, or multiple steps — including the entire pulse measurement — may be performed by a single analyzer. A general overview of the process is shown in Figure 3.
Acquisition hardware can take several forms, including both baseband and IF sampling, and can be performed by standalone instruments and modular systems. The most important performance characteristics of the hardware are frequency bandwidth and dynamic range, though memory depth, the number of channels, and other factors are also important.
Analysis algorithms turn the digitized signals into measurement data in the form of displays and result tables as needed. The algorithms may be part of general spectrum or VSA functions, or they may be embedded in dedicated pulse analysis applications. The applications are especially powerful when more comprehensive pulse analysis is needed, such as pulse parameter statistics or signal environment characterization.
Deep data storage is critical for some applications — generally where a large number of contiguous pulses must be analyzed from gap-free capture, or where access to the signal under test is limited and analysis must be performed later. Sampled data storage is combined with post-processing to generate the analysis results needed, and may also be used for signal playback.
Triggering operations can initiate or synchronize pulse acquisition, or can be used to time-align existing samples for pulse analysis. Since triggers can be taken directly from the input signal or can be the result of signal processing, such as real-time analysis or post-processing from data storage, they can be part of any of the main measurement blocks.
Challenges of Complex Pulse Analysis
Finding the signal of interest and aligning measurements with the desired timing are the first steps in pulse analysis, and can be some of the most challenging. This is particularly true of complex pulse environments that can include frequency- and amplitude-agile emitters, and multiple signal sources with widely varying amplitudes. Though it provides no frequency selectivity narrower than the analyzer span, the time and level parameters' IF magnitude trigger provide enough flexibility and specificity for many pulse measurements. By combining the selectable positive and negative (pre-trigger) delays with appropriate holdoff values and types, a single pulse can be selected from among many. If there is a repeating pattern, the largest signal can be used for triggering, and positive or negative time delays used to select any other single pulse or time interval. The holdoff function can also be used to avoid false triggers from pulse amplitude variations due to modulation.
FMT is valuable in finding and measuring transient or interfering signals, or in capturing specific signal behavior that can be best identified in the frequency domain. FMT processing is real-time or gap-free, providing confidence that any signal or behavior that matches the criteria will result in a trigger for measurement or time capture operations. Frequency masks can be generated manually or constructed from example spectrum measurements with offsets, or subsequent editing of amplitude/frequency breakpoints. Selectable trigger criteria make FMTs very powerful for identifying particular signal or environment behavior. Triggers can be generated when a signal enters or leaves the mask, and even for more complicated behavior such as leaving the mask after an entry event. These logical trigger criteria can be useful for capturing signals that are switching channels or are using frequency-hopping techniques.
The triggers and signal processing described so far will meet most pulse measurement needs; however, pulse durations and duty cycles are important variations in some signals, and a time-qualified trigger (TQT) can help isolate them for measurement. The TQT is a supplement to the FMT and IF magnitude triggers, continuously tracking the duration of events in the acquisition bandwidth. Thus, the TQT establishes a time qualification parameter in addition to the amplitude or spectrum parameters already available (Figure 4). Three properties are associated with TQT: Time 1, Time 2, and Time Criteria. These properties allow both open-ended and fully-bounded time durations to be set. In TQT operation, samples are acquired for analysis after an event has lasted for the specified time duration. If analysis of prior events is needed, a negative (pretrigger) delay can be used.
As baseband samplers, oscilloscopes typically lack the sophisticated combination of time and frequency domain triggers described so far; however, they do offer trigger capabilities that can be useful with RF pulses. One example is basic edge triggering, when combined with trigger holdoff. A trigger occurs when the input signal crosses a voltage threshold, as is the case with the beginning of an RF pulse as it grows in magnitude. By selecting a holdoff time longer than the longest expected pulse, the holdoff ensures that triggers will happen only at the beginning of RF pulses. This technique works most predictably for signals with consistent pulse duration.
Dynamic Range and Bandwidth Tradeoffs
Wide and ultra-wide bandwidths are increasingly important in pulse applications for several reasons. Ultra-wideband radars provide fine range resolution, plus increased resistance to detection and jamming. Frequency-hopping transmitters operate over wide ranges, requiring wideband capture to fully characterize the signal and avoid missing hops. Signal intelligence applications require acquisition of wide, contiguous bandwidth to identify targets.
Though the specifics of the tradeoffs improve over time, sampling with wider bandwidths inherently imposes performance limits. These limits arise primarily from the increased noise inherent in wider bandwidths and the decrease in ADC effective bits as sampling rates increase. These limits must be weighed against performance needs such as dynamic range, sensitivity, distortion, amplitude accuracy, and phase noise.
Capture length, or the amount of data to be acquired for a measurement, is a critical issue in many pulse analysis applications. Capture length — in terms of time — is especially important in analyzing dynamic environments, where there is a need to capture a time segment long enough to represent the dynamics in question. Every hardware platform has a limit to its memory size, and efficient memory use will provide the longest possible capture and the best signal measurement.
For a sampled data system, the maximum capture length for a given memory size is a linear function of the acquisition bandwidth. This favors the signal analyzer over the oscilloscope, since the signal analyzer samples only the IF bandwidth. The oscilloscope must perform baseband sampling of the entire signal spectrum — with later data reduction to convert to a band-limited IF — and the result is a much shorter gap-free signal capture. As noted, the transfer and processing of this baseband data can also result in slower measurement throughput. For baseband sampling of signals with wide bandwidth and low duty cycle, the memory issues can be a problem.
For many applications, the solution to this problem is in the segmented memory feature of some oscilloscopes. When this feature is enabled, the acquisition memory of the oscilloscope is broken up into many smaller equal-length segments. The segment length is chosen to be slightly longer than the widest pulse to be captured. Segmented memory also makes it much easier to replay or review the captured pulses in the time domain in the oscilloscope. The user can step through the segments manually or automatically to understand pulse sequences before processing the memory segments.
The combination of oscilloscope-based time domain analysis, signal analyzer-based real-time spectrum analysis, and VSA software that can make comprehensive measurements from both platforms meets many needs for pulse analysis. However, some applications require more macro-scale measurement capability that gathers information from hundreds or thousands of pulses and organizes analysis results in tabular or graphical form. Some typical applications include transmitter and component testing, characterizing pulse modulation stability, characterizing threats (SIGINT), verifying threat simulations, and verifying responses to EW jamming.
Understanding and quantifying the stability and repeatability of multiple characteristics of pulsed signals are critical tasks in making effective use of them. The collective analysis of large numbers of pulses can reveal behavior that is otherwise difficult to spot or to quantify. Time-aligned pulse modulation measurements are especially useful for diagnosing problems, and the software can use best-fit algorithms to provide tabular summaries of parameters such as FM slope and peak-to-peak deviation.
These time-aligned pulse modulation measurements are especially useful for diagnosing problems, and the software can use best-fit algorithms to provide tabular summaries of parameters such as FM slope and peak-to-peak deviation. With large numbers of measured pulses, the application can use statistics derived from them to produce histograms and trend lines. The statistics can be gathered from single or multiple acquisitions and, in comparison to measurements of individual pulses, provide much greater sensitivity to defects and a more comprehensive view of transmitter performance.
This article was contributed by Keysight Technologies, Santa Rosa, CA. For more information, Click Here .