Radar Measurement Comparison:
Swept and FFT-Based Signal Analyzers
By John Barfuss, Agilent Technologies

Modern pulsed radar systems use advanced waveform and modulation characteristics to obtain greater range resolution, enhanced clutter suppression and superior target recognition. Specifically, performance improvements are achieved using radar signals with wide bandwidths, low duty-cycles, high linearity and sophisticated modulation. The nature of these waveforms makes them difficult to measure and synthesize. Understanding how different types of instruments respond to these radar signals is crucial when designing high-performance and cost-effective radar systems. This article will review several measurement approaches for characterizing the spectral content of low duty-cycle, wideband radar signals.

The basic tool for characterizing radar signals is the spectrum analyzer or signal analyzer, which measures the power content of radar signals as a function of frequency. This capability is important because an incorrect spectral profile indicates a number of problems that result in wasted power and emission of undesired signals. Traditional spectrum analyzers use a swept-tuned architecture to achieve high dynamic range and wide frequency measurement ranges [1]. Advances in digital signal processing have resulted in two significant technological developments: 1) the inclusion of digital IFs in traditional swept-tuned spectrum analyzers, and 2) the emergence of Fast Fourier Transform (FFT)-based analyzers as an alternative to the traditional spectrum analyzer architecture. The inclusion of digital IFs in traditional spectrum analyzers has greatly enhanced the accuracy, repeatability and speed of these instruments. FFT-based signal analyzers provide unprecedented modulation analysis capability and can result in much faster measurements in some cases, but much slower measurements in other cases.

Selecting the optimal measurement approach requires some knowledge of the instrument capabilities and signal under test. To understand how the instrument’s architecture affects the displayed frequency response, measurement speed and dynamic range, the spectral responses of a wideband low duty-cycle radar waveform will be compared using swept-tuned and FFT-based analysis techniques. Fortunately, some signal analyzers contain both swept and FFT capabilities, allowing a direct comparison of these two techniques within the same instrument. In addition, some signal analyzers can also be used as a vector signal analyzer for measurement of phase profiles, modulation, transient analysis and spectrograms.

Chirped Radar Signals
A pulsed radar transmits a periodic sequence of narrow pulses and receives target echoes between the pulses during the transmitter-off time. Pulse compression techniques using linear frequency modulated (LFM) or “chirped” waveforms can improve range resolution with a relatively higher average transmitted power as compared to narrow pulsed waveforms operating with a similar operating bandwidth [2]. Whether accomplished by implementing pulsed or chirped waveforms, increasing the range resolution and unambiguous range requires wide bandwidth and low duty-cycle waveforms, respectively. The combination of wide bandwidths and low duty-cycle creates unique measurement challenges, as there is a lower probability of intercepting these signal types during a typical measurement.

Wideband Chirped Radar Measurements Using Swept-tuned Analyzers
As an example, Figure 1 shows the measured frequency responses of a wideband chirped signal for a 1 GHz frequency deviation with a 2-microsecond pulse width and a 100-microsecond period. The two measurements shown in this figure were captured using a swept-tuned analyzer configured with different sweep times in order to compare the effects of instrument configuration on low duty-cycle waveforms. The measurement on the left was captured using an analyzer configured with a sweep time automatically set to 2.4-milliseconds. In this case, the analyzer’s sweep time is too fast to capture all of the spectral energy in the waveform. Spectral energy is captured every 100-microsecond during the sweep, but nothing is captured as the sweep advances between pulses. As swept-tuned analyzers continuously measure the signal during a sweep, these periodic pulses appear as individual frequency energy components with equal spacing. These are often referred to as pulse repetition frequency (PRF) lines. Note that these signals have no specific frequency domain meaning and will move around with each sweep [2]. To measure the complete spectral content of the chirped waveform, slow the analyzer’s sweep time so that a pulse occurs in each measurement bucket or point of the sweep. In this example, peak detection was turned on and the sweep time was increased to 100-milliseconds. The desired chirped spectrum is the result as shown on the right in Figure 1.

Wideband Chirped Radar Measurements Using FFT-based Spectrum Analyzers
Measuring this same chirp with an FFT-based spectrum analyzer is less optimal. The reasons for this have to do with how the FFT-based spectrum analyzer “sweeps” or measures across spans greater than the FFT analysis bandwidth of the instrument. Essentially, the analyzer must measure the spectrum a section at a time and then concatenate or “stitch” the results. This approach works reasonably well for continuous signals but is less effective for pulsed signals due to measurement efficiency. The time required by the analyzer to retune between each segment of the desired spectrum is long relative to the short time data is sampled for each FFT computation. The result is a low probability of intercepting the signal, especially for signals with low duty cycles.

In the swept analyzer example, the sweep time was slowed to increase the number of times the pulsed energy was intercepted during the sweep. This resulted in a better view of the signal, as shown in Figure 1. However, this approach will not work with an FFT-based spectrum analyzer as it may not even have a sweep time control. If it does, its function is not the same even though it may simulate a conventional sweep control.

An alternate approach to improving a pulsed RF measurement that does work with an FFT-based spectrum analyzer is to reduce the RBW setting. As the RBW is reduced, the measurement slows, increasing the probability of intercept. If the RBW is reduced enough, missing the signal is no longer an issue since the analyzer sees the spectral components of the signal as continuous waveforms, the sum of which forms the pulsed signal.

With a narrow RBW setting, the FFT-based spectrum analyzer actually becomes more efficient at measuring spectrum than a swept analyzer with an equivalent narrow RBW setting, despite the fact that the analyzer still must “stitch” together the spectrum from several computed FFT segments. For example, the time required to measure 1.4GHz of spectrum with 1 kHz RBW is 91s on the PSA in FFT mode vs. 1688s with the PSA sweep mode.

Nonetheless, reducing the RBW to measure a pulsed RF signal has its costs in both measurement speed and dynamic range.

Figure 2 shows a measurement of the same chirped radar as measured in Figure 1, this time using the FFT mode of the Agilent PSA spectrum analyzer and different RBW settings. With the default RBW of 3 MHz, the measurement is sporadic. By decreasing the RBW to 1 kHz, we are able to measure the spectrum. However, the measurement now takes 91s vs. the 100 ms for the swept mode shown in Figure 1. In addition, reducing the RBW also increases the amount of pulse desensitization, resulting in less dynamic range. Comparing the figures, the swept mode achieves about 15 dB more dynamic range for this example. An explanation of pulse desensitization can be found in the Agilent Radar Measurements Application Note. [2]

Chirped Radar Measurements Using a Vector Signal Analyzer (VSA)
As shown in the previous examples, an FFT-based analyzer has limits when the span and/or signal of interest extends beyond the analysis bandwidth (FFT bandwidth) of the analyzer. However, for signals within the instrument’s analysis bandwidth, an FFT-based signal analyzer can provide rich analysis when implemented as a vector signal analyzer.

For instance, modulation analysis is possible because a vector signal analyzer measures both the magnitude and phase of the signal over time and frequency. Figure 3 shows the amplitude, phase and frequency responses for a radar chirp as a function of time measured simultaneously using the Agilent 89601A VSA software connected to the Agilent PSA spectrum analyzer.

A vector signal analyzer’s performance is determined largely by the capacity of its digitizer [2]. For instance, the Agilent PSA uses a 200 MSa/sec digitizer with 14 bits of resolution for 80 MHz of analysis bandwidth and 78 dBc distortion-free dynamic range.

Wider bandwidths can be achieved using an ultra-wideband (UWB) vector signal analyzer such as the Agilent VSA90000A oscilloscope- based VSA. The VSA90000A analyzer samples at 40 GSa/s with 8 bits of resolution and is therefore capable of analysis bandwidths up to 13 GHz (though with less dynamic range than the PSA). This can be very useful for analyzing wideband chirps such as the 1-GHz chirp shown in the previous examples. In this case, the oscilloscope uses the same VSA software as used by the PSA. Since the oscilloscope is sampling at 40 GSa/s, it can measure this X-Band radar signal directly without requiring down conversion. (See Figure 4.)

Conclusion
Signal analyzers use different approaches to measuring a signal’s spectrum. Each approach has its advantages and limitations. These differences are most apparent and critical when measuring wideband pulsed signals such as radar. The pros and cons are summarized in Table 1. Instruments that perform as swept, FFT and vector signal analyzers, such as the Agilent PSA and MXA signal analyzers, provide optimal performance and capability regardless of the characteristics of the signal being measured.

Additional information concerning measurements for radar can be found in Agilent’s Radar Measurements Application note mentioned previously. This application note can be ordered from www.agilent.com/find/radarprogram or downloaded from www.agilent.com/find/ad.

References
[1] Agilent Application Note 150, “Spectrum Analyzer Basics”, Literature number 5952-0292, August 2006.
[2] Agilent Application Note, “Radar Measurements”, Literature number 5989-7575EN, January 2008.
[3] Agilent PSA Series High Performance Spectrum Analyzers, Literature number 5980-1283E, May 2007.
[4] Agilent Application Note 150-2, “Spectrum Analysis – Pulsed RF”, Literature number 5952-1039, May 2005.

AGILENT TECHNOLOGIES
www.home.agilent.com
TXTLINX.COM99
Email this article to a friend!