- Buyers Guide
Measuring Modern Frequency Chirp Radars
Modern radars using frequency chirp modulation present many measurement challenges to the designer and system engineer. Complex test systems involving many instruments have traditionally been required to validate radar performance. Automatic measurements in a single instrument can be the key to simplifying radar test and improving test reproducibility. This article discusses a variety of short-frame and long-frame measurements that are available to cover most all modern chirp radar systems with bandwidths up to 20 GHz.
Modern wideband pulses, both short CW as well as longer FM chirped pulses, present difficulties for traditional pulse measurement techniques. Shortening a pulse increases the bandwidth and chirped pulses include a bandwidth as wide as the deviation of the FM modulation used, even if the timing is slower. The primary measurements needed for a chirp is the linearity of the frequency change—a measurement not previously needed.
There are RF converted instruments now available with digitizing bandwidths up to 110 MHz that include fully automated internal pulse measurement routines. Some of the measurement software routines used in these RF analyzers can also be used on wideband digitizing oscilloscopes, providing measurements of pulses and chirps up to 20 GHz bandwidth.
Such an instrument can perform multiple tests using data in a single acquisition from the device under test. Measurements can include RF power, spectrum analysis, spurious signal detection, pulse timing, frequency and phase variations, chirp linearity, digital modulation accuracy and more. These measurements enable root cause analysis for a deficiency.
Discovering Hidden Frequencies
Before performance or repair deficiencies can be documented, the problem must be identified and its nature defined. The need is for a trigger in the frequency domain. One recent development is the “Frequency Mask Trigger.” This technology allows the equipment operator to define a limit mask in both frequency and amplitude that will automatically be compared to a Fast Fourier Transform (FFT) spectrum of the incoming signal. Any signal intrusion into the mask area will trigger a capture of the error condition. A traditional level trigger will always trigger on the large signals that are within the IF bandwidth and cannot find the small one surrounded by large ones (see Figure 1). This requires an FFT process that can run fast enough that it will never fall behind the incoming signal. It processes one FFT faster than the time necessary to digitize a buffer full of time-domain data for the next FFT. A specified frequency event (even if smaller than many other nearby signals) can reliably trigger an event capture. Such a trigger mechanism will be specified for the shortest frequency event that can be captured with 100 percent probability.
Discover Low Level Spurious Signals within a Pulse Spectrum
Another improvement has been the addition of frequency transform processing power sufficient to display the time-varying nature of undesired signals that may be included with a desired chirp or pulse. Traditional spectrum measuring instruments performed one spectrum measurement for each display screen update, usually only 30 to 50 per second. Dedicated hardware processing now allows 48,000 or more spectrum measurements per second, which can be combined into a color-graded display.
In Figure 2, there are two representations of the same CW radar pulse with suspected intermittent interference. To see intermittent signals, the spectrum measurement must be made faster than has traditionally been done. The first plot in Figure 2 is a display with an update rate about 50 spectrum measurement updates per second. Since there are infrequent updates, the spectrum display will jump up and down. The most frequent plot is the one shown. It is extremely difficult to see a low level signal that is within the radar signal. Triggering on the pulse can stabilize the display. The second plot is of the same radar signal using 48,000 contiguous spectrum updates per second combined into the one, color-graded display. This way the low level signal becomes quite clear, even buried well below the radar pulse. No triggering is needed to see both signals at the same time. This interference sweeps through the pulse in a few milliseconds, repeating at two-second intervals. Discovering that a problem exists is critical to finding a solution.
Figure 3 is the spectrum display of a traditional signal analyzer. It has 30 to 50 spectrum updates per second, which is considerably faster than the older swept spectrum analyzer technology. Both the main display and a max hold trace are selected. Even with the addition of the max hold trace, the instrument can only build up an outline of the largest signal that is present. Both the outline of the chirp and even a large interfering carrier on the left are visible, but no other interference to the chirp pulse can be seen.
However, with fast spectrum updates, the radar chirp can be seen as “Live RF.” The complete character of the transmitted signal becomes visible. The smaller spurious chirp underneath the main signal is now visible, as well as several other spurious signals that were otherwise not visible at all. With 48,000 spectrum updates per second contributing to this display, fast time-varying signals are clearly visible.
Automated Pulse Measurements
It is important for those working on radars to think in both frequency and time domains. Depending on the parameters needed to test, there are three main types of automated tests available:
• Frequency domain tests: These include spectrum measurements of the transmitted signal, spectrum power measurements of the assigned channel or adjacent channels, and spurious signals that can be caused by a great variety of problems, including software problems either in control computers or in the DSP that generates the signals.
• Single pulse timing measurements (short-frame measurements): These reveal the quality of the individual pulses such as width, pulse repetition interval, rise and fall times, and modulation errors.
• Measurements of multiple pulses (long-frame measurements): Parametric trend analysis shows differences between individual pulses that can cause false radar readings.
Traditional manual pulse measurements require visually locating the features of the shape of the pulses. Modern automated pulse measurements must internally determine the shape of each pulse. The start and end times, the top and bottom of the pulse, and carrier frequency are all determined first. Then the remaining parameters are measured and can be included in a table of measurement results.
A Pulse Statistics display performs a combined analysis on a series of pulses, offering either a measurement trend view or an FFT of a series of measurement results. The pulse measurement trend graphs one measurement result for each pulse, versus its pulse number, automatically eliminating the “off” time between pulses. This enables easy inspection of the trend of a measurement over time. For example, a constantly decreasing phase could indicate a negative frequency offset present in the pulse, or could represent a constant-velocity Doppler shift in a received pulse.
Troubleshooting with Pulse Trend Data
Unintentional phase or amplitude modulation of radar pulses can be a problem. For example, insufficiently filtered aircraft power supplies converting 400 Hz AC to a high voltage DC power source can modulate the microwave power amplifier used to transmit radar pulses, resulting in pulse-amplitude variations that occur at the rate of the AC power source.
Quickly isolating a power supply modulation problem from a myriad of other possibilities can be difficult. With a typical swept tuned spectrum analyzer it would be necessary to try to discern a low level of narrow-band modulation on a broadband pulse in the frequency domain. A more advanced method that graphs the Pulse Statistics can easily discern a 400 Hz modulation on a pulse spectrum many MHz wide.
For example, a trend of Average On Power measurements can be scaled to show small variations in the pulse-to-pulse amplitude. Even small variations (e.g., 0.2 dB peak-to-peak) might go unnoticed unless they were plotted as a trend. Parameter variations that are periodic can be seen. But even then one cannot easily determine the frequency at which they occur. Thus, some problems are best viewed in the frequency domain.
Conversion of trend data into the frequency domain allows easy viewing of the nature of the modulation and reveals key information about its source (see Figure 4). The spectral view can show if the modulation is at a single frequency or contains several frequencies. These advanced measurements enable this capability by performing a Fast Fourier Transform on the trend measurement results. In this case, the FFT provides a spectral view of the amplitude trend data showing 400 Hz modulation.
Compressed Pulses (Chirps)
Many advanced radars use one form or another of a chirp. The most common is a linear “FM” chirp. A chirp allows “Pulse Compression” in the receiver processing, which provides detailed range resolution, while using a long duration transmitted pulse (providing good sensitivity with less transmitter power).
There are now automated pulse measurements available for chirped pulses. Overall chirp frequency width is the basic metric. For advanced measurements, linearity of frequency and particularly phase during the chirp are the critical parameters that relate to the radar’s ability to see individual targets without false responses causing ambiguities.
To see and measure the frequency errors more closely, the Frequency Error measurement subtracts off a perfect linear chirp from the frequency deviation display. Now only the frequency error difference from the perfect ramp is plotted. Also, the numeric answer can be selected to calculate either the peak error or the RMS error over the pulse measurement time. Figure 5 shows a chirp frequency error plot, which displays a very good linearity with no slope, but some transient ringing on both ends of the pulse.
Chirp Phase Measurements
Even more important than frequency linearity is phase linearity. For compressed pulses, the prime determinant of target movement is phase of the return pulse throughout its width, and multiple targets within the pulse duration produce multiple variations of the phase. The phase deviation plot measures the phase across the entire pulse, and then unwraps the phase so that the parabolic shape of the phase change is easily seen. This shows the full extent of the phase change in the frequency chirp. The most important characteristic is how accurately the phase of the chirp follows the phase of an ideal chirp. In the lower graph of Figure 6, the phase error is plotted. One can easily see the nonlinearity (quadratic error) of this chirp and slight frequency error. Note that after the ideal chirp phase is subtracted, the 19 degrees of peak error out of the total of 4959 degrees of chirp phase shown in the upper graph is visible.
Frequency and particularly phase are very sensitive to noise in the measurement. The signal-to-noise ratio of the measurement is proportional to the bandwidth in which the measurement is made. Therefore, if the signal is captured in a wide bandwidth to get all of the pulse, then the user may select a narrower filter for the signal processing to use for the measurement so as to reduce noise in the plot.
Measurements of differences from one pulse to the next can discover modulations of the transmitted pulse that will cause errors in the radar receiver. This comparison of phase or amplitude differences between pulses can be programmed to be measured at any selected time within the pulse duration.
Some radar may use other forms of pulse compression. For the receiver to process a chirp it usually requires many powerful DSPs. A simple receiver can be made using a bank of SAW filters, which could be each at a discrete frequency. A hopper is an ideal transmitter to work with this simple receiver. Each transmitted pulse hops from the first through the last frequency, to match the receiver’s filter bank. A further advancement would be to use fully randomized sequences that can now make analysis of the radar more difficult, and prevent spoofing of the radar during operation.
Measurements of a radar pulse containing multiple frequency hops within each pulse require wideband frequency-versus-time capabilities. The spectrogram shown in Figure 7 reveals the time-varying nature of the frequency spectrum and maintains the simultaneous display of amplitude variations (as color). There are seven microsecond duration pulses with a linear frequency progression of one microsecond hops shown. The total frequency width is 90 MHz. Overlapped FFT processing allows the spectrogram to clearly show the time nature of the frequency variations. In the right photo, there is a pulse with randomly hopped frequencies. The spectrogram easily shows the sequence that the hops follow.
Figure 8 shows a signal that is considerably smaller in duration. Each segment is only 150 nanoseconds, for a total pulse time of about one microsecond. The time sequence is still clearly shown by using overlapping FFT records. For this display each FFT overlaps the previous one by 99 percent.
For accurate measurements of both frequency and timing, markers can be used on a frequency-versus-time measurement plot. Very detailed frequency trajectories can be seen, including transients due to transmitter overshoot as well as any filters in the signal path. By using time samples that each have in-phase and quadrature values, frequency changes can be measured with time resolution as tight as 7 ns per sample at 110 MHz bandwidth.
Very Wideband Pulses
Some of the same software that performs pulse measurements in a dedicated RF instrument can also be used to analyze the output from an oscilloscope. This can extend some of the measurements to include pulses up to 20 GHz bandwidth. While this method lacks the real-time nature of the fast update hardware processor, it does have extraordinary bandwidth. It can also export the captured data (as can the RF analyzer) for further analysis by external programs such as Matlab.®
Figure 9 shows a table of measurements made on a 2 GHz wide chirp. The frequency width measurement shows each pulse at 2.048 GHz width. The measurement software can also optimize a spectrum measurement to exactly the width of one of the acquired pulses, as seen in the bottom of the plot in Figure 10.
Selecting the Correct Test Instrument
The selection of test equipment is driven by the extremes of the parameters of the signals that are being measured. Table 1 contains generalized requirements listed with the type of instrument needed as an overall guide. In conclusion, test instruments are keeping up with the measurement needs of radar designers including specialty tests that are available for advanced chirp radar./
Tom Hill earned his BSEE degree from Clemson University. He is currently principal engineer in the RF test group at Tektronix. He has over 37 years experience with RF test and measurement, 33 of those years at Tektronix. He has held various engineering and technical marketing positions with Tektronix since 1974, including general purpose and RF specific measurement equipment. He develops measurement techniques and is responsible for translating end-user’s needs into measurements and test equipment that solve those needs. He has specific expertise in radar, surveillance, EW and communication systems, and holds several patents for test and measurement equipment.