Wideband RF measurements are changing and along with them, the tools you need to make sense of the signals. Today’s radar systems require higher target tracking resolution, communications systems require higher data throughput—and to meet the demands, you require wider modulation schemes on related signals to validate prototypes and production units.
Gone are the days when an instantaneous measurement bandwidth of 510 MHz, the longtime standard in signal and spectrum analyzers, could handle this modulation bandwidth. Some systems have crossed beyond 1- GHz and even 2-GHz-wide formats. You need a different approach to make high-quality, insight-providing wideband RF measurements.
How different an approach? One that uses high-bandwidth, real-time oscilloscopes. Digitizers and oscilloscopes offer enough bandwidth and sample rate to directly sample the carrier plus modulation either alone or with the use of down-converters in front of the scopes.
The trick is knowing what to use when. One way to consider your options for wideband measurements is to plot the possibilities on a chart. The vertical axis represents analysis bandwidth of the solution and the horizontal axis representing carrier frequency that you can measure. Don’t worry about doing the plot—we’ve taken care of it:
Applicable tools as a function of signal carrier frequency and spectral width
As you see, classic signal analyzers offer analysis bandwidths up to 1 GHz and handle carrier frequencies up to around 50 GHz. As an alternative, mid-range oscilloscopes offer bandwidths in the 8-GHz range, letting you measure signals with carrier frequencies approaching 8 GHz, and with very wide modulation bandwidths, approaching 8 GHz. As long as the carrier plus modulation spectrum fits within the oscilloscope bandwidth, you can make meaningful measurements.
But even that may not be enough. In wideband aerospace/defense applications, including electronic warfare, radar, and surveillance, signals of interest may have carrier frequencies higher than 8 GHz. Cue the high-performance oscilloscopes. These scope families have higher bandwidths up to 33 GHz and 63 GHz and, as you might guess, corresponding higher prices. But they offer impressive performance in areas like frequency response flatness and low noise. An alternative is to place a down converter in front of a mid-range oscilloscope. You pay less but can handle high carrier-frequency signals with wideband modulation—provided you’re willing to make some tradeoffs in amplitude and phase linearity.
As a first down-converter option, you can place a standard signal analyzer in front of a mid-range oscilloscope and use the IF down-conversion path in the signal analyzer. You’ll typically need calibration to flatten the overall system amplitude and phase response over frequency. But a solution like this can address a wide range of carrier frequencies, typically up to 50 GHz.
A second down-converter option is to place a lower cost harmonic mixer in front of a mid-range scope. This results in a “banded” solution: Very high carrier frequencies can be analyzed, but there is generally a “band” of carrier frequencies that a particular harmonic mixer can handle. That makes this option especially convenient for applications like 5G, Wigig, and automotive radar.
Typical RF performance for high-bandwidth real-time oscilloscopes
So what do you need to know before making FFT or wideband RF measurements with an oscilloscope or scope combined with vector signal analyzer (VSA) software? You need to know that the RF characteristics can have a major influence on the measurement results—so you’ll need to evaluate this first.
Today you can find oscilloscopes that incorporate amplitude and phase correction for excellent absolute amplitude accuracy and low deviation from linear phase across their frequency range. This in turn contributes to high-quality RF measurements. These oscilloscopes also offer excellent noise densities, in the vicinity of -160 dBm per hertz, and high dynamic range and signal-to-noise ratios, considering the wide bandwidth capability they offer.
What does that do for you? You can look at wideband signals with very small amplitude adjacent in time to large signals. You can also boost scope sensitivity to measure isolated, small-amplitude signals. The time-base circuitry in these oscilloscopes also means good, close-in phase noise, which corresponds to low jitter in very deep memory traces. If you want more details, see the RF characteristics of a high-performance 33-GHz oscilloscope in Table 1.
Table 1. Typical RF performance in a high-bandwidth oscilloscope
Wideband pulsed RF time-domain measurements of envelope, frequency, and phase chirp
Now that we know what our high-bandwidth scope is capable of, let’s see how it handles time-domain measurement and analysis of wideband pulsed-RF signals with no help. The choice of which oscilloscope to use depends on the maximum frequency content of the carrier plus modulation. Consider an example where a signal under test is supposed to have 1-usec-wide pulses, with a pulse repetition interval of 100 usec. It also has an RF carrier frequency of 15 GHz and linear FM chirping that is 2-GHz wide.
Figure 2 shows a variety of measurements on a single RF pulse, including envelope parameters and the frequency chirp across the pulse. Stable triggering on this pulse is accomplished with trigger “holdoff” set to a value slightly longer than the 1-usec RF pulse width.
Figure 2. Time-domain measurements on 1-usec wide, 15-GHz carrier, 2-GHz-wide linear FM chirped RF pulse with a 33-GHz bandwidth oscilloscope
To make amplitude measurements, we use the “Envelope” math function and then pulse measurements are dropped down onto the visible RF pulse envelope. A “Frequency” measurement is dropped down onto the RF pulse (not onto the envelope), and a “Measurement Trend” math function is defined with the frequency measurement as a source. Next we perform a smoothing math function on the measurement trend with the resultant linear ramp display of the linear FM chirp modulation, also shown in Figure 2. The oscilloscope magnitude linearity over the frequency span of interest has a direct effect upon the quality of the envelope measurement. To see the effect, take a look at the magnitude plot over frequency of the 33-GHz bandwidth scope in Figure 3.
Figure 3. Typical magnitude linearity over frequency on four individual 33-GHz channels
Wideband pulsed RF-gated FFT measurement of spectrum
You can create a wideband FFT by defining an “FFT Magnitude” math function with “Rectangular” windowing. Then create a time-gated FFT using the (you guessed it) “Timing Gate” math function. Once the time-gating math function is defined, you can define an FFT math function that is calculated from the time record within the time gate, as shown in Figure 4.
Figure 4. View of normal and time-gated FFT and display with time gate at the beginning of the RF pulse
Wideband pulsed-RF time- and frequency-domain measurements with a scope plus VSA software
But that’s not all. You can further enhance RF and FFT measurements made with high-bandwidth oscilloscopes by importing scope-captured signals into VSA software. Some advantages of using VSA software include:
- many built-in RF measurements;
- ability to bandpass-filter oscilloscope input samples and decimate prior to the FFT calculation to reduce noise and speed the calculation;
- variety of digital and analog demodulation options like QAM16 and FM demodulation;
- time-domain baseband view of pulse with reduced noise through processing gain;
- frequency and phase shift across the pulse through a demodulator.
If the oscilloscope-captured data is imported to VSA software, it can be digitally down-converted into I and Q baseband data, bandpass-filtered, and then resampled. This can greatly decrease the amount of noise in the measurement. Essentially the process is “tuning” to the center frequency of the signal and “zooming” into the signal to analyze the modulation. This is also referred to as “processing gain.”
In this example, the original 8-GHz-wide measurement with the associated noise is reduced to a 500-MHz-wide measurement, centered on the 3.7-GHz carrier with an instantaneous measurement bandwidth slightly wider than the width of the signal modulation. This corresponds to an improvement in signal-to-noise (SNR) ratio of:
10log*(ScopeBW/Span) = 10log*(8E+09/500E+6) = 12 dB.
SNR is improved by 10log*(ScopeBW/Span).
By taking advantage of this processing gain, combined with the VSA software’s capability to use a log-magnitude scale, and using averaging, you can now see the 50-dB down pulse, as shown in Figure 5. It wasn’t visible in the scope display with the 8-GHz wide measurement.
Figure 5. 50 dB down pulse seen with VSA software “Center Frequency” and “Span” set
The secret to long target-time capture and statistical pulse analysis
When an oscilloscope samples a wideband RF signal, it must do so at a fast enough rate to accurately capture the carrier plus modulation. Often a very fast sample rate is required. In a normal real-time sampling mode, the oscilloscope memory will not allow for a long capture period.
But there is a work-around: oscilloscope segmented memory. This can greatly increase the target activity time when there is a low-duty-cycle signal, such as a common pulsed RF radar signal. The scope memory is divided into smaller segments of fixed time width, chosen to be a little wider than the widest RF pulse. The scope triggers on an event, such as the beginning of the RF pulse, and then places one RF pulse in a memory segment. The scope then stops capturing data, rearms the trigger, and waits for the next RF pulse. A second RF pulse is put into the second segment of memory. This process continues until all the scope memory segments are used.
Modern pulse-analysis software can let you take advantage of the scope segmented memory and then offers built-in measurements for pulsed RF signals. Figure 6 shows a capture of many RF pulses via segmented memory, combined with pulse-parameter measurements in the pulse-analysis software. Here a 1-GHz linear FM chirp and related phase shift across pulses is compared to a best-fit ideal linear ramp and ideal parabola, respectively. A close-up view is made of the delta between measured and reference for frequency in trace S and for phase in trace J.
Figure 6. Pulse analysis software calculations based on measurements taken on oscilloscope segmented memory
The bandwidth limitations of signal and spectrum analyzers are driving designers to use digitizers and oscilloscopes, with or without down-converters. Math functions like envelope, measurement trend, and FFT all prove helpful in understanding target system operation and issues. Combining an oscilloscope with VSA software creates a powerful RF-measurement suite to perform measurements, including demodulation, extended SNR time-domain views, and statistical RF pulse analysis. Yes, there’s a tradeoff between dynamic range/SNR and the instantaneous bandwidth available, but you can still access many useful wideband measurements to evaluate a prototype or production unit.