Spectrum analyzers are among the most essential of RF/microwave instruments, showing in the frequency domain what a high-speed oscilloscope might show in the time domain. Spectral measurements can reveal a great deal about the performance of a system and its components, provided that those measurements are performed realistically and with a proper understanding of a spectrum analyzer’s capabilities and limitations. As an aid to spectrum-analyzer owners—and for those considering the addition of a spectrum analyzer to their instrument arsenal—Agilent Technologies recently made a free 12-page application note (No. 1286-1; “8 Hints for Better Spectrum Analysis”) available on its website. Those hints, combined with solid fundamental knowledge of a spectrum analyzer’s key performance parameters, can help make the most of spectrum analyzer measurements whether in the field or in the laboratory.

Modern spectrum analyzers are available with a wide range of analog and digital signal-processing functions across frequency ranges extending from the high-frequency (HF) range through millimeter-wave frequencies. The N9010A EXA signal analyzers from Agilent Technologies, for example, are available with a number of different frequency-range options, with the broadest-frequency model spanning 10 Hz to 44 GHz (see figure). Spectrum analyzers provide their operator with a number of different adjustments that can impact the accuracy and reliability of a measurement, including resolution-bandwidth (RBW) filters, sweep times, input attenuation, and even preamplifier and intermediate-frequency (IF) amplifier gain.

The N9010A EXA analyzers have RBW filters with 3-dB bandwidths as narrow as 1 Hz to as wide as 3 MHz. The selection of the filter depends on the type of signal under study and the frequency span occupied by the signal. Sweep times for frequency spans greater than 10 Hz can be set from 1 ms to 4000 s, while sweep times for zero (0-Hz) spans are adjustable from 1 μs to 6000 s. The EXA signal analyzers offer an input attenuation range of 0 to 60 dB in 10-dB steps in standard instruments, and the same 0-to-60-dB attenuation range in 2-dB steps as an option. At the lower end of their dynamic range, the N9010A EXA signal analyzers achieve a typical displayed average noise level (DANL) of −140 dBm or better with high IF amplifier gain and 0-dB input attenuation, and the DANL can be dropped to −160 dBm by using a preamplifier with high gain.

By adjusting a spectrum or signal analyzer’s settings, such as the RBW filters, the measurement accuracy can be optimized. A narrow RBW filter can make it possible to display low-level signals at the expense of sweep speed. But if the RBW filters are too narrow, sideband information for signals with wideband modulation can be lost. Wider RBW settings yield faster sweep speeds for a given spectrum analyzer, but narrower RBW filters lower the DANL, improving the signal sensitivity of the analyzer and expanding its dynamic range. If noise may be possibly obscuring the measurement of a signal, switching to a narrower RBW filter will lower the noise floor and better extract the signal from the noise.

Changing a RBW filter can impact the measurement accuracy of a spectrum analyzer, depending upon the quality of an instrument. In better-quality analyzers, amplitude uncertainty will be consistent across the different filter bandwidths, but variations can exist for different filter bandwidths in lower-cost analyzers. Additional passive components in a test setup, such as coaxial cables and external filters, can also influence the measurement accuracy of a spectrum analyzer. Any amplitude errors that are introduced in a test setup (due to coaxial cables and other passive components) can be cancelled with the aid of an analyzer’s built-in amplitude correction function, along with an external test signal source and a power meter. Most newer spectrum analyzers allow a user to store the amplitude response of different passive components and measurement setups so that calibration is not necessary every time that test setup is used.

When measuring low-level signals with a spectrum analyzer, any noise generated within the analyzer will limit its capability to detect and measure small signals. The instrument’s low-level sensitivity and DANL can be improved by minimizing the input attenuation, using the narrowest RBW filter practical, and using a low-noise preamplifier with suitable frequency range. By using a preamplifier with low noise figure and adequate gain, the measurement sensitivity of the analyzer can be improved. If the gain of the preamplifier is sufficiently high, the noise floor of the analyzer/amplifier combination will be determined by the noise figure of the preamplifier.

Spectrum analyzers are often called upon to measure distortion, and distinguishing distortion products from fundamental-frequency signals can sometimes pose measurement challenges. The analyzer’s dynamic range indicates the maximum separation that the instrument can distinguish between signal and distortion or signal and noise. Any improvement in dynamic range is a matter of switching to a narrower RBW filter, albeit with some sacrifice in sweep speed. Measuring a low phase noise for a test signal, for example, may require setting a narrow RBW filter width, such as the use of a 1-kHz RBW filter. The compromise in using such a filter will be the limit in sweep speed.

For measurements of transient signals, the spectrum analyzer’s sweep speed is critical and the instrument parameters that impact sweep speed must be adjusted accordingly. The setting of the RBW filter will largely dictate the spectrum analyzer’s sweep time. Narrower RBW filters typically require longer sweep times, which translate into a tradeoff between sweep speed and sensitivity. For example, a ten times change in RBW on a modern spectrum analyzer can translate to an approximate 10-dB improvement in sensitivity. By switching to narrower RBW filters, the measurement sensitivity for continuous-wave (CW), as well as for transient signals, can be dramatically improved.

Newer spectrum analyzers may offer Fast Fourier Transform (FFT) functions that provide a good balance between sweep speed and sensitivity. When using FFT analysis, the sweep time is dictated by the frequency span rather than the setting of the RBW filter. Using a spectrum analyzer’s FFT mode, where available, can significantly speed measurements of relatively narrow spans (such as 20 MHz).

Spectrum analyzers function with a number of different display detection modes and, depending upon the type of signals to be analyzed, the choice of mode can have a bearing on the accuracy of the measurement. Following an analyzer’s input RF/microwave circuitry, with downconversion mixers, input signals are digitized by means of a high-speed analog-to-digital converter (ADC) at either the IF stage following the mixer or following the spectrum analyzer’s video filtering. The analyzer’s display is driven by this digitized data, and the choice of which portion is emphasized on the instrument’s display depends on the display detector following the ADC. Display detectors include positive peak, negative peak, and sample detectors.

As the Agilent application note states, these data are considered to fall within a “bucket,” and the detectors work with the data within each bucket. A sample detector uses data within the center of the bucket to feed the analyzer display. As a result, sample detection mode is effective for showing noise, but not as accurate when displaying CW signals with narrow RBWs, and can miss portions of such signals. Positive peak detection, which is effective for CW signals, displays the highest-power portion of the data bucket. Negative peak detection, which is good for signals with amplitude modulation (AM) or frequency modulation (FM), shows data from the lowest-power part of the bucket.

Some spectrum analyzers include a detection mode known simply as “normal” detection, which is a sampling mode in which inputs are dynamically classified as either signals or noise. Such a detection approach will not miss signals or portions of signals in the manner of conventional sample detection, and can provide a better visual display of random noise when using peak detection mode on the spectrum analyzer.

Spectrum analyzers that offer average detection can show signals based on their average power or voltage levels. This type of detection can be useful for measuring the power levels of complex signals, in which their power is time varying, or for electromagnetic-interference (EMI) testing, where voltage levels can vary over time. In general, average detection is effective for analysis of a wide range of modulated signals, such as pulse-modulated signals in radar systems or signals with AM.

The Agilent application note provides clear details on eight different measurement scenarios with a spectrum analyzer and can be instructive even for regular users of these instruments. RF/microwave signals used in modern communications, radar, and electronic-warfare (EW) systems are increasingly complex, using time gating to vary the information carried by a transmitted signal from moment to moment. Understanding some of the measurement capabilities of a modern RF/microwave spectrum analyzer can at least reduce the time required to capture these signals, and can contribute a great deal toward improving the accuracy of those measurements.