Spectrum analyzers have changed drastically in recent years, due largely to the use of digital components. Once predominately based on a superheterodyne receiver architecture to downconvert input signals to intermediate-frequency (IF) signals that were then filtered and processed, newer spectrum analyzers are just as likely to be called "signal analyzers" and employ sampling techniques with a high-speed analog-to-digital converter (ADC).

As with computers, spectrum analyzers are available in large and small versions, for benchtop or portable applications, respectively. Although portable instruments were once considered lightweight in terms of performance, portable spectrum and signal analyzers from a number of manufacturers now offer impressive performance and accuracy. (For more information on portable test equipment.)

Although modern spectrum analyzers may incorporate a good deal of digital processing, including Fast Fourier Transform (FFT) analysis capabilities, the signals they display are in the frequency domain, magnitude as a function of frequency. As a result, many of the key performance specifications for a spectrum analyzer have to do with amplitude or frequency characteristics. The user selects the range of frequencies for the analyzer to sweep across, and different sweep speeds and filters can be applied to capture and analyze a signal of interest. In addition to showing continuous-wave (CW) and modulated signal waveforms, a spectrum analyzer with adequate bandwidth can also display information about pulsed signals. Understanding the key performance parameters and functional controls of a spectrum analyzer can help when choosing a spectrum analyzer for different requirements.

The frequency range may seem like an obvious starting point in choosing an analyzer, since it will ultimately limit the number of applications that can be handled by the instrument, but the instantaneous analysis bandwidth is also an important consideration when comparing instruments, especially for communications applications. The frequency range sets the limits on low and high frequencies, but the analysis bandwidth determines the amount of signal modulation that can be analyzed. In LTE systems, for example, the total range of available frequencies is 700 to 3000 MHz, but the channel bandwidth allocations can be as wide as 20 MHz, requiring a spectrum analyzer with an instantaneous or real-time bandwidth of 20 MHz or more. (For more on LTE testing, see p. 110.) In addition, harmonics may be of interest, so that when specifying a frequency range for a given application and band of interest, the total range may have to be doubled or tripled in order to evaluate second and third harmonics. In amplifier studies, for example, these additional harmonic signals can force an amplifier to work overtime, and ultimately will limit the power-added efficiency possible for a given power amplifier (PA) design.

For any spectrum analyzer to effectively display a signal, the proper frequency span must be selected and the unwanted energy around the signal of interest must be rejected by means of the analyzer's filters. The choice of span must be wide enough to accommodate a signal's full bandwidth, and the filters must be set appropriately. A spectrum analyzer provides two sets of adjustable filters: resolutionbandwidth (RBW) filters and its videobandwidth (VBW) filters. The RBW filters are like the IF filters for a receiver, determining how much information about a signal will be displayed on the analyzer's screen, and with what frequency resolution. The VBW filters are used to remove noise from a displayed signal. The bandwidths of these filters are usually specified in term of 3-dB points from the center frequency, with typical bandwidths as narrow as 1 Hz and as wide as 20 MHz or more. Narrower filters can provide greater detail and resolution, but will result in slower sweep times.

A spectrum analyzer's filters are related to one of the instrument's key dynamic-range characteristics, its displayed average noise level (DANL), essentially the noise floor for a given sweep speed and filter settings. The narrowest RBW filter in an analyzer will give the lowest displayed average noise level (DANL) as well as the longest sweep time. Selecting wider RBW filters result in faster sweep speeds but with higher DANL. The type of signal to be studied will usually determine the trade off to be made. For a modulated signal, for example, the RBW filter must be set wide enough to include the signal's center frequency and sidebands. Similarly, the widest possible RBW setting is usually required to pass the harmonic information contained within a pulsed signal.

A spectrum analyzer's amplitude measurement range is set by the DANL at the low-level end and the largest signal that the analyzer can accept for analysis without distortion, usually determined by the instrument's maximum safe input power level. Higher power levels may be possible at the input to the analyzer through the use of built-in attenuators, although attenuation will also degrade the frequency response of the instrument. Analyzers typically display amplitude in either logarithmic or linear scales, depending upon the need of the measurement, and in a wide range of units, including V, W, dBm, and dBmV. The amplitude or level accuracy is usually specified in terms of a decibel variation (such as 0.3 dB) for a given frequency range and under a given set of conditions, such as with or without internal attenuation and with or without an internal preamplifier. When comparing different analyzers, it is important to match these conditions for each instrument.

Some analyzers may include or offer an option for EMI measurement bandwidths, compliant to either Comite International Special des Perturbations Radioelectriques (CISPR) or MIL-STD- 461E military requirements. For EMI measurements, the CISPR bandwidths are 200 Hz, 9 kHz, 120 kHz, and 1 MHz, while the MIL-STD-461E bandwidths are 10 Hz, 100 Hz, 1 kHz, 10 kHz, 100 kHz, and 1 MHz.

When comparing specifications for different spectrum analyzers, it is important to remember that some functions, such as sweep speed, depend on how other settings, such as RBW filter, are adjusted. The time of a sweep also depends on the frequency span. Other factors to consider in choosing a spectrum analyzer include the inherent noise of the analyzer, in the form of single-sideband (SSB) phase noise and spurious signal products, and the sensitivity of the analyzer. When testing oscillators for phase noise, for example, the instrument's own phase noise will set a limit on the measurement capability.

With their generous digital content, some newer spectrum analyzers offer triggering functions much like an oscilloscope, allowing operators to capture a signal based on timing rather than just on frequency, as in the case when examining pulsed signals. The RSA6000 series of real-time spectrum analyzers from Tektronix (www.tek.com) can trigger on events as short as 10 ns in the time domain and boast capture bandwidths as wide as 110 MHz. These spectrum analyzers, with 6.2-, 14-, and 20-GHz versions, also provide timequalified and frequency-mask triggers for capturing specific signal content according to predetermined parameters. All of the instruments digitize signals at a standard sampling rate of 100 MSamples/s with an optional sampling rate of 300 MSamples/s.

In addition to pure performance, modern spectrum analyzers typically pack a number of automatic measurements into their firmware. The R&S FSU spectrum analyzers from Rohde & Schwarz (www.rohde-schwarz. com), for example, include models from 20 Hz to 3.6 GHz, 8 GHz, 26.5 GHz, 43 GHz, 46 GHz, 50 GHz, and 67 GHz. The RBW filters can be set from 1 Hz to 50 MHz, with a DANL of -158 dBm for a RBW setting of 1 Hz. The phase noise is -133 dBc/Hz offset 10 kHz from any carrier. In addition to the sterling performance, the analyzers include automatic measurement routines for a wide range of tests, including for adjacent- channel power (ACP), occupied bandwidth (OBW), and third-order intercept (TOI). In addition, add-on software can arm the analyzers for performing standards-based testing on wireless communications designs, greatly simplifying both research and production-line testing.