Understanding Spectrum and Network Analyzer Applications

Signal analysis is invaluable for applications ranging from amplifier characterization to signal intelligence.

In a recent series, we discussed specs such as 1-dB compression and third-order intercept points. At the end, I pointed to some application notes that explain how to use specific spectrum, signal, and vector network analyzers to make these measurements. These instruments’ usefulness extends to many other applications, ranging from component scattering-parameter (S-parameter) characterization to signals intelligence (SIGINT).

Q: First, can you define “spectrum analyzer,” “signal analyzer,” and “vector network analyzer” and explain the differences?
A: Good question. Let’s start with the spectrum analyzer. Analog versions build on early AM-radio superheterodyne technology. In fact, if you have an old AM radio with a mechanical tuning knob, you can try an experiment. First, disable any automatic level control (ALC). Then, start at the low end of some frequency range of interest and slowly turn the dial clockwise. When you come to a station, record the frequency and whether the volume seems soft, medium, or loud. Continue until you reach the end of the frequency range of interest. Then, plot your results. If your old radio has a volume-unit (VU) meter, you can even put some numbers on the y-axis. If you do this from 1 MHz to 1.2 MHz in the Boston area where I live, you will get a plot like Figure 1.

Figure 1. This frequency-domain diagram covers Boston-area AM radio stations in the 1 MHz to 1.2 MHz range.

Q: Does a spectrum analyzer automate this process?
A: It automates an analogous process. In Figure 1, we located the RF carrier frequencies and estimated the amplitudes of modulated audio signals they carry. If we could get each station to transmit the same audio tone during our experiment (and if we make sure our radio’s ALC is off), we could indirectly infer the strength of each carrier. An RF analog spectrum analyzer can directly locate and measure the signal strength of each carrier and express the power in units such as decibels relative to 1 mW (dBm) or 1 µW (dBµ).

Q: How does it work?
A: Like a superheterodyne AM radio receiver, the spectrum analyzer downconverts the input signal to an intermediate frequency (IF). However, the AM radio employs a local oscillator (LO) whose frequency is controlled by an auxiliary variable capacitor mechanically coupled to the main variable capacitor used for tuning (which you adjust using the tuning knob). In contrast, the spectrum analyzer uses a sweep generator to apply a voltage ramp to a voltage-controlled oscillator (VCO) that serves as the LO (Figure 2). Although not shown in Figure 2, the spectrum analyzer may use several mixers and filters to reach the final IF stage.

Figure 2. A spectrum analyzer uses a tracking generator to control a local oscillator.

Q: How do we determine the appropriate sweep time?
A: That depends on the resolution bandwidth (RBW), which is the bandpass filter bandwidth shown in the IF stage of Figure 2. If you’ve ever used a cheap radio, you may have noticed that closely spaced stations seem to interfere with each other — KDKA Pittsburgh at 1.02 MHz and WBZ Boston at 1.03 MHz come to mind from my childhood in central Pennsylvania. If you want your spectrum analyzer to be able to distinguish these two stations’ carriers, you’ll need an RBW much lower than the 20 kHz that separates the two.

Q: Tell me more about RBW
A: RBW might be the most important spectrum analyzer specification. It affects not only frequency resolution but also noise performance. Typical RBW specs can range from 1 Hz to tens of megahertz, and most spectrum analyzers come with selectable RBW filters appropriate to the instruments’ rated maximum frequency and typical applications. Figure 3 shows an example of an RBW filter response.

Figure 3. This IF filter response curve indicates the 6-dB RBW and 60-dB bandwidth, from which a shape factor can be derived.

In this case, relative to the peak at center frequency f_C, the response falls off by 6 dB at ±1 kHz offsets from f_C, so the 6-dB RBW is 2 kHz. We could specify a 3-dB RBW. You should also consider the shape factor — the ratio of the 60-dB bandwidth to the 6-dB bandwidth:

Figure 3 shows the locations of f_60-HIGH, f_60-LOW, f_6-HIGH, and f_6-LOW for the solid curve, whose 60-dB bandwidth is 8.5 kHz. Its shape factor is, therefore, 8.5/2 = 4.25. Note that the dashed curve has the same 2-kHz 6-dB RBW but a wider 11-kHz 60-dB bandwidth, and its shape factor is 11/2 = 5.5. With shape factor, a lower number is better.

Q: So, how do we relate the RBW to the sweep time?
A: In part 2, we’ll investigate the interactions of the input signals, the IF, the RBW, the sweep-generator output, and the VCO output. Then, we will look at several shortcomings of the analog approach and investigate how digital technology overcomes these drawbacks, enabling today’s sophisticated vector signal analyzers and vector network analyzers.