Working with scope-based spectrum analyzers

The spectrum analyzer is a member of the oscilloscope family. An electrical signal is fed, by means of a BNC cable or probe, into an input port on the front panel, and its graphic image appears on the LCD screen. But rather than the oscilloscope’s familiar time domain graph, in which amplitude in volts is plotted against the Y-axis and time against the X-axis, in the spectrum analyzer a frequency domain graph appears on the spectrum analyzer screen.

Oscilloscope users can access the frequency domain either by pressing Math>FFT or by connecting the signal to the RF input, using an RF adapter. The spectrum analyzer, in contrast, operates exclusively in the frequency domain with the exception of spectrogram and similar displays. Except for the PC-based instrument, it is substantially more expensive, but it has greater spectral range, specifications and analytic capabilities.

In the frequency domain, like the time domain, amplitude is plotted against the Y-axis. But rather than volts, amplitude is quantified as power, in dB. The user can choose either a linear or logarithmic scale. And in the frequency domain, rather than time, frequency is plotted against the X-axis. The time-domain waveform as shown in the oscilloscope display provides an immediate, highly intuitive image of the signal as it propagates through a conductor, and in separate channels shown on the same screen, it displays the difference between any input and its associated output. This is not to say that the electrical signal “looks like” the waveform. What you see on the screen is a graph plotted against Cartesian co-ordinates, whereas an electromagnetic wave consists of changes in flux density or electrical current (number of electrons passing a given point on a conductor per second).

The frequency domain graph of that same signal provides far more concrete information. Harmonics, for example, in the time domain appear as small irregularities in the trace and noise shows up as a barely noticeable thickening of the line. In the frequency domain, in contrast, the spectral distribution of all harmonic energy associated with the signal’s fundamental is displayed as a series of peaks, each located at its precise frequency and with its own bandwidth and amplitude denoted by the height of its peak. All this is in addition to the usual array of broad-spectrum noise. Moreover, the instrument’s intrinsic noise floor is shown, below which no signal can be displayed.

Properly configured, the spectrum analyzer can reveal rare, short duration events, weak signals masked by stronger signals or noise, transient and dynamic signals, burst transmissions, glitches and switching transients. In addition, in the frequency domain, the spectrum analyzer is adept at displaying spread-spectrum and frequency-hopping signals, monitoring spectrum usage and detecting rogue transmissions, sensing and diagnosing EMI, characterizing time-variant modulation schemes and displaying software and hardware interactions.

Compared to the oscilloscope, the spectrum analyzer has a somewhat steeper learning curve. At first, it may not do what you want and expect it to do. This is partly because the terminology used to label front-panel controls and on-screen menus in some instances presupposes instrument-specific knowledge that the oscilloscope user may not possess. Secondly, some signals may not display in a meaningful way.

As an example, you know that a square wave, due to fast rise and fall times, which are high-frequency components, should be encumbered with powerful harmonics. But where are they?

From your arbitrary function generator, begin by conveying to the spectrum analyzer via BNC cable a square wave. Depending on the default settings in both instruments, you will probably see that the fundamental is difficult to distinguish because it is crammed against the left side of the display. Also, discrete harmonics are not shown. For a meaningful display in the spectrum analyzer, press the Frequency/Span button. The first task is to center the fundamental. Disregard the R to Center button. (It places a reference marker at the center of the display.) To center the fundamental, take notice of the frequency of the square wave coming out of the AFG. It is the default 1 MHz. In the spectrum analyzer, accordingly, set Span to twice that amount, 2 MHz. Now the strong fundamental is displayed at the center of the screen, but still there are no harmonics. Why?

The problem is that the span is too small. The third harmonic, as you can see from the label at the top of the display, is actually 3 MHz. (The fundamental is considered the first harmonic.) Setting Span for a start to 20 MHz, the harmonics are now quite prominent, but there are not many of them displayed because we quickly run out of screen space. Changing Span to 200 MHz, many more of them are shown, but not the full range. Boosting Span to 400 MHz, there are more harmonics, but now it becomes more difficult to resolve them. Such are the trade-offs inherent in waveform display! Notice that as you change Span, Start and Stop frequencies adjust accordingly. Of course, the center frequency stays the same.

There are three basic spectrum analyzer architectures: swept spectrum (SA), vector signal (VSA) and real time (RTSA). The most recent, RTSA, is the most advanced and best choice in today’s laboratory or shop, but the others are valuable in many applications and may be economical alternatives.

The swept-tuned, superheterodyne spectrum analyzer is suitable for displaying, in the frequency domain, controlled, static signals. Using superheterodyne circuitry, it down-converts a signal under investigation and then sweeps it through the pass band of its resolution bandwidth (RBW) filter. The signal is then fed to a detector, which determines the amplitude at each frequency. The downside is that for a variable signal there will be gaps where it is not acquired. Transient signals may be lost. Moreover, impulse signals as in radar and communications systems are essentially invisible. Current SA’s are no longer analog machines, but even the latest digital circuitry is not capable of dealing with dynamic events within the signal of interest.

VSAs are used when digitally modulated signals are displayed because both phase and amplitude information are required. To begin, an internal ADC digitizes the wide-band IF signal, followed by down conversion, filtering and detection. Discrete Fourier Transform (DFT) algorithms perform transformation of time to frequency domain. The VSA is capable of measuring modulation parameters including FM deviation, Code Domain Power and Error Vector Magnitude. Additionally, constellation diagrams, channel power, power vs. time and spectrograms can be displayed.

As a digital instrument, the VSA can store waveforms in memory, but it lacks transient event analysis functionality. Like the SA, the VSA cannot perform operations involving transient events. The VSA is limited insofar as all computations cannot be performed at a rate sufficient to keep up with all possible changes in the input signal.

The RTSA is a relatively recent improvement over the SA and VSA. It is the preferred spectrum analyzer because it is better able to measure and display transient and dynamic RF signals. To be considered real time, spectrum analysis must be processed without gaps. All time-domain information must move intact into the frequency domain. Signal processing must have sufficient bandwidth to support analysis of the signal of interest. The ADC clock rate must exceed the Nyquist criteria for the capture bandwidth. The analysis interval must support the narrowest RBW of interest. DFT rates exceeding the Nyquist criteria for RBW require overlapping DFT frames.

Spectrum analyzers are not feasible for some because of the price, which is substantially higher than that of oscilloscopes with comparable specifications. A practical solution is the PC-based real time spectrum analyzer, such as the Tektronix RSA306B. The reduced price of this PC-based spectrum analyzer arises from the fact that it makes use of the computing power and display capability of a user-supplied PC.

The Tektronix RSA306B USB PC-Based Spectrum Analyzer uses SignalVu-PC RF signal Analysis software, available from the manufacturer’s website. The instrument can display and analyze signals from 9 kHz to 6.2 GHz with a measurement range of +20 dBm to -160 dBm. It performs two sweeps per second over the entire frequency span. The acquisition bandwidth is 40 MHz, enabling wide-band vector analysis of current standards. The minimum signal duration is 27 μs/sec with 100% probability of interception.