SPECTRUM ANALYZER

This document contains some general information about spectrum analyzers, and some specific information about the spectrum analyzers you have available in the lab, namely:
• Tektronix 2712
• HP 8590 and the HP 8592

Remember that the spectrum analyzers User's Guides are also located in the lab. Like an oscilloscope, a spectrum analyzer produces a visible display on a screen. Unlike an oscilloscope, however, the spectrum analyzer has only one function-to produce a display of the frequency content of an input signal. (But it is possible to display the time waveform on the spectrum analyzer screen with the proper settings.) And also like an oscilloscope, the spectrum
analyzer will always produce a picture on the screen; but if you do not know how to properly use the spectrum analyzer, that picture may be completely meaningless.

CAUTION: The input of the spectrum analyzer cannot tolerate large signals; before you connect a signal to the input, be sure you know that the signal will not exceed the maximum allowable input rating of the spectrum analyzer. (For example, the Tektronix scope has an RF input power limit of +20dBm, the HP limits at +30dBm. Also keep in mind that if you apply more than one input signal, the maximum allowable amplitude per signal must correspondingly decrease – see the Owner’s Manual for more information.)

Signal Acquisition in a Spectrum Analyzer

Most spectrum analyzers (including the models in lab) are heterodyne1 spectrum analyzers (also called scanning spectrum analyzers). A heterodyne analyzer is essentially a radio receiver (a very sensitive and selective reciever). Radio receivers, including those based on the heterodyne principle, will be covered later in lecture. For now we will provide a simple description of the
basic ideas.
Given a voltage signal x(t), we need to somehow extract the frequency content out of it. As we know, the digital storage oscilloscope provides one solution as it can calculate the FFT of the signal from stored samples. Another solution would be to pass x(t) through a long series of very narrow bandpass filters, having adjacent passbands, and then plot the amplitudes of the filter
outputs. That is, if filter 1 has passband f1 - BW/2 < f < f1 + BW/2, and filter 2 has passband f 2-BW/2 < f < f2 + BW/2, where f1 + BW/2 = f2 - BW/2, and so on, and if BW (the bandwidth) is small enough, then the filter outputs give us the frequency components X(f1), X(f 2), . . . and so on . This is, of course, not a practical solution. A better solution is suggested by a simple
property of Fourier transforms: recall that if we multiply (in the time domain) a signal by a sinusoid, the spectrum of the signal is shifted in frequency by an amount equal to the frequency of the sinusoid.

That is,

Now instead of a bank of narrow filters, we shall have one narrow filter centered at a fixed frequency, say fI, and we shall scan the signal spectrum across this filter by multiplying x(t) by a sinusoid of varying frequency f0 . See Figure 1. The filter is a narrow bandpass filter at a fixed center frequency, fI, (called the intermediate frequency); in a spectrum analyzer, its bandwidth is selected by the user. The oscillator frequency, f0, is adjustable, as indicated in Figure 1. In an ordinary AM or FM radio, when you tune the receiver you are selecting this frequency so that the
desired signal will pass through the filter; in a spectrum analyzer, this frequency is automatically scanned (repeatedly) over a range, which must be selected so that the frequency component X(f) is shifted to fI and passed by the filter. For example, if we want to view the frequency content of x(t) from f1 to f2 , then we must select f0 to scan from f1 + fI to f2 + fI .

Of course, much more signal conditioning is going on inside the spectrum analyzer than is indicated in Figure 1; but the frequency mixing is the fundamental step. In particular, the signal first is passed through a lowpass filter whose bandwidth is chosen to eliminate image frequencies, more on this concept later in the course. Also, most scanning spectrum analyzers are multiple conversion analyzers - they have multiple intermediate frequency stages, at successively lower frequencies. The reason is that we have two conflicting goals to achieve; we would like to
have the filter bandwidth as small as feasible, and we would like to be able to scan over large frequency ranges. It is hard to build sharp narrow filters at high frequencies, but it is also hard to build multipliers that will work over large frequency ranges. Therefore, we achieve narrow filters at low intermediate frequencies by shifting the frequency down in several steps. The User’s
Manual for the Tektronix Spectrum Analyzer has a nice overview diagram its internals it you are interested – you will learn about many of the components given on this diagram at the end of this quarter, or if you continue with 145B/218B.

You may naturally ask why we have a spectrum analyzer if the oscilloscope will display an FFT of a signal. The DSO's display of the FFT has the advantage of capturing one-shot events, as well as being able to store the FFT in memory or on a floppy. But the scanning spectrum analyzer usually holds the advantage over the FFT in frequency range, sensitivity, and dynamic range. If you find yourself working in communications, especially in RF and microwave communications,you will probably find that you will frequently be using a spectrum analyzer for spectral measurements.

more...

RF SPECTRUM ANALYZER