Principles of Spectrum Analysis

Spectrum Analysis Principles

Not everyone has access to a spectrum analyser. This is an expensive and used to be a strangely heavy piece of equipment similar to an oscilloscope but that measures signals in the frequency domain instead of the time domain on the horizontal axis.

The frequencies available to us range from DC (direct current) 0 Hz right upto visible light and beyond. Within this great range of frequencies lies the radio frequency spectrum. Within that lies our television and radio frequencies, and our satellite frequencies. It also contains radar and communications for aircraft, police, fire and ambulance, military forces and shipping to name but a few. This is a crowded part of the spectrum. These days nearly everything is wireless and that means more transmissions.

Why do we need to analyse the frequency domain?

Well because RF (radio frequency) waves are invisible, the only way we can see them and then measure them accurately is to analyse the spectrum of frequencies that they are in. Just like an oscilloscope measures voltage against time.

To do this we use what is called a spectrum analyser. This peice of equipment is two things: expensive and complicated. They work by plotting the amplitude of a signal against the frequency. If you look at the picture above, you can see a signal in the middle of the analyser screen.

The noise floor (level of background noise) is visible 2 squares up from the bottom and covers all frequencies across the screen. The signal is 5 squares tall and about half a square wide. This all tells us a lot of information about the signal. The reason is that the screen is divided into squares just like an oscilloscope. Instead of time being on the horizontal axis, it is frequency. Amplitude is on the vertical axis measured in dB.

If our screen is set to 10 GHz wide (span) and our center frequency (CF) is 6 GHz, then our signal is at 6 GHz, the far left of the screen is 1 GHz and the far right is 11 GHz. The height of our signal is 5 squares and if our screen is set to 5 dB per square (dB/Div) then our signal is 25 dB above the noise floor. As you can see knowing all about our screen gives us the information we need to measure our signal.

The noise floor can also be given a value and this allows us to reference the signal to the noise floor. This then lets us see and measure the signals to detrmine if they are the correct size and at the correct frequency. When compared with the expected results from the link budget this is a very useful way of checking all is well.