FFT "I" or real input is n samples from the sample buffer (preferably with window weighting applied before conversion). Set "Q" or imaginary input for all zeros.
The output is a complex spectrum with a direct current in the position index = 0. Index = 1 corresponds to the sampling frequency divided by the length of the FFT N, index = 2 is twice the frequency, and so on to the N / 2 index. To get power at a specific frequency, you need to add the square real and imaginary parts.
Usually you want to display power in a dB scale, which is calculated as 10 * log10 (power), each block on the display corresponds to, for example, 3dB. You can also add some averaging or peak detection.
More advanced displays also have a logarithmic frequency axis, for example. each column corresponds to 1/12 octave. Ideally, each output has its own filter, but you can achieve something similar by adding outputs with a higher FFT index instead of showing all of them.
source share