FFT Analyzer Basic FAQ

The mean of a spectrum can be divided into two types: [1] power spectral mean and [2] Fourier spectral mean.
The main purposes of averaging the power spectrum in a single-channel measurement are to average the fluctuating spectral values for deterministic signals (e.g., periodic signals) and to improve the accuracy of statistical spectral estimation for irregular signals. Averaging methods include [1] averaging, [2] exponential averaging, [3] peak averaging, and [4] sweep averaging.

Averaging is a simple average (collective average) of power values for each frequency band, also known as RMS averaging. Exponential averaging is a weighted power average where the number of averaging cycles corresponds to the time constant. Peak averaging is a method that retains the maximum power value in each frequency band (after N averaging cycles), and is not strictly an averaging process. Sweep averaging is a method that uses a sinusoidal signal and sweeps it (i.e., sequentially changes the frequency) to obtain the spectrum, and is divided into internal signal mode and external signal mode. In internal signal mode, a sine wave that is perfectly synchronized with the resolution of the FFT is used with the built-in oscillator equipped in the FFT analyzer, so measurements can be taken with high accuracy.

It's important to note that power spectral averaging is based solely on power values (i.e., EU ² values). For specific averaging methods, please refer to the related sections.

Fourier spectral averaging is an average over complex numbers, and when used in conjunction with a trigger function for synchronous averaging, it has the same effect as time waveform averaging. Although it has the disadvantage of requiring a trigger signal, it has advantages over power spectral averaging, such as [1] reduction of noise components and [2] averaging of the phase spectrum.

	Averaging processing (FFT basic glossary)