FFT on unequal spaced data.

Hi,

What is the effect of apply FFT on unequal space data? The frequency domain will also unequally spaced? Why some of the signal processing, they remain the DC component?

Should i remove DC component?

Thank you.

Answer 1

[Star Strider]

The d-c component is the mean of the signal, so it is up to you to keep it or discard it (setting it to zero).

The Fourier transform requires equally-spaced data, at least in my experience. One spectral estimation function that does not require equally-spaced data is the Lomb-Scargle periodogram, implemented in the Signal Processing Toolbox plomb function.

[Walter Roberson]

Generally speaking, the Fourier Transform does not require equally spaced data, but the MATLAB fft() routine does, as is common for most Fast Fourier implementations that do not specifically say they accept non-uniform data.

[Ken W]

Thank you.

Could you explain more detail to me or give me some examples for the effect of DC component? Especially in spectral analysis.

[Star Strider]

Th d-c component is the first value (zero frequency) of the FFT vector and as opposed to the other components, is always a real (not complex) value. It has no effect on the other frequency components.