Power Spectral Density Estimates Using FFT

http://www.mathworks.com/help/signal/ug/psd-estimate-using-fft.html

In the code example, psdx = (1/(Fs*N)) * abs(xdft).^2;

why isn't it instead (Fs/N) such that psdx = (Fs/N) * abs(xdft).^2? Isn't the period equal to the length of the signal, which is N, multiplied by the sampling interval, which is 1/Fs ?

Answer 1

[Jeremy]

There are different ways of looking at it but the PSD is often written as:

(fft*conj(fft))/(bin_width*NFFT^2)

(fft*conj(fft)) = abs(fft) and bin_width = Fs/NFFT, so you can see how it reduces down.