5.2  FFT Frequency Resolution Considerations

The frequency resolution of the FFT (sometimes called a resolution element) which describes the frequency spacing of the FFT is given by Equation 5.2.1.

f₀=1/(NT_s)                                                   (5.2.1)
with  N = number of samples
and   T_s= sample interval (inverse of sampling rate)

It is obvious from Equation 5.2.1 that an increase in the spectral resolution can be achieved by increasing N (data interval) or increasing Ts (decrease sampling rate).  Care must be taken when increasing Ts so as not to introduce aliasing by reducing the sampling rate below the required minimum Nyquist rate of 2·f_max.

In this study, the sampling rate is fixed at 950Hz.  Therefore, frequency resolution is strictly a function of the data interval (N) as shown in Equation 5.2.2.

f₀=950/N                                                     (5.2.2)

Also, the data window type (rectangular, Hanning, Hamming, etc.) influences the ultimate resolution of the FFT by "blurring" the transform.  Each window type has its own characteristics, but in general, longer windows produce less spectral blurring.

<PREVIOUS

NEXT>