User Manual

- 31 -
Select FFT window
■ There are six FFT windows. Each window makes trade-offs between frequency resolution and amplitude
accuracy. Choose the window based on what you want to measure and the characteristics of your source
signal. The following table will help you choose the best window:
Art
Characteristics
Windows
Rectangle
(Rectangle)
This window is best for frequency resolutions, but is the worst for
accurately measuring the amplitude of these frequencies. It is the
best window for measuring the frequency spectrum of non-
repetitive signals and measuring frequency components near DC.
 Use the rectangular window for measuring transients or peaks
where the signal level before and after the event is almost the
same.
 Also usable for sine waves with the same amplitude and with
fixed frequencies
 Broadband noise with relatively slowly varying spectrum.
Hanning
This window is well suited for measuring amplitude accuracy, but
less so for frequency resolutions.
 Use the Hanning window to measure sine, periodic and
narrowband noise.
 Best suited for transients or peaks where the signal levels
before and after the event differ significantly.
Hamming
This is a very good window for frequency resolution with slightly
better amplitude accuracy than the rectangular window. It has a
slightly better frequency resolution than the Hanning window.
 Use the Hamming window to measure sine, periodic and
narrowband noise.
 Best suited for transients or peaks where the signal levels
before and after the event differ significantly.
Blackman
This is the best window for measuring the amplitude of
frequencies, but offers the poorest frequency resolution.
 Use the Blackman-Harris window for single frequency signals
and finding higher order harmonics.
Bartlett
The Bartlett window is a slightly narrower version of the triangular
windows, with "zero weight" at both ends.
Emperor
The frequency resolution when using the Kaiser window is
adequate, the spectral leakage and amplitude accuracy are both
good.
The Kaiser window is best when the frequencies are very close
but have very different amplitudes (the sidelobe level and form
factor are close to the traditional Gaussian RBE). This window is
also good for the random signals.