TABLE OF CONTENTS The FFT is inherently a baseband transform. In other words, the frequency range of the FFT always starts at 0 Hz and extends to some maximum frequency, f S/2. This can be a significant limitation in measurement situations where a small frequency band (not starting at DC) needs to be analyzed.
For example, suppose an FFT analyzer has a sample rate, f s = 256 kHz. The frequency range of the FFT would be 0 Hz to 128 kHz ( f s/2). If N = 1024, the frequency resolution would be f s/ N = 250 Hz. Spectral lines closer than 250 Hz could not be resolved. [4]
One way to increase the frequency resolution is to increase N, the number of samples in the time record, which also increases the number of bins in the FFT output. Unfortunately, this increases the size of the arrays that the FFT has to deal with and the computation time increases accordingly. The computation time of the FFT algorithm often limits the performance of the instrument (in the form of update rate to the display), so increasing the size of the FFT is often undesirable.
Reducing f s will also improve the frequency resolution but at the expense of reducing the upper-frequency limit of the FFT and ultimately the instrument bandwidth. This is a worthwhile trade-off that gives the user control over the frequency resolution and frequency range of the...
