KutupASP DijitalThe FFT (Fast Fourier Transform) is an algorithm that splits a recorded vibration signal into its frequency components. The result is a spectrum: a plot of how much amplitude sits at each frequency. On a machine, this makes peaks at running speed, gear mesh or defect frequencies visible one by one.
An accelerometer records vibration by sampling it thousands of times per second. In the time domain, the record is a curve of amplitude against time. That curve can be the sum of several sinusoids, and picking them out by looking at the waveform is hard.
In the frequency domain, the same record becomes a plot where each frequency appears as its own bar. The FFT takes N samples and computes this transform with a fast algorithm. For a real signal, the usable result is about N/2 frequency bins, and each bin covers a narrow band of frequencies.
Bin width depends on record length. A longer record gives narrower bins and separates close components more easily. The FFT tells you which frequencies are present; it does not tell you why they are there.
Consider a hypothetical pump. The accelerometer samples at fs = 1,024 Hz and records for 4 seconds, so N = 1,024 × 4 = 4,096 samples.
If the same pump is recorded for only 1 second, Δf becomes 1 Hz. Components at 30 Hz and 30.5 Hz then differ by half a bin and may look like one wide peak.
Most misreadings come from three habits:
The Fourier transform is the mathematical definition for a continuous signal. The discrete Fourier transform (DFT) does the same job for a finite set of samples. The FFT is a fast way to compute the DFT. The result is the same; the computation is much quicker.
Usually not. A short record shows one moment of the machine. Taking several records under the same load and speed, then comparing them, is more reliable.
Example numbers are hypothetical, not measured field results.
After reading, you can compute frequency resolution for a vibration record, justify your window choice, and read spectral peaks as clues rather than diagnoses, using a worked example.
Data from a vibration sensor cannot be interpreted without knowing which quantity it measures, which frequency range it sees and how it was mounted.
