How to use
Enter the signal frequency and analyser settings (Fmax, number of lines).
Pick a window; the table shows the amplitude error of every window for the same signal.
Move the frequency between lines (e.g. 24.6 → 24.75 Hz) to see how the error changes; "snap" shows the best case.
Theory & formulas
fs = 2.56·Fmax, N = 2.56·lines, Δf = Fmax/lines, record length T = N/fs = 1/Δf.
Worst-case amplitude (picket fence) error: rectangular ≈ −36%, Hann ≈ −15%, Hamming ≈ −18%, flat-top < 0.1%. Hann is the general-purpose standard in vibration analysis.
ENBW (equivalent noise bandwidth) N·Σw²/(Σw)² in lines; a wide ENBW raises the noise floor and makes close frequencies harder to separate.
Impact tests use no window or a force/exponential window; continuous signals use Hann, and flat-top is preferred for calibration and accurate amplitude.
Frequently asked questions
Which window should I use?
Hann for routine machine vibration. Flat-top for calibration and tolerance checks where amplitude must be exact. To separate two close frequencies, increase resolution; flat-top makes peak separation harder.
Why does the peak look wide in the spectrum?
The main-lobe width of the window and a signal falling between lines spread the peak over several lines. More lines or a lower Fmax make it narrower.