FFT Resolution & Sampling Calculator
Configure your vibration analyzer’s data acquisition settings. Calculate frequency resolution from sampling parameters, or work backwards from the resolution you need.
Sampling Parameters
Results
Enter your parameters and click Calculate.
Understanding FFT Parameters
The frequency resolution Δf determines the smallest frequency difference you can distinguish in your spectrum. It is governed by a simple relationship: Δf = fs / N, where fs is the sampling rate and N is the number of samples in the time block. A longer time record (more samples at the same rate) gives finer resolution.
For bearing analysis, resolution matters because defect frequencies like BPFO and BPFI can be close to shaft harmonics or electrical frequencies. If your Δf is too coarse, these peaks blur together and become impossible to identify. As a rule of thumb, your effective resolution should be at least 3–5 times finer than the smallest frequency gap you need to distinguish.
The Nyquist frequency (fs / 2) is the absolute upper limit of what the FFT can represent. In practice, most analyzers use a 2.56× oversampling factor — meaning fs = 2.56 × fmax — to leave room for the anti-aliasing filter’s roll-off slope.
Why Window Functions Matter
The FFT assumes the sampled signal repeats infinitely. Since real signals don’t, discontinuities at the block edges cause spectral leakage: energy from a single-frequency peak smears into neighboring bins, masking nearby lower-amplitude peaks.
Window functions taper the signal to zero at the block edges, reducing leakage at the cost of widening the main lobe. The effective resolution accounts for this widening. For example, a Hanning window multiplies Δf by 1.5, meaning you need 50% finer raw resolution to achieve a given effective resolution.
- Rectangular: No windowing. Best raw resolution but worst leakage. Only appropriate when the signal is perfectly periodic within the block (synchronous sampling).
- Hanning: The standard choice for vibration analysis. Good balance of resolution and leakage suppression.
- Hamming: Similar to Hanning with slightly better sidelobe rejection but a wider main lobe in the first sidelobe.
- Flat Top: Sacrifices resolution for amplitude accuracy (<0.01 dB error). Used for calibration and precise amplitude measurements.
- Kaiser (β=6): Tunable window. At β=6, it offers a practical compromise between Hanning and Flat Top.
Common Analyzer Configurations
Commercial vibration analyzers typically specify their FFT size in “lines of resolution,” which equals N / 2. The table below shows standard configurations and their resulting parameters.
| Lines | N (samples) | fs | Δf | Record Time | Typical Use Case |
|---|---|---|---|---|---|
| 400 | 1,024 | 1 kHz | 1.25 Hz | 0.80 s | Quick route-based screening |
| 800 | 2,048 | 2 kHz | 1.25 Hz | 0.80 s | General vibration analysis |
| 1,600 | 4,096 | 5 kHz | 1.56 Hz | 0.64 s | High-frequency gear mesh / blade pass |
| 3,200 | 8,192 | 10 kHz | 1.56 Hz | 0.64 s | Detailed bearing defect analysis |
| 6,400 | 16,384 | 20 kHz | 1.56 Hz | 0.64 s | High-resolution bearing / ultrasonic |
Practical Tips
- Power-of-2 rule: FFT algorithms are most efficient when N is a power of 2. If your record length produces a non-power-of-2 N, most analyzers will either truncate or zero-pad to the next power of 2. This calculator flags when your N isn’t a power of 2 and suggests the nearest one.
- Averages vs. resolution: More averages improve signal-to-noise ratio but don’t improve frequency resolution. Only increasing N (or decreasing fs) improves Δf.
- Low-speed machines: At low RPM, defect frequencies drop below 10 Hz. You need very fine resolution (0.05–0.1 Hz) which demands long records (10–20 seconds). Consider order tracking or synchronous time averaging for these applications.
- Memory constraints: Larger N means more data per acquisition. At 24-bit resolution, a single 65,536-sample block is 256 KB. For continuous monitoring with multiple channels, file sizes add up quickly.
Related Tools
Use the Bearing Defect Frequency Calculator to determine the exact BPFO, BPFI, BSF, and FTF frequencies for your bearing, then return here to configure your analyzer settings to resolve them.
For more on vibration analysis techniques and bearing monitoring, visit iotbearings.com.