Spectra

Inspect a vibration record with a spectrum, an envelope spectrum and a kurtogram, and mark 1×, 2× and bearing fault frequencies on the chart. The maths runs in your browser; your record is not uploaded.

Spectra in half a minute

A short tour from the waveform to the bearing frequencies, and how to read the results.

Read the transcript

Narrates how a vibration record moves from the waveform and spectrum to the bearing frequencies in the envelope.

Narration: synthetic voice (Piper, voice “Joe”, CC0 public domain). Captions are burned into the video.

In three steps.

1

Read the record

Pick a synthetic example or load your own CSV. Enter the sampling rate and unit; the record summary appears at once.

2

Split it into frequencies

The waveform, FFT spectrum and spectrogram come together. Running speed and its harmonics are marked on the spectrum.

3

Match bearing frequencies

The envelope spectrum is taken from the most impulsive band. Peaks that match BPFO, BPFI, BSF or FTF are listed as candidates.

1

Data

Pick a record. All are mathematically generated examples, not real plant data.

My own vibration record

Drop a CSV or TXT file here

Pick one numeric column; if there is a time column, the sampling rate can be inferred from it.

2

Settings

The window and FFT length they change.

1000 Hz

Example bearing (deep-groove, illustrative): check your bearing’s data sheet and enter your own values.

(mm/s RMS)

illustrative boundaries: replace them with the limits for your machine class from ISO 20816

RMS
Effective size of the signal; relates to energy.
Peak
Largest absolute value.
Impulsiveness: about 1.41 for a pure sine.
About 3 for Gaussian noise; impacts raise it.
Envelope kurtosis
Impulsiveness of the envelope; rises with a defect.

Waveform

First 0.25 seconds, mean and trend removed. Drag to zoom.

Amplitude spectrum

Each line is a frequency component; its height is the amplitude. Dashed lines are running-speed harmonics; coloured lines are bearing frequencies. Drag to choose the envelope band.

Spectrogram

Time left to right, frequency bottom to top; brighter means stronger.

Bright cells are impulsive bands; the brightest one is used for the envelope. Each row has more bands as the level rises.

Envelope spectrum

Impact repetition frequencies. A bearing defect shows a peak at frequencies such as BPFO or BPFI.

Overall velocity RMS (10–1000 Hz). The needle shows the zone against your boundaries.

    What this suggests

    These are hints, not a diagnosis.

      The decision is the engineer’s. For the limits, see the reading note.

      Strongest 8 peaks

      f/fr is the ratio to running speed. A match is written when the peak is near a harmonic or bearing frequency.

      #HzAmplitudeMatch

      Nominal values. Enter your catalogue data.

      ComponentHz

      Export

      How do I read the result?

      1. 1Verdict card. Colour and wording together give the state: Normal, Watch or Attention. It sets a priority; it is not a diagnosis.
      2. 2Spectrum. 1×, 2× and 3× lines are running-speed harmonics. Bearing frequencies have their own colours. Drag to select a band.
      3. 3Kurtogram and envelope. The band of the brightest kurtogram cell is applied to the envelope. Peaks at BPFO, BPFI, BSF or FTF and their multiples are listed as candidates.
      4. 4Severity. The overall velocity RMS is compared with the A/B, B/C and C/D boundaries you enter. The defaults are illustrative; enter the ISO 20816 limits for your machine class.
      5. 5What a match means. It is only a frequency match. Confirm with trend, temperature, oil analysis and inspection.
      Spectra gives hints, not a diagnosis. Bearing frequencies are nominal: slip can shift real peaks by a percent or two. The sampling rate and the accelerometer bandwidth limit the usable range, so check mounting and the sensor’s frequency range. An acceleration record suits envelope analysis best.

      Frequently asked

      What files and units does Spectra accept?

      A CSV or TXT file, or a table pasted as text. You choose one numeric column as the signal. Enter the sampling rate in Hz, or let it be inferred from a time column in seconds or milliseconds. The unit can be g, m/s² or mm/s. At most 16,384 samples are analysed; longer records use their first 16,384 samples.

      Why an envelope spectrum?

      A rolling-element defect makes a short impact at each contact. The impacts excite a high-frequency structural resonance. The envelope extracts how that resonance’s amplitude changes, so the repetition rate of the impacts, such as BPFO or BPFI, shows directly in the envelope spectrum. It is a way to separate the defect frequency from high-frequency noise.

      Why velocity for overall severity and acceleration for bearings?

      Overall severity scales use the RMS of vibration velocity, which relates to the energy that causes wear and heating. Bearing defects appear as short impacts at high frequency, which an accelerometer captures better. Spectra derives velocity from acceleration by dividing each component by 2πf.

      Is a zone A–D a diagnosis?

      No. The zones only place the overall velocity value relative to the boundaries you enter. The default boundaries are illustrative and do not represent ISO 20816; enter the limits for your machine class from the standard yourself. A zone does not say what the cause is.

      Is my data uploaded?

      No. The record is processed in your browser by a WebAssembly module written in C. Nothing is sent anywhere, and the record is gone when you close the page. The CSV, JSON and PNG files you export stay with you.

      Go deeper.

      Full library
      Instrumentation

      Nyquist and aliasing: how fast to sample sensor data

      After reading, you can set a sampling rate from your analysis band, recognise a false low-frequency peak caused by aliasing, and see why slow PLC or SCADA logs fall short for spectra.

      October 7, 20265 min

      Related questions

      All questions
      ASP Dijital

      Start with a production problem.

      Let’s review your data source, your existing system and the decision you want to improve.

      Discuss your project ↗