Understanding Demodulation (Envelope Analysis)

Devices

Portable balancer & Vibration analyzer Balanset-1A

€1,975.00 + VAT (if applicable)

Parts

Vibration sensor

€90.00 + VAT (if applicable)

Parts

Optical Sensor (Laser Tachometer)

€124.00 + VAT (if applicable)

Devices

Balanset-4

€6,803.00 + VAT (if applicable)

Parts

Magnetic Stand Insize-60-kgf

€46.00 + VAT (if applicable)

Parts

Reflective tape

€10.00 + VAT (if applicable)

Devices

Dynamic balancer “Balanset-1A” OEM

€1,735.00 + VAT (if applicable)

Demodulation is a signal-processing technique used in vibration analysis to detect repetitive, low-frequency impacts that are effectively “hidden” inside a machine’s high-frequency vibration. It is the engine behind the more familiar term Envelope Analysis, and the two are often used interchangeably. The method isolates a high-frequency band of vibration that behaves as a carrier, then extracts the envelope of that carrier — revealing the underlying repetition rate of tiny, periodic impacts such as those produced by microscopic faults in bearings or gears.

1. Definition: What is Demodulation?

Every defect in a rolling-element bearing or a meshing gear produces a brief mechanical shock each time a loaded surface passes over it. That shock excites the structure’s natural frequencies, making the machine “ring” at frequencies far above running speed. The impacts themselves carry very little energy, but they repeat at a precise, predictable rate tied to the geometry of the component. Demodulation discards the high-frequency ringing and recovers only this repetition rate — the information that actually identifies the fault.

The result is closely linked to the idea of an envelope spectrum: a frequency display computed not from the raw waveform but from its demodulated envelope. Where a conventional vibration spectrum shows the energy in the signal, the demodulated spectrum shows the rhythm of the impacts buried within it.

2. The Process of Demodulation

Demodulation is a three-step chain, applied to the raw signal from an accelerometer before any final transform:

1. Band-Pass Filtering: The raw vibration signal is first passed through a high-frequency band-pass filter. This removes the strong, low-frequency content — unbalance, misalignment, looseness — and keeps only a high-frequency region where the stress waves from bearing or gear impacts excite structural resonances. Choosing this band well (often centred on a known structural resonance) is the single most important setup decision in the whole method.
2. Rectification: The filtered, high-frequency signal is then rectified — the negative half of the waveform is flipped to positive — producing a signal that represents the absolute amplitude of the carrier.
3. Low-Pass Filtering (Enveloping): Finally, the rectified signal is passed through a low-pass filter. This smooths away the high-frequency carrier and leaves behind only the slow-moving “envelope” that traces the peaks of the rectified signal. That envelope directly represents the repetition rate of the underlying impacts.

An FFT is then performed on the envelope signal. The resulting spectrum — the envelope spectrum, or demodulated spectrum — shows clear peaks at the exact fault frequencies of the bearing or gear components, even when those peaks would be invisible in an ordinary spectrum of the raw data.

3. Why is Demodulation So Effective?

Demodulation is one of the most valuable techniques for early fault detection precisely because of how it handles impact signals.

• Early Warning: When a tiny spall on a bearing race is struck by a rolling element, it produces a small, low-energy impact. That impact causes a very brief, high-frequency burst of vibration as the machine structure rings at its natural frequencies — long before the damage is large enough to raise the overall vibration level.
• Separating the Signal from the Noise: In a normal FFT spectrum, the minuscule energy from these early-stage impacts is completely buried under the massive energy of low-frequency vibration such as unbalance. The fault is present in the data, but drowned out.
• Focusing on the Repetition Rate: Demodulation ignores the powerful low-frequency signals entirely. It concentrates on the high-frequency ringing and, crucially, on the repetition rate of that ringing. It is this repetition rate that corresponds directly to the bearing fault frequencies — BPFO, BPFI, BSF — and to the gear mesh frequency (GMF) and its sidebands.

Because demodulation reacts to impacts rather than amplitude, it can flag a defective bearing months before that bearing shows up on a standard velocity spectrum — a decisive advantage in predictive maintenance.

4. Applications and Field Use

The primary applications for demodulation are:

• Rolling-Element Bearing Analysis: It is the definitive method for detecting and diagnosing faults in ball and roller bearings, often providing warning months before the fault becomes critical. The presence of energy at BPFO, BPFI or BSF in the envelope spectrum is a near-unambiguous fingerprint of a localised defect.
• Gearbox Analysis: It is highly effective at detecting cracked or broken gear teeth, which generate a clear impact at 1× the affected gear’s rotational speed in the demodulated spectrum, often accompanied by sidebands.
• Other Impacting Events: It can also detect other repetitive impacting phenomena — steam traps cycling open and closed, or reciprocating-engine valve-timing issues.

In the field, the same instrument used for balancing doubles as a diagnostic tool. A portable two-channel analyser such as the Balanset-1A captures the broadband signal from an accelerometer at each bearing, so a technician can review the ordinary spectrum and the demodulated envelope side by side and decide whether a 1× peak is true imbalance or the first sign of a failing bearing. Related approaches such as the shock pulse method and spike energy exploit the same high-frequency impacts, but demodulation remains the most diagnostic because it preserves the full repetition-rate spectrum rather than collapsing it to a single number.

5. Setup Pitfalls and Good Practice

• Wrong filter band: If the band-pass filter is placed away from a genuine structural resonance, the impacts are not amplified and the envelope spectrum looks empty even when a defect exists. Many instruments offer preset bands; a bump test can confirm where the structure rings.
• Mounting matters: High-frequency impact energy is easily lost through soft mounts. A stud- or adhesive-mounted sensor preserves the carrier far better than a magnet on a painted surface — see ISO 5348 on accelerometer mounting.
• Interpretation, not just detection: A peak in the envelope spectrum should be matched against the calculated fault frequencies for the specific bearing before a diagnosis is made; harmonics of running speed can otherwise be mistaken for a defect.

← Back to Main Index