Envelope Analysis (Demodulation) for Early Fault Detection
Envelope analysis — also called demodulation or high-frequency enveloping — is a signal-processing technique in vibration analysis that extracts the faint, repetitive impacts of an early-stage defect from the noisy background vibration of a running machine. It is the single most powerful tool available for detecting incipient damage in rolling-element bearings and gearboxes. Microscopic cracks, spalls and surface flaws each generate a burst of low-energy, high-frequency stress waves every time a rolling element or gear tooth strikes the defect, and envelope analysis is the method that recovers those bursts and reveals the frequency at which they repeat.
1. Definition: What Envelope Analysis Detects
When a rolling element rolls over a tiny pit or crack in a bearing raceway, it does not produce a smooth sine wave — it produces a sharp, hammer-like impact. Each impact is brief and carries very little energy, but it excites the natural ringing (the resonance) of the bearing, the sensor and the surrounding structure at high frequency, typically several kilohertz. These impacts repeat at a precise rate governed by the bearing geometry and shaft speed. Envelope analysis treats those high-frequency rings as a carrier that is being switched on and off — modulated — by the repetitive impacting, and it works backward to recover the modulation pattern. The result tells the analyst not just that something is impacting, but how often, and therefore which part of the bearing is damaged.
2. Why a Standard FFT Is Not Enough
The energy from these initial impacts is usually too small, and sits at too high a frequency, to be visible in an ordinary velocity spectrum produced by a standard FFT. In a routine measurement the impact energy is buried in the broadband noise floor and completely swamped by the large, low-frequency peaks from unbalance, misalignment and mechanical looseness. A plain spectrum, in other words, is dominated by the machine’s healthy 1× and 2× vibration, while the diagnostic information about an emerging bearing fault hides up in the high-frequency region where nobody is looking. Demodulation exists precisely to strip away that low-frequency clutter and lift the modulating fault signal out of the noise.
3. The Envelope Analysis Process
The technique isolates the high-frequency ringing and then measures its repetition rate. In practice it proceeds through four steps:
- Band-pass filtering: The raw signal from the accelerometer is first passed through a high-pass or band-pass filter. This removes the strong low-frequency vibration (typically everything below about 1 kHz or 5 kHz) and keeps only the high-frequency ringing and stress waves produced by the impacts. Choosing the band so that it sits on a structural resonance maximises sensitivity.
- Rectification: The filtered high-frequency signal is then rectified, flipping its negative half upward so that only the magnitude of the ringing remains. This step prepares the signal for enveloping.
- Enveloping (low-pass filtering): A low-pass filter is applied to the rectified signal. It smooths away the fast carrier oscillation and leaves behind only the slowly varying outline — the “envelope” — which traces the amplitude-modulation pattern, i.e. the repetition rate of the original impacts.
- FFT of the envelope: Finally, an FFT is performed on this envelope time waveform. The resulting envelope spectrum displays clean peaks at the frequency of the repetitive impacts, free of the low-frequency machinery vibration that masked them before.
4. Diagnosing Faults with the Envelope Spectrum
The peaks in the envelope spectrum line up with the bearing’s calculated bearing fault frequencies. By matching a measured peak to a known frequency, an analyst can pinpoint exactly where the fault lies:
- BPFO (Ball Pass Frequency, Outer race): a defect on the stationary outer race.
- BPFI (Ball Pass Frequency, Inner race): a defect on the rotating inner race. This peak usually carries sidebands spaced at 1× running speed because the defect moves in and out of the load zone once per revolution.
- BSF (Ball Spin Frequency): a defect on one of the rolling elements themselves.
- FTF (Fundamental Train Frequency): the slowest of the set, pointing to a fault in the cage that holds the rolling elements.
The same logic applies to gearing: a cracked or broken gear tooth impacts once per revolution, so the envelope spectrum shows a peak at that gear’s running speed, often surrounded by sidebands. To turn a bearing’s bore, ball count and speed into the exact target frequencies before you measure, an analyst can use the Bearing Defect Frequency Calculator; for gear meshes the Gear Mesh Frequency Calculator serves the same purpose. Reading the harmonic pattern is itself a form of bearing defect diagnosis: the number and height of harmonics in the envelope spectrum scales with how advanced the damage has become.
5. Where Envelope Analysis Fits in the Field
Envelope analysis is a core capability of any serious condition monitoring programme, and modern portable analysers compute it routinely alongside the ordinary spectrum. In day-to-day field work a maintenance team will often arrive at a machine first to balance it and check its overall health: an instrument such as the Balanset-1A measures broadband vibration and the 1× amplitude and phase needed for field balancing, while a complementary enveloping channel confirms whether the bearings underneath are sound before a balancing job is signed off. Catching the bearing problem first matters, because balancing a machine whose bearings are already spalling only masks the symptom.
6. The Power of Early Detection
The defining advantage of envelope analysis is sheer sensitivity. It can flag a bearing or gear fault months — sometimes a year — before the same defect would grow large enough to register in a routine velocity spectrum or radiate enough heat to show up under thermography. That long lead time is exactly what gives an early warning its value: maintenance can be planned, parts ordered, and the repair slotted into a convenient outage instead of forced by a sudden breakdown. In the wider context of predictive maintenance, the extended warning that demodulation provides is what prevents catastrophic failures and the expensive secondary damage they cause.
