Vibration Fault Signatures: Frequencies and What They Mean
Key Takeaways
- Every mechanical fault type produces a distinct spectral pattern. Knowing the expected frequency, amplitude relationship, and directional behavior of each fault allows analysts to identify the problem without disassembly.
- Synchronous faults (imbalance, misalignment, looseness) appear at integer multiples of shaft speed. Non-synchronous faults (bearing defects, gear mesh) require separate calculation from component geometry.
- Fault severity changes both the amplitude and the complexity of the spectrum. Early faults produce subtle high-frequency indicators. Late-stage faults generate broadband noise and rapidly rising overall levels.
What Are Vibration Fault Signatures?
When a mechanical component degrades, it imparts repetitive force impulses into the machine structure at a rate governed by its geometry and rotational speed. Those impulses produce vibration energy that concentrates at specific frequencies in the FFT spectrum. The pattern of peaks, their relative amplitudes, their directions, and their relationships to running speed together constitute the fault signature.
Vibration analysis relies on recognizing these patterns the same way a clinician reads an ECG: the shape of the signal tells you what is happening and roughly how far along the fault has progressed. A trained analyst can distinguish imbalance from misalignment from a bearing defect without removing a single fastener, based solely on where energy appears in the spectrum and how it behaves under phase measurement or load change.
Synchronous vs. Non-Synchronous Frequencies
Vibration frequencies are classified relative to shaft running speed (1x), which is the fundamental reference for all spectral interpretation.
Synchronous frequencies occur at exact integer multiples of running speed: 1x, 2x, 3x, and so on. They are caused by faults that produce one or more force impulses per shaft revolution. Imbalance, angular misalignment, looseness, and shaft bow all fall into this category. Because their frequency shifts proportionally when shaft speed changes, they remain at the same harmonic integer under any operating condition.
Non-synchronous frequencies are not integer multiples of running speed. They include bearing defect frequencies, gear mesh frequencies, and sub-synchronous features such as rub. They are calculated from component geometry rather than directly from shaft speed, so they drift as speed changes but do not lock onto clean harmonics. Seeing energy at a non-synchronous frequency is an immediate indicator that the fault involves a rolling element, gear tooth, or fluid film interaction rather than rotor mass distribution.
Major Fault Signatures and Their Spectral Patterns
Imbalance
Imbalance is the most common fault in rotating machinery. It occurs when the mass center of a rotor does not coincide with its geometric center, producing a centrifugal force that rotates with the shaft.
Spectral pattern: Dominant peak at exactly 1x running speed in the radial direction (both horizontal and vertical planes). Axial vibration remains relatively low. The 1x amplitude scales with the square of shaft speed, so doubling speed quadruples amplitude, which is a useful diagnostic check.
Phase indicator: Radial phase readings taken 90 degrees apart on the same bearing are approximately 90 degrees apart. Phase across two bearings on the same shaft is in-phase (0 to 30 degrees difference for static imbalance) or 180 degrees out of phase (couple imbalance).
Common causes: Material buildup or erosion on fan blades or impellers, missing balance weights after maintenance, component eccentricity, and shaft bow from thermal effects.
Misalignment
Misalignment occurs when two coupled shafts are not on the same centerline, either angularly or in parallel offset. Both types appear in the vibration spectrum, though with different emphasis.
Spectral pattern: Elevated 2x peak is the most reliable indicator. Angular misalignment produces strong axial vibration at 1x and 2x. Parallel offset misalignment tends to emphasize radial 2x. Severe misalignment adds energy at 3x and higher harmonics, creating a harmonic comb.
Phase indicator: Axial phase measured across a coupling is typically 180 degrees out of phase for angular misalignment. Radial phase across the coupling is also 180 degrees for parallel offset misalignment.
Common causes: Improper coupling alignment during installation or after maintenance, thermal growth differential between driver and driven machine, soft-foot on the baseplate, and pipe strain forcing the shaft off-center under operating conditions.
Mechanical Looseness
Looseness introduces nonlinearity into the system, causing the shaft motion to clip or truncate. This truncation generates a family of harmonics from a single input frequency.
Spectral pattern: Multiple harmonics of running speed (1x, 2x, 3x extending out to 10x or beyond) with no single dominant peak. The spectral pattern often looks like a "forest" of harmonics. Sub-harmonics at 0.5x or 1.5x sometimes appear when looseness is severe. Time waveform truncation (flat-topping) is diagnostic.
Looseness is categorized by location. Type A is a loose rotor on the shaft (shows 0.5x and harmonics). Type B is loose bearing in the housing (prominent harmonics at 1x through several multiples). Type C is structural looseness in the machine mount or baseplate (harmonics at running speed with amplitude that varies with measurement direction).
Common causes: Worn or loose bearing fits, loose hold-down bolts, deteriorated grout or base, and loose coupling elements.
Rolling Element Bearing Defects
Rolling element bearings produce four mathematically distinct defect frequencies based on bearing geometry and shaft speed. These are non-synchronous, falling between integer harmonics of running speed.
The four frequencies are:
- BPFO (Ball Pass Frequency Outer Race): The rate at which rolling elements pass a defect on the outer race. Typically 3x to 8x shaft speed for most bearings. This is usually the first detectable bearing defect because the outer race is stationary and the impact energy is consistent in direction.
- BPFI (Ball Pass Frequency Inner Race): The rate at which rolling elements pass a defect on the inner race. Inner race defects modulate with shaft speed because the defect rotates through the load zone, producing sidebands spaced at 1x around BPFI.
- BSF (Ball Spin Frequency): The rate at which a ball or roller spins on its own axis. BSF defects are relatively rare and can be difficult to detect because a ball defect only contacts each race twice per spin, producing low-amplitude, non-synchronous peaks.
- FTF (Fundamental Train Frequency): The cage rotation frequency, typically 0.3x to 0.5x shaft speed. Cage defects appear as low-frequency, non-synchronous peaks and often indicate lubrication breakdown.
Spectral pattern evolution: Early-stage defects appear in the high-frequency envelope spectrum (typically above 5 kHz) with no change in overall vibration. As the defect grows, bearing defect frequencies become visible in the standard low-frequency spectrum with sideband families. Late-stage defects produce broadband noise, disappearance of discrete peaks, and rapidly rising overall levels. See envelope demodulation below for early detection technique.
Resonance
Resonance occurs when an excitation frequency coincides with the natural frequency of a structural component or the rotor system itself. It amplifies whatever vibration exists at that frequency, potentially by 10x or more depending on damping.
Spectral pattern: A sharp amplitude peak at a fixed frequency that does not shift when shaft speed changes. Alternatively, broadband elevated vibration across a frequency band when excitation is continuous. Structural resonance can be excited by any fault signature that produces energy near the natural frequency.
Confirmation method: A bump test (impact hammer test) with a spectrum analyzer identifies natural frequencies from the structural response. A coast-down or run-up plot (Bode plot of amplitude and phase versus speed) shows a 180-degree phase shift and amplitude peak precisely at the resonant frequency as the machine passes through critical speed.
Why it matters: Resonance does not indicate a defective component, but it does accelerate wear. A failure mode that would normally produce low-amplitude vibration can become destructive if the fault frequency coincides with a resonance. Structural modifications, mass changes, or stiffness changes (dampers, isolation mounts) are the corrective paths.
Rubs
A rub occurs when a rotating component makes intermittent contact with a stationary part. The contact forces alter the rotor dynamics and introduce sub-synchronous energy.
Spectral pattern: Sub-synchronous peaks at 0.5x and sometimes 0.333x running speed (fractional harmonics). The 0.5x peak appears first. As the rub worsens, the spectrum fills with both sub-harmonics and multiple running speed harmonics. The time waveform shows a characteristic spike at the moment of contact followed by decaying oscillation.
Distinction from looseness: Looseness rarely produces sub-synchronous energy below 0.5x, while rubs are defined by the sub-synchronous component. Phase analysis also differs: looseness shows in-phase harmonics, while a rub often produces unstable or shifting phase due to nonlinear rotor dynamics.
Gear Mesh Faults
Gearboxes produce mesh frequency energy under normal operation. Fault signatures appear as changes in the amplitude, sideband structure, and harmonic content around gear mesh frequency.
Gear mesh frequency (GMF): Number of teeth on the gear multiplied by the shaft's rotational speed (in Hz). A gear with 48 teeth on a shaft running at 25 Hz produces a GMF of 1,200 Hz.
Healthy gear spectral pattern: GMF peak with low amplitude and minimal sidebands. Second harmonic (2x GMF) is present but significantly lower than the fundamental.
Fault signatures:
- Tooth wear (uniform): Increased GMF amplitude with sideband families spaced at the shaft speed of the worn gear. Sidebands grow as wear progresses.
- Single-tooth defect: Prominent sidebands at exactly 1x shaft speed around GMF. Time waveform shows a repeating impact at once-per-revolution intervals.
- Eccentricity or pitch line runout: Amplitude modulation of GMF produces sidebands spaced at 1x. The modulation pattern is symmetric in a Bode plot of GMF amplitude versus time.
- Ghost frequencies: Peaks at non-integer multiples of running speed related to tooth-spacing errors from manufacturing. These are fixed in frequency regardless of load and appear even on new gearboxes.
Fault Signature Comparison Table
| Fault Type | Dominant Frequency | Synchronous? | Key Additional Indicators |
|---|---|---|---|
| Imbalance | 1x (radial dominant) | Yes | In-phase radial, low axial, amplitude scales with speed squared |
| Angular misalignment | 1x + 2x axial | Yes | Strong axial, 180-degree axial phase across coupling |
| Parallel misalignment | 2x (radial dominant) | Yes | 180-degree radial phase across coupling, elevated axial |
| Looseness | 1x + multiple harmonics | Yes | Harmonic series, waveform truncation, possible 0.5x |
| Outer race bearing defect | BPFO (non-synchronous) | No | Consistent amplitude, detectable in envelope spectrum early |
| Inner race bearing defect | BPFI (non-synchronous) | No | 1x sidebands around BPFI (load zone modulation) |
| Cage defect | FTF (0.3x to 0.5x) | No | Sub-synchronous, often associated with lubrication failure |
| Resonance | Natural frequency (fixed) | No | Does not shift with speed, 180-degree phase shift at resonance |
| Rub | 0.5x (sub-synchronous) | No | Impact spike in waveform, unstable phase, sub-harmonics |
| Gear tooth wear | GMF + sidebands | Yes (GMF is synchronous to gear shaft) | Sidebands spaced at shaft speed of worn gear |
How Fault Severity Changes the Signature
Fault severity follows a progression that changes both the amplitude and the spectral complexity of the vibration signature. Understanding this progression is essential for distinguishing a nascent defect from one approaching failure.
Stage 1 (early): The fault is detectable only through specialized techniques. For bearings, envelope demodulation reveals defect frequencies at high sensitivity. Overall vibration levels are within normal range. Standard low-frequency FFT shows no anomaly. Kurtosis (a statistical measure of impulsiveness) may begin to rise above 3.0.
Stage 2 (developing): Defect frequencies become visible in the standard FFT spectrum. Harmonic series begin to develop around the primary defect frequency. Sidebands appear. Overall vibration levels start to rise above baseline but remain within warning thresholds. This is typically where most condition-based maintenance programs intervene.
Stage 3 (advanced): Multiple harmonic families are prominent. Broadband noise floor rises as the defect damages surrounding surfaces. Overall vibration levels exceed alarm thresholds. Discrete peaks may begin to merge into broadband noise. The failure mode is well established and repair timelines become urgent.
Stage 4 (failure imminent): Discrete spectral peaks disappear into broadband noise. Overall vibration levels rise sharply and unpredictably. Kurtosis may paradoxically decrease as impacts become continuous rather than impulsive. Bearing temperature rises. The machine requires immediate shutdown to prevent secondary damage.
Envelope Demodulation for Early Bearing Detection
Standard FFT analysis of raw vibration data can miss early-stage bearing defects because the low-energy impact pulses from a small defect are masked by the dominant low-frequency content (1x, 2x, GMF) in the spectrum.
Envelope demodulation solves this by processing the high-frequency content where defect impacts excite structural resonance:
- Bandpass filter: The raw signal is bandpass-filtered to isolate the resonant frequency band of the bearing housing (typically 2 kHz to 20 kHz, chosen where bearing resonance is known or detected via bump test).
- Rectification: The filtered signal is rectified (absolute value taken) to convert bipolar oscillations into a unipolar signal whose peaks represent impact amplitude.
- Low-pass filtering: A low-pass filter smooths the rectified signal to produce the envelope waveform.
- Second FFT: An FFT of the envelope waveform reveals bearing defect frequencies (BPFO, BPFI, BSF, FTF) with high signal-to-noise ratio, even at early fault stages.
Automated vibration monitoring systems perform envelope demodulation continuously, enabling detection weeks to months before the defect becomes visible in standard spectral analysis. This is the primary method used in predictive maintenance programs targeting bearing failures.
Reading Multiple Fault Indicators Together
Real machines rarely present with a single perfectly clean fault signature. Misalignment accelerates bearing wear, so both a 2x peak and bearing defect frequencies may coexist. Looseness can amplify imbalance signatures. Resonance can magnify any fault frequency that falls near a natural frequency.
Effective spectral analysis therefore follows a systematic sequence:
- Identify all dominant peaks and classify each as synchronous or non-synchronous.
- Compare current spectrum to a baseline taken when the machine was known to be healthy.
- Use phase analysis and directional measurements to confirm the primary fault type.
- Check bearing defect frequencies against manufacturer geometry data and assess severity via envelope spectrum.
- Integrate root cause analysis findings with maintenance history to determine whether the fault is progressive or related to a recent intervention.
Industrial vibration analysis programs formalize this sequence into data collection routes with defined measurement points, sensor orientations, and alarm limits so that analysts compare equivalent data at each collection interval.
Instrumentation and Sensor Selection
The fault signatures detectable in any given measurement depend on the sensor type, its mounting method, and its frequency response range.
Accelerometers are the standard sensor for most rotating equipment vibration analysis. They measure acceleration across a broad frequency range (typically 1 Hz to 10 kHz for general purpose; to 20 kHz or beyond for envelope demodulation). A vibration sensor mounted with a stud provides the flattest frequency response; magnet mounting introduces a low-pass effect that attenuates high-frequency bearing content and can cause misleading readings above 1 kHz to 2 kHz.
Velocity sensors (geophone-type or converted accelerometer output) are used for overall vibration severity measurements per ISO 10816 and ISO 20816 standards. They are less suited for high-frequency bearing envelope work but provide reliable low-frequency fault signature data.
Proximity probes measure shaft displacement directly and are used for fluid-film bearing machines (large steam turbines, compressors) where rolling element bearing signatures do not apply. Sub-synchronous instabilities (oil whirl at ~0.45x to 0.48x, oil whip locking to rotor critical speed) are only detectable with proximity probe measurements.
Frequently Asked Questions
What is a vibration fault signature?
A vibration fault signature is the characteristic frequency pattern that appears in a machine's vibration spectrum when a specific mechanical defect is present. Each fault type produces energy at predictable frequencies related to shaft speed, component geometry, or resonant properties, allowing analysts to identify the fault type and affected component before failure occurs.
What is the difference between synchronous and non-synchronous fault frequencies?
Synchronous frequencies occur at exact integer multiples of shaft running speed (1x, 2x, 3x) and are associated with faults like imbalance, misalignment, and mechanical looseness. Non-synchronous frequencies are not integer multiples of running speed and include bearing defect frequencies (BPFO, BPFI, BSF, FTF) and gear mesh frequencies with their sidebands. Distinguishing between the two categories is the first step in narrowing down the likely fault type.
How do you identify an imbalance fault from a vibration spectrum?
Imbalance produces a dominant peak at 1x running speed in the radial direction (horizontal and vertical). The phase reading across the bearing span is typically in-phase (0 to 30 degrees difference). Axial vibration remains low relative to radial. As imbalance worsens, the 1x amplitude increases proportionally to the square of speed, which distinguishes it from looseness or misalignment that also show 1x energy.
What bearing fault frequencies should I monitor and how are they calculated?
The four standard bearing defect frequencies are: BPFO (ball pass frequency outer race), BPFI (ball pass frequency inner race), BSF (ball spin frequency), and FTF (fundamental train frequency, also called cage frequency). Each is calculated from shaft speed and bearing geometry factors provided by the bearing manufacturer. BPFO typically falls between 3x and 8x shaft speed for most bearings. These frequencies appear in the high-frequency spectrum and are most reliably detected using envelope demodulation (high-frequency resonance technique) rather than raw waveform analysis.
What is envelope demodulation and when is it used?
Envelope demodulation (also called high-frequency resonance technique or HFRT) is a signal processing method used to detect early-stage bearing defects. When a bearing defect strikes a race or rolling element, it produces a brief high-frequency impact that excites the natural resonance of the bearing housing. Envelope demodulation bandpass-filters the signal around that resonant frequency, rectifies it, and applies a second FFT to reveal bearing defect frequencies in the resulting spectrum. It detects defects weeks to months earlier than standard low-frequency FFT analysis.
How does fault severity change the vibration signature?
Fault severity progression follows a predictable pattern. Early-stage faults typically show defect frequencies in the high-frequency envelope spectrum with little change in overall vibration levels. Intermediate faults add harmonic series and sidebands to the spectrum while overall vibration begins rising. Late-stage faults show broadband noise, loss of discrete peaks, and rapidly increasing overall levels. For bearings, the ISO 10816 severity zones and trending of statistical indicators like kurtosis help quantify severity independent of the frequency pattern.
What does gear mesh frequency indicate?
Gear mesh frequency (GMF) equals the number of gear teeth multiplied by shaft rotational speed. A healthy gear produces GMF with low amplitude and minimal sidebands. Worn or damaged gears show increased GMF amplitude with prominent sideband families spaced at the shaft speed of the affected gear. Eccentricity or uneven tooth wear produces sideband asymmetry. Ghost frequencies at non-integer multiples of running speed may indicate manufacturing errors in tooth spacing.
Can vibration signatures detect resonance?
Yes. Resonance appears as a sharp amplitude peak that occurs at a fixed frequency regardless of running speed changes, or as broadband elevated vibration when excitation is continuous. A coast-down or run-up Bode plot (amplitude and phase versus speed) confirms resonance by showing a 180-degree phase shift and amplitude peak at the resonant frequency. Resonance alone is not a defect but amplifies other fault signatures and accelerates component wear if the operating speed is close to a natural frequency.
The Bottom Line
Vibration fault signatures give maintenance teams a non-invasive window into the mechanical condition of rotating assets. Each fault type, whether imbalance, misalignment, looseness, bearing defect, resonance, rub, or gear mesh anomaly, produces a distinct spectral pattern. Recognizing these patterns from frequency, amplitude, direction, and phase data allows analysts to confirm what is failing, estimate how far along the fault has progressed, and plan corrective action before the machine reaches a failure state.
The practical value of fault signature analysis depends on consistent data collection from calibrated sensors, baseline spectra taken under known-good conditions, and systematic comparison over time. Organizations that move from periodic route-based data collection to continuous automated monitoring can detect bearing defects at Stage 1, misalignment within days of a maintenance event, and looseness before secondary component damage occurs. The earlier the detection, the lower the repair cost and the smaller the production impact.
Identify Vibration Faults Before They Cause Downtime
Tractian's condition monitoring platform applies automated spectral analysis across your entire rotating equipment fleet, detecting imbalance, misalignment, bearing defects, and gear faults continuously without manual data collection routes.
See How Tractian Identifies Vibration FaultsRelated terms
Robot Maintenance
Robot maintenance covers scheduled and condition-based activities to keep industrial robots operating reliably. Learn key components, failure modes, maintenance schedules, and how condition monitoring applies.
RTU (Remote Terminal Unit)
An RTU (Remote Terminal Unit) is a ruggedized field device that acquires I/O signals, runs local logic, and communicates with a SCADA master over DNP3, IEC 60870, or Modbus.
SAE JA1011
SAE JA1011 is the SAE International standard defining minimum criteria a process must meet to qualify as Reliability-Centered Maintenance (RCM).
Salvage Value
Salvage value is the estimated amount a company expects to recover from an asset at the end of its useful life through sale, trade-in, or disposal.
Schedule Compliance
Schedule compliance measures the percentage of planned work orders completed on time. Learn the formula, benchmarks, causes of low compliance, and how to improve it.