From FFT fundamentals to fault diagnosis: learn to read vibration spectra, calculate bearing defect frequencies, assess severity per ISO 10816, and diagnose unbalance, misalignment, looseness, bearing and gear defects — with the Balanset-1A.
What is Vibration Analysis?
Quick answer: Vibration analysis is the process of measuring and interpreting mechanical oscillations of rotating machinery to diagnose faults without disassembly. Using FFT (Fast Fourier Transform), the complex vibration signal is decomposed into individual frequency components. Each fault produces a characteristic spectral "fingerprint": unbalance at 1× RPM, misalignment at 2×, looseness as multiple harmonics, bearing defects at non-synchronous frequencies. The Balanset-1A performs both balancing and spectrum analysis in one portable instrument.
Every rotating machine vibrates. In a healthy machine, vibration is low and stable — its normal "operating signature." As defects develop, vibration changes in predictable ways. By measuring and analysing these changes, we can identify the root cause, predict failure, and schedule maintenance before catastrophic breakdown. This is the foundation of predictive maintenance.
Fault Identification at a Glance
Each mechanical fault produces a characteristic "fingerprint" in the vibration spectrum:
| Fault | Primary Frequency | Harmonics | Direction | Phase Behaviour | Key Distinguishing Feature |
|---|---|---|---|---|---|
| Static unbalance | 1× | Low / none | Radial (H,V) | In-phase both bearings | Pure 1× sinusoid. Amplitude ∝ ω². |
| Dynamic unbalance | 1× | Low / none | Radial (H,V) | ~180° between bearings | 1× dominant, bearings out of phase (couple). |
| Parallel misalignment | 2× (≥ 1×) | 1×, 3× | Radial | 180° across coupling | 2× often > 1×. High radial at coupling. |
| Angular misalignment | 1×, 2× | 3× | Axial dominant | 180° across coupling (axial) | High axial. Axial ≥ 50% of radial. |
| Component looseness | 1×, 2×…10×+ | Many (~10×) | Radial | Erratic | "Forest" of harmonics. Possible 0.5× sub. |
| Structural looseness | 1× or 2× | Few above 2× | Vertical | Unstable | Strong vertical. Responds to bolt check. |
| Outer race (BPFO) | BPFO, 2×BPFO… | Multiple BPFO | Radial | N/A | Non-synchronous. No 1× sidebands. |
| Inner race (BPFI) | BPFI, 2×BPFI… | Multiple BPFI | Radial | Modulated at 1× | BPFI harmonics with ±1× sidebands. |
| Rolling element (BSF) | BSF, 2×BSF… | Multiple BSF | Radial | N/A | 2×BSF often > 1×BSF. Non-synchronous. |
| Cage (FTF) | FTF ≈ 0.4× | 2, 3× FTF | Radial | N/A | Sub-synchronous (< 1×). |
| Gear mesh | GMF = N × 1× | 2, 3× GMF | Radial + axial | Modulated at 1× | GMF with sidebands. N = teeth. |
| Electrical (motor) | 2× line freq | — | Radial | Drops on power-off | 100/120 Hz. Instant drop test. |
FFT: The Core of Spectrum Analysis
A vibration sensor (accelerometer) converts mechanical oscillation into an electrical signal. Displayed over time, this is the waveform — a complex, seemingly chaotic curve when multiple faults are present. FFT (Fast Fourier Transform) decomposes this complex signal into individual sinusoidal components, each with its own frequency and amplitude.
Think of FFT as a prism splitting white light into a rainbow. The complex waveform is "white light" — FFT reveals the individual "colours" (frequencies) hidden inside. The result is the vibration spectrum — the primary diagnostic tool.
f_1x = RPM / 60 (Hz)
Key Spectrum Parameters
- Frequency (X-axis, Hz): How often oscillations occur. Directly linked to the source. 1× = shaft speed. 2× = twice shaft speed.
- Amplitude (Y-axis, mm/s RMS): Vibration intensity at each frequency. Higher peaks = more energy = more serious condition.
- Harmonics: Integer multiples of the fundamental: 2× (2nd), 3× (3rd), 4×, etc. Their presence and relative height carry diagnostic information.
- Phase (°): Timing relationship at different measurement points. Essential for distinguishing unbalance (in-phase) from misalignment (180°).
Vibration Measurement Units: Displacement, Velocity, Acceleration
Vibration can be measured as three different physical parameters. Each emphasises different frequency ranges, making them suited to different diagnostic tasks. Understanding when to use which parameter is fundamental to effective analysis.
| Parameter | Unit | Frequency Range | Best For | Standards |
|---|---|---|---|---|
| Displacement | µm pk-pk | 1–100 Hz | Slow machines (< 600 RPM), shaft orbit, proximity probes, journal bearings. 1 mil = 25.4 µm. | ISO 7919 (shaft vibration) |
| Velocity | mm/s RMS | 10–1000 Hz | General machinery monitoring — unbalance, misalignment, looseness. Default parameter. Balanset-1A measures in mm/s RMS. | ISO 10816, ISO 20816 |
| Acceleration | g or m/s² RMS | 500 Hz – 20 kHz | Early bearing defects, gear mesh, impacts, high-speed machinery. 1 g = 9.81 m/s². | ISO 15242 (bearing vibration) |
Conversion at a single frequency:
v = 2πf · d a = 2πf · v = (2πf)² · d
Rule of thumb: If you have only one sensor and one parameter to choose — choose velocity (mm/s RMS). It covers the broadest range of common faults with flat response. The Balanset-1A uses this as its native parameter. Add acceleration measurement only when you need to catch early-stage bearing or gear defects at high frequencies.
Measurement Technique with Balanset-1A
Sensor Placement
The quality of diagnosis depends entirely on measurement quality. Vibration forces are transmitted through bearings, so sensors must be mounted on bearing housings — as close to the bearing as possible, on the load-bearing structure (not covers or cooling fins).
- Surface preparation: Clean, flat, free of paint flakes. Magnetic base must sit flush.
- Radial horizontal (H): Perpendicular to shaft, horizontal plane. Often highest amplitude.
- Radial vertical (V): Perpendicular to shaft, vertical plane.
- Axial (A): Parallel to shaft. Critical for detecting misalignment.
Two-channel diagnostic trick: The Balanset-1A has 2 channels. For diagnostics, mount both sensors on the same bearing — one radial, one axial. This gives simultaneous radial + axial spectrums, enabling instant misalignment detection.
Balanset-1A Modes for Diagnostics
- F1 — Spectrum Analyser: Full FFT display. The primary diagnostic mode.
- F5 — Vibrometer: Quick assessment. Compare V1s (total RMS) vs. V1o (1×). If V1s ≈ V1o → unbalance. If V1s ≫ V1o → other faults.
- F8 — Charts: Detailed spectrum + time waveform. Best for harmonic patterns and bearing frequencies.
Warning — V1s vs. V1o, the first diagnostic check: Before balancing, compare V1s with V1o. If V1s ≫ V1o (e.g., 8 vs. 2 mm/s), most vibration is NOT from unbalance. Balancing won’t solve it — examine the full spectrum.
Phase Analysis — The Diagnostic Differentiator
Frequency tells you what is vibrating; phase tells you how. Two faults can produce identical spectrums (both dominated by 1×) — only phase analysis distinguishes them. Phase is the angular relationship between vibration at different measurement points, measured in degrees (0°–360°).
| Phase Relationship | Measurement Points | Diagnosis | Explanation |
|---|---|---|---|
| 0° (in-phase) | Bearing 1 ↔ Bearing 2 (radial) | Static unbalance | Both bearings move together in sync — single heavy spot in centre of rotor. Single-plane correction. |
| ~180° (anti-phase) | Bearing 1 ↔ Bearing 2 (radial) | Dynamic (couple) unbalance | Bearings rock in opposition — two heavy spots at different planes create a rocking couple. Two-plane correction needed. |
| ~90° | Horizontal ↔ Vertical (same bearing) | Unbalance (any type) | Normal for unbalance — force vector rotates with shaft, producing ~90° between H and V at same point. |
| ~180° | Across coupling (radial) | Parallel misalignment | Coupling forces push shafts apart in opposite radial directions. 180° across coupling with high 2× is the signature. |
| ~180° | Across coupling (axial) | Angular misalignment | Shafts alternately push/pull axially. 180° axial across coupling with high 1× and 2× is definitive. |
| 0° | Across coupling (axial) | Not misalignment | Both sides moving same axial direction — likely thermal growth, piping strain, or soft foot. Not angular misalignment. |
| Erratic / unstable | Any consistent points | Mechanical looseness | Phase readings jump randomly between measurements — characteristic of impacts in loose joints. Unstable phase = looseness. |
| Slowly drifting | Any point, over time | Resonance or thermal effects | Gradual phase shift during warmup suggests structural stiffness changing with temperature (thermal misalignment). |
| Consistent, non-0/180° | Bearing 1 ↔ Bearing 2 | Combined static + couple unbalance | Phase between 0° and 180° indicates a mix of static and couple components — requires two-plane balancing. |
Phase measurement with Balanset-1A: The instrument displays phase at 1× using the tachometer as reference. To compare phase between two bearings, measure each bearing in the same direction (e.g., horizontal) with the tachometer on the same reference mark. The difference in phase readings reveals the fault type. No special software needed — just subtract the two readings.
Fault 1: Unbalance
Cause: Centre of mass displaced from rotation axis. Manufacturing tolerances, deposit buildup, erosion, broken blade, lost weight.
Spectrum: Dominant peak at exactly 1× RPM. Very low harmonics. Radial vibration. Amplitude increases with speed² (quadratic). Phase is stable and repeatable.
Static Unbalance (Single-Plane)
Pure 1× peak, sinusoidal waveform. Both bearings in-phase. Single-plane correction.
Typical spectrum: dominant 1× at 25 Hz (1500 RPM), minimal harmonics.
Dynamic Unbalance (Two-Plane / Couple)
Also 1× dominant, but bearings ~180° out of phase. Two-plane correction required.
Typical spectrum: similar to static unbalance — the difference shows in phase between bearings, not in the spectrum.
Action: Perform rotor balancing with the Balanset-1A — see the dynamic shaft balancing instruction. G-grade tolerance per ISO 21940-11 / ISO 1940-1.
Fault 2: Shaft Misalignment
Cause: Axes of coupled shafts do not coincide. Can be parallel (offset) or angular (tilted), usually both.
Parallel Misalignment (Radial)
High 1× and 2× in the radial direction. 2× often ≥ 1×. 180° phase shift across coupling.
Typical spectrum: strong 1× and 2× with minor 3×, radial direction.
Angular Misalignment — Radial
1× and 2× present in radial, but 2× typically dominates.
Angular Misalignment — Axial
Axial vibration ≥ 50% of radial. 180° phase across coupling in axial. This is the key distinguishing measurement.
Typical spectrum: very high 2× in the axial direction.
Action: Balancing will NOT help. Stop the machine and perform shaft alignment. Re-check vibration after.
Fault 3: Mechanical Looseness
Cause: Loss of structural stiffness — loose bolts, cracks in foundation, worn bearing seats, excessive clearances.
Component Looseness
"Forest" of harmonics — 1×, 2×, 3×, 4×… up to 10×+ with decreasing amplitude. May show 0.5× subharmonics.
Typical spectrum: many harmonics 1× through 10×, often with a 0.5× subharmonic.
Structural Looseness
1× and/or 2× dominant. Few higher harmonics. Strong vertical vibration.
Action: Inspect and tighten mounting bolts. Check foundation. Always check looseness before balancing.
Fault 4: Rolling Bearing Defects
Cause: Pitting, spalling, wear on raceways, rolling elements, or cage.
BPFO = (n/2)(1 − Bd/Pd·cos α) · f_s
BPFI = (n/2)(1 + Bd/Pd·cos α) · f_s
BSF = (Pd/2Bd)(1 − (Bd/Pd·cos α)²) · f_s
FTF = ½(1 − Bd/Pd·cos α) · f_s
Outer Race Defect (BPFO)
Series of peaks at BPFO, 2×BPFO, 3×BPFO… No 1× sidebands (stationary ring). Most common bearing fault.
Inner Race Defect (BPFI)
BPFI harmonics with ±1× sidebands (rotating ring, load zone modulation). Sideband pattern is the key identifier.
Rolling Element Defect (BSF)
BSF harmonics. 2×BSF often dominant. Non-synchronous. Often accompanied by race damage.
Cage Defect (FTF)
Sub-synchronous peaks (FTF ≈ 0.4× shaft speed). Low frequency. Often accompanies other bearing damage.
Bearing defect progression (4 stages):
Stage 1 — Subsurface: Ultrasonic zone (> 5 kHz). Not visible on standard FFT. Detectable by spike energy / enveloping.
Stage 2 — Early defect: Bearing frequencies appear (BPFO, BPFI). Low amplitude. This is where Balanset-1A begins detection.
Stage 3 — Progressed: Multiple harmonics. Sidebands develop. Noise floor rises.
Stage 4 — Advanced: Broadband noise. Bearing frequencies may disappear into noise. Replacement urgent.
Envelope (Demodulation) Analysis — Early Bearing Detection
Standard FFT spectrum analysis detects bearing defects from Stage 2 onward. But in Stage 1, bearing impacts are too weak to appear above the noise floor. Envelope analysis (also called demodulation or high-frequency detection, HFD) extends detection to much earlier stages.
When a rolling element hits a defect, it generates a short impact pulse that excites high-frequency structural resonances (typically 5–20 kHz). These resonances "ring" briefly at each impact. Envelope analysis works in three steps:
- Band-pass filter: Isolate the high-frequency resonance band (e.g., 5–15 kHz) where the impacts ring.
- Rectify and envelope: Extract the amplitude modulation pattern — the "envelope" that follows the peaks of the ringing.
- FFT of the envelope: Apply FFT to the envelope signal. The result shows the repetition rate of impacts — which equals the bearing defect frequencies (BPFO, BPFI, BSF, FTF).
Why envelope detects earlier: In the raw spectrum, a weak impact at BPFO might produce 0.1 mm/s — invisible among machine noise of 2 mm/s. But that same impact excites a resonance at 8 kHz where there is no other vibration source. After demodulation, the BPFO repetition pattern emerges clearly from a clean background.
Related parameters:
- Spike Energy (SE): Overall measurement of high-frequency impact energy. Scalar trending value. Good for "go/no-go" screening.
- gSE / HFD / PeakVue: Vendor-specific names for envelope-derived parameters. All based on the same principle.
- Acceleration enveloping: The Balanset-1A measures in velocity (mm/s). For full envelope analysis, a dedicated analyser with acceleration input and band-pass filtering capability is ideal. However, the Balanset-1A’s FFT can still detect Stage 2+ bearing defects effectively in the standard velocity spectrum.
Action: Check lubrication. Plan bearing replacement. Increase monitoring frequency. See the bearings glossary for defect frequency definitions.
Fault 5: Gear Defects
Cause: Worn, pitted, or broken teeth. Gear eccentricity. GMF = number of teeth × shaft RPM / 60.
Gear Eccentricity
GMF with sidebands at ±1× shaft speed. Gear’s 1× may also be elevated.
Gear Tooth Wear / Damage
Multiple GMF harmonics with dense sidebands. Severity tracks with sideband count and amplitude.
Action: Check gearbox oil for metallic particles. Schedule inspection. Monitor GMF sideband trend.
Fault 6: Electrical Faults (Motors)
Electromagnetic faults produce vibration at 2× line frequency (100 Hz on 50 Hz grids, 120 Hz on 60 Hz). Critical test: vibration disappears instantly when power is cut. Mechanical faults decay gradually.
- Stator eccentricity: 2× line frequency, steady amplitude.
- Rotor bar defects: Sidebands around line frequency at slip frequency intervals.
- Soft foot: Vibration changes when individual motor feet are loosened.
Fault 7: Belt Drive Problems
Cause: Worn, misaligned, or improperly tensioned belts. Belt drives generate vibration at the belt pass frequency, which is typically a sub-synchronous frequency (below 1× shaft speed) since the belt is longer than the pulley circumference.
f_belt = (π · D · RPM) / (60 · L)
Common belt signatures:
- Belt wear / defect: Peaks at belt frequency (fbelt) and its harmonics (2×, 3×, 4× fbelt). These appear below 1× shaft speed — sub-synchronous peaks are the key indicator.
- Belt misalignment: Elevated axial vibration at 1× and 2× shaft speed. Similar to shaft misalignment but restricted to the belt-driven machine.
- Improper tension: High 1× vibration that changes dramatically with belt tension adjustment. Overtight belts increase bearing load; loose belts cause slapping and belt-frequency peaks.
- Resonance: Belt natural frequency (belt "flutter") can be excited if belt span resonance coincides with operating speed. Visible as broad peak at belt natural frequency.
Action: Check belt condition, tension, and pulley alignment. Replace worn belts. For recurring issues, verify pulley alignment with a laser tool or straight-edge.
Fault 8: Pump Cavitation
Cause: Vapor bubbles form and collapse violently when local pressure drops below the liquid’s vapor pressure — typically at the pump suction. Each bubble collapse creates a micro-impact. Thousands of collapses per second generate a characteristic broadband noise.
Spectral signature:
- Broadband high-frequency energy: Unlike mechanical faults (which produce discrete peaks), cavitation generates a raised noise floor across a wide frequency range, typically above 2–5 kHz. The spectrum looks like a "hump" or elevated plateau rather than sharp peaks.
- Random, non-periodic: No harmonics, no relationship to shaft speed. The noise sounds like "gravel" or "crackling" — audible even without instruments.
- Low-frequency effects: Severe cavitation may also cause instability at 1× and broadband low-frequency noise from flow turbulence.
Action: Increase suction pressure (lower pump, open suction valve, reduce suction pipe losses). Check NPSHavailable vs. NPSHrequired. Reduce pump speed if possible. Cavitation causes rapid erosion damage — do not ignore.
Fault 9: Oil Whirl & Oil Whip (Journal Bearings)
Cause: Fluid-film instability in journal (sleeve) bearings. The oil film wedge forces the shaft to orbit within the bearing clearance at a sub-synchronous frequency. This is distinct from rolling element bearing defects and occurs only in plain/journal bearings.
Oil Whirl
- Frequency: Approximately 0.42× to 0.48× shaft speed (often cited as ~0.43×). This is a sub-synchronous peak that tracks shaft speed — if RPM increases, the whirl frequency increases proportionally.
- Spectrum: A single peak at ~0.43× that shifts with speed. Amplitude may be moderate.
- Condition: Precursor to oil whip. Usually not immediately destructive but indicates instability.
Oil Whip
- Frequency: Locks onto the rotor’s first natural frequency (critical speed). Unlike whirl, it does NOT track shaft speed — the frequency remains constant as RPM changes.
- Spectrum: Large sub-synchronous peak at the rotor’s first critical speed. Amplitude can be very high — destructive.
- Condition: Dangerous. Immediate action required. Can lead to bearing wipe-out and shaft damage.
Oil whirl vs. looseness — how to distinguish: Both produce sub-synchronous peaks, but: oil whirl is at ~0.43× (not exactly 0.5×) and tracks with speed. Looseness produces peaks at exactly 0.5×, 1.5×, 2.5× and does not track with speed (stays at fixed fractions of 1×). Oil whirl only occurs in journal/sleeve bearings — if the machine has rolling element bearings, it cannot be oil whirl.
Action: For oil whirl: check bearing clearance, oil viscosity, and load. Increase bearing loading or change oil viscosity. For oil whip: reduce speed immediately below the critical threshold. Consult a rotor dynamics specialist.
ISO 10816 Vibration Severity — Complete Classification Table
ISO 10816 (superseded by ISO 20816 but still widely referenced) defines vibration severity zones for four machine classes. Vibration is measured as velocity in mm/s RMS on bearing housings. The table below shows all zone boundaries for all four classes — use it as a quick reference when evaluating measurements.
| Machine Class | Zone A (Good) | Zone B (Acceptable) | Zone C (Alert) | Zone D (Danger) |
|---|---|---|---|---|
| Class I — small machines ≤ 15 kW (pumps, fans, compressors) | ≤ 0.71 | 0.71 – 1.8 | 1.8 – 4.5 | > 4.5 |
| Class II — medium machines 15–75 kW (without special foundation) | ≤ 1.8 | 1.8 – 4.5 | 4.5 – 11.2 | > 11.2 |
| Class III — large machines > 75 kW (rigid foundation) | ≤ 2.8 | 2.8 – 7.1 | 7.1 – 18 | > 18 |
| Class IV — large machines > 75 kW (flexible foundation, e.g. steel frame) | ≤ 4.5 | 4.5 – 11.2 | 11.2 – 28 | > 28 |
How to use this table:
- Determine your machine class by power and foundation type.
- Measure overall vibration velocity (mm/s RMS) on each bearing housing in radial direction.
- Find the zone. Zone A = newly commissioned or excellent. Zone B = unrestricted long-term operation. Zone C = acceptable only for limited periods — schedule maintenance. Zone D = damage is occurring — stop machine as soon as possible.
Remember: trends matter more than absolute values. A machine running at 3.0 mm/s (Zone B for Class II) that was previously at 1.5 mm/s has doubled — investigate the cause even though it’s still "acceptable." The Balanset-1A’s vibrometer mode (F5) displays overall velocity V1s for instant zone assessment.
ISO 10816 vs. ISO 20816: ISO 10816 was formally superseded by ISO 20816 (published 2016–2022). The zone boundaries remain similar for most machine types, but ISO 20816 adds evaluation criteria for displacement and expands machine-specific parts. In practice, ISO 10816 values remain the industry-standard reference. Both the Balanset-1A and most industrial vibration programs still use ISO 10816 zones.
From Measurement to Monitoring
Trend Analysis
A single spectrum is a snapshot. The power of vibration analysis is trend analysis — tracking changes over time.
- Create a baseline: Measure new or known-good equipment. Save spectra.
- Establish intervals: Critical: weekly. Standard: monthly. Auxiliary: quarterly.
- Ensure repeatability: Same points, same directions, same operating conditions.
- Track changes: A 2× increase from baseline is significant even if in ISO Zone A.
Decision Algorithm
- Get a quality spectrum (F8 Charts, radial + axial).
- Identify the highest peak — this is the dominant problem.
- Match to fault type:
- 1× dominates → Unbalance → Balance with Balanset-1A.
- 2× dominates + high axial → Misalignment → Realign shafts.
- Many harmonics → Looseness → Inspect and tighten.
- Non-synchronous peaks → Bearing → Plan replacement.
- GMF + sidebands → Gear → Check oil, inspect gearbox.
- Fix the dominant fault first — secondary symptoms often disappear.
Frequently Asked Questions — Vibration Analysis
What is vibration analysis?
The process of measuring and interpreting mechanical oscillations to diagnose machinery faults without disassembly. FFT decomposes complex vibration into frequency components. Each fault produces a characteristic spectral "fingerprint" — unbalance at 1× RPM, misalignment at 2×, looseness as multiple harmonics, bearing defects at non-synchronous frequencies.
How do I tell unbalance from misalignment?
Unbalance: dominant 1× peak, radial, stable phase, amplitude ∝ speed². Misalignment: significant 2× (often ≥ 1×), high axial vibration, 180° phase across coupling.
What are bearing defect frequencies?
Frequencies generated when rolling elements pass over defects. BPFO = outer race, BPFI = inner race, BSF = ball/roller, FTF = cage. They are NOT integer multiples of shaft speed — see the formulas in the bearing defects section above.
What is a good vibration level?
ISO 10816 zones for Class II (15–75 kW): A < 1.8, B < 4.5, C < 11.2, D > 11.2 mm/s RMS. Trends matter more — a 2× increase from baseline is significant even in Zone A.
Can Balanset-1A do vibration analysis?
Yes. 2-channel FFT spectrum analyser (F1), vibrometer (F5), detailed charts (F8). Mount both sensors radial+axial on one bearing for instant misalignment detection.
Time waveform vs. FFT spectrum?
Waveform shows amplitude vs. time — complex when multiple faults overlap. FFT decomposes into individual frequencies (like a prism splitting light). Each spectrum peak = one vibration source.
How often should I measure vibration?
Critical: weekly. Standard: monthly. Auxiliary: quarterly. After maintenance: immediately. Consistency is key — same points, directions, conditions.
What causes 0.5× (subharmonic) vibration?
Severe mechanical looseness, oil whirl in journal bearings, or cracked shaft. Half-order peaks (0.5×, 1.5×) arise from impacts every other revolution. Serious condition — prompt attention required.
Diagnose First — Then Balance
The Balanset-1A is both a 2-channel vibration analyser and a precision field balancer. Identify the fault by spectrum, then fix it — all with one instrument. When the spectrum says "unbalance," follow the dynamic shaft balancing instruction to correct it on-site.