Bearing Fault Frequencies: BPFO, BPFI, BSF, and FTF

Definition: Bearing fault frequencies are the characteristic vibration frequencies at which defects on each bearing component (outer race, inner race, rolling element, and cage) produce measurable impulse patterns. Each frequency is calculated from the bearing geometry and shaft speed, allowing analysts to identify which component is damaged by locating matching peaks in a vibration or envelope spectrum.

What Are Bearing Fault Frequencies?

Bearings fail by developing surface defects such as spalls, pits, or cracks on their raceways, rolling elements, or cage. Each time a rolling element strikes a defect, it generates a brief mechanical impulse. These impulses repeat at a rate determined by the bearing geometry and shaft speed, producing predictable frequency signatures in the vibration signal.

Bearing fault frequencies allow a reliability engineer to compare measured spectrum peaks against calculated values and determine precisely which bearing component has failed, often weeks or months before the bearing reaches a catastrophic state. This is the foundation of vibration analysis for rotating machinery health.

Bearing Geometry Terms

Four geometric parameters govern all bearing fault frequency calculations:

  • Pd (Pitch Circle Diameter): The diameter of the circle passing through the centers of all rolling elements. Also called the pitch diameter.
  • Bd (Ball Diameter): The diameter of one rolling element. For cylindrical roller bearings, this is the roller diameter.
  • Alpha (Contact Angle): The angle between the line connecting ball-to-raceway contact points and a plane perpendicular to the bearing axis. For deep-groove ball bearings under pure radial load this is typically near zero; for angular contact bearings it may be 15 to 40 degrees.
  • N (Number of Rolling Elements): The count of balls or rollers in the bearing. This number is usually available from the bearing manufacturer's datasheet.

The ratio Bd/Pd and the cosine of alpha together determine what fraction of shaft speed each component sees. Bearing manufacturers publish these values, or they can be measured directly from the bearing geometry.

The Four Fundamental Bearing Fault Frequencies

BPFO: Ball Pass Frequency Outer Race

BPFO is the rate at which rolling elements strike a defect on the outer race. Because the outer race is stationary in most applications, the load zone is fixed, and each defect impact occurs at a steady interval.

Formula:

BPFO = (N/2) x S x (1 - (Bd/Pd) x cos(alpha))

Where S is shaft speed in Hz (revolutions per second). The result is in Hz.

BPFO is typically less than shaft speed multiplied by N/2, and it is not an integer multiple of shaft speed. Spectra showing BPFO usually produce clean harmonic series (1x BPFO, 2x BPFO, 3x BPFO) without sidebands, because the outer race defect stays stationary relative to the load zone.

BPFI: Ball Pass Frequency Inner Race

BPFI is the rate at which rolling elements contact a defect on the inner race. The inner race rotates with the shaft, so the defect periodically passes through the loaded zone and then out of it, causing amplitude modulation of the impact signal.

Formula:

BPFI = (N/2) x S x (1 + (Bd/Pd) x cos(alpha))

BPFI is always higher than BPFO for the same bearing. The characteristic spectral signature is BPFI harmonics surrounded by sidebands spaced at shaft rotation frequency (1X), reflecting the modulation effect of the inner race passing through the load zone.

BSF: Ball Spin Frequency

BSF is the spin rate of the rolling element itself about its own axis. A defect on a ball or roller produces impacts at BSF as the damaged surface contacts each raceway in turn.

Formula:

BSF = (Pd / (2 x Bd)) x S x (1 - (Bd/Pd)^2 x cos^2(alpha))

BSF defects are the hardest to detect reliably. The impact energy is split between the two contact points (inner and outer raceway), and rolling elements can exhibit slip rather than perfect rolling contact. This slip causes the actual BSF to drift from the calculated value, spreading energy across a frequency band rather than a sharp peak.

FTF: Fundamental Train Frequency

FTF is the rotation rate of the cage, which holds the rolling elements evenly spaced around the raceways. A cracked, worn, or loose cage produces impacts at FTF and its harmonics.

Formula:

FTF = (S/2) x (1 - (Bd/Pd) x cos(alpha))

FTF is a sub-synchronous frequency, always less than shaft speed. It is typically in the range of 0.4x to 0.45x shaft speed for most standard bearing designs. Because cage defects are uncommon, FTF peaks are often associated with contamination-induced wear or severe lubrication failure rather than surface fatigue.

Comparison Table: Fault Frequency Summary

Fault Frequency Bearing Component Relative to Shaft Speed Typical Harmonic Pattern Sidebands?
BPFO Outer race 3x to 5x shaft speed (typical) 1x to 5x harmonics, rising amplitude with fault growth Rarely
BPFI Inner race 4x to 7x shaft speed (typical) 1x to 5x harmonics with 1X shaft sidebands Yes, at 1X shaft spacing
BSF Rolling element 2x to 3x shaft speed (typical) 1x to 3x harmonics, often modulated at FTF Sometimes at FTF spacing
FTF Cage 0.4x to 0.45x shaft speed (typical) Sub-synchronous, low amplitude unless cage is severely damaged Rarely

Worked Numerical Example

Consider a deep-groove ball bearing running at 1,800 RPM (30 Hz) with the following geometry:

  • Number of rolling elements (N): 9
  • Pitch circle diameter (Pd): 52 mm
  • Ball diameter (Bd): 11 mm
  • Contact angle (alpha): 0 degrees (cos 0 = 1.0)

The Bd/Pd ratio is 11/52 = 0.2115.

BPFO:

BPFO = (9/2) x 30 x (1 - 0.2115)
     = 4.5 x 30 x 0.7885
     = 106.4 Hz

BPFI:

BPFI = (9/2) x 30 x (1 + 0.2115)
     = 4.5 x 30 x 1.2115
     = 163.6 Hz

BSF:

BSF = (52 / (2 x 11)) x 30 x (1 - 0.2115^2)
    = 2.364 x 30 x (1 - 0.04473)
    = 2.364 x 30 x 0.9553
    = 67.7 Hz

FTF:

FTF = (30/2) x (1 - 0.2115)
    = 15 x 0.7885
    = 11.8 Hz

An analyst reviewing the vibration spectrum for this bearing would look for peaks at 106 Hz (and harmonics at 212, 318 Hz) for outer race damage, peaks at 164 Hz with sidebands at 134 Hz and 194 Hz (164 ± 30 Hz) for inner race damage, peaks near 68 Hz for rolling element defects, and a sub-synchronous peak near 11.8 Hz for cage damage.

How Fault Frequencies Appear in Vibration Spectra

A single defect on the outer race produces a periodic impulse each time a rolling element contacts the damaged area. In the FFT spectrum, this appears as a peak at BPFO. As the defect enlarges, the impact amplitude grows and harmonics emerge at 2x BPFO, 3x BPFO, and so on.

The shape of the spectrum tells the story of fault severity. Early-stage faults may show only a slight rise at the fundamental frequency, potentially buried in the noise floor. With a standard vibration monitoring setup measuring velocity or acceleration, early bearing faults are often invisible in the low-frequency spectrum.

This is why envelope analysis is preferred for bearing diagnostics: it demodulates high-frequency structural resonances (typically in the 1 to 20 kHz range) that are excited by the impacts, bringing bearing fault frequencies into view before they are detectable in the conventional velocity spectrum.

Sidebands and Harmonics

Sidebands appear when a fault frequency is amplitude-modulated by another frequency. For inner race faults, modulation occurs at shaft rotation frequency because the damaged inner race surface passes in and out of the bearing load zone once per shaft revolution. The result is a family of peaks at BPFI plus or minus integer multiples of shaft speed (1X, 2X).

The number of visible sidebands indicates fault severity. One pair of sidebands at BPFI plus or minus 1X shaft speed is an early indicator. Five or more sidebands spreading across the spectrum, combined with an elevated broadband noise floor, signal an advanced defect approaching replacement threshold.

Harmonics (2x, 3x, and higher multiples of a fault frequency) develop as a defect grows and the impact shape changes from a narrow spike to a broader impulse. A bearing showing five harmonics of BPFO with no sidebands still points to an outer race defect, while a bearing showing five harmonics of BPFI surrounded by dense sidebands is in a more severe condition.

Envelope Analysis

Standard vibration analysis in the velocity domain (10 to 1,000 Hz) is effective for imbalance, misalignment, and looseness. Bearing faults produce low-amplitude, broadband impacts that are often masked by these dominant machine frequencies.

Envelope analysis solves this by targeting the high-frequency structural resonances excited by bearing impacts. The process has four steps:

  1. Bandpass filter: Select a frequency band around a known structural resonance, typically 2 to 20 kHz. The resonance amplifies bearing impacts, making them much larger relative to the background.
  2. Rectify the signal: The filtered signal is rectified (absolute value taken) so the impacts appear as positive pulses.
  3. Smooth (low-pass filter): A low-pass filter extracts the envelope of the rectified signal, leaving only the amplitude modulation pattern.
  4. Apply FFT to the envelope: The resulting envelope spectrum shows clear peaks at the bearing fault frequencies and their harmonics, now visible far earlier than in a standard spectrum.

Envelope analysis is the most reliable method for early failure mode detection in rolling element bearings and is the standard technique used by automated asset condition monitoring systems.

Early vs. Advanced Fault Indicators

Bearing defects progress through four recognized stages, each with distinct spectral signatures:

Stage 1 (Incipient): High-frequency energy rises in the ultrasonic band (20 to 60 kHz). No visible peaks in the standard velocity or acceleration spectrum. Only detectable with ultrasound instruments or high-frequency envelope analysis. This stage can last months on a well-lubricated bearing.

Stage 2 (Early): Bearing fault frequencies appear in the high-frequency envelope spectrum (2 to 20 kHz). Harmonics may appear. The standard velocity spectrum still shows no bearing-specific peaks. This is the optimal window for intervention: the bearing can be scheduled for replacement at the next convenient maintenance window.

Stage 3 (Moderate): Fault frequencies and harmonics are now visible in the standard velocity spectrum. Sidebands appear around BPFI harmonics. The broadband noise floor begins to rise. A replacement window of two to eight weeks is typical at this stage, depending on operating conditions.

Stage 4 (Advanced): Multiple harmonics with dense sidebands. Elevated broadband noise floor across the full spectrum. Discrete peaks begin to merge into a broad hump. Bearing surface temperature rises measurably. Replacement is urgent; risk of sudden failure is high.

Connecting this progression to a predictive maintenance program gives maintenance teams the information needed to schedule replacements before Stage 4.

Limitations of Bearing Fault Frequency Analysis

Calculated bearing fault frequencies are theoretical values based on ideal rolling contact. Several real-world factors reduce their precision:

  • Rolling element slip: Under light loads, rollers can slip rather than rolling at the theoretical speed. This causes actual fault frequencies to deviate from calculated values by 1 to 2%, smearing energy and making peak identification harder.
  • Missing geometry data: When the exact bearing designation is unknown, geometry must be estimated from physical measurements or generic tables. Errors in Bd or Pd of even 0.5 mm shift calculated frequencies enough to cause misidentification.
  • High background noise: Heavily loaded machines, gearboxes, or machines with imbalance generate high vibration levels that mask early bearing signals even in the envelope spectrum.
  • Distributed defects: Widespread spalling produces a broad rise in the noise floor rather than discrete frequency peaks, making pattern identification ambiguous.
  • Speed variation: Variable-speed drives cause shaft speed to change continuously, shifting fault frequencies in real time. Order-based analysis (tracking fault frequencies as multiples of shaft speed rather than fixed Hz values) is required for accurate diagnosis on variable-speed machines.

Despite these limitations, bearing fault frequencies remain the most reliable diagnostic tool available for industrial vibration analysis, particularly when combined with trend data from a vibration sensor collecting continuous measurements.

Frequently Asked Questions

What are the four fundamental bearing fault frequencies?

The four fundamental bearing fault frequencies are Ball Pass Frequency Outer race (BPFO), Ball Pass Frequency Inner race (BPFI), Ball Spin Frequency (BSF), and Fundamental Train Frequency (FTF). Each corresponds to a specific bearing component and is calculated from shaft speed and bearing geometry: pitch circle diameter, ball diameter, contact angle, and number of rolling elements.

How do you calculate BPFO?

BPFO = (N/2) x S x (1 - (Bd/Pd) x cos(alpha)), where N is the number of rolling elements, S is shaft speed in Hz, Bd is ball diameter, Pd is pitch circle diameter, and alpha is the contact angle. This frequency predicts how often a rolling element strikes the outer race defect per unit time.

What is the difference between BPFO and BPFI?

BPFO represents the contact rate on a stationary outer race defect. BPFI represents contacts on the rotating inner race defect. Because the inner race moves relative to the load zone, BPFI produces sidebands at shaft frequency in the vibration spectrum, while BPFO typically does not. BPFI is numerically higher than BPFO for the same bearing.

What is envelope analysis and why is it used for bearing diagnostics?

Envelope analysis isolates the low-energy impulse patterns produced by bearing defects from the broader vibration signal. It bandpass-filters the signal around a structural resonance, rectifies the result, then applies an FFT to the envelope. This reveals fault frequencies that are invisible in the standard velocity spectrum, making it far more sensitive for early-stage bearing damage detection.

What are sidebands in a bearing vibration spectrum?

Sidebands are spectral peaks appearing symmetrically around a fault frequency at spacing equal to shaft rotation frequency (1X). They result from amplitude modulation of the fault impulse as the inner race defect passes through the load zone. More sidebands or uneven sideband spacing indicates a more advanced defect and greater fault severity.

What are the limitations of bearing fault frequency analysis?

Key limitations include: rolling element slip shifting actual frequencies 1 to 2% from calculated values; missing or incorrect bearing geometry data; high background noise masking early-stage signals; distributed spalling producing broadband noise instead of discrete peaks; and variable shaft speed requiring order-based analysis rather than fixed-frequency tracking.

How many harmonics of a bearing fault frequency indicate an advanced fault?

Early-stage bearing defects typically show one or two harmonics. As the fault grows, three to five harmonics become visible. Advanced faults generate five or more harmonics along with elevated broadband noise floor and, for inner race faults, dense sideband families around each harmonic. The transition from one to five-plus harmonics, tracked over time, indicates fault progression.

The Bottom Line

Bearing fault frequencies turn the physics of rolling contact into actionable diagnostic data. By calculating BPFO, BPFI, BSF, and FTF from shaft speed and bearing geometry, analysts can identify exactly which component is failing and track how quickly the defect is progressing. Early detection in Stage 1 or Stage 2 transforms an emergency replacement into a planned maintenance event, eliminating unplanned downtime and reducing bearing replacement costs.

The technique works best when combined with continuous data collection and envelope analysis, allowing the full harmonic and sideband picture to build over time rather than relying on a single snapshot measurement. When bearing fault frequency tracking is part of a broader asset health program, maintenance teams move from reacting to failures to anticipating them weeks or months in advance.

Detect Bearing Faults Before They Cause Downtime

Tractian continuously monitors bearing fault frequencies using envelope analysis and machine learning, surfacing BPFO, BPFI, BSF, and FTF anomalies in real time across your entire asset fleet.

See How Tractian Detects Bearing Faults

Related terms