🔍

Bearing Fault Frequency Calculator

Enter a rolling-element bearing’s geometry — number of balls/rollers, ball diameter, pitch diameter, contact angle — and the shaft RPM to compute the four classic defect frequencies: BPFO (outer race), BPFI (inner race), BSF (ball/roller spin) and FTF (cage), each with its 2X and 3X harmonics. These are the frequencies to look for in a vibration spectrum when you suspect a bearing fault.

ℹ These are the exact published rolling-element formulas, but every result is only as good as the geometry you type in — read the ball count, ball/pitch diameter and contact angle from the bearing manufacturer’s datasheet. This tool deliberately has no built-in part-number database: it will not invent SKF/NSK/FAG/Timken dimensions for you. Results assume the listed geometry is exact and ignore slip, so real fault peaks land near, not exactly on, these values. Matching a spectral peak to a defect frequency is a strong indicator, not a diagnosis — confirm with a calibrated accelerometer and a trained analyst.

Bearing geometry & shaft speed

Count the rolling elements (Z).
Rolling-element diameter.
Centre-to-centre circle ≈ (bore + OD)/2.
0° for deep-groove; e.g. 15–40° for angular-contact.
Inner race rotates, outer race fixed.

Fault frequencies & harmonics

No geometry? Rule-of-thumb estimate

When you cannot get the ball/pitch diameter, these approximate shortcuts use only the ball count and shaft speed. They are rough — expect ±10–20 % error — so always prefer the geometry calculator above when a datasheet is available.

How It Works

A rolling-element bearing has four moving parts that each generate a characteristic repetition rate when a localised defect — a spall, pit or crack — is struck by a passing ball or roller. The shaft turns at fr = RPM/60 hertz (the inner race rotates while the outer race is held in the housing). From the geometry the four defect frequencies follow directly, where Nb is the ball count, Bd the ball diameter, Pd the pitch diameter and θ the contact angle:

  • BPFO — ball pass frequency, outer race = (Nb/2)·fr·(1 − (Bd/Pd)·cosθ)
  • BPFI — ball pass frequency, inner race = (Nb/2)·fr·(1 + (Bd/Pd)·cosθ)
  • BSF — ball spin frequency = (Pd/(2·Bd))·fr·(1 − ((Bd/Pd)·cosθ)²)
  • FTF — fundamental train (cage) frequency = (fr/2)·(1 − (Bd/Pd)·cosθ)

Each defect frequency comes with harmonics (2X, 3X …) because the impact is a sharp, repeating pulse rather than a pure sine wave, so it is rich in overtones. In practice you look for a peak at the calculated frequency and a train of harmonics above it. An inner-race fault (BPFI) and a ball fault (BSF) also produce sidebands spaced at the shaft speed (BPFI) or the cage rate (BSF), because the defect moves in and out of the load zone once per revolution — the sideband pattern is itself a clue to which element is damaged.

What each defect tends to sound and look like: an outer-race fault is the most stationary — a steady, periodic rumble or growl with a clean BPFO peak and strong harmonics, little sideband activity. An inner-race fault rises and falls once per shaft revolution as it rotates through the load zone, giving a BPFI peak flanked by 1X sidebands and an amplitude that “breathes.” A ball/roller fault produces BSF with FTF-spaced sidebands and can be intermittent as the element re-orients. A cage fault (FTF, always below 0.5X shaft speed) is usually a late-stage, low-frequency looseness that often shows up only as sidebands on the other peaks. These order→fault and sideband→defect associations are widely used diagnostic conventions, not guarantees — bearing slip, multiple defects, mounting and load all blur the picture.

Two honest cautions. First, the calculator assumes the geometry is exact and ideal; in reality the contact angle changes with load and rolling elements slip by a percent or two, so measured peaks land a little off the computed values — analysts search a small window around each frequency, not a single bin. Second, you must verify your inputs: get Nb, Bd, Pd and θ from the bearing datasheet (if you only know bore and OD, Pd ≈ (bore + OD)/2 is a reasonable estimate, Bd less so). We deliberately do not ship a part-number lookup table, because publishing invented SKF/NSK/FAG/Timken geometry would be worse than no data — the geometry calculator is the honest tool here.

Frequently Asked Questions

What are BPFO, BPFI, BSF and FTF?
They are the four bearing defect frequencies. BPFO (ball pass frequency, outer race) is how often rolling elements pass a point on the fixed outer race; BPFI is the same for the rotating inner race; BSF (ball spin frequency) is how fast each rolling element spins about its own axis; and FTF (fundamental train frequency) is the rotation rate of the cage that holds the elements. A defect on a given surface excites a peak at that surface’s frequency, plus harmonics.
Where do I find the ball count, ball diameter and pitch diameter?
From the bearing manufacturer’s datasheet or engineering drawing for the exact part number — that is the only reliable source. This tool intentionally does not include a part-number database, because publishing made-up geometry for real catalogue numbers would give confidently wrong answers. If you only know the bore and outside diameter, pitch diameter is roughly (bore + OD)/2, but ball diameter and exact ball count really do need the datasheet.
What is the no-geometry rule of thumb, and how accurate is it?
When the geometry is unknown you can approximate BPFO ≈ 0.4 × Nb × fr and BPFI ≈ 0.6 × Nb × fr, where fr = RPM/60. These come from the fact that (Bd/Pd)·cosθ sits near 0.2 for many common bearings. They are clearly labelled APPROXIMATE here and can be off by 10–20 %, so use them only to get into the right ballpark; the exact formulas with real geometry are always preferred.
Will a peak land exactly on the calculated frequency?
Not quite. The formulas assume pure rolling with no slip and a fixed contact angle, but real bearings slip by a percent or two and the contact angle shifts under load, so measured peaks fall in a small window near the computed value — typically within a few percent. Look for a peak near the frequency together with its harmonics and sidebands, rather than expecting a hit on a single exact bin.
Does a peak at a fault frequency prove the bearing is failing?
No. Matching a spectral peak to a defect frequency is a strong indicator and a standard diagnostic convention, but it is not a diagnosis on its own. Trends over time, the amplitude, the harmonic and sideband pattern, the envelope (demodulated) spectrum, load and temperature all matter. Confirm a suspected bearing fault with a calibrated accelerometer, envelope analysis and ideally a trained vibration analyst before acting.
Can I capture vibration with my phone or laptop microphone for this?
A microphone hears the airborne sound a machine radiates, which can reveal the frequency content to compare against these calculated values, but it is not a calibrated vibration measurement and gives no true amplitude in mm/s velocity or g acceleration. For real vibration measurement — and for any ISO 10816 severity assessment — you need an accelerometer mounted on the bearing housing and a proper analyzer.