Bearing defect frequencies & ISO severity
Enter a bearing designation and a shaft speed to get BPFO, BPFI, BSF and FTF. Enter a measured velocity to get its ISO 10816 / 20816 zone. Every geometry value is editable, because a 6205 is not the same bearing at every manufacturer.
Which frequency belongs to which fault.
Bearing & shaft
You measured 4.2 mm/s. Is that a problem?
Enter an overall velocity reading (RMS, 10–1000 Hz) and pick the machine class. The answer is a zone, and each zone maps to a decision — not just a colour.
Picking the right class
| Your machine | Use |
|---|---|
| Pumps, small motors, spindle drives up to ~20 hp | Class I |
| Motors and machines 20–100 hp on standard mounts | Class II |
| Large motors, compressors, mill drives on rigid concrete foundations | Class III |
| Turbines, turbo-compressors, large sets on flexible or spring mounts | Class IV |
| Gearboxes 15–300 kW, grouted to a rigid base | Group 2, rigid |
| Gearboxes and drives 15–300 kW on skids or soft mounts | Group 2, flexible |
| Mill drives, large gearboxes, sets above 300 kW | Group 1 |
A number is a snapshot. A trend is a decision.
2.0 rising is worse than 4.0 steady
A machine that has sat at 4.0 mm/s for two years is telling you less than one that moved from 1.8 to 2.6 in a month. Absolute limits catch the obvious; slope catches the developing.
A healthy total can hide a screaming BPFO
Overall velocity is broadband. An early outer-race defect can add very little to the total while being unmistakable in the envelope spectrum at the frequency you just calculated above.
Variable-speed spindles need live RPM
Every frequency on this page scales with shaft speed. On a spindle that runs 800–12,000 rev/min, defect bands move constantly — which is why we read live RPM from the controller rather than assuming it.
This is one reading. MachineWise watches these frequencies every minute.
You just did by hand what our vibration monitoring does continuously on every critical bearing — tri-axial sensing, live FFT with envelope analysis, per-machine learned baselines, and a plain-language alert naming the component and an estimated time to fault.
See vibration monitoring → Bearing life tracking