A bearing’s quality is invisible until it is measured. Two bearings can carry the same number, look identical and cost the same, yet one runs quiet for years while the other vibrates and fails early, because the difference lies in micrometres of roundness, runout and width variation. Incoming inspection therefore does not stop at whether the part fits. It checks the deviations that ISO 492 defines, so that a substitute part, a worn gauge or a counterfeit can be caught before it reaches the machine.
Why micrometres of difference decide life
The catalogue numbers that sell a bearing - dynamic load rating, limiting speed - are properties of the design. The micrometres that decide whether a particular unit behaves like its design are properties of the individual part, and they are exactly what the tolerance classes govern. A ring whose bore varies by a few micrometres changes the interference and therefore the running clearance, as the shaft and housing fits assume; a ring whose raceway runs out by a few micrometres adds a once-per-revolution forcing function. Neither shows up on a visual inspection, and both show up on a gauge.
The ISO 492 feature set
ISO 492 sets out the tolerance classes - Normal, P6, P5, P4 and P2 - and, for each, the permitted deviations of the critical features. The precision classes are defined entirely by these numbers, so measuring them is the only way to confirm which class a part actually is.
| Feature | Symbol | Reference condition | What a bad reading reveals |
|---|---|---|---|
| Mean bore deviation | Δdmp | 20 °C, light contact | Fit wrong; interference too tight or too loose |
| Bore variation | Vdp | Several angular positions | Ovality or a three-lobed bore |
| Mean outside diameter deviation | ΔDmp | 20 °C, multi-position | Housing fit wrong |
| Inner ring width deviation | ΔBs | Comparator stand with master | Matched-set mismatch |
| Inner ring radial runout | Kia | Ring rotated against indicator | Vibration and noise source |
| Outer ring radial runout | Kea | Ring rotated against indicator | Housing-borne vibration |
The symbols are the shared language of every maker’s tolerance table; a class is only a set of limits on these quantities.
Bore: mean value plus variation
Bore is not a single diameter. The ring is measured at several angular positions, the readings are averaged to give the mean bore deviation, and their spread gives the ovality. Because a thin ring can be distorted by the gauge itself, bore gauges apply a controlled, light contact, and comparative instruments are set with a master ring of known size.
Setting a bore gauge with a master ring
A master ring of certified size sets the zero. Every subsequent reading is a comparison against that master, so a worn or mislabelled master silently offsets every measurement taken with it. The master should be checked against its own certificate on the same schedule as the gauge.
Two-point versus three-point measurement
A plain two-point micrometer can miss a three-lobed form, which reads the same across every diameter yet is out of round. Multi-position or three-point methods catch what a single diameter cannot, which is why bearing inspection never rests on one reading.
Outside diameter and the three-lobe trap
The outside diameter is measured the same way, at multiple positions, to obtain a mean value and a variation figure. For a 6205 the outside diameter must measure within the class tolerance of 52 mm, against the ISO 15 nominal, and both the mean and the variation matter. A plain micrometer across one diameter can accept a lobed ring that a multi-position check would reject, so a single diameter is never sufficient evidence.
Width and the matched-set problem
Width tolerance is generous compared with bore and runout - tens of micrometres at Normal class - but it becomes critical in matched sets, where two or more bearings must share the load. That is why matched pairs are ground and measured as a set and should never be mixed with unmatched bearings of the same number. The width deviation ΔBs is the number that decides whether two parts can be paired.
Reading radial and axial runout
Radial runout is measured by rotating one ring relative to the other and reading a dial indicator against the raceway. It is the index of how far the ring’s centre wanders as it turns, and it is the quantity that most affects vibration and noise, so it is the first thing a defect-frequency analysis will point back to. Axial runout reads the ring face and governs face-located and thrust-loaded arrangements. Higher tolerance classes halve these values step by step, which is precisely what a precision spindle pays for.
Temperature, force and the written procedure
Measurements are referenced to 20 °C, because steel size changes with temperature, and the instrument, the master and the bearing must all be at that temperature to compare fairly. Magnetic chucks and heavy clamps distort thin rings and must be avoided. The measuring force, the number and position of readings, and the cleanliness of the surfaces all change the number, so a written procedure matters as much as the instrument.
The reference temperature
A ring measured warm is larger than the same ring measured cold. Because the difference is a few micrometres per ten degrees, temperature control is not a formality at precision classes, and a part straight out of a hot wash should be allowed to settle before it is gauged.
Gauge calibration is part of the method
A bore gauge or comparator that has been dropped, or a master that has worn, gives confident but wrong readings. Checking the gauge against its master before a batch, and keeping calibration current, decides whether the rest of the inspection means anything at all.
Incoming checks that catch a counterfeit
Counterfeit bearings rarely fail on boundary dimensions, which are easy to copy. They fail on internal geometry, steel quality, heat treatment and roundness. That is why a check that goes beyond the basic fit is valuable: measuring variation and runout on a suspect part often reveals a product that is the right size but the wrong bearing. Roundness, width consistency and runout are much harder to fake than a 25 mm bore, and a recorded series of measurements over months turns a subjective judgement about a supplier into evidence. The same discipline underpins the checks described in the counterfeit risk review.
| Check on receipt | Instrument | What it decides |
|---|---|---|
| Bore and outside diameter | Bore gauge, micrometer | Whether the fit is per class |
| Ovality at several positions | Two- and three-point gauges | Roundness, not just size |
| Width | Comparator stand | Usability in matched sets |
| Radial and axial runout | Dial indicator on a rotating ring | Vibration quality |
| Cleanliness and markings | Visual, side-by-side comparison | Traceability and anti-counterfeit |
Records are part of the method. A single measurement tells you about one part; a series over months tells you about a supplier. Keeping the results of incoming checks, together with the maker, batch and date, lets a plant notice a drift in quality before it becomes a failure on the line, and gives the buyer evidence when a shipment is not what was ordered. The relevant class limits are in the technical library, the terms are defined in the glossary, and the practical acceptance route runs through the deep groove and angular contact families, where a supplier such as FAG publishes the tolerance tables to compare against.
A short incoming-inspection procedure
A written procedure removes the guesswork and makes results comparable between operators and over time. The steps below suit a normal industrial receipt; precision parts add the runout and temperature controls already described.
- Condition the part. Let the bearing and the master settle to 20 °C before any reading, and keep magnetic chucks away from thin rings.
- Zero the gauge. Set the bore and outside-diameter instruments against their masters and record which master was used.
- Measure the bore. Take at least 3 readings at roughly 120 degrees apart and average them; the spread is the ovality.
- Measure the outside diameter. Repeat the multi-position routine at 3 or more positions to catch a lobed ring.
- Check the width. Use a comparator stand, especially where the part is destined for a matched set.
- Check runout on critical parts. Rotate each ring against a dial indicator and compare the reading against the class limit for that bore.
- Record and trend. File the readings with the maker, batch and date so a drift across months becomes visible.
Take a worked case: a 6205 with a 25 mm bore and a 52 mm outside diameter, received in a batch of 40. Measuring 3 bore readings and 3 outside-diameter readings on a sample of 5 parts takes a few minutes and yields 30 numbers to file against the class limits. The point is not the individual figures but the pattern they build. A single part that measures at the edge of its tolerance is unremarkable; a whole shipment that sits at the same edge, batch after batch, is a signal that the process or the gauge has drifted, and it is the trend that turns inspection into procurement evidence.
Nothing in this procedure is expensive. A calibrated bore gauge, a micrometer, a comparator stand and a dial indicator cover every check above, and the discipline that makes them useful is simply consistency - the same positions, the same temperature, the same records, every time. A plant that measures this way can defend a rejection, compare two suppliers on facts rather than impressions, and spot a counterfeit long before it reaches a spindle.
- ISO 492 defines every class. Confirming a class means measuring its features.
- Bore is a mean plus a variation. One reading is never enough.
- Runout drives vibration. It is measured with the ring rotated against an indicator.
- 20 °C and controlled force. Temperature and gauge care change the number.
- Measuring catches fakes. Roundness and runout are far harder to counterfeit than a bore.
Frequently asked questions
What temperature should bearing measurements be taken at?
All dimensional checks are referenced to 20 °C, and the instrument, the master and the bearing must all be at that temperature. Steel grows and shrinks a few micrometres per ten degrees, which is enough to shift a precision-class reading, so a part straight out of a wash should be allowed to settle first.
Why is one micrometer reading not enough to accept a bore?
A single diameter cannot reveal a three-lobed form, which measures the same across every diameter yet is out of round. The mean bore deviation and the bore variation are separate quantities, so the ring is measured at several angular positions and both the average and the spread are judged.
How can dimensional checking expose a counterfeit bearing?
Counterfeits copy the boundary dimensions easily but rarely match the internal geometry, steel quality and roundness. Measuring variation, width consistency and runout often reveals a part that is the right size but the wrong bearing, which a basic fit check alone would pass.
Should I check runout on every incoming shipment?
Runout matters most for precision, high-speed and vibration-sensitive positions, where a check on a sample from every batch is worthwhile. For ordinary industrial stock, checking bore, outside diameter and ovality on a sample, with runout reserved for suspect or critical parts, keeps the effort proportional to the risk.
What makes a matched pair different from two identical bearings?
A matched pair is ground and measured as a set so the two parts share load evenly, which depends on width deviation and preload as well as class. Two bearings of the same number and class but from different batches can differ in width, so a set should never be mixed with loose stock of the same number.
How often should a bore gauge be checked against its master?
Before a batch, and on the calibration schedule for the instrument. A dropped gauge or a worn master gives confident but wrong readings and quietly removes the value of every measurement taken with it, so the check is part of the method rather than an optional final step.