P0 vs P6 vs P5 vs P4 Bearings: What the Step-Up Actually Buys

Stop buying a class to fix noise

The precision class is a tolerance-band specification. It covers boundary dimensions and running accuracy — radial and axial runout of the rings, perpendicularity of the inner ring face and outer ring outside surface, and thrust raceway thickness variation^11,15.

That is the whole scope. The same sources state plainly that the class does not control noise and vibration levels, lubrication, ball grade, radial internal clearance, surface finish, material, heat treatment, cage quality, speed ratings or load capacity¹⁴.

So when a pump running at 2,500 rpm is noisy and the engineer steps P0 to P5, the runout band tightens and the noise does not move. The premium is spent on a dimension the symptom never touched.

This is the most common way money leaves a bearing budget: the complaint is acoustic or thermal, and the lever pulled is dimensional. Before you write a tighter class on the drawing, name the symptom in the units it is measured in.

If it is decibels, acceleration or temperature, the class is the wrong lever and the RFQ should be rewritten around lubrication, ball grade, cage and fit instead. If it is runout of the rotating assembly, the class is exactly the right lever.

Ask your supplier which of the two your symptom belongs to before the drawing is released.

Does your duty sit in a band that pays?

The published mapping ties class to a speed band and a vibration or runout requirement, and that mapping is the only decision rule these sources support¹⁸. Below roughly 3,000 rpm the tolerances are described as irrelevant, and ABEC1 is stated to serve about 90% of industrial applications adequately^14,18 (supplier-reported figure).

A CNC spindle at 18,000 rpm needing low vibration sits in the 10,000-30,000 rpm band mapped to ABEC5/P5, which is where the step-up starts to pay¹⁸. Read the table as a filter, not a ladder: if your duty is not in a band, the tighter class has nothing to exploit.

Ask your supplier to confirm the band before you commit.

A tighter class does not raise speed

Limiting speed is a separate catalogue parameter. It is published separately for oil and grease lubrication, and it is not set by the tolerance class^12,14.

GMN catalogues limiting speed separately for oil and grease lubrication, which is the shape of the data you should expect from any precision supplier¹². This matters because the speed bands in the previous section can be misread as a class property.

They are not. The band tells you which class is typically fitted to a duty at that speed; it does not tell you the bearing will survive that speed.

A motor specified P5 instead of P0 for a 12,000 rpm duty still has the same limiting speed as the P0 bearing of that size, because the rating is a lubrication and geometry figure published on its own line of the catalogue. The class letter changes the tolerance band and nothing on that line.

So the RFQ needs two separate numbers from the supplier: the class, and the limiting speed for the lubrication you will actually run. If the supplier quotes only the class, you have not yet been told whether the bearing can turn at your duty.

Ask for the oil and grease limiting speed figures in writing before you accept the quote.

Tight classes often arrive as angular contact

The bearing type that typically carries a tight class is angular contact, and that type brings its own mounting bill. Angular contact bearings require paired installation and a defined axial preload — often 1-2% of the dynamic load rating — and the sources state this demands specialised expertise^1,22,24 (supplier-reported figure).

Deep groove bearings are simpler to install and have less stringent shaft and housing fit requirements²². The misalignment tolerance is where the gap is widest: angular contact tolerates under 2 minutes of arc, against 0.5-1 degree for deep groove and up to 3 degrees for self-aligning types⁷ (supplier-reported).

A machine tool spindle specified P4 arrives as a matched angular contact pair needing preload around 1-2% of dynamic load rating and alignment inside 2 minutes of arc^7,24 (supplier-reported). That is a mounting and shaft-preparation cost, not a bearing cost, and it lands on your assembly line rather than the supplier's invoice.

Budget for it before the class goes on the drawing. If your team cannot hold that alignment, the tight class will be wasted by the installation, and you should ask the supplier what mounting precision the pair actually needs.

What paperwork will you actually get?

Only the highest class is stated to be individually serialized and sold with inspection certificates showing measured rather than class values²¹. No equivalent documentation is stated for ABEC5, and none is stated for P6, P5 or P4.

That gap matters because a class marking certifies a band, not a measurement: if you need the actual bore or runout value for your assembly record, the class letter will not supply it. Write the documentation into the purchase specification, or you will be paying a premium for a class whose measured values never arrive.

• ✓Ask whether the ordered class ships with a certificate showing measured bore and runout values, or only a class marking.
• ✓Confirm whether the bearings are individually serialized, since that is stated only for the highest class.
• ✓State the measured values you need on the purchase order, not just the class letter.
• ✓Ask what traceability the supplier provides for P6, P5 and P4, since no source states a scope for these.
• ✓Decide whether your assembly record needs measured values at all before paying for documentation.

Where the cost curve actually bends

The cost curve is not linear, and the sources describe its shape rather than its numbers. ABEC3 is described as a marginal improvement over ABEC1 at a small cost premium, while higher tolerance classes and matched duplex sets require specialised grinding and are significantly more expensive^16,18.

The first step up is cheap; the later ones are not. The market data supports the same reading from the demand side.

Automotive accounts for 42.5% of ABEC-rated bearing revenue, yet most automotive bearings stay at ABEC3 rather than stepping to ABEC5 or ABEC7¹⁸ (supplier-reported). A high-revenue sector looked at the further step and did not take it.

Meanwhile ABEC5 holds 49.7% of ABEC-rated revenue, and the same source states most engineers over-specify by at least one class¹¹ (supplier-reported). Revenue share is not a recommendation: ABEC5 dominates because high-value industries concentrate there, not because it is the correct default¹¹.

So treat the step from P0 to P6 as the cheap one to justify, and every step above it as one that needs a named speed, runout or vibration requirement behind it. Ask your supplier to quote P0 and P6 side by side before you consider P5.

Pull the numbers before you release the RFQ

The class tables, limiting speeds and fit calculations live in the manufacturer catalogue, not in the class designation. An RFQ that names only P5 leaves the supplier free to quote a bearing whose speed rating and fit do not match the duty, and you will not see the mismatch until the drawing is fixed.

Pull the published figures first, then write the class.

• ✓Obtain the supplier's tolerance table for the exact class you intend to order, since P6, P5 and P4 are published as distinct classes.
• ✓Get the limiting speed for your lubrication type, oil or grease, for the specific bearing number.
• ✓Run the shaft and housing tolerance calculation for the fit you plan, rather than assuming the class sets it.
• ✓Ask for the preload figure if the bearing is angular contact, and confirm your assembly can hold it.
• ✓Confirm the supplier's installation guidance covers the mounting precision your duty needs.

Where the sources disagree

These are not errors to resolve — they are the honest edges of the evidence. One source warns that most engineers over-specify by at least one class, while another calls ABEC5 the sweet spot for CNC spindles at 10,000-30,000 rpm^11,18.

Both can be true: the sweet spot exists for that duty, and buyers outside it over-specify. The same split runs through noise and speed.

One source states the class governs running accuracy including runout, while another states it explicitly does not control noise, vibration or speed ratings^14,15. Read together, the class controls the runout band and nothing acoustic or thermal.

Treat every disputed figure as a question for the supplier, not a settled fact. Ask which side applies to your duty before you commit the class to the drawing.

What the sources do not establish

• No source reproduces the ISO 492 / ISO 199 numeric tolerance limits for P0, P6, P5 and P4 — pull these from the manufacturer catalogue
• No source gives a numeric limiting speed per precision class for a specific bearing — verify from catalogue data
• No source gives a numeric runout, vibration or noise value per precision class
• No source gives a price multiplier, absolute price difference, MOQ or price-break structure per precision class
• No source states stock status, made-to-order status or delivery lead times per precision class
• No source states what certificates or measured values are provided for P6, P5 or P4 — only the highest class is stated to ship with measured values
• No source states whether P6, P5 and P4 are dimensionally interchangeable with P0 for the same bearing number, or whether the class itself changes shaft/housing fit