Superfinishing Bearing Races: Ra Targets Without Rework

Ra progression through superfinishing

What Ra means on a raceway — and what it doesn't

Ra is the number bearing makers quote, and it is also the number most likely to fool them. Ra averages the profile: a raceway with three 0.5 μm furrows and an otherwise mirror finish can still report 0.05 μm Ra — and will still die early in service. That is why serious bearing lines control the full parameter set: Ra, Rz, Rpk, Rvk and the material-ratio curve. Ra sets the headline; Rpk — reduced peak height — decides whether the elastohydrodynamic oil film survives. On a typical precision bearing the EHL film runs 0.05-0.5 μm thick; a single 0.3 μm peak pierces it, and every metal-to-metal contact event starts a micro-pit that becomes a spall. Halving the finishing Ra typically buys 10-30 % on rated fatigue life, which is why the superfinishing station — not the grinder — is where bearing life is earned.

Superfinishing itself is a stone-plus-abrasive process: an oscillating stone rides the race on a thin slurry film, cutting only the peaks while the valleys stay untouched. The result is a plateaued profile — Rpk collapses toward zero, Rvk holds oil, and the bearing ratio climbs past 90 % at a 2 × Rz depth. A ground race at Ra 0.4-0.8 μm enters the station; a chromium-oxide-finished race at Ra 0.02-0.05 μm leaves it, typically in a single 10-40 second pass. Every extra pass is rework, and rework on a race that has already consumed 90 % of its machining cost is the arithmetic of margin destruction.

The role of the abrasive inside this physics: the slurry does the cutting; the stone only carries it. If the slurry's cutting depth per grain varies lot to lot, the plateau shifts and the pass count grows. The abrasive specification is therefore a surface specification written in powder terms.

One more parameter deserves a line on the chart: the λ ratio, oil-film thickness over composite roughness. Bearing engineers design for λ ≥ 3 in the running band; below 1, boundary lubrication and wear accelerate. Ra 0.02 μm instead of 0.08 μm is frequently the difference between a λ of 1.5 and 4 — same bearing geometry, same oil, double the safety margin. Roughness is load capacity you can buy at the finishing station.

The parameter set through the finishing sequence

Micro-fracturing: the abrasive that sharpens itself

Every abrasive cuts by one of two mechanisms, and the choice between them decides how a superfinishing line behaves through a shift. Blocky grains — silicon carbide, fused alumina — cut until they round over; the edge dulls, the grain glazes, and the stone starts burnishing instead of cutting. Removal rate decays through the batch, Ra drifts upward, and the operator compensates with pressure until the stone loads and the pass count climbs. Friable grains do the opposite. Chromium oxide green at Mohs 8.5-9 micro-fractures under the 1-3 kg/cm² pressure of a superfinishing stone: each grain sheds fragments and continuously exposes fresh 0.5-2 μm cutting edges. The abrasive sharpens itself, which is why chromium-oxide lines hold flat removal rates across an entire campaign while alumina lines sag.

The self-limiting depth of cut is the second gift of friability. A freshly fractured edge cuts shallow by definition; before a groove can deepen into a scratch, the grain fragments again. That is the physical reason chromium oxide delivers Ra 0.02-0.05 μm without the single deep scratches coarse-tail powders leave behind. Micro-fracturing also keeps the stone open: the fine debris acts as a mild lapping medium, preventing the loading that turns stones into burnishers.

On the line, the mechanics translate into three measurable behaviours. First, the Ra chart stays flat across a three-shift week at constant pressure — no periodic pressure increases chasing a dying stone. Second, stock removal holds within ±10 % through a batch, so cycle times stay valid and the line keeps its takt. Third, stone consumption falls, because the stone's job is reduced to carrying the slurry, not generating the cut.

Practical limits: micro-fracturing trades peak removal rate for surface quality. Chromium oxide is not a stock-removal brute; it is the final step. Pair it with a proper grind — Ra 0.4-0.8 μm in — and it removes the last 0.5-1 μm of peaks in one pass; ask it to erase a bad grind and it will do the job slowly while the line waits. An abrasive is only as good as the surface it inherits.

Running the line on spec powder

A superfinishing station buys three things from its abrasive supplier: the powder, the certificate and the repeatability. The first two are commodities; the third is the product. A lot-to-lot D50 shift of 1 μm in a commodity powder changes the cut rate and moves the Ra chart; on a line tuned to 0.02 μm the result is either rework passes or a finished race that measures out of band. GREEN 1025 removes the variable: every lot ships with D10/D50/D90/D99 laser-diffraction data and ≤0.8 % batch variance, SPC-controlled, so the machine settings outlive the pallet.

Incoming QC on a professional line is a 30-minute screen, and it should be treated as non-negotiable: wet PSD against the certificate, an oversize pad test through a 5 μm membrane, and a visual contamination check. Lots that pass go to the tank farm; lots that fail never see a machine. The cost of the screen is trivial against the cost of discovering a bad lot through the Ra chart four hours later.

Slurry management is where lines win or lose quietly. Water-based superfinishing slurry runs at 5-10 % solids with concentration checked twice per shift by density; pH sits at 7-9, and a biocide belongs in every tank because bacteria eat dispersants and flocculate the oxide overnight. Filtration runs 10 μm down to 2 μm in the recirculation loop, and the whole tank is dumped and recharged every 40-80 machine hours. Oil-based slurries skip the biocide but demand temperature control at 20-30 °C for viscosity stability.

The discipline that pays for everything else: one abrasive family per line. A single silicon-carbide grain carried over from a prep station becomes a scratch generator on chromium-oxide races. Dedicated tanks, hoses and stones, clearly labelled — the cheapest insurance a bearing plant buys. When the Ra chart eventually drifts, and it will, the certified powder lets the engineer rule out the abrasive in one glance and find the real cause: stone wear, pressure drift or dilution. That is what predictable means in practice.

Two line practices separate the plants that hold 0.02 μm Ra from the ones that only visit it. First, measure the slurry, not just the part: a density check and a pH strip twice per shift cost minutes and catch the dilution and bacterial drift that silently change cut rates. Second, log stone pressure against the Ra chart — when pressure has to climb to hold the number, the stone or the slurry has degraded, and the fix is cheaper today than tomorrow. Certified powder makes both signals readable, because the abrasive is the one variable that never changes.

On stone selection, pair the oxide with the right stone for the pass: a soft, fine-grained stone carries the slurry well and wears into the race profile, while a hard stone concentrates the same load on fewer grains and deepens their cut. The abrasive supplies the finish; the stone supplies the contact.

FAQ

What Ra can GREEN 1025 hold on steel races? 0.02-0.05 μm on hardened bearing steel (58-62 HRC); on ceramic and nitride races 0.01-0.03 μm is routine. Water or oil slurry? Water-based at 5-10 % solids is the default for steel races with filtration; oil where corrosion control or long queue times before washing matter. Both ship as pre-dispersed liquid. How is GREEN 1025 different from commodity polishing powder? Double classification removes the coarse tail, D99 is certified per lot, and batch variance is ≤0.8 % — commodity powders typically drift 5-10 % between lots. How long does a slurry charge last? 40-80 machine hours with filtration and concentration control; watch pH and bacterial growth in water-based tanks. Can it remove deep grinding marks in one pass? Normal grinding marks (Ra 0.4-0.8 μm) clear in one pass. A single stray deep scratch from a damaged wheel is not a powder problem — fix the wheel, and give the race the stock the scratch needs.