Key takeaway: The counting method is part of the number — ask which method produced a count before you compare two of them. Match the method to the fluid: clear oil to a light-extinction counter, dark or viscous gear oil to a microscope, grease to per-mass size bands and never an ISO code. And remember that a low elemental result does not mean a clean bearing, because large wear particles are invisible to it.

Two labs, one oil, two counts — both right#

Send a split of the same oil to two competent laboratories and you can get back two different ISO 4406 cleanliness codes. Neither lab made a mistake. They measured the particles with different instruments, and those instruments size particles on different bases.

A light-extinction counter reports the diameter of a circle with the same cross-sectional area as the particle — a value in µm(c), tied to a calibration standard (ISO 4406:2021, Clause 4.1). A microscope reports the longest physical dimension of the particle (ISO 4407:2025, Scope). For a round particle those two numbers nearly agree. For an elongated sliver — say 25 µm long and a few µm wide — the microscope calls it "25 µm" while the counter, seeing only its modest cross-section, files it at roughly half that. Same particle, two honest answers.

That is the whole thesis of this article. A particle count is never just a number. It is a number produced by a specific instrument that physically sees only part of the particle population. The skill is not reading the code — it is choosing the counting method to match the fluid and the component, then reading every count as a method-qualified figure.

The instruments — what each one physically sees#

There are four counting classes in routine use, and each answers a slightly different question.

Light-extinction particle counter (the PAMAS-class laboratory workhorse). A narrow light beam passes through a thin stream of the sample; each particle reduces the intensity of the beam as it crosses the sensing volume (ISO 11500:2008, Clause 3.3), and from that signal the instrument derives the size of an equivalent particle based on its cross-sectional area, reported in µm(c) (ISO 4406:2021, Clause 4.1). Calibration against a certified reference standard makes "6 µm(c)" mean the same thing on any brand of instrument (ISO 11171:2016, Clause 6.8). It is fast, analyses the whole sample, and reports the three-number ISO 4406 code at 4 / 6 / 14 µm(c) (ISO 4406:2021, Clauses 4.4.2–4.4.4).

Optical microscopy on a membrane. A measured volume of fluid is vacuum-filtered onto a membrane, and each particle is imaged and sized by its longest dimension — though for gear and pitch-hydraulic oils TriboTech sizes on an area-equivalent basis to stay comparable with the counter (see the doctrine section below). The current edition defines both an automated image-analysis route and the legacy manual count (ISO 4407:2025, Scope). It reports a two-number code at 5 / 15 µm, written with a leading dash — --/X/Y (ISO 4406:2021, Clauses 4.5.2–4.5.4). Crucially, it works on dark and viscous fluids where the beam counter fails, and it shows morphology.

Direct-imaging (flow-imaging) counter. This class photographs each particle in flow and classifies both its size and its shape — five wear-mode classes plus soot — while flagging water droplets and air bubbles by their optics (ASTM D7596-24, Section 1.1). It is the bridge between count-only and morphology, recovering shape information without the hours-per-sample cost of manual microscopy. But it is a triage signal, not a diagnosis: the shape classifier is trained on lab-generated wear modes (ASTM D7596-24, Section 6.9).

Coarse-filter mesh counting for grease. Grease does not circulate, so a weighed aliquot is dispersed and filtered, and particles are counted per unit mass in size bands. Mechanistically it is microscopy — but with a different denominator: mass, not volume.

And a warning that spans the counters: a light-extinction counter counts anything that blocks light. Air bubbles, water droplets and gel produce the same signal as hard metal particles — a fluid interface obstructs the light beam and gives false signals (ISO 11500:2008, Clause 1, NOTE 2) — and in practice those false counts land in the small-size channels where droplets are most numerous. A water-containing sample should not be counted at all (ISO 11500:2008, Clause 6.5.4.3), and dark or viscous fluids force heavy dilution (ISO 11500:2008, Clause 7.2.1). This is exactly what a microscope and an imaging counter can screen out — because they see each object rather than infer it from a light blockage.

Method chosen per component — the doctrine#

At TriboTech the counting method is not arbitrary. It is a fixed routing with a stated rationale.

Gear oil is always microscope-counted, never by particle counter. Dark, viscous, heavily additised gear oil defeats light extinction; on the membrane, every counted particle is a confirmed solid and its morphology is read, not just tallied. That confidence is precisely why the tighter recommended limits for gear oil can be held.

There is a subtlety worth making explicit. A microscope's default sizing basis — longest dimension — runs coarser than a counter's area-equivalent µm(c) basis, which is exactly why the two methods disagree (above). TriboTech closes that gap on purpose: our automated microscope measures the full projected area of each imaged particle and sizes it as an equivalent-circle diameter — the same basis the light-extinction counter uses — so a gear or pitch-hydraulic count keeps the confirmed-solid confidence of the microscope while staying comparable to the µm(c) code scale the fleet trends on. It is genuinely best-of-both: we image the particle to be sure it is a solid, then size it the way the counter would.

High counts on pitch-hydraulic circuits in modern turbines are always verified by microscope before reporting. The counter is the fast fleet screen. Where it reads high, the microscope reads a lower, more stable number, because droplets, air and gel that the counter tallied as particles are excluded. We report the confirmed count.

The through-line: the darker, more additised or more static the fluid, the more the count must come from a confirmed, imaged particle rather than a light-blockage inference — gear oil at one extreme, clear hydraulic oil at the other.

Size bands are component physics#

The numbers on the axis are chosen, not generic.

For oils, the ISO 4406 code. The three-number code sits at 4 / 6 / 14 µm(c) for a counter, and the microscope code at 5 / 15 µm written --/X/Y (ISO 4406:2021, Clauses 4.4.2–4.4.4 and 4.5.2–4.5.4). Read a code like 18/16/13 straight off the scale: each step doubles the count range (ISO 4406:2021, Clause 4.3.2), so 18/16/13 is a normal, well-filtered distribution. The spacing between the three numbers is itself a distribution-shape signal — a flat code means coarse ingress, a steep one means the fluid is dominated by fines.

For grease, per-mass bands chosen by component. TriboTech's standard wind-turbine grease programme — a programme design decision, not a standard — trends a continuously-rotating main bearing on fine bands (>5, >15, >25, >50, >100 µm), because its earliest wear signal sits in the fine-to-medium range. An oscillating blade/pitch bearing is screened for large spall debris on coarse bands (>100, >200, >300, >400, >500 µm) on the more-loaded sample.

Grease adds one preparatory twist the oils do not need: the soap thickener clogs a filter membrane before any usable particle loading is captured, so TriboTech first lifts the ferromagnetic wear debris out of a weighed grease aliquot with neodymium magnets, then deposits it on a capture substrate — an 80 µm mesh for blade bearings, a 1.2 µm membrane for main bearings — chosen deliberately far finer than the smallest band we count (>100 µm and >5 µm respectively). Because the substrate holds far more than it reports, the shape of every retained particle is read in full.

One hard rule follows: per-gram grease counts never convert to an ISO 4406 code — the two rest on different denominators (mass versus volume) and different retention. And there are no published in-service grease particle-count limits; the bands support within-fleet trending and relative standing, never pass/fail against a standard.

Try it yourself: decode or build an ISO 4406 code with our ISO 4406 particle-count tool — enter counts at each threshold and read the code, or read a code back into its count ranges. And because every count is only as good as the sample behind it, see the physics behind correct oil sampling.

The elemental blind spot#

A particle count tells you how many and how big — not what the particles are. Elemental analysis answers the composition question, but it has its own ceiling. ICP gives low results for particles larger than a few micrometres (ASTM D5185-26, Clause 1.4); wind-gearbox guidance states the limit crisply as an eight-micron cut-off for elemental analysis (AWEA RP 108, p. 7). So a bearing shedding coarse spall can read near-zero ICP iron while it is actively failing. The complement is ferrous-debris magnetometry (ASTM D8120-25), which is size-insensitive and catches the large-particle fraction ICP misses — the reason counting and elemental methods are complementary, not redundant. The full transport-physics and grease-digestion story lives in our ICP guide.

Read every count as a method-qualified number#

The advisory skill is not memorising a threshold table. It is asking which instrument produced a count, choosing the method that matches the fluid and the component, and reading the result as the method-qualified figure it always was. Get that right and two labs disagreeing stops being a puzzle and becomes information.

If you commission cleanliness reports across a fleet and want the counting method matched to each fluid — and read correctly against the right baseline — get in touch for a cleanliness-programme review, or explore our diagnostic tools.