Semiconductor · test and burn-in floors

Half a channel width is the whole specification

A fab already filters harder than a data centre. What changes is the consequence: a particle excursion is a yield event, not a downtime event. The loop is many parallel small ones, and the sensitive item is the test socket.

Cold-plate channel 100–300µm

Finest absolute protection rating 0.45µm

Ask for the specification sheet

Burn-in bay with trays of loaded boards running into racks under a filtered ceiling

Burn-in bay · trays into racks, filtered ceiling

01 / Particle limit

Why half a channel width is the number

Above half the channel width, blockage probability at an entrance or a bend rises materially. A blocked microchannel loses local heat rejection, so the part runs slower before anything fails.

Particle size against channel width

Every micron figure carries its category.
Figure	Value	Category and basis
Cold-plate channel	100–300 µm	A channel dimension, not a rating.
Blockage threshold	>150 µm	Half the channel width, at entrances and bends.
Supply-chain specification	100 & 200 µm	A tightening specification band moving toward 50 µm and below. Not a filter rating.
Cold-plate cleanliness	<5 µm	A protection rating on the operating element.
Verification	VDA 19.1 · ISO 16232	Internal flushing extraction onto a 2 µm analysis membrane.

02 / Instrumentation

Panel-mounted conductivity controller on a fabrication instrument panel, sensor cable leaving the enclosure

The instrument has to span both loop types

A band without its loop type is not a target. The measurement range spans both, so one part number serves a de-ionised technology loop and a glycol or facility-water loop.

Item	Figure
Conductivity range	0.04–20 µS/cm at 0.01 cm⁻¹ · 10–200 µS/cm at 0.1 cm⁻¹
De-ionised technology loop	0.2–20 µS/cm at pH 8.0–9.5
Glycol or facility-water loop	50–300 µS/cm at pH 7.0–9.0
Nominal media	Quoted at 90 % retention
Absolute media	Reaches 99.98 % retention
Sampling and calibration	Laboratory panel every 50–100 days · calibration interval printed per instrument

03 / Detection

Where a drip is a contamination event

A mop is not acceptable on a gowned floor. Point probes sit in catch trays and rope runs on the manifolds. Either detecting raises an alarm, and both detecting triggers intervention.

Setpoint against contamination current

Point probes in catch trays

One probe per tray, 0 to 4.8 mm above the surface. Local, unambiguous.

Rope on the manifolds

Coverage along the run, so an unlisted joint is still supervised.

Thresholds agreed at commissioning

Set against the measured site noise floor, not accepted from a factory default.

Contamination current

1 to 300 µA. Flux or swarf crosses the 150 µA setpoint with no water present.

Response window

150 µA held 20 s separates a drip from a splash.

The cost of a false alarm

Every entry is gowned and logged, so a nuisance trip costs gowning time twice.

04 / Model codes

Pick the codes here

EMS-CF-F0.45-20 Pleated depth cartridge, 0.45 µm absolute protection rating, 20 in The rating a 100–300 µm channel asks for. PDF
EMS-CF-F1-10 Pleated depth cartridge, 1 µm absolute protection rating, 10 in Shorter housing for a burn-in bay manifold. PDF
EMS-CF-F5-10 Pleated depth cartridge, 5 µm nominal protection rating, 10 in A protection rating, not a commissioning staging step. PDF
EMS-CF-WQ-PC Inline particle counter, 3-channel, ISO 4406 The reading an acceptance sheet can name. PDF
EMS-CF-WQ-C Contacting conductivity cell, 0.01 and 0.1 cm⁻¹ Range spans both loop types on one instrument. PDF
EMS-CF-WQ-PH Process pH sensor, 1–12 pH Band printed with its loop type beside it. PDF
EMS-LD-P1 Surface point probe, 4.3 m leader For the catch tray under each test socket bank. PDF
EMS-LD-C8 8-channel distance-read leak panel One panel per probe-and-rope group. PDF
EMS-LD-R100 Locating sensing rope, 100 m Position resolution 0.1 m, stated separately from repeatability. PDF