Sealing a pH probe through an enclosure wall without leaks

Short answer: take the probe through a machined round bore in a bolted-on boss, and seal it with a clamped elastomer. A 12 mm [0.47 in] combination-electrode barrel seals with a 12 x 2 EPDM O-ring squeezed 17.5% in a bore 0.2 mm oversize; a 2-3 mm round machine pin seals with a rubber bushing bored 15-20% under the pin. A square header pin buried in solid sealant never seals, because sealant only holds where it bonds to both surfaces or where you squeeze it. Order the bore instead of drilling it: the cutout designer puts the boss into the wall before the box is built, and the wall arrives rated NEMA 4X / IP66 by the manufacturer; not UL listed.

Why a square pin fails and a round barrel seals

A static seal against liquid holds in one of two ways. Either a thin adhesive layer grips both surfaces — epoxy on a clean, round pin — or an elastic ring is squeezed between them and pushes back. Solid potting compound does neither: it grips a square pin's corners at best, it shrinks away from the barrel as it cures, and it cannot follow the metal when the metal moves. That is why a pin cast in a blob weeps at the pin-to-blob line within weeks, while the same pin in a clamped rubber bore runs for years.

The movement behind that leak is calculable. Austenitic stainless steel expands about 16 ppm per kelvin; assume about 5 ppm per kelvin for the electrode glass (electrode glasses vary, and the exact figure changes the total, not the conclusion). Across a 50 K swing between a hot washdown and a cool night, a 12 mm steel bore and a 12 mm glass barrel move apart by (16 - 5) x 10⁻⁶ x 50 x 12 = 6.6 μm — six thousandths of a millimetre, every cycle. An EPDM ring stretches that far and comes back. A rigid epoxy collar cracks along the interface within a few dozen cycles, and the leak shows up later as fogging inside the box.

Treat this as a static liquid seal, not a weather seal: the wet side sits in the process fluid for as long as the probe is in service. That moves the material choice from weather to chemistry, which is why the fittings below are picked by fluid and temperature rather than by code alone.

Match the fitting to the probe

Pick the fitting by what actually passes through the wall. Four routes cover nearly every pH or conductivity probe:

Round pins are easy to buy once you know the name: they are sold as machine pins or instrument pins, and 2 mm is the common size — the tip that fits small multimeter sockets. A sealed panel-mount 2 mm socket, the little cousin of a 4 mm banana socket, then carries the probe pin from outside, with its own O-ring doing the sealing.

Every fitting you add has to carry the rating you bought, because an assembly sits at its lowest part's level. Read three lines of the datasheet before you pick: the IP code and the jet test behind it, the clamp or grip range in millimetres, and the printed hole size and torque. An IP68 gland on its own does not make an IP66 wall jet-tight: the fitting only carries the rating when its manufacturer publishes it as a Type 4X / IP66 accessory with a printed hole size and torque, and you bore and tighten to that print within its stated grip range. A drip-drained or louvered part sets the whole wall at its own level instead. The jet behind the second digit 6 runs 100 L/min from a 12.5 mm nozzle under IEC 60529; Type 4 under NEMA 250 is the hose-directed water case, so the fitting has to survive both on a washdown wall. The wall vocabulary — hole sizes, gland styles, display openings — lives in the cutouts, glands and windows guide, and the NEMA 4X vs IP66 guide explains what each of those codes actually tests.

What you need

• A boss machined from 316L stainless where the process carries brine, chlorides or chemicals, 304 for water, food and general washdown — the same grades as the wall; custom 304 and 316L enclosures take the boss as a drawing line.
• An EPDM O-ring, 12 x 2 (12 mm inside diameter, 2 mm cross-section), 70 Shore A, the compound O-ring catalogs stock as standard.
• An EPDM flange gasket, 2 mm thick, cut to the boss flange; flat face or raised lip per the gasket-face guide.
• Calipers reading to 0.02 mm, a hole saw, a deburring tool, a torque wrench, M4 stainless screws with washers, and feeler gauges.
• Soap solution and a hand pump with a gauge reading to 0.1 bar [1.5 psi] for the bubble test.
• Eye protection: the electrode bulb is thin glass and lets go without warning.

Cut the wall and bolt the boss down

1. Lay the boss on a flat face at least 20 mm [3/4 in] from any bend, weld seam or existing cutout, so the gasket lands on flat stainless — a rule of thumb that keeps the clamp pressure even.
2. Cut the wall hole to the boss drawing; for a boss with a 16 mm spigot, cut 16.3 mm and deburr both faces.
3. Offer the boss up from outside with its gasket, bolt it down from inside, snug the screws diagonally, then pull each M4 screw to about 2 N·m. Screws through an outside face carry water along their threads — the outdoor-screw sealing guide covers washers and thread sealers.
4. Check the flange gap with feeler gauges: a 2 mm gasket should read a 1.5 mm gap, its 25% squeeze. Write the number on the drawing so the next builder repeats it.

Sizing the gland for a 12 mm electrode barrel

Here is the arithmetic behind the headline numbers, for the common industrial electrode.

• Barrel: 12.0 mm — the common industrial combination-electrode size; caliper yours before you cut.
• O-ring: 12 x 2 EPDM, 70 Shore A.
• Bore: 12.20 mm — the barrel plus 0.20 mm of diametral clearance, a fit you can hold with a drill and reamer, leaving all the sealing to the ring.
• Groove, cut into the boss bore: 1.55 mm deep, 2.8 mm wide — about 0.78 and 1.4 times the ring's cross-section.
• Limits: 15-30% squeeze and a gland fill no higher than 85%, the values O-ring handbooks publish for static seals.

Squeeze is the share of the ring's cross-section you remove radially: half the gap between groove root and barrel, taken from the cross-section, divided by the cross-section. Fill is the ring's cross-sectional area over the groove's area.

• Groove root diameter: 12.20 + 2 x 1.55 = 15.30 mm.
• Radial gap at the root: (15.30 - 12.00) / 2 = 1.65 mm.
• Squeeze: (2.00 - 1.65) / 2.00 = 17.5%.
• Ring area: 0.785 x 2.00² = 3.14 mm². Groove area: 1.55 x 2.8 = 4.34 mm².
• Fill: 3.14 / 4.34 = 72%.

The result passes: 17.5% squeeze sits mid-band, and 72% fill leaves room under the 85% ceiling for grease, tolerances and thermal movement. Hold the bore at 12.20 ± 0.05 mm. At a 12.30 mm bore the gap is (15.40 - 12.00) / 2 = 1.70 mm, so squeeze falls to (2.00 - 1.70) / 2.00 = 15% — already the floor, where a ring weeps at the first hose-down. Above 30% the ring takes a permanent set and the gland opens within a few hot-cold cycles. If the bore drifts past +0.05 mm, step the ring's cross-section up rather than patching the fit.

Seat the probe and dress the cable

5. Wipe the groove clean, grease the ring lightly with the grease the drawing names, and roll it into the groove with your fingers.
6. Support the barrel in both hands and slide it through square; never side-load glass, which cracks long before the ring feels it.
7. Hand-start the gland collar to its face, then tighten to the drawing's torque. On glass-barrel electrodes the clamp lands on metal collars at each end; the glass only slides through.
8. Form the cable below the boss into a 50 mm [2 in] drip loop, so washdown water runs down the cable and drops clear of the seal.
9. Hold any cable gland leaving the box to the same three datasheet numbers — passing a connector through a cable gland without leaks covers that side in detail — and plug every spare bore: an open 16 mm hole sits below the wall until it is capped, and a misplaced one has a fix in the wrongly-drilled-hole guide.

Prove the wall before the first washdown

• Bubble test: cap the cable end and any breather, pressurize the box through a spare gland to 0.1 bar [1.5 psi], and brush soap over the boss, collar, screws and flange. A steady stream of bubbles marks the leak; a wall that holds clean for 5 min passes.
• Feeler gauges again after the final torque: 1.5 mm on a 2 mm gasket, so the 25% squeeze survived tightening.
• Wipe the boss bore dry, run your normal hose-down, and look inside within the hour: droplets at the bore came through the seal, because the box was dry going in.
• Re-run the bubble test after every probe swap. The gland is the part you disturb most, so it is the part that leaks first.

Mistakes that cost the rating

• A fitting below the wall sets the wall: one IP44-rated socket turns a Type 4X / IP66 assembly into an IP44 assembly, whatever the box itself passed.
• Rigid potting around any pin, square or round: the 6.6 μm per-cycle movement above opens the bond line within a season, and the first sign is fogging, not a drip.
• A drain or vent punched as a field fix: put drains on the drawing of a Type 3R design, where they belong; punched into a Type 4X wall, they take the rating with them.
• Chloride splashes above about 60 C (140 F): both 304 and 316L risk stress-corrosion cracking there, so the fix is a cooler location or a different material, not a thicker flange.
• Tightening the gland onto the glass: hand-tight plus the printed quarter-turn is enough, and an over-clamped bulb shows up as drift before it shows up as a leak.

The decision, case by case

• Serviceable combination electrode: bolted boss, 12 x 2 EPDM ring at 17.5% squeeze in a 12.20 mm bore, flange gasket held at 25%, drip loop below.
• Round machine pin you want to unplug: rubber bushing bored 15-20% under the pin — 0.3-0.4 mm on a 2 mm pin, 0.45-0.6 mm on a 3 mm pin — clamped metal to metal.
• Round pin, permanent: flexible adhesive annulus in a bore 0.3-0.5 mm oversize, with a spare probe on the shelf.
• Square header pin: swap it for machine pins first; the round surface is the whole trick.

The same bore-and-squeeze logic carries other devices too — a PCB push button and a USB pass-through seal by the same two rules of bond or squeeze. If you would rather receive the wall finished, send the barrel diameter, ring size and torque with the drawing through the custom enclosure builder, and the box arrives with the bore already cut, delivered in 10 business days from your approval of the drawing.

Catalog dimensions to compare

Compare these catalog dimensions with your component layout. Open each enclosure for its rating, material and usable space, then check those specifications against your application.

Questions