Oil Seal, O-Ring & Sealing Guides

The short answer: lay the seal on a flat surface and check which way the lip points, whether the outer face is flat or has a heel, and measure three numbers — rod diameter, groove outside diameter and groove width. Those four observations narrow almost any rod seal to one profile family.

1. Which way does the lip point?

A rod seal always points its sealing lip towards pressure, which is towards the inside of the cylinder. A wiper points outward, towards the atmosphere. If the part has two lips of unequal length with the long one facing the cylinder, you are holding a U-cup rod seal. If one lip is stubby and sits in a metal case, it is a wiper or a double-acting scraper.

2. Flat back or stepped heel?

A plain U-cup has a flat back that sits against the groove wall. A stepped or notched heel usually means the seal was designed to take a back-up ring, which in turn tells you the cylinder runs above roughly 160 bar or has a wide extrusion gap. Replacing a stepped seal with a plain one is the most common cause of a rebuild that leaks again within weeks.

3. What is it made of?

Polyurethane is amber to deep red, noticeably stiff, and rebounds slowly when you press a fingernail into it. Nitrile is matt black and softer. PTFE is white or bronze-filled, does not rebound at all, and is almost always paired with an energising O-ring behind it. Fabric-reinforced chevron sets show visible weave on the cut edge.

4. The three dimensions

Measure the groove in the gland, not the old seal. A seal that has run for years has taken a compression set and will read two to five per cent under its true size, and on a worn part the lip may have lost a further half millimetre to abrasion.

When the part is destroyed

If the seal came out in pieces, measure the gland instead: rod diameter with a micrometer, bore of the seal housing with an internal gauge, and groove width with a depth gauge or a strip of feeler stock. Send those three numbers and a photograph of the cylinder head, and we can normally identify the profile without the sample at all.

The short answer: decide on the fluid first, then the temperature, then the pressure. Getting the order wrong is how an O-ring that meets the temperature spec dissolves in the first week.

Start with the medium

Then the temperature band

Standard nitrile is comfortable from about -30 to 100 degrees C, and will harden and crack above that. Viton runs to around 200 degrees C and is the usual answer near engine exhausts and process heat. Silicone spans roughly -55 to 200 degrees C but has poor tear strength, which is why it belongs in static joints rather than on a moving rod. EPDM is good to about 150 degrees C and handles steam that would destroy nitrile in days.

Pressure and the extrusion gap

Above roughly 100 bar, compound hardness starts to matter more than compound type. A 70 Shore A ring in a groove with a generous clearance will nibble — you will see a ragged crescent on the low-pressure side. The fix is either a harder 90 Shore ring, a back-up ring, or a tighter gland. Reaching for a more exotic compound will not help, because extrusion is a geometry problem wearing a chemistry costume.

Reading a failed ring

• Flattened, shiny, no rebound: compression set — too hot for the compound, or over-squeezed.
• Swollen and soft: chemical attack — wrong compound for the fluid.
• Radial cracks, brittle: thermal ageing or ozone.
• Ragged edge one side only: extrusion — hardness or clearance.
• Spiral cut around the ring: the ring rolled in its groove — it wanted a quad ring.

That last one is worth acting on. An X-profile quad ring resists rolling because it seals on four points rather than two, and it is a straight swap into most O-ring grooves.

The short answer: a correctly chosen seal fitted into a damaged, dirty or misaligned gland will fail again. Check the rod, the gland surface, the back-up ring, the installation method and the breather before you blame the kit.

1. The rod is scored

Run a fingernail along the rod. If it catches, the chrome is broken and every stroke drags that edge across a fresh lip. A score deep enough to feel will destroy a new rod seal in weeks. The rod needs polishing, re-chroming or replacing — there is no seal that survives it.

2. The gland or groove is damaged

Burrs on the groove edge, corrosion pitting in the bore, or a sharp lead-in chamfer will cut the seal during assembly. The cut is often invisible until the cylinder is pressurised. Deburr, clean and chamfer before anything goes in.

3. The back-up ring was left out or fitted backwards

Back-up rings go on the low-pressure side, always. Fitted on the wrong side they do nothing, and the seal extrudes exactly as it would have with no back-up at all. In a double-acting gland there are two, and both face outwards from the seal.

4. The seal was stretched cold

Polyurethane and PTFE do not enjoy being forced over a rod at ambient temperature. Warm the seal in clean oil at around 60 degrees C, fit it with a smooth plastic tool, and let it recover its shape for a few minutes before the gland goes together. A seal installed with a screwdriver has a nick in it, whether or not you can see it.

5. The reservoir cannot breathe

A blocked breather puts the whole return side under pressure it was never designed for, and the rod seal is the weakest point in that circuit. If several cylinders on one machine start weeping at once, check the breather before you order five kits.

Before you reassemble

• Every component washed in clean solvent and dried — no lint, no grit.
• All grooves measured, not eyeballed.
• Seal orientation confirmed against the old parts, laid out in removal order.
• First cycle run at low pressure, no load, full stroke, several times.

The short answer: the mounted system only isolates when the disturbing frequency is at least 1.4 times its natural frequency, and it works properly at around three times. Below that ratio the table amplifies vibration instead of reducing it.

Find the disturbance first

In a typical Indian industrial building the traffic is predictable: air handling at 8 to 20 Hz, rotating machinery at the shaft speed and its harmonics, footfall around 2 Hz, and passing traffic in a broad band below 10 Hz. Pick the lowest significant frequency, because that is the one that dictates the design.

Then the natural frequency

The natural frequency of an isolated mass falls as static deflection rises. More deflection under load means better isolation, which is why a stiff pad under a heavy instrument does almost nothing.

To get below about 5 Hz you need an air spring. Rubber mounts run out of useful deflection before they get there.

Load distribution matters more than total load

A coordinate measuring machine or a microscope rarely loads its four corners evenly. If one mount carries twice its neighbour's share, that corner sits at a different natural frequency and the whole table rocks rather than heaves. Weigh the corners, or select mounts individually, and keep the centre of gravity as low and as central as the instrument allows.

The things that undo a good table

• A rigid conduit, cable or air line bridging the tabletop to the wall — a short circuit around the isolators.
• Levelling feet screwed down onto the floor after installation, bypassing the mounts entirely.
• A tabletop too light for the instrument, letting it rock on its own mounts.
• Isolators selected for the table's weight rather than the loaded weight.

The short answer: choose a keyless locking device when the drive reverses, when it shock-loads, when the hub position needs adjusting, or when the shaft is already close to its fatigue limit. Keep the key for steady unidirectional drives where cost dominates.

What a keyway costs you

Cutting a keyway removes roughly 10 per cent of a shaft's section but a good deal more than 10 per cent of its fatigue strength, because the corner of the slot is a stress raiser sitting in the zone of maximum torsional stress. Under reversing torque the key also works in its slot, and the resulting fretting widens the keyway until there is measurable backlash.

How a shrink disc works

Tightening the bolts on the outer rings drives them along matching tapers, which contract the inner ring onto the hub and clamp it to the shaft by friction alone. Torque transfers around the full circumference, there is no backlash, and the connection can be released and repositioned. A locking assembly does the same job from inside the hub bore rather than outside it.

Where each one belongs

Fitting them properly

Three rules decide whether the connection holds. Mating surfaces must be clean and dry — oil on the shaft can cut transmissible torque by half, and anti-seize is not lubricant here, it is a failure. Bolts are tightened in several passes around the circle to the stated torque, never to feel. And the surfaces need the specified roughness and fit tolerance, because friction is the entire mechanism.

The coupling behind it

A keyless hub removes backlash at the shaft, which then makes any backlash in the coupling obvious. If you are upgrading one, look at the other: a polyurethane spider in a jaw coupling takes more torque and damps more than the rubber element it replaces, and it is usually a drop-in.

The short answer: in AS568 the first digit of the dash number tells you the cross-section, and the last two tell you the position in that series. A -2xx ring has a 3.53 mm section; a -1xx has 2.62 mm.

How the series break down

Measuring without instruments

Lay the ring flat on a sheet of paper and trace the inside circle, then measure the traced diameter with a steel rule — it is repeatable to about half a millimetre, which is enough to shortlist. For the cross-section, roll the ring between two flat faces and compare it against a drill shank or a known ring. If it matches a 2.5 mm drill it is almost certainly a -1xx.

Metric rings are not AS568

Much of the machinery in Indian plants runs metric ISO 3601 rings, sized directly as inside diameter by cross-section — a 25 x 3 ring is exactly that. They interchange with AS568 only by coincidence, and a near-miss substitution is a slow leak rather than an obvious failure, which makes it worse.

Three numbers we always need

• Inside diameter, relaxed and flat.
• Cross-section diameter.
• Fluid and working temperature — the size tells us what to cut, the duty tells us what to cut it from.