A gasket is the cheapest part on an enclosure and the one that decides whether an IP66 claim survives the first field season. Most water ingress we see on returned cabinets is not a gasket material failure — it is a compression, flatness or spacing problem fixed on the drawing before anyone chose a rubber. Here are the decisions in the order they should be made.
Water gets in through six paths, not one
List every path before specifying a seal. A door gasket closes one of them, and it is rarely the one that fails first.
- Door perimeter. Sealing quality is set by compression uniformity, not gasket thickness.
- Cable entries. The most common leak point on real installations. A gland rated IP68 in an unsealed knockout still leaks around the thread.
- Vents. Added for condensation control and then forgotten. A vent without a membrane is an open hole.
- Welds and seams. Spot and stitch welds leave a capillary path through the joint.
- Fasteners and mounting holes. Through-bolts need bonded sealing washers.
- Condensation. No leak exists and the enclosure still fills with water. This needs a different fix.
Choose the elastomer for the environment, not the price list
Every material below seals water adequately when new. The differences appear after twelve months of UV, ozone, temperature cycling and compression set.
| Material | Working range | Best suited to | Watch out for |
|---|---|---|---|
| Silicone sponge, closed cell | −60 to +200 °C | Outdoor cabinets, wide temperature swing, UV and ozone | Tears at thin sections; not for oil mist |
| Solid silicone, 40–60 Shore A | −55 to +200 °C | High-reliability doors on heavy duty cycles | Needs much higher closing force; accurate groove required |
| EPDM sponge | −40 to +120 °C | General weather, water, ozone, washdown without oil | Swells in oil and fuel and loses section |
| Co-extruded TPV on frame | −40 to +120 °C | IP66/67 doors, no adhesive, long production runs | Frame profile must be designed for it; extrusion tooling |
Two rules follow. Use closed-cell sponge for doors, and solid elastomer only where the closing force exists to compress it — a 40 Shore A gasket on a 600 mm door needs latches, not a single lock. And if the enclosure sees oil or fuel mist, EPDM is the wrong answer despite being the cheapest good outdoor material.
Compression is the specification that matters
A gasket seals by being squeezed until it fills every surface irregularity. Target 30–50% of free height for closed-cell sponge and 15–30% for solid elastomer. Below 30% on sponge the material does not fill the roughness of the paint or the waviness of a formed door, and water tracks along the interface. Above 50% the cell walls take a permanent set, the gasket stops recovering when the door opens, and the joint leaks the second time it is closed.
Two consequences. Specify compression as a measured value on the assembled unit, never as a groove depth on a drawing — welding distortion, paint thickness and panel flatness all consume the margin. And replace field gaskets with the identical free height; a thinner gasket that "closes more easily" is how a sealed cabinet becomes an unsealed one. Compression force drives the structure too: sponge runs about 0.2–0.5 N per millimetre, so a 2 m perimeter puts 400–1,000 N on the hinges and latches.
Groove design and the fill calculation
The groove locates the gasket and limits compression. Size it so the gasket occupies 70–85% of the groove cross-section when the door is closed. Cross-section area matters, not width:
Fill ratio = gasket area ÷ groove area
A 10 × 8 mm sponge gasket with a 5 mm hole compresses to 6 mm: free area ≈ 66 mm², compressed ≈ 48 mm². A groove 12 mm wide × 6 mm deep gives 72 mm², so the ratio is 67% — reduce the groove depth to reach the 70–85% band.
Put the gasket on the frame, not the door. The frame is usually a formed flange with continuous stiffness; the door is a large unsupported panel that bows under latch force. Sealing against the stiffer surface halves the compression variation around the perimeter.
The sealing face is part of the seal
- Flatness. Target ≤1.5 mm bow per metre and ≤0.3 mm local step. Welding distortion is the usual cause of a bowed door; the fix is weld sequence and fixturing, not a thicker gasket. A 10 mm gasket absorbs only 3–5 mm of compression, so a 4 mm bow cannot be sealed at any thickness.
- Roughness. Ra 1.6–3.2 µm is comfortable for sponge. Above 6.3 µm the texture itself becomes a set of capillary channels.
- Coating. Powder coat at 60–100 µm both reduces groove depth on the gasket land and leaves a hard mask edge that acts as a leak channel. Mask the gasket land and all grounding points, and never chase a leak by sealing over a painted joint.
Latch spacing and door stiffness
The most common cause of a failed first-article hose test is a door that bows between latches. Deflection scales with the cube of the unsupported span, so spacing matters far more than latch strength: keep mid-span deflection below 10% of gasket free height, which means 300–400 mm spacing on a 1.5–2.0 mm door with a 6–8 mm gasket. A 600 mm door with a single central lock will weep at mid-height on both sides — use a three-point latch.
Condensation: the leak that is not a gasket problem
A sealed enclosure with electronics inside is a closed volume of moist air, and when it cools, water condenses on the coldest surface. No gasket prevents this, and drilling a hole in the bottom trades a condensation problem for a dust and water problem. Pressure makes it worse: at constant volume a 20 °C rise produces roughly 7 kPa of differential, and cooling the reverse. That cycling flexes the gasket and, on the cooling stroke, pulls vapour through any marginal joint — the enclosure is a pump. Fit a membrane breather vent rated at or above the enclosure IP level: it equalises pressure in seconds, blocks liquid water and lets trapped moisture escape as vapour. Mount it on a side wall, never facing upward.
Verification: what to test, and when
An IP rating is a type test on a sample, not a production guarantee. Test at first article and again after any change of gasket supplier, groove geometry, tooling or coating process.
| Design parameter | Target value | What happens when it is wrong |
|---|---|---|
| Gasket compression | 30–50% of free height (closed-cell sponge), 15–30% (solid elastomer) | Under-compression leaks; over-compression takes a permanent set and leaks later |
| Groove fill ratio | 70–85% of groove cross-section area | Above 90% the gasket goes hydrostatic and pushes the door open |
| Sealing face flatness | ≤1.5 mm per metre bow, ≤0.3 mm local step | Error larger than the gasket can absorb becomes a continuous leak path |
| Latch spacing | ≤400 mm, mid-span deflection under 10% of gasket height | The door bows between latches and weeps along the seal line |
| Coating on the gasket land | Masked — 0 µm dry film (60–100 µm elsewhere) | Coating kills groove depth and the mask edge becomes a leak channel |
Two things catch people out. Orientation matters: a jet at 30° to the seal is far harsher than one at 90°, which is why IPX5 specifies a 6.3 mm nozzle at 12.5 L/min for 3 minutes at 2.5–3 m and IPX6 a 12.5 mm nozzle at 100 L/min for 3 minutes at 3 m. And thermal cycling before the water test exposes compression set that a room-temperature test never shows: a cabinet that passes IP66 cold and fails after ten cycles between −40 °C and +85 °C is a normal result.
The gasket is typically 1–3% of enclosure BOM cost, which is why it is the last thing specified and the first thing value-engineered. The expensive decisions are the frame profile, the flatness of the door and the number of latches — all of which cost more and matter more than the rubber.
At Xingyuan Electric we build sheet metal enclosures to print for energy storage, EV charging, semiconductor and industrial customers, with IP66/IP67 sealing designed at the drawing stage rather than tested for at the end. Our facility runs a 52-person team, a 5,000+ m² production floor and 20+ laser cutting, CNC bending and welding machines, with standard orders delivered in 7–15 days. Build-to-print and sample-based customisation are both supported, and gasket selection, groove geometry, latch layout and first-article water testing are handled in-house. Send us your drawings and target IP rating and we will return a sealing design review before quoting.
Frequently asked questions
Is a thicker gasket better for IP66?
No. Sealing depends on compression percentage, not size. A 12 mm gasket compressed 15% seals worse than an 8 mm gasket compressed 40%, and it loads the frame more heavily. Set free height so compression lands in the 30–50% band for sponge.
Can a single-point lock hold IP66 on a 600 mm door?
Rarely. The door bows between the latch and the corners and the seal weeps at mid-height. Use a three-point latch, or two latches plus a stiffened latch edge, and keep spacing at or under 400 mm.
Does IP66 mean the enclosure can be washed down continuously?
No. IPX6 is a 12.5 mm nozzle at 100 L/min for 3 minutes at 3 m. Continuous high-pressure, high-temperature washdown is IP69K (80–100 bar at 80 °C from four angles), which needs solid or co-extruded gaskets.
Silicone or EPDM for an outdoor enclosure?
EPDM sponge for general outdoor weather — cheaper, and it handles water, UV and ozone well. Move to silicone for extreme temperature range, minimal compression set, or exposure to oils and fuels, where EPDM fails.
Why is there water inside a sealed enclosure with no visible leak?
Condensation. The enclosure is a closed volume of moist air that cools at night and the water has to go somewhere. Fit an IP66 or IP68 rated membrane breather vent to equalise pressure. Do not drill a drain hole, which turns a condensation problem into an ingress problem.
How do I know when a gasket needs replacing?
Check for permanent set: if it does not recover to roughly 85% of free height within a minute of the door opening, it is spent. Typical life is 5–10 years for silicone or EPDM sponge and 3–5 years for PU foam. Replace with the identical section.