IP20 vs IP65 vs IP67: Choosing an Enclosure Rating Using Six Field Failures

A customer told us our IP67 rating was fake. Their 150 W IP67 supply had been installed at a car wash and died in three months, full of water.

We got the unit back. The housing was intact, the potting compound was intact, and the water had come in through the output cable entry. They had laid the supply flat on the floor with the cable gland pointing up, so every wash-down sent water running down the cable and straight inside.

That unit had genuinely passed IPX7 immersion testing. It had never been tested against a spray nozzle, because IPX7 and IPX5 are two different tests in two different rigs.

This article covers what the IP code actually measures, why IP67 is not an upgraded IP65, why a better-sealed enclosure is more likely to take water outdoors, and why the most common outdoor failure mode isn’t water ingress at all — it’s condensation.

The two digits are independent

IP (Ingress Protection) is defined in IEC 60529 as a code with two digits. The first covers solids, the second covers liquids. The critical point: the two digits are entirely independent. A high second digit says nothing about the first, and vice versa.

IP65 and IP67 both start with 6, which means their dust protection is identical — eight hours in a dust chamber under negative pressure with zero ingress. Only the second digit differs.

Here’s what the second digit actually involves:

Read that table again and the car wash failure explains itself:

IPX7 tests still water. It does not test spray. An IP67 unit has never been proven to survive a pressure washer.

And it runs the other way too: an IP65 unit that handles a hose fine will flood if it ends up in a puddle. One test uses a nozzle cabinet, the other uses a tank.

If a site has both wash-down and standing water, specify dual IP66/IP67 — and ask for two test reports. Don’t let an IP67 certificate stand in for an IP66 requirement; the manufacturer has to be able to show both. Our IP67 waterproof power supply range is built and tested to the immersion spec; where a site also sees hose-down, dual-rated units are the ones to ask for.

Six failures, grouped by actual cause

1. IP67 at a car wash, water in three months. Wrong rating pressed into service (IP67 substituting for IP66/IP69K) plus wrong orientation. A cable gland facing up is a rainwater funnel. Fix: dual-rated unit, gland facing down, and a drip loop in the cable.

2. IP20 in an outdoor light box, batch failure in the rainy season. “Under a cover” got treated as “won’t get wet.” The box had a roof, but wind-driven rain comes in through side seams, and the internal day/night temperature swing condenses. IP20’s first digit is 2 (protects against solids over 12.5 mm) and its second is 0 — no liquid protection at all. One drop finishes it. Those sites need IP65 minimum, or IP66 plus drainage where condensation is heavy. For genuinely dry indoor panels, the IP20 indoor power supply range is the correct — and considerably cheaper — choice; where the unit sits outside but under a soffit, IP65 rainproof supplies usually cover it.

3. IP67 potted units on street lighting, failing in batches after two years. Heat. Potting compound conducts at roughly 0.6–1.5 W/(m·K); aluminium manages 200+. Same 150 W, same losses: 65 °C case temperature in a vented aluminium shell can become 82 °C inside a potted plastic one. Apply the 10 °C rule for electrolytic capacitors — every 10 °C halves life — and 82 °C versus 65 °C costs you about two-thirds of the life. These units were rated 50,000 hours and made it to under 20,000.

4. IP65 enclosure in an underground car park during the rainy season, full of droplets inside. Condensation, not ingress. IEC 60529 tests liquid water; it has nothing to say about water vapour. Humidity cycling between 60% and 95% puts the inner wall below dew point and it rains inside the box. Fix: an ePTFE breather vent to equalise humidity, or a desiccant pack — though desiccant has a service life of 2–3 years and then has to be replaced.

5. IP67 at 3,000 m altitude, creepage breakdown at the terminals. An IP rating was treated as an electrical safety rating. It isn’t. IEC 60529 covers solids and liquids and says nothing about creepage or clearance. At 3,000 m, air dielectric strength is down roughly 30% versus sea level, so clearances that were adequate aren’t anymore. Altitude correction lives in IEC 60664-1, not in the IP code.

6. IP40 in a woodworking shop, fan seized in six months. IP4X stops solids above 1 mm. Wood dust is far smaller. Those environments need IP5X at minimum — limited dust ingress permitted, without affecting operation — or preferably IP6X fully sealed combined with fanless convection cooling. In any dusty environment the fan is always the first thing to die.

Three things that run against intuition

1. The better you seal it, the more likely it takes water outdoors

This sounds wrong and it’s the single biggest cause of outdoor power supply failure. In the industry it’s called breathing.

Air is trapped inside a fully sealed enclosure. Sun pushes it to 50 °C during the day; night drops it to −10 °C. Treated as an ideal gas, that swing (263 K → 323 K) is a pressure change of about 22%, roughly 22 kPa, acting on every seal and cable entry. Day pushes out. Night pulls in.

That inward pull is the problem. Any microscopic defect in a sealing face draws damp outside air in; once inside, it condenses on a cold wall, and it can’t get back out. Repeat that for a season and an IP67 enclosure that has never been rained on can hold visible water.

The standard fix is a waterproof breather vent — an ePTFE microporous membrane that equalises pressure continuously while the pore structure blocks liquid water. A typical spec is airflow ≥300 ml/min at 7 kPa, and a correctly fitted vent preserves the IP67/IP68 rating. The part costs 0.50–3.

Rule of thumb: sealed enclosure, installed outdoors, day/night swing over 20 °C — fit a vent.

2. IP67 thermal performance is worse than IP20, and not by a little

Higher protection gets read as a better product. On thermal performance it’s the opposite.

IP20 indoor supplies usually use an extruded aluminium housing that is the heatsink, with a short thermal path. IP67 has to be sealed, which generally means one of two routes: aluminium with a silicone gasket (decent thermal path, but the gasket adds resistance) or plastic fully potted (worst case).

Measured comparison, 150 W output, 88% efficiency (20.5 W dissipated), 25 °C ambient:

At 92 °C you’re at the usable ceiling of a 105 °C electrolytic and life collapses.

Which is why high-IP outdoor supplies need a much bigger power margin. A 150 W load can run on a 150 W supply indoors. In a potted IP67 enclosure outdoors, specify 200 W — not for inrush, but to pull the case temperature down. Power supply efficiency ratings and thermal derating shows how to read the derating curve that produces that number for your specific ambient.

3. IP says nothing about condensation, salt spray, UV, corrosion or impact

Writing “IP67” on a PO and expecting it to survive a coastal site is a routine mistake. IEC 60529 tests two things. Everything below sits outside it:

• Condensation / vapour — high-temperature steam penetrates seals; IP can’t help

• Salt spray corrosion — IEC 60068-2-52, must be specified separately

• UV ageing — IEC 60068-2-5 or UL 746C (f1 rating) for plastic housings

• Mechanical impact — the IK code (IEC 62262, IK08 = 5 J, IK10 = 20 J), a completely separate system

• Corrosive gas — IEC 60068-2-60 for mixed flowing gas

A coastal project should read “dual IP66/IP67 + salt spray IEC 60068-2-52 severity 3 + housing UV rating f1.” Not “IP67.”

What the extra protection costs

Relative to an IP65 baseline:

A meaningful slice of that IP67 premium isn’t material — it’s inspection. IP67 lines typically run 100% leak detection (differential pressure decay, 15–60 seconds per unit). That’s a fixed cost on the line, which is also why IP67 MOQs are usually higher than IP20’s: the test fixture isn’t worth building for 100 pieces.

A three-step selection process

Step 1 — What form does the water take?

• Never contacts water (indoor panel, dry room) → IP20

• Dust but no liquids (woodworking, textile) → IP40 / IP54

• Rain or low-pressure rinse (light boxes, signage, building facades) → IP65

• Pressure washer (food plants, car washes, decks) → IP66

• Standing water (buried, basement pump rooms, flood zones) → IP67

• Both jet and immersion → dual IP66/IP67, two reports required

• Permanent submersion → IP68, and get the declared depth and duration in writing

Each of those branches maps to a live product family: slim IP67 waterproof supplies for wash-down and immersion, plastic-case IP67 units where unit cost matters more than footprint, outdoor rainproof supplies for soffit and facade mounting, and IP40 dustproof supplies for dry but dirty interiors.

Step 2 — How much thermal headroom?

• IP20 / IP65 convection cooled → size at 100% of load, keep 10–20% spare

• IP67 potted → keep 30–50% spare. A 150 W load gets a 200 W supply

Step 3 — Any stress the IP code doesn’t cover?

• Sealed enclosure with a swing over 20 °C → fit a breather vent

• Coastal or chemical plant → add salt spray or corrosive gas requirements

• Plastic housing outdoors → confirm the UV rating

• Above 2,000 m → re-check creepage, clearance and thermal derating separately

FAQ

Is IP67 always better than IP65? No. It means the unit passed immersion, not that it passed spray. On a pressure-wash site, a unit rated only IP67 can be less reliable than an IP65 one. Select by the form the water takes, not by which number is bigger.

Why does water collect inside a sealed box? Breathing. Day/night temperature cycling drives pressure in and out; the cooling half pulls humid outside air in through any microscopic seal defect, where it condenses and can’t escape. Bigger swings and tighter sealing make it worse. An ePTFE breather vent is the fix.

Can a potted IP67 supply run continuously at full load? It will run, but life drops sharply. Potted construction has high thermal resistance — case temperature runs 15–25 °C higher at the same load. By the 10 °C rule, +20 °C costs you roughly three-quarters of capacitor life. Leave 30–50% power headroom outdoors.

Are IP and IK the same thing? Completely separate systems. IP (IEC 60529) covers solids and liquids; IK (IEC 62262) covers impact energy — IK08 is 5 joules, IK10 is 20. Anything mounted where it can be struck (in-ground luminaires, low-level car park installs) needs both numbers.

Is IP67 more reliable than IP20 indoors? Backwards. In a dry indoor panel, IP67 sealing is pure cost and thermal penalty: higher thermal resistance, 20–40% more expensive, and the seals themselves age. A dry indoor installation gets better life from IP20 aluminium — best cooling, lowest cost.

How do I verify a supplier’s IP67 claim? Ask for three things: a third-party lab report rather than a self-declaration, a sample model number on that report matching what you’re buying, and lab accreditation for IEC 60529 (CNAS, ILAC, or equivalent). Then ask one more question — whether they run 100% outgoing leak testing. An IP67 line that only samples will show noticeably higher batch failure rates.

The short version

Pick the rating by answering “what form does the water take,” then “where does the heat go.” Most outdoor supply failures aren’t caused by a rating that’s too low — they’re caused by picking the wrong form of water, or by not leaving thermal headroom for a sealed structure.

Not sure which rating your site needs? Send two photos — one wide shot of the installation, one close-up of where the supply mounts — plus power, voltage and whether you need dimming, and get in touch with engineering — a recommendation usually comes back within half a day. Full mechanical drawings and ingress test reports for each family are in the download centre.