Solutions
IP65 outdoor DTU enclosure: condensation, glands and heat
IP65 stops sprayed water, not the humidity sealed inside. Dew point, gasket pressure and solar gain, worked through for a pole-mounted DTU.
A pole-mounted data terminal dies in its second monsoon, and the post-mortem finds a bead of water sitting on the PCB inside a housing marked IP65. The rating is not a lie. It answers a narrower question than the one the installer asked: IP65 says a jet of fresh water aimed at the outside of the box for a few minutes does not get in. It says nothing about the air that was closed inside at 08:40 on a humid morning, nothing about the 14 kPa the gasket sees when the afternoon sun heats the lid, and nothing about a cable gland tightened onto a jacket a millimetre thinner than its sealing range.
Those three numbers — dew point of the trapped air, gauge pressure across the seal, solar gain on the housing — decide whether an outdoor enclosure survives. Each is a one-line formula. Together they explain most of the failures that get blamed on the IP rating itself.
The reference hardware here is a real outdoor DTU: the TP5500, 116 × 85 × 41 mm, IP65, rated −40 °C to +70 °C, drawing 1 W or less from a 12 V supply. It shares its shell with the TP5510 gateway, so the same arithmetic applies to both. Everything below is either quoted from those specification tables or computed from a stated assumption; the assumptions are labelled so you can substitute your own.
What IP65 actually certifies
IP65 is two independent digits from IEC 60529. The first, 6, is the dust scale: dust-tight, verified by a dust-chamber test. The second, 5, is the water scale: protection against water jets, tested with a 6.3 mm nozzle delivering 12.5 litres per minute at 30 kPa from three metres away, for at least three minutes, from any direction. Level 6 would be a 12.5 mm nozzle at 100 kPa; level 7 immersion. Both digits come from IEC 60529, whose publication record is on the IEC webstore — worth reading in full before you quote a rating in a tender.
Test element | What the standard actually does | What it leaves untested |
|---|---|---|
First digit 6 | Dust chamber, no dust ingress afterwards | Dust trapped at assembly; air movement through a cable core |
Second digit 5 | 6.3 mm jet, 12.5 L/min, 30 kPa, 3 m, ≥3 min, all directions | Condensation, immersion, steam cleaning, gasket ageing |
Sample condition | A sample tested at the factory | Three years of UV, thermal cycling and vibration in the field |
Read the third row carefully. The IP test is a pass/fail on a sample, not a service-life statement. A gasket that passes at 23 °C in a laboratory and a gasket that is still sealing after 2,000 daily thermal cycles are different engineering problems, and only the second one matters on a pole.

The air you close inside is the water that eventually kills it
Air at 30 °C and 60 % relative humidity carries 18.2 g of water per cubic metre. The TP5500 enclosure holds 404 mL, so closing it at that condition traps 7.3 mg of water — a quantity you cannot see. That is not the problem, and it is worth saying so plainly, because the usual remedy (throw in a desiccant sachet) is sized for a much smaller problem than the one that actually occurs.
The problem is two-fold. First, the dew point: at 30 °C and 60 % RH, water starts to condense on any surface below 21.4 °C. Second, and far more important over months, the enclosure is not gas-tight in the way the marketing implies. A stranded cable is a capillary: moisture migrates along the interstices between strands, through the gland, into the box. Every thermal cycle also pumps air past the gasket lip as the internal pressure swings — the subject of the next section.
Air at assembly | 50 % RH | 60 % RH | 70 % RH | 80 % RH |
|---|---|---|---|---|
20 °C | 9.3 °C | 12.0 °C | 14.4 °C | 16.4 °C |
25 °C | 13.9 °C | 16.7 °C | 19.1 °C | 21.3 °C |
30 °C | 18.4 °C | 21.4 °C | 23.9 °C | 26.2 °C |
35 °C | 23.0 °C | 26.1 °C | 28.7 °C | 31.0 °C |

The table is computed with the Magnus formula, a = 17.62 and b = 243.12 °C, which is accurate to a few tenths of a degree over the range that matters here. Its practical reading is simple: a box assembled in a workshop at 30 °C and 60 % RH will start condensing on the underside of its lid as soon as that lid falls below 21.4 °C. On a clear night the lid radiates to the sky and runs 2–5 K below air temperature, so in a subtropical climate that threshold is crossed routinely between November and March — and indoors in an unheated workshop it is crossed every winter.
Once condensed, the water does not leave. It sits in the low corner, wicks into the terminal block, and by the second season the RS485 pair has a leakage path to ground. This is why "sealed" and "dry" are not synonyms, and why the enclosure design has to move water vapour out rather than pretend it is not there.
Every warm afternoon pushes on the gasket
A sealed box obeys the gas law whether or not anyone planned for it. With the volume fixed, gauge pressure scales with absolute temperature: ΔP = P₁(T₂/T₁ − 1). Closing the lid at 20 °C (293 K) and letting the sun take it to 60 °C (333 K) leaves 13.8 kPa of overpressure inside. On the 116 × 85 mm face of the TP5500 that is 136 N — roughly the weight of a 14 kg bag — pushing the lid away from the gasket.
Closed at | At −10 °C | At +60 °C |
|---|---|---|
0 °C | −3.7 kPa | +22.3 kPa |
20 °C | −10.4 kPa | +13.8 kPa |
40 °C | −16.2 kPa | +6.5 kPa |

The negative column is the dangerous one. Underpressure pulls the gasket lip inward and draws water and humid air through any imperfection — a scratch on the sealing face, a gland whose elastomer has taken a set, a cable jacket that shrank 0.3 mm in the cold. A box closed on a hot afternoon at 40 °C sees −16.2 kPa overnight at −10 °C, which is more than one and a half times the pressure a 30 kPa jet applies from outside. That is the mechanism behind the classic complaint: the unit passed IP65 at the factory and flooded in its first cold snap.
There are three honest ways to deal with it. Use a pressure-equalisation vent with an ePTFE membrane, which passes air and water vapour but not liquid water; fill the enclosure with dry air or nitrogen at assembly and accept a finite life; or design the gasket and gland for the pressure range and test it thermally, not just with a hose. Drilling a 2 mm hole "to let it breathe" is not on the list, because it also lets in the jet, the dust and the ants.
Sizing the housing before you buy it
Combine the two effects and you get the number that decides whether a housing works: internal air temperature at the hottest hour. For a first-order estimate, the surface sits α·G/h above ambient, where α is the solar absorptivity and h the combined convection and radiation coefficient — 12 W/m²K is a reasonable value for a still day. Inside, 1 W of electronics into roughly 0.05 m² of internal surface at h ≈ 5 W/m²K adds about 4 K.
Ambient at 14:00 | White housing (α 0.30) | Light grey (α 0.60) | Dark grey (α 0.90) |
|---|---|---|---|
35 °C | 59 °C | 79 °C | 99 °C |
40 °C | 64 °C | 84 °C | 104 °C |
45 °C | 69 °C | 89 °C | 109 °C |
Assumptions: 800 W/m² irradiance, no wind, no shade, unvented box. The 70 °C operating limit in the TP5500 specification is the column to compare against — and the table says something most integrators never check. A white housing keeps a 1 W terminal inside its rating at 45 °C ambient with 1 K to spare; the same electronics in a dark grey housing is 34 K over the limit. Colour is not a cosmetic choice on an outdoor box.
Three practical consequences follow. Mount the box so the lid faces away from the midday sun rather than bolting it flat to a south-facing wall. Leave an air gap of at least 20 mm between the housing and the mounting surface, because a box flush against concrete loses its back-face convection path. And if the installation is unavoidably in full sun in a hot climate, add a ventilated sunshade — a plate 40 mm above the lid does more than any change you can make inside.
Matching the enclosure to the job
Not every node needs IP65, and paying for it where it is not needed buys nothing.
TP5500 | TP5510 | TP5301 | |
|---|---|---|---|
Role | Outdoor RS485/RS232 data terminal | Gateway, up to 100 terminals | Indoor RS485 data terminal |
Protection class | IP65, outdoor | IP65, outdoor | IP30, indoor |
Dimensions | 116 × 85 × 41 mm | 116 × 85 × 41 mm | 79 × 54 × 30 mm |
Operating temperature | −40 to +70 °C | −40 to +70 °C | −20 to +70 °C |
Power | ≤1 W | ≤1 W, ≈0.25 W average | ≤0.5 W |
Supply | DC 12 V (5–28 V) | DC 12 V (5–28 V) | DC 12 V or AC 220 V |
Mounting | Wall or DN35 DIN rail | Wall or DN35 DIN rail | Wall |
The TP5301 is the case where the enclosure question disappears: put it inside a meter box or a control cabinet and the IP30 rating is sufficient, the −20 °C limit is irrelevant, and the AC variant removes a power supply from the bill of materials. The mistake is using it outdoors because it is smaller and cheaper, then discovering that the 30 mm depth has no room for a gland and the lid has no sealing face at all.
Failure modes worth designing out
- Gland range mismatch. Every gland has a clamping range, typically 4–8 mm or 9–14 mm. A cable at the edge of the range seals on the jacket; a cable below it seals on nothing, and the gland body still looks tight.
- Gland on the top face. Water pools on the gland and works past the seal. Enter from the bottom, always, and add a drip loop so the cable runs below the entry before rising.
- A single cable entry for two cables. Splitting one gland between power and RS485 halves the compression on both and leaves a gap between the jackets.
- Desiccant used as a substitute for sealing. A 1 g silica sachet holds roughly 200 mg of water at 40 % RH — 27 times the 7.3 mg trapped at assembly, which is why it works at first and is exhausted within weeks once vapour is migrating through the cable core.
- Screws over-torqued on a plastic lid. The gasket takes a permanent set, and after two summers the compression that passed the factory test is gone. Use the torque in the housing datasheet, not the wrist.
- UV-unstabilised glands and cable ties. Nylon without UV stabiliser embrittles in two to three seasons; the gland snaps at the thread root during the next cable pull and the box is open until someone notices.
- Condensation on the cold lid, not the warm board. Water forms on the inside of the lid at dawn and drips onto the terminals. A PCB mounted with the connector side down, plus a lid that is not thermally isolated from the sky, makes this worse.
Verifying an installation in one afternoon
You do not need a climate chamber to catch most of these. Five checks, in order:
- Log temperature and humidity inside and outside for 72 hours. Put a small logger in the box next to the board. Compute the dew point of the internal air from the logged T and RH, then compare it with the minimum internal surface temperature — in practice the minimum logged air temperature, which is close enough at night. If the margin is under 2 K, the box will condense and you need a vent or a drier assembly routine.
- Do the hose test the standard will not do for you. Close the box at a known 25 °C / 50 % RH with a paper towel inside, then run low-pressure water over it for 30 minutes, including from below. Open it immediately: any damp patch on the towel is a real leak, and it will show where the factory test — done on a dry sample — could not.
- Do the cold test. Close a second sample at the same condition, put it in a freezer at −10 °C for two hours, then bring it out and open it within a minute. Condensation on the lid proves the dew point margin is negative; this is the single most predictive test for subtropical deployments.
- Measure the solar gain in service. Point an infrared thermometer at the lid at 14:00 on a clear day and compare with shaded ambient air. Above a 30 K rise, the housing colour or the mounting position is costing you operating margin, and no amount of internal heatsinking fixes it.
- Check the entry hardware by hand. Twist each gland body — it should not rotate — and pull each cable 20 N; a cable that slides is a cable that will leak. Confirm the drip loop drops at least 50 mm below the gland before turning up.
A terminal that passes all five has a defensible installation. One that passes only the factory IP test has a rating, which is a different thing.
FAQ
Is IP65 enough for a pole-mounted DTU in a monsoon?
For direct rain and sprayed water, yes. IP65 is rated for jets at 30 kPa, which is well above driving rain. What it does not cover is water arriving through the cable entry, condensation from air trapped at assembly, and gasket ageing. Treat IP65 as the starting condition and design the cable entries and the thermal path as if the rating did not exist.
Should I fit a breather vent or drill a drain hole?
Fit a membrane vent if the enclosure sees large temperature swings — it equalises pressure and passes vapour while blocking liquid water. Never drill a plain hole: it converts an IP65 enclosure into an IP20 one and gives insects a route to the terminals. If the box has no vent, at minimum assemble it in dry, cool conditions and accept that a desiccant sachet buys time rather than solving the problem.
Why did my unit fail after the first cold night rather than in the rain?
Because the failure was condensation, not ingress. A box closed at 30 °C and 60 % RH has a 21.4 °C dew point, and a lid radiating to a clear sky drops below that on the first cool night. Water then sits on the board until the next warm day, when it evaporates and the cycle repeats. The cold test in the checklist above reproduces it in two hours.
Do I still need a sunshade if the housing is white?
Usually not, but measure. At 800 W/m² and α = 0.30 the surface runs about 20 K above ambient, which leaves a 1 W terminal at 69 °C in a 45 °C ambient — inside the 70 °C rating but with almost no margin. A shade, or a mounting position that keeps the lid out of the midday sun, buys back 15–20 K for the cost of a bracket.
If the enclosure is being specified alongside the radio, the product range lists the dimensions, protection class and operating temperature for each model, and the RS485 side of the same installation is covered in the meter-room wiring guide.



