A SanJun customer ran a 600 W power supply inside a sealed 19-inch rack enclosure last summer. The supply was rated for 70 °C ambient; the enclosure interior hit 92 °C within an hour and the supply began derating within two. Nothing was wrong with the supply and nothing was wrong with the sheet metal — the enclosure simply had no path for the heat to leave. This guide covers the four heat paths and how much each carries, the load bands that tell you when passive cooling stops being enough, how to size vent patterns for the airflow you will really get, fin design rules, when material choice matters, how to move air without giving up IP protection, and what solar load and condensation do to a budget that looked fine on paper.
Key takeaways
- Conduction and radiation through the walls carry only about 5–10 W per m² per 10 K of temperature difference. Enough for low-power electronics; nowhere near enough for inverters, industrial PCs or battery racks.
- Material only helps if there is somewhere for the heat to go. Aluminium conducts roughly three to four times better than cold-rolled steel, but a sealed aluminium box traps heat just as effectively as a sealed steel one.
- A fan never delivers its datasheet airflow. Real CFM lives where the fan curve crosses the system impedance curve, and on a choked inlet that is commonly 30–50% below the catalogue figure.
- Solar load is a heat source, not a background condition. Direct sun deposits roughly 800 W/m² at noon in temperate latitudes and up to 1,100 W/m² in desert climates, which can add 20–40 K to a black enclosure.
- SanJun supplies the metal, the machining and the supporting evidence. We are a sheet metal fabricator: we do not issue listings or certificates of compliance, and we do not run UL 9540A.
Why does an enclosure overheat when nothing is wrong with the hardware?
Because the enclosure is a boundary, and every watt dissipated inside has to cross it. Components do not care what temperature the room is; they care what temperature the air immediately around them is. Put a supply rated for 70 °C ambient into a box that has no way to reject 600 W, and within an hour the air around the supply is 92 °C regardless of what the datasheet promised.
Three reasons this catches teams out late:
- The rating was for ambient, not for the enclosure interior. An industrial PC rated for 70 °C ambient drops to a 50 °C rating when it sits inside a sealed, sun-heated cabinet. The number on the datasheet is not the number inside the box.
- The thermal budget was never written down. Nobody summed the dissipation, so nobody knew what the design needed.
- Cooling and sealing were specified by different people. The IP requirement closed every opening the thermal design needed — see our enclosure ventilation guide.
The fix starts with one number: total internal dissipation in watts. Everything else follows from it.
What are the four heat paths, and how much does each carry?
Heat leaves an enclosure through exactly four mechanisms, and every thermal decision trades one against the others:
| Path | What drives it | How much it carries | When it dominates |
|---|---|---|---|
| Conduction through the wall | Material conductivity, wall thickness, contact area | Small on its own; sets how well heat reaches the outer surface | Low-power sealed enclosures, conduction-cooled designs |
| Natural convection through vents | Buoyancy, vent area, height between inlet and outlet | Carries the majority in passive designs | 30–150 W with low and high vents |
| Forced convection with fans | Airflow and the fan's operating point | Multiplies natural convection roughly 5–10× | 150 W and up; trades noise, dust and reliability |
| Radiation from the surface | T⁴, surface emissivity and finish | 10–20% of the total below about 80 °C | A secondary contributor — and a large one for a sun-facing surface |
The number to internalise is this: conduction and radiation through the walls carry only about 5–10 W per square metre of cabinet surface per 10 K of temperature difference. A 1 m² cabinet with a 20 K internal rise sheds perhaps 10–20 W through its skin. That is fine for a controller drawing 15 W and hopeless for a 600 W supply.
The most common mistake is over-investing in one path and forgetting the others: a sealed aluminium enclosure with no vents traps heat as effectively as a sealed steel one.
How much heat can a sealed enclosure shed without vents?
Less than most drawings assume, and the arithmetic is worth doing before you commit to a sealed design. Take the surface area available, take the temperature rise you can tolerate above ambient, and apply the 5–10 W per m² per 10 K figure.
A worked example: a 600 × 400 × 300 mm enclosure has roughly 1.1 m² of external surface. With 20 K of internal rise, conduction and radiation together shift somewhere in the region of 11–22 W. That is the entire passive budget of a sealed box that size. Anything above it makes the interior temperature climb until either something derates or something fails.
Three ways to raise the ceiling without opening the box:
- External fins. They multiply the surface area doing the radiating and convecting. Fabricated fin stacks are the cheapest version and are covered below.
- A conduction path to a larger sink. Mounting the hot component to the wall with a thermal pad or a machined boss moves heat into the sheet instead of the internal air.
- Colour and finish. A light-coloured, high-reflectance exterior changes the solar term, which matters outdoors.
If the dissipation exceeds what those three can shift, the enclosure has to breathe — passively or with a fan.
When do you need vents, and when do you need a fan?
The decision is straightforward once you know total internal dissipation and the internal temperature rise you can tolerate:
| Total dissipation | Strategy | What it looks like | Trade-offs |
|---|---|---|---|
| ≤ 30 W | Conduction through the wall | Sealed; 2.0 mm or thicker aluminium, or external fins | No vents, no fans, no filter maintenance, best IP performance |
| 30–150 W | Conduction plus natural convection | Filtered IP54 vents low and high; chimney effect | Needs height between inlet and outlet; modest dust ingress |
| 150–500 W | Forced convection | Internal fan or blower, filtered intake and exhaust | Noise, filter maintenance, fan-fail derating plan required |
| 500 W and up | Forced convection plus heat exchanger, or thermal zoning | Air-to-air heat exchanger, heat pipes, or split zones | Highest cost; keeps IP rating on the outer loop |
Three rules apply inside every band. Put the cool inlet low and the hot outlet high — buoyancy does the rest. Filter the intake on anything outdoors, and plan the filter as a maintenance item with an access panel. And write down what happens when the fan fails: a forced-convection design with no derating plan fails on the first hot day after it stops.
For multi-kW architectures, splitting the enclosure into thermal zones — keeping the power stage separate from the control electronics — is usually cheaper than trying to cool the whole volume. Our energy storage cabinet guide shows an example of that approach at scale.
How do you size a vent pattern for real airflow?
Start from the fan's operating point, not from its datasheet. A fan never delivers its catalogue airflow once it sits behind a perforated panel, a filter and a board full of components. Air moving through an enclosure has to overcome the vent pattern, the filter media, the internal boards and any restriction at the outlet; that resistance is the system impedance curve, where pressure drop rises roughly with the square of flow. The fan adds its own curve, where available pressure falls as flow increases. Real CFM lives where the two curves cross.
When open area is insufficient, the vent sheet becomes the dominant restriction and the operating point slides up and left: pressure demand climbs and airflow falls faster than intuition suggests. On real builds we measure delivered CFM 30–50% below the datasheet value, with higher noise, higher power draw and local hot spots downstream of a choked inlet.
Open area ratio — the share of the panel that is hole — is the number to control:
- 30–40% — the minimum for effective natural convection. Rarely enough for forced air.
- 40–50% — the sweet spot for general electronics enclosures: good airflow, plenty of stiffness left.
- 50–60% — high-power racks and inverter cabinets. Needs stiffening ribs or return flanges.
- 60% and above — only where the outer skin carries the structure. A bare panel at 60% will oil-can under finger pressure.
Pitch drives the percentage: a 5 mm round hole on a 7 mm staggered pitch gives about 46% open area; the same hole on a 6 mm pitch gives 64%. Keep at least 2–3× sheet thickness between any hole edge and a bend line. The geometry, cost and CFM picture is in our vent pattern cost and CFM guide, and the pattern options in the perforated vent pattern guide.
How do you design sheet metal heat sink fins?
Finned sheet metal is the cheapest way to multiply the surface area of a passive enclosure. The rules of thumb that hold up on real parts:
| Parameter | Natural convection | Forced convection | Why |
|---|---|---|---|
| Fin pitch | 4–8 mm | 2–4 mm | Natural convection needs room for the boundary layer to develop |
| Fin height | 15–30 mm | 10–20 mm | Tall fins starve at the tip without flow |
| Fin thickness | 1.0–1.5 mm | 1.0–1.5 mm | Thinner fins add cost and weight, not heat transfer |
| Material | Aluminium 6063 / 6061, or folded sheet | Same | Extrusion transfers best; folded sheet is cheaper at low volume |
Fabricated fin stacks — folded sheet, tack-welded or bonded to the wall — work well for low and medium loads and cost far less than an extrusion at prototype volumes. Two things decide whether they perform. The bond to the wall is the whole thermal path, so a stack spot-welded at four points performs like a much smaller fin. And the fins must be oriented so buoyancy works with them: vertical fins on a vertical wall, with an unobstructed path for air to enter at the bottom.
Orientation matters more than fin count. A horizontal fin stack on a vertical wall performs badly because the rising air cannot get between the fins. If the enclosure is outdoors, an external fin stack on the shaded side is worth more than a larger one in the sun.
Which material moves heat best, and when does it matter?
Aluminium, clearly — roughly 160 W/m·K for 6061 against roughly 50 W/m·K for cold-rolled steel, so about three to four times better. That number matters in exactly two places and nowhere else:
- Getting heat from a component into the wall. A conduction-cooled design lives or dies on this, and an aluminium subplate under the hot part is the standard move. See our subplate and mounting plate guide.
- Spreading heat across the outer skin. A high-conductivity wall distributes a local hot spot over the whole surface, so the area that radiates and convects is larger.
It does not matter at all if there is no path out. A sealed aluminium enclosure with no vents, no fins and no conduction path traps heat just as effectively as a sealed steel one — the conductivity only helps once the heat has somewhere to go. That is the single most common way a material upgrade gets spent without buying any thermal performance.
Wall thickness is the other overlooked variable: a thicker wall spreads heat better and adds thermal mass, which flattens short-term peaks. It also adds weight and cost, so spend it where the hot spot is. Material selection across steel, stainless and aluminium is covered in our material selection guide.
How do you keep IP protection while moving air?
Ventilation and ingress protection are usually specified as if they were competing, and they are not — a vent that maintains protection is engineered hardware, not a hole. The families that matter:
- Membrane vents (ePTFE). Hydrophobic membranes pass air and water vapour while blocking liquid water, typically rated IP66 to IP68 depending on the housing. Excellent for pressure equalisation and condensation control; low airflow, so they are not a cooling solution on their own.
- Mesh or louver vents. Stamped or laser-cut openings with a wire mesh or baffle behind, typically IP54 and occasionally IP66 with fine mesh. Higher airflow, lower protection.
- Labyrinth and baffled vents. A convoluted path that lets air through and stops direct water jets. Good compromise for IP54–IP55 at reasonable airflow.
- Honeycomb panels. Reach 70–80% open area, and on the EMC side each cell acts as a waveguide below cutoff — useful when shielding and cooling have to coexist.
One rule decides whether a vented enclosure stays dry: every penetration gets sealed at assembly, and the vent goes where water cannot run in. A membrane vent on the side of a cabinet is fine; the same vent facing upward on the top surface is not. The sealing details are in our gasket sealing to IP65 guide, and the rating definitions in the outdoor waterproof IP guide.
What do solar load and condensation do to the thermal budget?
Both are outdoor terms that indoor budgets omit, and both are large enough to invalidate a design that passed on the bench.
Solar load is a heat source. An enclosure in direct sun absorbs roughly 800 W/m² at noon in temperate latitudes and up to 1,100 W/m² in desert climates. On a black enclosure that raises the internal temperature 20–40 K above ambient — which is more than most electronics have in headroom. A sunshade is the cheapest fix: a well-designed one with a light-coloured exterior (solar reflectance at or above 0.85) reduces absorbed flux by 60–85%, which is roughly the improvement you would get from doubling the heat-sink area. On one 36-inch display enclosure in Phoenix, a 250 mm shade at 10° tilt took the case from 25 K above ambient to 13 K — for about 4% of the enclosure cost. Our sunshade and hood design guide covers the sizing rules.
Condensation is the other half of the same problem. A sealed enclosure is not a dry enclosure: IP66 stops water getting in and does nothing about moisture already in the trapped air. Warm air holds more vapour than cold air, so a roof that radiates to a clear night sky drops below dew point and manufactures water. A cabinet sealed at 30 °C and 80% relative humidity holds roughly 24 g of water per cubic metre; cool the roof to 22 °C and the air beneath it can hold about 19 g, and the missing five grams appear as liquid water on the underside. The controls are a breather, a drain at the true low point, and an anti-condensation heater sized from P = k × A × ΔT with a ΔT of 3–5 K. Full treatment in our condensation control guide.
How do you verify a thermal design before tooling?
Five checks, in the order they catch problems:
- Sum the dissipation. Every rail, every driver, every display backlight. The number is usually higher than anybody's estimate, and it is the input to everything else.
- Name the allowable internal temperature for the most sensitive component, not for the enclosure as a whole.
- Choose the strategy from the load band and check the surface area actually available for it — wall area for passive, vent area for convection, duct area for forced air.
- Add the environmental terms: solar load, ambient extremes, altitude, dust, and whether the enclosure is in sun or shade.
- Measure a prototype at the worst case. Thermocouples on the hottest component, the internal air and the external skin, at maximum ambient and full load, in a settled steady state.
Environment classes and how they are verified are written up in standards such as IEC 60068-2 for environmental testing, and temperature-rise verification of assemblies is handled under IEC 61439 — both are covered in our qualification test plan guide. What we supply is the metal, the machining and the supporting evidence: we build to the agreed procedure and measure the first article. We do not issue listings or certificates of compliance, and we do not run UL 9540A; a witnessed third-party test or a certificate naming a standard is agreed at RFQ and run by the named laboratory.
Once those numbers point to a vented aluminium shell rather than a sealed steel box, our custom power supply chassis enclosure page shows how we size honeycomb vent arrays against a stated airflow figure, and how coolant ports, indicator windows and masked grounding points are placed on the same shell.
Frequently Asked Questions
How much heat can a sealed sheet metal enclosure dissipate?
Conduction and radiation through the walls carry only about 5–10 W per square metre of external surface per 10 K of temperature difference. A 600 × 400 × 300 mm enclosure has roughly 1.1 m² of surface, so at 20 K of internal rise it sheds on the order of 11–22 W. Above that you need fins, a conduction path to a larger sink, or vents.
Do I need a fan, or will vents be enough?
Work from total internal dissipation: up to 30 W can usually be handled sealed; 30–150 W needs filtered vents low and high for natural convection; 150–500 W needs forced air with a fan-fail derating plan; above 500 W you are looking at a heat exchanger or thermal zoning. Check the vent area you actually have before committing.
Why does my fan deliver far less airflow than its datasheet?
Because the datasheet figure is free-air delivery. Real airflow is the intersection of the fan curve and the system impedance curve — the resistance of the vent pattern, filter, boards and outlet. With insufficient open area the vent sheet becomes the dominant restriction, and we commonly measure 30–50% below the catalogue value.
What open area should a vent panel have?
30–40% is the minimum for effective natural convection; 40–50% is the sweet spot for general electronics; 50–60% suits high-power racks and needs stiffening; above 60% only works where the outer skin carries the structure. Pitch drives it — a 5 mm hole on 7 mm staggered pitch gives about 46%.
Does an aluminium enclosure cool better than steel?
Aluminium conducts roughly three to four times better than cold-rolled steel — about 160 W/m·K against 50 W/m·K — but only if the heat has somewhere to go. A sealed aluminium box with no vents traps heat just as effectively as a sealed steel one. Spend the material where heat enters the wall or spreads across the skin.
How much does direct sun add to the internal temperature?
Roughly 800 W/m² at noon in temperate latitudes and up to 1,100 W/m² in desert climates, which on a black enclosure adds 20–40 K above ambient. A sunshade with a light-coloured exterior reduces absorbed flux by 60–85% — comparable to doubling the heat-sink area.
Can an enclosure be both IP-rated and ventilated?
Yes. Membrane vents pass air and vapour while blocking liquid water and are rated IP66 to IP68; mesh and louver vents reach IP54, occasionally IP66 with fine mesh. The rule that matters is placement — a vent on a vertical face is fine, one facing upward on the top surface is not.
Does SanJun run thermal testing or certify the result?
We build the enclosure, do the machining and supply supporting evidence, and we measure the first article. We do not issue listings or certificates of compliance, and we do not run UL 9540A. Environmental testing is specified against standards such as IEC 60068-2 and temperature-rise verification under IEC 61439, and a witnessed test is agreed at RFQ and run by the named laboratory.
Heat has to have somewhere to go
Every thermal failure we are asked to fix comes back to the same thing: the heat was real, the budget was not written down, and the path out was never designed. Sum the watts, name the allowable temperature, pick the strategy from the load band, add the environmental terms, and measure a prototype at the worst case. Do that and the enclosure stops being the reason a product derates.
Send us the drawing, the dissipation figure and the installation environment. We will return a thermal proposal — vent area and placement, fin geometry, material and finish, and a prototype measurement plan — with a quotation. For worked examples of multi-kW thermal architecture, see our thermal solutions case studies.
Need an enclosure that keeps its electronics inside rating? Send your drawings and heat load for a thermal proposal and quotation within 24 hours.