Heat sink thermal resistance in sealed enclosures

Yes, an internal finned heat sink can collect heat from enclosure air and conduct it to an external heat sink. However, you cannot automatically reuse its published thermal resistance with the heat flow reversed. Build a branched model, assign each heat source to its actual path, and calculate temperature rises using the watts through each branch. Keep component-specific losses separate from shared enclosure losses. The result predicts junction temperature only as well as the airflow, contact and wall-spreading assumptions behind it.

What reversing heat flow really changes

Thermal resistance relates a temperature difference to heat flow: R = ΔT / Q. Its units are °C/W or K/W; both give the same temperature-rise calculation.

For conduction through a metal part, reversing the temperature difference reverses heat flow. The resistance remains approximately unchanged if geometry, contact pressure and material properties remain comparable. Contact interfaces can also be modeled this way over a validated operating range.

A heat sink's published sink-to-air resistance is different. It describes the complete exchange with air under particular conditions, not just conduction through its metal.

Those conditions can include:

• Fin orientation and clearance around the heat sink.
• Natural convection or a specified forced airflow.
• Heat load and temperature difference above ambient.
• Surface finish and radiation to surrounding surfaces.
• Air properties, pressure and the manufacturer's test arrangement.

A cooler internal fin assembly develops different buoyant circulation from a hotter assembly. Nearby walls and components further restrict that circulation. A resistance measured for a hot sink in open air therefore needs validation before use as an internal air collector.

The distinction between internal and external heat-sink convection matters more than the direction of conduction through the metal.

Draw the branches before adding resistances

Start with four temperature nodes: component junction, internal air, shared enclosure wall and external ambient. Add separate wall nodes if substantial temperature gradients exist across the enclosure.

For a component that rejects its heat into internal air, the simplified path is:

junction → component heat sink → internal air → enclosure wall → external ambient

The internal air can reach the wall through two parallel routes:

• Convection into internal fins, followed by conduction through their attachment into the wall.
• Heat transfer to the remaining exposed internal enclosure surfaces.

The wall can also reject heat through parallel routes:

• Conduction through the external heat-sink attachment, followed by heat rejection from its fins.
• Heat rejection from the remaining exposed outside enclosure surfaces.

Do not count wall area covered by a heat sink again as bare surface area. Also distinguish convection to air from radiation to surrounding surfaces when those surrounding temperatures differ.

Other devices mounted directly to the wall inject heat at the wall node. Their watts cross the shared external paths, but not necessarily the internal air-to-wall paths.

For the basic model, define:

• Qc: component heat entering the internal air.
• Qa: additional heat entering that air.
• Qw: heat conducted directly into the shared wall.
• Rin: equivalent resistance from internal air to the wall.
• Rout: equivalent resistance from the wall to external ambient.
• Rlocal: validated component junction-to-internal-air resistance.

Then:

Twall = Tambient + (Qc + Qa + Qw) × Rout

Tair = Twall + (Qc + Qa) × Rin

Tjunction = Tair + Qc × Rlocal

These equations assume steady conditions and an approximately uniform wall node. The sealed-enclosure conduction article explains the physical conduction path; this network determines which watts use each part.

Calculate junction temperature with shared loads

Consider an illustrative 1 W processor and another 5 W conducted directly into the enclosure wall. Assume no additional air load, solar heating or other heat input.

Use the following trial resistances. These are calculation assumptions, not measurements, catalog specifications or values inferred from enclosure dimensions.

The internal fin branch could comprise 3 °C/W for air-to-fin exchange and 0.25 °C/W for its attachment. The external branch could comprise 0.1 °C/W for its attachment and 1 °C/W for sink-to-ambient exchange. Include wall conduction and spreading in the appropriate path rather than silently omitting them.

For parallel branches, add conductances:

Rin = 1 / (1/3.25 + 1/6.5) = 2.17 °C/W

Rout = 1 / (1/1.1 + 1/4.4) = 0.88 °C/W

The external paths carry a combined 6 W. Their wall temperature rise is:

ΔTwall = 6 W × 0.88 °C/W = 5.28 °C

Only the processor's 1 W crosses the internal air-to-wall network:

ΔTair-to-wall = 1 W × 2.17 °C/W = 2.17 °C

The processor's local temperature rise is:

ΔTjunction-to-air = 1 W × 10 °C/W = 10 °C

The total junction rise above external ambient is therefore approximately 17.45 °C. At 85 °C ambient, the predicted junction temperature is 102.45 °C.

For an assumed 105 °C junction limit, the calculated zero-margin ambient ceiling is:

Tambient,max = 105 − 17.45 = 87.55 °C

That is not a qualified operating limit. Choosing an illustrative 5 °C junction margin lowers the calculated ambient ceiling to 82.55 °C. Select the actual margin from component requirements, uncertainty and reliability needs.

If the bare-wall branches contributed nothing, the same assumptions would produce 10 + 3.25 + 6 × 1.1 = 19.85 °C of junction rise. This demonstrates why branch structure matters.

Never add resistance values and watt-weighted temperature rises together. The final result above is a temperature rise in °C, not a resistance in °C/W.

Why an air gap is not a fixed resistor

A distance between the processor heat sink and the wall does not define an air resistance. Pure conduction also requires a defined cross-sectional area and suitable boundary conditions.

For illustration, consider a 3 in [76 mm] air layer with a 4 × 4 in [102 × 102 mm] cross-section. Using air conductivity of approximately 0.026 W/(m·K) near 25 °C:

Rconduction = L / (k × A) ≈ 284 °C/W

That is a hypothetical conduction-only slab calculation. It is not the resistance of an open space inside a working enclosure.

Real enclosure air circulates. Hot surfaces create plumes, cool surfaces create descending flows, and radiation transfers heat directly between surfaces. These mechanisms do not form a simple series resistor equal to the physical separation.

Do not add an arbitrary gap resistance after a sink-to-air resistance. Instead, establish the local air temperature around the component and its relationship to the collector inlet air. A large gradient may require several air nodes rather than one bulk-air node.

Use the guidance on component layout inside a sealed enclosure to avoid placing the processor in another device's hot plume.

What a thermal rating establishes and what it does not

A heat-sink data sheet establishes performance under its stated conditions. It does not automatically establish performance inside a sealed enclosure with blocked fin entrances or recirculating hot air.

Check whether the quoted resistance includes radiation. Also check airflow direction, fin orientation, heat-source footprint and the temperature reference used for ambient.

For the component, use thermal parameters appropriate to its actual mounting. Do not add a published junction-to-ambient resistance to junction-to-case resistance. They describe different thermal relationships and can double-count portions of the heat path.

Enclosure environmental ratings answer a separate question. Our stainless wall-mount enclosures are rated NEMA 4X / IP66 by the manufacturer; not UL listed. Those ratings do not establish processor junction temperature or validate an added heat-sink assembly.

For thermal qualification, check the completed assembly under representative worst-case conditions:

• Run the maximum credible simultaneous component losses until temperatures stabilize.
• Reproduce the installed orientation, surrounding clearances and external airflow.
• Measure local air, wall and heat-sink base temperatures at the modeled nodes.
• Determine junction temperature using the component manufacturer's documented method.
• Repeat at the intended hot ambient, with solar loading included when applicable.
• Check fan-failure conditions where continued operation is required.

A claimed operating range of −40 to 85 °C also requires separate cold-start validation. Power supplies, interfaces, fans, seals and other components must support their actual local temperatures.

What the model changes in a real panel

The worked example gives an immediate design priority: reducing the processor's local 10 °C/W path saves more junction temperature per unit resistance than improving its internal collector path alone.

The shared external path also deserves attention. With 6 W entering the wall, reducing Rout by 0.1 °C/W lowers the modeled wall temperature by 0.6 °C.

An internal fan can improve air circulation and fin heat transfer. It does not eliminate the external rejection requirement, and its electrical losses add heat. The article on internal fans in a sealed enclosure explains that distinction.

Stainless walls need particular attention at small heat-source footprints. Local spreading resistance can invalidate an assumption that the entire enclosure wall has one temperature. Create separate collector, device-attachment and external-sink wall nodes when measurements show meaningful gradients.

Directly attached devices can also warm the internal air through exposed hot surfaces. Assign that portion of their heat to Qa, rather than assuming all their power enters the wall directly.

If passive rejection cannot meet the hot-ambient requirement, compare other enclosure cooling methods. A passive heat sink cannot maintain an internally heated component below external ambient without another colder heat sink or active refrigeration.

What to specify for the enclosure

Use the stainless enclosure size finder for available enclosure dimensions. Reserve space for fin clearance, cable routing and access before selecting the outside size.

For custom enclosures, specify the wall and attachment geometry needed by the thermal model. We build custom enclosures with customer-defined material, width, height, depth and cutouts, from one piece.

Include these details in the enclosure specification:

• Maximum external ambient, installation orientation and expected external airflow.
• Component losses, identifying loads entering air and loads conducted directly to the wall.
• Required junction limits, temperature margin and shutdown strategy.
• Heat-sink footprints, masses, mounting loads and fin-clearance envelopes.
• Material grade, wall construction and any required spreading plates.
• Interface material, contact area, fastener arrangement and assembly requirements.
• Penetration locations, sealing details and required environmental rating after assembly.
• Temperature measurement locations and thermal acceptance conditions.

Do not treat thermal paste as a water seal or assume attachment screws preserve the enclosure rating. Resolve sealing and thermal contact together on the drawing. Keep assumed resistances marked as assumptions until the installed assembly supports them.

Enclosures for this job

Catalog sizes that suit the job this article answers, picked from its subject; open one for its full specifications.

Questions