Cree states white XLamp LEDs are over 40% efficient, with roughly 50-60% of input power leaving as heat. Not all of it appears at the junction: a share is generated in the phosphor. If that heat doesn't leave the chip, the chip dies. The aluminium housing isn't decoration, it's the lifeline.
Four ideas to carry into the next specification conversation you have.
LM-80 reports the LED case temperature Ts, not Tj. And there is no 10 °C halving rule for LEDs: that factor of two belongs to electrolytic capacitors. On one published module, L90/F10 falls from 38,000 h at tp 45 °C to 26,000 h at tp 75 °C, about 12% per 10 °C. Specify Ts or tp measured in the finished luminaire at your design ambient, never a maximum Tj.
1100-O ≈ 222 W/m·K, extruded 6063-T6 ≈ 201, die-cast A380 ≈ 96. Same metal, a bit over twice apart, and the low end is where most housings are made. Quote the temper: it moves conductivity the opposite way from intuition.
For a fixed alloy on a fixed base, geometry decides. Crowd the fins past the optimum and total dissipation falls even though fin area rises. Spacing 6 to 10 mm for natural convection, about 8 mm as a default; forced air is designed tighter because the fan supplies the pressure the buoyant plume cannot.
Thermal Interface Material between LED MCPCB and heatsink. In one bench test a dry joint measured 2.9 °C/W case to sink against 0.9 °C/W with grease, taking the whole junction-to-ambient path up 22%. Serious, but not everything. Specify resistance at a stated bond line, pressure and area, never bulk W/m·K.
Typical values near room temperature. Producer data is stated at 20 °C and the die-cast table prints no test temperature. The application column is NLC practice, not an industry rule.
| Alloy | Conductivity · process | Typical use |
|---|---|---|
| 1100 (pure) | 222 W/m·K · sheet, O temper | Non-structural formed parts and covers |
| 6063 (extruded) | 201 W/m·K · extruded, T6 | Extruded profiles and housings, including linear heatsink extrusions |
| 6061 (T6) | 167 W/m·K · machined, T6 | Custom heatsinks · brackets |
| A380 (die-cast) | 96 W/m·K · pressure die-cast | Highbay shell · outdoor housings |
| A360 (die cast) | 113 W/m·K · pressure die-cast | Premium downlight bodies |
| ADC12 (JIS H5302) | ~96 W/m·K · die-cast (illustrative) | Asian-market fixtures |
| Copper (Cu) | 391 W/m·K · C11000 ETP | PCB inserts · premium projects |
Five thermal-design ideas that separate good fixtures from great ones.
Heat travels junction → solder point → board dielectric → thermal interface → heatsink → ambient. Rth, also written R-theta, names ONE link; add them, then Tj = Ta + Rth j-a × P. A well-cooled fixture totals < 5 °C/W, a poor one > 12 °C/W. At 10 W that is a 50 °C rise against 120 °C, and on a 45 °C in-fixture ambient the second one lands past the 145-150 °C absolute maximum on typical emitter datasheets.
Conductivity and thermal resistance are different quantities: a conductivity ratio only holds if both layers are the same thickness, which no real board is. Compare insulance, R = t/k. On Cree's measured comparison the metal-core board wins: 3.81 °C/W against 8.40 °C/W for 1.6 mm FR-4 with 14 vias in 4 oz copper, and lower thermal resistance is better. A well-designed via array is still not automatically hopeless: in a SEPARATE Texas Instruments experiment on a different product, one particular thin FR-4 design was estimated at 3.76 °C/W against 4.23 °C/W for that specific IMS construction, whose dielectric was thicker than needed (TI reference design TIDA030). That is a different test, not part of Cree's figures. Cree does not recommend the via technique above 5 W per LED.
Natural convection wants vertically oriented fins at 6 to 10 mm spacing. The optimum is a formula, not a number: S_opt = 2.714 L / Ra_L^0.25, with L the fin length along the airflow. Fin HEIGHT is capped by fin efficiency, not by any ratio to spacing, and the payback is sublinear: doubling height buys about 60% more heat. Orientation matters through whether air can chimney through the channels; rotating an extrusion so the channels run horizontally is put at roughly 25% worse by one heat sink maker.
Bare aluminium is not one number: a fresh extrusion measured 0.02 in one study, handbooks put commercial sheet near 0.09, and highly polished aluminium is 0.04 to 0.06. A real anodic film reaches about 0.83 to 0.86, and the oxide does it, not the dye, so clear anodising radiates essentially as well as black. Radiation is not a rounding error: on a natural-convection heatsink with a high-emissivity finish it carried 28 to 30% of the total across 5 to 25 W in one worked example, and going bare cost close to 30 °C. Do not specify mirror-polished outdoor heatsinks. Inside a sealed housing the heatsink radiates to a hot inner wall, so the finish that governs heat rejection has moved outward to the housing skin.
Avoid modifying the enclosure where possible. If the applicable test method requires an access hole, keep it minimal, reseal and document it, and do not infer IP compliance from the modified sample. The tc point is in any case a location on a component case marked by the component maker, and the accepted attachment methods are mechanical clamping, minimal solder or minimal adhesive. Run the test in a draught-proof enclosure, not on an open bench, and hold it to thermal equilibrium rather than a stopwatch. Ask instead for the in-situ measured temperature at the LED maker's designated point, on the hottest LED, in that exact luminaire, at rated current and your design ambient, measured per ANSI/IES LM-98-24 and with the test report attached. Ask separately for the ANSI/IES LM-80-21 report: that one covers light output and colour maintenance, not temperature.
Every NLC outdoor and industrial fixture is thermally validated at Tj target before mass production, not just calculated.
A 14-slide explainer, ready to post straight to LinkedIn as an image carousel. Same content, faster reading.
14 slides · LinkedIn-ready. Download all 14 as high-res PNGs and post them straight to LinkedIn as a carousel. PDF version still available.
Producer datasheets, standards and manufacturer-published measurements, each fetched and checked, with the date it was checked and the edition it was read in.
Practical lighting know-how from NLC Academy. No spam, unsubscribe anytime.
By subscribing you agree to receive NLC emails.