Put in the operating point and get the sink-to-ambient resistance you actually need, with solar loading, installed airflow, and fan life at real internal temperature. The arithmetic is shown, so you can argue with it.
Specified at 25 °C. Running at 70 °C.
About 31% of bench capacity survives here: 20 K of headroom against 65 K on the bench.
Required sink-to-ambient resistance is 0.45 K/W. Natural convection will not reach it at this ambient; air has to be moved.
Costs youAir through the box means an intake, an intake means a filter, and a filter in this dust class clogs and then fails silently. Prefer a sealed internal loop moving air over an internal sink coupled to external fins, so no outside air ever enters.
The CFM on the box is a free-air figure with no grille, no filter and no fin stack. Specify against the second line, not the first.
First-order hand calculation for sizing and sanity checking, not a CFD result and not a guarantee of compliance. No transient, no board conduction, no radiation credit. Verify against vendor data and a measurement before committing a design.
Want the part, not the number?
Send the spec across and we will come back with a sourced or built solution for this operating point.
Effective ambient 70 degrees. Required sink-to-ambient resistance 0.45 kelvin per watt. Recommended class: Forced air, but the sealing conflict has to be resolved.
None of this is difficult. It is just the step that gets skipped, because a heatsink that looks big enough usually is, at 25 °C, in still clean air, on a bench.
Every passive path is proportional to the temperature difference it has to work with. At 48 C against a 90 C limit there is 42 K to spend. On a 25 C bench there was 65 K. Same heatsink, roughly a third less capacity. That is before dust and before sun.
An unventilated enclosure in direct sun sits well above air temperature. A dark absorptive box is the worst case and a shield is the cheapest fix anyone will ever buy. Design to the surface the electronics actually see, not to the forecast.
The airflow number on a fan assumes no grille, no filter and no fin stack, which describes no real product. Installed airflow lands well below the rating, and a filter that has to survive fine sand costs the most of all.
Bearing life is quoted at a bench temperature and roughly halves for every 7 to 10 K above it. A fan rated for 50,000 hours at 40 C is not a 50,000-hour fan inside a sealed box at 80 C. This is why units fail in their second summer.
The calculation picks one of these and tells you what it costs you. Cheaper is further up. Most Gulf outdoor work lands on the second.
Enough headroom, clean air, fins free to convect.
Dust present. No openings, no filter, nothing to clog.
Natural convection will not reach the required resistance.
Air cannot find the area at any practical size.
Short runs with a cool-down. Size thermal mass, not a sink.
The limit is reached before a heatsink is attached. Usually the wrong answer.
This is a first-order steady-state estimate for sizing and sanity checking. It has no transient model, no board conduction, no radiation credit, and no knowledge of your mechanical layout. It will tell you which class of solution you are in and roughly how much margin you have. It will not certify a design.
The solar and airflow figures are stated engineering estimates, not measurements. They are printed with the result so you can substitute your own. Vendor data and a thermocouple beat all of it.