Active vs Passive Cooling for Computers

Active vs passive cooling is the choice between cooling a computer with fans and cooling it with no moving parts: active cooling uses a powered device, usually a fan or blower, to force air over the hot components and heatsinks, while passive cooling removes heat by conduction into heatsinks, heat pipes and the chassis and then by natural convection and radiation into the surrounding air.

Both follow the same physics. Heat from the processor, graphics and power circuitry has to reach the air around the computer, and the difference lies in that last step. Forced air carries away far more heat from each square centimetre of fin than still air does, so active cooling handles more power in a given volume. Passive cooling gives up some of that in exchange for a sealed enclosure, silence and the removal of the part most likely to wear out.

In industrial computing, passively cooled PCs are the usual choice for cabinets, machines and vehicles. Active cooling comes in where the processor or GPU power is beyond what a sealed chassis can shed, or where the computer lives somewhere clean and easy to maintain. This guide compares the two side by side, explains the techniques behind each, works through how much heat a fanless chassis can handle, and sets out how to choose.

The quick answer, for computers in industrial use. Figures are typical; individual designs vary.

There is no single best answer. The right choice depends on how much sustained power the application needs, what the air around the computer contains and how easily someone can reach it to maintain a fan.

An actively cooled computer moves air through its enclosure with one or more fans. Axial fans push air straight through a heatsink or case; blowers, also called centrifugal fans, turn the air through 90 degrees and suit ducted designs and slim enclosures. A typical system has a fan on the processor heatsink, one or more case fans, a fan in the power supply and, where there is a graphics card, fans on the GPU.

Fan speed is usually controlled by temperature, so the fans spin up as the load rises. That gives a lot of cooling capacity when it is needed, at the cost of noise and of air drawn in from wherever the computer sits. Filters keep the worst of the dust out but need cleaning, and once fins clog the system runs hotter, which can lead to thermal throttling long before anything fails outright.

Liquid cooling is a form of active cooling too: a pump carries heat from a cold plate on the processor or GPU to a radiator, where fans pass it to the air. It suits very high power densities in servers and workstations, but adds a pump, fittings and coolant to the list of things that need attention, so it is rare in industrial edge computers.

A passively cooled computer turns its chassis into the heatsink. The processor sits against a copper block or thermal pad, heat pipes carry the heat to the case, and fins on the outside pass it to the air by natural convection and radiation. Our guide to how a fanless PC works traces that path step by step.

Conduction cooling is a close relative used in rugged and military systems. The chassis is bolted to a cold plate, a vehicle structure or a cabinet wall, and heat leaves through the mounting face rather than through fins in open air.

The performance ceiling of a passive design is set by the fin area, the heat path and the ambient temperature. The chassis designer sizes these for a maximum processor power at a maximum ambient, and the thermal mass of the metal absorbs short bursts above it. Mounting matters more than with a fan: vertical fins with space around them let warm air rise freely, while fins boxed in or laid flat cool less well.

A modern fanless computer uses several techniques together to move heat from small, hot components to a large, cool surface. Our FAQ on what a heat sink is covers the basic part.

Fin orientation is the free variable. Natural convection relies on warm air rising between the fins, so vertical fins with open space above and below cool best; the same chassis on its back can lose a large part of its capacity. Where the computer must go into a closed box, a flat-top design that uses the enclosure wall, described in our guide to electrical enclosure cooling, is usually the better answer.

Still air is a poor coolant. Natural convection typically moves around 5 to 10 W for each square metre of surface and each degree of temperature difference, while forced air over a heatsink can move several times that, roughly 25 to 250 W/(m²·K) depending on air speed. That ratio is why a fan lets a smaller box carry more power, and why fanless chassis are larger and finned.

Makers express the result as a rating: a maximum processor power at a maximum ambient. When a manufacturer offers the same computer with and without a fan, the effect is easy to see.

Two patterns stand out. Adding a fan buys roughly 15 °C of ambient headroom for the same processor, and the same chassis rated fanless at 35 W needs a fan to carry 65 W at a similar temperature. Our guide to processor TDP explains how the processor power classes are defined.

Fans are the part of a computer most likely to need attention. Their bearings have a finite life, which fan makers usually state as an L10 figure, the hours by which one in ten fans is expected to fail, at a stated temperature. Heat shortens it, and dust and fibres in the air wear bearings and clog the fins downstream.

In a server room with filtered air and scheduled maintenance that is easy to manage. On a factory floor, in a vehicle or at a remote site it is the part most likely to need a visit, which is why the advantages of fanless computers go beyond cooling. Fewer moving parts also improve the mean time between failures of the system.

Neither approach can cool a computer below the temperature of the air around it. A fan moves more air, but if that air is already hot, the margin between the air and the processor shrinks just as it does for a fanless chassis. In a hot location the answer is the same for both: choose a processor with headroom at the maximum ambient, and where that is not enough, cool the enclosure itself with a heat exchanger or cabinet air conditioning.

Cold behaves differently. A fanless chassis has nothing to stall or seize, which is one reason passive designs are common in outdoor and vehicle installations that start from well below freezing. Fans have to be rated for the low end of the range, because lubricants thicken and some bearings struggle to start in deep cold.

A passive chassis also runs warm to the touch under load, because its outer surface is doing the cooling. That is normal, but it is worth planning where the computer is mounted, keeping cables and plastics off the fins and leaving space for warm air to rise away from it.

Small ones can. Integrated graphics, NPUs and NVIDIA Jetson modules run fanless every day, and a low-power MXM GPU can be conduction-cooled in a rugged chassis. Our Jetson thermal design guide shows how module power modes are matched to a fanless case. A high-power discrete graphics card is different: at 70 W and above it usually needs more cooling than a compact sealed chassis can provide.

Some industrial GPU computers answer that with a split design. The processor and motherboard stay in a fanless, sealed section, and the GPU sits in a separate compartment with its own managed airflow, often with a fan tray that can be swapped without opening the main enclosure. Fan speed and health can be monitored by the system, so a failing fan is flagged before it causes a problem.

Other approaches include fanless chassis with an optional bolt-on fan kit for higher ambient temperatures or higher power modes, and rack-mount servers that use front-to-back airflow inside a filtered cabinet. In each case the aim is the same: keep fans only where the power genuinely needs them, and make them easy to service.

Our industrial GPU computers cover the split and active designs, and a fanless GPU computer built on a custom Jetson carrier board shows how far passive cooling can go with the right module.

Choose passive cooling when

• the air carries dust, fibres, oil mist or moisture, or the site is washed down;
• the computer is mounted on a machine or vehicle that vibrates;
• it is hard to reach, so every site visit costs time and money;
• silence matters, as in clinics, studios and offices;
• the processor the application needs sits within the chassis rating at your maximum ambient.

Choose active cooling when

• the sustained processor or GPU power is beyond what a fanless chassis can shed;
• the computer sits in a clean, filtered space such as a server room or control room;
• someone can clean filters and replace fans on a schedule;
• space is tight and the power cannot come down.

Consider a hybrid when

• a GPU needs fans but the rest of the system faces a harsh environment;
• a fanless system needs extra headroom only in the hottest weeks of the year, which an optional fan kit can provide.

Can someone reach this computer to clean a filter or change a fan? If the honest answer is no, start with passive cooling and size the processor to suit. Our guide to the benefits of fanless computers covers the wider case for going fanless, fanless mini PCs covers the smallest designs, and for weight- and power-limited platforms our FAQ on SWaP-C explains the trade-offs.

We specify the cooling as part of the whole system, from the processor and GPU to the chassis and the installation. Our range of fanless industrial computers covers most cabinet, machine and vehicle installations, and we configure, image and test every build before it ships.