Heat Pipe and Fan Cooling Design Guide

When a device dissipates more heat than a small copper area can spread, the answer is usually to move the heat somewhere else rather than to make the board bigger. A heat pipe does that without moving parts, and a fan does it by forcing air through a path that the enclosure defines. Together they can multiply the heat that a given volume can reject.

When Air Alone Is Not Enough

Natural convection from a board depends on the surface area, the orientation and the temperature difference to the surrounding air. Beyond a few watts in a closed enclosure, the required area grows faster than the space available, and the temperature rise becomes unacceptable no matter how the copper is arranged.

A heat pipe breaks that relationship by decoupling the area from the source. A small area at the device collects the heat, and a large area elsewhere rejects it, with the pipe carrying the heat between them at a temperature difference of only a few degrees. That allows the rejection area to be placed where the air is coolest, which is rarely where the hot device sits.

forced convection with a fan raises the surface heat transfer coefficient by an order of magnitude compared with natural convection, so the area required for a given dissipation falls by the same factor. The two techniques are usually combined: a heat pipe to move the heat and a fan to reject it.

How a Heat Pipe Moves Heat

The pipe contains a working fluid and a wick, sealed under partial vacuum. Heat at the evaporator section vaporises the fluid, the vapour travels to the condenser section where it gives up its latent heat, and the wick returns the condensate to the evaporator by capillary action. There are no moving parts and the cycle continues as long as the temperature difference exists.

The performance is characterised by the temperature difference between the two ends for a given heat load, which is equivalent to a thermal resistance that is usually much lower than that of a solid copper bar of the same dimensions. The limit is reached when the wick cannot return the condensate fast enough, a condition called dry out, which produces a sudden rise in thermal resistance.

Working fluid and wick type follow the application temperature. A water based pipe suits electronics, an ammonia pipe suits lower temperatures, and a methanol pipe is used where freezing is a risk. Using a pipe outside its intended range is a common mistake that produces either poor performance or a dry out at high power.

Mounting and Orientation Effects

Orientation affects the capillary limit, because gravity either helps or opposes the return of the condensate. A pipe with the evaporator below the condenser works against gravity and has the lowest limit, while one with the evaporator above works with gravity and has the highest. Horizontal operation sits between the two.

The performance figure in a datasheet is normally quoted for a specific orientation, and applying it to a different one produces a design that looks adequate on paper and fails in the product. Where the orientation cannot be guaranteed, the design should use the worst case figure and keep margin below it.

Mechanical mounting matters as much as orientation. The pipe must be clamped or bonded to the source with a defined pressure and a thermal interface material, and it must be supported so that vibration cannot fatigue the joint. Bending a pipe too sharply reduces the vapour path and can damage the wick, so the minimum bend radius from the manufacturer should be respected.

Fan Selection and Airflow Path

airflow path is the first thing to define, and it is usually constrained by the enclosure rather than by the board. Air enters at a low point, passes over the surfaces that need cooling and leaves at a high point, with the fan placed so that it does not work against natural convection. A path that forces the air through a narrow gap between two hot components cools neither of them well.

The fan is selected for the operating point on its curve rather than for its free air rating. A fan rated at fifty cubic feet per minute delivers far less when it has to push the air through a filter and a fin stack, and the actual operating point is the intersection of the fan curve and the system impedance curve. Measuring that impedance, or using a manufacturer tool, avoids a surprise.

Pressure drop and flow rate are traded against noise. Doubling the flow through the same restriction quadruples the pressure drop and increases the noise substantially, so reducing the restriction by rearranging the ducts often gives a better result than fitting a larger fan. The thermal design rules for the board itself are described in our guide to PCB thermal management.

Fan Control and Noise

fan control by pulse width modulation is common because it is efficient, but the switching frequency must be chosen carefully. Below about twenty kilohertz the modulation can be audible as a tone, and at the fan natural frequency it can excite a mechanical resonance in the blades. Frequencies above the audible range are preferable, provided the fan tolerates them.

Driving a fan from a modulated supply requires care with the return path. The fan current is pulsed, and a shared return with a sensor or an analogue circuit injects noise into it. The layout rules for a switching load, described in switching regulator layout, apply to the fan drive as much as to a converter.

A fan tachometer output is the usual way to detect a stalled or failed fan, and the firmware should treat the loss of tachometer pulses as a fault and reduce the load or shut down. Testing the response deliberately, by holding the blades or disconnecting the fan, confirms that the thermal design has a defined behaviour in that case rather than an unmanaged rise in temperature.

Dust, Filters and Long Term Performance

A fan that moves air also moves dust, which accumulates on fins, on filters and inside the enclosure. The thermal resistance rises as the passages become restricted, and a design that has no margin will overheat after a year in a dusty environment even though it was comfortable when new.

Filters protect the fins but add pressure drop and require maintenance. Where the equipment cannot be serviced, the design should use a filter with a large area so that the pressure drop rises slowly as it loads, or should be designed with enough thermal margin to tolerate a restricted path. The maintenance interval should be part of the specification rather than an afterthought.

Fans themselves wear out. The bearing is the limiting component, and its life is quoted at a temperature that may be much lower than the air inside a warm enclosure. Using a fan rated for the actual internal temperature, and making it replaceable, is worth more than a slightly lower specification on the heat sink. The cost decisions involved are the same as those in manufacturing cost reduction.

Layout and Mechanical Integration

The heat pipe and the fan are mechanical parts that interact with the board, so the layout has to leave space for both and for the airflow between them. Fins placed immediately behind a board edge have no air path, and a fan that discharges into a flat surface recirculates its own exhaust.

Components that are sensitive to temperature, such as electrolytic capacitors and crystals, should be placed upstream of the hot components in the airflow path rather than downstream. The difference can be twenty degrees or more in a compact enclosure, and it costs nothing but a placement decision.

The assembly sequence matters for serviceability. A heat pipe that has to be removed to reach a socket underneath it makes every repair expensive, and a fan mounted behind the board cannot be replaced without removing the board. Reviewing the assembly from the point of view of a service technician catches most of these problems before the enclosure is tooled.

Verification and Common Faults

Verify the design with the enclosure closed, the fan running at its production setting and the load at its maximum. Measuring at the device case, at the air inlet and at the air outlet gives a picture of the whole path and shows which element is limiting: a small difference between the device and the fin stack points to the interface, while a large difference between the inlet and outlet air points to insufficient flow.

A device that is cool when the enclosure is open and hot when it is closed is a flow problem rather than a conduction problem. Checking the actual airflow with a vane anemometer, or simply feeling the exhaust, shows whether the fan is moving air in the intended direction or recirculating it.

Where the design has a temperature dependent behaviour, such as a fan that only starts above a threshold, the transition should be tested as well as the steady states. A fan that hunts on and off produces more noise and more wear than one that runs continuously at a low speed, and the hysteresis in the control loop is what prevents it.

FAQ

Can a heat pipe work against gravity? Yes, but its capacity is much lower. The datasheet figure applies to a specific orientation, so use the worst case value for the installation and keep margin below it.

How much margin should a fan cooled design have? Enough to tolerate a partially blocked filter and a worn fan. A design that is at its limit when clean has no allowance for the conditions it will actually see.

Is a larger fan always quieter? For the same airflow, usually yes, because it can run at a lower speed. The noise depends on the operating point rather than on the size alone.