Heat pipes are a well-known thermal management solution used in computing applications. Nearly all laptop computers contain some number of heat pipes to effectively remove heat from heat generating components like the CPU and GPU.

The heat pipes move heat from the CPU and/or GPU to the frame of the computer or a fin stack for heat rejection to the ambient. Naturally, heat pipes have made their way into high- power embedded computing systems. Modular embedded computing systems often utilize modular circuit card assemblies mounted to aluminum carrier cards. These CCA carrier card assemblies are then mounted into one of several slots in a chassis.

Figure 1 shows a small embedded computing chassis with a conduction card assembly mid-installation. The chassis provides structural mounting for the card assemblies as well as a method to remove heat from the heat generating board components via conduction through the aluminum card and into the chassis. This improved heat removal by conduction allows these systems to operate at higher powers. More recently, the addition of heat pipes to conduction cards has further expanded the power capabilities of these systems. Similarly, embedded heat pipes within the chassis spread heat for more effective heat removal by natural or forced convection from the exterior of the chassis. The embedded heat pipes can also be used to transport the heat to a distant area for heat rejection such as liquid cooling the chassis from a single side.

An embedded computing system, like any system dissipating heat, can be analyzed at a high level using a thermal resistance network. The thermal resistance network is analogous to a simple DC electrical resistive network where thermal resistance, Rt, is analogous to electrical resistance, Re; temperature gradient, ΔT, is analogous to electrical potential (voltage), V; and heat transfer rate, Q, is analogous to electric current, I. Each is therefore governed by a linear equation known as Newton’s law of cooling for thermal networks (Equation 1), and Ohm’s law for electrical networks ( Equation 2).

Equation 1: Newton's Law of Cooling

Equation 2: Ohm's Law

Each major component of the embedded computing system: conduction card, card retainer (wedgelock), and chassis, can be represented by a thermal resistance or combination of resistances as shown in Figure 2. Each of these components resists the flow of heat from the heat generating component through the conduction card and into the chassis. The final resistance considered in this thermal network is the thermal resistance of heat transfer to the final heat sink.

Generally, embedded computing systems reject heat to the ambient with fins machined into the outer surfaces of the chassis or with a liquid cold plate bolted to an outer surface of the chassis. Either method requires the heat be “dumped” to the ambient air. As evident in Newton’s law of cooling, any reduction in these resistances will reduce the temperature gradient between the ambient and heat generating components. Likewise, in order to remove more heat (increase heat transfer rate) and maintain the same temperature gradient the thermal resistances must be reduced.

Heat pipes are a very effective method of reducing thermal resistance. Heat pipes are essentially high thermally conductive cylinders that, when embedded in aluminum structural components, increase the bulk or effective thermal conductivity of the base aluminum material. As evident in Equation 3 for conductive thermal resistance, increasing the thermal conductivity decreases thermal resistance. Figure 3 demonstrates the improvement attributed to the implementation of heat pipes within an aluminum conduction card. A reduction in maximum temperature of almost 20°C is realized by embedding heat pipes in this design. Similarly, Figure 4 shows the improvement attributed to the addition of heat pipes to the chassis wall utilizing a bottom rail cooling method. Here an improvement of over 50°C is realized by the addition of the heat pipes.

Equation 3: Conductive thermal resistance

ACT has branded this heat pipe embedded thermal solution the HiK™ (high conductivity) technology. One of the simplest methods for evaluating the benefit of a HiK™ solution is to model the component utilizing an increased thermal conductivity for the base material that contains the heat pipes. Experience has shown an effective thermal conductivity in the 600-800 W/m-K is easily achieved by HiK™ solutions.

Heat pipes have been effective in increasing the power capabilities of embedded computing systems. Current and future efforts to further increase capabilities will focus on reducing the thermal resistance of the card retainers and/or alternate heat transfer technologies such as vapor chambers, pump liquid cooling, and pumped two-phase cooling.