Electrical issues traditionally take precedence in the design of notebook computers. But, as system functionality per square inch shoots skyward, thermal issues are now moving closer to the forefront. With heat-generating microprocessors and other electronic components being packed into a limited space with restricted airflow, heat inevitably becomes a problem. It can cause anything from erratic behavior to complete system failure.
To date, methods to alleviate thermal effects range from an extensively modeled, optimum component layout that minimizes system cost to a sophisticated high-power fan with speed control to maximize battery life. Since every notebook is different, however, a unique thermal solution exists for each system. Heatsink manufacturers, thermal-modeling software providers, and others active in thermal issues can assist the designer in this complex area. In addition, the global requirements for notebooks may suggest the use of a thermal consultant with global expertise and support. Whatever method is chosen, space constraints, battery life, weight, and system cost all demand that the thermal issues be addressed during the early phases of system design.
Thermal design in notebooks is particularly challenging because system heat tends to increase the temperature of all components—even those that generate little heat themselves. The heat-removal rate is driven by the temperature differential between the inside and the outside of the notebook. The chassis itself can dissipate only a limited amount of heat, as it must remain comfortable to the touch. Furthermore, long battery life is a selling feature for notebooks, so optimal solutions must minimize the draw on the battery. Portability also is necessary, so the thermal design must not add significant weight to the package. To add to this issue's complexity, notebooks restrict the maximum height of the thermal solution to under one inch.
Heat from low-end notebooks can be eliminated with proper component placement and inexpensive heatsinks. Yet the high-end notebooks, with their greater power draw, require high-performance fan systems as well as strategic vent placement to optimize airflow. To preserve battery life, the smallest fan possible must be used. This minimal airflow requires that the heatsinks and pipes conform to tight operational tolerances. Ease of installation also is important. Attachment of the heatsink and heat pipe usually occurs after the CPU has been installed. These thermal components must align with both the CPU and the chassis, so they must also conform to tight physical tolerances.
Heat pipes and heat plates: Solutions for low-end notebooks can be very cost-effective. Here, the primary heat generator is the CPU. Simply keeping the CPU away from the hard drive maintains hard-drive reliability at essentially no cost. The heat from the CPU can then be piped out using a 3-mm-diameter heat pipe that's connected to a stamped aluminum plate on the back of the keyboard. The aluminum plate can typically absorb 2 to 4 W without significant temperature rise. The heat is then dissipated through the keypad.
A straight, round heat pipe can dissipate up to 15 W. Maintaining the low profile of the more expensive notebooks requires more intricate designs. These designs increase the need to bend and flatten the heat pipe, decreasing the amount of heat dissipated. Adding 90° bends or flattening the pipe can reduce effectiveness to as little as 6 W. These low-profile notebooks require additional thermal solutions.
Heatsinks: For low-end to mid-range notebooks, small, inexpensive extruded heatsinks provide sufficient additional cooling when the heat pipe cannot carry the whole load. Finned heatsinks dissipate the most power in the smallest amount of space, incrementally increasing the ability of the designer to dissipate heat passively. This is a low-cost solution. But for some applications, more fin density may be needed. This, in turn, requires more space. Folding the fin increases the surface area available without increasing the space occupied. The folded-fin heatsinks are more expensive than the simple extruded variety. Still, the amount this adds to the cost of manufacturing is more than compensated for by the increase in the retail price of the higher-end notebook.
Airflow control: Properly placed vents provide airflow to remove the remainder of the heat from the system. As the notebook's complexity increases, simple fans can raise the airflow, thereby improving heat transfer out of the system. A variable-speed control fan is an innovative solution to battery life limitations, offering additional margin on high-end systems. Whenever the user accesses high levels of computing capacity, and the CPU is generating the maximum amount of heat, this type of high-performance fan operates at maximum speed. When computing levels decrease, the fan slows down to conserve battery power. This solution also reduces acoustic noise, since the noise associated with the fan is proportional to fan speed. Though these fans are approximately 10% more expensive than the conventional ones, the extension of the battery life increases the marketability of the notebook, justifying the added cost.
Thermal materials: Pliable, compressible thermal-interface materials can also play a role in keeping notebooks cool. The aluminum plate is located on top of the chassis under the keypad, while the IC is mounted to the bottom of the chassis. Tolerance stack-up between the CPU and the chassis creates a variable gap in the thermal-conduction path. By compressing gap-filling, compliant, interface materials between the CPU and the chassis-mounted heatsink, thermally conductive material replaces the air gaps. Doing this ensures good thermal contact between the CPU and the heatsink. Even though the compliant interface material isn't as thermally conductive as the commonly used 5-mil-thick pad, it also isn't subject to the tolerance mismatch that might occur with that pad. If the pad is too thin for the space, there will be no contact. If it's too thick, it will put pressure on the IC.
A custom cooling solution for one high-end notebook design is shown in Figure 1. The large, stamped plate connects directly to the keyboard, providing 2 to 4 W of passive cooling. Though not shown, the gap-filling thermal-interface material creates a thermal pathway from the CPU to both the stamped plate and the heat pipe. The heat pipe conducts heat to a low-cost heat exchanger, where directional airflow from the custom, low-profile fan can take it out of the system. Efficient component placement minimizes cost.
Modeling: With the addition of heat-generating ICs other than the CPU, the thermal issues become much more involved. Thermal solutions for these more complex notebooks typically cost two to three times that of a low-end notebook. The standard, 3-mm heat pipe isn't large enough to dissipate the heat generated in these systems. By using a thermal modeling package, such as Icepak from Fluent Inc., Lebanon, N.H., the designer can optimize board layout to take advantage of the natural thermal convection and directional airflow provided by a fan.
Thermal modeling can predict the heat generated by each component in a particular system and model the effectiveness of the thermal solution. Such programs are capable of taking existing CAD files and determining the airflow direction and velocity at any point in the notebook. They also calculate thermal resistance, offering an indication of thermal transfer and cooling.
The same tools can help develop an optimal heatsink for the notebook. The heatsink design can be modeled in terms of volume resistance. Then, the modeling package can analyze each face of the heatsink for the free-area ratio characterizing the amount of air that can flow past the face. Together with the thermal resistance calculations, volume resistance is a powerful tool to optimize heatsink design for a specific application in the shortest amount of time.
By modeling different low-cost and passive thermal solutions, the designer can reduce the overall cost of the thermal solution. Without modeling, the designer will need to prototype the system and try "what-if" scenarios in hardware. Often, "what-if" scenarios are difficult or impossible to cover with a single prototype. So, additional prototypes must be built and tested. Although modeling increases up-front design time, it allows for multiple "what-if" analyses—typically reducing system test time and often eliminating the need for redesign.
In systems with multiple heat generators, designing a heatsink to address several hot components in close vicinity may minimize the height of the heatsink itself. Now, the layout and heatsink design phase must occur simultaneously. The complexity of the thermodynamics with two heat generators makes thermal modeling of the new heatsink imperative. Without it, the design-cycle time increases. Much more time is spent in prototype testing. It's also easy to miss a critical heat-generating component without thermal modeling. Any shortcuts to these processes can result in missed data.
Figure 2 shows a layout in which the heatsink effectively cools the CPU. Yet it also shows significant heat transfer from the CPU to the hard drive. Unless the layout is modified, the hard drive will be subject to frequent failure due to the additional heat load.
In a layout that restricts airflow, there must be a strategy to move heat out. In the case of Figure 3, the CPU will suffer from reliability issues caused by excessive heat.
Figure 4 shows a strategy where the heatsinks pipe the heat to a localized area where active airflow can pick it up and eliminate it. The temperature chart shows that this strategy is effective for the chosen layout and component mix.
These examples display how thermal modeling identifies thermal problems before the hardware is built. Such modeling exemplifies how the lowest-cost solutions, such as layout and passive heatsinks, can be tested and compared with more expensive solutions to generate the most robust, reliable package.