Are Wireless Devices Too Hot To Handle?

June 1, 2004
Merely Increasing A Cabinet's Fan Speed Won't Always Cool Today's High-Density, Wireless-Equipment Enclosures.

Cooling electronic enclosures was once a fairly straightforward job. First, one had to calculate the maximum heat that was dissipated in the enclosure. The next step was to calculate the maximum allowable temperature rise above the maximum expected ambient temperature. Using the cooling equation, one would then calculate the airflow that was required to remove the heat and stay within the allowable temperature rise:

The fourth step was to calculate (or guess) the backpressure that the calculated airflow would develop as it flowed through the enclosure. Lastly, one had to select a fan or blower that would provide the required airflow against the expected backpressure.

This method worked okay as long as the heat-generating components were separated from each other and airflow was relatively unrestricted. Today's electronic enclosures and card cages present a much greater challenge: Circuit cards generate more heat. Plus, they are mounted closer together.

For an example of this trend, look to the recent introduction of the packet-switched backplane (PSB or switch-fabric) architecture for CompactPCI. A CompactPCI switch-fabric system might consist of more than 20 cards. One or two of those 20+ cards are switching cards. Each card can dissipate more than 70 W. Even worse, these systems are often installed in environments with maximum ambient temperatures that are far above 25°C.

The total amount of air that's flowing through an enclosure may seem adequate. In addition, the temperature of the air that's leaving the enclosure can be about the expected value. But it's still unlikely that the cards will be cooled properly.

The reason for this inadequate cooling is poor airflow distribution. The air that's leaving a fan isn't all moving at the same speed. Instead, it tends to form a rotating ring downstream from the fan blades. The area just behind the hub, for instance, is barely moving (FIG. 1).

It was once thought that with distance from the fan, the airflow would become more uniform. People also believed that the turbulence that was created when the air blast struck the card cage would even things out. Experience, however, has shown that neither of these things actually occur. As the air passes through or over the card cage, it still flows faster in some areas than in others. In some places, it stops or actually reverses direction (FIG. 2). This reversal can lead to hot spots and component failure.

The result of these effects can be surprising (FIG. 3). In this application, the maximum device temperature changes by 20°C as the card is moved from slot to slot. The life of electronics, as estimated using the Arrhenius function, is doubled by operating at 10°C cooler. Of course, that lifetime also is halved by operating 10°C hotter. Essentially, the life of a circuit board can be doubled simply based on its slot location in the card cage and the resulting air-velocity profile.

Simply installing a more powerful fan isn't the answer. That fan will increase noise and power requirements. It also may fail to get air to the hot spots. In addition, remember that backpressure increases as the square of the velocity. The power that's required to run the fan therefore increases as the cube of the velocity. Plus, the fan's own heat adds to the total system heat load. Clearly, the industry desperately needs a way to make the airflow uniform as it passes over the cards. It might even settle for a method of directing faster-moving air to the spots with greatest dissipation.

WHAT CAN BE DONE? Thankfully, there are several ways to ensure proper cooling in densely packed card cages. One approach is to tune the system: First, experiment with internal baffles. Then, build plenums or add internal fans and ducts to direct air to where it's needed. The only way to tell if this method works is through measurements.

The measurement methods that can be used range from placing an air-velocity probe (Pitot-type or hot-wire anemometer) at various places throughout the card cage to flow-visualization methods like smoke streams or even laser Doppler velocimetry. A possible indirect method might be to measure the temperatures at numerous points on the cards. For this approach, one can use temperature probes; surface-applied indicator materials that brush, rub, or stick on and change appearance at known temperatures; or infrared thermal-imaging systems.

Another option is to consult with the enclosure manufacturer. Some enclosure makers provide cooling-system design services that can be of great help in keeping things cool. Make sure, however, that the design services that are being purchased are worth having. To eliminate hot spots, some enclosure-cooling design services use proprietary (generally patented) methods for ensuring uniform airflow. Others do little more than simply calculate the overall heat load and required airflow. Those results may seem impressive when, for example, cooling assemblies are specified all the way down to the part numbers of the individual components. To make sure that the air gets where it's needed, though, it's a good idea to check what steps they take.

Some methods are based on computational-fluid-dynamics (CFD) modeling. As long as they're detailed (i.e., expensive) enough, they're capable of good accuracy. But it's important to remember that their results are only as good as their input values, which can be only guesses. Still other approaches use proprietary methods that run faster.

Make sure that the design service uses a mathematical model of the actual system chassis. Results should be dependent on the total heat load as well as the shapes and locations of the specific cards used in the system. In addition, check that the cooling design isn't so general that it ignores the details. At the same time, make sure that the solutions that it generates are sufficiently robust. A seemingly minor change in board placement or configuration should cause a major change in temperature profiles.

A recent advance in the cooling area is Advanced Vector Controlled Air Flow (AVCAF). Raytheon developed this technology for rigorous military aircraft and marine applications. Essentially, AVCAF controls airflow through the use of a specially designed fan shroud and perforated baffles, which are inserted between the fan and card cage (FIG. 4).

Normally, those baffles are designed to provide a uniform flow of air across and along each slot in the backplane. They can be modified, however, to concentrate airflow over specific portions of the system. While conventional wisdom might indicate that this arrangement would be more restrictive to airflow, the actual result is greater flow uniformity with very little additional backpressure.

Obviously, modern methods can make the job of cooling much easier. In general, however, today's high-density equipment makes cooling design a greater challenge than it was in the past. Going forward, keep in mind that all cooling methods aren't created equal. Their effectiveness and expense both vary. So choose well and know what you're doing.

Sponsored Recommendations


To join the conversation, and become an exclusive member of Electronic Design, create an account today!