Heat sinks, Part 2: Available components and performance

Part 1 of this FAQ looked at the basic issues associated with cooling and heat sinks. Part 2 focuses on the various heat sinks available.

Q: Are heat sinks limited to a single component such as an IC?

A: No, there are heat sinks which fit entire unit or chassis cool such as those which stretch across an entire enclosure or rack (Figure 1).

Fig 1: Heat sinks range in size from fingernail size to the full length of a large chassis (here, a high-power audio amplifier), and even larger in some specialty situations. (Image source: Takachi Takachi Electronics Enclosure Co, Ltd)

Q: A simple but obvious question: what is used as the material for heat sinks?

A: They are usually made of aluminum, but sometimes copper is used. While copper has better thermal conductivity than aluminum, it is heavier, more costly, and cannot be extruded (a low-cost and common way to make fabricate many heat sinks). Aluminum has favorable thermal, fabrication, weight, and other desirable attributes which make the preferred choice over copper in most cases

Q: How is the heat sink “attached” to the component it is cooling?

A: The answer, of course, depends on the heat sink style. For most IC heat sinks, there is a flat area which rests in the flat top of the IC or module package to be cooled. The heat sink may be fastened with small screws, a clip, a support strap, or even a single small screw directly to a tab on the component (if it has a thermal tab, as do many including this TO-220 package) (Figure 2).

Fig 2: The heat sink for a small, discrete TO-220 power transistor can be a simple extrusion (or even folded sheet metal) attached by a single screw to a integral tab; if the power dissipation is low, the tab alone may be sufficient. (Image source: Addicore.com)

Q: So, all it takes is to just attach the heat sink?

A: No, it requires more than simple placement. The heat sink must be in good thermal contact with the package, which requires an interposed thermal interface. This can be a very thin layer of special non-adhesive paste (often called thermal grease), or a special elastomeric pad, among other choices. The objective to have as little thermal resistance as possible between the IC or module and the heatsink, without even microscopic voids (they are thermal resistors).

Q: What are the sizes and styles of heat sinks?

A: They can be as small as a fingernail and as large as a chassis in a rack or an entire rack. The most common heat sink is an array of fins or pin, BFigure 3); the function of these “obstacles” is to interfere with airflow such that their heat can transfer from its solid surfaces to the surrounding air. These heatsinks are available in countless sizes, with many pin/fin configurations and shapes to meet the somewhat conflicting goals maximizing surface-to-air area while not excessively impeding the critical air flow.

Fig 3: This finned heat sink is designed specifically to fit on top of industry-standard “brick” DC/DC power supplies, with screw holes which match corresponding threaded inserts in the brick corners. (Image source: Advanced Thermal Solutions)

Q: What other types of heat sinks are available?

A: Heat sinks are available from dozens of vendors. Some are simple “wings” which clip onto a discrete device such as a TO-5 package transistor BFigure 4), which is a simple, folded, low-cost sheet metal stamping. There are also the multi-finned extrusions shown previously; finned devices but with bonded fins (more costly than a simple extrusion, but used when an extrusion cannot provide the desired greater fin height-to-gap aspect ratio; and cast or forged versions.

Fig 4: This heat sink for a TO-5 “canned” metal-package transistor package is a stamped strip of sheet metal which has been folded into a circle with “wings” and blackened to enhance emissivity; it is low cost and effective. (Image source: apprize.info)

For unique applications such as a spacecraft, the heat sink may be a custom casting or machined to fit the component and the available geometry of the area, obviously at far greater cost and lead time. Some very high-volume applications such as autos also use custom heat sinks to optimize both function and fit but fabricated using low-cost techniques.

Q: Is the necessary thermal analysis difficult?

A: As with most engineering issues, the answer is both yes and no. For the “no” part, basic thermal analysis is fairly straightforward, with clear equations that can be worked by “hand” using a basic spreadsheet or simple modeling program. These allow a rough estimate of the heat sink situation and challenge, often to within 10 or 20 percent of the final figure.

The “yes” part is where the heat sink must be considered as part of the overall board, enclosure, or chassis. Here, a thermal model of the overall system must be created and analyzed using thermal and fluid-flow equations. This model can be relatively simple or fairly complicated, depending on the mechanical and thermal design complexity as well as the fidelity and accuracy desired. It must take into account the airflow path, airflow “shadowing” of the heat sink by nearby larger components (even if they are cool), and the dissipation nearby components even if not blocking airflow.

Q: Do all ICs need a heat sink?

A: No, most ICs by themselves do not as long as the ambient temperature is within limits, as their own dissipation is negligible. Some ICs having modest dissipation have a thermal (metal) pad on their top, to reduce the thermal resistance to the extremal heat sink. Finally, some ICs (such as power regulators) have a thermal pad undeneath, but this is not intended to act as a heat sink. Instead, it is used to provide a thermal path between the IC and the copper tracks of the PC board which are routed underneath the IC. These tracks then function as conduits for the IC heat to flow to other copper areas of the PC board which then act as a “remote” heat sink.

Q: Given the thousands of basic heats available and their many variations, how do you make the choice as as to which how much heat sink is needed and which type? Is this difficult?

A: It used to be, but now, thermal modeling and simulation tools are available from many vendors which it a reasonably straightforward process. Many engineers start with a quick, rough analysis to get a sense of how much heat-sink performance they need and use that as a starting point.

Q: What about vendor support?

Vendors of heat sinks provide thermal and mechanical models of their heat sinks. These are used with the overall simulation to analyze if the heat sinking is physically compatible and provides the needed thermal performance in the application. This thermal analysis creates a thermal map of the component, PC board or product to assess performance (Figure 5).

Fig 5: A simulation program takes the thermal and mechanical model of a board (or box, module, or IC) and shows its 2-D temperature profile along with specific readings. (Image source: Quickfield/Tera Analysis Ltd.)

This FAQ has explored some of the many aspects of heat sinks and their role in the thermal management of an IC, board, or module. As with most simple-looking issues, their reality is both simple and complex at the same time. The heat sink is a passive, single-piece component without moving parts, but choosing the right one in terms of function, performance, cost, and size requires consideration and study.

EE World References

References