Heatsinks are devices which increase the heat transfer from a component by increasing
the surface area associated with the transfer process. There are a number of parameters
which will impact heatsink performance such as material selection, surface treatment,
surface area, number and spacing of fins, thermal interface material, air speed,
and air bypass.
A typical heatsink is shown in figure 1. The following discussion
covers some of the issues faced by the designer when defining an appropriate heatsink.
Figure 1: Typical heatsink
The heat transfer process is a function of the surface area of the heatsink
as defined by equation: Q = hA∆T, where Q is the amount of heat being extracted,
h is the heat transfer coefficient, A is the surface area of the heatsink, and ∆T
is the temperature difference between the heatsink and ambient temperature. In general
terms it is easy to see that as the surface area of the heatsink is increased, the
temperature difference between the heatsink and ambient is decreased (assuming constant
h). Hence we can generally say that increasing the surface area of the heatsink
is a good thing, although there is a limit how much we can increase surface area
since other parameters will come into play to reduce the heatsinks effectiveness.
These parameters include heat spreading, fin efficiency, and air bypass, which will
be covered later.
Material selection and heat spreading
Material selection is also an important aspect
of heatsink design. The thermal conductivity of the heatsink material plays an important
role in how well the heat is conducted to the end of the fins. If we imagine a heatsink
with very long fins, then it is clear that such a heatsink would have a large temperature
from fin base to fin tip. The greater this temperature difference is, the less efficient
the heatsink will be. Hence under certain circumstances, it may be critical to design
a heatsink with a very high thermal conductivity such as Copper or even Silver. Another
scenario where this is true as well is when we are looking to spread the heat out
from a small source to a much larger heatsink. The greater the thermal conductivity
of the material the less the spreading resistance of the heatsink will be and the
better it will perform.
Surface treatment is important but mainly for cooling in natural
convection or without the use of fans. Typically, the influence of radiation is minimal
at high air speeds such as generated with the use of fans. However, in natural convection
where the air speeds are low, the impact of radiation heat transfer becomes important
as compared with the convective heat transfer. For this reason, applying a high emissivity
coating on a heatsink in a natural convection application will increase the amount
of heat transfer from the component and further help reduce component temperatures.
Air speed is an important factor in heatsink performance. Generally, a heatsink
will perform better as the airflow is increased, since the higher speed generates
greater heat transfer coefficients along the heatsink fins. Typical heatsink performance
as a function of air speed is shown in figure 2 for a ducted heatsink. The curve
is typical, although the extent the performance improvement will depend on whether
a number of heatsink characteristics such as ducted/unducted, and how tall the fins
Figure 2: Typical heatsink performance vs airspeed
The number of fins is also an important factor in the design of heatsinks. One would
think that having more fins is always beneficial as we are increasing the surface
area of the heatsink, however, this is not always the case. As the number of fins
is increased the spacing between the fins is reduced. As a result, the air flow between
the fins will be reduced, leading to a lower effective heat transfer coefficient
h. In our equation Q=hA∆T, if h reduces at a faster rate than the increase in A (surface
area), then the heatsink will see reduced performance. A typical heatsink performance
graph looks as shown in figure 3 for a non-ducted heatsink application. Clearly there
is an optimum number of fins for this specific design.
Figure 3: Typical unducted heatsink performance
Air bypass is another important factor and is closely related to item 3 above regarding
number of fins and fin spacing. As the air approaches a heatsink, the lower resistance
route is to flow over and around the heatsink rather than right through the fins.
This is due to the fact that the fin blockage and boundary layer growth between the
fins represent a greater pressure drop than simply flowing over and around the heatsink.
As a result, the heat transfer coefficient associated with the heatsink is lower
than if we had full airflow. Very often heatsink performance is quoted based on “ducted”
airflow or air velocity. What this means is that the data is obtained with a duct
or cowling which forces all the air through the fins. Unless the heatsink application
is also ducted, we are not likely to see that level of performance in an actual heatsink
sitting on top of an IC on a large board. It is therefore important to understand
the application and the data that is provided by the heatsink manufacturer. Typically,
taller fins and tighter spacing generate more bypass. By comparison, a shallow (shorter)
heatsink will not suffer quite as much air bypass.
Ducted heatsinks are often used in cases where we want to ensure
that the air directed towards a heatsink does not bypass around it. Basically, this
consists of a heatsink and fan with a duct joining them, such that all the air being
moved by the fans is forced to flow through the heatsink. This is sometimes seen
in low aspect applications (ex. 1U server) where the height restrictions are important.
In this type of heatsink design, it is important to understand that the amount of
air being moved will be a function of the heatsink pressure drop. Since the pressure
drop that the fan will work against is mainly generated by the heatsink, the greater
the number of fins the lower the actual amount of air that will be blown by the fan.
The example shown in figure 4 demonstrates how the fan and heatsink performance curves
can be overlaid in order to obtain the approximate operating point of the fan in
a ducted configuration. In this example, the flow rate expected from the fan would
be approximately 9 CFM. We can use this information and the heatsink performance
data to establish whether this would result in sufficiently high temperature margins
for the IC being cooled.
Figure 4: Finding the operating point of a fan / ducted heatsink system
Thermal Interface material
The thermal interface material is another important factor
in heatsink design. The choice of thermal interface material is of particular importance
especially when the heat flux (W/cm2) is high. If we consider that Q = A/Rth*∆T,
we can then write: ∆T = Q*Rth/A. If we let Q = 10W, Rth = 2 C*cm2/W (as provided
from thermal interface material supplier), and A = 1 cm2 (heat flux of 10W/cm2),
it follows that ∆T will be 20C across the thermal interface material. If on the other
hand, the heat source is spread over a much larger surface (say 6 cm2 - heat flux
of 1.66 W/cm2), then the ∆T will only be 3.3C. It can be seen then that higher the
heat flux, the greater the impact of the thermal interface material will be. In the
case where we have a heat flux of 1.66 W/cm2, we can probably not expect more than
a few degrees reduction in IC temperature by optimizing the thermal interface material.
On the other hand, in the case where we have a heat flux of 10 W/cm2, it is very
possible that we can reduce IC temperatures 10C or more by using a better thermal
interface material. There are various types of thermal interface materials from thermal
paste, to gap fillers, thermal adhesive pads, and phase change materials. All have
their strengths and weaknesses depending on the application in question.
If you have any questions regarding the above or for a specific design question which
you may have, please contact us and we will be pleased to help you.