|
The Thermal Territory Concept, part 2 |
|---|
Home page...................: Frigus Primore home page
Calculator....................:
Thermal territory for a component
Back ground article....:
The thermal territory concept, part 1
Figure 1
A component that has a large thermal territory may create
difficulties in the layout phase.
Introduction
The essence of the thermal territory method is to assign an area
around a component for its cooling. The size of this area, which
here is called a territory, depends on many factors. The heat
dissipated by the component is obviously the most essential one
but properties such as the thermal conductivity of the PCB, the
heat transfer coefficient and the available temperature
difference are also very important.
This, the second article about the issue, brings up some typical
applications. They are quite a few and several of them address
problems that thermal designers face almost daily. The method
therefore has somewhat of a working horse character. This does
not mean that it always yields the answers wanted but it always
clarify things and it often points in a direction where solutions
can be found.
The thermal territory method is by its nature approximate and can
therefore never replace more accurate calculations. As this
article hopefully will show it is nevertheless both simple to
apply and very valuable for front-end purposes. The latter aspect
is very important, because it is in that phase that most
critical decisions are taken.
Single component application
It is important to realise that the size of a thermal territory
not only depends on the properties of the component. The
sensitivity to other parameters, such as the local heat
transfer coefficient, can therefore result in large size
differences depending on where the component is placed on
the PCB. Figure 1 basically shows that a component potentially
could create layout problems if it is placed at the top of
the PCB but that there is a much better chance of success
if it is placed at the bottom.
When comparing these two placement options it could also be of
interest to note that the top placement dissipates 20% more of
its heat through the PCB but has a thermal territory size that
is 200% larger than for the bottom placement! The non-linear
character of thermal territory method essentially causes this
effect. The reverse side of this tendency is that only a modest
increase of the allowed temperature difference is needed to
radically decrease the territory for the top component. The
method therefore has a tendency to exaggerate differences. This
is both an advantage and a disadvantage but it is nevertheless
something important to be aware of.
Even thought a thermal territory assessment only is a first
design approach, it does provide very helpful information. A
straightforward use would be to estimate if the size of the
territory could be handled in the layout process. Applied
front-end design is however rarely that simple. The input data
is in most cases only preliminary and sometimes even based on
pure guesses. There are also often several alternatives to
examine. The most frequent use of the method is therefore as
help for PCB designers to focus on potential problems. They
might be real but they might also merely be the effect of poor
input data quality.
Large thermal territories are in the initial design phase often
caused by too conservative heat dissipation estimations. Excesses
on the 30% level are common and factor 2 errors appear
occasionally. Discrepancies on that level will of coarse have a
radical impact. The temperature criterion can have a similar
effect. Another important concern is the quality of the thermal
component model and not the least the thermal resistance between
the pads of the component and the inner layers of the PCB.
A typical characteristic for applications in this design phase is
therefore not that the territories are calculated once and for
all. They are rather re-assessed over and over again with
different input data. This nature of things requires the
software tool used must be smooth, fast and have the ability to
create overview diagrams. Used in this way the method definitely
has an important role to play between crude rules of the type
"no more than XXX W on any component that is smaller than
XXxXX mm", which often are found in design guidelines and
accurate finite element calculations.
Figure 2
A heat sink can decrease the thermal territory size radically.
This type of calculation also estimates the proportion of the
heat that is absorbed by the PCB.
Heat sink application
A heat sink is a device that usually increases the
component-to-air heat dissipation radically, particularly if the
air velocity is above 0.5 m/s. This fact must therefore also have
an impact on the thermal territory size, which also is the case,
figure 2.
A very frequent question to thermal experts is to what extent
the PCB con-tributes to the total heat dissipation when a heat
sink is used. It is true that there is no exact answer to this
question until the total layout environment is known. The thermal
territory method can however always provide a good
estimation.
A key issue in this context are the temperatures that can be
allowed on the chip and on the PCB. There are arguments that
limit both of these. It is evident that the chip temperature
not can exceed the safe function limit. It is equally evident
that the PCB temperature not can exceed a certain limit because
that would jeopardise the safe function of the surrounding
components, particularly the passive ones. Whether the actual
temperature limit is set by the chip or by the PCB depends on the
circumstances. The thermal territory method can nevertheless be
used for both cases.
This type of application is very convenient for component
manufacturers that need to know if a component requires a heat
sink or not. If so, they also need to know for what
circumstances. Not knowing the environment is of coarse always a
great handicap for these considerations. They must therefore
cover a multitude of options and also endure the fact that there
always will be grey zones. The thermal territory method is ideal
for exposing this complexity. Alternative methods are in this
case neither more accurate nor smoother.
Another typical application is custom circuit design. If the
package used not can be combined with a heat sink and if the
heat dissipation is on a critical level, there are great risks
involved, see: Not a success story.
Figure 3
The thermal territory method can be used for PCB level estimates
once a component list is available.
Front-end application
The method can also be use for estimates on the PCB level. In the
simplest case this can be done on the bases of a component list,
figure 3. The idea is to compare the sum of all thermal
territories with the surface of the PCB. The method is of
coarse crude but it serves well as a first feasibility indicator.
The rule of thumb is that cover ratios below 50% definitely are
manageable and that ratios above 70% create great difficulties.
For PCBs that tend to have a few domination heat sources, for
example power applications, it is recommendable to decrease
these limits by 10%.
An alternative to thermal territory method is the cooling
efficiency method. This method is PCB oriented as oppose to the
thermal territory method, which is component oriented. The two
methods are complementary. If the heat source distribution is
fairly uniform they will essentially create the same result but
the advantage with the thermal territory method is that it is
more sensitive to PCBs with hot spots.
Figure 4
A layout example before and after improvement.
Layout application
Figure 4 shows an example of how easy it is to make a thermal
improvement of a layout when the thermal territories are visible.
The cover ratio was in this case initially 70% but since some of
the components in the improved layout were pushed towards the air
inlet, the cover ratio for the latter decreased to 68%. The
changes were made disregarding all layout restrictions and would
probably not be possible for an applied case.
To use the thermal territory method in the layout phase usually
results in considerable timesavings. Conventional methods, that
are based on the idea of "make changes and calculate", are much
slower. The economical aspect of this difference should not be
neglected. If one hour of work time can be saved every week it
can be estimated that the pay back time for a program that can
handle thermal territories is well below the 3 year limit that
usually is regarded as the attractive investment horizon.
The example shown is somewhat exceptional in the sense that the
improved layout has no thermal territory overlapping. In most
cases this is not possible. When overlapping does occur it is
however important to ensure that the territories have free
spaces around them in some other direction.
Overlapping of thermal territories tend to become increasingly
common in modern electronics. The reason is that
component-to-component communications both has a tendency to be
made with an increasing number of connectors and at a higher
signal frequency. The general remedy for both these
complications is to place the components closer together. A
situation in which the thermal territory method needs to be
applicable not only for single components but also for groups
of components is therefore fast approaching. This is both a
theoretical and a numerical challenge. How should these kinds
of territories be defined and how should they be calculated?
Figure 5
The thermal territory method is not exact. There will always be
an accuracy problem when two territories of different size meet.
Another layout related problem is that the method not is exact.
Each thermal territory has by definition adiabatic borders,
which of course is a simplification of the actual conditions on
a PCB. When two territories meet there will consequently always
be an accuracy problem, figure 5. This discrepancy has a
tendency to increases with the thermal territory size
difference. The same phenomenon appears if the territories have
been calculated with different temperature criterions. It can
therefore be concluded that the method works best on territories
that roughly have the same size and the same temperature
criterion.
This is one of the reasons why the method not should be
overdone. To spend hours on trying to place all thermal
territories exactly edge to edge might not be particularly
rewarding. The method is therefore at its best when it is used
for a coarse first layout attempt. Fine-tuning is better done
with other means.
Conclusions
The thermal territory method can be used to indicate if thermal
problems can be expected on the individual component level.
The method can be used to approximately expose the impact of a
heat sink even if the layout environment not is known.
A procedure that compares the sum of all thermal territories
with the size of the PCB can considerably improve front-end
thermal design estimates.
The method is ideal for creating a first layout attempt.
Ake Malhammar