Cirrus Logic AN Integrating CobraNet into Audio Products - Manual (Page 11)

Thermal P RELIMINARY
9
CS18101 (CM-2) AppNote1 - rev 1.1 Jan, 2004
w w w . c i r r u s . c o m
There are not many other parts on the CM-2, and all of them are low-power enough to not
worry about.
How To Test
There are so many factors that were not, and could not be, considered in this thermal
analysis that testing should be considered.
The first question to answer is, "Do I need to test at all?"
It seems that if the ambient temperature is below approximately 55°C then you probably
do not need to extensive tests. (The ambient temperature should be measured about 0.25
inches above the FPGA on the CM-1 or the DSP on the CM-2.) If past experience can be
a guide then it would indicate that units with fans don't need testing while units without
fans do. However, using the internal ambient temperature is a better benchmark for this
sort of thing.
If you have determined the necessity to test, the next question is, "How do I test?"
Measuring things like the case temperature of the individual chips is not very helpful.
Usually the temperature probes are taped to the chip (which insulates it, making it hotter),
and we're not really interested in the case temperature anyway since it is the junction
temperature that directly reflects the failure point. And it is difficult to calculate the junction
temp from the case temp unless you know exactly how much power the chip is dissipating
(which we don't know exactly).
A basic, but effective, test would be to put the unit (case and all) inside an environmental
test chamber. One or more temperature probes would be placed inside the unit to
measure the ambient temps at various locations.
The temp inside the chamber (inside the chamber, but outside of the unit) would be raised
to about 50°C (122°F) and kept there for several hours while the unit was run through
some functional tests. If all goes well the temperature would be raised 5 or 10°C and the
test repeated. Tests would be run at ever increasing temperatures until the unit failed.
If the temperatures recorded during the tests are acceptable then you're done! Otherwise,
you need to fix things (which is beyond the scope of this document).
What if you don't have a thermal test chamber? Then it's time to improvise. A basic one
can be built using a cardboard box lined with thick foam insulating sheets (available at
most home improvement stores). Inside the box put an incandescent lamp, a small fan,
and some sort of thermal mass (a large bench vise or a jug of water in a well sealed
container works well). This sort of setup is cheap, but it works. There isn't a thermostat,
but it heats up and cools down slowly so a "manual thermostat" is good enough. If buying
a real test chamber for thousands of dollars is out of the question then this might be just
the ticket.