Cirrus Logic AN Integrating CobraNet into Audio Products - Manual (Page 10)

Thermal P RELIMINARY
8
CS18101 (CM-2) AppNote1 - rev 1.1 Jan, 2004
w w w . c i r r u s . c o m
T
ja
(max) = 33.6°C/Watt.
T
ja
(typ)= 33.6°C/Watt.
Therefore:
T
a
(max) = T
j
(max) - (T
ja
(max) * P
d
(typ)) = 100 - (33.6 * 1.0) = 66.4°C
According to the math we are not above the 70°C spec on the CM-1. That is because the
math above does not take into account the heat-sink effect of the PCB. Under the FPGA
are four copper planes and 49 vias that transport the heat to those copper planes. There
are four more signal layers that also aid in heat conduction away from the FPGA.
There is not enough data to allow us to calculate the cooling effects of the PCB. And
getting this data would be extraordinarily difficult. In this case we are assuming that the
PCB gives us that extra 3.6°C of cooling that we need. While such an assumption might
seem cavalier, 3.6° is within the margin of error for the other numbers used in our
calculations. This is why we can only use the numbers for a ballpark estimate and need to
do testing for final acceptance.
Another issue is the T
j
(max) spec of the FPGA. Above it was listed as 100°C, but the data
sheets list it as 85°C. The truth is that the absolute maximum is 125°C, but the timing
specifications are only guaranteed to 85°C. Above that the FPGA slows down at a rate of
0.35% for every degree above 85°C. We have designed the FPGA to run at 100 MHz,
85°C. But, except for some custom firmware, we are running the FPGA at 95 MHz. So if
we account for this thermal de-rating then we can calculate that a T
j
(max) spec of 100°C
is correct.
If we take the lowest common denominator of all the chips on the CM-1 we find that our
ambient temperature spec is 70°C. In the case of the FPGA we know that we're talking
about still air, but we really don't know what "ambient air" is with regard to the other chips
on the CM-1. This is where testing can really clear things up. Testing with a still air
ambient temp at 70°C and above can help to verify proper operation.
In our labs we have done elevated testing of the CM-1 to well beyond 70°C ambient
without failures. Unfortunately, this doesn't mean that the CM-1 will work in your system at
temperatures beyond 70°C. As has been pointed out before, there are lots of variables
and only system testing can prove or disprove that it works. So we encourage all users to
do their own testing.
CM-2
Since the CM-2, and the main chip on it (the CS18101) have not been completely
characterized it is difficult to do any sort of analysis on it at this point. But we do have
some preliminary information:
The designers of the CS18101 have said that the part will be able to operate with an
ambient temp of 70°C. Based on the information available, this seems like an achievable
goal.
The Ethernet Mac/Phy, the Davicom DM9000E, is rated at 0-85°C case temperature with
zero air flow. Further information is needed to calculate an ambient temperature from that
case temperature. But based on the chips power consumption (0.33 watts) it seems very
likely that this part will work fine with a 70°C ambient environment.