Heat Sink Technology
Heat Tunnels, Thermal Gradients and More
By John Roberts
 Over
the last several years we've seen a number of power amp ads
purporting to educate the market about heat sink technology.
More often than not, the chosen approach is neither novel nor
superior. The advertisement is more creative than the design
being pitched. The following description of Peavey's new heat
sink approach will describe a real advance on the state of
the art.
Forced Air and Heat Tunnels
Above a fairly modest power level, passive or convection cooling
is inadequate to deal with the heat generated. While spreading
out the heat is desirable in passive cooling, once you move
up to forced air, you often need to bring the heat sources
closer together to better control the cooling. A classic technique
is known as a "heat tunnel". The general configuration of a
classic heat tunnel is a fan pulling cool air from outside
the chassis into a passage. The cool air inside this channel
flows over a series of heat sink fins. While the air begins
at ambient temperature, it extracts heat from the heat sink
fins with every collision. By the time it exits the heat tunnel,
it is tens of degrees hotter than when it entered.
I find it a little amusing that one competing approach uses
a center-driven heat tunnel, essentially two short tunnels
driven by a single fan. The remarkable claim is that because
the heat tunnel is short there is no thermal gradient (temperature
difference) along the length of it. This claim is obviously
questionable because if there were no thermal gradient, the
air would have to exit at the same temperature it came in;
consequently, there would be no heat transferred to the cooling
air.
A classic problem when using forced air to cool a multiplicity
of devices arrayed along a heat tunnel is to equalize the device
temperatures along the length of the heat tunnel. In high-power
amplifiers you routinely use several power devices in parallel.
If one receives less cooling than the others, it will run hotter.
Like the weak link in a chain, it will reach its maximum temperature
and fail first. This technique can make a huge difference in
total work output available from the amplifier because you
are limited by that one, hottest device. Thermal protection
must shut down the entire amp to protect that one hottest device.
Even if all of the power devices have the exact same access
to the cooling air- flow, the air picks up heat as it passes
over preceding devices. This hotter air can't absorb heat as
readily as the cool air. The classic solution for equalizing
the thermal gradient in a heat tunnel is to tune the length
of the heat sink fingers. Just look inside the industry standard
CS800X, The heat sink at the hot, exit end of the heat tunnel
has much more aluminum and longer fingers than the cool inlet
end.
Multi-stage, graduated-heat tunnels are complex and difficult
to manufacture. While you could just ignore the heat rise phenomenon
and invest in advertising, we chose to apply our 30+ years
of experience design power amps to find a better way.
"Turbo-V Cooling"
I apologize for the name but these days we need to cut through
the marketing clutter. Our new heat sink approach is deceptively
simple. By angling two heat sink extrusions toward each other,
forming a truncated "V", the heat transfer is significantly
increased at the narrow end.
The cool ambient air enters at the wide end, top of the "V." As
the cool air moves through the heat tunnel gaining heat, the
walls move closer together increasing collisions with the air
and helping the hotter air extract the same amount of heat
the cooler air did in the previous section. By the time the
hottest air gets to the end of the tunnel, the heat transfer
is at it's maximum. The net result of tuning the heat sink
for air temperature, the devices all operate at close to the
same temperature and the amplifier will be able to do more
work before the hottest power device gets too hot.
Note: Turbo V cooling is currently being used
in the GPS 1500 and GPS 900 power amplifiers. This approach
is novel and a US patent has been applied for
|
|
 |
| |
|
