eddy currents in metal core PCB's

They have failure mechanisms, and early generation heat pipes did fail a lot. 20 years is considered to be a good MTBF nowadays.

Is that more like a trapezoidal block, a better local heat spreader?

John

I can imagine various ways that you could make a heat pipe fail in 20 years, e.g. by using cotton for the stuffing. If you seal it correctly, use some more durable fill, e.g. glass wool, and manage the electrolytic problems with due care, I can't think of a short-term failure mechanism. Do you know of one?

The ones I've seen in Apple servers are simple loops between the heat spreader and the middle of the fins.

Cheers

Phil Hobbs

google heat pipe reliability for some gory details.

John

I'm seeing references to leakage and corrosion, which are not very difficult to avoid with intelligent materials choices and decent manufacturing procedures. I find it hard to believe that there's anything fundamental that prevents a heat pipe from lasting centuries at reasonable temperatures.

Of course the application has to be able to bear the amount of engineering required.

Cheers

Phil Hobbs

pe

at

Whoever said they did?

Correct. Lots of heat through a rather narrow crossection.

It often is, but you also get parallel arrays of narrow pipes and other slightly baroque structures.

Of course.

Fins are pretty much inevitable if you want to dump the heat into air, and a fan lets you put the fins closer together.

Then you wouldn't add it, would you. They cost money and you've mostly got to get them custom-made to solve your particular problem.

True, but - like every other exotic trick - they can make practicable something that was otherwise impossible.

I've had the same kind of reaction to proposals to solve specific timing problems by using a couple of ECL parts.

-- Bill Sloman, Nijmegen

Would Aavid sell them if they were a "pipe" dream?

Thanks, Rich

Hi,

Heat pipes are quite common in PC's now, mainly on motherboards and on high end CPU/GPU heatsinks:

motherboard with heatpipe connecting northbridge IC heatsink to buck switchers heatsink:

CPU heatpipe cooler:

(above images are from these links)

PS. I don't work for NCIX :)

cheers, Jamie

--- Posted via news://freenews.netfront.net/ - Complaints to snipped-for-privacy@netfront.net

Shhhhh, don't spill the beans :-)

One method would be to goose the gate but it is risky because if the "peaker pulse" is too long the gate blows.

Because you can't build things such as iPads with it?

They are useful in some cases, like cooling CPUs, and where they can be manufactured in fairly high volume. In a lot of situations, some machined copper will work as well.

Again, they are only moving heat a short distance. If one has the option of spreading out the heat, by using multiple transistors for example, or where the heat is already distributed, they usually aren't worth it.

Intel causes a lot of heat pipes.

John

In fact,their particular virtue is that they move heat - as latent heat of evaporation emboddied in vapour - over relatively long distances, through narrow cross-sections and quite fast.

Conduction needs a thermal gradient. Heat pipes just need vapour whistling along. The pressure drops required to move the volumes of vapour involved are quite modest.

You've missed the point. Heat pipes allow you to put your - bulky - radiating structure quite a lot further from your heat source than any scheme that relies on conduction or convection.

There's a lot to be said for keeping terminated transmission lines on silicon, inside a single package, but the package does end up having to dissipate a lot of heat.

-- Bill Sloman, Nijmegen

(...)

"

Careful, the solder holding the FET and pin header has only ~17% of the thermal conductivity of copper and the brass header pins are only

See IRF's heatsink recommendations:

--Winston