How it's made: heat sinks

Most heat sinks are extruded. Some are cast, like pin fins.

Some are machined.

There's also a process where the fins are pressed into slots

Skiving is real slow. The only value is you get thinner fins than can be extruded. More machining is needed if you want holes though any of that.

This gets you more fins for better transfer to the air. People focus on silly points like diamond heat sink grease, when they often lose far more performance at other points in the heat path. Ultimately there is a limit on how larger a heat sink you can attach to a CPU/GPU directly.

If size and cost are not an issue, heat pipes to connect the heat block on the CPU/GPU to a much larger thermal air interface. Bigger fins, bigger fan and much better performance.

Then water cooling can get even better performance, and the noisy bits can be somewhere else, even in another room. I remember discussing this with a guy who ran the tubes to a drum in his garage where he didn't even need to use an air interface. The thermal mass of the drum was good enough to absorb the heat for the time he ran the computer. Zero noise other than the power supply fan, I suppose he still had one of those.

Those long skinny fins don't look efficient to me. And they would need a huge air blast.

Does anyone know what he is talking about?

The large surface area of the long, skinny fins are perfect coupling between the low thermal resistance of the heat sink and the relatively high thermal conductivity of the fin/air contact. Lots of surface area gives a low thermal resistance at the point of contact. That's the point of using them.

It's the extruded heat sinks with much fewer fins and a lot less surface area, that require a lot more air flow to get the same thermal conductivity.

======================

Also: Scived heatsinks suffer the disadvantage of being limited to internal use. Extruded and cast ones are often structural parts of a case.

Horses for courses....

..... Phil

The question hangs on how much heat you have to get rid of.

The heat capacity of air is well defined. If you need to move a lot of heat you have to heat up a lot of air and move it away as soon as it has got hot.

Fewer fins with lots of space between them let you blast a lot more air past them than you can squeeze between the thin fins of a skived heat sink. If you don't need to get rid of as much heat, a slower air flow will serve, and you can get away with a less voluminous heat sink

I don't think so: a strong air blast would bend the fins.

The sheet thermal resistance of a thin plate can be high, especially aluminum alloys that skive well.

If you blow air at a thin, tall, dense array of fins, it will naturally prefer to go around, not through, the fins. And you'll get most of the air flow at the tips of the fins, not near the baseplate, so the thermal resistances hurt. Short skived copper fins are good if the air is really blasted into the fins from above, like in a CPU cooler.

I've measured a lot of this stuff lately. That trumps theories.

That strong would destroy the enclosure and kill bystanders.

Cydrome Leader snipped-for-privacy@MUNGEpanix.com wrote in news:t3f6os$9d4$ snipped-for-privacy@reader1.panix.com:

Ricky snipped-for-privacy@gmail.com wrote in news: snipped-for-privacy@googlegroups.com:

But at that close spacing, due to boundary layer effects, it would have to be strongly forced air. or it would simply "walk over top of" the whole thing, doing less work in the end.

All that is needed is intimate, coplanar contact. The grease takes up voids. What *should be* "micro-sized voids", but voides nonetheless... errr without it the heat tranfer IS less. Those voids contain the insulative gas "air".

The first series of AMD dual core CPUs back in oh... 1998 or so would fry the CPU AND the motherbaord almost instantly if the heat sink was not well coupled, and that included void filling paste.

That is why contained flowing water was the choice for a long time.

I always wanted to make an entire PC immersed in dielectric fluid

But it too has to be moving to carry the heat away.

Still water is better.

Oh... I see you are aware.

Ricky snipped-for-privacy@gmail.com wrote in news: snipped-for-privacy@googlegroups.com:

The word(s) for today is "boundary layer".

Slow air would pass right over a close fin spaced sink. High speed forced air is required when the fins get that closely spaced.

It is not only about their condution from the heat source, it is about how well it gets transferred to a MOVING air mass. That is why most heats sinks had wadire fin spacing.

"Carlos E.R." <robin snipped-for-privacy@es.invalid wrote in news:ot6tii-7un.ln1 @Telcontar.valinor:

Silly.

A strong air blast must have blown the only two firing neurons you have between your ears to a new location.

Right. Viscous drag will keep air from flowing between tall, closely spaced fins. It will have to be ducted and forced, or it will go around.

The limiting case, more and more denser and thinner fins, volumetric air flow will approach zero.

My general rule is that a heat sink should reduce the native air flow by about half. Neither zero nor 100% does any cooling.

And of course, the tips of tall thin fins have a high thermal resistance to the baseplate, so run at about inlet air temp, so restrict air flow without contributing much coolong. My 50% number is useless if the air flow is restricted without corresponding cooling.

There is no limit to how bad a heat sink you can design. A solid aluminum brick is pretty bad.

That's where heat pipes are nice.

What moves in heat sink is just water vapour. The water gets wicked along to the area that needs cooling, then evaporates - rapidly, because there's only water vapour inside the heat pipe, and the vapour pressure in side the heatpipe is that of the water at the point where it is condensing, plus the pressure drop to move it along to the point where it does condense.

It really whistles through the heat pie on the way to the condensing end, and water vapour carries a lot more heat per gram that a noncondensable gas - the latent heat of evaporation of water is 2256 kiloJoules per kilogram. A kilogram of water is a lot of water - about 1250 litres of water vapour at atmospheric pressure.

Even one gram of water is a 1.25 litres of water vapour, and the pipe work on an ordinary heat sink would be more like 0.1 litres, and you might get a tenth of an atmosphere pressure inside, so you might have 10 milligrams of water on the move, so about 22 joules. But if that got from one end of the pipe to the other in one second that would 22 watts. It seems to move faster than that

The heat pipe isn't bulky. The dissipating fins at the condensing end usually are pretty bulky. You are still dumping the heat into air, after all.

Only if you can get it to evaporate. Liquid water as a heat transfer medium is okay, if you can dump the warm water down the sink - and I have done that for a research set-up - but it isn't practical for long term use.

Not necessarily. Make the power supply physically big, and reasonably efficient, and free convection would work.

Our 1996 millidegree thermostat started off with circulating water, but it got airlocks in the water pipes after about six months, which needed a visit from our service man to fix. So the production manager switched it to a heat pipe, after I'd moved to the Netherlands (in 1993) but before I'd written up the job, so it's covered in the paper I wrote, which didn't get published until 1996.

It got his name (Douglas Stewart) onto the list of authors. The heat pipe setup was a custom part, and a pain to get designed and made but well worth the effort. I wished I'd thought of it.

Not exactly. true. Slow air would mostly bypass the gap between the fins but any pressure gradient along the length of the fins will move some air. If you haven't got much heat to dissipate, you don't need to move much air to carry it away.

It won't stop it flowing, but it won't flow fast in the narrow gap.

Most of it will.

But it won't get there.

But there's nothing magical about getting exactly 50%.

If you don't have to move much heat, a skived heat sink and slow moving air might be the way to go.

On a sunny day (Sun, 17 Apr 2022 09:25:41 -0700) it happened snipped-for-privacy@highlandsniptechnology.com wrote in snipped-for-privacy@4ax.com:

This works great: