Myth: heat sink saturation

[CalCoolage]

I have seen a number of articles debunking the myth that aluminum gives up heat better than copper. But a more obvious myth is currently popular with HS reviewers: Heat sink saturation. There is no such thing. These reviewers sometimes claim their data show the phenomenon. However, looking at the same data, I see no evidence of it.

Taking this a bit further, I have seen quite a few "how to" articles mentioning that "every heat sink eventually reaches a point where it can no longer keep up with the amount of heat it receives" Not before it melts. An aluminum sink could get hot enough to melt lead and there is still no saturatrion.

If there is such a thing as thermal resistance, then the temperature rise is proportional to the amount of heat, there is no leveling off. If you go from 70 watts to 100, the temperature rise is the same as when going from 100 to 130. No saturation. If anything, the apparent thermal resistance goes down due to a bit more radiation.

So where are the vigilantes when there is so obvious a myth to stamp out, one which is spread in the worst way by supposed authorities, who should know better? They are silent.

So I say bol*ox to saturation. :)

I think I can safely predict this message will scroll off the first page without a refutation or even a reply. It will disappear from view unnoted. The people spreading the myth will neither support it with facts and theory, nor cease spreading it.

[WyrmMaster]

Il reply.

Reallly, there are so many myths its not even funny.

[samuknow]

Let me get this right....


The myth is that a HS will reach a point of saturation.( no longer be able to remove heat). How are they proving this? This might be true in a vaccume where the HS could not radiate and lose heat built up.

[Mictlan]

Let me get this thing strait. The problem is not the HS per se, but the amount of fluid thats removing the heat. In a stedy state, the heatsink will recieve the heat from the die. Then the heat will rise along the HS until in reaches the fins. When the air flows across the HS it "removes" the heat from HS. So heat in from the die=heat removed with the air.

If not enough air is crossing the heatsink, some heat will remain in the HS, and the heat "stored" in the HS, which will begin to raise it's temperature. This won't be a steady state, so eventually you will reach a temperature at the bottom of the HS (where it contact the die) and eventually you'll be left with a cooked chip.

The trick here will be to calculate the amount of air you need to remove the heat from the HS. That's why you see the loudest and meanest fan you can slap over a HF.

[Badger]

I almost hate to say it, but I agree with you 100% (although I have never came across the saturation theory).
A heatsink temp will continue to rise until the heat it gives off to the air equals the heat input from the cpu (or whatever). There is no point where the heat input causes 'saturation' the HS just gets hotter until the heat transfer equilibrium is restored (or until the HS melts).

[Kingslayer]

I wholeheartedly agree that this is a myth. As a matter of fact it has been added to my list, right next to "Burn-In".

[Badger]

Yeah 'Burn-in' is the best myth of all.
Great if your itching for an argument though!

[UnseenMenace]

Originally posted by Badger
Yeah 'Burn-in' is the best myth of all.
Great if your itching for an argument though!

Myths can be disproved and while I have never seen proof that Burn-in works I have also never seen proof that it does not and as such I personally prefer to remain open minded to this.

[Monaco]

Burn-in most definately is not a myth. It is, however, way overrated as to it's importance.

Ask an EE major about electron migration. I can't explain it very well.

[Badger]

SEE !, I said it was good for an argument, but really I'm not in the mood today (most unlike me!)

[dimmreaper]

I'm not sure exatly what you mean by "saturation".

But every form of matter does absorbe heat. This is why a cast-iron skillet stays warm for many minuetes after you turn the burner off.

And every heatsink can only effectively remove so many watts heat. If the CPU generates more heat than this limit, CPU temps will clime to oblivian (or until the CPU frys, or the heatsink melts).

Think of it like this: a copper wire is only capable of conducting so many watts of electricity. Any more than that, and it will become a resistor, and eventually heat up till it melts. If you want to move more electricity than the wire permits you either need a larger wire, or one the same size but made of a better conductor.

Every heatsink has it's limit, just as every wire does.

If a CPU generates less heat than the heatsink can absorbe and remove (through convexion and radiation), but more than it can remove alone, eventually it will hit a wall.

[Badger]

Dimmreaper,
Have a read of the article on the front page it explains it all better than i can. But there is no 'wall' as such. The heat any heatsink is capable of giving off (by convection mainly) is proportional to the difference between the heatsink temp and the air temp. If you put more heat into a HS it heats up until the HS/Air temp difference means it dissipates as much heat as it takes in (until it melts).

The formula is:
q=h*(t1-t2)
q=energy transferred
h= the convection coefficient (practically constant for a given HS)
T1=air temp
T2=HS surface temp

You can see there is no limit to the amount of of heat a HS can transfer as long as the temp difference is great enough. Obviously in a good heatsink the h value should be high so the temp difference can be as small as possible for any given heat input.

In practical cpu cooling terms there is of course the limit at which the HS no longer adequately cools the cpu but we are talking HS theory here

[cjtune]

Badger, the convection heat transfer coefficient is not constant for all heatsinks. If so, we wouldn't need Glaciators in place of Orbs or Delta38s in place of YS Techs. Hmmm, I always thought that 'heat saturation' was just simply a point in time when an object's temperature stays constant given constant system characteristics maybe it's just the definition of it and I used it too liberally...gotta check the textbooks.

[cjtune]

Originally posted by Would71
I was saturated last night.. surprised I could even walk. But at least the room wasn't spinning. :D

Oh, we're talking about something else.. my bad. :D

Took a leak?

[Badger]

Originally posted by cjtune
Badger, the convection heat transfer coefficient is not constant for all heatsinks.

I never said it was constant for ALL heatsinks, just for a given heatsink (at least that's what I meant).

[Richard]

I always attested that "design was king" in the realm of heatsinks.

I've said it before and I'll say it again, a well designed aluminum heatsink will outperform a poorly designed copper. (duh..) However, a well designed copper will work better than the aluminum.

The thing is air-cooled heatsinks, like anything, have a point at which comparison becomes more and more difficult. What we had a year ago were a bunch of ineffecient heatsinks and a few good ones. Now we have a bunch of good heatsinks and a bunch of bad heatsinks. Take a look at the top 10 performing heatsinks around and you'll see the trend to bigger, more fins/pins, and copper playing a big role.

[CalCoolage]

Originally posted by dimmreaper
Think of it like this: a copper wire is only capable of conducting so many watts of electricity. Any more than that, and it will become a resistor, and eventually heat up till it melts.


This is a correct statement of the falsehood. In addition, you started a myth for the conduction of electricity that I never heard.

There is no saturation point for the conduction of eletriciity.

For a theory of heat saturation, I never saw one. I was trying to goad someone into stating one. As for someone who says that there is such a thing, check out the reviews from the excellent reviewer at overclockers.com. The saturation supposedly occurs under normal temperatures where the sink is not melting or anything like that. The guy at 2Cooltek also believes heatsinks suddenly fail, due to something like saturation, which is false. What happens is "poor contact", due to the way it's mounted.

Ohm's law, which is a statement involving electric resistance, is the model for thermal resistance.

Ohm's law is the backbone of electrical theory, and there is no leveling off of the amount of current that will flow through a given resistance. The voltage and current are always proportional for a given resistance. The truth of Ohm's law has been established beyond question. It's like Newtons laws of motion, except Ohm's law is even more universal. (remember Einstein?)

(BTW, electrical laws were "relativisticly invariant" before Einstein invented relativity, due to James Clark Maxwell. Al Einstein realized that had peculiar implications for matter. Maxwell, in his way, was more extraordinary than Einstein. )

For real wires, there are qualifications. As copper heats up, its electrical resistance increases a tiny percentage. For practical purposes, it is almost the same at -100C and +100C. This does not cause any saturation either.

Where does the ampere rating for wires come from? Wires generate heat due to their electical resistance. One rating comes from safety experts who determine how hot the wire will get in the worst possible conditions, so that a fire might start, or the insulation might melt. Putting 30 amps through a 15 amp rated wire might cause a fire for wires burried in 2 feet of already hot attic insulation. (It might, but it probably won't.)

Another rating, which includes the wire length, is determined by how much the voltage can be alowed to drop (Ohm's law), so that appliances and tools will still operate normally. The voltage drop is related to the amount of current (amps) an appliance will draw, and so a maximum wattage can be stated (at 120/240 volts). For resistances, Watts = Amps times Volts. Your electric lawn mower may crap out with 200 feet of number 18 extension cord, but the wire won't saturate.

[Nitestorm]

I had always assumed that myth about us hitting a wall, came from the idea there are limitations (ie a wall) of air cooling (forced air convection) with ever increaing heatloads? Which has forced heatsinks to use more surface area and big CFM fans to continue to perform well. However supposedly now heat loads are reaching a point where air-cooling isn't cutting it anymore because air isn't the most efficient way of removing heat. So even if you increase the airflow and/or surface area or use copper, the performace gains are becoming less and less. Thus many heatsinks perform very close to each other nowadays and most need the same powerful fans. There are also more unsual designs are appearing. I think Joe did an article where he saw little gain to justify using massively more airflow with current heatsink designs than the delta 38CFM, considering the extra noise and size. Plus increasing surface area is restricted by socket size and manufacturing costs. Everyone seems to keep saying water-cooling is the way to go because water is more efficient at removing heat than air and thus has more headroom to cope with higher heat loads.

[dimmreaper]

Originally posted by dimmreaper
I'm not sure exatly what you mean by "saturation".

Every heatsink has it's limit, just as every wire does.

In my above post I was not agreeing or disagreeing with the theory of saturation, I was mearly stating that every heatsink has a point at which it can no longer practically cool a CPU under a given set of circumstances. I was saying nothing more, and nothing less. Sorry if I wasn't making myself clear.

[dimmreaper]

Originally posted by CalCoolage
There is no saturation point for the conduction of eletriciity.
Agreed! But there is a point of impracticality (in wires it happens to be the point of self-destruction).

But it's still not practical (or safe) to use 16gauge wire for arc welding.

Just like it's not practical (or safe) to use a 486 heatsink on a 1.4GHz athlon.

This was the point of my admitedly rather poor analogy.

[dimmreaper]

Originally posted by Badger
Dimmreaper,
Have a read of the article on the front page it explains it all better than i can.Havn't read it, and can't seem to find it. Mind shooting me a URL?

[cjtune]

Originally posted by Badger


I never said it was constant for ALL heatsinks, just for a given heatsink (at least that's what I meant).

Errg, sorry there, what I meant was typed as something else. OK, for a given heatsink it's constant if the airflow is constant in magnitude and direction. Mounting a fan differently or a different fan will change the convection coefficient. Geez, I don't know about all of this saturation fuss. To me, the heat capacity of a body is saturated as in heat in = heat out and the temperatures don't change anymore. If heat in > heat out, the temperatures continue to increas and if a temperature reaches a certain point (melting, vaporisation, combustion, etc.) then that's too bad.

About conduction in wires, electrical resistance (and also thermal resistance) goes up with increasing temp. With all the plastic insulation around a wire, there is no way it can cool sufficiently once the I^2*R (electrical) heating becomes large enough and thus it melts. Note that the increasing electrical resistance accelerates the destruction of the wire. Electrical resistance is inversely proportional to the diameter and the conductivity of the wire material. That's why wires cannot carry currents safely above their rated max -unless you actively cool them. Note that the heating is a waste of energy and that's why using larger gauge wires is always better and that we use high voltage lines to ferry electrical power rather than high current lines. Same principle behind the electrical fuse.

[Badger]

Originally posted by dimmreaper
Havn't read it, and can't seem to find it. Mind shooting me a URL?

Here: http://www.amdmb.com/article-display.php?ArticleID=105

[CalCoolage]

Originally posted by Would71


Cal, please go out into my utility room and tell the fuses in my fusebox that they operate on just a myth, there's no saturation point for the conduction of electricity, and maybe I'll stop having to replace them so often. :D Thank you, come again.

(sorry, just couldn't let terminal eggheadedness rule the day.. getting too stuffy around here, and it annoys me)

OK. ... (waliking to your utility room) ... Fuses, if you operate on the basis that there is a saturation point for electicity, then you don't work.

I'll say it again: There is no saturatrion point for the conduction of electricity.

I could explain how fuses actually work to those who do not know, but I'd get another stuffy message from some unnamed person, whose content would be infected with terminal eggheadedness and which would ignore the point and the thread topic; I'll say it again: There is no saturation point for the conduction of heat, or electricity either. Its an important concept in engineering as well as amateur heat sink design.

If people get off calling me names like stuffy and terminally eggheaded, I'll live with it. Whoop-de-do.

Maybe some people are using the word saturation to mean something other than what the reviewers meant, as I described in my stuffy, terminally-eggheaded original post, which they may have found too tedious to read. Like maybe they want to say the sink let the CPU get too hot. How about just saying the sink reached the danger point? Saturation is one of those stuffy, terminally-eggheaded words that only stuffy, terminally-eggheaded people like to throw around to make it look like they know what they are talking about to people who have no idea what they are talking about.

Sorry, just couldn't let terminal eggheadedness rule the day.. getting too stuffy around here, and it annoys me.

Kelvin

[CalCoolage]

Originally posted by dimmreaper
Agreed! But there is a point of impracticality (in wires it happens to be the point of self-destruction).

But it's still not practical (or safe) to use 16gauge wire for arc welding.



Sorry to use your words to hammer on my point, but you were the only one that seemed to defend HS saturation. I'd still like to see someone try.

I thought everyone read all these reviews and articles. To reiterate, the reviewers and writers are talking about a normally operating heatsink. They are not saying the sink melts, vaporizes or otherwise self-destructs, and not that it just got too hot either. Within normal conditions, they say, the amount of heat any sink can get rid of levels off, and eventually the sink will not take up any more heat. Not true. Heat sinks actually get more efficient if you can get them really hot, too hot unfortunately for CPU's. In the case of reviewers, they are claiming some sinks actually saturated, or started to, in their tests! None of them let the sink melt. I don't think they could even get the aluminum to melt.

When AMD comes out with some future CPU, it could possibly operate normally at temperatures now considered too hot, and sinks that are now inadequate could work OK. But not if sinks were already into this mythical saturation.

At the risk of being stuffy and terminally-eggheaded, materials do not normally saturate with heat at transition points, like melting, vaporizing, "eutectic phase changes" or ionizing into a plasma. On the contrary, they ususally take up a lot MORE heat at these points. That's why they have condensers and evaporators in refrigerators and air-conditioners. They use the melting of ice, when salt is added, to freeze the cream in those ice cream makers you can buy .

I guess we all realize we woudn't want to melt the heat sink to (temporarily) cool the CPU a bit. :)

[Wahoogie]

Wow thats a laugh, the only point when it would become "saturated" is when the thing is melted all over your motherboard! From a CPU, there is no way a Copper HS or an Alluminum Heatsink would become close to melting point.

[CalCoolage]

Originally posted by Wahoogie
Wow thats a laugh, the only point when it would become "saturated" is when the thing is melted all over your motherboard! From a CPU, there is no way a Copper HS or an Alluminum Heatsink would become close to melting point.

bump