SPCR's 2010 CPU Heatsink Test Platform

Want to talk about one of the articles in SPCR? Here's the forum for you.
MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

SPCR's 2010 CPU Heatsink Test Platform

Post by MikeC » Mon Jan 25, 2010 12:48 am


Monkeh16
Posts: 507
Joined: Sun May 04, 2008 2:57 pm
Location: England

Post by Monkeh16 » Mon Jan 25, 2010 1:38 am

Any chance of getting the Xigmatek HDT-S1283 retested?

frenchie
Friend of SPCR
Posts: 1346
Joined: Tue Jul 08, 2008 4:53 am
Location: CT

Post by frenchie » Mon Jan 25, 2010 2:16 am

Interesting change Mike, thank you for the update.
Maybe the last paragraph should be mentionned in the very beginning.

I just have one point. When running a quiet CPU cooler (let's say with a fan running at 7 or 9V), the higher the temperature rise over ambiant, the more heat you have to exhaust from the case, ie the faster you have to run the exhaust fan(s), and the less quiet your system becomes. It is even more true today when you see the amount of heat those last generation GPUs produce at full blast.
I understand your point when you changed the system test plateform, it's like a worst case scenario for a silencer. May I suggest that you add the temperature results at idle, with the fan at 7V ? For some (most ?) people, this idle temp would probably give a good idea of temperatures to expect under light use of the computer, like word processing, internet browsing or even video watching (a small task for the CPU in the test plateform I believe, the video card doing most of the work ; please let me know if this is wrong).

cb95014
Posts: 109
Joined: Wed Jun 27, 2007 9:39 pm
Location: The insane State of California

Post by cb95014 » Mon Jan 25, 2010 2:37 am

Very interesting review, particularly for heavy-overclockers. Thank you for articulating your thought process in making the changes.

What was the reason for not including the HR-01+? AFAIK, Thermalright is still selling the HR-01+, which seems ideal for SPCR purposes - particularly if it can also handle extreme temperatures. At a minimum this seems like a better alternative than including the U120...

jmke
*Lifetime Patron*
Posts: 495
Joined: Thu Apr 15, 2004 3:53 am
Location: In Front of PC
Contact:

Post by jmke » Mon Jan 25, 2010 2:43 am

interesting article; I've been pondering about a similar change, I've tested with a dual core Prescott S775 CPU, seeing your results I can confidently say when you overclock a Prescott to 4ghz with enough vcore you match the heat output of the S1366 CPU:-)
What was the reason for not including the HR-01+? AFAIK,
time constraints most likely I'm guessing ;-)

K.Murx
Posts: 177
Joined: Tue Mar 17, 2009 10:26 am
Location: Germany

Post by K.Murx » Mon Jan 25, 2010 3:06 am

I'll second the wish for HDT-1283 (or 1284) figures. If you would do it, information regarding the heatspreader size of the i7 would be nice, too - I think it does make a difference with the HDT series.
frenchie wrote:I just have one point. When running a quiet CPU cooler (let's say with a fan running at 7 or 9V), the higher the temperature rise over ambient, the more heat you have to exhaust from the case, ie the faster you have to run the exhaust fan(s), and the less quiet your system becomes.
...What?
Once the dynamical equilibrium is reached, your CPU converts a set amount of electrical power to heat over time, and you need to exhaust that heat out of your case. If your CPU uses up 140 W of power, you need to exhaust 140 Joule per second, regardless of temperature.
And if the temperature of the air that you exhaust is higher, you can run the exhaust fans slower because the air will transport more heat per amount of air moved by your fan.

Ok, power usage goes up a little bit due to lower efficiency of the VRMs, but not significantly.

javitxi
Posts: 64
Joined: Thu Jan 01, 2009 7:30 am
Location: Madrid (Spain)

Post by javitxi » Mon Jan 25, 2010 4:08 am

First of all, nice article as always :)

About the analysis' methods, I have used prime95 a little bit, but I've read that to extress the CPU to the most it would be a great idea Prime Orthos 2004 instead of prime95 or the two programs running at the same time. Have you considered this option? I don't think i7-965 could not handle the 2 programs at the same time.

About undervolting tests, I'm interested and also I think the people at /reading Green Computing would be.

Also, it would be usefull to divide the heatsinks in two categories, because they blow (at least, they are designed to) the air in 2 different ways:

- towers like heatsinks/"not blowing the air to the motherboard-profile"
- low profile /"blow the air to the motherboard-profile" types, because all of them do the same task (Noctua NH-C12P, Kabuto, Nexus low-7000, Big Shuriken, etc)

Why? Well, the first thing I'm thinking about is that if you gather all this data (undervolted/clocked at load, normal OC/OV at load), maybe people could manage to build a cheap small form factor semi-passive Pc: e2xxx /e5xxx (i.e)/amd counterpart undervolted + small tower/low profile => less heat, less moving parts, less dust/need to air filters, less power and maintainance time consumption, less noise :) (mainly I'm thinking on electrodacus proyect).

And last but not least, Mike & staff, thanks for your time, help, suggestions and all the info and experience gathered :)

behr
Posts: 21
Joined: Sun Jun 05, 2005 9:12 pm

Post by behr » Mon Jan 25, 2010 4:14 am

you should also include results with no fan, maybe system being idle for a while and running some video and flash.

frenchie
Friend of SPCR
Posts: 1346
Joined: Tue Jul 08, 2008 4:53 am
Location: CT

Post by frenchie » Mon Jan 25, 2010 4:21 am

Hmmm...
I thought that the better the heatsink, the smaller the amount of air needed to dissipate the same amount of heat (hence the ability to run the fans slower on a good heatsink while maintaining low temps) ; all that because the heat is more effectively passed on to the air surrounding the fins.

Wait... I think we're saying the same thing... My wording was poor in my first attempt at an explanation.

Is is correct if I say the following ?
I just have one point. When running a quiet CPU cooler (let's say with a fan running at 7 or 9V), the higher the temperature rise over ambient, the warmer the air you have to exhaust from the case, ie the faster you have to run the exhaust fan(s), and the less quiet your system becomes.

PlanetOfTheApes
Posts: 103
Joined: Fri Jan 08, 2010 10:30 pm

Post by PlanetOfTheApes » Mon Jan 25, 2010 4:40 am

Suggestions

a) Test fanless performance - add an undervolted and fanless CPU test. Maybe underclocked to stock i7-920 or as required.

b) What I'd like to see tested:

Scythe Orochi Rev. B
Scythe Ninja 2
Thermalright IFX-14
Thermalright HR-01 Plus
Noctua NH-D14

Much appreciated thanks

jmke
*Lifetime Patron*
Posts: 495
Joined: Thu Apr 15, 2004 3:53 am
Location: In Front of PC
Contact:

Post by jmke » Mon Jan 25, 2010 4:51 am

you should also include results with no fan,
testing outside a case... makes passive cooling tests worthless, since inside a case with a single exhaust fan, you'll have more airflow and completely different end result.

K.Murx
Posts: 177
Joined: Tue Mar 17, 2009 10:26 am
Location: Germany

Post by K.Murx » Mon Jan 25, 2010 5:17 am

frenchie wrote:Is is correct if I say the following ?
I just have one point. When running a quiet CPU cooler (let's say with a fan running at 7 or 9V), the higher the temperature rise over ambient, the warmer the air you have to exhaust from the case, ie the faster you have to run the exhaust fan(s), and the less quiet your system becomes.
Not quite :)
The quality of the heatsink determines how efficiently you pass heat from your processor to the surrounding air, so far you are correct. Also, if you move less air, this air becomes warmer. But you do not need to run the exhaust fans faster.
Physics: Air can contain heat. The amount of heat in a certain volume of air is given by (Linearized, which is "good enough" for engineering):
Q=c_V * V * T
where Q: amount of heat, c_V: some material constant, V: Volume, T: Temperature.
The amount of heat per time t moved by a constant air current is thus:
Q/t = c_V * V/t * T
where the air flow V/t increases if you increase fan RPM and decreases if you decrease fan RPM. For the sake of the argument let us presume the dependance is linear, thus:
V/t ~ RPM
As mentioned before, once dynamic equilibrium is established, your fans must move the heat generated by the processor, thus Q/t is fixed. c_V is also fixed (material constant), thus:
RPM= const / T
so the RPM of your fan scale inversely with the temperature of the air.
Therefore, the higher the temperature the slower the fans (and the inverse holds true, too).
Until the electronics fail, anyway ;)

PlanetOfTheApes
Posts: 103
Joined: Fri Jan 08, 2010 10:30 pm

Post by PlanetOfTheApes » Mon Jan 25, 2010 5:39 am

jmke wrote:
you should also include results with no fan,
testing outside a case... makes passive cooling tests worthless, since inside a case with a single exhaust fan, you'll have more airflow and completely different end result.
Wouldn't a Cooler's fanless performance be relevant in a fanless or semi-fanless system. Semi-fanless system's fan(s) would be off until a preset temperature is reached.

Regards

jmke
*Lifetime Patron*
Posts: 495
Joined: Thu Apr 15, 2004 3:53 am
Location: In Front of PC
Contact:

Post by jmke » Mon Jan 25, 2010 5:40 am

in a case there is less "cool" air compared to an "open test" bed; so it would lead to different results;

PlanetOfTheApes
Posts: 103
Joined: Fri Jan 08, 2010 10:30 pm

Post by PlanetOfTheApes » Mon Jan 25, 2010 6:49 am

jmke wrote:in a case there is less "cool" air compared to an "open test" bed; so it would lead to different results;
No doubt a case will impede natural convection cooling to a degree. But the fanless results are still relevant in my opinion. A case such as Silverstone's FT02 (which has "stack effect cooling") could be used instead of the "open air" bed.

javitxi
Posts: 64
Joined: Thu Jan 01, 2009 7:30 am
Location: Madrid (Spain)

Post by javitxi » Mon Jan 25, 2010 6:52 am

K.Murx I have some doubts about what you have said. Hope to express myself correctly with my English :oops: :
Air can contain heat. The amount of heat in a certain volume of air is given by (Linearized, which is "good enough" for engineering):
Q=c_V * V * T
c_V in this formula, is the c_V of a monoatomic (3/2), diatomic (5/2), etc coeficient at constant pressure used on basics ideal gasses formulas?

In the formula you assume V is the V in litres of your case, isn't it?

Also, if the air surrounding is hotter than the heatsink, by the equilibrium of temperatures theory between 2 objects, the surruonding air will transfer heat to the heatsink although the sorruonding air is moving. Am I correct? So, the lesser temperature (then less Q it has) the airflow through the heatsink has, the bigger amount of heat (Q) the airflow can carry out (evacuate) from the heatsink.
As mentioned before, once dynamic equilibrium is established, your fans must move the heat generated by the processor, thus Q/t is fixed
As far as I know, copper has better thermal conductivity than aluminium and actually, then most of modern heatsinks use heatpipes based on copper. The little I know about them is that when your heatsink reaches the maximum amount of heat he can conduce by itself, the extra heat is not transferred to the heatsink, thus it is retained by the CPU's capsule, growing the temperatures of the CPU's capsule.

So establishing an airflow through the heatsink will have the ability to carry out so much Q as the equilibrium between the 2 corpses can. I mean, if your heatsink is at 80ºC and you have airflow at 0ºC, then the equilibrium between these temperatures are 40ºC, so at a constant 80ºC by the heatsink, the 0ºC airflow will move out 40ºC. I don't know what the exact formulas are because I have not studied so much about thermodinamics, so I can assume it is not linear but I think it works someway as I have said, does it not?
RPM= const / T
so the RPM of your fan scale inversely with the temperature of the air.Therefore, the higher the temperature the slower the fans (and the inverse holds true, too).
Until the electronics fail, anyway ;)
Ok if I have not misunderstanded, the more airflow (at less temperature) is moving around the heatsink, the more Q it can evacuate from the heatsink. So less airflow => more T at the heatsink.

The last thing I want to ask you is the following: I've read that the nearest the fins of the heatsink are, the more difficult for the airflow is to move. So, if you have closer fins, you have to blow air at higher pressure in order to circulate it through the fins. More pressure usually means more RPMs from the fan so higher noise. Then, a combo of a heatsink with the 'proper distance' between its fins + a common fan, makes better scores in temperature than a heatsink with closer fins at same fans' speeds. Am I correct?

So, if you have to score same temperatures, you have to increase your air pressure. Then, the better fan choice for the heatsink of before, would be one that can move a high air volume at the lowest rpm (which mean less noise), wouldn't it?

Many thanks for your/any other user help :)

kater
Posts: 891
Joined: Thu Sep 07, 2006 11:20 pm
Location: Poland

Post by kater » Mon Jan 25, 2010 7:28 am

A review/platform to end all "best CPU cooler out there" quibbling.

I'll also join the crowd asking for low airflow tests with undervolted CPU 8)

On a side not, taking into consideration results from these two tests (newest platform and Copper Ninja - am I correct to understand that the original Ninja (my bestest PC gadget so far...) is still in the top 3 (4?) coolers out there (equal to NH-U12P)? I mean, I can read, but it's just ... wow :) Praise be to the original mounting clips :)

ValueDriven
Posts: 1
Joined: Mon Jan 25, 2010 7:40 am
Location: Ohio

Multiple Fan Configurations?

Post by ValueDriven » Mon Jan 25, 2010 7:48 am

I believe previous SPCR testing indicated that the mugen-2 really shines in a two fan configuration. :D Other heatsinks might also change rankings using multi-fan arrangements.

Now if I could only decide for my new Q9550 build. :shock:

kosmonautbruce
Posts: 22
Joined: Thu Nov 01, 2007 4:12 pm
Location: San Diego, CA
Contact:

Post by kosmonautbruce » Mon Jan 25, 2010 8:50 am

Great job! I love getting all the technical information/concepts behind your testing methodology.

I think others are correct when they express skepticism of fanless testing. That is something that probably should be done more usefully in case reviews. That is, try fanless cooling in a particular case using some of the most popular heatsinks.

Das_Saunamies
*Lifetime Patron*
Posts: 2000
Joined: Tue May 15, 2007 1:39 am
Location: Finland

Post by Das_Saunamies » Mon Jan 25, 2010 9:19 am

Nice work bringing the rig into this decade.

Looking forward to two follow-up articles: baseline lineup in-case and same lineup with a 65W or even 45W CPU. These just to show folks what variation if any there will be in those cases.

My E8400 has a TDP of 65W, and was quite happy with both a Scythe Mini Ninja and a Xigmatek HDT-S964. Nowadays it's got a newer-revision Noctua NH-U12P on it for surefire mounting - don't want to repeat what happened after the case got dropped.

kjetil
Posts: 14
Joined: Mon Nov 09, 2009 3:05 am
Location: Oslo

Noctua DH-14

Post by kjetil » Mon Jan 25, 2010 9:39 am

Hopefully Noctua will send their latest heatsink to Mike for testing very soon.

I just got mine yesterday and it ran passively reaching only 75 degrees during prime95 testing! (i920, asus p6t, default settings)

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Mon Jan 25, 2010 9:57 am

Good to see all the feedback, it's all being considered. ;)

ValueDrive: regarding your Q9550, this apropos info was just added to the review:
Surprisingly, the 130W TDP Intel QX9650 drew only 66W at peak load (including losses in the VRM) on the existing Asus P5Q-EM platform — it ran 20W cooler than the old Pentium D 950. (This result was closely matched in a more elaborate test by Lost Circuits.)

danimal
Posts: 734
Joined: Mon Jun 08, 2009 2:41 pm
Location: the ether

Post by danimal » Mon Jan 25, 2010 10:34 am

xlnt upgrade to the testing methodology! however, either it's not quite right yet, or i am misunderstanding something?? could someone please clairify this for me???

there is a rather thorny problem with the same reference fan being used on all of the coolers, because it is inferior to the stock mugen 2 fan at higher rpms, by about three degrees, as your earlier tests proved:
http://www.silentpcreview.com/article961-page5.html

the xbit labs testing proved that as well:
"The tower-heatsinks – Prolimatech Megahalems and ThermoLab BARAM – were topped with a fan from Scythe Mugen 2 with the rotation speed varying between 250 and 1300RPM."
http://www.xbitlabs.com/articles/cooler ... up_18.html
using the same mugen 2 fan, the mugen 2 and the Prolimatech Megahalems were both at 42/72 degrees:
Image

the high-impedance mugen 2 has far more cooling surface area than any other cooler in this latest spcr test, so it needs a real fan, something that puts out more pressure than the quieter but rather gutless spcr nexus reference fan.

i think that spcr should be doing the cpu cooler testing with the fan that shipped with the cooler, because as it stands now, the cooling capacity of the mugen 2 has been mis-represented in this latest round of testing.

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Mon Jan 25, 2010 11:34 am

danimal wrote:xlnt upgrade to the testing methodology! however, either it's not quite right yet, or i am misunderstanding something?? could someone please clairify this for me???

there is a rather thorny problem with the same reference fan being used on all of the coolers, because it is inferior to the stock mugen 2 fan at higher rpms, by about three degrees, as your earlier tests proved:
http://www.silentpcreview.com/article961-page5.html
Not really. At the same SPL/rpm, there is no appreciable cooling difference between those fans. We will not change the reference fan unless and until we find a 120mm fan that's clearly quieter at the same airflow level. There are some contenders in the huge Scythe lineup, but we haven't proved it for ourselves yet. Never enough time.

dev
Posts: 43
Joined: Sun May 10, 2009 12:47 am
Location: Here

Post by dev » Mon Jan 25, 2010 11:56 am

Welcome to the new decade Mike. :)

I think the Megahalems will keep the crown for a fairly long time given it was engineered to work best at sub 1200 rpm.

As a suggestion: I think twin tower heatsinks should be tested with 2 fans. I know this might contradict the silent argument but they were designed as twin fans products. If you test them with only one 1000 rpm fan theres a good chance you won't get the best out of them.

Please lets not open the Padoras box with the whole Prescott vs Pressler vs modern cpus argument. Those are things of the past and that is where they should be. By now most of you should've realized 1366 chips are very hot and the new generation of heatsinks are build around them. If you're still not convinced heres what a fanless Noctua NH-D14 can do with a Q6600:

http://forum.lab501.ro/showpost.php?p=2 ... tcount=161

A small addendum: you forgot to mention the poor base quality on the Thermalright products.

RicaNeaga
Posts: 7
Joined: Wed Sep 03, 2008 10:31 am
Location: Romania

Post by RicaNeaga » Mon Jan 25, 2010 12:41 pm

Good to see that almost all my arguments from this thread were taken into account, since the result is a MODERN test platform.


I hope in the near future more heatsinks will be added to the new testbed (CM Z600 and TR HR-01+ come first to mind), and also that you will forgive burebista for his behaviour (he was right, but his tone was... harsh); I think the ultimate argument in his favor is that he is a mega-silent-freack. :lol:


Congrats once again!

Wibla
Friend of SPCR
Posts: 779
Joined: Sun Jun 03, 2007 12:03 am
Location: Norway

Post by Wibla » Mon Jan 25, 2010 12:42 pm

Its not fanless with that fan sucking air out, you know ;)

I'm impressed by the fact that the "mighty" TRU120E is hanging on as well as it is, as its quite old by now...

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Mon Jan 25, 2010 1:13 pm

A few points that may have been missed:

1) Most current CPUs actually run cooler than our old Pentium D 950. A 130W TDP C2 Extreme X9650 pulled just 66W at the AUX12V socket, compared to 86~88W with the P-D950. The 65W C2D pull 50W or less. Most 95W AMD processors pull no more than 90W, usually less than 85W.

2) Only the 125W and 140W Phenom IIs and the 130W TDP i7s run significantly hotter than our P-D950. These represent perhaps 10% of current CPU sales.

All the above info is based on weeks of hands-on testing of a huge # of CPUs. We'll pull a lot of this info (and other data) together into a new version of Power Distribution within Six PCs.

3) As it stands, the new test platform is most useful for the OC / gaming silencer, who represent a sizable but still minority portion of the SPCR audience. As a result...

4) A lower power test platform looks unavoidable. We may keep the P-D950 platform alive and add a lower power test to simulate typical 65W processors. This could be used for smaller, less ambitious heatsinks; the i7 platform would be used for the biggies. Just thoughts for now...

javitxi
Posts: 64
Joined: Thu Jan 01, 2009 7:30 am
Location: Madrid (Spain)

Post by javitxi » Mon Jan 25, 2010 2:30 pm

MikeC wrote:3) As it stands, the new test platform is most useful for the OC / gaming silencer, who represent a sizable but still minority portion of the SPCR audience. As a result...
Mike, one little thing about this. I don't know so much about heatsinks, how to dissipate heat and so on, but you have to consider that if a heatsink mounted on an OC CPU in this case -which means a great source of heat- can transfer better heat than others, probably at idle temperatures or with less greater sources of heat (other CPUs /non OC-CPU /underclocked/volted CPU), it will work better than the others.

Yes, it sound obvious but from the user's point of view, he/she will evaluate mainly two important aspects: money & quality (and for us as silent as possible in balance with quality and money, for others these 3 things + aesthetics, etc). So, the user may think: 'well, this is a high performance (can cool very well) cooler and is affordable, so although I'm not running so power hungry CPU /OC-CPU -so high source of heat- it will cool my CPU very well in the worst case scenario'. Like ~40-50% of Pc users doesn't clean 'so often' his/her Pc, so dust reduce the performance which lead us to the worst case scenario.

Besides this point, some users may think: 'well, probably a less performance heatsink that costs less would perform the same as a high performance one with my CPU (which is a less source of heat), so I really need a high performance one for my Pc?' Example of this reasoning-thinking: Megahalems might be overkill to my i3-540, why don't give a try to Scythe Kabuto (i.e)? (In the example we are assuming that the hypothetical user has enough room to Megahalems and uses a Mid or Full Tower)

In conclusion, as many users have said, it would be great that you put not only idle temperatures, but also the power that the i7 is 'eating' at that idle moment which can give us an approximate idea about how much heat we should evacuate (you have that amazing tools -which I don't know the name in English :oops:, that I have on my job- at home!)

Probably you might have consider this things, so sorry if I'm repeating some ideas /something you have though earlier.

PS: something I've learnt over the years is that when you design something, you have to design it to the top worst case scenarios (although with these considerations taken, some little thing you have forgotten would break the system -Murphy's law ;) ). Thus I will ussually take a little bit more of what I really need, because this way, you don't force to work so hard to your equipment and it will make longer their lifespan (i.e PSUs working 24/7) providing you do maintenance periodicaly -I think the word in English would be preventive maintenance, but I'm not sure)

K.Murx
Posts: 177
Joined: Tue Mar 17, 2009 10:26 am
Location: Germany

Post by K.Murx » Mon Jan 25, 2010 2:41 pm

javitxi wrote:
Air can contain heat. The amount of heat in a certain volume of air is given by (Linearized, which is "good enough" for engineering):
Q=c_V * V * T
c_V in this formula, is the c_V of a monoatomic (3/2), diatomic (5/2), etc coeficient at constant pressure used on basics ideal gasses formulas?
The value does not matter. It just needs to be reasonably constant.
javitxi wrote: In the formula you assume V is the V in litres of your case, isn't it?
Well... No, rather an arbitrary volume around the heatsink where the temperature is reasonably uniform ;)
javitxi wrote: Also, if the air surrounding is hotter than the heatsink, by the equilibrium of temperatures theory between 2 objects, the surrounding air will transfer heat to the heatsink although the sorruonding air is moving. Am I correct? So, the lesser temperature (then less Q it has) the airflow through the heatsink has, the bigger amount of heat (Q) the airflow can carry out (evacuate) from the heatsink.
Whoa! If your intake temperature is hotter than the heatsink, please take the case out of the oven! I think it is safe to assume that the air is cooler than the heatsink ;)
But actually, heat transfer works more efficiently/faster when the temperature difference is greater. Therefore you want a big temperature difference between the air and the heatsink, which you can by either providing cool intake air (thus the importance of ducts), or by letting the heatsink run comparatively hot (as the graphic card manufacturers do).
javitxi wrote:
As mentioned before, once dynamic equilibrium is established, your fans must move the heat generated by the processor, thus Q/t is fixed
As far as I know, copper has better thermal conductivity than aluminium and actually, then most of modern heatsinks use heatpipes based on copper. The little I know about them is that when your heatsink reaches the maximum amount of heat he can conduce by itself, the extra heat is not transferred to the heatsink, thus it is retained by the CPU's capsule, growing the temperatures of the CPU's capsule.
So far you are correct.
javitxi wrote: So establishing an airflow through the heatsink will have the ability to carry out so much Q as the equilibrium between the 2 corpses can. I mean, if your heatsink is at 80ºC and you have airflow at 0ºC, then the equilibrium between these temperatures are 40ºC, so at a constant 80ºC by the heatsink, the 0ºC airflow will move out 40ºC. I don't know what the exact formulas are because I have not studied so much about thermodinamics, so I can assume it is not linear but I think it works someway as I have said, does it not?
You might be confusing static and dynamic equilibrium. If you would bring two exactly equivalent systems into contact, and one had 80º and one 0º, the static equilibrium would indeed result in both systems having 40º.
The dynamic equilibrium I meant is that the heat flow out of the system (the case) is equivalent to the heat per time generated within the system (due to the CPU). For the argument I made above, the exact manners of transport are irrelevant.
Regardless, I'll try to explain what I vaguely remember from lectures long past - if there are any true experts here, please chime in :)
In the neighbourhood of the heatsink, a heat transfer process takes place between the metal and the surrounding air. These processes are more efficient/faster the greater the temperature difference between air and the heatsink metal is. In the beginning, when we switch the CPU on, there is no temperature difference and thus no heat transfer. Thus, heat builds up in the CPU/heatsink, and raises its temperature. Finally, if we wait long enough, the heatsink will reach a certain "dynamic equilibrium" temperature so that the air can actually carry all the heat away (and if that temperature is too high your CPU fries).
What that temperature exactly is depends on a lot of boundary conditions, material constants and whatnot in complex ways. But it should scale at least approximately linear with the amount of air that comes into contact with the heatsink surface during a given time frame, which is essentially the argument given above.
javitxi wrote: Ok if I have not misunderstanded, the more airflow (at less temperature) is moving around the heatsink, the more Q it can evacuate from the heatsink. So less airflow => more T at the heatsink.
Correct again :)
javitxi wrote: The last thing I want to ask you is the following: I've read that the nearest the fins of the heatsink are, the more difficult for the airflow is to move. So, if you have closer fins, you have to blow air at higher pressure in order to circulate it through the fins. More pressure usually means more RPMs from the fan so higher noise. Then, a combo of a heatsink with the 'proper distance' between its fins + a common fan, makes better scores in temperature than a heatsink with closer fins at same fans' speeds. Am I correct?
In principle, yes.
Yet.... Well, if you have closer (=more) fins, the contact area between the air and the heatsink is bigger. Thus, heat transfer from heatsink to air is more efficient, so you could theoretically get away with a lower equilibrium air current through the heatsink than with more sparse fins. Thus, more sparsely spaced fins are not always better, because at some point the contact area becomes too small for efficient heat transfer.
But then again, more fins make it harder for the air to move, you need more pressure for the same airflow, and if you missed the "optimal" spacing (which probably also depends on the current temperature of air/heatsink), you will not reach the equllibrium current with the same amount of fan RPMs.

Now, the problem is that the pressure and air current a fan can generate depend on the geometry of the fan (oh, and generated sound depends on the geometry, too...). Due to this, it is hard to find the optimal fan/heatsink design. Which is why we are here, and the SPCR guys measure these things for us :)

Post Reply