• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Cooling-The Definitive Guide

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
Thx for the rationale. I'm in the process of designing a custom case, and was thinking of reducing the size, but now I may not.

Thx for the rationale. I'm in the process of designing a custom case, and was thinking of reducing the size, but now I may not.

Thx for the rationale. I'm in the process of designing a custom case, and was thinking of reducing the size, but now I may not.

@Dr.Octopus: When you accidentally multi-post the same comment, you can go to edit, then put a check by the delete button, then click the delete button, and this will remove your multiple posts.

@Maximus Nickus: I have nominated this thread as a "most useful thread" in stompah's contest in GC, however I wanted to suggest an edit of the lapping process as described in your initial post. The best way to lap has been shown to be this: push straight in one direction, lift, repeat. This has been observed by Billa and noted as a known fact by machinists when attempting to get a flat surface. Using the method you mentioned, though the 90 degree rotation may help, can cause the surface of the heatsink to become convex. Here are some quotes:
Originally posted by Graystar
First, it is well known in the tool-sharpening industry that using such a procedure as he described will produce a convex surface. The reason is that when the block moves over the paper, the front edge will grab more than the center or rear. He even noted that this would occur. However, he didn't make the connection that his surface was not flat. Woodworkers have long known that in order to get a plane perfectly flat, you must push it in one direction only, *then lift it off the paper* to return to the starting position, then push forward again. Anything else will not get you a flat surface.

Originally posted by Billa
follow the procedure that Greystar described above
its the same that I use

one direction, lift and return - never rotate
close to impossible to actually get it flat by hand, I never have
(and I can, and do, inspect for such)

be cool

Another quote from Billa regarding the procedure you described, but in reference to someone elses description of it:
that procedure is really good for making a hump right in the center of your hsf or wb

Just a suggestion to help out. :)
Thankyou for your comments, :)

It raises a very interesting issue actually, you see the convex surface will be created using the Lapping Process which I describe but this doesn't actually matter in the right circumstances.
You see the center of the Heatsink will be flat, and as this is the part that is in contact with the chip the edges can be convex and it won't make a difference.

The reason why I rotate the Heatsink is to even out the 'grooves' created by the sandpaper, this is because when moving the Heatsink in one direction the 'grooves' created by the sandpaper are slightly more profound and thusly there is less of the Heatsinks surface area actually touching the CPU's core.
Using the rotating method makes sure these 'grooves' are evened out and thusly the Surface Area touching the chip is increased and as a direct result the tempratures will be lower.

I can see exactly where you are coming at, and it can be useful to use that method when working on a Heatsink that is the same diameter as the chip, to illiminate the convex affect happening. But in this case as long as the edges of the Heatsink are not near the chip the rotation method generally produces better results.

I Thankyou for your input as its a very interesting point and I'm glad you brought it up,


dudes...i'm totally frustrated...specs: 1.6a on an abit bd7raid with 512 corsair 2100........antec case. cooler: swiftech mcx 478...my temps before lapping: 29 for board and 42 for cpu...temps after lapping: 29 for board 42 for cpu....i want to scream. i have my 1.6a clocked to 2.0 completely stable; however, i can't get anymore out of it until i get my temps down....i believe i followed the instructions pretty well. am i not to see even the tiniest scratches...is the copper supposed to be a perfect mirror?...any suggestions for where i might be failing will be appreciated....thx
There can really only be two plausible reasons why your temperatures haven't gotten any lower.
Either the Heatsink was lapped when it left the factory or you didn't clean the base afterwards.

If you could tell me what you used to clean the Base of the Heatsink and whether or not it was Pre-Lapped it will be a great help in analysing why the temperatures are the same.

I'd also like to point out that if you are using Arctic Silver then it needs a day or two to fully cure, in this period the temperatures shoud drop a few degrees.

max, thx for the reply....i'm not sure whether it was pre-lapped or not. judging by it's condition, i don't think it was....i didn't clean the base after : ( i thought that alcohol would just ad another layer and that brass-o was an option. i did wipe down with a lint free cloth however. so, this could be where i went wrong. i did use arctic silver 3 per the manufacturer's instructions....thx again for your help.
To clean the base you can simply use Water, as Isopropyl Alchohol does leave a very small film, (although I've used it and noticed no difference).

By the looks of it I would assume that the Lapping Process simply didn't have much effect on the Heatsink, it may have looked rougher before you lapped it but it still may have been smooth.

Thinking about it that particular Heatsink has been found to come pre-lapped anyway, which is why you noticed no decrease in your temps.

max..why the high temps with a well-lauded heat sink? i know people that are 6 degrees cooler with the stock pentium cooler....i'm 42 no load and 47-48 full load......thx.
There are many variables which affect the temperature of your CPU at full load, the airflow through the case has a big influence on your CPU temperature, if you think your temperature is too high I would look there for solutions.
Taking off the side of your case panel (and monitoring the temperatures) should distinguish if there is a problem with your cases airflow.
If you let me know what case cooling you have I'll do my best to help you lower your temperatures.

You may also need to remember that you are running overclocked, which is why your temperatures are higher, I'd also suggest re-reading my guide.
Often things are overlooked, its worthwhile as pretty much everything you need to know is in there.

Let me add some bit to byte:

The preference in cooling must be low noise,
some theory:
due to it affects to your processor in your head
that breaks down total system performance.

The complex solution need be done:

Does processor is more important for cooling?
or HD with often critical information?
I think -HD. It is statefull,
others- stateless - can be changed immediately.
And information and your time for (and nerves for) problems cost most much.
Processor has some stock for temperature up to 65 C,
for real no problem use,
so it is better to use for lowering noise,
and not processor cooling much important as people writes.
(real house helping device, not terrorizing)

So the real need - to cool, stable HD,
make low noise in case.

The real biggest Second problem - to pass out hot air
from case with no noise -
usually can be done by 120mm propeller.

Big propeller can be with big performance- and slowed for
low noise while sufficiently performs the air out.

Case must open only from back side- for the sound that good reflects -remains mostly in case.
Same partially about front-side.
That's why I do not like HD coolers with front side in case
coolers. They make noise, and deliver it to the room immedialtely,
with no stands along with a cold air. Simple solution, and no good one, how I think.
Small coolers need big speed, for visible performance, that usually means big sound to you.

I prefer "cool case within + with good air float in case" with HD solution.
Ot intelligent air pump in air way- case back hole -> case cooler
blowing to within case, - to
HD cooler - to HD.
Due to assumed efficiency due to almost direct delivery,
no high speed of coolers in need,
cooles in such a way can be of big size.

I usually place HD into 5' placeholders with not solid attaching materials, (usually sound isolating ) - to prevent big sound transparency of solid materials.

Practical advices:
The best solution ever: Zalman 7000
Set it and forget all CPU cooling and noise problems.

Even some effective cooling solution from Intel appeared -
big copper sentral VERTICAL stright line as fast hot transport and
aluminum sinks around - any cooler made by this geometry is very effective.
Fun made from by big squared "pages" almost horisontally placed, to move it to more effective mode
some at low speed.
Big square of aluminum dissipator.

It not used out of center ait flows, that is the big, to differ from Zaman -not used.
(for video-card cooling - as said Intel)

It is bad for low noise solution, when every air move need be used.

Zalman is better.

It is good to make Heat Pipe attached to HD and passed out of case for effective cooling of that critical place.
It is possible to cool Heat Pipe by Peltier outside of case where little water is not a danger.
Or air in case may by be cooled by Peltier.
As : Cold side within hot- outside.

Use like CpuIdle program - it colds procesor to 35 C,
all time you not play,
stops much regulated fun on processor,
lowers accidently for that time HD voltage to lower than 12 V,
that makes not him making any additional heat.

It is a question – how to create good Heat Pipe with no much cost and not involving industry production methods ?
It there good instructions?

2 fans = 1.2 fan by noise but twice performance.
3 fans = 1.3 fan by noise but three times performance.
4 = 1.4

To compare sound level better NOT - with human voice - it has fully different spectra and nerve reactabilty by human.
Practically fully Incomparable things.

The neon lamps noise, some else around devices like cold conditioner- is better.
Last edited:
Maximus Nickus said:

Washing up liquid is that Greeny slimy solution that is used to clean up your dishes. It is used to provide a smoother surface to stop the Heatsink from sticking, providing a flatter smoother end product.

You could use Brass-O (I use it myself) to polish up the bottom of the Heatsink / Waterblock, but I mainly use it for looks. You can certainly do without it because it doesn't affect performance of a newly lapped Heatsink.
If the base has tarnished however it will certainly help performance.

Good Luck,

NEVER use metal polish on heatsink interfaces, it clogs the pores of the metal surfaces with non-conductive particles, preventing thermal grease from doing its job.
great thread helped me a lot i lowered my sys temps about 3 degrees, but it is still a lil high maybe u can help me some more.

47C cpu on load
38C case on load

here is a sketch of my case:
X = exhaust
O = intake

they are all 80mm fans.

I am thinking of making a duct from the top side fan to the heatsink. it should lower my cpu temps by 4-5 so i've read.

i want to bring my sys temp down 4-5 more. its hot down here in Florida about 83F outside this time of year and its about 75F in my house and even hotter then that in my room.

what do you suggest?
Hi dris!
I would suggest first off making sure your case is totally free of obtrusions and obstructions to airflow.
The Duct sounds like a good idea to me, if you really want to lower your case temperatures you could always introduce some higher airflow fans, or even go a little more extreme and install some 120mm fans, they are much more effective and they are quiet too.

Let me know how it goes, :)

which cases are better for cooling, full towers or mid towers?

i have an APEX ATX 100FT (http://sudhian.dealtime.com/xPF-Apex_ATX_100FT_ATX100FT) and i have the 2 80mm fans int he top at the back and one 80mm int he bottom front but i also at the moment have the sides off the case and am using right now stock HSF for an AMD 1800+ and have sn idle temp of around 54-55C. i have been messing around with adding a blowhole and a 92mm fan ont he side panel above the processor but that only lowerd the temp by like 6-7C and its nowhere near what i see people around here getting. so im thinking maybe full towers arent necessarily the way for cooling maybe. i used to have a 92mm fan in the middle of the case blowing upward but have removed that as i found out i was gonna use that for the duct idea as its the fan that moves the most air that i have without buying anythng new. i was also gonna add another 92mm fan behind the processor in the back panel fo the case but i have yet to get a holesaw as i dont have any dremel bits that cut metal right now.

any other suggestions i can do to get the temp down a bit more maybe?
Last edited:
Have anyobe heard about using
Delta funs of 140x40 mm in quiet mode ?

They can produce much ait flow very silently at low voltage?
Due to big sizes?
I'm just starting out here so this is all new to me and your article was really helpful although the section on Pressure confused me a bit.


When building an effective cooling system you have to take into account the affects pressure has on a case, positive pressure (more airflow entering) will create a pressure of cool air and less dust will accumulate, however negative pressure (more airflow exhausting) with be exhausting more air than is available and thus as a result increases the dust level.
Therefore positive pressure is often the most preferred.
To achieve this simply have more air arriving than leaving!

I think I understand why you would want a positive pressure but I didn't understand your explanation of this. You can't have more air arriving than leaving cause the volume of the case is constant, right? So here's what I'm guessing is happening. With a positive pressure you have ALL your air coming in your filtered intake holes and MOST of your air leaves through the exhaust hole. Some minute amount (or maybe not so minute) sneaks out through all the other little holes in your case. On the other hand if you have a negative pressure within your case, ALL the air goes out through your exhaust holes and MOST of your air comes in through the filtered intake hole. The rest of air that comes in sneaks in through those little cracks and seams in the case which are, of course, not filtered. If this is the case, then depending on how much the air is leaked by the case determines how important it is to have positive pressure. But, like I say, I'm new at this and this is just my thoughts on it. Are there other reasons that you would want positive pressure? If you had an airtight case, would it matter if it were positive or negative pressure?
Once again, thanks a lot for your write-up I found it super helpful in understanding what I'm trying to get into here.
no need in positive or negative pressure,
primarily -
But in flow of air through the case.