1 - 12 of 12 Posts

#### The_Nephilim

·
##### Registered
Joined
·
356 Posts
Discussion Starter · ·
Hey All,

I am getting a little nerdy here but I would like to do this a scientifically as possible.. Now what I am trying to determine is what is the required thermal conductivity needed (W/m-K) for a CPU running at 85degrees C with Copper as the heatsink.. I am really needing an answer to this not just I think this is the best..

here is the Formula I just need to know how I will find out the BTU's of said CPU??

thermal conductivity:

W/m-K = 1.7307 x Btu/hr-ft-F
W/m-K = 418.68 x cal/s-cm-C

Now I seen a few Shootouts of what they say is good and all I am just trying to take a more scientific approach to this problem, Perhaps someone here is as geeky..

I also want to point out if I only need a 3.8W/mK I will not bother with a higher amount, hopefully you see where I am going with this..

Or if there is an easier method to determine the required W/mK I am all ears..

#### cgipson1

·
##### Registered
Joined
·
824 Posts
I assume you are referring to the thermal interfaces ability to conduct heat, when you mention thermal conductivity?

If you know the watts that the CPU is rated at (or using in the 85C scenario you posted), you can determine BTU with the following formula:

P(BTU/hr) = 3.412141633 × P(W)

or the easy way: http://www.rapidtables.com/convert/power/Watt_to_BTU.htm

You can also convert Amps to Watts fairly easily if needed. Amps * Volts = Watts

There are other variables that come into play when trying to determine the exact amount of cooling something needs, such as heatsink / exchanger design / material and conductivity, cooling medium used, and its ability to absorb / transfer heat...

Unless you can specify all of the variables... you will not get an exact answer.

Due to that, the best answer is to use the most efficient parts possible in every variable...

In other words, don't use a crappy TIM if you can use a better one.

#### Lee17

·
Joined
·
1,061 Posts
Incomplete question... I would be happy to do an overall energy balance for your computer (since it is what I'm doing in my class, and my future job) but asking the thermal conductivity, is asking us how well your processor is delivering the energy from... I don't know, himself to something else, like a waterblock.

What you can do is to calculated the total output of your system in watts or whatever unit of power you like and you will know how many rad or which rad with which fan you need or which cpu air cooler fit your need.

Lee17

Ninja edit : If you are talking about thermal paste and want to know which one to choose base on the thermal conductivity, you have to know that better you thermal paste ( TIM for Thermal Interface Material), in other word, higher is the thermal conductivity, cooler your CPU will be with the same amount of paste.

#### cgipson1

·
##### Registered
Joined
·
824 Posts
Quote:
Originally Posted by Lee17

Incomplete question... I would be happy to do an overall energy balance for your computer (since it is what I'm doing in my class, and my future job) but asking the thermal conductivity, is asking us how well your processor is delivering the energy from... I don't know, himself to something else, like a waterblock.

What you can do is to calculated the total output of your system in watts or whatever unit of power you like and you will know how many rad or which rad with which fan you need or which cpu air cooler fit your need.

Lee17
Totally agree!

#### The_Nephilim

·
##### Registered
Joined
·
356 Posts
Discussion Starter · ·
Well I am basically just trying to determine the around average number (W/Mk) needed for my CPU..
I've seen thermal paste as low as 1.0W/Mk to a high of 20W/Mk..

What I am roughly trying to figure is what the sweetspot is in W/Mk for my Specific CPU?? does that make sense??

OK IF there are a lot of Variables what would you need to know?? I am NOT looking for an EXACT Defining number just a rought gestimate with known numbers??

I ran the BTU Conversion but what I got seems to high??

#### Lee17

·
Joined
·
1,061 Posts
Quote:
Originally Posted by The_Nephilim

Well I am basically just trying to determine the around average number W/Mk needed for my CPU..
I've seen thermal pase as low as 1.0W/Mk to a high of 20W/Mk..

What I am roughly trying to figure is what is the sweetspot is in W/Mk for my Specific CPU?? does that make sense??

OK IF there are a lot of Variables what would you need to know?? I am NOT looking for an EXACT Defining number just a rought gestimate with known numbers??

I ran the BTU Conversion but what I got seems to high??
The best answers is get the best you can! Also, don't waste too much money. Something like MX-4 is plenty for any purpose. You can go with something more fancy like Coolaboratory Pro/Ultra but , except for directly on the die of a CPU, it doesn't look like to worth its penny.

My 2 cents

Lee

#### ZealotKi11er

·
##### PC Evangelist
Joined
·
48,307 Posts
Quote:
Originally Posted by The_Nephilim

Well I am basically just trying to determine the around average number (W/Mk) needed for my CPU..
I've seen thermal paste as low as 1.0W/Mk to a high of 20W/Mk..

What I am roughly trying to figure is what the sweetspot is in W/Mk for my Specific CPU?? does that make sense??

OK IF there are a lot of Variables what would you need to know?? I am NOT looking for an EXACT Defining number just a rought gestimate with known numbers??

I ran the BTU Conversion but what I got seems to high??
It all depends if you are culling CPU die or CPU heat sink. The die will require much higher heat transferring material.

Taking VRMs of GPUs as an example. Stock pads did 5W/Mk, upgraded where 11W/MK. So a 25C drop in temps.

#### cgipson1

·
##### Registered
Joined
·
824 Posts
TIMS are one area where you should not skimp. The cooling efficiency of your CPU / TIM / Heatsink, etc... is not based just on one item, but how all the different components work together as a whole. Like Lee17 said... get the best you can, within reason. I like Gelid Extreme, myself...

Some reading (if you want to get scientific): http://electroiq.com/blog/2005/03/choosing-the-right-thermal-interface-material/

some decent testing of TIMs:

http://www.tomshardware.com/reviews/thermal-paste-performance-benchmark,3616.html

#### kaivorth

·
Joined
·
7,146 Posts
Quote:
Originally Posted by The_Nephilim

Hey All,

I am getting a little nerdy here but I would like to do this a scientifically as possible.. Now what I am trying to determine is what is the required thermal conductivity needed (W/m-K) for a CPU running at 85degrees C with Copper as the heatsink.. I am really needing an answer to this not just I think this is the best..

here is the Formula I just need to know how I will find out the BTU's of said CPU??

thermal conductivity:

W/m-K = 1.7307 x Btu/hr-ft-F
W/m-K = 418.68 x cal/s-cm-C

Now I seen a few Shootouts of what they say is good and all I am just trying to take a more scientific approach to this problem, Perhaps someone here is as geeky..

I also want to point out if I only need a 3.8W/mK I will not bother with a higher amount, hopefully you see where I am going with this..

Or if there is an easier method to determine the required W/mK I am all ears..
It looks like you're trying to find a TIM that isn't overkill to save money...

If that's the case, don't worry about it. 90% of TIM are under \$10 a tube. Get the best you can get for that amount. Plenty of reviews out showing the best and the worst.

#### AlphaC

·
##### ⤷ αC
Joined
·
11,278 Posts
No need to go overkill thinking about \$10 worth of thermal paste. Just grab some Gelid GC Extreme / Noctua NT-H1 / Arctic MX-4 / Prolimatech PK-3

The proper way is to setup a model as a thermal resistance network.

The thermal resistance network would be dependent on the surface area , contact (high pressure = less air gaps / less TIM), and the heatsink used (heatpipes greatly increase heat transfer) , fans on the heatsink, the ambient temperature & inlet air temp at heatsink

TIM is less thermally conductive than base to IHS (integrated heat sink on CPU).

Something like this:
http://www.electronics-cooling.com/2009/05/using-a-matrix-inverse-method-to-solve-a-thermal-resistance-network/

#### The_Nephilim

·
##### Registered
Joined
·
356 Posts
Discussion Starter · ·
ok sounds good will try and find a higher w/mk paste I can find thnx all..

#### Lee17

·
Joined
·
1,061 Posts
Quote:
Originally Posted by The_Nephilim

ok sounds good will try and find a higher w/mk paste I can find thnx all..
Look at what AlphaC suggest, you will be happy with that, I'm sure!

Quote:
Originally Posted by AlphaC

No need to go overkill thinking about \$10 worth of thermal paste. Just grab some Gelid GC Extreme / Noctua NT-H1 / Arctic MX-4 / Prolimatech PK-3

The proper way is to setup a model as a thermal resistance network.

The thermal resistance network would be dependent on the surface area , contact (high pressure = less air gaps / less TIM), and the heatsink used (heatpipes greatly increase heat transfer) , fans on the heatsink, the ambient temperature & inlet air temp at heatsink

TIM is less thermally conductive than base to IHS (integrated heat sink on CPU).

Something like this:
http://www.electronics-cooling.com/2009/05/using-a-matrix-inverse-method-to-solve-a-thermal-resistance-network/
Lee17

1 - 12 of 12 Posts