Fujipoly EXTREME Thermal pads

Well there are allot of threads dedicated about GPU/CPU Heat issues and i think allot of "stock" thermal pads are the main cause of bad heat transfer.

This thread will mainly concern the GPU thermal pad placement because thermal pads are mainly used for VRM's.
I doubt that the thermal pads supplied with my 680m are heat conductive enough (or too heat resistant) to get some proper cooling when (for example overclocking) pushing your GPU to a higher limit for a better game experience.

There isn't a single thread dedicated to proof real temp differences between stock and a "high end" thermal pads so i hope i can get some number crunching done.

If anyone else has FUJIPOLY pad experiences, please share!

I ordered 1/4 sheet of 150 x 100 x 0,5 - thermal conductivity of 11Wmk.
I also ordered a smaller sheet of 60 x 50 x 1,5 - thermal conductivity of 11Wmk.

There is also a more over the top sheet with a thermal conductivity indication of 17W/mk and that would be ideal except for a few big drawbacks:

1. it will not flex easily
2. it will not compress at all.

Upside:
Nothing except for copper will conduct heat faster than this..








The 11W/mk has way better flex quality and still has a minimal amount of compressibility, so that's why i wanted to go for the 11W/mk sheets.

For reference: the Arctic Cooling MX-4 TIM has a thermal conductivity of 8W/mk so just using the thermal pad in between the Die would conduct heat faster but the touch surface will not be as efficient so that won't work.

Conclusion:
Using a copper shim between the CPU/GPU Die with minimal highly conductive TIM would be the best option but then the conductivity would be as fast as the weakest link in the chain which of course is the TIM (rated at 8W/mk when referring to AC MX-4) so you would just need a miniminimini amount just to fill the microscopic gaps in the surfaces...anything more would work against the thermal conductivity and just be thermal insulating and thats the exact opposite of what you want.....you want heat going out as fast as you can, not keep the heat in

I'll try to get some work done with the benchmark numbers but i am not sure i can deliver the results today.

 

Good job and thank you for your effort mate, I hope your progress on the issue, since I'm also very interested in it.

 

Well done!! +1 Rep for the time taken to do this.

 

Well, as it turns out, i discovered that my heatsink is warped....this would explain the higher temperature
I am trying to find a way to compensate for the warped end....

 

Well i had to heat the "center island" that connects/binds the VRM and Die part and i think i managed to get a decent equal pressure back on to the Die.

Result


Before: 82 degrees Celcius
After: 77 degrees Celcius

So my results are of course not something that is guaranteed to work for everyone, i used 3 TIM's and oddly enough, not the MX-4 but the Zalman ZM-STG2 has a better result in cooling the Core.
Because the heat gets transferred quicker/more efficient in combination with the new pads on the VRM's there is less heat that remains stuck around the card which in turn keeps the core just a bit cooler i think.

Because i had some problems with the warped heatsink i can't really say for sure that one TIM is really better then the other, the Zalman is thicker and it could be that it makes better contact that way.

Benchmark:
I set 3dmark 11 to run the full experience 3 times over and note the highest temperature

TIM's used:
Zalman ZM-STG1
This is a high end TIM which is very easy to apply thanks to the brush that is included.
It cheaper and a very thin solution, and is less conductive then the other 2.

Zalman ZM-STG2
The bigger brother and the thickest solution of the 3.
This is classified as an extreme TIM and in my test had the best results.
This one is also the most expensive of the 3.

Arctic Cooling MX-4
The middle mode:
Not so thick and a good (and well known and proven) solution.
Funny enough i did not get the best results with this TIM, maybe because it did not touch the HS and the Die properly.
I think this one needs a longer "curing" or "setting" time...

I am using this solution on my 2600K (running @5Ghz WC) and does it's work (does not get hotter then 48 degrees Celsius).....that could also mostly be complimented on the high end parts of the res.pump and HS
I used the same clock settings on all (successful) pastes:
Core 888Mhz
Memory:2106Mhz
Volt modded: 1.000v (a higher volt would produce to much heat, i wanted to keep it in a "normal" spec.)

I hope this helps somewhat for people that want to use different cooling solutions, it was a nice little project for me and pointed out that my HS was warped so maybe this is something that you yourself can test.
Just apply the TIM and do a little mock up testing by re-assembling the HS in a normal manner and then disassemble again and see if the TIM has fully spread.

EDIT:
Reading this all back i concluded that it might just all have been related to the warped temps.
When i heated the HS back into it's normal position i never re-applied the original components....
Well when someone else tries....give me some feedback

Cheers.

 

Re-ordered 1/4 sheet of 150 x 100 x 1 (100% 16 months later), Gelid GC-Extreme Thermal Compound (TC-GC-03-A) 3.5g Thermal conductivity of 8.5 W/mK $10.99 + Tuniq TX-4 Extreme 3.0g Thermal conductivity of 6.53 W/mK $12.99.

[eXtreme-roundup] 26 Thermal Compounds tested - hwbot.org

http://i.imgur.com/tLAWRqZ.png

Thermal Compound Roundup - February 2012 | Hardware Secrets

 

this is probably a dumb question, but can u use thermal paste instead of thermal pads or both at the same time?.

 

Well stamatis from techinferno said that diamond 24 thermal compound is so thick and never bleeds that you can use on memory chips instead of thermal pads.

But its too expensive. ..

 

Pinoy, there is no sutch thing as dumb questions.

Yes you can use thermal paste in theory,but:
Thermal paste will liquify (or at least lose it sticky/grease state) when it introduced to a heat source.
This will mean it can run from between the heatsink and the VRM.

This will happen because the pressure on the VRM is not as high as the pressure between the HS and the Die.
As i stated in a earlier post, the thermal conductivity of a high end thermal pad can be twice that of a thermal paste so you must ask your self what you would gain from doing this.
The thermal pad also compresses and shapes around the VRM so the pressure does not need to be that high.

As for your second question:
Yes, you can combine, but again, what benefits would you expect from using this method, the thermal pad transfers the heat faster then thermal paste.

 

I`ve found these thermal pads only in ebay. But the seller does not sells to Latvia. Are any other possibilities on ordering to Europe?

 

Just order from Frozencpu

 

Thanks Ordered M4000 from Dell Precision for my Acer 5739G. This is damn hot thing x)))

 

You can buy fujipoly extreme or ultimate here: Ebay UK

 

Nice but not really required, standard stuff gets the job done even for extreme overclocking.