Copper Mod on Acer Aspire 4937G (NVIDIA G105M)

[chunlianghere]

i believe many of u experience gpu reaching near 100 degrees when gaming, causing laptop to jus turn off by itself.

a gd news here..copper shim solve the problem, which means it lower down temp to prevent tat from happening.

received copper shim in my mailbox yesterday.



thickness is 0.044", 1.11mm, 0.1cm (*actual fit thickness is 0.9mm, wan abit more thicker, get 1.0mm)


size is 0.75", 19.05mm, 1.9cm (kind of too big for gpu, becos the seller hav tat size for tat thickness, but nvm, look below)



wat i do next is to use a sand paper(fine one, not rough one), i use 1000cw since i had tat at home. lazy to go out to get more fine one.

do it on a flat surface, sand until like mirror effect.


now proceed to dismantle the whole laptop.


it seems tat the heatsink is too small for tat copper shim, but no need to worry, the surrounding alum is a little lower than tat copper.






using Arctic Cooling MX-2. apply on both cpu, gpu, n on copper shim after placing over gpu.



Stress test results

*Ambient Temperature: 32 degrees

Before copper mod, wit default thermal pad. i immediately stop the test b4 it blackout.



After thermal pad was replaced by Copper Shim.
temperature was still steady at 88 degrees after 11mins!

[BruBoo]

No thermal compound ?

[TehSuigi]

The copper shim acts as the thermal interface, and if done well can perform more effectively than compounds.

[BruBoo]

hmmm

I guess the best proof of that would be for someone to apply a micro-thin layer to both sides of the shim. . . . (I do mean micro thin - the aim of thermal compound is to fill the valleys on both sides if the interface while adding little or no distance between the highest peaks. It is not glue or bath sealant, but conducts heat much better than an air gap !)

In theory two mirror-flat surfaces perfectly aligned do not need or benefit from thermal compound. However 1000grit paper does not produce a mirror, and screw down GPU+CPU cooler assemblies do not produce perfect alignment. And with a shim there are two interfaces now compared to one before.

The 1mm size is interesting though . . wow. if the mechanical design (or some mod) creates a gap that big between the cooler and the GPU top when the CPU is down tight then filling 80- 90% % of that gap with Copper shim rather than thermal compound is a good idea (copper is a much better heat conductor) . I cannot see any scientific reason to have a shim otherwise... Many many Laptops do fix the CPU - cooler gap with screws and leave the GPU to take its chances with a thick pad or a non-pressure type thermal compound like akasa pro+ 5022. As there will be variations in the mounting height of the CPU socket and the CPU itself some float somewhere is clearly needed but 1mm sounds like poor engineering practice . . .

So I think these results nicely demonstrate the rubbish performance of thick thermal pads rather than the best cooling that is possible.

Also tempted to ask if any compound was (re?) applied to the CPU as that is not remarked on either !

[chunlianghere]

Originally Posted by BruBoo

hmmm

I guess the best proof of that would be for someone to apply a micro-thin layer to both sides of the shim. . . . (I do mean micro thin - the aim of thermal compound is to fill the valleys on both sides if the interface while adding little or no distance between the highest peaks. It is not glue or bath sealant, but conducts heat much better than an air gap !)

In theory two mirror-flat surfaces perfectly aligned do not need or benefit from thermal compound. However 1000grit paper does not produce a mirror, and screw down GPU+CPU cooler assemblies do not produce perfect alignment. And with a shim there are two interfaces now compared to one before.

The 1mm size is interesting though . . wow. if the mechanical design (or some mod) creates a gap that big between the cooler and the GPU top when the CPU is down tight then filling 80- 90% % of that gap with Copper shim rather than thermal compound is a good idea (copper is a much better heat conductor) . I cannot see any scientific reason to have a shim otherwise... Many many Laptops do fix the CPU - cooler gap with screws and leave the GPU to take its chances with a thick pad or a non-pressure type thermal compound like akasa pro+ 5022. As there will be variations in the mounting height of the CPU socket and the CPU itself some float somewhere is clearly needed but 1mm sounds like poor engineering practice . . .

So I think these results nicely demonstrate the rubbish performance of thick thermal pads rather than the best cooling that is possible.

Also tempted to ask if any compound was (re?) applied to the CPU as that is not remarked on either !

thread updated!

yes, i do apply thermal compound to both side of copper shim.

and also, after taking some time to look at the heatsink, it seems that the heatsink is not completely flat.

[BruBoo]

Cool (sorry couldn't resist that)

The whole idea of 1 cooler on two separate flat topped chips is crazy but the thermal compound usually manages to bridge the gap and make it all work.
There are some lovely stories of 'another' make of laptops where the gap from cooler to GPU is big enough to insert a coin! No surprise when some of these fail after a while :-)

[g9000z]

This is nice, I've tried and success...

can play dirt2 with medium smoothly without shutting down...

but still need cooler pad below it..

[ptrichardson]

Could anyone tell me if it would be a good idea to do this on my Aspire 7520 Northbridge (GPU)?

If so, what size would I need?

[chunlianghere]

Originally Posted by ptrichardson

Could anyone tell me if it would be a good idea to do this on my Aspire 7520 Northbridge (GPU)?

If so, what size would I need?

no point doing it on northbridge..

[yalcin19]

IC7 compound is claimed to have thermal conductivity of 2000-2500W/mK which is ~5x better compared to pure copper 406W/mK. Why then have a copper shim in between to create more transitional losses?
Innovation Cooling 7 Diamond CPU Thermal Paste Grease - eBay (item 260651877550 end time Feb-12-11 17:21:37 PST)