AMD Heatspreaders...

Toggle sidebar Toggle sidebar
T

Teq

Distinguished
Feb 16, 2003
1,519
0
19,780
Some time ago someone here (I forget who now) jokingly suggested using thermal epoxy to mount heatsinks so they wouldn't move against the cpu if jarred.

Well, they gave me an idea...

I have a couple of Duron 1200s here that I really don't much care about... so I decided to see if I could actually add a heatspreader to an AMD chip.

I got a piece of 2mm copper plate, cut it to the same size as the chip, lapped both sides to be sure it was flat and bonded it to the Duron core with Arctic Silver Thermal Adhesive. This produced a truly strange looking CPU with a nice shiny copper plate on top of it's rubber feet, kind of a "double decker".

I then took an aluminum heatsink and had a friend mill 2mm off the bottom... to keep the overall height and clip pressure the same for testing.

Ok... now I have two identical CPUs, one with the copper plate, one without. I also have two identical heatsinks, one milled for the plate the other unmodified.

Stick in the D1200 without the heatspreader, apply AS3, mount up the sink and take some readings... Ambient 22c, Case 28c, Core Idle 44c, Core Busy 51c

Now, switch to the new setup, heatspreader, AS3, modified heatsink and repeate the same test conditions... Ambient 22c, Case 29c, Core Idle 35c, Core Busy 42c

YIKES! It makes that much difference? So I repeated the test and got essentially the same results...


Out of curiosity I tried an all copper sink on the unmodified D1200 with the following results... Ambient 22c, case 28, Core Idle 40c, Core Busy 48c

Finally I put on a hybrid of about the same size as the aluminum ones I used in the main test... Ambient 22c, case 29, Core Idle 40c, Core Busy 47c.

Now... the question... has anyone else tried bonding heatspreaders to AMD chips and did they get similar results???

(Once again... I'm wondering if this is reliable or is it a fluke?)







--->It ain't better if it don't work<---
 
Nice experiment!

So you found that the use of a heatspreader reduces the CPU core temp. Now what did I do with that Aluminium plate? Might have to get some thin copper plate from the shop.

Only thing I can come up with is that the Heatspreader delays the heat soak at the Heatsink and therefore a more uniform cooling system is the result. The heat spots on the CPU can/lid are more likely to be spread around and therefore the cooling at the heat spots are delayed, all speculation of course. I'd like to know whether AMD placed the sensor in a hot spot of the cpu circuit or somewhere nearby, placement would be important for this design.

Might also explain why theres one in the Intel Pentium 4 graphite pad.

Just an add-on for above, the CPU can/lid is also larger in area than the hot cpu core, copper is a better conductor for heat than aluminium therefore the copper heatspreader can aid the transfer of heat from the centre of the CPU can/lid to the edges and therefore reduce core temperatures faster, hence a lower cpu core temperature.

The all copper heatsink is a real puzzler, why do the all copper heatsink make it lesser in cooling capacity? maybe the circulation of heat at high frequencies makes it up to a very complex science and I am definately remote in experience on this field. I have noticed on some very expensive chips the top of the chip core has a large copper top and the substrate is even larger and no lid on top, this could be the design of the future for our desktop CPU's.

The mind boggles.


<P ID="edit"><FONT SIZE=-1><EM>Edited by wired_geek on 04/24/03 06:05 AM.</EM></FONT></P>
 
are you sayin a copper heatspreader epoxied to the core, with an aluminum heatsink/fan combo on top, did better than just a copper heatsink by itself, or a hybrid??
that's real hard to believe (though i believe you did the experiments correctly), especially because of the two layers of TIM between core and heatsink, not just one
weird

<A HREF="http://www.tweaktown.com/document.php?dType=guide&dId=120&dPage=1" target="_new">WinXP tweak guide</A>
<A HREF="http://www.tweaktown.com/document.php?dType=guide&dId=145&dPage=1" target="_new">WinXP tweak guide 2</A>
 
The difference is that copper is better at absorbing heat and moving it throughout it's structure, while aluminum is better at transferring heat to different compounds. That's why you get those composite sinks, which effectively have a built-in copper heatspreader so that though they are slower to dissipate heat, they have the large surface area to the aluminum which gets rid of the heat.

XP 2500+ Barton @ 198x11.5@1.85v
Thermalright SLK-800
Thermaltake Smartfan II
A7N8X Dlx
2x512MB Corsair PC3200
BBA 9700 Pro AIW
2x80GB 7200RPM Maxtor
Soundstorm
 
noooooooooooo that's not true. i'm here to dispel the rumor. while Teq insists that his experience bears out that aluminum DOES dissipate heat better than copper, a solid conclusion based on the actual physics says it simply isn't true. <A HREF="http://forum.oc-forums.com/vb/showthread.php?s=c6fb812f131190d4b3142f164f8a4070&threadid=187162" target="_new">i started up a thread on it</A> in the technical forum at www.overclockers.com to clear up the matter. i posted arguments that aluminum DOES dissipate heat better for the sake of argument, though i was unsure myself. you can look at the thread and see the results. i'm now working to dispel the rumor =)

<A HREF="http://www.tweaktown.com/document.php?dType=guide&dId=120&dPage=1" target="_new">WinXP tweak guide</A>
<A HREF="http://www.tweaktown.com/document.php?dType=guide&dId=145&dPage=1" target="_new">WinXP tweak guide 2</A>
 
It's got me stumped.

Reminds me of the old Volcano 6 and Volcano 5 heatsinks. Both heatsinks were the same design except the 6 had a copper insert which I believe was epoxied in place. Despite the copper in one these two heatsinks they actually had very similar performance.

I realize your setup is different but you do have a copper slug and one surface epoxied while the other is not. This does make for a bit of a similarity.

The closest I ever came to your experiment is sandwiching a 2mm coldplate between a heatsink and the CPU. The coldplate wasn't bonded to either surface. I just used thermal compound. As I would expect (now but didn't know any better then) performance dropped. I can't remember how much but it was significant. (Granted, this was not very scientific).

<b>99% is great, unless you are talking about system stability</b>
 
I was of the same opinion as those guys, that having a medium with better thermal conductivity would perform better, because the rate at which heat is transferred is dependent more on the state of the matter (solid, liquid, gas) than the type of medium. This would be especially true of metals, because their conductive properties are fairly constant (all are conductive, malleable, and have lustre), so what I'm trying to say is that I'm mistaken for thinking someone who has more experience actually knows what they're talking about, and that I should go with my gut feeling more than people I don't know.

Sorry Teq, I had to take the jab ;-)

XP 2500+ Barton @ 198x11.5@1.85v
Thermalright SLK-800
Thermaltake Smartfan II
A7N8X Dlx
2x512MB Corsair PC3200
BBA 9700 Pro AIW
2x80GB 7200RPM Maxtor
Soundstorm
 
not many believers in science around! do you think your slk-800 would work better with a copper slap at the bottom and the top of the heatsink being aluminum?
i feel like thermalright would have explored this option...especially since aluminum is lighter (and cheaper?) do you think thermalright has done this research or no? i'm inclined to say yes, after all, it's their business to know what works best, and they're pretty much on top of the market. it would be easy enough to construct a heatsink like that.

<A HREF="http://www.tweaktown.com/document.php?dType=guide&dId=120&dPage=1" target="_new">WinXP tweak guide</A>
<A HREF="http://www.tweaktown.com/document.php?dType=guide&dId=145&dPage=1" target="_new">WinXP tweak guide 2</A>
 
Hey, that's what happens when you step into the big leagues my friend. "Duh, stay outta Riverdale."

XP 2500+ Barton @ 198x11.5@1.85v
Thermalright SLK-800
Thermaltake Smartfan II
A7N8X Dlx
2x512MB Corsair PC3200
BBA 9700 Pro AIW
2x80GB 7200RPM Maxtor
Soundstorm
 
Hey, I agree. They wanted to make the bloody best heatsink on the market, and I think they succeeded. That's why they got my business. Come on, you and I are on the same boat here. We BOTH think Teq was wrong ;-)

XP 2500+ Barton @ 198x11.5@1.85v
Thermalright SLK-800
Thermaltake Smartfan II
A7N8X Dlx
2x512MB Corsair PC3200
BBA 9700 Pro AIW
2x80GB 7200RPM Maxtor
Soundstorm
 
Y'know what... thinking about this... I didn't want to start a pissing war. I just wanted to see if someone else had done anything similar so we could compare results...

Suddenly I'm sorry I even mentioned it.


--->It ain't better if it don't work<---
 
Nah not here, my workshop is not that well equipped. Only thing I did was to rip out the graphite pad from between my Intel cpu and standard sink at the computer desk where I sit. It got replaced with AS3 paste.

I use Motherboard Monitor and the Max CPU temp has fallen from 62C down to 60C, it gets this hot while running DirectX Multiplayer game with Radeon 9700.

I have to get some cash together and do some renovating as I can see myself doing a bit more of the experimenting.

I think copper has more free electrons so it can store more heat energy than aluminium and can draw more heat from the cpu than aluminium, hope I got that right.
 
hey sorry umheint0, i thought you were bein sarcastic

and Teq, i think your results are interesting..strange but interesting. and i'm not trying to start a pissing war either, just gotta say what i think and, you'll notice, respectfully mention your opinion on the matter as well. i'd like to see if someone else can recreate the results (i personally have no means to do so), or if you had a few more spare heatsinks and whatnot, if you could try the experiment with a different set of processors etc - basically to see if you could reproduce them yourself

also - it looks like AMD is FINALLY integrating them themselves, as well. they'll be made out of ceramic though, not copper

<A HREF="http://www.tweaktown.com/document.php?dType=guide&dId=120&dPage=1" target="_new">WinXP tweak guide</A>
<A HREF="http://www.tweaktown.com/document.php?dType=guide&dId=145&dPage=1" target="_new">WinXP tweak guide 2</A>
 
You know from where I sit its quite easy to think cooling. It gets pretty hot here in my Hobby/Computer room so I got myself a wall mounted box type aircondtioner unit, r/c of course. Because I live next to forest I got myself a coastal unit, just over 12 months old now also in design.

It has remote and the outside fins are of course aluminium but also has gold plating on fins for anti-corosive use. This design apparently suits snow and coastal areas. Now is this thermally efficient, apparently so, especially with the new economics and real commodity pricing around the place.

Pretty soon my CPU will also have Gold plated fins.

When? because at 3Ghz cpu manufacturers need to be aware of the 80watt plus microwave oven cooking away under the lid.

<font color=blue>Q. What is a small shovel?</font color=blue>
<font color=purple>A. A long day.</font color=purple>
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom