COPPER CPU SHIMS: The Definitive Answer !!

  • Thread starter Thread starter rms
  • Start date Start date
rms said:
A typical lousy off-hand response. I'm running a 2yr-old xp, not a
mobile barton. And I suspect that *most* installations do not conform to
your definition of a 'properly installed heatsink'.

60C is where instabilities can begin to appear. Water cooling should do much
better.

--
Ed Light

Smiley :-/
MS Smiley :-\

Send spam to the FTC at
(e-mail address removed)
Thanks, robots.
 
BigBadger said:
Noozer said:
[...]

Uhm... If you touch somthing and it feels warm/hot then it IS
conducting heat and putting a sink on it will help keep it cool.

Depending on how you mean this, this is either very true or complete
bollocks :) Something "feels hot" if it conducts heat to your
hands/foot/whatever. In other words, the material you're touching is hotter
than your hand. However, this does not imply anything about how the material
is conducting heat to other destinations. For example, consider the
following system:
Ice ----- ceramic ----- copper bar ----- heat source
0 deg C ----- 18.8 deg C/W ----- 0.2 deg C/W ----- 200 deg C
(just making up numbers here). If you touch the copper bar, it will feel
hot, and you will get burnt. However, the bar will be conducting very little
heat. Why? Well., the total thermal resistance from the source to the water
is 20 deg C/W. So for the 200 deg C temperature delta, you get 10W of heat
flowing through the system. Hence at 0.2 deg C/W, the left had side of the
copper bar will be 200 - 2 = 198 deg C (with a linear gradient along it).

An somewhat related fact ... you can carry around shuttle heatshield plates
that are at close to 1000 deg C (internally) with your bare hands. They're
very hot, but their thermal resistance is so high that, after an initial
(very short) period where the top fraction of a mm cools, essentially no
heat is conducted to your hands (ie: you don't get burnt). I had an awesome
picture of this being demonstrated but I can't seem to find it any more.
However, at
http://www-pao.ksc.nasa.gov/kscpao/nasafact/tps.htm
There is the quote:
"Surface heat dissipates so quickly that a tile can be held by its corners
with a bare hand only seconds after removal from a 2,300 degrees F oven,
while the center of the tile still glows red with heat."

So, getting back to the topic :) , If you touch thte substrate and it feels
hot, then all you can say is that the material is hot. if the temperature is
close to the source temperature, then this implies that there's a higher
thermal resistance to a heat sink compared to the resistance to the heat
source.
yeah, everything conducts heat to some extent but not everything is a
'good conductor'..... Athlon XP's (which is what we are referring to)
do not have ceramic substrates. The substrate on an XP is some form
of resin, while I accept does conduct heat it does not do it very
well.

The thermal conductivity of the packaging is quite complex, since it's a
layered material, and each layer is actually broken up into a number of
areas by the tracing. The vertical thermal resistance should be fairly high
as there are quite a few interfaces that it needs to pass through, and a lot
of the distance travelled is through substrate. I suspect the horizontal
thermal resistance would be quite high as well, since a lot of the copper
groundplate in in a honeycomb form, and the traces are quite small. See
http://hill195.home.mchsi.com/Locked/carcass.jpg
(warning: link contains a graphic image of a mutilated CPU)
for details.
I'd hazard a guess that it conducts orders of magnitude less
well than the copper/aluminium of the heatsink that is in direct
contact with the CPU and therefore the quantity of heat that would
take the substrate>shim>heatsink path in favour of the easier direct
path to the heatsink is very very small.

The rough back-of-the-envelope calculations I did before showed that,
assuming the entire temperature difference was due to conduction through the
shim as opposed to improved mounting, the thermal resistance of the shim
path would have to be around 2.5 deg C/W (as opposed to about 0.6 deg C/W
for the direct path). This is the die->water thermal resistance btw. IMO
this is unlikely, so at least some of the difference would have come through
mounting improvements. Whether this was due to poor initial mounting or
general improved mounting due to the shim is not possible to say without
doing a few more remountings.
 
One way to settle this silly argument might be for others
to chip in with more data points ... so here's mine.

I did this with a system I can afford to lose if I botch
things: homemade system with an Athlon XP 2400+ on a
Asus A7N266-VM motherboard. Stock AMD heatsink/fan combo.
Not overclocked: 2 GHz = 15 x 133 MHz.

Temp's before and after adding a "copper cpu shim" (details
below.) Temperatures as reported by Asus PC Probe.

Before After
Idle temp: 36'C 35'C
Prime95 temp: 52'C 51'C

In other words, NO SIGNIFICANT CHANGE.

Ambient room temp was 22'C for both tests.

Prime95 was used because that was what the OP said he used.

The "copper shim" was cut from a sheet of jeweller's copper.
The results above were for 0.3 mm copper. I also tried
0.5 mm copper, but the additional thickness prevented me
from being able to clamp the heatsink down. And no, those
are not exact thicknesses - that is just the limits of the
accuracy of the cheap calipers I have at home.

In order to try to simulate the OP's setup, I also used
Arctic Silver on both sides of the shim - except it was
AS3 in my case because that is what I happened to have on
hand.
 
Rob said:
One way to settle this silly argument might be for others
to chip in with more data points ... so here's mine.

I did this with a system I can afford to lose if I botch
things: homemade system with an Athlon XP 2400+ on a
Asus A7N266-VM motherboard. Stock AMD heatsink/fan combo.
Not overclocked: 2 GHz = 15 x 133 MHz.

Temp's before and after adding a "copper cpu shim" (details
below.) Temperatures as reported by Asus PC Probe.

Before After
Idle temp: 36'C 35'C
Prime95 temp: 52'C 51'C

In other words, NO SIGNIFICANT CHANGE.

Ambient room temp was 22'C for both tests.

Prime95 was used because that was what the OP said he used.

The "copper shim" was cut from a sheet of jeweller's copper.
The results above were for 0.3 mm copper. I also tried
0.5 mm copper, but the additional thickness prevented me
from being able to clamp the heatsink down. And no, those
are not exact thicknesses - that is just the limits of the
accuracy of the cheap calipers I have at home.

In order to try to simulate the OP's setup, I also used
Arctic Silver on both sides of the shim - except it was
AS3 in my case because that is what I happened to have on
hand.

Your results are equal within measurement error and what one would expect
from the physics of it.
 
Temp's before and after adding a "copper cpu shim" (details
below.) Temperatures as reported by Asus PC Probe.

Let's put it this way, when somebody discovered grounding the heatsink
was going to give them a few tens Mhz in overclocks, even I heard
about it as it filtered through sites and online forums.

Copper shims aren't new especially before AMD started putting them
pads on the chip. If it can bring down temperature in any significant
way, I'm pretty sure them overclockers sites would had been onto it a
long while ago! :ppP

I'm of the opinion the originator probably didn't seat his HSF
properly in the first place. 60+ degrees celsius is pretty hot for
even an older XP. I had one and even at overclocked speeds with the
default heatsink and TIM, it was only in the high 50s. Full load with
my generally warm local ambient, no airconditioning.
--
L.Angel: I'm looking for web design work.
If you need basic to med complexity webpages at affordable rates, email me :)
Standard HTML, SHTML, MySQL + PHP or ASP, Javascript.
If you really want, FrontPage & DreamWeaver too.
But keep in mind you pay extra bandwidth for their bloated code
 
The said:
Let's put it this way, when somebody discovered grounding the heatsink
was going to give them a few tens Mhz in overclocks, even I heard
about it as it filtered through sites and online forums.

Copper shims aren't new especially before AMD started putting them
pads on the chip. If it can bring down temperature in any significant
way, I'm pretty sure them overclockers sites would had been onto it a
long while ago! :ppP

I think what people are trying to do with shims is
to emulate what some of the copper/aluminum heatsinks
do. The sinks with a copper bottom and aluminum fins
use copper's better thermal transport capability to do
a better job of moving heat to the outer fins so that
those fins contribute more towards cooling the cpu.
Some copper-based/aluminum finned sinks also have a heat
pipe in the copper base to help with this.

However, when you use a shim you probably negate any
such heat transport benefit because instead of having
a single layer of thermal compound between the cpu
and the sink, you now have three layers: compound,
shim, and compound.
 
Rob said:
I think what people are trying to do with shims is
to emulate what some of the copper/aluminum heatsinks
do.

The sinks with a copper bottom

Are doing what the integral heat spreader did.
and aluminum fins
use copper's better thermal transport capability to do
a better job of moving heat to the outer fins so that
those fins contribute more towards cooling the cpu.
Some copper-based/aluminum finned sinks also have a heat
pipe in the copper base to help with this.

However, when you use a shim you probably negate any
such heat transport benefit because instead of having
a single layer of thermal compound between the cpu
and the sink, you now have three layers: compound,
shim, and compound.

The bigger reason is it isn't in contact with the core whereas the copper
bottom of the heat sink is and already doing what the integral heat
spreader did.
 
I think what people are trying to do with shims is
to emulate what some of the copper/aluminum heatsinks
do. The sinks with a copper bottom and aluminum fins
use copper's better thermal transport capability to do
a better job of moving heat to the outer fins so that
those fins contribute more towards cooling the cpu.
Some copper-based/aluminum finned sinks also have a heat
pipe in the copper base to help with this.

That doesn't make much sense even if we ignore the point about there
being triple layers of compound/shim/compound to go through. One of
the key thing for the heatsink to work properly is strong pressure
pushing the sink and chip together. What's going to be pushing the
shim tight against the side of the core? If there's no way to get a
proper tight contact with the shim against the core, it would be
useless to transport heat away from the core. The packaging sure
doesn't conduct much heat since I know I can touch the package way
before it's not painful to touch the core :PpPp

--
L.Angel: I'm looking for web design work.
If you need basic to med complexity webpages at affordable rates, email me :)
Standard HTML, SHTML, MySQL + PHP or ASP, Javascript.
If you really want, FrontPage & DreamWeaver too.
But keep in mind you pay extra bandwidth for their bloated code
 
The said:
That doesn't make much sense even if we ignore the point about there
being triple layers of compound/shim/compound to go through. One of
the key thing for the heatsink to work properly is strong pressure
pushing the sink and chip together. What's going to be pushing the
shim tight against the side of the core?

The shim I experimented with was the exact size and shape
as the bottom of the heatsink. Putting the shim in contact
with the core is no different and no more difficult than
putting the bottom of a heatsink in contact with the core.

And because you are squeezing the shim between the heatsink
and the processor, there is actually more pressure - as is
evidenced by the increased difficulty in clamping the heatsink
down. Crushing the core or breaking the socket becomes a
significant risk.

Wish I had a camera. A pictures is worth a thousand
word, as the thriving porn industry amply demonstrates.

If there's no way to get a
proper tight contact with the shim against the core, it would be
useless to transport heat away from the core. The packaging sure
doesn't conduct much heat since I know I can touch the package way
before it's not painful to touch the core :PpPp

Contact with the packaging is irrelevant here.
The copper shim is used to more evenly distribute heat
from the core across the bottom of the aluminum heatsink.


And, of course, if the bottom of your heatsink is already
copper, then there is absolutely no point in trying this
silly experiment.
 
In comp.sys.ibm.pc.hardware.chips The little lost angel said:
That doesn't make much sense even if we ignore the point about
there being triple layers of compound/shim/compound to go
through. One of the key thing for the heatsink to work properly
is strong pressure pushing the sink and chip together.

The pressure is there to squeeze the thermal compound as thin
as possible. Both joints need to be good (thin and bubble-free)
What's going to be pushing the shim tight against the
side of the core?

I don't think there's much need to contact the die edge.
If there's no way to get a proper tight contact with the
shim against the core, it would be useless to transport heat
away from the core.

As I read it, the shim is sandwiched between the core and heatsink.
You don't want the shim near the package and those exposed jumpers
or caps. It basically makes up for a somewhat poorly designed HS
(too thin a base). Lateral heat transfer matters.
The packaging sure doesn't conduct much heat since I know I can touch
the package way before it's not painful to touch the core :PpPp

True enough.

-- Robert
 
Robert said:
The pressure is there to squeeze the thermal compound as thin
as possible. Both joints need to be good (thin and bubble-free)




I don't think there's much need to contact the die edge.




As I read it, the shim is sandwiched between the core and heatsink.

Unless he's doing something really strange it's a commonly available 'shim'
that sits on the chip carrier and not between the core and heatsink.
You don't want the shim near the package and those exposed jumpers
or caps.

The typical 'pre-made' shim is cutout around those areas to provide clearance.
It basically makes up for a somewhat poorly designed HS
(too thin a base).

Wishful thinking.
Lateral heat transfer matters.

What you're thinking would, at least, have a chance if you were talking
about an 1/8 inch think, or so, copper plate instead of an .027 inch thin
'shim' that could never be keep flat and in pressure contact across the
heatsink base. But, then, you wouldn't have a rat's chance in hell of
attaching the spring clip with a 1/8 inch plate under the base even after
the core was crushed flat.
 
Unless he's doing something really strange it's a commonly
available 'shim' that sits on the chip carrier and not
between the core and heatsink.

Why would anyone manufacture such a useless and potentially
dangerous thing? The chip carrier has very poor thermal
conductivity and you're never going to get the thickness
right. Either a too thick carrier will prevent die-contact
(horrors!) or a thin carrier won't contact the package (doing
no good).

What you're thinking would, at least, have a chance if you
were talking about an 1/8 inch think, or so, copper plate

Yes, that's what I thought. Nice slug of copper 3-4 mm thick,
shipped with a new bale or attachment mechanism to accommodate
the extra thickness.

The OP saw some improvement -- it must have been from a flatter
HS contact. I suspect s/he must have put the HSF on backwards,
or received an HSF with the bale switched.

-- Robert
 
Robert said:
Why would anyone manufacture

The original reason was the Athlon's tendency to get cracked cores from
heatsinks being 'rocked' on during installation. AMD responded by adding
rubber bumpers.

Then someone, long time before now, got the not very well thought out idea:
hey, if it's copper then maybe...
such a useless and potentially
dangerous thing? The chip carrier has very poor thermal
conductivity and you're never going to get the thickness
right. Either a too thick carrier will prevent die-contact
(horrors!) or a thin carrier won't contact the package (doing
no good).


That's been the point of the critiques.

Yes, that's what I thought. Nice slug of copper 3-4 mm thick,
shipped with a new bale or attachment mechanism to accommodate
the extra thickness.

There's a chance with a new bail but then not all heatsinks use the same
physical configuration so it's virtually impossible to make a 'universal' one.

The upshot is it's better to simply get a proper heatsink to begin with.

The OP saw some improvement -- it must have been from a flatter
HS contact. I suspect s/he must have put the HSF on backwards,
or received an HSF with the bale switched.

That's been the general speculation.
 
Back
Top