Opinions on Thermal Adhesive

Do you have a question? Post it now! No Registration Necessary.  Now with pictures!

Threaded View

Im working on passively cooling a VIA C3 I've got laying around and I'm
considering using a two part epoxy-like thermal adhesive to attach an
oversized heatsink to the CPU.  The CPU is integrated, so there is no
socket to clip the heatsink to, similar to a NB chip.  The only other
alternative is to drastically mod the heatsink so I can mount it to the
board.  I'm 90% certain that the oversize passive heatsink will work,
but I'd like to know how well the adhesive will perform before I
permanently attach it.  Anyone have experience with this stuff?  The
adhesive I have was provided with a Zalman NB heatsink, if it matters.

-Dylan C

Re: Opinions on Thermal Adhesive

On Wed, 11 Oct 2006 09:55:37 -0500, Dylan C wrote:

Quoted text here. Click to load it
Neven tried expoxy for a HS, but I've worked with it a lot building my on
golf clubs. The problem you may have with epoxy is that when it heats up
the bond is brokem. IOW's I use a heat gun blowing on the glubhead hossle
and the club shaft to remove remove the clubhead. I think most modren
cpu's would reach the temps required to break the bond. It's possible that
the epoxy supplied with the cooler might work, but I'd watch it closely. I
have heard of using superglue on the corners to hold a HS in place and
from  what I recall it works well. Just don't get too much on each corner
if you ever want to break the bond.

Want the ultimate in free OTA SD/HDTV Recorder? http://mythtv.org
http://mysettopbox.tv/knoppmyth.html Usenet alt.video.ptv.mythtv
My server http://wesnewell.no-ip.com/cpu.php
HD Tivo S3 compared http://wesnewell.no-ip.com/mythtivo.htm

Re: Opinions on Thermal Adhesive

Wes Newell wrote:
Quoted text here. Click to load it
Thats a decent idea.  I suppose I could try using two-sided tape as
well.  Thats what held the original heatsink on.

Maybe I could put a small drop of silicon adhesive on opposite corners,
and a thin coat of grease over the rest.  Or just omit the grease and
use all silicon.  I'm willing to take a few risks with the thing.  I
bought the board/CPU over 3 years ago for $5 at tigerdirect.  Since then
its just sat in my closet, so If i wreck it I'm not out much.

-Dylan C

Re: Opinions on Thermal Adhesive

Quoted text here. Click to load it

Do you see two holes next to the processor, like the C3 in this
picture ? The C3 in the upper right, has a hole on the left
and the right, and maybe those line up with a Northbridge
type cooler ?


Some coolers have adjustable arms, and they can adapt to a small
range of relationships between the holes and the chip to be


This is another example of a Northbridge cooler, only this has
a fan on it. This one has an exceptionally good theta_R
(thermal resistance) of 1.26C/W. Which is good enough to cool
a Tualatin, let alone a Northbridge. The mounting scheme on
this one (AMD version) can handle center to center on the
holes, of 52.1mm to 61.4mm.


The risk with thermal epoxy, is you only get one chance to
get it right. People who have tried to remove a heatsink
that used thermal epoxy, basically ripped the chip underneath
to shreds when trying. And some people managed to fit the
heatsink crooked (not parallel to the surface of the
chip), and regret the mess they made. It might help to
practice gluing two representative objects together first
with the epoxy, so you can get a feel for how tough it is
to get right.


Re: Opinions on Thermal Adhesive

Paul wrote:
Quoted text here. Click to load it
I have these holes, but the oversized heatsink covers them both up.

Quoted text here. Click to load it
This is very similar to the heatsink I installed on my Abit AN7.  The
problem is that If i want to passively cool it, I need to use a heatsink
substantially larger than one designed to cool a NB.  The one I have in
mind is ~60mm x 60mm CPU heatsink with the fan removed.  See a similar
80mm model here:

It would not be impossible to modify it to use the existing holes, but
I'd rather just glue it on if I can.

Quoted text here. Click to load it
What I'm worried about is how well the epoxy (designed for a NB) would
perform on a CPU.  I'd hate to fry the thing.  And if the temps did get
too hot...I'd like to be able to replace it.

-Dylan C

Re: Opinions on Thermal Adhesive

Quoted text here. Click to load it

The C3 at the bottom of this page, is listed as 10W max. I don't know
which C3 you might have, but it might be 10W or less. The Intel
865/875 generation Northbridges were 10-12W, so in fact that Northbridge
is in the same ballpark. And some of the Nvidia chipsets like Nforce4
are likely higher again in power (which is why they all tend to ship
with fans on them).


This is a table of data I've used a few times. Aavid has some heatsinks
for BGA packages, and these are some samples:

                   Length Width Height  still  with
                      mm    mm    mm    air    fan

35x35 374624B60024 35.00  35.00 10.00   23.40  7.55   Black anodize
35x35 374724B60024 35.00  35.00 18.00   15.30  5.15   Black anodize
35x35 374824B60024 35.00  35.00 25.00   12.00  4.27   Black anodize

The Zalman NB47J is taller than these heatsinks. As the fins get
longer, the added length contributes less to performance, so
there should be an asymtotic limit as the heatsink becomes higher.
You can plot the thermal resistance number (like the 23.40 number,
in units of degrees C per watt) in Excel, and extrapolate how
much height it would take to get good still air performance.
The Zalman NB47J is getting close to the right value, but it
would be really nice if Zalman would actually quote a number
themselves, instead of me trying to guess at it this way. The
NB47J only has taller central fins, which means the fin length
is not uniform across the whole 35x35 surface.

If we aimed for a 70C processor temperature, and ambient was 25C,
then we'd have 45C to play with. At 10W power, the thermal
resistance could not be higher than 4.5C/W. Then 25C + 4.5C/W * 10W
would give the 70C chip case temp. Plotting those values, in a
35mm x 35mm heatsink, and extrapolating, doesn't leave much room
for error.

So you are probably right to stick with your larger heatsink. If
you want to get some idea of the heatsink performance, have a look
at the Aavid web site. There is a section on extrusions, and looking
in there can give you some ideas on how effective your passive
heatsink will be.


And to fit your heatsink, all you need to do, is measure
the diagonal from hole to hole on your motherboard. Then
take a ruler, and eyeball with respect to the fins, whether
two holes can be placed between fins, such that the holes
have the same diagonal spacing. Purchase a couple of nylon
screws and nuts, as there is less chance of shorting something
by using Nylon. If the heatsink is big enough, the heatsink
temp should not be high enough to bother the nylon. Once the
first hole is drilled in the sink, you can transfer a centerline
from the trough between fins on one side of the sink, over
onto the flat side, allowing you to draw a diagonal line from
the first hole, to a point intersecting the centerline you
transferred. Drilling the second hole should be reasonably
accurate then. And if you get it wrong, there is plenty of
heatsink material to try again :-)

A couple of other notes on heatsinks. One performance issue
is called spreading angle. Basically, the base of the heatsink
has to be thick enough, to effective transfer heat from a
hot central spot, to the outer fins. Some heatsinks may look
impressively large, but if enough material was not used for
the base, the outer fins are not really that effective.

The other issue is mass. Years ago, for an application like
this, I was told not to adhere a heatsink larger than 50 grams
to the top of a 35x35 chip. If your heatsink really is as
large as you describe, securing it with screws might prevent
a sudden shock from decapitating the chip. Epoxying a heavy
heatsink, might mean damage if the board receives a shock
from being dropped a short distance.


Re: Opinions on Thermal Adhesive

On Thu, 12 Oct 2006 00:50:05 -0400, nospam@needed.com (Paul) wrote:

Quoted text here. Click to load it

The mfg states that overclocking may cause problems with that
heatsink.  They also state it's only for northbridges without
integral graphics to give you an idea of the limitations.  

We'd be better off with two chip designs for quiet motherboards.

Site Timeline