Re: Thermal Cement?

[John Doe]

Customer returned a Linksys 8-port Gigabit switch which clearly
had something rattling around inside it. When I open up the
unit, I discovered a heat-sync had dislodged from the board.

If you are very familiar with using super glue, you can use that.
That means the pieces must be flat and lightly sanded with fine
sandpaper. Then pressed hard to dry.

Some might say otherwise, but I have been using super glue for
ages, and I have used super glue for heatsinks in my computers
that have functioned well. You can tell when the heatsink gets
very hot

If you are not skilled at using super glue, trying is probably a
bad idea.

--

 

[SC Tom]

If you are very familiar with using super glue, you can use that.
That means the pieces must be flat and lightly sanded with fine
sandpaper. Then pressed hard to dry.

Some might say otherwise, but I have been using super glue for
ages, and I have used super glue for heatsinks in my computers
that have functioned well. You can tell when the heatsink gets
very hot

If you are not skilled at using super glue, trying is probably a
bad idea.

+1. I've used super glue a few times on chipset heatsinks. The old Dell's used to have a spring-style retainer that if
the anchors pulled free of the MB, the PC would shut down and not start until the anchor was repaired. For some reason
(maybe poor QC, I don't know), there was a rash of them having the anchors pull out. To fix the ones where the trace was
pulled free, I'd solder a jumper wire in place and super glue the heatsink onto the chip. Problem solved

 

[Paul]

Would this include 'Krazy Glue' ?

A_C

Cyanoacrylate.

"Solvents and debonders

Acetone, commonly found in nail polish remover, is a widely available
solvent capable of softening cured cyanoacrylate.

Other solvents include Nitromethane, Dimethyl sulfoxide, and Methylene chloride"

That means Crazy Glue can be removed, but the acetone would likely
do cosmetic damage to the IC as well. If you use methylene chloride,
use in a well ventilated area (makes carbon monoxide in your
blood stream). I'd probably just stick with the acetone, because
I've still got a can of that.

Crazy Glue would not be a good conductor of heat, so you'd want
as thin a layer as possible. If you have surfaces that are milled
flat, and have a mirror-like finish, then you stand a chance of
having a very thin glue layer.

The reason the thermal epoxy gets any mention at all, is
it's loaded with thermally conductive particles. And if
it goes on a little thick, it doesn't hurt anything. It
still conducts heat. If it wasn't for the conduction properties,
people wouldn't use it.

Paul

 

[Paul]

+1. I've used super glue a few times on chipset heatsinks. The old
Dell's used to have a spring-style retainer that if the anchors pulled
free of the MB, the PC would shut down and not start until the anchor
was repaired. For some reason (maybe poor QC, I don't know), there was a
rash of them having the anchors pull out. To fix the ones where the
trace was pulled free, I'd solder a jumper wire in place and super glue
the heatsink onto the chip. Problem solved

I have a motherboard here with the same problem.

Only, unlike a Dell, there is no "continuity check" across the heatsink
on mine, as a means of detecting it fell off.

The reason that generation of solution didn't work well, is the
part that soldered into the motherboard (anchor), is made from the wrong
kind of metal. Normally, QC at a factory, would detect that during visual
inspection of the prototypes (bad solder joint), and request a fix.
Possible fixes include a different kind of solder (manual assembly step),
or (preferably) an anchor with tin-lead outer finish. If it had tin-lead
on the outside, it could be auto-inserted and soldered the normal way.

The material selection was done in the name of strength,
but, it's still the wrong solution. What good is a stainless
steel anchor, if it just pulls out ? Yes, it's strong, but
now the solder joint is the weak link, not the anchor metal
itself.

Paul

 

[SC Tom]

I have a motherboard here with the same problem.

Only, unlike a Dell, there is no "continuity check" across the heatsink
on mine, as a means of detecting it fell off.

The reason that generation of solution didn't work well, is the
part that soldered into the motherboard (anchor), is made from the wrong
kind of metal. Normally, QC at a factory, would detect that during visual
inspection of the prototypes (bad solder joint), and request a fix.
Possible fixes include a different kind of solder (manual assembly step),
or (preferably) an anchor with tin-lead outer finish. If it had tin-lead
on the outside, it could be auto-inserted and soldered the normal way.

The material selection was done in the name of strength,
but, it's still the wrong solution. What good is a stainless
steel anchor, if it just pulls out ? Yes, it's strong, but
now the solder joint is the weak link, not the anchor metal
itself.

That's pretty much what I saw with them. Most of the time, it was the spring
pulling the anchor out of its socket. But on a few, it was actually the
pass-through that pulled free of the MB, taking part of the trace with it.
Those were the ones that I used the jumper wire in place (using some
insulated ~40 gauge wire).

This was basically what I did to fix them:
http://www.davidcortner.com/slowblog/20080525.php

Except that I used cheap on-hand superglue instead of the more expensive
adhesive that I *didn't* have on hand

 

[KR]

If you are very familiar with using super glue, you can use that.
That means the pieces must be flat and lightly sanded with fine
sandpaper. Then pressed hard to dry.

Some might say otherwise, but I have been using super glue for
ages, and I have used super glue for heatsinks in my computers
that have functioned well. You can tell when the heatsink gets
very hot

If you are not skilled at using super glue, trying is probably a
bad idea.

We used super glue in the early 1980's to attach small heatsinks onto (then) high speed RAM chips that used to get very hot. At the time, folding a piece of aluminium into a U shape and gluing it to the top of the chip worked very well, but the surfaces in contact had to be flat, and very clean.

If you carefully tap and move the heatsink, you can break the bond and remove the heatsink later if needed.


If the unit is going to be put in a case, you can also try putting the heatsink on, and putting some sort of packing between it and that case to hold it onto the ic, but this might not work well.

 

[John Doe]

You have to do it right, the first time. Using super glue takes
some practice. Since you don't really know what it is, you might
want to exhaust other options first.

That means Crazy Glue can be removed, but the acetone would
likely do cosmetic damage to the IC as well.

Yes, of course you do cosmetic damage, but it doesn't matter as
long as you get it right.

If you use methylene chloride, use in a well ventilated area

It will burn your eyes. Sometimes you can see superglue smoke
rising up from the object you are applying it to. I usually
promote ventilation, but superglue is so extremely volatile that
it doesn't stick around. The initial timeframe is what matters,
especially like when keeping yourself from getting stuck to
things.

Crazy Glue would not be a good conductor of heat

I know I've asked before, but you really need to provide some
citations for that. And I'm not talking about somebody else's
opinion from a hardware overclocking site.

If the two pieces to be glued are flat, there will be an extremely
thin layer of superglue between them.

so you'd want as thin a layer as possible. If you have surfaces
that are milled flat

Yes, flat matters. You don't really have to worry about using a
thin layer, Paul, because it's going to squeeze out when you clamp
it or press on it. Sounds like you're not at all familiar with
using superglue.

And I would not recommend using a mirror smooth piece of aluminum
as the heatsink. It should be lightly sanded with fine sandpaper.

and have a mirror-like finish, then you stand a chance of having
a very thin glue layer.

What you're saying would take some trial and error to prove. And I
doubt that you're aware of, or have personal experience with, any
such trials.

On glass, something superglued can be snapped off. And then the
residual superglue can be scraped off with a razor blade.

The aluminum heatsink should be flat as possible and lightly
sanded with fine sandpaper, then cleaned. The chip being clean and
flat doesn't need to be sanded, because the superglue will
probably melt the surface of the chip. With enough pressure, the
melted chip surface should meld into the finely sanded heatsink
particles.

The reason the thermal epoxy gets any mention at all, is it's
loaded with thermally conductive particles. And if it goes on a
little thick, it doesn't hurt anything. It still conducts heat.
If it wasn't for the conduction properties, people wouldn't use
it.

How much experience do you have using superglue as a heatsink
adhesive, Paul?

You keep saying to use a thin layer, but that's not important when
working with superglue, since it squeezes out with pressure. That
is, as long as you apply the pressure right after applying the
glue.

There's nothing wrong with superglue as a heat sink adhesive, as
long as you are skilled at using it.

If you are trolling for encouragement to actually try using
superglue as a heat sink adhesive, Paul, hopefully that helps.

--

 

[Paul]

Paul <nospam needed.com> wrote:

I know I've asked before, but you really need to provide some
citations for that. And I'm not talking about somebody else's
opinion from a hardware overclocking site.

If the two pieces to be glued are flat, there will be an extremely
thin layer of superglue between them.

Start with this chart, which has thermal grease and thermal epoxy
ranges on it, as well as air (which is leftmost). Air is a relatively
good insulator, especially air trapped in a matrix (downfilled jacket).
Industrial diamond is an excellent material, and there is a datapoint
for it on the right.

http://upload.wikimedia.org/wikipedia/commons/1/1e/Thermal_conductivity.svg

Now, let's take one example of a cyanoacrylate, which happens to
have a thermal conductivity spec. Value is 0.1 W/(m.K) . Now,
place that on the previous chart, and cyanoacrylate is a little
worse than mineral oils. It's a better conductor than air.

http://www.permabondllc.com/TDS/920_TDS.pdf

Yes, flat matters. You don't really have to worry about using a
thin layer, Paul, because it's going to squeeze out when you clamp
it or press on it. Sounds like you're not at all familiar with
using superglue.

And I would not recommend using a mirror smooth piece of aluminum
as the heatsink. It should be lightly sanded with fine sandpaper.

Next, some more reading material. There is a second reason for
wanting a thin coating.

"Handbook of Adhesives and Surface Preparation: Technology, Applications and..."

http://books.google.ca/books?id=MEt...ylate properties thermal conductivity&f=false

"In general, aging properties of the cyanoacrylates are good.
...

Metal-to-metal bonds generally age rather poorly, except under
special conditions, whereby the minimal glue line is exposed
to moisture. <--- keep glue away from atmosphere... thin glue line
...

Prolonged exposure to temperature in this range results in a
gradual but permanent breakdown of the bond. Generally, the
upper temperature limit for continuous exposure is about 77C"

So in a heatsink application, where there is a chance of exposure to
elevated temperatures, the glue doesn't do so well. The strength
of the bond, may be reduced with time.

Even epoxy isn't a "wonder adhesive". Epoxy changes color with time,
and exposure to UV. It's properties also change with time. I could
not consider that for a lot of long term applications either. It's
just not chemically stable.

These kinds of properties are taken into consideration when
looking for materials to fasten things together. And
cyanoacrylates simply don't have the stability properties
needed for long (15-25 year) lifetimes. In fact, very
few materials have proven to be effective for "permanent"
cooling applications. (My employer was in the telecom industry,
where a long product life is a requirement.)

Paul

 

[larry moe 'n curly]

Would this include 'Krazy Glue' ?

Krazy Glue is a brand of super glue. It will work, but I don't like
it because sometimes the heatsink comes off all of a sudden if
knocked.

While thermal epoxy may be the best stuff for this, it makes removing
the heatsink later on difficult because it sticks so well, and it can
be risky to remove it with the usual solvents, lacquer thinner and
acetone, because the chip packages are also made of epoxy. So your
best choice is probably silicone rubber sealer because it sticks and
conducts heat well, can easily stand something like 500 degrees F
(260C). I am including the no-name stuff from the 99-cent store. Yet
you can cut through it with a razor if you ever need to remove the
heatsink. Almost every power supply contains sheets of silicone
rubber on its heatsinks, both for electrical insulation and thermal
conduction.

 

[John Doe]

Krazy Glue is a brand of super glue. It will work, but I don't
like it because sometimes the heatsink comes off all of a sudden
if knocked.

That is not the superglue that I am familiar with.

While thermal epoxy may be the best stuff for this, it makes
removing the heatsink later on difficult because it sticks so
well, and it can be risky to remove it

Then don't remove it.

I cannot imagine why anybody would ever EVER want to remove a
functional heat sink.

To each his own...

--

 

[larry moe 'n curly]

That is not the superglue that I am familiar with.

It was explained that if it comes apart at one end, it becomes very
easy to keep coming off, and this is a worse problem with super glue
than with epoxy.

Then don't remove it.

I was just answering somebody else

I cannot imagine why anybody would ever EVER want to remove a
functional heat sink.

To each his own...

What if you have to replace a defective chip but don't have another
heatsink?

 

[DK]

I cannot imagine why anybody would ever EVER want to remove a
functional heat sink.

The idea of replacing CPU never crossed your mind?

 

[KR]

Paul &lt;nospam needed.com&gt; wrote:

&gt; Agent_C wrote:
&gt;&gt; John Doe &lt;jdoe usenetlove.invalid&gt; wrote:

&gt;&gt;&gt; If you are very familiar with using super glue, you can use
&gt;&gt;&gt; that.
&gt;&gt;
&gt;&gt; Would this include 'Krazy Glue' ?

You have to do it right, the first time. Using super glue takes
some practice. Since you don't really know what it is, you might
want to exhaust other options first.

&gt; That means Crazy Glue can be removed, but the acetone would
&gt; likely do cosmetic damage to the IC as well.

Yes, of course you do cosmetic damage, but it doesn't matter as
long as you get it right.

&gt; If you use methylene chloride, use in a well ventilated area

It will burn your eyes. Sometimes you can see superglue smoke
rising up from the object you are applying it to. I usually
promote ventilation, but superglue is so extremely volatile that
it doesn't stick around. The initial timeframe is what matters,
especially like when keeping yourself from getting stuck to
things.

&gt; Crazy Glue would not be a good conductor of heat

I know I've asked before, but you really need to provide some
citations for that. And I'm not talking about somebody else's
opinion from a hardware overclocking site.

If the two pieces to be glued are flat, there will be an extremely
thin layer of superglue between them.

&gt; so you'd want as thin a layer as possible. If you have surfaces
&gt; that are milled flat

Yes, flat matters. You don't really have to worry about using a
thin layer, Paul, because it's going to squeeze out when you clamp
it or press on it. Sounds like you're not at all familiar with
using superglue.

And I would not recommend using a mirror smooth piece of aluminum
as the heatsink. It should be lightly sanded with fine sandpaper.

&gt; and have a mirror-like finish, then you stand a chance of having
&gt; a very thin glue layer.

What you're saying would take some trial and error to prove. And I
doubt that you're aware of, or have personal experience with, any
such trials.

On glass, something superglued can be snapped off. And then the
residual superglue can be scraped off with a razor blade.

The aluminum heatsink should be flat as possible and lightly
sanded with fine sandpaper, then cleaned. The chip being clean and
flat doesn't need to be sanded, because the superglue will
probably melt the surface of the chip. With enough pressure, the
melted chip surface should meld into the finely sanded heatsink
particles.

&gt; The reason the thermal epoxy gets any mention at all, is it's
&gt; loaded with thermally conductive particles. And if it goes on a
&gt; little thick, it doesn't hurt anything. It still conducts heat.
&gt; If it wasn't for the conduction properties, people wouldn't use
&gt; it.

How much experience do you have using superglue as a heatsink
adhesive, Paul?

You keep saying to use a thin layer, but that's not important when
working with superglue, since it squeezes out with pressure. That
is, as long as you apply the pressure right after applying the
glue.

It might be very important if you want to minimise having it running out of the join when clamped and getting into places where it isn't wanted.