JB Weld copper shim to processor???????

[anvil]

I read, in OverClockers.com, a suggestion that applying a thin coat of JB Weld (Heat stable epoxy) to attach a cooler to the processor worked well. What's your thought on doing same to a copper shim. Then the contact area to the cooler can be large, while the contact between the shim and the processor is secure against misalignment.

anvil

 

[WarCon]

Should be ok unless you get it too thick and cause the core to make poor contact with the hsf.

 

[anvil]

I'm sure others have done similar things. In fact, I believe that this is also the same basic process that Step Thermodynamics used with their OC'd cpu's, but there may have been a more specialized "bonding" process used with the cooling plate.

anvil

 

[Mikewarrior2]

Other than bonding the shim to the cpu will do nothing for cooling(and in fact probably aid to heat production), you can probably do this...

But the fact that you don't need to begs the question: Why?


Mike

 

[anvil]

Why?

My understanding is that several heat sinks such as some of the Alphas as well as some Step Thermodynamics cooling systems, have (or have had) a copper plate bonded to the aluminum or the processor because the copper conducts heat better and the bonding process between the copper shim and the heat sink provides a better heat transfer than user applied liquid compounds which depend on the thickness of the coat and the eveness and pressure with which the heat sink is mounted to the processor. I do not know if the JB Weld compund compares favorably to other bonding processes, ... which is why I posted the question.

I do know that the attachment of the heat sink directly to the processor makes changing out either difficult, but a copper shim mounted on the processor would not prevent changing out system components, while providing the optimal heat exchange between the processor and the shim, assuming the JB Weld is a better and more reliable conductor of heat than what might be obtained with a misaligned heat sink. Since the shim can be made much larger than the processor face, there is more surface area contact between the shim and the heat sink face which improves the conveyance of heat from shim to heat sink and reduces the chance that a misaligned heat sink is interfering with heat reduction.

The primary question is whether or not the transfer of heat with an epoxy type bonding compound can be more efficient than a mechanical mount using one of the non setting liquid compunds. I am not claiming to know the answer. Just asking the question.

anvil

 

[Mikewarrior2]

Like I said before, you won't get any heat benefit..

Even though shim company's like to make yout hink that shim's make cpu's run cooler, due to increased contact surface with the heatsink, that is quite wrong:

Cool, highly conductive shim, one side touching hot heatsink base, the other side touching cooler cpu pcb... Where is the heat gonna go, its gonna go from heatsink to shim to pcb, which results in slightly higher core temps...

Copper embedded heatsinks are entirely different.

One thing you have to realize is that the core is solely the thing sticking up off the pcb... nothing else on the CPU is cpu core...

The real bottom line is, the shim is solely there for keeping a heatsink level, and some of those that are "too-tall" prevent even that... Only use them if you are worried and have made sure it doesn't interfere with proper heatsink-mounting... They do nothing for cooling whatsoever.


Mike

 

[Regalk]

I have to agree with Mike here - those shims are really meant for stabilizing the HS nothing more (some claim it prevents crushing the core and I can't prove this since I have never crushed a core even with a FOP38 which I have removed/installed several times). So I can't add to Mike's comments - its well put.

 

[anvil]

Forgive me Mike.

You have the commercial "shim" that is currently being marketed in mind and I did not mean that and never have seen one, but I intended the term as I have heard it used over the years in other applications. I apologize for the confusion. I simply meant a copper "plate" that sets entirely on the portion of the cpu that normally is in contact with the heat sink. That is, as you say, on the source of the heat that is conveyed to the sink. The function of this plate (or "shim," as I used it) bonded to the cpu face is to provide a more reliable and efficient conduit for the heat because of the bonding process, ... at least that was Step's claim. There need not be any additional conduction of heat "back" to the portion of the cpu board surrounding the raised chip surface. Ideally it should contact the cpu on its bonded face and the heat sink on its other face, and nothing else.

If the claim that the bonding process IS a more efficient conductor of heat to the copper plate (shim), then there is no reason why this would not reduce heat in the cpu more effectively, and no reason for it to return heat to the cpu board. (I will note, however, that Step did have a small amount of stabilizing material between the cpu board and the plate some distance away from the cpu, but it could have been a non conducting material, for all I know, or not even essential.)

I think it is just a question of whether or not the "bonding" process transfers heat more efficiently or not. I would think that the fact that others bond copper to aluminum suggests that they think this a better process for some reason.

anvil

 

[WarCon]

If you are really considering bonding a copper plate (cold plate?) to the core of your processor then you should at least consider a product made for endeavors like this. Arctic Silver has a thermal epoxy for mounting heatsinks to chip and if you read their website they recommend you thin it with regular Arctic Silver if you ever wanna get it off without breaking your core. But I believe alot of times those plates are used with peltiers though for cold distribution (or in reality a thermal vaccuum). Most good hsf's have a copper plate on the bottom for the same "heat spreading" effect. In addition you would have to bend your mounting clip or you will be putting too much force on your core and you will crush it. And if you are really gung ho about this I fully recommend a silver plate instead of copper. Almost twice the thermal conductivity of copper. Always wondered if a ounce of pure silver coin could be lapped well enough for this endeavor. Was reading a thermodynamics book today and saw something interesting. Compared to metal(Silver was something like 4.2), water(water was .5) sucks at removing heat, but silver ions in solution(was 7.6) are almost twice as effective as silver. Um new water cooler additive. hehe

 

[anvil]

WarCon,

Thanks! I was not aware of Artic Silver's epoxy. You would think that would be better than JB Weld. Who has it?

I had thought about silver too, but my ancient physics book states that silver conductivity is .97 versus copper at .92, which is not a lot of difference (aluminum comes in at .50 and iron at .16, just for comparison).

But a pure silver dollar would would do nicely, of course, and only cost a buck! If it works well, I don't guess it ever needs to come off. My cooler is homebrew water cooled BTW, so the clips are all custom anyway.

In the end, it is all about getting rid of the heat, ... and that is why I have considered these things.

anvil

 

[Mikewarrior2]

Anvil...

It might work okay if you intend to do a copper plate to the cpu core... but workmanship will make all the difference, as well interface materials...

Because:

You're introducing two new "gaps" between cpu and heatsink... If you have well-machined copper, with good thermal epoxy to bond the plate to the cpu to the copper, then try it..

You do run into one problem, that being it may be much harder to "mount" your heatsink due to increased cpu height...


Mike

 

[anvil]

Mike,

You are absolutely right about the mounting problem. It is a problem I am still fighting, but I have been trying hard to put together a near silent box since I acquired a video projector. DVD or Serious Sam at 80 inches is a sight to behold!

I am going after a quieter power supply next. If the GF2 is too noisy, I'll hook up a hose and water it too!

anvil