heat sink expansion idea

[skimzzz]

I thought of an idea about increasing the effectiveness of heat sinks. Someone tell me what is wrong with it. What if you took some copper wire, wrapped it around the heat sink, then extended it to the metal case and taped it.

I know this is crude, but it would extend the cooling area greatly. I don't think this would short the cpu/mobo since there isn't a circuit from the CPU to the heat sink.

If this is a viable idea, professional case makers could make efficient and nice looking versions of such a model.

 

[Saturn]

One of the fundamental ideas of heatsinks is to transfer heat from the heatsink to the air. It is already hard to transfer the heat from the processor to the heatsink, and that is why we use things like thermal paste. If you were to wrap wire around a heatsink, it would prevent airflow from the fan to circulate as well. Instead, the heatsink would act as a furnace and just heat up. The only way it could get rid of its heat would be to send it through this copper wire to the case. This would be very inefficient since heat does not travel through wires so well. What would happen when the heatsink and wire heated up? There would only be limited surface area to spread it to the case, and then the process would start over again. The case would heat all of your other components (in actuality the heat would never really transfer from the heatsink through wire to the case). Not an efficient, viable option.

<font color=blue> Blowing things up smells Aweful... </font color=blue>

 

[knowan]

I'm not so sure that what Saturn said is correct. In theory this could work, the main problem would be to get the heat to transfer from the single strand of wire to the case. I don't think that you could do it with just one strand.

Lets say that you have 1 foot of wire wrapped around the processor. You would then have 1 foot of wire collecting the heat. This heat then has to travel say, 8 inches to your case through a single strand of 1/16 inch diameter wire (all measurements are for illustration purposes only). It will transfer some heat, but will not be very efficient.

If you stick your spoon into a bowl of hot soup, does the end in your hand get hot immediatly? Will running a single copper wire from your freezer to your stove cool your stove effectivly?

I'll let someone with a physics degree quote that actual law of transference, but it is not the lack of air circulation that would cause this to fail. you would need to set up a heat pump, simular to what crashman proposes <A HREF="http://forumz.tomshardware.com/hardware/modules.php?name=Forums&file=viewtopic&p=641231#641231" target="_new">here</A>.

--------------
Knowan likes you. Knowan is your friend. Knowan thinks you're great.

 

[skimzzz]

It doesn't have to be a single wire. The point is that heat conducts through a metal like copper much more than it convects through air. If you could bridge your heat sink to the metal case (not that far a distance), then you would effectively be using a whole case as an additional heat sink for free. Plus, the case is already facing the outside air and has tremendous surface area exposure.

One issue would be whether this would heat the case too much.

 

[bum_jcrules]

The whole concept is a good idea but the physics is where the problems lie. On another thread there is the discussion of using heat pipes. This is also a good idea but again it is the physics that impedes the desired results.

The theory here is that if you add some kind of thermal conduit from the CPU to the case you could create a radiator with a large surface area.

Problem:

How do you get the heat out of the case? The thermal conductivity will have to exist with at a high enough rates to transfer it from the CPU to the case walls.

Please talk amongst yourselves.

<b>All for one and one for all...and 3 for 5! - Curly - The Three Stooges</b> :lol:

 

[knowan]

And how would you get a good connection between the case and the processor? We all know that a heatsink must have a good connection to the processor in order to be effective. That't why Artic Silver(tm) exists. How would you get a simularly good connection to the case? And if you do impliment some sort of screw down mechanism or such, how do you make it compatible with the large array of motherboards and cases currently on the market? A flexible hose perhaps?

Dont forget that the paint on a case will reduce the effectiveness of the heat dissipation.

I can see this possibly working for an individual setup, but it can't be marketed. In all likelyhood implimenting such a system would effectivly make it impossible to replace the mobo and processor.

Now the heat pipe that Crashman suggests could transfer the processor heat out of the case, but not to the case itself. It also suffers from some of the same design flaws. Mainly that it is only good for one case/mobo, and thus can't be marketed. It's not everyone that would be willing to cut holes in the top of their cases.

--------------
Knowan likes you. Knowan is your friend. Knowan thinks you're great.

 

[skimzzz]

I don't see why it would be hard to construct some type of metal bridge from the HSF to the case. In this setup, you would have a continous heat circuit from the CPU to the HSF to the metal case with 3 interfaces (CPU/HSF and HSF/metal and metal/CASE). If you use an unpainted metal case (esp aluminum), then this setup would use your entire case as part of the heat sink. Since the case is already exposed to the outside air on all sides, you dont need an extra fan.

With such a heat bridge to the case and if you ensure solid contact at all 3 interfaces, heat from the CPU would be RAPIDLY wicked away by the conductive metal to the case and outside air. Conduction is tremendously more effective than convection. True, convection would come into play on the outside of the case, but here, you have a tremendous amount of surface area AND you don't need a fan to draw the heat outside the case.

I don't know how efficient this would be, but this may even obviate the need for all those noisy fans inside. You would still have the fan on the CPU, but that would compound the effective heat transfer of this type of setup.

I don't see why this wouldn't work. It's based on the simple physical fact that conduction is more effective than convection.

 

[HonestJhon]

i think it is too great of a distance for the heat to be transferred to the case...
it depends on how fast the metal would loose the heat insiide the case...
and what you might want to do is look into making your own heatsink, because making it out of one piece, you would be able to make sure that you arent constricting the airflow on the heatsink, and make sure that you arent cutting off the heat transfer by just connecting two metals...if it is one piece, it is more likely to transfer heat efficiently...this thing has gotten me thinking about how possible it might be...and i think that it COULD work...but the distance from the cpu core to the case wall is so great, that i think that you would be not transferring the heat to the case, just transferring it to the air around the heatsink extension.......but it might not be too far...
it will take some designing, and some physics studyin...see which metal would move the heat fastest to the case wall, and not be too expensive for the common man to manufacture...hehe
lemme know what you come up with...maybe i can give some ideas on heatsink design, and figure out how to make this thing work.
i am very intrigued by this...

-DAvid

-Live, Learn, then build your own computer!-

 

[skimzzz]

I don't understand the questioning of the physics of this set-up. If you have doubts about conduction being better than convection, then try this:

Stick your hand out 10 feet from a fire with a fan behind it blowing your way. Then stand 10 feet from the fire holding a copper pipe which is also in the fire (fan optional).

 

[dejay]

This is an idea I'm sure a few of us have had at one time. The trouble I see is not the principal, but the implementation. The transfer medium from the heat sink to the case would have to be wide enough to transfer sufficient heat in order to be effective. A wire will obviously be too thin.

You either need something thick (which would require it being rigid), tubing with a liquid medium (water pumped to the case surface instead of a radiator), or lots of wires.

<font color=purple>I'm new here, I don't deserve a sig.</font color=purple>

 

[HonestJhon]

i am not really questioning the physics...
i am thinking along the lines of the implamentation.
like stated just before me, you might have a problem getting the heat to transfer efficiently through wires, since they are so thin...and it would be losing heat all the way to the case wall, which puts the heat back into the case...but this is going to happen anyways, so like it matters.
but i dont know how much heat is actually going to be transferred through the wires to the case wall...almost all will have been lost by the time it gets to the case wall.
could always just make a heatsink that is REALLY big, which will make contact with the case somewhere, and transfer heat efficiently..
i like this idea of utilizing the case as an extension of the heatsink....but implementation is the key.
i hope that you get it working, and if i can help at all, especially if you live in southern cali, let me know..i have some tools...and some spare parts laying around.

-DAvid

-Live, Learn, then build your own computer!-

 

[peteb]

okay logic, but flawed. In your fire example, the fire is many hundreds of degrees or more. Standing 10ft from the fire, you most likely get most heat from the radiated heat, not by air flow.

The copper pipe now has a huge temerature delta, it only needs to get to 60 or 70 degrees before it gets too hot to hold. The end in the fire though will probably be melting.

The issue is this. Your case is maybe 20-25 degrees (due to ambient temperature. You want your cpu under 60 degrees ideally (not overclocked) or way lower to overclock.

For heatsinks to work, the surface area must be as physically (thermally) close to the heat source as possible.

Look at your example - even if you dip the copper pipe in liquid nitrogen, is it going to cool the fire?

Heatsinks work to expand surface area as rapidly as possible as close to the cpu (heat source) as they can. They need a good base to act as the heat spreader, and then usually many fins from the base to dissipate the heat into an airflow. Your idea is okay, but it needs work.

Either:

Build/modify a PC such that the cpu can be strongly attached to a huge, highly thermally conductive sink (not by 8 inches of copper) such that that sink also has very good cooling, or increase the surface area and thermal conductivity of the materials that are close to the cpu.

You could, theoretically, build a sink larger and more effective than, say, a swiftec 462, however it would be custom for your board and be a devil to prevent cpu crushing. Potentially you could go to very low levels of active cooling even on an overclocked Athlon.

-* <font color=red> Under Offer </font color=red> *-
email for application details

 

[dejay]

Here is an idea, but I'm not sure how good it is.

Instead of plugging a cpu straight into the socket, you plug an extension into the socket, which consists of a high quality cable. You then plug the cpu into the other end of the extension, and in turn the cpu clips into a heat sink which is attatched to the case. This way the cpu core is in direct contact with the heatsink which in turn is in direct contact with the case.

The problems I can forsee with this is:

- The circuits in the mobo running to the socket may have to be a certain length and extending this length with an extension may not work. I don't know if this is the case, can anyone enlighten me.

- This would be something hard to do for the average tweaker. The quality of design required would be hard to achieve.

<font color=purple>I'm new here, I don't deserve a sig.</font color=purple>

 

[HonestJhon]

well, i think for the absolute max speed, you would want the traces coming from the cpu socket to be as short as possible, so i dont know if it would be defeating the purpose to have leads extending the cpu socket.
it might also allow interference with the signals running on the extensions.
but then again, i am not an expert on this subject, i am just thinking what could happen...anyone else?

-DAvid

-Live, Learn, then build your own computer!-

 

[peteb]

As you say, the traces all need to be a fixed, predictable length to elilminate crossover and predict latency.

Also, the CPUs all draw a huge amount of power. High amperage over a short wire/link is fine, but you'll need a hefty cable to extend it 8 inches off the board.

It needs to be cable to accomodate different distances to the case edge.

The case needs to be built with cpu clamping and heatsinking in mind.

Maybe possible using fibre in the future as the data connects.

Is it really going to be that much better than a regular heatsink to justify the effort? Just build a watercooler, which in thermodynamic terms does very much the same job.

-* <font color=red> Under Offer </font color=red> *-
email for application details

 

[knowan]

Here's a question everyone seems to be forgetting. Once the heatsink is attached to the case, how the heck are you going to open it?

--------------
Knowan likes you. Knowan is your friend. Knowan thinks you're great.

 

[dejay]

LOL.

You're right, but small points like this can be worked out after concepts are nutted out.

It reminds me of projects I've seen on the net where people put cooling fans in the sides of their case then realise that the cover won't fit back on because the fan hits part of pci cards or something.

At least they have the guts to admit their mistakes.


<font color=purple>I'm new here, I don't deserve a sig.</font color=purple>

 

[CALV]

how about a GIANT heatsink, sort of rightangled and finned, that went from the cpu to the case, then another large heatsink on the case, I bet it would work in theory, same as in a high powered amp where fans are a no-no , not practical but I bet it would work in theory, better off with a waterblock tho


Next time you wave - use all your fingers

 

[skimzzz]

I have an AMD and an Intel system and in both cases, the HSF is less than 4 inches from the case chasis.

I used some copper wire and some duct tape and was able to get a CPU temp reduction from 41 C to 34 C with this setup. And I can open the case just fine. It ain't pretty but it looks like it works.

 

[Saturn]

peteb made an excellent point with the fire example. The copper rod in the fire would reach hundreds of degrees at the one end before you could feel heat at the other. The heat you feel normally is most likely radiation at 10 ft away, not the air. Your processor and the wire close to it would be so hot by the time it went to the case that your processor would have already reached its thermal limit. I don't know what kind of setup you used where you stated a decrease in temperature with the wire. If you really want to test your hypothesis, take your fan off and connect your wire. Processors heat up rapidly, and there is no way it could spread through wire fast enough. If you do this it WILL kill your processor. Simple tests with a wire and heatsink could be putting your heatsink on a hot surface and seeing how fast you felt the heat at the other end (I am suggesting this with safety in mind- you won't get burned at the other end quickly because it takes forever for the heat to transfer!). Yes, some heat will transfer, but every time you connect two metals you lose conductivity. You can perform these kind of tests without using your computer. 1-Put a metal object in a hot liquid and see how fast the other end gets hot (when you can feel heat, take it out and feel the other end). 2-Put your metal end (however you have it set up:1 or many wires) in a freezer and hold your heatsink to feel coldness at the other end. These tests are not very professionally set up, but almost any body can do them. It is a good idea, but it is not practical. Paint on your case would interfere with the heat dissipation into the air. Heat would not transfer well between the processor and a metal medium. You would destroy your processor. I don't mean to be so blunt, but there are many huge problems. I once thought of something similar, but it had the same kinds of problems. Tell me what happens if you test any of the suggestions in this forum.

<font color=blue> Blowing things up smells Aweful... </font color=blue>