Direct contact waterblocks

Has anyone saw or built a waterblock that has a water seal and directly fits over the processor's heat spreader? No thermal paste, grease, etc.

I keep thinking about the physics of the whole cooling process and it seems to me that if you could remove any thermal junctions, your overall cooling effectiveness should improve.

I understand that it could be a messy process to remove a processor this way, unless you really did your homework with the drainage.

Anyone?

sounds interesting. i wonder what would happen if you covered a cpu with aquarium sealer, all but the core, of course, and then mount and seal a copper pipe over the top, and seal it too.

might look a bit messy if you are a slobb about it, but i kinda think it "MIGHT" work. i have an extra xp1600 that would be a good candidate of i was going to try it, but i doubt i want to sacrafice the rest of my pc in the event it does give away. lol

just some thoughts...

"edit" i am now thinking, what about a flat base on a copper pipe, and a rectangular hole cut into the bottom for the die, and the rest of the flat plate can be sealed to the rest of the cpu, and give a better seal than a pipe on the cpu like i was thinking before.

maybe if you have a fried chip....any kind....you can try it, and test for leaks before using it in the system

Actually, I'm sitting here looking at a plastic cube that has two holes in it and has one side gone. On the missing side is a channel with a rubber gasket embedded in it. I don't know about the angles of the inlet/outlet, but the darn thing is almost a perfect fit atop a P4.

Plastic is perfect because the thing never actually touches the hot part of the processor so it wouldn't melt, it's light weight; easy to install. It would be cheap to make a few million of these dudes--sell them for 3-4 bucks a pop.

I don't know, I might have a winner here.

My real question is, if this was any good, wouldn't there be a million of them in use already?

Dog_One said:

Has anyone saw or built a waterblock that has a water seal and directly fits over the processor's heat spreader? No thermal paste, grease, etc.

I keep thinking about the physics of the whole cooling process and it seems to me that if you could remove any thermal junctions, your overall cooling effectiveness should improve.

I understand that it could be a messy process to remove a processor this way, unless you really did your homework with the drainage.

Anyone?

Click to expand...

It is called 'direct die' cooling. Have a search at google: http://www.google.com/search?hl=en&ie=UTF-8&oe=UTF-8&q=direct+die+cooling

Dog_One said:

I don't know, I might have a winner here.

My real question is, if this was any good, wouldn't there be a million of them in use already?

Click to expand...

well, some of us asked the same thing about water , a year or two ago....

Re: Re: Direct contact waterblocks

nikhsub1 said:

It is called 'direct die' cooling. Have a search at google: http://www.google.com/search?hl=en&ie=UTF-8&oe=UTF-8&q=direct+die+cooling

Click to expand...

looks like a winner, according to that link. thanks....

I have seen some direct die experiment not so long ago. Surprisingly enough, the performances are not as good as a traditional WB. The big massive copper plate of a WB help the heat from the die to spread on a larger surface and also act as a 'heat buffer', therefore, it increase the surface in contact with the water. In a direct die solution, the surface in contact with the water is too small. That's what I understood...

about how much higher in temps was the on die compared to the block you used?

Maybe I misunderstood after all....

I tried it, higher temps and killed a cpu after a week..... better jet inpingement would have fixed the temps. and the cpu would still be dead... the cpu absorbed water...

Ouch. So I guess that probably nixes that idea, huh?

No, it should be possible.

The main issue is the thickness of the heatspreader: it's so thin that it'll require a fairly hefty pump to create a very good jet. Containing this "event" is a construction challenge on its own. Just make sure you plug the hole in the HS.

Correct me if I'm wrong...

What you guys are saying is that good heatsink compound between the WB and the processor actually beats direct thermal transfer.

(water<-->WB<-->Compound<-->CPU) Good
(water<-->CPU) Not so good

I swear I read someplace that a guy (I think he was Austrailian) tested thermal compounds and actually found that water by far was the best thermal conductor over AS3 or anything else.

Do you think the same would be true if the processor's heat spreader had a good fin arrangement like the inside of a WB?


As you were. Spode says,

"Direct Die rocks."

Click to expand...

No one was saying this:

What you guys are saying is that good heatsink compound between the WB and the processor actually beats direct thermal transfer.

Click to expand...

Dansdata has shown that water is a better heat conductor than thermal paste. thermal paste is designed partly for endurance and partly to simply be better than air at transfering heat... it is good at that. Water has no endurance in an application like that of thermal paste, it would evaporate or leak out.

Direct die has been done. It has been done with some success even (meaning better temps). The inherent problem of this cooling option was stated by JFETTIG but overlooked:

I tried it, higher temps and killed a cpu after a week..... better jet inpingement would have fixed the temps. and the cpu would still be dead... the cpu absorbed water...

Click to expand...

Direct die can be done and when done right it gets temps comparable to that of waterblocks. The concept is not as simple though as just letting the water touch the cpu and it works better... you still need turbulence, die area impingment and probably other things that I can't think of.

Direct die cooling is often used in phase change coolers also.

The cpu die is not completely a solid. It will absorb water over time. If you are using it in a phase change system the lifespan of your cpu would likely be even less because of the thermal expansion that takes place with large temperature differentials - this is extrapolation and conjecture as I have no data to show this, but it is what I would expect - moisture would be more likely to seep into the die.

So direct-die works, it has already been proven, but the lifespan of your chip is almost completely unpredictable.

Here is an interesting comment from one of the most technically accurate member of the forums:

BillA said:

research, key to everything

all organics are permeable to some extent, epoxies are well known in this regard due to their use as adhesives
common tests are weight gain, swelling, blistering, etc in addition to the mechanical properties tests
temperature accelerates all of these effects, which are all negative

is there a clue here ?

if the CPU package is not 'rated' by the mfgr for underwater service, why - or how - do you conclude that it is ?

now this has been discussed over and over and over
there are threads literally everywhere

try the relevant words plus leufkin or leufkins on the hard forum
he was certainly smarter than his customers, he made money and they lost their CPUs

geek, once sufficiently exposed it was just a question of the time to failure, one day less would not have saved it

be cool

Click to expand...

Click to expand...

Gotcha,

Good info guys, thank you all.

Dog_One said:

Gotcha,

Good info guys, thank you all.

Click to expand...

you got a headache too? LMAO...!!!

No, actually I feel fine.

I still think almost Direct Die would work nicely. I say almost, because in the case of say, a P4, you could stay away from the die and just use the processor's heat spreader. Certainly water would never harm that.

I also had another idea on how to control condensation when using coolant below the dew point. You could use coaxial tubing. You'd want fairly good thermal resistance on the inner jacket and good control of your flow rates as to keep the outside jacket from becoming too cold. That would work say what...?

Boy, I just have million dollar ideas coming out of my head all over the place today don't I. hehe Somebody better get rich reading this stuff.

You have a good idea about controlling condensation.

Of course, there are already numerous different ways to control condensation that work fine... normal tubing with thermal insulation tape is one. If you look on the main page in tips and reviews under peltiers I think there are something like 8 articles on controlling condensation.

The condensation issue is often the first thing to intimidate those inquiring about extreme cooling, however it is really not a big issue compared to the rest of the setup you have to perfect in order to get temps below the dew point. Insulation would be the easy part.