Water cooled heatsinks?

[Cupe|ix]

The watersink!


This might sound dumb, but...

What if you used a good all coper heatsink, and somehow seal off the top (the fins) into a plexi block or something (some kind of closed container), and then run water through it?

It seems that direct die cooling is not too safe, and water cools much better than air does. This might be a little hard to do, but I think it's a good idea. You might say "just use a waterblock" but I hear turbulence is a good thing. This should give lots of it, and more surface area to use for cooling should be better, right?

Here, would look something like this. Good idea or not?

 

[Monaco]

This idea works- WELL! I built one by basically enclosing a Pentium1 heatsink in plexi walls, bonded with JB weld. And it totally kicks butt.

Cost me nothing but time to build, and got my Celeron 566@995 to run in the high teens celsius under full load.

 

[-=UR=- Ranger]

This should work fine, and has been done several times before!
Give it a try

 

[IFMU]

One thing that alot have been wondering about is if this would be possible to do but with a few minor differences. The baby blue is air, the solid light gray is a fan. I among others have wondered how well this would perform. Granted, theres nearly no way for it to stand up to straight water cooling. However, how much protection could it give you? Would it keep the temp decent? How decent? This is something I think I will try out in time. But Im not quite ready here to give it a shot yet.

 

[Monaco]

IFMU, I've thought about that too- get a copper pipe and cut it just so , and slide it over the fins of a heatsink down to the base. And then, put the fan back on.

It would probably perform decently. But it would make up for it with coolness I imagine that if the water cooling part works well, it won't matter whether the fan is on or off. But good security!

 

[IFMU]

Basiclly the main reason most are interested in that, as well as myself, is if the pumps quit working would it remain at least decently cool enough to prevent the cpu from overheating to the extreme. ie damaging it. I am interested to see how well it could keep it cool and for how long. Theres alot of possiblities in my mind for it. But no chance for actually testing quite yet.

 

[3.49GHz]

watercooling

I think it's a good idea, and have been thinking about it. I want to try to overclock my AMD Athlon XP 1800+ to 3.49GHz (233 x 15)
and this is the only way to go. I've been thinking of doing this(for months now): pumping icewater through my heatsink like shown in the picture, and cooling the inside of my case to 10C.

 

[Cupe|ix]

Whoa, 3.49GHz LOL I wish!

Anyway, yeah, I'm thinking I'm going to try this, since it shouldn't be as expensive as building a traditional water-cooler system since you don't have to spend a lot on a good waterblock. My new T-bred system will be based on this, provided the actual modification to the heatsink is not too difficult to do.

Are there any links out there for good, easy designs? Would it be better to have it setup something like mine (with a really high flowrate of course) or to have the H20 in waterfall down onto the heatsink then out? What would be good materials to use?

Oh, and with this setup, I'm assuming that the heatsink will still stay fairly warm. So it would be okay to even chill the water below ambient temp and avoid condensation, right?

This is looking pretty promising, folks!

 

[Cupe|ix]

Does anyone else have any more information on this? I'm really anxious to see some results if anyone has any.

Keep the thread alive, this is an interesting idea, I think!

 

[res0r9lm]

this what I have in mind sort of a cross between a heatsink and direct die cooling. I was told it wouldn't work as good as direct die b/c water is a better conductor. water= .6 w/mk copper=401 w/mk

 

[Cupe|ix]

I didn't get that at first.

So your idea is to cool the die with water, but also leave on a heatsink and cool that with water also? That seems pretty cool, would the water interfere with the thermal compound between the sink and the die, or would you not even need any?

That seems like a good idea, but I think I'll stay away from direct die, until there is some solid evidence either way (on water eroding the die over time).

 

[res0r9lm]

no it wouldn't b/c you would have to epoxy it on. I'm going to try it with a duron 800 first b/c it will be perment.

 

[RoadWarrior]

I've got one nearly finished (been moving, now living in a temp rental looking for a place to buy, no shop space, time, etc etc) that uses a sink, see..
http://forum.oc-forums.com/vb/showthread.php?s=&threadid=49516

Gotta get back to that soon.

 

[The Overclocker]

there have been a few articles on the same idea here on overclockers.com some of them are listed:

http://www.overclockers.com/tips92/

http://www.overclockers.com/articles148/

http://www.overclockers.com/tips111/

http://www.overclockers.com/articles171/

 

[JFettig]

aboutthat whole 3.49ghz thing, its basicly impossible... maby if you liquid nitrogen cool it.... then itll be dead before you get that high........... basicly its impossible to get the fsb up that high so far, maby on a kt400 board, but getting the multi up WITH the fsb is virtually impossible....

 

[3.49GHz]

watercooling

Actually, ur not far off. I was going to use freon cooling, after i projected the possible loaded temps, (1000+ C)


I will use a small copper pipe wound around in a heatsink and arctic silver to fill the gaps, the pump freon through it.

We have an engineering lab at my college suited for things like this.

i need to unlock my chip first, to drop the multiplier to 6, and up the FSB on my PC3300 DDR to 233 (466DDR)

 

[JFettig]

well,... good luck

and welcome to the forums! and make sure you give us results!! pics uv everything eaven in the process.... that is if you can.....

 

[Cupe|ix]

I wish I could use Freon like that, that would be some extreme cooling power!

Pretty sure I'll stick with water the first time though


Actually though, this project will be on hold though, until June (when I get lots of hours at work ) when I actually build the system first.

Or I could build it soon, and shake it down, get it all working nice, and be ready to hitch it right up when I assemble the new comp. Yes, that sounds like a better idea. I guess I'll start on that in the free time this month.

It might be easier to have a machine shop do this. I'm thinking about an all copper heatsink, and sealing up the top with copper plating now. They would be able to nicely punch tubing holes for me