• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Direct Die Cooling

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

cheez

Registered
Joined
Jun 21, 2002
Location
West Lafayette IN
I'm working on setting some of this up to cool my CPU, GPU, and northbridge. I was wondering a few things about it.

1. I don't need special water or anything to keep it from shortening my components?

2. Obviously cutting out the holes for the die isn't perfect and there a bit too big so how do I go about sealing up the errors? Polyurethane?

3. What do you recommend for insulation to keep condenstation down.

Thanks
 
1. Hard to say for a lot of reasons. Depends on exactly what you're planning to do. Suffice it to say that if you intend to use an electrically inert fuild be prepared to spend several hundred dollars on it alone.

2. Well... again depends on your design. You could use some sort of silicone based sealant around the block to make sure water stays where it's supposed to.

3. Why would there be condensation, unless I'm totally not understanding what you're trying to do.
 
If you dont touch PCB, you dont need to worry about electrical stuff. However, AMD processor bridges would have to be sealed, so you'd have to use some conformal coating or nail polish :)
 
well... i dont suggest you try to direct die cool anything but the cpu,
you wont get condensation unless you have chilled water
and i dont suggest direct die cooling for a first timer, direct die cooling is dangerous..
 
I have tried Direct Die cooling, and it worked, kinda. I would only recomend DD cooling on a dedicated system. I ran one with a GPU and NB block, and they cut the flowrate too much and my temps were 45-48c, when my maze3 gave temps of 42-43c. because there isnt very much surface are being cooled, you need lots of water flowrate.

DO NOT use arcylic adhesive to hold the D2D block to the cpu. it does not stick well to the cpu.

Also, rumour has it that cpus will absorb water, so be cautious.

I think you might be betteroff with Peltiers as an extreme type of cooling. just my 2cents
 
yeah BUT....
copper is just an extra layer to the water....
read his article and find out how well it did
 
When I made direct die cooling for my Fiancee. I used Marine Goop to seal the block to the CPu and then used GE Silicone II all around the socket and half way up the block. Been running it that way for about 7 months with no problem.
 
I LAPPED my core and it still didn't take up any water. So as far as safety I think it's fine.

Also, if you design the block right, you can make it so when the pump stops, there is still some water to protect the core. Becuase it happened to me and it shot up to like 60c the plastic all heated up. As soon as I turned the pump on, the cold water caused the plastic to cool down at a different rate to something else in the block and it pulled apart and spewed! Basically the epoxy lost it's grip :S My fault I expect.

From my experience it is better than a waterblock.

And its cheap!

But anyone who thinks you can leave ANY watercooling system in a friends computer is wrong. Watercooling requires maintenance - if you don't understand it , youshouldn't have it.
 
I direct die cooled my cpu - it was a 800Mhz duron, and i kept it at almost ambient, this continued well (appart from the bridges disconecting) till i decided to change the water in the computer - after that my processor never worked again, i think the cpu because porus when it was exposed to air.

i used epoxy to stick the plug (that right - i said plug) to the cpu, when i came to take it off i was surprised how soft the epoxy was, it had also taken in some waterwetter and was a pinkish colour.

the morral of this story?

direct die cooling works incredably well, if it is done right - take a look at the early direct die articles on spodes site to see how not to do it, and it is very cheep to make (mine was a plug, spodes was a sunny D bottle ect.) however there is a large risk of something going wrong - and i dont think there has yet been a sutable glue found
 
Back