Direct die water cooling

cdabc123

Supreme [H]ardness
Joined
Jun 21, 2016
Messages
4,475
so I recently decided it was a good idea to shoot a jet of water directly on the cpu die and here is a build/result post on it!


https://image.************/g5eQFa/P_20170228_200257.jpg
water-proofing capacitors I used plastic epoxy from home depo for the majority of this build.
https://image.************/kEuaFa/P_20170302_164717.jpg
water-proofing capacitors
https://image.************/kcrNva/P_20170301_160144.jpg
results of water-proofing capacitors im actually pretty happy with it
https://image.************/nr2HTv/P_20170228_222959.jpg
lapping heat spreader to make it look good :p
https://image.************/gWpLgF/P_20170228_223007.jpg
cutting some channels in heat spreader because their is only ~.5 mm between the die and spreader.
https://image.************/koRpaa/P_20170301_160205.jpg
lineup of all the components.
https://image.************/mNZxTv/P_20170301_182049.jpg
drilled holes in the spreader
https://image.************/kFft1F/P_20170302_164723.jpg
more epoxy work
https://image.************/i2ULgF/P_20170303_202213.jpg
results of epoxy work
https://image.************/bC0hva/P_20170303_160015.jpg
initial text just to see if it was dead
https://image.************/cz8P8v/P_20170305_134642.jpg
it is now in my custom loop build

results:
cpu is a l5520 i may give this a try soon with a i7 920 courtesy of bluesynk

idle
39-42c

load
48-52

i actually mad a mistak installing it as the imput that was designed to shoot the water on the die is being used as the output :(


Updated pic link (original post untouched)-https://imgur.io/gallery/voeu6
 
Last edited:
lol i managed to rip it off a few years ago and it never grew back right :p
apparently, you mad a mistak! Seriously though, that's pretty hardcore mod. Now go and clean out the dust, damnit, or I can't sleep tonight
 
I love the idea.

Hope it is reliable for you.
 
apparently, you mad a mistak! Seriously though, that's pretty hardcore mod. Now go and clean out the dust, damnit, or I can't sleep tonight
I dont make mistaks! except for the three cpu's i may have killed deliding thease dam 1366 chips :p and na its not dusty enough.
 
I'd worry about stress directly on the CPU and socket.
Movement of the CPU even slightly in operation can cause problems and the socket pins could suffer damage.

I delidded my CPU and tried loosening the water block very slightly while it was still running.
It crashed weirdly and corrupted my SSD !

Fair play doing this though :)
 
Now, that is [H]ard mod.
Next time you do that, try soldering/brazing for fittings, way more reliable than glue.
 
Now, that is [H]ard mod.
Next time you do that, try soldering/brazing for fittings, way more reliable than glue.

sounds like alot of work :p with that said i did manage to break off one of the fittings halfway through the build so its only a matter of time before i bump it and water goes everywhere.
 
Wow... looks good and kudos for the balls out mod. Not easy to make it all look neat.
 
sounds like alot of work :p with that said i did manage to break off one of the fittings halfway through the build so its only a matter of time before i bump it and water goes everywhere.

check that:

 
That's a pretty [H]ard mode, for sure! Saves a bit on the water block, too!

Will be interesting to see how long that glue resists the coolant.
 
That's a pretty [H]ard mode, for sure! Saves a bit on the water block, too!

Will be interesting to see how long that glue resists the coolant.
im not too woryed about the glue as the epoxy i used is pretty dam strong but i am slightly concerned for the stock glue around the edge of the die. i cant imagine intel really planned for somthing like this.
 
im not too woryed about the glue as the epoxy i used is pretty dam strong but i am slightly concerned for the stock glue around the edge of the die. i cant imagine intel really planned for somthing like this.
I did wonder why the capacitor epoxy wasnt allowed to flow up to the edges of the die...
It would have increased flow rate (less turbulence around the die edges) and prevented wearing the side of the Die support.

You could slope the epoxy up to the edges of the die and create a forced route over it with epoxy channeling the water. Wouldnt be easy though.
 
Last edited:
That is absolutely badass.

I would havehve thought that it would:

a) Short something, and
b) Not work very well due to the lack of fins like the water blocks you buy have

You proved me wrong on both of those though. Not only are the results good, but it's not dead yet.

As far as my experience with epoxy and metals go - however - i'd be scared shitless that would break off. I'd want to either weld it and leak test the weld, or maybe the IHS is thick enough that some sort of threads can be tapped into it? Would be badass if you could screw G1/4 fittings straight into the IHS :p
 
Last edited:
Holy Shirt you are one step away from harnessing your cpu to create fusion type power off the steam.
 
I did wonder why the capacitor epoxy wasnt allowed to flow up to the edges of the die...
It would have increased flow rate (less turbulence around the die edges) and prevented wearing the side of the Die support.

You could slope the epoxy up to the edges of the die and create a forced route over it with epoxy channeling the water. Wouldnt be easy though.
originally I had it designed where the epoxy came up to the edge on all sides but the one by the outlet this would have created a bit of a dead area right there that in my mind would benefit me. However I made a mistake installing that and it ended up being the input putting unnecessary ware on the chip and putting the dead water spot directly over the die :(
 
That is absolutely badass.

I wiuld ahve thought that it would:

a) Short something, and
b) Not work very well due to the lack of fins like the water blocks you buy have

You proved me wrong on both of those though. Not only are the results good, but it's not dead yet.

As far as my experience with epoxy and metals go - however - i'd be scared shitless that would break off. I'd want to either weld it and leak test the weld, or maybe the IHS is thick enough that some sort of threads can be tapped into it? Would be badass if you could screw G1/4 fittings straight into the IHS :p
I accidentally bumped it when I first made it and had to reglew it so I have no doubt I will eventually hit it and send tap water everywhere.
 
Next up, direct die cascade phase change!
 
i was almost considering using gallium or liquid mercury but both would be abit difficault.
I'm not sure about the thermal transfer properties of mercury. But with gallium, if you let it cool to room temperature, it will solidify. Gallium's melting point is around 29°C.
 
I'm not sure about the thermal transfer properties of mercury. But with gallium, if you let it cool to room temperature, it will solidify. Gallium's melting point is around 29°C.
yea i figured i may be able to mess with the pressure to keep it a liquid but i also found out it may react with everything including the die :(
[H]ard Mod of the month nominee.
ohh, do i get a cookie?!?!?!?!?
 
This is one of the best things I have ever seen.

You are a great hero.
 
That is nuts. I thought I was a bit crazy when I did phase change sub zero cooling, but actually having water touch the die, crazy. Awesome job.
 
That is nuts. I thought I was a bit crazy when I did phase change sub zero cooling, but actually having water touch the die, crazy. Awesome job.
and im highly considering using either gallium murcury or ferrofluid for the next revition instead of lame water :p
 
and im highly considering using either gallium murcury or ferrofluid for the next revition instead of lame water :p
It was pointed out earlier that Gallium must be a fair bit above room temp otherwise its solid.
This can also make it difficult to push increasing pressure in the system greatly.
I wouldnt use glued or simple joints anywhere.

Ferrofluid reacts like iron to magnetic fields.
You will need a pump that does not have a magnetic field near the pump element.

Mercury vapour is poisonous, very!

All of those are much denser than water and will need a stern pump to prevent burnout.
They are also very heavy, normal piping will not do.
If any of those liquids leak even slightly or you accidentally spill, its goodbye equipment and you may need a clean up crew.
Not worth the risk.

Soz to be a doom monger but at least take care of yourself.
 
Could save a lot of money on thermal paste too. How soon till oems release a 'water cooling ready' cpu?
 
Could save a lot of money on thermal paste too. How soon till oems release a 'water cooling ready' cpu?
it could but I still need to up the performance a bit before its practical in any case. And I don't know how well it would work with a high tdp CPU.
 
Could save a lot of money on thermal paste too. How soon till oems release a 'water cooling ready' cpu?

Highly doubt this'll be a thing. Too many potential issues thay may arise, i doubt Intel or AMD would want to deal with them.
 
Back
Top