• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Direct contact water cooling?

C

cerebrex

[H]ard|Gawd
Joined
Sep 29, 2004
Messages
1,571
Has anyone ever given this a shot? Basically at the momemt your applying a copper waterblock to the copper IHS on the processor. Why not move that rubber O-ring seal from the top of the waterblock, to the bottom against the processor itself, and allow the water to directly contact the processor--completely removing thermal compounds, etc..?

Yeah, it's risky.. sounds scary... but it's no different than what we do currently.. except your waterblock is flipped upside down, and the acrylic cap is replaced with your processor... make sense?

Curious if anyone has ever tried that.
 
It would be far too difficult to keep the water pressure under control.

I mean, yeah, it would probably work a little better, (by eliminating the extra copper for the heat to travel throuth) but you'd risk spewing water all over the place if it gets out of control(poor contact or some sort of flaw in the mount with the IHS)
 
buffbiff21 said:
It would be far too difficult to keep the water pressure under control.

I mean, yeah, it would probably work a little better, (by eliminating the extra copper for the heat to travel throuth) but you'd risk spewing water all over the place if it gets out of control(poor contact or some sort of flaw in the mount with the IHS)
Click to expand...

pressure is a pretty constant thing, you apply a good amount of pressure from the backing plate, and it shouldn't be any different than a traditional system.. removing it would be the only down fall. I'm just curious to see JUST how much of a difference it would be.
 
Nice idea, but unless you perform a serious redesign of your processor and motherboard socket, I doubt this could ever work. I've been wrong on occasion though.
 
Justintoxicated said:
Wouldent it be easier to just remove the IHS?
Click to expand...

And soak your CPU's circuitry?

Sure. :)

EDIT wait I think I misunderstood what you're saying. You meant to remove the IHS and leave the copper on the block right? That would work.. but mounting without killing the CPU is still difficult. Then there's applying the tube once the block's...
 
Justintoxicated said:
Wouldent it be easier to just remove the IHS?
Click to expand...

well heres how current water cooling works on a conroe.
_____________
|_____________| - Acrylic Cap
|~~~~~~~~~~~ | - Waterblock
|_____________|
....................... - Thermal Compound
|''''''''''''''''''''''''''''''''''''''''''''''| - IHS
:::::::::: :::::::::: - Core(s)
=============== - Wafer


now, that's a lot of crap when all you need is:
______________
|~~~~~~~~~~~ | - Direct water cap - it seals directly to the IHS, removing the base.
|''''''''''''''''''''''''''''''''''''''''''''''| - IHS
:::::::::: :::::::::: - Core(s)
=============== - Wafer
 
cerebrex said:
Direct water cap - it seals directly to the IHS, removing the base.
Click to expand...

This is the part I am talking about.

It would be too difficult to keep the thing mounted on the IHS without slipping off or having a fountain inside your case.
 
buffbiff21 said:
This is the part I am talking about.

It would be too difficult to keep the thing mounted on the IHS without slipping off or having a fountain inside your case.
Click to expand...

I'm not scared, I can machine any part I need, I've lapped my cpu, I bet it wouldn't be a hazard at all, just have to make sure you apply completely even pressure across the entire thing. too much pressure and you could blow the seal. Probably should use 2 seals both made of thick silicone.
 
But what would you use to keep the water from spewing out from between the silicon and the IHS?

An O-ring? lawl
 
buffbiff21 said:
But what would you use to keep the water from spewing out from between the silicon and the IHS?

An O-ring? lawl
Click to expand...

exactly. If you grasp how they work, you'll realize your just being a little narrow minded, there's nothing to fear about doing that, its the exact same thing your current waterblock uses, just upside down.

the race is in the waterblock, the contact plane (being the IHS) requires no machine work, or any race - it will seal just fine. Even pressure is imperative - but there should not be any issues outside of that. My only concern is all the work might be a little overkill for the benefits.
 
How are you going to get the surface area to transfer heat effectively? Most modern water blocks have pins/fins to increase surface area and improve the heat transfer rate to take advantage of waters ability to store huge amounts of heat.
 
Gyrfalcon said:
How are you going to get the surface area to transfer heat effectively? Most modern water blocks have pins/fins to increase surface area and improve the heat transfer rate to take advantage of waters ability to store huge amounts of heat.
Click to expand...

THAT is a good question, and is the primary drawback. Will the water directly contacting the IHS overcome the lack of convection from a flat plane. You can maintain turbulence via the ceiling of the waterblock being machined with indentations, or fins. My belief is with enough surface area exposed, it will still see an increase. I'm talking directly from my ass here, but I just theorize even with that drawback, there will be a gain.
 
cerebrex said:
exactly. If you grasp how they work, you'll realize your just being a little narrow minded
Click to expand...
If you grasp English grammatical conventions, you'll realize you're spelling you're incorrectly. Sorry, couldn't resist.

the race is in the waterblock, the contact plane (being the IHS) requires no machine work, or any race - it will seal just fine. Even pressure is imperative - but there should not be any issues outside of that.
Click to expand...
Doesn't work that way. You don't realize that pressure inside a block is by no means "even." Look at the Storm. Look at the FuZion. Flow comes in one inlet, and leaves the outlet on a different side of the block. Even if your O-ring is mounted smug even, there may still be a chance it will slip under pressure. As such, it would take an immense amount of pressure to keep the O ring smug.

Good luck :)

My only concern is all the work might be a little overkill for the benefits
Click to expand...
Well, that's certainly an issue.. look at the temperatre difference when people just take off the IHS. Granted a block is comprised of more copper than an IHS, it still may only net you a 5C difference. (I cannot back this up of course - I've never tried direct flow to the IHS :p )
 
buffbiff21 said:
If you grasp English grammatical conventions, you'll realize you're spelling you're incorrectly.


Doesn't work that way. You don't realize that pressure inside a block is by no means "even." Look at the Storm. Look at the FuZion. Even if your O-ring is mounted smug even, there may still be a change it will slip under pressure.

Good luck :)
Click to expand...

great job on correcting my grammar, that's a last resort to piss someone off.

if you design the block correctly, there is no reason pressure cannot be evenly distributed, it's engineering - and I believe I have a fine grasp of it actually - so fear not... your skepticism is well received.
 
It might work. Maybe machine a grove into the IHS, 1 or 2mm thick, around the edges, and fit an o-ring. Piece of 1/2" thick polycarbonate, hollowed out completely, and another piece of polycarbonate to hold the barbs. That way, it would seal with the o-ring, while still direct contact ing.
 
Bbq said:
It might work. Maybe machine a grove into the IHS, 1 or 2mm thick, around the edges, and fit an o-ring. Piece of 1/2" thick polycarbonate, hollowed out completely, and another piece of polycarbonate to hold the barbs. That way, it would seal with the o-ring, while still direct contact ing.
Click to expand...

you actually increase the chances of a better seal without a race in the second surface. Leaks are much more probable if both surfaces have races.
 
cerebrex said:
Arc you come through again... thanks for the great information, I'll read up on it this evening.
Click to expand...

Any time! Just get a 1.5"x3" cube of plexiglass, hollow it out, tap it, and machine a groove into it for a thick Oring.

Apply the required amount of pressure (a back plate is REQUIRED for mounting this) then cover the outside of the Oring with a silicon seal.

The real problem though, is the flow of water. Picture this.

Water must be Jetted against the IHS, then it must exit the waterblock. If there's too much volume or empty space in the block, there will be a barrier of water between the jet and the IHS. The water has to jet against it, simple flow will not be enough... So...

You take the block of lucite, and machine away a ring of material leaving a column of plexi directly at the center of the removed ring... Sort of like removing a donut from a squareblock of dough... The middle column of material is then tapped, and serves as the jet, which accelerates the water, moving it directly against the IHS.

The "doughnut" that you removed serves as the collector, and it has 2 outlets (one on each side) which are then connected with a Wye fitting.

2 minutes in Sketchup...

Picture2.png


Would work for a single core design. The "greyish" translucent block at the very top is the IHS of the processor. The Toroid in the middle is bored out, and is the "jet" where the water enters.
 
Arcygenical said:
Any time! Just get a 1.5"x3" cube of plexiglass, hollow it out, tap it, and machine a groove into it for a thick Oring.

Apply the required amount of pressure (a back plate is REQUIRED for mounting this) then cover the outside of the Oring with a silicon seal.
Click to expand...

did anyone see any noticable performance increases? I ran through the articles 1 by 1, but didnt see any conclusive before/afters.
 
I thought that resistance was an important part of CPU waterblocks, but if you do direct water onto the IHS... where's the resistance? I don't know.... obviously I'm no expert, but it sounds funny to me...
 
less resistance = good. There are some modern blocks that use a restrictive design but the block is designed for this (lots of turbulance = pressure drop). The direct die coolers need massive amounts of flow to create the needed turbulance to keep heat transfer at a reasonable level, flat plane surface. If the top of the IHS could be groved out from the center to the sides you could probably improve the heat transfer assuming you have a good jet with low pressure drop.
 
Citizen86 said:
I thought that resistance was an important part of CPU waterblocks, but if you do direct water onto the IHS... where's the resistance? I don't know.... obviously I'm no expert, but it sounds funny to me...
Click to expand...

No, the resistance comes from the required "jet". In order for a direct die cooling setup to work properly, the velocity of the water has to be quite high. The resistance would come from a purposely designed accelerator, much as the TDX had a replaceable nozzle kit.

The direct die coolers need massive amounts of flow to create the needed turbulance to keep heat transfer at a reasonable level, flat plane surface.
Click to expand...

They need water of a high muzzle velocity more than high volumes of it.
 
Arcygenical said:
No, the resistance comes from the required "jet". In order for a direct die cooling setup to work properly, the velocity of the water has to be quite high. The resistance would come from a purposely designed accelerator, much as the TDX had a replaceable nozzle kit.



They need water of a high muzzle velocity more than high volumes of it.
Click to expand...

yeah basically you need highly directional velocity directly at the core.. makes sense... and it should be fun to design. Probably something better fit for a 4 axis CNC machine than my 3 axis machine. I don't see any reason to use plastic other than it's easier. Probably use a acrylic top just for a cool factor. The units they are using look to be overly large and cumbersome.
 
cerebrex said:
yeah basically you need highly directional velocity directly at the core.. makes sense... and it should be fun to design. Probably something better fit for a 4 axis CNC machine than my 3 axis machine. I don't see any reason to use plastic other than it's easier. Probably use a acrylic top just for a cool factor. The units they are using look to be overly large and cumbersome.
Click to expand...

Can you machine copper?
 
Arcygenical said:
Can you machine copper?
Click to expand...

yep! I modified a stock coolermaster waterblock a few months ago as an example:

wbmod_003.jpg


wbmod_006.jpg


wbmod_004.jpg


Copper Alloy 110 is pretty malleable metal... pretty soft and easy to work with. The excellent heat transfer means far less lubrication and expansion/contraction.
 
Somebody on here did a direct die cooler with the heat spreader removed and Flourinert, or something similiar, as the coolant. I can't find the thread, now...but I know it's around here somewhere...
 
As Arcygenical stated, it was all the rage for about a year, about 5 -10 years ago (the years they all fade together now.... ) Danger Den even sold a block for a little while.

Bottom line, lots of boards got soaked and in general it was abandoned as not worth the trouble, when it worked it worked OK, no huge leap however. Major issue was the ceramic substrate the exposed core was mounted on was not waterproof. With the IHS it might be worth another look. I would definitely find a dead board and dead CPU to play with first.

The Intel specification "Intel® Core™2 Duo Desktop Processor E6000Δ and E4000Δ Thermal and Mechanical Design Guidelines" would be very helpful with dimensions and clamping force the socket can withstand.

http://www.intel.com/design/core2duo/documentation.htm
 
It has been done before and posted on Procooling. Although it was a few years back.
 
Dan_D said:
Nice idea, but unless you perform a serious redesign of your processor and motherboard socket, I doubt this could ever work. I've been wrong on occasion though.
Click to expand...

Been there done that.

Back in the day I took a p3 700(when the p3s were the top end stuff) and used plexi and made a block with direct water to die cooling. Pretty much epoxied a few layers of plexi together and drilled out the center. We epoxied the block to the cpu. Ran great for a while. System was upgraded and switched back to air cooling(too lazy to upgrade cooling hardware).
 
cerebrex said:
So... what were the gains?
Click to expand...

I don't remember. Back then you didn't have on die temp sensors. No real good way of getting a good measurement. Was pretty quite though. Think the entire system had 2 fans.
 
Long time no post....

I wouldv'e thought it would work less well....

Remember, the copper waterblock has fins which increases surface area:

water
_/\/\/\/\_ <---- ridges
copper
-------------
IHS
-------------- <---- flat
Core

But if the water was directly on the IHS, there would be less surface area as the IHS is flat, so the water would remove less heat.
 
MikeHunt79 said:
Long time no post....

I wouldv'e thought it would work less well....

Remember, the copper waterblock has fins which increases surface area:

water
_/\/\/\/\_ <---- ridges
copper
-------------
IHS
-------------- <---- flat
Core

But if the water was directly on the IHS, there would be less surface area as the IHS is flat, so the water would remove less heat.
Click to expand...

That's what I was thinking too... but I guess if the water moves faster across the IHS surface it should get about the same effect.
 
Arcygenical said:
They need water of a high muzzle velocity more than high volumes of it.
Click to expand...

That's what I meant, high velocity sorry for not being clear.

Also if you round the edges of the jet you can keep pressure drop down to a minimum.

\ / <- rounded entry
| | <- straight exit
--------

This will help alot with pressure drop, any good fluids book will have the exact curvature to minimize the back pressure.

EDIT: Water flowing through a 1/2" orifice at 2 gal/min will give a Reynolds Number of 13159 which is turbulent flow.

The minimum flow rate through a 1/2" orifice is 0.304 gal/min (69.05 lph) this should be easily obtained in any moderate flow system.
 
Gyrfalcon said:
That's what I meant, high velocity sorry for not being clear.

Also if you round the edges of the jet you can keep pressure drop down to a minimum.

\ / <- rounded entry
| | <- straight exit
--------

This will help alot with pressure drop, any good fluids book will have the exact curvature to minimize the back pressure.

EDIT: Water flowing through a 1/2" orifice at 2 gal/min will give a Reynolds Number of 13159 which is turbulent flow.

The minimum flow rate through a 1/2" orifice is 0.304 gal/min (69.05 lph) this should be easily obtained in any moderate flow system.
Click to expand...

They are called hydrovelocity stacks, they work best if there is no adjacent interference within the the outermost radius equal to that of the inner circumference - if that makes sense to you - basically if the inner circumference of the tube is .5" than the velocity stack needs at minimum .5" from any interference - that is minimum optimal efficiency - water turbulence must be absolutely even across the entire flute - it's quite difficult to "tune" these without a micrometer and flow bench.
 
You must log in or register to reply here.
Back
Top