Overclocking Crazy submersion phase-change system

The CPU temperatures would be FAR better with a larger cooler such as a D14/Silver Arrow.

The extra surface area would really help.

 

Nah.....they don't work at sub zero....the heat pipes use distilled water to distribute the heat...trouble is the water freezes and the heat pipes just don't work......but good thought though.

The surface area I have is ample to ensure the surface heat density is not too great for this form of cooling (max 6-10watts/cm^2) so provided the heat transfer is efficient enough from the surface to the liquid then it should be fine, I've sufficient surface area to cool many hundred of watts and still be within that heat density .....I may need to try and copy that specialized coating...but in all probability it is the TIM.

 

Maybe try Cool Laboratory Liquid Pro/Ultra

 

They both contain gallium....gallium eats aluminium which the chamber is made of....I know it shouldn't come into contact with it in the normal run of things but just can't risk that.....I'll probably try Indigo Extreme....that doesn't contain Gallium.....I didn't use it first off as the the application process is a little bit touch and go....I wanted to use something that would just work for the first trials....it might be making my temps a bit higher but it's there and working for now....but I'll replace that at some point.

 

Ahh okay. Shows my ignorance.

I look forward to seeing more from this project.

 

Have you got a pump pushing water over the components? That would help if the gas is acting as an insulator. Given the relatively high temperatures that seems likely.

Your youtube video of the outside has been remove?

 

Yeah I'm embarrassed about the looks of the build...real mess and doesn't do the thought process and work I've gone through to build this justice.

There shouldn't need to be any liquid flow, the heat density at the heat sink surface is well below the maximal limit for this type of passive boil off cooling, granted if it were too high then flow would help to drive off the bubbles...but it really shouldn't need it.

 

Please keep us posted with any changes and take pics of work in progress. Function first then beauty.

 

I for one would like to see the mess

 

I'd love to see how you've executed this!

 

My grasp on the physics involved is not good enough to add anything meaningful, so instead have some rep

 

Okay for the purpose of explanation I've uploaded a diagram so I can explain how this work...you will get more from it than from a picture of my chaotic build.



Uploaded with ImageShack.us

So in the center you have the chamber, is hermetically sealed so no vapor or air enters or leaves.
It is made of 1.5cm aluminium plate and cut so all the plates interlock and can't be pushed together as this thing has to withstand compressive forces of up to -10psi.

At the bottom sits the PC...just the mobo all other components have to live outside...so as you can imagine running all the cables to it from the outside and still keeping the chamber gas tight is a task....you also have to seal the inside of the cables as air will be drawn through them because of the low pressure.
The mobo actually sits in a pressed steel seamless basin so there is no chance of leakage.

Top left in the chamber is the cold radiator of the air con unit (evaporator to use the correct term evap for short) its a 3.6KW cooling unit...that is shrouded in place and has 3 gentle typhoon 5400rpm fans to circulate vapor through the evap fins.

The fans are the only moving parts of the cooling system..apart from the air con unit itself...the rest is operated by temperature, pressure boil off and condensation.

Also not shown the chamber has a purging system so I can purge it with vapor to ensure as close as possible to 100% refrigerant vapor in the chamber vapor space.

Also entering the chamber is a flexible stainless steel hose, it is actually placed so the end comes from above and opens in the bottom of the basin, this transfers the liquid refrigerant back and fore the chamber and bottle and is made of stainless steel so there is little risk of it cracking at low temp and leaking out liquid refrigerant.

To the right is the chest freezer and pressure bottle, the sole purpose of these is to allow safe containment of the liquid refrigerant in the pressure bottle between uses.
At standard pressures the refrigerant has a boiling point of -1c so cannot be contained in the chamber...it is only designed to withstand compressive force not the large expansive forces that would be exerted by the gas at normal ambient temps.
So as stated the Stainless steel hose freezer and pressure bottle are simply concerned with moving the liquid back and fore by a process called cold line transfer, Which I'll explain shortly.

To the left is the ac unit, its a stipped down window type ac unit, it has a 3.6kw cooling power and can get temps as low as -33c on a cold day.


So how does all this work.....to start with we just have the chamber filled with refrigerant gas at ambient temps...so when you crank up the ac unit and fans the vapor space cools, the vapor starts to condense to liquid causing a partial vacuum in there....this is used to suck the liquid from the pressure bottle.....I actually pre chill the liquid in the pressure bottle a little first so that it doesn't boil off too violently as it enters the low pressure chamber....but apart from the pre chilling the prssure bottle is at this time out side the freezer.

So as the chamber temperature and hence pressure drops the bottle temp and pressure is rising as it warms towards ambient building pressure to push the bottle contents into the chamber basin.
I actually have the bottle upside down at this stage so the liquid phase is transferred rather than just sucking gas across which would take an age.

I have lighting and a web camera inside the chamber so I can see when the liquid level is sufficient at which point the bottle is placed back in the freezer to chill ready to recover the liquid.

So that's the fill procedure and I now have my mobo submerged in an 18 liter pool of liquified refrigerant.

Now the boiling point of the liquid at normal pressure is -1c...but the chamber is not at normal pressure so the liquids boiling point varies dependent on this pressure. In fact the as the pressure in the chamber is being determined by the low temperature produced by the evap and that in itself is being determined by the heat load in the chamber.

So in effect the liquid is held in a state of equilibrium between the vapor and liquid state....any heat input causes boil off absorbing that heat which increases pressure which causes condensation on the cold surface of the evap which removes the heat.

This entire design is really just to provide a very simple and efficient means of linking the cooling power of the ac unit to the pc components and it does it by a simple liquid/vapor phase change and re-condensation circulation with no moving parts save the fans.

It is one of the most efficient means of doing this, each 1ml of liquid as it transitions from liquid to gas removes 220 joules of energy so 1ml/sec of evaporation equates to 220watts of cooling and of course the reverse happens at the evap where it offloads the heat.....compare that to water cooling...1ml of water removes 4.2 joules of energy for a 1 deg c temp increase.

There are of course caveats with this form of cooling as with anything....firstly commercial tims wash out as they are based on silicone grease which dissolve in the liquid.

Secondly there is a maximum heat density this type of passive boil off can cope with...beyond that the bubbles generated start to insulate the component....the limit is 6 to 10 watts/cm^2

Also you should have a coating on the heatsink or whatever your using as the liquid component interface that encourages nucleation of the liquid to gas...that apparently increases the heat transfer by 15x (Thats something I don't have at this time)...the coating should consist of sub 10 micro particles...I'm working on a way that I can do this.

One thing that is difficult to take on board is that this really has nothing to do with liquid cooling....no flow is required...the cooling is solely due to boil off of the local liquid which is self refreshing as the bubbles rise through the liquid....the only occasion that flow is required is if you exceed the max heat density when you would need flow to drive away the bubbles.

Another misconception is that a large surface area is required....it's not, a 9cmx9cm cold plate even using the lower max figure of 6watts/cm^2 could dissipate 500watts...and that's just the top side of it the underside is also in-contact with the fluid.

So that's how it all works...at the end of the session to recover the liquid just open the valve on the pressure bottle in the freezer and turn off the chamber ac unit....the chamber temp gradually rises increasing the pressure and the pressure in the bottle in the freezer is low....so the fluid gets sucked across for safe containment.

I'm going to brag now....I've no HVAC background and I thought all this out myself I believe it to be a first....and....IT WORKS

 

I'd definitely like to see more pictures now you've explained how it all works, very intriguing.

 

Thank you for the break down I have a decent Imagination and with the video you did I definitely have the gist of it.

Web Cam that's how you did the video inside I was wondering.