Overclock.net › Forums › Cooling › Cooling Experiments › Subambient full submersion phase change cooled pc
New Posts  All Forums:Forum Nav:

# Subambient full submersion phase change cooled pc - Page 29

about the HS, leave as is, with the fan

The effort of creating the ideal HS with controlled nucleation is way to big/complex to pay off.

Ideally you would have to "coat" each fin of the HS unevenly, providing more nucleation spots near the bottom while having a smoother surface near the top ... or vice-versa

The more nucleation spots, the more bubbles can form, but the more they "stick" to those points untill they become large enough to have sufficient "buoyance" the rip themselves loose.
Which is why each fin needs to be unevenly coated to facilitate this.

Imagine, for the sake of illustration, that each fin of the HS is 10 mm tall, and each mm differs one degree (bottom "hottest", top "coolest"), by controlling the nucleation points with variable "coating" you are actually optimising the HS for a certain temp "range". So, to work out the ideal coating, you need to find the ideal working range of the heat dissipation. Therefore you need to know how much heat you are going to produce... and when its all said and done you'll have to live with the possibility that the "tuned" dissipating range is smaller as the production range... unless you can produce an "oversized/overextended" dissipation range.
While it is true that the "coolest" point would be the nucleation point, the duration of the bubbleforming (till it breaks free) could actually be counterproductive.
In other words, in order to find the best HS nucleation range/setup, you'ld need to setup a lab with controlled variable conditions.... or do all the involved maths taking into account that a mathematical model does not always behave as it should in real life

Oh, and you'ld also have the shorten each 2nd fin to half (or 3/4 or 1/4) the height of the adjacent fin. To avoid bubbles being "trapped" for too long.

My suggestion is to use the HS with the squirrel fan to move (lotsa) liquid as the easier solution to start off with. Its range will be wider but less efficient.
Yes, it means that it will cool the CPU with liquid which is substantially "hotter" as the shell of the bubble at the nucleation phase-change point.... but how many degrees are we talking about?

PS: if going passive (without fan) you could also crosshatch/file the HS so the fins have "teeth" at the top... which will act as nucleation points

PPS/ i could be wrong about all this
Edited by RnRollie - 10/3/13 at 6:44am
Quote:
Originally Posted by Butter Chicken

make sure you use authentic Canadian Maple

wouldn't deliding the proc be safest method to attach the addition? but I guess after a delid resoldiering the IHS would be a little difficult...

as I have seen on delidded cpu's the chips is actually pretty small under there, so there is an air gap trapping heat in there unless the solder doesn't have a good seal and the fluid is allowed to enter the IHS.

my 2011 processor has this little hole in the top corner of the IHS and I've wondered if that it to allow the release of heat? other procs I seen without it.

Lol...well BC...as I'm Brit I'll remain patriotic and go with good old "Tate & Lyle" golden syrup....main reason being it doesn't recrystallize.

As regards deliding.....well I'm currently on a 2700k which I believe have a soldered on IHS....so don't really fancy the risk of deliding that.

Quote:
Originally Posted by RnRollie

about the HS, leave as is, with the fan

The effort of creating the ideal HS with controlled nucleation is way to big/complex to pay off.

Ideally you would have to "coat" each fin of the HS unevenly, providing more nucleation spots near the bottom while having a smoother surface near the top ... or vice-versa

The more nucleation spots, the more bubbles can form, but the more they "stick" to those points untill they become large enough to have sufficient "buoyance" the rip themselves loose.
Which is why each fin needs to be unevenly coated to facilitate this.

Imagine, for the sake of illustration, that each fin of the HS is 10 mm tall, and each mm differs one degree (bottom "hottest", top "coolest"), by controlling the nucleation points with variable "coating" you are actually optimising the HS for a certain temp "range". So, to work out the ideal coating, you need to find the ideal working range of the heat dissipation. Therefore you need to know how much heat you are going to produce... and when its all said and done you'll have to live with the possibility that the "tuned" dissipating range is smaller as the production range... unless you can produce an "oversized/overextended" dissipation range.
While it is true that the "coolest" point would be the nucleation point, the duration of the bubbleforming (till it breaks free) could actually be counterproductive.
In other words, in order to find the best HS nucleation range/setup, you'ld need to setup a lab with controlled variable conditions.... or do all the involved maths taking into account that a mathematical model does not always behave as it should in real life

Oh, and you'ld also have the shorten each 2nd fin to half (or 3/4 or 1/4) the height of the adjacent fin. To avoid bubbles being "trapped" for too long.

My suggestion is to use the HS with the squirrel fan to move (lotsa) liquid as the easier solution to start off with. Its range will be wider but less efficient.
Yes, it means that it will cool the CPU with liquid which is substantially "hotter" as the shell of the bubble at the nucleation phase-change point.... but how many degrees are we talking about?

PS: if going passive (without fan) you could also crosshatch/file the HS so the fins have "teeth" at the top... which will act as nucleation points

PPS/ i could be wrong about all this

Hi Rollie....yeah I've moved away from the idea of the coating.....I figure that if an uncoated IHS is sufficient to cool a server cpu then the area provided by the copper HS will be more than adequate.

I am however in the process of removing half of the fins...every other one....and while as you say its far too complex to make any calculations on this I just feel from subjective comparison that this will be far better than having the fins as tightly packed as they were in respect of trapping the bubbles.

Even with half the fins removed on rough calculation the HS has about 30X the area of the IHS.

I may try coating the exposed underside of the HS just as an experiment....if I notice far more bubbles coming from that area then it may be worth having a rethink on the coating....but for now I'm going to try just the partially de-finned HS passively without the fan.
Well I'm stoked just transferred my gas/liquid from the one pressure bottle to the other....using nothing but the pressure differential caused by temperature difference......nice little proof of concept that I'll be able to recover the fluid from the chamber basin back into the pressure bottle.

I just put the propane cylinder that I had previously emptied into the freezer...had it connected to a full butane cylinder at ambient temp, turned the butane cylinder upside down so that the liquid phase covers the valve opening, and open the valves. The higher pressure in the ambient cylinder forces the liquid phase from the upturned cylinder into the cold cylinder, the pressure then builds in the cold cylinder but because it is cold and at pressure the gas liquifies so restoring the pressure differential to allow more liquid to be pushed through.

The cylinders have the same volume so there was no worry of overfilling.....also I didn't want to transfer it all across just 18 liters of liquid butane.....so I weighed it as I was transferring it.....works out to 10.8kg with the density of butane liquid being 0.6g/cm^3

On a safety note I purged out the connecting pipe with butane before I connected the cylinders so it shouldn't have transferred any air into the cylinder.
I'm also going to purge off an amount of vapor from the refilled cylinder once it has returned to ambient....that will again purge off any air that may have somehow got in (it's less dense than butane so will sit at the top of the vapor space and so be purged off first when you open the valve)....just belt and braces really as I'm sure none did.
Edited by technogiant - 10/4/13 at 4:46am
Things progressing along nicely.....I've finished messing with the cpu heat sink.....I decided against trying a coating on it....firstly because I was concerned that if I couldn't get it thin enough it may actually hinder heat transfer.

Secondly because I've been thinking out where the cooling bottle necks are in this type passive boil off system......the limiting factor is that at heat densities above 6 to 10 watts/cm^2 the gas bubbles generated insulate the component.....this limit is present regardless of the efficiency of the coating.

So there's no point having a coating that can pass 10 gazillion watts/cm^2/deg C as you are still limited by the 6-10 watts/cm^2 boil off limit.

This is fine for Novec cooled servers as their heat density is sufficient to be cooled by the IHS alone so the special coating reduces the temp delta between the silicon and the liquid.

But I have to use a HS as the heat density of an overclocked cpu is greater than the boil off limit.....so having accepted that this extra layer is essential and that heat distribution in the copper block will be very efficient then the larger surface area of the HS not only reduces the heat density but also reduces the requirement for the surface to be so efficient as the special novec coating as you have more of it.....so although they state their coating produces a 15x heat transfer improvement, I have 30x the area through which to transfer heat (even though I've removed every other fin to increase the fin gap so bubbles don't get retained there)

A problem with using a HS in submersion systems though is TIM wash out, most TIM's are based on silicone grease which either dissolves in the submersion liquid or is just mechanically washed out.

I toyed with the idea of liquid metals but most contain Gallium which my damage my aluminium case.

So for the moment I've gone for a home made TIM that I've been testing......a mixture of syrup and alumina powder.

I've been testing it out for the last month and it has been giving better results by about 5 deg C compared to arctic silver 5.
Also daily blitzing with Intel burn test and prime 95 have not reduced its efficiency and it did not appear dried out when I disassembled yesterday.

I think the reason its better than arctic silver 5 is that the carrier (syrup) has a thermal conductivity 6X greater than silicone grease which is the carrier for most commercial TIM's

But the features that make it useful to me are that it will not chemically dissolve in the submersion liquid and at low temps it becomes super viscous almost solid so will not mechanically wash away easily.

It is of course a test and an unknown....pretty much like the whole of this project....but having thought it up I want to trial it....and I'll have to use it when I change my graphics cards anyway, my current gpu HS appears stuck on with thermal epoxy.....but when I upgrade obviously indigo extreme isn't an option for that....so it will have to be syrup/alumina...so this will be a good test while I'm waiting on 20nm gpu's to come out ( also determined to squeeze 1GHz or greater out of my gtx 460's...lol)

I may at some stage use Indigo extreme liquid metal as this doesn't contain Gallium.....I was concerned that I wouldn't be able to test it once I'd burned it in but found yesterday when I installed my HS that I can still run the pc even with just the HS being passive and partially de-fined....so it would be possible to check that the indigo extreme had actually "flowed" properly as I understand that it sometimes doesn't go to plan.

So today I will be putting my mobo in the chamber testing then sealing the lid on.

After that not much further to go, insulation of the chamber, assembling the purging system, sealing the cables externally and I'll be ready to go.
Edited by technogiant - 10/5/13 at 1:10am
So exiting to see things fall into place, cant wait to see some results, when are you uploading some pics ? ^^
Hi MB, yeah I can't wait to get this thing running now......today I put the mobo in tested it and sealed the chamber lid on....I've kind of given up on taking pictures....as with all my builds it's a mass and mess of cables and pipes......but I have put a web cam in the chamber which will hopefully give me some footage of the the boiling off and filling/emptying process in action which I'll be able to share.

Tomorrow I'll be overlaying the lid joints with 90 degree aluminium angle to ensure they're completely sealed and starting on the insulation.....I've still got to stick the secondary chamber over the cables where they exit the main chamber and join to the extenders,...I'm still undecided about what fluid to use in there to prevent air getting sucked in....I've got several ideas floating around....but each have pro's and con's....just can't decide atm but sure I'll decide come the time to do it.
Quote:
Originally Posted by RnRollie

have you seen this?
http://www.versarien.com/technology/

I asked them some questions on their facebook page and as advised have sent an email to the sales team. Hoping that there is something of worth, something new

Not really holding my breath for a game changer though.
Interesting.....not sure I can think of a particularly viable usage scenario though?.....unless I'm missing something...which is more than possible.
So the lid is now completely sealed on with the aluminium angle sealed over all the joints.....I've bought my insulation board and cut that to size.....I've got commitments for the rest of this week but hopefully by the latter part of next week I'll be testing.
New Posts  All Forums:Forum Nav:
Return Home
Back to Forum: Cooling Experiments
• Subambient full submersion phase change cooled pc
Overclock.net › Forums › Cooling › Cooling Experiments › Subambient full submersion phase change cooled pc