Project Gallium (Liquid Metal Cooled PC)

[Ryoutarou97]

11 hours ago, ElementalComputers said:

 

Project Gallium

Hi I am the photographer and now i guess the pr guy too of a very small modding team named elemental computers. Another member of the team you guys might know is @Maybach123 he is pretty active on this forum. Anyways this is our first public build log and since this is our first we wanted to do something that we have not seen anyone else do before. We are going to be cooling a pc with liquid metal in a full custom loop. Even though this is our first build log it is still a company pet project and may take a while as we are working with somewhat restricted funds right now. None of the parts a set in stone we will most likely be changing them but once we have this system up and running we plan to see how long a gallium cooled pc will last and if it is a feasible long term solution.

 

Parts as of right now.

CPU:              AMD FX 8320

Motherboard: Asus m5A78L-M

RAM:             Corsair vengeance 8 to 16gb whatever we have lying around

Storage:         Samsung 850 120gb SSD

GPU:              Nvidia 970

Case:             Parvum M1.0

PSU:              Evga SuperNOVA GS 550w

Water Cooling Parts

Ek Supremacy evo plexi

Ek 970 plexi

 

Will Update the rest of the parts tomorrow

Thanks for checking this build log out

A quick followup about the price of the gallium itself. It has a density of 5.91 g/cm^3 (1ml = 1cm^3). A quick amazon search says $20 for 100g (I'm sure there is a better supplier). That's about $1.20 per ml. Let's say the loop has only 500ml of fluid, this loop will still  be something like half the budget and that's assuming there is no res. Still, the idea is amazing and I would love to see someone try it. Indium is even worse at something like $1.5/gram and a density of 7.31g/cm^3. That comes out to $11.34 per ml. It looks like you will be using galinstan and that stuff is expensive. Density of 6.44 g/cm^3 and I don't know where you plan to get it so you may have a cheaper supplier, but the stuff is $2.5 -$3 per gram. That's $18 for every milliliter in the loop. WHAT. That's a $9000 loop (@ 500ml loop which is being generous). Please tell me you have a way to get this stuff cheaply. 

[Ryoutarou97]

Oh, and if that wasn't enough, here's from the wikipedia page: 

 

Galinstan may be used as thermal interface for computer hardware cooling solutions, though its cost and aggressivity (it corrodes many other metals such as aluminium by dissolving them) are major obstacles for widespread use. It is also electrically conductive, thus needs to be applied more carefully than regular insulating compounds.

 

So make extremely sure that you have no aluminum in that loop. I'm not sure how it does with copper and whatnot either. But seriously, message the manufacturers of every part and make sure there is NO aluminum in there. 

[ElementalComputers]

11 hours ago, BanishedSovereign said:

I'm wondering what pump will you use? Also, will you add some sort of corrosion inhibitor to the coolant?

we will eventually be moving to an electromagnetic pump but this is still in the testing phases

11 hours ago, Ryoutarou97 said:

Well, damn. Like, DAMN! I sort of slept through chem so will that work in waterblocks? Corrosion? I'm assuming it's pretty viscous so it will flow easily because you wouldn't want to use to many pumps. How does it do with thermals? Water is nice because it has such a high specific heat capacity, how does this stuff do? Also, cool a freaking extreme edition i7 on that thing!

corrosion is a big issue that's why we are talking with a company to put a thin layer of a silicone like material that will protect the waterblocks from corroding while still maintaining a high heat transfer.

 

10 hours ago, Stefan1024 said:

Very interesting. But isn't the gallium and indium super expensive?

yes that's why the hardware isn't amazing.

29 minutes ago, AlienALX said:

The only MATX board currently that will support the FX 8320 without throttling at stock is the GA-78LMT-USB3  by Gigabyte. Sorry about the huge text there I just copied it from their site. It you go here -

 

http://www.gigabyte.com/products/product-page.aspx?pid=4305#ov

 

You will note to the left of the CPU socket there is a heatsink. This cools the VRMs but note, there are only 4 phases on this board so whilst overclocking is possible with it you will need to use super low voltages. As soon as you use too much voltage the board will throttle to stop the VRMs cooking themselves.

 

As such I would strongly recommend also changing the CPU with a 8320E model as it requires less voltage per clock than the standard 8320.

noted thanks . We may move to a cheap Intel chip but were still not sure yet.

8 minutes ago, Ryoutarou97 said:

A quick followup about the price of the gallium itself. It has a density of 5.91 g/cm^3 (1ml = 1cm^3). A quick amazon search says $20 for 100g (I'm sure there is a better supplier). That's about $1.20 per ml. Let's say the loop has only 500ml of fluid, this loop will still  be something like half the budget and that's assuming there is no res. Still, the idea is amazing and I would love to see someone try it. Indium is even worse at something like $1.5/gram and a density of 7.31g/cm^3. That comes out to $11.34 per ml. It looks like you will be using galinstan and that stuff is expensive. Density of 6.44 g/cm^3 and I don't know where you plan to get it so you may have a cheaper supplier, but the stuff is $2.5 -$3 per gram. That's $18 for every milliliter in the loop. WHAT. That's a $9000 loop (@ 500ml loop which is being generous). Please tell me you have a way to get this stuff cheaply. 

yes we have a whole sale supplier we calculated the fluid would cost around $3000 when its done but the prices can change.

4 minutes ago, Ryoutarou97 said:

Oh, and if that wasn't enough, here's from the wikipedia page: 

 

Galinstan may be used as thermal interface for computer hardware cooling solutions, though its cost and aggressivity (it corrodes many other metals such as aluminium by dissolving them) are major obstacles for widespread use. It is also electrically conductive, thus needs to be applied more carefully than regular insulating compounds.

 

So make extremely sure that you have no aluminum in that loop. I'm not sure how it does with copper and whatnot either. But seriously, message the manufacturers of every part and make sure there is NO aluminum in there. 

as previously stated we will be coating everything is a silicone like material that transfers heat very well. one of the challenges we are trying to overcome is the radiator as the channels are to small for the material but we have some ideas .

[vanished]

1 hour ago, Ryoutarou97 said:

A quick followup about the price of the gallium itself. It has a density of 5.91 g/cm^3 (1ml = 1cm^3). A quick amazon search says $20 for 100g (I'm sure there is a better supplier). That's about $1.20 per ml. Let's say the loop has only 500ml of fluid, this loop will still  be something like half the budget and that's assuming there is no res. Still, the idea is amazing and I would love to see someone try it. Indium is even worse at something like $1.5/gram and a density of 7.31g/cm^3. That comes out to $11.34 per ml. It looks like you will be using galinstan and that stuff is expensive. Density of 6.44 g/cm^3 and I don't know where you plan to get it so you may have a cheaper supplier, but the stuff is $2.5 -$3 per gram. That's $18 for every milliliter in the loop. WHAT. That's a $9000 loop (@ 500ml loop which is being generous). Please tell me you have a way to get this stuff cheaply. 

$9000 of coolant

Even the $3000 quoted sounds insane, especially if you factor in that it might have to be replaced or topped up from time to time. 

[iamdarkyoshi]

12 hours ago, Stefan1024 said:

Very interesting. But isn't the gallium and indium super expensive?

So are a shitload of heatsinks :3

[byalexandr]

Few problems here. The gallium alloy you plan on using is not only ridiculously expensive and holds heat less efficiently than water, but it eats aluminum. Yes it's technically corrosion, but it literally destroys the structure inside the aluminum within hours. Even the full copper radiators have aluminum housings and fittings, so plan on making your own radiation system along with your own blocks unless you obtain blocks with absolutely no aluminum that contacts the alloy. Applying silicon like you mentioned before will significantly reduce the heat transfer efficiency, and by then it's much better to use water anyway.

[pagani123]

1 hour ago, byalexandr said:

Few problems here. The gallium alloy you plan on using is not only ridiculously expensive and holds heat less efficiently than water, but it eats aluminum. Yes it's technically corrosion, but it literally destroys the structure inside the aluminum within hours. Even the full copper radiators have aluminum housings and fittings, so plan on making your own radiation system along with your own blocks unless you obtain blocks with absolutely no aluminum that contacts the alloy. Applying silicon like you mentioned before will significantly reduce the heat transfer efficiency, and by then it's much better to use water anyway.

its not silicone its a material called parylene its a bit like silicone accept its properties for transferring heat. The main reason isnt heat transfer for a higher overclock but its mostly for looks and just to show that you can do this. 

[iamdarkyoshi]

Whether it works or not guys, it will be a very cool experiment. That is what LTT is all about, trying new things and seeing what happens.

[pagani123]

this has been a personal project of mine that i have been researching for a little over half a year now. when some of my friends were talking about making a computer modding group i decided to join and they liked the idea. anyways i have a strong chemistry background and i am fairly certain we can achieve this.

[TheRandomness]

1 minute ago, Maybach123 said:

this has been a personal project of mine that i have been researching for a little over half a year now. when some of my friends were talking about making a computer modding group i decided to join and they liked the idea. anyways i have a strong chemistry background and i am fairly certain we can achieve this.

What are CPU IHSs made of?

[iamdarkyoshi]

3 minutes ago, TheRandomness said:

What are CPU IHSs made of?

Not entirely sure, they take solder fine...

[TheRandomness]

Just now, iamdarkyoshi said:

Not entirely sure, they take solder fine...

I'm just curious

[iamdarkyoshi]

1 minute ago, TheRandomness said:

I'm just curious

http://www.tomshardware.com/forum/184699-28-cover-material

Quote

Its copper coated in nickel. I'm quite certain every IHS is like that.

That explains why lapped CPUs look like copper

 

[pagani123]

10 minutes ago, TheRandomness said:

What are CPU IHSs made of?

they are usually made of a copper base with steel on the top (where you put the thermal grease on) not positive but pretty sure 

[byalexandr]

1 hour ago, Maybach123 said:

its not silicone its a material called parylene its a bit like silicone accept its properties for transferring heat. The main reason isnt heat transfer for a higher overclock but its mostly for looks and just to show that you can do this. 

Well good luck, I hope it works out

[fire219]

This is very cool. I was going to mention gallium's tendency to turn aluminum into a pile of goo, but you already have a plan.

 

Followed.

[vanished]

3 hours ago, iamdarkyoshi said:

Whether it works or not guys, it will be a very cool experiment. That is what LTT is all about, trying new things and seeing what happens.

For sure!  I mention potential issues because I'm interested and want to see this attempted (with the best possible chance of success of course), not because I think it's too hard and they should give up

It looks like they have in fact thought of everything though and have a plan so that's good

 

Once again, can't wait to see how this turns out!

[VerticalDiscussions]

Techrax like failure incoming :p? Well, i guess you have some sort of way, to prevent your catastrophe XD?

 

 

 

[AlexAlcala]

2 hours ago, VerticalDiscussions said:

Techrax like failure incoming :p? Well, i guess you have some sort of way, to prevent your catastrophe XD?

 

 

 

 

My iphone 4 and I will always hate that guy ¬¬

[Tmt97]

Gotta say, this looked pretty interesting when I first saw it. My reasoning was that if you had gallium flowing through your radiators and waterblocks you would have really good heat transfer, since it would be metal on metal. But then as I read further and it was explained that the mix of gallium, indium, and tin your using is less thermally conductive than water, and that it won't even have direct contact with the metal due to the thin layer of parylene. I no longer see the purpose of this. Your temps will most likely be crap even by air cooling standards, and it will cost thousands of dollars just for the coolant. Part of me still thought it would at least look cool to see the flowing metal, but then I remembered that gallium doesn't really look much different when it flows. Even in liquid form it just looks like solid metal in the shape of whatever its poured into. You won't even be able to tell when its flowing and when its not. So in reality it won't look any different than a custom loop with metal tubing except for a layer of acrylic around the metal which will be the only way to tell it's gallium inside and not just a solid metal tube. So you will essentially be paying $3,000 plus the price of parylene and components, for a PC with high temperatures, terrible performance, and a custom loop that cant be told apart from this:

Not trying to be mean, or discourage you. Just being honest. I'll probably still check it out when it's done just out of curiosity... I seriously hope someone tries to do this for real sometime in the future, with a liquid metal that's more thermally conductive than water, and with parts carefully chosen to prevent corrosion. I'd imagine the temps would be amazing if your main limiting factor is the thermal conductivity of metal on metal.

[pagani123]

7 minutes ago, Tmt97 said:

-snip-

well to be honest one of the main driving forces behind this is to just be able to say we did it first  I could always use straight gallium as it last a long time as a liquid even in colder temperature (a few hours from  my testing) again we are still in the early stages of this build so i am not 100% sure what we are going to do.

[Tmt97]

53 minutes ago, Maybach123 said:

well to be honest one of the main driving forces behind this is to just be able to say we did it first  I could always use straight gallium as it last a long time as a liquid even in colder temperature (a few hours from  my testing) again we are still in the early stages of this build so i am not 100% sure what we are going to do.

Well who am I to discourage someone to do something just to say you did it. I showed up to the new star wars movie in a spock costume just to say I did it. Just a suggestion(I don't know specifically what problems would come from this), but what if you used mercury? I think it has a similar problem with aluminum that gallium does, but it stays liquid at least.

[byalexandr]

17 minutes ago, Tmt97 said:

Well who am I to discourage someone to do something just to say you did it. I showed up to the new star wars movie in a spock costume just to say I did it. Just a suggestion(I don't know specifically what problems would come from this), but what if you used mercury? I think it has a similar problem with aluminum that gallium does, but it stays liquid at least.

It's poisonous, that's the problem.

[terrytek]

seriously wonder how you'll tackle this without potentially destroying something.

[Tmt97]

11 minutes ago, byalexandr said:

It's poisonous, that's the problem.

Not to the PC it isn't. Besides, that's never stopped anyone in the food industry. All jokes aside though, your gonna be pumping it through closed pipes, not ingesting it. As long as they are careful, and wear gloves it should be fine.