Am I crazy? Silicone amorphous waterblock

[nicozeg]

Dr Cooling: la tapa y los lados son de policarbonato de 1mm.

Porfa no me revuelvas el gallinero y trata de usar ingles nomas en este foro.

Cap. foo foo: Yes, I remember that, and studied several cooling solutions, but just wanted to see if this works.

Maybe i'm too optimistic, but I think my bigger risk is heat buildup and lower overclocks. Water spill in the mobo? bahh, It will not be the first time.

I still think just cooling the MOSFETs would have been enough, the capacitors and other components dont make any heat themselves but they are absorbing it from the mosfets

That's not qite wright; inductors and caps do heat a lot in high demand situation, but as those are passive components, heat is not critical in their operation. However, as are connected to the mobo with thick copper wires, they contribute to the overall warming of the area.

Now for the update:

The cover is ready, on the afternoon I´ll reinstall the system and make some overclock tests. For the cap I used 1mm policarbonate sheet, the same material used to structure the sides of the box. Is easier to see it on the second pic, as it have some condensation cause I blowed though the hose.

[MMZ_TimeLord]

nicozeg,

Soooo... how are you going to plug in your +12 connector... a quarter of it is blocked by silicone...

I do like Capt Foo Foo's solutions... compact and effective. I don't quite know about the back block though... most PCBs are multi-layer and are VERY good heat insulators... so the benifit of the back block is suprising. I would have thought that front block would have done a better job. :shrug:

[iakovl]

WOW
does it cool well?

[nicozeg]

Originally Posted by MMZ_TimeLord

nicozeg,

Soooo... how are you going to plug in your +12 connector... a quarter of it is blocked by silicone...

Nonono, It's just flush with the border, but as silicone is the same color as the connector, the pic is confusing.

[|kbn|]

have you leak tested it?

[redleader]

Is this supposed to be science? I want to see the text book. Interesting interpretation of the problem.

Get your own textbook and you'll see that heat is conducted by electrons. Not surprisingly things that conduct electrons also tend to conduct electricity pretty well for some reason

Also, amorphous silicon is an insulator.

Also 2, I would not have bothered to cover the caps or inductors, they should not be getting hot unless something is wrong.

[|kbn|]

whats amorphous mean? I thought we were talking about normal sillicone sealant? maybe thats what its called....?

As I said before, the caps and inductors get very hot but they do not generate heat, all of the heat is caused by the mosfets near to them.

Another sollotion would be to remove the MOSFETs from the mobo/turn them the right way up so the metal part is touching the wb. Any one done this before on a mobo where the MOSFETs back is soldered to the board? I dont think it would be at all easy to de-solder them. I have a dead board here so I have spares. I could cut the legs of the mosfets on my board and use the spares so I wouldnt have the problems of removing them first.
This shows the same idea on a psu from bladerunners site.

[Incoherent]

Originally Posted by redleader

Get your own textbook and you'll see that heat is conducted by electrons. Not surprisingly things that conduct electrons also tend to conduct electricity pretty well for some reason

Also, amorphous silicon is an insulator.

The implication being that any conductor of heat will also conduct electricity?
Not the case as I am sure you are aware.
Heat transfer in non-metallic solids can get pretty quantum mechanical, but suffice to say that movement of electrons is not the only way to shift heat. Although it is the most efficient way to do it.

Cheers

Incoherent

[bikr]

This is brilliant , I can't wait to see results..

[Grizzy]

You said you can't make a waterblock for them.... Why not just take that huge honking silicon thing, make a mold out of it, and pour molten copper in to form the block in the exact shape you want it?

[|kbn|]

Copper cannot really be cast sucsessfully on a small scale/without proer equipment.
I expect the MOSFETs used dont cost much, assuming they cant be removed you could just replacethem? I think a block like that would be a lot better

[Moparchris]

i cant be bothered taking the side of the case off to look atm but couldnt you un-solder the flat metal piece of the mosfets thats on the m/b and cool that surface with heatsinks or a w/block ?

also someone mentioned casting - its not really on topic but.. im gonna try sand casting my gpu block soon im just making the mold now, only thing im worried about is if i can melt the copper or not and if i can i dunno if i can get it hot enough becasue of its good heat dissipation its hard to pour it because it will solidify as you pour it if its not hot enough but oh well.. ill give it a go

[|kbn|]

Dont even bother casting copper. it melts at around 1083c iirc, so you need a flame temp of about 1300c. The container you melt it in has to withstand the temp and insulate it well or the heat wont stay in the copper.
the main problem with casting copper is it oxidises VERY whichly when molten, which means if its melted and poured around oxygen, you WILL get bubbbles in the final thing once its cooled down, which can lead to leaks and other problems. Unless you can remove the oxygen there is no point in trying with copper. Also iirc some one was killed when trying to cast copper as the mould was still wet and it exploded. If you attempt it make sure you think of how to do it safely.
Aluminuim is probably a lot easyer to cast, melts at 660c and afaik has no problems with oxygen (this wasnt mentioned when I was reseaching casting).

Yesterday I tryed to remove a MOSFET from a dead nf7-s but I couldnt. The soldering iron I had was 30w you probably need about 60w to do it. I expect it could damage the mobo or even the mosfet to remove it. If you can get spare MOSFETs from the maker of them then you could simply cut the original mosfet leads to get round this problem.

[Cptn. Foo Foo]

Originally Posted by MMZ_TimeLord

I don't quite know about the back block though... most PCBs are multi-layer and are VERY good heat insulators... so the benifit of the back block is suprising. I would have thought that front block would have done a better job. :shrug:

I was told by many people in the old AMDZone forums (which are now dead and burried ) and in emails from people who read my Overclockers article on the top-side block that the top of MOSFETs have various layers of materials(including an air layer) which end up acting as insulation. So I guess the "level of insulation" is less for the motherboard than the top layers of the MOSFETs.

I guess there's another thing to consider too which is the surface area I am cooling with the back-side block. Compare that to the combined area of each of the little MOSFETs and maybe that has something to do with it too. The back-side block is cooling a much bigger surface area.

[nicozeg]

I'm sorry for the lack of updates, but was so busy finishing the systen that didn't have time to post. :shrug:

The hurry was that it needed to be ready for an overclocking competition that happened yesterday. It was sponsored by corsair and the first prize was an athlon 64 CPU, Mobo and memory kit

Sooo, for the non believers : It worked very well, nothing died, and I managed to get the third place!

This block was a success considering that most bets were against it, but it managed to sustain the same level of overclock that previously was achievable with a powerfull fan directed to the mosfets. I had hopes of reaching a higher oc level, but considering that this was a quick and raw experiment, results were outstanding!

Now to the pics:

1- Leak testing, it was tested for a full night and everything looked ok in the morning.

2- First run! mobo is alive and the system is ok! it included a volt moded gf 440 mx

3- another view of the full system mess, including custom splitters made in about 1 hour!

4- My pc proudly shows his hard earned prize.

Some specs: Cpu Athlon xp 1700+ jiuhb dlt3c 0803 @ 2557 mhz

Gf4 mx 440; stock 275/525; oc 390/780

The competition was a series of benchmarks, and what counted was stock vs. oc speed jump, not final absolute score.

The place was incredibly hot due to more than 20 power hungry machines, ambient was at about 35°C!

[TerraMex]

Looks pretty good.
As long as you dont change your nick from Nicozeg to NicoZap.

[FizzledFiend]

simply marvelous baby! I was intreguied the whole time! now @ the begining you said you used oil to coat everything before you applied the silicone? just wondering how you got the silicon to stick...everytime I try that I have to glob it on so thick that gravity will take over? I have thought about a silica spray that dries to seal off electronice (comformal spray i think some call it)

anywho tis is simply amasing...it reminds me of the Borg the texture of the silicone gives is a biological look to it...add some color and do more elsewhere...this is great!

[G33k]

Nicely done - we never said it was impossible, just that there were a lot of reasons why it might fail. You did it though, so props to you - shame it didn't get you the higher OC you were looking for though

[MMZ_TimeLord]

nicozeg,

Good job... congrats on 3rd place. Now you have me thinking about something similar... /me runs off to research thremal properties of various adheasives/sealants.

[Tempus]

I'm still waiting to see the FSB OC results.

What are the before / after numbers?

[nicozeg]

Originally Posted by FizzledFiend

simply marvelous baby! I was intreguied the whole time! now @ the begining you said you used oil to coat everything before you applied the silicone? just wondering how you got the silicon to stick...everytime I try that I have to glob it on so thick that gravity will take over? I have thought about a silica spray that dries to seal off electronice (comformal spray i think some call it)

anywho tis is simply amasing...it reminds me of the Borg the texture of the silicone gives is a biological look to it...add some color and do more elsewhere...this is great!

Getting the silicone to stick in the first layer was difficult, it mostly was mechanical bond to all the surface shapes. the second layer was a lot easier cause it stick to the base one. It required a third layer to ensure that everithing was properly sealed.

I advice to anyone that wants to try this, that currently is a lot more effective a properly made waterblock, I think that a real success will require a very thin seal coat, something very difficult to do with silicone.

That thing of the conformal spray sounds intresting! Maybe it could be epoxy paint like the one for swimming pools.

About clocks: I didnt try high fsb's, as the chipset is still aircooled, and given the heat in the place, stability was very difficult. My final setup was at 204 *12.5 allowing for very low memory timings.

[MMZ_TimeLord]

nicozeg,

Here is a link to the Thomas Registry search for "Thermal Conductive Materials"

This document has a line of thermally and electrically conductive adheasives/paints.

Looks like the 805 one part system might be your best bet... Dielectric strength of 50 volts/mill and thermal conductivity of 12.5 Btu-in/hr-ft²-F°.

There are probably better thermal conductors with a higher dielectric strength... but this was just a quick search.

The idea I had was that you could spray/coat the FET area with something like this... then build up your walls around the coated area. Leaving a single material for the heat to go through as silicone is generally a thermal insulator.

Someone correct me if I'm wrong... ( It HAS happened before... )

[nicozeg]

Interesting. according to the maya thermal wizzard, silicone and epoxy have a similar conductivity of about 0.2w/mc, and a highly filled epoxy like the ones you found can be at 2 w/mc.
It sounds a lot better. If the seal layer can be made at a thickness of about 0.5 mm with one of those paints, things should be a lot better.

[BadPenguin]

Why not try this? (seems far fetched, but hey... it "could" work)

why dont you waterproof EVERYTHING eletronic in your case (cables, mb, hard drives, seal drive bays... etc EVERYTHING) and just fill your case up with water, and cycle the case as a whole through a bigger radiator... it could work if you think hard enough about it.... my 2 cents

[Butcher]

Been done with oil (though you can;t submerge HDDs), water would be somewhat more risky as one little leak and zap.