I've been researching for about a week now on how to make the most of a machine submerged in mineral oil.
The basic idea:
Q9550
EVGA 750i FTW
8 GB of 1066mhz dual channel memory- most likely G.Skill.
2x EVGA GTX 260- the revised edition with 216 processor cores and completely stock clocked. There's no need for factory overclocking as I'll be overclocking them myself.
A 750w power supply- the model is irrelevant as the system will be submerged in oil.
I will be overclocking the $%&^ out of it. I'd like to try to push 5 ghz. =D Bet that'd be nice to see on a Q9550. I've seen 4.7, so it might be possible. My real goal is just as high as I can get though; so if it's not 5, I'll still be able to sleep.
Most of the system will be submerged in mineral oil while optical and hard disks will remain dry.
The oil will be pumped through a 480mm radiator (I'm yet undecided on the volume of air I want to move over the radiator). No reservoir is needed.
Right now I'm toying with the idea of TECs sandwiched between a cold plate and a water block and fixed to anything hot. My main concern is the efficiency of TECs, they seem to not play well with anything that runs hot (ironic shortcoming for a method of cooling, I know). What I mean by this is that cooling efficiency decreases exponentially as temperature increases, that's a curve starting high and ending low on a graph. If we're not overclocking, TECs would be a fantastic option. However, if we're overclocking and increasing voltage, TECs soon lose their appeal and are outperformed by normal water cooling methods.
Now- this is all fine and dandy, but my system would be submerged in mineral oil. I'm trying to figure out how this would affect TECs. I'm looking for intelligent conjecture, with a rational explanation behind whatever idea is presented.
I'm not trying to attain sub zero temperatures. I don't see the point. I'm overclocking...
Food for though:
Condensation is not an issue when running TECs in a machine submerged in oil. Since there is no air (or extraneous amounts of) around any of the components being cooled, there is no water to condense.
In a normal submerged system there is pretty much just one system wide temperature, thanks to how wonderfully the oil distributes the heat. How do you think introducing water blocks and TECs will effect this trend?
The more TECs added, the more power required. I'll most likely require two additional power supplies for 4 TECs (north bridge, CPU, and GPUs). With three power supplies in the oil, would the temperature rise by a negligible amount?
I'd like to run the north bridge and CPU on it's own power supply, radiator, pump, and reservoir; the graphics cards on another set. Excessive, or efficient?
I'll be running only distilled water through the water blocks, with the exception of dye and something to kill all the tiny saboteurs. I toyed with the idea of methyl alcohol, but I couldn't find enough data on the long term effects.
I intend to suspend all electronics a good distance off the bottom of my enclosure. A distance proportionate to the volume of water circulating through both systems. This way, if ever a leak occurs, the water falls to the bottom of the enclosure and the electronic components are all left unharmed. Even if the leak was directly over a graphics card, the chance of a short circuit is pretty low, oil is amazing stuff. The water could then be pumped out without having to drain the oil from the enclosure.
While submerged in oil, previously unstable temperatures become stable- for whatever reason. I don't think it's been tested just how far temperatures could be pushed before a system lost stability while submerged. I like the idea of frying some chicken in my computer while playing Prototype....
Although it may seem otherwise, I'm not trying to construct some offensively expensive rig. We should only be looking at around $1,800.
Questions:
Are TECs worth it or would it be better to just stick to orthodox water cooling?
How much effect would water cooling have on top of the oil?
Are TEC/Waterblock kits worth the investment, or should I piece together my own?
What kind of overclock results do you guys think I could expect?
What kind of flaws do you see in my design?
The basic idea:
Q9550
EVGA 750i FTW
8 GB of 1066mhz dual channel memory- most likely G.Skill.
2x EVGA GTX 260- the revised edition with 216 processor cores and completely stock clocked. There's no need for factory overclocking as I'll be overclocking them myself.
A 750w power supply- the model is irrelevant as the system will be submerged in oil.
I will be overclocking the $%&^ out of it. I'd like to try to push 5 ghz. =D Bet that'd be nice to see on a Q9550. I've seen 4.7, so it might be possible. My real goal is just as high as I can get though; so if it's not 5, I'll still be able to sleep.
Most of the system will be submerged in mineral oil while optical and hard disks will remain dry.
The oil will be pumped through a 480mm radiator (I'm yet undecided on the volume of air I want to move over the radiator). No reservoir is needed.
Right now I'm toying with the idea of TECs sandwiched between a cold plate and a water block and fixed to anything hot. My main concern is the efficiency of TECs, they seem to not play well with anything that runs hot (ironic shortcoming for a method of cooling, I know). What I mean by this is that cooling efficiency decreases exponentially as temperature increases, that's a curve starting high and ending low on a graph. If we're not overclocking, TECs would be a fantastic option. However, if we're overclocking and increasing voltage, TECs soon lose their appeal and are outperformed by normal water cooling methods.
Now- this is all fine and dandy, but my system would be submerged in mineral oil. I'm trying to figure out how this would affect TECs. I'm looking for intelligent conjecture, with a rational explanation behind whatever idea is presented.
I'm not trying to attain sub zero temperatures. I don't see the point. I'm overclocking...
Food for though:
Condensation is not an issue when running TECs in a machine submerged in oil. Since there is no air (or extraneous amounts of) around any of the components being cooled, there is no water to condense.
In a normal submerged system there is pretty much just one system wide temperature, thanks to how wonderfully the oil distributes the heat. How do you think introducing water blocks and TECs will effect this trend?
The more TECs added, the more power required. I'll most likely require two additional power supplies for 4 TECs (north bridge, CPU, and GPUs). With three power supplies in the oil, would the temperature rise by a negligible amount?
I'd like to run the north bridge and CPU on it's own power supply, radiator, pump, and reservoir; the graphics cards on another set. Excessive, or efficient?
I'll be running only distilled water through the water blocks, with the exception of dye and something to kill all the tiny saboteurs. I toyed with the idea of methyl alcohol, but I couldn't find enough data on the long term effects.
I intend to suspend all electronics a good distance off the bottom of my enclosure. A distance proportionate to the volume of water circulating through both systems. This way, if ever a leak occurs, the water falls to the bottom of the enclosure and the electronic components are all left unharmed. Even if the leak was directly over a graphics card, the chance of a short circuit is pretty low, oil is amazing stuff. The water could then be pumped out without having to drain the oil from the enclosure.
While submerged in oil, previously unstable temperatures become stable- for whatever reason. I don't think it's been tested just how far temperatures could be pushed before a system lost stability while submerged. I like the idea of frying some chicken in my computer while playing Prototype....
Although it may seem otherwise, I'm not trying to construct some offensively expensive rig. We should only be looking at around $1,800.
Questions:
Are TECs worth it or would it be better to just stick to orthodox water cooling?
How much effect would water cooling have on top of the oil?
Are TEC/Waterblock kits worth the investment, or should I piece together my own?
What kind of overclock results do you guys think I could expect?
What kind of flaws do you see in my design?