Submerged Desktops
#1
Posted 07 February 2010 - 08:07 PM
TO be honest, right now, I am looking to do this on the dirt cheap - namely because I want to be out as little as possible should this not go well. Right now, I have a motherboard and processor on order.
AMD Sempron 140 2.7ghz its a $33 dollar CPU that should unlock to behave like an Athlon II X2... I know the Athlon 2 is cheap... but for a who knows what may cook the chip project.... $33 sounded much better.
GIGABYTE GA-MA78LM-S2H This I am going to hope with be a nice cheap ($60) board that will do the task, including unlocking the chip.
and I have some DDR2 800Mhz Corsair ram to toss in the board.
The power supply to be used will be a Silverstone ST-50F-ES 500Watt namely because I liked the price ($60), and 500watts will be more power than I need.
I am thinking about reusing my 9800GT EE card - as it does not need additional power, but intend to use this cooling:
ARCTIC COOLING ACCELS1 as it has a large radiator, and s passive cooling. I am trying to avoid fans... I know they say they don't mind the liquid... but I want to be sure.
The cpu cooling to be used will be ZEROtherm BTF95 CPU Cooler its $48, and should look good from the outside.... done correctly. Once again, it also has no fan.
now apart from this, there are (of course) other things that I will need to make this work well. obviously a fish tank, fluid, hard drive, and maybe even some way to cool the liquid. And this is where I start having questions. Will the fish tank really matter all that much? I mean does it NEED to be EITHER glass or acrylic? or will either one be fine? I was thinking about using this from walmart: Tetra Aquarium Starter Kit, 5 Gallons as it seems to be just the right size, and freaking cheap! But it is glass... not sure if that is a drawback or not.
I am trying to find an inexpensive hard drive solution, and at this point am simply thinking toss a Sata drive in the hood of the tank.
For some form of external cooling, do any of you think that the massive radiator the guys at puegot systems used is really necessary, or will a smaller 2x120 width radiator keep it cool enough? I have never done much with liquid setups, so if anyone can recommend an inexpensive pump/radiator setup please do. Remember though, while cheap, I do need it to be functional. So don't JUST look at the price please... I want this to have a solid chance of working.
I am also thinking about ordering the Lian-Li motherboard tray the recommend over at Puegot systems, but if there is another option, (better option?), please post it in here as well.
One other option I was considering for the hard drives was using a few CF to SATA adapters, and connect up 3 or 4 CF cards in raid.... I figure 4 16GB CF cards should be fairly quick... especially if they can read at 40MB/sec or more each.
so go ahead and post guys... I am ordering a few parts at a time, and will update as they come in, and even post pics as I get them.
#2
Posted 08 February 2010 - 11:51 AM
As far an aquarium made out of glass versus one made out of acrylic I can’t say from my experience, I used an acrylic tank for mine. I have seen some people claim that the mineral oil will degrade the silicon that seals a glass tank; I have my doubts about that though. My next oil submerged computer that I have planned will be with a glass tank. Glass conducts heat a lot better than acrylic and I can build a custom case easier with glass.
As far as cooling goes the 2x120 radiator should be sufficient to keep your system cool. The guys at Puegot used the large radiator because they were going for a silent system. I think it is necessary to have external cooling with the system you have planed. Not enough heat will be dissipated through the sides of the tank or the top, so the oil will just keep getting hotter.Feel free to post any questions you may have.
#3
Posted 08 February 2010 - 02:39 PM
#4
Posted 08 February 2010 - 03:27 PM
waldojim, on 08 February 2010 - 02:39 PM, said:
Having the back panel just above the oil may stop the wicking effect,I would say give it a try. If you submerge the PSU you will still have to have a junction in the line going to the wall and the power to your hard drive. Since you are using passive heat sinks I would have the return line from the rad split and return the cooled oil to the heatsink on the CPU and Video card, you want to create a bit of circulation in those areas. Using an air bubblier will also help to circulate the oil, just make sure that no bubbles come in contact with your components(they can create hot spots). You want to have a decent pump as the oil is harder to pump than water. I used a pump from a large aquarium that I had sitting around and it worked fine.Sorry I don't remember the make and I tossed it out awhile ago. You want the kind of pump that you submerge in the tank, it will work better and be quieter.
#5
Posted 08 February 2010 - 03:33 PM
and thanks for the heads up on the power cord... forgot about it.
I was also thinking I might grab a fountain pump - something designed for water that is not so clean, I found one at harbor freight with a 1/2" outlet, submersible, and pumps about 400gph
This post has been edited by waldojim: 08 February 2010 - 03:34 PM
#6
Posted 08 February 2010 - 03:53 PM
waldojim, on 08 February 2010 - 03:33 PM, said:
and thanks for the heads up on the power cord... forgot about it.
I was also thinking I might grab a fountain pump - something designed for water that is not so clean, I found one at harbor freight with a 1/2" outlet, submersible, and pumps about 400gph
I just did the aquarium pc as an experiment with some old components I had. I wanted to make sure it would actually work. I do have plans to build another one as part of an extreme case mod, a cross between steam punk and 60's scifi. I think the fountain pump is an excellent choice I will probably use a similar pump for my build, once a get all the parts for the case gathered up.
#7
Posted 10 February 2010 - 02:42 PM
lets assume in the end, I intend to use this rig (different CPU) to push some extreme OC.... What would happen if I introduced TEC cooling to the mix?
say I grab this 226 watt TEC and used what was normally a 80watt chip (dual core AMD for example)... how well do you think that can work out? I know that the TEC has a lot of limitations to start with, in an air cooled (even normal water cooling) it requires a lot of preparation to deal with condensation, and with the massive amounts of heat it dissipates. But with the oil cooling, condensation would no longer be a concern right? and with the oil it should be easy to keep the hot side of the TEC cool (using the heatsink I have planned for example).
This sounds almost too easy to implement... am I missing something?
-65C on the chip would be wonderful for overclocking... and I should think that extreme clocks would be much easier to achieve this way...
#8
Posted 10 February 2010 - 07:10 PM
http://www.pugetsyst...m/submerged.php
#9
Posted 10 February 2010 - 08:30 PM
waldojim, on 10 February 2010 - 02:42 PM, said:
lets assume in the end, I intend to use this rig (different CPU) to push some extreme OC.... What would happen if I introduced TEC cooling to the mix?
say I grab this 226 watt TEC and used what was normally a 80watt chip (dual core AMD for example)... how well do you think that can work out? I know that the TEC has a lot of limitations to start with, in an air cooled (even normal water cooling) it requires a lot of preparation to deal with condensation, and with the massive amounts of heat it dissipates. But with the oil cooling, condensation would no longer be a concern right? and with the oil it should be easy to keep the hot side of the TEC cool (using the heatsink I have planned for example).
This sounds almost too easy to implement... am I missing something?
-65C on the chip would be wonderful for overclocking... and I should think that extreme clocks would be much easier to achieve this way...
Interesting idea, I considered using TEC cooling as well but I never found anyone that had tried it in an oil submerged comp.My thought was to build a tank using double pain glass, if the oil is cooled bellow ambient air temps condensation is going to form on the outside of the tank. I also had concerns about how the oil would behave when cooled, it would definitely get thicker and harder to pump.Even if you are only running the TEC cooler to the CPU I would think the oil in the area of the CPU would be cooled bellow zero. I don't have a lot of experience with extreme cooling but it is my understanding that some components on the motherboard do not respond well to extreme cooling. Normally when using a TEC cooler it is just the CPU and/or the GPU that is cooled, but with the comp submerged other components would probably would be cooled as well. Oil submerged comps are not known for their ability to do extreme overclocks.There is a commercially available oil submerged comp and it does not overclock as well as some of the water cooled computers. The advantage of oil cooling is it is silent and the temps are consistent through out the system( makes the system stable at higher temps). If you decide to try the Tec cooling I would be very interested in how it works out.
#10
Posted 10 February 2010 - 08:40 PM
Being completely a noob with this topic, the only thing on the top of my mind is:
You overclock, generate lots of heat, and the liquid 'fluid' has a better thermal conductivity than the low density air? It is unclear to me if you will also submerge the HDD? If you do, I have apprehensions about that? Well, you won't overclock the HDD? You see HDD usually have a micron filter vented hole to equalize the barometric pressure differential inside the spinning platter area. Also these are called 'flying' heads. If for some reason or another this were to become contaminated with fluid (mineral oil or whatever inert non-conductive dielectric) I'm not sure HDD are designed to withstand these kinds of conditions?
Whatever, you sure captivated my curiosity? Go waldojim!
#11
Posted 10 February 2010 - 08:47 PM
Grr8008, on 10 February 2010 - 07:10 PM, said:
http://www.pugetsyst...m/submerged.php
If you have time Puget's forum has an interesting thread about the oil submerged computer. Their is a lot of information in the thread and a lot of questions are addressed.
#12
Posted 10 February 2010 - 08:49 PM
Now if I end up with the oil getting too thick to be moved, that would not be helpful at all. This will definitely need a bit of experimenting.
The bright side here, is that the TEC cooler is kinda cheap - so at least it is a cheap experiment for that portion!
#13
Posted 10 February 2010 - 08:59 PM
WinTard, on 10 February 2010 - 08:40 PM, said:
Being completely a noob with this topic, the only thing on the top of my mind is:
You overclock, generate lots of heat, and the liquid 'fluid' has a better thermal conductivity than the low density air? It is unclear to me if you will also submerge the HDD? If you do, I have apprehensions about that? Well, you won't overclock the HDD? You see HDD usually have a micron filter vented hole to equalize the barometric pressure differential inside the spinning platter area. Also these are called 'flying' heads. If for some reason or another this were to become contaminated with fluid (mineral oil or whatever inert non-conductive dielectric) I'm not sure HDD are designed to withstand these kinds of conditions?
Whatever, you sure captivated my curiosity? Go waldojim!
Defiantly you can not submerge the hard drive. Although there are sealed hard drive available they are designed for industrial and military applications, and they are very expensive. It would probably be cheaper to use an SSD if you wanted to submerge the hard drive. In addition to the pressure equalization vent some hard drives have ports that are used by the servo writer during manufacture, they are often just covered with sealing tape.
#14
Posted 10 February 2010 - 09:07 PM
waldojim, on 10 February 2010 - 08:49 PM, said:
Now if I end up with the oil getting too thick to be moved, that would not be helpful at all. This will definitely need a bit of experimenting.
The bright side here, is that the TEC cooler is kinda cheap - so at least it is a cheap experiment for that portion!
Ok I see what you thinking I was thinking of something like this. would work better, but its not cheap. Certainly would hurt to try out your idea, I think it would cause the system to actually run hotter as the Tec will produce a lot of heat.
#15
Posted 10 February 2010 - 09:13 PM
WinTard, on 10 February 2010 - 08:40 PM, said:
Being completely a noob with this topic, the only thing on the top of my mind is:
You overclock, generate lots of heat, and the liquid 'fluid' has a better thermal conductivity than the low density air? It is unclear to me if you will also submerge the HDD? If you do, I have apprehensions about that? Well, you won't overclock the HDD? You see HDD usually have a micron filter vented hole to equalize the barometric pressure differential inside the spinning platter area. Also these are called 'flying' heads. If for some reason or another this were to become contaminated with fluid (mineral oil or whatever inert non-conductive dielectric) I'm not sure HDD are designed to withstand these kinds of conditions?
Whatever, you sure captivated my curiosity? Go waldojim!
my intention with the hard drive, is to either go with some form of SSD, OR mount the drive in the aquariums hood.
The oil should do a better job cooling the system than air. Much like water cooling, the difference, is that everything is now kept at roughly the same temperature. This should actually lead to much more stable OverClocks... I think.
With an external radiator, I would like to see if I can maintain a steady 45C or so.... if I can get temps to stay down, I may be able to get a great deal of an OC out of it. One other solution that may work out ( not sure yet...) is using a true water cooling setup, and actually use 2 radiators, one inside the tank, one outside - and seal it off from the oil. I could then use fairly standard pumps, reservoirs etc, and not worry about killing pumps off. I just need to find a way to get the liquid to move freely through the radiators.
Which brings me to some interesting test results. While I am waiting on parts to arrive, I am testing with a few fans (and older parts) to see how they react to the oil. SO far, I would like to mention that Cooler Master 120mm fans are useless. They cannot maintain enough speed in the oil to actually DO anything. That said, I have a (very nice) ThermalTake 80mm 5000RPM fan (loud as hell BTW) That can maintain VERY STRONG movement, and creates an incredible current in there! and the oil certainly shuts it up! almost no noise out of it... for the first time in years.
#16
Posted 10 February 2010 - 09:14 PM
backbuster, on 10 February 2010 - 09:07 PM, said:
waldojim, on 10 February 2010 - 08:49 PM, said:
Now if I end up with the oil getting too thick to be moved, that would not be helpful at all. This will definitely need a bit of experimenting.
The bright side here, is that the TEC cooler is kinda cheap - so at least it is a cheap experiment for that portion!
Ok I see what you thinking I was thinking of something like this. would work better, but its not cheap. Certainly would hurt to try out your idea, I think it would cause the system to actually run hotter as the Tec will produce a lot of heat.
yeah.... I think I am spending plenty on this project to start with.... not sure if I want to dump a $1000 phase change unit in there just yet.
That would be nice to have though...
#17
Posted 10 February 2010 - 09:26 PM
waldojim, on 10 February 2010 - 09:14 PM, said:
backbuster, on 10 February 2010 - 09:07 PM, said:
waldojim, on 10 February 2010 - 08:49 PM, said:
Now if I end up with the oil getting too thick to be moved, that would not be helpful at all. This will definitely need a bit of experimenting.
The bright side here, is that the TEC cooler is kinda cheap - so at least it is a cheap experiment for that portion!
Ok I see what you thinking I was thinking of something like this. would work better, but its not cheap. Certainly would hurt to try out your idea, I think it would cause the system to actually run hotter as the Tec will produce a lot of heat.
yeah.... I think I am spending plenty on this project to start with.... not sure if I want to dump a $1000 phase change unit in there just yet.
That would be nice to have though...
LoL know what you mean. The cost can go up fairly quick when you get into extreme mods. I am in the same boat I don't think I will be spending the money to try out the phase change unit either, but I sure would like to see how it works.I found the site for the submerged computer that is available commercially it's called the Hardcore Reactor, check it out you might get some more ideas.
#18
Posted 11 February 2010 - 07:30 AM
#19
Posted 12 February 2010 - 03:17 AM
Guys, this is the second time I bought a gigabyte board... and the second time they failed me.
Unfortunately, this board is showing some stupid behavior... and once again, its something that is damn near impossible to troubleshoot. I had issues like this (impossible to troubleshoot that is) with an old socket 939 board.
in this case, I took the sempron tossed it in a box, with an Antec Earth Watts 550 watt PSU, and 4GB corsair XMS2 DDR 2 800Mhz ram, and a Hitachi 160GB hard drive.
Nothing special, nothing out of the ordinary, nothing uncommon. And yet, when windows OR Linux starts, I get a corrupted display. I tried using the 9800GT EE card that will be used in the final build. Same unusual effects. There are no known compatability issues. No known issues the the bios.... and Memtest86 says there is not a single issue.
So this means either the processor or the motherboard... I can run several basic CPU tests in the linux command line, CPU seems to hold up quite well... heck this chip will even unlock to the Athlon X2 VERY nicely, and remain stable....
The moment the graphics get used... its done. So this will be rma'd in the morning, and I will be waiting on a new motherboard... again. Glad I tested before I put it in oil.
Sorry guys if I sound upset, but it is because of the members here that I gave gigabyte another shot... and once again, the product has failed me. Never has an MSI board acted like this for me.
#20
Posted 12 February 2010 - 06:17 AM
Help






















