serpretetsky
2[H]4U
- Joined
- Dec 24, 2008
- Messages
- 2,180
Computer Oil Project.
So i've decided to do some oil submersion testing. I'm interested in which oil / fluid performs best per price.
I've been mesmirized by submersion cooling for quite a while because of two unique properties it has that other methods of cooling do not.
a) completely silent
b) complete cooling of ALL components (not unique peices like cpu, NB, gpu, etc)
it's also interesting to note that submersion cooling is not nececessarily exclusive of other types of cooling. You can still get better heatsinks, insulated water cooling loops, and possible below-zero cooling systems (choose your fluid wisely!).
===============================================================================================================================================================
TEST #1 "will engine oil kill my computer"
I'm beginning with testing of engine / motor oil. I've been going all over the internet and besides a tom's hardware article that recommends using engine oil instead of vegetable, i can't find anybody who's used engine or motor oil to cool computer. Most people have told me that it will probably eat away at plastic and rubber components (capacitor seals, the boards themselves, bla bla bla). Well... i'm hoping to find out if that's true.
I've assembled an extremely high end computer consisting of:
pentium 3 600mhz
asus p3b-f mobo
geforce 2 gts
250watt PSU
512MB pc100 sdram
computer passed 24 hour stability test before oil consisting of:
prime95 small fft
hci memtest
villagemark
OIL poured in.
,I have submerged this computer in formula shell 5-30 oil (the cheapest i could find...)
It is currently being stability tested in oil.
I am not monitoring temps too closely because i'm not concerned with that in this part of my testing. Temps did drop from 58C to 42C on cpu as i poured oil, but that was before i let the oil settle to final temps. I do also have three fans running in the setup: cpu, gpu, psu.
UPDATE 4/3/09:
24 hour stability test in oil has passed. The oil has grown pleasantly warm. Right after I closed the stability test i decided to check bios really quick for temps. CPU was hovering from 60C to 62C, NB was around 42C. The cpu temp is fine, however, im getting the feeling NB might be higher than it would be without oil because of being heated from oil.
I've also began to wonder about the Power Supply being submerged in oil. The only benefit PSU sees from being submerged is being silenced, i'm not sure any added cooling for the PSU benefits me at all. On the other hand, is the PSU heating up the oil noticably and causing all the other components to be warmer than i want them?
All of the electronics in a computer convert most of the energy they draw to heat. The PSU is inputing a certain amount of power from the wall. If the PSU is around 70% eifficient, that means 70% of the power will go to the electronics, which convert it straight to heat anyways, and around 30% of the power the PSU will turn into heat. So then anywhere from 1/3 to 1/4 of all the heat being generated in my little system is from the PSU. Is any of my math wrong here?
I also do not want to remove the PSU completely from the oil, because although this is all theoretical testing, I'm doing it from the standpoint of creating an actual case designed for custom hardware and oil. If such a case were to be built, trying to keep the PSU separate from the rest of the components would be a pain in the ass.
I've been considering creating some sort of an insulating barrier with holes for cables between the PSU and the rest of the components, but i would have to make sure that the PSU has less than 1/3 of the total surface area of the tub and surface of oil to be used for cooling. Otherwise, if i just create an insulating barrier down the middle, I would probably get the opposite of what i want, with my PSU being cooled even better than before, and everything else heating up more.
UPDATE 5/6/09
So the computer has been running strong with no issues. I try to use it like a normal computer but with a lot more stress testing. Usually I leave it on, sometimes i have different stress tests going, once in a while i'll turn it off overnight, and then boot it back up in the morning. I took it out of the oil yesterday to inspect it, no difference so far. All the fans spin just fine too (PSU , video , and cpu). Also, as a side note, i've noticed all of the other oil projects have trouble with oil seeping up the cables coming out of the oil, interestingly, i have not had this problem at all. I have felt the cables just two inches above the oil level and they are completely dry.
UPDATE 5/9/09
I have put the computer back in the tub for further testing. While i was doing this, i have felt the first effects of the motor oil. Sure enough, most of the plastic parts on my computer have become quite brittle. I didn't realize it at first, but after attempting to bend my round ide cables, i found they simple wouldn't bend. The round plastic cover they use to cover the 80 or 40 wires is not bendable at all anymore. I then checked the insulation for all the smaller cables (PSU wires , cpu fan wires, etc) and sure enough, they are all a little more brittle and "springy". Finally, I tried bending my video card (just a little) and found it to be harder than my ati card which it used to bend the same as.
UPDATE 6/30/09
have lost some of my enthususiasm for this project, none the less, i still have the computer up and running. Passes prime95 small fft + HCI memtest + villagemark running at the same time over 24 hour period no problem. No issues, will need to check to see if all the fans are still running in the future. Last stress test was started during 100 degree fahrenheit weather.
UPDATE 10/6/09
So i came to check out the computer about 3 weeks ago and it wasn't booting anymore. Due to lack of time, i have delayed messing with it. Today i have finally taken it apart to see what was up. The machine wasn't posting, nothing was being displayed. I pulled the whole thing out of the oil and tested it. All the fans were still working fine, but for some reason no video is going to the monitor. I grabbed the setup i was using for the overclocking section and found that it didn't boot anymore either (using the same PSU). Unfortunately i can't really investigate what exactly failed. The overclocking motherboard was never properly cleaned after it's overclocking run, so it's been sitting in a layer of the same oil as the endurance run. The PSU obviously provides power to the boards (all the fans start up) and the PSU fan works as well, however, i suppose it's possible something in the PSU fried somehow, and now it's provided proper voltages on one of the lines.
This test has come to its end. The computer, after 6 months, has failed. I could only conclude that one of two things happened.
1) the oil messed up the PSU somehow
2) the oil messed up a critical component (cpu or mobo) on BOTH the endurance machine and the overclock machine.
I'm sorry i didn't get more indepth, i had to clean out my test setup as its in the way of some other stuff, plus i don't have more equipment to try to and isolate which component failed exactly.
I would, through personal experience, not recommend using motor oil to submerge your computer
p.s. as a sidenote, it's interesting that all the plastic has gotten quite brittle. Different plastics seem to react differently. The AC cable still has some flex to it, the plastic cover that coevered my round ide cables is completely solid (it was very soft before). All the pcb boards are a little stiffer.
===============================================================================================================================================================
test #2 Does engine oil affect any thermal grease used?
The other concern i had was whether or not engine oil has any effect on thermal paste (dissolves it? chemically reacts with it?) For this test i simply took three different samples: Arctic Silver 3, Arctic Silver ceramique, and generic thermalright grease, and submerged them in a tube of engine oil. I will be checking on it once every two days to see if the mass of grease has changed or if the consistency is different.
UPDATE 4/4/09
I have only been giving the greases visual inspections, i decided i don't want to mess around with them for about 3 weeks, when i can do a direct before and after photo comparison and then play around with them. The Silver ceramique and thermalright grease (both generic white looking) are visually the same. The Arctic Silver 3 has taken on the appearance of dried mud.
UPDATE 5/8/09
Pulled the thermal paste out of the test tube and also took photos, here's some before and after photos
BEFORE------------------------------------------AFTER
Both the white pastes haven't changed, however, the arctic silver has noticeably changed. The arctic silver has become very liquid like, it was dripping off of the stick as i pulled it out and put it to drain.
UPDATE 6/30/09
The only thing left of the arctic silver is a silvery puddle, not the original dab that used to be there. Both the white pastes are still "dabs".
UPDATE 10/6/09
arctic silver is just a stain now,both the white pastes look exactly the same as when i put them on. I am concluding this test. Arctic silver seems to be a no go for regular motor oil.
===============================================================================================================================================================
TEST #3 : Overclocking Test
I am going to test maximum overclock of a computer in air, as well as in motor oil (same oil as in TEST 1). As usual, i have gathered amazing hardware:
amd k6-2 (unknown stock clock)
unknown motherboard
64mb 100mhz sdram
(yeah, im missing some information, give me a break, it's 1:00)
maximum overclock achieved in air is 95mhz * 4.5 @ 2.9V cpu (tried 3.1V and 3.2V, didn't help. These things can take a lot of beating though)
maximum overclock achieved in oil is 95mhz * 4.5 @ 2.9V cpu (couldn't do anything higher, no matter how much i played around with voltage or fsb/multiplier settings)
As a side note, trying to overclock a computer with jumpers for fsb, mult, and Vcore while it's submerged in dark oil is the most annoying thing possible. I will have to try this test again later with a slightly more modern machine with more voltage controls.
===============================================================================================================================================================
While this test is underway, I'm gladly accepting advice on testing as well as different fluids to test. I know i wanto compare engine oil to mineral oil (assuming engine oil passes my first tests...). Tell me if there's any other fluids you want tested. I'm also gladly accepting computer hardware ( i'm going to need 2 or more identical computers, same PSU's, same CPU's, same mobo's... it can be old hardware...)
I'm still not 100% sure how i will do a comparison test of different fluids because I
a) do not have control of environment temperature
b) am cheap.
I know i will be focusing on stable overclockability and NOT temperature, but I don't know if i will be using multiple identical computers or if i find some other method.
So i've decided to do some oil submersion testing. I'm interested in which oil / fluid performs best per price.
I've been mesmirized by submersion cooling for quite a while because of two unique properties it has that other methods of cooling do not.
a) completely silent
b) complete cooling of ALL components (not unique peices like cpu, NB, gpu, etc)
it's also interesting to note that submersion cooling is not nececessarily exclusive of other types of cooling. You can still get better heatsinks, insulated water cooling loops, and possible below-zero cooling systems (choose your fluid wisely!).
===============================================================================================================================================================
TEST #1 "will engine oil kill my computer"
I'm beginning with testing of engine / motor oil. I've been going all over the internet and besides a tom's hardware article that recommends using engine oil instead of vegetable, i can't find anybody who's used engine or motor oil to cool computer. Most people have told me that it will probably eat away at plastic and rubber components (capacitor seals, the boards themselves, bla bla bla). Well... i'm hoping to find out if that's true.
I've assembled an extremely high end computer consisting of:
pentium 3 600mhz
asus p3b-f mobo
geforce 2 gts
250watt PSU
512MB pc100 sdram
computer passed 24 hour stability test before oil consisting of:
prime95 small fft
hci memtest
villagemark
OIL poured in.
,I have submerged this computer in formula shell 5-30 oil (the cheapest i could find...)
It is currently being stability tested in oil.
I am not monitoring temps too closely because i'm not concerned with that in this part of my testing. Temps did drop from 58C to 42C on cpu as i poured oil, but that was before i let the oil settle to final temps. I do also have three fans running in the setup: cpu, gpu, psu.
UPDATE 4/3/09:
24 hour stability test in oil has passed. The oil has grown pleasantly warm. Right after I closed the stability test i decided to check bios really quick for temps. CPU was hovering from 60C to 62C, NB was around 42C. The cpu temp is fine, however, im getting the feeling NB might be higher than it would be without oil because of being heated from oil.
I've also began to wonder about the Power Supply being submerged in oil. The only benefit PSU sees from being submerged is being silenced, i'm not sure any added cooling for the PSU benefits me at all. On the other hand, is the PSU heating up the oil noticably and causing all the other components to be warmer than i want them?
All of the electronics in a computer convert most of the energy they draw to heat. The PSU is inputing a certain amount of power from the wall. If the PSU is around 70% eifficient, that means 70% of the power will go to the electronics, which convert it straight to heat anyways, and around 30% of the power the PSU will turn into heat. So then anywhere from 1/3 to 1/4 of all the heat being generated in my little system is from the PSU. Is any of my math wrong here?
I also do not want to remove the PSU completely from the oil, because although this is all theoretical testing, I'm doing it from the standpoint of creating an actual case designed for custom hardware and oil. If such a case were to be built, trying to keep the PSU separate from the rest of the components would be a pain in the ass.
I've been considering creating some sort of an insulating barrier with holes for cables between the PSU and the rest of the components, but i would have to make sure that the PSU has less than 1/3 of the total surface area of the tub and surface of oil to be used for cooling. Otherwise, if i just create an insulating barrier down the middle, I would probably get the opposite of what i want, with my PSU being cooled even better than before, and everything else heating up more.
UPDATE 5/6/09
So the computer has been running strong with no issues. I try to use it like a normal computer but with a lot more stress testing. Usually I leave it on, sometimes i have different stress tests going, once in a while i'll turn it off overnight, and then boot it back up in the morning. I took it out of the oil yesterday to inspect it, no difference so far. All the fans spin just fine too (PSU , video , and cpu). Also, as a side note, i've noticed all of the other oil projects have trouble with oil seeping up the cables coming out of the oil, interestingly, i have not had this problem at all. I have felt the cables just two inches above the oil level and they are completely dry.
UPDATE 5/9/09
I have put the computer back in the tub for further testing. While i was doing this, i have felt the first effects of the motor oil. Sure enough, most of the plastic parts on my computer have become quite brittle. I didn't realize it at first, but after attempting to bend my round ide cables, i found they simple wouldn't bend. The round plastic cover they use to cover the 80 or 40 wires is not bendable at all anymore. I then checked the insulation for all the smaller cables (PSU wires , cpu fan wires, etc) and sure enough, they are all a little more brittle and "springy". Finally, I tried bending my video card (just a little) and found it to be harder than my ati card which it used to bend the same as.
UPDATE 6/30/09
have lost some of my enthususiasm for this project, none the less, i still have the computer up and running. Passes prime95 small fft + HCI memtest + villagemark running at the same time over 24 hour period no problem. No issues, will need to check to see if all the fans are still running in the future. Last stress test was started during 100 degree fahrenheit weather.
UPDATE 10/6/09
So i came to check out the computer about 3 weeks ago and it wasn't booting anymore. Due to lack of time, i have delayed messing with it. Today i have finally taken it apart to see what was up. The machine wasn't posting, nothing was being displayed. I pulled the whole thing out of the oil and tested it. All the fans were still working fine, but for some reason no video is going to the monitor. I grabbed the setup i was using for the overclocking section and found that it didn't boot anymore either (using the same PSU). Unfortunately i can't really investigate what exactly failed. The overclocking motherboard was never properly cleaned after it's overclocking run, so it's been sitting in a layer of the same oil as the endurance run. The PSU obviously provides power to the boards (all the fans start up) and the PSU fan works as well, however, i suppose it's possible something in the PSU fried somehow, and now it's provided proper voltages on one of the lines.
This test has come to its end. The computer, after 6 months, has failed. I could only conclude that one of two things happened.
1) the oil messed up the PSU somehow
2) the oil messed up a critical component (cpu or mobo) on BOTH the endurance machine and the overclock machine.
I'm sorry i didn't get more indepth, i had to clean out my test setup as its in the way of some other stuff, plus i don't have more equipment to try to and isolate which component failed exactly.
I would, through personal experience, not recommend using motor oil to submerge your computer
p.s. as a sidenote, it's interesting that all the plastic has gotten quite brittle. Different plastics seem to react differently. The AC cable still has some flex to it, the plastic cover that coevered my round ide cables is completely solid (it was very soft before). All the pcb boards are a little stiffer.
===============================================================================================================================================================
test #2 Does engine oil affect any thermal grease used?
The other concern i had was whether or not engine oil has any effect on thermal paste (dissolves it? chemically reacts with it?) For this test i simply took three different samples: Arctic Silver 3, Arctic Silver ceramique, and generic thermalright grease, and submerged them in a tube of engine oil. I will be checking on it once every two days to see if the mass of grease has changed or if the consistency is different.
UPDATE 4/4/09
I have only been giving the greases visual inspections, i decided i don't want to mess around with them for about 3 weeks, when i can do a direct before and after photo comparison and then play around with them. The Silver ceramique and thermalright grease (both generic white looking) are visually the same. The Arctic Silver 3 has taken on the appearance of dried mud.
UPDATE 5/8/09
Pulled the thermal paste out of the test tube and also took photos, here's some before and after photos
BEFORE------------------------------------------AFTER
Both the white pastes haven't changed, however, the arctic silver has noticeably changed. The arctic silver has become very liquid like, it was dripping off of the stick as i pulled it out and put it to drain.
UPDATE 6/30/09
The only thing left of the arctic silver is a silvery puddle, not the original dab that used to be there. Both the white pastes are still "dabs".
UPDATE 10/6/09
arctic silver is just a stain now,both the white pastes look exactly the same as when i put them on. I am concluding this test. Arctic silver seems to be a no go for regular motor oil.
===============================================================================================================================================================
TEST #3 : Overclocking Test
I am going to test maximum overclock of a computer in air, as well as in motor oil (same oil as in TEST 1). As usual, i have gathered amazing hardware:
amd k6-2 (unknown stock clock)
unknown motherboard
64mb 100mhz sdram
(yeah, im missing some information, give me a break, it's 1:00)
maximum overclock achieved in air is 95mhz * 4.5 @ 2.9V cpu (tried 3.1V and 3.2V, didn't help. These things can take a lot of beating though)
maximum overclock achieved in oil is 95mhz * 4.5 @ 2.9V cpu (couldn't do anything higher, no matter how much i played around with voltage or fsb/multiplier settings)
As a side note, trying to overclock a computer with jumpers for fsb, mult, and Vcore while it's submerged in dark oil is the most annoying thing possible. I will have to try this test again later with a slightly more modern machine with more voltage controls.
===============================================================================================================================================================
While this test is underway, I'm gladly accepting advice on testing as well as different fluids to test. I know i wanto compare engine oil to mineral oil (assuming engine oil passes my first tests...). Tell me if there's any other fluids you want tested. I'm also gladly accepting computer hardware ( i'm going to need 2 or more identical computers, same PSU's, same CPU's, same mobo's... it can be old hardware...)
I'm still not 100% sure how i will do a comparison test of different fluids because I
a) do not have control of environment temperature
b) am cheap.
I know i will be focusing on stable overclockability and NOT temperature, but I don't know if i will be using multiple identical computers or if i find some other method.
Last edited: