submersion/pelt chilled

wymjym · 08-03-2003, 09:47 AM

Hi,
I really want to build a submersion cooling project using mineral oil along with peltier chilling.
I know there will be quite a few hurdles to jump but one that I am most concerned with is capacitor failure.
I want this project to be reliable.
I know that many tin - can caps use a rubber sealing washer in their base. Oil normally will wipe out rubber in short order.
If I seal the caps to prevent oil from touching them then I will also be sealing in the gases that are going to want to escape when they get hot and expand….right?
So…is my concern valid (caps failing because of the oil, gases being trapped)?
Any work arounds?
wj

hydrogen18 · 08-03-2003, 10:41 AM

mineral oil submersion has been done many times before without this concern i beleive, so why bother? Anyhow, if you really want to get serious, extend the caps on the mobo outside of the container with preferably large wires(14 gauge solid)

wymjym · 08-03-2003, 12:00 PM

Quote:

Originally posted by hydrogen18
mineral oil submersion has been done many times before without this concern i beleive, so why bother?

are there any long term (a year or so) systems still running submerged? If so then it is a non-issue. A link or two would be appreciated.

Quote:

[i]Anyhow, if you really want to get serious, extend the caps on the mobo outside of the container with preferably large wires(14 gauge solid) [/b]

I'm serious, but not to the extent that I'll be moving components from the mobo.
wj

hydrogen18 · 08-03-2003, 03:48 PM

umm, get that serious?

]JR[ · 08-04-2003, 09:52 AM

Theres a bigger problem than caps leaking, thats the fact they dont like being cold at all. Also mosfets like it quite toasty too, the capacitors in particular would need to be insulated.

And then onto the next problem, chilled oil gets very viscous which leads to another problem of how you move it about, which youll need to do to stop the cpu/gpu frying.

Also if your serious enough to dunk the whole friggin setup into oil whats moving a few caps? Your never gunna be able to rma it really are you...

If your 100% going for immersion cooling, and your going to chill the immersion fluid its almost certainly going to have to be floreneit (sp.)...

Just my opinion...

]JR[

hydrogen18 · 08-04-2003, 10:09 AM

Mosfets are simply regulators, which prefer to be cold, so they are a moot point. Extending capacitors isnt hard, just use good wire and get some soldering practice prior to doing it if u havent soldered b4. My only other though about the caps would be to make small molds that cold be placed over the caps and injected with epoxy. As for circulating the mineral oil, why? Just place it all in a sealed box and put the evaporator of a phase change system in the box too, voila, totally chilled system, no pump, no condensation.

wymjym · 08-04-2003, 11:06 AM

Sorry everyone, lots of variables and not enough typing speed to explain my thoughts.
Basically....ambient or thereabouts oil temps are desired. So the pumping problems are not an issue. Condensation will not be an issue as I'll handle the temps as I have with my water chiller.

Mosfets do like cooler rather than 'toasty' temps.

If I seal the caps and they are then submerged...what type of temp rise should they see? and is it enough that I would have a mini bomb (because of the internal pressure) or will the oil bath keep them so cool that the expanding internal gases will not be an issue?
Does anyone know of a submerged system that has been running for months?
I plan on using an external pelt chiller to cool the oil and pump it into the 'tank'...similar to my present water system. I will have almost limitless options for directing the cool oil. I can use multiple hoses directed at specific hot areas, I can pump it into 'water' blocks on the cpu, gpu, nb etc and then let the heated oil exit into the pool of oil, yadaydayadaydayada.
wj

iggiebee · 08-04-2003, 11:50 AM

Quote:

Originally posted by wymjym
Does anyone know of a submerged system that has been running for months?
I plan on using an external pelt chiller to cool the oil and pump it into the 'tank'...similar to my present water system. I will have almost limitless options for directing the cool oil. I can use multiple hoses directed at specific hot areas, I can pump it into 'water' blocks on the cpu, gpu, nb etc and then let the heated oil exit into the pool of oil, yadaydayadaydayada.
wj

Look here

wymjym · 08-04-2003, 12:12 PM

Quote:

Originally posted by iggiebee
Look here

thanks, is there any information available about this setup?
wj

sevisehda · 08-04-2003, 12:14 PM

The only issue I've heard with using mineral oil is the fact that it will dissolve some PCBs over time. I don't put too much credit to this because I've never heard about this actually happening. I've never heard about any problem with capacitors so I wouldn't give them any specail treatment. As for mosfets they like it cool. In fact as they heat up they output more heat because resistance increases. The Max3 series actually cools the mosfets.

The one thing you should keep in mind, and its in the picture, is to use standoffs under the board. Overtime water can condense and form 'bubbles' on the bottom of your system, the standoffs will prevent those bubbles from shorting your board.

hydrogen18 · 08-04-2003, 12:31 PM

you could use a non conductive thermal epoxy on the capacitors...

wymjym · 08-04-2003, 01:15 PM

I am building some new copper chiller blocks so there goes the next few weeks but.....
I have an old slot A system that I'm going to dunk in a vat of oil and let it run for a while to see what hiccups I might run into.
Thanks for the heads up on the standoffs.

I noticed that in the pictures (click on the old link within the photo page) the original system had the ps supported above the oil and the later one has it submerged.

I plan on slanting the mobo with the mem slots at the bottom and the i/o ports on top.....rather than add/buy a remote i/o relocator I'm going to try to simply keep the connections either out of the oil or at the very top of it.
This is one of those build it right the first time affairs because it will be way too messy to keep changing it around.
wj

hydrogen18 · 08-04-2003, 02:53 PM

wow, i just looked at that link and with the volume of oil used there is no reason for cooling at all really....

sevisehda · 08-04-2003, 04:12 PM

Quote:

wow, i just looked at that link and with the volume of oil used there is no reason for cooling at all really....

Nooooooooo...............................

It may take a while to heat up but it will. Also keep in mind its hard to move the oil around. Without somekind of agitation the oil will get very hot near the CPU/GPU and stay cool near the exterior. A design that I've seen used is to use a submerisable pump to push the coolant through a rad then split the flow and have it 'blow' onto the hotspots.

SexyMF · 08-08-2003, 08:22 PM

I don't think the molecule size of mineral oil can seep into the rubber of the capacitors. I have heard this from multiple sources (reputable). MOSFETs have a positive temperature coefficient meaning Rds (on-state resistance) increases with heat. Device such as BJTs have negative co-efficients (not that this helps you!).

The only other thing I can comment on is don't submerse any PCI cards that have relays on them (ie mechanical).

I say go for it. I've been curious about this myself.

Oh, and if the oil gets hot enough just throw in some chips

hydrogen18 · 08-08-2003, 10:52 PM

most relays found on modern pci cards are small and are completely sealed, so oil entering them is not a problem.

08-03-2003, 09:47 AM	#1
wymjym Cooling Neophyte Join Date: Sep 2002 Location: near Austin Posts: 96	submersion/pelt chilled Hi, I really want to build a submersion cooling project using mineral oil along with peltier chilling. I know there will be quite a few hurdles to jump but one that I am most concerned with is capacitor failure. I want this project to be reliable. I know that many tin - can caps use a rubber sealing washer in their base. Oil normally will wipe out rubber in short order. If I seal the caps to prevent oil from touching them then I will also be sealing in the gases that are going to want to escape when they get hot and expand….right? So…is my concern valid (caps failing because of the oil, gases being trapped)? Any work arounds? wj

08-08-2003, 08:22 PM	#15
SexyMF Cooling Neophyte Join Date: Jul 2003 Location: NZ Posts: 26	I don't think the molecule size of mineral oil can seep into the rubber of the capacitors. I have heard this from multiple sources (reputable). MOSFETs have a positive temperature coefficient meaning Rds (on-state resistance) increases with heat. Device such as BJTs have negative co-efficients (not that this helps you!). The only other thing I can comment on is don't submerse any PCI cards that have relays on them (ie mechanical). I say go for it. I've been curious about this myself. Oh, and if the oil gets hot enough just throw in some chips __________________ A7N8X Deluxe AMD 2600+ 512MB Corsair XMS DDR333

08-03-2003, 10:41 AM	#2
hydrogen18 Cooling Savant Join Date: Feb 2003 Location: USA Posts: 248	mineral oil submersion has been done many times before without this concern i beleive, so why bother? Anyhow, if you really want to get serious, extend the caps on the mobo outside of the container with preferably large wires(14 gauge solid)

08-03-2003, 03:48 PM	#4
hydrogen18 Cooling Savant Join Date: Feb 2003 Location: USA Posts: 248	umm, get that serious?

08-04-2003, 09:52 AM	#5
]JR[ Cooling Neophyte Join Date: Apr 2003 Location: UK Posts: 74	Theres a bigger problem than caps leaking, thats the fact they dont like being cold at all. Also mosfets like it quite toasty too, the capacitors in particular would need to be insulated. And then onto the next problem, chilled oil gets very viscous which leads to another problem of how you move it about, which youll need to do to stop the cpu/gpu frying. Also if your serious enough to dunk the whole friggin setup into oil whats moving a few caps? Your never gunna be able to rma it really are you... If your 100% going for immersion cooling, and your going to chill the immersion fluid its almost certainly going to have to be floreneit (sp.)... Just my opinion... ]JR[

08-04-2003, 10:09 AM	#6
hydrogen18 Cooling Savant Join Date: Feb 2003 Location: USA Posts: 248	Mosfets are simply regulators, which prefer to be cold, so they are a moot point. Extending capacitors isnt hard, just use good wire and get some soldering practice prior to doing it if u havent soldered b4. My only other though about the caps would be to make small molds that cold be placed over the caps and injected with epoxy. As for circulating the mineral oil, why? Just place it all in a sealed box and put the evaporator of a phase change system in the box too, voila, totally chilled system, no pump, no condensation.

08-04-2003, 11:06 AM	#7
wymjym Cooling Neophyte Join Date: Sep 2002 Location: near Austin Posts: 96	Sorry everyone, lots of variables and not enough typing speed to explain my thoughts. Basically....ambient or thereabouts oil temps are desired. So the pumping problems are not an issue. Condensation will not be an issue as I'll handle the temps as I have with my water chiller. Mosfets do like cooler rather than 'toasty' temps. If I seal the caps and they are then submerged...what type of temp rise should they see? and is it enough that I would have a mini bomb (because of the internal pressure) or will the oil bath keep them so cool that the expanding internal gases will not be an issue? Does anyone know of a submerged system that has been running for months? I plan on using an external pelt chiller to cool the oil and pump it into the 'tank'...similar to my present water system. I will have almost limitless options for directing the cool oil. I can use multiple hoses directed at specific hot areas, I can pump it into 'water' blocks on the cpu, gpu, nb etc and then let the heated oil exit into the pool of oil, yadaydayadaydayada. wj

08-04-2003, 12:14 PM	#10
sevisehda Cooling Savant Join Date: May 2003 Location: NY Posts: 234	The only issue I've heard with using mineral oil is the fact that it will dissolve some PCBs over time. I don't put too much credit to this because I've never heard about this actually happening. I've never heard about any problem with capacitors so I wouldn't give them any specail treatment. As for mosfets they like it cool. In fact as they heat up they output more heat because resistance increases. The Max3 series actually cools the mosfets. The one thing you should keep in mind, and its in the picture, is to use standoffs under the board. Overtime water can condense and form 'bubbles' on the bottom of your system, the standoffs will prevent those bubbles from shorting your board.

08-04-2003, 12:31 PM	#11
hydrogen18 Cooling Savant Join Date: Feb 2003 Location: USA Posts: 248	you could use a non conductive thermal epoxy on the capacitors...

08-04-2003, 01:15 PM	#12
wymjym Cooling Neophyte Join Date: Sep 2002 Location: near Austin Posts: 96	I am building some new copper chiller blocks so there goes the next few weeks but..... I have an old slot A system that I'm going to dunk in a vat of oil and let it run for a while to see what hiccups I might run into. Thanks for the heads up on the standoffs. I noticed that in the pictures (click on the old link within the photo page) the original system had the ps supported above the oil and the later one has it submerged. I plan on slanting the mobo with the mem slots at the bottom and the i/o ports on top.....rather than add/buy a remote i/o relocator I'm going to try to simply keep the connections either out of the oil or at the very top of it. This is one of those build it right the first time affairs because it will be way too messy to keep changing it around. wj

08-04-2003, 02:53 PM	#13
hydrogen18 Cooling Savant Join Date: Feb 2003 Location: USA Posts: 248	wow, i just looked at that link and with the volume of oil used there is no reason for cooling at all really....

08-08-2003, 10:52 PM	#16
hydrogen18 Cooling Savant Join Date: Feb 2003 Location: USA Posts: 248	most relays found on modern pci cards are small and are completely sealed, so oil entering them is not a problem.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)