Thermal compound + oil

C

CyborgTMT

n00b
Joined
Jul 1, 2008
Messages
47
I've got an issue and, after being a visitor here for years, figured this would be the best place to ask. I'm currently building a submerged pc using mineral oil. The problem is my Arctic Silver I used happens to be oil based. I now have a lovely grey pile below my cpu/heatsink. Fortunately this is only a test build using older parts to work out things like this. I've tested every thermal compound in the house, but all 'melt' away when placed in oil. Anyone have a good option to use for this?

-thanks
 
The Puget system used Arctic Silver also. I've been using this system for a couple of months now and the temperatures on the CPU are slowly rising under load. When I first built it I ran a 24hr stress test and the temp never climbed over 48C. Now after just a few hours of gaming I'm in the high 50s. Still acceptable, but I plan on overclocking for the final build and need better heat transfer.
 
Actually, we didn't use Artic Silver. We used the thermal pad on the heatsink, no additional paste added...because if you add paste, it ends up floating around in the tank or in a big glop at the bottom of the aquarium.
 
Jon Bach posted that Arctic Silver was used, which was part of my reason for using it myself at first. I lap both the heat sink and the spreader on the CPU so I use very little TIM anyway in my builds. Even so, what was there is now gone. I pulled the tower off and only a very small bit of a grey stain remains. I've tested a stock pad on an old 939 stock heat sink and it melts away too. Looks like I'm going to have to check out pastes that are use for applications other that PCs.

Some pics if interested
http://forums.7thes.com/index.php?showtopic=2812

I'm redoing the outside wood work since the 'stainable' filler failed to mention that it stained purple!
 
You don't need (and should'nt use) TIM in a submersion based computer.

Lapping would be good , but probably isnt required.

Keep in mind : The submersion fluid will act as it's own thermal interface material.

:D
 
CyborgTMT said:
Jon Bach posted that Arctic Silver was used, which was part of my reason for using it myself at first.
Click to expand...

Yeah, Jon was the one I asked when I saw your question, because the aquarium PC was built before I came here. Perhaps that was an earlier version of the build--he said his reasons for omitting Artic Silver in our current build were to keep the system as clean as possible.

Also, another interesting bit---our blue cathode tube in the top has turned pink. I'm not sure why that is, exactly, but there you go.
 
The submersion fluid will act as it's own thermal interface material.
Click to expand...

True, as this is the only thing currently between my CPU and Heat sink. But the down side is that the oil by itself is not as efficient as a TIM. The build currently has an AMD 4400 X2 which will get replaced by a Intel quad overclocked.

@Heather
Thanks for the input from Puget, your build was actually part of my inspiration for doing a submerged pc project. I've had cathodes that have changed their colors in the past over time so I don't think the oil played a part in that. Last I checked you guys/gals didn't have a circulation pump on you system, but if you decide to put one back on I recommend picking up a cheap Fram fuel filter. The Fram G3 fuel filter is 3/8" and, after a few days of cycling the oil, cleared all the haze from the Arctic silver and other stuff that found it's way into the system.

@BrainEater
I dropped a pair of UV cathodes into my tank and they worked fine, but I didn't like the look of the lighting. I've switched to LED strips and they create better illumination.

Current system specs:
Asus A8N-SLI Deluxe
ADM 4400 X2 @ stock
Corsair PC3200 2 Gigs
BFG 8800GT OC

Here are the temperatures of idle then after running 3 X Prime95 and Furmark for 1 hour.
Idle temps / 1 hr:
Oil: 32C / 36C
CPU: 38C / 54C
Sys: 34C / 38C
GPU: 42C / 52C
Ambient: 25C

Prior to the loss of the Arctic Silver the same test resulted in the oil/38, CPU/46, and Sys/41. The chipset has no fan on it so the measured system temp evenly corresponds with the oil temp. Without a TIM less heat is being pulled off the CPU into the oil giving me a 8 degree increase. This doesn't seem like much now, but with a quad running @ 4Ghz the difference will be much more dramatic.
 
Yep.

When I put together the Thinktank , one of the things I discovered a long the way , Is the incredible need to examine every single piece from a 'materials compatibility' standpoint.I spent literally months examining each and every little bit that was going into submersion , and asking myself ' What is my submersion fluid going to do to this piece...' ...

One of the things I discovered , is that all of the TIM's I looked into , from AS5 to plain old silicone TIM , will dissolve in a non-polar (read : oil/dielectric fluid) solvent.

If you wanted to actually use a TIM , you'd need a water based , non-metallic TIM.I'm not sure such a thing exists.

GL !

On a side note : I'm adding a new criteria to my 'materials compatiblity' charts on the thinktank rebuild.The ability to safely withstand -100 C. ;)
 
My UVs sat in the oil for almost 4 weeks, but I did seal the ends with silicon so that might be the solution. Or it could be just different brands reacting that way. I have noticed that the mineral oil does have an impact on plastics/vinyls such as Tygon tubing. After a few weeks the tubing has shrunk a bit and has a significant loss of flexibility (significant meaning hard as rock!). Depending on the construction of the cathode, this may also have played a part of it. I'll drop an unsealed tube in and see what happens over the next few weeks.

BTW BrainEater, I loved that build. A bit too much for me to pull off personally, but kick ass none the less. I had contemplated using a TEC, but I couldn't justify the benefit to power use ratio. I still have to be able to pay the electric bill each month :D
 
Thanx mang ! I did'nt wanna jack the thread with that linky but I was'nt sure if you'd seen that or not.....And yea the TECs were an outrageous power hog.I've fixed that problem.

:D

-------

I'm very curious about the whole cold cathode color change.I'm leaning towards a high voltage effect.....

I think you'll like the LED's better anyways.

Have you seen the Luxeon LED's ? ... Check em out.They make 1 watt RGB models....Cree X-lamp is another LED co to check out.

:D
 
CyborgTMT said:
Think I found a solution:
http://www.coollaboratory.com/en/product_metalpad.shtml
Click to expand...

:D :D

I was waiting to see who posted that first. gj.

Here's the deal with that.

It's an all-metal (Gallium-Indium alloy) TIM pad.

If you like that , try gooogling liquid gallium cooling.They have actually attempted full liquid metal coolant systems.

Here's the caveat ; That liquid metal , will dissolve nearly all other metals.It's like metal solvent.

I said no to this myself.

;)
 
For the LEDs I'm reusing a couple of strips I already own - LEDs. A few builds ago I went nuts with the UV so I've got a few spare parts left over.

As for the metal pad, I knew about the solvent effect which is why I also over looked the Liquid Pro they offer. But from what I've read the pad acts similar to a solder, crank up the heat to spread then cool. The pad in solid form after the burn-in should be safe compared to the liquid version. I would be using a very sparse amount anyway - just enough to keep the oil from getting between the CPU heat spreader and the heat sink. I'll pick some up and play with it on another old proc I got laying around.
 
IC diamond 7 wasnt around when I did my testing ( 4 years ago now).

It might work it might not.All depends what the binding agent is.
 
You must log in or register to reply here.
Back
Top