3M Flourinert FC-6003

[ronacevedo]

I've fried several XT's due to condensation. Although we've been able to resurrect a couple, this is the end. I have acquired 15 gallons of this stuff and will go completely submerged.

http://products3.3m.com/catalog/us/en001/m...tymaterials_3_0 (http://products3.3m.com/catalog/us/en001/manufacturing_industry/specialty_materials/node_21TTSKXBQWbe/root_GST1T4S9TCgv/vroot_FG8FTD9L7Wge/gvel_RN2BNC4D33gl/command_AbcPageHandler/theme_us_specialtymaterials_3_0)

After that, if I'm still not satisfied - the next step will be a GN2/LN2 vacuum, environmental chamber that will be hermetically sealed and sustain as low as -192C, complete with Watlow F4 controller for any desired temperature. After that, if I'm still not happy - a cryogenic pump that is cooled with liquid helium will get me to 10 kelvin. The same Watlow F4 will be employed for any desired temperature above 10K.

-Ron

[Slickthellama]

just so you know flourinert turns to gel at about -90C.

[ronacevedo]

Its just condensation prevention for my current cascade setup. I will retain the evaporator insulationn for gpu/cpu to concentrate cooling for the dies only. If it becomes a problem, I can always recirculate the fluid through a warmer/chiller.

The other steps will not be using flourinert at all. Does that makes sense?

-Ron

[PyroTeknik]

Originally posted by Slickthellama
just so you know flourinert turns to gel at about -90C.

whered you pick up that tid-bit?

[PyroTeknik]

what hes doing is using evaps as normal and just having flourinert in there to protect against condensation... so the flourinert wont get terribly cold

[chilly1]

I have a 14K cryopump and it will cool only 1 watt at that temp you are gonna need one helava enviro chamber to dissipate the 350 watt load.... I saw one once .5 kw the system had its own 300 Amp panel and it drew 155 amps 3 phase 440 V under full load, it had 12 cryo torr pumps and full vacume. The problem it the vacume basically insulates the devices you will be trying to cool.

[ronacevedo]

That's crazy. But so is Gavula Engineering. They already have a working -192c GN2/LN2 chamber. They're going to micronize it just for my benching rig. They engineer space simulation systems for NASA and other big gov't contractors. They thought my application was a novel idea and very intrigued.

In order to use the cryo pump, they mentioned liquid helium under vacuum.

-Ron

[Gleep]

10 degrees kelvin is way too low for silicon, below a certain temp it's resistance goes up instead of down and loses it's ability to function as a semiconductor. In fact the LN2 guys have had P4 cpus stop working below ~-150C. So lower temp isn't always better.

Using Flourinert to eliminate condensation is a good though, even chilling the flourinert would help as it would keep all the motherboard components temps down. Having the cpu power circuits at -20C I'm sure would help overclocking.

[zabomb4163]

Originally posted by ronacevedo
That's crazy. But so is Gavula Engineering. They already have a working -192c GN2/LN2 chamber. They're going to micronize it just for my benching rig. They engineer space simulation systems for NASA and other big gov't contractors. They thought my application was a novel idea and very intrigued.

In order to use the cryo pump, they mentioned liquid helium under vacuum.

-Ron

despite their resources, i still doubt their ability to bring 100watts to 10K.

[ronacevedo]

Originally posted by zabomb4163
despite their resources, i still doubt their ability to bring 100watts to 10K.

I'm not an ME and cannot refute this, and so it remains to be seen. I'm a long ways from even resorting to that cooling option. Baby steps. We'll see how the flourinert submersion works with the cascade.

I just finished the tank. Here are some pics:

http://sr510.50megs.com/3dmark/flourinert_tank1.JPG

http://sr510.50megs.com/3dmark/flourinert_tank2.JPG

http://sr510.50megs.com/3dmark/flourinert_tank3.JPG

-Ron

[jfr0]

This seems like a good idea.

[DaBit]

I'm using a box filled with dry air to prevent condensation. That's a much easier than fluorinert

[TheDude]

I will be watching with great interest. I have wanted to try this for a long time but have never been able to afford to buy enough flourinert to do it. Do you mind telling how much it set you back?
If I can't play with it, watching you do it is the next best thing! Thanks! :D

[ronacevedo]

I found them listed as surplus goods and got it for $10/lb. It weighs more than water, so each 5 gal. container is 44lbs. Since I bought the lot of 3 containers, I got it for an even $1200.00.

Brand new its like $30/lb.

-Ron

[Dissolved]

Originally posted by ronacevedo
I found them listed as surplus goods and got it for $10/lb. It weighs more than water, so each 5 gal. container is 44lbs. Since I bought the lot of 3 containers, I got it for an even $1200.00.

Brand new its like $30/lb.

-Ron


thats insane :|

[TheDude]

Insane.....maybe, but I've been pricing that stuff for a year and that's a hell of a deal! Good find;)
:toast:

[ronacevedo]

Yeah it was a really good score, cause when the stuff first hit the market - it was $1000/gal!!!

-Ron

[TheDude]

Yes it certainly was, I remember that. ;)

[duKe]

why is this stuff so expensive? isnt it possible to use some sort of oil?

[sandman]

iirc, vegetable oil ha sbeen used before for submersion passive cooling.

but, I'm not 100% sure on that.

If that isn't it, I thinkit might have been mineral oil.

[DaBit]

I have seen many fluids being used, amongst them various alcohols, vegetable oil, mineral oil, DI water, etc. So far all tests were unsuccesful.

I still wonder if submersion in liquid R134a (or another common fairly cheap HFC) would work..
It would require dedicated pumpdown mechanics before opening the box with mainboard in it is possible, but it's price tag is reasonable.

R11 would have been quite an ideal liquid....

[Redwolf]

Some of you have way to much money to burn

Cheeper to just use Dry Nitrogen gas or mineral oil. With some better sealing technics.
Dont see the point in using Floro when you've alrdy got the insulation on the blocks.

Yep R-11 would be great.... Accually its pretty cheep in the States now. Just dont think of using R-123. Its known to cause brain tumors in lab mice.

[ronacevedo]

-92c on the gpu block is very hard to fully insulate, condensation is a terrible nuissance. I've lost patience with coming up with a consistent insulation method, so I said screw it. I give much credit to those who are able to prevent condensation at these temps, they are truly masters at the art. I am not virtuous in this regard.

-Ron

[TheDude]

Ron, you're not saying that you gave up on the submersion project are you? You are talking about a different cooling method?

[Hobocrow]

Originally posted by ronacevedo
-92c on the gpu block is very hard to fully insulate, condensation is a terrible nuissance. I've lost patience with coming up with a consistent insulation method, so I said screw it. I give much credit to those who are able to prevent condensation at these temps, they are truly masters at the art. I am not virtuous in this regard.

-Ron

I totally agree and can only imagine what the problems are @ -92c! I hope in the future to have these same problems. :eek:

I do understand that @ -40c condensation is a major PITB! That's a Looong way from -92c! Don't give up .... take a break! All your work will be worth it! :)

[herefishy]

Originally posted by ronacevedo
-92c on the gpu block is very hard to fully insulate, condensation is a terrible nuissance. I've lost patience with coming up with a consistent insulation method, so I said screw it. I give much credit to those who are able to prevent condensation at these temps, they are truly masters at the art. I am not virtuous in this regard.

-Ron


Hmmm... I read something... that's dangerous huh? :p:


How 'bout employing a flow of say dry nitrogen across the cooled surfaces of the affected components. The dry (gas) would serve to displace the moist air which contains the water that will condense.

I think of it as much like a shield gas in a metal welding application. That's it!!!! Use a shield gas of dry air (or nitrogen), in order to prevent the moist ambient air from contacting the cold surfaces. You could seal your case with an entrance and an exit, and flow dry nitorgen throught the case while you are cooling.

Hence, there would be no moisture to condense on your components.


Cheers!

[FUGGER]

Ron, take my advice on GPU proofing.

You did read the PM I sent right?

[herefishy]

Originally posted by DaBit
I'm using a box filled with dry air to prevent condensation. That's a much easier than fluorinert


OH! :p:

I guess I know where I read it... That's just typical though.. the most profound observations are overlooked or ignored....


Cheers!

[captaincascade]

im sorry ron, condensation is not what killed any of your cards.

[ronacevedo]

maybe for the cards I've left with you - but the ones I have here, I know for a fact that it was condensation. The proof is that the cards would get wet after awhile and then the vertical banding and funky encrypted characters start showing. I would shut down, then dry the card out with compressed air, reinsulate and then be able to resume benching. I've also observed that between bench sessions, if the insulation is not changed - there is residual moisture trapped in the armacell foam. This also has the same effect and causes the shorting. I didn't notice the pattern until it was too late for a few xt's. I don't deny the fact that you may have uncovered something else, and that's good to know. We can learn from these mistakes hopefully in the future, should I regain motivation.

-Ron

[ronacevedo]

Originally posted by FUGGER
Ron, take my advice on GPU proofing.

You did read the PM I sent right?

just got it thanks.

-Ron

[ronacevedo]

Originally posted by herefishy
You could seal your case with an entrance and an exit, and flow dry nitorgen throught the case while you are cooling.

Hence, there would be no moisture to condense on your components.


Cheers!

Yes, pressurized nitrogen chamber is another option and excellent alternative!

-Ron

[ronacevedo]

Originally posted by TheDude
Ron, you're not saying that you gave up on the submersion project are you? You are talking about a different cooling method?

No I'm not, I was referring to "traditional" insulation - as it has royally kicked my ass.

-Ron

[TheDude]

Originally posted by ronacevedo
No I'm not, I was referring to "traditional" insulation - as it has royally kicked my ass.

-Ron

Good, Just wanted to make sure I was following you. DaBit has some pics on his website of his pressurized chamber....nice setup!or actually...he might have a vaccumn...pressure would be better imo.

[ronacevedo]

Here's a video clip of the system running submerged in the 3M FC-6003. Please right click and save to disk. You'll need the divx codec.

FC-6003_test01.avi (http://sr510.50megs.com/3dmark/fc-6003_test01.avi)

My cpu no longer performs very well and does not like the high volts, during testing I managed to corrupt windows prior to shooting the vidclip. I'm sorry that I could not show the system running completely.

-Ron

[TheDude]

I don't know if it's just me but it's very hard to see the video clearly.:(

[ronacevedo]

The native file was 42mb, I guess the compression was too harsh. I can make it a 10mb file and see how that goes.

-Ron

[TheDude]

Thanks, I would like to see a clearer pic if possible. I'm pretty excited about it! :D

[ronacevedo]

FC-6003_test01_10mb.avi (http://sr510.50megs.com/3dmark/fc-6003_test01_10mb.avi)

-Ron

[berkut]

When i had a radeon 32mb ddr (a long time ago) @ sub -40C temps i sprayed the card with poliurethane laquer (paint). (later used it on mobos)

That stuff's very similar to the thing they use to laminate mobo's, it doesent crack and if you apply it correctly no amount of humidity will kill any card unless the AGP port is soaked...

If you guys want i can make a walkthrough with pics featuring my geforce ddr :D

[ronacevedo]

That should be similar to silicone dielectric conformal coating, right? I haven't had any luck with it, perhaps my coats weren't thick enough or not cured well. When I found that stuff, I was sure that would've done the trick. I'm happy to see that alternative works well when properly implemented.

-Ron

[berkut]

Thats much different than silicone. For me it worked MUCH better, the best thing is the fact its poliurethane - practicly indestructeble and forms a "second skin" on the card, after covering the card with it i could run the card totally wet and no problems were encountered, probably the card could be operated under water :D

Worked the same on my mobos

The biggest problem is applying it as if some if it gets on the agp port you can kiss the card goodbye unless you are brave enough to pile it off with a razor knife

[ronacevedo]

at this stage, I think courage is no longer is a factor! You do what you have to when you encounter problems, am I wrong?

That's great berkut.. keep the info flowing, this isn't for everyone and its good to have as many options available to all who seek it.

-Ron

[paul007]

RON did you use this stuff here

http://www.hmcelectronics.com/cgi-bin/scripts/pd/1530-0098_040306/1530-0098_1830/1530-0098

or here?

http://www.hmcelectronics.com/cgi-bin/scripts/pd/1530-0097_040306/1530-0097_1830/1530-0097

[berkut]

You are right but theres a giant difference between you guys and how's it in Poland... If i smoke my card i have to use a GF256DDR :]

What kind of silicone coating did you use, the greasy one, the "window silicone" one or the liquid electric-tape ? (these are the ones i know)...?

[berkut]

Mine is more like the acrylic one

[ronacevedo]

Originally posted by paul007
RON did you use this stuff here

http://www.hmcelectronics.com/cgi-bin/scripts/pd/1530-0098_040306/1530-0098_1830/1530-0098

or here?

http://www.hmcelectronics.com/cgi-bin/scripts/pd/1530-0097_040306/1530-0097_1830/1530-0097

Nope, similar though. Here's some pics of what I used:

http://sr510.50megs.com/3dmark/conformal001.JPG

http://sr510.50megs.com/3dmark/conformal002.JPG

This stuff dries to a clear shiny, hard coat - just like the pcb laminate. I wonder where I went wrong with it.

-Ron

[berkut]

Hmm i had a complettely different type... Came in syrynges

[ronacevedo]

did you have to bake your card for an extended period of time? I was reading up on some other conformals, and they listed that as the proper curing procedure. So, I passed on those types - I'm not very patient.

-Ron

[berkut]

No, it had to cure, i gave her 12h for that, if i remember well it was required.

If you want to bake it preety fast you can use a heat gun, but look out not to fry it and unsolder any elements

[captaincascade]

geez