Project thinktank

braineater · Post by **braineater** » Wed Jul 21, 2004 4:50 pm

Hi !

I've just spent 6 months working on my latest casemod.
(and just bought my 2nd mx2 series display as a controller)

Here's some info on my mod .It is an Intel p3 800 EB proc on an Asus p3v4x mobo.
w/
768 MB cas2 pc-133
Elsa Gladiac geforce2 gts pro 64MB
Intel pro1000MT gigabit nic
300 watt psu

All of this is submerged in a biodegradeable ,non-toxic transformer insulating fluid , called Midel 7131.The fluid is circulated with a submersable pump.

It is capable of operating completely passively and maintaining a temp of 47 degrees C in the coolant and 54 degrees on the cpu.

The back plate of the tank is a 450 watt peltier assy. attached to a 500 sq inch pure copper heatsink (17 lbs),designed for convective cooling.

Now , I am just starting testing with the pelts , but the initial goal is to maintain ambient temp inside the tank (i.e. 25 C) regardless of thermal load (fully overclocked)

The whole idea of the mod is to provide completely uniform cooling to EVERY component on the computer , for an overclocked machine , as close to passively as possible (no fans)
Technically , the peltiers on the backplane could freeze the liquid coolant solid (-50) if they were water cooled.So cooling isnt really an issue......it's how quiet i can make it.

The active cooling can be fully controlled via the matrix orbital mx2 lcd , by regulating the duty cycle of the peltiers.....

I don't know the minimum temps yet , I'm still doing testing/tweaking of the entire system...I can tell you that under maximum load , the machine draws just under 1 Kw of power.......

Now , I do not have a host for pictures yet (real soon) but I wanted to share this mod with the Matrix Orbital staff , as it is one of your displays that controls most of it.

Can I email someone a couple of pics ?

Johnny · Post by **Johnny** » Thu Jul 22, 2004 8:48 am

yeah go for it ill host them for you!

braineater · Post by **braineater** » Sat Jul 24, 2004 5:37 pm

Well I emailed ya Johnny......

S'ok tho, It would appear I can post this link :

MTB gallery

lemme know what y'all think of it.

Stinger · Post by **Stinger** » Sun Jul 25, 2004 3:23 am

You have got to be kidding me!!!!

That, my friend, IS AWESOME!!!!!!!!!!!!

mtbeercan · Post by **mtbeercan** » Sun Jul 25, 2004 1:44 pm

Well thought out and very impressive!

PS, a note about controlling peltiers... You want to use pwm for controling and not switching them off and on.

Alexander · Post by **Alexander** » Sun Jul 25, 2004 3:38 pm

Thats liquid in there right?

Tom · Post by **Tom** » Sun Jul 25, 2004 4:55 pm

Very Impressive!!!

braineater · Post by **braineater** » Sun Jul 25, 2004 5:53 pm

Yes , that is liquid in there.

PWM eh ?

I had intended to use a 12v gpo to actuate a relay , which in turn would use the atx power on (green wire to ground)......and just use the strobe function.......ie....10 secs on....50 off etc....

I will reconsider.

Johnny · Post by **Johnny** » Mon Jul 26, 2004 5:18 am

OMG!

I heard about peeps using submersed systems for testing.... but not running one as their system! amazing!

All you need are a couple of fish! lol

Isn't that coolant about $400 a litre?

braineater · Post by **braineater** » Mon Jul 26, 2004 7:27 am

hehe

I'd hafta say , about 50% of the people that see this mod , say 'all you need is some fish'

Actually , the coolant you are thinking of Johnny is 3M Flourinert.It's 400$ US a gallon.
This coolant : Midel 7131 , was 400$ CDN for 25L.....(mostly shipping from the UK)

While fluorinert has better thermal properties , Midel is superior in others.
-Midel is both non-toxic and biodegradable.
-While Midel will burn above 300c.....it just burns....Fluorinert decomposes releasing a gas just like phosgene (thats bad)

Paradigm · Post by **Paradigm** » Mon Jul 26, 2004 11:09 am

WOW

I've got to agree with everybody else, this case is way too cool. Amazing idea using just peltiers and a heat sink to cool the fluid. Way to many questions to post on the forurms all at once, so here are my biggest ones:

Can't tell by the pics, but I assume that you have a heatsink still on the cpu?

Do you have any kind of agitator (besides the pump) around the cpu to move the coolant around?

What kind of temp differences to you get from idle to full load?

I haven't seen stuff like this since since working with massive RF transmitters for a previous job. The amplifiers on those were power hogs and required massive cooling. Who would of thought that that someone would have done something similar for a PC case, and made it look soooooo good.

More detailed pictures of your setup would be greatly appreciated. I'm sure it would be possible talk Henry into having some pix on the lcdforums.com server.

Amazing job!

mtbeercan · Post by **mtbeercan** » Mon Jul 26, 2004 3:26 pm

Here is a site that may help with questions and considerations to peltiers...
http://www.tellurex.com/12most.html
Anyways peltiers have a considerably shorter life span when switching them on and off due to thermal expansion and contraction. PWM gets you past this problem. I'm currently experimenting with peltiers using a matrix display and LCDC.

braineater · Post by **braineater** » Mon Jul 26, 2004 10:30 pm

First off , I just wanna say thx for the positive feedback.Thank you.

I want to extend an Invitation to the Matrix Orbital staff.We're in Calgary....u guys wanna see it in person....C'mon over......PM me , we'll get in touch and arrange something.

-----------------------

-Yes , there is still a heatsink on the cpu.....gpu as well.....just no fans.There is no other agitation , other than the jets for cpu/gpu.circulation provided by the pump.

-As far as temps go......I'm still testing all of this.
I can tell you this : I cannot operate it passively with the current heatsink (17 lbs pure copper ain't enuff .... wtf ? )....I have fans on it now.
I can now maintain 50 C 'easy peasy' on the pelt hotside (heatsink) , with one 226 W Pelt running full on.......but this is just slightly not enuff....it appears it dissapates 250 W mebbie.......really slow temp rise.Testing with both pelts begins when I can afford another psu. (or find a spare car battery)

I have a 240 picture worklog....but it's film (ugh) and it's not vetted/scanned yet.

mtbeercan wrote:Here is a site that may help with questions and considerations to peltiers...
http://www.tellurex.com/12most.html
Anyways peltiers have a considerably shorter life span when switching them on and off due to thermal expansion and contraction. PWM gets you past this problem. I'm currently experimenting with peltiers using a matrix display and LCDC.

Cool....I've been to that site and never seen that ....lol.
I had considered thermal exp/contr. tho......It was totally one of my concerns.....
I am very interested about your experiment....let me know Please about your current data/etc.

cheers !

mtbeercan · Post by **mtbeercan** » Wed Jul 28, 2004 1:31 pm

braineater, thanks for the interest in my work. I'm happy to share info on what I've learned playing with peltiers. I built a test bed using old processor heat sinks that I had laying around and tied it to a matrix LK-204-24 using temp sensors and a keypad which allows increase / decrease of variables "on the fly" under LCDC. I have set up cooling fans for the sinks that I change RPMs on to see the different effects on the heating / cooling properties. I havn't quite finished mapping this out yet but I have found out some interesting info. If I cool the heat side too much I loose out on the cold side. If I don't cool the hot side enough I still loose on the cold side. All my tests are done with air, not liquid so your results will be different. If I can keep the hot side air temp around 88* to 89* F, the coolside will get down into the lower 20's. I use a 12 volt lawn tractor battery and a 2/10 switchable charger. If I increase the amps with the charger to 10 amps the cool side will fall into the teens, I've seen as low as 10*F while the hot side runs aroun 89*F. I've set up LCDC to give me 5% change on fan rpm with the keypad so I'm able to watch the different effects on the peltier when changing air flows. It seems to me to be a balancing act. Too much heatsink can give undesired effect just as much as too little heatsink. Available watts or amps to the pelt also will alter the cooling ability. It seems that pelts are very power hungry. It also seems that they require a little coaxing to get the max differential out of them through cooling changes on the hot side. Perhaps if you used a smaller heatsink you would achieve a better temp differential across the chip? Also how many peltiers are you using and how are they wired together (series / parallal)? I'm a nubee at this and in the learning stage but I do see a very bright future for the development of peltiers and the many uses that they can serve. I think your computer design and application are great. I mostly use PIC micros for my work but LCDC allows a simple easy to use interface and I couldn't resist trying it out. I don't know yet about PWM with LCDC, I might just port all this over to a PIC after I work a few details out. If you would like to compare notes further please pm me and I can send pics and so forth.

braineater · Post by **braineater** » Thu Jul 29, 2004 9:18 pm

Imma PM ya after saturday mtbeercan ! (u need a full beercan !)...gotta do the full on test.Just got the power for a full test....so I'll let ya know.
===================================================
Tom,Paradigm ,......I'm havin beers at my place sat night......Lemme know if ya wanna come see the rig.I'll be doing the first 2 pelt test.