Anyone ever tried something like this?

[Kevin]

I had this weird idea... On Slot processors, one was able to use 2 - 4 pelts side by side. Now we are limited basically to one w/ these socket chips. How about someone making a block of copper w/ a base about as wide as wide/long as a standard hsf. Have it a bit taller than the surrounding capacitators. Then put a cold plate on top of it that is like flat, but w/ a lot of area. Slap like four pelts on it in a square fashion. Then put a big ole custom H2O block on that. Wouldn't that work? Its all theoretical, and probably more pain than its worth, but I was just thinking about it...

[Cryonosis]

hmmmm. . . NO,NO,NO

just stack the damnded pelts, the cascade effect would be a hell of a lot better than your suggested setup. But you do get some kudos from me for being creative

keep thinking up new stuff,
Cryo

[Kevin]

It would be cool though

Originally posted by Cryonosis:
hmmmm. . . NO,NO,NO

just stack the damnded pelts, the cascade effect would be a hell of a lot better than your suggested setup. But you do get some kudos from me for being creative

keep thinking up new stuff,
Cryo

[Cryonosis]

hmm, after thinking about it for a day, you might get it to work, you would have to use a trapazoidal spacer between the chip and the pelts, the small end matches the chip and the big side matches the pelts, you would also need massive insulation to reduce wasted energy, but it might actually work. It would be more reliable than cascading but would only cool to the same level as one of them. Basically all of them have a lowest possible temp, since the cold side of all of them would be the same temp the chip would only get that cool, but then on the reverside, much more heat could be pulled off.

in a nutshell I will just compare the two methods:
cascading - colder cold side but with hotter hot side
Kev's idea - greatly increased heat removal but will only get so cold.

hmm, makes me wonder now, someone will have to try this out, but making a trapazoidal spacer might be a tad difficul, especially making one that will fit.

well, that's my days worth of thought's on this, damn you for making me think about this all day

with care,
Cryo

[This message has been edited by Cryonosis (edited 11-28-2000).]

[This message has been edited by Cryonosis (edited 11-28-2000).]

[Kevin]

You should try it!

[Cryonosis]

give me the parts and some time and I will .

[JC]

I think you will have a hard time insulating all those peltier units. It would be COOL though. Cool in both ways...

[Joe]

Damn JC's beat'n me to the punch twice now

Yeh insulaing those would be... fun at best.

Unfortunately a cascade would work easiest.

Well thinking about it... buy one BIG ASS 180watt + Mutha, strap that to the cpu, put cooling block on it, run the coolant to a pelt/heatexchanger sandwich with like 6 120watt pelts, then have the hot side cooled with.. a small car radiator ( cause.. 6x120watt+180watt + a 80watt CPU = just a hair cooler then the surface of the sun ).

if you do that I am betting not only will you be pulling a HUGE amount of power, but that cpu would be well below -100DegF
( You'd need flourintert, or industrial anti-freeze, and REAL good hoses, and insuilation)

------------------
C-ya
Joe - Owner/Editor www.ProCooling.com
Where the Completely Addicted Come to Cool Off
Somebody set up us the bomb.

[This message has been edited by Joe (edited 11-30-2000).]

[syran]

SOunds like you need a combo of both, a nice cascade, then kev's idea for getting rid of the heat caused by the cascaded pelts... Any way you put it, it would be sweet.

[Freakyfrank]

well of what i read.. maybe this will help you out :

socket proc = about 50W (thats a OC'D PIII)
TEC's = say 156W -> 117W (cause of 75% = maxW rule)
MaxTdelta = 70C (depends on type of TEC)

2 TEC's CASCADED :

W endW C
PIII 50W
TEC1 117W 167W 9C
TEC2 117W 49C
ambient 20C
Tdelta PIII->TEC1 = 1 - (50/117) * 70C = 40C
Tdelta PIII->TEC1 = 1 - (167/117) * 70C =-29C

the second TEC only adds temp to the cpu-die, however if you place 2 TEC's parallel, u can add the wattages and have a stronger 1 pelt..


2 TEC'S PARALLEL :

W endW C
PIII 50W
TEC1/2 234W (284W) -25C
ambient 20C
Tdelta PIII->TEC1 = 1 - (50/234) * 70C = 55C

u see this has far better results than de cascade one..

better have this setup:
1 TEC CASCADED WITH 4 TECS PARALLEL:

W endW C
PIII 50W
TEC1 117W 167W -65C
TEC2/3/4/5 468W -25C
ambient 20C
Tdelta PIII->TEC1 = 1 - (50/117) * 70C = 40C
Tdelta PIII->TEC1 = 1 - (167/468) * 70C = 45C

-65C is a very CUWL temp .. but u have to keep in mind that a TEC becomes less effective when dropping to the -120C barrier.
and 156 TEC's arent that cheap.. say about 5x$30 = $150..

I think the best setup is a phase-changer on top of 2 or 3 156W TEC's .. something like -70C to -90C , depending on your phase-changer and compressor, etc


------------------
OC'ing is my middlename

[This message has been edited by Freakyfrank (edited 01-08-2001).]

[mfpmax]

Don't know if you've seen it, but the Japanese company that produces the Kendo Raditor Heatsink produces a watercooling kit that has the trapzoid spacer for the CPU.

------------------
Watercooled K6-2 450 at 600

[Joe]

Yeh I have seen that around a long time ago, nice design

------------------
C-ya
Joe - Owner/Editor
www.ProCooling.com
Where the Completely Addicted Come to Cool Off

Somebody set up us the bomb.

[eTiMaGo]

hmm stacking pelts = not good idea. from 2 sources, overclockers.com's experimentation and tedist.com. the best way would be to get a preassembled multistage pelt, as there are many variables that need to be taken into account for max. efficiency.