PDA

View Full Version : possibly a GREAT cooling idea


JFettig
02-11-02, 08:50 PM
this could be a GREAT cooling idea! if someone wants to try it, i would but i dont have the cash and resources.....

basicly water goes from pump/rez to block that cools the pelt to rad that cools the water that was warmed up by pelt,
then to the other block being cooled by the pelt, then to the cpu and back!
it sounds like a great idea, basicly its a inline waterchilling system without need to worry about motherboard insulating and whatnot.... may be a great idea, and may not work...but its worth giving it a shot

some of you have tried a pelt in a rez and had some luck but not much but this may work great!

JoeCrappa
02-11-02, 09:02 PM
uhh...thats called a water chiller. many people have tried that and FAILED MISERABLY. seriously i wouldn't try that crap. its a waste of money. they have an article at procooling.com that shows how inexective it is.

here's the article: http://www.procooling.com/articles/html/kev_explores_pelt_water_chille.shtml

gerby123
02-11-02, 09:03 PM
good idea. that fact that you can use the peltier without a completely separate cooling system (pump,rez,raditor ect.) is nice. someone who has 3 blocks should try this

JFettig
02-11-02, 09:47 PM
im sure you tried the rezpelt thing...
this is diffrent its basicly cooling a second cpu with or something....

JFettig
02-11-02, 09:56 PM
oh i see what that means, but another thing that mite effect in that is if it was like 1/8in passages and a LOT of them i think that the water would chill decently?

JoeCrappa
02-11-02, 09:56 PM
oh i see what you're getting at. you'd have to have one hell of a radiator to cool off the water from the peltier. i doubt that the water would cool enough to even make a difference. it could even yield you with higher temps.

gerby123
02-12-02, 05:27 PM
wait, there are two things that would make this not work
1. the fact that the pelt/block combo doesnt cool the water much.
2. after the CPU block, the water is hot and goes to the rez hot. it sits in the reservior hot, then is sent to the pelt/block combo, expecting to cool it. the water is already hot/warm it wouldn't be able to absorb much more heat, therefore hindering the pelt's cooling abilities

omaticrail
02-12-02, 05:57 PM
I believe the concept can work, although not necessarily as described above. The original work, referenced in the link, used only water, and due to the block design, had very little time to chill. I think that's called "wet time" or something, in that radiator article (very nice read).

Anyway, heat transfer rate improves as the temperature delta increases. That is, I believe, why peltiers are so effective in the first place.

Has anyone tried a peltier in the standard configuration, but using a large rad/water setup to chill the hot side of the peltier?

dream caster
02-12-02, 08:38 PM
Having been discussed at large and many times said this cannot work
it HAS WORKED,
so it CAN work.
Unluckily the complete story I know was posted in the late ForumOC. The owner of the rig is Below Ambient now moderator in www.speedforums.com , the maker is 72510 @[H]ardforum now

A CHILLER THAT WORKS

from 72510
the blocks are 1" thick with 3/4" deep slots which leaves a 1/4" plate to transfer the heat, from what I've read a 1/4" plate is the best thickness
it is like a. but 13 of them connecting the inlet and outlet, massive surface area but little restriction, the slots are 0.082" wide 0.750" deep and 6.000" long with 0.082" wide fins in between
the inlet and outlet?, they are both 7/16" drilled to the bottom slot

from 72510
damn, to many questions Archmage, lol, basically I take the solid block and drill 2 holes down to within 0.500" from the bottom, then tap them for fittings, next mount them in a mill and cut the slots with a circular saw blade thats 0.0625" wide, it takes about 5 minutes per pass after the slots are cut then the end caps and top plate is welded on, I use a special method of tig welding to insure massive penetration (wouldn't want a leak), then its back to the mill for drilling of the holes for clamping and counterboring of the Socket head cap screws, then drill and tapping for the heli-coils, then I face mill the bottoms and give them a good lapping, the waterblock was a maze 2.1 I modified for proper pelt clamping, he's not using a pelt on the cpu though, any more questions?

from Below Ambient
first of all im not using air, im using water... second of all, if you use a smaller pump and not a 700 gph behemoth you will get results!! I have done this!! lol
that one article is by a place that condones a bunch of crappy waterblock dealers !! The objective is not to cool the water to minus whatever..
it is to cool the water that goes INTO the peltier thats on the CPU.. why anyone would want frozen water is beyond me... I know for sure nylon cant withstand too much expansion and the pumps will give out as soon as water turns to slush....
this is what im doing...
from chiller (hot) ->to pump (pullling)-> to bong->from bong-> to pump (pulling)-> to chiller (cold) -> to waterblock-> from waterblock -> to chiller (hot)....

they were really good temps not below freezing though, using very nicely done aluminum water blocks and three pelts.(in the pic there is one pelt but they did put three when they installed it.

dream caster
02-12-02, 08:55 PM
and a picture from the guts of block

JFettig
02-12-02, 09:31 PM
wow thats really cool! now thats what im talkin about!!!!! totally what im talkin about!!!

but one thing is that i would think it would be coolest to have it small so that its just tiny if at all possible and without any rad or anything so that it goes pump>chiller>cpu>grfx>nb>pump
now that would be sweet!

i also built a waterblock for my voodoo3 and am gonna put that on when i build a bong and make it all bong cooled wich will be cool

dream caster
02-13-02, 08:47 AM
Pelts need some kind of cooling so you keep the hot side near ambient temp and not way up, and you have to dissipate in that cooling not only the CPU power but also pelt power.

The quotes and pictures come from ForumOC and i didnīt save more data. 72510 told that he thought the problem with previous instances of this kind of setup was improper pelt mounting, specifically not enough pressure on them. Notice how strong those waterblocks are and how many bolts is he using.

Owenator
02-13-02, 09:59 AM
I built (and am using currently) an inline water chiller with a 122W pelt and a HSF cooled by a 92mm 56CFM fan. I lowered my load temps by maybe 2C but didn't seem to effect my idle temps at all, huh. I attached a pic. I can't cool the pelt hot side as effectively with air so I am considering either adding a pelt water cooling loop or direct cooling my cpu and nb with watercooled pelts.

f155mph
02-13-02, 10:08 AM
Seem like an good idea. But what will happen if the pelt die?

dream caster
02-13-02, 02:35 PM
Owenator:
Really i do not manage pelt specs cause i've never been really interested in using them, but precisely, how do you clamp the heatsink on your TEC, and wich ones do you have? I remember you have to use about 150 pounds force between pelt and ,in this case, the heatsink.

I understand that the power disspation is so high that most of times you have to use water cooling if you use a pelt; and if not the pelt but the pump dies with a watercooled pelt on your CPU, . . . BOOM! Water boils, probably your block breaks and you fry your processor.(you have to cope with processor plus tec heat).

Owenator
02-14-02, 08:38 AM
dream caster ,

I bolted a socket A water block to a large heatsink. the waterblock had holes for the four bolt mounting method and I drilled holes in the heatsink. I tightened the bolts very tightly to get a high contact pressure.

O

Originally posted by dream caster
Owenator:
Really i do not manage pelt specs cause i've never been really interested in using them, but precisely, how do you clamp the heatsink on your TEC, and wich ones do you have? I remember you have to use about 150 pounds force between pelt and ,in this case, the heatsink.

I understand that the power disspation is so high that most of times you have to use water cooling if you use a pelt; and if not the pelt but the pump dies with a watercooled pelt on your CPU, . . . BOOM! Water boils, probably your block breaks and you fry your processor.(you have to cope with processor plus tec heat).

takiwa
02-14-02, 09:53 AM
this could be a GREAT cooling idea! if someone wants to try it, i would but i dont have the cash and resources.....

basicly water goes from pump/rez to block that cools the pelt to rad that cools the water that was warmed up by pelt,
then to the other block being cooled by the pelt, then to the cpu and back!
it sounds like a great idea, basicly its a inline waterchilling system without need to worry about motherboard insulating and whatnot.... may be a great idea, and may not work...but its worth giving it a shot

some of you have tried a pelt in a rez and had some luck but not much but this may work great!

Not that it hasn't been answered already, but a correct diagram would probably look like this for what you want to accomplish (this way the water is cooled after the cpu and before the res, then after the res and before the cpu):

dream caster
02-15-02, 09:13 AM
look at this diagram

Owenator
02-15-02, 10:08 AM
Probably the best way to do this would be with two loops/pumps. That way there is no hot water in the cold side of the pelt. I don't know how much heat a bong can disapate but to do it combined it would probably need to be 200 Watts plus. What about like the pic below? I suppose you could substitute a radiator if you like instead of the bong.

dream caster
02-16-02, 09:42 AM
All bongs are not created equal, but most of them may dissipate around 600-800W (two or three pelts plus one CPU)
I don't know if it is better to have a cold and separated hot loop, or just one loop.
I figure out that probably it does not matter too much,(but two loops =two pumps, some more pump heat)
The bong will receive about the same heat load in both cases. It has to dispose all heat received by water, so the temperature of water exiting it will be the same (equal heat load equal ambient temp and humidity);
and what pelt produces is a temp diff and since hot water output will have the same temp I think all water temps should be equal.

Owenator
02-16-02, 07:46 PM
dream caster,

Dissapates 600-800W no wonder bongs are so popular! I may have to make one of those! I'll have to figure out how much moisture it adds to the air because my computer room is pretty small and I don't really want to make a sauna;) Thanks!

O

mustangman
02-17-02, 03:29 AM
Originally posted by Owenator
Probably the best way to do this would be with two loops/pumps. That way there is no hot water in the cold side of the pelt. I don't know how much heat a bong can disapate but to do it combined it would probably need to be 200 Watts plus. What about like the pic below? I suppose you could substitute a radiator if you like instead of the bong.
I am making a setup now like the one you have a diagram of. I am just waiting on the pelts to show up! :( I have been kicking around this idea for some time now but originally I was going to use a heater core with fan and home made shroud to cool the pelt water but after making a bong I decided that the bong would be a better cooling method for the pelt water. I will post results once I get everything working. If the pelts ever show up!