boiled peltier sandwich...

nhusby

Limp Gawd
Joined
May 1, 2006
Messages
303
ok... so you guys are probubly sick of my stupid idea's... but I have alot of time at work to sit around thinking.

how efficient are peltiers sandwiched together?
also, how hot can most peltiers safely get?

here is what I am thinking... would it be possible to passively cool some peltiers by boiling water?

it'd be quiet, and it'd look cool as hell in my case... (steam exhausted of cource)
 
hmmmmmmmmmmmmmm............................. sounds like it might work, but you would use alot of electricy in the process,at least 400 watts for the peltlers. if i were you i would just do it to a test rig for kicks and for the look on peoples face when they see it.
 
yeah... I'm thinking it would just be for cool factor... maybe when I build my next rig I'll do it to my old one... I'll tell people; "its so old, its actually steam powered" LoL
 
hmm indeed .

Very interesting idea....I might steal it. :p (j/k)

Peltiers are inefficient at the best of times , so Stacked units are no better , and the watt load is too small.....

The typical max operating temp is 125 C , so that should be no problem.

Some things to consider :

-A single ~430 watt peltier takes , at peak wattage ,over 800 watts of electricity (most people run em with 24 volt psu's , that drops it to 600ish watts) , and unlike a computer power supply that may only actually draw 100-300 watts even at peak loads , this'll be 600 -Watts fulltime...I've designed some multi-peltier computers , and when you see your A.C. loads go into the multikilowatt range , thats bad.hehe.

-If you want to make it a visual piece , you need a water resevoir that has ; a) a good heat conduction to the water (a 'hotplate') and , b) a portion thats clear so you can see the water boiling.This implies a resevoir thats constructed out of two materials : metal and glass/plastic....and with the thermal varience ( ambient to ~100 C ) , it'll hafta be made correctly.

-If you intend to cool more than the processor in this manner , you have to contend with the fact that the processor and the cards are at 90 degrees to each other.This probably involves either a water loop or heatpipes.
------------------

some stuff to think about.

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

I am unsure of the dissapation of a 'boiling water' resevoir , but it's certainly NOT a stupid idea...Kudos Sir....

I've got some pelts lying around . I might test this theory out if I can get the time.I'll let ya know. :D
 
nobi125 said:
Not sure why I was expecting to see a sandwich heated with a pelt.

hmm... I wonder how many more MHz I can sweeze out of these celerons with some grilled cheese sanwiches???

and BrainEater thanks for the input... I was planning on using a water cooling loop... maybe try and submerge a water block with the peltiers attached to it? I would have to insulate the electric parts of the peltiers... otherwise I would probubly end up constructing a plexi-glass res on top of the peltier with a heatsink and displaying it in the top 2 5.25" drive bays with an exhaust duct comming out the top

It was more or less just an idea, if you want to try it, I would love to hear about it. I dont have the $$$ to get power supplies and such... Maybe if I came accross some for cheap...
 
and trying to pull that much heat off with water cooling would take min. to 120.3's and loud a$$ fans. If you have the parts go for it. If not look into a chiller or direct die for it.
 
cre8chaos said:
and trying to pull that much heat off with water cooling would take min. to 120.3's and loud a$$ fans. If you have the parts go for it. If not look into a chiller or direct die for it.

actually it wouldnt require any fans... the act of the water boiling would cool the hot side of the peltier. Depending on the efficiency of the heatsink (in the water) the hot side shouldnt get much over 100*C and I dont amagine it would get above 120*C.

It wouldnt be a practical system because of how much power it would take to make an 80-100*C difference in temperatures with the peltiers... maybe 3x260W peltiers sandwiched together?

The beuty of this setup would be an infinite cooling capacity (so long as you keep up the water level), and it would have major cool factor...
 
nhusby said:
The beuty of this setup would be an infinite cooling capacity (so long as you keep up the water level), and it would have major cool factor...
It would turn your computer room into a sauna, too! That way you could achieve ultimate relaxation while you got your game on.
 
hehe... :D

Yea you would definitly need a method to continously add water to the setup.

At this point I still say you don't want stacked pelts , but im not sure.I'll work on the math a bit and try some experiments.

I went thru my parts bins and actually found the correct materials for a decent 2x stack , so I can try that if a single 226 watt fails... :

pelt70226.jpg


A 70 watt peltier and a 226 watt peltier......

peltstack.jpg


hehe

All I need is the time to setup a test rig.......don't have it right now ,as I'm preparing for a LAN right now..... , but I might find some beerspiration in the next few days. :D

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

nhusby said:
maybe 3x260W peltiers sandwiched together?

It does'nt work that way.

The first peltier in the stack (the one that contacts the cpu) , needs to be capable of dissapating the heat from it , in my test rig the max would be 70 watts.The next peltier needs to be able to dissapte that 70 watts + the heat the heat from the first peltier's work.....in this case about another 110 watts......so the next peltier in the stack needs to dissapate at least 200 watts....if you wanted to make a 3-stack , the next pelt would need to dissapate ~650 watts.....
 
as I am sure is obvius to you my knowledge of peltiers is limited...

but as far as I know most pelts max out at about 65*c difference between the hot and cold side, and the colder you can get the hot side, the cold side will be that much colder....

I made a simple assumption that each pelt could maintaine an average 50*c difference. I could be very wrong, and you would more likely know that than I would.

let me know how the experiment goes...
 
I did a little bit of digging. It appears that the only way to get a difference over 110*C is to use a three stage peltier. Maybe these are low quality pelts, and a higher quality pelt could do with a 2 stage, but I've never heard of a single pelt going over 60-80*C...

here I found a 3 stage with a 111*C difference (If I am reading the data correctly)
three stage peltier
 
Yes, and you would, as a previous poster stated, keep your rig at about 30C minimum.

Have fun making steam come out of your computer, and condense, and leak into, and melt, your computer :) :D
 
mwarps said:
Yes, and you would, as a previous poster stated, keep your rig at about 30C minimum.

Have fun making steam come out of your computer, and condense, and leak into, and melt, your computer :) :D


it wouldnt be my first fried electronic... but it'd be cool... call it anything but efficient...
 
The thing you all seem to forget....peltiers are only rated to pump so many watts at a certian temerature differential. Mine are advertised to drop the temperature up to 40F lower from ambient air temperatures. So if each peltier can only drop it "up to" 40F each, that is nearly 6 peltiers to get 0F under a very small load. Using a compressor and refrigerant (phase-change sysem), you could do this idea, though it still wouldnt work that great.

Each peltier has to be rated to pump the first peltiers input, plus heat generated due to the fact that peltiers arnt all that efficent, and more heat from losses in insulation. By the time you get done, might as well give up and use a good heatsink or W/C setup.
 
I dont follow... 40*6 = 240...

unless you are saying the same thing brain eater was... yes, each pelt in a series (or sandwich) must move the heat of the one before it. so if each pelt (not that it is possible) could make a 40 degree difference it would only reqire 3...

I'm not completely ignorant... I wouldnt expect that perfect of a result... I'm just kicking around the idea for a mod...

I did find an example of a 3 stage peltier that does 110*C. I do not know its power consumption (probubly insane) or its thermal capacity...
 
nhusby said:
I dont follow... 40*6 = 240...

unless you are saying the same thing brain eater was... yes, each pelt in a series (or sandwich) must move the heat of the one before it. so if each pelt (not that it is possible) could make a 40 degree difference it would only reqire 3...
Last time I checked, water boils at 210, correct? 3x40=120. 210-120=90.

90F... Slightly warm for computer cooling in my opinion.


nhusby said:
I'm not completely ignorant... I wouldnt expect that perfect of a result... I'm just kicking around the idea for a mod...

I did find an example of a 3 stage peltier that does 110*C. I do not know its power consumption (probubly insane) or its thermal capacity...

It could be rated for very little, 110C is a very large temerature drop even for 3 peltiers.
 
I believe it would work, but it would be rediculiously innefficient. If I understand correctly, you are chilling water and that is your only goal. You have to take the wattage from the CPUs out of the circulating water into the "boiler". Peltiers work @ a coefficient of about .6. A triple peltier "sandwich" would proably be about .4 - about 4 watts transported to the hot side for every 10 watts put in. Assume you have two 60 watt CPUs. You have to use 300 watts to cool the water in the loop.


Problems:
Inefficiency of heat transfer from water to "boiler" through submerged passive heatsink.

The peltier may go above "rated" temperatures because the passive heatsink might not be able to dissipate that much heat into water. I'd suggest 2 peltier "sandwiches".
 
bob said:
Last time I checked, water boils at 210, correct? 3x40=120. 210-120=90.

90F... Slightly warm for computer cooling in my opinion.

It could be rated for very little, 110C is a very large temerature drop even for 3 peltiers.

sorry.. I didnt notice that you were talking about farenheit...



Silvermirage: thanks for the input... going on a figure of .6 to 10 watts it would take over 325 watts. ((70w/.6)/.6)/.6 = 324w

but that doesnt sound right to me... each peltier must dissipate the heat from the one before it... my calculation is flawed...

first pelt 70w/.6 = 116w
second 116/.6 = 194.5w
third 194.5/.6 = 324w

add them together... 635 watts. Thats alot of power...

as far as the issue of heat dissipation into the water, I am hoping that the water would boil fast enough to cool the pelts to less than 120*C and preferably less than 110*c.

mwarps may well be correct with his prediction of 30*C depending on how big of a difference each pelt will make...
 
Couldn't you just set up a closed loop H2O cooling system, with an enormous amount of water in it that could keep the whole thing below boiling temperature? If it was in a big ass container with alot of surface area you might be able to keep it a good bit cooler than boiling. Or is this somehow against the whole point of the thread?
 
Mark305TBI said:
Couldn't you just set up a closed loop H2O cooling system, with an enormous amount of water in it that could keep the whole thing below boiling temperature? If it was in a big ass container with alot of surface area you might be able to keep it a good bit cooler than boiling. Or is this somehow against the whole point of the thread?
And I mean without a pump or anything...

There was an article in the [H] News about a guy that had a regular watercooling setup like that. No pumps or fans. Just really big tubing. I can't find the article now, though.
 
Mark305TBI said:
Couldn't you just set up a closed loop H2O cooling system, with an enormous amount of water in it that could keep the whole thing below boiling temperature? If it was in a big ass container with alot of surface area you might be able to keep it a good bit cooler than boiling. Or is this somehow against the whole point of the thread?

I could hook up a radiator for normal oporation, just let it steam for show...

EDIT: not that it would normally be oporated LoL
 
mwarps said:
Yes, and you would, as a previous poster stated, keep your rig at about 30C minimum.

Have fun making steam come out of your computer, and condense, and leak into, and melt, your computer :) :D


Or buy some silicon waterproofing spray. I used that when i built a watercooling rig inside a fridge. Some condensation built up around the waterblock and on the chipsets, but i just let it build up, it never hurt anything.

My next idea, Is to build it inside a freezer, and instead of water..... some 190 proof alcohol so it doesnt freeze =p
 
I'm going to use beer... miller highlife...

why? because... it will be rediculus, and thats the idea...
 
How about setting up a phase-cooled condensor right above the boiler? Then you could catch the water and send it through again, and almost never have to fill it up.
 
HOCP4ME said:
How about setting up a phase-cooled condensor right above the boiler? Then you could catch the water and send it through again, and almost never have to fill it up.

good idea, but then I minaswell use it to cool the boiler directly...
 
Why not just use a bong? It'll evapourate off the water, it'll cool down just about any load, it's cheap, and you get the pleasure of using the word "bong" while talking about your computer. Just stick an aquarium UV light in it and you won't need to worry about Legionnaire's disease.
 
THE JEW (RaVeN) said:
Why not just use a bong? It'll evapourate off the water, it'll cool down just about any load, it's cheap, and you get the pleasure of using the word "bong" while talking about your computer. Just stick an aquarium UV light in it and you won't need to worry about Legionnaire's disease.

no cool factor points for you!
 
Back
Top