waterblock + aircooled Chiller - Radiator=Good idea?

[the_master]

I have a rather small miditower chassi and I found that a radiator mounted inside the case would not give suffient airflow to cool the water from the watertank. I was thinking of having a plasic watertank with a square hole in it, in that hole a thought of putting two copper coldplates in direct contact with the water, and on the inside of the tank an old p90 HS. On the outside I'm thinking of using a cheap-ass 40-60w peltier from my local radioshack plus my present HSF (Taisol CGK-something-copperbase) on top of that to cool it down. That, I figured would act as a good chiller for the water.. The Watertank is positioned under the 3.5"internal slots where the HDD's are and the HSF is supposed to stick out of the front of the case - for better cooling.
I know that my Athlon XP 1500+ generates about 60w heat and my existing HSF is barely managing to cool it. Do you think that my HSF is enough to cool down the peltier, cus I don't know how much heat in watts that a peltier generates.<---No experience with peltiers!.

/The Master

[Monaco]

I've seen a few attepmts at what you describe, but it never really works. Turns out peltiers aren't too good at what you describe- just too inefficient to work. You'd need a bunch of peltiers to get any worthwhile effect, and a bunch of good heatsinks to remove all the waste heat, and by that time you could've bought a better radiator in the first place.

Try sending a PM to fuzzba11; I know he tried exactly the same thing you are thinking of. He can tell you his results-

[the_master]

The HSF in this case is mounted outside the chassi ,and it only uses cool, ambient air. And the water circulation in this case is very low about 0.5Litres a minute. So if I have a 1-2Litre tank and have a circulation of 0.5litres per minute, I think the peltier would have sufficient time to cool the water. - Depending on the effect of the peltier. Don't ya think so?
How much heat does a 40 or 60w peltier generate?
/The Master

[The Overclocker]

you will need to keep a small cold bit of the pelt out of the water in order for the water to cool down properly otherwise it will be ineffective

[ButcherUK]

Originally posted by the overclocker
you will need to keep a small cold bit of the pelt out of the water in order for the water to cool down properly otherwise it will be ineffective

Why would that help, surely you'd just be wasting some of the peltier cooling the air instead of the water, the block on the pelt might be cooler but the water wouldn't (overall).

[Monaco]

So if I have a 1-2Litre tank and have a circulation of 0.5litres per minute, I think the peltier would have sufficient time to cool the water. - Depending on the effect of the peltier. Don't ya think so?

Nope. It's all about efficiency. Pelts work best when a very large delta-T is involved. But what do I know, try it and see what happens. I think there was an article about this on the front page a while back....

I've heard that anywhere from 5 to 35% of a peltier's wattage rating is used up as waste heat, depending on ambient temps, load, voltage, and the wattage of whatever you are trying to cool.

IF your CPU puts out 60watts, and the pelt is rated for 40-60w as you say, you'll be shopping for a new CPU very soon. 60w on a peltier does NOT mean 60W of cooling power, unless I'm wildy mistaken. Moreover, all pelts do is move heat around- you'll need to remove the heat from the water and I don't think a single pelt and HSF combo will be enough to do it. Water tends to gain heat much easier than it loses it.

Well that's enough of me dumping on your idea- nobody ever learned anything by listening to the pessimists (me) so I say you just give it a try. Keep your eyes on temps ,tho. And keep us posted on the results!:D

[ButcherUK]

Originally posted by Monster of Rock

IF your CPU puts out 60watts, and the pelt is rated for 40-60w as you say, you'll be shopping for a new CPU very soon. 60w on a peltier does NOT mean 60W of cooling power, unless I'm wildy mistaken. Moreover, all pelts do is move heat around- you'll need to remove the heat from the water and I don't think a single pelt and HSF combo will be enough to do it. Water tends to gain heat much easier than it loses it.


Actually that does mean 60w of cooling power... under ideal conditions (hotside temp of ~20C is usually one). The pelt actually consumes more power than this due to inefficiencies.
Also water gains an loses heat with equal ease, it;s just that usually the heating element has a larger delta and better conductivity than the cooling element (as the cooling element is usually air at ambient temps).

[Monaco]

Actually that does mean 60w of cooling power... under ideal conditions (hotside temp of ~20C is usually one). The pelt actually consumes more power than this due to inefficiencies.

hmm I don't think so man, but then again I don't know everything. Anybody got some facts to back this up, either way?

Also water gains an loses heat with equal ease, it;s just that usually the heating element has a larger delta and better conductivity than the cooling element (as the cooling element is usually air at ambient temps).

This I agree with totally. In practical applications, it's easier to heat than to cool for us, regardless of the physics.

[the_master]

God damn it!!! Why do you people have to smash my dreams!!
If I think it's going to work I'm not going to let the Laws of physics STOP me! Never! Never! ;)

- Just kidding ;) I really appreciate your edvice guys.
I think I'm just going to use a Black Ice Radiator mounted on the front of the case...humm... ;(
It's more expensive, but what can ya do....

[docJ]

Just a few things that I would like to clear up for ya!

1) Water gains or loses heat equally well, this depends on delta T.

2) TEC's pump no heat at delta Tmax, as delta T increases heat pumping decreases.

3) TEC's can be made to operate so they consume less power than they pump or a higher COP.

[ButcherUK]

Originally posted by Monster of Rock

hmm I don't think so man, but then again I don't know everything. Anybody got some facts to back this up, either way?


Yes, this quote "Regarding the module specifications, the Qmax is the maximum heat pumped at the maximumVmax and Imax applied", from this page (http://www.tedist.com/overclk.html). Qmax is the quoted rating of modules (also verifiable on that site).

[ButcherUK]

Originally posted by docJ
2) TEC's pump no heat at delta Tmax, as delta T increases heat pumping decreases.

Kindof, what it really is is that to achive deltaTmax you have to pump no heat (this is impossible as you always gain heat through insultaion etc.), and as you pump more heat the delta decreases. Same thing really just said in a different way (seems more clear to me ;))

[docJ]

It works for me ! Just as long as people understand that at Qmax,
delta T is near zero and at delta Tmax, Q is near zero.

[ButcherUK]

Yes the formula is
deltaT = (1 - Q / Qmax) * deltaTmax
Where Q is heat to be pumped.
so for an 80W cpu on a 156W TEC you have
deltaT = (1 - 80/156) * 69 = 33.6C