i need a place to post some questions

[XeroHouR]

My mind is back to work on contrived cooling methods...just for fun mostly. until I hit something I like that i can implement on my real system. So I wanna ask questions here..keep it organized instead of having multiple threads per question, they will all be along the same lines.

1.) anyone know the heat dissipation (W, or BTU or anything? i can do the conversions if you dont want to) of a heatpipe. i'm sure size matters but just on a basic guess. could i just use the specific heats of the materials and water to figure it out?

2.) heat pipes dont work in reverse right...coldpipes? Like they wont transfer cold from one place to another right? From what I know of heatpipes(which is pretty limited) its basically sort of a closed convection cell...or something, where the water inside evaporates from the lower(physically) side then goes to the top where it cools and goes back to the bottom, and in the process of evaporating and condensing it transfers heat. Now I can't even say for sure heatpipes have ANY liquid in them at all...but thats what I've read. So if this is true...it cant transfer cold at all(well short of what any hunk of metal would do)...am I right?

[cadaveca]

Yes. heatpipes are filled with a gas at pressure, and there is a wick inside to allow the moisture to travel back to the hot side once cooled. Quite obviously, they work best when the cool side is the highest.

Your assumptions are correct in the "cold"...it takes a TEC for what you are looking for i think.

[XeroHouR]

yes, those are in the works here too this is all just a theoritcal plan im working out...a learning experience, but if some how it works it'll be implimented. Its all a fun learning process. I'm loving every minute of it.

[cadaveca]

I have both my cpu and gpu heatpipe cooled...andi could not be happier for the results on air....heck, my gpu is a full 20c lower @ full load than it was when idle with the stock heatsink, and the memory 12(although there is not heatpipes directly on the memory).

I am looking at getting an si-97, or a hyper6 for my next athlon build....and i really like how most heatpipes can be used across multiple platforms.

[XeroHouR]

cooling the GPU makes the ram cooler...though I'm not sure why. When I wasn't quite cooling myGPU well enough, the whole card, including the ram got way hotter. When I had my Giant 3 on my 5900 it dropped the temps like 30+oC and it only went up ~10oC load. They are marvelous inventions.

[robbie]

They are marvelous inventions

That they are, I've seen heat pipe technology used in high volume copiers as well. They use it as a cooling roller with big fins on the end and ported air aimed at the fins. I'll have to get a picture of one sometime.
Rob

[XeroHouR]

ok, i know this post is better served in the general hardware section, but I been thinking about trying this mod out, looking at the dominator pro. This is pending the cost of having my friends make me a psu for my pelts(given I go with this idea), does anyone think this is really viable for use in a TEC setup? Would it maybe be a better idea to just have them build a PSU with that kind of power? Given the dominator requires 24+V, adn this only give 17V(if i remember my electronics right), but I would assume that you could mod it to run all the rails together? This or maybe get one to run the upgraded pelt for my video card which is in the works. Basically, I just need to know if this mod is a viable means of getting more of these types of power through the wires for use.


http://www.overclock.net/showthread.php?t=1933


btw my friends are EE majors, and they say building a simple PSU, for powering simple stuff like tecs(not quite computers) is easy, and wouldn't be too difficult or costly to produce(esp because of the connections they have with componets).

[XeroHouR]

oh dear god...i hadn't seen this :
http://www.frozencpu.com/cgi-bin/fro...x-blc-134.html
my brain hurts, and I'm scared.

[Gizmo]

I'd tend toward building a PSU specifically for the pelt. The mod mentioned at overclock.net would only work well with a PSU that regulates the +12v rail; PSUs that regulate from the +5v rail wouldn't be able to deliver the full rated load without also placing a load of some significant size on the +5, so you would tend to get less than the full 28 amps from the PSU mentioned, and you'd also tend to get less than the full 12v.

That Dominator looks nice, but you are talking almost 700w of heat at maximum output JUST FOR THE PELT. It should definitely be able to cool your cpu, but it wouldn't be cheap.

[XeroHouR]

yea I realize how much heat that is, and man is it an intimidating task, which is why it appeals to me. I was considering dedicating my water loop to just the CPU now anyways, since the video card is now seeing the same temps as air cooling...which is quite unfortunate. It was -5oC ambient when I had my XP in there, this prescott musta warmed the loop an awful lot. Maybe someone will come out with a superior air cooler for it in the mean time. I'm pretty sure with a pump upgrade my water system can handle 700W of heat disspiation from the hot side of the pelt.

[surlyjoe]

Originally Posted by cadaveca

Your assumptions are correct in the "cold"...it takes a TEC for what you are looking for i think.

wouldn,t it be the same thing but at a lower temp and reversed?

[Aedan]

Originally Posted by XeroHouR

2.) heat pipes dont work in reverse right...coldpipes? Like they wont transfer cold from one place to another right? From what I know of heatpipes(which is pretty limited) its basically sort of a closed convection cell...or something, where the water inside evaporates from the lower(physically) side then goes to the top where it cools and goes back to the bottom, and in the process of evaporating and condensing it transfers heat. Now I can't even say for sure heatpipes have ANY liquid in them at all...but thats what I've read. So if this is true...it cant transfer cold at all(well short of what any hunk of metal would do)...am I right?

You're wrong, because you're looking at it the wrong way. In exactly the same way you can't make a room dark by putting the opposite of a light bulb in there, you can't transfer cold per se, as cold is just the absence of heat, as dark is just the absence of light. However, you can still transfer heat. As you've already sussed out, the evaporation and condensation makes them passive phase change devices. They can move heat from one place to another.

If I put the bottom end against a CPU, and the top end against a TEC, what's happening? That's right, heat is being transferred in the correct direction.

Heatpipes need some form of liquid and a space for the vapour inside them. The liquid and vapour pressure inside determine the efficiency at a set of given temperatures.

[surlyjoe]

err,, what he said!!

[Aedan]

Originally Posted by XeroHouR

I'm pretty sure with a pump upgrade my water system can handle 700W of heat disspiation from the hot side of the pelt.

Nononono. The peltier itself will GENERATE Umax*Imax watts of heat itself. That's 26.7 times 32.8, which is 876W of heat at full power (not necessarily full efficiency). On top of this, you also have the heat that the peltier is pumping (437W). That's a worst case total of 1312W of heat you need to get rid off via your water cooling.

Then, you'll also have the peltier's power supply generating a fair bit of heat. Assuming it's a good switcher, and you're probably only losing some 20% of the power, which leaves you another 120W to get rid of.

Also remember that your system's energy consumption will probably jump by about four times too!

Given the cost of the device, the assosiated components to go with it (upgrade of water cooling system, probably including radiator) and the power supply, you might well towards price of a phase change system. A phase change system should offer better performance, lower temperatures and far higher efficiency!

[XeroHouR]

Ahhh, I was thinking that the total heat was the pelt+the proc and then a bit more because of the inefficency. And you can't base how much heat your water system can take based off the heat that your rad can supposedly dissipate either I assume?

As for the heatpipes thing I just phrased it wrong, I was treating "Cold" as something other than the lack of heat energy. I guess what I meant was, for lack of better wording so wrap ur mind around this for me, carry the cold down to the bottom and make the bottom activly cooling the bottom processor side. But now like you said in the theoretical tec sitution(tec on heatpipes) thats limited by the pressure of the gas inside for its operating range? Say that the upper surface has basically no limit to the heat it can dissipate(just for the what if factor), the heatpipes would have a limit to how much they could carry away right? And I'm guessing that limit is right around what a really good heatsink could do?(on average for the heatpipes that are actually used)

btw this is all theory so feel free to theorize on your own, I'm just searching for answers.

btw does anyone have the specific heat of water on hand? I wanna look at how close that 1200W of heat brings the water directly above the tec. Wow my chemestry class this semester IS good for something...

[Aedan]

Originally Posted by XeroHouR

you can't base how much heat your water system can take based off the heat that your rad can supposedly dissipate either I assume?

Not on its own. It's all about thermal resistance. Basically, the more heat you push into the radiator, the hotter it will get, all things being the same. Now, I don't know the exact relationship, so I can't even begin to guess at how much hotter the radiator would be. The big gotcha here is that the peltier is a heat pump. The amount of heat it can pump is strongly dependant on the delta (or difference) between the cold side and the hot side. The bigger the difference, the harder it is to pump heat across the gap. This is exactly the same sort of idea as a water pump and the back pressure. If there's a lot of back pressure against the pump, it can't pump very much water, as it's spending most of it's energy 'fighting' the pressure. So, ideally, you want your radiator to be as close to ambient as possible.

The bad news on this front is that radiators get less efficient the closer they are to ambient. Radiators have a log relationship - the hotter they are compared to ambient, the more heat they can shift. Unfortunately, this is the opposite situation of the peltier!

Originally Posted by XeroHouR

Say that the upper surface has basically no limit to the heat it can dissipate(just for the what if factor), the heatpipes would have a limit to how much they could carry away right? And I'm guessing that limit is right around what a really good heatsink could do?(on average for the heatpipes that are actually used)

Like everything, the heatpipe does have a limit to how much heat you can push along it. However, this depends on the fluid inside the pipe, the width of the pipe, the construction of the wick (Technically, heatpipes work in any orientation due to the wick. Thermosyphons have to be placed the right way up) and a whole bunch of other things. I know of heat pipes that are designed to move 1200W with a delta T of 20C. You will always have a temperature gradient when you've moving heat - it's just a fact of life.

If you actually use water, then your heat pipe won't really move any heat until the hot end is at 30C. If you want it to start moving heat before that, you need to switch to a different working fluid, such as ammonia.

There is a lot of information about heatpipes on the internet - Google is your friend!

[cadaveca]

All of my observations have been just in practice...if i place my system on it's side, the gpu gets hotter, and the cpu gets cooler, only a few degrees in both places, but we all know how bad just a few degrees can be!


I've been messing with the idea of having a heatpipe inside a rad, and making up for the differences for pelt cooling that way (trying to make the rad more efficient), but once i got the thermaltake heatpipe-cooled PSU, i thought it might be a misplaced idea....

[Kaitain]

Originally Posted by Áedán

Not on its own. It's all about thermal resistance. Basically, the more heat you push into the radiator, the hotter it will get, all things being the same.

Actually yes you can. The radiator is (usually) the rate-limiting step since air is a pretty useless heat transfer material. In thermal system design one either works forwards, specifying the size of radiator needed to dissipate a given thermal load or to work in reverse, specifying the maximum acceptable load for a given radiator.

The calculations to determine the temperatures of the system at or near steady-state are pretty simple (for a chemical or mechanical engineer)

[Daniel ~]

Aedan was wrong? The foundations of my world are crumbling!! ":O}

[XeroHouR]

sorry im a computer tech major...and only in my second year of study...I don't have a firm gasp on ANYTHING yet. I just got through enthalpy changes in chemestry(btw I was pretty good at it ). So what are the average heat dissipations of radiators? I'm using a heatercore, I think its one from a 77 Bonneville, its larger than 2 120mm fans on all sides, its a lot of surface area...I guess its just wishful thinking to hope that just water would cool such a pelt and give the kind of low temps I think would help the prescott...but I'm not beat yet, I'm gunna work something out.