A little part of a much larger project I cant make up my mind on i'm after opinions even if their totally unfounded

A picture speaks a thousand words so
http://blackhawk.rrss.co.uk/Thinking.JPG

I'm trying to think which would be the best way to arrange this,

As you can see the base of the heatsink is less of that of the peltiers surface so if it was to be arranged like that of type 2 only the centre of the peltier would really get any cooling making it very inefficient possibly as the surface is a poor heat conductor I am led to believe

So then comes in type 1 having two cold/hot plates on either side to take the heat from all areas of the peltier and allow it to be cooled easier by the heatsink as it is a much better thermal conductor however! would the use of a second plate decrease the efficiency enough for me just to go with type 2 and only cool the centre of the peltier??

Also if I only cool the centre of the peltier as in type 2 there is a possibility of the rest of the peltier overheating to the point of failure for the whole unit

So what's you're opinions i'm currently leaning towards idea 1

My opinion is that option 1 would be a better option, to as you say prevent the peltier itself from overheating and reducing its cooling effect. But then how much does the heatsink plate cover the peltier, and if your putting another coldplate on top wont that increase the surface the heat sink is not covering?

Can I just say Peltier is one of the liquid cooling options I really wanted to do
when I first built my liquid cooled system. You have to keep us updated on the progress!!!

Your going to love this then Tam....think Sub Zero(hopefully) two pelts one on GPU one on CPU, totally submerged cooling

But you didn't tell me your opinion!

Peltiers are really inefficient imo considering the amount of power you have to use for the cooling you obtain from the peltier.

They output so much heat, that cooling the peltier consumes so much power to defeat the purpose of having a really good cooler.

However, you should have the hot plat of the peltier in direct contact with the heatsink, as adding another plate - unless its copper maybe, would really only add to trap the heat there, rather than conduct it away.

Also, you want to be careful about how powerful of a peltier you use, as you could get a lot of condensation around vital components, as you have extreme cold & heat in very close proximity.

Condensation wont be a problem but ill explain that when its finished, infact its impossible

In regards not to having a second plate see this is my problem I am thinking what you are about using a second plate it will build more heat BUT the edges of the hot side of the peltier that are not in contact with a heatsink if I don't use a second plate will then become very very very hot to the point of failure id imagine,

So its a trade off either an extra layer of copper adding inefficiency or running the risk of the uncooled edges of the peltier overheating

A Larger heatsink would be idea one that had a greater surface area on the bottom but unfortunately its not possible in this build.

EDIT: Yes the plates are both as pure as I could lay my hands on copper and are lapped to a mirror finish

Go with type 2. direct heatsink onto the heat side of the peltier would be more efficient. in regards to your heatsink only contacting the center of the peltier.... easy fix... you need to grind down the raised portion of the heatsink. What you want is a total flat surface across the entire area of the heatsink.

as far as condensation.... yes, this is an area that needs to be monitored. depending on how low, as in temp on the cold side of the peltier, the % humidity of the air running through your case, you can reach the dew point which can cause condensation on any exposed surfaces and more so with exposed metal surfaces. You can power the peltier to the point where the cpu temp gets below 0C.... but if the humidity at any given point in time reach high enough to cause you to reach dew point, you will start to condensate... high rate of condensation will depend on how much humidity is in the air. it's just something you need to be aware off.

GG
Due to the design of the heatsink there is no easy way to make the bottom of the heatsink larger by grinding it......I supposed I could solder the larger copper plate onto the existing parts though......
http://i684.photobucket.com/albums/vv201/blackhawk1uk/Peltier.jpg

Condensation wont be a problem
http://i684.photobucket.com/albums/vv201/blackhawk1uk/Case.jpg
Can you see where this is going?

I'll have to admit, condensation was a thought on my mind way back when I thought
about it. Also tbh, I seen it as more hassle. Liquid cooling is a hassle too, but not
so much as Peltier. I seen it as high maintenance
Whatever you decide to do, I don't have time to look at the graphic link you posted
as I'm at work, but i'll check it out later tonight when I get home.

ahahah it looks like something is going to take an oil bathe.....

Shawn Langley said:


from your original drawings.... the bottom of the heatsink looked like it had a raised portion... i.e. sorta like a heatsink bottom with a smaller raised portion in the center to make contact with the core. the sink is raise becuase it has to accomodate a shim that is around the core.

didnt realized that the raised portion in the drawing was the total surface of the bottom of the heatsink..................


three things.... you either can modify the heatsink to make the surface larger to fit all the surface of the peltier, find another heatsink with a bigger block, or get a smaller peltier.

An Athlon XP 2500+ Barton Mobile and an NF7 v2

http://i684.photobucket.com/albums/vv201/blackhawk1uk/EOS.jpg
Has the viscosity and clearness of water infact id drink it if it wasn't a potential laxative.

EDIT: My mistake, paint is hard to make things clear with, heatsink I'm having a hard time finding another one suitable (needs screw mountings from a website I trust in the UK) nor am I willing to spend much more money on this, smaller peltier considered but im going for sub zero here and it took quite some time for these to arrive, I'm impatient lol

I think ill solder a copper plate to the bottom of the heatsink seems the best and quickest way to go.

Shawn mentioned "totally submerged cooling" so I'm guessing that's why condensation won't be a problem.

What I don't get, is why use peltier if submerged? Why not just use very efficient heat-sinks (lapped of course) and super-cool the liquid?

Also, even if you do use peltier and don't have "edges" 100% heat-sinked, wouldn't submersive cooling fluid substantially mediate any over-heating effects?

I think your best bet (while not risking an actual CPU/GPU) would be to obtain some empirical data by doing a test bed using something to emulate the hot processor and add some thermal sensors at various points on the peltier plate edges and your heat-sinks. Vary the config until you're satisfied with results. If you're going to experiment...then do some hard-core experimenting.

Well that will teach me to not refresh the page before I post...looks like you guys already had the whole "submerged" thing figured out.

john albrich said:


My other concern would be the performance of the peltier under submerged conditions - I'm not sure I've seen this done before.

Depending on the temperature you drop the peltier to, you risk a massive change in the viscosity of the liquid being used, which could cause all sorts of problems.

You raise some good points John

And your also on the same path as my thinking, I already thought the delta fan i'm using to force oil through the heatsink anyway is going to be moving over the surface of the pelt fairly quickly anyway so it may just alleviate the problem.....but there's the problem and why i'm asking for opinions its all guesswork till I actually fill the tank.

In regards to using pelts, well its two fold firstly I really wanted to use peltiers, secondly its easier super cool a small point with a peltier than the 25 litres of oil to the same temperature that the pelt can achieve, also the tank would then start to gather condensation on the outside if I cooled the oil. But yes ideally I would have cooled the oil but it just wasn't a practical idea.

Finally as much as i'd like to experiment with different setup's seeing which is best I don't have the time or the patience unfortunately. Besides this is an experiment hence using the cheaper parts if all goes well ill take what I learnt to a Quad Core setup

Dublin
I have insulated around the Socket (up to where the hot side would start and the back of the socket) with neoprene as if it was going to be used in air to keep as much of a temperature difference as I can, and also this will mostly keep the cooled oil within the gasketing separate from the rest of the tank which will be substantially warmer. Also I was sent the data sheet for the oil and I don't believe the peltier can cool it anywhere near enough for it to be a massive problem.

That was always my understanding. I haven't given them a second look since I considered using them in my first Athlon Thoourghbred rig. <3

Fortunately efficiency isn't what I'm going for here, if it takes a few hundred watts and a hell of a cooling solution to reach negative temperatures I think its more than worth it, especially if you compare it to the cost of phase change or impracticality of nitrogen.

Quick update specialy for Tam

I decided to take Johns advice and test out the differnce setup's and turns out my inital thoughts were wrong using a second copper plate made things worse; even in air. So for the record a single cold plate and the hotside directly in contact with the heatsink even if its only 2/3 of it does work better.

GPU Peltier running at 5v
http://i684.photobucket.com/albums/vv201/blackhawk1uk/GPUat5V.jpg

Also I managed to chip the GPU core slightly, only half a millimeter off the corner but its visable....however so far the chip seems to be working fine strangely..... I assume not all of the silicone that makes up the die actualy has somthing etched onto it then? Or im just very lucky not to have found the problem yet?

http://i684.photobucket.com/albums/vv201/blackhawk1uk/SneakPeak.jpg
Just a quick one of the overall setup in testing, all this is mounted to a self constructed "dremmel made" motherboard tray, alowing the whole thing to be lowered in and out of the tank on "guide rails" iv'e added to the tanks walls.