Sciencemadness Discussion Board

Liquid heat exchangers

mnick12 - 10-7-2010 at 17:23

So I recently ordered some Peltier chips, and as you may know to run properly they need a good way of disspaiting heat. For small chips a heat sink is generally enough, but for larger ones you need a massive heat sink or something better. I was thinking of using a liquid heat exchanger also known as a water block. The problem I have run into is a good water block is generally $60.00 to $90.00. Which is ridiculous considering I can get a 540w chip for 1/3 the price. I dont even understand why they are so pricey all they are is a machined piece of aluminum or copper with two nozzles.

My idea for the water block was to have it run through a radiator where a fan would cool the coolant back down. But considering how expensive water blocks are I am forced to go a different route. My question is what would be a good alternative to a water block? Or could a water block be made without to much trouble?

Here some stuff http://en.wikipedia.org/wiki/Water_block , http://en.wikipedia.org/wiki/Peltier_cooler .

Thanks

watson.fawkes - 11-7-2010 at 06:12

Quote: Originally posted by mnick12  
My question is what would be a good alternative to a water block? Or could a water block be made without to much trouble?
If you're handy with a brazing torch, you can make an adequate one without too much trouble. Unlike a water block for a computer, yours doesn't have to be so compact. Start with a slab of copper about 1/4" thick and the same plan outline as your Peltier device. Now take a length of refrigeration tubing of convenient diameter, say 1/4". Roll it up into a coil the same height as one of the slab dimensions. Carefully flatten the coil so that one side is straight for the length of the slab. It may be easier for you to wind it on a former, a block of wood with rounded ends to avoid pinching the tubing. Braze it all up in a single operation, jeweler's style, where you pre-flux the entire joint, slab and pipe both, lay down your fusion alloy (hard solder or brazing rod). Heat from the bottom until everything melts.

Add compression fittings if you want to make the block demountable. Or else get a swaging tool for your tubing and just braze the whole thing together.

Mr. Wizard - 11-7-2010 at 07:16

I agree making one with copper tubing wouldn't be too tough. I would use solder though instead of brazing. Solder will work just as well and can be done with more commonly available materials and tools. I agree with watson-falkes' suggestion to pre flux and tin everything before you go for the final bonding. You might even tin the tubing before you form it.

There might be some confusion as to what is soldering and what is brazing. Brazing to me involves a much higher temperature close to red heat, and a non lead / tin alloy.

[Edited on 11-7-2010 by Mr. Wizard]

Sedit - 12-7-2010 at 18:51

What about just forgoing the solder or brazing and go for epoxy? Make the block itself out of either Plexi glass or metal and for the base tha acts on the hot surface use a piece of Aluminum pie tin. It will be thin enough for rapid heat exchange yet strong enough to not tear at first sight.

It does not seem like a design that needs overthinking at all. Just glue some tin to a block and run some water thru it as fast as you can.

Soldering/Brazing

ciscosdad - 12-7-2010 at 21:09

I have had considerable experience both soldering and brazing. They are the same thing but with alloys of different temperature ranges.
The keys to success are:
1 clean the base metals (abrasive)
2 apply flux to where you want the braze to go and
3 control the temperature (to avoid oxidizing the substrate or the brazing alloy).
This is why you would choose copper as the plate material. Much easier to braze or solder.
Glue will be a lot less efficient at heat transfer as the epoxy (or whatever) is actually a poor conductor of heat and you will be depending largely on whatever direct contact points there are between the 2 pieces. Solder or brazing alloy will transfer heat very well for a much better result.

I could go on, but you get the idea.

Wizzard - 13-7-2010 at 06:07

Those 540W modules are a joke. I have one, and if I can get 6V/150W into it without the wires melting, it's a nice day. And I put solid-core copper wire leads on, about 10AWG. Most peltiers are rated for 12V DC. It's still anice module, but 540W is some rediculous energy output density- Could likely rival those perfect chips of solicon we pump 200W into :)

watson.fawkes - 13-7-2010 at 06:28

I recommended a "brazing torch" to indicate that it takes a lot of heat to get a block of copper and bunch of tubing up to temperature enough to liquefy your fusion alloy, whatever it is you use. A soldering iron just isn't going to work. Another way of getting it up to heat would be to use an electric hot plate. Protect the hot plate with a sheet of aluminum, which won't react with ordinary soldering/brazing materials.

I second the recommendation to pre-tin everything. On large pieces like this, it sure helps to get everything to fuse all at once.

I don't think it matters a huge amount here between brazing and soldering. I wouldn't use a lead-based solder, though, out of concern for both the environment (your own) and bond strength. I would specifically recommend a silver-bearing fusion alloy, though, because they have much lower surface tension when liquid. There's a lot of surface contact area in a cooling block, which is the whole point, after all, and it all needs fusion with your alloy so that it's actually, not apparently, in thermal contact. There are tin/antimony/silver solders available as lead-free plumbers solder available at home stores. I already have a supply of 2% silver brazing alloy, though, which is what I'd use if I were doing it today.

mnick12 - 13-7-2010 at 08:53

Thanks for the ideas.My chips are 40mmX40mm , but they are still in the mail.
I am thinking about using aluminum blocks instead of copper since they are quite a bit cheaper. My idea is to order some of that thin copper tubing you see on refrigeration units wind it into tightly packed zigzags and then bond to the block with a thermally conductive epoxy or braze it. My torch is a plumbers propane torch so I hope that will work.

Thanks.

unionised - 13-7-2010 at 09:51

Quote: Originally posted by Wizzard  
Those 540W modules are a joke. I have one, and if I can get 6V/150W into it without the wires melting, it's a nice day. And I put solid-core copper wire leads on, about 10AWG. Most peltiers are rated for 12V DC. It's still anice module, but 540W is some rediculous energy output density- Could likely rival those perfect chips of solicon we pump 200W into :)

Are they rated for the power that you feed them, or the power they pump?
In any event, good luck trying to solder aluminium. It's not impossible, but it's tricky. Thermal epoxy might be a better bet.
Also, why not cut out the middle man and glue the coiled tubing straight to the Peltier?

densest - 13-7-2010 at 14:55

@unionised - the Peltier units have extremely thin alumina plates for top and bottom - strong, but very brittle. The metal in contact with them must be level to an extraordinary degree or they'll shatter. Also, the heat is dumped in in an array of point sources and the lateral heat conductivity is very low so a metal heat spreader is very useful when trying to couple such a unit to anything else.


[Edited on 13-7-2010 by densest]

franklyn - 19-7-2010 at 06:24

Quote: Originally posted by densest  
Peltier units have extremely thin alumina plates for top and bottom - strong, but very brittle. The metal in contact with them must be level to an extraordinary degree or they'll shatter. Also, the heat is dumped in in an array of point sources and the lateral heat conductivity is very low so a metal heat spreader is very useful when trying to couple such a unit to anything else.
Lapping offers no particular advantage.
Products like this are what are used to mate Heat sinks to CPU processors.
http://www.google.com/search?hl=en&source=hp&q=arcti...

@ mnick12

Water cooling is quiet but heat sinking capacity depends on the
capability of a particular system employed not the method used.
The computer case is the major obstacle to cooling adequacy.
Simply leaving the case door off will improve the ability of any
chip heat sink fan. A work around is to channel air directly from
outside the case with a PVC pipe duct attached to a hole cut
into the case door directly over the heat sink fan .
There are commercially available units employing a flexible duct.

- Fan duct kits
http://2cionline.com/nova/Products/computers/cpu_vent.htm
http://www.directron.com/badongs60.html
http://www.highspeedpc.com/Merchant2/merchant.mv?Screen=CTGY...

- Fan adpters and ducts
http://www.jab-tech.com/Fan-adpters-and-ducts-c-171.html
3 inch flexible duct
http://www.grainger.com/Grainger/items/3C590?Pid=search
4 inch inch flexible duct
- http://www.grainger.com/Grainger/items/3C591?Pid=search

.

JohnWW - 19-7-2010 at 10:53

Here are a few PDFs on the subject of Peltier (effect) coolers:

http://www.designer-iii.com/cco/Peltier.pdf

http://www.solstice-analyse.com/docs/produits/prelevement_co...

http://ankersmidsampling.com/cooler%20APC%203xx%20m&c.pd...

http://www.freepatentsonline.com/5960142.html

http://saaubi.people.wm.edu/TeachingWebPages/Physics252_Spri...

http://www.micropelt.com/down/datasheet_mpc_d403_d404.pdf

http://www.bradford-space.com/pdf/be_paper_peltier_glovebox....

http://web.mit.edu/nanoengineering/publications/PDFs/NPO3055...