Homebrew Heatpipes

[Olen]

I've been working on this project for a good week now and though there isn't anything usable yet but I have eventually managed to demonstrate that its possible. I don't know if this has been done before so appoligies if it has. Also this is my first write up so bear with me.

As the topic suggests I've been building my own heatpipes. Eventually I might even make a heatsink with them.

Before I start I'll briefly explain how a heatpipes works, if you know/ aren't interested then skip this paragraph. Its a sealed pipe :eek: with a little volitile liquid in the bottom and nothing but the vapour of that liquid above. Depending on the liquid there will normally be at a mild vacuum as the liquid/vapour will balance at the pressure where its about to boil. Heating up the bottom where the liquid is will tip this balance so the liquid rapidly evapourates and raises the pressure so vapour at the other end condenses. These phase changes quickly cool the bottom and warm the top (it takes the same energy to evapourate 1 gram of water as to heat it by 540*C - phase change is very effective!). Commercial ones work in any orientataion but to keep them simple mine take heat from the bottom to the top. For more info go to wiki (http://en.wikipedia.org/wiki/Heat_pipe).

http://i69.photobucket.com/albums/i58/Seidr/Modding/soldering.jpg
No its not acne - the sun went for me a few days ago

I wanted to know what would be most effective as a coolent, the most likely ones which are used commercially are ethanol and plain old water. So I made three - one with water, one with ethanol and one with just air as a comparison. To make them I took three equal lengths of copper pipe and squashed one end. This squashed end was soldered and tested (see above) and then a compression fitting was attached to the top. This was opened and the fluid (about 1.5ml) was put in. The fluid was then boiled until the air had been forced from the pipe when the compression fitting was quickly reassembled. As it cools the vacuum develops, then I shook them to ensure that there was some fluid left (you need very very little - it expands well over 1000 times when it vapourises).

Three finished pipes

http://i69.photobucket.com/albums/i58/Seidr/pipes.jpg

Bottoms labled

http://i69.photobucket.com/albums/i58/Seidr/pipeends.jpg

I tested them today but I'll wait until tomorrow to post more. Build the suspense :lick:.


If you feel like copying then please do but beware that sealed pipes (among other things) are dangerous and can burst forcefuly enough to kill you.
Also sorry if this is the wrong board but its cooling related...

[Zephik]

Interesting. Can't wait to see some result!

[Spawn-Inc]

hehe i tried that (http://www.thebestcasescenario.com/forum/showthread.php?t=7223) but i wasn't aware at the time that they had gas inside them.

did you solder or braze them?

also you first picture doesn't work/isn't showing

[Computer-Geek]

he soldered it .
This squashed end was soldered and tested

[Olen]

Yea soldered unfortunately. My torch isn't good enough to braze with and I don't have brazing rods for the welder, also it might be difficult with more heat to seal...

@Spawn-inc - What? Did you just put water and air in them?

Sorry about the broken pic - photobucket had, for its own reasons, deleted the picture... Fixed now.


Part 2

So having produced them I wanted to test them. This was a bit of a problem but I settled for heating one end with a steady heat source while sinking the transfered heat into a can of water at the other. I haven't really got many pics... sorry.

http://i69.photobucket.com/albums/i58/Seidr/Modding/jacketed.jpg

I attached a can to the top of each pipe so the top was inside it then sealed with silicone and wrapped it in lagging to minimise the heat lost to surroundings and not to the water in the top bowl (this was a big inaccuricy anyway).

The I put them into a shallow (only the very bottom is effective) basin of water at 60*C this was kept there by removing a little as it cooled and adding from the kettle. Each can at the top held 400ml of cold water, the idea was that the transferred heat would go into the cold water and from the temperature rise there I could calculate the pipes' thermal conductivities.

However, the seal on the can round the meths one didn't work so no results there I'm afraid. The seal on the air one died too but after a few minutes so I can safely say it was conducting VERY solwly. The water one held up to a full 30 minute test.

With a temperatute difference of ~40*C and being 38cm long and dT/dt being about 1*C every 360s I could get a rough answer. To be accurite it would require a slightly different set up and much more difficult maths but the fundemental inaccuricies made such maths pointless.

Anyway the number *drum roll*: 960 w/mK give or take.

Now bearing in mind the thermal conductivity of solid copper is only 400 w/mK my first attempt at a heatpipe performed over twice as well as solid copper would have! Success I'd say.

In other less useful but more understandable terms the 13 mm pipe transfered heat at a rate of 23w over 38cm after losses. The actucal heat removed, especially if the lagging were removed would be considerably more.

Another point about heatpipes is that they get more effiecnt the hotter they are up to a point. At a certain temperature just after they hit peak efficency all the liquid is boiled off and they stop working (the copper pipe still conducts but only about 1/80th as fast). This could be a problem so I tested mine at about 70*C by immersing it in water at that temperature then taking it out and shaking it. I could hear a very little water in the bottom so its fine up to 70 and a bit above.

Sorry for the lack of photos. Will be a few days before the next update when I hope to have data for meths early next week and then to build a heatsink... But thats for later.

[MaxOC]

wow this is cool,

[Spawn-Inc]

@Spawn-inc - What? Did you just put water and air in them?



i had some 20 year old copper from a old ac unit i was changing and i just used that. i didn't put anything in it, but there was a littl bit of water inside i think. i never properly tested it becuase i didn't understand the bios on the old computer i put it on. i was board and saw the old copper so i decided to try and make a heatsink.

[Olen]

Part 3

Did some more tests today.

Sorted the seal on the meths one and did the same test as the water one got. Odd results - it worked better until the whole thing got hot when it worked considerably worse. It wasn't boiling dry either from what i heard shaking it so more research is needed there. I plan to make a couple of thinner ones using varying amounts of fluid and perhaps a better testing mechanism (any thoughts anyone?).

Then I'm going to make heatsinks of varying designs and a heatsink tester to test them on.

Current plans for the tester are a metal plate the same contact area as a processor heated by two high power wire wound resistors with a combined power of about 60w. I'd be interested in thoughts on this power - is it too low for more modern processors?

Here's a sketch up of the first heatsink design:
http://i69.photobucket.com/albums/i58/Seidr/heatpipesinkgraphic.jpg

It's a copper block (assuming I can locate one) with two pipes going though iand each end of each pipe will go up to a heatsink. Another option is to have the four pipes come vertically out - this could give better performance at the cost of less even cooling (it should even itself out in a big block of copper though).

Not much to report but should have more results at the end of the week. Then on to heatsink design.

[rakuza]

heat pipes need a wick in them to work better eg. the heat pipes made here wont work upside down mabe not even on the side up on a upwards gradient they will work ( the steeper the gradient the better) where as a wicked one will work better on lower gradient and even horizontal.

(btw as a demo the teacher in my chemistry class made a sodium heat pipe which wont start to work till it gets quite hot and the seal broke and made a big flame) 8)

[jdbnsn]

Olen, you should put this in the tutorial section, it's great. Just open a thread in there and write an intro post (i.e. what they are, what they do,how they work, etc...) and I'll move your other posts in there.

[Olen]

@rakuza - I know but a wick makes no difference if its running verticle, it might even impair the workings. As a wick is quite difficult to produce I'll just run them verticle.

jdbnsn - not yet. Once I have a working prototype yes but this is more or less useless to anyone who actually wants to make something. More reaserch is needed into how much water or meths is required for best results. There was a flaw in my maths which will have given a falsely high result so its not good enough yet. Next I will raise the pressure a bit and hopefully increase performance, I might try a lighter substance.

But first I need a better way to measure efficency, my last method was rubbish. Maybe a small thermistor would be better but I don't have one. I might do some more fiddling in a couple of days.

I also need an old heatsink or block of copper to attach them to and some sheets of preferably copper to sink the heat at the top of them.

[Olen]

Part 4

Ok quite a while sonce I last updated but I have some progress. New heatpipes, new test, new solvent.

So the last heatpipes prove my point, but the results were totally wrong and they were hardly a conveniant size or easy to test.

So I made new ones with 8mm pipe and a better testing method. I decided that they needed to be openable to change the working fluid without taking them apart so heres what I did.

Cut 4 lengths of 8mm copper pipe about 300mm long. The red thing is a pipe cutter and it vital for this project.
http://i69.photobucket.com/albums/i58/Seidr/Modding/4lengthsandpipecutter.jpg

Then, as with the last ones, crush one end and solder. This might be the trickiest bit but once you can do it its easy. Those solder joints look alright to me and they hold air fine. Because they are smaller and more delicate I tested them by blowing in the other end into a bowl of water. No bubble = good.
http://i69.photobucket.com/albums/i58/Seidr/Modding/endsSoldered.jpg

Now the clever bit - have a jacket of water around the top of the pipe while still being able to easily undo the compression stop end there. Well, a 8mm to 22mm reducer made from an 8 to 13 then a 13 to 22 soldered a bit from the top with a nipple which slots on (but not soldered) sealing with vaseline gives a jacket of water from 50mm down from the top to 30mm above it.
http://i69.photobucket.com/albums/i58/Seidr/Modding/filePipeandFitting.jpg

Unfortunately the fittings are designed to sit on the end of the pipe so some filing was needed - my fake dremmel is dead so thats lots of elbow grease. But I persevered and put it together with compression top:
http://i69.photobucket.com/albums/i58/Seidr/Modding/endsFinished.jpg


Now a short diversion into soldering the perfect joint:
Step 1
Get the parts you will need, solder any close joints at the same time as you will melt close ones anyway and can overdo a joint you've already done if you then add one next to it (not to metion damaging presoldered fittings which aren't being used).
http://i69.photobucket.com/albums/i58/Seidr/Modding/solder1.jpg

Step 2
Solder won't hold onto muck. Take some fine steel wool and polish.
http://i69.photobucket.com/albums/i58/Seidr/Modding/solder2.jpg
Nice and shiny :banana: < always wanted to use him

Step 3
Still not good enough. If the solder is going to really stick with airtightness we need it to run freely. We need flux. Not like electronic flux - plumbing flux is acid and burns you skin a bit so be careful. Rub a little on both sides of the joint and put together and twist.
Applying flux with a bit of card:
http://i69.photobucket.com/albums/i58/Seidr/Modding/flux.jpg

Step 4
I'm going though a presoldered joint as they are easier to do because the solder is inside the ridge on the outside fitting. All we need is heat. The best (though slightly more difficult) method is to put the outside fitting down and heat the top and just above the joint to pull the solder up. Just as it starts to appear at the top(all the way round at once if you heated evenly enough) stop heating. If it doesn't appear everywhere help it a little with very localised heat but aviod over cooking any part of the joint as this ruins it becuase the solder all runs out.
It looks a bit yellow under the flux but this is a good joint.
http://i69.photobucket.com/albums/i58/Seidr/Modding/solder.jpg

Step 5
Wash off the flux in warm soapy water. Flux casues endless problems down the line - such as eating though your joints - so wash it well.

NB For joints which aren't presoldered like the crush one (a solder endstop would be best here but I don't have any) flux more heavily melt the flux and get it hot with the flame then draw the solder into the joint by playing the flame under it. Don't heat directly at this point or if you do go easy. Give it another little blast if it looks like it needs it at the end.


Now you know how to solder and have seen the nipple which goes on top to hold water. Heres a view down into the heatpipe:
http://i69.photobucket.com/albums/i58/Seidr/Modding/inside.jpg

I opened the compression fittings and put in the desired working fluids. One has a little water, one has more, one has ethanol and one has some petrolium distilate (you know the solvent cleaner used on electronic boards which evapourates REALLY fast - that one). With the water and ethanol I put the top on lightly and blowlamped the bottom until steam came out so there was no air inside. Then I did it up tighty. For the solvent one I warmed it in my hands then gave it another blast then shut it. Flame + solvent = unwise.

I'm still finishing the nipples (note - a nipple is a short bit of pipe... don't get too excited) as I was only using one end of a presoldered joint so there is solder all over the other end and lots of filing. Fun.

I'll test them soon.

The finished product:
http://i69.photobucket.com/albums/i58/Seidr/Modding/completeWithScale.jpg


EDIT:

Part 5

The results.

To be honest they weren't particulalrly inspiring. The thin ones didn't function as well as the thick ones even after maths errors were sorted (turnes out the first maths was out by a factor of six (a couple of wee mistakes multiplied)) so was more like 190 w/mK.

These ones varied between 57 for the one with a little water in it to 85 for the one with meths in it. However I couldn't find anything suitable to lag them in so the true value will be better as heat will have been lost to surroundings. Will get some thermistors soon for even better testing and I will focus on meths and maybe take one more look at spray cleaner or petrol. Then it will be onto making and testing heatsinks... but thats later.

I have had another think about heatsink design which could prove interesting... I have two quite ... different ideas based on this sort of phase change