New aluminum blocks

[FUGGER]

We just finished the design and started testing of our new captubeless phase change blocks, and we may go into production of a small run if there is interest.

Estimated cost to build is $225 in time, parts, weld, and lapping.

64 pins inside that extend from base, 50/50 expansion to surface area, 2"x2"x 1.34"

http://fugger.netfirms.com/alu.jpg

[afireinside]

Wow that RED X is really sexy :rolleyes:

[Marci]

Mefinks thats it in his avatar...

[Gary Lloyd]

Methinks he misspelled the name of the jpg file

[chilly1]

Here is the pic I promised. I am currently lapping the block, I will put a mirror surface and the plate it with gold....
http://www.blairwing.com/images/P1010010.jpg

[bowman1964]

man they are nice...would be cool to play with...:D

now guys remember if you are looking to sell this has to be placed in the sale section or services area...just a reminder.:D

[FUGGER]

Not available, approximate cost to make for those who might have wanted to know.

Netfirms ftp upload didnt go, and still down.

[DaBit]

Nice TIG welds. But why aluminium?

And I know you guys are not stupid, but did you use the 1050 alloy? I hope you did; heat conduction of high magnesium/zinc content aluminium is bad.

[Holst]

Id also be interested in some more info on the why's and wherefores of your design..

TBH I dont quite understand how a "captube less" block works..

[FUGGER]

We found that the aluminum blocks are superior to the copper and brass ones.

I cut the base from solid 2x2x1 aluminum stock with a function of the router call area removal, I tell the router that the bit is half the size of actual and it will remove exactly 50% of the material in the area I setup as the center of the block. Since the pins inside are delicate I have to make a lot of passes to get the depth and not bend or break pins inside. I take my time and every one so far has been a perfect block

The top half has a distributor with 6 or 10 ports machined into the top by Chilly1. We use an adjustable expansion vavle to flood the evap head and maintain coolant levels. This also gives us the ability to adjust the coolant getting back to the compressor. In a idle situation the compressor may get too much coolant and start to frost on the side, so we can adjust the valve for less coolant return to solve easy. on the other end, we can set it for max temp on the evap head and less cooling back to the compressor. We can tweak a lot with the current config.

[Pimpsho]

do u have ne pics of premod...like b4 u put it together? i cant remeber if uhad another thread about ithis...

[berkut]

You mean that aluminium evaporators work better than copper ?

[FUGGER]

I took a few pictures of one of the blocks in progress.

http://fugger.netfirms.com/alu1.jpg
http://fugger.netfirms.com/alu2.jpg
http://fugger.netfirms.com/alu3.jpg
http://fugger.netfirms.com/alu4.jpg
http://fugger.netfirms.com/alu5.jpg

[nas]

Are you sure alum is better evaps then copper ? sounds wired to me, but it will solve me lots of problems coz the damn copper is so soft

[DaBit]

Heat conduction is heat conduction. The problem must be somewhere else with copper.

What exactly did you observe when comparing copper with aluminium? And what aluminium alloy do you use?

[chilly1]

We haven't tested this one yet, the first one worked better, it got cold instantly and pulled the processor down to the same temps as the copper block only in a few seconds instead of a few minuets. This was on my single stage test rig. We sent the first out to be polished and they removed too much material, when I pressure tested it to 350psi, the bottom expanded like a soda can, But maintained it's shape (almost) under refrigerant load so the tests are a little skewed. This aluminum is softer than the copper. Brass is the hardest weve used so far and appears to conduct heat ok, I will get back to you on the alloy, I just picked up the the only stuff they had available, not many choices there.
It appears to be fairly easy to tig so I supect it is a fairy good alloy.

[nas]

what hight is its base plate ?

[iboomalot]

softer ?? what alum did you use?? might try some T6

[DaBit]

Use the aluminium with the least amount of copper, zinc, magnesium etc. in them. The heat conduction of the aluminium drops very fast, even with a single percent of 'impurities' in them.

The 1050 alloy is soft, and consists of 99.5% Al. This is definitely the one you want.

The fact that the alu block drops so fast; isn't that a side effect from it being light? Thermal mass is a lot less than that of a copper block.

Personally I would consider a slow but steady temperature drop a good thing since it reduces thermal stress buildup on the processor and mainboard.

[chilly1]

I took a piece of AL rod,Brass rod and a Copper rod from the same stock I make the block from and put a piece of wax on the ends of the equal length/dia rods same amount of wax and set them under the burner on my stove with the same conditions.
I controlled all the variables and insured they were the same except the material of the rods. turned the burner on high The AL one melted in 45 sec. the copper in 68sec and the brass in 75sec. Seeing they were all the same diameter and length the aluminum has a faster conduction, but I haven't tested the actual btu's the indivigual rods will carry, I am devising a better test of the materials and I will document this. I did not take into account the thermal carrying capacity of the materials and I assumed that it is directly related to the speed at which the wax melted. but after your question I am now in doubt and I must devise a more comprehensive test.

[nas]

i would sure love to see some pics of this test :)

anyway.. again , what size is the base plate is ? and does alum wont reshape after the havy pressure with 2-3mm base like copper ?

[Holst]

Originally posted by chilly1
I took a piece of AL rod,Brass rod and a Copper rod from the same stock I make the block from and put a piece of wax on the ends of the equal length/dia rods same amount of wax and set them under the burner on my stove with the same conditions.
I controlled all the variables and insured they were the same except the material of the rods. turned the burner on high The AL one melted in 45 sec. the copper in 68sec and the brass in 75sec. Seeing they were all the same diameter and length the aluminum has a faster conduction, but I haven't tested the actual btu's the indivigual rods will carry, I am devising a better test of the materials and I will document this. I did not take into account the thermal carrying capacity of the materials and I assumed that it is directly related to the speed at which the wax melted. but after your question I am now in doubt and I must devise a more comprehensive test.

You may find htis is due to the copper and brass having more mass... so they took longer to heat up.

you need to test in equilibrium... then the copper should win.

[chilly1]

Originally posted by Holst
You may find htis is due to the copper and brass having more mass... so they took longer to heat up.

you need to test in equilibrium... then the copper should win.
Exactly I am developing a test rig for this BTU through put is something I need to know and it may be a month or two til it is up...

This is 2in X 2in X 1" high, the fittings add 3/4 in. Building cascade for it now.... Hand lapped to 200 grit wetted with cutting oil last passes with rouge and terrycloth, all done on flat stone...

http://www.blairwing.com/images/Block.jpg

[DaBit]

Originally posted by chilly1
I took a piece of AL rod,Brass rod and a Copper rod from the same stock I make the block from and put a piece of wax on the ends of the equal length/dia rods same amount of wax and set them under the burner on my stove with the same conditions.

Those rods have different masses, and thus, they need a different amount of BTU's (or kJ's, whatever you prefer) to get heated to the same temperature. Thus, you are basically testing heat capacity here instead of heat conduction.

Here is how I would test it. I would take the rods, insulate them, make one side always the same temperature, and apply a fixed amount of heat to the other side.

Then, the temperature the 'heated' end of the rod will rise to is a measure of heat conductivity. The more conductive, the lower the temperature.

To keep the other end at a fixed temperature, I would dump a nice heatsink into a bath of ice water, and firmly clamp/screw the rod to that heatsink.

To apply heat I would use a power resisitor.


I think aluminium blocks would perform slightly less than copper, but i doubt the difference is very large. I have used alu for waterblocks, and they worked fine.

[chilly1]

Originally posted by nas
i would sure love to see some pics of this test :)

anyway.. again , what size is the base plate is ? and does alum wont reshape after the havy pressure with 2-3mm base like copper ?

The first prototype had pins and a 0.1in base this under 250 psi test expanded 0.15in
The new base is 0.3 with 0.2 support ridges between the pins.
I will post a whole series on the testing of materiels and procedures when I get them developed.

[chilly1]

Originally posted by DaBit
Those rods have different masses, and thus, they need a different amount of BTU's (or kJ's, whatever you prefer) to get heated to the same temperature. Thus, you are basically testing heat capacity here instead of heat conduction.

Here is how I would test it. I would take the rods, insulate them, make one side always the same temperature, and apply a fixed amount of heat to the other side.

Then, the temperature the 'heated' end of the rod will rise to is a measure of heat conductivity. The more conductive, the lower the temperature.

To keep the other end at a fixed temperature, I would dump a nice heatsink into a bath of ice water, and firmly clamp/screw the rod to that heatsink.

To apply heat I would use a power resisitor.


I think aluminium blocks would perform slightly less than copper, but i doubt the difference is very large. I have used alu for waterblocks, and they worked fine.

I am going to use a known sink and source for the material being tested. If I use a stable and known temperature and insulate all the materials, keep the surrounding temperature the same and use the same cubic area and square surface, I will be able to tell which material conducts heat better by The sink temperatures and their rise over time.

[Conflict]

Originally posted by chilly1
I am going to use a known sink and source for the material being tested. If I use a stable and known temperature and insulate all the materials, keep the surrounding temperature the same and use the same cubic area and square surface, I will be able to tell which material conducts heat better by The sink temperatures and their rise over time.

Hey, had some information to tell you...sign into ICQ or AIM one of these nights!!!!

[gokickrocks]

you should just look at the diffusivity of each metal...

Al (aluminum) - 97.1
Cu (copper) - 117
Ag (silver) - 174

...copper should perform better than aluminum

[nas]

hmmm 7.5mm base ? that pritty alot , but then again.. alum transfer heat faster .. :o

[chilly1]

Originally posted by nas
hmmm 7.5mm base ? that pritty alot , but then again.. alum transfer heat faster .. :o
needed to have as much internal volum as possiable for the evaporator and heat transfer surfaces.... Fits inside the bolt pattern for both AMD and INtel

[chilly1]

Originally posted by gokickrocks
you should just look at the diffusivity of each metal...

Al (aluminum) - 97.1
Cu (copper) - 117
Ag (silver) - 174

...copper should perform better than aluminum
How about gold plating on the aluminum?

[]JR[]

Or you could go to webelements and look at the thermal conductivity?

http://www.webelements.com/webelements/properties/media/tables/cityscape-x/thermal-conductivity.jpg

the highest three peaks are silver copper and aluminium

or

430 /W m-1 K-1 silver
400 /W m-1 K-1 copper
235 /W m-1 K-1 aluminium

]JR[

[nas]

how about a silver base plate with the rest of the block alum ?
and how would you connect the plate to the block ? so there will be the best contact possible ? (brazing ?)

[berkut]

Silver is not worth it

[gokickrocks]

Originally posted by chilly1
How about gold plating on the aluminum?

dont know the answer to that, since there are a lot variables

[Russell_hq]

How about sand blasting the inside? This will give you more surface area and increase the roughness which should increase the rate of heat transfer.

[nas]

Originally posted by berkut
Silver is not worth it

why is that ? i have a big 3mm silver board that i can cut it from..
i thought the silver can spread the heat better to the all plate and the block will have biger and better contact area

[JToolz]

Originally posted by nas
why is that ? i have a big 3mm silver board that i can cut it from..
i thought the silver can spread the heat better to the all plate and the block will have biger and better contact area

I think Berkut is referring to cost of silver.

[chilly1]

News.... Aluminum can be plated ....but.... you need to coat it with nickel first which would defeat the purpose of gold plating in the first place, However the copper can be electroplated with gold and the next block I do in copper will be tig welded and electroplated.

[PyroTeknik]

dang a gold plated block.... to bad its gotta be insulated thats some serious bling bling....

[chilly1]

20 us to plate a 2x2 block...

[Gary Lloyd]

I'm thinking gold is less conductive than copper. If you are going to plate the block face, it should be with silver, not gold.

[chilly1]

Maybe still trying to find stats on thermal conductivity of plated surfaces...

[]JR[]

gold is worse than copper, look on webelements.

Its also fairly well known that plating really does nothing for you, a few microns on silver wont help one iota, cpu heatsink manafacturers just do it for a gimmick.

]JR[

[Russell_hq]

Yeah, what you want is a nice slab of diamond the same size as your evap base and put that between the evap and the core. That will spread the heat somewhat better :D Now all you need to do is find yourself a slab of diamond and you will have some real Bling Bling there :D

[DaBit]

gold plating has one advantage: you get rid of the oxide layer on the alu/copper. And I think that a few microns of gold conducts heat better then a few microns of oxide.

Trying is the only way to know for sure..

[nas]

any one have spare golden tooth for the test ? :P

anyway... if i do wona connect an silver/gold plat .. how do i do it ? braze them with flux ? there should be smt to make contact between them

[Popcicle]

This Gold/Copper *debate* has all been worked out

with much testing by Cathar (of LRWW and Cascade

waterblock fame). Cathar is the waterblock Guru.

His research and testing results can be taken to the bank.

Cathar does build an SS Cascade that has a silver base

and gives a slight temperature drop when compared to

the copper base version Cascade. Price is the killer.

[PMM]

Is this not just the case of Aluminium being a better radiator of
heat than copper thats a better conductor ?

I always thought copper was the metal for moving heat from
place to place and aluminium to give up and radiate heat away
from itself.

Seems logical to me that aluminium would be the better for
this purpose ans possible why such results have been seen.

[sky]

Originally posted by Russell_hq
Yeah, what you want is a nice slab of diamond the same size as your evap base and put that between the evap and the core. That will spread the heat somewhat better :D Now all you need to do is find yourself a slab of diamond and you will have some real Bling Bling there :D

a diamond or some diamond type carbon proably won't do as much good as one would think. for all i know these carbonbased materials are very good spreading heat, but too bad it's only in 2 dimensions. what that implies is that it may spread heat to it's sides - serving as some kind of hs - but not vertically to the cooling device. so it's bascially worthless or at least the effect will be close to non-existant compared to maybe silverplating, ect.
there's been some very thin diamondized (tm) ;) carbonsheets that were offered as replacement for thermal paste - reusable and all, but they just didn't perform as expected. sure they did give an ever so slight improvement, but nothing compared to what was expected having its thermal conductivity in mind (and don't forget it's pretty pricey).
and the thing with these carbon bits is that you'd have to make sure you get the block align just right while manufacturing it, so it actually works in the two dimensions needed for our purpose.. and i guess that will make it even more expensive.

[lalPOOO]

Originally posted by Popcicle
This Gold/Copper *debate* has all been worked out

with much testing by Cathar (of LRWW and Cascade

waterblock fame). Cathar is the waterblock Guru.

His research and testing results can be taken to the bank.

Cathar does build an SS Cascade that has a silver base

and gives a slight temperature drop when compared to

the copper base version Cascade. Price is the killer.

But the difference between that and what someone suggested above is that the entire bottom of cathars block was silver. What was suggested above was a silver plate on the bottom of the block, which imho would be pretty much useless because of the layer of 'whatever' connecting it to the block, and the increased distance between the bottom of the block and the heat source.

[chilly1]

Originally I suguested only a molecule thin layer of silver or gold and I wondered what differance it would make as it surface to surface would conduct heat faster.... But aluminum cannot be electroplated.

[]JR[]

Originally posted by sky
a diamond or some diamond type carbon proably won't do as much good as one would think. for all i know these carbonbased materials are very good spreading heat, but too bad it's only in 2 dimensions. what that implies is that it may spread heat to it's sides - serving as some kind of hs - but not vertically to the cooling device. so it's bascially worthless or at least the effect will be close to non-existant compared to maybe silverplating, ect.
there's been some very thin diamondized (tm) ;) carbonsheets that were offered as replacement for thermal paste - reusable and all, but they just didn't perform as expected. sure they did give an ever so slight improvement, but nothing compared to what was expected having its thermal conductivity in mind (and don't forget it's pretty pricey).
and the thing with these carbon bits is that you'd have to make sure you get the block align just right while manufacturing it, so it actually works in the two dimensions needed for our purpose.. and i guess that will make it even more expensive.

What physical process allows heat to travel in x and y but not z exactly? This I got to hear ;)

The reason diamond is the best thermal conductor is becasue the lattice which propgates the vibrations is perfect (there are no free electrons contributing here either).

Agreed aligining the lattice directly above and perpendicular to the heat source wouldnt be particularly effective, cleaving the lattice diagonally would work extremely efficently. Likewise with you previously mentioned 'diamond type carbon' that would be carbon then since thats all it is :rolleyes: but seriously a bundle of carbon nanotubes arranged as set of pipes vertically on the core, such that the cross section is perpendicular to the core surface would almost make the best heatsink possible.

The main problem with diamonds is theyre pretty hard, so increasing the surface area would take a *lot* of machining with a diamond saw.

When it comes to this kind of technology, cost is no issue to the military, or nasa, both of which are developing nanotube type heatsinks and pipes. Which inevitably means they will be main stream eventually.

]JR[

[chilly1]

How about buckminsterfullrene?

[PMM]

I'm sure Diamond was considered bad in terms of thermal management.

check out Graftech, they are at the cutting edge of thermal
management with some very insteresting carbon based products
inc Heatsinks / TIMs

Some of there PDFs can make for interesting reading.

http://www.graftech.com/GrafTech/Industry+Portals/Electronic+Thermal+Management.htm

[]JR[]

Originally posted by chilly1
How about buckminsterfullrene?

A carbon nanotube is an elongated bucky ball :)

]JR[

[chilly1]

Do you think it may absorb heat better?

Or graphite foil? Is this what is on the cryo cooler's heat absorbers?

[]JR[]

At the temperatures were talking about you need to pic your TIM (thermal interface material) fairly carefully.

Requirements for a good TIM at any temerature would be...

Low Thermal mass
High Thermal conductivity
High Degree of order (or low degree of entropy)
Strong Bond between TIM molocules
Strong Bond between TIM and ajoining surfaces

Low thermal mass is easy, just use the smallest amount of TIM material possible
High thermal conductivity and High degree of order are closely linked, achieved by crystaline structures, which is due to strong bonds between molocules
Strong bond between ajoining surfaces, i.e. evaporator and die, connected by TIM.

Ideally you want a gel like substance that etches the surface of both die and evaporator and then bonds, and sets into a crystalline form.

The physics is good, but my chemistry aint, so how you implement this i dont know.

]JR[

[DaBit]

Originally posted by chilly1
Do you think it may absorb heat better?

Or graphite foil? Is this what is on the cryo cooler's heat absorbers?

Cryocoolers often use a thin sheet of indium (a very soft metal) as TIM.

[nas]

anyone thought about block from copper/silver or mixture ? how well will it work ? lets say 80-20 copper-silver

[chilly1]

I got some of that. You think it will improve heat transfer?

[sky]

Originally posted by ]JR[
What physical process allows heat to travel in x and y but not z exactly? This I got to hear ;)

The reason diamond is the best thermal conductor is becasue the lattice which propgates the vibrations is perfect (there are no free electrons contributing here either).

Agreed aligining the lattice directly above and perpendicular to the heat source wouldnt be particularly effective, cleaving the lattice diagonally would work extremely efficently. Likewise with you previously mentioned 'diamond type carbon' that would be carbon then since thats all it is :rolleyes: but seriously a bundle of carbon nanotubes arranged as set of pipes vertically on the core, such that the cross section is perpendicular to the core surface would almost make the best heatsink possible.

The main problem with diamonds is theyre pretty hard, so increasing the surface area would take a *lot* of machining with a diamond saw.

When it comes to this kind of technology, cost is no issue to the military, or nasa, both of which are developing nanotube type heatsinks and pipes. Which inevitably means they will be main stream eventually.

]JR[

a well, i guess i didn't put it quite right ;). i didn't mean to say that the heat is only transported in two dimensions and not in the third. of course it is transported there as well, just that the total amount moving in that dimension is less than the others. and that's no special process but just the way the structure of that material influences the transportation of heat. and yup, something like those nanotubes would quite possibly make one hell of a heatsink - but at what price?
even if nasa and what other cutting edge researches were done developping (commercial) heatsinks like that, we'd still be a few years shy of a reasonable massmarket-priced product - if ever. but i guess it won't get much better than that..


nas, what good would that do? silver and copper just aren't that far apart, so the results would probably be within the margin of error. in my opinion it would be better to go all silver or just stay with copper - at least regarding the bottom plate - the remaining part of an evap could very well be copper or anything else.


pmm, i think the tim i was talking about earlier ("diamondized carbonsheets") was made by graftech. some guy in a german board had them sent him a sample that he tested later on. his results weren't overwhelming, tho mostly ok.. from that tests rose a discussion about tims in general and carbon-based blocks, ect... the results were roughly what i tried to put in my last post (2 dimensions stuff, ect., see above)

[nas]

but since silver conduct heat better then silver , a mixture would improve the heat conductivty , no ?

[sky]

yes, but only slightly... see on page 2 there's a list of the thermal conductivity of the 3 materials.
silver tops at 430, copper at 400 and aluminium somewhere around 235 (iirc). so just do a quick calculation of what 80% copper and 20% silver would give you with that figures. sure it will increase, but by very small amounts. your 80-20 mixture would give you a - theoretical - conductivity of 406. not that big a rise - and as i said you might not actually notice it due to the margin of error with each new measurement... the idea is fine, but the real world effect might be too small - that's all i'm saying ;)

[PMM]

Originally posted by sky
pmm, i think the tim i was talking about earlier ("diamondized carbonsheets") was made by graftech. some guy in a german board had them sent him a sample that he tested later on. his results weren't overwhelming, tho mostly ok.. from that tests rose a discussion about tims in general and carbon-based blocks, ect... the results were roughly what i tried to put in my last post (2 dimensions stuff, ect., see above) [/B]

Actually got some Graftech Grahite TIM pads here for testing at some point, time permitting 16W/m.K through-plane 120W/m.K
In-plane.

I don't think enough pressure can be made without a custom
mounting and I think thats the key to this stuff(Pressure)

http://brittech.co.uk/graftech.jpg
http://brittech.co.uk/egraf2.jpg
http://brittech.co.uk/egraf3.jpg

[8-ball]

Originally posted by PMM
Is this not just the case of Aluminium being a better radiator of
heat than copper thats a better conductor ?

I always thought copper was the metal for moving heat from
place to place and aluminium to give up and radiate heat away
from itself.

Seems logical to me that aluminium would be the better for
this purpose ans possible why such results have been seen.

It is quite true that aluminium radiates thermal energy better than copper, but this piece of information, along with the "HSF manufacturer use copper bases and aluminium fins" argument have been used to promote the ability of aluminium to "give up heat" better.

However, as I mentioned, aluminium "radiates" heat better. But in the case of both water cooling and phase change, the thermal energy is dissipated largely by convective heat transfer in water cooling, and, I would imagine, a mix of convective heat transfer and conduction in phase change applications. Both of these modes of transfer of thermal energy from the dissipating surface into the coolant are entirely independent of the material of construction, and yet, account for the vast majority of thermal energy transfer.

Put simply aluminium's better radiating capabilities give rise to little or no advantage in the transfer of thermal energy from the material to the coolant while severely hampering the conduction of thermal energy from the chip to the surface in contact with the coolant.

Despite what you have read / been told, the ONLY reasons that heat sink manufacturers use aluminium in preference over copper for the fins are as follows:

1. Cheaper
2. Cheaper
3. Cheaper
4. Much lighter
5. Much easier to process than copper - hence, cheaper
6. Cheaper

The only scenario where aluminium has a distinct advantage over copper is when the amount of thermal enegy transferred by radiation is more significant than by other modes of transfer, and this only really happens when the radiating surface is in a vacuum, explaining why, along with the reduced weight, almost all heat dissipating units in satellites are aluminium.

Hope that makes sense

8-ball

[8-ball]

Originally posted by sky
yes, but only slightly... see on page 2 there's a list of the thermal conductivity of the 3 materials.
silver tops at 430, copper at 400 and aluminium somewhere around 235 (iirc). so just do a quick calculation of what 80% copper and 20% silver would give you with that figures. sure it will increase, but by very small amounts. your 80-20 mixture would give you a - theoretical - conductivity of 406. not that big a rise - and as i said you might not actually notice it due to the margin of error with each new measurement... the idea is fine, but the real world effect might be too small - that's all i'm saying ;)

Not true.

It is not a linear relationship.

Adding small amounts of silver to pure copper, even though the silver is a better conductor of heat, actually reduce thermal conductivities.

The silver acts as an impurity distorting the crystal lattice of the copper matrix reducing it's ability to conduct heat.

It is very rare, in fact I have never seen an example, of an alloy exhibiting greater thermal conductivity than it's major component in pure form.

8-ball

[chilly1]

Originally posted by 8-ball


The only scenario where aluminium has a distinct advantage over copper is when the amount of thermal enegy transferred by radiation is more significant than by other modes of transfer, and this only really happens when the radiating surface is in a vacuum, explaining why, along with the reduced weight, almost all heat dissipating units in satellites are aluminium.

Hope that makes sense

8-ball
Almost all of the phase change systems run in a vacume. Aluminum is still untested as an internal radiating surface in an evaporator in computer coolers, It is used extensively in the refrigeration industry for condenser and evaporator fins. It's manafacture is more costly. Carrier corporation did make aluminum coils for evaporators and condensers in the 1980's but found that external factors such as acidic organic compounds building on their surfaces would cause too much corrosion. These airconditioners were more effecient than their simmilary sized counterparts.

[DaBit]

Originally posted by 8-ball
[B]Not true.

It is not a linear relationship.

Adding small amounts of silver to pure copper, even though the silver is a better conductor of heat, actually reduce thermal conductivities.

There exists an alloy called 'CuSil', which is exact that: an alloy of copper and silver. This alloy features a higher thermal conductivity than both copper and silver.

[8-ball]

Originally posted by DaBit
There exists an alloy called 'CuSil', which is exact that: an alloy of copper and silver. This alloy features a higher thermal conductivity than both copper and silver.

If I remmeber correctly, the results published by the manufacturer of Cusil (WESGO metals) were found to have been conducted 30 years ago, with very inaccurate equipment. The figures weere retested (By WESGO and subsequently updated on their website (http://www.wesgometals.com/physdata.html)) and it was found that CuSil has a lower conductivity than copper, as would be predicted. It's actually 371W/m/K

Hope that clears this up.

8-ball

[DaBit]

Finally someone who knows the facts about this mystery alloy. Thanks for clearing this up :)

[8-ball]

It just goes to show how quickly scientific results can be accepted should they show what everyone is wanting to see.

It almost becomes a harder job to disprove what "must" now be "true".

I guess it happens everywhere.

8-ball

[DaBit]

I'm a quite stubborn guy, because of that. Don't take everything for granted.

[8-ball]

When I first heard about CuSil, I was in my second year studying Metallurgy and the Science of Materials at Oxford, and I still fell for it, though not without asking a few questions. People still managed to produce "the answers" mind.

I learnt a lot about trusting scientific data which hasn't been reproduced by independent sources. Having nearly finished my course now, I can see why it should never work. I just hadn't covered the relevent stuff at the time.

8-ball

[sky]

Originally posted by 8-ball

Originally posted by sky
yes, but only slightly... see on page 2 there's a list of the thermal conductivity of the 3 materials.
silver tops at 430, copper at 400 and aluminium somewhere around 235 (iirc). so just do a quick calculation of what 80% copper and 20% silver would give you with that figures. sure it will increase, but by very small amounts. your 80-20 mixture would give you a - theoretical - conductivity of 406. not that big a rise - and as i said you might not actually notice it due to the margin of error with each new measurement... the idea is fine, but the real world effect might be too small - that's all i'm saying
Not true.

It is not a linear relationship.

Adding small amounts of silver to pure copper, even though the silver is a better conductor of heat, actually reduce thermal conductivities.

The silver acts as an impurity distorting the crystal lattice of the copper matrix reducing it's ability to conduct heat.

It is very rare, in fact I have never seen an example, of an alloy exhibiting greater thermal conductivity than it's major component in pure form.

8-ball

thanks for clearing that up!

oh, i didn't want to make it appear as a given fact or state it as one - just seen from a theoretical point of view, without having the knowledge you have.
if you can put it into laymens terms, i'd like to know why it is that the combination of two materials gives worse results - so to speak - as the indviduals. if it can't easily be explained like that, don't bother - might scare away people ;)

[8-ball]

Sky,

I will assume that you you know nothing about metals. (this may not be true, but it gives me somewhere to start;))

Here is a quote I found while checking that I did have my facts straight. (annoyingly, all of my notes are at home as I am now specialising in the chemical analysis of biological materials)


The amount of reduction due to alloying does not depend on conductivity or any other bulk property of the alloying element, but only on the effect that the particular foreign atoms have on the copper lattice.

Essentially, metals, unlike non metals, don't really form specific bonds between atoms (like you would see in diamond for example) In a single crystal of copper the atoms arrange into a face-centred-cubic arrangement of unit cells. (the unit cell is a repeating unit which makes up the crystal structure. Each copper unit cell is represented by 4 atoms, though this is a little hard to describe.

FCC is a close packed crystal structure, meaning, the atoms are as close to one another as possible while maintaining an ordered structure. The highest energy electrons from each atom become dissociated and form a "sea of electrons". It is this sea of electrons in the conduction band which both hold the material together and provide the medium for the transport of thermal energy and electrical conductivity. At a given temperature, the electrical and thermal conductivity of a material are in direct relation, but as the temperature increases the thermal conductivity increases while the electrical conductivity decreases. Essentially, at higher temperatures, the free electrons become more agitated, a good thing for thermal conductivity, but the electrical conductivity suffers as random collisions between electrons divert them off their required direction for electrical conductivity.

Anyway, why do impurities affect conductivity?

Any defects in the copper lattice, such as an atom missing from a lattice site (a vacancy), or a line defect where a half plane of atoms is missing (a dislocation), a grain boundary, or even an impurity atom can disrupt the lattice. With regard to impurity atoms, there are two types, depending on the size of the impurity relative to the size of the matrix element. If it is a similar size, then it will take the place of one of the matrix elements, however, all atoms are of varying sizes, so it will not be a perfect match. The alternative, is an interstitial, where a very small (relative to the matrix atoms) impurity will reside in the gaps between matrix atoms. Think of three spheres arranged into a triangle with another balance on top in the middle, all the same size. There will be a small gap right in the middle, and this is where you will find interstitials.

However, due to size differences they will create a strain field around the defect, as will the other defects I mentioned. These defects can impede the motion of free electrons, providing a low energy site for them to reside. So the more defects are present, the harder it is for the electron to provide conduction. (this is why modern semiconductors have rediculously low defect densities. A single dislocation in close proximity to a silicon transistor can prevent it from operating.)

As an example, consider oxygen free copper, of a very high purity (min 99.99%)

now add 0.03% silver. The conductivity will drop by around 1%

That's quite a drop for such a small amount of impurity.

On another topic, the subject of the other types of defects, namely vacancies, dislocations and grain boundaries has been brought up before. Cold worked copper, the typer preferable for machining has a lower conductivity than annealed, soft copper, due to the vastly higher proportion of defects introduced by cold working. Worse still, the machining process itself intorduces further defects.

It has been suggested that annealing could be used to reduce the defect concentration of a finished waterblock/evaporator, thus returning it to it's soft, annealed, low defect density state, with a slightly higher thermal conductivity.

However, what was not ascertained was the effect of annealing on the actual shape of the finished product. IE, would the flat base stay flat. It's worth looking into though.

8-ball

[8-ball]

Originally posted by ]JR[
What physical process allows heat to travel in x and y but not z exactly? This I got to hear ;)

I think he is probably referring to pyrolitic graphite rather than any form of diamond.

It is formed by a very carefully controlled vapor deposition technique. It has conductivities going up to thousands of W/m/k in the horizontal plane but often less than a hundred W/m/k in the vertical direction.

It is down to the layerd microsturcture of the material.

8-ball