Better Performance???

[R4fo]

Waterblocks made from copper, aluminium or bronze, which of this has a better performance/heat transfer???

[pauldenton]

Originally Posted by R4fo

Waterblocks made from copper, aluminium or bronze, which of this has a better performance/heat transfer???

copper - only silver is better iirc

[Butcher]

And diamond.
Some thermal conductivities:

Material            Conductivity (W/m-K)
Copper C110         391.1
Brass C260          121.2
Brass C270          116.0
Bronze C220         189.0
Ph Bronze C510       69.2
Al Bronze C614       67.5
Mn Bronze C674      100.4
Nickel silver C752   32.9
Al Alloy 1060-O     234.0
Al Alloy 6061-T6    166.9
Al Alloy 6063-T6    201.0
Silver (pure)       419.0
Silver (sterling)   410.0
Gold (pure)         301.0
Gold (commercial)   300.0
Diamond(natural)   2000.0
Diamond(synthetic) 2000.0

The most commonly used (by enthusiasts) of each material are:
Copper C110, Brass C260, Bronze C220, Al 6063-T6/6061-T6.

[|kbn|]

didnt relise brass and bronze were used (except for the barbs?)
alu is als not nice to to corosion, though its easy to machine...
silver is very easy to machine aswell but expensive and wastefull to get in large quantitys... (goto factorys that use silver powder in there products and get any scrap they dont want, and get them to melt it into a useable shape. wont cost much that way.. where my dad works they give away a few £100 worth of matirals for about £30 becuase its scrap and they dont need it....

[Butcher]

Brass is common for barbs and various parts of heatcores. I've also seen it used for blocks as it's cheaper and easier to work than copper and has good compatibility with copper. Al corrodes less than copper in an all-Al system, however most people have brass or copper in their loop somewhere so it is less often used. Also Al has less conductity though it's much lighter and cheaper.
For maximum corrosion resistance gold would be the thing to go for, however cost is generally prohibitive.
I don't recall anyone ever using bronze, it's not that cheap and it's quite a poor material to use really. However it was mentioned in this thread so I included it.

[|kbn|]

btw, would you know the material used for heatspreaders? i havent got any from a p4 chip, but i have several k6-2 chips, and im confused as to which material it is. i assume it would be one with good heat conductiviity (or what would be the point in using them..). the metal is not ferrous (well it doesnt stick to magnets, that would rule out a few more) and at the thichnes about 1.5mm at most, it is very hard to bend (by hand :o )
not really sure how to find out what it is (btw it looks like steel/aluminium in colour but i doubt it is either as i said above). maybe its an alloy, but dont they usually have worse propertys than the base metals?
i you wonder why i ask, im am thinking of using one in a direct die/heatspreader cooling block

[Cathar]

Originally Posted by |kbn|

btw, would you know the material used for heatspreaders?

Heatspreaders are typically copper with a very thin nickel plating layer which gives the silvery appearance. Nickel does not tarnish like copper does, which is why the IHS are plated with it. Due to the super-thin nickel plated layer this does not adversely affect thermal conductivity in any significant fashion.

[|kbn|]

i dont think its copper?... you might be right on the nickel plating though.. to remove the heatspreader i used a junior hacksaw to cut off two of the corners so it would lever off easly - was easyer than freezing it like most sites suggested
the metal used is shiny silvery colour all the way through.. though that might be what happens to plated copper? i dont know...

nickel plated aluminuim?

[Cathar]

Originally Posted by |kbn|

i dont think its copper

Sand the surface of the IHS back (remove the nickel plating) and you'll very quickly see copper underneath. I've done this. Hundreds of others have done this. It is nickel plated copper.

At the edges where it attaches to the packaging there may be something else going on, but that's largely unrelated to the thermal function of the IHS.

[|kbn|]

the heatspreader doesnt sand very well atall (meaning it didnt scratch it atall), so instead i done two things. firstly i drilled all the way through the center of the IHS, and using i file i removed the surface. both left it shiny silvery in colour. this was from a k6-2 400. i cant make out the stepping as thats where i drilled
that heatspreader was from a chip i recived second hand DOA so i was unsure if it was a real or faked chip?, but i had a working k6-2 350 AFR stepping which i had removed the IHS, so i done the same to that, with the same results. i can take picture of it if you like, im quite sure its not copper, unless its an alloy of some kind? i will ask my dad as he works with lots of metals in R&D so he might know..

<edit>my dad thinks its some kind of alu alloy with probably 5% copper at most, explains the shape of it aswell as it makes it stronger. i expect p4 and amd64 heatspreaders are as you said, copper with nickel plating, as alu wouldnt be much good with them.. </edit>

[Cathar]

Sorry, I was referring to modern CPU's (P4's/AMD64's). I have no idea what heat-spreaders from CPU's dating back 5 years were made with. It could very well be aluminium as the CPU's back then generally exhibited fairly low heat densities.

[Butcher]

On an old chip like a K6 it's probably just annealed Al of some sort. I doubt it's alloyed with copper.

[|kbn|]

makes sence...
so theres little point in keeping the heatspreaders on chips atall then - all there for is protecting cores from bad mountings so they dont get chiped etc.. oh and keeping warrentys

[Butcher]

Bear in mind old K6 chips aren't flipchips, so the core is actually mounted on the underside of the ceramic package...

[superart]

Originally Posted by Butcher

And diamond.
Some thermal conductivities:

Material            Conductivity (W/m-K)
Copper C110         391.1
Brass C260          121.2
Brass C270          116.0
Bronze C220         189.0
Ph Bronze C510       69.2
Al Bronze C614       67.5
Mn Bronze C674      100.4
Nickel silver C752   32.9
Al Alloy 1060-O     234.0
Al Alloy 6061-T6    166.9
Al Alloy 6063-T6    201.0
Silver (pure)       419.0
Silver (sterling)   410.0
Gold (pure)         301.0
Gold (commercial)   300.0
Diamond(natural)   2000.0
Diamond(synthetic) 2000.0

hmm, so your saying that diamond is like 500% better at transferring heat than copper?

If so, what do you think the queen of england would say if we asked her to burrow the hope diamond for a couple days? You think she'd be a good sport and let us use it for a while?


no but seriously, I figured materials like diamond would have bad heat transfer properties, since their not metals. Or am I mistaken in thinking that only metals can have good heat transfer properties?

[Cathar]

Originally Posted by superart

Or am I mistaken in thinking that only metals can have good heat transfer properties?

Many materials that have a good/strong lattice structure are good conductors of heat.

Silicon, for example, is almost as good as aluminium.

Various types of graphite have excellent thermal conductance (up to 5x better than silver) along the plane that the lattice structure is strongest (and somewhat poor between planes). Diamond has a super-strong/cohesive lattice structure in all 3 dimensions. It's been hard to quantify diamond, but some sources place isotopically pure diamond's thermal conductivity up around 12x that of silver. (~5000 W/m-K)

[Butcher]

Diamond's superiority as a thermal conductor is matched only by it's huge cost.

[|kbn|]

rare things usually cost more
industrial diamond though, afaik is cheap (my dad works with it, and i think i have one somewhere - its about 1mm big!). its made from crushed up bigger diamonds. if crushed up more then it may be good to use for heatsinks/pastes etc.. still expensive though..
i expect those graphite types are very expensive too, as they proably arnt around in large quanitiys, or we would be useing them by now

[quote=Butcher]Bear in mind old K6 chips aren't flipchips, so the core is actually mounted on the underside of the ceramic package...[/qupte]

can you explain this please, what does flipchip mean, thoguht it was just a marketing name, or somehting about cpu's using slots instead of sockets...
the core on my k6-2's doesnt go all the way through, so you didnt mean that. maybe you mean the core AND the pins are flipped round, so whats left the right way up? what does that have to do with heatspreaders?

[Shwag]

Flip-Chip is a slang kinda term for Zero Insertion Force.
Flip the lever.
Put the chip in.
Flip it back.

[Althornin]

Originally Posted by Shwag

Flip-Chip is a slang kinda term for Zero Insertion Force.
Flip the lever.
Put the chip in.
Flip it back.

ummm, no?
Flip chip is short for FC-PGA - Flip Chip Pin Grid Array, which refers to the package of the CPU and the fact that the chip is "flipped" from the old traditional place under the ceramic/organic substrate package to on top of it. The PGA part is kinda odvious - it means the chip has a grid of pins rather than a grid of say, balls, like BGA would mean.

anyways, you could have been joking, in which case, sorry.

[msv]

Aren´t diamonds quite bitchy to work with?
Sounds like a large investment in tools and machinery to make a diamond WB.
regards
Mikael S.

Edit: Had to correct the bad grammars

[Butcher]

Althornin is correct, see these pics:



Notice how the core is "flipped" between the two versions of the chip.

On diamond - yes it's really really hard to work - you'd basically have to grow the diamond synthetically into the shape you wanted.
As for pastes, people have tried, they are less efficient than artic silver or such. One theory for this is that the silver particles in AS are all regularly shaped (discs I believe) so they stack together well giving good inter-particle contact. The particles of diamond tend to be more irregular giving poor inter-particle contact. So while you get good conductivity across each particle, as they only have a few atoms touching each other overall performance is low.

[R4fo]

I have heard that teflon has a high thermal conductivity too and very resistance to corrosion, this material is very used in industries, but can it be use on WB's??

[|kbn|]

my k6-2 chip has nothing on the underside apart from loads of pins and the remains of a warrenty vioding sticker
the top has the core and a few capacitorrs around it. that would make it a flipchip then? there is nothing on the underside like the top right pic....

[superart]

No, i think the k6 chips were flip-chips. That was the whole point of using the IHS. The pentiums, on the other hand, were not flip chips.