I've noticed that with cupper heat sinks, many times the fins and attached or soldered to the base. This seems incredibly stupid to me. Do they have a reason not to make it from one piece? I honestly think the fins are only guiding the air to the base and that's where the cooling is happening sometimes. Doesn't it make sense to just make fins a little wider (so you get more cross-section for the heat to transfer) and make it from the same block as the base (not unlike the cheap stock aluminum ones)? The stupides ones I've seen are the orb-type ones where the fins were held with 2 screws to the rest of the the heat sink? Is it just asthetic? Is it me?
I know the Zalman (flower) is a lottle different because I think they just pressed a lot of sheets together and cut it on one end (contact area with the CPU) and spread apart the rest. But when I feel for temps, the outside of the flower is cold and the core is hot (not scientific measurement but there is a relatively big temperature difference). Again wouldn't thickker fins be more useful here.
I doubt anybody needs to have a loud heatsink fan with a little better product designing.
stupid question about heat sink manufacturing (little rant)
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
Well on a superficial level I think that Al is a lot more machinable than copper (I know cause I HATE tapping Cu, unless I have the right tap forming bit). This being said, Al alloys tend to be a lot harder than copper. Transporting shaped Al requires much less packing than shaped Cu. Making the HS from parts made from sheet and machined materials is much cheaper than precision molding (most non-brass copper alloys have very poor colding working characteristics or have excessive creep problems).
Cu is much more expensive than Al. While the decrease in radiated heat to cost ratio is not economical, every little bit counts when it comes to something like a CPU.
By constructing HSes from assembled fins and bases does decrease the effective transport of heat. I remember seeing a single piece molded Cu Hs with bent fins, I wonder why I haven't seen more of these. The soldering material can go a long way to making heat transfer better, but EVERY single interface decreases the amount of heat transfered. You are right on when you talk about single piece HSes.
Al and (even more so Cu) have thermal conductivities that even in extreme cases do not show much temperature gradients. The fins on a Zalman flower only need to be as thick as they are to retain their mechanical (read as handling and installation) properties.
------- Lengthy read, unrelated to computers ---------
==================================
To illustrate my point, in my lab we are water cooling a 10W continuous wave laser. A 10W laser head will easily put out 300W of heat. We do light scattering and can have NO vibrations. Convective heat transfer (fans and heatsinks) do not move the heat we produce and the air flow against heat sinks causes too much vibration.
We water cool the sucker with computer parts =) The cross drilled Al heat exchanger (on 3 peltiers and 1 redudant, 4 modded enermax whisper PSUs) we use has house water on one side and have the laser on the other. It has 3 tiers of 40 3/16" diameter holes. The block weighs a @#$^ load. The water goes in 1 hole at 15C and comes out at 25C at a flow rate of about 380gal/hr. If you measure the block temperature at different places (I did this to check the heat profile of the laser) one finds that the temperature differs by about 50-70 mK. This value is well within bounds of convective transport in air.
Even with effective heat transfer between the block and the water, no temperature gradients are seen. In a system of metal and air, the convective transport is much less than that seen in watercooling. In this case the temperature gradient along the fins of a heatsink (say a Zalman) can be expected to be even smaller. The heat profile in the metal at a metal-metal interface is linear, whereas the heat profile at a metal-gas interface is exponential. These models are made with stagnant air, but it is obvious from molecular arguments that a denser phase is the preferred transfer medium.
This says nothing for fins that do not have a direct point of contact with the heat source (thermalright comes to mind). The dissipating power of these heatinks is highly dependent on the solder used. Generally in design, the number of interfaces is kept to a bare minimum as each interface (even between metal and metal) a good percentage (read as ~15% in copper-copper interface using Ni-Au solder) of the heat transfer capacity is lost at interfaces.
Hope that helps.
Cu is much more expensive than Al. While the decrease in radiated heat to cost ratio is not economical, every little bit counts when it comes to something like a CPU.
By constructing HSes from assembled fins and bases does decrease the effective transport of heat. I remember seeing a single piece molded Cu Hs with bent fins, I wonder why I haven't seen more of these. The soldering material can go a long way to making heat transfer better, but EVERY single interface decreases the amount of heat transfered. You are right on when you talk about single piece HSes.
Al and (even more so Cu) have thermal conductivities that even in extreme cases do not show much temperature gradients. The fins on a Zalman flower only need to be as thick as they are to retain their mechanical (read as handling and installation) properties.
------- Lengthy read, unrelated to computers ---------
==================================
To illustrate my point, in my lab we are water cooling a 10W continuous wave laser. A 10W laser head will easily put out 300W of heat. We do light scattering and can have NO vibrations. Convective heat transfer (fans and heatsinks) do not move the heat we produce and the air flow against heat sinks causes too much vibration.
We water cool the sucker with computer parts =) The cross drilled Al heat exchanger (on 3 peltiers and 1 redudant, 4 modded enermax whisper PSUs) we use has house water on one side and have the laser on the other. It has 3 tiers of 40 3/16" diameter holes. The block weighs a @#$^ load. The water goes in 1 hole at 15C and comes out at 25C at a flow rate of about 380gal/hr. If you measure the block temperature at different places (I did this to check the heat profile of the laser) one finds that the temperature differs by about 50-70 mK. This value is well within bounds of convective transport in air.
Even with effective heat transfer between the block and the water, no temperature gradients are seen. In a system of metal and air, the convective transport is much less than that seen in watercooling. In this case the temperature gradient along the fins of a heatsink (say a Zalman) can be expected to be even smaller. The heat profile in the metal at a metal-metal interface is linear, whereas the heat profile at a metal-gas interface is exponential. These models are made with stagnant air, but it is obvious from molecular arguments that a denser phase is the preferred transfer medium.
This says nothing for fins that do not have a direct point of contact with the heat source (thermalright comes to mind). The dissipating power of these heatinks is highly dependent on the solder used. Generally in design, the number of interfaces is kept to a bare minimum as each interface (even between metal and metal) a good percentage (read as ~15% in copper-copper interface using Ni-Au solder) of the heat transfer capacity is lost at interfaces.
Hope that helps.
-
powergyoza
- Patron of SPCR
- Posts: 543
- Joined: Mon Oct 21, 2002 1:01 am
- Location: Vancouver, BC, Canada
- Contact:
counterpt, I understand your concerns. Some manufacturers have tried to deal with the issue of eliminating those metal-metal junctions.
I can think of a few: Dynatron/Cooljag/Bitspower have resorted to a skiving manufacturing technique where the fins are literally shaven/chiseled from the metal base. From what I can tell, it works well, except that no one has produced an 80mm version. HS manufacturers, do you hear me? Forget 60/70mm heatsinks, 80mm is the minimum for today's CPUs!
I also have seen some fancy extrusions and milled HS that worked well too. Ever heard of the Canopus Firebird R7? It had the most excellent idea (before the thermalright AX-7/SLK-series, before the vantec aeroflow) arranging the fins (in the R7's case, pins) around a big chunka metal that mimicked a heatpipe. Spire has the same idea with their falconrock heatsink (a decent one i hear).
There is also this other company, ToCools that offers a milled Al-HS called the Novasonic that also uses the "heatpipe" technique with good results. Unfortunately, I don't think there's any availabilty outside of asia.
You can find a review of the HS's by dynatrons & ToCools at http://www.frostytech.com.
Sadly enough, except for maybe the Novasonic, none of the above in their current form perform as well as the high-end Thermalrights, Zalmans, Swiftechs & Alpha's at low fan speed.
I can think of a few: Dynatron/Cooljag/Bitspower have resorted to a skiving manufacturing technique where the fins are literally shaven/chiseled from the metal base. From what I can tell, it works well, except that no one has produced an 80mm version. HS manufacturers, do you hear me? Forget 60/70mm heatsinks, 80mm is the minimum for today's CPUs!
I also have seen some fancy extrusions and milled HS that worked well too. Ever heard of the Canopus Firebird R7? It had the most excellent idea (before the thermalright AX-7/SLK-series, before the vantec aeroflow) arranging the fins (in the R7's case, pins) around a big chunka metal that mimicked a heatpipe. Spire has the same idea with their falconrock heatsink (a decent one i hear).
There is also this other company, ToCools that offers a milled Al-HS called the Novasonic that also uses the "heatpipe" technique with good results. Unfortunately, I don't think there's any availabilty outside of asia.
You can find a review of the HS's by dynatrons & ToCools at http://www.frostytech.com.
Sadly enough, except for maybe the Novasonic, none of the above in their current form perform as well as the high-end Thermalrights, Zalmans, Swiftechs & Alpha's at low fan speed.
Thanks for the reply (I'm sorry about all the typos in my earlier message).B wrote:Al and (even more so Cu) have thermal conductivities that even in extreme cases do not show much temperature gradients. The fins on a Zalman flower only need to be as thick as they are to retain their mechanical (read as handling and installation) properties.
I'm not sure I agree with you completely. You can heat one end of aluminum foil and it wouldn't get hot on the other end and you get a gradient. So I would think there is a sweet spot for thickness of fins.
Thanks again for the replies.
-
MikeC
- Site Admin
- Posts: 12285
- Joined: Sun Aug 11, 2002 3:26 pm
- Location: Vancouver, BC, Canada
- Contact:
Interface losses notwithstanding, the proof is always in the pudding. The Thermalrights ALL have soldered joins and they have consistently ranked at the top for the past 6-12 mos. On the otherhand, both Swiftech and Alpahs use force-fit designs which seems to work well, too, and Zalman is only using pressure to hold their fins together. The skived one-piece copper Thermaltake 7+ is a powerful cooler, but the fin spacing almost demands a high pressure, high airflow fan for good performance. There have also been instances of great heatsinks undermined by a bad clip.
It just seems that design and implementation tend to overshadow other factors in HS performance, especially with low airflow.
It just seems that design and implementation tend to overshadow other factors in HS performance, especially with low airflow.
I'll make a philosophical arguement.MikeC wrote:Interface losses notwithstanding, the proof is always in the pudding. The Thermalrights ALL have soldered joins and they have consistently ranked at the top for the past 6-12 mos.
1. Just because it's the best, doesn't mean it can't be better.
2. If we just say this is good enough, we won't improve.
Well, I'm off to be a hypocrite and buy a thermalright....
Heatsink making
Well, maybe an "old" topic, but although it would be nice to have a single piece CU machined out for a HS, that would be expensive. I have the Zalman "flower" and like how efficient the design is. Almost no wasted material in the manufacturing process since they just stamp a whole bunch of the same fin and change the bend angle, then press them together between two pieces of aluminum extrusion. If you want to machine a single copper piece you have to start with a larger block and machine off about half of the material as scrap. Would be nice, but expensive.
Re: Heatsink making
Not really, in any metalworking maunfacturuer the trimmings are collected and recycled.fmah wrote:If you want to machine a single copper piece you have to start with a larger block and machine off about half of the material as scrap. Would be nice, but expensive.
-
Bluefront
- *Lifetime Patron*
- Posts: 5316
- Joined: Sat Jan 18, 2003 2:19 pm
- Location: St Louis (county) Missouri USA
Maybe I'm reading this wrong.....You want a heatsink machined or extruded from a single block of copper?
The heat transfer ability of the two metals differs quite a bit. Copper works best as the base metal with heat transfered into it by direct contact. Aluminum transfers heat better when used as a fin...air blowing over aluminum transfers heat better than air over copper. I think that's the reason many heatsinks are made this way.
Using pins such as an Alpha has always been the the best fin design. It has to do with the total surface area available for the airflow. So for a set size, a pin design has the largest area.
When I look at an Alpha, it combines the best overall features of any heatsink. Only way to beat it is with a larger sized heatsink....which may be dangerously heavy, and may not fit your case/m-board.
Automobile radiators are an example of this....they used to be big, heavy, and made out of copper. Now they're all aluminum, lighter and cool better than the old setup. Of course cost plays a part here also...heh.
The heat transfer ability of the two metals differs quite a bit. Copper works best as the base metal with heat transfered into it by direct contact. Aluminum transfers heat better when used as a fin...air blowing over aluminum transfers heat better than air over copper. I think that's the reason many heatsinks are made this way.
Using pins such as an Alpha has always been the the best fin design. It has to do with the total surface area available for the airflow. So for a set size, a pin design has the largest area.
When I look at an Alpha, it combines the best overall features of any heatsink. Only way to beat it is with a larger sized heatsink....which may be dangerously heavy, and may not fit your case/m-board.
Automobile radiators are an example of this....they used to be big, heavy, and made out of copper. Now they're all aluminum, lighter and cool better than the old setup. Of course cost plays a part here also...heh.
Nope. Completely totally wrong.Bluefront wrote:Copper works best as the base metal with heat transfered into it by direct contact. Aluminum transfers heat better when used as a fin...air blowing over aluminum transfers heat better than air over copper. I think that's the reason many heatsinks are made this way.
The idea that aluminum is somehow better at transferring heat to the air is a myth that gets repeated all too often. Thermal conductivity is thermal conductivity. There is no scientific property of materials that controls how quickly they give up their heat to the air. Well, that's not completely true, there is some radiation of heat as infrared, and that varies, but its effect on heatsink performance is nil.
Copper is used for just the contact patch on heatsinks with aluminum fins for one reason only. Cost. Copper is 4 times as expensive as a raw material thn aluminum, plus it's more difficult to machine, and the final product is more expensive to ship since it weighs more.
Look at it this way: Cheap heatsinks (like OEM's) use all aluminum. Medium price heatsinks use aluminum fins with a copper base. But the most expensive, highest performance sinks are all copper.
heatsink stuff
That's true, scrap is recycleable, but what you get from scrap is not going to be as much as the original cost of the stock material.
I don't think there is any copper extrusion that is thin walled like the thin fins on these heatsinks. Aluminum fins can get down to 1/16" but as they become longer, going that thin or any thinner is tough. The thickness of the copper fins on these heatsinks is less than that, nearly half that thick. Nearly impossible to extrude that kind of thickness, at least on normal machinery.
As far as copper versus aluminum, well the fins on copper can be thinner because the heat will transfer to the tip better than aluminum. If all things are equal, then the cross sectional area of the fin for copper needs to be only around 60% as that for aluminum fin since the copper thermal conductivity is about 50% greater.
However, if copper has a greater heat capacity than aluminum, the actual temperature rise of the copper will be a little less than the aluminum for a given amount of heat. And since the heat transfer is directly proportional to the temperature difference between "hot fin" and ambient air that makes some difference. But, I suspect the greater thermal conductivity is more noticable in copper's favor.
I don't think there is any copper extrusion that is thin walled like the thin fins on these heatsinks. Aluminum fins can get down to 1/16" but as they become longer, going that thin or any thinner is tough. The thickness of the copper fins on these heatsinks is less than that, nearly half that thick. Nearly impossible to extrude that kind of thickness, at least on normal machinery.
As far as copper versus aluminum, well the fins on copper can be thinner because the heat will transfer to the tip better than aluminum. If all things are equal, then the cross sectional area of the fin for copper needs to be only around 60% as that for aluminum fin since the copper thermal conductivity is about 50% greater.
However, if copper has a greater heat capacity than aluminum, the actual temperature rise of the copper will be a little less than the aluminum for a given amount of heat. And since the heat transfer is directly proportional to the temperature difference between "hot fin" and ambient air that makes some difference. But, I suspect the greater thermal conductivity is more noticable in copper's favor.
-
Bluefront
- *Lifetime Patron*
- Posts: 5316
- Joined: Sat Jan 18, 2003 2:19 pm
- Location: St Louis (county) Missouri USA
By saying the highest performance heatsinks are made out of copper, rather than aluminum, Rusty implies the material (copper) is the source of the high performance. I don't think so....
Marketing is at play here, "our copper heatsink is better than your otherwise identical aluminum heatsink". Or "Our aluminum case cools much better than your steel case". Look past the hype. Sure...a better designed copper heatsink cools better than a poorly designed aluminum one. So what?
Marketing is at play here, "our copper heatsink is better than your otherwise identical aluminum heatsink". Or "Our aluminum case cools much better than your steel case". Look past the hype. Sure...a better designed copper heatsink cools better than a poorly designed aluminum one. So what?
Copper is the source of the higher performance. If the marketing guys were going to hype a material even though it doesn't perform as well they would hype aluminum, or even steel heatsinks. There's no point in falsly hyping a material that costs the manufacturer 4 times as much. Where's the logic in that?
It is also true that the copper heatsinks tend to have better design. If the company is going to pay through the nose for the material and the machining, they are going to make sure that the resulting product makes as efficient a use of the stuff as they can.
Think about it.
If you can find me some scientific documentation to back your claim that aluminum is better for heatsinks I'll gladdly retract my statements. But you won't find any.
It is also true that the copper heatsinks tend to have better design. If the company is going to pay through the nose for the material and the machining, they are going to make sure that the resulting product makes as efficient a use of the stuff as they can.
Think about it.
If you can find me some scientific documentation to back your claim that aluminum is better for heatsinks I'll gladdly retract my statements. But you won't find any.
-
Bluefront
- *Lifetime Patron*
- Posts: 5316
- Joined: Sat Jan 18, 2003 2:19 pm
- Location: St Louis (county) Missouri USA
This is my idea of an almost perfect heatsink...add your choice of fan. It has a 4mm thick copper plate contacting the cpu, and aluminum pin fins. Would copper fins make this thing work any better? Doubtful....but with all the copper-hype going on, maybe some one would be more inclined to buy one.
http://www.busitek2000.com/forms/bsttes ... l8942.html
http://www.busitek2000.com/forms/bsttes ... l8942.html
Yes, the aluminum/copper transistion is a weak point.
And yes, the design would be even more efficient with all copper fins. But it would be heavier and more expensive.
Bluefront, as an example of how copper improves heatsink performance compare the Zalman flower coolers that come in both AlCu and Cu versions. For Example:
The CNPS6500B AlCu has a Surface area of 3400cm2 and aThermal resistance: 0.30C/W
The CNPS6500-Cu has a Surface area of 3400cm2 and a Thermal resistance: 0.28C/W
The all copper version is more efficient. Lower thermal efficiency with the same fan is the only measure of heatsink efficiency. There is no magical pixie dust that makes Aluminum fins or pins better than copper. They're not. They're used because they're cheaper and lighter, and that's it.
It's not hype, it's science. Until you understand the science keep your erroneous opinions to yourself. It only serves to confuse more people.
It's not that I don't think you should share your thoughts, and I'm not trying to flame you for your ignorance. But the internet is full of people who spout off like they know what they're talking about when they don't.
Now having said that, if you know of some information on this topic that I don't, please share. I'm always interested in being proven wrong.
And yes, the design would be even more efficient with all copper fins. But it would be heavier and more expensive.
Bluefront, as an example of how copper improves heatsink performance compare the Zalman flower coolers that come in both AlCu and Cu versions. For Example:
The CNPS6500B AlCu has a Surface area of 3400cm2 and aThermal resistance: 0.30C/W
The CNPS6500-Cu has a Surface area of 3400cm2 and a Thermal resistance: 0.28C/W
The all copper version is more efficient. Lower thermal efficiency with the same fan is the only measure of heatsink efficiency. There is no magical pixie dust that makes Aluminum fins or pins better than copper. They're not. They're used because they're cheaper and lighter, and that's it.
It's not hype, it's science. Until you understand the science keep your erroneous opinions to yourself. It only serves to confuse more people.
It's not that I don't think you should share your thoughts, and I'm not trying to flame you for your ignorance. But the internet is full of people who spout off like they know what they're talking about when they don't.
Now having said that, if you know of some information on this topic that I don't, please share. I'm always interested in being proven wrong.
-
Bluefront
- *Lifetime Patron*
- Posts: 5316
- Joined: Sat Jan 18, 2003 2:19 pm
- Location: St Louis (county) Missouri USA
From what I've heard about this heatsink.....the copper slug is a press fit into the aluminum. Then as you can see in the pictures, the bottom is machined flat, with cut-outs for the mounting points and the fan mounts.
Due to it's size and weight, it requires direct mounting to the board....using four studs, and spring loaded mount-screws, which maintain even pressure on the cpu. It won't come loose or fall off when transported.
I don't know if there is any thermal compound or adhesive between the copper and aluminum. Probably it's just a press-fit. This has been the #1 socket 478 heatsink for a year or more. I don't think you can beat it unless you're willing to go bigger/heavier.
Due to it's size and weight, it requires direct mounting to the board....using four studs, and spring loaded mount-screws, which maintain even pressure on the cpu. It won't come loose or fall off when transported.
I don't know if there is any thermal compound or adhesive between the copper and aluminum. Probably it's just a press-fit. This has been the #1 socket 478 heatsink for a year or more. I don't think you can beat it unless you're willing to go bigger/heavier.
-
Bluefront
- *Lifetime Patron*
- Posts: 5316
- Joined: Sat Jan 18, 2003 2:19 pm
- Location: St Louis (county) Missouri USA
Rusty....you're making a mistake when you assume results from your one comparison, apply equally to all heatsinks. Were the exact Alpha made in an all-copper version, and it truly tested better, I would be inclined to believe you.....but it's not and I don't.
If all-copper would make that heatsink better, I'm certain Alpha would build it. The cost of the materials is a minor part of the manufacturers total cost....so if a copper version would work better, it no doubt would sell better.
Hell they might make a copper version anyway, hoping somebody like you would buy one, replacing his "inferior" aluminum Alpha.
And no.....I'll refrain from calling you ignorant.
If all-copper would make that heatsink better, I'm certain Alpha would build it. The cost of the materials is a minor part of the manufacturers total cost....so if a copper version would work better, it no doubt would sell better.
Hell they might make a copper version anyway, hoping somebody like you would buy one, replacing his "inferior" aluminum Alpha.
And no.....I'll refrain from calling you ignorant.
Bluefront,
The aluminum and copper in the Alpha heatsink are the same as the aluminum and copper in any other heatsink. I used the Zalman as an example since they happen to make identical heatsinks in both copper and aluminum versions.
The reasons Alpha chooses not to make an all copper version of their pin heatsinks are more complicated than just the cost of materials, although that is a component. (And since when is the cost of materials a "minor" part of the cost?) The manufacturing process would most likely have to be changed, as would their tooling and machining equipment.
But the most likely reason they don't is the weight of the finished part. As the weight goes up so does shipping costs. If the weight is too far above the Intel specs System Integrators and OEM's are less likely to buy the heatsink. It's the big customers like those that Heatsink manufacturers really court. Alpha probably did the testing and decided that the increase in performance wasn't worth the increase in cost and weight.
But we digress.....
If Alpha did make a all copper version of that HS it would perform better. It's simple physics. There is nothing about Aluminum that makes it better than copper for heatsinks. That's the point I'm arguing with you. I agree that the Alpha is an excellent heatsink, but your contention that Aluminum is better than Copper because of its supposed magically better ability to transmit heat to the air is just total bunk. So far you've offered nothing to say otherwize. If you can find anything that says otherwise, please post it here.
As soon as you find it, post it. Until then, be careful about spreading your uneducated opinions as fact.
As always, I look forward to being proved wrong.
The aluminum and copper in the Alpha heatsink are the same as the aluminum and copper in any other heatsink. I used the Zalman as an example since they happen to make identical heatsinks in both copper and aluminum versions.
The reasons Alpha chooses not to make an all copper version of their pin heatsinks are more complicated than just the cost of materials, although that is a component. (And since when is the cost of materials a "minor" part of the cost?) The manufacturing process would most likely have to be changed, as would their tooling and machining equipment.
But the most likely reason they don't is the weight of the finished part. As the weight goes up so does shipping costs. If the weight is too far above the Intel specs System Integrators and OEM's are less likely to buy the heatsink. It's the big customers like those that Heatsink manufacturers really court. Alpha probably did the testing and decided that the increase in performance wasn't worth the increase in cost and weight.
But we digress.....
If Alpha did make a all copper version of that HS it would perform better. It's simple physics. There is nothing about Aluminum that makes it better than copper for heatsinks. That's the point I'm arguing with you. I agree that the Alpha is an excellent heatsink, but your contention that Aluminum is better than Copper because of its supposed magically better ability to transmit heat to the air is just total bunk. So far you've offered nothing to say otherwize. If you can find anything that says otherwise, please post it here.
As soon as you find it, post it. Until then, be careful about spreading your uneducated opinions as fact.
As always, I look forward to being proved wrong.
Excellent link SpyderCat!!! That's just the sort of information source we've been looking for.
I'll quote the segment of it that relates to our topic here:
Bluefront, any comments you'd like to add?
I'll quote the segment of it that relates to our topic here:
Dave Smith's General Heat Transfer Guide wrote:Contrary to popular belief, aluminum is not superior to copper where convection is concerned. I will not claim to know how this rumor started, but I’ll do my best to lay it to rest. I can think of three facts that may have led to the conclusion that aluminum is superior to copper. Let’s take a look at these facts and explain what they really mean.
Fact 1: Aluminum has a higher specific heat and lower density than Copper. Specific heat is a measure of how densely a material can store thermal energy. If you add thermal energy to a specific mass of material, it will increase in temperature.
The specific heat of aluminum is 903 joules/kg*K at 300K while copper is only 385. The density of aluminum is 2702 kg/m^3 at 300K while copper’s is 8933. The product of specific heat and density determines how much energy may be stored in a given volume. For a given volume of material, copper will store more energy at a given temperature change than aluminum.
False Perception: Since aluminum stores less energy per volume, it must be more efficient at getting rid of heat.
Truth: During steady-state operation, there is no net energy storage in the heat sink or fins/pins; hence, specific heat plays no part in steady-state performance.
Fact 2: Aluminum has lower density than Copper. Volume for volume, aluminum is much lighter than copper.
False Perception: Weight acts as a “sink” for heat. Since copper is more dense, it absorbs heat well from the die. Since aluminum is light, it gets rid of heat more effectively than copper.
Truth: Density has no direct relationship with steady-state heat transfer.
Fact 3: A small volume of aluminum will cool more quickly than an equal volume of copper once the heat source is gone. This is due to the same reason as fact #1, namely there is less energy stored per unit volume is aluminum than copper. This is, however, a transient condition. Heat transfer from a computer is a steady-state condition where the temperature of the heat sink remains relatively constant. The specific heat of a material partially determines how a material responds to transient conditions but has no effect at all on steady-state operation.
False Perception: Since aluminum cools more quickly once a heat source is removed, it must be more efficient at convection.
Truth: The heat source driving energy into the heat sink remains in effect until you turn off your computer. If you have aluminum pins or fins, congratulations, they will cool off more quickly than copper ones after you shutdown your PC.
The only properties belonging to the solid that affect convection are geometry and surface temperature. The fluid stream has no knowledge of what lies beneath the surface of the material. If an aluminum and copper item have the same precise geometry including microscopic surface details and they have the same temperature then they will have precisely the same convection.
Bluefront, any comments you'd like to add?
-
Bluefront
- *Lifetime Patron*
- Posts: 5316
- Joined: Sat Jan 18, 2003 2:19 pm
- Location: St Louis (county) Missouri USA
Nice read on the subject....however, I have to disagree with his conclusion regarding al/cu thing. Read over his fact #3. In it he admits that aluminum releases it's heat faster than copper, once the heat source is removed(or lowered which he doesn't seem to figure in). And this heat is released by convection....air currents over the metal.
So we get back to two heatsinks, one copper, one aluminum. Suspend them both in air, heat them to the same temperature, then remove the heat source. The aluminum heatsink will cool back to ambient quicker...he admits it as such. It has to do with density, and the ability of copper to retain or store the heat longer.
The source of heat (cpu) never remains constant in a computer. So the cpu temp is always going up and down, and the heat transfered into the heatsink by conduction (at which copper is superior) varies at the same rate. Then the convection takes place(airflow), removing the heat from the heatsink. At this point his argument fails in my opinion. He admits aluminum releases heat more quickly than copper, but he states it is only of benefit when you turn off the computer. I don't come to that conclusion.
If aluminum is superior at releasing it's heat by means of airflow, it stands to reason in a computer the fins of a heatsink will perform better if made of aluminum rather than copper. Back to my original post.....
So we get back to two heatsinks, one copper, one aluminum. Suspend them both in air, heat them to the same temperature, then remove the heat source. The aluminum heatsink will cool back to ambient quicker...he admits it as such. It has to do with density, and the ability of copper to retain or store the heat longer.
The source of heat (cpu) never remains constant in a computer. So the cpu temp is always going up and down, and the heat transfered into the heatsink by conduction (at which copper is superior) varies at the same rate. Then the convection takes place(airflow), removing the heat from the heatsink. At this point his argument fails in my opinion. He admits aluminum releases heat more quickly than copper, but he states it is only of benefit when you turn off the computer. I don't come to that conclusion.
If aluminum is superior at releasing it's heat by means of airflow, it stands to reason in a computer the fins of a heatsink will perform better if made of aluminum rather than copper. Back to my original post.....
Bluefront,Bluefront wrote:So we get back to two heatsinks, one copper, one aluminum. Suspend them both in air, heat them to the same temperature, then remove the heat source. The aluminum heatsink will cool back to ambient quicker...he admits it as such.
So we get back to two heatsinks, one copper, one aluminum.
Exactly the same design, same volume.
Suspend them both in air, heat them...
We notice it will take longer for the Cu HS to heat up, but finally we get them to the same temperature.
We can't come to another conclusion than that the CU HS can store more thermal energy for the same delta T.
Then remove the heat source.
We notice the Al HS cools down much quicker.
Does it cool down quicker because it transers it's energy more easily than CU ?
NO, it cool down quicker because it has less thermal energy stored in the HS !
I rest my case.
Regards, Han.
-
Bluefront
- *Lifetime Patron*
- Posts: 5316
- Joined: Sat Jan 18, 2003 2:19 pm
- Location: St Louis (county) Missouri USA
Yeah you can look at it differently. But get down to nuts and bolts. With the 478 boards mounted vertically, we have recommended maximum weights for heatsinks. The Alpha pushes that limit, even though it's mostly aluminum. If it were all copper it would be well over the limit, perhaps dangerously so. To maintain the same weight, the fin height would be reduced by 1/2 or more.
I'll stick with my statement....To be much better than an Alpha, a heatsink would have to be bigger or weight a whole lot more.
I'll stick with my statement....To be much better than an Alpha, a heatsink would have to be bigger or weight a whole lot more.
The max weight recommendation for heatsinks really relates only to possible damage during shipping. Once installed in a machine it would very hard to pull even the heaviest heatsink off a motherboard. There are plenty of heatsinks out there that are 2 or 3 heavier than the max recommended, and I've never seen a report of one of them coming off because of their weight.Bluefront wrote:With the 478 boards mounted vertically, we have recommended maximum weights for heatsinks. The Alpha pushes that limit, even though it's mostly aluminum. If it were all copper it would be well over the limit, perhaps dangerously so.
But that's not your original statement now is it? At first you said:Bluefront wrote:I'll stick with my statement....To be much better than an Alpha, a heatsink would have to be bigger or weight a whole lot more.
Are you now saying that your original statement was wrong?Bluefront wrote:Aluminum transfers heat better when used as a fin...air blowing over aluminum transfers heat better than air over copper.
I think the only factor that matters it the heat conductivity of the heat sink material.
If copper can store more heat per temp change then aluminum that means it that it'll take a little bit longer for the copper to reach the same equilibrium tempeature than aluminum, everything else being equal, assuming good surface area/fins/etc.
If copper can conduct heat better than aliminum, then the heat transfr to the fins will be faster and the temperature lower.
If the fins are not part of the base of the heat sink (ie soldered, screwed on) then that advantage is lost.
This is all my opinion only and how this makes sense to me, and why I started the rant.
One of my computers has a Zalman flower (I think I mentioned this before) which I don't think is a bad design. When I touch the tips they're almost room temperature but the core is hot. This is why i suggested making the fins a little thicker.
I don't have the means to test this but I bet if you take a stock Al heat sink and make an identical Cu one (with the same fin thickness), it'll blow the socks off most fancy Al/Cu and Cu/Cu (ie welded) heat sinks out there.
Again, I have no basis. I am a physical chemist though, so I'm not completely ignorant.
If copper can store more heat per temp change then aluminum that means it that it'll take a little bit longer for the copper to reach the same equilibrium tempeature than aluminum, everything else being equal, assuming good surface area/fins/etc.
If copper can conduct heat better than aliminum, then the heat transfr to the fins will be faster and the temperature lower.
If the fins are not part of the base of the heat sink (ie soldered, screwed on) then that advantage is lost.
This is all my opinion only and how this makes sense to me, and why I started the rant.
One of my computers has a Zalman flower (I think I mentioned this before) which I don't think is a bad design. When I touch the tips they're almost room temperature but the core is hot. This is why i suggested making the fins a little thicker.
I don't have the means to test this but I bet if you take a stock Al heat sink and make an identical Cu one (with the same fin thickness), it'll blow the socks off most fancy Al/Cu and Cu/Cu (ie welded) heat sinks out there.
Again, I have no basis. I am a physical chemist though, so I'm not completely ignorant.