Heatsink Mod for MBA

[evsp341995]

I know, I mean how would you be able to transfer the heat from the heatsink to the panel.

You can try that, but in theory that would be counter productive because again, the air flow is being blocked from exhausting the warm air into the outer environment. Instead, the warm air might just be pushed around and swirling around inside rather than exiting.

How does the hot air leave the system now? You said that air flows accross the heatsink, but there's nothing on the far side of the heatsink to blow the air out of the vents. When the fans running 7200 rpm, I can't feel much air coming out of the vents

 

[ayeying]

How does the hot air leave the system now? You said that air flows accross the heatsink, but there's nothing on the far side of the heatsink to blow the air out of the vents. When the fans running 7200 rpm, I can't feel much air coming out of the vents

The air being blown across the heatsink when its at 100% load for both GPU and CPU is approx 70-80 deg C. However, since the ambient air is cooler, the exiting air wouldn't be near that temp, closer to 40-50 deg C but it can build up.

What happens is the heat generated by the diode is being pushed to the edge of the heatsink, like how you have water on a desk and you're scraping it off the edge. The air just takes a ride across the heatsink then exit at the vents in the back. I'll post a diagram of the heatsink air flow later in the day when I have access to a scanner.

The fan blows air horizontally across the heatsink instead of directly out of the vents. Thats why air flow seems very small, then again, we only have a very small fan running at 6200RPM which is not fast at all.

 

[evsp341995]

The air being blown across the heatsink when its at 100% load for both GPU and CPU is approx 70-80 deg C. However, since the ambient air is cooler, the exiting air wouldn't be near that temp, closer to 40-50 deg C but it can build up.

What happens is the heat generated by the diode is being pushed to the edge of the heatsink, like how you have water on a desk and you're scraping it off the edge. The air just takes a ride across the heatsink then exit at the vents in the back. I'll post a diagram of the heatsink air flow later in the day when I have access to a scanner.

The fan blows air horizontally across the heatsink instead of directly out of the vents. Thats why air flow seems very small, then again, we only have a very small fan running at 6200RPM which is not fast at all.

How is tghe hot air leaving the system? If thre is little airflow from the vent, then isn't the air just circulating around inside the system?

 

[mac jones]

Thanks for the very informative original post (and follow ups)

I work with aluminum all the time. It's great stuff for projects because it's easily bendable and it's easily cut.

fantastic material and it's even a good conductor (cheap also).

I have some sheets that would be perfect for this so i'm going to try it. I've got an idea that this may work a bit.

It's odd though, I was afraid to do any kind of mod like this as these things are so expensive, but now that some time has (reduced the value) I feel fine about it

should be fun.

Note: Got the slop in hinge. Do they still replace these defective parts at the Apple stores? (I heard it was a ongoing commitment , but you know how that goes )

 

[evsp341995]

Thanks for the very informative original post (and follow ups)

I work with aluminum all the time. It's great stuff for projects because it's easily bendable and it's easily cut.

fantastic material and it's even a good conductor (cheap also).

I have some sheets that would be perfect for this so i'm going to try it. I've got an idea that this may work a bit.

It's odd though, I was afraid to do any kind of mod like this as these things are so expensive, but now that some time has (reduced the value) I feel fine about it

should be fun.

Note: Got the slop in hinge. Do they still replace these defective parts at the Apple stores? (I heard it was a ongoing commitment , but you know how that goes )

what are you going to do? are you just going to lay more sheets of aluminum down on top of the original heatsink?

 

[mac jones]

what are you going to do? are you just going to lay more sheets of aluminum down on top of the original heatsink?

I've been inside and poked around a lot when swaping out SSD's and doing the paste at one time.

I'll try what the OP did as thankfully now we have measurements ( and the old adage applies "If it ain't broke....." )

If I can think of a better options I will try it and post.

Everyone else do the same.

note: As I mentioned above, I used to be nervous about this, but I imagine that it would be pretty hard to screw this up if you don't break anything as the heat sink in there now is a joke. They went for extreme weight reduction, which is nice but maybe not what I want, presently.

 

[chopper dave]

some corrections

Ayeying, I think you have a lot of good ideas, but I'm going to have to be a science stickler and correct some common misconceptions you have spread about heat transfer.

Aluminum is actually a great heat radiator and that's the primary reason it is used here. Copper is a great heat conductor, however, it's not good at all at radiating the heat away. It likes to store heat when Aluminum likes to get rid of it as fast as possible.

Actually, they are equally good radiators. The ability of a metal to radiate heat is entirely dependent on its surface area, geometry, fluid type, and fluid velocity. See the following article for a description of how engineers approximate heat transfer from a solid to a fluid/gas, and notice how there is lots of attention paid to the fluid, and NONE to the material of the solid.

http://en.wikipedia.org/wiki/Heat_transfer_coefficient

You are correct of course that copper is a good conductor, and has a greater heat capacity than aluminum. In the steady-state case, heat capacity does not enter into the equation, though it definitely changes the time required to reach steady-state conditions. These factors may have obscured what was really causing the results of your trials.

What happens is the heat generated by the diode is being pushed to the edge of the heatsink, like how you have water on a desk and you're scraping it off the edge.

Heat always flows from a hot area to a cold area, that's just the 2nd law of thermodynamics, so your analogy of heat being pushed is incorrect. A better one would be of water flowing downhill, with the higher ground symbolizing the hot area of the heatsink around the diode.

 

[ayeying]

Ayeying, I think you have a lot of good ideas, but I'm going to have to be a science stickler and correct some common misconceptions you have spread about heat transfer.

I welcome anyone to question my theories, that's how we learn. Considering I never took physics in College (yet) or even in High School (Seriously, I never took it), I'm doing this through logic and wiki pages and Google.

Actually, they are equally good radiators. The ability of a metal to radiate heat is entirely dependent on its surface area, geometry, fluid type, and fluid velocity. See the following article for a description of how engineers approximate heat transfer from a solid to a fluid/gas, and notice how there is lots of attention paid to the fluid, and NONE to the material of the solid.

http://en.wikipedia.org/wiki/Heat_transfer_coefficient

You are correct of course that copper is a good conductor, and has a greater heat capacity than aluminum. In the steady-state case, heat capacity does not enter into the equation, though it definitely changes the time required to reach steady-state conditions. These factors may have obscured what was really causing the results of your trials.

I'm a bit confused here. For the sake of argument, if both metals of same density and mass were in a controlled environment and both were required to dissipate heat, which would go faster? I understood that Aluminum would dissipate faster?

Heat always flows from a hot area to a cold area, that's just the 2nd law of thermodynamics, so your analogy of heat being pushed is incorrect. A better one would be of water flowing downhill, with the higher ground symbolizing the hot area of the heatsink around the diode.

Yeah, I kinda figured that one was a bit off.

Here's a picture that I randomly drew up that shows the air flow of the stock heatsink/fan. I'm not an artist so screw my crappy drawing but it shows how the air enters and exits out of the MacBook Air (This applies to the Rev A models also, just a different heatsink style but same concept)

From 1) It draws in cold air from the environment
2) The fan pushes the air to a horizontal exit of the Fan assembly
3) The warmed up air exits through the vents in the back

The air flow works because there are foam based walls directing the air flow on the heatsink.

 

[evsp341995]

I'm still confused as to what the proper material for this heatsink would be, copper or aluminum. I plan to cnc cut a much better heatsink with some short fins on it to replace the shoddy apple one. Does anyone have measurements of how much space is between the bottom case and the heatsink when the bottom is on the computer? I estimate about 2mm. I also still haven't gotten a clear answer as to whether attaching the. Heatsink to the bottom case (so the bottom could act as a radiator) would be benefitial.

 

[chopper dave]

I'm a bit confused here. For the sake of argument, if both metals of same density and mass were in a controlled environment and both were required to dissipate heat, which would go faster? I understood that Aluminum would dissipate faster?

Interesting question. It is not very useful to compare Al and Cu and say they have the same density- it defeats the purpose of a comparison. Let's say they are both the same geometry. It is important to distinguish between short-term and long-term (steady-state) performance. It is also important to define "dissipate." Do you mean watts from the chip, or watts from the heatsink?

In the very short term, the Cu sink will probably take more watts from the chip because it has more heat capacity and better conduction. it is possible for the Al sink to dissipate more heat to the air at first, though, because it will heat up faster due to its low heat capacity.

In the long term, steady state solution heat capacity does not enter into the equation (it is like a mass in a physical system, the conductivity like a spring). Heatsinks usually have a maximum temperature determined by the processor they are cooling. The tradeoff is that the hotter the sink gets, the more heat it can dissipate. You can imagine then that an ideal heatsink will be uniformly hot, enabling maximum dissipation to the air across its entire geometry.

Cu will almost always work better because its high conductivity will keep the heatsink temperature more uniform, while the Al sink will tend to be cooler at its extremes. We choose Al over Cu when weight, cost, or corrosion is a concern.

to evsp341995: I have had a hard time machining copper- it is very gummy and you need to get the right alloy for it to be easy to work with. Still, if you have the experience and don't mind the extra few ounces, it is probably the superior material. I think a lot of manufacturers treat their Cu heatsinks with a thin film to protect against corrosion- you might want to look into this as well. No coatings are necessary for Al and as long as you choose a 6 or 7000 alloy it will machine nicely.

I think attaching the sink to the bottom, probably through a thermal interface pad, is a good idea, as long as your lap doesn't mind the extra heat. It is hard to measure spaces when you can't get a micrometer in there. I would put shims in between the heatsink and bottom and just keep adding them until you feel some resistance putting the bottom back on. Measure the shims and plan for that distance.

 

[ayeying]

I'm still confused as to what the proper material for this heatsink would be, copper or aluminum. I plan to cnc cut a much better heatsink with some short fins on it to replace the shoddy apple one. Does anyone have measurements of how much space is between the bottom case and the heatsink when the bottom is on the computer? I estimate about 2mm. I also still haven't gotten a clear answer as to whether attaching the. Heatsink to the bottom case (so the bottom could act as a radiator) would be benefitial.

I would say at best there's approx 1.5 to 2mm max of clearance between the heatsink itself and the bottom enclosure. Remember, the bottom plate actually warps slightly when it's screwed on in a non ordered fashion so the clearance would be slightly different +/- 0.02mm.

I'd say a combination of Copper and Aluminum would be the best since those are the 2 mostly used metals in conventional heatsinks for all computer hardware.

Attaching the heatsink to come in contact with the bottom panel will increase radiation of heat and mass, but if you block the airflow, wouldn't that just make the fan literally useless since it's being blocked from exiting the warm air?

 

[evsp341995]

Interesting question. It is not very useful to compare Al and Cu and say they have the same density- it defeats the purpose of a comparison. Let's say they are both the same geometry. It is important to distinguish between short-term and long-term (steady-state) performance. It is also important to define "dissipate." Do you mean watts from the chip, or watts from the heatsink?

In the very short term, the Cu sink will probably take more watts from the chip because it has more heat capacity and better conduction. it is possible for the Al sink to dissipate more heat to the air at first, though, because it will heat up faster due to its low heat capacity.

In the long term, steady state solution heat capacity does not enter into the equation (it is like a mass in a physical system, the conductivity like a spring). Heatsinks usually have a maximum temperature determined by the processor they are cooling. The tradeoff is that the hotter the sink gets, the more heat it can dissipate. You can imagine then that an ideal heatsink will be uniformly hot, enabling maximum dissipation to the air across its entire geometry.

Cu will almost always work better because its high conductivity will keep the heatsink temperature more uniform, while the Al sink will tend to be cooler at its extremes. We choose Al over Cu when weight, cost, or corrosion is a concern.

to evsp341995: I have had a hard time machining copper- it is very gummy and you need to get the right alloy for it to be easy to work with. Still, if you have the experience and don't mind the extra few ounces, it is probably the superior material. I think a lot of manufacturers treat their Cu heatsinks with a thin film to protect against corrosion- you might want to look into this as well. No coatings are necessary for Al and as long as you choose a 6 or 7000 alloy it will machine nicely.

I think attaching the sink to the bottom, probably through a thermal interface pad, is a good idea, as long as your lap doesn't mind the extra heat. It is hard to measure spaces when you can't get a micrometer in there. I would put shims in between the heatsink and bottom and just keep adding them until you feel some resistance putting the bottom back on. Measure the shims and plan for that distance.

Thanks for the very informative post. I don't have any experience machining copper, I just read a few forums about it. So you think I should use aluminum? Do you have any experience machining aluminum? What bit should I use? Carbide? Do I need any cooling?

I would say at best there's approx 1.5 to 2mm max of clearance between the heatsink itself and the bottom enclosure. Remember, the bottom plate actually warps slightly when it's screwed on in a non ordered fashion so the clearance would be slightly different +/- 0.02mm.

I'd say a combination of Copper and Aluminum would be the best since those are the 2 mostly used metals in conventional heatsinks for all computer hardware.

Attaching the heatsink to come in contact with the bottom panel will increase radiation of heat and mass, but if you block the airflow, wouldn't that just make the fan literally useless since it's being blocked from exiting the warm air?

We won't be blocking airflow. Were only going to cover a small area of the original heatsink with the heat transfer pads.

 

[PinkyMacGodess]

I'm still confused as to what the proper material for this heatsink would be, copper or aluminum. I plan to cnc cut a much better heatsink with some short fins on it to replace the shoddy apple one. Does anyone have measurements of how much space is between the bottom case and the heatsink when the bottom is on the computer? I estimate about 2mm. I also still haven't gotten a clear answer as to whether attaching the. Heatsink to the bottom case (so the bottom could act as a radiator) would be benefitial.

Intel started using a solid copper core surrounded by a finned aluminum structure a number of years ago because the copper would draw the heat out and the aluminum would radiate it out through the air moving through the vanes.

I replaced the heat sinks in our 1U Sun Fire server and it also used a copper slab with an aluminum finned structure to diffuse the heat. It was a rather interesting find. After seeing no copper in any heat sinks and now seeing a lot of uses of it.

Could someone machine a groove into the existing heat sink, slip a copper piece into position and then use either thermal paste or a thermal pad to transfer the heat to it. I'd think that should work wonders for keeping things cooler. Hiding it from the prying eyes of an Apple surgeon would be far more difficult...

 

[evsp341995]

Intel started using a solid copper core surrounded by a finned aluminum structure a number of years ago because the copper would draw the heat out and the aluminum would radiate it out through the air moving through the vanes.

I replaced the heat sinks in our 1U Sun Fire server and it also used a copper slab with an aluminum finned structure to diffuse the heat. It was a rather interesting find. After seeing no copper in any heat sinks and now seeing a lot of uses of it.

Could someone machine a groove into the existing heat sink, slip a copper piece into position and then use either thermal paste or a thermal pad to transfer the heat to it. I'd think that should work wonders for keeping things cooler. Hiding it from the prying eyes of an Apple surgeon would be far more difficult...

We couldn't machine a groove into the original one as it is WAY too thin. It is about .2mm thick.

 

[ayeying]

Yeah, machining Copper that's barely a mm thick isn't possible. I actually ended up cutting it with scissors and flattening it out as much as I can.

Your idea of using thermal pad or something to transfer the heat from the heatsink to the lower panel is interesting. I'll see if I can find some extra materials and try it out.

 

[phogstenjr]

Any Luck?

Hello,

Has anyone had any luck with a working heatsink modification for the 1st Generation MacBook Air?

Paul

 

[Drag'nGT]

A better heatsink would do really well on the forums. Anyone able to make these to sell?

 

[nicklausi]

possible design of heatsink of mba rev. A

Hi there,
i just registered to participate in this discussion. Btw: I think its really interesting!
I also thought about many possible ways, improving my mba's heatsink.
As i read the whole thread and some other threads, i came to the conclusion, that it would be the best attempt to concentrate on the airflow and how to make use of it the most efficient way.

How would it be, if you just take a thin layer of aluminum - but with the double size of the actual heatsink. So you cut out a piece of aluminum with the same thickness, the same width - but at least the doubled height.
Then you bend it over 180°, where the height exceeds the former heatsink.

This new heatsink touches the bottom of the aluminum macbook body, but doesn't get into the way of the airflow.
At last you could cut some fins at the end, for fitting it between the airflow openings.

I will make a small drawing and attach it, as my english is not sufficient enough to explain it properly.

 

[nicklausi]

heatsink draft

the draft