Guide: MacBook Air Heatsink and Heat Pad Mod

[macking99]

Thanks OP for the great write up. I did the same thing by adding the shim and thermal pad but I went a step further. I soldered the 0.3mm shim to the heatsink using a heatgun and some solder. The goal was to get better thermal conductivity and to only have a single application of thermal paste on the CPU/GPU and heatsink.

TG Pro was used to monitor temps with room temp 25C/76F and results are based on average web browsing and watching YouTube videos:
Base line stock temps web browsing: 62-75C
Base line stock temps watching YouTube: 75-85C

Shim and Thermal Pad mod temps web browsing: 58-70C
Shim and Thermal Pad mod temps watching YouTube: 70-80C

Soldered Shim and Thermal Pad mod temps web browsing: 49-60C
Soldered Shim and Thermal Pad mod temps watching YouTube: 60-70C

If anyone thinking about soldering the shim to the heatsink, I recommend purchasing a spare heatsink for peace of mind. In my case, that wasnt necessary since I have years of soldering experience. A new heatsink only cost $10 on Ali Express.

 

[corentin-p]

Hello !

I registered just to say thank you, your guide gave me enough confidence to try this mod myself on my brand new core i5 256 / 8Gb MacBook Air. I used generic 0.3 15x15mm copper shims and a generic heat pad from eBay (coming from uk), plus arctic mx-4 thermal paste.

heat pad : https://www.ebay.fr/itm/100MM-x-100...e=STRK:MEBIDX:IT&_trksid=p2057872.m2749.l2649

copper shims : https://www.ebay.fr/itm/1-5cm-x-1-5...var=553590392526&_trksid=p2057872.m2749.l2649

I did not attain full performance I think, my scores are a little under what others seem to attain (I must not have done the shim mod correctly, or the heat pad of a generic brand does not have enough thermal conductivity), let me know what you think of the scores below)

before mod GB5 : 1080 / 2500
after shim mod = around 1100 / 3300
after shim mod + heat pad mod = 1180 / 3700

I must say the shim mod was a great improvement already, the heat pad only added a lot more stability. I did not achieve completely steady temperatures during the Geekbench test, but the temperature only attained 100°C after 30+ s of the test which is really good and far from the 2s 40° -> 100° that it was before. my idle temperatures in my 19/20°C living room are around 29 low / 35 high.

I use my MacBook for light tasks, occasional YouTube videos, working with Flac / apple lossless files (conversion, tags) and noticed a lot less fan noise. I also game but light titles only (openRA - open source rewrite of Command and conquer - red Alert - check it out it's awesome ! - and Starcraft I remastered through battle.net)

I have a Windows based work-provided machine for the upcoming (hopefully partial) lockdown that is getting more and more chance of happening tomorrow evening in France, so I don't really need my MacBook for teams / zoom calls that are ahead :-(

stay safe and enjoy this wonderful machine !

Corentin

 

[LED]

I did not attain full performance I think, my scores are a little under what others seem to attain (I must not have done the shim mod correctly, or the heat pad of a generic brand does not have enough thermal conductivity), let me know what you think of the scores below)

before mod GB5 : 1080 / 2500
after shim mod = around 1100 / 3300
after shim mod + heat pad mod = 1180 / 3700

It's been a very successfully modification for me as well.. as for your scores, your pre-mod scores were very similar to mine. Seemed a bit lower than the norm. My post mod scores are however a bit higher than yours. I'm at about 1160/3950. I used 0.5mm shims, so I wonder if you would've gotten better results with the slightly thicker shim.

 

[throAU]

I've ordered 0.8mm and 0.5mm shims, to anyone who has done it, do you reckon the screws would be long enough for 0.8mm? Currently I'm running a thermal pad under the heatsink and if 0.8mm will fit I might break the machine open again this weekend, otherwise I'll wait for the 0.5mm to arrive

 

[corentin-p]

Sooo i've re-run the tests and what I can say with my setup is, in fact, the heatpad changed my idle temps, because with the heat pad on, temps are climbing way less rapidly, only hitting 100°C for a brief period of time during the test, check intel power gadget capture below, but cools slower and does not get back to almost under 30°C as before.It's now more around the 32-37°C.

the benchmark scores stay around 1170 / 3750

I also found one culprit of some way lower scores : Bluetooth eating 60% of my cpu from time to time, a well documented problem with Microsoft sculpt mouse : https://forums.macrumors.com/threads/blueotoothd-10-15-1.2210403/ . Only linking macrumors thread because I don't know if other links are accepted, but well documented on other sites, even apple's own discussions platform. deactivating Bluetooth for my tests now.

bottom line is : better than stock performance, I'm not gonna try other mods for now, we'll see in about a year's time when time has come to change the thermal paste, maybe I'll try the 0.5 copper shims, I don't need some more perf for now, let's stay on the safe side because maybe I'll need this computer in the next months, I don't wan't to ruin it ;-)

Thanks for your answers anyway

Corentin

 

[macking99]

Here is a better solution if anyone is interested to try it. I cut a sheet of 0.3mm copper and make it into a heat transfer plate. The plate is 200mm x 36mm and covers a large surface area. High Temp Kapton tape was used to prevent any short or electrical damage when the heat transfer plate is installed.

After the installation, I've noticed very low temps than before with only the shim mod+pad. I am seeing a more steady and consistent temperature. With only the shim mod+pad, I was seeing a temp range of 49-60C with fan on when doing web browsing.

Now, I am getting 42-45C temps with no fan at 77F/25C room temperature when doing web browsing. Watching YouTube videos is now giving me 55-60C temps and no fan. That is very impressive.

The bottom of the laptop is no longer very warm to the touch when on my lap. I am very happy with the final results and I think this is it for me when it comes to getting better thermals from this 2020 MacBook Air.

Please Note:
This mod is not to be taken lightly and you are doing this at your own risk. I shorted my battery when test fitting the copper heat transfer plate and ended up having to buy a replacement battery. This is why the High Temp Kapton tape is so important and was applied to the bottom and sides of the copper heat transfer plate to prevent electrical conductivity.



Mockup of the copper heat transfer plate.



Kapton tape applied to the copper plate and thermal pad applied to the heatsink.

The final installation.

Benchmark result after the install. CPU is an i5

 

[corentin-p]

Very Interesting thank you. It’s interesting to note that score don’t really go higher than on my shim+pad setup, I wonder if the OS is capping the perf somewhat. Like on my setup you seem to note much better thermals without a tremendous boost in perf (some benchmarks got around 4000 on multi core) I wonder if your external shim is really conducting the heat to the fan or just increasing the dissipating surface. Good point for the casing external temp though, well done !
What all this seem to show is that the fan does not really matter for cpu temps, maybe for the aluminium casing / other internal components though.
On my end I removed the pad by fear of getting the components around the CPU too hot / blocking internal air circulation. Went back to 1184/3380 and a little more time at 100°C during the test but That is the price of my peace of mind . I also noted that conducting the heat to the casing with the pad resulted in the black insulator having lifted a little bit, pointing to the adhesive of said insulator that keeps it glued to the aluminium casing potentially melting. That could have less chance of happening with your custom heatpipe.

 

[macking99]

Very Interesting thank you. It’s interesting to note that score don’t really go higher than on my shim+pad setup, I wonder if the OS is capping the perf somewhat. Like on my setup you seem to note much better thermals without a tremendous boost in perf (some benchmarks got around 4000 on multi core) I wonder if your external shim is really conducting the heat to the fan or just increasing the dissipating surface. Good point for the casing external temp though, well done !
What all this seem to show is that the fan does not really matter for cpu temps, maybe for the aluminium casing / other internal components though.
On my end I removed the pad by fear of getting the components around the CPU too hot / blocking internal air circulation. Went back to 1184/3380 and a little more time at 100°C during the test but That is the price of my peace of mind . I also noted that conducting the heat to the casing with the pad resulted in the black insulator having lifted a little bit, pointing to the adhesive of said insulator that keeps it glued to the aluminium casing potentially melting. That could have less chance of happening with your custom heatpipe.

Correct, I believe the copper plate is absorbing most of the heat from the CPU heatsink and dissipating it via mass. The fan seems to turn on to compensate for when the copper plate reaches its maximum efficiency. So far, the fan only comes on when running very high intensity benchmark apps like Geekbench or Cinebench. But only between 2500-2800 RPM ranges and never goes beyond that.

The fan is no longer coming on during regular tasks like: email, web browsing with many tabs opened, watching videos etc. After noticing that the fan was no longer on during normal operation, I set my eye on tweaking the copper plate a tiny bit. The goal would be to help the copper plate dissipate the heat a little better.

So, I have drilled a few holes in the copper plate to increase the surface area and help shed the heat. I will be monitoring it for a few days to see if it helps.

 

[christophosphorus]

Correct, I believe the copper plate is absorbing most of the heat from the CPU heatsink and dissipating it via mass. The fan seems to turn on to compensate for when the copper plate reaches its maximum efficiency. So far, the fan only comes on when running very high intensity benchmark apps like Geekbench or Cinebench. But only between 2500-2800 RPM ranges and never goes beyond that.

The fan is no longer coming on during regular tasks like: email, web browsing with many tabs opened, watching videos etc. After noticing that the fan was no longer on during normal operation, I set my eye on tweaking the copper plate a tiny bit. The goal would be to help the copper plate dissipate the heat a little better.

So, I have drilled a few holes in the copper plate to increase the surface area and help shed the heat. I will be monitoring it for a few days to see if it helps.
View attachment 1414701

A bit late in replying here, Macking, but that's a helluva mod you did. Awesome job! If I ever make the time for it, I'll give this a go.

To date, since the writing of this thread, my i7 MBA has been performing flawlessly and provides an infinitely better experience compared to the stock setup. I still get high temperatures in the 90*C+ region if I'm videoconferencing for long periods of time (and that's mainly due to my external 49" 2K Ultrawide being powered by the laptop), but I disable Turbo Boost and temps drop 10-20*C immediately. Something to keep in mind for any power users out there.

 

[PArker93]

My Single score decreased from 1181 to 1165. My multi score increased from 3253 to 3853. And compute from 7582 to 9603. Applied both mods.
Edit: second attempt with mod was 1187 and 3894.

 

[Roland_Nem]

Hi all, I did this modification on my MBA 2020 i5, first I did not toch the heatsink, I only used the thermal pad between the heatsink and the bottom lid - it increased the thermal performance. Now I made the complete modification with the coper shim and new thermal paste + thermal pad. It became worse! My laptop heets up a bit slower but once it hits 75 degrees the fans starts to spin at around 5000-6000 Rpm (and stays so) which was maybe 2000-3000 before. It happens during a zoom call with chrome open 2-3 tabs. I starting to regret this mod. I just reset SMC no results. Any tips?

 

[Micheluzzo02]

Some of you after having modded your macbook air with copper shims have encountered strange electrical noises coming from the cpu area?
I usually hear them during the final part of the geekbench test.
I assumed that the components around the cpu were under stress for an extended period of time.
Do you think it is dangerous?