2020 Air Heatsink Modification Thread

[Loog]

Well, at the risk of repeating myself, it seems to me to be a failure of execution and not design.

Also, looking at the way the 2020 heatsinks have been step-milled, I wonder if the original heatsinks delivered to Apple were not milled that way, and were in fact meant to sit closer to the CPU/GPU cluster with minimal thermal paste – which is effectively what you guys are doing by shimming (replacing copper that was taken out during the milling process). Maybe Apple made the decision late in the day to step-mill the heatsinks for extra clearance because when they tested them on the production line they discovered there wasn't enough tolerance. Prototypes are usually built to finer tolerances than production models.

I'm only guessing, and maybe I'm wrong, but isn't the step in the heatsinks roughly the same thickness as the shims you guys are using? Something to think about.

@Mopar, the area that's milled out is the same dimension as the CPU die, I guess to take account for thermal paste to take up the rest, it makes it easier for automation during assembly.

For some/many I'm sure the machine may operate within expected tolerances, some have no issues with heat, good for them, I guess tolerances are in their favour. For those who have a machine which doesn't perform well feel a little miffed at spending £1000 + on a device that run out of steam doing some real basic tasks such as video conferencing with fans spinning at 8000rpm for short Teams/Zoom call or watching online videos, that's not too much too ask. We're not asking for pro power, nor are we pro users or suggesting the device is used for such.

If you look at the very early threads you'll see @srkirt who used an ocean of thermal paste on his 2019 to fill the void in early tests on the 1st gen Retina model before moving to shim it out. Apple has a track record of poor thermal transfer which is widely documented over numerous devices and thermal paste replacement often help.

There are some good machines so I agree the design does work, however there are plenty which don't operate in the same manner, if there is no consistency in manufacture then I see this as a design or QA issue and not execution in the manufacturing process.

 

[thepowerfulpotato]

Just did the shim mod on my MBA i5/8/512 and I have to say i'm deeply impressed with the results. To add to the discussion on why apple chose to design the heatsink in this way I think it's clear that they had to have a larger heatsink -> die clearance because they are using a positive mounting mechanism with no 'give' or springs in it. If the heatsink were right against the die then there would be a potential risk of cracking the die when installing the heatsink. That said there are better ways to work around that problem, such as installing springs under the screws or using a spring loaded backplate like they do in all their other models.

That said does anyone know of a source for the heatsink screws? I damaged one of mine as it had an ungodly amount of loctite on it and would like to have new ones to put in if the machine ever goes in for warranty work. Sorry if this. has been mentioned previously in the thread.

 

[kazune_karin]

So where I tend to agree with @jackiez is that the cooling system design is not the failure many here are tying to make it out to be. Certainly we now know a heat pipe is not an easy fix – despite all the shrill posts in the early part of this thread claiming otherwise.

Heatpipe is easy to be modded by yourself
Your bottom case does not have to be warm.

MacBook Air 2020 i5/16GB/256GB heatpipe mod

Background and acknowledgement
Hi guys. Here is a new Air user from Japan. Thanks to all brilliant challengers from this thread (including srkirt, kinchee87, vyruzreaper, RiaKoobcam, Robotronic and DanSilov). I tried the shim mod which is great. The performance is better and more quiet, but the heat dissipation still worries me. I tried the heat pad mod but the bottom is too warm for me and the mod seems to stress the insulating material at the bottom case too much. So, I tried to transfer the heat from the heatsink to the spinning fan without touching the bottom case.

Process
I read this thread for guys who used metal pipe to transfer the heat to somewhere near the fan. I think the improvement is limited because the heat conduction from the metal pipe is not efficient and most air to the fan does not need to pass through the metal pipe. My plan is to make a metal punching mesh to force the spinning fan to cool it.
Here is the overall idea before installation.
View attachment 921971
I used the remaining 0.3mm copper plate (same as shim mod) to make the copper pipe. And I bought an aluminum punching mesh 0.5mm (I think) because I cannot find a copper mesh on Amazon Japan here.

View attachment 921973 View attachment 921974
I used heat insulating double-side adhesive to seal the fan border and brown heat insulating tape to seal the copper plate not to overheat other components. Between the heatsink and copper pipe I used a heated (0.5mm, 6mW only) to hold the pipe in place. On top I used the brown heat insulating tape again to ensure no contact and heating up of bottom case.

Results
I found that the startup CPU temperature with shim mod alone still reaches 100 degree. But with heated mod it was 70 degree. This one is 80 degree but the temperature drops very quickly with fan spinning. I used TG Pro to setup a custom profile (4500 rpm always and maximum 7000 rpm over 90 degree).

I ran Geekbench Multicore CPU test and found that the score is very good
3169 (shim mod)
3595 (shim + heatpipe mod)
View attachment 921976

The heat profile also proves that heat is transferred properly. The frequency (light blue) during early low utilization part of the test caught up the requested one (pink) than shim pad alone. And heavy load part is better also (though heatpad mod is better because the heatpipe and fan speed capacity). You can find the comparison quoted from DanSilov thread below.
View attachment 921978



Conclusion
I am very happy to have this little reversible fix to help my Air performing better. I don't want the touch bar, and this mod helps me to have a comfortable use of my Air for years (at least I think). Thank you all.

Thanks. What paste did you used? I only have Arctic MX-4 on my hands.
I repasted again and with shim + heat pipe mod (I am too lazy to remove the heat pipe),
my score rises to 3713. I did it 3 times, and its all above 3700. Wow.
View attachment 922650

And more importantly, the frequency profile is perfect for me. The pink (requested frequency) line completely overlapped with the blue (average frequency) line THROUGH the test. That means the CPU provided what was needed for the test without thermal restricted. And the temperature never reached 100 degree.
View attachment 922651

 

[Mopar]

Well I don't know if they dropped the ball with mine, but I just ran GB5 on my unmodded i5/16/256 2020 MBA and I don't have much to complain about. Idle temps afterwards good, too.

 

[adrianstuartt]

Well I don't know if they dropped the ball with mine, but I just ran GB5 on my unmodded i5/16/256 2020 MBA and I don't have much to complain about. Idle temps afterwards good, too.

View attachment 927390

Yeah, that looks very good. A lot of the people here are only getting ~1100 and ~2600 for SC/MC, respectively, unmodded (i5). When did you get that and where from ?

I got my i3 model via Amazon about 14jun and the thermals were just trash in mine.

 

[Mopar]

I got mine April 8. It was a BTO, so it was ordered through Apple in Australia. I ran GB5 on it the first day I had it and it scored 1203/3116, so there has been some multi-core improvement since then, although the single processor score has gone down <shrug>.

 

[adrianstuartt]

I got mine April 8. It was a BTO, so it was ordered through Apple in Australia. I ran GB5 on it the first day I had it and it scored 1203/3116, so there has been some multi-core improvement since then, although the single processor score has gone down <shrug>.

any chance you'd be willing to take apart a perfectly functioning MBA for science? xD

 

[Mopar]

I've been following this thread since the start, so I was thinking about doing the shim mod, but the reality is my machine appears to be fine and I don't have any real compliants. I'm a bit loathe to mess with it. It would mean not only opening it, but removing the heatsink and replacing the thermal paste. Not a hard job, but if it doesn't need it . . . However, I'm now thinking of getting another one for the wife, so if that one doesn't perform then yeah. Sorry. I'll think about it.

 

[jackiez]

that level of comprehension and blind stubbornness, whatever ?‍♂️

?‍♂️

 

[beyondmeat]

Hello all,

I finally built up the courage to do the shim mod. I received my MacBook Air i5/8GB/256GB on the 17th and have been using it regularly since. I watch a lot of twitch and I use firefox as my main browser. I was frustrated with the high temps while watching streams and I have now read the majority of this thread. I bought 15x15x0.3 mm shims from amazon and used scissors to cut one of them to be 7x15x0.3 mm. I used Arctic Silver 5 for the thermal paste. My screws for the bottom casing were P5 and the heatsink screws were T3.

I did a few tests prior to doing the mod using Geekbench 5. My best single-core was 1176 and my best multi-core was 3244 (from waking after a while of no activity and on battery). Post mod my best single-core was 1185 and my best multi-core with fans set to max was 3699 (again on battery but after being used briefly). Post mod multi-core without changing the fan profile was 3576. I was reaching 100 degree temps during the post mod tests (not sure if this should happen based on other users tests).

Now I'm wondering if this is normal expected increase in performance or if I messed anything up. I notice when I watch twitch post mod, I'm still getting up in the 90 degree range sometimes around 100, but the requested frequency is being met whereas prior to the mod the requested frequency was nowhere near being met.

I can see that certain things are running a bit better and the frequencies are getting higher when requested, but I'm unsure if I should be seeing better results.

Is it possible that too much thermal paste could make this mod not as effective as it should be? Should open it up and check that I didn't "overpaste" it.

I've attached some screenshots. Any comments would be greatly appreciated.

 

[adrianstuartt]

Hello all,

I finally built up the courage to do the shim mod. I received my MacBook Air i5/8GB/256GB on the 17th and have been using it regularly since. I watch a lot of twitch and I use firefox as my main browser. I was frustrated with the high temps while watching streams and I have now read the majority of this thread. I bought 15x15x0.3 mm shims from amazon and used scissors to cut one of them to be 7x15x0.3 mm. I used Arctic Silver 5 for the thermal paste. My screws for the bottom casing were P5 and the heatsink screws were T3.

I did a few tests prior to doing the mod using Geekbench 5. My best single-core was 1176 and my best multi-core was 3244 (from waking after a while of no activity and on battery). Post mod my best single-core was 1185 and my best multi-core with fans set to max was 3699 (again on battery but after being used briefly). Post mod multi-core without changing the fan profile was 3576. I was reaching 100 degree temps during the post mod tests (not sure if this should happen based on other users tests).

Now I'm wondering if this is normal expected increase in performance or if I messed anything up. I notice when I watch twitch post mod, I'm still getting up in the 90 degree range sometimes around 100, but the requested frequency is being met whereas prior to the mod the requested frequency was nowhere near being met.

I can see that certain things are running a bit better and the frequencies are getting higher when requested, but I'm unsure if I should be seeing better results.

Is it possible that too much thermal paste could make this mod not as effective as it should be? Should open it up and check that I didn't "overpaste" it.

I've attached some screenshots. Any comments would be greatly appreciated.

View attachment 927661 View attachment 927662 View attachment 927663

Looks good. You can check out the Google Sheet to compare.

You can bring temps down quite a bit further by thermal coupling the heatsink to the chassis cover with a thermal pad. remove the cushion sticker thing behind the heatsink from the chassis cover and stick a thermal pad cut to size on the heatsink. I went with the 2mm wathai blue. Yes, the heatsink area of the bottom cover gets quite warm under heavy loads if you do it The rest of the cover stays cool.

 

[Supra Mac]

Gentlemen, thanks for this thread and your work. I had my eye on the MacBook Air but the thermal issue moved me over to the MBP so I got more machine that I needed.

Now with the timeline of Apple silicon and the good work done here to address thermals, I was lucky enough to be in the return window. So I have an i3 air expect tomorrow!

I will post soon my stock numbers. I have parts on order to do the paste, shim and after testing, heat pad then more testing.

Basic needs here, email, web browsing, YouTube in safari, zoom (rarely), basic excel and word. Hope it will stay quiet pushing the laptop display and 1080p monitor.

I got a mechanical engineering background so I’m ready to crack this bad boy open!

 

[nill1234]

@Supra Mac with an i 3 you wont have any real problems. The cooling seems sufficient, but if you do the mod the shim mod shouöd work but the heatpad not so much because you wont have a corrugated heatsink like the i5 and i7.

 

[agaskew]

I've been following this thread since the start, so I was thinking about doing the shim mod, but the reality is my machine appears to be fine and I don't have any real compliants. I'm a bit loathe to mess with it. It would mean not only opening it, but removing the heatsink and replacing the thermal paste. Not a hard job, but if it doesn't need it . . . However, I'm now thinking of getting another one for the wife, so if that one doesn't perform then yeah. Sorry. I'll think about it.

The over riding rule in all of this: If It Ain't Broke, Don't Fix It.

If your machine is performing how you want it to, great

 

[adrianstuartt]

@Supra Mac with an i 3 you wont have any real problems. The cooling seems sufficient, but if you do the mod the shim mod shouöd work but the heatpad not so much because you wont have a corrugated heatsink like the i5 and i7.

the thermalpad does actually help quite a bit for what the i3 would consider 'heavier' tasks, actually, especially if you absolutely never want to hear the fan. Not nearly as important as the shim, though.

 

[beyondmeat]

Looks good. You can check out the Google Sheet to compare.

You can bring temps down quite a bit further by thermal coupling the heatsink to the chassis cover with a thermal pad. remove the cushion sticker thing behind the heatsink from the chassis cover and stick a thermal pad cut to size on the heatsink. I went with the 2mm wathai blue. Yes, the heatsink area of the bottom cover gets quite warm under heavy loads if you do it The rest of the cover stays cool.

Thank you for your comments! That actually makes me feel a lot better lol. I'll add some of my benchmarks to the Google Sheet for reference.

Maybe I'll give the thermal pad a try. If it gets too hot, it's very easy take the pad back out.

 

[srkirt]

I am going to return the Huawei 13 2K intel to buy the fresh out of the oven Huawei 13 2K AMD Ryzen ™ 5 3500U with the same power of the Intel 8 Generation, for only 649 € !!! Ohhhh mama !!! More power than a 2020 MBA Less heat and half the price !!! It is difficult after weeks testing this brand, that I spend money on a MacBook ...

Comprar HUAWEI MateBook 13 AMD- HUAWEI ES

Compra ahora el nuevo Huawei Matebook 13 con la potente tecnología de AMD, descubre su pantalla full view con carga rápida.

consumer.huawei.com

 

[excelsior.ink]

I am going to return the Huawei 13 2K intel to buy the fresh out of the oven Huawei 13 2K AMD Ryzen ™ 5 3500U with the same power of the Intel 8 Generation, for only 649 € !!! Ohhhh mama !!! More power than a 2020 MBA Less heat and half the price !!! It is difficult after weeks testing this brand, that I spend money on a MacBook ...

Comprar HUAWEI MateBook 13 AMD- HUAWEI ES

Compra ahora el nuevo Huawei Matebook 13 con la potente tecnología de AMD, descubre su pantalla full view con carga rápida.

consumer.huawei.com

Look good on paper, but too bad it doesn't run macOS

 

[vyruzreaper]

I got mine April 8. It was a BTO, so it was ordered through Apple in Australia. I ran GB5 on it the first day I had it and it scored 1203/3116, so there has been some multi-core improvement since then, although the single processor score has gone down <shrug>.

Run it back to back to back and see if your temps stay below 100C and your score don't drop

 

[srkirt]

Look good on paper, but too bad it doesn't run macOS

The one that Intel installs theoretically if you can ...

 

[Supra Mac]

Alright, I'm up and running. Some stock numbers on fresh install nothing brought over.

2020 i3/8/256

Geekbench run average:
single 1088, multi 2251
max fan during test 4000 rpm @100C

Ran another with fans at full blast:
single 1095, multi 2360

Cinebench R20 run average:
multi 685
max fan during test 7700 rpm @100C

With fans at full blast:
multi 722

Will follow up with some observations after more daily usage.

 

[adrianstuartt]

Alright, I'm up and running. Some stock numbers on fresh install nothing brought over.

2020 i3/8/256

Geekbench run average:
single 1088, multi 2251
max fan during test 4000 rpm @100C

Ran another with fans at full blast:
single 1095, multi 2360

Cinebench R20 run average:
multi 685
max fan during test 7700 rpm @100C

With fans at full blast:
multi 722

Will follow up with some observations after more daily usage.

You have very good numbers if that's stock. It took modding for mine to get there. while i don't care for the temps, it's technically within spec. where did you buy that unit from ?

 

[Supra Mac]

I got it from Amazon. I am not expecting any spectacular post mod from benchmarks but hopefully it helps with normal usage temps. Right not it idles around 62-65C with ~2700 rpm hook up to dell 1920x1200 monitor. If I disconnect the monitor it drops to 47-50C with fans off.

 

[srkirt]

 

[Saul Giordani]

This is the average temp during a Zoom VC with 12 participants