2020 Air Heatsink Modification Thread

[jackiez]

A trained monkey could design better thermals than Apple's current engineering staff.

Proper thermal design requires one simple thing: MAKE IT THICKER.

There is simply insufficient thermal mass to handle the heat. There is simply insufficient airflow to handle the heat. And there is simply insufficient battery to handle the power requirements.

Making it as thin as Apple idiotically keeps doing is DESTROYING performance.

MacBook Air series was never meant to be a powerhouse but portability, the ability to run most daily or even certain heavy tasks smoothly and its unique thin teardrop design

if u want high and consistent performance go for Pro series, it is really that simple, just the matter that if you could or willing to fork out extra few hundreds for it. get over it

 

[fcracer]

Happy Fathers Day! I decided to do the shim mod today and the pad again.


Stock i7 16GB 256 SSD
View attachment 925752

Artic 1.5mm Thermal Pad only on heatsink
View attachment 925753

Thermal Pad and .3mm Shim 20X16 with MX4 paste
View attachment 925754

Hi there, sorry if you’ve already answered this somewhere, but what scores were you able to get with just the shim and no pad? For the paste, is there any difference between MX4 and something like Kryonaut (it’s easier for me to get the Kryonaut).

 

[nill1234]

@fcracer Get the Kryonaut if its easier, between all the thermal pastes there is no real difference, except if you arw using liquid metal but that’s definitely nits recommended for mobile usage.

 

[kazune_karin]

MacBook Air series was never meant to be a powerhouse but portability, the ability to run most daily or even certain heavy tasks smoothly and its unique thin teardrop design

if u want high and consistent performance go for Pro series, it is really that simple, just the matter that if you could or willing to fork out extra few hundreds for it. get over it

I am yes and no for your view. Yes, Air is not designed to be powerful as Pro. So extracting all potential CPU performance (more than Air overall design) is an enthausiatic action.

However, the fan of my i5 Air does turn on very frequently and suddenly with simply opening a browser or email app. Since i3 Air does not have this severe problem, so it is clear Apple fails to let i5/i7 to overreact even in very light tasks. Apple uses a new heatsink for i5/i7 Air, so they know i5/i7 is hotter than i3.
My point is, if they make i5/i7 available as an upgrade option, they should make sure it works no worser than i3. Otherwise, they should not make this option available and pushes people to Pro, instead of forcing users to rely on a returning policy.

 

[srkirt]

@srkirt, I'm quite tempted by Huawei and even the AMD version which is even cheaper for a windows feel but on a mac type platform. I still won't let the MBA go as I love the OS and settled with the performance of the 2018 model post mod (it was ok before but better now). The Matebook AMD is on sale here in the UK for £579, 8GB/256GB
[automerge]1592772983[/automerge]
Does the MX250 work ok MacOS ?

I have not installed it at the moment, but for it to work it must be Intel cpu not Amd.
This machine is amazing !!! it starts from when you press the button in just 3 seconds you already have the desktop !!!
[automerge]1592828236[/automerge]
[QUOTE = "fcracer, publicación: 28581447, miembro: 1082852"]
Hola, lo siento si ya has respondido esto en alguna parte, pero ¿qué puntajes pudiste obtener solo con la cuña y sin almohadilla? Para la pasta, ¿hay alguna diferencia entre MX4 y algo como Kryonaut (es más fácil para mí obtener el Kryonaut).
[/CITAR]
MX4 8 Kryonault 7 y Noctua 9... metal liquid 10.

 

[jackiez]

I am yes and no for your view. Yes, Air is not designed to be powerful as Pro. So extracting all potential CPU performance (more than Air overall design) is an enthausiatic action.

However, the fan of my i5 Air does turn on very frequently and suddenly with simply opening a browser or email app. Since i3 Air does not have this severe problem, so it is clear Apple fails to let i5/i7 to overreact even in very light tasks. Apple uses a new heatsink for i5/i7 Air, so they know i5/i7 is hotter than i3.
My point is, if they make i5/i7 available as an upgrade option, they should make sure it works no worser than i3. Otherwise, they should not make this option available and pushes people to Pro, instead of forcing users to rely on a returning policy.

the intention of putting different heatsink between i3 and i5 simply prove apple engineers have put effort in making sure the i5 would not perform badly over the i3 using the same heatsink. Besides, the i3 doesn't seems to has this kind of problem is because of i3 dual core itself doesn't allow it to happen due to its own power constrain or limit, while i5 with quad core is somehow more than what the macbook air could take under certain situation, thus i5/i7 users are facing the problem that wouldn't have appeared on the i3, not really apple engineers failed at something or whatsoever in my perspective. Also bear in mind this is just a 3 month old baby with merely 1 update so far, so we shall see further optimisation through os update to fix/balance these issues, just like how the benchmark of 10.5.5 has improved over the 10.5.4, wont be a drastic change just improvement to what it should be over time.

i somehow agree with you that the upgrade option is kinda like pushing people to the pro, but for me the option given by apple is very clear

1.you are a casual user & would like to spend less, we have an entry Macbook Air with i3 which will perform most of the tasks just as smooth and fast
2.you are a semi-power user and would like to have the option to upgrade with small amount of money, we have i5/i7 that will perform faster and 'smoother' than the base i3 model
3.you are a power user and willing to spend a little more but not too much? there you go the base macbook pro model
4.truly power user+want consistent performance and willing to pay more? there you go our high end options

 

[adrianstuartt]

it's not just the i5/i7s having issues, it's the entire 2020 MBA lineup. However, the i5/i7s can handle a bit more load before the fans spin up to 8k. So, while the i5/i7s have more raw performance to be gained with a simple fix to the heatsink, i'd argue that it's _less_ of an issue for those models than the i3 because it takes even less load before the i3 spins up to 8k. 1080p60 with hardware accelerated h264 via safari should NOT cause these machines to take flight. If you want to have this very minimal capability without a loud fan, i do believe you are looking at an i5+ with the stock configured heatsink

[edit/] i swear to god, if anybody tries to suggest that a $900 laptop shouldn't be capable of 1080p60, i'm going to suggest you run, not walk, to your nearest psychiatrist. or at minimum, to stop choking on the kool-aid

 

[kazune_karin]

Besides, the i3 doesn't seems to has this kind of problem is because of i3 dual core itself doesn't allow it to happen due to its own power constrain or limit, while i5 with quad core is somehow more than what the macbook air could take under certain situation, thus i5/i7 users are facing the problem that wouldn't have appeared on the i3, not really apple engineers failed at something or whatsoever in my perspective.

If the work can be handled easily by dual core i3, why the work has to lead to unnecessary overheat on i5/i7? Isn't their FAILURE TO CONTROL the more powerful CPU? Some users here undervolt a bit to keep it control under stock heatsink. So is Apple too aggressive to boost unreasonable maximum power?

Also bear in mind this is just a 3 month old baby with merely 1 update so far, so we shall see further optimisation through os update to fix/balance these issues

Aren't they supposed to make sure no major issues before releasing a product? Apple has access to hardware parts several generations ahead of the public. I don't think they do not have time to check their design. IF THEY CANNOT FIND A SOLUTION, THEY SHOULD NOT RELEASE IT. Just look at the AirPower charging mat, Apple claims their product before they know they can make it.

Maybe I am critical on Apple. But I think since APPLE IS BRANDED AS SUPERIOR, THEY DESERVE TO BE MORE PERFECT THAN OTHERS. I like Apple, so I want to be great again.

2.you are a semi-power user and would like to have the option to upgrade with small amount of money, we have i5/i7 that will perform faster and 'smoother' than the base i3 model

It is a common logic (acceptable?) also to MacBook Pro 16 as well. If Apple can offer more power as an option, does it mean Apple does not need to upgrade to a SUFFICIENT thermal design as well? Are we assumed to bare less controlled thermals if we want more power?

 

[SuperMatt]

Just think how great it will be when we have ARM chips in the MacBook Airs. My iPad Pro seldom even gets warm.

 

[jackiez]

If the work can be handled easily by dual core i3, why the work has to lead to unnecessary overheat on i5/i7? Isn't their FAILURE TO CONTROL the more powerful CPU? Some users here undervolt a bit to keep it control under stock heatsink. So is Apple too aggressive to boost unreasonable maximum power?

isn't this referring to what i was talking about Optimisation? as ive said it's still too early for us to simply jump into conclusion, especially when this is just a 3 month old baby with merely 1 update so far, who knows if it's just optimisation problem? we are not engineer nor designer and common sense is not always common practice, theres no need to be unnecessarily aggressive / over-react over this kind of 'issue', to the extend where creating a problem when there isn't one, just because it is APPLE, seriously.

ive tried using turbo boost switcher, it works great only when u are really not doing any intensive task, but when it suddenly kicks in, everything started to get laggy, warm, and fan spins, struggling to get the extra power that has been disabled on purpose.

 

[adrianstuartt]

isn't this referring to what i was talking about Optimisation? as ive said it's still too early for us to simply jump into conclusion, especially when this is just a 3 month old baby with merely 1 update so far, who knows if it's just optimisation problem? we are not engineer nor designer and common sense is not always common practice, theres no need to be unnecessarily aggressive / over-react over this kind of 'issue', to the extend where creating a problem when there isn't one, just because it is APPLE, seriously.

ive tried using turbo boost switcher, it works great only when u are really not doing any intensive task, but when it suddenly kicks in, everything started to get laggy, warm, and fan spins, struggling to get the extra power that has been disabled on purpose.

i dont see how software fixes a hole in the heatsink lol

 

[jackiez]

i dont see how software fixes a hole in the heatsink lol

oh you thought it can be?

 

[christophosphorus]

Doesn't the laptop feel too warm at the bottom now?

I think the case itself is hotter to the touch, but to be fair, I rarely if ever tested this prior to the mods. My laptop sits on my desk 99% of the time, so the "hot on my lap" issue isn't something that'd bother me.

I wonder if getting a much thinner heat pad would fix that.

 

[Loog]

Just think how great it will be when we have ARM chips in the MacBook Airs. My iPad Pro seldom even gets warm.

@SuperMatt, Big Sur and Apple silicon here we come for the fall!! So glad I shipped my MBA 2020 I7 back and reverted to the MBA 2018 ... I feel a new unit coming for the end of the year

 

[srkirt]

[QUOTE = "Loog, publicación: 28586261, miembro: 1212829"]
@SuperMatt, Big Sur y Apple de silicio ¡¡¡¡¡Aquí venimos para el otoño !! Estoy tan contento de haber devuelto mi MBA 2020 I7 y volver al MBA 2018 ... Siento que vendrá una nueva unidad para fin de año
[/ CITAR]
Huawei X pro.... hahahahaha !!!

 

[kazune_karin]

oh you thought it can be?

I do not mind if they offer a hardware optimization: free logic board swapping program for Air ?

 

[jackiez]

I do not mind if they offer a hardware optimization: free logic board swapping program for Air ?

'hardware optimisation', you and that guy must be cute ?

 

[kazune_karin]

'hardware optimisation', you and that guy must be cute ?

I think you exactly agree how impossible a software optimization can do on hardware design failure, e.g. too wide gap in the heatsink

 

[IngerMan]

Hi there, sorry if you’ve already answered this somewhere, but what scores were you able to get with just the shim and no pad? For the paste, is there any difference between MX4 and something like Kryonaut (it’s easier for me to get the Kryonaut).

1st I did the pad first and then added the shim. So I don't know what the shim only does for me. But what I seen the Pad only is the biggest gain. 3500 Stock, 4000 Pad, 4200 Pad and paste and shim.

I can say technique is everything when doing the Paste and Shim. The heat sink screws are like the size of dog hair. The T4 end is invisible for a 55+ eyes.

I recommend very good lighting, extra lighting. A hand held magnify glass was so useful finding the fit of the bit and the screw head.

After I cleaned all the older paste and cleaned it with an approved alcohol cleaner on the inner heat sink and the CPU & GPU die, careful not to have excess liquid. I dampened my tips on a coffee filter.

A I put a thin layer of paste with a smudge tool provided with the paste in the cutout of the heat sink.

B I placed the shim on the paste of the inner heat sink.

C I put a thin layer of paste on the other open side of the shim that I placed in the heat sink.

D I put a medium layer of paste on the CPU & GPU. My .3mm shim is not displacing that much thermal paste I had to remove. I don't want to do this twice.

E I cleaned up excess blobs on everything where I got sloppy with the help of Q-tips. No left over paste on any edges just the inner heat sink and inner CPU and GPU dies.

F I lined up the heat sink over the screw holes and set it down. Using the magnifying glass again I put it back together with decent torque, no strippey, winner winner Chicken Dinner. I cleaned the top of the heat sink before I put the pad back on.

 

[fcracer]

1st I did the pad first and then added the shim. So I don't know what the shim only does for me. But what I seen the Pad only is the biggest gain. 3500 Stock, 4000 Pad, 4200 Pad and paste and shim. Do the math.

I can say technique is everything when doing the Paste and Shim. The heat sink screws are like the size of dog hair. The T4 end is invisible for a 55+ eyes.

I recommend very good lighting, extra lighting. A hand held magnify glass was so useful finding the fit of the bit and the screw head.

After I cleaned all the older paste and cleaned it with an approved alcohol cleaner on the inner heat sink and the CPU & GPU die, careful not to have excess liquid. I dampened my tips on a coffee filter.
View attachment 926260 View attachment 926261
A I put a thin layer of paste with a smudge tool provided with the paste in the cutout of the heat sink.

B I placed the shim on the paste of the inner heat sink.

C I put a thin layer of paste on the other open side of the shim that I placed in the heat sink.

D I put a medium layer of paste on the CPU & GPU. My .3mm shim is not displacing that much thermal paste I had to remove. I don't want to do this twice.

E I cleaned up excess blobs on everything where I got sloppy with the help of Q-tips. No left over paste on any edges just the inner heat sink and inner CPU and GPU dies.

F I lined up the heat sink over the screw holes and set it down. Using the magnifying glass again I put it back together with decent torque, no strippey, winner winner Chicken Dinner. I cleaned the top of the heat sink before I put the pad back on.

You're awesome, thank you for the detailed instructions. The Kryonaut and 0.3MM shim arrive later this week and I'll do my best not the break the MBA

For the pad, on my i7 I found the bottom case getting too hot during my photo editing sessions in Capture One. Interestingly, it's not during sustained workloads that the case heats up so much, it's during continuous high frequency bursty work that engages the GPU.

When I was running Cinebench back to back, the case got hot, but not worse than my 2018 MBP 13" i7, however when flipping through images in Capture One where the CPU and GPU hit max utilization, the temp gets hot enough to be uncomfortable on my thighs (I primarily use this laptop on the sofa on my lap).

I cut the pad down to a size sufficient to just touch the four screws of the heat sink. It's a very small sized pad that sits on top of where the CPU/GPU reside, leaving the rest of the heat sink to dissipate the heat through the air. Two things of note, the bottom case is cooler, and the Geekbench scores went down consistently by 100pts (from a cold start, it's now 3800 vs. 3900 and when warm, it's now 3400 vs. 3500). Cinebench surprisingly didn't change much with ~1132 scores when warm, which must mean there is some hard frequency limit that I've hit.

The fan also spins up more frequently now which is something I wanted because it rarely was audible with the larger pad. Why would I want to have the fans spin up when everyone else wants them to be silent? Under very heavy loads, I'd prefer that the fan dissipate the heat through the air rather than the thermal pad to my thighs.

Assuming I'm successful in the shim installation, I'll report back on any improvements. The ideal situation would be to replicate your increase in performance (+100-200) and get a consistent 3,600 when warm and 4000 when cold with the smaller pad and the shim mod. Perhaps I can even consider dumping the pad to keep my thighs nice and cool...

 

[jackiez]

I think you exactly agree how impossible a software optimization can do on hardware design failure, e.g. too wide gap in the heatsink

the only person who thinks thats possible is you and that guy, who failed to simply understand whats os optimisation and being blindly stubborn over something you don't really know (sorry to burst your bubble?) yet insist in twisting other people's words saying 'software could physically change hardware' bs.

 

[ilikewhey]

In what way ? Are they slower than your 2014 mba ?

didnt have my old mba for direct compare but, from recalling my old mba didn't ramp up its fan nor hot to the touch as quick.

 

[kazune_karin]

the only person who thinks thats possible is you and that guy, who failed to simply understand whats os optimisation and being blindly stubborn over something you don't really know (sorry to burst your bubble?) yet insist in twisting other people's words saying 'software could physically change hardware' bs.

Sorry to put it a bit bold.
I just want to point that there are software issues which can be fixed by OS update; and very likely hardware design failures/restrictions (intended restriction against MBP, ease of assembly but not making things so tight, etc.) which cannot.

The fantastic guys here who applied shim mod agreed that the heatsink gap is too wide to make the annoying spiking thermal/fan profile in especially i5/i7 Air 2020. Even if there is a design failure Apple will (and obviously) cannot admit, as you see for the butterfly keyboard design.
P.S. By the way I owned a 2015 MacBook 12 also, and my keyboard was always find. Maybe my environment is not so dusty.

Since the heatsink-related design problem cannot be fixed by software (macOS) optimization (well, Apple can lower the turbo boost setting to minimize that though), a shim mod (actually make the heatsink gap less) fixed the problem. Some recent new buyers found that they do not have that severe problem (other than OS version), so maybe Apple quietly improved the design which however we are difficult to know. And even if Apple improves the design we early adopters will not likely receive an official fix in the near future.

In the end, Apple may be right for most of the time, but it may not always be right.

 

[agaskew]

isn't this referring to what i was talking about Optimisation? as ive said it's still too early for us to simply jump into conclusion, especially when this is just a 3 month old baby with merely 1 update so far, who knows if it's just optimisation problem? we are not engineer nor designer and common sense is not always common practice, theres no need to be unnecessarily aggressive / over-react over this kind of 'issue', to the extend where creating a problem when there isn't one, just because it is APPLE, seriously.

ive tried using turbo boost switcher, it works great only when u are really not doing any intensive task, but when it suddenly kicks in, everything started to get laggy, warm, and fan spins, struggling to get the extra power that has been disabled on purpose.

Just re read the thread . There are numerous, documented examples posted here that show one or more modifications to the default cooling system deliver tangible and measurable improvements to the performance of these laptops. This is not 'creating a problem when there isn't one'. There are many reports in this thread of some use cases, notably video conferencing, that induce high system usage, fan spin up, noise and throttling. You seem to be implying that it is OK to ship products with sub-par performance or extra power that has been disabled on purpose, and just wait until some mysterious 'Optimisation' is delivered that will fix the issues.

 

[jackiez]

Just re read the thread . There are numerous, documented examples posted here that show one or more modifications to the default cooling system deliver tangible and measurable improvements to the performance of these laptops. This is not 'creating a problem when there isn't one'. There are many reports in this thread of some use cases, notably video conferencing, that induce high system usage, fan spin up, noise and throttling. You seem to be implying that it is OK to ship products with sub-par performance or extra power that has been disabled on purpose, and just wait until some mysterious 'Optimisation' is delivered that will fix the issues.

this is a heatsink modification thread not 'fixing apple's failed engineering'. you only see what you want to see perhaps you should also read Erehy Dobon and few other comments regarding the thermal system then you know if its 'creating a problem when there isn't one' or not. We all can agree that butterfly keyboard and macbook pro flexgate issue is the real failing design or oversight from apple engineers which eventually lead to a huge issue, but it is different story here.

all these hiccups that have been mentioned over and over again could possibly be optimisation problem for now, and ive said it wont be a drastic change even after optimisation but improvement to what it should be over time and how funny y'all complicate yourself and twist my words where i find it hilarious and also speechless at the same time. This happens is all due to over-expectation over an ultra thin entry level laptop and overreact on how a 3 month old baby behave.

by one of the forumer Erehy Dobon:
Every design choice is a compromise: power, weight, bulk, cost, noise, convenience, battery life, etc.

The point of a MacBook Air isn't to harness the maximum capability of the CPU 100% 24x7. Maybe they could add a heat pipe and a CPU fan but that would force an increase in the case thickness. Maybe you need a bigger battery to handle the power needs. All of sudden you end up with a MacBook Pro.

this says it all, Macbook air is Macbook air and it was only built to target at casual user for casual use in the very first place, its a bonus where it still capable of performing some power intense task, just same like any other windows laptop at the same category. if you want power and consistent performance, seriously i believe an extra few hundred bucks is not really a matter for yall, go for pro series already and most of your problems will be solved instantly, simple as that. ?‍♂️