2020 Air Heatsink Modification Thread

[deeddawg]

Hello everyone! I'm waiting for my first ever Macbook Air 2020 (i3 model) and reading all this staff I'm still confused should I get my order or cancel it. I read this topic a lot of times, watch many videos about heating issues and now have a question.

On YouTube guy TheJuan&Only made direct comparison between three models: base 2019 Air, base 2020 Air (i3) and 2020 Air with i5. So he did some casual tests first, launched browser with several tabs open and Apple TV+ running video. Both 2019 and 2020 i3 Airs had similar CPU temperatures (56 versus 61 degrees (check screenshot 1)). Next he added more apps: calculator, text editor etc. In this situation 2020 Air i3 CPU was 88 degrees, but 2019 model was 80 too (check screenshot 2)! Pretty similar, huh? Next step: he launches final cut, photoshop and other more demanding apps. On both models once again CPU temperatures are similar (74 vs 77, 89 vs 92, 99 vs 94, check screenshots 3,4,5)). I did not screened Air i5 2020, because in heavy tasks CPU temperature was about 95-99 all time.

So what is my question is. I guess Air 2019 model and Air 2020 model have similar cooling systems, right? During same tasks both Air 2019 and Air 2020 (i3) have similar CPU temperatures. But Air 2019 was released about a year ago and if this kind of pressure really hurts processor wouldn't we see across all the internet dozens of messages about malfunction and failure, like it was with butterfly keyboard, for example? And this kind of workload is pretty normal and does not kill your notebook? Why topic of overheating was not raised when 2019 model was released nevertheless data pretty the same?

Thanks for this forum, all your discussions and posts. They are very helpful and it have big impact on my final choice. P.S. Sorry for my English, it's not my native language.

Get the machine, use it for a week to ten days, decide whether or not it meets your needs, then keep or return it based on that assessment.

 

[KrinklesMontgomery]

I had the 2013 MacBook Air, and don't think that the 2020 Air perform worse. What exactly did you try running on it? I have run Windows 10 on a virtual machine, and it is fine. It seems your complaints are primarily around the CPU temperature and battery life. I haven't had it long enough to observe the battery life (particularly since I'm not traveling right now), but the CPU temperature doesn't translate into a hot chassis. I have only noticed the fans when running something CPU-intensive, such as running a Windows Update in Parallels.

In any case, the 13" Pro is coming soon, along with the upgraded keyboard and presumably a boost in base storage, so that may be better suited for you while not being that much more expensive.

It would have taken over the same role as my old 2014 MBA, which I use for Zoom, video/movie streaming, virtual machines etc. By 'worse' I do mean that the 2020 got hotter and louder and with worse battery life than my 2014 MBA, not that there was noticeable lag or stutter. 100 degrees as a one off doesn't mean a hot chassis true, but when it's sitting near there for hours it does get hot. Luckily I managed to restore my 2014 so I don't need the 2020. If I needed to use the 2020 it probably would have been, to quote others here, 'fine'.

 

[Loog]

Yes I have a 2020. I consume video, music, excel, word, powerpoint, edit pdfs, light photo editing, outlook, mail, zoom, terms, messages, VPN, safari, chrome and a couple proprietary work related apps. All with no heat issues and minimal to no fan. Fan does come on (inaudible) with zoom and teams.

Really ... you're lucky, my experience with teams was horrendous, 15% CPU load, 100 deg and 8000 rpm after less than 10 mins. Zoom was fine so this may be issues in software, but never the less

 

[Bolanders]

Really ... you're lucky, my experience with teams was horrendous, 15% CPU load, 100 deg and 8000 rpm after less than 10 mins. Zoom was fine so this may be issues in software, but never the less

Zoom and Teams definitely pushes mine. Got to approaching 100 degrees but the fan was inaudible unless you put your ear up to it. i5/16/1TB. Not sure which you had but even pushed when using those apps I can live with.

 

[srkirt]

Hi Jordi!

I must be the only person skipping through this thread looking for your updates, because you seem to be the only person proactively trying to improve the thermals of your 2019 Air. I am really tempted to see if I can do the same mod on my 2020 Air, but I need to order some tools and materials first.

• What are the dimensions of the "cavity" in the heatsink that needs to be filled in? and how thick is the gap?
• What screws are holding the heatsink in place? Are they Philips or Pentalobe, and what screw size?
• Do you think using foam to restrict the airflow has made a noticeable difference?

Hi !! Most of the things you ask me I cannot give you since I have a 2019 ... but I will help you in some.
The screws are extremely small and pentalobe, they resemble those of the Iphone. With the pair of plates you have to tighten crosswise because the screw does not reach the end, I have tightened it strongly and it has not broken ... The thing about covering the gaps, of course it shows in the outflow, practically I have Reinforced the shape of the original tunnel plus cover gaps, even in the fan I have put a rubber with a semicircle shape.
Otherwise you will find photos of me on the channel.

 

[srkirt]

I'm going to tell you about a horrible experience ...
We already know that touching these things is very delicate and that until today I have not seen the battery connector, so when I work there is always a white mini LED on the right of the CPU ... the connector has a flap to stretch and disconnect. the batteries and that light goes out.
Well, since I carry thermal paste everywhere I have dedicated myself to cleaning it with a cloth with soapy water (which is difficult to remove) and without realizing it I have crossed the plate where the battery cable near the turbine ends. The mini led has started to flash and turned off !! no longer started !!! I have removed the plastic protection and I have seen wet components ¿?? I don't know ... with a brush and WD40 I have cleaned the components well, I have dried and I have reconnected the batteries. Well it didn't work !!!
**** !!! I was sweating and it would not start because I had dampened it a bit (although I have seen matte welds from the factory ...) Without being able to eat and crazy I was given to do another test, I have put the power supply and the led has turned on !!! I have started the machine and it has worked !!! I have stopped it and I have disconnected the power supply and then the led if it worked !! I have started it with the batteries and so far ...
Warning: if you remove the battery connector to do something when putting it back it DOES NOT WORK !!! you have to connect it to the power supply and once it has started it is ready. It should be a type of protection for the power supply of the batteries.
What a day...
Well one more experience and with finishing lining the entire tunnel I have done a fabulous performance !!! lowest temp !!! and at 8,000rpm you can feel the air in your hand, something that you didn't before.
Photo 1: conector batery.
Photo 2: microled.
Photo 3: sealed with adhesive rubber.
Photo 4: Temp Idle browsing a single folder.

 

[kinchee87]

Hi !! Most of the things you ask me I cannot give you since I have a 2019 ... but I will help you in some.
The screws are extremely small and pentalobe, they resemble those of the Iphone. With the pair of plates you have to tighten crosswise because the screw does not reach the end, I have tightened it strongly and it has not broken ... The thing about covering the gaps, of course it shows in the outflow, practically I have Reinforced the shape of the original tunnel plus cover gaps, even in the fan I have put a rubber with a semicircle shape.
Otherwise you will find photos of me on the channel.

Thanks for your answers. You said that you had to tighten the heatsink crosswise --> Is that because your metal shims are too thick and they lift up the heatsink?

I guess I will have to find copper shims of different thicknesses, and probably get the ifixit repair tools for macs.

There was one other person (Robotronic) who managed to mod their 2020 Air, but unfortunately they didn't take photos for reference: https://forums.macrumors.com/threads/2020-air-heatsink.2227066/page-27?post=28373009#post-28373009

That was a scary story you had with the battery connector! But now I can see the wind tunnel you've created.

 

[AirBud]

Well it looks like I’m going to give this mod a try on my 2020 i5. 😳

The 20x20x0.5mm copper shim, pentalobe screwdriver kit, and thermal paste were $25 after tax. The shipping for the shim is 2 weeks so I won’t have anything to report for a while but I’ll be back.

 

[RiaKoobcam]

For the sake of everyone who tries it, please either take a video or lots of pictures! Godspeed to you AirBud! See you on the other side of the mod!

 

[KPOM]

Purely as an experiment, I placed my MacBook Air (i7) on a synthetic “ice pack” while it was running a CPU-intensive task (CPU around 99%) I noticed a few things:

• The temperature dropped from 98-100 to around 94-96%
• Average core speed went from 1.55-1.6GHz to around 1.7-1.8GHz. Interestingly, the “requested” CPU speed eventually dropped from 3.8GHz (suggesting a requested maximum boost) to 1.7-1.8GHz. I’m guessing that this is the voltage limiter kicking in and noting that it has “normalized” at a slightly higher level.
• Power consumption increased from about 10W (total) to about 11-12W
• The fans turn off

I don’t recommend this long term (the case gets cold to the touch and could cause condensation), but thought I’d post it, since it’s a lot less complicated than rigging a heat sink. I think this shows with a more aggressive cooling system, perhaps Apple could have squeezed another 10-15% CPU performance, but it wouldn’t turn the Air into the Pro since the Y-series processors just aren’t built to run at high speeds for long periods of time.

 

[srkirt]

[QUOTE = "AirBud, publicación: 28390141, miembro: 1213549"]
Bueno, parece que voy a probar este mod en mi 2020 i5. 😳

La cuña de cobre de 20x20x0.5 mm, el kit de destornillador pentalobe y la pasta térmica costaban $ 25 después de impuestos. El envío para la cuña es de 2 semanas, así que no tendré nada que informar por un tiempo, pero volveré.
[/CITAR]
👍
[automerge]1587510044[/automerge]
[QUOTE = "KPOM, publicación: 28390166, miembro: 505893"]
Simplemente como un experimento, coloqué mi MacBook Air (i7) en una "bolsa de hielo" sintética mientras ejecutaba una tarea intensiva de CPU (CPU alrededor del 99%) Noté algunas cosas:

• La temperatura bajó de 98-100 a alrededor de 94-96%
• La velocidad promedio del núcleo pasó de 1.55-1.6GHz a alrededor de 1.7-1.8GHz. Curiosamente, la velocidad de la CPU "solicitada" finalmente se redujo de 3.8GHz (lo que sugiere un aumento máximo solicitado) a 1.7-1.8GHz. Supongo que este es el limitador de voltaje que entra en acción y noto que se ha "normalizado" en un nivel ligeramente más alto.
• El consumo de energía aumentó de aproximadamente 10 W (total) a aproximadamente 11-12 W
• Los ventiladores se apagan

No recomiendo esto a largo plazo (el estuche se enfría al tacto y podría causar condensación), pero pensé en publicarlo, ya que es mucho menos complicado que manipular un disipador de calor. Creo que esto se muestra con un sistema de enfriamiento más agresivo, tal vez Apple podría haber exprimido otro rendimiento de CPU del 10-15%, pero no convertiría el Air en Pro ya que los procesadores de la serie Y simplemente no están diseñados para funcionar a un alto rendimiento velocidades por largos períodos de tiempo.
[/CITAR]
you should have done a Cinebench test !!!
[automerge]1587510274[/automerge]
When this cover comes to me, I remove the original and make my own tunnel ... it will also be very original to see the interior ... What do you think of the idea?
[automerge]1587510554[/automerge]
Original test MBA 2020.
Posible: 1050 y 400

 

[srkirt]

Well it looks like I’m going to give this mod a try on my 2020 i5. 😳

The 20x20x0.5mm copper shim, pentalobe screwdriver kit, and thermal paste were $25 after tax. The shipping for the shim is 2 weeks so I won’t have anything to report for a while but I’ll be back.

Paddy Juego de Destornilladores,25 en 1 Precisión Destornilladores Set para iPhones,PC,Xbox,Cámara,Reloj,Tablet,PC,Gafas y otros Electrónicos etc Screwdriver Set

25 en 1 Mini Destornillador Precisión Kit de Destornillador Herramientas de Reparación para Para Smartphone,Laptop,Reloj,Joyería,Juguetes,iPhones,Gafas etc Destornillador incluye: 1 x destornillador de mango Herramienta de extracción de tarjetas SIM. (0.8) Destornillador Pentagonal (0.8;1.2) Des...

www.amazon.es

 

[srkirt]

Solution total... jajajaja

 

[Robotronic]

There was one other person (Robotronic) who managed to mod their 2020 Air, but unfortunately they didn't take photos for reference: https://forums.macrumors.com/threads/2020-air-heatsink.2227066/page-27?post=28373009#post-28373009

The 20x20x0.5mm copper shim, pentalobe screwdriver kit, and thermal paste were $25 after tax. The shipping for the shim is 2 weeks so I won’t have anything to report for a while but I’ll be back.

Again, sorry for not taking photos, although it wouldn't have help trying to get a measurement for the thickness of shim needed. I don't have a pair of digital calipers unfortunately. I do have a set of feeler gauges though, so I made another small piece of foil-paste-foil-paste-foil sandwich just now, and compared the best I could visually and by feel to the feeler gauges. My best guesstimate is that the thickness is about 0.35mm-0.40mm. It definitely feels smaller than 0.5mm, although a 0.5mm shim may still work. My concern with a shim any thicker than what I rigged up is that the screws securing the heatsink are very short, so there won't be many threads holding it in place, the thicker the sim, the fewer threads will grab. And speaking of screws, the bottom case ones are pentalobe, but the heatsink ones are torx, very small torx.

 

[KPOM]

[QUOTE = "AirBud, publicación: 28390141, miembro: 1213549"]
Bueno, parece que voy a probar este mod en mi 2020 i5. 😳

La cuña de cobre de 20x20x0.5 mm, el kit de destornillador pentalobe y la pasta térmica costaban $ 25 después de impuestos. El envío para la cuña es de 2 semanas, así que no tendré nada que informar por un tiempo, pero volveré.
[/CITAR]
👍
[automerge]1587510044[/automerge]
[QUOTE = "KPOM, publicación: 28390166, miembro: 505893"]
Simplemente como un experimento, coloqué mi MacBook Air (i7) en una "bolsa de hielo" sintética mientras ejecutaba una tarea intensiva de CPU (CPU alrededor del 99%) Noté algunas cosas:

• La temperatura bajó de 98-100 a alrededor de 94-96%
• La velocidad promedio del núcleo pasó de 1.55-1.6GHz a alrededor de 1.7-1.8GHz. Curiosamente, la velocidad de la CPU "solicitada" finalmente se redujo de 3.8GHz (lo que sugiere un aumento máximo solicitado) a 1.7-1.8GHz. Supongo que este es el limitador de voltaje que entra en acción y noto que se ha "normalizado" en un nivel ligeramente más alto.
• El consumo de energía aumentó de aproximadamente 10 W (total) a aproximadamente 11-12 W
• Los ventiladores se apagan

No recomiendo esto a largo plazo (el estuche se enfría al tacto y podría causar condensación), pero pensé en publicarlo, ya que es mucho menos complicado que manipular un disipador de calor. Creo que esto se muestra con un sistema de enfriamiento más agresivo, tal vez Apple podría haber exprimido otro rendimiento de CPU del 10-15%, pero no convertiría el Air en Pro ya que los procesadores de la serie Y simplemente no están diseñados para funcionar a un alto rendimiento velocidades por largos períodos de tiempo.
[/CITAR]
you should have done a Cinebench test !!!
[automerge]1587510274[/automerge]
When this cover comes to me, I remove the original and make my own tunnel ... it will also be very original to see the interior ... What do you think of the idea?
[automerge]1587510554[/automerge]
Original test MBA 2020.
Posible: 1050 y 400

Cinebench R20
- 1.62GHz, 9.8W, 1084
- With Ice Pack: 1.9GHz, 12W, 1204

Consistent with my previous experiment. Apple could have extracted maybe another 10-15% with better cooling, but this isn't a MacBook Pro.

 

[magbarn]

Cinebench R20
- 1.62GHz, 9.8W, 1084
- With Ice Pack: 1.9GHz, 12W, 1204

Consistent with my previous experiment. Apple could have extracted maybe another 10-15% with better cooling, but this isn't a MacBook Pro.

I don’t understand the huge discrepancy in performance. I’m going to have to run it again on my i7 as you’re getting better numbers than me. Are you running in windows or macOS?

 

[KPOM]

I don’t understand the huge discrepancy in performance. I’m going to have to run it again on my i7 as you’re getting better numbers than me. Are you running in windows or macOS?

MacOS.

 

[throAU]

Solution total... jajajaja

Sure looks like an ultraportable thin/light machine to me!

Cinebench R20
- 1.62GHz, 9.8W, 1084
- With Ice Pack: 1.9GHz, 12W, 1204

So I guess we confirmed that even with additional, sub ambient cooling, you're looking at maybe 10% additional performance on a CPU bound benchmark (that isn't really relevant to real world use that this machine is aimed at), and that the CPU is power limited by design in any case as per intel TDP ratings and boost behaviour in late model processors.

 

[dylin]

This was all to see how much performance could be gained.

I'm not sure if someone posted this already, or if they did, if they mentioned the fact that even with a high end cooling solution, the MacBook Air still wouldn't allow for a higher all core turbo when the CPU was at 50º Celsius.

Even with the CPU at 50º, it locked itself to 1.8 GHz.
I'm not sure if that was what intel intended for its sustained all core boost, but if there was a way to get a cooling mod to be cool enough just to keep the CPU at 1.8 GHz for an all core turbo, I would say that would be great.

He mentioned that with the stock cooling, it was turbo-ing to about 1.5 GHz.

 

[RiaKoobcam]

Purely as an experiment, I placed my MacBook Air (i7) on a synthetic “ice pack” while it was running a CPU-intensive task (CPU around 99%) I noticed a few things:

• The temperature dropped from 98-100 to around 94-96%
• Average core speed went from 1.55-1.6GHz to around 1.7-1.8GHz. Interestingly, the “requested” CPU speed eventually dropped from 3.8GHz (suggesting a requested maximum boost) to 1.7-1.8GHz. I’m guessing that this is the voltage limiter kicking in and noting that it has “normalized” at a slightly higher level.
• Power consumption increased from about 10W (total) to about 11-12W
• The fans turn off

I don’t recommend this long term (the case gets cold to the touch and could cause condensation), but thought I’d post it, since it’s a lot less complicated than rigging a heat sink. I think this shows with a more aggressive cooling system, perhaps Apple could have squeezed another 10-15% CPU performance, but it wouldn’t turn the Air into the Pro since the Y-series processors just aren’t built to run at high speeds for long periods of time.

I don't like MaxTech's videos/sensationalism at all - I think KPOM summed it up far better with far less clickbait and theatrics.

 

[KPOM]

This was all to see how much performance could be gained.

I'm not sure if someone posted this already, or if they did, if they mentioned the fact that even with a high end cooling solution, the MacBook Air still wouldn't allow for a higher all core turbo when the CPU was at 50º Celsius.

Even with the CPU at 50º, it locked itself to 1.8 GHz.
I'm not sure if that was what intel intended for its sustained all core boost, but if there was a way to get a cooling mod to be cool enough just to keep the CPU at 1.8 GHz for an all core turbo, I would say that would be great.

He mentioned that with the stock cooling, it was turbo-ing to about 1.5 GHz.

Yes, this video and a follow up pretty much started this thread.

 

[throAU]

I don't like MaxTech's videos/sensationalism at all - I think KPOM summed it up far better with far less clickbait and theatrics.

That's because @KPOM isn't a YouTuber who relies on angry clickbait reactions to pay his mortgage.

 

[Loog]

Zoom and Teams definitely pushes mine. Got to approaching 100 degrees but the fan was inaudible unless you put your ear up to it. i5/16/1TB. Not sure which you had but even pushed when using those apps I can live with.

i7/16/1TB, the noise is very auditable and the device was very hot, I can even hear this above the Dell R230 which is sitting at the side of me. Noise levels drop to inaudible around 7k mark. Very disappointing to get so hot so fast from just basic tasks such as VC.

 

[ipos]

some say i3 heat and fan noise is not as bad compared to i5 model
any i3 users can attest to that?

 

[RegularGuy09]

Purely as an experiment, I placed my MacBook Air (i7) on a synthetic “ice pack” while it was running a CPU-intensive task (CPU around 99%) I noticed a few things:

• The temperature dropped from 98-100 to around 94-96%
• Average core speed went from 1.55-1.6GHz to around 1.7-1.8GHz. Interestingly, the “requested” CPU speed eventually dropped from 3.8GHz (suggesting a requested maximum boost) to 1.7-1.8GHz. I’m guessing that this is the voltage limiter kicking in and noting that it has “normalized” at a slightly higher level.
• Power consumption increased from about 10W (total) to about 11-12W
• The fans turn off

I don’t recommend this long term (the case gets cold to the touch and could cause condensation), but thought I’d post it, since it’s a lot less complicated than rigging a heat sink. I think this shows with a more aggressive cooling system, perhaps Apple could have squeezed another 10-15% CPU performance, but it wouldn’t turn the Air into the Pro since the Y-series processors just aren’t built to run at high speeds for long periods of time.

This means two things in my humble opinion :

- The Air performs like its supposed to with this CPU. Since its a Y series processor, its meant to work in bursts, and not meant for sustained loads.

- With better cooling, the Air would have run quieter. Not really sure why Apple didnt go for that. I mean its not really market segmentation either is it?

My question for the Air is very different. For me, Im not really getting whats the point of having the 13" Pro with the 15W processor, and the 13" Air with the 10W processor. The Pro has better performance, and similar battery life, form factor and weight.

Why even have the Air then? Or, why not just stick the 15W processor in the Air? What is the tradeoff really going to be?