Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
2020 Air Heatsink Modification Thread
- Thread starter ilikewhey
- Start date
- Sort by reaction score
Slightly off topic: I know from @srkirt that the 2019 Air also has this gap between the CPU and the heatsink. Does anyone know if that's true for the 2018 Air also?
I'm currently testing a 2018 Air for light "on the road" use, and if I decide to keep it, I want to know if I should do the copper shim mod on this as well …
I'm currently testing a 2018 Air for light "on the road" use, and if I decide to keep it, I want to know if I should do the copper shim mod on this as well …
From what I've read, the heat sink design on the 2018 and 2019 is identical, so, yep.
Thanks very much rba1989 (post #1149 above). Very interesting!
My initial reactions to that are:
1. Your 2020 Air i5 reduces from about 3.1 GHZ to 2.0 GHZ after two minutes, compared to 3.3 down to 2.9 GHZ after a shorter time on my MacBook, which supports what Dcicer1 said...it throttles more than the MacBook.
2. The pkg power in the 2020 Air stabilises at 6-8 W in the throttled back state, while my MacBook flattens at 8-9 watts. This is a bit surprising to me as I thought the 2020 Air was higher wattage processor than the MacBook. I don't know what the light and dark blue lines signify.
3. The 2020 i5 reaches 100C in a few seconds, while it takes the fanless MacBook about a minute to reach 100C
4. The fan in the 2020 Air is working hard while the above is going on, which does rather support the idea that the fan isn't helping the CPU much, maybe keeping the case cooler?
Most of this long thread has been comparing the Air with MBP, I just wanted to see if its performance made sense in relation to its lesser cousin the MacBook. Seems the i5 is not as as well cooled as the Macbook (for their respective loads).
It would be interesting to see how the two-core i3 measures compared to the MacBook as they are much more similar. Dcicer1's comment that it never throttled in 10 minutes is consistent with the MacBook. Sounds like a silent fanless i3 2020 Air would be viable and have similar throttling to the MacBook. OTOH since users report it doesn't come on much, maybe not a lot of point.
My initial reactions to that are:
1. Your 2020 Air i5 reduces from about 3.1 GHZ to 2.0 GHZ after two minutes, compared to 3.3 down to 2.9 GHZ after a shorter time on my MacBook, which supports what Dcicer1 said...it throttles more than the MacBook.
2. The pkg power in the 2020 Air stabilises at 6-8 W in the throttled back state, while my MacBook flattens at 8-9 watts. This is a bit surprising to me as I thought the 2020 Air was higher wattage processor than the MacBook. I don't know what the light and dark blue lines signify.
3. The 2020 i5 reaches 100C in a few seconds, while it takes the fanless MacBook about a minute to reach 100C
4. The fan in the 2020 Air is working hard while the above is going on, which does rather support the idea that the fan isn't helping the CPU much, maybe keeping the case cooler?
Most of this long thread has been comparing the Air with MBP, I just wanted to see if its performance made sense in relation to its lesser cousin the MacBook. Seems the i5 is not as as well cooled as the Macbook (for their respective loads).
It would be interesting to see how the two-core i3 measures compared to the MacBook as they are much more similar. Dcicer1's comment that it never throttled in 10 minutes is consistent with the MacBook. Sounds like a silent fanless i3 2020 Air would be viable and have similar throttling to the MacBook. OTOH since users report it doesn't come on much, maybe not a lot of point.
Last edited:
We don't know for sure, but it's either intentional (to maintain a gap to the 13" MBP) or just an oversight. The surface of the heatsink that's supposed to touch the CPU is milled out slightly, as if the CPU was thought to build up higher.
I somehow don't really believe it's intentional, but on the other hand - you'd think that they would've corrected it by now if it was just an oversight/mismatch …
I think it is more valid to compare the 2020 MBA with the 2020 MBP, this because they both mount the tenth generation of Intel processors. Compared to your Macbook, would you be able to test with geekbench 5? With the results obtained, we can check with certainty whether your i7 is less performing than my i3. The dissipation system is fundamental, and since both are not equipped with a dedicated fan (we have ascertained that the MBA fan is not needed), we can observe which of the 2 makes it better for the same load ...
Attachments
Thats very good for an i3, the only 2020 i3 I could find in the database was 1003/1998. Is that because of the cooling mod ?
Yes more valid to compare with the MBP, but it is also interesting to compare the two core partially cooled Air with the two core uncooled MacBook.
Attached is my MacBook i7. There is a bigger difference than I thought.
Attachments
Yes, I think so.. and also because it never throttled..
So, 2020 i3 is more powerful than 2018\2019 i5 MBA and 2017 i7 Macbook, I'd say not bad..
If only it had a heat pipe.. damned apple
Last edited:
Ok on the way..
MBA 2020 i5/16/256
Gelid solutions GC Extreme
0,4mm shim
iFixit toolkit
I’ll post results as soon as i can ?
MBA 2020 i5/16/256
Gelid solutions GC Extreme
0,4mm shim
iFixit toolkit
I’ll post results as soon as i can ?
I simply wonder if a copper shim is really needed. There are silicone thermal pads 0.5 mm thick already available on the market. Why copper? Also, if we go the metals route, why not gold? It's not really as expensive as you imagine, and it's available for gold plating in some shops. Gold has better thermal conductivity than copper, AFAIK. I am just asking here, I am by no means an expert. My 2020 MBA i5/16/512 is on it's way and I am already thinking to mod it. So am I am looking for the best possible solution. If it's an intended/not intended design flaw it seems it can be fixed somehow.
Silicone pads aren't even close in thermal conductivity to copper. If you go for a silicone pad, you might as well do nothing and keep the stock blob of thermal paste.
Hi guys I have tried removing all the tires from the wind tunnel and it has dropped from 700 more or less to 650 !!! you have to close all the holes that there are with foam rubber so that the flow that enters is distributed part above the plate and part below ... I add to that we add that it has small entrances and exits only half passes through the cpu of the little air that arrives ...
Attachments
Pre production samples are now made and in the post so will be with me soon. I'l be looking for some good people to give this a test and provide their feedback on this thread.
These are a limited run for now until I can establish if they are just drop in or if additional modifications are needed to the program. All are machine cut via water jet and then hand finished. I have 10 in 0.3mm and 10 in 0.5mm both thicknesses appear to work in all retina MBA.
Shim are at production cost with postage at fixed international rates for UK, EU and world zone 1-2 only.
To be transparent here, material has cost me £52 (some surplus and waste for trials), machine time was a fixed rate of £100 + vat (£120). Postage is £2.75 with hard back envelope for most of the world. The Project to date has cost me £172 in layout costs and with postage will come to £227 so these work out at £11.35 each.
I'm happy to ship these out at a flat rate of £10 as postage will be less for some countries i.e. UK and EU so I hope that it all balances out. I work full time so I'm trying to keep things simple. I can get to the post office twice a week so there may be a delay in me getting these out. Be cognisant of this before replying if you are in a rush.
Interest by PM please with your thickness requirement, however some are likely to go quicker than others but both should fit. If you would like to try both then this is £17.25 with flat postage. For now just register your interest and I'll let you know the next steps once I have them in hand and have obtained envelopes.
I appreciate that you can buy the 20mm x 20mm for considerably less and that may be your preferred route. These are custom made 16.5mm x 22mm +- tolerance with radius edges so should be a straight forward install without any fuss. I'm looking to cover my outlay here not make a living. If successful and enough interest I may get some more made which by volume would be more cost effective, not eBay china prices but then I'm not planning to make a living out of this, just feed back into the community.
Let me know your thought ...
These are a limited run for now until I can establish if they are just drop in or if additional modifications are needed to the program. All are machine cut via water jet and then hand finished. I have 10 in 0.3mm and 10 in 0.5mm both thicknesses appear to work in all retina MBA.
Shim are at production cost with postage at fixed international rates for UK, EU and world zone 1-2 only.
To be transparent here, material has cost me £52 (some surplus and waste for trials), machine time was a fixed rate of £100 + vat (£120). Postage is £2.75 with hard back envelope for most of the world. The Project to date has cost me £172 in layout costs and with postage will come to £227 so these work out at £11.35 each.
I'm happy to ship these out at a flat rate of £10 as postage will be less for some countries i.e. UK and EU so I hope that it all balances out. I work full time so I'm trying to keep things simple. I can get to the post office twice a week so there may be a delay in me getting these out. Be cognisant of this before replying if you are in a rush.
Interest by PM please with your thickness requirement, however some are likely to go quicker than others but both should fit. If you would like to try both then this is £17.25 with flat postage. For now just register your interest and I'll let you know the next steps once I have them in hand and have obtained envelopes.
I appreciate that you can buy the 20mm x 20mm for considerably less and that may be your preferred route. These are custom made 16.5mm x 22mm +- tolerance with radius edges so should be a straight forward install without any fuss. I'm looking to cover my outlay here not make a living. If successful and enough interest I may get some more made which by volume would be more cost effective, not eBay china prices but then I'm not planning to make a living out of this, just feed back into the community.
Let me know your thought ...
Attachments
Last edited:
So it was going to be local ... you are unpresentable and I hope you spend all the benefits on medicines.
You have taken advantage of my nobility, an idea of mine without counting on me, trying to make a miserable business, are you not ashamed?
Calm down, everyone gets their Karma ...
I don't think what you're doing is correct.
we are doing tests on tests to get a result and from all this you want to get an economic benefit, without even having tried anything. I'm sorry but I absolutely don't agree. The copper plate in the photo attached cost me € 10 including shipping.
Attachments
I'm at a loss here @srkirt, please explain what money I'm making here?
I can understand the cost of processing the machine, I can understand the effort and the lost time, but I do not find it justifiable to affix such a high cost for each plate, if I told you that it cost me € 10 an entire plate how can you say that did it cost you 52L (only the material) to you?
And I add: in Italy nobody would work 10 plates for you, as the labor cost is very high and turning on a cutting machine for so little work is counterproductive ... this may mean that you have created the processing maybe where you work .. here this is not justifiable
Shims have been used in CPU cooling for years.
Loog outlined the costs in details in post #1168. They don't sound unreasonable to me. And £10 is really not much if it can make your £1299 5%-20% faster. To me it's quite obvious that Loog isn't doing this in order to get rich.
Don't wanna turn the thread into an emotional dispute, but from my point of view all contributions are GREAT.
Both @srkirt and @Loog (and a lot of other users here) have been working big to improve our machines.
No one is branding the improvement as their own, please don't take it personally... I mean, we are talking about the disruptive discovery of copper shims thermal properties (WTF?!).
If someone makes a business out of it I don't think posting it on a public forum is the way to go... Or at least I'd not complain if someone tries to go for it lol. Welcome to opensource development.
That said, if you think it's an opportunity, make a business. It's a free world. At the end you'll make money if people perceives value in your product. Otherwise you'll go broke.
If you think it's not an opportunity to make a business, but to help the community, you can try to manufacture it at zero profit (this is what @Loog is doing btw. If you don't trust him you can get material supply, cut your own sheets, setup all delivery, handle all the logistics of the process and report the costs).
And if you don't like other people working on your ideas don't make them public, it's that easy.
Thinking about the community it seems a big improvement from Loog to spend his time and money to get a perfect cut shim to fit our machine. Some users might be doubting about implementing the mod or not. This will hopefully motivate them to do it with a more 'out of the box' solution that you don't have to cut on your own.
I'm expecting more people to test it, share their process and standardize it somehow, get to the perfect thickness between 0.5-0.3. Get a bigger sample of users posting their improvements in performance/temperatures.
Thanks everybody for your time and energy!
Both @srkirt and @Loog (and a lot of other users here) have been working big to improve our machines.
No one is branding the improvement as their own, please don't take it personally... I mean, we are talking about the disruptive discovery of copper shims thermal properties (WTF?!).
If someone makes a business out of it I don't think posting it on a public forum is the way to go... Or at least I'd not complain if someone tries to go for it lol. Welcome to opensource development.
That said, if you think it's an opportunity, make a business. It's a free world. At the end you'll make money if people perceives value in your product. Otherwise you'll go broke.
If you think it's not an opportunity to make a business, but to help the community, you can try to manufacture it at zero profit (this is what @Loog is doing btw. If you don't trust him you can get material supply, cut your own sheets, setup all delivery, handle all the logistics of the process and report the costs).
And if you don't like other people working on your ideas don't make them public, it's that easy.
Thinking about the community it seems a big improvement from Loog to spend his time and money to get a perfect cut shim to fit our machine. Some users might be doubting about implementing the mod or not. This will hopefully motivate them to do it with a more 'out of the box' solution that you don't have to cut on your own.
I'm expecting more people to test it, share their process and standardize it somehow, get to the perfect thickness between 0.5-0.3. Get a bigger sample of users posting their improvements in performance/temperatures.
Thanks everybody for your time and energy!
It is great for all the contributors in this thread to share the results.
I don't think anyone is going to make a business out of this. You can order a 10 pack of 20*20*.5 from Amazon for around $10 with shipping USA. You can cut one to fill the space recommended of 22*16.5*.5, @20*16.5*.5
Thank you to all those who took photos and conducted the test.
![Screen Shot 2020-05-11 at 10.48.57 AM.png]( "Screen Shot 2020-05-11 at 10.48.57 AM.png")
I don't think anyone is going to make a business out of this. You can order a 10 pack of 20*20*.5 from Amazon for around $10 with shipping USA. You can cut one to fill the space recommended of 22*16.5*.5, @20*16.5*.5
Thank you to all those who took photos and conducted the test.
Last edited: