Just checked the spreadsheet again, and sadly someone has gone and "vandalised" it :-(
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
this is why we can't have nice things, I removed the spreadsheet for now.
Thanks for the heads up. Luckily, Google and M365 both allow for restoring backups. I got the data back and decided to edit the spreadsheet.
Though I still feel the need to apologize for the content.
it should be the same link, but I generated it again, just in case.
2020 Macbook Air Benchmarking
it should be the same link, but I generated it again, just in case.
2020 Macbook Air Benchmarking
docs.google.com
Thank you for restoring it. But - I tried finding the last values (by @silvia_18) and couldn't find them in this thread (or anywhere else). Could you point to their origin please?
I am on vacation and had time to ponder. I been running the Pad and New paste and shim for over a month. I did the pad first then added the shim a week or so later. Never tested with out the pad.
.3mm 16x20 shim, MX4 paste, 1.5mm Arctic pad.
I took the pad off today and happily it came off with no marking what so ever. It was a clean peal off and I had removed the plastic film prior to insulation.
My GB5 Paste Shim only is very close to what it was with the pad and shim, I would have scores over 4100 but many just over 4,000, I scored 4039 and 4024 paste shim only. You could say from the highs and lows it seemed equal. The bottom case felt cooler than with the pad attached.
The Cinebench R20 also scored about the same as when I had the Pad with Shim, I got a high of 1226 pad and shim and shim only just ran a 1213 1x. But the temps running R20 did hit 100 with out the pad and with the pad it would max mid 80's.
So the tests show the Pad does dissipate the heat better then no Pad, although making the bottom case a little warmer. But output from the device hardly changes from New Paste and Shim vs Paste Shim Pad. It seems to me that Apple has it restricted to not be able to benefit a greater than just the paste alone or just the pad alone. A Apple Governor
My conclusion is if you do the Paste Shim no need to add the pad. If you are not comfortable putting new paste on then Just do the 1.5mm Pad.
I recommend the 1.5mm Arctic pad, not that its the best thermal conductive (not sure that it matters a lot for this mod) but it clearly is a peel off pad leaving no residue or any signs of modification.
.3mm 16x20 shim, MX4 paste, 1.5mm Arctic pad.
I took the pad off today and happily it came off with no marking what so ever. It was a clean peal off and I had removed the plastic film prior to insulation.
My GB5 Paste Shim only is very close to what it was with the pad and shim, I would have scores over 4100 but many just over 4,000, I scored 4039 and 4024 paste shim only. You could say from the highs and lows it seemed equal. The bottom case felt cooler than with the pad attached.
The Cinebench R20 also scored about the same as when I had the Pad with Shim, I got a high of 1226 pad and shim and shim only just ran a 1213 1x. But the temps running R20 did hit 100 with out the pad and with the pad it would max mid 80's.
So the tests show the Pad does dissipate the heat better then no Pad, although making the bottom case a little warmer. But output from the device hardly changes from New Paste and Shim vs Paste Shim Pad. It seems to me that Apple has it restricted to not be able to benefit a greater than just the paste alone or just the pad alone. A Apple Governor
My conclusion is if you do the Paste Shim no need to add the pad. If you are not comfortable putting new paste on then Just do the 1.5mm Pad.
I recommend the 1.5mm Arctic pad, not that its the best thermal conductive (not sure that it matters a lot for this mod) but it clearly is a peel off pad leaving no residue or any signs of modification.
I took the pad off today and happily it came off with no marking what so ever. It was a clean peal off and I had removed the plastic film prior to insulation.
My GB5 Paste Shim only is very close to what it was with the pad and shim, I would have scores over 4100 but many just over 4,000, I scored 4039 and 4024 paste shim only. You could say from the highs and lows it seemed equal. The bottom case felt cooler than with the pad attached.
The Cinebench R20 also scored about the same as when I had the Pad with Shim, I got a high of 1226 pad and shim and shim only just ran a 1213 1x. But the temps running R20 did hit 100 with out the pad and with the pad it would max mid 80's.
So the tests show the Pad does dissipate the heat better then no Pad, although making the bottom case a little warmer. But output from the device hardly changes from New Paste and Shim vs Paste Shim Pad. It seems to me that Apple has it restricted to not be able to benefit a greater than just the paste alone or just the pad alone. A Apple Governor
My conclusion is if you do the Paste Shim no need to add the pad. If you are not comfortable putting new paste on then Just do the 1.5mm Pad.
I recommend the 1.5mm Arctic pad, not that its the best thermal conductive (not sure that it matters a lot for this mod) but it clearly is a peel off pad leaving no residue or any signs of modification.
IngerMan, first off thanks for all the posts you put up as it inspired me to go down this path of modifying my MBA.
For the most part, I agree with your view and for 99% of people, the shim and paste is the best modification as there is a real upside benefit (+15%) with no downsides. I believe Apple designed it to perform the same as the shim and paste mod but somehow got a manufacturing issue.
However, the performance is not the same with the paste+shim vs. paste+shim+pad. In a single test, the results as you noted are the same with ~4000 GB MC scores, however in back to back tests, the system will heat soak without the pad and the scores will come down to ~3500. With the pad, the scores come down as well but not as severe, to ~3950.
The tradeoff to that extra heat capacity is a hot bottom cover which can range between 46C and 50C depending on the pad used and size. You can read the results of extensive testing at this post. I've had a few people email me with their results which match what my computer produced and I would appreciate if you can test it to confirm as well.
For Cinebench, I agree there is some restriction in the software/hardware that limits it to 1.9GHz/1.8GHz in a perfect square wave. Interestingly, with reduced heat transfer, if you can keep the computer operating just below 1.9GHz, it doesn't cycle down to 1.8GHz and can actually produce a higher Cinebench R20 score, even though the CPU is running hotter.
I think what we're finding is that it's a very complex balancing act to design a computer! We are trying to balance performance, battery life, and heat and realizing that some tradeoffs are not worth it (excess heat baking our thighs). It gives us some insight into the challenges engineers face and why the easy fixes are not really so easy after all.
Last edited:
The document was restored to just before it was "vandalized". So all data previously entered should be there.
Sorry, I guess I wasn't clear. I can of course see everything in the restored spreadsheet. But when I try to find the original post by silvia_18 with the data seen in the spreadsheet, I can't find it, so I'm a bit confused.
EDIT: I guess there was a misunderstanding. I thought the spreadsheet is a summary of data posted in this thread. It seems though that the data was collected from direct Geekbench uploads that I don't know how to access. Please correct me again, if this is still wrong.
Last edited:
Hi! Can I help? Not sure I understand, but I can see my scores are back on the sheet?
Stripped a screw on the heatsink for the i5 2020 -- does anyone what options I have for replacement?
Very frustrating because I was so, so careful. Every other screw is fine, the screw bits are fine as well.
Very frustrating because I was so, so careful. Every other screw is fine, the screw bits are fine as well.
What do you think Arm MBA will bring. I just took delivery of my 2020 iPad 11” Pro. It scores 1118 SC 4687 MC GB5. No new paste, No thermal pads or No fans ?
I think the change will be positive and I will embrace it.
I think the change will be positive and I will embrace it.
Finally, give my feedback on testings.
Below are results:
- Using a single custom 56x36 heat sink to replace original one almost reached the power limitation. But the issue is the bottom case very hot.
- The 22 x 22 copper heat sink also much improvement, but still can't reach to the limitation.
- So I decide still go with Shim plus pipe. It's more suitable. I am not sure what happen, after I done final adjustment, the performance looks better compare to above Shim+Pipe record.
Here are final mod and review, you can see the GB performance very stable with 5 times continue testing. The CPU temperature at 83~85.
One more strange thing is my I5+8+256 Air the max power limited to 12.5W(12~13W), and CB20 testing CPU running at 1.7G(50%) and 1.8G (50%). Compare another I5+8+512, I have tested that's 13.5W and 1.8G(90%) and 1.9G (10%).
Some other photos:
============26/7/2020 Update===========
Small adjustment.Using heat-resistant tape to cover the heat pipe, move the pipe a bit to cover fan to bring more heat outside. Now the bottom case temperature really same as original. Daily operations just warm and bottom case acceptable to put on leg during heavy workload. I am thinking add second short pipe under the 18cm one, maybe 12~14cm. It should be helpful for heat dissipation through more area of bottom case and of course should lower the bottom case temperature.
Below are results:
- Using a single custom 56x36 heat sink to replace original one almost reached the power limitation. But the issue is the bottom case very hot.
- The 22 x 22 copper heat sink also much improvement, but still can't reach to the limitation.
- So I decide still go with Shim plus pipe. It's more suitable. I am not sure what happen, after I done final adjustment, the performance looks better compare to above Shim+Pipe record.
Here are final mod and review, you can see the GB performance very stable with 5 times continue testing. The CPU temperature at 83~85.
One more strange thing is my I5+8+256 Air the max power limited to 12.5W(12~13W), and CB20 testing CPU running at 1.7G(50%) and 1.8G (50%). Compare another I5+8+512, I have tested that's 13.5W and 1.8G(90%) and 1.9G (10%).
Some other photos:
============26/7/2020 Update===========
Small adjustment.Using heat-resistant tape to cover the heat pipe, move the pipe a bit to cover fan to bring more heat outside. Now the bottom case temperature really same as original. Daily operations just warm and bottom case acceptable to put on leg during heavy workload. I am thinking add second short pipe under the 18cm one, maybe 12~14cm. It should be helpful for heat dissipation through more area of bottom case and of course should lower the bottom case temperature.
Last edited:
Tried the Arctic pad mod and NOT the shim mod on a 2020 i5 16GB, because it doesn't look like anyone has tried that mod configuration.
Received the 2020 i5 in early July of this year.
Got the 1.5mm Arctic pad from Amazon, cut it to 35mm x 40mm, applied it to the heat sink, and gave it the IngerMan nipple rub. No other mods were made.
Here are my before and after GB5 benchmarks, deliniated by running on the laptop screen only, an while driving an external 4k monitor.
Reading row across shows you the penalty for the external 4k monitor, reading down shows you the gains from the Arctic pad:
I did order the shims and paste from Amazon, may go back in there and mod that later.
Received the 2020 i5 in early July of this year.
Got the 1.5mm Arctic pad from Amazon, cut it to 35mm x 40mm, applied it to the heat sink, and gave it the IngerMan nipple rub. No other mods were made.
Here are my before and after GB5 benchmarks, deliniated by running on the laptop screen only, an while driving an external 4k monitor.
Reading row across shows you the penalty for the external 4k monitor, reading down shows you the gains from the Arctic pad:
I did order the shims and paste from Amazon, may go back in there and mod that later.
MacBook Air 2020 (i5, 16GB ram) - cooper shim pad 0.3mm + arctic MX-2 + 150 x 12.1 x 1mm heat pipe sitting on double-sided thermal tape (0,3mm, 1.4 W/mK). Idle temperature 40-45°C, under heavy load 75-85°C with bottom case only slightly warm/fully comfortable on lap. Fan is completely silent under load. Happy with this cheap and easy mod and everyday use for programming/code compiling.
Nice and clean install! If you conduct back to back GB5 tests with no waiting in between, what are the MC scores And CPU temps?
How durable is the double sided tape? I worry about it coming loose in a portable environment where the laptop might get knocked around on flights.
If I might say what I'm thinking, I don't think that heatpipe is doing much at all in terms of what a heatpipe should do, in order to act as a heatpipe it needs to be connected to some kind of metal on the other side, something with fins.. But it adds mass to the heatsink and it helps a lot, cause it's a lot of mass (compared with the heatsink). The MBA desperately needs a third party heatsink with more mass and better design.
Yes, agree that a real heat-pipe needs fins on fan side for full function. But that needs total redesign whole cooling system, include fan, heat sink, heat-pipe.
To add a heat-pine can do two things to cooling the MBA.
- through the long heat-pipe can transfer the heat to more area of bottom case
- bring few more heat to the fan and outside. you know put heat-pine under a fan, hot air will more easy up and enter the fan and blow out.
Actually, you can see people just simple add a thermal pad can improve the MBA max performance
or
You can refer to my two testings of direct put a copper heat sinks. Small one (22x22x3) can push MBA to 95% performance and custom big one (56x46x3) can push MBA to it's limitation with 100% performance. The problem is all heat inside the machine, and it made the small area of bottom case which close to the heat sink very hot. just like thermal pad.
This is why blow out more heat through fan and transfer heat through more big area of bottom are better solution.
Last edited:
More things that I tried on my machine has not been done by anyone from this forum ... in the end he died of so much tinkering ...
BTW I have found a better paste than the NX4, Thermalright TF8 Thermal Composite Paste 13.8 W / mK.
In my Hauwei I have reached almost 1500 points in R20 with an AMD Ryzen, (by the way much colder and faster than an i5)
BTW I have found a better paste than the NX4, Thermalright TF8 Thermal Composite Paste 13.8 W / mK.
In my Hauwei I have reached almost 1500 points in R20 with an AMD Ryzen, (by the way much colder and faster than an i5)
Attachments
I have been looking around for them, I think the best bet will eventually be iFixit, but last I checked they didn't have them.
To be honest, your execution was messy and I was counting days, when you kill it. Ideas were excelent point, which crossed many minds, but when I look at those pictures, I wouldnt ever take them as a benchmark, its your machine and there are alredy clean executions.
Transfer of heat with just coper foil is ineadequate.