A list of successful iMac 27" (2012-2019) SSD upgrades

[mbosse]

A few things here I'd like to note:

1. Heatsink:

With so little space, I want as much air movement as possible. Reduced air space limits efficient cooling. I understand airflow is a more important factor for heat dispersal than total heatsink surface area (fins). It is airflow which is essential for removing radiated heat. A greater volume and faster movement of air = greater heat dispersal and a cooler running SSD or chip, fins or no fins. That is my opinion. The flat copper helps absorb heat from the controller and diffuse it more evenly throughout the heatsink. Fan then blows excess radiant heat away from the copper heatsink thus preventing heat build up in the NVMe operation. Generally, when NVMe temperature rises to 75° or above (depending on each models design tolerance), NVMe will throttle back to avoid damage.

Excellent reasoning on using copper instead of the usual aluminium heat sinks, as copper has roughly twice the conductivity coefficient than (most) aluminium alloys. What would interest me would be the actual reduction in temperature with/without copper heatsink with no interference by Macs Fan Control etc., because I am not entirely convinced about your statement that airflow is more important than heatsink surface: This is because the airflow reduces temperature because it flows over the surfaces to be cooled, the higher the touched surface the higher hence the heat removal. And here the fins serve to enlarge the available surface area such that the net effect between a flat copper heatsink vs. the commercial aluminium heatsink with fins could be actually identical.

Really curious!

2. Fan control software:
There is absolutely no need to use such as macOS perfectly reads the temperature from the NVMe SSD. Fan control software was necessary back in the 2009 - 2011 days of iMacs when Apple used (i) special versions of HDDs with a separate connector for a temperature data cable (2009/2010) or (ii) special firmware on their HDDs. Vendors like OWC offered (expensive) special cables with temperature sensor. Fun fact: for the 2009/2011 iMacs one could easily use a €2 Apple DVD Drive temperature cable to reinstate perfect temperature reading from any HDD/SSD, just the 2011 iMac absolutely needed the OWC cable if a third-party drive would be use. No such drive cable or fan control software is needed for the slim 2012 or later iMacs for any of the SSDs confirmed to be compatible.

If you want to achieve lower than the macOS-set temperature range at the cost of some additional noise then of course you can resort to fan control software.

3. Migration from TM backup of FusionDrive:

Also, did anyone make a Fusion drive clone before the drive replacement? If yes, please let me know how it works, since I do not want to make a clean installation on a new drive because I would like to install Sonoma from scratch via OCLP some time after the upgrade, restore the backup via Time Machine (if it would be possible, of course) and upgrade OS to Sequoia (since the restoration does not work on Sequoia properly according to OCLP website). And I do not want to make some extra steps during or right after the upgrade, so I can have some time to test the updated hardware.

In terms of migration and TM there is no difference between a FusionDrive or a normal HDD/SSD. You can migrate from or to a FusionDrive just like a normal drive provided the available drive space allows and the target macOS is the same version or younger. For purposes of macOS backup and migration, a FusionDrive is seen just like a standard drive.

BTW, @Terraaustralis, Howard Oakley does no longer recommend full macOS install and then migration later - he does it now directly at setup: https://eclecticlight.co/2024/11/12/migrating-to-a-new-mac-and-claiming-time-machine-backups/

Cheers and good luck to all!

 

[Terraaustralis]

Diagnostic light won’t tell him what the issue is regarding audio. Likely issue is due to the headphone port not being recognised and the cable being damaged during disassembly

Understood mbosse.
Read my remarks of Monday 11.15 and you will appreciate the context of my comments.
Cheers,

 

[mbosse]

Understood mbosse.
Read my remarks of Monday 11.15 and you will appreciate the context of my comments.
Cheers,

Are you mixing up replies? I did not comment on the diagnostic lights... ;-)

 

[Terraaustralis]

Diagnostic light won’t tell him what the issue is regarding audio. Likely issue is due to the headphone port not being recognised and the cable being damaged during disassembly

That is understood TwoH.
See my original post of Monday at 11.55 to understand context.
Cheers

 

[TwoH]

That is understood TwoH.
See my original post of Monday at 11.55 to understand context.
Cheers

Always appreciate extra ideas from people and I apologise I missed your initial post regarding it.
I hope we can get an update from OP regarding his issue to know if there has been any change and/or further diagnosing of the issue.

I do think we can interpret from their post of the machine's sound not working that they are able to POST and boot fine into their machine, and we are purely looking at a sound-related issue, to which I think we are both correct in looking towards the headphone port being the issue - either due to not being seated correctly or due to damage on the actual cable itself.

But we are eagerly awaiting any further information from OP to know for certain!

 

[Terraaustralis]

Are you mixing up replies? I did not comment on the diagnostic lights... ;-)

I am indeed! just saw the mbosse ‘like’ while scanning. My apologies.

A few things here I'd like to note:

1. Heatsink:



Excellent reasoning on using copper instead of the usual aluminium heat sinks, as copper has roughly twice the conductivity coefficient than (most) aluminium alloys. What would interest me would be the actual reduction in temperature with/without copper heatsink with no interference by Macs Fan Control etc., because I am not entirely convinced about your statement that airflow is more important than heatsink surface: This is because the airflow reduces temperature because it flows over the surfaces to be cooled, the higher the touched surface the higher hence the heat removal. And here the fins serve to enlarge the available surface area such that the net effect between a flat copper heatsink vs. the commercial aluminium heatsink with fins could be actually identical.

Really curious!

2. Fan control software:
There is absolutely no need to use such as macOS perfectly reads the temperature from the NVMe SSD. Fan control software was necessary back in the 2009 - 2011 days of iMacs when Apple used (i) special versions of HDDs with a separate connector for a temperature data cable (2009/2010) or (ii) special firmware on their HDDs. Vendors like OWC offered (expensive) special cables with temperature sensor. Fun fact: for the 2009/2011 iMacs one could easily use a €2 Apple DVD Drive temperature cable to reinstate perfect temperature reading from any HDD/SSD, just the 2011 iMac absolutely needed the OWC cable if a third-party drive would be use. No such drive cable or fan control software is needed for the slim 2012 or later iMacs for any of the SSDs confirmed to be compatible.

If you want to achieve lower than the macOS-set temperature range at the cost of some additional noise then of course you can resort to fan control software.

3. Migration from TM backup of FusionDrive:



In terms of migration and TM there is no difference between a FusionDrive or a normal HDD/SSD. You can migrate from or to a FusionDrive just like a normal drive provided the available drive space allows and the target macOS is the same version or younger. For purposes of macOS backup and migration, a FusionDrive is seen just like a standard drive.

BTW, @Terraaustralis, Howard Oakley does no longer recommend full macOS install and then migration later - he does it now directly at setup: https://eclecticlight.co/2024/11/12/migrating-to-a-new-mac-and-claiming-time-machine-backups/

Cheers and good luck to all!

Hi mbosse,

Heatsink thoughts:
I entirely agree with your considerations regarding cooling radiant solids such as heatsink. Only a test can tell us the best balance between air flow and radiant surface area. Not an easy test to configure.

As any mechanic knows with a motor vehicle radiator, increased surface area increases exposure to wind pressure and disperses heat to cool water - and therebye, engine temperature. However for the iMac, the passage of air in the restricted interior between iMac back and NVMe heatsink suggests to me that greater heatsink surface (fins), simultaneously obstruct airflow which is not adequately compensated for with increased air flow speed due to turbulence. Therefore, I suspect heat dispersion would be less efficient.

Alternatively It can be argued in the reverse: That if given air pressure from a fan source remains constant, the physical restriction of fins will create a faster air flow for the same volume of air over the heatsink therebye removing heat more efficiently. The determinant is how airflow actually behaves in the circumstances of the iMac.

Perhaps I favour free airflow in volume because I enjoy sailing and understand something about lift and the passage of a hull through water. That is in no way parallel to our heatsink except regarding turbulance. Whether air or water, turbulance reduces efficiency. How significant that is with our heatsink is an open question. I welcome comments.

In practice, the saving grace is the SN770 being a PCIe 4.0 NVMe working at PCIe 3.0 speeds. So heat cannot build up to a critical level in the Intel iMac.

Fan Control Software:
Thank you for the information. I appreciate your comments. I have taken the subject of fan temperatures at face value while asking myself how efficient are these fan readings actualy are. Being software, I assume they are derived from the iMac OSX sensors and data.

Would you care to expand upon ‘need’. Does that mean that for any post 2011 iMac I can trash Macs Fan Control and rely entirely on Apple? When I use a more powerful NVMe than the small Apple supplied SSD’s I would expect to be reaching or exceeding Apples’ built in temperature tolerances. If this is so it would seem reasonable to increase cooling via fan control software?

Howard Oakley:
I shall have to send Howard a stiff reprimand for changing tack! - Joke.

Thanks for your time mbosse. always something to learn from you.

 

[Terraaustralis]

Always appreciate extra ideas from people and I apologise I missed your initial post regarding it.
I hope we can get an update from OP regarding his issue to know if there has been any change and/or further diagnosing of the issue.

I do think we can interpret from their post of the machine's sound not working that they are able to POST and boot fine into their machine, and we are purely looking at a sound-related issue, to which I think we are both correct in looking towards the headphone port being the issue - either due to not being seated correctly or due to damage on the actual cable itself.

But we are eagerly awaiting any further information from OP to know for certain!

Yes. From what I could make out he was upgrading for a client which leads to a very difficult situation. I hope a replacement audio cable is the solution. I was concerned that his logic board may be damaged, hence my remarks.

 

[Terraaustralis]

A few things here I'd like to note:

1. Heatsink:



Excellent reasoning on using copper instead of the usual aluminium heat sinks, as copper has roughly twice the conductivity coefficient than (most) aluminium alloys. What would interest me would be the actual reduction in temperature with/without copper heatsink with no interference by Macs Fan Control etc., because I am not entirely convinced about your statement that airflow is more important than heatsink surface: This is because the airflow reduces temperature because it flows over the surfaces to be cooled, the higher the touched surface the higher hence the heat removal. And here the fins serve to enlarge the available surface area such that the net effect between a flat copper heatsink vs. the commercial aluminium heatsink with fins could be actually identical.

Really curious!

2. Fan control software:
There is absolutely no need to use such as macOS perfectly reads the temperature from the NVMe SSD. Fan control software was necessary back in the 2009 - 2011 days of iMacs when Apple used (i) special versions of HDDs with a separate connector for a temperature data cable (2009/2010) or (ii) special firmware on their HDDs. Vendors like OWC offered (expensive) special cables with temperature sensor. Fun fact: for the 2009/2011 iMacs one could easily use a €2 Apple DVD Drive temperature cable to reinstate perfect temperature reading from any HDD/SSD, just the 2011 iMac absolutely needed the OWC cable if a third-party drive would be use. No such drive cable or fan control software is needed for the slim 2012 or later iMacs for any of the SSDs confirmed to be compatible.

If you want to achieve lower than the macOS-set temperature range at the cost of some additional noise then of course you can resort to fan control software.

3. Migration from TM backup of FusionDrive:



In terms of migration and TM there is no difference between a FusionDrive or a normal HDD/SSD. You can migrate from or to a FusionDrive just like a normal drive provided the available drive space allows and the target macOS is the same version or younger. For purposes of macOS backup and migration, a FusionDrive is seen just like a standard drive.

BTW, @Terraaustralis, Howard Oakley does no longer recommend full macOS install and then migration later - he does it now directly at setup: https://eclecticlight.co/2024/11/12/migrating-to-a-new-mac-and-claiming-time-machine-backups/

Cheers and good luck to all!

Further considerations regarding radiation and heatsink.

Quote:
‘This is because the airflow reduces temperature because it flows over the surfaces to be cooled, the higher the touched surface the higher hence the heat removal.’

Pondered on your question a little longer mbosse. Your statement above is both true and untrue: Key is your use of the word flow. In my motor car illustration, the radiator cools hot water by the passage of air. If the radiator remains in still air it eventually fails at some point when it cannot dissipate heat as fast as an idling engine creates it. There is probably a physics equation which defines this behaviour.

When the car is moving, airflow through the radiator improves and water temperature is managed within safe levels.

With our hot NVME, Fins will disperse heat only at the specific rate that a given air temperature can absorb excess heat - no matter how large the area. This rate varies according to circumstance, but like the car, unless airflow is active to remove hot air and introduce cool air the NVME will overheat.

Measuring vectored radiation is very tricky and quite beyond me. All I can see is that defining rate of airflow and volume of air is key to establishing a safe rate of heat dissipation for the NVME, with or without fins.

I agree that low level fins (1.5 mm) or a corrugated surface would improve the efficiency of heat dispersal on my copper heatsink. However for the SN770 it is not necessary. My central point here is that air movement is the critical cooling factor as I understand the physics of the situation. Especially so in a confined iMac space such as we are discussing.

I would appreciate an electrical design engineer to weigh in on this topic as I am sure there are formulaes which better answer both our points of view. The answers would be enlightening.

 

[JustBlogg]

Тhank you for a quick reply.

Heatsink:
I chose copper for two reasons;
One, it is far better at conducting and dissipating heat than any metal other than Silver with a conductivity of 400 watts per Kelvin per metre. Graphite compositions can be good with Applied Nanotech announcing their CarbAl heatsinks supplying conductivity at 390 W/mK, good if it performs as claimed and it is not too costly.
In metals, Aluminium, with with 235W/mK. ranks fourth down the list of conductivity after Gold.

Two: With only about 3-4mm of space for a heatsink in an iMac, the heatsink must be thin to allow effective circulation of air from the fans. See: https://computer.howstuffworks.com/heat-sink.htm
Like you, I have a WD Black NVMe SN770 which is a cool running SSD and I wish to keep it that way. Macs Fan Control offers sensor readings for SSD, CPU and GPU cores so you can reliable adjust airflow to reduce operating temperatures.

With so little space, I want as much air movement as possible. Reduced air space limits efficient cooling. My understanding of heatsink function regards airflow as a more important factor for heat dispersal than total surface area (fins). It is airflow which is essential to removing radiated heat. A greater volume and faster movement of air = greater heat dispersal and a cooler running SSD or chip, fins or no fins.

Got it. Thanks.
And as I see, you used a single-sided heat sink instead of a double-sided one. Could you please clarify why? Because the SSD is single-sided itself?
And according to the shared photos in the topic, most users applied the full-sized thermal pads to the SSD surface without cutting, not only to NAND chips, right? As I know, the WD SN770 has different heights of chips on the SSD, therefore, did you use a double-sized thermal pad for some chip, for example?

As for the Fusion, you are absolutely right, it seems I did not make my question very clear. I will remove the HDD in any case, however, since I prefer to clone the data to the SSD before the upgrade, I do not know if cloning will be made correctly or if I should split the "Fusion" first.

As for apps, they include Adobe PS and Illustrator, MS Teams, MS Remote Desktop, Samsung Dex, Zoom, Chrome, Slack, VPN, etc. It's my wife's iMac, and it freezes up during work in Chrome sometimes when many tabs are opened. I have increased the RAM from 8 GB to 32 GB, and it has almost solved the issue, however, the usual read speed of about 450 MB/s sometimes drops to 250 MB/s, and it causes some "freezes". Read speed is about 1200-1300 MB/s. Therefore, I have decided to maximize the performance with a CPU and drive upgrade.

As for Mist, I am not familiar with this software. Is this some analog of OCLP? And it is not quite clear how I can install an unsupported Sequoia OSX on a 2015 iMac using this app. In a regular way, without any dancing?

As for the clean installation, got it. I will backup the data using Time Machine on an external drive and restore it after the clean Sequoia OSX installation.

Thanks.

There is absolutely no need to use such as macOS perfectly reads the temperature from the NVMe SSD.

Got it, thanks.

As for the heatsink, it seems that I need to make a test. It would be really interesting to find the right answer.

 

[Terraaustralis]

Тhank you for a quick reply.


Got it. Thanks.
And as I see, you used a single-sided heat sink instead of a double-sided one. Could you please clarify why? Because the SSD is single-sided itself?
And according to the shared photos in the topic, most users applied the full-sized thermal pads to the SSD surface without cutting, not only to NAND chips, right? As I know, the WD SN770 has different heights of chips on the SSD, therefore, did you use a double-sized thermal pad for some chip, for example?

As for the Fusion, you are absolutely right, it seems I did not make my question very clear. I will remove the HDD in any case, however, since I prefer to clone the data to the SSD before the upgrade, I do not know if cloning will be made correctly or if I should split the "Fusion" first.

As for apps, they include Adobe PS and Illustrator, MS Teams, MS Remote Desktop, Samsung Dex, Zoom, Chrome, Slack, VPN, etc. It's my wife's iMac, and it freezes up during work in Chrome sometimes when many tabs are opened. I have increased the RAM from 8 GB to 32 GB, and it has almost solved the issue, however, the usual read speed of about 450 MB/s sometimes drops to 250 MB/s, and it causes some "freezes". Read speed is about 1200-1300 MB/s. Therefore, I have decided to maximize the performance with a CPU and drive upgrade.

As for Mist, I am not familiar with this software. Is this some analog of OCLP? And it is not quite clear how I can install an unsupported Sequoia OSX on a 2015 iMac using this app. In a regular way, without any dancing?

As for the clean installation, got it. I will backup the data using Time Machine on an external drive and restore it after the clean Sequoia OSX installation.

Thanks.



Got it, thanks.

As for the heatsink, it seems that I need to make a test. It would be really interesting to find the right answer.

1. Mist.
Using my link read about and download Mist. Use it to choose an OSX. Mist will download the OSX direct to any preferred drive.

2. Heatsink.
You do not need to test. Mbosse and I where merely discussing principals of comparative efficiency. My heatsink is simply polished copper adhered to NVME with single adhesive thermal pad. Mbosse uses a finned heatsink. Max thickness including thermal pad 2.0mm - 2.5mm. 2280mm Pad will accommodate various height differences of controller and NAND chip. Finned heatsinks are available from Ali Express. Take your pick. The object of covering the full length of the NVMe is to offer greater surface for heat absorption and more even dissipation of heat prior to removal via airflow.

The SN770 is a cool running NVME working at half its capacity on PCIe 3.0. No need for anxiety in choice of heatsink. It is not so much about ‘right’ answers. It is about using judgement. Copper is simply more efficient than Aluminium but both will work.

3. Fusion
When you erase your drive you erase Fusion. After that just forget Fusion and load the OSX as I describe above.

4. Cloning
Cloning is not recommended for OSX on APFS formatting. Again, see link I provided. Use Migration Assistant to import Account files from TM.

5. Tabs
Having a number of open browser tabs consumes your computers resources very quickly. Delete unused tabs. Open Activity Monitor in Utilities and watch the numbers change as you open tabs without closing unused tabs.

 

[Terraaustralis]

Hello my friend

All connections are ok, I already checked and nothing.
they are simply not recognized
I don't know if reinstalling OS is an option.

Go to this site: https://archive.org/details/appleservicemanuals/imac-27-2017/page/n71/mode/2up
where you will find Apple technical repair manuals with diagrams and procedures. This will help identify parts and shows block diagrams for circuitry.
I hope this helps solve your problems.
Cheers,

 

[sebasq26]

Hello Hello Hello Hello Hello Hello

Gentlemen, news, solved

I would like to mention each of you for your help, after almost crying hahaha, worrying, I went to a colleague technician in microsoldering, from now on a SemiDios, and found the fault quickly, a microchip or small piece, which had to be checked under a microscope came off, I managed to get a board for spare parts and the change was made and we have sound, my soul returns to the body. I will proceed with great care to reassemble everything and then I will share with you.

PS: I leave a picture of my headache.

 

[Terraaustralis]

Hello Hello Hello Hello Hello Hello

Gentlemen, news, solved

I would like to mention each of you for your help, after almost crying hahaha, worrying, I went to a colleague technician in microsoldering, from now on a SemiDios, and found the fault quickly, a microchip or small piece, which had to be checked under a microscope came off, I managed to get a board for spare parts and the change was made and we have sound, my soul returns to the body. I will proceed with great care to reassemble everything and then I will share with you.

PS: I leave a picture of my headache.

I am very happy for you. I know the feeling well. I hope the cost was reasonable. When a computer I was upgrading showed a dead monitor, I suspected the cable linking monitor to Logic board. I swapped cable and no change. Diagnostic lights showed 3 green. Then I suspected the cable fitting on the logic board. Eventually I found a micro soldering technician and asked him to test the board. He did. It was not the cable connection. He found the problem and fixed it for AUD$300. I was very happy to have the board working again. A recycled logic board would cost around AUD$900 and Apple wanted AUD$1200. Like you, I had no idea how the fault occurred.

The lesson from my experience, is to never underestimate how sensitive components are to damage. Apple Tech Service manuals warn against touching soldered parts. That is why I wear surgical gloves. Far safer than wrist straps. Better for me and the logic board. Every caution is justified.

I understand your feeling of relief.
Congratulations.

 

[mbosse]

Hello Hello Hello Hello Hello Hello

Gentlemen, news, solved

I would like to mention each of you for your help, after almost crying hahaha, worrying, I went to a colleague technician in microsoldering, from now on a SemiDios, and found the fault quickly, a microchip or small piece, which had to be checked under a microscope came off, I managed to get a board for spare parts and the change was made and we have sound, my soul returns to the body. I will proceed with great care to reassemble everything and then I will share with you.

PS: I leave a picture of my headache.

Wow - finding (and being able it repair it) is something - congrats!

 

[Ruadh]

If you feel competent to handle the above you should find the procedure straight forward. First time around, I had the OWC video on the monitor of a second computer while I worked on the first. I prefer to work with the iMac on its back rather than standing.

So, I'm back here after a month's delay. I finally did the install replacing my NVME PcIE drive in my 2019 iMac (19,1) last week. In short, it was not a success. Everything went smoothly, I thought, but when I turned the power on, nothing happened - black screen, no response to on/off button, dead.

So, I opened my temporarily fixed screen (masking taped) and did a check of the Diagnostic LEDs and found, oddly, that only No. 2 was alight. Which, checking against the Apple 27" 2019 iMac manual (thanks for the link here) is not a listed outcome. If LED 1 is off, no power should be available. If No. 2 is on, the computer is turned on.

So, I googled people that had this diagnostic combination ....

And I got the diagnosis of a CPU failure ....

So, I'm going to reopen the machine tomorrow and go through and see if that checks out.

 

[Terraaustralis]

So, I'm back here after a month's delay. I finally did the install replacing my NVME PcIE drive in my 2019 iMac (19,1) last week. In short, it was not a success. Everything went smoothly, I thought, but when I turned the power on, nothing happened - black screen, no response to on/off button, dead.

So, I opened my temporarily fixed screen (masking taped) and did a check of the Diagnostic LEDs and found, oddly, that only No. 2 was alight. Which, checking against the Apple 27" 2019 iMac manual (thanks for the link here) is not a listed outcome. If LED 1 is off, no power should be available. If No. 2 is on, the computer is turned on.

So, I googled people that had this diagnostic combination ....

And I got the diagnosis of a CPU failure ....

So, I'm going to reopen the machine tomorrow and go through and see if that checks out.

Checking the thread on your project, you referred to replacing thermal paste on CPU. mwidjaya and myself recommended leaving it alone. Subsequently you posted reference to finding a video showing CPU heatsink assembly technique which implies you were set on thermal paste replacement.

Given popular articles about thermal paste life being limited according to intensity of use, replacement seems a logical initiative after five years or so. Yet many computers continue to function without overheating for well over a decade. My venerable 2007 iMac is one example. It has had little use over the last five years but remains perfectly operational. Gamers or intensive video/graphics users have reason to be more attentive.

When I removed CPU to replace thermal paste in my 2019 iMac two years ago (4yrs old) I found that the original paste appeared entirely sound and was not dried out. Once the heatsink is removed obviously new paste is required and I happened to use silicone based MX4 for what that may be worth.

The iFixit post was a valuable link. Good find. So you intend checking the CPU pins?

Regarding refitting CPU bracket/spring plate screwing technique; diagonal - corner to corner, in rotation, and equal number of screw turns at a time. You simply need to avoid over-tightening the screws. The purpose of the spring plate is to provide even compression whilst locking CPU in place.

From the iFixit/Michael Murphy experience, I assume the user over-tightened one corner more than another - unless the pin was bent through other circumstances. I look forward to your success. Please tell us which pin/pins are bent if any?

Good Luck.