We spend a lot of time criticizing Apple for their anti-consumer practices like monopolizing its repair ecosystem with parts pairing, engaging in psychological obsolescence, and lying to congress—among many examples of poor corporate behavior—but here’s one thing they’re finally doing that shows a lot of promise— the Self Service Repair program.
With the release of the nominal upgrade that is the M2 Pro SoC inside the 14” MacBook Pro, we dug a little deeper with repairability in mind and with one of the gems of the Self Service Repair program in hand: the service manual for the M1 MacBooks. We asked ourselves, “can we follow the old M1 repair manual for the new M2 Pro MacBooks?”
Spoiler alert: Yes. Yes we can.
Tried and True, or Old and Busted?
Being able to follow the prior generation’s manual is neither surprising, nor a bad thing. Okay, yes, “refreshing” the physical design of a product does artificially increase consumption—there’s a century of marketing research that agrees. But there are cost saving benefits to maintaining the same manufacturing processes, too, as most famously shown by Henry Ford with the Model T.
Aside from working out any kinks in a product, maintaining the same production methods reduces retooling overhead, and some of those savings get passed down to consumers. It also helps the would-be repairer by retaining a familiar design and expanding the pool of available parts. Not only is the OEM producing more of the same part, there’s more time and incentive for third-party manufacturers to make parts, too. Oh and minimal-to-no retooling of factories means reduced waste and turnaround time. That’s a lot of wins.
The M1 MacBook Pro’s and the M2 MacBook Pro’s share similar manufacturing processes. This means more profit for Apple, lower costs for consumers, and most importantly for us, the ability to use the same repair manual nearly 2 years down the line.
My general impressions of the repair manual were positive. While convoluted and intimidating at first due to the length of the document (162 pages!) and the sheer number of warnings, the internal logic to make sense as I learned the layout of the manual and understood the methodology. Apple’s repair manual does set you up for success—provided you have the patience and time to carefully follow the instructions contained therein.
Spot the Difference: M2 MacBook Pro Changes
As for the M2 notebook itself, there are of course some changes but those changes are restricted to the Logic Board. The most interesting of which is the apparently reduced size of the heatsink on the M2 Pro SoC. This prompted a bit of investigation since we would have expected the M2 Pro to have a similar if not larger heatsink compared to the M1.
With the SoC deshielded, the reason for the size disparity is immediately apparent. The M1 Pro has an 8GB Samsung LPDDR5 RAM module on either side of the core while the M2 Pro has two SK Hynix 4GB LPDDR5 RAM modules on either side of the core—a total of four. These are the very same RAM modules we found in the M2 MacBook Air.
Could Apple’s Supply Constraints be Shrinking RAM?
So the SoC’s are of comparable size, they’re just arranged differently. But why is Apple using four RAM modules this time around instead of two larger ones? I posed this question to Dylan Patel, Chief Analyst at SemiAnalysis, who said “ABF substrates were in very short supply when Apple made the design choice. By using four smaller modules rather than two larger ones, they can decrease routing complexity within the substrate from the memory to the SoC, leading to fewer layers on the substrate. This allows them to stretch the limited substrate supply further.”
And the supply woes don’t stop there. While Apple is busy splitting the RAM modules down into smaller packages, they’re moving in the opposite direction with the NAND modules, going from four smaller 128GB modules on the 14″ M1 MacBook Pro up to two larger 256GB modules on the 14″ M2 MacBook Pro. We saw a similar move on the base model of the M2 MacBook Air which saw a significant drop in read/write performance and we’re reportedly seeing the same again with the base model of the 14″ M2 MacBook Pro. Only the base models suffer from this 20-40% performance hit.
What gives? According to Patel, the reason for this is simply because the smaller 128GB modules are becoming harder to come by and more expensive to purchase as the smaller dies are phased out and the industry moves towards ever larger NAND die densities. Which raises the question, why is Apple even bothering with a 512GB version?
Overall, the teardown provided us with some very interesting insight into the Apple way of performing repairs.
Apple’s Self Repair Still Has a Big “But”
There is of course the elephant in the room: parts pairing. As it stands, despite the increasingly repairable designs, the software locks that Apple maintains will result in waste as otherwise-useful components end up in landfills instead of being repurposed. The useful life of our devices will also be limited to Apple’s hardware support—whatever they decide that may be. Once support is dropped for a device, those software locks will remain in place which means even if a third-party manufacturer is willing to step in with replacement parts, those parts may be restricted in functionality.
It took us 20 years to get these manuals but momentum is building fast. We’ll continue to fight for our right to repair our stuff and that means tackling parts pairing head on, whether that’s in a phone, laptop, or tractor.
You guys always do a great job for us. Thank you - Big Time
Warren Fitch - 回复
Any chance of getting full shots of the logic board for component identification purposes? In particular, I was wondering about the DisplayPort to HDMI 2.1 protocol converter chips...
repoman27 - 回复
Absolutely, chip ID will follow soon.
Shahram Mokhtari -
Do the 1TB models use two 512GB chips, or a single 1TB chip? Does the 2TB model use four 512GB chips or two 1 TB chips?
Robbed Developer - 回复
The ball pitch for the NAND packages has changed significantly between generations, going from 110 total pads to 315. There are likely internal changes beyond simply moving to higher density NAND dies, so just looking at the number of packages may not tell the whole story as far as performance is concerned.
For anyone with access to an Apple Silicon Mac, you don't need to tear down the machine to find out how many NAND packages / dies are present in the SSD. You can get fairly extensive information about the SSD architecture using the ioreg tool from Terminal:
ioreg -lrw0 -d1 -c AppleANS3NVMeController
The "Controller Characteristics" key contains several properties of interest. "dies-per-bus" shows the number of channels and how many NAND dies are connected to each one, and "bus-to-msp" shows the number of memory signal processors (MSPs) and which channels are connected to each one. Historically, each NAND package contains a single MSP, so the number of MSPs is also the number of NAND packages.
repoman27 - 回复