See the video UPGRADING a Brand NEW M4 Mac mini by dosdude1. That was fast. Kudos to this hardware hacker for planning ahead and ordering the SSD chips in advance. The thumbnail for this video is priceless:
Video spoilers: I love that @dosdude1 failed with the 1TB chips and simply re-tried with the 500GB chips he had on hand. Maybe he cooked one or both of those big chips. In any case, fully documenting the failure and re-try was great. Using a G4 iMac as a spare monitor left me with a feeling of euphoria. That black goo that Apple uses around the soldered memory chips looks dubious. #whatwouldifixitsay
I’d love to see Jason Snell or other tech journalists ask Tim Apple about this kind of field upgrade during a quarterly earnings call. IMHO, the EU should be paying attention to Apple’s absurd upgrade pricing. And Google must love throwing a spanner at Apple by publishing videos like this.
2 Likes
Also see iBoff RCC (a company in Malaysia) and their YouTube channel. They have produced a tiny PC board to replace the soldered Mac memory with removable memory. Their board and port have the same geometry as NVMe, but those chips are NOT compatible. iBoff will make/market memory modules that use their pinouts; other manufacturers could do the same. These and other issues are addressed in their video No More Storage Limits : Answering All Your MacBook Questions!
That is a wonderful Q&A video – highly recommended even if one has no intention of ever modifying any of your Macs. They provide broad brush strokes about how they designed and created their lose-the-soldered-memory adapter board. They even show the University/Industry lectures providing them the theory for making a correctly-behaving high-performance interface:
- transmission line phenomena
- impedance matching
- trace routing to minimize crosstalk
Others could follow their path. They also give a nod to @dosdude1’s work in their video.
At school In the 80s, I took a course taught by Tom Knight where he discussed various software/hardware issues in modern computer design. He threw us a problem set where we calculated the voltage on signal ground where all the (32?) address lines coming into a chip go from 1 (a high signal voltage) to 0 (a low signal voltage). In the problem set, there was a single pin on the chip providing that signal ground. We quickly learned why this was a nightmare scenario – why one would never use a single line to remove all of the stored charge from all those lines. The [theoretical] voltage spike on that one ground line was spectacular. I never did any hardware design, and I recall doing badly on that problem set. OTOH, Knight left me with valuable lessons on impedance that have broad applicability elsewhere. As an aside, I’m somewhat sad to see that MIT’s EECS department has dropped requirements that software engineers take any basic electronic and signal-processing courses. Knowing RC/RLC may not be essential for programming, but it sure as heck is a useful skill for understanding the world – and it goes far beyond understanding electronics.