EDIT: Whoops! Turns out my isopropyl-dry-clean-out of the spindles and blades wasn’t quite enough — particularly with the left fan. First (second) fix was adding a small puff of graphite powder to the spindle hole. This worked for about a day (for the left fan; the right fan is working fine once more) before briefly making that loud, whining noise, followed by just stopping entirely. The second (third) fix, for the left fan only: leaving in the residual graphite powder but adding a very small amount of dry-ceramic lubricant (engineered for higher-end road/touring bicycle drivetrains). It is now running quietly and its low speed before shutting off is no longer in the 1000–1200rpm range; rather, it’s a bit slower 500–700rpm (the ceramic lubricant’s viscosity is likely why). Not perfect, but it’s enough for now.
Over the weekend, I did the same for my A1139 — except after cleaning, I decided to explore something different.
The reason for doing so is the A1139 fans are engineered to run at up to nearly 7000rpm (whereas the fans in the A1138 top out at around 3500rpm). The two models use very different fan designs, despite both models being released on the same day, both with brand-new logic board designs.
There is a lot of DIY advice on using specific lubrication methods for some of the faster fans. For the A1138, I chose to keep them dry after cleaning them thoroughly. For the A1139, I opted not to use mineral oil or conventional sewing machine oil (usually, a lower viscosity mineral oil), but a synthetic oil designed for sewing machines, typewriters, and power tools (namely, drills).
The thing about working on my A1139 is I find it to be one of the most stressful to deal with. One wrong thing, more so than on other Mac laptops I have, and the system kernel panics almost immediately or the display has issues (one of at least three different ways I’ve documented in the past).
This time was no different:
the fan design uses an adhesive gasket to hold the fan and motor to its enclosure, and that adhesive, now 16 years old, isn’t strong enough to re-seal perfectly after working on the fan rotors (enter kapton tape to the rescue). The A1138, by contrast, uses screws to hold the two halves together. So after cleaning and lubricating the fans, which on the A1139 first required removal of the logic board and display assembly, I put in fresh paste (it’s amazing what three years of always-on use can do to the internals of a PowerBook).
After assembly, which itself was stressful (particularly with trying to get the display assembly reconnected correctly with the base), I tried powering on and hit both instant kernel panic
and one of the previously documented display issues (
a single, horizontal line which acts like the “equator” of the display). The kernel panic was virtually identical to
this one, except it was Darwin 9.8.0 (not 8.1.0). But with little else to go on, other than the logic board may have been ruined inadvertently, I came back to it a day later and a night rested, with “don’t be panic” lulling me to sleep.
Next day, I disassembled again, and I backlight-inspected the board with a very bright torch (flashlight). I also inspected everything on both sides. Nothing looked burnt and no cold/bad solders could be seen. I reassembled again, and after a couple of hang-ups with the pre-inserted PC Card (an 802.11n card) preventing full start-up, the system finally booted completely. I don‘t know what was preventing it before, but I think it was probably a poorly seated bus cable (most likely the PC Card cage or the Airport/Bluetooth cable).
The “equator” display line, annoyingly, remains, which may or may not go away at some point (sometimes it’s there, and sometimes it’s not, but I think it’s probably a failing LCD; the line doesn’t show up on screen caps or Remote Desktop). Although the fans did spin prior to servicing them, they were extremely loud and would often spin to less than a third of their rated top speed, making loud, groaning noises like the laptop was dying somewhere out in a field. I’ve more or less dealt with this noise for most of the time I’ve had it.
Now, with the synthetic oil in place, they’re so much quieter, and I hope they stay that way. When I manually set them to max, they’re now running at the fastest I’ve ever seen them (roughly 200–300rpm faster than the pre-servicing max):
In hindsight, and there might be some blasphemy here, I think I prefer to work on the A1138 over the A1139, despite some of the assembly features inside the A1138 feeling like more of a kludge (given tighter space constraints). I also think I prefer to work on the internals of my 17-inch MacBook Pro (an A1261) far more than inside the A1139. The internal design of the final aluminium MBP reveals just how much refinement they could muster with this form factor after five years of incremental improvements, and I don’t get the same kind of anxiety when working on the A1261 as I do with the A1139.
Postscript: The very last upgrade I plan to do with this A1139 at some point, provided it continues to keep working, is to find and bring over a working LG 1920x1200 LCD — glossy or matte, I’m fine with either.