On Thursday 26th October, a new SMC firmware update appeared in the Software Updates window of my week 36 Macbook. Version 1.0 of the SMC firmware fixed the "moo" and overheating issues, and version 1.1 ostensibly fixes the random shutdowns that have been plaguing some of us for months.

My first reaction to this firmware update is very positive; Apple have acknowledged the RSD issue and have taken steps to fix the issue without having to send in the machine for repair! My second reaction is slightly less positive; it makes me wonder how they've managed to fix this in software...

The reason for my hesitation in rejoicing is primarily due to personal experience with RSD. My first macbook (week 26, if I recall correctly) started randomly powering down in Augus, and at the time such issues were largely unheard of. I found a few websites where disgruntled users were discussing potential causes and fixes, but nothing major. The machine underwent logic board repair and all seemed well.

A short while later, in September, the random shutdown problem reoccurred. I took the machine in, had the logic board replaced again, and went home. The very next day, the machine shut down all by itself. Less than 12 hours had elapsed. I took the machine in for re-repair, aware that this, my 3rd repair, would be the decider in whether Apple would replace the machine outright.

Now it gets complicated. Various independent sources have confirmed that, if a machine with the issue is checked, the sensor wires that run under the heatsink melt and short on the metal heatsink over the processor. Flicker user thatsiebguy has pictures of his own attempt to reroute the sensor wires and thus fix the issue (image from thatsiebguy's photostream.)

The third repair on my Macbook was delayed due to Apple recalling and remanufacturing all heatsinks for Macbooks. This new heatsink solved the issue of the sensor wires melting, and prevented the need to replace the entire logic board. Unfortunately, the parts were taking so long (and considering I'd just been without the machine for a previous unsuccessful repair) that it seemed nobody knew when the parts may arrive.

At this point, the Genius at the Apple store who had been dealing with my machine decided that enough was enough, and replaced the whole machine with a week 36 Macbook, complete with new heatsink. This was new stock that had arrived to be sold to new customers, and so the 1.0 SMC firmware update was already applied.

This raises the issue I alluded to earlier; how have Apple fixed this in software? I can only imagine that the new firmware somehow ignores readings given by a shorting sensor cable. This could be very simple (ignore readings outside a given range), or more complex (ignore readings that do not fit a sane temperature curve). Either way, it would seem that compensating for faulty sensor readings is the only way this could work.

I have installed the update, and some people are reporting increased maximum volume, which will be nice. I would hope that, if somebody has made a terrible mistake and the week 36 heatsink replacements do not fix the issue, that the new firmware will prevent a repeat of August's fiasco, but only time will tell. At least I am confident in my new machine, and any issues I hear about should come from other people who have not been through the repair-process wringer quite so often as I have!

Update: one theory kicking about the web is that there are actually several sensors around the CPU; up to 4 by anyone's guess. If that's the case, turning one sensor off doesn't sound as bad - there are still the three backing it up. If this is true, then there's really nothing to worry about. Sure, there'll be useless wires under your heatsink, but so what? It's not like you can run it with the lid off!