Continuing on from the last salvage post – these were salvaged at the same time and place. Unusually, there was an Apple Macintosh G4 tower there – lovely case too. If it weren’t so heavy, I would have carried the thing away. Alas, anything that’s been left at the dump spot is basically discovered and picked apart by vultures almost immediately – so if I had left it there and bought the trolley in the next day, it would have been gone.
So, instead, I took the chance to take it apart and examine its bits. Interestingly, those towers weren’t too different from regular ATX towers – the power supply had an almost identical shape, as did the motherboard power connector. The back I/O panel of the motherboard was pretty similar too, but the mounting was entirely different.
One of my friends wanted the power supply for a project – so I unscrewed it and let him have it. The RAM from all the machines always gets pilfered first – probably by the staff here who always have another elderly machine to support. And then, normally the hard drives are taken and crushed for privacy reasons …
… but not this one! It is the original drive fitted to the machine, sizing up at 20.4Gb, it’s a Maxtor 92049U3. A classic Maxtor label and shape – but you can see the Apple logo indicating the OEM nature of this drive, and possibly a different firmware. One of the nice features of the Maxtor label was the warranty end date – December 2003, for a manufacture date of 19th September 2000.
The underside of the drive shows the PCB which also has some other Maxtor characteristics. For one, Maxtor almost always used edge connectors that soldered on both sides of the PCB. As a result, only half of the IDE pins are visible from this side. They also gave this PCB a “codename” of Fish’n. I wonder if they were fishin’ for compliments – because I rather like this PCB design!
There’s a large contact area at the top for manufacturer calibration, and underneath that there is a section for flash. No flash was populated, because it is likely that the drive was in the “transition” period between hardware flash for firmware or the use of a service cylinder and some mask ROM to load the firmware from the disk on startup.
The cache RAM is visible top left, along with the main CPU, and a chip from Lucent (maybe a VCM driver?). Then there’s a TI motor driver chip for the spindle motor which drives separate FETs for each of the phases. How antique! Integrated driver chips were already available in that era – but some of the early ones did suffer from failure (Quantum ones especially).
Am I going to take this drive apart? In a word, no. This drive is a special one because it has a heap of test points scattered around. Just from a cursory glance at the PCB, you can see them marked by the “square” around them and the silkscreened text – things like VCM (voice-coil motor), SRVGATE (servo gate), POR (power-on reset?), ZCROSS (zero-crossing?), MOTA (motor-phase A), INDEX, etc. This is a treasure trove of signals just waiting to be explored. Rarely do you have such a well-marked and test-pointed specimen to probe.
The other thing – this drive still works perfectly. Smooth throughput, no strange noises.
Another thing I managed to grab was the AGP graphics card. This was an ATi Rage 128 Pro graphics card – and no, that’s not 128Mb, but rather 128-bit memory bus. It’s probably an 16-32Mb class card with AGP 4x support. The one thing to notice here is that the graphic cards for Mac machines were a bit different – while it does have the regular AGP edge connectors, there’s also another tab just a bit forward of the AGP slot. This was used to provide power through the proprietary Apple display connector to power the monitor from the PC itself.
You can see the contact was on the other side as well – one polarity on each side it seems.
And there’s the propietary connector – it looks like a DVI with a rounded shell, and it has a but of a strange clipping system, but it’s also quite different. Next to it, on the backplate, it’s written Rated 28v DC 4A Max.
Here’s something else I managed to grab. It wasn’t easy to get either but I worked it out :). Care to guess what it is?
…
You probably guessed right – it’s the CPU module with the heatsink on it. Secured by two metal clips, branded Foxconn.
Here’s the side profile of the module. It’s a 933Mhz G4. I call it a module because it’s more than just a plain CPU. See the inductor and surface mount capacitors? It suggests that there’s at least an onboard VRM (voltage regulator module) which means that upgrading between different CPUs with different voltage/current requirements could be easily done, and it’s reminiscent of what Sun did with their CPUs as well.
The other thing to notice was the lack of a fan on the CPU heatsink and its strange profile – which was designed to fit the chassis perfectly. More on that in a moment.
Having never seen one of these CPUs before, it was definitely interesting to take a closer look at it. The CPU itself was screwed down into the motherboard – that’s what the three screw holes are for. In order to get the CPU off the board, you had to undo the clips and get the heatsink off to get access to the screws. This is the underside of the CPU and we get a little bit more information about it – in this case, what you can see is one of the off-die cache chips – very “Pentium 2/3” slot-like. There’s also a very complex socket to connect it to the motherboard.
The top side without the heatsink reveals the other cache chip, the CPU itself (covered by a piece of plastic and self-adhesive foam as a shim to stop the capacitors from potentially shorting out on the heatsink). The rest of the VRM components are visible too, but not all footprints are populated.
I removed the plastic and foam shim and got a closer look at the CPU.
Exciting? Okay, probably not. I don’t know what you were expecting, but this was what I was expecting. It’s a bit smaller than the socketed cousins of the PC world – but that’s because it doesn’t need to have as much spacing between pins since it was a BGA chip, soldered onto the PCB permanently.
As for no fan on the heatsink? Well this is why …
A Delta Electronics Sensflow WFC1212B temperature sensing 12cm fan. This was pointed directly at the heatsink – and would have meant impressively quiet acoustics under load. Their design engineers definitely did the “smart’ thing – a large fan, with temperature sensing.
