It’s been another busy week – and as I wait for the autoclave to cool, here’s something from a salvage done earlier in the week …
But before I talk about the item itself, the circumstances surrounding the salvage deserve a mention. It was a Tuesday afternoon, roughly 2pm, at the “it’s not a dump” dumping ground – there were three discarded machines, and I was sitting there alone with my totally rubbish screwdriver set.
A former student of mine walks up to me and joins in – he needed a power supply, so we cooperatively undid the screws, and liberated the cables from the chassis. And then more students join in – it was a salvage party. Most of them were third year students of the Electrical Engineering program, all aware of the “salvage spot” around here at UNSW. I encourage them not to let anything go to waste – after all, one man’s trash is another man’s treasure. Taken to the extreme – some of these things could end up being worth money – when they’re antiques. (On the other hand, it’s hoarder speak – so one has to be selective about what they salve).
Then, one of them says to me – “Hey … uh … aren’t you that guy who salvages things and posts it up on a website or something?” “Yes … yes, I am!” Recognition at last! I didn’t think anybody would end up recognizing me because of my random posts on my blog, but apparently there’s a few followers.
I did manage to grab quite a few items before leaving – but I haven’t had the time to test them yet, so I will progressively post them up for you to all marvel at.
Today, it’s a Quantum Fireball 1280Mb AT hard drive. 1280Mb isn’t much nowadays, but back then, 1280Mb was too much for your old regulars running DOS (528/540Mb CHS limit).
This one was an IBM OEM drive – likely originally shipped in an IBM computer and then moved to the chassis we found it in. You can tell because it has IBM P/N (Part Number) and FRU (Field Replaceable Unit) numbers on it. It is dated September 1996 which makes it almost 17 years old. The geometry is listed on the label as 2484 Cylinders, 16 Heads by 63 Sectors per Track – but it’s important to note that the geometry itself no longer bore any resemblance to the physical geometry used by the disk itself at the time. IDE electronics would translate the requested CHS values (logical CHS) to physical sector locations on the disk. There would be even more trouble, as different BIOSes had different ways of translating CHS values to ensure the OS could access the full capacity of the disks … but that’s another hell very well covered here.
Anyway, I’ve had fond memories of Quantum drives. They were generally reliable enough, very light (alloy chassis) compared to the counterparts, and had a cover design which remained the same for almost all of their drives. They also had very low power consumption figures – whereas other drives advertised close to 1A on each rail, Quantum drives regularly were below 0.5A on 12v. Indeed, this one is 680mA on 5v, and 300mA on 12v.
They typically had very clean PCBs and undersides, although early Quantum drives had a large massive set of jumpers – many of them reserved. Some were for self testing which resulted in a very nice track-to-track seek of the surface, followed by a butterfly seek. This one doesn’t seem to have that anymore. There was an annoyance – the jumpers themselves were miniature sized, so jumpers from most other vendors just won’t fit.
Not easily evident was that each Quantum drive PCB had its own hard disk indicator light (bottom left area) which let you know which drive was working at any given time. Quantum drives also featured (normally) generous cache RAM to improve performance (top, centre Sanyo branded chip). This design is a very integrated design – with very few chips.
What you might not know is that Quantum drives might as well have been branded Panasonic. Why? Because most of it was made by Matsushita – Panasonic’s parent company. It’s evident just above the cache RAM, there is a PCB inscription of a (M) logo which is for Mastushita.
The drive itself has a standard IDE connector interface, and a 4-pin Molex style power plug. The 3 pin connection next to it is not for master/slave, nor for a serial console. In fact, it’s a different power connector for some other systems, wired for 5v, 12v and ground over 3 pins instead.
There’s some red-printing on the other side of the base as well. Unusual choice of colour. You can see the recess where the flexible flat cable enters the “clean” platter chamber with very-compressed beige coloured foam.
Well, I plugged it in and powered it on. It wasn’t happy.
It was loud – as was the tendency for ball-bearing based spindle motors after they’ve been run for a long time. The sound varied a little depending on what track it was on, ranging from a loud whine, to a cordless angle grinder.
Then, after a few minutes, the spindle stopped. The activity LED on the bottom blinked out a code – four blinks, two blinks. And it repeated.
Obviously, the internal system has spotted something wrong. Irregular spindle speed? Excessive runout from spindle bearing wear? Missing servo marks? Failed head? I wasn’t sure, but I bet that we’d see something interesting if we opened it up because it sounded like media damage.
Another reason to like Quantum drives was the fact they were easy to tear apart. There were only two things hidden under labels – a nut which you could undo with pliers, and a “security” screw over the spindle motor itself which needed a Torx screwdriver. Everything else was Phillips (in general, aside from the platter clamp).
Lid off, the drive showed its true form –
The platter clamp only has three screws – the lowest number I’ve seen in a while. There are two platters and four heads in total. The drive has auto park (as later non-Winchester based IDE drives always had) which lands the heads in the inner area and locks them with a red plastic “wind vane” style lock which unlocks when the airflow around the disks reaches a certain pressure.
If you’re eagle eyed, at this point, you would already notice the concentric ring around near the outer edge of the platter and go “Oh my god, what is that!”
The answer would be, it’s a good old classic head crash, although it probably happened really slowly. It’s on every platter surface, top and bottom.
My guess? This drive has done a long time in a machine that was never turned off. The drive probably had very little to do, but remained powered up almost all the time. The drive’s firmware probably idled the head tracking the outer track (a bad idea) and over time, the air-pressure of the head riding the air bearing, along with an occasional bump or vibration caused the magnetic media along that specific track to get eroded to the point where it just lost enough media to make it fail. It could have roughened the surface so much that it literally “grinds” the head and causes more head crashes, or maybe wore through the diamond-like coating causing the magnetic material to be lost and clog the heads.
Whatever it was, this drive is toast.
Interestingly, idling the heads over one specific region of the disk was one of the reasons behind the early failure of the deathstars … and was one of the reasons for the need for some firmware upgrades.
Some more pictures of the heads, arms, etc.
The dust? well most of it was because it was sitting open in my room for a bit too long …
Taking off the PCB, we find another Quantum-specific surprise – the spindle motor connection. Commonly made by spring contacts – this one is made from a silicone elastomer connector, commonly used for low currents – like connecting an LCD screen to a PCB. This sort of connector relies on pressure against its spongy body where little “filaments” of conductive material run in parallel, insulated from one another, from the bottom surface where the PCB is, to the motor itself.
And here’s the other Quantum drive surprise – a label which has MKE Made in Japan on it. MKE in this instance stands for Matsushita Kotobuki Electronics.