Been a bit busy lately, so didn’t have a chance to post any more nifty things for a bit. Hopefully I can make some time to show a few more interesting things.
For this post, a follow on from my previous post about a beast of a 486 – this one is a 486 that wasn’t quite a 486. People that go to UNSW would probably recognize this machine.
Yes, that’s a blurry photo, no need to adjust your monitor. But yes, it’s another one of those machines abandoned at the Electrical Engineering Building, that nobody disposed of for weeks, so I decided to take it home. Already, we can see the complement of 5.25″ and 3.5″ drives, one in an off-grey colour that was popular for larger desktops of the time, beige being the other main colour. The case was decidedly miniature for its time – this would have been the “small-form factor” of the time.
From the back, we can notice the new-fashioned “smaller” AT form factor power supply (identical dimensions to modern ATX supplies). Two vacant expansion card slots suggest that someone had cannibalized the machine before it was dumped – a network card and a sound card was probably what was taken (by guess, from the configuration common of the time). Thus we are left with a graphic card and a multi I/O card to play with.
Here’s a view from the side. Already, it’s beginning to look tight. Lets quickly go through the components:
Motherboard: AC-59610 Made in Taiwan
Unfortunately this motherboard is not otherwise identified. It is probably a low-cost budget motherboard with no real reason to shout, especially since it’s based on the SiS chipset. Warranty labels inside date the machine 16th June 1993, celebrating its 19th birthday quite soon. It interestingly advertises it has been under “Dynamic Burn-In Test”. Not an often heard claim. It’s probably an early 486 board, since latter boards used 72-pin SIMMs.
Unfortunately, it’s not quite perfect. The CMOS battery in these machines were often rechargeable Ni-Cd packs soldered to the board, and as is common for these boards, they would leak and corrode the traces around it. I ordered a new Ni-Mh replacement and soldered it in – seems to work okay despite the corrosion!
Power Supply: Godspeed GS-27B 200W
I don’t recognize the brand, and if anything, I’d have to say it’s probably a “cheap” no brand supply. The insides don’t look too special either, but hey, it still works!
RAM: 4 x 1Mb Motorola and Texas Instruments + 4 x 1Mb Mitsubishi, all Parity 70nS
Looks like these guys started with just 4Mb and upgraded at a later date. Interestingly, they seemed to have broken a cardinal rule – don’t mix gold contacts and plated tin contacts. If the board has gold contacts, you use gold fingers. And if the board has tin contacts, you use tin fingers.
CPU: Kingston Turbochip (AMD Am5x86-P75 133Mhz (33Mhz x 4))
This is an unusual CPU to find. First of all, this is supposed to be an upgrade CPU. Upgrade CPUs promise you the ability to keep that system going for a bit longer past its normal “expiry date” by offering a better CPU for an older socket. In this case, we’re shoehorning a 133Mhz AMD 5×86 core into a 486-class board. Of course, there are several challenges, notably compatibility, and voltages. As the 486’s ran on 5V core, these upgrade CPUs were innovative – this case converting a surface mount 5×86 into a pinned CPU based on a PCB with a voltage regulator on it. As the 5×86 got somewhat hot, those guys just glued a fan on top of the package, with no attention to adding any heatsink to increase surface area. Hmm.
Unfortunately, these upgrade CPUs often ran high costs, and weren’t economical for many users. They offered much lower performance than would otherwise be implied as they were hobbled by the legacy components around them. Some were rather unreliable and getting them to work stably on any board was a challenge. So I guess this is a noteworthy system, for the fact it has an unusual upgrade CPU.
Hard Disk: Maxtor 7131AT
This drive is 131Mb of goodness. Works perfectly still, was formatted by the last owner and sys’d with MS-DOS 6.22. Geometry isn’t printed on the label (aww), but it’s 1002/8/32. Interesting is the underside, which shows an Adaptec chip – they were famous for their SCSI controllers.
Multi I/O: Unknown Brand DLW-200
Based on the Winbond chipset, nothing special. Just like every other Super I/O cards of the day.
Graphics: UMC 85C408AF 512kB
A rather unknown card, with virtually no drivers available. Not a good performing card, and had artefacts when used. Eventually, we lost one bank of memory resulting in 256kB being the memory reported, and had flashing, blinking dots.
So yes, the system itself, it works. Another success story – and here’s the Landmark result:
Note how the video speed is only about 1/4 of that of the ET4000 based cards. But because this machine has an upgrade CPU, the speed is much much faster than the beast.
I decided to upgrade this machine by decking it out –
Moved the floppy drive into a 3.5″ bay, changed the hard disk out for something bigger, added a CD-ROM drive, sound card (Crystal), SMC Ultrachip NIC, dual serial port card, and a 33.6k USRobotics Modem. Lots of I/O base address and IRQ juggling but it really works! And it runs Windows 95, although the RAM is very limiting.