Hi guys.

I have in my collection an ISA card that I tough was a 486 powered System on Board. I've had it for a few years now, and only recently did I think to use it in one of my P3 rigs - to build the ultimate Combo Win98 - DOS gaming machine.

My plans were ruined after I googled the board trying to identify it and find a manual / instructions on how to use it. It turns out that the board is actually NOT a SoB, but some sort of communications interface used by Telecom companies in the mid-90s. A proper System on Board should have connectors or headers for VGA and I/O, witch this board does not. In stead, it uses two 25 pin D-SUB connectors to interface with proprietary comm equipment.

A bit frustrated by my discovery, I started googling the chips on the comm board to confirm my findings - and confirmed they were. As it is, the board has no value to me, albeit for spare parts - but even so, it has no discernible NB or SB chips, and the RAM chips are soldered to the board. I pried the heatsink off the CPU using heat and cleaned off the glue with nitro and a guitar pick, to see what kind of 486 I'm dealing with, and behold:

... the board is powered by an Intel DX4 100MHz 486 with 16KB of Write Back cache. I don't have one of these - in fact, the only Intel DX4 I've ever see was inside of an IBM Aptiva. Most 486 machines I've seen were either AMD or Cyrix powered, with the exception of a few early Intel 33MHz DX desktops, making the DX4-100WB a very rare chip in my country.

This seems to be an OEM cpu, A80486DX4-100 model, part number SK096 - more details can be found here: http://www.cpu-world.com/CPUs/80486/Intel-A80486DX4-100.html

Not wanting to let the CPU go to waste, I broke out my trusty Lukey Tech 852D+(yes, I spelled that correctly... it's chinese made...) SMD soldering station + solder sucker, and proceed to carefully remove all the solder form the back of the Pin Grid Array. After a couple of hours, I managed to set the CPU free from the PCB:

(I have no ideea what the black gunk is, but good thing I didn't plug that board into any of my computers 😁)

As a result, the CPU's pins are all messed up - some bent, and most covered in excess solder and the black gunk that was under the CPU. The black stuff looks like the side effect of some sort of explosion, but it's in fact organic in nature and smells a little like plum jam 😜.

I spent another hour cleaning as much of the excess solder off the CPU's pins as I can, using de-soldering wick and my solder pump, then another 30 minutes or so to straighten the pins w/o braking off any of them. I used a 10 bani coin witch is EXACTLY the same thickness as the space between a 486 cpu's pins and the blade from a box cutter to align them. Results were surprisingly good.

Now all that was left is to test the CPU. I pulled out my Pine PT-432b test board (UMC Chipset), and carefully seated the CPU in:

SUCCESS! Well, all-most:

As you can see, some solder is still left on the pins, so the CPU does not go all the way into the socket. Despite this, the machine powered on and POSTed just fine. In fact, it will also post and run quake 1 sw at 120MHz! (3x40MHz).

I then installed my 500MB Test HDD with DOS 6.22 and benchmark suite, and ran speedsys at both speed settings. Here are the results:

@100MHz - 3x33MHz - Pine PT-432b, 32MB EDO (1 stick)

@120MHz - 3x40MHz - Pine PT-432b, 32MB EDO (1 stick)