My first 'PC' used an early non-turbo XT-clone
motherboard I bought at a
swapmeet. At the same swapmeet I bought an original 63.5 watt IBM
supply, a used 'Leading Edge, model D' case (just the sheet metal bits),
and an IBM MDA card. I already had an IBM keyboard that had no case on
it. When I got the whole mess home, the only way to fit the XT
motherboard in the case was to cut away all but one of the 'card edge
slots' in the LE case as it used different card connector spacing. The
only way to fit the power supply was to strip it entirely out of it's
original IBM case and mount the circuit board on standoffs in the
Leading Edge case. So that's what I did. Add two 360K floppy drives
and an original IBM MDA card and you've got my first PC. What a
conglomeration that system was. It lasted for years and years. It
makes the 'case modding' that enthusiasts dabble in today look like
childsplay.
If we're having those sort of tales...
This machine started out as an IBM PC/AT (8MHz version. It's got a 486
kludgeboard in place of the 286, but said kludgeboard was designed to fit
the PLCC type of socket, not the PGA one. You were supposed to plug a
PLCC socket into the PGA one if neccessary, but doing that made the board
sit so high up that it blocked one of the expansion slots. Out with a
soldering iron, off with the PGA socket, and in with a PLCC one.
Then there was the drive table. I wanted to use a large hard disk than
IBM had an entry for. I found out that the last 8K of the ROMs contained
the various data tables. Programmed a couple of 2764s with the new
tables, pluged them into the 'spare' ROM sockets on the motherboard,
added a couple od TTL chips and kludgewires to do the address decoding.
Clone MDA card. It got the timing of the wait signal wrong. More
kludgewires and an unused gate in one of the chips fixed that.
Early IDE card. TTL, a couple of PALs, and a floppy controller chip (I
forget which one). A bit of stripboard as a daughterboard, fitted under a
couple of the ICs, and carrying a couple more TTL chips to stretch the
read and write signals to the IDE port, to drive the 'in use' LED. Only
RC delays I'm afraid, but good enough for this. Fortunately this card can
go in the slot next to the PSU, so the overhanging daughterboard doesn't
waste a slot.
One of the RAM cards started out life as an IBM 128K card. I
reverse-engineered the bits IBM didn't document and worked out that it
could take 2 rows of 256K chips (1M byte), not just 1 row of 64K chips
(128K). So that's what I did.
And the real time clock battery. I am not paying ridiculous amounts for
the right lithium battery. In went a battery holder, a series diode, and
4 AA cells. They last about 3 years, and they're cheap to replace. Easy
to get too.
There must be more I've forgotten about....
-tony