Tony Duell wrote:
I
remember a little shop here in Santa Ana which had the first clone of
the IBM PC. It was a single board which had 640k memory, allowed
using 64k memories, instead of the 16K memory that the PC and XT
earlier models used.
DId any IBM PC/ST motherboards use 16K DRAMs. Yes, I know the standard
memory mapping PROM could be set up to use 4 rows of 4816s (64K o nthe
mainboard), but did IBM ever do that? There's no mention of it in my TechRef.
You may be right, I think due to fog bank between ears that it was 16k in
apple ][ then 64k in original xt, then 256k in the superboard.
IIRC, early PCs used 16K DRAMs (1 row soldered, sockets for 3 more, so
16K .. 64K on the mainboard)
Later PCs used 64K DRAMs, so 64K .. 256K on the mainboard.
PC/XTs also used 64K DRAMs, and from what I've seen all 4 rows were
socketed. There's a well-knwon modification, I am sure everyone here
knows it, to put 2 rows of 256K and 2 rows of 64K on the PC/XT mainboard
for the full 640K. This was later sort-of supported by IBM, my TechRef
has an entry for the 256K to 640K system board.
Early PC/AT machines used those special IBM piggyback 128K 'chips'. Up to
512K on the mainboard, you could flip a jumper if you only wanted 256K.
Later PC/AT boards used 256K DRAMs, again 512K on the mainboard. The
jumper is still present, flipping it wastes half of each IC (you still
have to have all 18 chips fitted, it's a 16 bit data bus + parity on each
byte).
I _think_ the PC/XT-286 had 4 off 64K*4 and 2 off 64K * 1 DRAMs on the
mainboard for 128K RAM and then SIMM sockets for 2 256K * 9 30 pin SIMMs.
The PC used either EProms, (16K I think) and the
Roms that were
shipped with the BIOS were registered. The standard Data I/O would
Do you mean there were internal data latches in the IBM ROMs?
Yes, as I understand it and saw, the roms could be enabled, using a
line that the data I/O didn't drive (29b variety any did not). So you got
FF out when you read them.
I can believe there were some odd enables on those ROMs, but I wasn't
aware of any internal latches. I will have a look again.
not read them since they were not programmable,
and needed their
output enabled to read the data.
But once someone had them in the 2716's, it was easy to get them
running in your superboard.
Surely it was trivial to use DEBUG or similar to dump the appropriate
area of memory to disk...
Yes the dump was trivial, but the R232 or other to an eprom programmer
was not.
When I built by first microcontroller system (around this time), I had to
build my ownEPROM programmer and emulator. 3 wire-wrapped boards of TTL,
distretes, a UART, etc. Plugged into an RS232 port. I couldn't, of
course, use a microprocessor in the programmer, since there was the
obvious bootstrap problem of how would I program the firmware for that?
If you could dismantle the PC far enough to pull the ROMs to attempt to
read them out in your programmer, then surely you could have added a
homebrew serial card or similar.
A real kludge would have been to use a couple of printer port cards
(which would give you 24 output lines) and a little external circuitry to
program an EPROM. These EPROMs were not hard to program 'by hand'.
-tony