31 Jan
1998
31 Jan
'98
11:57 a.m.
1) They're hardly original, which may bother some
people
The reason for using FPGAs (or whatever) is good if you can not find a
donor machine.
2) You'll probably have to make up a kludge-board
to get the power pins
and clock inputs in the right places. Doing that for a chip with pins on
a 0.5mm spacing will be interesting to say the least.
I did not say it would be _easy_. Anyway, many FPGAs are available in the
same styles of packages.
3) Remember you probably don't have complete
timing specs of the chip
you're trying to replace. You may not even have a functional spec. Trying
to design a replacement will be interesting...
Incredibly hard today, perhaps easy in the future. Envision a logic
analysis program that one could clip on a working chip (from another
machine, obviously) that would think a while then spit out several hundred
pages of HDL describing what it saw.
There's a chip in my HP Omnigo made by Vadem.
It's a complete PC/XT apart
from the memory (but including the CPU, DMA, interrupt controller, serial
port, keyboard interface, parallel port, etc) together with an LCD
controller. Trying to emulate _that_ in an FPGA would be an interesting
task, even with the data sheet (which I have, all 230 pages of it).
In ten years, what will FPGA densities be like? Probably way beyond what
they are now. How about core sets that could be dropped into (or added to)
FPGAs? The Z80 still lives in this fashion, with gate arrays, and probably
will continue to for ten more years.
William Donzelli
william(a)ans.net