12 Nov
2005
12 Nov
'05
1:47 a.m.
While I, like a lot in this group, tinker with old computers, I have
started another project that involves more than old computers. Hope this
is close to being on topic, some of you may find it interesting.
I purchased parts of a Nicolet Fourier Transform Spectrometer with the goal
of putting together a working Michaelson Interferometer. What I have is
What fun!
[...]
Any tips on how to efficiently reconstruct circuit
diagrams from inspection
of the electronics? There are two cards, with standard TTL and analog
parts. The printed circuit has a design date of 1985, while the IC's have
88-89 dates.
Good luck! I've traced out a fair few schematics over the years, so can
possibly give you a few hints.
Firslty, difficulty of reverse-engineering depends on the complexity of
the components. Easiest first, it goes like this :
Large LSI chips that are known (e.g. microprocessors, LSI I/O devices).
These can essentially only be used in one way, so yu can identify a lot
of signals from one chip (like address and data buses)
TTL gates and flip-dlops. Harder. A NAND gate could be doing a million
different things (it could be half of an SR, too...).
Disvretes. Even harder. A transistor might not even be a digital function
(it might be part of a power regulator even on an essentially digital
board, for example). Do not pick the HP9100 as your first
reverse-engineering project :-)
Unknown/custom LSI parts. Painful. You don't know what the pins are. If
you can power the thing up you can make some good guesses as to some of
them by examining the signals (I did this for some of the HP calculator
chipsets, I did know roughly what signals to expect, I just had to find
out which was on each pin).
Assuming you know what all the ICs are, start by indentifying the power
and ground rails (shouldn't be too hard on a board of TTL). Do watch out
for chips not powered from the main rails, though.
Make a list of all the ICs and the sections in them. It will look
something like this :
U1 a b '74
U2 a b c d e f '04
U3 '163
U4 a b c '10
Then, as you draw out each gate/chip, cross it off the list. That way
you'll know what you've not done, and conversely you'll not draw out the
same stuff twace.
If the ICs are house-coded and you know the equivalents, you put the
standard number on the list.
If necessary, list the transistors (and even R's and C's) too.
If there are no 'U1' etc numbers on the board, you have to assign them.
Either sketch out the layout of the board and write a number for each
chip, or stick labels on the actual ICs. I normally do the latter -- the
local stationary shop sells sticky lables numbered 1-160 (3 times over in
different colours) about 6mm in diameter. They sitck nicely onto most ICs.
Make sure you have data on all the chips. A TTL databook is essential,
often asking here will get pointers to info (it's worked for me in the
past :-)).
Use any info you have about external connections to identify signals. Not
a lot of help in your case, but if, say, you're doing an ISA card,
obviously use the slot pinout.
This way you'll get a schematic of the board. How much help that is in
linking up to the mechanical/optical side is another matter.
-tony