6 Jan
2006
6 Jan
'06
3:13 p.m.
I hate to say this again, but the most important tool
for debugging is a
brain :-). I feel that a good engineer/hacker armed with an LED+resistor
logic probe is likely to do rather better than an idoit or novice (these
are NOT THE DAME THING!) armed with the most expensive 'socpe and
analyser available.
An excellent point - My favorite related story when working at a local telecom
company, happened to visit a guys desk just as he was getting up - said he was
going to get the logic analyzer. Told me what was "going wrong", and went off
to
get it (other side of lab) - by the time he came back I pointed out the cause...
Told him I had used the LA between my ears. He just hadn't even tried to figure it
out, relying on the LA as his first action... I rarely need the LA (it's so much work
to hook up) ... Course - when I do decide I need it, I *really* need it!
1) A multimeter. Digital or analogue, the choice is
yours (I have, and
use, both). It is rare to need to make accurate measurements in classic
computer work, so analogue is OK, and in fact better when you want to
'peak' the voltage at some test point or something like that. If you go
digital, I would certainly consider a Fluke.
A Very useful feature is a continuity buzzer, a beeper that sounds if the
resistance between the probes is less than a certain, fixed, value. But
make sure it repsonds quickly (and doesn't, for example, take the time
for the autoranging system to work and then for the instrument to take a
couple of samples). You will want, quite often, to clip one probe onto,
say, a wire at one end of a cable, and run the other probe down the pins
at the other end. You don't want to have to stop and wait on each pin.
#1 tool, and I agree completely on fluke/quality. Continuity buzzer is essential,
and I find a LOT of the cheaper meters (even the fixed range ones) take too
long detect continuity - basically, they do it by taking a measurement and only
after it has done it's "digital approximation" does it figure out the
connection
is below 'x' ohms. Thats a related problem - I find the cheaper meters are fixed
at 200-500 ohms as being "continuity". Sometimes you need to know that it's
zero (or almost so), and sometimes you want to know that higher value paths
exist - thats why I built a separate continuity buzzer with very low in-circuit
current, and an adjustable threshold.
Hold function is handy, so you can take a tricky measurement without having
to take your eyes off it, but otherwise the basics are all that you need.
2) A logic probe. HP have made some nice ones over the
years (I've seen
them on E-overpay from time to time). Actually, a cheap one (Radio Shack
used to sell them) is all you need for most work. This is very useful for
fioding a signal that's stuck high, or something like that. If you are a
rich enthusiast, consider attempting to find an HP 'Advanced Logic Probe'
aka LogicDart. It's a handheld thing that acts as a digital voltmeter,
frequency meter, logic probe and 3-channel logic analyser. A word of
warning, if you ever use one of these you will be 'hooked'....
Disagree on the cheaper ones - Tried the RS one - it doesn't represent a
valid TTL input (it is supposedly switchable for TTL/CMOS) but don't
believe that you are actually close to representative of a real input.
These days I tend to use the scope when I need a logic probe. With a
slower sweep you can see even short pulses, you can also see the
level and what the pulse looks like if you need to. Yeah, it's overkill, but
I have them and they do the job well. My latest one (the TDS) has a
nifty "autoset" function that will get "most anything" viewable very
quickly
which makes it just about as easy to use for the task. Also, the scope
can see the relationship between two signals much easier than a probe.
3) A 'scope. I would always consider Tektronix
here. Actually, I rarely
use a 'scope, other than for disk drive alignments and PSU repairs,
neither of which need a particularly highly spec'ed instrument. Mine is a
very old, valved, Tektronix 555 with an assortment of plug-ins. If I was
buying now, I'd consider getting a second-hand 7000 series or 460 series.
Nothing much more recent, IMHO Tektronix went way downhill when they
stopped putting schematics in the user manual.
A typical spec for most classic computer work would be dual trace, at
least 20MHz. External trigger -- and a good trigger system -- is
essential, if you can't keep the trace still you can't measure from it.
Delayed timebase (or a second timebase which can be used as a delay) is
very useful. Storage is useful, but by no means essenital.
I went for years with scopes like you describe, however I think it's like your
"Advanced Logic Probe" - once you get used to a good DSO you get
"hooked".
I find the ability to look at a single shot (or very short duration) event to be
most helpful. I'd say a scope is my #2 instrument.
4) A logic analyser. Tektronix made a reasonable one
as a plug-in for the
7000-series 'scopes. HP and Gould also made analysers, either stand-alone
or as plug-ins for 'scopes.
I would say at least 50MHz and preferably 100MHz is fine for classic
computers. 16 channels is enough (you can get away with fewer if you have
to). Make sure you get the probes/pods with the instrument, they are
_hell_ to find on their own and often get lost/separated from the unit.
The ability to transfer the captured data to another computer for more
analysis (often via an RS232 or GPIB port) is very useful.
Don't know where to put it on my "list" but it's a ways down. Most
repairs
can be figured out quicker by other means, but when required, it can be
exceptionally handy. Whatever you pick, put some time into learning it's
capabilities as there is a lot you can do with these things.
5) An EPROM Emulator. I put this under test equipment
because it's very
useful to replace the ROMs in a system with an emulator containing a
little test program (even something as simple as a jump to itself), and
see what happens. I built my own, they are not complicated.
This one is pretty high on my list, perhaps #3 or 4 when fixing a computer.
Once you get the CPU and ROM to operate, you can start doing diagnostics
from "inside" with an eprom emulator - often
in conjunction with the other
tools listed above (use the CPU to generate a
repeating access and then
follow it through etc.)
Also agree with the bench power-supply you posted in another message.
I built one with a nice variable section, as well as fixed 5V and +/-12V
outputs. Panel volt/amp meters are essential.
Another tool I use a lot is a little analog amp meter - the old fashioned
"wedge" kind with two terminals on top - very handy from time to time.
I also have a little "clock generator" I put together which has DIP switches
to set the frequency and generates square waves - occationally it has been
very useful to clock a circult at a certain rate - often to slow down "fast"
stuff
so I can see what it's doing.
Course in our line of work, the "light bulb box" and Variac rate pretty high
on the bench tool list too!
--
dave04a (at) Dave Dunfield
dunfield (dot) Firmware development services & tools: www.dunfield.com
com Collector of vintage computing equipment:
http://www.parse.com/~ddunfield/museum/index.html