21 Jul
2000
21 Jul
'00
12:08 p.m.
You're right, Tony, many of them are cheap because they're incomplete. I
paid over $1k for mine, but it was complete, right down to the manuals and
GPIB/RS232/Epson Printer adapters. It also came with ROM packes and
self-test jumpers.
I've seen plenty of LA's without these, and even a number of
pods/lead-bundles without the LA, in each case worth little without the
"other-half." I wonder why surplus dealers let this happen? They'd get
10x
the money if the equipment were complete.
Dick
----- Original Message -----
From: Tony Duell <ard(a)p850ug1.demon.co.uk>
To: <classiccmp(a)classiccmp.org>
Sent: Friday, July 21, 2000 1:45 PM
Subject: Re: Other useful test equipment (was: RE: Scope use...)
instrument.
Another useful piece of equipment is a multi-channel
logic analyzer. Most of these units can read out
Agreed. My Logic Analysers get a _lot_ of use.
Now, before anyone flames me, I will admit
that
new versions of these can cost multiple thousands
of dollars, but I've noticed that a significant
number of older units are showing up at hamfests
for << $200. (I ran across a Tek. DAS analyzer at
the Trenton Computerfest for <$25.00)
I picked up an old Gould unit for \pounds 55.00 over here. It's 16
channels, 100 MHz.
One thing, make sure the pods/probes are with it. These probes are
'active' -- they include a high speed comparator circuit. They're
expensive to buy (if they're available at all!), and non-trivial to
build. They're also specific to a particular model (or range of models)
of analyser.
Mine was _missing_ the probes (probably why it was cheap). Fortunately
there was a service manual with it, and I discovered that the inputs to
the analyser were differential ECL signals. I made something using 10124
(TTL-ECL translator chips) that gives me TTL inputs (which is all I ever
need), but I don't get the fun-but-useless-most-of-the-time variable
thresholds mentioned in the manual.
These older units are becoming useless for
new and/or cutting-edge hardware development,
and are being dumped by a number of hardware
development firms, since most of them can't
sample systems with clocks > ~10-20 Mhz reliably,
There are some reasonable 100MHz instruments out there which aren't _too_
expensive. Those are easily good enough for any classic computer repair.
And don't feel that because the master clock on the PCB-in-question is
100MHz or something of that order that a 100MHz (or even slower) analyser
is no use to you. There's a lot you can test even with a slower
Some instruments will are designed to sample faster if you use fewer
channels (perhaps 16 channels at 25 MHz, 4 channels at 100MHz --
obviously they load the 4 channels into 4 shift registers and read them
out in parallel (16 bits) every 40ns. Not as nice as a true 100MHz
analyser, but better than nothing. This trick _can_ be used externally to
'push' an anlyser that's too slow for the circuit-under-test, but it's a
bit of a fiddle to get working properly.
Modern analysers have all sorts of fancy display modes (for example they
can be connected to a CPU bus and they'll display a disassembly of the
code that's being executed). But that, IMHO, is secondary to getting the
data in the first place. If you can read in the bits, you can figure out
what they mean later. Even older analysers often have a IEEE488 (or maybe
RS232) port so you can link them to a computer to interpret the data.
Also, I suggest that you stick to analyzers
made by
larger companies such as HP and Textronix. You'll
Gould/Biomation is another one to look for.
-tony