Setting up a VAXstation

[Tony Duell]

> > A logic analyser is a bit lile a 'scope in that it displays a grpah of
> > signals against time. The differnces are (a) it records the signals and
> > displays the recorded version (some 'scopes do that too -- storage
> > 'scopes), (b) it only works with digital signals (it doesn't display th=
> e
> > voltage, only whether they are high or low), and (c) it has many more
> > input channels (even a good 'scope rarely has more than 4 channels, a
> > logic analyser will have 16 or more).
> 
> Fascinating. The snag is, I know very little about electronics below
> the level of a broad knowledge of TTL, the rudiments of circuits and

OK, a higher-level description would be 'something that records a number 
of digitla signals and displays their relative timing'. 

The sorts of things you might use one for would be to trace machine code 
(or microcode!) instructions/addresses on a processor bus (but cache 
memory and even a pre-fetch buffer makes that a little hard!), checking 
things like handshake sequencies on an interface, recording bit-seiral 
signals so you can decode them later, checking state machine sequences, 
and so on.

What it is _not_ is a majic box that finds all digital electronic faults. 
It's a tool, which if used intellegently will help, though.

> gates and so on. I have a bit of theory, no practice. I could not
> diagnose a faulty chip or anything; my troubleshooting consists of
> swap bits until it works. And at this, I know more than most people I

You mean oyu've not been on the receiving end of my flames about this 
:-). Suffice it to say I've just written a presentation where I describe 
that as a 'ridiculous method'.

I tried it twice when I was young-and-foolish and both times it not only 
didn't find the fualt, it actually left me more confused. I ended up 
tracing the fault using test gear in the traditional way, and it took me 
a lot less time.

The main prolems are : 

1) You have to know the module you're swapping in works. This is not 
always certain, particulalry on classic machines. Whre would you find 
known-good boards for an obcure 30-year-old system?

2) E very careful if the old and new boards are not the same revision. 
This caught me once, I swapped in what I thought was the same board (same 
part number, etc), only to find (after _much_ testing) it was a later 
version that needed a backplane modification to work.

3) A fault elsewhere in the machine could have damaged the origianl 
board, and might damage the replacemetn too. An obvious example of this 
is a defective power supply that's damaging chips on the logic boards

4) Even if you replace a board and the machine works, you may not have 
found the fault. One example of this (which happened to me) can be 
simiplified into a system of 2 modules linked by some interface. Let's 
call them 'A' and 'B'. You replace 'A' and the machine works. 
Unfortunately, the old 'A' was working fine _and was withing tolerance 
for things like interface timing (but towards the edge of that 
tolerance_, module 'B' was failing and couldn't handle the timing of the 
old 'A'. The new 'A' was on the other side of that tolerance, so it 
worked with the fialing 'B', but in time 'B' got worse and worse and the 
machine failed again.

> I'm interested in things like the Sun and the DEC - and my couple of
> IBM PS/2s and even the 9" mono compact Macs - because I hugely admire
> the engineering and design of these machines from before the rise of
> the mass-market PC. Things were /different/ then.

Things were even more different before that IMHO. In many of my machines 
(and those of others on the list), theres isn't a processor chip. The 
processor is several boards of fairly simple chips -- simple gates, 
flip-flops maybe some small RAMs or PROMs. 

> I am, a bit like Chuck, mainly a software person, but one who's
> competent with hardware to a basic level. I know bugger-all about
> electronics and while I regret that, I'm not inclined to fix it now. I

I actually wonder how you can be 'competent with hardware' and 'know 
bugger-all about electronics' To me those are contradictory statements.

We also ssem to have different attidudes about learning new things. I 
tend to spend the time to learn whatever I need to fix the problem. I 
don't claim to be a programmer, I;d never write an OS or a compiler or an 
emulator, or.. from scratch. But when I had a problem which was clearly 
due to a device driver not correctly hadnling the somewhat odd hardwre 
in my PC, I learnt enough C to understand how said driver worked, and 
then editied the sources to get it to work.
> Well, I would offer it here, but my "vintage" kit is old ISA boards,
> dead MC680x0 Macs and suchlike. Probably not of much interest to
> classic collectors, I fear.

Actually, there are some ISA boards I am still looking for. Top of the 
list is an origianl IB< PGC.

I don't suppose your Apple bits include Laserwriter spares, do they? I am 
looking for partially-dead boards to raid for custom chips (PALs, 
microcontrollers, etc)

-tony