anyone do vaxstation repairs?
ard at p850ug1.demon.co.uk
Mon Feb 25 16:41:53 CST 2008
> Sure, I agree with those points, but I don't necessarily think that
> means people are *incapable* of component-level repair.
> I recently got a new job (I was laid off awhile back; the ISP
> industry isn't what it used to be) in which I'm doing hardware design.
> I was hired to bring some experience to the team; the other two
> engineers are frighteningly intelligent but are still in college as EE
> students...lots of knowledge, but very little practical experience. I
> believe that either of these guys (for example) could easily do
> component-level repair if they were asked to do so.
There is a lot of differnce between design and repair (I've done both,
and FWIW, I do not regard the former as being necessarily more difficult
than the latter...). Of course there are some common skills for both --
you'd better know how the components behave, how to solder, etc. But
equally, repair (or more precisely diagnosis) of a design that's worked
once is IMHO very different from working out why your newly-created
design doesn't work -- for one thing in the latter case I would hope you
know how you inteded it to work in the former case you may well not fully
understnad the circuit.
> The ASIC problem is a biggie; there's no getting around that. But
> lots of things are still quite repairable.
ASICs bring 2 problems. THe obvious one is getting replacements, but the
obvious answer to that is to raid them off old boards...
The lkess obvious problem is that the behaviour of an ASIC will probably
not be fully understood. Which means you may well not be able to tell
which signals on the pins are correct, which are incorrect, and wheter
the incorrect ones are caused by incorrect inputs or a fialure inside the
ASIC. That makes fualtfinding a lot harder...
> > YEs, AFAIK there were never published schematics for the VAXstations.
> > And
> > since (from the few I've looked inside) they're full of custom silicon,
> > it's going to be very hard to produce meaningful schematics of such
> > machines.
> Well, schematics including the custom silicon as blocks with pins
> (much like a Z80 or something) that could allow one to narrow down the
> failure to (say) one particular chip, which could then be replaced.
If you take a VAXstatiuon board, you could produce a sort-of scheamtic.
It would show the ASIC as a 200-pin (or whatever) device and show what
each of the pins linked to. Some might be very understnadable (power,
ground being the really simple ones). But then some might not,
particularly if there are several ASICs with traces that just link them.
Given that sort od scheamtic, you could, I suppose, find broken tracks (2
pins not connected that the schematic shows should be) or shorts (the
reverse :-)). But other than that, it';s not a lot of help. If you stick
a 'scope (or whataver) on a pin and see a particualr signal, you can't
know if it's right or not, You can't know if the timing is reasonable. So
how do you know what to replace?
A _meaningful_ schematic (and in my experience a lot of commercial
schematics are not 'meaninful' in this sense) firstly names all useful
signals (clocks, address/data buses, bus strobes, etc) and secondly
shoews circuits in the conventional way (e.g. a pair of cross-coupled
NAND gates froming an SR flip-flop is shown just like that, you don't end
up with the 2 gates on different sheets). Having done it a few times, I
am convinced that producing a meaningful schematic is a lot of work...
More information about the cctech