No one seemed to notice this, but I thought I'd post it for Tony's
edification. It's a story in the October 7 EDN from the "Tales from
the Cube" back section and deals with solving a problem with the I/O
board of the HP9845:
http://bit.ly/cNwokC (it's a PDF)
Interesting... Why specifically for me, though (other than I have an
HP9845B which AFAIK uses the same I/O backplane PCB).
A few comments...
For those who don't know the machine, the I/O backplane fits at the rear,
behind the PSU. It has 4 edge connectors for interface modules which plug
into the rear of the machine. The PCB carries the connecotrs, some bus
buffer chips and not a lot else.
The problems with the origianl design seem to indicate that 'there is no
such thing as groudn' )(with thin tracks, 2 points that both claim to be
'ground' woukld not always be at the same voltage), and that the most
complciated part of a digital circuit is the wire :-)
I am also a little suprised by the comment that logic analysers didn't
really exist at that time (not that a logic analyser would have been a
lot of help fo this fault). I am wondering how the older HP machines
were actually debugged. Having repaired several of them, I find a logic
alauyser to be next-to-essential. Even more so if the design is not known
to be sound.
-tony