10 Mar
2005
10 Mar
'05
4:02 p.m.
On Thu, 10 Mar 2005 13:19:31 -0800 (PST), Eric Smith <eric at brouhaha.com> wrote:
Yes, using the DS3662. Unfortunately it has been
discontinued, as was the
DS8641. Surplus DS8641 chips seem to be not that hard to find, I'd
probably use those.
We used DS8641s with our Unibus COMBOARD line with great satisfaction.
Of course,
we also use DC013s for the interrupt logic, and the only place I know
to get those is
from dead boards.
I don't know of any current production parts that
meet the Qbus specs.
Most drivers don't meet the Vol spec at full current. The receiver Vil
and Vih specs are hard to meet, as they aren't typical of logic parts.
The leakage specs are also a problem for many parts. DEC used specially
selected DS8641 chips.
It's not a DEC Qbus design, but I did see SMT 7438s on CompuServe's 80386
X.25 node processors. I couldn't ever get my hands on one, but I did get to
stare at one on the bench once... the cages were true Qbus, 9 slots per
processor (3 independent backplanes per enclosure), with modern, not DEC,
peripheral cards. From that standpoint, I guess one would more accurately
describe them as "a bus standard in the style of the DEC Qbus", since by
the end of the CompuServe Qbus X.25 Node era, it was 100% purpose-built
hardware and 0% DEC, removing the requirement for enforced backwards
compatibility. But that having been said, I think the 7438 comes closest to
satisfying most of the required bus characteristics, and would probably work
in anything short of a maxxed-out arrangement (which is why DEC's spec
are as stringent as they are - they guaranteed a wide performance range).
In short - the 7438 isn't a perfect hit, either, but it was used back
in the day,
and might be good enough for most applications.
-ethan