27 Aug
2009
27 Aug
'09
5:46 p.m.
Don North wrote:
The 26s10 has an input threshold of 2.0V +/- 0.25V,
where the original
DS8640/41 had an
input threshold of 1.5V +/- 0.2V. UNIBUS terminates all the bussed
signals to 3.5V, so
you indeed lose 0.5V of high side noise margin using the 26s10 vs an
8641 (but you also
gain 0.5V of low side margin). Indeed, with the 26s10 the high and low
margins are now
symmetric (3.5-2.0=1.5; 2.0-0.5=1.5); with the original 8641 they were
very asymmetric
(3.5-1.5=2.0; 1.5-0.5=1.0).
If Unibus used totem-pole drive, that would be a good
thing, but it
doesn't. On a heavily loaded Unibus, the signals may take significantly
longer on deassertion to reach 2.0V than 1.5V. Like I said before, you
can probably get away with it on smaller systems, but it's likely to be
less reliable.
Eric