7 Feb
2006
7 Feb
'06
10:47 p.m.
vrs wrote:
I'm a little concerned about the input thresholds on the Am261S0, though.
The 1.8V Vil is actually in the allowable range for a UNIBUS high! OTOH,
the DS8838/8T38/MC3438 that were used in the RX emulator won't sense
voltages that low as a high, either!
(I don't know why the bus spec Vih is so low; is it just because the DS8641
was that way? Is it is really possible to have that much leakage on a real
world bus?)
Looking at the bus specs, 105ua of leakage are allowed per bus load. I
calculate that over 9ma of load would be required to drag the bus down
to the Vih 0f 2.3V for the Am26S10. That works out to 87 bus loads!?
For the DS8838 (2.5V), I get 71 bus loads. Am I doing this right? If
so, I am guessing that the bus can't get that big, and the Am26S10 should
work fine?
So to shed more light on the Am26S10 receiver threshold I decided to do
a little lab work, since I have both DS8641 and Am26S10 physical devices
available. The Vbus vs Vrcvr input/output plots can be seen here:
http://www.ak6dn.com/stuff/26S10vs8641.jpg
As can be seen in the plot, the classic DS8641 has a very well defined
receiver switching threshold in the 1.35-1.5V range. The Am26S10
receiver threshold is in the 1.85-2.25V range, about 0.5-0.75V higher
(not unexpected, this is what the data sheet indicates).
So the Am26S10 parts are somewhere in between the DS8641 at 1.7V and the
older DS8838/DEC380 parts at 2.5V. They should work on a reasonably
sized UNIBUS system (not pushing loads and/or lengths) but may not be
optimal for some configurations (Appendix B of the UNIBUS
Troubleshooting Guide on bitsavers has some good background info on the
threshold issue).
