UNIBUS extension card/cable sets

Noel Chiappa jnc at mercury.lcs.mit.edu
Fri Jun 26 07:57:35 CDT 2015

    > From: tony duell

    >> So every other wire on the 40-conductor flat cables should be ground -
    >> that's even better than the classic BC11A, where almost every other
    >> line is, from what I can see, simply left floating (which is better
    >> than nothing, but not as good as grounding them, is my understanding).

    > I am surprised. DEC didn't waste copper like that. It's been a long
    > time since I worked on a BC11A, but I thought alternate wires were
    > grounded. Maybe a track right along the edge of the PCB where the cable
    > comes off (so you can't see it).

You're right, the alternates are grounded (ohmmeter shows it). I cannot see
how they did it; I think there must be a comb-shaped trace along the top of
the card, where it's hidden once the Flexprint cable is soldered down. The
intermediate ground conductor on one trace, on one end, _is_ connected to
ground, so the rest could pick it up via a comb-shaped trace.

    >> I would have assumed that it's the _change_ from one impedance level
    >> to another that's the issue (you can get a reflection off the
    >> junction), so whether one's using long or short cables between a pair
    >> of M9014's, it shouldn't be _that_ big a deal (modulo propagation
    >> delays, which _are_ an issue with length).

    > Well, Unibus is terminated into 180 Ohms and 390 Ohms, isn't it?


    > The thevenin equivalent is thus around 123 Ohms.

DEC spec for UNIBUS is 120 +/- 18 ohms.

    > Most ribbon cables have a characteristic impedance when used with
    > alternate wires grounded of around 100 Ohms (I seem to remember that is
    > certainly right for the twist-n-flat ones).

What's the number for the regular flat? (I have a ton of the latter, but none
of the twisted kind. And speaking of the twisted kind, I've always wondered
what kind of machine they used to produce it - the mind reels!)

By definition, regular flat must work 'OK', because DEC created these cards,
and specified the use of ordinary BC05L-xx cables, so whatever its number
is, it must be acceptable! :-)

    > That's a small mismatch, but I don't think it is going to cause big
    > problems.

BTW, is my understanding that the issue is the _junction_ of the two
different impedences, and not so much the length of the section with a
different impedance, correct? (The sound-based mental model I'm using is two
different diameters of pipe - going from a larger cross-section to a smaller
could produce echos - aka reflections - from the junction, but after that, it
should be OK.)


More information about the cctalk mailing list