6 Apr
2013
6 Apr
'13
10:59 a.m.
On Sat, Apr 6, 2013 at 12:34 PM, Arno Kletzander <Arno_1983 at gmx.de> wrote:
The only existing application I know of (besides PMI) is the RL11
board set. There were probably quite a few other board sets that used
the CD lines for inter-board communication; the slots are routed
pin-to-pin for the most part, but slot-to-slot instead of bussed
straight through, so as long as you have the cards in the right order,
they won't be talking to other ones in the backplane. For example,
the MSV11-J boards (PMI memory) have the top pins connected straight
through to the bottom on the card, but the KDJ11-B only has the top
fingers populated, so the PMI bus terminates at the card. You could
put an RL11 board set directly below it with no negative consequences,
for example.
I'd thought so; perhaps I'll happen across some esoteric cards that
use them, or I'll hack something custom in there.
Pretty much. There are plenty of quad-height boards that use the CD
slots for power but nothing else.
I wouldn't swear to the fact that there weren't 11/23s sold as 18-bit
systems; my fuzzy memory says that 11/23 might have been 18-bit while
11/23+ would have been 22-bit. Other people doubtless know better
than me; my other recollection is that the "plus" distinction has to
do with the KDF11-B board (which has the boot ROM and SLUs).
- Dave
David Riley <fraveydank at gmail.com> wrote:
Well, you *could* work around it that way. I guess Kapton would
probably be a good bet. You're very unlikely to be using early LSI-11
boards unless you're a true masochist, IMO, but I wouldn't call it
impossible. The issue is that the very early boards used those as
test lines, which has the tendency to cause problems with other 22-bit
devices. Be aware that if you're using the system as 18-bit, you'll
need to make sure that your peripheral cards are jumpered to 18-bit if
necessary (boards that use DMA often need to know).
The 22-bit conversion is also easy to undo, assuming you're handy with
a soldering iron; you're just bussing 4 lines straight down the
backplane. For me, undoing that would be about as easy as looking up
which pins to cover and then covering them with tape, though I guess
it would involve removing the backplane from the chassis. It's a
shame that the H9273 doesn't have wire-wrap stakes coming out the back
like the H9270 does, because that makes the conversion much easier to
perform and to reverse.
The H9273 is perfectly compatible with an 11/23
CPU; there
is a jumper you can set for 18-bit use. Obviously, though,
turning it into a 22-bit backplane is preferable.
Fortunately, that's pretty easy.
Meh...I've been hearing that I might run into problems with some other processors
(LSI-11(/2), very early KDF11-A versions), should I ever need to use the backplane as a
test platform for one of those in the future. Is that problem the one simply worked around
with some insulating tape over the BDAL18:21 connector pads? > The M8189
is a quad-width board, but with the CD part configured
> for a serpentine backplane (the manual mentions two jumpers to be removed when using
in another
> environment to avoid shorting signals together), so again no "meaningful"
use of the CD lanes.
Correct. The Qbus PDP-11s only used the CD lane
for PMI
(or custom logic, if you had that).
Sounds interesting, could you elaborate?
What are these intended for anyway (I was thinking of
some sort of PMI predecessor, probably
wrong by now), and what boards do I need to take advantage of them in this machine?
For an 11/23, those lanes will just go unused. It's not
so bad, really; you have to try pretty hard to fill up 9
slots. If it's
in
a box labelled 11/23, though, check to make sure someone
hasn't already performed the modification.
OK, thanks!