PDP 11 Line Time Clock (LW11L) M787 and EIS board, (KE11E) M7238
jnc at mercury.lcs.mit.edu
Mon Feb 13 12:22:46 CST 2017
> From: William Degnan
> PDP 11 KE11E M7238 EIS board (for PDP 11/40) which causes the CPU to
> crash when installed; front panel not responsive
> I installed a removable jumper so I can flip jumper configs back and
> forth between EIS installed/not installed. Without the EIS the system
> works fine
To understand this symptom, one needs to understand how the EIS interacts
with the main CPU. Both include microcode, and what is supposed to happen is
that when an EIS instruction happens, control is passed to the microcode on
the EIS board (the actual microcode words being fed back to the main CPU
through those three over-the-back jumper cables). The microcode on the EIS
board can then control the data paths, etc in the main CPU, to feed the EIS
data, and take back the results of the computation performed on the EIS card.
I'm trying to understand what W1 does, but I'm not there yet. It's shown on
the KD11-A print K3-8 (pg. 48), in the lower left corner, but its effects are
To start with, the array of odd chips E6-E7 (74H60's) and E17 (74H53) are
expandable AND-OR gates. I'd never seen these before, but the lines running
to and from pins 11 and 12 on the 'H53 join the other three gates below it
into it - i.e. that whole array of AND gates all feed into one NOR gate
(output on pin 8 of the 'H53).
So far, so good, but from there I'm still lost. When W1 is inserted (no EIS)
it grounds the signal ECIN00, which comes in from off-board (as shown by the
"A05S2", which is the pin it arrives on). The output of that giant NOR gate
is CIN00, which is immediately sent off-board (pin 'A05P1'). I have yet to
try and chase these signals down, and work out what they do; the KD11-A Tech
Manual is fairly cryptic on the subject.
Note also that, IIRC, the front console operates under control of microcode.
So I'm _guessing_ that what is happening is that somehow the EIS is, when
enabled, messing up the operation of the microcode in the main CPU, causing
it to freeze.
> Thought - I don't have a LW11L, M787 installed. ... Do you think that
> maybe the EIS board requires this for some reason
No. I've looked at the KW11-L prints in the past, and it's just a very simple
UNIBUS device. I don't see any way it not being there could cause the
symptoms under discussion.
> I could swap out the current 11/40 backplane with a backplane that has
> the jumpers for the M787 already removed
I think it's only one jumper - for BG6, no?
> I neglected to mention I had no M787
To run Unix V6 you'll need either a KW11-L or KW11-P (see previous
discussion about how Unix needs a clock - both at a low level, because
it will panic() if it doesn't find one, and at a high level, because
even if we patch the panic, stuff won't 'work right' without one).
More information about the cctalk