19 Apr
2011
19 Apr
'11
1:45 p.m.
Well, a few more bizarre things to report:
I tried to disable interrupt sources as suggested. The only one, =
apparently, was the DL11-W SLU/LTC card which was set for both functions. =
The card was reconfigured to be "SLU Only". No change in behaviour. The =
DL11-W console is set for address 177560, vector 60.
The console will geneate 2 iterrupt vectors (60 and 64 I think) for
transmit and receive. They cannont be set separately, so if the DL11-W is
working properly, that's OK.
The RX11 will also generate interrupts. What's it set to?
Next, there are three memory cards: one M7847-DJ (16K*18 MOS, MS11-JP), =
one "Motorola memory systems" card with 48K words, and one "MOSTEK memory
=
systems" card with 96K words. Both the two later cards use 4116 DRAM =
chips.
Hang on a second. You've got 16Kw + 48Kw + 96kw there. That's 160Kw
total. The PDP11/34 (it is an 11/34 IRIC) has 18 bit addressing and can
access a total of 128kw. 4Kw are used for the I/O space, so you have have
a maximum of 124kw of memory (practically, most memory boards wil
lautomatically disable themselves in the I/O space, so you could fit a
pair of 64Kw boards, say,m without problems). But with your configuration
you must have some locations addressing 2 boards, which is not a good thing.
Tried to use the M7847-DJ card alone (after all, RT-11SJ should run on =
16KW of memory). Reconfigured the card address to start at 0 and verified =
that it occupies addresses 000000 to 077777 via the front panel. When the =
machine it turned on, the display reads "054207" (this was not expected). =
Tried booting using the bootstrap on the M9301-YB card: the heads don't =
engage at all and the program halts rapidly displaying "173764" ... =
BIZARRE since that is a non-existent address.
Is it? Remeebr the address show is the 16 bit program counter address. If
the MMU is disabled (which it will be on power-up, and nothing is going
to enable it), addresses starting 16 or 17 are mapped to the I/O space at
76 or 77. So that address is 773764. Which I think is in the bootstrap
ROM.
Next, tried the Motorola memory card which was already configured to start =
at 0. When the CPU is powered-on, displays "000002" on the display. =
Tried to boot the disk, heads engage, a few steps of the disk heads, and =
halts again at "005134".
Finally, tried the MOSTEK memory card. Reconfigured to start at 0, and =
verified that addresses 000000 through 377777 are occupied by RAM. Powers =
up with a display of 000002 and halts at 005134 when a boot is attempted.
The system is now stripped during testing, consisting of the following:
Slot 1/2 =3D CPU, Slot 3=3DM7859 Panel interface and M9301-YB bootstrap/ter=
minator, Slot 4=3Dmemory, slot 5&6&7 empty (with a grant continuity card =
in D), slot 8=3DM7856 SLU/LTC, slot 9=3DM9302 at the unibus end and M7846 =
RX-01 interface.
That sounds OK.
As I was typing this, had a thought about slots: I
recall reading =
something about not putting memory into slot 9. So, I moved the RX-01 =
Yes, The reason is that memroy boards are MUD (Modified Unibus Device)
boards and take their signals from connectors A,B (this matters on
machines like the 11/44 which have a 22 bit address bus, and where the
addres lines on A,B are not the same as those on E). Device controllers
(SPCs -- Small Peripheral Controller) take their signals from C-F. On
slot 9, the A/B conenctors are the Unibus output to the next backplane
(or for a terminaotr) and are not qurie the same signals as the ones on
an MUD slot. In fact I think on some machines you can short out a power
rail by putting an MUD card in the Unibus Out slot or a termintor in an
MUD slot.
controller to slot 6 (and moved the grant card in slot
9D). Now, when =
powered-up with the Motorola or Mostek memory cards, displays "000000" but =
otherwise no changes in booting behaviour. When the MS11-JP RAM was =
installed instead, displays "177777" once, "163776" the next time
(when =
powered up), and still absolutely nothing when boot is attempted (i.e. no =
heads engaging at all).
Is there something about RT-11 5.04 that I am not aware-of? An it was =
suggested that there is a difference between the Unibus and Q-Bus boot ... =
any thoughts on that one? =20
I've never heard of that being a problem...
I can't see memory being the issue anyway: sure, 16KW might be an issue =
(and that card might even be 'flaky'), but the other two cards are large =
enough.
Silly question... Whenever I hear of a PDP11 wit flaky memeory I think of
the time I was led a merry dance by my 11/45. It turns out the problem
ther was power supply related (one ofthe 5V lines was sitting at 4.4V).
Have tyoy checked all the power supply voltages with the machine in
operation?
-tony