25 Apr
2015
25 Apr
'15
4:58 a.m.
From: Josh Dersch
the power supply was completely dead. I've rebuilt
the supply
Just out of curiousity, what was the failure mode? (Also, trying to figure
out of the failure could have killed things...)
BTW, is it in a BA11-K box, or a BA11-L?
the AC LO and DC LO signals are high (these are active
low, right?)
Yes. All UNIBUS signals are active low _except_ for grants, if memory
serves...
The status register display is dark.
Hmmm.
The system does not respond to any keypresses on the
panel.
Sounds like you've got the variant with the real ("Programmer's") front
panel
- that's a plus. Although it's too bad that variant doesn't have a
"HALT/CONTINUE" switch - it would have been good to power it on with the HALT
on, which would have prevented it from trying to do any bus cycles.
No response to HALT/CRTL, though? Although I'm not sure if it would respond
to that if it was in a double-bus fault loop (NXM on trying to respond to a
NXM).
I've taken it down to a pretty minimal
configuration, sans any memory
or other stuff.
Always wise... :-)
Slot 3: (AB) M9301 (bootstrap terminator) / (CDEF)
M7859 (front panel board)
Most 11/34's run those cards in slot 4, not 3 (probably to leave room for the
FPP/Cache in slot 3), but slots 3 and 3 are (AFAIK) identical, so it should
be OK to run them there.
the NPG jumpers on CB1-CB2 are all present on the
backplane.
Err, I hope that's a typo for "CA1-CB1" - tying CB1 to CB2 (+15V) would not
be good (although if you did do it, it should be harmless in this
configuration).
I don't really have much to add to Don's message, about the next things to
try. There doesn't seem to be a way to configure the M9301 to have the
machine halt on power up, which would to my mind be the ideal. It might be
worth throwing a 'scope probe on MSYN, to see if it's trying to do bus cycles.
Noel
25 Apr
25 Apr
9:30 a.m.
On Apr 25, 2015, at 4:58 AM, Noel Chiappa <jnc at
mercury.lcs.mit.edu> wrote:
After replacing the capacitors, everything was working except the H745 (-5, +20V) which
immediately smoked a resistor in a very spectacular fashion (flames). This was due to a
shorted output transistor, and now things are OK.
From: Josh Dersch
the power supply was completely dead. I've
rebuilt the supply
Just out of curiousity, what was the failure mode? (Also, trying to figure
out of the failure could have killed things?)
BTW, is it in a BA11-K box, or a BA11-L?
It?s in a BA11-K.
the AC LO and DC LO signals are high (these are
active low, right?)
Yes. All UNIBUS signals are active low _except_ for grants, if memory
serves...
The status register display is dark.
Hmmm.
The system does not respond to any keypresses on
the panel.
Sounds like you've got the variant with the real ("Programmer's") front
panel
- that's a plus. Although it's too bad that variant doesn't have a
"HALT/CONTINUE" switch - it would have been good to power it on with the HALT
on, which would have prevented it from trying to do any bus cycles.
No response to HALT/CRTL, though? Although I'm not sure if it would respond
to that if it was in a double-bus fault loop (NXM on trying to respond to a
NXM).
No response at all to HALT/CTRL (or anything else). I kind of expected that I?d at least
be able to halt the processor and punch in numbers even if the CPU faulted, but I?ve never
dealt with an 11/34 before. The run light is on, the other lights are all halfway-lit
(dim). If I pull the bootstrap terminator and the unibus terminator at the other and and
fire it up, the run light is on but the other lights are completely off (and the panel is
still unresponsive).
I've taken it down to a pretty minimal
configuration, sans any memory
or other stuff.
Always wise... :-)
Slot 3: (AB) M9301 (bootstrap terminator) /
(CDEF) M7859 (front panel board)
Most 11/34's run those cards in slot 4, not 3 (probably to leave room for the
FPP/Cache in slot 3), but slots 3 and 3 are (AFAIK) identical, so it should
be OK to run them there.
the NPG jumpers on CB1-CB2 are all present on the
backplane.
Err, I hope that's a typo for "CA1-CB1" - tying CB1 to CB2 (+15V) would
not
be good (although if you did do it, it should be harmless in this
configuration).
I don't really have much to add to Don's message, about the next things to
try. There doesn't seem to be a way to configure the M9301 to have the
machine halt on power up, which would to my mind be the ideal. It might be
worth throwing a 'scope probe on MSYN, to see if it's trying to do bus cycles.
The machine came with an SLU and an 128KW memory board and with those installed (SLU in
slot 4, memory in slot 5, which is precisely how it was configured when I got the machine)
I get exactly the same behavior. Just in case I hooked up the SLU to a terminal and I get
nothing there either?
Thanks (and thanks to Don as well) for the tips?
- Josh
Noel
9:54 a.m.
My memory is not clear, so best to do is some manual reading.
For example, can you put the M9301 (or M9312) and M7859 in slot 3?
Or is slot 3 sort of reserved for M8267 or M8268, and should the
terminator and console interface go in slot 4?
No effects at all from the console is (IIRC) strange. CTRL/HALT should
work. I once had a dead M7859, as soon as I hit the numeric keys
strange things happened, and (for example) "LAD" turned the RUN
LED on and the system hung. The M7859 is "just another UNIBUS device".
On my interface board it was a dead RAM chip.
As yours is doing practically nothing, check whether all ICs are in
the sockets ... some vandal may have ripped the 8008 processor to sell
is as @*LOOK RARE *@
On my 11/34, a lamp (RUN, SR DISP, BUS ERR, MAINT) is either bright
ON or 'very' OFF. I never used MAINT, should do that some time ...
Grant problems can hang the bus (BUS ERR ON), but as you know,
you can check that by removing the M9302 at the far end. I think you
should keep the M9301 next to the CPU installed.
If you want the 11/34 to do something you probably *need* the
M7856 at 777560 as console. The CPU wants to dump it registers
there at startup ...
- Henk
1:09 p.m.
I only have time to glance at this now, but if you said -5, + 20 vdc, isn't
that a H754 ? that sounds like a core regulator. The H745 is for mos. Am I
not awake yet and I missed something?
Paul
On Sat, Apr 25, 2015 at 11:54 AM, Henk Gooijen <henk.gooijen at hotmail.com>
wrote:
My memory is not clear, so best to do is some manual
reading.
For example, can you put the M9301 (or M9312) and M7859 in slot 3?
Or is slot 3 sort of reserved for M8267 or M8268, and should the
terminator and console interface go in slot 4?
No effects at all from the console is (IIRC) strange. CTRL/HALT should
work. I once had a dead M7859, as soon as I hit the numeric keys
strange things happened, and (for example) "LAD" turned the RUN
LED on and the system hung. The M7859 is "just another UNIBUS device".
On my interface board it was a dead RAM chip.
As yours is doing practically nothing, check whether all ICs are in
the sockets ... some vandal may have ripped the 8008 processor to sell
is as @*LOOK RARE *@
On my 11/34, a lamp (RUN, SR DISP, BUS ERR, MAINT) is either bright
ON or 'very' OFF. I never used MAINT, should do that some time ...
Grant problems can hang the bus (BUS ERR ON), but as you know,
you can check that by removing the M9302 at the far end. I think you
should keep the M9301 next to the CPU installed.
If you want the 11/34 to do something you probably *need* the M7856 at
777560 as console. The CPU wants to dump it registers
there at startup ...
- Henk
7:34 p.m.
On 4/25/15 1:09 PM, Paul Anderson wrote:
I only have time to glance at this now, but if you
said -5, + 20 vdc, isn't
that a H754 ? that sounds like a core regulator. The H745 is for mos. Am I
not awake yet and I missed something?
There's both a 754 (-5V, +20V) and a 745
(+15V) supply in this thing.
When I acquired the machine (which had long since been decommissioned,
it was equipped with 128KW of MOS memory, I assume this is how it had
been used so I hope everything's kosher...
- Josh
Paul
On Sat, Apr 25, 2015 at 11:54 AM, Henk Gooijen <henk.gooijen at hotmail.com>
wrote:
> My memory is not clear, so best to do is some manual reading.
> For example, can you put the M9301 (or M9312) and M7859 in slot 3?
> Or is slot 3 sort of reserved for M8267 or M8268, and should the
> terminator and console interface go in slot 4?
>
> No effects at all from the console is (IIRC) strange. CTRL/HALT should
> work. I once had a dead M7859, as soon as I hit the numeric keys
> strange things happened, and (for example) "LAD" turned the RUN
> LED on and the system hung. The M7859 is "just another UNIBUS device".
> On my interface board it was a dead RAM chip.
> As yours is doing practically nothing, check whether all ICs are in
> the sockets ... some vandal may have ripped the 8008 processor to sell
> is as @*LOOK RARE *@
>
> On my 11/34, a lamp (RUN, SR DISP, BUS ERR, MAINT) is either bright
> ON or 'very' OFF. I never used MAINT, should do that some time ...
>
> Grant problems can hang the bus (BUS ERR ON), but as you know,
> you can check that by removing the M9302 at the far end. I think you
> should keep the M9301 next to the CPU installed.
>
> If you want the 11/34 to do something you probably *need* the M7856 at
> 777560 as console. The CPU wants to dump it registers
> there at startup ...
>
> - Henk
>
>
26 Apr
26 Apr
12:32 a.m.
On 4/25/2015 7:34 PM, Josh Dersch wrote:
On 4/25/15 1:09 PM, Paul Anderson wrote:
The 754 -5V/+20V regulator is only used for the core memory subsystem (like an
MM11-D board set) for the 11/04 or 11/34.
You can leave it or pull it, it won't matter in your configuration.
I only have time to glance at this now, but if
you said -5, + 20 vdc, isn't
that a H754 ? that sounds like a core regulator. The H745 is for mos. Am I
not awake yet and I missed something?
There's both a 754 (-5V, +20V) and a 745
(+15V) supply in this thing. When I
acquired the machine (which had long since been decommissioned, it was
equipped with 128KW of MOS memory, I assume this is how it had been used so I
hope everything's kosher...
25 Apr
25 Apr
7:39 p.m.
On 4/25/15 9:54 AM, Henk Gooijen wrote:
My memory is not clear, so best to do is some manual
reading.
For example, can you put the M9301 (or M9312) and M7859 in slot 3?
Or is slot 3 sort of reserved for M8267 or M8268, and should the
terminator and console interface go in slot 4?
I assume it's OK to put the
M9301 in slot 3 only because there's a
DEC-applied label running along the edge of the cardcage labeling what
goes in what slots, and slot 3 is labeled "M9301/SPC." It's also where
the card was situated when the machine entered my possession (not that
that means much :)).
No effects at all from the console is (IIRC) strange. CTRL/HALT should
work. I once had a dead M7859, as soon as I hit the numeric keys
strange things happened, and (for example) "LAD" turned the RUN
LED on and the system hung. The M7859 is "just another UNIBUS device".
On my interface board it was a dead RAM chip.
As yours is doing practically nothing, check whether all ICs are in
the sockets ... some vandal may have ripped the 8008 processor to sell
is as @*LOOK RARE *@
The 8008 is still there :). Nothing else is socketed.
I'll note that
running the system with just the console card and RAM installed (no
CPU), the behavior is identical. I'm beginning to think that there's
something wrong with the console, but I'll need to do some more
debugging tonight.
On my 11/34, a lamp (RUN, SR DISP, BUS ERR, MAINT) is either bright
ON or 'very' OFF. I never used MAINT, should do that some time ...
Thanks -- good to know that what I'm seeing is atypical.
Grant problems can hang the bus (BUS ERR ON), but as you know,
you can check that by removing the M9302 at the far end. I think you
should keep the M9301 next to the CPU installed.
If you want the 11/34 to do something you probably *need* the M7856 at
777560 as console. The CPU wants to dump it registers
there at startup ...
Thanks, that's good to know -- I hadn't considered the
requirements of
the bootstrap ROMs w.r.t. the SLU.
- Josh
- Henk
3527
days inactive
3528
days old
6 comments
5 participants
participants (5)
-
derschjo＠gmail.com -
henk.gooijen＠hotmail.com -
jnc＠mercury.lcs.mit.edu -
north＠alum.mit.edu -
useddec＠gmail.com