14 Jul
2012
14 Jul
'12
2:14 a.m.
I have the power supply tested and reassembled and am ready to power
up the 11/04.
I am looking to limit the number of installed boards initially.
Certainly the RK05 controller boards can go as I haven't got the
drives operational yet.
The configuration of the 11/04 as I received it is as follows
(hopefully this is readable):
M7257 M7257 M9302
M7256 M7256 -
M7255 M7255 -
M7254 M7254 M920
M7258 M7258 M920
M7856 M7856 -
M7860 M7860 -
GRANT M9202
GRANT M9202
M7856 M7856 M7850
GRANT
M7847 M7847 M7847
GRANT
M7847 M7847 M7847
M7859 M7859 M9301
GRANT
M7263 M7263 M7263
Looking for some guidance if the following is a valid configuration:
GRANT M9302
M7856 M7856 M7850
GRANT
M7847 M7847 M7847
GRANT
M7847 M7847 M7847
M7859 M7859 M9301
GRANT
M7263 M7263 M7263
I am also looking to confirm my understanding of the M9301 card. I
have read the maintenance and operators manual for the M9301. I think
it is saying that the Console Emulator startup message from the M9301
and the console emulator commands will be available on the terminal...
which I take to mean the terminal connected to the M7856 in this
machine. Am I understanding that right.
Thanks everyone for your help.
Regards
Andrew
15 Jul
15 Jul
1:51 p.m.
I am also looking to confirm my understanding of the
M9301 card. I
Basiclaly, the M9301 contains a small amount of ROM containing a PDP11
machine code program to provide the ocnsole emulator and device
bootstraps. The PDP11/04 CPU on its own will not talk to a terminal at
power-up, it'll just execute machine instructions. Of course, if said
intructions are a program that acesses the terminal port then the system
talks to the terminal. And that's what happens here.
have read the maintenance and operators manual for the
M9301. I think
it is saying that the Console Emulator startup message from the M9301
and the console emulator commands will be available on the terminal...
which I take to mean the terminal connected to the M7856 in this
machine. Am I understanding that right.
It'll display the output and take the conoslue emulator commands from a
terminal (conented to a DL11-something-or-other card) at the standard
console terminal address, which IIRC is octal 777560 yo 777566. Note that
hardware addresses are 18 bits lont (even on a machine without an MMU)
and that word addresses are always even. The DL11 port takes up 4 word
addresss, for thee receive control/status register (CSR, it's a term
you'll come across time and again), the receive data register, the
transmit CSR and the transmit data register.
Progrm adresses, are, of course, always 16 bits long. On a machine with
out an MMU (like the 11/04), program addresses 000000 to 157777 are left
unchanged (The extra 2 bits for the 18 bit address are set to 0), while
160000 to 177777 are converted to 760000 to 777777 The latter range of
addresses (4K words) are gernally used for I/O devices, not memory.
Now, the M7856 is certainyl a possible terminal itnerfce card for this,
but the address can be set by DIP swithces on said board, so it's not
certain that' sthe right one. It wuld be perverse in the extreme _not_ to
have a DL11 at the normal conosle terminal address, but if you get noting
on the connected terminal, checkign the address would not be a bad idea
-tony
6:21 p.m.
Thanks Tony. I did some more research. The info on
http://www.pdp-11.nl/peripherals/comm/interface/dl11w-info.html was
particularly useful and I have determined which of the two M7856
boards in the machine is configured as the system console.
I need to check the cable wiring to ensure it is configured for EIA
serial IO (the attached cable has a molex connector rather than a DB25
or DB9) and I will give it a try.
I did power up the machine with a minimum set of boards installed.
The KY11-LB front panel came up, address 000000, Run and DC OK leds
on. Some of the front panel button work. I can load an address into
the address register and read it back but the buttons on the right
side are very stiff and do nothing or provide the wrong digit. Will
need to pull the front panel off and have a look at the keyboard but
it certainly seems that the controller card for the front panel is
working.
Before I pull the front panel apart I will get a terminal connected to
the system console port and see if the console emulator comes up.
That should tell me if the CPU is running.
Regards
Andrew
On Mon, Jul 16, 2012 at 8:51 AM, Tony Duell <ard at p850ug1.demon.co.uk> wrote:
I am also
looking to confirm my understanding of the M9301 card. I
Basiclaly, the M9301 contains a small amount of ROM containing a PDP11
machine code program to provide the ocnsole emulator and device
bootstraps. The PDP11/04 CPU on its own will not talk to a terminal at
power-up, it'll just execute machine instructions. Of course, if said
intructions are a program that acesses the terminal port then the system
talks to the terminal. And that's what happens here.
have read the maintenance and operators manual
for the M9301. I think
it is saying that the Console Emulator startup message from the M9301
and the console emulator commands will be available on the terminal...
which I take to mean the terminal connected to the M7856 in this
machine. Am I understanding that right.
It'll display the output and take the conoslue emulator commands from a
terminal (conented to a DL11-something-or-other card) at the standard
console terminal address, which IIRC is octal 777560 yo 777566. Note that
hardware addresses are 18 bits lont (even on a machine without an MMU)
and that word addresses are always even. The DL11 port takes up 4 word
addresss, for thee receive control/status register (CSR, it's a term
you'll come across time and again), the receive data register, the
transmit CSR and the transmit data register.
Progrm adresses, are, of course, always 16 bits long. On a machine with
out an MMU (like the 11/04), program addresses 000000 to 157777 are left
unchanged (The extra 2 bits for the 18 bit address are set to 0), while
160000 to 177777 are converted to 760000 to 777777 The latter range of
addresses (4K words) are gernally used for I/O devices, not memory.
Now, the M7856 is certainyl a possible terminal itnerfce card for this,
but the address can be set by DIP swithces on said board, so it's not
certain that' sthe right one. It wuld be perverse in the extreme _not_ to
have a DL11 at the normal conosle terminal address, but if you get noting
on the connected terminal, checkign the address would not be a bad idea
-tony
16 Jul
16 Jul
11:57 a.m.
I need to check the cable wiring to ensure it is
configured for EIA
serial IO (the attached cable has a molex connector rather than a DB25
or DB9) and I will give it a try.
If the 'Molex' connector is actually a flat white thing with 8 possible
contacts in a row, then it's actually an AMP connecotr ;-). It's also
almost certainly current loop, since that's what DEC used for corrent
loop conenctors.
AFAIK, most, if not all, DL11-Ws had the RS232 driver/receiver chips
fitted, if not, it's easy to add them (they're 1488s and 1489s, easy to
get). You then need to make up the right cable. Note that there's a
jumper in the cable between 2 pins at the RL11 end, it connects the
output of the appropriate receiver (RS232 or current loop) to the input
of the UART. Apart form that, there are no gotchas.
I did power up the machine with a minimum set of boards installed.
The KY11-LB front panel came up, address 000000, Run and DC OK leds
on. Some of the front panel button work. I can load an address into
Can you haldt the CPU (IIRC that's ctrl-halt on the keypad panel),
turning the RUN lamp off? If not, it's likely you have an open grant
chain somewhere.
-tony
4:33 p.m.
If the 'Molex' connector is actually a flat
white thing with 8 possible
This one is 2 rows of 3 pins. The Berg connector is wired for EIA as
described on pdp11.nl and the DL11-W's have the driver chips so all
good there.
Can you haldt the CPU (IIRC that's ctrl-halt on
the keypad panel),
You are right with ctrl-halt but I get the Bus Error LED rather.
My current configuration is as follows (note that the Grant cards are
all in Row D).
GRANT M9302
GRANT
GRANT
M7847 M7847 M7847
GRANT
M7847 M7847 M7847
M7859 M7859 M9301
GRANT
M7263 M7263 M7263
When I first received the machine the configuration for this 9
connector section was:
GRANT M9202
M7856 M7856 M7850
GRANT
M7847 M7847 M7847
GRANT
M7847 M7847 M7847
M7859 M7859 M9301
GRANT
M7263 M7263 M7263
so the only difference is the removal of the M7856 and replacement
with a Grant card. I will reinstate the M7856 and see it if makes any
difference.
Regards
Andrew
17 Jul
17 Jul
11:54 a.m.
If the 'Molex' connector is actually a
flat white thing with 8 possible
This one is 2 rows of 3 pins. The Berg connector is wired for EIA as described on pdp11.nl and the DL11-W's have the
driver chips so all
good there.
OK, you have 2 choices I guess. Try to find a mating plug for that
conenctor, or chop it off and conenct the wires to soemthing else. In
either case you need to trace where the wires go to, all you ned are TxD,
RxD and Ground.
Can you haldt the CPU (IIRC that's ctrl-halt
on the keypad panel),
You are right with ctrl-halt but I get the Bus Error LED rather.
My current configuration is as follows (note that the Grant cards are
all in Row D).
GRANT M9302
GRANT
GRANT
M7847 M7847 M7847
GRANT
M7847 M7847 M7847
M7859 M7859 M9301
GRANT
M7263 M7263 M7263
When I first received the machine the configuration for this 9
connector section was:
GRANT M9202
M7856 M7856 M7850
GRANT
M7847 M7847 M7847
GRANT
M7847 M7847 M7847
M7859 M7859 M9301
GRANT
M7263 M7263 M7263
so the only difference is the removal of the M7856 and replacement
with a Grant card. I will reinstate the M7856 and see it if makes any
difference.
Are yuo using the single-height grant boards (little square things) or
the dual-height ones (with handles)? If the former, check you've got them
the right way round, the traces towards the M9302 end of the backplane.
What I normally do is use a logic proble to see which grant signal is
high (yes, grants are active-high, unlike most Unibus signals) at the
M9302. Then trace that particular signal trhough all the boards. Most fo
the time it's the NPG signal because the jumper is not fitted o nthe
backplaen and I'm usin a board that doesn't use or pass through the NPG
signal.
-tony
6:35 p.m.
OK, you have 2 choices I guess. Try to find a mating
plug for that
I checked the wiring with a meter and identified the TXD, RXD and GND
wires and have just connected them to a DB9. I believe this machine
was last used as part of a medical device which may explain the
interesting connector choices.
Are yuo using the single-height grant boards (little
square things) or
the dual-height ones (with handles)? If the former, check you've got them
the right way round, the traces towards the M9302 end of the backplane.
Yes it certainly looks like a grant issue. I have the little square
boards and they are all correctly inserted.
Time to get out the logic probe!
11:43 p.m.
snip snip
You might suffer from an open NPR/NPG line on one of the open slots.
This is cured by using dthe double width cards (green handles).
The short cards only loop the bus grant lines, hence the 4 loops, but
not the NPR/NPG line.
Some non DMA cards loop this line on the board itself, others don't.
So, you may have to place a small wire on the open NPR/NPG grant line
(pin CA1 & CB1) or find a few G7273 cards.
Ed
--
Dit is een HTML vrije email / This is an HTML free email.
Zeg NEE tegen de 'slimme' meter.
Are yuo using
the single-height grant boards (little square things) or
the dual-height ones (with handles)? If the former, check you've got
them
the right way round, the traces towards the M9302 end of the backplane.
Yes it certainly looks like a grant issue. I have the little square
boards and they are all correctly inserted.
18 Jul
18 Jul
1:49 a.m.
Thanks for the feedback. One of the slots where I added a short grant
card previously had an M7856 board. I will have a look at this board
in the morning. I understand it doesn't connect the NPG line so it is
possible if this is not connected on the backplane that board was
wired to connect it.
On Wed, Jul 18, 2012 at 6:43 PM, E. Groenenberg <quapla at xs4all.nl> wrote:
snip snip
You might suffer from an open NPR/NPG line on one of the open slots.
This is cured by using dthe double width cards (green handles).
The short cards only loop the bus grant lines, hence the 4 loops, but
not the NPR/NPG line.
Some non DMA cards loop this line on the board itself, others don't.
So, you may have to place a small wire on the open NPR/NPG grant line
(pin CA1 & CB1) or find a few G7273 cards.
Ed
--
Dit is een HTML vrije email / This is an HTML free email.
Zeg NEE tegen de 'slimme' meter.
Are yuo
using the single-height grant boards (little square things) or
the dual-height ones (with handles)? If the former, check you've got
them
the right way round, the traces towards the M9302 end of the backplane.
Yes it certainly looks like a grant issue. I have the little square
boards and they are all correctly inserted.
1:56 p.m.
OK, you have 2 choices I guess. Try to find a
mating plug for that
I checked the wiring with a meter and identified the TXD, RXD and GND
wires and have just connected them to a DB9. I believe this machine was last used as part of a medical device which may
explain the
interesting connector choices.
Then it's most likely a missing NPG jumper. This is a wire-wrapped jumper
on the backplane between pins CA1 and CB1 of a SPC slot. It has to be cut
if there's a NPR device (one that does DMA, like an RL11 or RX211) in
that slot, it has to be fitted if the slot is empty (other than the grand
contiuity card in connector D), or if there's a device there that maks no
connection to those pins. And finallym, it doens't matter whether it's
there or not if there's a device there that simply shorts the pins
together (like a DZ11)
If you find the misisng jumper you have sevral choices. One is to pull
the backplane and fit a wire-wrapped link between the pins. I normally
use red or orange wire for this so it's easy to spot if I want to remvoe
it later. If it's an empty slot, yuo could use one of the later
dual-height continuity cards if you have one. Or cut a sliver or
single-sided copper clad board ot slid into that conenctor (copper side
towards the CPU) and just short those pins togehter.
If it's a slot with a card in it that maks no connection to pins CA1 and
CB1, then you could solder a link between those edge fingers on the card.
-tony
Are yuo using the single-height grant boards
(little square things) or
the dual-height ones (with handles)? If the former, check you've got them
the right way round, the traces towards the M9302 end of the backplane.
Yes it certainly looks like a grant issue. I have the little square
boards and they are all correctly inserted.
21 Jul
21 Jul
12:44 a.m.
Spent a few hours today with the unibus documentation, meter and scope
and have concluded there are no breaks in the NPG / NPR / BG4-7 or
BR4-7 lines on the 9 socket backplane.
It is possible that some of the HALT / Bus Error problem might be with
the KY11-LB.... some of the keys on the keypad don't return the
correct values on the display when pressed. Specifically 3 and 6
which cause the display to show 1 and 4. These happen to be on the
same READ cycle as the CTRL and DEP buttons. My initial thought was a
problem with the 7404 that controls the READ signal but that looks OK
on the scope. I will pull the keypad off tomorrow and make sure it
isn't something in the switch matrix.
Not sure if anyone else has seen similar issues with the KY11?
I need to get back to the M7856 serial cable and replace the Molex
with a DE9 (well spotted Tony... I never realized that DB9 was not the
right name). It may be that the machine is powering up fine and the
console emulator works even if the KY11-LB is misbehaving. Will try
that tomorrow.
Tony - to answer your question.... the 6 pin molex had 2 ground pins,
RXD, TXD. Also RTS and DTS are wired back to the Berg connector. I
thought I read somewhere that these are not used.
On a positive note... all the power supply voltages on the backplane
looks good after the rework on those.
22 Jul
22 Jul
2:23 a.m.
From: "Andrew Quinn" <jaquinn2001 at gmail.com>
Sent: Saturday, July 21, 2012 9:44 AM
To: "General Discussion: On-Topic Posts Only" <cctech at classiccmp.org>
Subject: Re: Ready to Power up the 11/04
It is possible that some of the HALT / Bus Error
problem might be with
the KY11-LB.... some of the keys on the keypad don't return the
correct values on the display when pressed. Specifically 3 and 6
which cause the display to show 1 and 4. These happen to be on the
same READ cycle as the CTRL and DEP buttons. My initial thought was a
problem with the 7404 that controls the READ signal but that looks OK
on the scope. I will pull the keypad off tomorrow and make sure it
isn't something in the switch matrix.
Not sure if anyone else has seen similar issues with the KY11?
The KY11-LB is "just another" peripheral on the UNIBUS. So if the KY11-LB
creates a problem on the UNIBUS the rest of the system will have problems
too. I once had a KY11-LB that misbehaved. What the symptoms were, I
can not remember, maybe that's described on my website.
Anyway, it turned out to be a faulty RAM chip. After I had found the cause,
I have replaced (sorry Tony, swapped) the module, because at the time I
could not find a replacement 4 bit, 16 nibble RAM chip. I am no longer
searching, but one day I hope to replace the RAM chip, if only to prove that
I did find the fault ...
- Henk, PA8PDP
4:11 p.m.
Hi Henk,
Thanks for the info. I also appreciate the detail on your webpage
about the KY11-LB.
I think I have got to the bottom of the front panel problem.
Initially I thought that it was related to the scanning technique that
is used to read the keypad and update the 7 segment displays given the
keypad gave the wrong values and sometimes the display was missing
segments.
After further research it appears not to be the case (which is good).
The scanning signals all look good on the scope and the keypad matrix
works correctly when tested with a meter.
The insight came when I looked at which keys on the number pad worked
and which didn't. I came up with the following table
Pressed = Displayed
1 = 1
2 = 0
3 = 1
4 = 4
5 = 5
6 = 4
7 = 5
0 = 0
If you convert this to binary it becomes pretty obvious where the
problem lies. The invalid display is for the items marked with * and
all have the second bit set. If you mask out the second bit you get
the displayed value.
0 = 0000
1 = 0001
2 = 0010 *
3 = 0011 *
4 = 0100
5 = 0101
6 = 0110 *
7 = 0111 *
The front panel uses a 7447 BCD to 7 Segment decoder to drive the LED
modules. If pin 1 (B) on this is not being set when required then the
behaviour will be what I am seeing..... it isn't that the keypad is
getting it wrong... but the display is showing the wrong value due to
a missing bit.
Need to do a bit more digging here but given that this line is pulled
to +5V via a resistor on the front panel board I am hoping that the
problem is either
a connection problem on the front panel board with the line not being
pulled up or the 7447 needs to be replaced. If there was a break in
the ribbon table to the controller board then the bit would always be
set and a different set of incorrect values would be displayed.
Hopefully there isn't a fault on the controller board that is pulling
the pin to ground.
Apart from incorrect values, other behaviours of the front panel seem OK.
The other display problems with odd segments missing on the 7
segment displays are due to them being socketed (clearly there was
some concern about reliability in 1979) with dirty connections. I
need to pull each 7 segment display and clean the pins. Pressing on a
display that is missing segments will cause them to appear.... might
need to solder joins touched up too.
So that is all very positive about the front panel. If I can get down
to the workshop
tonight I will try to validate if it is the 7447 that is at fault.
Tony - OK on testing the Grant lines. I will do that. It seems that
if the CPU was
already halted that a similar BUS Error would result. Apart from
checking for address
bus activity is there any other way of confirming if the CPU is
running or already halted?
I am going to recheck the DC LO and AC LO signals on the unibus as
well using a scope.
The "Hints on testing a dead PDP11" at has a comment that these can
cause problems
in the BA11-K chassis and cause the processor to halt.
Regards
Andrew
On Sun, Jul 22, 2012 at 9:23 PM, Henk Gooijen <henk.gooijen at hotmail.com> wrote:
From: "Andrew Quinn" <jaquinn2001 at
gmail.com>
Sent: Saturday, July 21, 2012 9:44 AM
To: "General Discussion: On-Topic Posts Only" <cctech at classiccmp.org>
Subject: Re: Ready to Power up the 11/04
It is possible that some of the HALT / Bus Error
problem might be with
the KY11-LB.... some of the keys on the keypad don't return the
correct values on the display when pressed. Specifically 3 and 6
which cause the display to show 1 and 4. These happen to be on the
same READ cycle as the CTRL and DEP buttons. My initial thought was a
problem with the 7404 that controls the READ signal but that looks OK
on the scope. I will pull the keypad off tomorrow and make sure it
isn't something in the switch matrix.
Not sure if anyone else has seen similar issues with the KY11?
The KY11-LB is "just another" peripheral on the UNIBUS. So if the KY11-LB
creates a problem on the UNIBUS the rest of the system will have problems
too. I once had a KY11-LB that misbehaved. What the symptoms were, I
can not remember, maybe that's described on my website.
Anyway, it turned out to be a faulty RAM chip. After I had found the cause,
I have replaced (sorry Tony, swapped) the module, because at the time I
could not find a replacement 4 bit, 16 nibble RAM chip. I am no longer
searching, but one day I hope to replace the RAM chip, if only to prove that
I did find the fault ...
- Henk, PA8PDP 
11:46 p.m.
On Mon, Jul 23, 2012 at 1:11 AM, Andrew Quinn <jaquinn2001 at gmail.com> wrote:
The insight came when I looked at which keys on the
number pad worked
and which didn't. I came up with the following table
Pressed = Displayed
1 = 1
2 = 0
3 = 1
4 = 4
5 = 5
6 = 4
7 = 5
0 = 0
If you convert this to binary it becomes pretty obvious where the
problem lies. The invalid display is for the items marked with * and
all have the second bit set. If you mask out the second bit you get
the displayed value.
0 = 0000
1 = 0001
2 = 0010 *
3 = 0011 *
4 = 0100
5 = 0101
6 = 0110 *
7 = 0111 *
The front panel uses a 7447 BCD to 7 Segment decoder to drive the LED
modules. If pin 1 (B) on this is not being set when required then the
behaviour will be what I am seeing..... it isn't that the keypad is
getting it wrong... but the display is showing the wrong value due to
a missing bit.
Need to do a bit more digging here but given that this line is pulled
to +5V via a resistor on the front panel board I am hoping that the
problem is either
a connection problem on the front panel board with the line not being
pulled up or the 7447 needs to be replaced. If there was a break in
the ribbon table to the controller board then the bit would always be
set and a different set of incorrect values would be displayed.
Hopefully there isn't a fault on the controller board that is pulling
the pin to ground.
It's likely there's a driver chip on the controller board with an open
collector output (something like a 7438). That output could very well
be shorted to ground. That would in fact be my first suspect for this
error.
Camiel.
23 Jul
23 Jul
3:26 p.m.
Right. These lines are driven by a 7417's on the controller board.
DEC seemed to like non-inverting buffers. The resistors on the front
panel board hold the line high and the 7417 needs to pull this to
ground to get the display segments to light up. Checked with the
scope on the front panel board and the second bit of the incoming BCD
value isn't going low. Pulling the line to ground on the front panel
board causes the 7 segment displays to read correctly so the BCD to 7
segment converter is OK. The cable to the controller board checks out
OK so the next step is to check the input and output on the 7417 gate
and see if the problem is there. Unfortunately I don't have extender
boards so will have to go with the old "solder wires to the chip"
trick.
On Mon, Jul 23, 2012 at 6:46 PM, Camiel Vanderhoeven
<iamcamiel at gmail.com> wrote:
On Mon, Jul 23, 2012 at 1:11 AM, Andrew Quinn
<jaquinn2001 at gmail.com> wrote:
The insight came when I looked at which keys on
the number pad worked
and which didn't. I came up with the following table
Pressed = Displayed
1 = 1
2 = 0
3 = 1
4 = 4
5 = 5
6 = 4
7 = 5
0 = 0
If you convert this to binary it becomes pretty obvious where the
problem lies. The invalid display is for the items marked with * and
all have the second bit set. If you mask out the second bit you get
the displayed value.
0 = 0000
1 = 0001
2 = 0010 *
3 = 0011 *
4 = 0100
5 = 0101
6 = 0110 *
7 = 0111 *
The front panel uses a 7447 BCD to 7 Segment decoder to drive the LED
modules. If pin 1 (B) on this is not being set when required then the
behaviour will be what I am seeing..... it isn't that the keypad is
getting it wrong... but the display is showing the wrong value due to
a missing bit.
Need to do a bit more digging here but given that this line is pulled
to +5V via a resistor on the front panel board I am hoping that the
problem is either
a connection problem on the front panel board with the line not being
pulled up or the 7447 needs to be replaced. If there was a break in
the ribbon table to the controller board then the bit would always be
set and a different set of incorrect values would be displayed.
Hopefully there isn't a fault on the controller board that is pulling
the pin to ground.
It's likely there's a driver chip on the controller board with an open
collector output (something like a 7438). That output could very well
be shorted to ground. That would in fact be my first suspect for this
error.
Camiel.
7:26 p.m.
Correction to that.... it should read:
The 7417 needs to pull this to ground to get the display segments to switch off.
So what I am seeing is the line coming in to the front panel from the
control board is pulled to ground and does not change.
So it could be:
1. The pullup resistor on the front panel board has failed.
2. The 7417 on the controller board has failed and is pulling the
line to ground.
3. Something feeding the 7417 on the controller board has failed.
That said the next step is still to is to check resistor and then the
input and output on the 7417 on the control board.
On Tue, Jul 24, 2012 at 10:26 AM, Andrew Quinn <jaquinn2001 at gmail.com> wrote:
Right. These lines are driven by a 7417's on the
controller board.
DEC seemed to like non-inverting buffers. The resistors on the front
panel board hold the line high and the 7417 needs to pull this to
ground to get the display segments to light up. Checked with the
scope on the front panel board and the second bit of the incoming BCD
value isn't going low. Pulling the line to ground on the front panel
board causes the 7 segment displays to read correctly so the BCD to 7
segment converter is OK. The cable to the controller board checks out
OK so the next step is to check the input and output on the 7417 gate
and see if the problem is there. Unfortunately I don't have extender
boards so will have to go with the old "solder wires to the chip"
trick.
On Mon, Jul 23, 2012 at 6:46 PM, Camiel Vanderhoeven
<iamcamiel at gmail.com> wrote:
> On Mon, Jul 23, 2012 at 1:11 AM, Andrew Quinn <jaquinn2001 at gmail.com>
wrote:
>> The insight came when I looked at which keys on the number pad worked
>> and which didn't. I came up with the following table
>>
>> Pressed = Displayed
>> 1 = 1
>> 2 = 0
>> 3 = 1
>> 4 = 4
>> 5 = 5
>> 6 = 4
>> 7 = 5
>> 0 = 0
>>
>> If you convert this to binary it becomes pretty obvious where the
>> problem lies. The invalid display is for the items marked with * and
>> all have the second bit set. If you mask out the second bit you get
>> the displayed value.
>>
>> 0 = 0000
>> 1 = 0001
>> 2 = 0010 *
>> 3 = 0011 *
>> 4 = 0100
>> 5 = 0101
>> 6 = 0110 *
>> 7 = 0111 *
>>
>> The front panel uses a 7447 BCD to 7 Segment decoder to drive the LED
>> modules. If pin 1 (B) on this is not being set when required then the
>> behaviour will be what I am seeing..... it isn't that the keypad is
>> getting it wrong... but the display is showing the wrong value due to
>> a missing bit.
>>
>> Need to do a bit more digging here but given that this line is pulled
>> to +5V via a resistor on the front panel board I am hoping that the
>> problem is either
>> a connection problem on the front panel board with the line not being
>> pulled up or the 7447 needs to be replaced. If there was a break in
>> the ribbon table to the controller board then the bit would always be
>> set and a different set of incorrect values would be displayed.
>>
>> Hopefully there isn't a fault on the controller board that is pulling
>> the pin to ground.
>
> It's likely there's a driver chip on the controller board with an open
> collector output (something like a 7438). That output could very well
> be shorted to ground. That would in fact be my first suspect for this
> error.
>
> Camiel.
24 Jul
24 Jul
11:47 a.m.
Hi Andrew,
On Tue, Jul 24, 2012 at 4:26 AM, Andrew Quinn <jaquinn2001 at gmail.com> wrote:
Correction to that.... it should read:
The 7417 needs to pull this to ground to get the display segments to switch off.
So what I am seeing is the line coming in to the front panel from the
control board is pulled to ground and does not change.
So it could be:
1. The pullup resistor on the front panel board has failed.
2. The 7417 on the controller board has failed and is pulling the
line to ground.
3. Something feeding the 7417 on the controller board has failed.
Number 2 is a very common mode of failure for open-collector 7400's
like the 7417. My money's still on that one. Time to desolder it and
put in a new one.
Camiel.
22 Jul
22 Jul
11:27 a.m.
Spent a few hours today with the unibus documentation, meter and scope
and have concluded there are no breaks in the NPG / NPR / BG4-7 or
BR4-7 lines on the 9 socket backplane.
What I would do next is, with the machien assembled and powered up, check
if any of BG4-7 or NPG are asserted (high) at the M9302 terminator. It's
_possible_ the CPU is asseerting one of them due to a problem in the
logic, it's also possible (and more likely) that one of the device cards
is not correctly passign on the grant (I have had this).
It is possible that some of the HALT / Bus Error problem might be with
the KY11-LB.... some of the keys on the keypad don't return the
correct values on the display when pressed. Specifically 3 and 6
which cause the display to show 1 and 4. These happen to be on the
same READ cycle as the CTRL and DEP buttons. My initial thought was a
problem with the 7404 that controls the READ signal but that looks OK
on the scope. I will pull the keypad off tomorrow and make sure it
isn't something in the switch matrix.
Yes, mechanical problems, like a defective switch are the most common
:-). You;'d better hope the rpoblem isn't in the KY11-B logic, IIRC
that's 8008-based so debugging it is 'fun'. I don't think I've ever
seen
lisitns of its firmware either.
I need to get back to the M7856 serial cable and
replace the Molex
with a DE9 (well spotted Tony... I never realized that DB9 was not the
right name). It may be that the machine is powering up fine and the
Sorry... this is a pet rant of mine. The letter after the 'D' is the
shell size. DB25 (and DB44) are correct. It is _amazing_ how many peopel
get this wrong, though.
console emulator works even if the KY11-LB is
misbehaving. Will try
that tomorrow.
Tony - to answer your question.... the 6 pin molex had 2 ground pins,
RXD, TXD. Also RTS and DTS are wired back to the Berg connector. I
thought I read somewhere that these are not used.
No DEC didn't use hardwre flow control on these interfaces. Certainly on
some DL11s, the RTS and DTR pins on the Berg connecotr are conencted to
the +15V line through suiotable resistors so they're always asserted.
In any case, theyr;'e not inputs to the PDP11, so you can ignore them.
The machien can't know if you've connected them or not :-)
On a positive note... all the power supply voltages on the backplane
looks good after the rework on those.
Excellent!
-tony
15 Jul
15 Jul
2:06 p.m.
I picked up an 11/04 this weekend, thanks to the generosity of another list
member. It has a known faulty PSU, although I don't yet know if it produces
any output at all or not. Any tips on common failures?
Regards
Rob
-----Original Message-----
From: cctalk-bounces at classiccmp.org [mailto:cctalk-
bounces at classiccmp.org] On Behalf Of Andrew Quinn
Sent: 14 July 2012 10:15
To: cctech at classiccmp.org
Subject: Ready to Power up the 11/04
I have the power supply tested and reassembled and am ready to power up
the 11/04.
I am looking to limit the number of installed boards initially.
Certainly the RK05 controller boards can go as I haven't got the drives
operational yet.
The configuration of the 11/04 as I received it is as follows (hopefully
this is
readable):
M7257 M7257 M9302
M7256 M7256 -
M7255 M7255 -
M7254 M7254 M920
M7258 M7258 M920
M7856 M7856 -
M7860 M7860 -
GRANT M9202
GRANT M9202
M7856 M7856 M7850
GRANT
M7847 M7847 M7847
GRANT
M7847 M7847 M7847
M7859 M7859 M9301
GRANT
M7263 M7263 M7263
Looking for some guidance if the following is a valid configuration:
GRANT M9302
M7856 M7856 M7850
GRANT
M7847 M7847 M7847
GRANT
M7847 M7847 M7847
M7859 M7859 M9301
GRANT
M7263 M7263 M7263
I am also looking to confirm my understanding of the M9301 card. I have
read the maintenance and operators manual for the M9301. I think it is
saying that the Console Emulator startup message from the M9301 and the
console emulator commands will be available on the terminal...
which I take to mean the terminal connected to the M7856 in this machine.
Am I understanding that right.
Thanks everyone for your help.
Regards
Andrew
6:11 p.m.
There are a series of posts on my blog
(http://www.quicktrip.co.nz/jaqblog) about the process I went through
getting the power supply running on my 11/04. This is in a BA11-K
chassis so may differ if you have the smaller BA11-L. For me it was
all about capacitor replacement. They work well now that is done.
On Mon, Jul 16, 2012 at 9:06 AM, Rob Jarratt
<robert.jarratt at ntlworld.com> wrote:
I picked up an 11/04 this weekend, thanks to the
generosity of another list
member. It has a known faulty PSU, although I don't yet know if it produces
any output at all or not. Any tips on common failures?
Regards
Rob
-----Original Message-----
From: cctalk-bounces at classiccmp.org [mailto:cctalk-
bounces at classiccmp.org] On Behalf Of Andrew Quinn
Sent: 14 July 2012 10:15
To: cctech at classiccmp.org
Subject: Ready to Power up the 11/04
I have the power supply tested and reassembled and am ready to power up
the 11/04.
I am looking to limit the number of installed boards initially.
Certainly the RK05 controller boards can go as I haven't got the drives
operational yet.
The configuration of the 11/04 as I received it is as follows (hopefully
this is
readable):
M7257 M7257 M9302
M7256 M7256 -
M7255 M7255 -
M7254 M7254 M920
M7258 M7258 M920
M7856 M7856 -
M7860 M7860 -
GRANT M9202
GRANT M9202
M7856 M7856 M7850
GRANT
M7847 M7847 M7847
GRANT
M7847 M7847 M7847
M7859 M7859 M9301
GRANT
M7263 M7263 M7263
Looking for some guidance if the following is a valid configuration:
GRANT M9302
M7856 M7856 M7850
GRANT
M7847 M7847 M7847
GRANT
M7847 M7847 M7847
M7859 M7859 M9301
GRANT
M7263 M7263 M7263
I am also looking to confirm my understanding of the M9301 card. I have
read the maintenance and operators manual for the M9301. I think it is
saying that the Console Emulator startup message from the M9301 and the
console emulator commands will be available on the terminal...
which I take to mean the terminal connected to the M7856 in this machine.
Am I understanding that right.
Thanks everyone for your help.
Regards
Andrew
6:36 p.m.
BA11-K (H765) or BA11-L (H777)? The H765 are easier to work on and
the parts are easier to find.
Paul
On Sun, Jul 15, 2012 at 4:06 PM, Rob Jarratt
<robert.jarratt at ntlworld.com> wrote:
I picked up an 11/04 this weekend, thanks to the
generosity of another list
member. It has a known faulty PSU, although I don't yet know if it produces
any output at all or not. Any tips on common failures?
Regards
Rob
-----Original Message-----
From: cctalk-bounces at classiccmp.org [mailto:cctalk-
bounces at classiccmp.org] On Behalf Of Andrew Quinn
Sent: 14 July 2012 10:15
To: cctech at classiccmp.org
Subject: Ready to Power up the 11/04
I have the power supply tested and reassembled and am ready to power up
the 11/04.
I am looking to limit the number of installed boards initially.
Certainly the RK05 controller boards can go as I haven't got the drives
operational yet.
The configuration of the 11/04 as I received it is as follows (hopefully
this is
readable):
M7257 M7257 M9302
M7256 M7256 -
M7255 M7255 -
M7254 M7254 M920
M7258 M7258 M920
M7856 M7856 -
M7860 M7860 -
GRANT M9202
GRANT M9202
M7856 M7856 M7850
GRANT
M7847 M7847 M7847
GRANT
M7847 M7847 M7847
M7859 M7859 M9301
GRANT
M7263 M7263 M7263
Looking for some guidance if the following is a valid configuration:
GRANT M9302
M7856 M7856 M7850
GRANT
M7847 M7847 M7847
GRANT
M7847 M7847 M7847
M7859 M7859 M9301
GRANT
M7263 M7263 M7263
I am also looking to confirm my understanding of the M9301 card. I have
read the maintenance and operators manual for the M9301. I think it is
saying that the Console Emulator startup message from the M9301 and the
console emulator commands will be available on the terminal...
which I take to mean the terminal connected to the M7856 in this machine.
Am I understanding that right.
Thanks everyone for your help.
Regards
Andrew
16 Jul
16 Jul
10:44 a.m.
It is a H777 PSU.
Regards
Rob
-----Original Message-----
From: cctalk-bounces at classiccmp.org [mailto:cctalk-
bounces at classiccmp.org] On Behalf Of Paul Anderson
Sent: 16 July 2012 02:36
To: General Discussion: On-Topic and Off-Topic Posts
Subject: Re: Ready to Power up the 11/04
BA11-K (H765) or BA11-L (H777)? The H765 are easier to work on and the
parts are easier to find.
Paul
On Sun, Jul 15, 2012 at 4:06 PM, Rob Jarratt <robert.jarratt at ntlworld.com>
wrote:
I picked up an 11/04 this weekend, thanks to the
generosity of another
list member. It has a known faulty PSU, although I don't yet know if
it produces any output at all or not. Any tips on common failures?
Regards
Rob
terminal...
-----Original Message-----
From: cctalk-bounces at classiccmp.org [mailto:cctalk-
bounces at classiccmp.org] On Behalf Of Andrew Quinn
Sent: 14 July 2012 10:15
To: cctech at classiccmp.org
Subject: Ready to Power up the 11/04
I have the power supply tested and reassembled and am ready to power
up the 11/04.
I am looking to limit the number of installed boards initially.
Certainly the RK05 controller boards can go as I haven't got the
drives operational yet.
The configuration of the 11/04 as I received it is as follows
(hopefully
this is
> readable):
>
> M7257 M7257 M9302
> M7256 M7256 -
> M7255 M7255 -
> M7254 M7254 M920
>
> M7258 M7258 M920
> M7856 M7856 -
> M7860 M7860 -
> GRANT M9202
>
> GRANT M9202
> M7856 M7856 M7850
> GRANT
> M7847 M7847 M7847
> GRANT
> M7847 M7847 M7847
> M7859 M7859 M9301
> GRANT
> M7263 M7263 M7263
>
> Looking for some guidance if the following is a valid configuration:
>
>
> GRANT M9302
> M7856 M7856 M7850
> GRANT
> M7847 M7847 M7847
> GRANT
> M7847 M7847 M7847
> M7859 M7859 M9301
> GRANT
> M7263 M7263 M7263
>
> I am also looking to confirm my understanding of the M9301 card. I
> have read the maintenance and operators manual for the M9301. I
> think it is saying that the Console Emulator startup message from the
> M9301 and the console emulator commands will be available on the > which I take to mean the terminal connected
to the M7856 in this
machine.
>> Am I understanding that right.
>>
>> Thanks everyone for your help.
>>
>> Regards
>>
>> Andrew
>
11:57 a.m.
It is a H777 PSU.
You sure know how to pick them. First you have an H7140 that has startup
chopper problems and now you have an H777...
Anyway, the first thing to do is to check the fuse (in the obvious holder
on top), then to see if it prodcues any outputs and if the internal relay
operates. That relay is part of the inrush limiter circuti and you can
have soem very odd problems if it doesn't work.
-tony
12:50 p.m.
Can't say I get much of a choice when it is given to me free.... :-)
It will be a little bit of time before I get to look at it properly, but I
will start with these things and try to narrow down the problem area.
Regards
Rob
-----Original Message-----
From: cctalk-bounces at classiccmp.org [mailto:cctalk-
bounces at classiccmp.org] On Behalf Of Tony Duell
Sent: 26 June 2012 20:51
To: cctalk at classiccmp.org
Subject: Re: Ready to Power up the 11/04
on
It is a H777 PSU.
You sure know how to pick them. First you have an H7140 that has startup
chopper problems and now you have an H777...
Anyway, the first thing to do is to check the fuse (in the obvious holder top), then to see if it prodcues any outputs and if
the internal relay
operates.
That relay is part of the inrush limiter circuti and
you can have soem
very
odd problems if it doesn't work.
-tony
17 Jul
17 Jul
11:47 a.m.
[Hopefully the time/date for this message will be correct...]
Can't say I get much of a choice when it is given to me free.... :-)
I know... Given the choice, I'd never recomend the PDP11/45 as an
introduction to PDP11s, that machine is complciated. But it's what I
started on, for much the same reason :-)
It will be a little bit of time before I get to look at it properly, but I
will start with these things and try to narrow down the problem area.
Right...
Look at the fuse. If it;s violently blown (shattered glass, or
blackened), don't just replace it and try again. There is a major problem
i nthe masin side of the PSU.
If the fuse has blown softly, replaced it and try the PSU. Of course if
the fuse is OK. then try the PSU.
The outputs are on those 15 pin AMP connectors on the angled side of the
PSU. Standard (DEC) pinout, I can look it up. You want to see if there
are any outputs at all, if one particualr output is missing/incorrect, if
the entire PSU seems dead, or what.
-tony
16 Jul
16 Jul
11:57 a.m.
I picked up an 11/04 this weekend, thanks to the generosity of another list
member. It has a known faulty PSU, although I don't yet know if it produces
any output at all or not. Any tips on common failures?
IIRC the 11/04 can be (officially) fitted in several different
mounting boxes.. I have experience of2 o them
The 5.25" one uses an H777 PSU to the right of the backplane. This is a
somewhat evil SMPSU at least if you don't fully understand the scheamtic.
The 10.5" one Has the PSU o nthe back. It has a large mains transformer
in the middle feed DEC power bricks (low-votlage swtiching regualtors)
for the 5V and -15V outputs. The +15V supply is a simple linear regulator
on a PCB under the transforrmer. This one is relatively easy to sort out.
I cant' thin kof any common failures for either one, but let us know what
you have and I'll think of things to check.
-tony
4534
days inactive
4544
days old
25 comments
7 participants
participants (7)
-
ard＠p850ug1.demon.co.uk -
henk.gooijen＠hotmail.com -
iamcamiel＠gmail.com -
jaquinn2001＠gmail.com -
quapla＠xs4all.nl -
robert.jarratt＠ntlworld.com -
useddec＠gmail.com