2 Jun
2009
2 Jun
'09
8:52 a.m.
Paul Anderson <useddec at gmail.com> wrote:
I always used a M930 in the 11/40 and older machines,
and the M9300 in the
11/34 and newer machines. I have no way to check on the differences
now. now. Also, there may be a trick using MOS in an 11/40.
I can't remember that there should be any functional difference between
a M930 and a M9300. The later is just an improved design.
As for the original posters problems. When you get a stuck machine when
the terminator is in, but a somewhat more functional machine when the
bus terminator is out, you have a problem on the bus. Most likely a bus
grant or NPR grant. A third possibility is a problem in the CPU with the
logic related to these signals.
I'd go over the bus grants once more. Check that they *really* are the
right way, since it's so easy to put them in the wrong way, and there is
basically no visible clue to which way they should go.
Look at some other card to get help. They will also have connections on
the same places the BG cards have, except they will likely run them up
to the priority plug. If you have some Unibus memory, they are nice,
since they have the BG lines wired straight through, just like the BG cards.
I would have started by removing *all* cards from the Unibus (with the
exception of some memory), and make sure all bus grants and NPRs were
in, the terminator was in place and that all power supplies were
delivering the right voltage. Also check that all cards in the CPU box
are in place, and are firmly seated.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt at softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
2 Jun
2 Jun
6:06 p.m.
Paul Anderson <useddec at gmail.com> wrote:
I thought that was only true if you were using an M9302 termintor. That
board will assert SACK if it gets a grant (that is, if a grant goes all
the way along the Unibus and isn't 'taken' by some device). The older
terminators (M930, and I think M9300, don't. THey're just resistors to
terminate the bus.
I always used a M930 in the 11/40 and older
machines, and the M9300 in the
11/34 and newer machines. I have no way to check on the differences
now. now. Also, there may be a trick using MOS in an 11/40.
I can't remember that there should be any functional difference between
a M930 and a M9300. The later is just an improved design.
As for the original posters problems. When you get a stuck machine when
the terminator is in, but a somewhat more functional machine when the
bus terminator is out, you have a problem on the bus. Most likely a bus grant or NPR grant. A third possibility is a problem
in the CPU with the
logic related to these signals.
I think it's time to stop guessing -- I think we've tried all the obvious
things -- and start logical faultfinding. What test gear does the OP
have?
What _I_ would do is first check all the power voltages, with the boards
in (it's too late to care about a rogue PSU damaging boards :-)). A low
+5V line, or a missing supply to the terminator, will casue all sorts of
problems.
Then I'd look at all the Unibus signals both with the terminator out
(they will still be terminated by the resistors at at the CPU end -- IIRC
on an 11/40 these are on the Unibus jumper between the CPU and first
expansion backplane) and with it fitted. I would guess something is
changing state, let's find out what.
-tony
7:27 p.m.
Tony Duell wrote:
I have a nice Tektronix 1241 logic analyzer and a DMM (and a really
flaky old Tek O-scope). I said in my original mail on this thread that
I just wanted to be sure my Unibus config _looked_ sane before I started
digging. (No sense spending hours debugging if all it is is a misplaced
board...)
Paul Anderson
<useddec at gmail.com> wrote:
I thought that was only true if you were using an M9302 termintor. That
board will assert SACK if it gets a grant (that is, if a grant goes all
the way along the Unibus and isn't 'taken' by some device). The older
terminators (M930, and I think M9300, don't. THey're just resistors to
terminate the bus.
I always used a M930 in the 11/40 and older
machines, and the M9300 in the
11/34 and newer machines. I have no way to check on the differences
now. now. Also, there may be a trick using MOS in an 11/40.
I can't remember that there should be any functional difference between
a M930 and a M9300. The later is just an improved design.
As for the original posters problems. When you get a stuck machine when
the terminator is in, but a somewhat more functional machine when the
bus terminator is out, you have a problem on the bus. Most likely a bus
grant or NPR grant. A third possibility is a
problem in the CPU with the
logic related to these signals.
I think it's time to stop guessing -- I think we've tried all the obvious
things -- and start logical faultfinding. What test gear does the OP
have?
What _I_ would do is first check all the power
voltages, with the boards
in (it's too late to care about a rogue PSU damaging boards :-)). A low
+5V line, or a missing supply to the terminator, will casue all sorts of
problems.
The voltages seem to be fine with the boards installed (5.2V for the +5,
-5.3V for -5). I fixed the ACLO/DCLO problems I was having awhile back
(turned out to be a bad contact on one of the many molex connectors.
That was nice, since I didn't want to pull the supply out again :)).
Then I'd look at all the Unibus signals both with
the terminator out
(they will still be terminated by the resistors at at the CPU end -- IIRC
on an 11/40 these are on the Unibus jumper between the CPU and first
expansion backplane) and with it fitted. I would guess something is
changing state, let's find out what.
I will start looking at this tonight (assuming I have the time tonight
:)). Thanks.
Josh
-tony
3 Jun
3 Jun
6:13 p.m.
I think
it's time to stop guessing -- I think we've tried all the obvious
things -- and start logical faultfinding. What test gear does the OP
have?
I have a nice Tektronix 1241 logic analyzer and a DMM (and a really
flaky old Tek O-scope). I said in my original mail on this thread that I just wanted to be sure my Unibus config _looked_
sane before I started
digging. (No sense spending hours debugging if all it is is a misplaced
board...)
Sure. It is always best to look for the obvious things first (says the
guy who once spent a long time fioguring out a low contrast fault on a
VT100 video board, only to find the CRT screen needed cleaning...). But
after a bit, it's best to attack the problem logically.
What _I_ would do is first check all the power
voltages, with the boards
in (it's too late to care about a rogue PSU damaging boards :-)). A low
+5V line, or a missing supply to the terminator, will casue all sorts of
problems.
The voltages seem to be fine with the boards installed (5.2V for the +5,
-5.3V for -5). I fixed the ACLO/DCLO problems I was having awhile back (turned out to be a bad contact on one of the many
molex connectors.
That was nice, since I didn't want to pull the supply out again :)).
How many +5V lines are there? Certainly in my 11/45 there are 4 or
5 +5V regulator bricks for the CPU, and you need to check the outputs of
all of them. They are not parelleled, rather one feeds the CPU boards,
another the first expansion backplane, and so on.
Then I'd look at all the Unibus signals both
with the terminator out
(they will still be terminated by the resistors at at the CPU end -- IIRC
on an 11/40 these are on the Unibus jumper between the CPU and first
expansion backplane) and with it fitted. I would guess something is
changing state, let's find out what.
I will start looking at this tonight (assuming I have the time tonight
:)). Thanks. 
5 Jun
5 Jun
10:46 p.m.
Tony Duell wrote:
Sure. It is always best to look for the obvious things
first (says the
guy who once spent a long time fioguring out a low contrast fault on a
VT100 video board, only to find the CRT screen needed cleaning...). But
after a bit, it's best to attack the problem logically.
Long ago, I bought a Heathkit color television -- it was gorgeous.
Needed some work. Also, while I was contemplating buying it, the owner
offered to throw in a 'scope, too. It had never worked he said. The
screen was "grainy" and a new tube cost too much. I accepted. The
television was easy to fix (Heathkit, I said).
But, when I fired up the 'scope, I was shocked. He was right. The
scope image could be focused, but it never got clear. It looked like
the phosphor was painted on the tube with a roller. Then, I had to
laugh....
I took off the bezel, pulled off the graticule screen, and peeled the
tyvek sticky paper off the back of the plastic screen, and voila! A
nice clear trace.
Warren
6 Jun
6 Jun
5:30 p.m.
Long ago, I bought a Heathkit color television -- it
was gorgeous.
Needed some work. Also, while I was contemplating buying it, the owner
offered to throw in a 'scope, too. It had never worked he said. The
screen was "grainy" and a new tube cost too much. I accepted. The
television was easy to fix (Heathkit, I said).
But, when I fired up the 'scope, I was shocked. He was right. The
scope image could be focused, but it never got clear. It looked like
the phosphor was painted on the tube with a roller. Then, I had to
laugh....
I took off the bezel, pulled off the graticule screen, and peeled the
tyvek sticky paper off the back of the plastic screen, and voila! A
nice clear trace.
ROFL!!
I wodnered why complex electronic instruments came with such losing
manuals, after all, surely the sort of person to buy a 'scope or logic
analuyser has more than 2 working braincells. Now I know...
-tony
6:59 p.m.
New subject: Stupid problems / was Re: Further 11/40 unibus questions...
Warren Wolfe wrote:
But, when I fired up the 'scope, I was shocked.
He was right. The
scope image could be focused, but it never got clear. It looked like
the phosphor was painted on the tube with a roller. Then, I had to
laugh....
I took off the bezel, pulled off the graticule screen, and peeled the
tyvek sticky paper off the back of the plastic screen, and voila! A
nice clear trace.
As a kid, my first scope (a low-end tube-based TV servicing scope) developed a
problem with 60Hz noise in the trace, that is, even with the V input shorted,
there would still be a slight 60Hz sine wave on the trace rather than flat. The
problem was intermittent as well. Opened it up (of course the problem went away
when opened), tried to diagnose it as much as possible (now that I didn't have
a scope to diagnose with).
No luck.
I suffered with the problem for ages, then finally clued in that my bench/lab
power supply - which of course contained a power transformer which of course
would generate a magnetic field - was sitting on top of the scope, just above
the CRT.
8 Jun
8 Jun
11:32 p.m.
New subject: Stupid problems / was Re: Further 11/40 unibus questions...
Brent Hilpert wrote:
I suffered with the problem for ages, then finally clued in that my bench/lab
power supply - which of course contained a power transformer which of course
would generate a magnetic field - was sitting on top of the scope, just above
the CRT.
Back in the old days, I got a call from a friend who was being driven
nuts by his CP/M computer, which was a current, nice machine at the
time. His 8" disks were randomly flaking out on him, and he had taken
the whole system in to the store several times (a difficult task, back
then) and when they set it up, there were no problems. He was going
crazy, keeping multiple backups of each session, and still losing a lot
of work. He had had numerous people over after business hours to try to
find the problem, but had had no luck. With a techie there, there was
never a problem.
I went over to his place, and we watched for a while, with no effect. I
wrote a program which wrote pseudo-random crap on the disk, and then
randomly skipped around reading various tracks and sectors, checking the
data, and then re-writing it, ringing the bell like a hunchback on crack
if it found one of the sectors did not have the proper information or if
a write failed. (Program: Quasi.com)
We ran that sucker for hours, with no problems detected. I told him
that my wife and I had plans for dinner that I didn't want to mess with,
so when she called, that was it, I had to leave. It was getting near
the time when she would call, and he was getting desperate. He appeared
to be close to tears with frustration. "Look, I know there's no problem
NOW, but, really, it fails all the time when I'm doing actual work.
What's different about that?" Whiner. <Grin>
Finally, as things happen, my time was up -- my wife called.... And the
terminal bell started going nuts at the same time. We both started, and
I reached to pick up the phone.... which was sitting on top of his disk
drive enclosure. The (Old Bell System) phone had a real bell, and the
magnetic field generated when the phone rang was scrambling any disks in
the drive at the time. During work, of course, he got lots of calls,
hence had lots of problems. After work, nobody called. Sheesh. I
moved his phone to the other side of the desk... problem solved.
Warren
9 Jun
9 Jun
1:39 p.m.
New subject: Stupid problems / was Re: Further 11/40 unibus questions...
From: lists at databasics.us
Brent Hilpert wrote:
Hi
I'd seen similar problems from setting the video monitor on
top of the PC. The color screens had the de-guassing coil
that came on when turned on. It would get the disk to.
Dwight
_________________________________________________________________
Lauren found her dream laptop. Find the PC that?s right for you.
http://www.microsoft.com/windows/choosepc/?ocid=ftp_val_wl_290
I suffered with the problem for ages, then finally clued in that my bench/lab
power supply - which of course contained a power transformer which of course
would generate a magnetic field - was sitting on top of the scope, just above
the CRT.
Back in the old days, I got a call from a friend who was being driven
nuts by his CP/M computer, which was a current, nice machine at the
time. His 8" disks were randomly flaking out on him, and he had taken
the whole system in to the store several times (a difficult task, back
then) and when they set it up, there were no problems. He was going
crazy, keeping multiple backups of each session, and still losing a lot
of work. He had had numerous people over after business hours to try to
find the problem, but had had no luck. With a techie there, there was
never a problem.
I went over to his place, and we watched for a while, with no effect. I
wrote a program which wrote pseudo-random crap on the disk, and then
randomly skipped around reading various tracks and sectors, checking the
data, and then re-writing it, ringing the bell like a hunchback on crack
if it found one of the sectors did not have the proper information or if
a write failed. (Program: Quasi.com)
We ran that sucker for hours, with no problems detected. I told him
that my wife and I had plans for dinner that I didn't want to mess with,
so when she called, that was it, I had to leave. It was getting near
the time when she would call, and he was getting desperate. He appeared
to be close to tears with frustration. "Look, I know there's no problem
NOW, but, really, it fails all the time when I'm doing actual work.
What's different about that?" Whiner.
Finally, as things happen, my time was up -- my wife called.... And the
terminal bell started going nuts at the same time. We both started, and
I reached to pick up the phone.... which was sitting on top of his disk
drive enclosure. The (Old Bell System) phone had a real bell, and the
magnetic field generated when the phone rang was scrambling any disks in
the drive at the time. During work, of course, he got lots of calls,
hence had lots of problems. After work, nobody called. Sheesh. I
moved his phone to the other side of the desk... problem solved.
Warren
6:10 a.m.
Tony Duell wrote:
OK, let us know how you get on :-)
-tony
Ok, finally stopped being distracted by other shiny objects for long
enough to do some more fiddling with the 11/40.
And of course, instead of hooking up the logic analyzer, I decided to
play around with the Console SLU/LTC board. Because I evidently don't
follow suggestions well.
But this has a good ending, sort of. Maybe.
So the SLU was unresponsive no matter what I did. Tried it at 9600
baud, no go. Dialed it down to 300, no dice. Checked the continuity of
the dip switches, of which there are approximately 500. No problems
there. Checked, and double-checked the wiring on the serial cable I
built. No go. Stole the cable from my 8/e... still no good.
So I moved it out of the 9th slot on the processor backplane and into
the first slot on Unibus backplane. (And put a grant card in the 9th
slot...) And hey, it works. Toggled in a short "echo" program and what
I type on the terminal keyboard is echoed back, at a blistering 300 baud.
So... clearly there's something wrong with the SPC slot on the processor
backplane. A couple more questions:
1) Is the NPG grant on the unibus slot on the processor backplane (slot
9) supposed to be connected to the NPG grants on the Unibus expansion?
That is -- right now if I set my DMM to continuity mode and put one
probe on CA1 on the first slot of the unibus expansion, and the other on
CB1 on the last slot of the unibus expansion, since all NPG grant
jumpers are in place, the DMM shows the circuit as closed. This is as
I'd expect. However, if I move the probe from CA1 on the first slot of
the expansion to CA1 on slot 9 of the processor backplane, the circuit
is then open. I'm guessing this is not correct. (There is currently an
NPG jumper installed on slot 9.)
2) Where is the +15V to the processor backplane supposed to be
connected? (I suspect this may be the reason the SLU won't function in
slot 9...). Right now it's plugged into pin CV1 on slot 9 (if I'm
reading the Unibus pin chart right :)) but the docs I have found say
this should be ACLO_L...
Thanks as always...
Josh
Then I'd look at all the Unibus signals both with
the terminator out
(they will still be terminated by the resistors at at the CPU end -- IIRC
on an 11/40 these are on the Unibus jumper between the CPU and first
expansion backplane) and with it fitted. I would guess something is
changing state, let's find out what.
I will start looking at this tonight (assuming I have the time tonight
:)). Thanks.
5691
days inactive
5698
days old
9 comments
6 participants
participants (6)
-
ard＠p850ug1.demon.co.uk -
bqt＠softjar.se -
derschjo＠mail.msu.edu -
dkelvey＠hotmail.com -
hilpert＠cs.ubc.ca -
lists＠databasics.us