20 Nov
1998
20 Nov
'98
3:35 a.m.
On Fri, 20 Nov 1998, Bill Yakowenko wrote:
] The amazing lace and tiny rings are long gone now.
But the
] legacy still lives on in the word "stack" and "core memory".
Of course the coolest thing about 'core' is it's the original
non-volitile memory; you could carry one around with a diag
utility already loaded, just plug it in and run.
One of the myth's they always told us was they all had to be
made by hand - there wasn't a machine that could thread the
row/column and sense wires for mass production. Is there any
rumor to this truth?
Chuck
cswiger(a)widomaker.com
20 Nov
20 Nov
7:28 a.m.
Chuck wrote:
One of the myth's they always told us was they all
had to be
made by hand - there wasn't a machine that could thread the
row/column and sense wires for mass production. Is there any
rumor to this truth?
The row/column drive line insertion was eventually automated, although
amazingly enough there was a period of time for which manual assembly was
less expensive than automatic assembly.
The cores were poured onto a metal plate with indentations where they were
supposed to reside. Each indentation had a pinhole through which a vacuum
pump provided suction to keep the cores in place. The plate was vibrated
until all the cores were properly seated.
Once the cores were corrrectly positioned, long hollow needles with wire
threaded through them were driven through the rows (or columns). The wires
were attached to the frame, and the needles withdrawn.
AFAIK, however, the sense lines were still threaded by hand.
Eric
7:59 a.m.
New subject: Core planes
The row/column drive line insertion was eventually automated, although
amazingly enough there was a period of time for which manual assembly was
less expensive than automatic assembly.
The cores were poured onto a metal plate with indentations where they were
supposed to reside. Each indentation had a pinhole through which a vacuum
pump provided suction to keep the cores in place. The plate was vibrated
until all the cores were properly seated.
Once the cores were corrrectly positioned, long hollow needles with wire
threaded through them were driven through the rows (or columns). The wires
were attached to the frame, and the needles withdrawn.
AFAIK, however, the sense lines were still threaded by hand.
My first core stack is in need of repair - a 13-bit PDP-8/L stack made
of 3 panels of 4 planes and one panel with 1 plane (parity). I got the
machine in high school - it was defective, so I cut apart the wires
holding the planes together. There are several dozen broken cores on
a couple of the planes. I have always contemplated repairing it, after
all, what can I do, break it? With no source of replacement cores, I was
considering scavenging the parity plane. If someone had a similarly broken
PDP-8/L or PDP-8/i core stack I would love to sort the planes for three
good panels. I do not relish the thought of threading the sense/inhibit
wire through 4096 cores. Does anyone have any idea as to the wire diameter?
(In case I can't remove and reinstall the same wire)
One problem I had when disassembling it was that the PCB material is
very temperature sensitive. It blisters under a normal iron. I tried
to desolder the top plane and did a bit of damage to the diode-matrix
board.
-ethan
8:39 a.m.
New subject: Core planes
Ethan wrote:
I do not relish the thought of threading the
sense/inhibit
wire through 4096 cores. Does anyone have any idea as to the wire diameter?
(In case I can't remove and reinstall the same wire)
It is my understanding that when core planes were commercially repaired,
typically the row/column lines were replaced, but the sense/inhibit line was
spliced.
8:43 a.m.
New subject: Core planes
Ethan wrote:
That doesn't surprise me. I think I can thread a wire stright through the
matrix (64 cores, each way), but the thought of installing the sense/inhibit
wire always made me postpone the project. At least the old cores are embedded
in some kind of pliable substrate (hot glue?)
Fortunately for me, I found a brand-new core stack for sale on the net
a few years ago - $20 in the factory box. Probably one of the greatest
buys I've ever made. The guy even had the one M module I was misssing
I do not relish the thought of threading the
sense/inhibit
wire through 4096 cores. Does anyone have any idea as to the wire diameter?
(In case I can't remove and reinstall the same wire)
It is my understanding that when core planes were commercially repaired,
typically the row/column lines were replaced, but the sense/inhibit line was
spliced. from the 8Kw upgrade for the PDP-8/i, some kind of
buffered delay line (M305?)
When I move the -8/i next year, I should be able to
boost it to 8Kw at
last. If only I could get a better disk for it than the DF32's it came
with - then I would load OS/8 on it.
-ethan
8:56 a.m.
New subject: Core planes
Ethan wrote:
When I move the -8/i next year, I should be able to
boost it to 8Kw at
last. If only I could get a better disk for it than the DF32's it came
with - then I would load OS/8 on it.
I've been trying to find a DF32 for some time. Would you like to trade
them for an RL02?
9:49 a.m.
New subject: Core planes
Ethan wrote:
I really don't need any more RL02 drives than I have. I am interested
in a trade for PDP-8 peripherals. I have no RK8E, for example. I have
two entire DF32 systems - cables, master and slave drives. Racks available
as well. I am located in Columbus, OH.
-ethan
When I move the -8/i next year, I should be able
to boost it to 8Kw at
last. If only I could get a better disk for it than the DF32's it came
with - then I would load OS/8 on it.
I've been trying to find a DF32 for some time. Would you like to trade
them for an RL02?
9:05 a.m.
New subject: Core planes
Ethan wrote:
At least the old cores are embedded
in some kind of pliable substrate (hot glue?)
Since the cores are fragile, that serves to protect them from mechanical
stress, in handling, and perhaps in operation. I don't know to what extent
it affects 18/30 and smaller cores, but there were problems with larger cores
vibrating in operation.
Of course, it took a lot of amps to switch the big cores, so they also had
problems with heating due to the power dissipation. And the hysteresis
curves of the big cores were more temperature sensitive, so the stack
had to be maintained at a constant temperature. Some early core systems
like the IBM 7302 (used on the 7090) were immersed in oil for this reason.
The later 7302A was apparently air-cooled.
8:01 a.m.
At 06:28 PM 11/20/98 -0000, Eric Smith wrote:
The row/column drive line insertion was eventually automated, although
amazingly enough there was a period of time for which manual assembly was
less expensive than automatic assembly.
More recently, when did chip companies stop using manual labor
in far-away places for the affixing of the gold wires from the
bonding pads on the silicon die to the metal legs on the chip case?
Isn't this still happening?
- John
9377
days inactive
9377
days old
8 comments
4 participants
participants (4)
-
cswiger＠wilma.widomaker.com -
erd＠infinet.com -
eric＠brouhaha.com -
jfoust＠threedee.com