Some Core Memory
  

An early 1960's core memory

A never-used spare part

Part of the Core Memory pages
by Douglas W. Jones
THE UNIVERSITY OF IOWA Department of Computer Science

Contents

Overview

thumbnail image thumbnail image
This module contains 8k bits, physically arranged as 2 planes of 64 × 64 cores. 4 wires penetrate each core, 2 parallel wires in the X direction, and 2 parallel wires in the Y direction. The wire pairs are 0.1 inches (2.54 mm) apart at their anchorages in both the X and Y directions, so the nominal area of the actual core plane is 6.4 × 6.4 inches (162.6 × 162.6 mm). Within the plane, the wires wander, so the local spacing within the plane is not uniform. The individual cores appear to be 0.077 inches (1.96 mm) in diameter, 1/4 that in width, and 1/6 that in rim thickness, measured with a vernier caliper on a single core.

thumbnail image
The core planes hang suspended from the wires in the center of a green fiberglass-epoxy printed circuit board frame. The opening in the frame is 6.5 inches (165 mm) square, and the frame is 5/16 inches (7.9 mm) wide. The fiberglass-epoxy board is 1/16 inch (1.59 mm) thick.

The two planes are separated by aluminum spacers 1 1/4 inches (31.7 mm) long, and protected on the two outside surfaces by plastic squares 7 3/8 inches (187 mm) on a side and 3/32 inches (2.4 mm) thick. 4 6-32 UNC screws 2 1/8 inch (54 mm) long hold the entire assembly together, including mounting ears that serve as spacers between the core plane and plastic on one side.

thumbnail image thumbnail image
A rectangle of printed circuit board is mounted across one side of the assembly, screwed to two of the posts supporting the core planes. At first, this looks like it merely connects the two planes, but there is a slot in the board that supports a single row of 64 cores. These may be the same type of cores as are used in the two planes, but they are wired differently. On close inspection, some of the wires through each of these cores are wrapped at least once around the core, making it clear that these cores serve, somehow, as transformers.

Close examination of the close-up photos of where the wire pairs meet the fiberglass-epoxy circuit boards shows an occasional plastic film filling the triangle formed where the wires diverge as they meet the circuit board. This suggests that the entire core module has been dipped or sprayed with some kind of varnish after assembly in order to stabilize it.

thumbnail image
Bundles of stranded hookup wire runs from each of three edges of the frame to silver-plated tapered connector pins that presumably were press-fitted into the backplane connectors where they mated with the electronics. The actual connector pins are 1/4 inch (6.35 mm) long, with a 3/8 inch long (9.5 mm) tube at the rear crimped to the wire and encased in a 1/2 inch (12.7 mm) length of vinyl tube 1/8 inch (3.17 mm) outside diameter. information to confirm any guesses about the memory module described here.

Provenance

This memory was found in late October 1991 in the University of Iowa Physics and Astronomy Department's Van de Graff accelerator lab. This was shortly before the contractor took posession of the lab to demolish it to make way for the new Biology annex. Based only on the low density of this memory and its unusual organization, we guess that it is a very early core module, probably from the early 1960s.

When found, it was in what appears to have been its factory original packaging, suggesting that it was a never-used spare part. Each bundle of connector wires remains separately wrapped in a polyethylene baggy, and the entire module was wrapped in heavy polyethylene sheeting and held shut with decayed fiberglass-reinforced cellophane tape. The sheeting was removed shortly after the memory was found, but the photos here were taken in 2019.

thumbnail image
One side of the wrapper had, handwritten on it in black marker, the following text:

MEMORY ASSY
  128 × 64 Linear Sel

968603

  S/N 2826

thumbnail image thumbnail image
The other side of the wrapper had handwriting on it in blue marker, this was difficult to read without suppressing the blue and green channels and bringing up the red channel, as shown in the image to the right. following text:

22305200
MEMORY Stack
64 × 64
8865.02
USA

If the 8865.02 figure is a price, it comes to something more than one dollar per bit of memory capacity. The serial number 2826 appears nowhere on the module, but only on the wrapper. The other large numbers appear to be inventory numbers, but in whose inventory systems?

The U of the USA at the end is extremely faint, and the S is faint. The other possible interpretation is NASA, since the U of Iowa Physics Department has a relationship with NASA that goes back to the very beginning (they built the instrtument package for Explorer 1). The remaining legible part of the letter before the S is a point, like the upper left point of a U, and does not resemble the arched top of the A, making NASA unlikely.

We have no documentation to indicate who made this core module. The fact that the dimensions are all round numbers or binary fractions of inches suggests North American or British manufacture (the bulk of the move to metric in Britain occurred in between 1970 and 1980). The use of 6-32 screws points solidly to North American manufacture.

The tapered pin connectors point away from IBM and Digital Equipment Corporation. IBM used wire-wrap connections for their backplanes starting in the 1950s. DEC used soldered wiring in their very first computers, and then moved to wire-wrap. Burroughs Corp. seems to have preferred brown printed circuit boards in the early 1960s, and moved to wire-wrap interconnection at an early date.

The University of Iowa Physics department had several CDC 160 computers acquired in the early 1960s. It is hard to find documentation on the internals of these machines, but fragmentary information suggests that they may have had a tapered-pin interconnect system on their backplanes. This is not enough

Condition

thumbnail image
thumbnail image
When we received the core memory from the Physics Department, it was still wrapped in its plastic wrapper. The adhesive in the fiberglass reinforced cellophane tape had hardened, so that the cellophane came free from the fiberglass, and when that was removed, powdery yellow residue remained on the polyethylene wrapper.

When the memory was unwrapped in 1991, the polyethylene wrapper was extremely brittle. Every time it is handled, flakes of polyethylene sheet break off, and it is breaking apart along the creases.

We never removed the baggies that are taped around the bundles of wires from the memory. These were intact when we unwrapped the memory, but they have since begun to flake. Our best guess is that the wrapper around the entire memory excluded oxygen for many years, so that the decay processes in the baggies only began when the outer wrapper was removed.

thumbnail image thumbnail image
When we removed the outer wrapper in 1991, the smell of acetic acid was overwhelming. Looking at the two plastic sheets that enclosed the core, one side was clear and flat, while the other side was yellow, cloudy, warped and obviously smaller. In the years since the module was unwrapped, the shrinkage has continued. The shrinkage was enough that it was seriously distorting the memory plane nearest it, and in fact, enough that it cracked one corner off of the still-flat plastic sheet protecting the other side.

thumbnail image thumbnail image
thumbnail image
After removing the decayed plastic in early 2019, several things were apparent. First, the decayed plastic had been under sustained tension sufficient to stretch the corners around the screw holes. Second, there were numerous cracks and bubbles in the plastic. The longer cracks appear to be perpendicular to the direction of maximum tension on the plastic, but some of the shorter cracks look almost random and the bubbles can be felt as lumps. Perhaps some of the cracks began as bubbles? Apparently, the plastic is outgassing some decay product. In any case, the bubbles and cracks were not apparent in when the memory module was first unwrapped.

thumbnail image
After the decayed plastic was removed in 2019, it was set aside. The decay process continued. The photo here shows it as it was in mid 2022. By that point, the remains of this once flat sheet of plastic had separated into two extraordinarily delecate layers, each of which was slowly shattering under the internal tensions caused by the decay process. The shattered remains were discarded after taking this last photo.

Note that the entire core memory appears to have been stored at room temperature with little exposure to direct sunlight since it was obtained by the University of Iowa. The storage area in the Physics Department's Van de Graff lab was windowless, and the bookshelf where it was has been stored in the U. of Iowa Computer Sience Department is in a faculty office. Prior to the construction of Van Allen Hall in 1964, the memory may have been stored in un-air-conditioned space, but since then, it is unlikely to have been exposed to temperatures outside the range 65 — 90°F (18 — 32°C). (The high temperature in this range is the result of heating system malfunction in the winter, but it is well within the standard range for commercial electronics, where temperatures up to 70°C are considered normal in the interior of electronics enclosures.)

thumbnail image
The shrinking plastic cover did more than merely distort the core plane, it tore some of the wires loose along the edge where the core plane was jumpered to the single row of cores joining the two planes. In the photo to the right, the 8 wires on the left side are pulled loose, while the wires on the right are still connected. After the warped plastic cover was removed, the fiberglass frame of the core plane sprung back up, pushing some of the loose wires up while bending others.