HN
Today

Cores in space: The core memory module from a 1980 Spacelab computer

This detailed exploration uncovers the magnetic core memory system from a 1980 Spacelab computer, revealing the intricate engineering that enabled data storage in space decades ago. The article meticulously explains the mechanics of core memory, from its historical development to the ingenious 2½D architecture used in this specific module. It's a captivating read for hardware enthusiasts, demonstrating the sophisticated, robust, and now obsolete technology that powered early space missions.

9
Score
1
Comments
#2
Highest Rank
9h
on Front Page
First Seen
Aug 30, 8:00 PM
Last Seen
Aug 31, 4:00 AM
Rank Over Time
3225446711

The Lowdown

This fascinating deep dive examines the core memory module from a 1980 Spacelab computer, offering a rare glimpse into the advanced, yet ultimately superseded, technology that powered space missions before the age of ubiquitous semiconductor memory. The story not only showcases the physical components of this space-faring memory but also meticulously explains the fundamental principles and intricate engineering tricks that made core memory viable and robust enough for the harsh environment of space.

  • Spacelab, a European-built reusable laboratory for the Space Shuttle, utilized French-built Mitra 125 MS minicomputers, distinct from the Shuttle's IBM mainframes.
  • Its 128 kilobytes of RAM relied on magnetic core memory, where each bit was stored in a tiny ferrite ring, chosen for its non-volatility and radiation resistance.
  • The core memory system is physically a stack of seven boards, including driver boards, four core plane boards, and an interface board, cooled by direct conduction to the computer's side panel.
  • Core memory operates by magnetizing tiny toroidal cores clockwise or counterclockwise to store bits, with currents through wires flipping their magnetic state.
  • The crucial innovation of 'coincident current addressing' allowed selection of a single core in a grid by applying half-currents through intersecting X and Y wires, leveraging the core's hysteresis property.
  • Reading a core is a destructive process: the core is forced to a '0' state, and a change in magnetic field (indicating it was previously a '1') induces a current in a sense wire, after which the bit must be rewritten.
  • The Spacelab's memory, manufactured in 1980, represented a late and advanced iteration of core memory, boasting high density and utilizing a 2½D architecture that eliminated the traditional 'inhibit' line by using separate X driver circuitry for each bit.
  • Further optimization included a phase reversal technique, connecting pairs of vertical wires in U-shaped loops, effectively halving the number of required vertical drivers.
  • Each core plane board held 16K of 18-bit words (32 KB total), comprised of 294,912 lithium ferrite cores, and featured intricate wiring, including twisted pairs and 'bow tie' sense line crossings to minimize electrical noise.

Core memory's longevity in aerospace, owing to its inherent non-volatility and radiation resistance, extended its use well into the integrated circuit era, as exemplified by the 1980 Spacelab module. While ultimately surpassed by semiconductor memory due to advancements in density, cost, and power efficiency, this detailed look underscores the ingenuity of early computer engineering and the lasting legacy of core memory, even if only in the phrase 'core dump'.