As the dashing officer shown above will tell you, early data processing machines and ADP systems employed two types of magnetic cores for memory and other purposes. This 1961 U.S. Army training film is an introduction to the properties of ferrite cores, which are commonly made from nickel alloy and other magnetic materials. As this is only part one of a series, the metallic ribbon type of magnetic core is covered in some other segment we have yet to locate.
The use of magnetic cores for random access memory was built upon transformer theory and provided a rugged and low-power solution until the semiconductor came into vogue. Before that time, the humble ferrite core served many uses and did so very well. The Apollo Guidance Computer had erasable magnetic core memory, and much of its software was stored in core rope memory.
The film covers a lot of theory and does so clearly and concisely. It begins by explaining what a magnetic core is and why it’s used, and then moves on to describe how the cores are used to store bits and the method by which they can transfer information to other cores. Along the way, it provides background on bi-stable devices and provides explanation of magnetization behavior in terms of magnetizing force and flux density.
Continue reading “Retrotechtacular: Core Competencies”
We’ve seen NAND and NOR logic gates – the building blocks of everything digital – made out of everything from marbles to Minecraft redstone. [kos] has outdone himself this time with a logic circuit we’ve never seen before. It’s based on magnets and induction, making a NOR gate out of nothing but a ferrite core, some wire, and a diode.
The theory of operations for this magnetic NOR gate goes as follows: If two of the input windings around the core have current passing in different directions, the fields cancel out. This could either be done by positive or negative voltages, or by simply changing the phase of the winding. To keep things simple, [kos] chose the latter. The truth table for a simple two-input, one-output gate gets pretty complicated (or exceedingly cool if you’d like to build a trinary computer), so to get absolute values of 1 and 0, a separate ‘clock’ winding was also added to the core.
One thing to note about [kos]’ gate is its innovation on techniques described in the relevant literature. Previously, these kinds of magnetic gates were built with square ferrites, while this version can work with any magnetic core.
While this isn’t a very practical approach towards building anything more complex than a memory cell, it is an exercise of what could have been in an alternate universe where tube technology and the transistor just didn’t happen.
[Kos] tipped us about an article he made presenting his experiences in designing and implementing a core rope memory. This magnetic read-only memory (ROM), contrary to ordinary coincident-current magnetic core memories (used for RAM), uses the ferrite cores as transformers. If you look at the picture above, you’ll count 7 of them. This sets the memory word size (7bits). A new word is added to the memory by passing (or not) a wire through the ferrite holes. If you then pass an alternating current through this wire, a current will be induced (or not) in the other wire turned 30 times around the ferrite (alias transformer secondary).
In [Kos]’s setup, an input pulse of 5V generates output pulses of 15V. For demonstration purposes, he “wrote” a simple program that lights up digits in a seven segment display. Therefore, different numbers will light up depending on which wire he uses to pass the AC current.
These days core memory hacks are few and far between. But looking at this one, and the one we saw in August, makes want more. If you know of any others don’t hesitate to send us a tip.