I would hope not!. To make a core memory you need a magneitcally hard material that will stay magnetised. But for a transformer or inductor, you want a soft (magneticaly) material. -tony

Grodon asks about making core memory: There was a construction article in one of the electronics magazines a few years ago about making a one-bit core memory using a steel hex nut. It takes a lot more energy to flip the magnetization of a steel hex nut than a "proper" magnetic core, so it's not very practical. Eric

On 23 May, 2005, at 09:20, Tom Jennings wrote: At the end of the core era (or maybe a little after the end), NCR used "knitting machines" to make core planes. They even had an accounting machine where the entire logic was core based (inhibit core was their name for the technology). The last magnetic memory (I think) for the NCR Century was "plated wire". They used cobalt-plated wires instead of rings, it was much easier to manufacture but the S/N ratio on the read wire was terrible. NCR was much into cobalt, they also used cobalt-plated disks instead of the usual iron oxide coatings of the day for some high-end drives. The disks were shiny blueish black. If you just want to play and make a 4 x 4 core or something, I My guess is that the ferrite rings made by TDK for noise suppression should work OK. They will obviously be lossy (they are designed to be), but they should have fairly high hysteresis. As long as you do not work them too hard, they should not get hotter than a P4 Prescott :-) -- -bv

Tore wrote about threading core memory: Actually they used automation for the X and Y wires starting fairly early on, because that was cheaper than human labor. The cores were vibrated into positions on a metal plate with indentations, then long needles (like hypodermics) were propelled through a row or column of cores. The wire that was threaded through the needle was soldered to the frame, the needle withdrawn, and the wire soldered at the other end. It's the zig-zag sense/inhibit wire that was done manually for many years, because it was much harder to automate. A great book; I'd go so far as to call it "required reading" for anyone interested in computer history. But for more details on core memory, see "Memories that Shaped an Industry: Decisions Leading to IBM System/360" by Emerson Pugh, one of the coauthors of the former book. Eric

Tore> On Mon, 2005-04-25 at 13:39 -0400, Allison wrote: Tore> Are these cores suitable for use in making data memories? Tore> Ie. could one hypothetically make a core store today, using Tore> those cores? Has anyone done this? Do the current-day cores Tore> have better properties for use as memory than the ones used in Tore> past-day cores? Tore> Forgive me if this is a stupid question. Not a stupid question at all. No, those cores are not suitable for core memories. The issue is the "hysteresis curve" -- which shows the magnetization of the core as a function of the applied field, or something along those lines. Cores used for transformers have hysteresis curves that are very nearly a straight line. Cores for memory have a curve that looks more like a rectangle. The difference tells you whether the material remains magnetized after the applied field goes away. For transformers you want that NOT to be the case. For memory cores, you do. So the material is more like something you'd use to make permanent magnets. paul