Relic Information query matches
Matching on query: Unique id = 6/97, Name = any, Associated machine = any, class = any
Unique id/year of acquisition: 6/97
Name: Ferrite core memory plane
Ferrite core memory plane
Other nos on object: none
Description: Square Plessey ferrite core memory plane. Each side of the square bakelite (?) frame has around 60 metal contacts on the outside. Inside the frame is a grid of thin wires, with each junction surrounded by a thin wire ring. Several contacts are bent and the frame is slightly scratched.
DW: It's got 1949 on it. 1949 is a mislabling - it didn't exist in 49. Yes, perhaps we could cross this 1949 out. This is a typical ferrite core memory plane. It contains 32 x 32 cores with probably 3 wires through each core. These were the first reliable fast memories larger cores made for computers. The Williams store was about as fast but the tubes had a life of about 3000 hours and they were extraordinarily sensitive to noise. Every single bit was stored on a ring of ferrite and there were three wires passing logically with an X drive, a Y drive and a sensor and if both the X and the Y drive were applied the core would switch if it could and the third wire picked up the signal. The cores had an inherent square loop property which made this possible, if you only half disturbed the core and did not switch magnetisation but if you gave a full write it could switch magnetisation to the way you were switching. So, the mode of operation was to read a bit and if there was a signal coming out you would then write that bit back so it was a so-called disruptive read. But the cores could all be tested before they were assembled and they're made of ferrite and the only thing of remain of ancient civilisation are the ceramics and ferrites are ceramics so they are very long lasting and it's just made of and ferrites so they were extraordinarily reliable by the standards of the day.
Q: Where would be machines be used?
DW: Essentially they were used on all machines starting round about the late fifties. There was nothing to compare with them and I can remember during the following two decades about every year somebody that announced a new way of storing information which would supercede the ferrite core but in fact it took the arrival of integrated circuits before this happened because the ferrite became cheaper and they were hand wired and then they were wired in the far East and so their cost came down continually and their size reduced and their speed increased and so they lasted a surprisingly long time.
Q: Into the late nineteen sixties?
DW: I can't actually remember the date, I think they went on to the seventies, the early space rockets used ferrite core because they were reliable.
Q: Were there any one person or team behind the original design of these?
DW: They were designed at MIT by someone whose name I forget and is well known for it in I think the late fifties. And the Whirlwind I think was the first machine ever to have a ferrite core which replaced the electrostatic storage which was not very good. For memory this plane would store one bit and you would have possibly 32 parallel planes like this assembled into a compact store.
Q: Would this have been made locally?
DW: The EDSAC 2 ferrite stores were wired locally, they even made a little tester to test every individual core automatically and verify its properties. The EDSAC 2 also used ferrite cores in two different ways: one was for the straight memory, the second was by passing an additional inhibiting wire through a core, you could prevent it switching and this allowed a single core to potentially store up to four bits and this was used for the read only memory where the information was given by the wiring and this was attractive because they'd use the same amplifiers, the same type of drivers and everything else as a main store which is why EDSAC 2 has 768 words of fixed-in store. That store was used to hold typical routines like cosine, logarithms and assembly routines, printing routines, solution of differential equations and other things like that. Some of which you might even find on a hand calculator nowadays. The third use of the ferrites were the 13 mm, much which were assembled into a frame of I think 1024 cores and these were used to form the memory of the microprogramme which ran, and it gave control to the rest of the machine. The microprogramme cores worked rather differently to memory cores, again it used the threshold to enable switching on one core out of an entire plane. There were little turns of three wires to pick up the signal these drove directly the gates were used in the control.
See also: 28/97
See also: 29/97
See also: 58/97
See also: 59/97
See also: 150/01
Number of matches = 1
Copyright © University of Cambridge Computer Laboratory, 1999. All rights reserved.