On 18/04/2019 05:47, Grant Taylor via cctalk wrote: This is great and I will look into this. But my original request was for something a little bit different. That is, a properly specified core memory plane in working condition, because I would very much like to interface with it from a modern computer. Yes, I agree. And somebody from a DEC list I'm on said that the number of connectors and their spacing isn't anything they've personally seen before. Andrew -- OpenPGP key: EB28 0338 28B7 19DA DAB0 B193 D21D 996E 883B E5B9

On 2019-Apr-17, at 9:47 PM, Grant Taylor via cctalk wrote: It doesn't plug into a system bus, it would plug into a dedicated slot connected to other dedicated boards. There's piles of circuitry necessary for it be fully functional: decoded x/y drivers, inhibit drivers, sense amplifiers, word latch, cycle sequencer. It's a 4-wire 3D planar array. By topology and construction I would guess it date it from the 60s. Into the 70s most (American) core production had moved on to simpler weave topologies (the crosshatch sense winding, as in this unit, was very laborious to weave), (although it is not precluded from being from the 70s). although here the earlier physical stack for the 3rd dimension is flattened to a planar array.

Jussi Kilpelainen's page cited above ( https://www.tindie.com/products/kilpelaj/core-memory-shield-for-arduino/) refers to the work of Ben North and Oliver Nash to create another core memory shield for Arduino Unos. Their site inspired Jussi to create his shield kit, which can be viewed at: http://www.corememoryshield.com <http://www.corememoryshield.com/report.html> Ben's and Oliver's home page contains a link to the files: http://www.corememoryshield.com/Core-Memory-Shield.zip It also contains a link to a page with the detailed explanations and schematics for core memory circuitry needed to read and write to core: http://www.corememoryshield.com/report.html Wayne Holder shows how to build the most minimal circuit using an SN754410 driver IC (http://focus.ti.com/docs/prod/folders/print/sn754410.html, about $2.50 each for one, and a bit over $2.00 each for 10) and an Arduino Nano (about $5.00 each for one, and around $3.50 each for 10) to address one bit of core memory and, more importantly, how to expand the circuit to address larger numbers of bits/cores using more SN754410s: https://sites.google.com/site/wayneholder/one-bit-ferrite-core-memory There's enough information on Wayne's page to build the circuitry to access any arbitrary number of cores, with the number of cores/bits being addressable increasing by a factor of four for each doubling of the number of driver ICs. ePrey has lots of old core memory planes available for as little as $30 each for about 500 bits, with intact wiring (from Russia or the Ukraine, which developed much of the Soviet Union's weapons systems and advanced commercial electronics). So, you can just wire the above circuits to their pins instead of going through the hate and discontent of testing a ziplock bag of 500 ~2,000 teensy-weensy cores (from Bulgaria) to find enough with the right properties, and then having to weave your own plane. All the Best and Enjoy! Jim

On 4/18/19 9:30 AM, Chuck Guzis via cctalk wrote: The 1401 guys at CHM were working on one using a real 701 tube. I don't think it was ever finished. @tubetimeus built a small core array with Bulgarian cores https://twitter.com/TubeTimeUS/status/1053424445463752704 Discussion of the EDSAC rebuild https://www.youtube.com/watch?v=xGEAPVCuwvY Haven't heard if anyone has tried to get one of the Russian MR lines on eBay to do anything, or if anyone has a running Packard-Bell PB-250

On 4/18/19 9:42 AM, Al Kossow via cctalk wrote: I don't expect that any EBAM has survived--I think all of the stuff I saw at CDC ADL was scrapped. Seems that the technology is all but forgotten today: https://bit.ly/2KOOl82 It's mentioned in a proposal by Neil Lincoln back in the 1970s: https://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/19770084287.pdf I've long thought that a scan converter tube (e.g. 7912) might make a good electrostatic memory. --Chuck

Well, I am. And I have a LOT of 8K core system modules (planes and drivers) from old Stewart-Warner (I think) vector graphics terminals -- Will

I don't believe there is a simple non-destructive way to read the state. If you could remove the cores, I believe you could put each core in a weak magnetic field. As the field passes from side to side, one should be able to determine the direction of the saturated cores because one side would allow more of the field to enter the core while the other side would act as non-magnetic. As for reading the data, you'd need to experiment to determine the minimum current that particular core required to flip a bit. With careful adjusting, I suspect one could sample the first bit. Once knowing the levels used, the entire array can be read. Dwight ________________________________ From: cctalk <cctalk-bounces at classiccmp.org> on behalf of Jim Brain via cctalk <cctalk at classiccmp.org> Sent: Thursday, April 18, 2019 10:19 AM To: cctalk at classiccmp.org Subject: Re: Plane of core memory I am the enviable owners of a plane of memory (procured a few years back at VCF-East, when there were a bunch of 32K? boards int he consignment pile. (Sorry, not currently interested in selling :-) But, I am thankful for the links, as I have wanted to interface this with a CPU or PC of some kind. A few questions, though: * Is there a way to "read" the core non destructively using any kind of passive method (I know, it would be tedious, no doubt, but I just feel like I should "backup" the core before I go messing with it)? * Along with the above, might there be a way to extend the passive read to be a worthy "exhibit" I could take to shows? Core memory is impressive just to look at, but reading it out using the PC and displaying the contents is so easy to fake that I think people will assume the core memory is not really being used. Some way of showing the actual magnetic changes in a small matrix (the large plane I have is probably not a good candidate unless there is a way to show such minute cores fields) would I think make the exhibit far more interesting, especially if I arranged the grid in a square and created a really slow version of something like "Tetris" on the plane. Ideas appreciated. Jim

I'm having trouble figuring out what typical magnetic field strengths on the cores would be, but I wonder if you could wave some ferrofluid over individual cores and get movement. Some ferrofluid particles are apparently on the order of 10 nM in diameter, so wouldn't be all that hard to move with a weak magnetic field. Showing the read pulse and sense transition waveforms on a 'scope might be of some interest. De

How was the CDC EBAM different from the other memory tubes, like the Radechon? -- Will

On 2019-Apr-18, at 9:30 AM, Chuck Guzis via cctalk wrote: - The Manchester Baby replica recreated Williams tube memory (aka Williams-Kilburn tube memory). Williams tube memory uses conventional oscope CRTs, so it wasn't that difficult. (i.e. it's Williams 'tube memory', rather than 'Williams tube' memory.) Holding-beam CRT memory ala Whirlwind would be another matter as the tubes were highly specialised. - The EDSAC replica is using a new-production magnetostrictive delay-line memory to substitute for the mercury tanks.

On 04/18/2019 04:49 AM, Brent Hilpert via cctalk wrote: Make that EARLY '60s. As soon as somebody figured out that you could combine the sense and inhibit wires, everybody immediately went to 3-wire planes. Jon

I believe I read they weaved the planes this way to minimize crosstalk, EMI or heat. =] On Thu, Apr 18, 2019, 1:13 PM Paul Koning via cctalk <cctalk at classiccmp.org> wrote:

? inverse cube (?) ? depending on the geometry you are talking about? Real physicists please set me right if I have this wrong, but I think only radiating and static point fields (like electric or gravitational) from a finite source fall off as inverse square. That?s easy to see, it?s the same effect crossing the total area of a sphere centered at the source. Area on the sphere goes up like square of radius, so intensity has to go down like the square. Magnetic field (from a finite source) I think goes down like the cube of the distance - north and south poles of the source tend to cancel better as the apparent angle between them gets smaller, in addition to the above effect. (That is a mnemonic, not a real explanation.) On the other hand, if you were talking about field around a wire with current in it (a non-finite source, at least locally), then it *is* inverse square for the magnetic field, where inverse square refers to distance from the axis of the wire. - Mark

On 2019-Apr-18, at 8:47 AM, Jon Elson wrote: Well, not everybody - I have a 4-wire, diagonal-sense module with IC date codes extending from 1974 to 1978. Certainly it's an outlier, on the tail-end of the distribution curve of production of such design, but that's why I expressed dating it to the 60s as a likelihood, not an absolute surety. (The ICs are 711 comparators and diode arrays. The 711 does go back to at least 1969 so it's conceivable the design of the module goes back to the late 60s, although that's seems less likely considering production into 1978.)