Loading ...
Sorry, an error occurred while loading the content.

Re: [midatlanticretro] Found some Core Memory

Expand Messages
  • Dave McGuire
    ... They are probably (chips full of) selection diodes for the X/Y array. See DEC-8E-HR1C-D_8eMaint_Feb73.pdf pp.3-75 to 3/78 for a description of how that
    Message 1 of 12 , Jun 2, 2011
    • 0 Attachment
      On 6/2/11 12:59 PM, Dan Roganti wrote:
      > when I took a closer look, it does appear to be a raw core plane - is
      > that PG-rated ;)
      > As there's no support logic on there that you typically see in a memory
      > controller circuit. Other than what appears to be some kind of buffer or
      > level translator chips sparsely located on the edges of the core and
      > routed to the backplane card edge connectors.

      They are probably (chips full of) selection diodes for the X/Y array.

      See DEC-8E-HR1C-D_8eMaint_Feb73.pdf pp.3-75 to 3/78 for a description
      of how that works, and a few pages before that for schematics. (only
      using PDP-8 as an example because it's well-documented)

      -Dave

      --
      Dave McGuire
      Port Charlotte, FL
    • Jim Scheef
      I would expect anything from IBM to say I B M on it somewhere. In those days, IBM made everything themselves, so another company name would indicate this is
      Message 2 of 12 , Jun 2, 2011
      • 0 Attachment
        I would expect anything from IBM to say I B M on it somewhere. In those
        days, IBM made everything themselves, so another company name would
        indicate this is from somewhere else.

        Jim

        On 6/2/2011 1:42 PM, Dave McGuire wrote:
        > On 6/2/11 12:59 PM, Dan Roganti wrote:
        > > when I took a closer look, it does appear to be a raw core plane - is
        > > that PG-rated ;)
        > > As there's no support logic on there that you typically see in a memory
        > > controller circuit. Other than what appears to be some kind of buffer or
        > > level translator chips sparsely located on the edges of the core and
        > > routed to the backplane card edge connectors.
        >
        > They are probably (chips full of) selection diodes for the X/Y array.
        >
        > See DEC-8E-HR1C-D_8eMaint_Feb73.pdf pp.3-75 to 3/78 for a description
        > of how that works, and a few pages before that for schematics. (only
        > using PDP-8 as an example because it's well-documented)
        >
        > -Dave
        >
        > --
        > Dave McGuire
        > Port Charlotte, FL
        >
        >
      • s100doctor
        ... The brand name on the core board is Fabri-Tec . They manufactured cord planes for a number of memory board manufacturers. YOu ve got the cores but not the
        Message 3 of 12 , Jun 2, 2011
        • 0 Attachment
          , Dan Roganti <ragooman@...> wrote:
          >
          > I found some Core memory. It's from the same guy who has the mylar
          > papertape. He gave me 2 core memory cards. He mentioned that he remembers
          > this coming from an IBM system. The really cool thing about this is that it
          > assembled in a way where it folds up. I have some pics here. I'm hoping to
          > find out what system this is from.

          The brand name on the core board is "Fabri-Tec". They manufactured cord planes for a number of memory board manufacturers. YOu've got the cores but not the drivers and receivers. This board would plug into a card set with those electronics.

          I suggest you note the model number and look at all the "core memory board" pictures you can until you find a match for that model number. Then you'll have an idea of what "drove" these core planes.

          I don't advise trying to build analog core memory read/write electronics. Ask me privately about it if you want, I think a better "deal" would be to trade it for a complete core memory board set.

          Herb Johnson
        • Dave McGuire
          ... Hey, it s finicky and fiddly, but it can be done (and has been), and there are many design examples out there. -Dave -- Dave McGuire Port Charlotte, FL
          Message 4 of 12 , Jun 2, 2011
          • 0 Attachment
            On 6/2/11 7:54 PM, s100doctor wrote:
            > I don't advise trying to build analog core memory read/write
            > electronics. Ask me privately about it if you want, I think a better
            > "deal" would be to trade it for a complete core memory board set.

            Hey, it's finicky and fiddly, but it can be done (and has been), and
            there are many design examples out there.

            -Dave

            --
            Dave McGuire
            Port Charlotte, FL
          • Jeff Jonas
            ... But can they handle the temperate range as things heat up, and the density of the one being considered? The Arduino core project I saw recently is only 8 x
            Message 5 of 12 , Jun 5, 2011
            • 0 Attachment
              > > I don't advise trying to build
              > > analog core memory read/write electronics ...

              > Hey, it's finicky and fiddly, but it can be done (and has been),
              > and there are many design examples out there.

              But can they handle the temperate range as things heat up,
              and the density of the one being considered?

              The Arduino core project I saw recently
              is only 8 x 8 and very sparsely populated.

              Yes, today's DSPs can do things
              that required dedicated hardware not long ago,
              but the algorithms and reliability looks tricky.
              If that's handled properly, I'd love to see the new ways!
            • Jeff Jonas
              ... The core I salvaged from the IBM 1620 and IBM 1130 were merely green frames that stacked together. No IBM anywhere. Even back then, IBM outsourced SOME
              Message 6 of 12 , Jun 5, 2011
              • 0 Attachment
                > I would expect anything from IBM to say I B M on it somewhere.

                The core I salvaged from the IBM 1620 and IBM 1130
                were merely green frames that stacked together.
                No "IBM" anywhere.

                Even back then, IBM outsourced SOME things.
                The IBM system 1130's plotter is from Calcomp
                and the paper tape reader from Tally.
                The semiconductor RAM on the IBM system 38 is from EMM.

                > In those days, IBM made everything themselves,

                I wish I could find photos from the IBM Kingston facility's
                Historical Society. It showed a HUGE foot-locker sized
                core memory module for the IBM 360,
                with photos of all the steps of manufacturing, weaving
                and even re-weaving to replace a failed core!

                -- jeffj
              • Dave McGuire
                ... Core runs hot, sure, but this is well understood and easily handled. If Dan has trouble getting that core plane running, it won t be because it runs too
                Message 7 of 12 , Jun 5, 2011
                • 0 Attachment
                  On 6/5/11 6:46 AM, Jeff Jonas wrote:
                  >>> I don't advise trying to build
                  >>> analog core memory read/write electronics ...
                  >
                  >> Hey, it's finicky and fiddly, but it can be done (and has been),
                  >> and there are many design examples out there.
                  >
                  > But can they handle the temperate range as things heat up,
                  > and the density of the one being considered?

                  Core runs hot, sure, but this is well understood and easily handled.
                  If Dan has trouble getting that core plane running, it won't be
                  because it runs too hot.

                  > The Arduino core project I saw recently
                  > is only 8 x 8 and very sparsely populated.

                  The sparseness of that design surely had little to do with heat.

                  > Yes, today's DSPs can do things
                  > that required dedicated hardware not long ago,
                  > but the algorithms and reliability looks tricky.
                  > If that's handled properly, I'd love to see the new ways!

                  Nah. Core is pretty reliable when you get a design running. Getting
                  it to that point takes work, though. Most, if not all core systems have
                  a thermistor near a core plane to adjust X/Y drive currents...I suppose
                  one *could* do that with a DSP, but you'd be replacing a simple,
                  easily-modeled circuit with maybe half a dozen transistors and a handful
                  of resistors with a very complex chip and hundreds, if not thousands of
                  lines of code.

                  -Dave

                  --
                  Dave McGuire
                  Port Charlotte, FL
                Your message has been successfully submitted and would be delivered to recipients shortly.