Name that Ware August 2012

The Ware for August 2012 is shown below.

This is yet another ware recovered from my childhood basement stash. Unlike the previous ware, I actually have no idea what computer this comes from. Clearly, an IBM of some type; probably an old mainframe, from the early ’80s. But these boards were rescued from a dumpster with no sign in sight of what machine it comes from. I’m hoping someone can definitively resolve the question of which machine this came from, and what function this particular board had.

I also had one other question I’m hoping someone can answer — why did all the boards from this machine feature a regular array of holes, as opposed to vias drilled on demand? I have my own hypotheses, but it’d be nice to hear from someone who knows the answer!

23 Responses to “Name that Ware August 2012”

  1. J. Peterson says:

    One guess on the regular grid of holes: It made the manufacture of the PCB quicker. Rather than wait for an CNC drill head to move from hole to hole, they just punched the base material with a regular grid and routed around it. Coming up with a custom die punch for each board would likely have been prohibitively expensive, even for IBM.

  2. Adrian says:

    From a quick search, this looks like an exemplar of http://en.wikipedia.org/wiki/IBM_Solid_Logic_Technology or successors like Monolithic System Technology.

    The examples at the bottom of this gallery have similar PCBs:
    http://ibmcollectables.com/gallery/album128

  3. Qwghlm says:

    When I was in school – late 80s – I’ve got a bunch of very similar boards as a cheap source for TTL chips. They were sold per kg in a second hand electronics shop. Pretty hard to unsolder, but most important: cheap.

    Don’t know what devices they really came from – but thanks for refreshing the memories of my early hardware hacks

  4. Brian says:

    Don’t know much about IBM hardware, but the crystal frequency
    suggests baud rate generator; that plus the custom marked Motorola
    parts suggest an IBM video terminal or terminal controller.

    From the early eighties date codes, my guess would be something
    from the 3270 terminal family.

    I used those in college in the early eighties, but never dissected one.
    ( However, writing and debugging LISP programs in a 3270 screen
    mode editor on MTS still occasions parenthetical nightmares.)

    The via array could be for testability ( all pins and via layer changes
    pass through big honkin’ 0.1″ spaced pads ideal for probing ).

    – Brian

    • Brian says:

      Following up on my earlier 3270-family-terminal-or-controller guess,
      the card sizes used within a 3278 terminal are too small [1] to match,
      haven’t tracked down the terminal controller card outlines yet.

      [1] 3278 card illustrations, page 130 of :
      http://bitsavers.org/pdf/ibm/3270/fe/SY27-2510-6_3278_Display_Station_Models_1-5_Maintenance_Information_Jan81.pdf

    • Brian says:

      No luck finding an 8-connector wide candidate for the board yet,
      but I did find a 1981 IBM Journal describing the PCB technology
      with a timeline:
      http://www.kmitl.ac.th/~kbittibh/home2/BOOKS/Packaging/IBM.pdf

      page 7 of pdf:
      ” A back panel (a printed circuit board) of the same
      ” construction was used to accept six of the cards
      ” with 24 contacts in the width dimension and
      ” 78 cards total (Fig. 3). Later, wider cards with
      ” 96 terminals were developed.

      ” The grid [all holes on a pitch of 0.125 in. (0.317 cm)]
      ” was standardized for mass production drilling.

      – Brian

    • Brian says:

      ( hit send too soon…)
      from page 8 of pdf:

      – Standard pattern of holes & pads helped with plating.

      – PCB evolution, switching from 0.125″ to 0.1″ :
      ” printed circuit board required three lines per 125-mil
      ” grid and four layers of interconnection.

      ” These MST developments were largely refinements,
      ” a growth from four layers to six layers, including
      ” power distribution planes. This was reduced to
      ” a 100-mil grid by 1973 to accommodate the new
      ” metallized ceramic modules with 48 to 76 pins
      ” per square inch and the vendor DIPS

      And a good closeup of a MST PCB in Figure 13-c, page 9

      -Brian

    • Andy Johnson says:

      It looks a bit too long to fit inside the casing of a 327x.

    • Brian says:

      Andy wrote:
      >
      > It looks a bit too long to fit inside the casing of a 327x.
      >
      The service manual agrees with you, showing a four-wide
      board as the biggest 3278 card (see follow up post of 8/16).

      My personal recollection is that the 3277/3278 family
      of terminals were large enough to hold the ‘ware board,
      had they needed to :)

      Other notes:

      ——————–
      IBM Part numbering

      It would help to find a cross reference showing logic
      functions for the IBM MCMs and custom marked DIPs.

      The best I’ve found so far is a manual for the older
      16-pin MST devices:

      http://www.bitsavers.org/pdf/ibm/logic/MST-2_Module_Data_Apr74.pdf

      ( Aside: I can’t recall ever seeing a BGA device with
      a spiral pin numbering scheme before, see page 8 of pdf;
      oddly, there’s also an alphanumeric reference grid on
      the same drawing )

      ——————–
      Similar vintage ( by appearance )

      The closest boards I’ve found, technology-wise, are from
      the mid-70’s IBM 5100 family:

      http://www.classiccmp.org/dunfield/ibm5100/disass.htm

      The 5100 boards have similar mix of custom marked
      standard logic + IBM logic as does the ‘ware board.

      But they are also too skinny to match the ‘ware.

      ——————–
      ‘Ware board design and manufacture date range

      – Packaging article dated 0.1″ board introduction to 1973
      – oldest IC date codes are from early 1981

      Mainframe IBM CPU’s proper within this timeframe appear to
      have used much higher density packaging.

      So I think some sort of I/O or peripheral interface card
      still seems a likely function.

      Unless someone comes along who recognizes the card, an exact
      ID seems tough from the online resources I’ve found so far.

      – Brian

  5. Tracy Hall says:

    Looks like a technique I used on a couple boards in that era – it’s not a PCB, as such, but the base for an automated point-to-point wiring system – used thin insulated wires channel-routed by machine to make the interconnects.

    A follow-on to manual/machine wire-wrap; Insanely expensive compared to PCB’s ($100’s to $1000’s per board), but pretty damn stable, customizable and fast-produced, as long as bus speeds are fairly low (10MHZ and much less). Rework consisted of finding the tiny wire you wanted to get rid of, and cutting it with a VERY fine xacto knife.

    I’ll see if I can find the name, but this is pre-internet (barely BBS’s, ARPAnet time)

    • michael says:

      no, it is a pcb, I used to do solder rework on those in the mid 80s. no pins, no wires, ‘cept the printed ones.

      if you look very carefully between the pads, you can see the traces.

      • Tracy Hall says:

        As I said, it’s really a hybrid: PCB for the holes and planes, but very thin wires machine channel routed between the pins, then sealed. Those “traces” are the wires. Like I said, I built a few short pre-production runs with these. No pins, true. wires “tacked” to the pads.

        Since each was “custom built” (other than the interface/connector pins), you could make design changes rapidly; you could rework (cut the wires). Great stuff, at the time.

  6. Arargh says:

    Looks something S/360 SLT cards:

    http://members.optushome.com.au/intaretro/Packages.htm
    http://en.wikipedia.org/wiki/IBM_Solid_Logic_Technology

    But I haven’t a clue as to what the thing in the picture is. :-)

  7. dan says:

    I’d guess that it is a printer head driver card from a RPL IBM system.

  8. Dre says:

    I took apart boards like this in my childhood. The chips with the silvery top have a strange transparent liquid substance in them, I hope it’s not poisonous…

  9. SP says:

    The transparent gel is just silicone. I dismanted one client when i was young. You can see an image of how is a decapped chip (the silverish things) at the image http://ummr.altervista.org/IBM_ferrite_core_storage_rings_102649703_lg.jpeg near the ferrite cores. These chips are IBM chips, this is a board for the IBM System 360.

  10. olle says:

    Looks very much like an old RS/6000 server board… http://en.wikipedia.org/wiki/POWER1

  11. Mike says:

    http://ummr.altervista.org/ibmtechviews.htm shows a memory board with the PGA sockets that this one has

  12. Steve Shockley says:

    I’m going to take a shot in the dark and say it’s from a Selectric, maybe an IBM MC Composer. I found http://www.weller.to/wcc/ibm/wccibm.htm (search for 6212) which has a board with the 8-wide connector top and bottom.

  13. Ali Khurram says:

    So at least I have found that I am not the only one in this universe who kept a similar board scavenged from a scrap yard (early 90’s) for such a long time to find from what machine it came from. I thought (in those days) that maybe it is from some IBM mainframe computer used for aerospace or military purpose.

  14. Alex says:

    Three processors and seven ram chips, weird. I’d love to see the other side.

  15. Nick says:

    I used to work at IBM as a logic designer in the 70s and 80s. That card would be an 8W3H card in the vernacular. The pcb technology I used was up to 4 signal planes, 2 power planes and a wiring density of three lines/channel – where a channel is the gap between two square pads 0.1″ apart. I know the big CPUs used more signal planes. The connectors top and bottom were 0.125″ spacing since they were from a heritage of IBM SLT technology .. that spacing for the motherboard kind of stuck. The cards plugged into boards, the boards which carried up to 20 cards were put into gates, the gates into frames .. an elegant and consistent heirarchy for the design.

    I used a number of IBM technologies in my time .. Dutchess was a low density TTL family gate array of about 100 gates per module. (one of those aluminium cans there). Golf extended to about 300 gates per module, Tango increased by another factor of three or so. There were separate product lines packaged in the same way for read-only memory and for random access memory. My machine designs used ESROS and CPROS, (read only store – using memory was a little too close to human for IBM), which reached about 72K or 144K bits. My favourite RAM was Concorde – I used the 8K x 9 modules (which consisted of four internal chips), a 200ns or so DRAM. For 1977 that wasn’t too shabby – I borrowed six modules for a 48K RAM card for an S100 machine at home .. I probably had ten times as much RAM as anyone else in their homebrew!