Image Map Image Map
Page 5 of 6 FirstFirst 123456 LastLast
Results 41 to 50 of 60

Thread: Wire wrap

  1. #41
    Join Date
    Jan 2007
    Location
    Pacific Northwest, USA
    Posts
    30,345
    Blog Entries
    20

    Default

    I remember when that came out. It struck me as hard to rework, especially on prototypes. But I can't claim to have used the stuff.

    For another take on related prototyping, have a look at Elm-Chan's work here

    When I was experimenting with early IC logic, I etched a single-sided PCB with pads that corresponded to the early flatpack and TO-100 layout (DIP was still very very new) and then wired them with 24AWG magnet wire. Something like this:



    IC integration was pretty low then; I remember being very proud that I scored a bunch of Fairchild 923 RTL in flatpack. Gotta make registers from something.
    Last edited by Chuck(G); March 2nd, 2019 at 01:51 PM.

  2. #42
    Join Date
    Feb 2017
    Location
    Chilliwack, BC, Canada
    Posts
    152

    Default

    Quote Originally Posted by Chuck(G) View Post
    For another take on related prototyping, have a look at Elm-Chan's work here
    Oh wow! Makes me wanna have a go at it!
    My vintage systems: Tandy 1000 HX, Tandy 1000 RSX, and some random Pentium in a Hewitt Rand chassis...

    Some people keep a classic car in their garage. Some people keep vintage computers. The latter hobby is cheaper, usually takes less space, and is less likely to lead to a fatal accident.

  3. #43
    Join Date
    Sep 2012
    Location
    Kent, England
    Posts
    129

    Default

    My attempt to build a replica of a Honeywell 200 computer is my first wire-wrapping project, so I hope that I'm getting it right. Certainly so far the circuits seem to be reliable. Here is a portion of the wiring for the 8k main memory ...

    Wirewrap detail 2.JPG

    ... and here is the entire two foot square standard 200 series backplane. The original transistorised basic Honeywell 200 central unit contained three or four of these. Note the candy-striped backing paper that Honeywell thoughtfully provided as a guide for locating particular rows of logic modules. Without that one could easily go mad trying to avoid making the wrong connections. In fact even with it the pins are long enough for parallax errors to creep in if one doesn't view them directly on. The plastic backing paper also covers up the uninsulated power bus lines hidden beneath it.

    H200 backplane 2.JPG

    One backplane accommodates 160 logic modules and over sixty transistorised ones are needed just to operate the magnetic core memory, so they alone occupy one half of the space. I am currently designing and developing the modules for the control memory that contains the address registers, which I am making using 240 flip-flop ICs. I have a second backplane to use when this one fills up but I will have to be careful choosing which signals can be passed from one to the other over relatively long lines.

    These backplanes came from old equipment, so I had to strip off all the original wiring. I did this laboriously using a simple finger-twirled pencil-style hand tool, although I discovered that Honeywell had attached some wires anticlockwise around the pins instead of clockwise, so even that wouldn't work on them. My wife and I sorted the wires into bundles by length so that I could reuse them and I have a custom computer application that I wrote that helps with doing that as well as other problems that could easily arise. The application is something like those used to design printed circuit boards except that the components are plug-in logic modules and the connections are wires instead of PCB tracks. It has the equivalent of the schematic design and physical layout parts encountered in PCB design applications, so I can do the logical design of the connections between modules as an armchair exercise on my laptop and then do the actual wiring in my workroom. The key issue was to make the wiring task as mindless as possible to avoid errors, so for each signal net a list of the pins to be connected is shown for me to choose in which order to wire them together. To permit later changes it is important to avoid daisy-chaining the connections so that ones in the middle can't be removed, the preferred method being to create two distinct layers of overlapping connections on successive pairs of pins like brick-work. The graphic display of all connections on the pins in order up them makes it clear whether I've done this right and also this record makes it possible to identify which wire on any pin is which. If one wants to disconnect a wire from a pin with several attached later and the wires are all the same colour it helps to know the order in which one put them on it because tracing a wire through the dense rat's nest is virtually impossible.

    I did have to make some changes in the wiring to get the memory working, not because I made the wrong connections but because I didn't know that my later core memory modules manufactured in Ireland had to have the current in some windings running in the opposite direction from the earlier ones first used in America. Core memories needed elaborate wiring patterns to cancel out unwanted induction between windings and evidently a later improvement in the design required some windings to be connected in reverse to make the noise cancelling work correctly. After I'd worked out the correct wiring by trial and error involving several rewiring operations, including putting some connections back the way that I had them originally, a former Honeywell field engineer with whom I was corresponding mentioned to me that the Irish memory modules were different, so I told him that I'd already discovered that!

    With so many connections to make I use an OKI electric wire-wrapping gun with a bit that does "modified wraps", which means that the first turn around the pin is insulated wire to take any strain in the connecting wire itself. My computer application tells me not only which pins to connect, working from my signal net schematics, but also when I choose two pins to connect it tells me what colour the zone in the candy-striped backing is for each and the length of wire needed for a straight connection, an orthogonal one or a semi-diagonal one following the natural lie of the pins. This enables me to select a wire from my presorted bundles, trim it to the right length and strip the insulation from both ends before attaching either to a pin. If the wire appears to be the wrong length then I may be trying to connect the wrong pins, so this approach reduces the possibility of errors. As soon as a wire is connected I notify the application so that it updates its records. By following this strict regime of schematic design first and then recording each wire as I connect it I can be confident that my computer record always corresponds to the physical backplane wiring. If I have to disconnect wires the application records these actions as I do them as well, so I can interrupt my work at any time at all without losing track of what I've done. I can also ask the application to check the record of the physical wiring against the schematic plan to identify differences and work outstanding.

    Programming and testing this application took me a while but it was worth it as I have had no wiring errors apart from the Irish problem and we Brits are used to having those occasionally. Anyone following our current efforts to leave the EU will be well aware of that. Making this replica computer would be pointless if I didn't also create all the documentation required for it as well, so having all the backplane wiring reliably documented in a standard relational database, as it is, is in itself valuable for the long term as well as for my day to day construction work. Yes, it's fun to create these gargantuan wiring rat's nests and discover that they actually do the job required of them, but if they ever don't then ... pray that your documentation is well organised.
    Last edited by RobS; March 14th, 2019 at 09:44 AM.
    Rob - http://www.honeypi.org.uk
    The Internet is a winch to get your project off the ground ... but always have a parachute handy.

  4. #44
    Join Date
    Jan 2007
    Location
    Pacific Northwest, USA
    Posts
    30,345
    Blog Entries
    20

    Default

    Thank your lucky stars that you didn't have to deal with taper-pin wiring mats!

  5. #45
    Join Date
    Dec 2005
    Location
    Toronto ON Canada
    Posts
    7,075

    Default

    [QUOTE=RobS;561810]
    Quote Originally Posted by RobS View Post
    ... I have had no wiring errors apart from the Irish problem and we Brits are used to having those occasionally.


    Great description; thanks! Kudos for tackling this project and the way you're going about it!

  6. #46
    Join Date
    Sep 2012
    Location
    Kent, England
    Posts
    129

    Default

    Quote Originally Posted by Chuck(G) View Post
    Thank your lucky stars that you didn't have to deal with taper-pin wiring mats!
    They sound more like a bed of nails than even my backplanes and as the pins on those do have bevelled ends I have to use a mat to protect my hand when it is up against them while working or else it gets perforated.
    Rob - http://www.honeypi.org.uk
    The Internet is a winch to get your project off the ground ... but always have a parachute handy.

  7. #47
    Join Date
    Jan 2007
    Location
    Pacific Northwest, USA
    Posts
    30,345
    Blog Entries
    20

    Default

    No, they're not sharp, but they can get very deep. Modify using both hands to push away the wires, whilst holding a penlight in your mouth so that you can see what you're doing.


  8. #48
    Join Date
    Feb 2006
    Location
    in the basement
    Posts
    878

    Default

    Quote Originally Posted by RobS View Post
    .....To permit later changes it is important to avoid daisy-chaining the connections so that ones in the middle can't be removed, the preferred method being to create two distinct layers of overlapping connections on successive pairs of pins like brick-work....
    Sounds very interesting......; would you please explain more!
    Mebe a close-up photo would be very informative...

    Thank you

    ziloo

  9. #49
    Join Date
    Feb 2017
    Location
    Chilliwack, BC, Canada
    Posts
    152

    Default

    Nice! Put it up on YouTube once you get it going. That'd be interesting to see!
    My vintage systems: Tandy 1000 HX, Tandy 1000 RSX, and some random Pentium in a Hewitt Rand chassis...

    Some people keep a classic car in their garage. Some people keep vintage computers. The latter hobby is cheaper, usually takes less space, and is less likely to lead to a fatal accident.

  10. #50
    Join Date
    Sep 2012
    Location
    Kent, England
    Posts
    129

    Default

    The easiest way for me to explain the daisy-chain problem is to show virtual examples using my wirewrap documentation application. The real examples are deep in the rat’s nest on my backplane and not clear to see. That’s the whole point though, that one can’t easily see the problems that one may be creating in the real thing.

    The daisy chain issue arises when existing wires have no slack in them, which is a desirable way of working in many situations. For example, in the Honeywell 200 the logic drawers can be pulled out of the cabinet for servicing and any slack wire might get caught on something in the process, so it is necessary that all the wires lie snugly flush with the pins and are reasonably taught. In this situation when a wire is unwrapped from a pin and the crumpled end is cut off so that it can be wrapped onto a pin again it will most likely end up too short. Therefore assuming that the existing connections already take the shortest path to join all the necessary pins any changes to the wiring will necessitate removing wires at both ends to replace them with new ones.

    Below is a screen grab from my wirewrap application. All the pins to be connected to a specific signal are listed down the screen, so the thin horizontal lines represent the pins and the thick vertical ones the wires joining them. The tops of the pins are on the right and the numbers indicate the order of the connections up each pin from the base. The wires have been connected as a daisy chain.



    Now I will remove pin A3J-22 from the logical schematic as I have decided that it shouldn’t have been connected to this signal. The result is shown below. The application has detected that the physical signal net now has two invalid connections (marked in red) that need to be removed and it is clear that when that is done the two portions of the net will also have to be joined with a new connection. (As a matter of fact red connections also appear when I add temporary connections to test circuits and I actually always use red wire for these on the backplane so that it is visually clear that they must be removed later.)



    It’s clear from the diagram that the unwanted wire on pin A3K-21 can’t be unwrapped because there is another one on top of it and if that one were removed the problem would just move further along the daisy chain. Another solution is to cut the wire and leave the stub end wrapped around the pin. The end result after also removing the other unwanted wire, which can be unwrapped, and making the new connection is then as shown below.



    My application marks any trailing wire unattached at the other end in blue, so the stub appears blue. This solution is fine so long as there is room on the pin for an extra wire and no more are likely to need to be attached in the future, but it’s a messy solution.

    The alternative is from the outset to attach the wires to pairs of pins in two layers as shown below. When the diagram is turned to make the pins vertical the wires look much like courses of bricks.



    This time when the same pin is removed from the signal net the situation is different as shown below.



    I have manually marked in grey the one extra wire that needs to be removed so that the problem one can be unwrapped entirely. Once that is done the two new wires can be added to complete the signal net. Below the second of these is being added and the dialogue box that provides me with useful information to physically do that is shown.



    The dialogue indicates that the wire is number two on both pins, so there should already be one wire attached to each and that rows K and E both have black stripes on the candy-striped backing paper for location. The wire lengths indicated allow extra length for wire-wrapping both ends rather than soldering them. If there is any discrepancy in what I find then something is wrong. When I have attached the wire hitting the “Add” button tells the application to add it to the diagram, as shown below.



    Compared to the daisy-chained method this one leaves the connections in a better state for any further alterations with no more than two wires on any pin.

    The problem is that the natural tendency is to make all the connections in one pass, which results in a daisy chain, instead of making alternate connections in two passes. Having an active diagram of what one is doing such as my application provides is certainly an advantage as you can no doubt see.

    My ambitious construction project is primarily a pastime and meant to be enjoyable, but wire-wrapping can demand a lot of attention, so any aids such as the one that I use are desirable to reduce both stress and errors. So far it has worked.
    Last edited by RobS; March 16th, 2019 at 03:23 AM.
    Rob - http://www.honeypi.org.uk
    The Internet is a winch to get your project off the ground ... but always have a parachute handy.

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •