Apollo AGC backplane simulator

We used to do wire-wrap backplanes and boards. It was horrible.

Does anybody still use wire-wrap?

John Larkin wrote in news: snipped-for-privacy@4ax.com:

Decades ago I pumped numbers into a VAX/VMS machine that then ran a nesting app to reduce scrap on a stainless plasma cutter.

They had three, and the farthest one from the computer did not work. They were connected with 9 pin serial cable and hdw. I could see easily why the number three machine did not work. The cable was ran right next to power conduits leading to huge power consuming machines, and it had several 'breaks' (splices) along the way.

The machine was 100% wire wrap half rack enclosure. It was a huge rat's nest and that is where the term came from (maybe not).

It still gets used per se in 'peg board' set ups where a wiring simulation of open wires or shorted wires can be simulated to see how redundant systems kick in.

We made such test racks for the then McDonnell Douglas C-17 program, which had 52 on board computers, 17 of which were considered "mission critical".

We made 17 racks of test trays where two of each computer gets mounted and the peg board is used for system tests. We even made one for the HUD that had a 3 foot long dual mount tray sticking out of the front of the rack! We made two full set of racks that then went up to Long Beach the now Boeing owned C-17 mfg and test facilities. The peg boards allowed one to simulate any short or open in the cabling between different aircraft systems, which then was observeed to see how the redundant systems would kick in and take over... or fail to do so. That was back in the late '90s

Wirewrap is a stable gas tight connection. I am seriously considering it for a new design

Cost is super cheap, only high investment in robotics

It's terrible for fast stuff, and can be flakey. Are wires still gas-tight after they break?

It's not cheap! PCBs are cheap.

Optimizing wire-wrap is the "traveling salesman" problem. I've written programs to re-order wire-wrap lists to approximately minimize total wire length. Never again!

mandag den 22. juli 2019 kl. 18.45.04 UTC+2 skrev John Larkin:

how is that different from a PCB?

For one thing, wire-wrap doesn't have power pours!

A PCB can make a tee or plop a via anywhere; it doesn't need a post to make a junction. That is not the traveling salesman problem, or the wire-wrap problem, which have fixed nodes.

I recall only being able to make two automatic wraps on a post, so a node was topologically one long string. A PCB has no such constraints.

DEC's early PDP-11 boxes used a wire-wrap backplane for the unibus. It was a slow signal-integrity nightmare. It was shocking that they would do something so stupid.

What is the impedance of a wire-wrap wire? How repeatable is it?

That technology has been highly automated for at least 55 years. The proces s has as much reliability and integrity as any available at the time. Net l ist verification, and probably design rule checking, was the least of its f unctions.

Apparently still the interconnection technology of choice for communication s wiring: "In telecommunications wire wrap is in common high volume use in modern com munications networks for cross connects of copper wiring. For example, most phone lines from the outside plant go to wire wrap panels in a central off ice, whether used for POTS, DSL or T1 lines. Typically at a main distributi on frame Internal Cross Facilities Assignments and External Cross Facilitie s Assignments, are connected together via jumpers that are wire wrapped. Wi re wrap is popular in telecommunications since it is one of the most secure ways to attach wires, and provides excellent and consistent data layer con tact. Wirewrap panels are rated for high quality data services, including C at 5 grade wiring. The principal competitor in this application is punch bl ocks, which are quicker but less secure."

And to make things even worse, all wire wraps were single color :-)

A wire-wrap backplane is inviting shorts.

I once had to repair a Data General Eclipse computer, where a paper clip had been dropped into the backplane mess. It had landed between the core memory +15V supply and one line of the address bus, destroying all IC's connected to the address line and even some connected to the unlucky first order victim chips.

Pots? DSL? T1? Some ancient language?

No, we used colors for manually wrapped things. Not full auto.

I use wire-wrap wire, but not for wire wrapping.

Reportedly, that sort of thing happened to a whole mainframe, once. An FE had taken one of the back cabinet doors off of its hinges to gain better access to something on the backplane, and let the door fall against the backplane... it shorted the high-voltage supply to the +5. BANG.

I was told that the manufacturer's refurb facility received "boxes and boxes of cards, where the lids had been blown off of the cans."

For low frequency stuff, the PCB is 1 cent per cm2

If you have few big leaded components, quote wrap is much cheaper and more reliable

Cheers

Klaus

Quote wrap = wire wrap (sorry, autocorrection crap)

Hi,

Wire wrap is really susceptible to common mode noise compared to a PCB with ground plane.

cheers, Jamie

Easy. 100-150 ohms.

Assuming a close-by ground plane, which is a necessity for anything of more than minuscule scale.

Not bad considering a proto PCB is +/- 20%. More than good enough for TTL. They used twisted pair wire wraps for ECL backplanes, too (which would be closer to 100 ohms all the time).

The crosstalk is the worse problem.

Tim

Plain Old Telephone Service Digital Subscriber Line (eg VDSL ADSL) T1 trunk (a type of telephony trunk invented by Bell Labs) Some ancient language? not yet, but it seems that VOIP is the future.

We will probably wind up with a single, universal, worldwide network for everything from light bulbs to data centers, 8G or something.

I find that the #30 Kynar wire is great for reworking SMT boards.

Cheers