6 Jun
2006
6 Jun
'06
6:27 p.m.
Brent Hilpert wrote:
There was another technique for making circuit boards
in the 70s: a
programmed/automated machine moved over the initially bare board and laid down
a fine wire from a spool to form the circuits. I think the wire stuck to the
board initially and the board was later lacquered or coated with something to
seal/hold the wires better. I forgot how the wire traces were terminated
around IC pin locations, probably a pressure weld.
I think the economics were such that it was too slow for large production
runs, but was suitable for prototypes and small production runs of larger,
more-complex boards. Last example I saw was a disk-controller for a TI-990
mini circa 1980.
Anybody remember the name for the technique?
Multiwire was the trade name from one vendor I am familiar with. It was
used to do the prototype for
the first pass of the DEC 11/74 commercial instruction set CIS
processor. It was a disaster.
The selling point was we could get complex multilayer-equivalent PCB in
1/3 the time it took to layout
and fabricate a real four layer PCB. That claim was true, unless you
wanted working, reliable boards.
Multiwire was very prone to the wire-hole connection breaking, leaving
an open. Each time the PCB
was flexed, connections would come and go. By the time we finished
debugging all the boards (three of
them) they were covered in a ratsnest of red wire wrap jumper wires to
fix all the open connections,
and the real PCB fabs arrived about a week later.
I think that was the first (and last) time DEC used Multiwire to do
'quick-turn' prototypes, at least in
mid-range PDP-11 land.