On Sat, 20 Sep 2014 07:41:50 -0700, Jeff Liebermann
wrote:
On Fri, 19 Sep 2014 11:14:09 +0100, Tom Gardner
wrote:
I can't seem to sleep. Good time for a rant.
Back in the 70s people used blue and red tape taped out on the
same side of the same piece of plastic sheet, typically at 2:1
scale. The blue was the top copper, the red was the bottom
copper (or vv?!). It was projected through coloured filters
onto the light-sensitive etch resist.
At the risk of starting a nostalgia thread, we didn't use the
different colored tape method. Everything was done 2:1 or preferably
4:1. We used mylar sheets and an Xacto knife, Brady tape "donuts" for
pads, Brady black tape for traces, red rubylith for ground planes and
solder masks. Red photo opaque paint for touchup. I was marked for
an early death when I used one of draftings sacred Xacto knives to cut
traces on a PCB board. They had to be very sharp to work well for
working with rubylith.
We did black pads (padmaster) and black traces. The mylar sheets were
padmaster, top traces, bottom traces, assembly, and often a ground
plane thermals. Sometimes more layers. The photographer could make
ground plane film from the padmaster - all copper, clearances for the
pads, thermals added from the thermal sheet.
We did biggish boards, so worked 2x.
At Data General, only one person had a reserved parking spot: the
layout guy.
The best layout people I have worked with were women. True today.
I still have my seriously expensive E size 0.1" mylar alignment grid
somewhere. However, the glue on the tape and pads would dry out after
a few years. Most of my early layouts and layout supplies have long
ago dried out and were thrown out. The lifetime of these original
layouts sometimes defined the lifetime of the product as making
changes to a layout using a photographed enlargement or the negatives
or a PCB was not easy.
While computer layout to Gerber plots were common, I found myself
making changes and corrections to old PCB layouts using these methods
well into the 1990's. Old tech dies hard:
"How It Was: PCB Layout from Rubylith to Dot and Tape to CAD"
http://www.eetimes.com/author.asp?section_id=14&doc_id=1285442
A good light table, and a young body, were necessary for hand-taped
layout.
You also needed a flat table with an overhead UV light, for burning
sepia assembly and fab drawings from the various mylar layers. And a
blueline machine of course.
In RF, it was common for the design engineers to participate directly
in the layout process. Most managers didn't want to waste expensive
engineering time on "menial" tasks, such as PCB layout and checking.
However, those with an interest in getting things right the first time
had other ideas and allowed direct involvement. For RF boards I would
locate the major RF components on a PCB, mark the location of
grounding holes, and make sure the RF path was reasonably straight,
didn't loop back on itself, devices were properly bypassed, and often
supplied the prototype PCB to the layout person. Just handing them
the schematic and parts list was an invitation to start over from
scratch. In honor of my involvement in this system, the drafting
department presented me with a "Change Everything" rubber stamp.
Four layer boards? No.
Yes, although getting them right was difficult. Without computerized
rule checking, it was easy to create problems and not find them until
the prototype was built.
We didn't have trouble with multilayers. We just checked the layouts
(and the film!) a lot. Most boards worked first time; still do. A
couple days of overboard checking pay off.
Poured copper areas? Tedious.
Not really, at least for RF. The real PCB would be fair accurate
clone of the hand made prototype board. The ground plane was always
on top of the PCB. Where the prototype used routed clearances for
non-grounded areas, the PCB layout used rubylith with those areas cut
out with a swivel knife compass. It was a bit tedious, but not very
difficult. The hard part was reconnecting the "islands" of ground
with Brady black tape.
I tried to find examples of such layouts using Google image search and
found nothing. I'll see if I dig out some old photos.
I still have a few mylar layouts around. I'll post pics if anyone is
interested.
Lifting a blue track layed under a red track? expletive deleted
Yep. That's why we didn't use that method. Instead, we had multiple
layers of transparent mylar, with the layers aligned by punched holes
with "pins" and targets. With a 2 layer PCB, there would be 2 sheets
for the traces, and one each for the solder mask and silk screen. For
digital PCB's, we would use 3 layers. There would be a "pad master",
which was used for both the component and circuit side pads. The
other two sheets were just the traces for the component and circuit
sides. When photographed and reduced, the pad master was combined
with the traces to form the final image. The component outlines (silk
screen) were done by hand with an elevated template and india ink.
Every time components were moved due to a design change, the silk
screen had to be redone from scratch as making changes to the original
were difficult.
With luck everything would fit. With my luck, there could be
duplicate reference designators, test points under parts, traces
shorting to component cases, interchanged circuit and components side
copper, and a myriad of other layout mistakes that never seemed to
completely disappear. I never could get anything right the first.
There would always be mistakes. Even when everything seemed perfect,
somone might do something stupid, like leave the original "tape ups"
in a hot car, and have the pads and traces drift when the glue melts.
Traces and pads falling off on the way to the photographers was
common.
I once worked on a very simple design where I decided I was going to
have one PCB that worked the first time. Everything was triple
checked by 3 different people. Everything looked good until the PCB
arrived. The PCB fab shop had gotten the component and circuit side
reversed. Life was hell.
All of that changed when computerized layout and schematic capture
arrived. The term "capture" is rather interesting as I participated
in several ordeal processes of converting pencil drawings on velum to
vector line drawings on a computer. I think my first was in 1979(?)
on an Applicon CAD system running on a PDP11/34. Having an RF design
engineer doing board layout, mechanical design, and drawing schematics
was initially deemed a waste of time, so I had to do it after hours. I
wanted to experience the entire process, which proved worthwhile.
When I first jumped into this newsgroup in about 2011, I got an
initial surprise when the other JL (John Larkin) announced that he
doesn't do full breadboards of complete products.
I never have!
In the 1970's and
80's, I always did breadboards because there were so many unknowns
that could only be answered by building a prototype. Today, those
problems are anticipated by simulations and better characterized
parts. In other words, most of what I did back then is now totally
obsolete.
--
John Larkin Highland Technology, Inc
jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com