HAM Radio 1/HamRadio.cdr/tech/droege2/intro.doc

Text File | 1986-10-14 | 9.0 KB | 241 lines

                                 Page 1-1

INTRO.DOC

INTRODUCTION

This program is designed to aid the manual preparation of printed circuit
phototooling.

It runs on an IBM PC, XT, AT and anything else that runs PC DOS we think.  A
minimum configuration is one disk,  320k memory, and a color card.  The free
version uses 320 k memory.  The licensed version will make use of 640k on a PC
or a megabyte or so on an AT.  While it will run with a mono monitor on a color
card,  the program is not very useful without a color monitor.  A printer and
an additional disk are useful.

We describe this as a manual system because there are now many programs which
purport to automatically generate phototooling from circuit diagrams or logic
statements.  We have been designing mostly analog circuits for thirty years and
are yet to know enough about design to direct a skilled layout professional,
much less to design a computer program for the task.  We would not know where
to begin to set up design rules.  We often redesign the circuit during layout
to overcome a problem introduced by the topology of the corner we have worked
ourselves into.  We do not see how a program could be designed to make such
opportunistic decisions.

If you have not used a computer aided layout system before you have an
important decision to make.  Will such a system save you time (money) in the
long run?  There is no doubt that it will be more expensive at first to go to a
computer aided system.  Even one like this that is free! (Ha!).  Everyone
underestimates the effort it requires to become productive.  Your friendly
marketeer will talk about factors like 5 in productivity.  You will be lucky to
make .5 the first year.

In our design group, the engineers do their own layout.  We have watched some
of the fancy automatic systems in operation and we sometimes wonder why anyone
would want to work so hard to do a layout.  We watch with awe the problems
associated with the introduction of a new component into an automatic system.
I have completed an entire design while my neighbor in the CAD room was trying
for several weeks to properly specify a new component.  If you want an machine
to be really automatic you have to tell it everything about its components.
This is tough enough with digital chips, as my CAD neighbor discovered.  For
analog chips or for new concepts unanticipated by the CAD designers, it is very
difficult indeed.

We are not so reactionary as to think that computers will never be able to
perform this task.  We do feel that with the aid of a simple design tool like
this program that today we can outperform the computers on all but the most
routine digital layout tasks.  Marketeers extol the virtues of such features as
rubberbanding - yet try it on a reasonable sized circuit and it is almost
impossible to make any sense out of the jumble.

Computer aided systems win big if there is repetition.  Some of the boards we
design contain 32 identical amplifiers, and some control circuitry.  Typically
a thousand mixed analog and digital components.  These designs are very cost
effective for a design system.  When the customer comes back for a different
design we can produce a whole layout just by changing the amplifier symbol.

We have set out to design a program to aid us overcome the drudgery of the
layout task.  It has been written while under constant use by several very
                                 Page 1-2
senior designers.  The general plan has been to add the feature most griped
about next.  This will explain to CAD professionals the otherwise mysterious
selection of features.  For example, there is no mouse or joystick input
because we were able to work just fine without it.  There was no marketing
department to force us to add features for marketing purposes only.  Before you
decide to pass up this system because it does not have some favored feature,
give it a chance.  You may find it workable.  We have used the full version of
this program to produce 13" x 17" printed circuit boards containing over one
thousand components and 3500 holes.  The boards are mostly mixed analog-digital
designs where we are concerned with microvolt signal levels and nanosecond rise
times.

Some of the designs have been done on a team basis.  It is important to us to
be able to pass parts of a design between work stations with several designers
working at the same time.  This is accomplished by the extensive and flexible
symbol features of this program.  We have not yet networked our terminals but
only because it is still easier to walk across the hall with a floppy disk.

A symbol is anything that might be put down as part of a printed circuit
master.  A very simple symbol might be a pad for the location of a component.
This program allows nesting of symbols so that once, for example,  a symbol for
a printed circuit component pad is created it can be duplicated to form an
integrated circuit package outline.  At the next level a group of package
symbols might be combined with wiring to form a memory symbol.  Some examples
of this are included in the demonstrations.  We are gradually accumulating a
library of such symbols so that we can rapidly produce rather large designs.

The program allows entry of lines, pads, symbols, labels, and arcs on 12
different layers and on a 65000 x 65000 point grid.  Lines correspond to
printed circuit tape which is used to interconnect components.  Lines come in
fifteen widths.  Pads correspond to the doughnut shaped cutout used to indicate
that a drill hole was to be made to mount a component.  Symbols are arbitrary
collections of everything else which can be located and deleted as a single
object.  Labels are lettering.  Arcs allow drawing circles and partial circles.

While the program is labeled as a 65.000" square grid, there is nothing special
about the label and the output files can be scaled arbitrarily.

Data is entered by selecting an object type and locating it on the grid through
use of a screen display and the arrow keys.  We had planned for a mouse, but
our users never complained.  A color display is provided so that objects can be
distinguished on different levels ( our users absolutely demanded color) though
the "free" version shows only the last color entered for intersecting objects.

Three different colors can be assigned to the layers.  This is usually enough,
since more is just confusing.  An upcoming version will support the EGA with
more colors but early work indicates that the resolution is more interesting
than the additional colors.  We seldom work on more than two or three layers at
once, even when doing boards with many more layers.

Parameter sets can be stored for frequently used combinations of working
conditions.  One set allows instant redraw to any preselected part of the
layout.  Another set allows switching between entered object types.  For
example a few keystrokes are all that is needed to switch between entering red
100 mil wide lines on level 3 to green 15 mil wide lines on level 7 if these
parameters have been preset.

Our goal has been a simple system that works on a minimum PC configuration.  We
have resisted making special monitor calls or doing anything that was not
                                 Page 1-3
available on any system with a color card and color monitor, even if it would
make operation better.  This further explains why there is no mouse.  The
program was written originally in compiled BASIC though we are gradually
converting sections to assembly language.  We have run it on a variety of IBM
PC, XT, and AT computers, and a number of clones.  Except for saving and
loading jobs, and creating plot files, the program runs entirely in memory.
When we started this development several years ago, we made the decision that
by the time that we would get our program written, memory would be cheap enough
to hold it.  It is almost an axiom that memory will be cheap enough to hold any
program you can conceive of writing by the time you can get it written.

This version is being distributed free, primarily to advertise our
photoplotter, which was the reason for it being written.  As distributed it has
restricted memory, enough for about a thirty chip circuit board.  One of the
demos, ALRM, is a 1000+ component board where advantage is taken of the
repetitive nature of the circuits to save memory.  Print FEATURES.DOC for the
specifications of the extended version of this program which is available for
$100.00.  Work started on the free version of the program is transferable to
the pay version which has a number of additional features which serious users
will want.  We hope to get desperate letters asking us to rush the larger
version as users run out of memory.  Plan ahead and buy now.






































                                 Page 1-4