Do yourself a favor and get the latest edition
of "The Art of Electronics" by Horowitz and Hill.
It's generally considered by most of us to be the
pre-eminient book on practical electronic design
that's currently available. It should be widely
available...I got mine (5 copies so far? I keep
giving them to people..) at Powell's Technical
Books (http://www.powells.portland.or.us).

I was very disappointed with the way that electronics
is taught at the university level. It really bears
very little on what working engineers need to know,
and the laughable idea that you're going to look
up the h-parameters of a part for design is just silly!
For one thing, you can't count on those numbers! Yes,
they'll give you a "typical" value, but over a spread
of manufactured units, they're of little more than
order-of-magnitude use.

A simple test; design your circuit, then ask yourself
what happens if all the resistor values get changed by
a factor of 10. Oops..loaded a 1K where 100 ohms should
go.....does it still work? If not, why not? Try it several
ways, with the idea that if any part "makes a difference"
if you push its value around, then you need to think about
*why* that makes a difference and see if you can reduce
its effect. You won't end up with a highly-optimized circuit,
but that's not really what we want out here in the real world.

When I'm designing discrete-component circuitry, one of my
main concerns is what parts I'm already using elsewhere on
the board. If I *have* 180 ohm resistors already, I'm going
to use them everywhere I *can*, even if a proper design would
call for 220 or 270 ohms. Sure, it's not optimum, but if it
works and it minimizes the number of reels I have to kit up
and load on the SMT placement machine, that counts for more
than a perfectly-centered output voltage. When I finish a
design, the first thing I do is to run a parts list, line up
all the resistor values and see how many I can condense into
a single value. Sometimes I'll use a high-precision part in
a low-precision application, because I *have* to use it in
one spot, and it costs more to use a 5% part if I have to
load a reel of them separately. Not to mention the risk that
the assembly guys will put the *wrong reel* on, and lay down
5% parts where I *must* have a 0.1% part. I generally avoid
such parts, but there are a few places (Ethernet filter tuning
and such) where the manufacturers of the ICs specify what
should be used and how.

My point is, what they teach you in school doesn't
really have much to do with designing products. If
you get the chance, work in the repair business for
a while. You'll get to see everybody's mistakes, and
learn how to do things right by seeing what doesn't
work. Do some interning in manufacturing engineering
or test engineering; they're traditionally considered
'boring' or 'unglamorous', but you'll be a much better
designer for having the experience. Those areas of
electronics don't pay as well, but you'll learn a lot
that will make your products work right and be easy
to build. That leads to profits and success, which is
part of why we're all doing this!

Mark G. Forbes, R & D Engineer  |  Acres Gaming, Inc.    (541) 766-2515
KC7LZD                          |  815 NW 9th Street     (541) 753-7524 fax
forbesm@peak.org                |  Corvallis, OR 97330
http://www.peak.org/~forbesm
mforbes@hq.acresgaming.com

"There has been an alarming increase in the number of things I know nothing
about."
---Anomalous