Re: electronics texts

["John S. Denker" <jsd@MONMOUTH.COM>]

At 09:41 AM 11/7/01 -0500, Bob Muir wrote:
> I have been teaching (for a long time) a course entitled "Electronics
> for Scientists"
....
> Now, I would like to have a book that has primarily an applied
> orientation.  It should contain problem sets and lab activities that
> are integrated with the textual materials. I would like it to treat
> diodes and transistors to a minimal extent, op-amps & digital IC's to
> a large extent, noise and grounding issues, signal processing, etc.
> I don't want it to get bogged down with detailed treatment of the
> properties of solid state materials and semiconductors.  I want
> students who have completed the course to be able to design simple
> anlog, digital, and combintation circuits using IC devices and to be
> able to analyze more complex circuits from a circuit diagram (as well
> as to breadboard them).

1) The following suggestion may sound nit-picky at first, but I think it
has merit:  Consider changing the title of the course.  I like the
description of the course a lot better than the title;  a new title along
the lines of
   "Physics 512: Measurement, signal processing, and control"
would seem preferable for two reasons:  It is more informative, and doesn't
sound so "segregationist".

Partly on practical grounds and partly on philosophical grounds I object to
courses aimed at specific segregated audiences, such as:
  -- Chemistry for women.
  -- Electronics for tall people.
  -- Mathematics for people with green hair.


2) Related point:  Under ordinary circumstances, such a course should be
co-listed in the Electrical Engineering department.  In this case things
are not so simple, because UNCG doesn't offer any Engineering to speak of:
  http://www.uncg.edu/phy/preengineering/preeng.htm

However, there is a point to be made that applies in most cases (but not,
alas, UNCG).

In general, it is good to have a course in "measurement, signal processing,
and control" and it would be good if a few Electrical Engineering majors
would take the course.  It's amazing how many people graduate with EE
degrees without ever having heard of a ground loop, or an instrumentation
preamp, or a bridge, or a ratio transformer, or a lockin amplifier.

The best approach is to have a "cognate" course, i.e. to have it listed
both places:
   "Physics 512: Measurement, signal processing, and control"
   "EE 512: Measurement, signal processing, and control"

That way it's obvious that EE majors can take it and have it count toward
their major.  That makes it clear that it isn't a course in "music
appreciation for people with green hair who aren't qualified to take a real
music-theory course".

Similar remarks apply to courses called "mathematical methods of physics":
they should be co-listed in the math department.  This reassures the
students that the faculty has its act together.

3) I have my doubts about the course description above, when it speaks of
building circuits from IC devices.  Nobody does that anymore, not in a real
measurement lab.  For almost all measurements, there are three basic steps:
  a) Physics and analog stuff:  Shielding, and using a bridge to make a
null or differential measurement.
  b) Electronics: slap a preamplifier module on it and digitize it.
  c) Software: do all the signal processing, and all the control, in software.

Note that the electronics is usually modular.  If you want to teach a
course in how to design high-frequency low-noise instrumentation preamps,
that's OK, but that's a totally different course.

And why mess with digital ICs if you can do what you want with a Stamp?
  http://www.stampsinclass.com/
If it's so tricky you can't do it with a Stamp, you probably can't do it in
a one-semester class at all.

========================

To contribute toward answering the original question:  IMHO everybody
should own a copy of Horowitz and Hill.  Its chapter 7 (Precision Circuits)
is on-target for 1/3rd of the course description.

http://www.amazon.com/exec/obidos/tg/stores/detail/-/books/0521370957/conten
ts/ref%3Dpm%5Fdp%5Fln%5Fb%5F2/107-4763737-6771759

IMHO everybody should own a copy of Gardner, _Phaselock Techniques_

http://www.amazon.com/exec/obidos/ASIN/0471042943/qid=1005152003/sr=8-1/ref=
sr_8_3_1/107-7291191-7601350

As for the other parts of the course, signal analysis and control, those
are vast subjects unto themselves.  Probably the best you can do in a
one-semester course is to pick a package such as labview and teach 'em how
to use that.  Loads of texts are available

http://www.amazon.com/exec/obidos/search-handle-url/index=books&field-keywor
ds=labview&bq=1/102-4148522-0261761
including for example

http://www.amazon.com/exec/obidos/ASIN/013833949X/qid=1005155744/sr=1-4/ref=
sr_1_9_4/102-4148522-0261761

For the analysis (but not control) part, the de_facto standard is Matlab.
Scilab is similar (almost but not quite identical), works well, and has a
very attractive price ($0).
  http://www-rocq.inria.fr/scilab/