Page 105 - Analog Circuit Design Art, Science, and Personalities

Propagation of the Race (of Analog Circuit Designers)

ested in design opt to take the undergraduate subjects early in their graduate pro-
gram.) In this way we ensure a small, very competent and enthusiastic group of
students. The course enjoys essentially perfect attendance, and virtually all partici-
pants earn A’s.
The subject is ajoy to teach. We discuss (and “we discuss” is really a better de-
scription than “1 lecture”) how designers have accomplished some function during
two 1 X hour sessions a week for several weeks. The specific topics vary from term
to term, but the selection generally includes sample-and-holds, digital-to-analog
converters, and analog-to-digital converters. There is no serious attempt at theoret-
ical rigor in any part of this; it’s not necessary since we have right on our side!
We then hand out an assignment that is effectively a spec sheet and ask the stu-
dents to conduct a detailed paper design of a circuit that they think meets the speci-
fications. There are usually two or more sets of specifications offered for each topic,
often a high-accuracy set and a high-speed set, and the students may chose either. In
keeping with the spirit of the subject, they may work to their own set of specifica-
tions as long as they are of comparable difficulty. The design can be either IC or
discrete. and they have about two weeks to complete it, during which time a new
topic is being discussed in class. Either 1 or the teaching assistant (who is always the
most senior one working in the sequence) read each design and discuss it with its
author. Because of differences in the backgrounds of the participants, ranging from
seniors and early graduate students to practicing design engineers, no absolute scale
is used for evaluation. We reserve any severe criticism for errors that a particular
student should know enough not to make. If our students disagree with any negative
comments we may make about their circuit, they can always prove us wrong by
building it!
It is interesting to compare the approaches of various students. Many directly
adapt some topology we have discussed. Considering the difficulty of invention,
this is a fine approach if the details are filled in correctly. I’m sure most of us do
much of our design by combining topologies we have seen before rather than via
completely original configurations.
An occasional student will try what to him or her is a completely new approach.
For example, one student designed an incredibly complex circuit using an inductor
as the memory element in a sample and hold. Since he did not use superconductors,
extraordinary means were necessary to achieve the required self-time constant. 1
think it might even have worked. He chose the approach not because of naivetk, but
just to prove he could do it. Needless to say, he was the teaching assistant the next
time the course was offered!
The first time I offered the course as described above, about 20 years ago, several
participants mentioned that the format resulted in a rather “lumpy” work load, with
a major effort required preceding each assignment due date. After some considera-
tion, wc dccided that the best way to remedy this situation was not by leveling the
peaks, but rather by filling in thc valleys. These fillers are not directly correlated
with the topic being covered in class, but do offer a way of expanding coverage to
include other important material.
There is, of course, an associated laboratory. We generally do not ask students to
build their designs because of the time commitment that would be required. How-
ever: we may suggest building a portion of it. For example, if a student chooses the
high-speed design for the digital-to-analog convcrtcr, we expect him or her to
breadboard the most and least significant bits and demonstrate setting timc. This is
a very worthwhile exercise for students who have limited experience with ground
plane and ‘kettle-box” circuits.

86