Page 101 - Analog Circuit Design Art, Science, and Personalities

Propagation of the Race (of Analog Circuit Designers)

These latter constraints are included primarily to reduce difficulty with spurious
oscillations during the breadboard phase of the project. Most students just don’t
have the experience necessary to effectively use ground plane and ferrite beads in
the alloted time!
The exact specifications change from year to year, but we try to maintain a con-
stant degree of difficulty. Intentionally missing from all specifications is anything
that suggests a circuit topology or limits the total number of devices used.
Students are expected to complete the following multifaceted solution to this
problem.
1. Guided by anything you know, find a topology and associated component
values that you think will meet the specifications.
2. While we assume that the above will be determined in part by the estimation
methods suggested, show us why you believe your design will meet specs.
3. Now simulate the circuit and see if the computer confirms your optimism.
(If not, decide why the simulation is wrong or redo your design.)
4. Build and test the circuit. If there are problems, iterate. (This phase is usually
accomplished in protoboard form. Laboratory handouts and hints from the
staff have suggested efficient layout and stressed how to include parasitics in
earlier steps.)
5. Talk with your teaching assistant about the above, and convince him or her
that you have done a good job.
The important difference between this assignment and many of the students’
earlier experiences involves the quantity of good answers. For many students. all
problems they have been given earlier have only one correct answer. (Example:
What is the integral of e‘.‘? Not too much choice on this one!) Suddenly this unique-
ness disintegrates. A common characteristic of design problems is that there are an
infinite number of solutions to all problems; some of these work; some work much
better than others.
The reaction of students to this situation is interesting. (I feel I have enough expe-
rience to justify the following anecdotal observations.) A few students who have
easily jumped through all the academic hoops previously presented to them are very
uncomfortable in this situation. The subset of this group that doesn’t adapt as the
term progresses drops this subject and presumably lives happily ever after doing
something else. Ignoring the large group in between, another fraction of the students
love this sort of thing. These folks may become our kind of people!
We evaluate with only course quantization. It works or it doesn’t work, with little
gradation in between. This approach is appropriate for the first real design experi-
ence of the group. However, it is clear that finer value judgments are possible. One
discriminator, since the designs must eventually be built by their designers, is the
number of transistors used. Most designs require four to six devices. Some require
more-good if the resultant performance far exceeds specifications and not so
good otherwise. A few students usually design and successfully implement three-
transistor solutions.
The year we assigned the gain and bandwith specifications mentioned above, a
copy of the assignment found its way to Bob Pease in Silicon Valley. I’m not sure
how this happened, but 1 suspect Jim Williams may have been involved. Bob sub-
mitted a design that met specs using two transistors. (He would have gotten a very
good grade had he been taking the course.) His basic trick was to use positive feed-
back to reduce the effective input capacitance.
Bob’s performance has become a benchmark. The teaching assistants who select

82