Page 102 - Analog Circuit Design Art, Science, and Personalities

Jim Roberge

the specifications each year make sure that they can meet them with two transistors
and present their design to the class after the due date.
As mentioned earlier, we also look at the dynamics of switching circuits in this
subject and use change-control methods for estimate in this case.
In addition to the material on circuit dynamics, which represents more than half
the content and which is included every time the course is offered: we select several
other topics from a menu that includes DC amplification, high-voltage-gain stages
wing dynamic loads, band-gap circuits, translinear circuits, noise considerations,
and power handling stages. The exact mix changes from year to year, reflecting our
belief that the specific choice of examples doesn’t matter; the important feature is
that design, rather than analysis, methods are used.
The use of real-time demonstrations to illustrate many ideas was mentioned
carlier. We also use many examples drawn from available integrated circuits. For
example, open-circuit time constants can be used to show that the collector-to-base
capacitance of the transistors used in the 733 (an earlier linear integrated circuit)
must be less than 0. i pF. Similarly, charge-control can be used to justify the inclu-
sion of the “Miller-killer” portion of FAST logic.
The second undergraduate course is one in classical feedback. I include a discus-
sion of this subject in a book on analog circuit design because I believe that B
thorough understanding of this topic is the single most important prerequisite for
the effective design of many analog circuits. I occasionally encounter designers
who know so little about feedback that they should be prohibited legally from
using it. ‘The areas in which these individuals can do effective circuit design are
quite limited.
The prerequisite for this course is a good understanding of basic linear-systems
ideas. Students should have a reasonable appreciation of the importance of poles
and zeros, and be able to sketch Rode plots. We do not require an in-depth under-
standing of Laplace stuff such as partial fraction expansions and contour integration
for tiiking inverse transforms. although many of the participants have completed
that part of the EE core program. There is no prerequisite requirement linking this
and thc active-circuits subject, although most students feel that each provides
excellent background for the other.
The emphasis for many years was OR the electronic feedback systems, using pri-
marily operational-amplifier configurations as examples. However, the discussion
was at the block-diagram level, iis opposed to developing the tranfer functions of
the blocks from the innards of a particular amplifier. This approach was Esed so that
systenis types and computer scieniists from our department, as well as students in
other disciplines, could take the course without first acquiring a circuit background.
Topics covered included:
Modeling and block-diagram representation.
9 Approximating responses. Under what condition can the transient response of
a system be approximated as an appropriately chosen first- or second-order
system? (Answer: Almost always.)
Stability unalysis via root-locus and Nyquist diagrams.
Analysis and design of nonlinear systems via linearization and describing
functions. (Describing functions is an excellent tool when one actually M3ants
to ticsign an oscillator.)
e Cornpcnsation.
The associated laboratories were design oriented and used appropriately config-
wed operational amplifiers as the vehicle. However, there is a difficulty associated

83