 Excerpt from the Introduction in the Technical
Therapy DVD Owner’s Manual:
Analog design engineers find that they are not able to translate a large part of what they
thought they learned in college into practical application.
If this condition is thought of as a “technical disability,” the possibility arises that
suitable “technical therapy” can alleviate the condition, and that is what this Course is
about.
Despite the fact that electrical engineering graduates have been exposed to a large volume of
technical theory and practice, they still find themselves poorly equipped to handle the design
problems they are faced with in industry. Even though for four or more years they have been
“drinking from a firehose,” they often feel as though they “fall off a cliff” when presented with
their first job assignment. Their plunge in confidence comes when they realize how few tools they
seem to possess that can readily be applied to real-life problems. They actually know a lot, but
they find it difficult to apply their education on the job. Employers in the electronics field
continually call for more “design” training at the university level, yet when it comes to practical
specifics, the design process is all too often ill-defined or not well understood.
In the course I taught for many years at Caltech, I developed a design-oriented approach to circuit
analysis in which the design objective is kept in view from the start and maintained throughout the
process. Some of these techniques resulted from my consulting experience “putting out fires” caused
by nonoptimal designs that usually were supported by little or no analysis. I began to change the
emphasis of my approach until finally the essence is in the methods and the techniques, and the
subject material is secondary. It’s really a course in how to se up and solve design problems, using
analog circuits as a vehicle, although of course the methods are applicable in other fields as well.
Nature of the Technical Disability
Analog design problems are admittedly ill-defined from the point of view of a mathematician in
that there are never enough equations to solve for the number of unknowns. Nevertheless, the
designer must solve the problem anyway, by substitution of missing exact equations with inequalities
in the form of approximations and tradeoffs. Even the design of something as technical as an
electronic circuit is as much an art as a science. A good designer is skilled in the art of
approximation and creates a result that meets the specification in a way that is, in various senses,
optimum.
The typical electronics engineer graduate has been drilled in solving simplified, sanitized
analysis exercises that have unique answers. One answer is right and the rest are wrong. Of course
there are good reasons for why this has to be true to some extent in a school setting. However, at
the first job, an engineer is confronted with a requirement to design something to meet a
specification that is much more complex than was generally presented in school. This brings the
realization, often for the first time, that design is the reverse of analysis. In design
you start with the specification and must work backward to get the circuit configuration that
meets it.
In analysis you are given the circuit configuration and the component values and you work
forward to get the specification, or the performance characteristics of the circuit.
In design you are given the specification and must determine both the configuration
and the component values.
Causes of the Technical Disability
Engineers, as students, learn a conventional problem-solving paradigm that is not taught
explicitly, but seems to be implicitly instilled by our current education system. The message
received is remarkably uniform from student to student, from school to school, and even from country
to country.
Here is an attempt to illuminate the conventional paradigm:
Premise
The world is a complex, unmanageable, uncertain place; to make any headway in design, we have to
harness the problem with equations and beat it into submission with mathematical rigor. The emphasis
is on exactitude, regardless of how long it takes.
Conventional Problem-Solving Approach, or Stupid Analysis Tricks (apologies to David
Letterman)
-
Put everything into the model and simplify later.
-
Postpone approximation as long as possible, and don’t even dare to make
an approximation unless you can justify it on the spot.
-
The “answer” is acceptable in whatever form it emerges from the algebra.
-
The more work you do, the more valuable the result.
-
Every problem is a brand-new problem, and requires a brand-new strategy
to solve it.
Consequences of the Conventional Approach
This approach is a recipe for failure. It engenders a syndrome of Technical Disability,
algebraic diarrhea: write down all the equations you can think of, and after a page or two of
manipulation, during which the equations get longer and longer, progress comes to a sudden halt as
algebraic paralysis sets in.
Another syndrome of Technical Disability, less obvious, is fear of approximation. Students
infer that if they are unable to solve a problem exactly, they have somehow “failed.” Conversely, an
exact answer should be worth more (credit) than an approximate answer.
It is very difficult to arrive at a “good” answer in the presence of these debilitating
syndromes. The student is aware that the odds are stacked against him, so is under-confident and
fearful of attacking real-world problems. Since school problems are much simpler than real-world
problems, the negative results of the conventional paradigm are often masked while the student is in
school.
Once the student graduates and confronts a real problem on the job, the weaknesses of this
approach are exposed (the “falling off the cliff” syndrome), and the novice engineer must begin a
long process of re-education to develop a new problem-solving paradigm. Unfortunately, some
engineers never recover from this setback; they throw out analysis altogether and become “cut and
try” engineers dependent on simulation, cookbook recipes, or pure “knob-twiddling.”
In the course I developed at Caltech, and later adapted as a short course for industry, I found
it necessary to assign names to the various methods and techniques. The short-course version was
titled Structured Analog Design, Part 1 and this DVD version is titled Technical Therapy
because the methods and techniques offer ways to alleviate the Technical Disability that is so
widespread.
The most basic method is to present the results of analysis in a low entropy form.
A Low Entropy Expression is defined as one in which terms are ordered, or grouped, so that
additional insight is gained into the relative importance of the various contributions to the
result. This is the source of the additional information needed for design, and substitutes for the
missing equations that would be needed to solve formally for the number of unknowns.
This is in contrast to a High Entropy Expression, which is typically obtained by “blind”
application of algebraic manipulations, usually leading to sums of products of circuit elements that
provide no insight into how the relative values of components affect the result.
Since design is the reverse of analysis, it is essential that every analysis result be expressed
in a low entropy form. Such a paradigm is called Design-Oriented Analysis, or D-OA
(don’t forget the hyphen!).
Design-Oriented Analysis is the principal strategy of Technical Therapy.
Design-Oriented Analysis is the only kind of analysis worth doing.
Unfortunately, conventional loop and node methods do not lend themselves well to forming low
entropy expressions, and alternatives are developed in detail throughout the course.
Technical Therapy
Design-Oriented Analysis is a positive approach to displace the conventional negative approach.
Premise
The world is a complex place, but is manageable and solutions can be found with reasonable
certainty. To make headway in design, we can attack the problem by dividing it into manageable
pieces (“Divide and Conquer”), then applying suitable models, approximations, assumptions, and
physical reasoning until the system is understood well enough to achieve the design objectives.
It is basically understood by most that real-world analysis requires assumptions and
approximations: however, D-OA says “make lots of assumptions and approximations and make them now.”
In other words the emphasis is shifted heavily towards these shortcuts being employed at the very
beginning and being employed as copiously as possible.
D-OA Problem-Solving Approach (Rules)
-
Put only enough into the model to get the answer you need.
-
Make all the approximations you can, as soon as you can, justified or
not. Plow through the problem leaving a wake behind you of assumptions and approximations. You
can’t lose by trying.
-
Figure out in advance as many of the quantities as you can that you want
to have in the answer, and put them into the statement of the problem as soon as possible – even
into the circuit model.
-
The less work you do, the more valuable the result. You control
the algebra. You make the algebra come out in low entropy form by applying strategic mental
energy before and during the math.
-
Every problem is not unique. There are problem solving strategies
and techniques that apply to almost all engineering problems.
These five Rules are the direct opposites of those of the conventional approach
Consequences of the D-OA Approach
This approach is a recipe for success. It fends off algebraic paralysis, and even when
complexity increases you have an array of assumptions, approximations and analysis tools in order to
remain algebraically mobile.
Approximations are not “bad” things; on the contrary, assumptions and approximations are the key
to arriving at a “good” answer, namely, a low entropy expression that can be worked backwards for
design. Technical Therapy lets you overcome “fear of approximation” by giving you not only
permission, but also encouragement, to employ assumptions and approximations as the primary method
of controlling the algebra.
Overall, the D-OA paradigm bestows a sense of empowerment: the algebra is malleable, you have
choices; it can be made to take on whatever forms are most useful for design. The math becomes the
designer’s slave, rather than his master. It clearly shows that dependence on circuit simulation
without fundamental understanding of the intuitive, physical sources of circuit behavior is a
mistake, and can waste hours of fruitless trial and error churning.
The five Rules listed above describe the methods, or broad strategies, of D-OA. To implement
them, you also need Tools, the specific detailed techniques that constitute the bulk of this course.
Chapter 1 presents the Rules in more detail, and then lists many of the Tools with emphasis on the
motivation for their introduction. These are listed below, and a more detailed summary, Tools of
Design-Oriented Analysis (D-OA), is in a separate section for convenient reference .
D-OA Problem-Solving Approach (Tools)
How to avoid most of the math, yet get more useful design information from less work.
Doing the Algebra on the Circuit Diagram
Doing the Algebra on the Graph
Using Normal and Inverted Poles and Zeros
Using Numerical Values to Justify Analytic Approximations
Improved Formulas for Quadratic Roots
The Input/Output Impedance Theorem
The Extra Element Theorem (EET)
The Feedback Theorem
Loop Gain by Signal Injection into the Closed Loop
Most of us can benefit from a process of Technical Therapy. As in any therapy, the first step is
a “regression,” in this case all the way back to high school, and not only to accept, but believe,
that many of the things we learned were of little use. They weren’t wrong of course, they just
weren’t tailored to engineering applications. We don’t get to hear from engineers until the
mathematicians and the scientists have already been at us for several years. For example, we are
taught the conventional quadratic root formula without any mention of its deficiencies or how they
can be avoided; we are taught that if you have a certain number of unknowns, you’d better have an
equal number equations, otherwise you can’t solve the problem.
The difficulty is that engineers have to solve the problem in the face of not enough equations.
In fact, not only are there not as many equations as the number of unknowns, there aren’t anywhere
near enough equations. Design-Oriented Analysis tells us that the few equations we do have can be
made to work harder, that is, you can get more than one piece of useful design information from one
equation: this is the main benefit of a low entropy expression in which you can track separate
components of the result back to their circuit origin.
Once we believe that the conventional approach carries with it severe but unnecessary
limitations, our minds become receptive to improved approaches. This DVD offer a Technical Therapy
Course based on the positive approach of Design-Oriented Analysis.
As with any therapy, a Technical Therapy participant must first recognize his Technical Disability,
and then must want to overcome it. It’s usually harder to break an old habit than to adopt a new
one, and in the present context the hardest step is merely to remember to use the improved methods.
In Chapter 10 there is an Exercise in using Rule 1, “Put only enough into the model to get the
answer you need.” Invariably, almost every participant fails to employ this rule, in spite of its
use only minutes earlier in the preceding example!
Middlebrook’s Structured Analog Design Course has been given many times in North America and
Europe to design engineers and managers in open-to-public seminars and company sponsored in-house
courses.
Experienced engineers are extremely receptive to these ideas, because they’ve already tried it
the “hard way,” and they know that doesn’t work. After participating in this Technical Therapy
course, many engineers say,
“I wish I’d known these things years ago! Why wasn’t I taught them in college?” | 
