定理至此證畢。

讀書筆記——Microelectric Circuits (電子學課本)
Feb. 22, 2016
(originally posted on Amazon)

A. Sedra and K. Smith (2011), Microelectronic Circuits, 6th ed. New York: Oxford U. Press.

This book is the required reading of the 3-semester “microelectronic circuits (電子學)” offered by dept. of Electrical Engineering of National Taiwan University (台灣大學), that I took two years ago. This is not the course which I get the poorest grade of, but is certainly the one it seems most tedious to learn. Many of my classmate of EE dept. too find this book extremely dry. (EE dept. of NTU is claimed to be one of the leading engineering institutions — This is not meant to offend the reader, but to supply some background. Also, my edition is international one.) 

It is true that this book has its merits — the circuit diagrams are all easy to follow, and show steps very clearly. And there are some interesting material, like operation of SRAM (sec. 16.3.1, p.1320), or ingenious design of current mirrors (sec. 6.5.4, p.518). And it is not the case that the authors lacks physical insights — Sedra / Smith frequently point out design experience, for example that source follower is suitable for buffering, and intuitive way to memorize results, for example that resistance attached to BJT emitter seen looking into the base is β times of the original value. 

So what’s wrong with it that makes the read so painful? 

To be sure, in many places the presentation seems so sketchy that it is difficult to assimilate for first time readers. I would occasionally suggest insertion of just one or two more pages explaining key points that would greatly facilitate understanding and memorization. 

To take an example, in p.834, fig.9.32, the mysterious diagram indicates by breaking of loop how one determines its loop gain. I think the reason behind this is straightforward. With loop broken, the output is then amplified by a multiple that is just the loop gain, as if the whole system were simplified to be a Thévenin equivalent, where, by inserting a equivalent impedance, the circuit performance is not changed. 

Likewise, derivation of two RC networks in p.1005, fig. 11.28 are also omitted, which actually took me very long to verify. Do they think Butterworth or Chebyshev filter’s transfer function (sec. 11.3) is easy to show, too? Even Wikipedia does a better job at explaining. 

Take p-n junction for further example, which later development lies heavily on. Readers are told that the built-in voltage given by eqn. (1.48) is such-and-such, depletion width given by (1.50) is such-and-such, and so on. Indeed, physics is not the main theme of this course. But some derivation seems short enough to justify their inclusion, actually. By assuming particle-in-the-box in doped silicon and Maxwell-Boltzmann approximation of energy levels, we may express carrier concentration in terms of energy. Moreover, the Gauss law of electric field, when integrated, relates charge to voltage; this way, charge conservation gives (1.50). 

As another example, in chap. 12, it was stated that (p.1043) the pole should lie at right half of s-plane (where s is Laplace dummy variable), and that loop gain, Aβ, should be made slightly greater than unity. Sedra / Smith did not even mention the physical significance of Laplace transformation, which would take less then one page. It goes like this: a rational function of s in frequency domain is a product of step function and an exponentials in time domain; in order that this does diverge, poles are made to have real part positive. 

Okay, maybe this is supposed an introductory book so some formula are not shown. But why then always is the end-of-chapter problems suddenly become incredibly difficult? At times, I can stare at the solution manual for twenty minutes and still not knowing what it is mainly about. All in all, some of the involved design examples should really have been placed in the main text, while omitting trivial circuit analyses. 

And why, on the other hand, Sedra / Smith can spend twenty pages on very similar and straightforward large-signal BJT calculation (p.248 to 264), and fifty pages on routine calculation of four configurations of feedback (sec. 9.4, 9.5, 9.6, 9.7, p.782 to 831). Besides these, endless in-text examples that mainly is just plugging in numbers. 

Take for another example chap. 8, where Sedra / Smith spent some dozens of pages on frequency responses of common-emitter and of common-source MOSFET. First they analyzed the low frequency response by considering separately each capacitance while setting others opened (sec. 8.1.1, p.659). As of the high frequency response, on one hand they analyzed by applying Miller formula on the gate-to-drain direction only (sec. 8.3.1, p.682), and again by doing the same on both gate-to-drain and drain-to-gate directions (sec. 8.5.2, p.702). But they also, alternatively, considered separately each capacitance while setting others shorted (sec. 8.5.3, p.705), and lastly calculated exact transfer function as well (sec. 8.5.4, p.702). And everything done on common-source amplifier is done again to common-emitter one. 

And various parasite capacitance are treated at great length (p.670ff). However, their estimated values offers little insight, and seem to be merely experimental, fitting numbers. For an introductory book, it would probably suffice to take account of all kinds parasite capacitance for by introducing a single equivalent C placed at the output node — students may well look up tables to find C_gd or C_gs, if needed. 

This is very fine, but If the reader can read to up to page 700, he or she probably could perform the circuit analysis himself. Meanwhile, valuable comments on their performance features and comparison are scattered among trivial details.

Being nearly 1500 pages long, a relevant presentation is not easy. Positive feedback chap. 12 harks back to chap. 9 when the negative feedback concept was introduced. SRAM, as the application of differential amplifier in chap. 7 and current steering circuits in chap. 6, is introduced as late as chap. 16. 

I am not to dogmatically insist what material must be included in such course, but merely to point out this seem to be an unsuccessful mixture of reference-oriented book and introductory book. Sedra / Smith can be explaining simple things to extreme detail, yet be extremely condensed in exposition of difficult ones. They might have attempted to incorporate both reference-book depth while aiming at novice. At the end, I think, neither side is pleased. Sedra / Smith would be too lengthy for one already having had exposure to IC design, and quite often impossible to learn for just-beginning students of circuits. 

I am actually a student of communication; nevertheless, integration circuit, I believe, is too an engaging field. It is a pity that this book, so commonly assigned as textbook, might actually put students off, especially when, with considerable organization and trimming, it could have been a very great textbook.