cmos digital integrated circuits a first course pdf

CMOS Digital Integrated
Circuits: A First Course

www.TechnicalBooksPDF.com


www.TechnicalBooksPDF.com


CMOS Digital Integrated
Circuits: A First Course
Charles Hawkins,
Jaume Segura,
and
Payman Zarkesh-Ha
University of Florida
University of Balearic Islands
University of New Mexico

Edison, NJ
scitechpub.com

www.TechnicalBooksPDF.com


Published by SciTech Publishing, an imprint of the IET.
www.scitechpub.com
www.theiet.org

Copyright © 2013 by SciTech Publishing, Edison, NJ. All rights reserved.

No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means,
electronic, mechanical, photocopying, recording, scanning or otherwise, except as permitted under Sections 107 or 108 of
the 1976 United Stated Copyright Act, without either the prior written permission of the Publisher, or authorization through
payment of the appropriate per-copy fee to the Copyright Clearance Center, 222 Rosewood Drive, Danvers, MA 01923,
(978) 750-8400, fax (978) 646-8600, or on the web at copyright.com. Requests to the Publisher for permission should be
addressed to The Institution of Engineering and Technology, Michael Faraday House, Six Hills Way, Stevenage, Herts, SG1
2AY, United Kingdom.
While the author and publisher believe that the information and guidance given in this work are correct, all parties must
rely upon their own skill and judgement when making use of them. Neither the author nor publisher assumes any liability
to anyone for any loss or damage caused by any error or omission in the work, whether such an error or omission is the
result of negligence or any other cause. Any and all such liability is disclaimed.

10 9 8 7 6 5 4 3 2 1
ISBN 978-1-61353-002-3

Typeset in India by MPS Ltd
Printed in the USA by Sheridan Books, Inc.

www.TechnicalBooksPDF.com


CONTENTS
Preface

xiii

Introduction

1

xix

Transistors and Computers—Until Death Do They Part

xx

Transistors and Computers—How Deep Can the Friendship Go?

xxi

Computers—Is There a Limit?

xxiii

Future

xxiv

Basic Logic Gate and Circuit Theory

1

1.1 Logic Gates and Boolean Algebra

1

1.2 Boolean and Logic Gate Reduction

5

1.3 Sequential Circuits

7

1.4 Voltage and Current Laws

9

1.4.1
1.4.2
1.4.3
1.4.4

Terminal Resistance Analysis by Inspection
Kirchhoff’s Voltage Law and Analysis by Inspection
Kirchhoff’s Current Law and Analysis by Inspection
Mixing Voltage and Current Divider Analysis by Inspection

9
12
14
16

1.5 Power Loss in Resistors

18

1.6 Capacitance

21

1.6.1
1.6.2

Capacitor Energy and Power
Capacitive Voltage Dividers

22
24

1.7 Inductance

26

1.8 Diode Nonlinear Circuit Analysis

27

1.8.1

Diode Resistor Analysis

28

1.9 Some Words about Power

31

1.10 Summary

32
v

www.TechnicalBooksPDF.com


vi

2

Contents

Semiconductor Physics

39

2.1 Material Fundamentals

39

2.1.1
2.1.2
2.1.3

Metals, Insulators, and Semiconductors
Carriers in Semiconductors: Electrons and Holes
Determining Carrier Concentrations

2.2 Intrinsic and Extrinsic Semiconductors
2.2.1 n-Type Semiconductors

2.2.2 p-Type Semiconductors
2.2.3 Carrier Concentration in n- and p-Doped Semiconductors

2.3 Carrier Transport in Semiconductors
2.3.1
2.3.2

45
48
49

51
51
52

56

2.5 Biasing the pn Junction

59

The pn Junction under Forward Bias
The pn Junction under Reverse Biasing

60
60

2.6 Diode Junction Capacitance

62

2.7 Summary

64

MOSFET Transistors

67

3.1 Principles of Operation

67

3.1.1 The MOSFET as a Digital Switch
3.1.2 Physical Structure of MOSFETs
3.1.3 MOS Transistor Operation: a Descriptive Approach

4

45

2.4 The pn Junction
2.5.1
2.5.2

3

Drift Current
Diffusion Current

39
41
43

68
69
70

3.2 MOSFET Input Characteristics

73

3.3 nMOS Transistor Output Characteristics and Circuit Analysis

74

3.4 pMOS Transistor Output Characteristics and Circuit Analysis

83

3.5 MOSFET with Source and Drain Resistors

89

3.6 Threshold Voltage in MOS Transistors

90

3.7 Summary

93

Metal Interconnection Properties

99

4.1 Metal Interconnect Resistance

100

4.1.1
4.1.2
4.1.3

Resistance and Thermal Effects
Sheet Resistance
Via Resistance

www.TechnicalBooksPDF.com

103
104
106


Contents

vii

4.2 Capacitance

4.2.1
4.2.2

110

Parallel Plate Model
Capacitive Power

4.3 Inductance

113

4.3.1
4.3.2
4.3.3

113
115
116

Inductive Voltage
Line Inductance
Inductive Power

4.4 Interconnect RC Models
4.4.1
4.4.2

C-model for Short Lines
RC Model for Long Lines

4.5 Summary
5

117
117
119

122

The CMOS Inverter

125

5.1 The CMOS Inverter

125

5.2 Voltage Transfer Curve

127

5.3 Noise Margins

129

5.4 Symmetrical Voltage Transfer Curve (VTC)

131

5.5 Current Transfer Curve

132

5.6 Graphical Analysis of VTC

134

5.6.1
5.6.2

Static Transfer Curves
Dynamic Transfer Curves

134
138

5.7 Inverter Transition Speed Model

140

5.8 CMOS Inverter Power

143

5.8.1
5.8.2
5.8.3

6

110
112

Transient Power
Short-Circuit Power
Quiescent Leakage Power

143
145
147

5.9 Power and Power Supply Scaling

147

5.10 Sizing Inverter Buffers to Drive Large Loads

150

5.11 Summary

152

CMOS NAND, NOR, and Transmission Gates
6.1 NAND Gates
6.1.1
6.1.2

157

157

Electronic Operation
NAND Noncontrolling Logic State

6.2 NAND Gate Transistor Sizing

www.TechnicalBooksPDF.com

158
159

162


viii

Contents

6.3 NOR Gates
6.3.1
6.3.2

164

Electronic Operation
NOR Noncontrolling Logic State

6.4 NOR Gate Transistor Sizing

168

6.5 Pass Gates and CMOS Transmission Gates

173

6.5.1
6.5.2
6.5.3

Pass Gates
CMOS Transmission Gates
Tristate Logic Gates

6.6 Summary
7

173
175
176

177

CMOS Circuit Design Styles

185

7.1 Boolean Algebra to Transistor Schematic Transformation

185

7.2 Synthesis of DeMorgan Circuits

190

7.3 Dynamic CMOS Logic

194

7.3.1
7.3.2

8

164
165

Dynamic CMOS Logic Properties
Charge Sharing in Dynamic Circuits

194
196

7.4 Domino CMOS Logic

199

7.5 NORA CMOS Logic

202

7.6 Pass Transistor Logic

203

7.7 CMOS Transmission Gate Logic Design

205

7.8 Power and Activity Coefficient

206

7.9 Summary

211

Sequential Logic Gate Design and Timing
8.1 CMOS Latches
8.1.1
8.1.2

219
221

Clocked Latch
Gated Latches

221
222

8.2 Edge-Triggered Storage Element

223

8.2.1
8.2.2
8.2.3

The D-FF
Clock Logic States
A Tristate D-FF Design

8.3 Timing Rules for Edge-Triggered Flip-Flops
8.3.1
8.3.2

Timing Measurements
Timing Rule Violation Effect

www.TechnicalBooksPDF.com

224
225
226

228
228
230

Contents

ix

8.4 Application of D-FFs in ICs

231

8.5 tsu and thold with Delay Elements

232

8.6 Edge-Triggered Flip-Flop with Set and Reset

235

8.7 Clock Generation Circuitry

236

8.7.1

237

8.8 Metal Interconnect Parasitic Effects

241

8.9 Timing Skew and Jitter

241

8.10 Overall System Timing in Chip Designs

242

8.10.1
8.10.2
8.10.3
8.10.4

9

Phase Locked Loop Circuit

Period Constraint
Period Constraint and Skew
Hold Time Constraint
Period Constraint with Skew and Jitter

243
244
245
246

8.11 Timing and Environmental Noise

249

8.12 Summary

251

IC Memory Circuits

261

9.1 Memory Circuit Organization

262

9.2 Memory Cell

264

9.3 Memory Decoders

266

9.3.1
9.3.2

Row Decoders
Column Decoders

266
267

9.4 The Read Operation

269

9.5 Sizing Transistor Width to Length Ratio for Read Operation

270

9.6 Memory Write Operation

272

9.6.1
9.6.2

Cell Write Operation
Latch Transfer Curve

272
273

9.7 Sizing Transistor Width to Length Ratio for Write Operation

273

9.8 Column Write Circuits

277

9.9 Read Operation and Sense Amplifier

278

9.10 Dynamic Memories

281

9.10.1 3-Transistor DRAM Cell
9.10.2 1-Transistor DRAM Cell

9.11 Summary

283
284

285

www.TechnicalBooksPDF.com


x

Contents

10 Programmable Logic—FPGAs
10.1

10.2

10.3

10.4

287

A Simple Programmable Circuit—The PLA

288

10.1.1 Programmable Logic Gates
10.1.2 AND/OR Matrix Gates

288
290

The Next Step: Implementing Sequential Circuits—The CPLDs

292

10.2.1 Incorporating Sequential Blocks—The Complex Programmable
Logic Device (CPLD)
10.2.2 Advanced CPLDs

292
294

Advanced Programmable Logic Circuits—The FPGA

300

10.3.1

10.3.2
10.3.3
10.3.4
10.3.5

Actel ACT FGPAs
Xilinx Spartan FPGAs
Altera Cyclone III FPGAs
Today’s FPGAs
Working with FPGAs—Design Tools

302
304
307
309
309

Understanding the Programmable Technology

310

10.4.1 Antifuse Technology
10.4.2 EEPROM Technology
10.4.3 Static RAM Switch Technology

311
313
314

11 CMOS Circuit Layout

317

11.1

Layout and Design Rules

317

11.2

Layout Approach: Boolean Equations, Transistor Schematic,
and Stick Diagrams

319

11.3

Laying out a Circuit with PowerPoint

321

11.4

Design Rules and Minimum Layout Spacing

322

11.5

Laying out a CMOS Inverter

323

11.5.1
11.5.2
11.5.3
11.5.4

323
325
325
326

11.6

pMOS Transistor Layout
Revisiting the Design Rules of the pMOS Transistor Layout
nMOS Transistor Layout
Merging Transistors to a Common Polygate

Completed CMOS Inverter Drawn to Design Rule
Minimum Dimensions

327

11.7

Multi-Input Logic Gate Layouts

328

11.8

Merging Logic Gate Standard Cell Layouts

331

11.9

More on Layout

333

www.TechnicalBooksPDF.com


Contents

xi

11.10 Layout CAD Tools

334

11.11 Summary

334

12 How Chips Are Made

335

12.1

IC Fabrication Overview

336

12.2

Wafer Construction

336

12.3

Front and Back End of Line Fabrication

337

12.4

FOL Fab Techniques

338

12.4.1
12.4.2
12.4.3

12.4.4

338
339
340
341

Oxidation of Silicon
Photolithography
Etching
Deposition and Implantation

12.5

Cleaning and Safety Operations

342

12.6

Transistor Fabrication

344

12.7

Back End of Line BOL Fab Techniques

344

12.7.1 Sputtering
12.7.2 Dual Damascene
12.7.3 Interlevel Dielectric and Final Passivation

345
346
347

Fabricating a CMOS Inverter

347

12.8.1 Front End of Line Operation
12.8.2 Back End of Line Operation

347
348

Die Packaging

349

12.8

12.9

12.10 IC Testing

350

12.11 Summary

350

Answers to all Even Numbered end of Chapter Exercises

351

Index

365

www.TechnicalBooksPDF.com


www.TechnicalBooksPDF.com


PREFACE
This book teaches introductory complementary metal-oxide-semiconductor (CMOS) digital electronics for electrical and computer engineering undergraduates. For many years
the CMOS technology has dominated the method of designing and manufacturing digital
(computing) integrated circuits. The selection of material here is not significantly different from the graduate texts by J. Rabaey et al. (Digital Integrated Circuits, 2003, Prentice
Hall), N. Weste and D. Harris (CMOS VLSI Design, 2011, Addison-Wesley), or J. Baker
(CMOS: Circuit Design, Layout, and Simulation, 2010, Wiley-IEEE Press), but the style
is introductory with many examples, self-exercises, and end-of-chapter problems.
This book initially reviews material relevant to digital electronics that students learned
in previous circuit and logic courses. The book then moves through chapters on basic
physics of semiconductor materials and diodes; nMOS and pMOS field effect transistors
circuit analysis; electronic properties of the metal interconnections; the CMOS inverter;
the CMOS NAND, NOR, and transmission gates electronics; transformation from Boolean

equations to CMOS transistor schematics and domino circuits; timing electronics; memory
circuits; field-programmable gate arrays (FPGAs); CMOS layout; and CMOS fabrication
basics. The emphasis is on transistor level electronics.
The principles of power dissipation are introduced with numerical examples. Lowering
circuit power has special urgency today where total Internet power consumes about 10%
of US electrical power generation.
Other features and objectives include:
• There are abundant examples, self-exercises with answers, and many problems at the
end of chapters to give students reflexive skills in transistor circuit analysis.
• This course can be taught before or after a companion class in introductory analog
electronics.
• The book strives for clarity and self-learning in an undergraduate presentation.
• The book doesn’t overwhelm students with too much details; it defines teaching goals
consistent with what they will take forward to the next level of electronics.
• Students are provided with an education that serves as a prerequisite for graduate or
senior courses in digital electronics and allows entry level into the digital electronics
industry.
• The book is light enough for students to carry to class.
xiii

www.TechnicalBooksPDF.com


xiv

Preface

Chapter Summaries
Chapter 1 reviews relevant logic theory that includes Boolean equation to logic gate
schematics, DeMorgan’s theorem, logic equivalence, and logic gate reduction. Basic circuit

theory is next, with emphasis on analysis of terminal impedance, node voltages, and branch
currents by inspection. Nonlinear circuit analysis techniques are introduced using the diode
and its nonlinear current-voltage expression. Capacitor and inductor properties and circuits
are reviewed, as is the power wasted in resistive and capacitive circuits. These topics are
a few among many but are selected for their relevance in the digital circuit analysis that
follows.
The second chapter introduces the semiconductor physics that underlie device operation. The goal is to impart a good visual model of the physics of materials and diodes, and to
use basic equations for better understanding. Semiconductor physics is a complex subject
that can involve more than one course at the graduate level, and Chapter 2 cannot replicate this. However, visual models of semiconductor materials and diodes are important
because engineers often use qualitative language to communicate important properties
of the physics of semiconductor diodes and transistors. Students should be able to answer the question, “How do diodes (and transistors) work and perform basic parameter
calculations?”. This chapter leads directly into Chapter 3 on field effect transistors.
CMOS circuits use two transistors types; the nMOS and pMOS field effect transistors.
Chapter 3 describes how these transistors work, followed by numerical analysis of circuit
node voltages and currents. Many examples, self-exercises, and end-of-chapter problems
give students the reflexive response to analyze transistor digital circuits. Equal treatment
is given to each transistor type.
Chapter 4 deals with metal properties, which are especially relevant in modern circuits
since chip total metal length may be on the order of several miles, and minimum metal
dimensions can be 22 nm or smaller. Metal properties are a major concern in attaining
maximum IC frequency and minimum noise operation, and metal physics deserves as
much study as does the transistor.
In Chapter 5, the CMOS inverter is discussed. The CMOS inverter is the most abundant
logic gate in any digital integrated circuit (IC). It has one nMOS and one pMOS field
effect transistor. This chapter introduces about a dozen important electronic properties,
with numerical examples. Inverter properties are inherently important but are also the
basis for electronic properties of NAND, NOR, and sequential logic circuits such as the
master-slave flip-flop. Inverter power dissipation properties are emphasized.
Chapter 6 covers NAND and NOR gates, which build on the inverter by placing transistors in parallel and series to the inverter pair. These multi-input logic gates have all of the
electronic properties of the inverter and a few that are unique. The electronic basis for the

noncontrolling logic state is described in this chapter as it relates to circuit debugging, test
engineering, and schematic reading. Pass transistor and CMOS transmission gate properties conclude the chapter. Transmission gates are abundant, comprising half of the logic
gates in the master-slave flip-flop.

www.TechnicalBooksPDF.com


Preface

xv

Chapter 7 develops design styles that assemble transistors into logic gates. It begins
with a relatively simple technique to transform Boolean equations into a CMOS transistor
schematic that performs the logic. Other design styles are presented along with the reasons
for having different styles. Power dissipation is analyzed with a technique that allows a
power comparison of different combinational logic configurations.
Chapter 8 discusses the accurate design and placement of timing signals, which may be
the most challenging task for a designer. This often neglected undergraduate course topic
is emphasized, giving it the importance it deserves. The edge-triggered flip-flop (FF) has
a complexity that must be mastered. Timing parameters and rules must be exact otherwise
circuits will fail. System-level timing builds on these foundations and introduces system
timing parameters and constraints.
Chapter 9 covers memory circuits. They have always been embedded within the computing chips, but today microprocessor chips may dedicate more than 70% of the total
transistor count to these memory circuits. Therefore, special emphasis is given on these
static random access memory (SRAM) designs. Transistor sizing of SRAM cells is developed with numerical examples. Another high-volume memory design is dynamic random
access memory (DRAM). This single transistor memory cell has different properties.
Chapter 10 looks at a unique and popular design style using field-programmable gate
arrays (FPGA). This material follows from other design styles described in the preceding
chapter. The electronics and method of operation are different, but FPGAs are common
and abundant enough to devote a chapter.

In Chapter 11, the CMOS layout is discussed. A conversion occurs in the design process
when transistor schematics are transformed to rectangular images on a photographic mask.
The images represent transistor and metal line geometries. Masks are drawn for each of
several layers in the buildup of the IC. Layout is not electronics but is the necessary first
step in using photolithography to make the tiny transistors and metal interconnections.
The mask layout step is introduced using manual layout of the inverter, NAND, and NOR
gates. Several commercial layout tools exist, but cost and training time led us to consider
the Microsoft PowerPoint program to draw the layouts. PowerPoint is typically available
on all computers, training time is minimal, it appears to have long-term stability in the
market, and students get a better grounding in design rules. PowerPoint has been successful
as a teaching tool in the classroom for layout of simple logic gates circuits.
Chapter 12 describes the chemical, physical, and photolithography techniques that
actually make the final circuit. This chapter is qualitative but sufficient enough to allow
students to converse on the various sequenced fabrication techniques that achieve the end
circuit result of the chip.

Comments for Instructors
The book uses long channel models for MOSFET analysis, even though short channel
models are common in industry. The reasons are twofold. First, the short channel models are often simplified for undergraduate presentation where they lose accuracy. Also,

www.TechnicalBooksPDF.com


xvi

Preface

full short channel models become too complex for hand calculations. Although the long
channel models are also not accurate, they allow manual problem solving insight into
the various bias regions of the transistors. We originally designed the book using short

channel models, but found the simplified analytical expressions clumsy and inaccurate. A
second observation is that modern industry electronic papers and oral presentations often
refer to long channel models despite the use of short channel transistors. It is part of the
language. The more accurate short channel models are best left to graduate courses and
detailed computer models.
Other choices were made to avoid overly complex material at this undergraduate stage.
The subjects and their depth were a trade-off between designing a one-term course and covering the important topics. For example, combinational logic power analysis uses the truth
table analysis rather than logical effort. Chapter 9 on memory keeps the timing description
simple but to the point. Memory design deserves a whole book for a full description.
The problems in this book most efficiently use the modern equation solving ability of
scientific calculators. One great learning advantage is that time is spent on the problem
itself and little on the grind of manually solving with quadratic equations or iteration. An
unknown variable can be embedded anywhere in the equation, and the scientific calculator
doesn’t care. It solves for the unknown variable in seconds. Students and instructors can
solve these problems any way they desire, but the scientific calculator is truly an advance
in modern digital circuit teaching.

A Suggested Semester Chapter Order
CHAPTER

TITLE

TIME IN CLASS

Chapter 1
Chapter 2
Chapter 3
Chapter 5
Chapter 6
Chapter 7

Chapter 11
Chapter 12
Chapter 4
Chapter 8
Chapter 9
Chapter 10

Basic Logic Gates and Circuit Theory
Semiconductor Physics
MOSFET Transistors
CMOS Inverter
CMOS NAND, NOR, and Transmission Gates
CMOS Design Styles
CMOS Circuit Layout
How Chips are Made
Metal Interconnection Properties
Sequential Logic Gate Design and Timing
Memory Circuits
FPGAs

1 week
1 week
2 weeks
1.5 weeks
1 week
1.5 weeks
1 week
1 week
1 week
2 weeks

1 week
0.5 week

Chapter 4 on metal interconnects logically fits with device descriptions, but it interrupts
the flow of electronic circuitry so it was put later. Chapters 11 and 12 continue the emphasis
on circuitry and the IC before returning to Chapter 4.

www.TechnicalBooksPDF.com


Preface

xvii

Author Background
This book reflects the experience of the authors, who have taught this material at the
graduate and undergraduate level and have worked closely with the digital electronic
industry in their careers. Hawkins and Segura did sabbaticals with the Intel Corporation:
Segura at the Intel campus in Portland Oregon, and Hawkins in Rio Rancho, New Mexico.
Segura also did sabbatical work at Philips Semiconductor and received numerous research
contracts from industry. Hawkins worked closely with the Sandia National Laboratory in
New Mexico for over 20 years in its CMOS integrated circuits group. Both authors have
long histories of committee work for the European DATE conference, the International
Test Conference, and the VLSI Test Symposium. Hawkins was editor of the Electron
Device Failure Analysis magazine.
Payman Zarkesh-Ha is professor in the ECE Department at the University of New
Mexico (UNM). He teaches graduate and undergraduate VLSI, digital, and analog electronics. Prior to joining UNM, he worked for five years at LSI Logic Corp, where he
worked on interconnect architecture design for the next ASIC generations. He has published more than 60 refereed papers and holds 12 issued patents. His research interests are
statistical modeling of nanoelectronics devices and systems, and design for manufacturability, low power, and high performance VLSI designs. All of these activities outside of
the classroom influenced our choice of material and style in the book. It is long overdue for

electrical and computer engineering undergraduate students to rid themselves of outdated
logic circuits and receive a course dedicated to digital CMOS electronics.

www.TechnicalBooksPDF.com


www.TechnicalBooksPDF.com


INTRODUCTION
Any sufficiently advanced technology is indistinguishable from magic.
Arthur C. Clarke’s Third Law
The goal of this book is to prepare you to contribute to the computer evolution in the 21st
century. It is about the electronics that propel the incredible surge in human communication and knowledge capability. The foundation of a computer is the transistor. Computer
electronics deals with transistor-level behavior of circuits that perform all of the computer logic operations such as adding, multiplying, storing, comparing, and any operation
described by Boolean equations. Billions of transistors and their wire connections are
embedded in small, thin, rectangular silicon computer chips. The total wire connections
on these tiny chips may be several miles in length, and power dissipation may range from a
few microwatts to over 200 watts. The chip is also referred to as an integrated circuit (IC).
Chips are complicated, and electrical and computer engineers must understand computing
at this circuit level.
Engineers face challenges. How would you blend digital circuit knowledge with computer architecture to design a chip? How fast do we want to clock the computer, and where
do we start? How do you interface chips on a circuit board? How much heat from chip
power loss can you stand—how do you minimize it? As a customer, how do you talk to a
chip designer? When your first chips are returned from the factory to evaluate and something is wrong, where do you begin to solve the problem? Failures may be temperature or
power supply dependent and not simple static Boolean errors. What skills and knowledge
do you need to identify and correct these failures? Whether you are an engineer at the chip
level or you design at the higher board and system level, the solutions often reside with
knowledge of chip properties at the transistor level.
A knowledge hierarchy exists in electronics. Semiconductor physics describes diode

and transistor action using model equations that allow calculation of transistor circuit node
voltages and path currents. Specific transistor configurations then form the different logic
gates, such as the inverter, NAND, NOR, transmission gates, the D-flip-flop, and more
complex combinational logics gates derived from arbitrary Boolean statements. These
logic gates electronically perform the Boolean operations that define the computer, and
xix

www.TechnicalBooksPDF.com


xx

Introduction

we must understand their properties. What are the voltage, current, temperature, power
drain, propagation delay time, and noise margins properties?
A master clock oscillator drives sequential circuits with pulses that synchronize data
movement of the Boolean operations in the computer. Clock speed is an important parameter and often the first specification that a buyer looks at when shopping for a computer. The
amount of computer memory may be the next question. Memory subcircuits are extensively
built into the computer chips. So, what is a standard memory cell and how are memories
organized? Modern computer chips may dedicate over 70% of the total transistor count to
embedded memory. Memory embedded in the chip allows faster computing as opposed
to sending signals back and forth to external memory chips mounted on a circuit board.
We might take for granted our computer-based miracles, such as the Internet, cell
phone magic, email, Google, automobile electronics, biomedical instrumentation, GPS,
YouTube, instant news, weather, and sports, automobile electronics, e-books, Facebook,
and, yes, video games. You might ask, “Hasn’t it always been like this?” The answer is
no—the applications didn’t really get rolling until the early 1990s, and all of these modern
products depended on fast, cheap, and small computer chips.

Transistors and Computers—Until Death Do They Part
To get a better sense of our subject, let us track electronic progress in digital computer
development and then its role in the Internet. We see not only the march of computers
to smaller, faster, and cheaper but also the fascinating interplay of diverse forces. The
Internet did not grow in a vacuum, and neither did computers.
The first computer circuit we are aware of was called the flip-flop by its English inventors, Eccles and Jordan, almost 100 years ago. A flip-flop remains stable in one of
two voltage states until triggered to the other state by an external electrical pulse. The
flip-flop stores a voltage state. Computers were not thought of at that time so the flip-flop
remained dormant for many years. But today up to millions of flip-flops exist in every
computer chip from the advanced Internet server chip to the chips in modern coffee makers
or dishwashers. Flip-flops are at the heart of synchronizing data transfer.
In the late 1930s primitive computers combined Boolean algebra with mechanical
switches to demonstrate simple computing machines. The Second World War sparked an
interest in using computers for scientific calculations. The first vacuum tube computer
was the ENIAC at the University of Pennsylvania in 1946. By the standards of its day, the
100 kHz clock was fast. It weighed 30 tons, was 80×8.5×3.5 feet, and dissipated 150 kW of
power. The old flip-flop was now an integral part of computer electronics. But the vacuum
tube was a relatively large device requiring a glass enclosed vacuum and a heated metal filament. Tubes had poor reliability and were a challenge to cool. Something better was needed.
Bell Labs had a vision in the 1930s that a small, switching device could be constructed
in a pure solid material. Bell Labs was thinking of replacing the slow, clunky mechanical
relays in their telephone switching centers and not about computer development. In 1947,
they struck gold with demonstration of a small, solid-state device called the transistor.

www.TechnicalBooksPDF.com


Introduction

xxi

Approximately five years later, transistor computers emerged in production from several
companies. Transistors were a giant step toward smaller, cooler, and more reliable computers. These computers used discrete (individual) transistors that were mounted in small
metal cans and were not the small, integrated circuit chips with billions of transistors that
were to follow. These mainframe computers as they were called still required a cooled,
dedicated room, but steady progress was made into the 1970s when another revolution
occurred.
Actually several things happened at the transistor level. The first was a rapid transition
from the original Bell Labs transistor called a bipolar junction transistor to a newer device
called a metal oxide field effect transistor (MOSFET) transistor. The MOSFET was blended
in a unique design style called CMOS that was markedly cooler. The cooler CMOS allowed
more transistors to be placed on a single chip without overheating thus increasing the
computer functionality. CMOS also had the unusual property that if the transistor size
were shrunk, the transistor would operate faster.
A third feature was that the smaller size of a CMOS transistor allowed more chips to
be manufactured in a single operation than before because the total chip size could now
be reduced. More chips could be accommodated per process run, and that drives the cost
down. Often industry left the chip the same size and just added more transistors to increase
functionality.
A final feature is that if the small particle defects that kill the chips in a production run
remain the same density, then packing more chips in the same area will increase the fraction
of good chips, (i.e., the yield). This gives a marked cost savings. CMOS has dominated
computing chip design since around 1980, and CMOS technology today remains the focus
of intense development.
It is a manufacturing miracle that next-generation chips could be sold for a lower price
if the next-generation transistor was smaller. That was huge, and today you still pay about
the same price (or less) for a personal computer as one that is a few years older. And these
newer chips go faster and give more functionality while keeping the chip temperature
under control. These CMOS features really fueled the development of computing chips.
The reader should pause and dwell on the significance. What other product offers more
dramatic performance each year for the same price or less?

Transistors and Computers—How Deep Can the Friendship Go?
In the early 1970s, Intel brought out the first microprocessors, first the 4-bit and then the
8-bit. Product innovation leaped on these transistor level advances. In 1974, the MITS
Corporation in Albuquerque, New Mexico, offered the first personal computer, the PC.
The MITS Altair 8800 was a primitive PC requiring code to be entered by toggle switches,
but it had a video monitor and was the size of a typewriter. It had a 2 MHz clock and
cost $498 assembled. It was also the first computer to be personally owned. It used a
single microprocessor chip, the Intel 8080, to perform the computer function, and many
engineers bought the Altair out of curiosity. Interestingly, Bill Gates and Paul Allen of

www.TechnicalBooksPDF.com


xxii

Introduction

the new Microsoft Corporation in Albuquerque wrote BASIC for the MITS Altair PC.
In 1977, Apple launched the Apple II PC for $1200. No one had ever had a computer at
that price, size, and capability—especially one they could call their own. But the IBM PC
launched in 1980 had more impact because it brought in the business sector. There was no
looking back. Businesses were being freed from the tedium of the big, central computer
room, and later travelers found they could do work on the road with the coming of laptops.
The ubiquitous typewriter was on the way out.
The PC launched a revolution in information accessibility that could not have been
imagined. Technology and novel business enterprises were beginning to move together.
A partnering of technology, business enterprise, and government support at crucial points
drove this revolution. But a monster enterprise called the Internet lay quietly awaiting its
entrance.

In the 1960s and 1970s, Internet development was marching in the background to its
own beat driven by engineers and scientists who wanted to use each other’s specialty
computers across the country. It was government funding through the Advanced Research
Projects Agency (ARPA) that allowed a mainframe computer from the University of
California at Los Angeles (UCLA) to use an interface unit, called the Interface Message
Processor (IMP), to talk to a similar hookup at Stanford University in October 1969. Long
distance sharing of computer resources had happened. Messages, later called email, were
exchanged, but the ability to do this was regarded then as a secondary feature and not a
big deal. In fact, the first Internet exchange was not widely publicized. The response was
sort of, “Isn’t that nice that scientists and engineers can use each other’s computers, but
that won’t affect my life.” What an understatement.
The next necessary development occurred in 1989 when PC manufacturers began
bundling internal modems in the PC. The Internet was now open to anyone. Email grew
at a tremendous rate as users found it a good business tool, and as true today it was just
plain fun to use. The mouse and graphical displays were huge steps toward friendly computers. And computer chips doubled their speed and transistor density about every two
years following what is called Moore’s law. Then spam, viruses, and hackers showed their
ugly heads. Spam is expensive in system bandwidth and the required electrical energy
generation to support its Internet hunger.
The Internet went global with introduction of the World Wide Web. “www” was a
concept from CERN in Europe that was demonstrated in 1991. We now see “www” in
our URL addresses. Browsers quickly followed with MOSAIC from the University of
Illinois and the NETSCAPE browser from Netscape Corporation. Yahoo and Microsoft
entered the competition, and the famous browser wars were on. Two students from Stanford
introduced Google in 1998 with a novel concept in searches that became so successful
that Google is now a verb.
Although clearly visible in these early applications, it was the special talents of the
business entrepreneurs that carried the World Wide Web into its most recent surge. The
incredible innovations now seem endless. The list of Amazon, eBay, PayPal, Google,
Wikipedia, YouTube, bloggers, Refdesk, Facebook, email delivery each morning of your

www.TechnicalBooksPDF.com


Introduction

xxiii

favorite newspaper, instant check on stocks and weather, instant Google satellite maps
of the earth, online business carried on across the globe, tweeters, e-books, and many
more brings us to our present state of information availability. These applications required
computer chips that were faster, smaller, and cheaper. The miracle applications needed
the base technologies.

Computers—Is There a Limit?
Computer chips depend on many disciplines. Electrical and computer engineers, computer scientists, mathematicians, physicists, chemical engineers, chemists, mechanical
engineers, statisticians, manufacturing engineers, and marketing people work in harmony
to achieve these miracle products. Technology products typically develop from an idea
and a prototype demonstration. If the idea is sound, the product undergoes continuous
improvement until performance limits are reached. How far can we push performance?
Let’s look at three other technologies to see their performance trajectory and how that
might provide clues to our electronic future.
Train development spanned an increasing performance era from around 1820 to the
1950s. Then, except for the bullet train, it was over. Automobile development spanned
from the 1890s to about the 1960s. Speed, comfort, and engine power peaked for trains
and automobiles. Commercial aircraft basically spanned from 1903 to the 1960s when the
Boeing 747 was produced. Speed and passenger carrying capability maxed out. Later in
all three areas, the integrated circuit caused a second revolution in the 1980s, but if you
lived in the height of these technology rushes, there was a feeling that “progress” would
never end. The basic speed, power, and transport capability did end, though. What does
this imply for computers?

Will the CMOS computer technology rush end? Will our computers provide dramatically more functionality each year? For several reasons, we believe that CMOS technology
will hit a performance limit. If history is a guide, we will then squeeze every last design
and manufacturing detail from our chips, and improvements will lie in more efficient
manufacturing and using multiple processors on one chip. When will we see that soft end
point? We see signs of reaching some of the limits now, so we hazard an educated guess
that the CMOS technology limit may be reached by 2025. That is a guess. The exact date is
immaterial to the thought that a performance limit exists within our professional lifetimes.
When CMOS performance development ends, we expect research will continue seeking
another manufacturable electronic technology. The urge is strong to build faster, smaller,
cheaper computers, and it will be novel transistor or transistor like devices that pushes us
even further.
One significant challenge deals with electrical power. Tom Friedman in his book “Hot,
Flat, and Crowded” (p. 31) quoted a Sun Microsystem engineer who put future power
demand in perspective. He observed that the earth will add about one billion persons in
the next 12 years. If each person were given a 60-watt light bulb then 60 billion watts of
power generation would be required. If each person used that bulb for an average of four

www.TechnicalBooksPDF.com


xxiv

Introduction

hours per day, then that average 10 billion watts would require about 20 coal-fired power
plants. If each of the billion persons were also allowed to use a 120-watt computer for four
hours then we would need an additional 40 power plants. The earth is power limited, and
the pressure on low power computer chip design is huge.
The concept of technology is little different from early mankind using a wheel to support
a cart to carry heavy weights. Technology is the use of materials and natural laws to ease

our burdens. Electronic technology is no different, but we know that all technology has a
downside. The Internet has done miracles, but that hasn’t stopped hackers, spammers, and
swindlers from peddling their dark objectives. The Internet can bring instant and accurate
news, but it can also bring instant and inaccurate propaganda. These are issues to deal
with as it has always been with technology. We should keep our eye on the benefits and
continue the historic human battle of fighting the misuse of technology.

Future
We won’t speculate much on the future other than that there is one. The Internet brought
rapid changes in business and technology that we now take for granted. Startling new
products will appear, and some old product names will disappear. This book addresses the
education of the next generation of engineers who will continue to move this historic epic
in information accessibility.

www.TechnicalBooksPDF.com