How to use your abacus Reading a number on the abacus. The abacus works on the place value system. Reading it is almost like reading a written numeral. The five beads below the bar each have a value of 1. The two beads above the bar each have a value of 5. The beads which are pushed against the bar represent the number. The number on the abacus is 2,364. Thousands Hundreds Tens Ones

2 3 6 4 Adding on the abacus. Suppose you want to add 2,364+3,473. To do this put 2364 on the abacus. You need to move 3 to the center on the right-hand string. There aren't three singles. Instead, you can move a 5 to the center and move 2 1's back. Move 2 ones and one 5 to the center on the tens string. You have two 5's on the tens string so you can regroup. Move the two 5's on the tens string away from the center and move a 1 to the center on the hundreds string. Now you need to move 4 to the center on the hundreds string. To do this move 5 to the center and one away from the center on the hundreds string. Last step: move 3 ones to the center on the thousands string. You now have five ones at the center on the thousands string, so you move them away and replace them with a 5. Here's the result:

5 8 3 7

  1  

Four important concepts have shaped the history of computing:! the mechanization of arithmetic;! the stored program;! the graphical user interface;! the computer network.

The following timeline of the history of computing shows some of the important events and devicesthat have implemented these concepts, especially the first two. Additional information about theseand the other two important concepts follow the timeline.

MACHINES TO DO ARITHMETICThe term computer dates back to the 1600s. However, untilthe 1950s, the term referred almost exclusively to a humanwho performed computations.

For human beings, the task of performing large amounts ofcomputation is one that is laborious, time consuming, anderror prone. Thus, the human desire to mechanize arithmeticis an ancient one. One of the earliest “personal calculators”was the abacus, with movable beads strung on rods to countand to do calculations. Although its exact origin is unknown,the abacus was used by the Chinese perhaps 3000 to 4000years ago and is still used today throughout Asia. Early mer-chants used the abacus in trading transactions.

The ancient British stone monument Stonehenge, locatednear Salisbury, England, was built between 1900 and 1600B.C. and, evidently, was used to predict the changes of theseasons.

In the twelfth century, a Persian teacher of mathematics inBaghdad, Muhammad ibn-Musa al-Khowarizm, developedsome of the first step-by-step procedures for doing computa-tions. The word algorithm used for such procedures isderived from his name.

In Western Europe, the Scottish mathematician John Napier(1550–1617) designed a set of ivory rods (called Napier’sbones) to assist with doing multiplications. Napier also devel-oped tables of logarithms and other multiplication machines.

The videotape series entitled “The Machine That Changed The World” is highlyrecommended by the authors. For information about it, see http://ei.cs.vt.edu/~his-tory/TMTCTW.html. A Jacquard Loom, Hollerith’s tabulator, the ENIAC, UNI-VAC, early chips, and other computer artifacts can also be viewed at the NationalMuseum of American History of the Smithsonian Institution in washington, D.C.Also see this book’s Web site for more information about the history of computing.

EARLY CALCULATORS

3000 B.C. ABACUS

1900-1600 B.C. STONEHENGE

12TH CENTURY:AL-KHOWARIZM

1612 NAPIER’S BONES

0.2 Part of the Picture The History of Computing #

ANChp0wTLV5_052002.qxp 5/20/02 10:53 AM Page 5

     

  2  

The English mathematician William Oughtred invented a cir-cular slide rule in the early 1600s. Slide rules were based onNapier’s logarithms, and more modern ones like that shownhere were used by engineers and scientists through the 1950sand into the 1960s to do rapid approximate computations.

The young French mathematician Blaise Pascal (1623–1662)invented one of the first mechanical adding machines to helphis father with calculating taxes. It used a series of eight ten-toothed wheels (one tooth for each decimal digit), which wereconnected so that numbers could be added or subtracted bymoving the wheels.

The “Pascaline” was a digital calculator, because it repre-sented numerical information as discrete digits, as opposed to agraduated scale like that used in analog instruments of meas-urement such as nondigital clocks and thermometers. Eachdigit was represented by a gear that had ten different positions(a ten-state device) so that it could “count” from 0 through 9and, upon reaching 10, would reset to 0 and advance the gearin the next column so as to represent the action of “carrying” tothe next digit.

Although Pascal built more than 50 of his adding machines,his commercial venture failed because the devices could notbe built with sufficient precision for practical use.

The German mathematician Gottfried Wilhelm von Leibnizinvented an improved mechanical calculator that, like thePascaline, used a system of gears and dials to do calculations.However, it was more reliable and accurate than the Pascalineand could perform all four of the basic arithmetic operationsof addition, subtraction, multiplication, and division.

A number of other mechanical calculators followed that fur-ther refined Pascal’s and Leibniz’s designs, and by the end ofthe nineteenth century, these calculators had become importanttools in science, business, and commerce.1673 LEIBNIZ’ CALCULATOR

1642 PASCALINE

1630 SLIDE RULE

#

ANChp0wTLV5_052002.qxp 5/20/02 10:53 AM Page 6

 

  3  

THE STORED PROGRAMThe fundamental idea that distinguishes computers fromcalculators is the concept of a stored program that controlsthe computation. A program is a sequence of instructions thatthe computer follows to solve some problem. An income taxform is a good analogy. While a calculator can be a useful toolin the process, computing taxes involves much more thanarithmetic. To produce the correct result, one must execute theform’s precise sequence of steps of writing numbers down(storage), looking numbers up (retrieval), and computation toproduce the correct result. Likewise, a computer program is aprecise sequence of steps designed to accomplish somehuman task.

The stored program concept also gives the computer itsamazing versatility. Unlike most other machines, which areengineered to mechanize a single task, a computer can be pro-grammed to perform many different tasks—that is, the choiceof task is deferred to the user. This is the fascinating paradoxof the computer: Although its hardware is designed for a veryspecific task—the mechanization of arithmetic—computersoftware programs enable the computer to perform a dizzyingarray of human tasks, from navigational control of the spaceshuttle to word processing to musical composition. For thisreason, the computer is sometimes called the universalmachine.

An early example of a stored program automaticallycontrolling a hardware device can be found in the weavingloom invented in 1801 by the Frenchman Joseph MarieJacquard. Holes punched in metal cards directed the actionof this loom: A hole punched in one of the cards would enableits corresponding thread to come through and be incorporatedinto the weave at a given point in the process; the absence ofa hole would exclude an undesired thread. To change to a dif-ferent weaving pattern, the operator of this loom would sim-ply switch to another set of cards. Jacquard’s loom is thus oneof the first examples of a programmable machine, and manylater computers would make similar use of punched cards.

The punched card’s present-or-absent hole also marks theearly occurrence of another key concept in the history of com-puting—the two-state device, which refers to any mechanismfor which there are only two possible conditions. Within adecade, thousands of automated looms were being used inEurope, threatening the traditional weaver’s way of life. Inprotest, English weavers who called themselves Luddites riot-ed and destroyed several of the new looms and cards. Some ofthe Luddites were hanged for their actions. (The term Ludditeis still used today to refer to someone who is skeptical of newtechnology.)

1801 JACQUARD LOOM

#

ANChp0wTLV5_052002.qxp 5/20/02 10:53 AM Page 7

 

  4  

The two fundamental concepts of mechanized calculation andstored program control were combined by the English mathe-matician Charles Babbage (1792–1871). In Babbage’s lifetime,humans involved in almost any form of computation relied heav-ily upon books of mathematical tables that contained the resultsof calculations that had already been performed by others.However, such mathematical tables took far too long for humansto produce and were typically rife with errors. Moreover, worldtravel, the Industrial Revolution, and other new scientific andeconomic realities had produced an explosion in the need formathematical computations. It was clear to Babbage that “humancomputers” were simply not up to the task of supplying thedemand.

In 1822, supported by the British government, Babbage beganwork on a machine that he called the Difference Engine.Comprised of a system of gears, the Difference Engine wasdesigned to compute polynomials for preparing mathematicaltables. Babbage continued this work until 1833, when he aban-doned this effort having completed only part of the machine.According to Doron Swade, curator of the London ScienceMuseum, the cantankerous Babbage argued with his engineer,ran out of money, and was beset by personal rivalry.

In 1833, Babbage began the design of a much more sophisti-cated machine that he called his Analytical Engine, which wasto have over 50,000 components. The operation of this machinewas to be far more versatile and fully automatic, controlled byprograms stored on punched cards, an idea based on Jacquard’searlier work. In fact, as Babbage himself observed: “The analo-gy of the Analytical Engine with this well-known process is near-ly perfect.”

The basic design of Babbage’s Analytical Engine correspond-ed remarkably to that of modern computers in that it involved thefour primary operations of a computer system: processing, stor-age, input, and output. It included a mill for carrying out thearithmetic computations according to a sequence of instructions(like the central processing unit in modern machines); the storewas the machine’s memory for storing up to 1,000 50-digit num-bers and intermediate results; input was to be by means ofpunched cards; output was to be printed; and other componentswere designed for the transfer of information between compo-nents. When completed, it would have been as large as a loco-motive, been powered by steam, and able to calculate to six dec-imal places of accuracy very rapidly and print out results, all ofwhich was to be controlled by a stored program!

Babbage’s machine was not built during his lifetime, but it isnevertheless an important part of the history of computingbecause many of the concepts of its design are used in moderncomputers. For this reason, Babbage is sometimes called the“Father of Computing.”

MECHANICAL COMPUTERS

1822 BABBAGE’S DIFFERENCEENGINE

1833 BABBAGE’S ANALYTICALENGINE

#

ANChp0wTLV5_052002.qxp 5/20/02 10:53 AM Page 8

     

  5  

Ada Augusta, Lord Byron’s daughter, was one of the fewpeople other than Babbage who understood the AnalyticalEngine’s design. This enabled her to develop “programs” forthe machine, and for this reason she is sometimes called “thefirst programmer.” She described the similarity of Jacquard’sand Babbage’s inventions: “The Analytical Engine weavesalgebraic patterns just as the Jacquard loom weaves flowersand leaves.” In the 1980s, the programming language Ada wasnamed in her honor.

During the next 100 years, little progress was made in realiz-ing Babbage’s dream. About the only noteworthy event dur-ing this time was the invention by Herman Hollerith of anelectric tabulating machine that could tally census statisticsthat had been stored on punched cards. There was a fear that,because of growing population, it would not be possible tocomplete processing of the 1890 census before the next onewas to be taken. Hollerith’s machine enabled the UnitedStates Census Bureau to complete the 1890 census in 2 1/2years. The Hollerith Tabulating Company later merged withother companies to form the International Business Machines(IBM) Corporation.

Much of Babbage’s dream was finally realized in the “Z”series of computers developed by the young German engineerKonrad Zuse in the 1930s. Ingeniously, Zuse designed hiscomputers to mechanize arithmetic of binary numbers ratherthan that of decimal numbers. Because there are only twobinary digits, 0 and 1, Zuse could construct his machine fromtwo-state devices instead of ten-state devices, thus greatlysimplifying the engineering of his computer. The two-statedevice Zuse deployed was the electromechanical relay, a two-position switch that would either complete or break the circuitconnecting two phone lines. This mechanism was in wide usein the telephone industry to automate connections previouslymanaged by human operators.

However, Zuse ultimately grew dissatisfied with the slowspeed at which the relay switched from one state to the other.His assistant, Helmut Schreyer, made the brilliant suggestionof using vacuum tubes, which could switch between states—on and off—electronically, thousands of times faster than anymechanical device involving moving parts. In the middle ofWorld War II, however, Adolf Hitler was convinced that vic-tory was near and refused to fund Zuse’s proposal to build thefirst fully electronic computer.

1842 ADA AUGUSTA

1890 HOLLERITH’S TABULATINGMACHINE

1935-1938 KONRAD ZUSE

ELECTROMECHANICALCOMPUTERS

#

ANChp0wTLV5_052002.qxp 5/20/02 10:53 AM Page 9

 

  6  

 

In addition to building electromechanical computers, KonradZuse in 1945 designed a high-level programming languagethat he named Plankalkül. Although Zuse wrote programsusing this language, it was never actually implemented due toa lack of funding. As a result, it lay in obscurity until 1972when Zuse’s manuscripts were discovered. This language wasamazingly sophisticated for its time—over 15 years passedbefore its features began to appear in other languages. Zusedesigned programs to perform tasks as diverse as integer andfloating-point arithmetic, sorting lists of numbers, and playingchess.

World War II also spurred the development of computingdevices in the United States, Britain, and Europe. In Britain,Alan Turing developed the universal machine concept, form-ing the basis of computability theory. (See Chapter 4.)During World War II, he was part of a team whose task was todecrypt intercepted messages of the German forces. Severalmachines resulted from this British war effort, one of whichwas the Collosus, finished in 1943.

The best-known computer built before 1945 was the HarvardMark I (whose full name was the Harvard–IBM AutomaticSequence Controlled Calculator). Like Zuse’s “Z” machines,it was driven by electromechanical relay technology.Repeating much of the work of Babbage, Howard Aiken andothers at IBM constructed this large, automatic, general-pur-pose, electromechanical calculator. It was sponsored by theU.S. Navy and (like Babbage’s machines) was intended tocompute mathematical and navigational tables.

The first fully electronic binary computer, the ABC(Atanasoff–Berry Computer), was developed by JohnAtanasoff and Clifford Berry at Iowa State University during1937-1942. It introduced the ideas of binary arithmetic,regenerative memory, and logic circuits.

Unfortunately, because the ABC was never patented andbecause others failed at the time to see its utility, it took threedecades before Atanasoff and Berry received recognition forthis remarkable technology. Although the ENIAC(1943–1946) bore the title of the first fully electronic comput-er for many years, a historic 1973 court decision ruled thatAtanasoff was the legal inventor of the first electronic digitalcomputer.

1937 ALAN TURING

1944 MARK I

1936-1939 ATANASOFF’SELECTRONIC DIGITALCOMPUTER (ABC)

EARLY ELECTRONICCOMPUTERS

#

ANChp0wTLV5_052002.qxp 5/20/02 10:53 AM Page 10

 

  7  

Grace Murray Hopper (1907–1992) began work as a coder—what we today would call a programmer—for the Mark I in1944. In the late 1950s, “Grandma COBOL,” as she has affec-tionately been called, led the effort to develop the COBOL pro-gramming language for business applications.

The actual physical components that make up a computer sys-tem are its hardware. Several generations of computers can beidentified by their type of hardware. First-generation com-puters are characterized by their extensive use of vacuumtubes. Although they could do calculations much more rapidlythan mechanical and electromechanical computers, the heatgenerated by large numbers of vacuum tubes and their shortlifetimes led to frequent failures.

The ENIAC (Electronic Numerical Integrator and Computer)is arguably the best known of the early electronic computers(and long thought to be the first). It was designed by J.Presper Eckert and John Mauchly, who began work on it in1943 at the Moore School of Engineering at the University ofPennsylvania. When it was completed in 1946, this 30-tonmachine had 18,000 vacuum tubes, 70,000 resistors, and 5million soldered joints and consumed 160 kilowatts of electri-cal power. Stories are told of how the lights in Philadelphiadimmed when the ENIAC was operating.

This extremely large machine could multiply numbersapproximately 1000 times faster than the Mark I, but it wasquite limited in its applications and was used primarily by theArmy Ordnance Department to calculate firing tables and tra-jectories for various types of artillery shells. The instructionsthat controlled the ENIAC’s operation were entered into themachine by rewiring some of the computer’s circuits. Thiscomplicated process was very time consuming, sometimestaking a number of people several days; during this time, thecomputer was idle. In other early computers, the instructionswere stored outside the machine on punched cards or someother medium and were transferred into the machine one at atime for interpretation and execution. Unfortunately, becauseof the relative slowness of the moving parts of mechanicalinput devices in comparison to the electronic parts of the com-puter dedicated to processing, such computers would alwaysfinish executing the instruction long before the next instructionwas finished loading. Thus, again, the processing portion ofthe computer was sitting idle too much of the time.

FIRST-GENERATIONCOMPUTERS

1945-1956 FIRST-GENERATIONCOMPUTERS—VACUUM TUBES

1943-1946 ENIAC

1944 GRACE HOPPER

#

ANChp0wTLV5_052002.qxp 5/20/02 10:53 AM Page 11

     

  8  

In 1945, John von Neumann wrote “First Draft of a Reporton the EDVAC (Electronic Discrete Variable AutomaticComputer)” computer in which he described a scheme thatrequired program instructions to be stored internally beforeexecution. This led to his being credited as the inventor of thestored-program concept. The architectural design hedescribed is still known as the von Neumann architecture(although there is evidence that others including Eckert andMauchly and Zuse had similar ideas several years beforethis).

The advantage of executing instructions from a computer’smemory rather than directly from a mechanical input deviceis that it eliminates time that the computer must spend wait-ing for instructions. Instructions can be processed more rap-idly and more importantly; they can be modified by the com-puter itself while computations are taking place. The intro-duction of this scheme to computer architecture was crucialto the development of general-purpose computers.

While working on the Mark II computer, Grace Hopper foundone of the first computer “bugs”—an actual bug stuck in oneof the thousands of relays that has been preserved in theNational Museum of American History of the SmithsonianInstitution. She glued it into the logbook, and subsequentefforts to find the cause of machine stoppage were reported toAiken as “debugging the computer.”

Eckert and Mauchly left the University of Pennsylvania toform the Eckert–Mauchly Computer Corporation, which builtthe UNIVAC (Universal Automatic Computer). Started in1946 and completed in 1951, it was the first commerciallyavailable computer designed for both scientific and businessapplications. The UNIVAC achieved instant fame partly dueto its correct (albeit unbelieved) prediction on national televi-sion of the election of President Eisenhower in the 1952 U.S.presidential election, based upon 5% of the returns. UNIVACsoon became the common name for computers.

Soon afterward, because of various setbacks, Eckert andMauchly sold their company to the Remington–RandCorporation, who sold the first UNIVAC to the CensusBureau in 1951.

1945 COMPUTER BUG

1945 JOHN VON NEUMANN’S“FIRST DRAFT OF AREPORT ON THE EDVAC”

1951 UNIVAC

#

ANChp0wTLV5_052002.qxp 5/20/02 10:53 AM Page 12

     

  9  

SECOND-GENERATIONCOMPUTERS

Second-generation computers, built between 1956 and1963, used transistors in place of the large, cumbersome vac-uum tubes, marking the beginning of the great computershrinkage. These computers were smaller, faster, required lesspower, generated far less heat, and were more reliable thantheir predecessors. They were also less expensive.

Early computers were difficult to use because of the com-plex coding schemes used to represent programs and data. Akey development during the late 1950s and early 1960s wasthe development of programming languages that made itmuch easier to develop programs.

In 1957, after three years of work, John Backus and his col-leagues delivered the first FORTRAN (FORmulaTRANslation) compiler for the IBM 704. Their first reportcommented that a programmer was able to write and debug infour to five hours a program that would have taken severaldays to complete before. FORTRAN has undergone severalrevisions and remains a powerful language for scientific com-puting.

In 1958, IBM introduced the first of the second-generationcomputers (the 7090 and other computers in their 7000series), vaulting IBM from computer obscurity to first placein the computer industry.

Also in 1958, as part of his work in developing artificial intel-ligence, John McCarthy developed the programming lan-guage LISP (LISt Processing) for manipulating strings ofsymbols, a non-numeric processing language.

Since 1952, Grace Hopper had been developing a series ofnatural-language-like programming languages for use inbusiness data processing. This culminated in 1960 with thedevelopment of COBOL (COmmon Business OrientedLanguage) by an industry-wide team. Since then, more pro-grams have been written in COBOL than in any other pro-gramming language.

Another language that appeared in 1960 was ALGOL 60(ALGOrithmic Language), which became the basis of manyprogramming languages that followed, such as Pascal.

1956–1963 SECOND GENERATIONCOMPUTERS—EARLY TRANSISTORS

1957 FORTRAN

1958 IBM 7090

LISP

1960 COBOL

ALGOL 60

#

ANChp0wTLV5_052002.qxp 5/20/02 10:53 AM Page 13

 

  10  

1964-1971 THIRD-GENERATIONCOMPUTERS—CHIPS ANDINTEGRATED CIRCUITS

THIRD-GENERATIONCOMPUTERS

Third-generation computers used integrated circuits (IC,chips), which first became commercially available from theFairchild Corporation. These ICs were based on the pioneer-ing work of Jack Kilby and Robert Noyce.

It was also during this period that, in addition to improvedhardware, computer manufacturers began to develop collec-tions of programs known as system software, which madecomputers easier to use. One of the more important advancesin this area was the third-generation development of operat-ing systems. Two important early operating systems still usedtoday are Unix (1971) and MS-DOS (1981).

The IBM System/360, introduced in 1964, is commonlyaccepted as the first of the third generation of computers.Orders for this family of mutually compatible computers andperipherals climbed to 1000 per month within two years

In 1965, Digital Equipment Corporation introduced the PDP-8, the first commercially successful minicomputer. Becauseof its speed, small size, and reasonable cost— $18,000, lessthan 20% of the six-digit price tag for an IBM 360 main-frame—it became a popular computer in many scientificestablishments, small businesses, and manufacturing plants.

1964 THE IBM SYSTEM/360

1965 PDP-8#

ANChp0wTLV5_052002.qxp 5/20/02 10:53 AM Page 14

     

  11  

In 1968, Douglas Engelbart and his research team worked atdeveloping a more user-friendly form of computing, usableby average persons and for purposes other than numericalcomputation. Engelbart’s inventions anticipated many of theattributes of personal computing, including the mouse, wordprocessor, windowed interfaces, integrated “Help,” and linkedtext that would later be termed “hypertext.”

Disillusioned by how work on the multiuser operating systemMultics was proceeding, Ken Thompson of Bell TelephoneLaboratories began work in 1969 on a simpler OS aimed atthe single user. His first implementation of Unix was writtenin the assembly language of a spare Digital EquipmentCorporation PDP-7 computer. In a pun on the name Multics,the new operating system was named Unix.

Unix is still undergoing development today and has becomeone of the most popular operating systems. It is the onlyoperating system that has been implemented on computersranging from microcomputers to supercomputers.

Another noteworthy event began in 1969 when the AdvancedResearch Projects Agency (ARPA) of the U.S. Department ofDefense introduced the ARPANET, a network linking com-puters at some of the department’s university research centers.Transmissions between the ARPANET computers traveled inthe form of packets, each of which was addressed so that itcould be routed to its destination. As more and more hostswere added to the ARPANET backbone, it became known asthe Internet.

1968 DOUGLAS ENGELBART:COMPUTER MOUSE, TWO-DIMENSIONAL DISPLAY, EDITING, HYPERMEDIA

1969 KEN THOMPSON: UNIX

ARPANET—THE BEGINNING OF

THE INTERNET

#

ANChp0wTLV5_052002.qxp 5/20/02 10:53 AM Page 15

 

  12  

Computers from the 1980s on, commonly called fourth-gen-eration computers, use very large-scale integrated (VLSI)circuits on silicon chips and other microelectronic advancesto shrink their size and cost still more while enlarging theircapabilities. A typical chip is equivalent to millions of tran-sistors, is smaller than a baby’s fingernail, weighs a smallfraction of an ounce, requires only a trickle of power, andcosts but a few dollars.

The first chip was the 4004 chip designed by Intel’s TedHoff, giving birth to the microprocessor, which marked thebeginning of the fourth generation of computers. This, alongwith the first use of an 8-inch floppy disk at IBM, ushered inthe era of the personal computer.

Robert Noyce, one of the cofounders of the IntelCorporation (which introduced the 4004 microprocessor in1971), contrasted microcomputers with the ENIAC as fol-lows:

An individual integrated circuit on a chip perhaps a quarter ofan inch square now can embrace more electronic elementsthan the most complex piece of electronic equipment thatcould be built in 1950. Today’s microcomputer, at a cost ofperhaps $300, has more computing capacity than the firstelectronic computer, ENIAC. It is twenty times faster, has alarger memory, consumes the power of a light bulb rather thanthat of a locomotive, occupies 1/30,000 the volume and costs1/10,000 as much. It is available by mail order or at your localhobby shop.

To simplify the task of transferring the Unix operating systemto other computers, Ken Thompson began to search for ahigh-level language in which to rewrite Unix. None of the lan-guages in existence at the time were appropriate; therefore, in1970, Thompson began designing a new language called B.By 1972, it had become apparent that B was not adequate forimplementing Unix. At that time, Dennis Ritchie, also at BellLabs, designed a successor language to B that he called C,and approximately 90 percent of Unix was rewritten in C.

Other noteworthy events in 1973 included the follow-ing:

! Ethernet, the basis for LANs (Local Area Networks), wasdeveloped at Xerox PARC by Robert Metcalf

! A district court in Minneapolis ruled that John Atanasoffwas the legal inventor of the first electronic digital computer,thus invalidating Eckert’s and Mauchly’s patent.

1971 INTEL 4004 CHIP

FOURTH-GENERATIONCOMPUTERS

1973 DENNIS RITCHIE: C

ETHERNET

HISTORIC COURT DECISIONREGARDING FIRST ELECTRONIC

COMPUTER#

ANChp0wTLV5_052002.qxp 5/20/02 10:53 AM Page 16

     

  13  

Noteworthy in 1974:

! The MITS Altair 8800 hobby-kit computer was inventedby Edward Roberts (who coined the term personal comput-er), William Yates, and Jim Bybee. It was driven by the 8-bitIntel 8080 chip, had 256 bytes of memory, but no keyboard,no display, and no external storage. It sold for $300–400.

! Bill Gates and Paul Allen wrote a BASIC compiler for theAltair.

! Working in a garage, Steven Jobs and Steve Wozniakdeveloped the Apple I.

One of the most popular early personal computers was theApple II, introduced in 1976 by Steven Jobs and SteveWozniak. Because of its affordability and the availability ofbasic software applications, it was an immediate success,especially in schools and colleges.

The first supercomputer and the fastest machine of its day,the Cray I, developed by Seymour Cray, was also introducedin 1976. It was built in the shape of a C so components wouldbe close together, reducing the time for electronic signals totravel between them.

Also in 1976, Apple Corporation and Microsoft Corporationwere founded.

1974

ALTAIR

BASIC

JOBS & WOZNIAK: APPLE 1

1976 APPLE II

CRAY 1

APPLE CORP.MICROSOFT CORP.

#

ANChp0wTLV5_052002.qxp 5/20/02 10:53 AM Page 17

     

  14  

In 1981, IBM entered the personal computer market with theIBM PC, originally called the Acorn. Driven by the Intel 8-bit8088 chip, it used Microsoft’s DOS operating system underan agreement that gave Microsoft all the profits in exchangefor their having borne the development costs. MS-DOS thusbecame the most popular operating system for personal com-puters, and the PC established a microcomputer standardadopted by many other manufacturers of personal computers.

The IBM XT debuted the following year, sporting a 10-megabyte hard disk drive. The IBM AT followed in 1983,driven by the 16-bit Intel 80286 microprocessor, the first inthe line of Intel’s “80x86” chips.

By the late 1970s, a new approach to programming appearedon the scene—object-oriented programming (OOP)—thatemphasized the modeling of objects through classes andinheritance. A research group at Xerox’ Palo Alto ResearchCenter (PARC) created the first truly object-oriented lan-guage, named Smalltalk-80.

Another Bell Labs researcher, Bjarne Stroustrup, began thework of extending C with object-oriented features. In 1983,the redesigned and extended programming language C WithClasses was introduced with the new name C++.

Also in 1983

! Novell Data Systems introduced NetWare, a network oper-ating system (NOS), which made possible the construction ofa Local Area Network (LAN) of IBM PC-compatible micro-computers.

! Transmission Control Protocol/Internet Protocol(TCP/IP) became the official protocol governing transmittingand receiving of data via the ARPANET. Later that year, theUniversity of California at Berkeley released a new version ofBSD (also known as Berkeley UNIX), which includedTCP/IP, thus providing academic computing systems nation-wide with the technology to tie into the ARPANET. Explosivegrowth in the ARPANET resulted.

1981 IBM PC

1983 BJARNE STROUSTRUP: C++

NOVELLANNOUNCES NETWARE

TCP/IP

#

ANChp0wTLV5_052002.qxp 5/20/02 10:53 AM Page 18

     

  15  

Using a renowned Orwellian advertisement parodying thedowntrodden masses subservient to the IBM PC, Appleannounced in 1984 the Macintosh, a new personal computerdriven by the 32-bit Motorola 68000 microprocessor. Inspiredby Steve Jobs’ visit to Xerox PARC in 1979, the “Mac”brought the graphical user interface (GUI) to personal com-puting.

In 1985, Microsoft introduced Windows 1.0, its graphicaluser interface for IBM-PC compatibles. It was not until therelease of Windows 3.0 in 1990, however, that it gained wide-spread acceptance.

In 1986, Intel released the 32-bit 80386 chip (better knownas the “386” chip), which became the best-selling micro-processor in history. It contained 275,000 transistors. The80486, released in 1989, had more than a million.

In 1991, CERN (European Organization for NuclearResearch) introduced the World Wide Web, developed byTim Berners-Lee.

In 1992, Linus Torvalds developed Linux, a free version ofthe Unix operating system for PCs.

1984 MACINTOSH

1985 WINDOWS

1986 INTEL 386 CHIP

1991 TIM BERNERS–LEE: WWW

1992 LINUX

#

ANChp0wTLV5_052002.qxp 5/20/02 10:53 AM Page 19

     

  16  

Several noteworthy things happened in 1993:

! Intel introduced the 64-bit Pentium chip containing morethan 3 million transistors. The Pentium Pro released in 1995had more than 5.5 million. The Pentium II followed in 1997with 7.5 million transistors, and the Pentium III in 1999 withmore than 10 million.

! Motorola shipped the first PowerPC chip.

! The National Center for Supercomputing Applications(NCSA) at the University of Illinois released the first versionof Mosaic, the first graphical Web browser.

! Apple introduced the Newton, the first “palmtop” computer.

In 1994

! Netscape Navigator 1.0 was released.

! Yahoo!, the first major Web index, went online. It wasstarted in April 1994 by Electrical Engineering Ph.D. candi-dates at Stanford University, David Filo and Jerry Yang, as away to keep track of their personal interests on the Internet.

! Jeff Hawkins and Donna Dubinsky founded PalmComputing. The first Pilot was shipped in 1996.

In 1995, the new C++-based object-oriented programminglanguage Oak, developed at Sun Microsystems by JamesGosling, was renamed Java and burst onto the computerscene. Applications created in Java can be deployed withoutmodification to any computing platform, thus making ver-sions for different platforms unnecessary.

Other important events in 1995

! Microsoft introduced Windows 95.! Microsoft released Microsoft Internet Explorer 1.0 tocompete with the unforeseen popularity of Netscape.

! The U.S. Government turned the maintenance of theInternet backbone over to commercial networking companies.Commercial traffic was now allowed on the Internet. AmericaOnline, Compuserve, and Prodigy brought the Internet to thepublic.

1993 PENTIUM CHIPS

POWER PC CHIP

MOSAIC

APPLE NEWTON

1994 NETSCAPE NAVIGATOR 1.0

YAHOO!

1995 JAMES GOSLING: JAVA

WINDOWS 95INTERNET EXPLORER

INTERNET GOES COMMERCIAL

#

ANChp0wTLV5_052002.qxp 5/20/02 10:53 AM Page 20

 

  17  

! In 1999 more than $300 billion was spent worldwide in theyears leading up to Jan. 1, 2000 to solve the Y2K problem(also known as the millennium bug) —the inability of oldhardware and software to recognize the century changebecause years were stored with only two digits.

! Apple released the PowerMac G4

! In 2000 Microsoft launched Windows 2000

! AMD's Athlon and Intel's Pentium III broke the 1GHz barrier.

! In 2001 Apple released MacOS X

! Microsoft released Windows XP

! IBM's Almaden Research Center unveiled a quantum computer.

POWERMAC G4

2000WINDOWS 2000

1GHZ PROCESSORS

2001MAC OS X

WINDOWS XP

1999Y2K PROBLEM

2002QUANTUM COMPUTER

This summary of the history of computing has dealt mainly with the first two importantconcepts that have shaped the history of computers: the mechanization of arithmetic and thestored program concept. Looking back, we marvel at the advances in technology that have,in barely 50 years, led from ENIAC to today’s large array of computer systems, rangingfrom portable palmtop, laptop, and notebook computers to powerful desktop machinesknown as workstations, to supercomputers capable of performing billions of operationseach second, and to massively parallel computers, which use thousands of microprocessorsworking together in parallel to solve large problems. Someone once noted that if progressin the automotive industry had been as rapid as in computer technology since 1960, today’sautomobile would have an engine that is less than 0.1 inch in length, would get 120,000miles to a gallon of gas, have a top speed of 240,000 miles per hour, and would cost $4.

We have also seen how the stored program concept has led to the development of largecollections of programs that make computers easier to use. Chief among these is the devel-opment of operating systems, such as UNIX, Linux, MS-DOS, MacOS and Windows, thatallocate memory for programs and data and carry out many other supervisory functions.They also act as an interface between the user and the machine, interpreting commandsgiven by the user from the keyboard, by a mouse click, or by a spoken command and thendirecting the appropriate system software and hardware to carry them out.

The Graphical User InterfaceThe third key concept that has produced revolutionary change in the evolution of the computeris the graphical user interface (GUI). A user interface is the portion of a software programthat responds to commands from the user. User interfaces have evolved greatly in the past twodecades, in direct correlation to equally dramatic changes in the typical computer user.

#

ANChp0wTLV5_052002.qxp 5/20/02 10:53 AM Page 21

     

  18  

In the early 1980s, the personal computer burst onto the scene. However, at the outset,the personal computer did not suit the average person very well. The explosion in the amountof commercially available application software spared computer users the task of learningto program in order to compose their own software; for example, the mere availability of theLotus 1-2-3 spreadsheet software was enough to convince many to buy a PC. Even so, usinga computer still required learning many precise and cryptic commands, if not outright pro-gramming skills.

In the early 1980s, the Apple Corporation decided to take steps to remedy this situation. The AppleII, like its new competitor, the IBM PC,employed a command-line interface, requiring users to learn dif-ficult commands. In the late 1970s, Steve Jobs had visited Xerox PARC and had viewed several tech-nologies that amazed him: the laser printer, Ethernet, and the graphical user interface. It was the last ofthese that excited Jobs the most, for it offered the prospect of software that computer users could under-stand almost intuitively. In 1995 interview he said, “I remember within ten minutes of seeing the graph-ical user interface stuff, just knowing that every computer would work this way some day.”

Drawing upon child development theories, Xerox PARC had developed the graphicaluser interface for a prototype computer called the Alto that had been realized in 1972. TheAlto featured a new device that had been dubbed a “mouse” by its inventor, PARC researchscientist Douglas Engelbart. The mouse allowed the user to operate the computer by point-ing to icons and selecting options from menus. At the time, however, the cost of the hard-ware the Alto required made it unfeasible to market, and the brilliant concept went unused.Steve Jobs saw, however, that the same remarkable change in the computer hardware mar-ket that had made the personal computer feasible also made the graphical user interface areasonable possibility. In 1984, in a famous commercial first run during half-time of theSuper Bowl, Apple introduced the first GUI personal computer to the world: the Macin-tosh. In 1985, Microsoft responded with a competing product, the Windows operating sys-tem, but until Windows version 3.0 was released in 1990, Macintosh reigned unchallengedin the world of GUI microcomputing. Researchers at the Massachusetts Institute of Tech-nology also brought GUI to the UNIX platform with the release of the X Window systemin 1984.

The graphical user interface has made computers easy to use and has produced manynew computer users. At the same time, it has greatly changed the character of computing:computers are now expected to be “user friendly.” The personal computer, especially, mustindeed be “personal” for the average person and not just for computer programmers.

NetworksThe computer network is a fourth key concept that has greatly influenced the nature of mod-ern computing. Defined simply, a computer network consists of two or more computers thathave been connected in order to exchange resources. This could be hardware resources suchas processing power, storage, or access to a printer; software resources such as a data fileor access to a computer program; or messages between humans such as electronic mail ormultimedia World Wide Web pages.

As computers became smaller, cheaper, more common, more versatile, and easier to use,computer use rose and with it, the number of computer users. Thus, computers had to beshared. In the early 1960s, timesharing was introduced, in which several persons make si-multaneous use of a single computer called a host by way of a collection of terminals, eachof which consists of a keyboard for input and either a printer or a monitor to display out-

# Chapter 0 Beginning Snapshots

ANChp0wTLV5_052002.qxp 5/20/02 10:53 AM Page 22

     

  19  

put. With a modem (short for “modulator/demodulator,” because it both modulates binarydigits into sounds that can travel over a phone line and, at the other end, demodulates suchsounds back into bits), such a terminal connection could be over long distances.

Users, however, began to wish for the ability for one host computer to communicate withanother. For example, transferring files from one host to another typically meant transport-ing tapes from one location to the other. In the late 1960s, the Department of Defense beganexploring the development of a computer network by which its research centers at variousuniversities could share their computer resources with each other. In 1969, the ARPANETbegan by connecting research center computers, enabling them to share software and dataand to perform another kind of exchange that surprised everyone in terms of its popularity:electronic mail. Hosts were added to the ARPANET backbone in the 1970s, 1980s, and1990s at an exponential rate, producing a global digital infrastructure that came to be knownas the Internet.

Likewise, with the introduction of microcomputers in the late 1970s and early 1980s,users began to desire the ability for PCs to share resources. The invention of Ethernet net-work hardware and such network operating systems as Novell NetWare produced the LocalArea Network, or LAN, enabling PC users to share printers and other peripherals, disk stor-age, software programs, and more. Microsoft also included networking capability as a majorfeature of its Windows NT.

The growth of computer connectivity continues today at a surprising rate. Computersare becoming more and more common, and they are used in isolation less and less. With theadvent of affordable and widely available Internet Service Providers (ISPs), many homecomputers are now “wired” into a growing global digital infrastructure.

Exercises1. What are four important concepts in the early history of computation?

2. Match each item in the first column with the associated item in the second column

_____ John von Neumann A. early high-level language

_____ Charles Babbage B. first commercially available computer

_____ Blaise Pascal C. developed first fully electronic computer

_____ Herman Hollerith D. stored program concept

_____ Grace Murray Hopper E. Difference Engine

_____ Konrad Zuse F. designer of FORTRAN language

_____ Alan Turing G. Harvard Mark I

_____ Howard Aiken H. an early electronic computer

_____ John Backus I. integrated circuits (chips)

_____ Joseph Jacquard J. vacuum tubes

_____ Ada Augusta K. transistors

_____ John Atanasoff and Clifford Berry L. Apple Computer

_____ Bjarne Stroustrup M. automatic loom

_____ Steven Jobs and Steve Wozniak N. developed the UNIX operating system

_____ Ken Thompson O. developed the World Wide Web

0.2 Part of the Picture: History of Computing #

!

ANChp0wTLV5_052002.qxp 5/20/02 10:53 AM Page 23

 

Vacuum Tube Theory, a Basics Tutorial – Page 1Vacuum Tubes or Thermionic Valves come in many forms including the Diode, Triode, Tetrode, Pentode, Heptode and many more. These tubes have been manufactured by the millions in years gone by and even today the basic technology finds applications in today's electronics scene. It was the vacuum tube that first opened the way to what we know as electronics today, enabling first rectifiers and then active devices to be made and used.

Although Vacuum Tube technology may appear to be dated in the highly semiconductor orientated electronics industry, many Vacuum Tubes are still used today in applications ranging from vintage wireless sets to high power radio transmitters.

Until recently the most widely used thermionic device was the Cathode Ray Tube that was still manufactured by the million for use in television sets, computer monitors, oscilloscopes and a variety of other electronic equipment.

Concept of thermionic emission

Thermionic basics

The simplest form of vacuum tube is the Diode. It is ideal to use this as the first building block for explanations of the technology. It consists of two electrodes - a Cathode and an Anode held within an evacuated glass bulb, connections being made to them through the glass envelope.

If a Cathode is heated, it is found that electrons from the Cathode become increasingly active and as the temperature increases they can actually leave the Cathode and enter the surrounding space.

When an electron leaves the Cathode it leaves behind a positive charge, equal but opposite to that of the electron. In fact there are many millions of electrons leaving the Cathode. As unlike charges attract, this means that there is a force pulling the electrons back to the Cathode. Unless there are any further influences the electrons would stay in the vicinity of the Cathode, leaving the Cathode as a result of the energy given to them as a result of the temperature, but being pulled back by the positive charge on the Cathode.

Vacuum Tube Theory, a Basics Tutorial – Page 2The Diode – the simplest tube

In a Diode Vacuum Tube there is also another electrode called the Anode. If a positive potential is applied to this electrode, the electrons will be attracted by this potential and will move towards it if it is at a higher potential than the Cathode.

For the optimum performance the space between the Cathode and the Anode should be a vacuum. If there are any gas molecules in the space in which the electrons travel, collisions will occur and this will impede the flow of electrons. If an appreciable amount of gas is present, the electrons will ionise the gas, giving rise to a blue glow between the electrodes. In the early days of valves, it was thought that a certain amount of gas was necessary in the envelope. Later this was discovered that this was not the case and new "hard valves” were made that had a superior performance to the older "soft valves”. Very early radio receivers often used a soft valve for the detector stage and hard valves for the other stages.

Space charge

The electrons flowing between the Cathode and the Anode form a cloud which is known as the "space charge". It can tend to repel electrons leaving the Cathode, but if the potential applied to the Anode is sufficiently high then it will be overcome, and electrons will flow toward the Anode. In this way the circuit is completed and current flows.

As the potential is increased on the Anode, so the current increases until a point is reached where the space change is completely neutralised and the maximum emission from the Cathode is reached. At this point the emission can only be increased by increasing the Cathode temperature to increase the energy of the electrons and allow further electrons to leave the Cathode.

Concept of vacuum tube diode with cathode and anode

If the Anode potential is reversed, and made negative with respect to the Cathode it will repel the electrons. No electrons can be emitted from the Anode as it is not hot and no

Vacuum Tube Theory, a Basics Tutorial – Page 3current flows. This means that current can only flow in one direction. In other words the device only allows current in one direction, blocking it in the other. In view of this effect, the inventor of the Diode vacuum tube, Professor Sir Ambrose Fleming called it an "oscillation valve" in view of its one way action ?.

Control of current flow

Although the basic concept of the vacuum tube enabled a rectifier to be made, it does not allow for another form of control of the flow of electrons in the Anode circuit. However it was discovered that is a further potential was placed between the Cathode and the Anode this could be used to control the flow of electrons between the Cathode and Anode. Once the theoretical idea was devised, it was necessary to implement a way of placing this potential in the right place. An electrode known as a Grid, in the form of a thin mesh or wire through which the electrons could pass, was inserted between the Cathode and Anode.

It was found that by varying the potential on the Grid, this could alter the flow of electrons. The Grid is normally placed at a voltage below that of the Cathode so that it repels the electrons and counteracts the effect of the pull on the electrons from the potential on the Anode. If the voltage on the Grid is varied then it will vary or control the level of current flowing between the Cathode and the Anode. As such, this form of grid is known as a Control Grid. It makes the vacuum tube into an active device that is capable of amplifying signals.

Further grids

The basic thermionic tube with three electrodes is called a Triode in view of the number of electrodes. To improve the performance of the tube, further grids may be added. These tubes

are given generic names that describe the number of electrodes, and thereby giving an indication of the type of tube and performance.

Number of grids Total number of electrodes Generic name1 3 Triode2 4 Tetrode3 5 Pentode4 6 Hexode5 7 Heptode6 8 Octode

The basic concept of the vacuum tube outlined here enables signals to be rectified and amplified. Many refinements have been added in the form of further grids to enable much better performance to be obtained, but the principles involved are all the same.

Vacuum Tube Theory, a Basics Tutorial – Page 4Vacuum tube electrodes

The Cathode

There is a variety of different types of Cathode that are used in vacuum tubes. They differ in the construction of the Cathode and the materials used.

One of the major ways in which Cathodes can be categorised is by the way they are heated. The first type to be used was what is termed directly heated. Here a current is passed through a wire to heat it. In addition to providing the heat it also acts as the Cathode itself, emitting the electrons into the vacuum. This type of Cathode has the disadvantage that it must be connected to both the heater supply and the supply used for use in the Cathode - Anode circuit itself. This has disadvantages because it limits the way the circuit can be biased unless each heater is supplied separately and isolated from each other. A further disadvantage is that if an alternating current is used to provide the heating, this signal can be superimposed upon the main Cathode - Anode circuit, and there is a resultant hum at the frequency of the heater supply. The second type of Cathode is known as an indirectly heated Cathode. Here the heater is electrically disconnected from the Cathode, and heat is radiated from the heater to heat the Cathode. Although as a rule it takes longer for these types of tubes to warm up, they are almost universally used because of the flexibility this provides in biasing the circuits, and in isolating the Cathode - Anode circuit from the effects of hum from the heater supply.

The earliest type of Cathode is known as a bright emitter Cathode. This type of Cathode uses a tungsten wire heated to a temperature of between 2500 and 2600 K. Although not widely used these days, this type of Cathode was used in high power transmitting tubes such as those used for broadcasting. It suffers a number of drawbacks, one being that it is not particularly efficient in terms of the emission gained for the heat input. The life of the Cathode is also limited by the evaporation of the tungsten with failure occurring when about 10% of the tungsten has gone.

A further type of Cathode is known as a dull emitter. These Cathodes are directly heated and consist of thoriated tungsten. They provide more emission than a tungsten Cathode and require less heat, making the overall efficiency of the tube greater. Typically they run at a temperature of between 1900 and 2100 K. Although these Cathode normally have a relatively long life, they are fragile and any valves or tubes using them should be treated with care and they should not be subjected to technical shocks or vibration.

The type of Cathode that is in by far the greatest use is the oxide coated Cathode. These may be used with indirectly heated cathodes, unlike the tungsten and dull emitter Cathodes that must be directly heated as a result of the temperatures involved. This type of Cathode is normally in the form of nickel in the form of a ribbon, tube or even a small cup shape. This is coated with a mixture of barium and strontium carbonate, often with a trace of calcium added. During the manufacturing process the coating is heated to reduce it to its metallic form and the products of the chemical reaction are removed when the tube is finally evacuated. In this Cathode it is the barium that acts as the primary emitter and it operates at a much lower than the other types being in the region of 950 - 1050 K.

Vacuum Tube Theory, a Basics Tutorial – Page 5A wide variety of electron tubes have used radioactive material as a cold cathode - voltage regulators, spark-gap tubes, voltage sensitive switching tubes, glow lamps, etc. In general, such tubes consist of a gas filled glass envelope, a radioactive source, an Anode and an unheated (cold) Cathode. An interesting cold cathode tube was the 0Z4, a rectifier tube that was often used in tube car radios in the 1950’s. This tube did not use a radio active Cathode, it utilised a starter electrode and an Ionically heated cathode

More information on radio active cold Cathode devices on this web site:

https://www.orau.org/ptp/collection/consumer%20products/electrontubes.htm

The Anode (called the Plate in the early days of tube technology)

The Anode is generally formed into a cylinder so that it can surround the Cathode and any other electrodes that may be present. In this way the vacuum tube can be constructed in a tubular fashion and the Anode can collect the maximum number of electrons.

For the smaller tubes used in many radio receivers, the Anodes are generally made of nickel plated steel or simply from nickel. In some instances where larger amounts of heat need to be dissipated it may be carbonised to give it a matt back finish that enables it to radiate more heat out of the tube.

For applications where even higher powers are required, the Anode must be capable of dissipating even more heat, and operating at higher temperatures. For these tubes, materials including carbon, molybdenum, or zirconium may be used. Another approach is to build heatsink fins into the Anode structure to help radiate the additional heat. This approach is naturally limited by the construction of the device and the fact that the tube needs to be contained within its glass envelope. However a large heatsink structure will require the glass envelope to be much bigger, thereby increasing the costs.

To overcome this problem the Anode may be manufactured so that heat can be transferred outside the valve and removed using a forced air or a water jacket. Using this approach the envelope of the tube can be made relatively small, while still be able to handle significant levels of power.

The Grid

We have already discovered the Grid is the electrode by which the current flowing in the Anode circuit can be controlled by another potential. In the most basic form a vacuum tube may have one Grid. It is possible to use more than one to improve the performance or to enable additional functions to be performed. Accordingly tubes are named by the number of electrodes they contain that are associated with the electron flow. In other words the filaments or heaters and other similar elements are omitted.

A Grid is normally constructed in the form of a gauze mesh or a wire helix. If made of wire, it normally consists of nickel, molybdenum or an alloy and is wound using supporting rods that keep it clear of the Cathode. As such they may be wide, possibly oval in shape and they are generally made from copper or nickel.

Vacuum Tube Theory, a Basics Tutorial – Page 6To achieve a high level of performance that is repeatable, the tolerances within the vacuum tube must be maintained from one device to the next. In addition to this it is often necessary to mount the Grid only fractions of a millimetre away from the Cathode or other Grids. To be able to maintain these dimensions one approach that is adopted is to use a stiff rectangular frame and then wind the grid wire onto this under tension. This structure then needs to be fixed by the use of glazing or even gold brazing so that it remains firmly in place. Under some circumstances it may even be necessary to grind the cathode surface coating to ensure its flatness. This form of Grid is known as a Frame Grid. Look inside most tubes and you will see mica structures the support the elements.

One important aspect of the design of vacuum tubes is to ensure that the Grid does not overheat. This could lead to mechanical distortion and failure of the whole tube. To assist in the removal of heat the Grid wire may be carbonised, and often cooling fins may be attached to the Grid supporting wires. These supporting wires may also be welded directly to the connection pins in the base of the tube so that heat may be conducted away through the external connections.

A wide variety of vacuum tubes are available even today. Using the techniques that have been developed over many years they are able to offer excellent repeatability, performance and reliability.

The above information was adapted from the site http://www.radio-electronics.com/info/data/thermionic-valves/vacuum-tube-theory/tube-tutorial-basics.php

Naming the Grids

The first Grid is called the Control Grid (Grid 1), the second grid is called the Screen Grid (Grid 2) and in a Pentode the third grid is called a Suppressor Grid (Grid 3). With tubes with more than three Grids the other grids are usually named in the same way, Grid 4 Grid 5 etc. In many special purpose tubes with more Grids, some of the Grids may be internally connected to other elements.

The Triode

The Triode has a Cathode, a Control Grid and an Anode. Any where you have two conductors separated by an insulator you have capacitance, As a result, the Anode and Grid in a Triode Tube have capacitance, ( referred to as parasitic capacitance) between them. Because the tube inverts the signal the capacitance appears to be much bigger than it actually is. This is known as the Miller effect and accounts for the increase in the equivalent input capacitance of an inverting voltage amplifier due to amplification of the effect of capacitance between the input and output terminals.

The Miller capacitance between the input and the output of active devices like Vacuum Tubes is a major factor limiting their gain at high frequencies. Miller capacitance was identified in 1920 in T vacuum tubes by John Milton Miller. The Miller Capacitance also can cause instability in high frequency/high gain circuits. This same effect also applies to Transistor circuits.

Vacuum Tube Theory, a Basics Tutorial – Page 7The Tetrode, Pentode and Beam Tetrode

To combat the stability problems and limited voltage gain due to the Miller effect, the physicist Walter H. Schottky invented the Tetrode tube in 1919. He showed that the addition of a second Grid, located between the Control Grid and the Anode, known as the Screen Grid, could solve these problems.

"Screen" in this case refers to electrical "screening" or shielding, not physical construction - all "Grid" electrodes in between the Cathode and Anode are "screens" of some sort rather than solid electrodes since they must allow for the passage of electrons directly from the Cathode to the Anode).

A positive voltage slightly lower than the Anode voltage was applied the Screen Grid, and was bypassed (for high frequencies) to ground with a capacitor. This arrangement decoupled the Anode and the Control Grid, essentially eliminating the Miller capacitance and its associated problems. Consequently, higher voltage gains from a single tube became possible, reducing the number of tubes required in many circuits. This two-Grid tube is called a Tetrode, meaning four active electrodes, and was common by 1926.

However, the Tetrode had one new problem. In any tube, electrons strike the Anode with sufficient energy to cause the emission of electrons from its surface. In a Triode this so-called secondary emission of electrons is not important since they are simply re-captured by the more positive Anode. But in a Tetrode they can be captured by the Screen Grid also acting as an Anode, since it is also at a high voltage, thus robbing them from the Anode current and reducing the amplification of the device.

Since secondary electrons can outnumber the primary electrons, in the worst case, particularly as the Anode voltage dips below the Screen voltage, the Anode current can decrease with increasing Anode voltage. This is the so-called "Tetrode kink" and is an example of negative resistance which can itself cause instability. The otherwise undesirable negative resistance was exploited to produce an extremely simple oscillator circuit only requiring connection of the plate to a resonant LC circuit to oscillate; this was effective over a wide frequency range.

The solution was to add another Grid between the Screen Grid and the Anode, called the Suppressor Grid, since it suppressed secondary emission current toward the screen grid. This grid was held at the Cathode (or "ground”) voltage and its negative voltage (relative to the Anode) electrostatically repelled secondary electrons so that they would be collected by the Anode after all.

This three-grid tube is called a Pentode, meaning five electrodes. The Pentode was invented in 1926 by Bernard D. H. Tellegen and became generally favoured over the simple Tetrode. Pentodes are made in two classes: those with the suppressor grid wired internally to the Cathode and those with the Suppressor Grid wired to a separate pin for user access.

An alternative solution for power applications is the Beam Tetrode or "Beam Power Tube. This is a type of Tetrode vacuum tube with auxiliary beam-focusing Plates designed to

Vacuum Tube Theory, a Basics Tutorial – Page 8augment power-handling capability and help reduce unwanted emission effects. These tubes are usually used for power amplification, especially at audio-frequency.

Multifunction and multisection tubes

Superheterodyne receivers require a local oscillator and mixer, can use a tube that combines these two functions into a single Pentagrid Converter tube. Various alternatives such as using a combination of a Triode with a Hexode and even an Octode have been used for this purpose. The additional Grids include both Control Grids (at a low potential) and Screen Grids (at a high voltage). Many designs used such a Screen Grid as an additional Anode to provide feedback for the oscillator function, whose current was added to that of the incoming radio frequency signal.

To further reduce the cost and complexity of radio equipment, two separate structures (Triode and Pentode for instance) could be combined in the bulb of a single multisection tube. An early example was the Loewe 3NF. This 1920s device had three Triodes in a single glass envelope together with all the fixed capacitors and resistors required to make a complete radio receiver. As the Loewe set had only one tube socket, it was able to substantially undercut the competition since, in Germany, state tax was levied by the number of sockets. However, reliability was compromised, and production costs for the tube were much greater. In a sense, these were akin to integrated circuits. In the US, Cleartron briefly produced the "Multivalve" triple triode for use in the Emerson Baby Grand receiver. This Emerson set also had a single tube socket, but because it used a four-pin base, the additional element connections were made on a "mezzanine" platform at the top of the tube base.

By 1940 multisection tubes had become commonplace. There were constraints, however, due to patents and other licensing considerations (see British Valve Association). Constraints due to the number of external pins (leads) often forced the functions to share some of those external connections such as their cathode connections (in addition to the heater connection). The RCA Type 55 was a Double Diode Triode used as a detector, automatic gain control Detector and audio preamplifier in early AC powered radios. These sets often included the 53 Dual Triode Audio Output.

Other early type of multi-section tubes, the 6SN7 and 6SL7 Octal based "Dual Triodes" performed the functions of two Triode Tubes, while taking up half as much space and costing less.

The Miniature Tube Bases

Early tubes used a metal or glass envelope fixed to an insulating Bakelite or a ceramic base. In 1938 a technique was developed to use an all-glass construction with the pins fused in the glass base of the envelope. This was used in the design of a much smaller tube outline, known as the miniature tube base, having 7 or 9 pins.

The introduction of these miniature tube bases, more than previously available, allowed other multi-section tubes to be introduced. The 12AU7, 12AT7 and 12AX7 dual triodes in a nine pin Noval miniature envelope, became widely used audio signal amplifiers. The 12AX7

Vacuum Tube Theory, a Basics Tutorial – Page 9was the "high mu" - highest voltage gain device of the three. Another popular combination was a Triode-Pentode such as the 6BL8, 6U8 and 6GH8. These tubes became popular in domestic radio and television receivers.

The desire to include even more functions in one envelope resulted in the General Electric Compactron which had 12 pins. A typical example, the 6AG11, contained two triodes and two diodes. Compactrons were used in the last tube television receivers, built mostly for the American market and was the “last gasp” of the tube technology.

Subminiature tubes

Many very small tubes were constructed for specialised functions, for example, tubes roughly the size of half a cigarette were used in hearing-aid amplifiers. These tubes did not usually have pins plugging into a socket but were soldered in place.

The "Acorn" tube (named due to its shape) was also very small, and was developed during the 1940’s for very high frequency radio equipment being used during World War II.

There is also the metal-cased RCA Nuvistor from 1959, about the size of a thimble. The Nuvistor was developed to compete with the early transistors and operated at higher frequencies than those early transistors could. The small size supported especially high-frequency operation - nuvistors were used in UHF television tuners and some Amateur Radio receivers.

A look at this Wikipedia web site will show how tube sockets have evolved over the years – https://en.wikipedia.org/wiki/Tube_socket

Tube numbering http://www.r-type.org/articles/art-170.htmhttp://www.vintage-radio.com/repair-restore-information/valve_valve-numbering.html

List of vacuum tubes From Wikipedia, the free encyclopediahttps://en.wikipedia.org/wiki/List_of_vacuum_tubes

Bipolar Transistor Basics

In the Diode tutorials we saw that simple diodes are made up from two pieces of semiconductor material, either

silicon or germanium to form a simple PN-junction and we also learnt about their properties and characteristics. If we

now join together two individual signal diodes back-to-back, this will give us two PN-junctions connected together in

series that share a common P or N terminal. The fusion of these two diodes produces a three layer, two junction,

three terminal device forming the basis of a Bipolar Transistor, or BJT for short.

Transistors are three terminal active devices made from different semiconductor materials that can act as either an

insulator or a conductor by the application of a small signal voltage. The transistor's ability to change between these

two states enables it to have two basic functions: "switching" (digital electronics) or "amplification" (analogue

electronics). Then bipolar transistors have the ability to operate within three different regions:

• 1. Active Region - the transistor operates as an amplifier and Ic = β.Ib •

• 2. Saturation - the transistor is "fully-ON" operating as a switch and Ic = I(saturation) •

• 3. Cut-off - the transistor is "fully-OFF" operating as a switch and Ic = 0

Typical Bipolar Transistor

The word Transistor is an acronym, and is a combination of the words Transfer Varistor used to describe their

mode of operation way back in their early days of development. There are two basic types of bipolar transistor

construction, NPN and PNP, which basically describes the physical arrangement of the P-type and N-type

semiconductor materials from which they are made.

The Bipolar Transistor basic construction consists of two PN-junctions producing three connecting terminals with

each terminal being given a name to identify it from the other two. These three terminals are known and labelled as

the Emitter ( E ), the Base ( B ) and the Collector ( C ) respectively.

Bipolar Transistors are current regulating devices that control the amount of current flowing through them in

proportion to the amount of biasing voltage applied to their base terminal acting like a current-controlled switch. The

principle of operation of the two transistor types NPN and PNP, is exactly the same the only difference being in their

biasing and the polarity of the power supply for each type.

Bipolar Transistor Construction

The construction and circuit symbols for both the NPN and PNP bipolar transistor are given above with the arrow in

the circuit symbol always showing the direction of "conventional current flow" between the base terminal and its

emitter terminal. The direction of the arrow always points from the positive P-type region to the negative N-type

region for both transistor types, exactly the same as for the standard diode symbol.

Bipolar Transistor Configurations

As the Bipolar Transistor is a three terminal device, there are basically three possible ways to connect it within an

electronic circuit with one terminal being common to both the input and output. Each method of connection

responding differently to its input signal within a circuit as the static characteristics of the transistor vary with each

circuit arrangement.

• 1. Common Base Configuration - has Voltage Gain but no Current Gain. •

• 2. Common Emitter Configuration - has both Current and Voltage Gain. •

• 3. Common Collector Configuration - has Current Gain but no Voltage Gain.

The Common Base (CB) Configuration

As its name suggests, in the Common Base or grounded base configuration, the BASE connection is common to

both the input signal AND the output signal with the input signal being applied between the base and the emitter

terminals. The corresponding output signal is taken from between the base and the collector terminals as shown with

the base terminal grounded or connected to a fixed reference voltage point. The input current flowing into the emitter

is quite large as its the sum of both the base current and collector current respectively therefore, the collector current

output is less than the emitter current input resulting in a current gain for this type of circuit of "1" (unity) or less, in

other words the common base configuration "attenuates" the input signal.

The Common Base Transistor Circuit

This type of amplifier configuration is a non-inverting voltage amplifier circuit, in that the signal voltages Vin and Vout

are in-phase. This type of transistor arrangement is not very common due to its unusually high voltage gain

characteristics. Its output characteristics represent that of a forward biased diode while the input characteristics

represent that of an illuminated photo-diode. Also this type of bipolar transistor configuration has a high ratio of output

to input resistance or more importantly "load" resistance (RL) to "input" resistance (Rin) giving it a value of

"Resistance Gain". Then the voltage gain (Av for a common base configuration is therefore given as:

Common Base Voltage Gain

The common base circuit is generally only used in single stage amplifier circuits such as microphone pre-amplifier or

radio frequency (Rf) amplifiers due to its very good high frequency response.

The Common Emitter (CE) Configuration

In the Common Emitter or grounded emitter configuration, the input signal is applied between the base, while the

output is taken from between the collector and the emitter as shown. This type of configuration is the most commonly

used circuit for transistor based amplifiers and which represents the "normal" method of bipolar transistor connection.

The common emitter amplifier configuration produces the highest current and power gain of all the three bipolar

transistor configurations. This is mainly because the input impedance is LOW as it is connected to a forward-biased

PN-junction, while the output impedance is HIGH as it is taken from a reverse-biased PN-junction.

The Common Emitter Amplifier Circuit

In this type of configuration, the current flowing out of the transistor must be equal to the currents flowing into the

transistor as the emitter current is given as Ie = Ic + Ib. Also, as the load resistance (RL) is connected in series with

the collector, the current gain of the common emitter transistor configuration is quite large as it is the ratio of Ic/Ib and

is given the Greek symbol of Beta, (β). As the emitter current for a common emitter configuration is defined as

Ie = Ic + Ib, the ratio of Ic/Ie is called Alpha, given the Greek symbol of α. Note: that the value of Alpha will always

be less than unity.

Since the electrical relationship between these three currents, Ib, Ic and Ie is determined by the physical construction

of the transistor itself, any small change in the base current (Ib), will result in a much larger change in the collector

current (Ic). Then, small changes in current flowing in the base will thus control the current in the emitter-collector

circuit. Typically, Beta has a value between 20 and 200 for most general purpose transistors.

By combining the expressions for both Alpha, α and Beta, β the mathematical relationship between these

parameters and therefore the current gain of the transistor can be given as:

Where: "Ic" is the current flowing into the collector terminal, "Ib" is the current flowing into the base terminal and "Ie"

is the current flowing out of the emitter terminal.

Then to summarise, this type of bipolar transistor configuration has a greater input impedance, current and power

gain than that of the common base configuration but its voltage gain is much lower. The common emitter

configuration is an inverting amplifier circuit resulting in the output signal being 180o out-of-phase with the input

voltage signal.

The Common Collector (CC) Configuration

In the Common Collector or grounded collector configuration, the collector is now common through the supply. The

input signal is connected directly to the base, while the output is taken from the emitter load as shown. This type of

configuration is commonly known as a Voltage Follower or Emitter Follower circuit. The emitter follower

configuration is very useful for impedance matching applications because of the very high input impedance, in the

region of hundreds of thousands of Ohms while having a relatively low output impedance.

The Common Collector Transistor Circuit

The common emitter configuration has a current gain approximately equal to the β value of the transistor itself. In the

common collector configuration the load resistance is situated in series with the emitter so its current is equal to that

of the emitter current. As the emitter current is the combination of the collector AND the base current combined, the

load resistance in this type of transistor configuration also has both the collector current and the input current of the

base flowing through it. Then the current gain of the circuit is given as:

The Common Collector Current Gain

This type of bipolar transistor configuration is a non-inverting circuit in that the signal voltages of Vin and Vout are in-

phase. It has a voltage gain that is always less than "1" (unity). The load resistance of the common collector transistor

receives both the base and collector currents giving a large current gain (as with the common emitter configuration)

therefore, providing good current amplification with very little voltage gain.

Bipolar Transistor Summary

Then to summarise, the behaviour of the bipolar transistor in each one of the above circuit configurations is very

different and produces different circuit characteristics with regards to input impedance, output impedance and gain

whether this is voltage gain, current gain or power gain and this is summarised in the table below.

Bipolar Transistor Characteristics

The static characteristics for a Bipolar Transistor can be divided into the following three main groups.

Input Characteristics:- Common Base - ΔVEB / ΔIE Common Emitter - ΔVBE / ΔIB Output Characteristics:- Common Base - ΔVC / ΔIC Common Emitter - ΔVC / ΔIC Transfer Characteristics:- Common Base - ΔIC / ΔIE Common Emitter - ΔIC / ΔIB

with the characteristics of the different transistor configurations given in the following table:

Characteristic Common Base

Common Emitter

Common Collector

Input Impedance Low Medium High

Output Impedance Very High High Low

Phase Angle 0o 180o 0o

Voltage Gain High Medium Low

Current Gain Low Medium High

Power Gain Low Very High Medium

In the next tutorial about Bipolar Transistors, we will look at the NPN Transistor in more detail when used in the

common emitter configuration as an amplifier as this is the most widely used configuration due to its flexibility and

high gain. We will also plot the output characteristics curves commonly associated with amplifier circuits as a function

of the collector current to the base current.

The NPN Transistor

In the previous tutorial we saw that the standard Bipolar Transistor or BJT, comes in two basic forms. An NPN

(Negative-Positive-Negative) type and a PNP (Positive-Negative-Positive) type, with the most commonly used

transistor type being the NPN Transistor. We also learnt that the transistor junctions can be biased in one of three

different ways - Common Base, Common Emitter and Common Collector. In this tutorial we will look more closely

at the "Common Emitter" configuration using NPN Transistors with an example of the construction of a NPN

transistor along with the transistors current flow characteristics is given below.

An NPN Transistor Configuration

Note: Conventional current flow.

We know that the transistor is a "current" operated device (Beta model) and that a large current ( Ic ) flows freely

through the device between the collector and the emitter terminals when the transistor is switched "fully-ON".

However, this only happens when a small biasing current ( Ib ) is flowing into the base terminal of the transistor at the

same time thus allowing the Base to act as a sort of current control input. The transistor current in an NPN transistor

is the ratio of these two currents ( Ic/Ib ), called the DC Current Gain of the device and is given the symbol of hfe or

nowadays Beta, ( β ). The value of β can be large up to 200 for standard transistors, and it is this large ratio between

Ic and Ib that makes the NPN transistor a useful amplifying device when used in its active region as Ib provides the

input and Ic provides the output. Note that Beta has no units as it is a ratio.

Also, the current gain of the transistor from the Collector terminal to the Emitter terminal, Ic/Ie, is called Alpha, ( α ),

and is a function of the transistor itself (electrons diffusing across the junction). As the emitter current Ie is the

product of a very small base current plus a very large collector current, the value of alpha α, is very close to unity,

and for a typical low-power signal transistor this value ranges from about 0.950 to 0.999

α and β Relationship in a NPN Transistor

By combining the two parameters α and β we can produce two mathematical expressions that gives the relationship

between the different currents flowing in the transistor.

The values of Beta vary from about 20 for high current power transistors to well over 1000 for high frequency low

power type bipolar transistors. The value of Beta for most standard NPN transistors can be found in the

manufactures datasheets but generally range between 50 - 200.

The equation above for Beta can also be re-arranged to make Ic as the subject, and with a zero base current ( Ib = 0 ) the resultant collector current Ic will also be zero, ( β x 0 ). Also when the base current is high the corresponding

collector current will also be high resulting in the base current controlling the collector current. One of the most

important properties of the Bipolar Junction Transistor is that a small base current can control a much larger

collector current. Consider the following example.

Example No1

An NPN Transistor has a DC current gain, (Beta) value of 200. Calculate the base current Ib required to switch a

resistive load of 4mA.

Therefore, β = 200, Ic = 4mA and Ib = 20µA.

One other point to remember about NPN Transistors. The collector voltage, ( Vc ) must be greater and positive with

respect to the emitter voltage, ( Ve ) to allow current to flow through the transistor between the collector-emitter

junctions. Also, there is a voltage drop between the Base and the Emitter terminal of about 0.7v (one diode volt drop)

for silicon devices as the input characteristics of an NPN Transistor are of a forward biased diode. Then the base

voltage, ( Vbe ) of a NPN transistor must be greater than this 0.7V otherwise the transistor will not conduct with the

base current given as.

Where: Ib is the base current, Vb is the base bias voltage, Vbe is the base-emitter volt drop (0.7v) and Rb is the

base input resistor. Increasing Ib, Vbe slowly increases to 0.7V but Ic rises exponentially.

Example No2

An NPN Transistor has a DC base bias voltage, Vb of 10v and an input base resistor, Rb of 100kΩ. What will be the

value of the base current into the transistor.

Therefore, Ib = 93µA.

The Common Emitter Configuration.

As well as being used as a semiconductor switch to turn load currents "ON" or "OFF" by controlling the Base signal to

the transistor in ether its saturation or cut-off regions, NPN Transistors can also be used in its active region to

produce a circuit which will amplify any small AC signal applied to its Base terminal with the Emitter grounded. If a

suitable DC "biasing" voltage is firstly applied to the transistors Base terminal thus allowing it to always operate within

its linear active region, an inverting amplifier circuit called a single stage common emitter amplifier is produced.

One such Common Emitter Amplifier configuration of an NPN transistor is called a Class A Amplifier. A "Class A

Amplifier" operation is one where the transistors Base terminal is biased in such a way as to forward bias the Base-

emitter junction. The result is that the transistor is always operating halfway between its cut-off and saturation regions,

thereby allowing the transistor amplifier to accurately reproduce the positive and negative halves of any AC input

signal superimposed upon this DC biasing voltage. Without this "Bias Voltage" only one half of the input waveform

would be amplified. This common emitter amplifier configuration using an NPN transistor has many applications but is

commonly used in audio circuits such as pre-amplifier and power amplifier stages.

With reference to the common emitter configuration shown below, a family of curves known as the Output

Characteristics Curves, relates the output collector current, (Ic) to the collector voltage, (Vce) when different values

of Base current, (Ib) are applied to the transistor for transistors with the same β value. A DC "Load Line" can also be

drawn onto the output characteristics curves to show all the possible operating points when different values of base

current are applied. It is necessary to set the initial value of Vce correctly to allow the output voltage to vary both up

and down when amplifying AC input signals and this is called setting the operating point or Quiescent Point, Q-

point for short and this is shown below.

Single Stage Common Emitter Amplifier Circuit

Output Characteristics Curves for a Typical Bipolar Transistor

The most important factor to notice is the effect of Vce upon the collector current Ic when Vce is greater than about

1.0 volts. We can see that Ic is largely unaffected by changes in Vce above this value and instead it is almost entirely

controlled by the base current, Ib. When this happens we can say then that the output circuit represents that of a

"Constant Current Source". It can also be seen from the common emitter circuit above that the emitter current Ie is

the sum of the collector current, Ic and the base current, Ib, added together so we can also say that " Ie = Ic + Ib "

for the common emitter configuration.

By using the output characteristics curves in our example above and also Ohm´s Law, the current flowing through the

load resistor, (RL), is equal to the collector current, Ic entering the transistor which inturn corresponds to the supply

voltage, (Vcc) minus the voltage drop between the collector and the emitter terminals, (Vce) and is given as:

Also, a straight line representing the Load Line of the transistor can be drawn directly onto the graph of curves above

from the point of "Saturation" ( A ) when Vce = 0 to the point of "Cut-off" ( B ) when Ic = 0 thus giving us the

"Operating" or Q-point of the transistor. These two points are joined together by a straight line and any position along

this straight line represents the "Active Region" of the transistor. The actual position of the load line on the

characteristics curves can be calculated as follows:

Then, the collector or output characteristics curves for Common Emitter NPN Transistors can be used to predict

the Collector current, Ic, when given Vce and the Base current, Ib. A Load Line can also be constructed onto the

curves to determine a suitable Operating or Q-point which can be set by adjustment of the base current. The slope of

this load line is equal to the reciprocal of the load resistance which is given as: -1/RL

In the next tutorial about Bipolar Transistors, we will look at the opposite or compliment form of the NPN Transistor called the PNP Transistor and show that the PNP Transistor has very similar characteristics to their

NPN transistor except that the polarities (or biasing) of the current and voltage directions are reversed.

The PNP Transistor

The PNP Transistor is the exact opposite to the NPN Transistor device we looked at in the previous tutorial.

Basically, in this type of transistor construction the two diodes are reversed with respect to the NPN type, with the

arrow, which also defines the Emitter terminal this time pointing inwards in the transistor symbol. Also, all the

polarities are reversed which means that PNP Transistors "sink" current as opposed to the NPN transistor which

"sources" current. Then, PNP Transistors use a small output base current and a negative base voltage to control a

much larger emitter-collector current. The construction of a PNP transistor consists of two P-type semiconductor

materials either side of the N-type material as shown below.

A PNP Transistor Configuration

Note: Conventional current flow.

The PNP Transistor has very similar characteristics to their NPN bipolar cousins, except that the polarities (or

biasing) of the current and voltage directions are reversed for any one of the possible three configurations looked at

in the first tutorial, Common Base, Common Emitter and Common Collector. Generally, PNP Transistors require a

negative (-ve) voltage at their Collector terminal with the flow of current through the emitter-collector terminals being

Holes as opposed to Electrons for the NPN types. Because the movement of holes across the depletion layer tends

to be slower than for electrons, PNP transistors are generally more slower than their equivalent NPN counterparts

when operating.

To cause the Base current to flow in a PNP transistor the Base needs to be more negative than the Emitter (current

must leave the base) by approx 0.7 volts for a silicon device or 0.3 volts for a germanium device with the formulas

used to calculate the Base resistor, Base current or Collector current are the same as those used for an equivalent

NPN transistor and is given as.

Generally, the PNP transistor can replace NPN transistors in electronic circuits, the only difference is the polarities of

the voltages, and the directions of the current flow. PNP Transistors can also be used as switching devices and an

example of a PNP transistor switch is shown below.

A PNP Transistor Circuit

The Output Characteristics Curves for a PNP transistor look very similar to those for an equivalent NPN transistor

except that they are rotated by 180o to take account of the reverse polarity voltages and currents, (the currents

flowing out of the Base and Collector in a PNP transistor are negative).

Transistor Matching

You may think what is the point of having a PNP Transistor, when there are plenty of NPN Transistors available?.

Well, having two different types of transistors PNP & NPN, can be an advantage when designing amplifier circuits

such as Class B Amplifiers that use "Complementary" or "Matched Pair" transistors or for reversible H-Bridge

motor control circuits. A pair of corresponding NPN and PNP transistors with near identical characteristics to each

other are called Complementary Transistors for example, a TIP3055 (NPN), TIP2955 (PNP) are good examples of

complementary or matched pair silicon power transistors. They have a DC current gain, Beta, (Ic / Ib) matched to

within 10% and high Collector current of about 15A making them suitable for general motor control or robotic

applications.

Identifying the PNP Transistor

We saw in the first tutorial of this Transistors section, that transistors are basically made up of two Diodes

connected together back-to-back. We can use this analogy to determine whether a transistor is of the type PNP or

NPN by testing its Resistance between the three different leads, Emitter, Base and Collector. By testing each pair

of transistor leads in both directions will result in six tests in total with the expected resistance values in Ohm's given

below.

• 1. Emitter-Base Terminals - The Emitter to Base should act like a normal diode and conduct one way only. •

• 2. Collector-Base Terminals - The Collector-Base junction should act like a normal diode and conduct one way

only.

•

• 3. Emitter-Collector Terminals - The Emitter-Collector should not conduct in either direction.

Transistor Resistance Values for the PNP transistor and NPN transistor types

Between Transistor Terminals PNP NPN Collector Emitter RHIGH RHIGH Collector Base RLOW RHIGH Emitter Collector RHIGH RHIGH Emitter Base RLOW RHIGH Base Collector RHIGH RLOW Base Emitter RHIGH RLOW

The Transistor as a Switch

When used as an AC signal amplifier, the transistors Base biasing voltage is applied so that it operates within its

"Active" region and the linear part of the output characteristics curves are used. However, both the NPN & PNP type

bipolar transistors can be made to operate as an "ON/OFF" type solid state switch for controlling high power devices

such as motors, solenoids or lamps. If the circuit uses the Transistor as a Switch, then the biasing is arranged to

operate in the output characteristics curves seen previously in the areas known as the "Saturation" and "Cut-off"

regions as shown below.

Transistor Curves

The pink shaded area at the bottom represents the "Cut-off" region. Here the operating conditions of the transistor

are zero input base current (Ib), zero output collector current (Ic) and maximum collector voltage (Vce) which results

in a large depletion layer and no current flows through the device. The transistor is switched "Fully-OFF". The lighter

blue area to the left represents the "Saturation" region. Here the transistor will be biased so that the maximum

amount of base current is applied, resulting in maximum collector current flow and minimum collector emitter voltage

which results in the depletion layer being as small as possible and maximum current flows through the device. The

transistor is switched "Fully-ON". Then we can summarize this as:

• 1. Cut-off Region - Both junctions are Reverse-biased, Base current is zero or very small resulting in zero Collector

current flowing, the device is switched fully "OFF". •

• 2. Saturation Region - Both junctions are Forward-biased, Base current is high enough to give a Collector-Emitter

voltage of 0v resulting in maximum Collector current flowing, the device is switched fully "ON".

An example of an NPN Transistor as a switch being used to operate a relay is given below. With inductive loads such

as relays or solenoids a flywheel diode is placed across the load to dissipate the back EMF generated by the

inductive load when the transistor switches "OFF" and so protect the transistor from damage. If the load is of a very

high current or voltage nature, such as motors, heaters etc, then the load current can be controlled via a suitable

relay as shown.

Transistor Switching Circuit

The circuit resembles that of the Common Emitter circuit we looked at in the previous tutorials. The difference this

time is that to operate the transistor as a switch the transistor needs to be turned either fully "OFF" (Cut-off) or fully

"ON" (Saturated). An ideal transistor switch would have an infinite resistance when turned "OFF" resulting in zero

current flow and zero resistance when turned "ON", resulting in maximum current flow. In practice when turned "OFF",

small leakage currents flow through the transistor and when fully "ON" the device has a low resistance value causing

a small saturation voltage (Vce) across it. In both the Cut-off and Saturation regions the power dissipated by the

transistor is at its minimum.

To make the Base current flow, the Base input terminal must be made more positive than the Emitter by increasing it

above the 0.7 volts needed for a silicon device. By varying the Base-Emitter voltage Vbe, the Base current is altered

and which in turn controls the amount of Collector current flowing through the transistor as previously discussed.

When maximum Collector current flows the transistor is said to be Saturated. The value of the Base resistor

determines how much input voltage is required and corresponding Base current to switch the transistor fully "ON".

Example No1.

For example, using the transistor values from the previous tutorials of: β = 200, Ic = 4mA and Ib = 20uA, find the

value of the Base resistor (Rb) required to switch the load "ON" when the input terminal voltage exceeds 2.5v.

Example No2.

Again using the same values, find the minimum Base current required to turn the transistor fully "ON" (Saturated) for

a load that requires 200mA of current.

Transistor switches are used for a wide variety of applications such as interfacing large current or high voltage

devices like motors, relays or lamps to low voltage digital logic IC's or gates like AND Gates or OR Gates. Here, the

output from a digital logic gate is only +5v but the device to be controlled may require a 12 or even 24 volts supply. Or

the load such as a DC Motor may need to have its speed controlled using a series of pulses (Pulse Width Modulation)

and transistor switches will allow us to do this faster and more easily than with conventional mechanical switches.

Digital Logic Transistor Switch

The base resistor, Rb is required to limit the output current of the logic gate.

Darlington Transistors

Sometimes the DC current gain of the bipolar transistor is too low to directly switch the load current or voltage, so

multiple switching transistors are used. Here, one small input transistor is used to switch "ON" or "OFF" a much larger

current handling output transistor. To maximise the signal gain the two transistors are connected in a

"Complementary Gain Compounding Configuration" or what is generally called a "Darlington Configuration" where

the amplification factor is the product of the two individual transistors.

Darlington Transistors simply contain two individual bipolar NPN or PNP type transistors connected together so that

the current gain of the first transistor is multiplied with that of the current gain of the second transistor to produce a

device which acts like a single transistor with a very high current gain. The overall current gain Beta (β) or Hfe value

of a Darlington device is the product of the two individual gains of the transistors and is given as:

So Darlington Transistors with very high β values and high Collector currents are possible compared to a single

transistor. An example of the two basic types of Darlington transistor are given below.

Darlington Transistor Configurations

The above NPN Darlington transistor configuration shows the Collectors of the two transistors connected together

with the Emitter of the first transistor connected to the Base of the second transistor therefore, the Emitter current of

the first transistor becomes the Base current of the second transistor. The first or "input" transistor receives an input

signal, amplifies it and uses it to drive the second or "output" transistors which amplifies it again resulting in a very

high current gain. As well as its high increased current and voltage switching capabilities, another advantage of a

Darlington transistor is in its high switching speeds making them ideal for use in Inverter circuits and DC motor or

stepper motor control applications.

One difference to consider when using Darlington transistors over the conventional single bipolar transistor type is

that the Base-Emitter input voltage Vbe needs to be higher at approx 1.4v for silicon devices, due to the series

connection of the two PN junctions.

Then to summarise when using a Transistor as a Switch.

• Transistor switches can be used to switch and control lamps, relays or even motors.

• When using bipolar transistors as switches they must be fully "OFF" or fully "ON".

• Transistors that are fully "ON" are said to be in their Saturation region.

• Transistors that are fully "OFF" are said to be in their Cut-off region.

• In a transistor switch a small Base current controls a much larger Collector current.

• When using transistors to switch inductive relay loads a "Flywheel Diode" is required.

• When large currents or voltages need to be controlled, Darlington Transistors are used.

The Field Effect Transistor

In the Bipolar Junction Transistor tutorials, we saw that the output Collector current is determined by the

amount of current flowing into the Base terminal of the device and thereby making the Bipolar Transistor a CURRENT

operated device. The Field Effect Transistor, or simply FET however, use the voltage that is applied to their input

terminal to control the output current, since their operation relies on the electric field (hence the name field effect)

generated by the input voltage. This then makes the Field Effect Transistor a VOLTAGE operated device.

The Field Effect Transistor is a unipolar device that has very similar properties to those of the Bipolar Transistor ie,

high efficiency, instant operation, robust and cheap, and they can be used in most circuit applications that use the

equivalent Bipolar Junction Transistors, (BJT). They can be made much smaller than an equivalent BJT transistor

and along with their low power consumption and dissipation make them ideal for use in integrated circuits such as the

CMOS range of chips.

We remember from the previous tutorials that there are two basic types of Bipolar Transistor construction, NPN and

PNP, which basically describes the physical arrangement of the P-type and N-type semiconductor materials from

which they are made. There are also two basic types of Field Effect Transistor, N-channel and P-channel. As their

name implies, Bipolar Transistors are "Bipolar" devices because they operate with both types of charge carriers,

Holes and Electrons. The Field Effect Transistor on the other hand is a "Unipolar" device that depends only on the

conduction of Electrons (N-channel) or Holes (P-channel).

The Field Effect Transistor has one major advantage over its standard bipolar transistor cousins, in that their input

impedance is very high, (Thousands of Ohms) making them very sensitive to input signals, but this high sensitivity

also means that they can be easily damaged by static electricity. There are two main types of field effect transistor,

the Junction Field Effect Transistor or JFET and the Insulated-gate Field Effect Transistor or IGFET), which is

more commonly known as the standard Metal Oxide Semiconductor Field Effect Transistor or MOSFET for short.

The Junction Field Effect Transistor

We saw previously that a bipolar junction transistor is constructed using two PN junctions in the main current path

between the Emitter and the Collector terminals. The Field Effect Transistor has no junctions but instead has a

narrow "Channel" of N-type or P-type silicon with electrical connections at either end commonly called the DRAIN

and the SOURCE respectively. Both P-channel and N-channel FET's are available. Within this channel there is a

third connection which is called the GATE and this can also be a P or N-type material forming a PN junction and

these connections are compared below.

Bipolar Transistor Field Effect Transistor Emitter - (E) Source - (S) Base - (B) Gate - (G)

Collector - (C) Drain - (D)

The semiconductor "Channel" of the Junction Field Effect Transistor is a resistive path through which a voltage Vds

causes a current Id to flow. A voltage gradient is thus formed down the length of the channel with this voltage

becoming less positive as we go from the drain terminal to the source terminal. The PN junction therefore has a high

reverse bias at the drain terminal and a lower reverse bias at the source terminal. This bias causes a "depletion

layer" to be formed within the channel and whose width increases with the bias. FET's control the current flow

through them between the drain and source terminals by controlling the voltage applied to the gate terminal. In an

N-channel JFET this gate voltage is negative while for a P-channel JFET the gate voltage is positive.

Bias arrangement for an N-channel JFET and corresponding circuit symbols.

The cross sectional diagram above shows an N-type semiconductor channel with a P-type region called the gate

diffused into the N-type channel forming a reverse biased PN junction and its this junction which forms the depletion

layer around the gate area. This depletion layer restricts the current flow through the channel by reducing its

effective width and thus increasing the overall resistance of the channel.

When the gate voltage Vg is equal to 0V and a small external voltage (Vds) is applied between the drain and the

source maximum current (Id) will flow through the channel slightly restricted by the small depletion layer. If a negative

voltage (Vgs) is now applied to the gate the size of the depletion layer begins to increase reducing the overall

effective area of the channel and thus reducing the current flowing through it, a sort of "squeezing" effect. As the gate

voltage (Vgs) is made more negative, the width of the channel decreases until no more current flows between the

drain and the source and the FET is said to be "pinched-off". In this pinch-off region the gate voltage, Vgs controls

the channel current and Vds has little or no effect. The result is that the FET acts more like a voltage controlled

resistor which has zero resistance when Vgs = 0 and maximum "ON" resistance (Rds) when the gate voltage is very

negative.

Output characteristic voltage-current curves of a typical junction FET.

The voltage Vgs applied to the gate controls the current flowing between the drain and the source terminals. Vgs

refers to the voltage applied between the gate and the source while Vds refers to the voltage applied between the

drain and the source. Because a Field Effect Transistor is a VOLTAGE controlled device, "NO current flows into

the gate!" then the source current (Is) flowing out of the device equals the drain current flowing into it and therefore

(Id = Is).

The characteristics curves example shown above, shows the four different regions of operation for a JFET and these

are given as:

• Ohmic Region - The depletion layer of the channel is very small and the JFET acts like a variable resistor. •

• Cut-off Region - The gate voltage is sufficient to cause the JFET to act as an open circuit as the channel

resistance is at maximum. •

• Saturation or Active Region - The JFET becomes a good conductor and is controlled by the gate-source

voltage, (Vgs) while the drain-source voltage, (Vds) has little or no effect.

•

• Breakdown Region - The voltage between the drain and source, (Vds) is high enough to causes the JFET's

resistive channel to break down and pass current.

The control of the drain current by a negative gate potential makes the Junction Field Effect Transistor useful as a

switch and it is essential that the gate voltage is never positive for an N-channel JFET as the channel current will flow

to the gate and not the drain resulting in damage to the JFET. The principals of operation for a P-channel JFET are

the same as for the N-channel JFET, except that the polarity of the voltages need to be reversed.

The MOSFET

As well as the Junction Field Effect Transistor, there is another type of Field Effect Transistor available whose Gate

input is electrically insulated from the main current carrying channel and is therefore called an Insulated Gate Field

Effect Transistor. The most common type of insulated gate FET or IGFET as it is sometimes called, is the Metal

Oxide Semiconductor Field Effect Transistor or MOSFET for short.

The MOSFET type of field effect transistor has a "Metal Oxide" gate (usually silicon dioxide commonly known as

glass), which is electrically insulated from the main semiconductor N-channel or P-channel. This isolation of the

controlling gate makes the input resistance of the MOSFET extremely high in the Mega-ohms region and almost

infinite. As the gate terminal is isolated from the main current carrying channel ""NO current flows into the gate"" and

like the JFET, the MOSFET also acts like a voltage controlled resistor. Also like the JFET, this very high input

resistance can easily accumulate large static charges resulting in the MOSFET becoming easily damaged unless

carefully handled or protected.

Basic MOSFET Structure and Symbol

We also saw previously that the gate of a JFET must be biased in such a way as to forward-bias the PN junction but

in a MOSFET device no such limitations applies so it is possible to bias the gate in either polarity. This makes

MOSFET's specially valuable as electronic switches or to make logic gates because with no bias they are normally

non-conducting and the high gate resistance means that very little control current is needed. Both the P-channel and

the N-channel MOSFET is available in two basic forms, the Enhancement type and the Depletion type.

Depletion-mode MOSFET

The Depletion-mode MOSFET, which is less common than the enhancement types is normally switched "ON"

without a gate bias voltage but requires a gate to source voltage (Vgs) to switch the device "OFF". Similar to the

JFET types. For N-channel MOSFET's a "Positive" gate voltage widens the channel, increasing the flow of the drain

current and decreasing the drain current as the gate voltage goes more negative. The opposite is also true for the P-

channel types. The depletion mode MOSFET is equivalent to a "Normally Closed" switch.

Depletion-mode N-Channel MOSFET and circuit Symbols

Depletion-mode MOSFET's are constructed similar to their JFET transistor counterparts where the drain-source

channel is inherently conductive with electrons and holes already present within the N-type or P-type channel. This

doping of the channel produces a conducting path of low resistance between the drain and source with zero gate

bias.

Enhancement-mode MOSFET

The more common Enhancement-mode MOSFET is the reverse of the depletion-mode type. Here the conducting

channel is lightly doped or even undoped making it non-conductive. This results in the device being normally "OFF"

when the gate bias voltage is equal to zero.

A drain current will only flow when a gate voltage (Vgs) is applied to the gate terminal. This positive voltage creates

an electrical field within the channel attracting electrons towards the oxide layer and thereby reducing the overall

resistance of the channel allowing current to flow. Increasing this positive gate voltage will cause an increase in the

drain current, Id through the channel. Then, the Enhancement-mode device is equivalent to a "Normally Open" switch.

Enhancement-mode N-Channel MOSFET and circuit Symbols

Enhancement-mode MOSFET's make excellent electronics switches due to their low "ON" resistance and extremely

high "OFF" resistance and extremely high gate resistance. Enhancement-mode MOSFET's are used in integrated

circuits to produce CMOS type Logic Gates and power switching circuits as they can be driven by digital logic

levels.

MOSFET Summary

The MOSFET has an extremely high input gate resistance and as such a easily damaged by static electricity if not

carefully protected. MOSFET's are ideal for use as electronic switches or common-source amplifiers as their power

consumption is very small. Typical applications for MOSFET's are in Microprocessors, Memories, Calculators and

Logic Gates etc. Also, notice that the broken lines within the symbol indicates a normally "OFF" Enhancement type

showing that "NO" current can flow through the channel when zero gate voltage is applied and a continuous line

within the symbol indicates a normally "ON" Depletion type showing that current "CAN" flow through the channel with

zero gate voltage. For P-Channel types the symbols are exactly the same for both types except that the arrow points

outwards.

This can be summarised in the following switching table.

MOSFET type Vgs = +ve Vgs = 0 Vgs = -ve N-Channel Depletion ON ON OFF

N-Channel Enhancement ON OFF OFF P-Channel Depletion OFF ON ON

P-Channel Enhancement OFF OFF ON

The MOSFET as a Switch

We saw previously, that the N-channel, Enhancement-mode MOSFET operates using a positive input voltage and

has an extremely high input resistance (almost infinite) making it possible to interface with nearly any logic gate or

driver capable of producing a positive output. Also, due to this very high input (Gate) resistance we can parallel

together many different MOSFET's until we achieve the current handling limit required. While connecting together

various MOSFET's may enable us to switch high current or high voltage loads, doing so becomes expensive and

impractical in both components and circuit board space. To overcome this problem Power Field Effect Transistors

or Power FET's where developed.

We now know that there are two main differences between FET's, Depletion-mode for JFET's and Enhancement-

mode for MOSFET's and on this page we will look at using the Enhancement-mode MOSFET as a Switch.

By applying a suitable drive voltage to the Gate of an FET the resistance of the Drain-Source channel can be varied

from an "OFF-resistance" of many hundreds of kΩ's, effectively an open circuit, to an "ON-resistance" of less than 1Ω,

effectively a short circuit. We can also drive the MOSFET to turn "ON" fast or slow, or to pass high currents or low

currents. This ability to turn the power MOSFET "ON" and "OFF" allows the device to be used as a very efficient

switch with switching speeds much faster than standard bipolar junction transistors.

An example of using the MOSFET as a switch

In this circuit arrangement an Enhancement-mode N-

channel MOSFET is being used to switch a simple lamp

"ON" and "OFF" (could also be an LED). The gate input

voltage VGS is taken to an appropriate positive voltage

level to turn the device and the lamp either fully "ON",

(VGS = +ve) or a zero voltage level to turn the device fully

"OFF", (VGS = 0).

If the resistive load of the lamp was to be replaced by an

inductive load such as a coil or solenoid, a "Flywheel"

diode would be required in parallel with the load to protect

the MOSFET from any back-emf.

Above shows a very simple circuit for switching a resistive load such as a lamp or LED. But when using power

MOSFET's to switch either inductive or capacitive loads some form of protection is required to prevent the MOSFET

device from becoming damaged. Driving an inductive load has the opposite effect from driving a capacitive load. For

example, a capacitor without an electrical charge is a short circuit, resulting in a high "inrush" of current and when we

remove the voltage from an inductive load we have a large reverse voltage build up as the magnetic field collapses,

resulting in an induced back-emf in the windings of the inductor.

For the power MOSFET to operate as an analogue switching device, it needs to be switched between its "Cut-off

Region" where VGS = 0 and its "Saturation Region" where VGS(on) = +ve. The power dissipated in the MOSFET (PD)

depends upon the current flowing through the channel ID at saturation and also the "ON-resistance" of the channel

given as RDS(on). For example.

Example No1

Lets assume that the lamp is rated at 6v, 24W and is fully "ON" and the standard MOSFET has a channel "ON-

resistance" ( RDS(on) ) value of 0.1ohms. Calculate the power dissipated in the MOSFET switch.

The current flowing through the lamp is calculated as:

Then the power dissipated in the MOSFET will be given as:

You may think, well so what!, but when using the MOSFET as a switch to control DC motors or high inrush current

devices the "ON" channel resistance ( RDS(on) ) is very important. For example, MOSFET's that control DC motors,

are subjected to a high in-rush current as the motor first begins to rotate. Then a high RDS(on) channel resistance

value would simply result in large amounts of power being dissipated within the MOSFET itself resulting in an

excessive temperature rise, and which in turn could result in the MOSFET becoming very hot and damaged due to a

thermal overload. But a low RDS(on) value on the other hand is also desirable to help reduce the effective saturation

voltage ( VDS(sat) = ID x RDS(on) ) across the MOSFET. When using MOSFET´s or any type of Field Effect Transistor

for that matter as a switching device, it is always advisable to select ones that have a very low RDS(on) value or at

least mount them onto a suitable heatsink to help reduce any thermal runaway and damage.

Power MOSFET Motor Control

Because of the extremely high input or Gate resistance that the MOSFET has, its very fast switching speeds and the

ease at which they can be driven makes them ideal to interface with op-amps or standard logic gates. However, care

must be taken to ensure that the gate-source input voltage is correctly chosen because when using the MOSFET as

a switch the device must obtain a low RDS(on) channel resistance in proportion to this input gate voltage. For

example, do not apply a 12v signal if a 5v signal voltage is required. Power MOSFET´s can be used to control the

movement of DC motors or brushless stepper motors directly from computer logic or Pulse-width Modulation (PWM)

type controllers. As a DC motor offers high starting torque and which is also proportional to the armature current,

MOSFET switches along with a PWM can be used as a very good speed controller that would provide smooth and

quiet motor operation.

Simple Power MOSFET Motor Controller

As the motor load is inductive, a simple "Free-wheeling"

diode is connected across the load to dissipate any back

emf generated by the motor when the MOSFET turns it

"OFF".

The Zener diode is used to prevent excessive gate-source input voltages.

Summary of Bipolar Junction Transistors

• The Bipolar Junction Transistor (BJT) is a three layer device constructed form two semiconductor diode junctions

joined together, one forward biased and one reverse biased.

• There are two main types of bipolar junction transistors, the NPN and the PNP transistor.

• Transistors are "Current Operated Devices" where a much smaller Base current causes a larger Emitter to

Collector current, which themselves are nearly equal, to flow.

• The most common transistor connection is the Common-emitter configuration.

• Requires a Biasing voltage for AC amplifier operation.

• The Collector or output characteristics curves can be used to find either Ib, Ic or β to which a load line can be

constructed to determine a suitable operating point, Q with variations in base current determining the operating

range.

• A transistor can also be used as an electronic switch to control devices such as lamps, motors and solenoids etc.

• Inductive loads such as DC motors, relays and solenoids require a reverse biased "Flywheel" diode placed across

the load. This helps prevent any induced back emf's generated when the load is switched "OFF" from damaging the

transistor.

• The NPN transistor requires the Base to be more positive than the Emitter while the PNP type requires that the

Emitter is more positive than the Base.

Summary of Field Effect Transistors

• Field Effect Transistors, or FET's are "Voltage Operated Devices" and can be divided into two main types:

Junction-gate devices called JFET's and Insulated-gate devices called IGFET´s or more commonly known as

MOSFET's.

• Insulated-gate devices can also be sub-divided into Enhancement types and Depletion types. All forms are

available in both N-channel and P-channel versions.

• FET's have very high input resistances so very little or no current (MOSFET types) flows into the input terminal

making them ideal for use as electronic switches.

• The input impedance of the MOSFET is even higher than that of the JFET due to the insulating oxide layer and

therefore static electricity can easily damage MOSFET devices so care needs to be taken when handling them.

• FET's have very large current gain compared to junction transistors.

• They can be used as ideal switches due to their very high channel "OFF" resistance, low "ON" resistance.

The Field Effect Transistor Family-tree

Field Effect Transistors can be used to replace normal Bipolar Junction Transistors in electronic circuits and a simple

comparison between FET's and transistors stating both their advantages and their disadvantages is given below.

Field Effect Transistor (FET) Bipolar Junction Transistor (BJT) 1 Low voltage gain High voltage gain 2 High current gain Low current gain 3 Very input impedance Low input impedance 4 High output impedance Low output impedance 5 Low noise generation Medium noise generation 6 Fast switching time Medium switching time 7 Easily damaged by static Robust 8 Some require an input to turn it "OFF" Requires zero input to turn it "OFF" 9 Voltage controlled device Current controlled device

10 Exhibits the properties of a Resistor 11 More expensive than bipolar Cheap 12 Difficult to bias Easy to bias

Introduction to Integrated Circuit

TechnologyFifth Edition

Written by: Scotten W. Jones

Introduction to Integrated Circuit Technology

Copyright © 2001 - 2012 IC Knowledge LLC, all rights reserved 2

1.0. IntroductionAt IC Knowledge LLC, we have found a wide diversity in our clients and web site visitors

with respect to their understanding of Integrated Circuit (IC) technology. Some of the people weinteract with have a strong understanding of IC technology, but there is also a substantial groupthat purchases or uses the technology without a strong understanding. For the later group, wethough it would be useful to produce a basic introduction to IC technology, and that is the objec-tive of this publication.

We have written this publication assuming no more technical background than a high schooleducation, and any technical terms will be defined when they are introduced. We have attemptedto provide a good high level overview of the technology in this document, if you have questionsabout the content or would like to provide us with feedback, please e-mail us at info@icknowl-edge.com.

2.0. Basic electronic conceptsElectronic circuits regulate and control the flow of electric

current. Electric current is the flow of electrons, the tiny sub-atomic particles that surround the nucleus of atoms. Electronscarry a fixed negative electric charge and the movement of elec-trons carries charge from one location to another - the flow ofelectrons is referred to as electric current. Electric current isdriven by a difference in potential from one location to anothermeasured in volts. Electric current flows easily through materialsthat are conductors, and is blocked by materials that are insula-tors. The amount of resistance that a material presents to the flowof electric current is logically called resistance. Conductors havelow resistance to the flow of current and insulators haveextremely high resistance (essentially infinite until the voltage isso high that the material breaks down). For a given voltage, thehigher the resistance the less current that will flow and the lowerthe resistance the higher the current that will flow. Conversely,for a given resistance, the higher the voltage the more current thatwill flow and the lower the voltage the less current that will flow.

3.0. Electronic circuit elementsElectronic circuits are made up of a number of elements used

to control current flow. There are a wide variety of different circuit elements, but for the purposeof this discussion the circuit elements will be restricted to the four most commonly used in ICs,these are, resistors, capacitors, diodes and transistors. Resistors, provide a fixed amount of resis-tance to current flow. Capacitors, store electric charge until discharged somewhat similar to a bat-tery. Diodes, allow current to flow in one direction but not in the opposite direction, a one wayvalve. Transistors, provides two major modes of action, one, a switch turning current flow on andoff, or two, act as an amplifier whereby an input current produces a larger output current.

Basic Definitions� Current - the flow of

electrons carrying electric charge.

� Voltage - the force driving the flow of current.

� Resistance - a mate-rial�s resistance to the flow of electric current.

� Conductor - a mate-rial that readily sup-ports the flow of electric current.

� Insulator - a mate-rial that blocks the flow of electric cur-rent.

Introduction to Integrated Circuit Technology

Copyright © 2001 - 2012 IC Knowledge LLC, all rights reserved 3

An Integrated Circuit, or IC, is nothing more than a numberof these components connected together as a circuit all formed onthe same substrate.

4.0. What is a Semiconductor?A semiconductor is a material that may act as a conductor or

as an insulator depending on the conditions. Diodes and transis-tors are made with semiconductor material and resistors andcapacitors may be made on or in semiconductor materials as well.As the scientific community began to understand semiconductormaterials, the transistor and later the IC were invented (see �His-tory of the IC� at www.icknowledge.com for more information).Resistors and capacitors as individual components are commonlymade without the use of semiconductor materials but the ability to make them with semiconductormaterial made it possible to integrate them with diodes and transistors. Semiconductors may bemade more conductive by adding other impurity elements to the semiconductor material and theability to do this selectively, i.e., add impurities to one part of a semiconductor material and not toother parts is what enables IC fabrication to take place. Areas of semiconductor material that arehighly pure and therefore have little or no impurities act as insulators. This is the key to IC fabri-cation and will be discussed further in the sections that follow.

5.0. Integrated Circuit Manufacturing Overview.At the highest level, the manufacture of ICs may be broken up into 5 major steps - see figure

1.

Figure 1. IC manufacturing.

IC Circuit Elements� Resistors - resists

current flow.� Capacitors - stores

charge.� Diodes - allows cur-

rent to flow in only one direction.

� Transistor - switches and or amplifies current.

Introduction to Integrated Circuit Technology

Copyright © 2001 - 2012 IC Knowledge LLC, all rights reserved 4

The five major steps are:

1. Starting substrate - the starting substrate is purchased by virtually all major IC producers. Starting substrates will be discussed further in section 6.0.

2. Wafer fabrication - the process of fabricating a numbers of ICs on the surface of the wafer simultaneously. Wafer fabrication will be discussed further in section 8.0.

3. Wafer sort/test - each IC (referred to as a die) on the wafer surface is tested and the bad die are marked with an ink dot or in an electronic map. The bad die are discarded after the wafer is sawn up for packaging to save the cost of packaging bad die. Wafer test will be discussed fur-ther in section 10.0.

4. Packaging - the wafer is sawn up into individual die and the good die are assembled into pro-tective packages. Packaging will be discussed further in section 11.0

5. Mark and class/final test - in order to insure that the die were not damaged during packaging, the packaged product is tested and marked with the product type. Final test will be discussed further in section 12.0.

6.0. Silicon WafersFar and away the most common material for IC fabrication is silicon (there are other materi-

als in use, but only for small niche applications). Silicon is an abundant material in the earth'scrust and relatively easy to obtain and refine. Silicon is a semiconductor, although silicon hasbecome the dominant IC material not so much because it is a great semiconductor material, butrather because it is relatively easy to work with.

The silicon used for IC fabrication has been highly purified, grown into nearly perfect crys-tals and sliced up into discs, called wafers, less than a millimeter (mm) thick and anywhere from100mm (4") to 300mm (12") in diameter (smaller sizes were used early in the development of theindustry but are now rarely used in production and 450mm wafers are currently in development).Silicon wafers are highly polished - appearing mirror-like, extremely flat, and extremely cleanand particle free at the start of fabrication.

100mm (4"), 125mm (5") and 150mm (6") wafers typ-ically have a flat section ground onto one or more edges tomark how the crystal planes are oriented in the wafer andallow consistent alignment of various layers built up on thewafer - see figure 2a. 200mm (8") and 300mm (12") wafersuse a small notch in place of a flat because a flat takes awayan unacceptable amount of wafer area on the larger wafers -see figure 2b.

Silicon wafers were at one time internally manufac-tured by the IC companies who then fabricated circuits onthem, but now virtually all IC manufacturers purchase thewafers from a third party.

There are three major types of silicon wafers currently in use for IC fabrication:

� Raw wafers, silicon wafers without any additional processing. For state-of-the-art ICs raw wafers are mainly used for memory such as DRAM and Flash.

Figure 2. Silicon wafer orienta-tion indications.

Introduction to Integrated Circuit Technology

Copyright © 2001 - 2012 IC Knowledge LLC, all rights reserved 5

� Epitaxial wafers, silicon wafers with a single crystal silicon layer deposited on them. The deposited layer typically has different properties than the underlying raw wafer. Epitaxial layers allow the properties of the layer in which the devices are formed to be more tailored than in a raw wafer and are widely used for the latest state-of-the-art Logic ICs. Epitaxial wafer costs are 1.4 to 2.5 times the cost of a raw wafer.

� Silicon on Insulator (SOI) wafers - silicon wafers upon which an insulating layer is formed with a thin single crystal silicon layer on top of the insulating layer. SOI wafers reduce the amount of power drawn by an IC when the circuit is switching at high speed. SOI wafers costs are 4 to 15 times the cost of a raw wafer. SOI is primarily used in low power and some high performance applications. We expect the use of SOI to increase and even become mainstream as linewidths continue to shrink.Figure 3 illustrates the basic

wafer manufacturing process.The process steps are:

� Pull crystal ingot - a small seed of single crystal silicon is dipped into a crucible of molten silicon. The crucible and seed are rotated in opposite direc-tions and the seed is slowly withdrawn from the crucible (figure 3a).

� Ingot grind - the silicon crystal ingot is ground to create a con-sistent diameter for the whole ingot (figure 3b).

� Saw off ingot ends - the two ends of the silicon ingot will not be usable and are sawn off using a diamond saw (figure 3c).

� Saw up the ingot into wafers - the Ingot is sawn up into wafers each approximately 1/2mm to 3/4mm in thickness (100mm to 300mm wafers) (figure 3d).

� Edge grind wafers - the edges of the wafers are ground to round off the sharp edges. Edge grinding minimizes chipping of the wafer edges during subsequent processing (figure 3e).

� Lap wafers - a process called lapping is used to flatten out the wafers and ensures the two wafer faces are parallel (figure 3f).

� Damage removal etch - a special wet etch is used to etch off the surface damage left from lap-ping (figure 3g).

Wafer Types� Raw - basic wafer

used to make ICs.� Epitaxial (Epi) - a

raw wafer with a single crystal film deposited on it.

� Silicon On Insula-tor (SOI) - a thin single crystal silicon film on an insulating film, on a handle wafer.

Figure 3. Silicon wafer manufacturing process.

Introduction to Integrated Circuit Technology

Copyright © 2001 - 2012 IC Knowledge LLC, all rights reserved 6

� Polish - the polish step removes the final residual damage layer on the wafers and creates a mirror polish surface (figure 3h).

� Final clean - the final clean step removes any contaminants left on the wafer surface from the previous steps (figure 3i).

The resulting wafer is a highly pure - nearly perfect single crystal, with crystal planes pre-cisely oriented to the wafer surface.

7.0. How small is small?Before delving into wafer fabrication - the

fabrication of incredibly tiny IC elements, it isuseful to review the size scale of the circuitelements being discussed. The units of mea-sure utilized in this publication are defined intable 1..

Once the units are defined, the size ofsome common objects may be put in perspec-tive, see figure 4.

The average height of people is measuredin meters, at the millimeter scale is the diameter and thickness of a penny and at the micrometerscale is the thickness of a human hair and size of bacteria, viruses are nanometer scale in size andfinally atoms are picometers in size.

The smallest processes currently in production are approximately 20nm processes. From fig-ure 4, 20nm may be put into perspective relative to some common objects. 20nm is on the order of

Table 1. Units of size

Unit Symbol Relationship

Meter m ---

Millimeter mm One thousand millime-

ters equals one meter.

Micrometer, also called a micron

m One million microme-ters equals one meter,

one thousand microme-

ters equals one millime-ter.

Nanometer nm One billion nanometers equals one meter, one

thousand nanometers equals one micrometer.

Picometer pm One trillion picometers equals one meter and

one thousand picome-ters equals one nano-

meter.

Figure 4. Size scale

Introduction to Integrated Circuit Technology

Copyright © 2001 - 2012 IC Knowledge LLC, all rights reserved 7

a large virus and much smaller than bacteria. The critical gate oxide in these processes is just over1nm in thickness or only a few atoms thick!

8.0. Wafer fabricationIC�s may have anywhere from a few components to >2 billion transistors and 2 billion capac-

itors integrated together on the latest 2Gb DRAM�s. Depending on the wafer size and IC com-plexity, there may be anywhere between tens and tens of thousands of die on a wafer. The key tosuccessfully fabricating all of these components on one substrate is the ability to selectivelychange the properties of silicon. In the next several sections, the key technologies behind waferfabrication will be described.

8.1. Cleaning The cleanliness of wafers during processing is so critical that

every wafer is cleaned prior to any high temperature or depositionstep. Complex sequences of acid and alkali solution are utilized toremove particles, organic films, metals and any pre-existing�native� oxide films. The most commonly used clean in theindustry is the RCA clean (see sidebar).

8.2. OxidationOne of the key reasons that silicon is the most commonly

used semiconductor, is that it is easy to work with and one of thekey factors in making silicon so easy to work with is the ability togrow a high quality insulating layer on silicon. If silicon isexposed to oxygen or water vapor at high temperatures, oxygencombines with silicon to form silicon dioxide, a glass. Silicon

dioxide is stable at high tem-peratures, an excellent bar-rier and an excellentinsulator; figure 5 illustratesthe basic silicon oxidation process.

RCA Clean� The most commonly

used clean - the RCA clean includes multi-ple steps:

� SC1 (standard clean 1) - removes organic films and particles.

� SC2 (standard clean 2) - removes metals.

� HF (hydrofluoric acid) removes silicon dioxide layers.

� May include SPM (sulfuric peroxide) - removes gross organic layers.

Figure 5. Silicon oxidation.

Silicon oxidation� Silicon (Si) combines

with oxygen (O2) at high temperature to form silicon dioxide glass (SiO2).

(1)Si O2 SiO2+

Introduction to Integrated Circuit Technology

Copyright © 2001 - 2012 IC Knowledge LLC, all rights reserved 8

8.3. PhotolithographyAt the heart of wafer fabrication technologies is photolithography. Photolithography defines

the patterns that when used in conjunction with etching can pattern deposited and grown thinfilms, and combined with ion implantation can selectively change the properties of silicon. Photo-lithography creates patterns in photoresist - a liquid photosensitive chemical that resists etchingprocesses.

The photolithography process works as follows:� Surface prime - a surface treatment to drive off moisture and improve adhesion of photoresist,

typically accomplished by heating the wafer in a primer chemical vapor.� Coat - a small amount of photoresist (a few milliliters) is dispensed onto the center of the

wafer and then spun at high speed to produce a uniform thin film - see figure 6

Figure 6. Photoresist coating process.

� Pre-bake - photoresist contains a solvent used to keep the polymer and photo sensitive chemi-cal in suspension. Once the photoresist is coated onto the wafer, the solvent is driven off by a pre-bake to stabilize the film.

� Exposure - the exposure step photographically transfers a pattern from a reticle to the photore-sist coating on the wafer surface. Reticles are glass plates with patterns of opaque and trans-parent areas. A reticle will typically have the patterns for a few die on it and will be stepped across the wafer exposing the pattern after each step to cover the wafer with patterns. In order to ease the task of reticle fabrication and make the process less defect sensitive, reticle pat-terns are either 5x or 4x the size of the desired feature on the wafer, and the reticle pattern is optically shrunk before reaching the wafer. Figure 7 illustrates pattern formation on a wafer by photolithography.

� Post exposure bake - the latest exposure tools use very short wavelength - deep ultraviolet light (DUV) to enhance resolution. Photoresist for DUV are chemically amplified and require a bake step after exposure to complete the chemical reaction initiated by exposure.

� Develop - the effect of the developing step depends on the type of photoresist being used. For positive photoresist the developer dissolves areas exposed to light more quickly and dissolves areas where the light was blocked more slowly. For negative photoresist the developer dis-solves the areas not exposed to light more quickly and areas exposed to light more slowly. The end result is that for a well designed process, at the end of the develop step the pattern from the reticle is replicated in the photoresist.

� Post bake - prior to developing, bake temperatures must be kept low enough to not break down the photosensitive chemical in the photoresist. Following the develop step this is no lon-

Introduction to Integrated Circuit Technology

Copyright © 2001 - 2012 IC Knowledge LLC, all rights reserved 9

ger a consideration and higher temperature bakes are used to stabilize the film prior to subsequent processing.

Following the photolithography process the photoresistpattern may be used to create selective processing, for exam-ple an etch that etches an underlying film but does not etchphotoresist. Following etching the photoresist pattern wouldthen be stripped off and the wafers sent on for further process-ing that might include further photolithography steps.

8.4. Ion Implantation

As was previously discussed, impurities may be used to change the electrical properties ofsilicon. Introducing impurities into silicon in a controlled manner is the key to forming integrated

Figure 7. Photolithography process.

Figure 8. Ion Implantation.

Photolithography� Prime - coat the surface

with an adhesion pro-moter.

� Coat - coat the wafer with a photosensitive liquid - photoresist.

� Soft Bake - bake the pho-toresist to dry it without breaking down the photo-sensitive chemical.

� Expose - expose the pho-toresist to ultraviolet light through a patterned reticle transferring the reticle pattern into the photore-sist.

� Post Exposure Bake - cer-tain types of photoresist require a bake after expo-sure to complete the exposure reaction.

� Develop - wash away the photoresist wherever light exposed a pattern into the photoresist and leave the photoresist wherever the light was blocked.

� Post Bake - bake the pho-toresist to stabilize the film prior to subsequent processing. The photore-sist no longer needs to be light sensitive so higher temperatures than soft bake may be used.

Introduction to Integrated Circuit Technology

Copyright © 2001 - 2012 IC Knowledge LLC, all rights reserved 10

circuits. Ion Implantation is currently the most commonly used method for introducing impuritiesinto silicon wafers. In an ion implanter, impurities to be introduced into silicon are ionized, i.e.stripped of one or more electrons giving the impurity ion a positive charge. A high voltage electricfield is then used to accelerate the ions to a very high energy. This acceleration process is done ina vacuum so that the ions don�t collide with any gases during acceleration. The accelerated ionsare then �implanted� into the silicon surface by virtue of their high energy causing them to pene-trate the surface they are aimed at before coming to rest.

The Ion Implantation process is made selective by using aphotoresist pattern to block impurity ions from reaching the sili-con surface where no impurities are desired. The selective intro-duction of impurities process begins with the growth of a thinsilicon dioxide layer. The silicon dioxide layer protects the sili-con surface, but must be thin enough not to block the implantedions. Photoresist is then applied and patterned as outlined in fig-ure 7, and ion implantation is performed. Following ion implan-tation, the photoresist is stripped off and a high temperaturefurnace process is used to anneal out the damage from the high energy ions impacting the silicon -see figure 8.

8.5. EtchingIn the following sections a variety of thin film deposition techniques and thin films will be

described. Whenever a thin film is patterned in IC technology, some form of etching step isalways involved, this even includes techniques such as damascene and dual damascene.

Early in the development ofIC technology all etches werebased on liquid chemicals (wetetching). In most cases wet etch-ing - etches in all directions atthe same rate (isotropic etch).The result of this is that whileetching down through a film, theetchant is also etching under-neath the edge of the photoresist - see the left side of figure 9. More recently etching processeshave used excited gas molecules to performing etching and can achieve faster etching in onedirection than in other directions (anisotropic etching) -see the right side of figure 9.

When linewidths were rela-tively large, isotropic etchingwas acceptable, but as lin-ewidths shrink isotropic etchingcan result in complete undercut-ting of the feature being printed- see figure 10. Notice how theright most line is almost com-pletely undercut away.

In addition to undercutting problems from wet etches, as IC technology advanced new mate-rials such as polysilicon and silicon nitride could either not be wet etched or could not be wet

Ion Implantation� Impurities with an

electric charge are accelerated to high energy and shot into the wafer surface.

Figure 9. Isotropic versus anisotropic etching.

Figure 10. Isotropic etching versus linewidth.

Introduction to Integrated Circuit Technology

Copyright © 2001 - 2012 IC Knowledge LLC, all rights reserved 11

etched with etchants that are compatible with photoresist. In cur-rent state-of-the-art processes, wet etching has been almost com-pletely supplanted by dry etching.

Dry etching processes use a variety of halogen containinggases such as fluorine, chlorine and bromine compounds. Highfrequency energy is used to split up the gas molecules in a lowpressure chamber creating highly reactive products. The chemi-cally active etch products may be combined with chemistries toproduce polymers and ions to create anisotropic etches.

8.6. Chemical Vapor Deposition (CVD)In CVD processes, gases or chemical vapors are reacted to

form a deposited film, most commonly at low pressure. Reactionsmay be induced by heat as in CVD, high frequency energy as inPlasma Enhanced CVD (PECVD) or light as in Photon AssistedCVD (PHCVD). If the chemicals being used in the reaction aremade up of molecules combining metals and organics, then theCVD process is referred to as Metal Organic CVD (MOCVD).There is also Low Pressure CVD (LPCVD) and Sub AtmosphericCVD (SACVD).

An example of a CVD process is illustrated in figure 11..

In figure 11, ammonia gas (NH3), and dichlorosilane (SiHCl2), are reacted to produce adeposited solid film of silicon nitride (Si3N4), and gaseous by products that are pumped away-hydrogen chloride (HCl), chlorine Cl2, hydrogen (H2) and nitrogen (N2). Some of the films usedin IC fabrication that are typically deposited by CVD are listed in table 2.

Figure 11. CVD process.

Etching� Wet etching uses

liquid chemicals, primarily acids to etch materials. Wet etching is predomi-nantly non-direc-tional.

� Dry etching uses gases in an excited state to etch materi-als. Dry etching may be non-direc-tional or directional.

� Isotropic - etches at the same rate in all directions.

� Anisotropic - etches in one direction faster than other directions.

Introduction to Integrated Circuit Technology

Copyright © 2001 - 2012 IC Knowledge LLC, all rights reserved 12

8.7. Sputter Deposition

Many metal films used in IC fabrication are deposited bysputtering. In the sputtering process, argon gas is excited by a high energy field to split up intopositively charged argon ions and free electrons. An electric field attracts the argon ions toward atarget made out of the material to be deposited, the argon ions physically knock loose atoms of thetarget material that then deposit out onto the wafer surface. Although CVD generally covers stepsbetter than sputtering, not all metals may be deposited by CVD. Figure 12 illustrates the sputter-ing process

8.8. Chemical Mechanical PlanarizationAs IC technology developed, more and more metal layers have been required to provide

interconnect with acceptable signal delays. As more and more metal layers are stacked up, topog-raphy becomes an overwhelming issue. CMP combines chemical and mechanical materialremoval to produce fully planar surfaces. The CMP process is illustrated in figure 13.

Table 2. IC thin films commonly deposited by CVD

Film name Chemical formula

Silicon dioxide SiO2

Silicon oxynitride SiOxNy

Silicon nitride Si3N4

Polysilicon Si

Titanium nitride TiN

Tungsten W

Fluorinated Silicon Glass (low-k) SiOxFy

Hydrogen and carbon doped oxide films (low-k).

Figure 12. Sputter process.

Thin Films� Grown or deposited

films a few microns to a few nanometers thick.

� Common films include:

� Silicon Dioxide - insulator.

� Silicon Nitride - protects the fin-ished IC.

� Polysilicon - used as a conductor and a control electrode for certain types of tran-sistors.

� Aluminum - used as a conductor.

� Copper - the newest type of conductor, has lower resistance than aluminum.

� Titanium Nitride or Tantalum Nitride - used as a barrier between films to prevent interactions.

Introduction to Integrated Circuit Technology

Copyright © 2001 - 2012 IC Knowledge LLC, all rights reserved 13

the exact mechanism of CMPdepends on the material being pol-ished and the polishing slurry used,but generally speaking a chemicalreaction is used to soften up the filmbeing removed and then mechanicalabrasion from the slurry particlesremoves the material. The advantageof CMP over a purely mechanicalprocess is that the chemicals used to perform the softening function can be tailored to attack onlyspecific materials so that the removal rate is relatively high for one material and relatively low forother materials. The difference in removal rates between materials allows a polishing stop layer tobe used in some cases.

8.9. Putting it all togetherThe unit steps described above are combined with other unit steps into complex process

flows with hundreds of steps where dozens to 50 or more reticles are used to print patterns ontowafers. The end result is a number of ICs on a single wafer, that depending on the wafer size andthe size of the IC may number, tens, hundreds, thousands or ten of thousands of ICs. Each IC mayhave tens of millions or even over a hundred million circuit elements.

Memory ICs now in production have over 2 billion transistors and 2 billions capacitors on asingle IC. Microprocessors are also reaching close to a billion transistors per die. The left side offigure 14 illustrates a technician holding a 300mm wafer and the right side illustrates an enlarged

photograph of one Pentium 4TM IC. The IC illustrated in figure 14 contains over 42 million tran-sistors in an area smaller than the surface of a dime and there are approximately 281 of these ICson a 300mm wafer.

Figure 14. 300mm wafer and Pentium 4TM IC. Photos courtesy of Intel.

Figure 13. Chemical mechanical planarization.

Introduction to Integrated Circuit Technology

Copyright © 2001 - 2012 IC Knowledge LLC, all rights reserved 14

9.0. Cleanliness and YieldDuring processing, ICs are very susceptible toany kind of small particle or defect landing onthe surface of the IC. ICs in wafer form arereferred to as die. Experience has shown thatparticles 1/2 to 1/3 the size of the smallest fea-ture on an IC can �kill� the die - referred to as akiller defect. For 65nm processing, particles assmall as 22 to 33nm in size can kill the circuit,and measurements of atmospheric air havefound that there are millions of particles thatsize or larger in a cubic foot of air. Themechanics of particle deposition from air arequite complex, but suffice it to say that with

millions of particles floating around, the likelihood of a killer defect landing on a given die is veryhigh. The number of killer defects is characterized by the defect density on the wafer surfacegiven in defects per unit area. The resulting yield depends on the size of the die and the defectdensity. Actual yield calculations require yield probability models, but the basic concept is illus-trated in figure 15. On the left side of figure 15, a wafer is shown with a defect pattern of 20defects and a relatively small die that results in 264 die per wafer. The number of die withoutdefects is 244 and the resulting yield is 92%. On the right side of the figure a wafer is shown withthe same defect pattern, but a larger die. There are 54 die per wafer of which 38 die have no defectresulting in a 70% yield!

Figure 16. Linewidth trends.

Figure 15. Defect density and yield concept.

0.01

0.1

1

10

Date

Intel X86

AMD X86

Power PC

AlphaIntel IA64

SparcTrend - 0.86091x/year

DRAM8080

8085

286

386

486

Pent

8086

PIIPIII

P4

Flash

1Kb DRAM 4Kb DRAM

16Kb DRAM 64Kb DRAM256Kb DRAM

1Mb DRAM

4Mb DRAM

64Mb DRAM256Mb DRAM

15001.04

4004 8008

Intel physical gate length

Introduction to Integrated Circuit Technology

Copyright © 2001 - 2012 IC Knowledge LLC, all rights reserved 15

As ICs have developed, the desire to pack more functionality onto individual ICs has drivenlinewidths to ever smaller sizes - see figure 16, and ICs have simultaneously gotten larger - seefigure 17. The net result is that lower particle levels are required for good yield, and the size ofparticles that can kill a circuit is also shrinking, requiring ever increasing cleanliness levels.

Figure 17. Die size trends.

In order to meet the cleanliness requirements and manufacture ICs with high yield, clean-rooms are used. Cleanrooms are rooms with continuous circulation of air out of the room, througha high quality HEPA or ULPA filter and back into the room - continuously sweeping particles outof the room.

Figure 18 illustrates a simple cleanroom. Cleanrooms used for IC production have the filterslocated in the ceiling and the air flows down the room to or near the floor before exiting the room.The down flow or vertical style of cleanroom keeps the dirtiest air down near the floor where nowork is exposed and has the benefit of gravity aiding in particle removal.The quality of clean-rooms is defined by federal standard 209E. Under 209E, a cleanroom with less than 100 particlesper cubic foot larger than 0.5µm is a Class 100 cleanroom, or less than 10 particles per cubic footgreater than 0.5µm is class 10, and so on. High yield on current state-of-the-art ICs requires betterthan Class 1 cleanrooms.

Water, chemicals and gases utilized in the manufacturing process must also be low in parti-cles and free of contaminants down to the parts-per-billion or even parts-per-trillion range.

10

100

1,000

Date

Intel X86AMD X86

Power PC

Alpha

Intel IA64

Sparc

Old trend - 1.14x/year

DRAM

Flash

1Kb DRAM4Kb DRAM

16Kb DRAM

64Kb DRAM

256Kb DRAM

1Mb

4Mb DRAM

4004

80088080 8085 8088

286

386 486

Pent

8086

15001.02

64Mb DRAM256Mb DRAM

PII

PIII

P4

Introduction to Integrated Circuit Technology

Copyright © 2001 - 2012 IC Knowledge LLC, all rights reserved 16

10.0. Wafer TestUpon completion of wafer fabrication, not all of the die on a

wafer will be fully functional. The yield loss at this step rangesfrom a few percent for mature processes, to 90% or more for newprocesses. In order to avoid adding additional value to defectiveunits during packaging, a 100% test of the die is performed.

Each die that has been fabricated on the wafer has a series ofpads referred to as bond pads where connections will be made tothe die during assembly. The die is covered with a protective pas-sivation layer everywhere except where the pads are located. Foreach type of die being tested a specialized �probe card� is fabri-cated with a set of tiny needles spaced apart so they line up withthe bond pad openings - see figure 19.

The wafer to be tested is held onto a chuck in a piece ofequipment referred to as a wafer prober - figure 20b. The proberalso holds the probe card and mechanically positions the probecard needles over the bond pads on a die, touches the needle downto make an electrical connection for testing, and following testinglifts the needles and positions them over the next die.

The wafer prober is connected to a tester, an automated pieceof equipment that performs electrical tests on each die - see figure20a. The tester is basically a computer and some power supplies,meters and function generators that can be programmed to perform a variety of electrical mea-surements. The tester communicates with the prober telling the prober when each die has beentested and whether the die is good or not. When a die tests �bad�, a tiny ink dot may be dispensed

Figure 18. Basic cleanroom configuration.

Cleanroom concepts� Cleanroom - a room

with continuos circu-lation of filtered air.

� Vertical cleanroom - a cleanroom with air flow from ceiling to floor - virtually uni-versal for IC produc-tion.

� Cleanroom class - specifies the maxi-mum number of parti-cles greater than or equal to 0.5µm in size. Class 10 has less than 10, Class 1 has less than 1, etc.

� HEPA filter - high efficiency particle air - blocks 99.97% of particles >0.3µm.

� ULPA filter - ultra low particle air - blocks 99.999% of particles >0.12µm.

� Raised floor - a type of floor used in most Class 10 or better cleanrooms. The floor is perforated to allow air to flow down through the floor. An under-floor air ple-num is used to return the air up to fans and filters for recirculation back into the ceiling of the room.

Introduction to Integrated Circuit Technology

Copyright © 2001 - 2012 IC Knowledge LLC, all rights reserved 17

onto the die to mark it bad or an entry may be made in an electronicmap to denote the location of the bad die.

11.0. PackagingSilicon ICs in �die� form are difficult to handle, fragile even though they have a protective

layer, and the tiny bond pads are difficult to connect to. In order to further protect the die andmake the parts easier to handle and connect, packaging is performed.

Historically, the most common packing method is as follows.� After wafer test, the wafer is mounted onto sticky tape stretched over a metal frame. The

backside - non circuit side of the wafer is stuck to the tape. An automated - high speed saw with a very thin diamond blade is used to saw apart the die. There is an area between each die that has no circuitry referred to as a street so that sawing does not damage the IC circuitry. The sticky tape serves to hold the individual die in place after sawing, see figure 21 - 1a through 3.

� Each good die on the wafer is now removed from the sticky tape and placed onto a metal frame referred to as a leadframe. This operation is performed by automated pick and place machines. The leadframe is etched or stamped into a pattern that will later become the electri-

Figure 19. Simple probe card.

Figure 20. Wafer test set-up.

Wafer Test Terminol-ogy� Probe card - a card

with tiny needles used to make elec-trical connections to IC die being tested.

� Bond pad - pads on the die where elec-trical connection may be made.

� Tester - a computer controlled system that performs elec-trical tests automati-cally.

� Prober - a piece of equipment that holds the probe card and wafer being tested and steps the card across the wafer contacting each die. Operates under control of the tester.

Introduction to Integrated Circuit Technology

Copyright © 2001 - 2012 IC Knowledge LLC, all rights reserved 18

cal connection pins that protrude from the package. The leadframe also includes a pad that the die is epoxy mounted to, see figure 21 - 4a through 4c.

� Tiny wires, typically gold, are now attached from each bond pad down to a corresponding leadframe pin. This is the electrical connections from the die to the pins that eventually connects the die to the outside world. Wire bonding is also performed by automated systems referred to as wire bonders, see figure 21 - 5.

� A rectangular area of black epoxy is now molded around the die and leadframe leaving the leadframe pins sticking out. Molding is accomplished in large presses under heat and pressure, see figure 21 - 6.

� A mechanical tool now punches out each individual packaged IC from the leadframe bars that held the units in rows. The tool also bends the leads forming them into their final configuration.

� The units are branded with a part number, company name, date, etc. on the epoxy plastic to identify the part.

Figure 22 illustrates the common package styles and their popularity.Recently, newer techniques have been developed to produce smaller packages and also

enable higher frequency connections to ICs. The relatively large pins used to connect plasticpackages to the outside world degrade high frequency signals. The newer techniques are generi-cally referred to as �chip scale� packaging, the idea being that the �package� is barely any biggerthan the silicon chip.

Figure 21. Plastic packaging process.

Packaging Flow� Saw up the wafer

into individual die.� Mount the die down

onto a leadframe.� Wirebond from the

die bond-pads to the leadframe.

� Mold epoxy around the die to protect it.

� Trim and form to break the individual packaged ICs apart.

� Mark the packages.

Introduction to Integrated Circuit Technology

Copyright © 2001 - 2012 IC Knowledge LLC, all rights reserved 19

One technique is called flip chip. In flip chip processing, tiny solder balls are fabricated onthe bond pads, the silicon chip is then flipped over and soldereddown to connecting pads on a substrate. The technique is compati-ble with high frequency, less expensive than wire bonding for highconnections counts, and results in a very small �package�. Addition-ally, for wire bonding, pads need to be at or near the edge of the ICto minimize wire length. In some cases the number of pads requiredfor electrical connections exceeds the number that will fit around theperiphery of the IC. The IC size must be increased just to accommo-date pads increasing the IC size and cost. With a flip chip approachpads can be placed in an array anywhere on the IC - see figure 23.

Figure 22. 2011 package types and market share.

Figure 23. Flip chip process

Chip Scale Packag-ing� Bare die are also

sometime referred to as chip scale packing - the package is no big-ger than the sili-con chip (die)

Introduction to Integrated Circuit Technology

Copyright © 2001 - 2012 IC Knowledge LLC, all rights reserved 20

12.0. Final TestDuring the packaging process, die may be damaged or packaging may not be correctly per-

formed. The defects introduced during packaging typically cause 1 percent or more of ICs to fail.Most customers today expect that only a few ICs per million will be non functional on arrival.Final test is a 100% test performed on each packaged IC prior to shipment to insure that any ICsimproperly packaged are not shipped. Final test is similar to wafer test except a handler is used to�handle� the IC packages and make connections - see figure 24b. The handler is connected to atester similar or identical to the one used at wafer test - see figure 24a.

At this point defective units are discarded and good units areready for shipment. Many handlers provide �bins� for good andbad parts and may also provide additional bins so parts can begraded by speed of operation or other criteria. It is quite commonto bin out parts at final test by grade. A common example wouldbe Pentium microprocessors. One process flow produces proces-sors that are then tested and sold as various speed grades. Thehighest speed grades are the rarest - highest performing parts andtherefore have the highest selling price. The variations in speedare due to process variations across a wafer and wafer to waferthat occur during wafer processing.

Figure 24. Final test system.

Final Test Terminol-ogy� Tester - a computer

controlled system that performs elec-trical tests automati-cally.

� Handler - a piece of equipment that moves parts from an input bin, makes electrical connec-tion to the parts for testing, and then moves the parts to a output bin or bins. Operates under con-trol of the tester.

� Binning - the prac-tice of sorting parts based on some mea-sured performance parameter.

Introduction to Integrated Circuit Technology

Copyright © 2001 - 2012 IC Knowledge LLC, all rights reserved 21

13.0. ConclusionIn conclusion, IC technology relies on the ability of semicon-

ductor materials to behave as conductors or insulators dependingon impurities selectively added to the semiconductor. The processof producing an IC is made up of:� A starting substrate - typically purchased.� Wafer fabrication - fabricates the IC - die, on the wafer surface.� Wafer test - tests each die.� Packaging - packages the die for easy handling and protection.� Final test - tests the packaged IC.

We hope you have found this publication interesting and infor-mative. Once again we would like to encourage you to give usfeedback, good or bad about content, depth of treatment or howunderstandable our explanations are. You may contact us by e-mailat info@icknowledge.com. There are also a variety of otherresources available at ICknowledge.com (click on any of theunderlined blue text to go there) including an extensive glossary ofIC terminology and a history of the development of the IC.