LECEDU EngineerVol 211 No 5 Feb Mar 1976

Manufacturing technologythe onward thrust

In the mainstream of the world-wide picture tube business, technical andmanagement personnel are confronted with the growing dichotomous crunchbetween inflationary increases in material, labor, and energy costs; and thelosses of profit margins that result from competitive pricing pressures. RCA'sPicture Tube Manufacturing Division is concentrating on reducing costs byimproving the productivity of our labor, the efficiency of our factories, and thequality of our raw materials. We are striving to enhance RCA's uniquecompetitive cost effectiveness through vertical integration by the efficientproduction of our own glass, masks, phosphor, mounts, stems, metal parts, andcoatings. An ongoing program is being formulated to further enhance ourproduct quality and manufacturing efficiency by utilizing computerized processcontrol technology and product flow transfer mechanization.

The cover picture of RCA's Circleville Glass Operation and the technical paperson process control in glass making and picture tube resident engineering in thisissue of the RCA Engineer tell part of this story. Future issues will enlarge uponthe technologists' contributions to RCA's picture tube business with a series oftechnical manufacturing papers covering the Circleville, Lancaster, Marion,Scranton, Juncos, and Barceloneta operations.

RCA's manufacturing engineers are in a unique position to improve productivi-ty. They work on the production line and are intimately familiar with the process,machinery, human, and cost elements of picture tube manufacturing. Togetherwith the industrial engineer's insight into the basic product flow and transferrequirements and the computer technologists' expertise, a team effort with theProcess and Equipment Development engineers is being made to furtheradvance the productivity through mechanization. Simple, reliable, and cost-effective alternate methods of color tube manufacture are being formulated,developed, and implemented. This action program is essential to maintain andenhance the profitability and leadership of RCA in the world-wide picture tubebusiness.

C. W. ThierfelderDivision Vice PresidentTechnical ProgramsPicture Tube DivisionLancaster, Pa.

F.L. Flemming

C.C. Foster

H. Rosenthal

P. Schneider

John PhillipsJoan DunnFrank StroblP.A. Gibson

Dottie SnyderJuli Clifton

RCA Engineer Staff

Editor

Art EditorContributing EditorComposition

Editorial SecretarySubscriptions

Editorial Advisory Board

Dr. J.J. Brandinger Div. VP, Television

Engineering, Consumer ElectronicsDr. D.J. Donahue Div. VP, Picture Tube Div.R.M. Cohen Dir., Quality and Reliability

Assurance, Solid State Div.VP, Engineering, NBC

Television Network Div.Mgr., Scientific PublicationsRCA LaboratoriesStaff VP, Engineering

Research and EngineeringExec. VP. Satcom System

Dr. W.J. Underwood Director, Engineering

Professional Programs

Our cover

focuses on RCA's Circleville. Ohio. plant whichmanufactures glass parts for color picture tubes. Thestorage silos (actually located at the rear entrance) holdraw materials to be used in the manufacturing process.The color picture tube face panel and funnel in theforeground are final products of the Circleville glassplant and are used by RCA's Scranton, Pa and Marion.Ind . plants in the construction of color picture tubes(see R Schneider's article. p.3)

Photo Credit: John Semonish and Bruce Hull. Solid StateDivision. Clark, N J

Vol. 21INo. 5Febl Mar1976

A technical journal published byRCA Research and EngineeringCherry Hill, N.J.Bldg. 204-2 (PY-4254)

Contents

alila Engineer To disseminate to RCA engineers technical inforsna-tion of professional value To publish in an appropri-ate manner important :ethnical developments at RCA,and the role of the engineer To serve as a medium ofnterchange of technical information between variousgroups at RCA To create a commur ity of engineer -rig interest within the company by stressing the in-terrelated nature of all technical contributions To

help publicize engineering achievements in a mannerthat will promote the interests and reputation of RCA inthe engineering fielc To provide a convenient meansby which the RCA engineer may review his professionalwork before associates and engineering manage-ment To announce outstanding and unusualachievements of RCA engineers in a manner most likelyto enhance their prestige and professional status.

Emphasis - manufacturing engineering

Editorial input 2

Cover feature Circleville glass operations R.K. Schneider 3

Manufacturingengineering

Computer utilization in a glass -forming operation G.W. Blair 10

Picture Tube Division resident engineering L.F. Hopenl C.E. Shedd 14

Development of compound for quadradiscs G.A. Bogantzl S.K. Khanna 17

SCOPE-the power testing paladin E. Cutshaw 22

Special Impressions of technology in China Dr. J. Hillier 27

General -interestpapers

Glass passivat on of IC's W. Kern 36

Thin metal films D.M. Hoffman 44

Engineer and the Corporation Technical education in RCA Dr. W.J. Underwood 48

General -interestpapers

Tungsten -matrix cermolox tubes R.E. Byrami J.T. Mark 52

Structured programming Dr. P.G. Anderson 55

Microwave parametric amplifier Dr. R.L. Camisal B.F. Hitch' S. Yuan 60

Software cost performance H.L. Fischer 63

Epoxy as a die attach in IC's B.M. Pietruchal E.M. Reiss 68

Television applications of PLZT ceramics B.M. Soltoff 72

Optical processing of wideband data G.T. Burton' R.F. Croce 76

Pulse doppler radar performance H.H. Behling 80

Engineering andresearch notes

Microstrip discriminator A. Schwarzmann 84

High reliability COS/MOS E. Reiss 84

Departments Pen and Podium 86

Patents granted 88

Dates and Deadlines 90

IEEE Fellows 91

News and Highlights 92

Copyright 1976 RCA CorporationAll rights reserved

editorial Transition

input a look back, a look ahead

The RCA Engineer was started twenty years ago as a journal "by and for" RCA engineers.Throughout those twenty years, the journal has responded to the changing needs of the RCAengineering community.

Initially, the journal emphasized technical information that was fairly easy to read and digest.However, through the early to mid sixties, much of RCA's engineering was directed toward thesophisticated technologies associated with satellites, defense systems, and computers, as well asadvanced components and materials developments. These influences tended to make the journalcarry articles with increasingly detailed, complex technical content. This direction continuedthrough the late sixties and early seventies. At the same time, many non -electronic businessesbecame a part of RCA. Profiles of the new subsidiaries and more business -oriented articles werepublished, and the technical material in the journal continued to follow the complex technology.

The mid seventies will probably be best remembered as a time of economic stress. As a result,recent issues have emphasized product design, process control, design to cost, automatic testing,and manufacturing engineering-subjects that deal with the basic, but always tough, engineeringquestion, "How do I make a better product for less money?"

Thus, Change earmarks the history of the magazine. My predecessor, Bill Hadlock, first editor of theRCA Engineer, who retired this February, did an excellent job of keeping the journal in step with thechanging environment. In fact, Bill's most valuable legacy is an editorial network that is designed torespond to change.

As the journal's new editor, I am particularly appreciative of Bill's example and fully aware of theneed for continuing change. Change, as always, to match your interests. After all, you-the RCAengineer- are the journal: its author, its reader, and its personality.

My major goals are to improve readability and usefulness and to produce a journal that moredirectly mirrors the personality and interests of the RCA engineering community.

To do this effectively, I need, and solicit, your help and advice. Through my years at RCA, I have metmany of you, and have always found you to be a fertile and ready source of new ideas andperspectives. Contact me directly or through your Division Editorial Representative (inside backcover). I value your input and will attempt to respond personally.

Address correspondence to:

John C. PhillipsRCA Bldg. 204-2Cherry Hill, N.J. 08101

2

Making glassfor television picture tubesR.K. SchneiderRCA staled producing glass parts for television picture tubes in 1970. thus becoming theonly fully integrated color television manufacturer in the Western Hemisphere. Makingglass parts is relatively straightforward: mix irg raw materials, netting, forming, process-ing and shipping. Making glass parts for picture tubes becomes complex. however, du? tothe requirement for large. dimensionally precise glass piec?s that approach opt calquality. For example. the typical face panel for a 25 -inch pidue tube weighs over 27pounds and measures 22'/2 X 171/2 inches and is roughly ' Anch-thick at its center; yet thecontour must be within 0.017 -inch of specifk:ation.

IN THE EARLY DAYS of black -and -white television, RCA con-sidered enterinc the picture -tube glass manufacturing busi-ness. At least two early feasibility studies were made, but

fo- one reason or another, the project did not take place. A 1965study, however, culminated in the construction of the Glass Plant in

Circleville, Ohio.

Construction of the Circleville plant began in September, 1969, and a little overone year later, the first face panels were shipped to RCA's picture tube plant inMarion, Indiana. Since that time, Circleville has established itself as a majorsupplier of both panels and funnels for RCA's Marion, Indiana; Scranton,Pennsylvania; and Midland, Ontario plants.

Although the Circleville operation is similar in many ways to other glassfactories, it has many unique features which set it apart. The stringent qualityrequirements imposed on the parts, both visually and dimensionally, make thisparticular glass manufacturing operation a very complex process requiringcareful adaptation of conditions, equipment, and techniques. The specializedglasses used for color television picture tube parts are part of a series of glassesdeveloped by the industry over the past years and are formulated so that theirproperties meet the current requirements of the tube and set maker.

3

Robert K. Schneider. Mgr , Mix and Melt Engineering,Glass Operations. Circleville, Ohio. received theBachelor of Ceramic Engineering in Glass Technologyfrom Ohio State University in June. 1953 After gradua-tion he joined RCA AS AN Engineering Trainee in thepicture tube activity at Marion. Indiana He has workedon product and process development engineering forboth black -and -white and color picture tubes and wasengineering leader responsible for glass andmetallurgical development at the Marion. Indiana plantMr Schneider helped in planning portions of theCircleville facility and in 1969 was assigned his presentresponsibilities which involved establishment andoperation of the laboratory facilities and the glassmaking processes He is a Registered ProfessionalEngineer and a member of the American CeramicSociety

Reprint RE -21-5-22Final manuscript received December 8. 1975

The processes used for making panels and funnels are basically similaralthough the glass composition and the parts shape are necessarily different. Itis interesting to note that RCA uses a "pressing" technique for forming bothpanels and funnels while other domestic glass manufacturers press only thepanel and employ a centrifugual casting or "spinning" process for formingfunnels.

The glass parts manufacturing process at Circleville is a "straight-line" flowoperation beginning with receipt of raw materials, continuing through batchpreparation, melting, forming, hot and cold processing of the formed parts, andending with the shipment of product at the opposite end of the line. (See theprocess flow diagram.)

[Editor's note: the numbers in the text refer to the photographs of the glassmaking operation in this article.]

Process Flow Diagram

Receipt and Storageof Raw Materials

IBatch Preparation

MeltingJ

ANEL

Stud Seal Moil Sear Off

COOling S,,il

Face Polish Button Seal

Edge Mill & Bevel Cooling

Acid Treat Edge Grind & Bevel

Wash & Dry Reference Point Grind

Inspect & Pack Acid Treat

1Wash & Dry

Inspect & Pack

Shipment

4

1. Raw material received by rail and stored in silos. 2 Ingredients for making glass are weighed itspecial weigh hopper scales lire the on.shown above.

Raw materials, all in granular or powder form, are received primarily by bulkshipment in rail cars. 1 Major ingredients are stored in large silos while minoringredients are stored in bags or drums and then transferred to storage bins asneeded. Among the major ingredients common to both glasses are sand, sodaash, pot ash, dolomite, and potassium nitrate. Strontium carbonate is used in thepanel glass and litharge in the funnel glass as the primary means of providing thenecessary x-ray attenuation properties.

Minor ingredients include oxides of titanium, cerium, antimony, arsenic, cobalt,and nickel.

A quantity of each raw material is weighed out according to a batch formulaprescribed so the resultant glass has the stringent physical propertycharacteristics required for color picture tube manufacture. 2 The weighedingredients are collected and delivered to either the panel or funnel "batch"mixer where they are blended together 3

The mixed batch is discharged onto a belt conveyor where a quantity of cullet isadded. (Cul let, stored in two separate silos, is crushed glass of nearly the samecomposition as the glass being made). The batch and cullet are then delivered tostorage bins ready to be charged into the respective glass meltingfurnance. 4 It should be noted that the entire batch preparation system isautomatic, being centrally controlled from a single instrument panel andactivated upon demand of the furnaces. 5

1.114:010 NM StompM Nem Mem.

PANEL

In.,/ P..

3. Weighed batch Ingredients are blended in large pan typ? misers. 4. Mixed batch is stored In bins ready to be charged into the furnace ondemand.

Ilkition Se.

CaNing

5. Batch is prepared automaticallywhen needed. This central control pan-el gives the operator a graphicrepresentation of batching operationas it is functioning.

1

6. Hot glass Is extruded through an orifice,cut to form a gob and dropped into a mold(for the funnel in this case.)

7. Hot glass in tunnel mold.

The melting furnaces are typical gas -fired, side -port, regenerative types, inwhich a large ceramic tank is heated to over 1500°C by the combustion of gasand preheated air over the surface of the glass within the space under the"crown" or roof of the tank. Several hundred tons of glass are contained in thetank allowing adequate melting or reaction time between the charging of thebatch into one end and the careful withdrawal of homogeneous glass from theother. The glass, particularly in the case of panels, must be of near opticalquality, i.e., essentially no visible defects. This includes not only inclusions,such as minute stones or bubbles, but cords which are the results of inadequatehomogenization of striae within the glass melt. These striae have very slightdifferences in index of refraction and, with certain intensity and orientation,disturb transmitted light in such a way as to distort an image viewed through thepanel face. Toward the withdrawal end of the furnace, the glass is cooled so itcan be extruded through an orifice, cut into a "gob" and dropped into a moldwhich is one of eleven mounted on a large circular metal table. 6,7 After amold receives a gob of glass, the table is quickly rotated into a position where ashaping plunger enters the mold, forcing or "pressing" the hot glass to conformto the contours of the mold on one side and the plunger on the other. 8 Theplunger is then withdrawn and, as the table rotates to other positions, blasts ofcool air are directed at the glass surface until the part is sufficiently cooled sothat it can be removed from the mold without changing shape. The precision towhich the shape of these pressed parts, both panel and funnel, must becontrolled is almost unheard of in the glass industry considering the overall sizeand weight of the parts. For example, the panel face contour must be held towithin 0.017 inch of specification at all points across what is to become theviewing area. This precise control of contour is necessary so that the criticalspacing between the glass and shadow mask can be achieved during the picturetube manufacturing operation. The periphery of both the panel and funnel mustbe kept within close tolerances to assure that pairs of parts will properly matchwhen selected at random for assembly at the picture tube plant.After the funnel and the panel are formed, they are processed in somewhatdifferent ways; these processes will be described separately.

Panel processing

After the pressed panel has been removed from the mold, special metal studs(for holding the tube's shadow mask) are sealed into the sidewall. 9 Then theassembly is sent through a lehr oven where controlled cooling takes place toprovide the required degree of thermal stabilization to the glass. The viscosity of

8. Pressing the gob takes place when aplunger forces the glass to conform to themold. The plunger forms the inside contourof the part.

9. Metal pins are sealed Into the sldevfalls of the hotpanel.

6

10. The face contour and periphery are carefully gaged. 11. Polishing the outside surface of the panel.

the glass varies continuously from the molten state temperature to that at whichit becomes, for all practical purposes, rigid, and oerceptible structuraladjustments stop. If this lower temperature is reached before the structure of theglass has reached its most stable arrangement, structural change will continue.Although in commercially annealed glass this change is imperceptible at roomtemperature, it is quite evident at temperatures well below the strain point (thestrain point is usually considered to be the temperature below which nopermanent structural changes take place within the glass). These changes, ifencountered during picture tube manufacture, can be sufficient to alter thecontour of the glass panel and thus effect the critical shadow -mask -to -glassregistry.

After this lehr treatment, the panels are checked for inside surface contour,periphery, stud location, and visual quality. 10 The resul:s of inspections andgaging throughout the process are important keys to process control for makinghigh quality glass parts.

Since the television picture is viewed through the panel, its outer surface mustbe ground and polished by a process similar to that used in the preparation oflenses. This allows the picture to be transmitted through the glass withoutinterference or distortion from surface aberrations. 11 The grinding andpolishing is accomplished using specially fabricated laps (contoured pads) inseveral steps utilizing successively finer abrasives. After the face is polished, theedge (which must mate with the funnel when the tube is assembled) is groundflat, 12 beveled to remove sharp edges, and given an acid fortificationtreatment. Finally the parts are washed, dried, given a firal inspection and thenpacked for a brief stay in the warehouse before being shipped to one of thepicture tube plants.

Funnel processing

The funnel part, after being removed from the mold, undergoes considerablymore "hot" processing than the panel. A moil, or solid plug of glass which closesthe small end of the funnel, is seared off with a carefully cirected and controlledflame. On another machine, a prepared piece of tubing is sealed to this small,now open, erd of the funnel. 13 Careful alignment of the funnel and this"neck" tubing is critical during this operation so the part will meet exactingrequirements.

Re.... end 1.of Pow 11.

WO,

Foe

Sp MN .

^ Pg

12. The sealing edge must be ground flat.

13. A section of 'ubing Is sealed to the funnel to fcrm theneck.

7

14. A metal button is sealed into the funnel sidewall.

FUNNEL

15. Parts are Inspected after being cooled in a lehr oven. 16. The seal edge of the tunnel must also be ground flat.

After the neck has been sealed to the funnel, a small metal buttonshaped part is inserted into the sidewall of the funnel to make provision for thetube's high voltage connection. 14 Immediately after this operation is per-formed, the funnel is passed through a lehr similar to that used for panels. 15

Funnels which pass the visual inspection at the cold end of the lehr aretransported by conveyor to the finishing operation where the rim of the funnel(which is to be mated with the flat edge of the panel) is ground flat and beveledby using diamond -impregnated metal wheels. 16 Other surfaces around thefunnel's rim, which are to be used as reference points during tube making, areground precisely also. 17 After an acid fortification treatment, 18 washingand drying, the funnels are given final gauging and inspection before packing.As with the panels, the funnels are stored in a warehouse before being shippedto the tube manufacturing plants.

Mold design, fabrication, and maintenance

A unique feature of the Circleville operation is the provision of complete molddesign, fabrication and maintenance facilities at the glass plant. Primarymachining of stainless steel castings used for new mold equipment is done onnumerically controlled milling machines. 19 Complete facilities to performnecessary grinding, polishing, drilling, and plating operations are also providedin the plant mold shop. 20,21

After fabrication, each new piece of mold equipment is measured on anautomatic measuring machine. 22 All glass -forming surfaces are checked toassure that necessary machining tolerances are maintained. The equipment isthen released for use in the glass -forming process.

Throughout its life, mold equipment is routinely returned to the mold shop forcleanup and minor repairs. After each repair, the equipment is remeasured andreleased. Mold equipment data obtained during these inspections, along withsubsequent product data, are used as the base for continual optimization andimprovement of current mold equipment designs- a vital step in the productionof these high quality, highly sophisticated glass parts.

8

17. Reference surfaces are precisely ground

Summary

18. The ground seal edges are acid treated or 19. New mold equipment W prepared by N/C machining."fortified".

The success of the Circleville Glass Operations represented a significant eventin the history of RCA. Today, RCA is the only color television manufacturer inthe Western Hemisphere to achieve this degree of vert cal integration.

The process used in the production of glass parts for color picture tubes iscomplex and sophisticated, requiring a meticulous adaptation of conditions,equipment, and techniques. The RCA plant is domestically unique in that apressing process is used to form both the face panel and the funnel. Theoperation is further enhanced by having a complete facility for design,fabrication, and maintenance of the mold equipment used in the manufacture ofthe parts.

Through the support, dedication, and effort of many people, the Circlevilleoperation has become a major supplier of both panels and funnels for the RCApicture tube manufacturing plants.

21 Facilities for chrome plating are available tor use on certain mold parts

20. The mold surfaces are carefully polished aftermachining.

22. Exacting measurements are made on both mold and glass for use in op'imizin1 moldequipment

O

Computer utilizationin a glass -forming operationG.W. Blair

The RCA Glass Plant at Circleville began the production of glass faceplates and funnels forcolor television picture tubes in the Fall of 1970. The concept of "distributed computersystems.' was chosen as the means of fulfilling the plant computer requirements.Presently. located throughout the plant, there are four computer systems performingprocess control, quality assurance. engineering, and plant accounting fuictions. Thispaper describes these systems and their use in the forming operation for picture -tubefaceplates.

COMPUTER SYS I EMS at theCircleville glass plant are being developedaround what is referred to as the"distributed computer concept." In thisconcept, "worker" minicomputers areused for control of the various areas ofthe production operation. Pertinent datais then transferred from these "workers"to a larger "supervisory" computer alsowithin the plant. The "supervisory" com-puter is then used for more extensive dataanalysis and design programs. The"supervisory" computer may then passdata on to yet higher level data processingsystems for further analysis.

N/CDESIGN

DATA

N/CDESIGN

DATA

CORDAxGAGINGSYSTEM

I()

PICTURE TUNE

DIVISION

SYSTEM

DATA

TRANSMISSION

The Circleville plant produces the glassfaceplates and glass "funnels" used toform color -television pictures tubes.Such a glass -forming operation involvesfirst, control of the mold equipment usedin the forming process and second, con-trol of the forming process itself. Thismust include the long-term control ofquality levels and design modifications tothe molding equipment, as well as themore immediate control of process equip-ment and product acceptance or rejec-tion.

Four computer systems located

N/C LISTINGS

U

MOLD

SUPERVISORY

SYSTEM

EQUIPMENTDATA

MOLD EQUIPMENTACCEPTANCE

4{5

NUMERICALCONTROLTAPES

QUALITY ANALYSISENGINEERINGMOLD DESIGN

U. S

MAIL

GAGE DATA

ACQUISITION

a CONTROL'ij GLASS

DATA

PROCESS CONTROL

QUALITY ACCEPTANCE

GAGE DATA

ACQUISITION

S CONTROL.2

Data flow through the Circleville 'distributed computer' system.

GLASSDATA

Si

PROCESS CONTROL

QUALITY

ACCEPTANCE

throughout the plant are used to fulfill thevarious operation requirements. One ofthe systems controls an automaticmeasuring operation, primarily for moldequipment. Two of these systems areutilized for data acquisition and controlon the production floor, and the fourth isthe "supervisory" system which is usedfor extensive quality analysis, mold -design programs, and engineeringcalculations.

Glass forming process

Formation or pressing of a picture tubefaceplate begins with three basic pieces ofmold equipment used to form the glass.These three pieces are the plunger, themold, and the shell. New pieces of moldequipment are machined, in plant, on anumerical -control (1%1 J C) tape -controlledmilling machine. After machining is com-pleted, the machined surfaces are handpolished to remove the marks left by themachining process.

Faceplates are formed on a "modified"Lynch ELP glass press. The press has ahydraulically driven ram on which theplunger is mounted. The ram is located atone station of an eleven -station rotaryindex table. Eleven molds are positionedon the index table. Through a doubleindexing of the table after each press cycleand a unit which transfers the shells from

Gary W. Blair, Manager, Technical ComputerApplications, Glass Operations, Circleville. Ohio,received the BSEE from Pennsylvania State University in1963. Mr. Blair's early experience with General MotorsCorporation involved the design of sequencing controlsfor automatic production equipment. In 1965 he joinedCorning Glass Works where he was project engineer forthe first real-time computer control system installed bythat company. From 1968 to 1970 he was affiliated withStruthers Energy Systems, a consulting firm specializingin the design of "total energy systems" for commercialand industrial complexes. He joined RCA at Circleville inlate 1970 and has been responsible for plant computerapplications planning and implementation since thattime.

Reprint RE -21-5-10Final manuscript received September 15, 1975.

Cutting a plunger on the N/C mill.

mold to mold, a total of five shells areused on the press. This arrangementprovides a total of fifty-five plunger -shell -mold combinations in the faceplate form-ing process at any given time.

To produce a faceplate, a gob of moltenglass (at approximately 1000°C) is

dropped into a mold which has a shellseated on top of it. The mold and shell arethen indexed into position under the ram.The ram drives the plunger downward,pressing the glass between the plungerand the mold -shell cavity, forming afaceplate. After a momentary dwell at thebottom of the stroke, the ram and plungerare retracted.

After the pressing operation is completedthe faceplate remains in the mold forseven indexes of the press table. Duringthis time, air is blown on the faceplatecooling it to approximately 500°C. Theshell is lifted from the mold at the thirdtable index after pressing. At the eighthtable index, the faceplate is lifted from thepress and transferred to a conveyor beltwhich transports it to the stud sealingarea.

There are a number of stud insertionmachines in the stud sealing area. Eachfaceplate is cycled across a stud machinewhere four metal studs are inserted intothe skirt of the faceplate. Insertion is doneby heating the metal studs to a

temperature of 1200°C and then pressingthem into the faceplate skirt.

After the studs are inserted, thefaceplates, which have now cooled to a

Cordons gaging system-measuring the periphery of a shell.

temperature of approximately 450°C, areplaced in an annealing lehr. Through theannealing cycle, the faceplates arebrought up to a temperature of 505°Cand then slowly cooled to a temperatureof 35 to 40°C. This cycle removes thestrain created in the glass during thepressing and stud insertion operations.

After annealing, the faceplates are visual-ly inspected and are conveyed to the gageroom where the final check of thedimensional characteristics associatedwith the forming process is made.

Computer systemsCordax gaging system

This system utilizes a Digital EquipmentCorporation PDP-8 computer with a 4kcore memory and 32k fixed -head discmemory. The computer is interfaced to aBendix Cordax 3000 measuring machinethrough a digital input/output systemand a 12 -bit analog -to -digital converter.The measuring machine may, at operatorinitiation, be operated in manual modewith the computer acting as a data -acquisition system or in automatic modewith computer control of machine posi-tion and data acquisition.

Mold equipment is measured after initialmachining on the N/C mill, after handpolishing of the surface, and periodicallythroughout its life.

Based on initial Cordax data, equipmentis either released for operation orreturned to the milling machine for sur-face remitting.

Data acquired on this system is used intwo ways: I) to determine the currentstatus of a piece of mold equipment(compared to dimensional limits) for usein manufacturing and 2) to provide datawhich can be valuable for product designmodifications.

Data output from this system is generatedas a hard -copy printout on the systemteletype and/or as a paper tape. Paper -tape outputs are taken to the supervisorysystem where they are input for moreextensive data analysis and designprograms.

Gage data acquisitionanc contra) systems

Two gage data acquisition and controlsystems provide data acquisition andcontrol for a faceplate production line,but can be expanded to service a funnelproduction line. One of the two systemsis, in fact, presently being used by afaceplate and a funnel line but only thefaceplate line application will be dis-cussed in this paper.

These systems are General AutomationSPC-I6/ 65 computers utilizing the RTX-16 real time executive operating system.The primary function of the systems is toprovide automatic on-line data acquisi-tion from the production gages and per-form immediate analysis required forprocess control and quality acceptance.

Their secondary function is to gather data

11

REMOTE

Tr

PAPERTAPEPUNCH

Periphery 8Seal Edge DataContour DataStud/Plane Data

STATUSDISPLAYS

System Software

PAPERTAPE

READER

CONSOLETTY

NOT ENDSAGE

/ HI- I :It EEns7dduData

Stud/Plane Surveys

SPC- 6

A/ 32K Core

STUD/PLANE

GAGE

Contour SurveysPeriphery SurveysSeal Edge SurveysQuality Statistical Data

STUD/PLANE

GAGE$2

PERIPHERY

GAGE

Gage data acquisition end control system layout.

which can be utilized as input to thesupervisory system for engineeringevaluation, quality assurance analysis,and equipment design modification.

The system hardware consists of aGeneral Automation SPC-I 6 / 65minicomputer with 32k of core memory,hardware priority interrupt structure,real-time' clock, power failuredetection/auto-re-start interrupts, andhardware multiply and divide. Machinecycle time is 960 ns. Input/output slotsare provided for use in interfacingperipheral equipment. A high-speedpaper -tape reader (400 char/ s) providessystem input so that the entire systemsoftware can be built using the facilities ofthe supervisory system. Printed outputcan be generated on the system console orremote teletype. Data output on papertape is also available via 75 char/ s papertape punch. A low-level analog inputsystem consisting of a 256 -channel mul-tiplexer and 12 -bit A/ D converter withfixed gain ±100 mV to ±10V inputsaccepts input signals from the sensorslocated on the various production gages.

The critical dimensions of a faceplate arethe inside contour, the peripherydimensions and thickness, and the loca-tion of the studs in reference to the facecontour and periphery. To provide dataacquisition and control of thesefunctions, data is obtained from a "hot -end" gage located in the stud sealing area

CONTOUR

GAGE

STATUSDISPLAYS

and two stud plane -to -contour gages, apanel contour gage, and a periphery sealedge -thickness gage located in the gageroom.

These systems provide automatic collec-tion and storage of quality acceptancestatus information for stud plane -to -contour for each stud machine. Throughcomputer -controlled displays located inthe gaging and stud machine areas, thestatus of each stud machine is indicated.Conditions included are I00% inspectionrequired, sampling to maintain accep-tance required, or acceptance withoutinspection.

Data necessary for contour control can begenerated for either a single piece of glassor as a summary of up to eleven pieces ofglass. The mode of operation desired isselected by a switch on the manual entrybox and then the faceplates to be includedin the survey are cycled across the contourgage. The output provides the averagevalue and standard deviation for allentries as well as a listing of the worstpoint and its location on the faceplate foreach faceplate gaged. Also included is anestimated percent of product withinspecification for any gage points wherethe data statistically indicates less than99% within specification.

Panel periphery and seal edge thicknessdata are normally output as a summary of

fifteen faceplates. The average deviationfrom design for sixteen gaging positionsaround the faceplate periphery is

provided. A listing for each piece is alsoincluded in the survey, indicating the shellwhich produced the piece and the posi-tion and value for the worst point on thefaceplate.

Data required for analysis beyond thelimits of the Gage Data Acquisition andControl System are generated on papertape by setting a switch on the manualentry box of the gage from which data isto be gathered. Data for weekly qualityassurance analysis and for engineeringevaluation and mold design modificationare obtained in this way. Data tapes thusgenerated are used as input for the dataanalysis and design programs on thesupervisory computer.Each gage has an associated manual data -entry box, which can be used to identifythe equipment used to process a piece ofglass. Data acquisition is initiated byentering the required information into themanual entry box and then depressing acomputer scan button. Display lights onthe entry box indicate the quality accep-tance or rejection of that particularfaceplate.

Data for control of stud -sealing machinesis generated both at the hot end and thecold end of the Lehr. Surveys can be runutilizing the "hot -end" gage, located inthe stud machine area, at any time. Toobtain a survey, the operator cycles threepieces of glass from a given stud machineacross the gage. The data is accessed bythe computer and corrections are made toaccount for dimensional changes whichwill occur as the faceplate goes throughthe Lehr cycle. The output, which isprinted on the KS R-35 "hot -end" teletypelocated in the stud machine area, is thenused to determine what control ad-justments are to be made on the givenstud machine.

At the cold end of the Lehr, a survey ofproduct is made for each stud machineonce per hour. To assure that surveys arerun for each machine, a computer -controlled display in the gaging area has alight for each stud machine which in-dicates the time at which a survey is to berun. The surveys from "cold -end" dataare printed out on the "hot -end" teletype.This data is then reviewed by the stud -machine operator to verify the precisionof the control moves which have beenmade on the basis of "hot -end" data.

12

Supervisory system

The Supervisory System is a GeneralAutomation 18/ 30 computer with 32k ofcore memory. The system I/ 0 includes a2.5 -million -word removable -pack discmemory, high-speed paper -tape reader(400 char s). high-speed paper -tapepunch (75 char/ s), 600 line/ min lineprinter, 500 char/ min card reader,magnetic tape transport, and Calcomp565 Incremental Plotter.

This system provides the capability forextensive analysis of mold equipment,which is not possible on the smaller"worker" systems. A complete mold -equipment data base is maintained on thissystem. Each time a piece of mold equip-ment is measured on the Cordax System,the data tapes are entered into thesupervisory system. Outputs generatedinclude present deviation from design aswell as deviation from originaldimensions, deviation from last measure-ment, and current mold componentlifehours. These data are utilized toprovide information regarding both wearand distortion effects on equipmentthroughout its life. Plots of the deviationfrom design are also provided each time adata tape is processed. A summary of thecurrent dimensions of all operating moldequipment is output each week.

Product data is obtained from the GageData Acquisition and Control Systemsthroughout each week. This data is thenprocessed on the 18/30 System, and is

used to provide overall process qualityanalysis.

Mold equipment data obtained from theCordax System for a specific piece ofmold equiment and the follow-up datafrom the Gaging Systems for productproduced by the equipment are utilized asinputs to the mold design programs.Through this procedure, product varia-tion from specifications can be assignedto errors in the mold equipment orprocess variations. These factors are thentaken into account in the modification ofmold equipment designs. When mold -equipment design modifications are to bemade, new design data is output tomagnetic tape. This tape is transmitted toPicture Tube Division MIS in Edison,NJ, where the major N/C mold -designprograms (which are too large for theplant computer system) are located on aSpectra 70/6 system. N/C tape outputsproduced at Edison are output onmagnetic tape, and the results aretransmitted back to Circleville. Programson the supervisory system check themagnetic tape data for transmissionerrors, convert it from 80 -charactertransmission records to N/C mill in-structions, and punch it onto Mylar tape.These tapes are used to control the N/Cmills in cutting additional mold equip-ment.

Utilizing cross -operating software, thesupervisory system is used to compile andbuild the software for the Gage DataAcquisition and Control Systems. Thisallows modifications of those systems to

Hot -end gage and remote teletype located in the stud -sealing area. This gage is d?signed to measurefaceplates while tie glass temperature Is around 450`C. Teletype is enclosed in the gage console forprotection from the environment.

be developed with a minimum amount of"worker" system downtime. This is asignificant, advantage since the produc-tion lines being serviced by those systemsoperate on a 24 -hour -per -day basisthroughout the year.In addition to being used for theEngineering and Process Controlprograms discussed above, thesupervisory system is utilized as the

Circleville plant MIS system. All

programs required for plant accountingare processed on this system.

Looking ahead

The "distributed computer concept" af-fords a production plant the ability todevelop long-range plans for plant com-puter applications without makingcapital committments four or five years inadvance. Using minicomputers dedicatedto specific applications, an integratedplant system can be implemented in astep-by-step manner, with economicjustification satisfied after each step.Initially major emphasis in the Circlevilleplant had been implementation of thoseapplications where the maximum returncould be realized in the minimum time.Future computer systems being con-sidered include direct digital control ofthe glass press, glass furnace data analysisand control, and automatic productionreporting. Enhancements to the existingsystems include implementation of real-time operating software for the"supervisory" systems and on-line com-munications between the "supervisory"system and the "worker" systems.

Stuc-plane-to-contour gage it he gage room. Manual entry box is on theright side of the gage

l3

Picture tube divisionresident engineeringL.F. Hopen C.E. Shedd

The role of Resident Engineering demands close liaison with the engineering functions ofboth product design and manufacturing. Resident Engineering must provide thenecessary product -design -on-location" support to all manufacturing activities. Thisrequires close cooperation and continuous support of associated manufacturingengineering activities such as plant engineering, quality control, and product safety. Thevarious engineering functions and activities that combine to assure a reliable and high -quality manufacturable product are described.

RCAhas established and main-tained leadership in the design, develop-ment, and manufacture of television pic-ture tubes. This leadership in a high-technology industry has resulted in aworldwide licensing arrangement withmost of the major manufacturers oftelevision picture tubes. The scope of the

RCA engineering effort to support thisworldwide responsibility is far more com-plex than those of competitive companieswho concentrate on more limited marketareas.

Reprint RE -21-5-17Final manuscript received November 26, 1975

L.F. Hopen, Mgr.. Picture Tube Process and MaterialsDevelopment Engineering. Lancaster, Pa.. received aBSME degree from Washington University. St. Louis.Missouri and has taken post graduate courses withPurdue University, Indiana University and MarionCollege He joined RCA in Camden in 1949 and wasassigned to the Marion Plant where he served in a varietyof engineering and supervisory capacities includingPlant Manger In 1971, he was transferred to France asOperations Manager of Videocolor SA (an RCA Jointventure) and in 1972 to Italy as Vice President andGeneral Manager of Videocolor SPA. His current assign-ment includes Resident Engineering responsibilities atthe Marion, Indiana and the Scranton. Pennsylvaniapicture tube plants

Clifford E. Shedd, Mgr of Color Picture Tube EquipmentDevelopment. Lancaster. Pa., received his B.S.degree inIndustrial Engineering I rom Penn State University in1940 He Joined RCA in 1946 as an Engineer in the Black -and -White Picture Tube Development area. From 1946 to1952. he held various lead engineering positions inprocess engineering for black -and -white. large -size pic-ture tubes. In 1953 he became Manager. EquipmentEngineering for Color Television. in 1959, he becameProduct Engineering Manager of Color Picture Tubes, in1960. he was made Manager. Process EngineeringEquipment Development for color picture tubes. Since1969. Mr Shedd has been Manager of EquipmentDevelopment for the Picture Tube Division

Engineering effort in picture tubes can becategorized into four major areas all ofwhich are highly important in achievingthe end result of a product which isinnovative and competitive for perfor-mance, cost, and quality. These areas are:

I) Research at RCA Laboratories, Princeton,N.J.

2) Picture Tube Division Engineering, Lan-caster, Pa.

3) Picture Tube Division Resident Engineering

4) Plant Associated Engineering includingManufacturing Engineering, IndustrialEngineering, Plant Engineering. Qualityand Reliability Engineering.

Role of Resident Engineering

In the classical view of product engineer-ing, the transition from product conceptto manufacture would follow the orderlisted above. In practice, however, theinformational flow on new or existingproducts is multidirectional and utilizesall areas of expertise to reduce the timingand cost of new product introduction orproduct improvement. This article dealsprimarily with the contributions of one ofthe four areas of expertise, the ResidentEngineering activity. Resident Engineer-ing is the term used to describe the "onsite" technical services and assistanceprovided by the product engineeringactivities (Product Support Engineeringand Equipment Development Engineer-ing) at the manufacturing plants. It is avital link in a dynamic high-technologyindustry where unit cost dictates max-imum control and minimum delay inproblem solving.

Product Support Engineering

Resident Product Support Engineeringcapabilities are structured to provide thebasic facilities and skills in manytechnical areas. These skills areas include:

Chemical and Physcial Laboratory Design Engineering Standardizing

Applications, Reliability, and Safety

Support Activities (Maintenance, Shops.Library, etc.)

The application of such skills to productdevelopment and processing is describedin the functions and responsibilities listedbelow:

I) Engineering Standards-Maintenance ofup-to-date Engineering Standards is a

14

major requirement in the RCA TelevisionPicture Tube Division. The complexity ofboth the processes and the products requiresstrict adherence to standardized proceduresin order to manufacture RCA product tospecification but also to fulfill Technical AidAgreement commitments to foreign com-panies. Resident Engineering is responsiblefor the proper editing, processing, approvalreview and distribution of standardizinginformation initiated at the manufacturinglocation or elsewhere. The EngineeringStandards activity at each manufacturingfacility works in close cooperation with thecentral Engineering Standards activity inLancaster.

2) Laboratory Assistance-Incoming materialinspection experience over years of picturetube manufacture has proven the criticalrole that assuring high quality materials canplay in performance and cost. The Chemicaland Physical Laboratory (CPL) performsspecified tests on each incoming lot of themore critical materials before their use inproduction. Non -routine tests are per-formed as requested or indicated whendeviations in materials quality aresuspected. The CPL participates in thedecision -making process when questions ofusability, availability, cost, and producteffect must be resolved.

3) Materials Support-In addition to theregular inspection of incoming materials,Resident Product Support Engineeringassists the plant functions in the selection,testing, and vendor approvals for a widevariety of materials used in tube manufac-ture or in general plant operation.

4) Plant Engineering Support-RCA plantshave responded positively to the increasedemphasis on the environmental aspects ofplant operation. Support Engineering worksclosely with the Plant Engineering functionto achieve and control plant effluent levels inaccordance with Federal and Stateregulations.

5) Plant Safety Support-Resident Engineer-ing actively participates and, in many casesprovides leadership in safety -relatedactivities such as the Central Safety Com-mittee and the subcommittees involvingchemical, electrical, and radiation safety.Members work directly with the Plant Safe-ty Director in the investigating and for-mulating of plant requirements, and inpublishing of safety standards and safetymanuals. Additional effort and testing, havebeen required to monitor plant conditions(air, noise, etc.) for compliance with OSHAstandards.

6) Product Safety Support-The activity inResident Engineering monitors and testsproduct for compliance in the major areas ofconcern for television picture tubes-fire,shock, implosion, and x-radiation. RCA'slong-standing commitment to product safe-ty and compliance with existing regulationshas resulted in considerable emphasis oncontinuing effort and coordination at alllevels of engineering and management.

7) Direct Manufacturing Support-Thisactivity accounts for the largest segment ofResident Product Support Engineering. Theefforts of this activity can be divided into fivemajor categories:

Author's note: Subsequent to the submittal of this article, Dr. D.J.Donahue, Vice President Engineering, Picture Tube Division, an-nounced organizational changes which included a furtherstrengthening of the role of the Resident Engineering activities at theMarion, Indiana and Scranton, Pennsylvania plants. The reorganiza-tion established three major Product Engineering activities:

Process and Material Development (includes C & P Lab) Product Design Applications, Reliability and Safety._

A functional reporting relationship was established between theabove Lancaster activities and their plant counterparts to promotemore direct communication and establish a broader base for skillsutilizatior in new products and improvement programs. Engineeringsupport for problem solving at the plants will also benefit from themore efficient utilization of skills. The administrative functions forresident engineering remain the responsibility of the ResidentManager, Product Engineering and Resident Manager, EquipmentEngineering.

-Len Hopen

a) Process Control: The monitoring of cer-tain critical areas in the processes for partsand tube making which require specializedlaboratory equipment or analysis tech-niques.

b) Process Development: The developmentof new or modified in -plant processesrequired as a result of changes in products,materials, or techniques.

c) Product Improvement: Efforts to improvethe product for increased customer accep-tance, improved performance, lower cost,or some combination of the three.

d) Problem Solving: The responsibility ofworking with the manufacturing organiza-tion when problem areas develop whichrequire added technical assistance.

e) Product Evaluation: The Applicationsfunction evaluates product from a

customer poirt of view on a daily samplingbasis. These evaluations include receivertesting, coil room (field -free) testing, lifetesting under various prescribed con-ditions, and environmental testing. Theresults of these tests can trigger significantchanges and improvements in the productthrough Support and ManufacturingEngineering.

8) Sales and Customer Assistance --The resi-dent Applications Engineering activityevaluates customer receivers, analyzesapplications problems, provides defectanalyses, and sets up product demon-strations to customers at the request of theSales or Marketing groups.

Equipment DevelopmentEngineeringThe Resident Equipment DevelopmentEngineering is also an integral part of thetotal plant operating structure but

functions under the guidance of Lan-caster Equipment Development Engi-neering. The Resident group sharesresponsibility for the design, run-in,proper use, and updating of all plantmanufacturing equipment. Thisresponsibility includes maintainingequipment design capabilities with con-tinuing emphasis on process im-provements, equipment improvements,cost reductions, and flexibility of setupamong tube types.

The dynamic nature of the picture tubebusiness requires that the manufacturingoperations maintain the capability toproduce established types of picture tubeson short notice but at the same time havethe expertise and capability to producenew types as they are introduced orrevised on an almost continual basis. TheResident Equipment DevelopmentEngineering activity plays a key role inthe design and installation of new equip-ment as well as the modification ofequipment required for the introductionof new types of picture tubes or therevision of established types. The Resi-dent Equipment Development groupdesigns the fixturing, tooling, gauges,testing equipment and process machinesthat fall within the areas of expertise ofthe groups engineers and designers.

Many problems arise on a daily basis.Not the least of these is the maintenanceof the specialized tube -making equip -

15

ment. It is a responsibility of ResidentEquipment Development Engineering toprovide technical help to the variousservice departments to help minimizedown -time and obtain maximummachine efficiency. Some of the problemsmay have to do with the equipment itself;others may relate to equipment partswhich are no longer available and forwhich. replacement must be devised orfabricated. New sources for replacementparts must be found and the new partsevaluated before replacement is im-plemented.

Continually, efforts are made to helpboth Production Engineering and

Product Engineering with productionproblems on the equipment, or the effectsof process variables on the equipment.Through mutual efforts, programs are setup to maintain controls on machinevariables and their effect on the product.New product innovations particularlyrequire a great deal of attention duringthe introductory stages particularly in theimplementation of new tube types orprocesses. Usually, the advent of a newtype brings many innovative re-quirements in terms of new equipmentand equipment revisions. Most of theeffort, including paperwork, for thesenew requirements is handled by andthrough Equipment Development

Various parts for the electron gun mount are assembled with a special fixture around a revolving work table.

Left

So

electron -gun mount assembly is check on a visual comparator. Right, all tubes are final tested before shipment.

Light -house stands (left) are used to expose phosphors in proper register. Centrally located electrical controls (right)allow the entire phosphor application and exposure operation to be monitored and checked from one position.

Engineering. Included are all the workfrom the basic facility study through theprocurement paperwork and finally theblueprints for construction. Then, afterthe equipment is completed, EquipmentDevelopment Engineering is responsiblefor its inspection and for any subsequentrevisions that may have to be made as theresults of production shakedownproblems.

If the facilities require a major engineer-ing design effort, the requirements areforwarded to the Lancaster EquipmentDevelopment Engineering group fordesign. Upon completion of the design ofthe new facilities, the plans are submittedto the local factory group for review and,if approved, are released for construc-tion. Upon completion, the equipment isinspected by Resident EquipmentEngineering prior to shipment and in-stallation. When production parts areavailable, an operating run is made to testthe equipment with the product it is tomanufacture.

The local manager of EquipmentDevelopment Engineering functionallyreports to the local Plant Manager. He ison his staff and provides assistance inplant layout, facility studies for new tubetypes, and new processes. He is usually amember of the plant committees such aselectrical safety, mechanical safety, laser,radiation, engineering, and education. Inaddition, ' members of the ResidentEquipment Development Engineeringgroup usually participate in other plantcommittees responsible for OSHA com-pliance, energy conservation, andenvironmental problems, as well as in theplant's suggestion systems and thevarious cost reduction programs.

Conclusion

The foregoing material describes thefunctions and responsibilities of the twoconstituents of the Picture TubeDivision's Resident EngineeringActivity-Product Support Engineeringand Equipment Development Engineer-ing. Because the responsibilities of theResident Engineers, both domesticallyand worldwide, are so varied and yetoften so vital to the establishment andmaintenance of our manufacturingoperations,it is hoped that this article willlead to a clearer understanding of theseefforts and an appreciation of their manycontributions.

O

Development of compoundfor quadradiscsG.A. Bogantz S.K. Khanna

The rapidly expanding use of 4 -channel sound requires apprcximately three times thebandwidth of stereo discs to be recorded on the discrete phonograph record. Thisincreased bandwidth demands a new fineness in the compound from which discrete discare made. This paper reveals the inadequacy of current stereo record compounds and theextent to which the compound requirements for discrete 4 -channel records have beenmet. It also reveals, in part, the basis for the bel of that records are the World's most precisemass-produced product.

THE ADVENT of discrete 4 -channeldisc records was based upon the exten-sion of the state-of-the-art in recordingand record manufacturing in severalareas. For the first time in the industry, acommercial record was marketed with fmsound. This innovation was used tosuper -audibly -impose two additionalchannels of sound upon the two stereosignals on the two walls of the recordgroove. -lre fm carriers are placed at 30kHz and each is modulated independent-ly of the other. In this way. four channelsare recorded upon one record groove,each wall of which carries one stereobaseband signal and one fm signal.

So that a discrete 4 -channel disc recordplayed on a stereo system will reproduceall of the sound recorded on the record, itis necessary to record as baseband stereothe sum of the left front and left backaudio signals on the left groove wall;likewise, the sum of the right -front andthe right- back audio signals is recordedon the right groove wall. In this way, thenew records are completely compatiblewith 2 -channel stereo. The differencebetween the front and back audio signalsis recorded on the fm channels respective-ly. On 4 -channel playback, algebraiccombinations of these signals producefour discrete channels.

Gregory A. Bogantz, Member, Engineering Staff, RCARecords, Indianapolis, Indiana, worked as ControlEngineer at radio station WVNO-FM during high school.and received the BSEE and MSEE degrees from PurdueUniversity in 1969 and 1970 respectively. From 1970 to1972. Mr. Bogantz worked as service manager of Produc-tion Audio Service in Lafayette, Indiana. where he gainedexperience in the consumer audio equipment field. InJanuary 1973 he Joined the staff of RCA RecordEngineering in Indianapolis. Indiana, specializing in discrecord technology, where he does mastering of all newtest and technical series records for RCA custom clients.Mr. Bogantz is currently concentrating on OuadraDisc(CD -4) cutting, processing, and playback, and haspublished an article on CD -4 pickups. He is co-author ofa patent pending on a simplified CD -4 modulatingsystem and a member of the Audio Engineering Society.

Sanvan Kumar Khanna, Senior Rheologist, RCARecords. Indianapolis. Indiana. graduated from PunjabUniversity in India and from Borough Polytechnic inEngland. He carried out postgraduate research inPolymer Rheology at Cranfield Institute of Technologyin England. and was rheologist with Associated Elec-trical Industries in the United Kingdom. He was seniordevelopment engineer with Gulf Oil (Canada) ChemicalDivision and is now engaged in rheological research invinyl resins and compounds.

Reprint RE -21-5-3Final manuscript received April 9. 1975

This greater information density wasachieved by the use of essentially one-halfspeed recording and a new cutting stylusdesign. To pick this information off therecord required a new design for theplayback stylus. [This stylus is namedafter the man who developed it. Mr.Shibata]. The Shibata stylus is shapedpyramidicall to give better contact withthe groove and uses a smaller contactradius to provide better tracing of thewaveforms in the groove.

Goals

While considerable effort was exerted ondeveloping phonograph record com-pounds with low surface noise, the in-troduction of the quadraphonic recordhas placed new demands on wearcharacteristics due to the decreasedwavelengths involved. The extensivewear testing procedure to be describedhere is the result of weeding throughseveral methods and proposals.

Originally, the intent of the program wasto develop a compound which wouldwithstand rather severe playback con-ditions in order to be compatible with thesingle inventory concept of stereo -

quadraphonic record production. Thus,the new compound would have to with-stand playback on rather unsophisticatedconsumer -type stereo or monophonicphonographs and be able to rendergood performance on a CD -4 playbacksystem that the customer may acquire inthe future. A little thought will show thatthis is a pretty rigorous demand to makeof an information storage medium likethe relatively "soft" vinyl phonographdisc which has traditionally been easilysubject to deterioration by simple scuff-ing and scratching. The first questionthen becomes, is this requirementrealistic? Is it even possible on a practicalbasis?

Initial testing indicated that this goal mayin tact be quite impossible to obtain. Wewere rather quickly able to determine thatexisting stereo compounds were totallyunsuited to the task on several counts. Wehad hoped to be able to play a

QuadraDisc upwards of 100 times on aconsumer phonograph equipped with a0.7 -mil radius conical stylus tracking atabout 5 -grams and then be able to getacceptable performance from the disc ona CD -4 playback system. Severalquestions become immediately obvious

17

at this point, like Why 100 plays and whatis acceptable quadraphonic performance.But, more about that later. Most stereocompounds failed miserably without hav-ing to answer the above questions.

After only five or ten plays on theconsumer phono, the quad performanceof the disc was unlistenable due to acutelyhigh noise or such a severe carrier loss asto cause the playback demodulator'smuting circuitry to activate, disabling theCD -4 function altogether. Anotherproblem with most stereo compoundswas unsatisfactory first -play perfor-mance on good quality CD -4 playbackequipment by means of a Shibata-tippedstylus tracking at 2 -grams. This was dueto poor molding characteristics and ex-cessive scanning loss and groove defor-mation during the first pass of theShihata stylus.

Review of wearing conditions

At this point we should stop and considerthe various conditions under which a discmay be played by a consumer and try todetermine a meaningful disc -wear for-mat. Modern playback cartridges can begenerally broken down into two majorcategories, I) consumer types withrelatively high tracking force and conicalstyli and 2) component quality typestracking at under 3 -grams and equippedwith conical, elliptical, or Shibata-typestyli. The typical conical stylus forstereo playback has a tip radius of about0.6-0.7 mil. Most elliptical styli havebiradial dimensions of about 0.3 mil by0.7 mil. The smaller dimension is thehorizontal scanning radius and the largerthe vertical bearing radius.

Henceforth, we will refer to all Shibata-type styli as simply Shibata styli; theseinclude the Quadrahedral, Obata,Pranianik and others. They all have thesame essential characteristics of smallscanning radii on the order of 0.2 to 0.3mil and quite large vertical radii of about3 mil.

We are then faced with four major wearconditions: I) high -force conical, 2) low -force conical, 3) low -force elliptical, and4) low -force Shibata.

Without belaboring the point, let us saythat after sampling several consumerproducts and based on past experience wehave further defined these categories as:

I) 4.5 gram conical, 2) 2 gram conical, 3)1.25 gram elliptical, and 4) 2 gramShibata. Of these, the 4.5 gram conicaltest is the most severe and the 2 -gramShibata test the least severe. Most of ourtesting involves these two tests with theother two performed only on compoundswhich yield satisfactory results on thefirst tests. We also further specify that the4.5 gram conical test is done with aconsumer -type player and pickup whichusually does not incorporate antiskatecompensation. The remaining tests areintended to simulate relatively goodplayback conditions and so are per-formed on good -quality component -typechangers with properly adjustedantiskate compensation.

Consider now one of the questions raisedearlier. How many playings must the discwithstand to be considered acceptable?This is certainly a speculative point, butone that should be defined. We felt that aconsumer would be likely to play a disc inquad mode probably at least 10 timesduring its useful life and maybe as manyas 25 or more times if he is in the habit ofshowing off his sound system to friends.Therefore, we want to be able to providehim with minimally about 25 plays withas little degradation as possible.However, to allow ourselves some leewayin meeting this requirement easily, we setout to see if we could achieve upwards of100 plays without serious degradation.We suspected that the 4.5 gram conicaltest would be much more severe andfrankly did not know initially how manyplays should be required of this test. Wesimply elected to shoot for 100 plays tosee if it was possible.

We feel that it is unrealistic to expect acompound to perform well even initiallyin the quad mode with a conventionalconical tip pickup due to the mechanics ofthe system. A 0.6 - 0.7 -mil scanningradius is simply too large to cope with thesmall carrier wavelengths at the innerdiameter of a QuadraDisc, even when thedisc is in perfect condition. We, therefore,chose to disregard the quad performanceof a disc when played with a standardconical tip, regardless of tracking forceand regardless of the number of times thedisc had been played. It is our feeling thathardware manufacturers must accept theresponsibility of equipping their playerswith the proper type of CD -4 playbackstylus Shibata. Hence all of ourevaluations of Quadra Disc performance

would he based on the use of a Shibata tipduring the measurement.

Another important aspect of wearing adisc is the amount of time allowedbetween successive passes of a givenportion of the record. This is a questionwhich has been discussed at length in theliterature as it relates to stereo perfor-mance, and the general conclusion is thatsome time should be allowed for vinylrelaxation between plays. Someresearchers state that as much as 24 hoursshould be allowed. Initial tests wereconducted using specially equippedchangers with timers to allow differingamounts of time delay between playingsof a QuadraDisc. We found that therewas some variability among compounds,but some did allow a relatively shortrecuperation time of about ten minutes.Not only was this desirable, it was almosta necessity for our testing program due tothe many compound formulations wewere investigating. We simply could notafford the time involved in allowing a discto relax for as much as a day betweenplayings. Since some early formulationsshowed promise of performing well withonly several minutes relaxation time, wechose to make this a requirement of theQuadraDisc formulation. We elected tooperate the wear testing changers con-tinuously with a set -down point at the 10 -inch (254 -mm) diameter. This allowedtwo advantages. An unworn portion ofthe 12 -inch (305 mm) disc at the outer-most diameter would be available forpossible future reference after the dischad completed the wear analysis. Startingat the 10 -inch (254 mm) diameter allowsabout 10 -minutes playing time for theduration of the record side, so a 10 -minute relaxation time is realizedbetween successive passes of any portionof the disc.

Another area of consideration regardingthe wearing process itself concerns themaintenance of the wearing styli. It wasfound that as the wearing stylus begins todevelop small flats on its contact faces.record wear actually decreases due to theincreased contact area with the groovewall. This was particularly true of theShibata tips. However, if the stylusbecomes excessively worn, the tip can nolonger navigate the groove properly andbegins chiseling and excessively wearingthe disc. It was determined that all stylishould be microscopically checked afterabout 100 plays. Conicals and ellipticals

18

are rejected when they develop obviousflats that are easily noticedmicroscopically.

Shibata styli are treated differently. Inaddition to the microscopic examination,they are fitted to a reference cartridge andturntable and checked using a special testrecord cut for this purpose. This disc(RCA #R L-1850 Test and TechnicalSeries) is a 30 -kHz signal recorded onboth groove walls using a constant cuttercurrent over the entire side of the recordending at about a 5 -inch (127 mm)diameter. The purpose of the test is todetermine the scanning loss of the stylusas its contact radius increases due towear. With a new Shibata tip operating atI -gram tracking force (J VC 4 M D -20Xpickup) both the left and right channelsshould show less than 3 -dB scanning lossfrom beginning to end of R L -I 850. Aworn Shibata tip is rejected when itdisplays greater than 6 -dB scanning losswith this test. At this point the wear isusually easily visible under a microscope.

Additionally, we have specified someother conditions on the actual disc -wearing procedure. The changers havebeen equipped with brushes which dustoff the stylus as the arm swings near thearmrest during the change cycle. Thechangers are operated with dust covers inplace to keep excessive dust from ac-cumulating on the disc during the wearprocess. However, the discs are notcleaned in any way before, during, orafter the wear process beforemeasurements are performed.

These requirements and procedures havebeen imposed to simulate as closely aspossible the field conditions the discs arelikely to encounter. We feel we haveavoided being unduly harsh or un-realistically pristine in arriving at thesewear conditions.

Evaluation of record quality

Ilaving discussed a good deal of historyleading up to our current procedure, wenow describe the actual process of ascer-taining the quality level of theQuadraDisc. There are four major testswe use to establish the quality of the disc:I) carrier level and its loss during wear, 2)spectral analysis of the noise content ofthe disc, 3) SEM photography of thehighly magnified grooves, and most im-portantly, 4) listening evaluations.

Carrier level is measured using a speciallyselected low -scanning -loss Shibata stylusfitted to a reference arm and turntable.The pickup is operated at I -gram track-ing force to further reduce scanning lossat the inner record diameters. The outputof the cartridge is fed without R IAA de -emphasis to a flat amplifier having 20 -dBgain and then to a 24-dB/octave high-pass filter with a turnover frequency of30 -kHz. This allows any portion of theQuadraDisc to be measured since thebaseband signal is filtered out leavingonly the constant -amplitude fm carrier.The filter output is fed directly to an acvacuum -tube voltmeter. This step is

calibrated using the first band on the "A"side of RCA #12-5-114 as a carrierreference. During testing, the referencestylus must be cleaned very frequentlywith a dry cotton swab to ensure accuratereadings. A number of compounds testedgenerate considerable residue left in thegroove after the wearing process. Thiscan cause gunking of the stylus within onerevolution of the disc, resulting in com-pletely false carrier level readings.Obviously, a compound less prone togunking with wear is desirable.

Real time spectral analysis of the hisslevel on the test disc is accomplished withthe aid of a modified CD -4 demodulatorfeeding a Hewlett-Packard 8054-Aanalyzer connected to an X- Y plottergiving a hard copy of the results. Thedemodulator has been equipped withswitches to deliver any of four signals to,the spectrum analyzer: I) baseband only,2) demodulated carrier only, 3)

demodulated carrier without ANRSnoise reduction, and 4) normal CD -4composite information.

The first two of these are normally used.Curves of the measuring system residualnoise, the test disc baseband noise, andthe demodulated carrier noise are plottedon one chart. Noise levels are relative toAmerican standard QuadraDiscrecorded levels of baseband and carriermodulation as represented by Band I onthe "A" side of RCA #I2-5-114.Measurements are taken at spirals on thedisc where no modulation is present.

The part of the record most subject towear is the innermost diameter. After thewearing process is complete and allmeasurements of the disc have beenobtained, a half -inch ( 12mm) circularsample is punched out of the last recordedband on the disc for scanning electron

microscope (SEM) analysis. SEMphotographs have proven quite helpful indeterming the best practical approach tosolving the single inventory QuadraDisccompound problem. More about thatlater.

With all of the foregoing investigation,we still haven't been able to eliminate theneed of actual listening tests in compoundevaluation. Defects like noisy swishes aredynamic in nature and hard to documentefficiently by spectral analysis. There aredifferent types of tics, pops, and clicks,characteristics of certain compounds asthey wear, but a tic -pop counter cannotdistinguish among them. Listenting testswere necessary at the very outset of thisprogram to determine where certainnoises were coming from on the first playof the disc. It is outside the scope of thispaper to discuss in detail here, but it wasdiscovered that most of the noise en-countered on the initial play of the disc isrelated to the metal parts and lacquerprocessing (matrix operations) ratherthan to the compound itself. It is ab-solutely imperative that stampers (astamper is the metal plate within the presswhich molds one side of the record) be ofsuperior quality before any judgment ofcompound quality can be made: TheQuadraDisc format is likely to show updeficiencies in record manufacturers'matrix facilities that they have beenunaware of in stereo record production.

After a few false starts, we arrived at threeaudio rating categories: I) hiss} noise, 2)sharp shattering breakup distortion, and3pics, pops, crackle, and spatter. Discsare auditioned on a high-qulaityreference playback system, and each ofthe three categories is rated by the follow-ing letter grades:

"A" - indicates superior performancewith no noticeable defect,

"B" - indicates a slight but noticeabledefect,

"C" - indicates a defect that is ob-jectionable, and

"D" - indicates still further degradation,resulting in rejectable performance.

Such a grading system is obviously highlysubjective, but several testers havebecome familiar with it and can providereasonably consistent results. We havehad occasion to test some compoundsseveral times and have been able to getrepeatable wear data from the listeningtest.

19

NEW DISCS

"FIRST -PLAY"LISTENING TEST

BAD

GOOD

SPECTRAL ANALYSIS

CARRIER MEASUREMENTWEARING CYCLES

IN FOLLOWING INCREMENTS

2g SHIBATA

25 PLAYS25 PLAYS25 PLAYS25 PLAYS

100 TOTAL

4 5g CONICAL

10 PLAYS15 PLAYS25 PLAYS25 PLAYS25 PLAYS

100 TOTAL

REJECTSTOP TEST

AUDIO FAILUREAT ANY INCREMENT

END OF 100 PLAYS

SPECTRAL ANALYSIS

SEM PHOTOGRAPHY

Fig. 1 - QuadraDisc wear testing procedure.

Wear testing procedure

Having now described each of the testsused in evaluating Quadra Discs, we willoutline the overall procedures (Fig. I). Anew test pressing is first auditioned todetermine if it is free enough of initialdefects to provide meaningful wear data.Usually one set of stampers of knowngood quality is used to press all wear testdiscs. A reference compound of known"good first -play quality is pressed fromtime -to -time to insure that the stampershave not deteriorated. The first -playquality of the new compound can then bematched against this standard com-pound. Audio ratings are noted for thefirst play of a new test disc. Carrier level ismeasured and documented at severaldiameters. Data are usually taken at thestart of the side, the end of the side, and atthe last two spirals. Spectral noise plotsare then made of the new discs at a spiralsomewhere around the 8 -in. (203mm)diameter. Usually two discs for each newcompound are tested. One goes on the 2 -gram Shibata wear tester for 25 plays.The other gets played 10 times on the 4.5 -gram conical consumer wear testchanger. Carrier levels are againmeasured and audio ratings given.

The first plays are usually the mosttelling, and many compounds have beenrejected at this point in the test. If thediscs have not incurred an audio rating

worse than "C" in any of the three areas ofnoise, breakup, or crackle, and thedemodulator has not automaticallymuted at any point of the audio test dueto excessively low carrier level, the discsgo back on the wear testers for 25 moreplays (Shibata) and 15 plays (4.5 -gramconical). At the end of this sequence theShibata-worn record has 50 plays and theother has 25 plays total. Carrier level andaudio tests are again made. If the discspass, they go on the wear testers again foranother 25 plays each.

This process continues until the discshave been worn by 100 plays. At this timeor when a disc has failed the audio test atsome earlier point by incurring a "D"audio rating, the spectral noise plots areagain made of the same portions original-ly tested, and the record is prepared forSEM photography.

Compounds that pass the 2 -gram Shibataand 4.5 -gram conical tests are also run onthe I.25 -gram elliptical and 2 -gram con-ical to make sure there are no surprisesdue to peculiar wearing patterns-and wehave been surprised on occasion. Severalcompounds have passed the 4.5 -gramconical test and failed the 2 -gram Shibatatest. This sounds unlikely at first, butanalysis of the SEM pictures indicates thecompounds trenched badly on both tests,but the Shibata stylus bridged the conicaltrench well enough the few times

necessary during measurement to yieldpassable results.

RCA's QuadraDisc compound

Based on our experience over the pastyear we feel we have identified the onlypractical type of compound whichsatisfies the single -inventory criteria. It isa somewhat softer formulation than wehad initially expected but with well-defined and controllable elastomericproperties.

The QuadraDisc compound wasdeveloped by studying the visco-elasticproperties of resins and various additivesused during compounding. As analyzedon suitable rheological measuring equip-ment, the shear rate -temperature -viscosity and melt -elasticity relationshipindicates a new range of properties. Acritical balance between melt viscosityand elasticity was built into this com-pound to have good moldingcharacteristics as well as improved wearproperties over stereo compounds. Newmixing techniques for the manufacture ofthis formulation had to be developed.The realization of the final properties ofthe RCA QuadraDisc compound is theresult of specific processing proceduresrequired to develop the full benefits ofthis heterogenous material. Therefore, amere listing of its ingredients is virtuallyuseless, but for those interested, theformulation is:

Primary resin 75.5-80.0%Secondary resin 20.8-16.3%Carbon black 0.2%Stabilizer 2.0%Antistatic agent 1.2%Modifier (EPO type) 0.5%Lubricant 0.4%

The general characteristics of a discpressed with this material are:

1) virtually no audible or visible deteriorationwhen played 100 times under ideal quadplayback conditions (Fig. 2a), and

2) very well -controlled, clean trenching of thesidcwalls with virtually no generation ofdebris due to abrasion and minimal "butter-ing over of the high -frequency modulation(Fig. 2b: compare with Fig. 3).

The radius of the trench created by the 0.6- 0.7 mil conical (or elliptical) stylus issmall compared with the 2 -3 mil verticalbearing radius of the Shibata tip and iseffectively bridged by the latter duringquad playback. Bridging this trench

20

allows the Shibata tip to ride on theunworn parts of the sidewalls and delivera sufficiently high carrier level to thedemodulator to produce acceptably quietquad performance. It is important,however, that the conically worn groovebe free of debris which could be trackedby the Shibata stylus resulting indegraded performance. The inclusion ofan antistatic agent in the formulation wasfound to be helpful in meeting thisrequirement (Fig. 4).

Notice now that we have not actuallydeveloped a compound which allows 100plays with a 5 -gram consumer -type con-ical stylus with little or no deterioration ofthe carrier modulation. We believe thatthis is virtually a physical impossibility. Itremains to be seen whether future pickupsusing conical tips of very small radiusspecially designed for QuadraDiscapplications (employing 30 -kHzmechanical resonances) will significantlychange this opinion. What we have doneis to devise a formulation which exhibitsvery well defined and controllable wear-ing of the carrier by taking advantage ofthe difference in the vertical bearing radiibetween the Shibata tip and all other tips.And second, the compound withstandsrepeated passes (100) of the Shibata tipwithout noticeable deterioration. This isnot a simple requirement to meet in itself,but it .is certainly easier than having tocope with the abrasion of conical andelliptical configurations which(disregarding mechanical resonances)produce between 2 and 4 times thecontact pressure respectively at the sametracking force as a Shibata tip.

Disc production

RCA recently made these findings knownto the chemical plants supplyingpolyvinyl resins and compounds to therecord industry. To date, both Keysor-Century and Tenneco Chemical havebeen able to duplicate the RCA formula-tion and processing, resulting in com-pounds that yield essentially the samecharacteristics.

In addition to the RCA formulation (Fig.2), the wear nature of other compounds isshown in Figs. 5 and 6. Both of thesecompounds are currently in use forregular QuadraDisc production by com-panies other than RCA. The number ofplays noted, if less than 100, is the point atwhich the disc failed the wear test.

/1/ /\\\\\\\\\

Fig. 2 - OuadraD.sc compound after 100 plays-a, leh) With a 2 -gram Shibata stylus-b, right) With 4.5 -gram conicalstylus. Notice clean, narrow trenching of the sideways.

Fig. 3 - Unsatisfactory formation after 25 plays with,* 5 -gram conical styk.s. Notice "bettered over naiu e ofwear pattern and deep wide trench leaving almost nourworn surface for Shibata tip to scan.

Fig. 4 - Cat.adraDisc compound without antistarc agentafter 100 plays with 2 -gram Shibata stylus Notic? deorisremaining n groove.

Fig. 5 - Compound C -a. lett)A1 ter 15 plays with 2 -gram Shibata stylus-b, right) After 10 plays with 4.5 -gram con calaltos.

.rig. 6 - Compound B-a, left) After 100 plays with 2 -gram Shibata stylus-b, right) After 25 plays with 4 5 gram conicalstylus.

21

Scopethe power testing paladinE. Cutshaw

SCOPE is an automatic test system that consolidates most of the previous standard andspecialized semiconductor testing requirements into a single high-speed systemcapable of categorizing and evaluating the millions of SSD power devices. Descriptions ofhow the system operates, what equipment is included, model test setups and the degree oftest system flexibility are given.

r-H IGH-SPEED

PAPER PUNCH1

TAPEREADER

TELETYPES z

HIGH-SPEEDPRINTER

r-- -VIDEOTERMINAL

IT- --IMINI -COMPUTER _ _WI DISK I

__J

TEST- SETINTERFACE

-10

A / DCONVERTER

PARAMETER TEST BOARDS

STAT IONMULTIPLEXER

TEST -SETLOGIC

0-

TESTSTATION

TESTSTATION

Fig. 1 - Block diagram of the SCOPE system.

SCOPE is an acronym taken from thewords semiconductor computer operatedparameter evaluation. It is a term used toidentify any one of several dozenautomatic test systems located in thevarious Solid State Division facilities.The SCOPE systems are the backbone ofthe RCA's power semiconductor testingcapability. Such systems are used tocategorize and evaluate the millions ofpower semiconductors manufactured

}earl} by RCA SSD. RF and Darlingtonpower transistors, bipolar transistors,rectifiers, SCR's, triacs, diacs, powerhybrids, and printed circuit modules en-compass the SSD manufacturing spec-trum. These devices may be examined bySCOPE systems as wafers, pellets and/ orin their final packaged form. SCOPEsystems are also used in engineeringfacilities for new product evaluation,competitive evaluation, failure analysis

and, new test -technique development.Each test system is tailored for itsparticular application without destroyingthe basic architecture. The SCOPE testsystems are capable of most standardparametric measurements and manyspecialized test procedures.

In-house systemdevelopment

Eery manufacturer, when producttesting is required, is perplexed by thedecision, make or buy. When a reputablecommercial piece of test equipment isavailable the obvious choice is to buy.Engineering and developmental costsalone will generally dictate this decision.The option, alas, is seldom this clear cut.This is the case with power semiconduc-tor automatic test equipment.

E. Cutshaw, Power Transistor Applications Engineering,Solid State Division, Somerville, N.J.. received the BSEEin1969 from Marquette University. He joined the RCASolid State Division in 1970 as a Thyristor ApplicationEngineer responsible for customer support and factorytesting. In 1971 he transferred to RCA Solid StateDivision. Mountaintop, Pa., as a Design Engineer inEquipment Development. His responsibilities includedcomputer controlled test set design in the areas ofthyristors and printed circuit fabrication. He transferredback to RCA Solid State Division, Somerville, N.J.. in1974 where he is presently responsible for advanced testequipment development and new testing techniques forpower semiconductors.

Reprint RE -21-5-18Final manuscript received October 18, 1975.

Small -signal transistors, rectifiers andintegrated circuits entertain a very largeproduction volume and generally speak-ing do not require as exotic test equip-ment as do power semiconductors. Thisallows commercial test equipmentmanufacturers to offer a good selection ofautomatic production testing systems to aprofitably sized market. With the impactof power semiconductors several yearsago, some of these commercial producersattempted to expand the generalcapabilities of their small signal systemsand exploit the power industry. In RCASSD their penetration was impotent forseveral reasons. Primarily the systemsoffered did not have the testingcapabilities of the semiautomatic test setswhich were produced in-house beforecommercial equipment manufacturershad even begun to consider such test sets.Also the technical expertise of product -line design and applications personnellinked with the experience of a matureequipment department is difficult com-petition for even the most aggressivecommercial producers. The advent of theminicomputer coupled with a readyengineering community lead the way tothe SCOPE test system. Today RCA stillholds leadership in high-speed power -semiconductor automatic testing.

How SCOPE works

Very well; as a matter of fact, the basicsystem architecture has not been changedsince day one (Fig. I is a block diagram ofthe SCOPE system). The heart ofSCOPE, or rather the brain, is a Hewlett-Packard minicomputer. Each systemcontains 8-K or I 6-K words of corememory, in some cases a disk is added.Paper tape is the primary programmingmedium with teletypes, high-speedprinters, video terminals, etc. asperipherals. Several models of the H/ Pminicomputer are in use as a result ofsystem evolution, however, all test sethardware and software are compatiblefrom set to set. The H P computer waschosen for its 1/O capability, model tomodel compatibility, vendor support, in-terest and availability.

The test set interfaces with the computerin the same way as all other peripherals. Apriority interrupt 1/O bus is the primaryintercommunication function in thesystem. The computer communicateswith the test set by way of the test setlogic. The test program is transferred to

the test set logic in the form of 16 -bit testwords which are decoded and distributedto the test boards to establish test con-ditions and circuit limits. The test setlogic interrupts the computer to requesttest words and to signal that informationis ready for transmission by the analog -to -digital converter. Each test board isdesigned to set conditions and makemeasurements for a particular type ofparameter (leakage, forward drops,betas, etc.). The test set logic establishestiming and controls the station multiplex-er. Each test station can request thatdevices be tested according to a testprogram which it can remotely select.Test stations may take the form ofmanual test sockets, wafer probers, pellethandlers, device handlers and hotsockets.

The SCOPE system software consists ofan executive program which maintainsoverall control and interfaces with the

2NXYLNPNV CEO.1 100V C BO

.1 1000VC ER.1 100 100VCES

1 100V C EX

1 100P I B5 1 1

PV BE

5 1

PV SAT5 .2PV BE S5 .2V BE05

VBCO1 1

V EBO 01 100PVC EO

1 100END

Fig. 2 (above) - Sampleprogram. Fig. 3 (right) -Sample data logging output.

operators in a simple Engligh languageconversation. Additional software in theform of test programs and user programsare entered and controlled by the ex-ecutive program. Fig. 2 is a sampleprogram which can be loaded into apower transistor test system. The firstentry is the program name, the secondidentifies whether the devices are p -n -p orn -p -n. The remaining entries are testnames followed by testing conditionswhich are generated from a schedule oftests for each test system. A sample datalogging output for this program is shownin Fig. 3.

The executive software allows up to tentest programs to be active in the testsystem. Each test program can test andcategorize up to 64 different types orcategories for each device tested. Theexecutive program allows softwarecounters to be opened up for each of the

1 ESTEL) 10

2r -a

:

se:z

i=1 1- 1-i

VCbri VC160 VCER VT:ti E: N b fVbE1.8100161 LHOMN 110018 1.88118 180MH 7 OH

1.141.1W 13" 1 ev18R

IJNI 1 V MN

1 68.4 93 81.5 81.6 83 5 193 1.082 72 1 1134 86 0 86 8 93 0 192 1.233 70 3 99 83 1 83 2 97 6 193 1 274 66 6 76 66.6 66 6 66 7 195 1.205 43 9 61 52 9 52 9 52.9 191 1.116 8 0 70 61 0 61 1 61 157 2 077 73 5 184 85 9 85 9 93 9 192 1.118 59 5 72 67 2 67 3 67 4 193 1 049 37 3 46 37 3 37 3 37 3 192 1 15

10 69 6 97 82 3 82 3 95 3 192 1.0811 37 4 47 37 4 37 4 37 4 193 1.0812 38 3 58 49 6 52 2 51.9 192 1 1213 45 3 53 45.5 45 5 45 5 191 1.0814 51 8 64 56 1 56 1 56 1 191 I. 1315 71 9 100 85 5 85 5 98 9 192 1.1016 59 5 68 613 8 60 8 60 7 192 1.8917 68 5 103 82 9 82 9 91 1 176 1.2018 35 1 42 35 3 35 3 35 3 192 1 8319 34 4 45 36 5 36 6 36 5 192 1 1170 68 8 97 79 7 79 8 54 9 198 1 1/

VBCO vE130 PVCE0 PVECE$ VbE011301 100114 /801414 5 ON 5 041 5 13914

MU'S .-dNfl defl

UNIT NV V

I. 703 7 9 41 0 .89 1.09 1.272 747 9 1 46 9 .00 1.24 1.393 703 7 5 88 8 91 1. 29 1. 434 751 7 J. 66 4 92 1.21 1. 355 700 8 3 25 8 89 1.12 1.296 698 19 0 39 8 63 1. 11 1. 317 697 7 2 92 5 88 1.12 1.318 696 7 6 64 6 90 1.05 1.249 760 7 9 33 9 91 1.16 1.29

10 6913 8 2 75 2 88 1.09 1.2611 752 8 6 35 9 91. 1.09 1 2612 697 8 1 29 4 89 I. 13 1 3013 690 7 1 40 8 87 1.09 1.2414 696 7 8 47 2 88 1 14 1. 2815 702 7 7 48 1 88 1 11 1.2716 694 7 1 54 1 89 1 10 1.2517 761 9 9 40 8 72 1 23 1.3518 691 7 0 31 7 89 1 04 1.2219 694 9 7 30 7 89 1 12 1. 3229 703 7 1 80 94 1.12 1. 31

23

64 types on each test program and allowsfor quantitative data to be gathered foreach type. While product is being tested,it is generally possible for a particulardevice to meet the requirements of severalcategories, therefore, each test programallows for up to 64 priorities to beestablished. If a required number ofdevices has been generated to a specifictype, the operator can simply load a newset of priorities to alter the categorizationof product being tested.

While the test system is operating,another form of software, called a userprogram, can be entered and exercised.User programs are FORTRAN programswhich can access the data being taken bythe test set. User programs can take datawhile product is being tested and presentit in the form prescribed by the operator.It may be output in the form of raw data,statistical summaries or histogram plots.User programs are also generated forcalibration and maintenance operations.

The production software operates in amode referred to as "by -type": eachdevice to be categorized is tested first tothe conditions of the highest priority typeuntil it fails to meet a limit. At this time,further testing is aborted for that priorityand the next priority type becomes thecriteria. If a test has already been per-formed for a higher priority, its data willbe retained rather than retesting in subse-quent lower priorities. Testing continuesin this manner until all requirements of aspecific type are met.

Testing occurs at the device -under -teststation on wafers, pellets, and finalpackages. Testing is performed eithermanually or with some type of automatichandler. In the wafer testing area,automatic probers equipped with a smallplatform to hold the wafer under test anda group of fine test probes incrementfrom pellet to pellet, and by means ofprecision inkers, identify and categorizegood and bad pellets. Each wafer canthen be data analyzed for purposes ofprocess feedback, final test yield projec-tion, and categorization for expecteddistribution requirements in the assemblyareas. As a result, only wafers best suitedto the current marketing requirementsare immediately processed.

After pellet separation, devices can behand probed manually or tested with apellet handler which can automaticallysort and bin each pellet. Wafer and pellet

testing is somewhat limited as to the typeof tests and the conditions of tests whichmay be performed because of the minimalheat sinking ability of an unmountedpellet under high power testing and thefact that some tests are designed todetermine the integrity of the pellet insidethe final package. However, byeliminating bad pellets at this stage, nofurther wasted effort or assembly cost isexpended on unusable product.

The final product testing areas takegreatest advantage of computerizedtesting. Power semiconductors in theirfinal packaged form take on many con-figurations which leads to a complex ofmechanical adaptations. Manuallyoperated computer stations make use of auniversal six -pin -socket receiver which isflush mounted in the counter top. Sixcontacts are required to each deviceunder test to accomodate Kelvin sensing.Each terminal of the device under test hastwo connections, one to carry the forcingfunction and the other as a non -current -carrying measurement contact. In-terchangeable sockets for each packageconfiguration can be plugged into thecounter -top receiver. This socket flex-ibility allows for station versatility and nosocket repair downtime. Just behind eachsocket receiver at each station is a smallcontrol console which contains a switchused to select the desired test programand a light emitting diode display whichidentifies the type number of each devicetested so that it can be appropriatelybinned. A series of lights also indicatesthe testing condition of each individualstation. In addition, manual sockets withheating elements and temperature con-trollers can be employed to provide high -temperature testing. The concept of high -temperature testing without heatingelements in which the device under test isheated internally through high powergeneration in the pellet is being developedfor factory testing. This test methodallows standard room -temperaturetesting of the component followed by self -heating to a prescribed temperaturefollowed by more parametric tests, etc.

Automatic mechanical handlers are usedfor high -volume final testing and typesorting. Several automatic handlers and aSCOPE test system provide a goodmarriage requiring only minimal atten-tion, primarily in the loading and un-loading of product bins. Hightemperature automatic handlers con-taining a handling mechanism inside of a

temperature chamber are also incor-porated.

Claims to fame

The first SCOPE testing system wasproduced in 1969. Since that time im-provements have occurred at an in-creasing rate as a resulting awareness ofthe system's versatility developed. Im-proved testing techniques and the cons-tant addition of more and diverse types ofmeasurements have been incorporated.Among the attributes of the family ofSCOPE test systems can be included thefollowing:

A system concept centering on the test board,which is the key to adaptability for factoryrequirements. Commercial systems are ex-citation and instrument oriented makinghardware modifications inflexible and ex-pensive and limiting the systems to theiroriginal power and sensitivity specifications.

Software operation under an interrupt struc-ture without which on-line data reduction ina factory environment for engineering pur-poses would be impractical.

A FORTRAN based compiler which allowsuser programming without specificknowledge of how the system actually per-forms the testing.

The selection of a high speed analog to digitalconverter as the data gathering mechanism.The traditional time consuming approach ofcommercial test equipment manufacturersfor data logging is successive go -no-goiterative evaluation.

Simple English language operating com-mands.

"By -type" testing to minimize test time andmaximize throughput by performing onlythe least number of tests necessary to classifyto the highest priority by use of automaticsoftware branching.

Ability to apply several hundred tests perprogram to a device under test.

Test programs that can sort up to 64 differenttype categories; these types can be arrangedinto any priority order by TTY command.

Software counters which can be opened,closed, purged and read under TTY com-mand for every type and on each testprogram.

Up to 32 test stations which can be multiplex-ed to one test console.

Photocell controlled manual test sockets forincreased throughput by elimination ofpushbuttons while still maintainingoperator safety.

Full Kelvin sensing. Typical test time for a single parameter is 10

microseconds.

2000 -volt leakage and breakdown testingcapability.

200 -ampere forward -current testing capabili-ty.

24

Below: this Is a TO -5 package automatic handler, also located Ina final testing area.

Above wafers are aligned on a novable X -Y tape by theoperator The prober then automatically Indexes from peltetto pellet testing and categcvizing the wafer. Below:automatic pellet handlers like this one are used in thethynstor product line for the testing and sorting of peletsprior to mechanical assembly.

Above (tcp): to the A lei twinge testing area automatichandlers ars connected to SCOPE test systems for hghvolume testig. Above (second from top) manual test stationsare uses to ow -volume selec.ions or where handlers are notyet available,. Directly above' automatic vacuum operatedmulti-p-obe sockets sucn as this are used for printed :irc uittesting or SCOPE systems

Above: compiler and peripherals being used for special solwsre developmentto be inco posted In CLiallty Ccntrol console under construction.

Below: rower -transistor test csole in final stages of checkprior le installation on factfloor

COMPONENTCOMPONENT MOUNTING PAD

COMPONENT LEAD HOLE

PEAS FREE OF SOLDER RESIST

CIRCUIT COPPER RUN

CIRCUIT TEST PAO

Fig. 4 - Configuration of component mounting holesand test pads.

High-speed automatic non-destructive 1S -Btesting capability.

Automatic ramped high voltage leakagemeasurement technique which protects evenlow -voltage product from breakover voltagedamage.

Room and elevated temperature single inser-tion automatic testing by device under testself heating.

Variations on a theme

Aside from discrete power semiconduc-tor testing, several SCOPE systems havebeen modified for multi -pin moduletesting. This includes power hybrids andprinted circuit modules. The testing ofthis type of product requires parametricmeasurements on passive components aswell as power semiconductors, with amuch more complex test -set to device -under -test interface problem.

The test set interfaces with the assembledmodule at up to 31 test points in thecircuit. Since it is advantageous to be ableto look at these electrical points in thecircuit in a multitude of combinations,the multiplexer makes all points availableto all test boards. Also since it is helpful inmodule repair and diagnosis to performtwo- and three -terminal testing, a 3- by31 -pin testing matrix is used to enableeach test board to access any nodedirectly through the multiplexer or in-directly through the matrix. The testprobes are part of a vacuum -operatedmechanical prober which can simul-taneously probe all circuit nodes on thecopper side of the printed circuit board. Anumber of steps are taken to ensurereliable operation of this method.

First, a suitable target is provided foreach test probe. To avoid probe deflec-tion and poor contact to the circuitboard, separate test pads are provided foreach circuit node, as shown in Fig. 4. Thetest pads consist of a blank copper padcoated with solder, as shown in Fig. 5.This type of test pad, which can be easilyintegrated into the printed circuit boardlayout with essentially no increase in

COMPONENT LEAD

COMPONENTMOUNTING PAD

SOLDER CIRCUIT TEST PAD

TEST PROSE

GASKET

PRINTED CIRCUITROARS

RAF

VACUUMCHAMBER

-cr

SUSPENDED PLATFORM

SPRING -LOADED'TEST PROSES

D=D3F)

Fig. 5 (above left) - Arrangement of component, test pads, and test probe. Fig. 6 (above right) - Circuit board and testsystem.

overall cost, provides a reliable circuitcontact and a considerable reduction inthe necessity for retesting modules re-jected for misprobe. The finely pointedand hardened probes that mate with thissoft target are spring loaded andsuspended in a platform which moves upto make contact with the printed circuitboard when vacuum is applied, Fig. 6.

In laying out the test pads, care is given tothe maximum deviation of the test probefrom its center, printed circuit boardinaccuracy, and module alignment withthe test fixture. In general, test pads aremade as large as possible and located soas to minimize bridging to adjacentcopper. Solder resist is used primarily inareas most susceptible to bridging, and isenlarged as much as possible aroundcomponent mounting pads and test pads.Once a successful contact is made to thecircuit board by all test probes, moduletesting may begin. Some additional ad-vances as a result of module testinginclude:

Adaptation of a basic three -lead devicetesting system to a multi -pin circuit moduletester.

Capacitance measurement using a singlepulse as opposed to cumbersome and slowac techniques.

Development of an electronic rolling techni-que to enable components connected tomultiple circuit nodes to be measured in-dependently.

Multiple digital signal output conversion tosingle analog reading for improved se-quential logic testing speed.

Is this utopia?

Power semiconductor testing is emergingfrom the dark ages. In the past, primarilybecause of the high -power levels andcritical timing requirements, most powertransistor tests were performed on uniquepieces of test equipment requiringseparate insertions for each parameter.The SCOPE computer controlled testingconcept consolidated most of the stan-dard and many of the specialized testingrequirements into a high-speed, non-

destructive, single -insertion means ofcategorizing power semiconductors. Italso enabled the use of mass data ac-cumulation and manipulation for im-proved product visibility.

The types of tests remaining which muststill be performed with separate in-sertions are predominately long durationtests or those impractical for multiplexedtesting techniques. The next test -set

generation should minimize the majorityof these difficulties.

The microprocessor is the next genera-tion of test technology. With it thepresent minicomputer that controls ahardware -multiplexed test set can bereplaced with intelligent test sets whichmultiplex through software to a

minicomputer. The minicomputer role is,then, to supervise rather than operate thetest stations. Intelligent stations can nowbe tailored to test a more narrow range ofproduct with a wider test capability. Asingle minicomputer will be able tooversee the operation of a large numberof intelligent stations and will be freed foruse in feeding back information toprocess control and marketing functions.

Test stations can now act as stand alonetesting functions with calculating anddecision making abilities. These testingstations may be grouped to an automatichandling mechanism with multipletesting locations for increasedthroughput. The possibilities are endlessand the future holds a challenge for thedesigner who keeps abreast of the state ofthe art .

Acknowledgment

The development of a family of testingsystems of this nature is the product ofmany minds and hands. Particularacknowledgment goes to Hal Ronan forhis multitude of technical breakthroughs,and to Vince Grobe for providing theenvironment; and most importantly to AlParsons, the originator and pioneer of theSCOPE test system.

26

Impressions of technology in ChinaDr. J. Hillier

ditur s Note Dr. Hillier, Executive Vice President for RCAResearch and Engineering, completed an 11 -day visit to thePeople's Republic of China (PRC) in July 1975. He was a memter ofa 10-mai delegation of Electronic Industries Associatior ex-ecutives touring electronics facilities in that country. In this article.Dr. Hillier shares with our readers brief excerpts from portions ofhis daily diary relating to the PRC technical centers visitec. Dr.Hillier's observations and impressions of Chinese technology.including a few suggestions on doing business with the PRC areincluded herein. For information on Dr. Hillier's observations on thesocial aspects in the PRC. the readers are referred to the article"China Diary" in the Oct/Nov 1975 issue of RCA Communicate.

Dr. James MOW, Executive Vice President, Research and Engineering, RCA, Pnnceton,N.J. studied at the University of Toronto. where he received a BA in Mathematics andPhysics in 1937. MA in Physics in 1938, and PhD in Physics in 1941. Between 1937 and1940, while Dr. Hillier was a research assistant at the University of Toronto, he and acolleague, Albert Prebus, designed and built the first successful high -resolution electronmicroscope in the Western Hemisphere. Following this achievement. Dr. Hillier joinedRCA in 1940 as a research physicist at Camden. N.J. Working with a group under thedirection of Dr. V.K. Zworykon, Dr. Hillier designed the first commercial electronmicroscope to be made available in the United States. In 1953, he was appointed Directorof the Research Department of Melpar, Inc. returning :o RCA a year later to becomeAdministrative Engineer, Research and Engineering. In 1955. he was appointed ChiefEngineer. RCA Industrial Electronic Products. In 1957. he returned to RCA Laboratoriesas General Manager and a year later was elected Vice President. He was Vice President.RCA Research and Engineering, in 1968, and in Janua-y 1969 he was appointed to hispresent position. Dr. Hillier has written more than 100 technical papers and has beenissued 40 U.S. patents. He is a Fellow of the American Physical Society, the AAAS. theIEEE. and Eminent Member of Eta Kappa Nu, a past president of the Electron MicroscopeSociety of America, and a member of Sigma Xi. He served on the Governing Board of theAmerican Institute of Physics during 1964. 65. He has served on the New Jersey l-igherEducation Committee and as Chairman of the Advisory CounCil of the Department ofElectrical Engineering of Princeton University. Dr. Hillier was a member of the CommerceTechnical Advisory Board of the U.S. Department of Commerce for five years He waselected a member of the National Academy of Engineering in 1967.

ONE DOES NOT VISIT the People's Republic of China(PRC) without an invitation horn the Chinese; this was truein our case. However, the initiative for obtaining such aninvitation was taken by the Communications Division of theElectronic Industries Association.

Our Chinese host group was the Machinery Import Export(MACHIMPEX) Corporation, one of several state cor-porations responsible for all trade between the People'sRepublic and the rest of the world. More specifically, ImportDept. 4 of Machimpex (w1ich handles all imports ofelectrical and electronic equipment-and instruments) wasour principal point of contact; this Department scheduledour visits while in China.

During our 11 -day visit to the PRC, we observed manydifferent kinds of facilities. Formal visits included aninstitute of technology, a semiconductor research institute,the Central Meteorological Observatory in Peking, theShanghai Permanent Indt.strial Exhibition, three elec-tronics manufacturing plants making radio and televisiontransmitters, teletype equipment, telephone carrier equip-ment, desk calculators and small and medium computers.

PRC economy and trade policies

Although the PPIC appears to be self-sufficient with regardto growing enough food for its population in good cropyears- continuance of an adequate food supply for agrowing population becomes the central critical theme ofthe PRC policies; thus, agricultural self-sufficiency isaccorded the highest priority. Accordingly, industry isdirected to the improvement of agriculture through theproduction of machinery and chemicals.

Foreign trade includes the acquisition of critical rawmaterials for production operations and the purchase offood (if needed). The policy is to buy only those items thatare critically needed to continue the development of theChinese economy and those materials and devices requiredin advance to enable the "start-up" of PRC development andproduction which can subsequently become self sufficient.Generally, such purclases fall into four main fields: com-munications, transportation, electric power and produc-tion.

Research, development and production

The research we witnessed could be described moreaccurately as development, involving the duplication ofknown accomplishments of Western research. Suchresearch or development is apparently executed as aneducational exercise to learn how to diffuse the knowledge

Reprint RE -2'-5-21. Final manuscript received November 5. 1975.

The author near one portion of the 4000 -mile Great Wall.

27

In a figurine factory in Wu Xi, a girl is painting one part of a small figurine.

and introduce it into production.

The diffusion of knowledge is an integral part of the PRCpolicy of self-reliance. In theory, there is a built-in com-pounding that ultimately could have the effect of rapidlyaccelerating Chinese industrialization.

Concerning the status of PRC technology and production,my observations and impressions must be consideredpreliminary and tentative. Nevertheless, some observationsseem to be consistent-the overwhelming dependence onhand -work, hand -fitting, the absence of intermediate in-spection steps, the absence of jigs and fixtures, the absenceof mechanical and visual aids, and the mystifying work -flowpatterns.

Many integrated circuits are incorporated in productionand development equipment. Such IC's are all made inChina, apparently produced in quite large quantities. Thesolid-state devices produced are mainly transistor -transistor logic and P -type metal oxide semiconductor (P-MOS)-and relatively simple circuits of one to few gates. Itrequires about 100 such devices to make a simple 12 -digitfour -function desk calculator.

We observed the demonstration of several computers usingintegrated circuits; the comptuers were all scientific typemachines. The largest computer was capable of a millionoperations per second with 2 microsecond access time to a128-k core memory with forty -eight -bit words. The morepopular computer provided 110,000 operations/s, a 32-kmemory and used 48 -bit words. In the areas of research anddevelopment, we also saw integrated circuit studiesprimarily directed to developing the technology of large-scale integration. A computer -controlled machine was usedto make masks for large-scale integrated circuits; this wasthe most advanced setup shown.

We visited several electronics factories and gained a strongimpression that they were overreaching their capabilities intrying to put high technology into production. This raisedthe question as to whether the approach was deliberate, in

order to force attention on the problems, or simply the resultof an unawareness of the infrastructure necesary to keepboth the throughput and the complexity on positive trends.

Daily -diary excerpts

Perhaps it will be instructive to the RCA ENGINEER readersin gaining their own impressions, to share with them daily -diary excerpts from some of our visits to certain points ofinterest-selecting only those having some relationship totechnical topics:

First meeting at MACHIMPEX Transmitter plant Qinghua University Seminar at MACHIMPEX Semiconductor Research Institute Central Meteorological Observatory Shanghai Permanent Industrial Exhibition Telephone Carrier Equipment Plant Computer Manufacturing Plant

First meeting at MACHIMPEX

At MACHIMPEX (about 20 minutes from our hotel) there wasvery little delay in ushering us into a large conference roomfurnished by a number of stuffed chairs and sofas that surroundedthe room. Shortly after we were seated, Cheng Chi-hsien, DeputyManaging Director of MACHIMPEX, entered with a retinueconsisting of several of the people who had greeted us at theairport. We were handed a duplicate list of the people in themeeting room. My copy promptly faded but I succeeded inrecording most of the names. In addition to Mr. Cheng there wereseven or more managers and interpreters. A Mr. Han was listed as"Staff Interpreter." He played that role throughout our visit.Later he visited the United States as a member of a tradedelegation from China. Then he was listed as Deputy Manager,Instruments and Telecommunications, confirming our suspicionsthat he had greater responsibilities than had been indicated to us.

The meeting opened with greetings, welcomes and small talkregarding how we traveled. Chinese tea was served continuously.Cheng (Deputy Manager) announced that he wanted us to see asmuch as possible in China in order to begin to understand them.

In response to a question, we were given a rundown of thefunction of their Import Departments. The head office ofMACHIMPEX in Peking is relatively small, about 500 people.The policy of operation was described to us in part. Two or threekey points were new to me. MACHIMPEX deals with exportsand imports only hut is only one of two organizations that importmachinery and instruments. The other organization was notidentified until several days later. The import activity is in reality apurchasing department for the state. The originator of arequirement is the "end user"; requirements are submitted toMACHIMPEX only after the state has approved the purchase.

However, MACH! M PEX can and does work closely with the end

28

user to advise what is available and to assist in making a selectionthat will meet the needs. When the end user is not sure how tosatisfy a requirement, "friends" from abroad are invited to "clarifyspecifications". As a result, there are large numbers of invitedvisits (100-200 per day) of which 50% are concerned with technicalservices while the other 50% are concerned with technicalspecifications and discussions. A special point was maderegarding the existence of "friends" in many countries, includingthose where the governmental relations with China had "artificialbarriers". The import and export departments of MACHIMPEXseemed to have little communication with each other.

Following this introduction, we were handed copies of anitinerary (it did not fade). In response to questions, we indicatedthe desire to see specific installations mostly in the telephoneswitching and transmitting equipment area. Notes were made butthere was no commitment. We made known our disappointmentregarding large amounts of sight-seeing relative to a few technicalvisits. Nevertheless, while the basic itinerary was followed, thetechnical visits desired by the majority of our group were inserted.

The Peking broadcast transmitter plant

We went to an old looking complex of buildings in North Peking.After passing a guard house, we went into an open area that had alarge statue of Chairman Mao plus a large red billboarddisplaying one of his quotations. We were ushered into a drabsecond -floor conference room with a long conventional table andtwo rows of chairs. We were introduced to a Mr. Hsu who wasidentified as the Deputy Chairman of the Revolutionary Com-mittee of the plant. He quickly introduced a number of people,mostly engineering managers. Mr. Hsu gave us a brief story of theplant which has a thirty-year history. Before liberation, it hadabout 100 people who assembled radios from imported com-ponents. After liberation, development had been rapid, nowrequiring 4,000 workers to fulfill production targets.

Since the beginning of liberation, low -power radio transmitterswere produced. Later, the product grew bigger and qualityimproved. Now, transmitters of many sizes are produced formany applications. Furthermore, all components are made inChina. The highest power is a 700 -kW shortwave transmitter! Thefactory produces several transmitters for hf and ssb telecom-munications, a.m./ f.m. transmitters, and tv transmitters in 1, 10,and 40 kW sizes (all less than 300 MHz).

The organization was described as consisting of a Chairman andVice Chairman of the Revolutionary Committee. The committeehad 25 members and included representation from all levelselected by the staff, including workers. Below the RevolutionaryCommittee there appeared to be two levels of management,designated as "group" and "workshop." The organization in-cluded 400 technical people.

We were next ushered into a large factory floor that reminded meof Camden of thirty years ago. The space was partitioned downthe middle by a row of steel cabinets with open empty shelves overwhich had been draped a semitransparent light blue clothmaterial. We were not taken behind the partition. In front of thedraped cabinets were rows of simple tables covering an area ofabout 50 x 200 feet. The tables (work benches) were covered with

Preliminary to Chinese Banquet: Glen Solomon. head of EIA delegation (with camera) isshown with Mr. Hsu (at his left) and Mr. Cb 3 ng (on his right( plus several key managersfrom MACHIMPEX.

very clean paper and were brighlty lit. About 150 men and womenworkers were at the tables assembling transmitter sub-assemblies. An amazing variety of tasks was being performedfrom inserting and soldering components in PC boards-to thefinal assembly of entire units that obviously fitted transmitterracks. There seemed to be no rhyme or reason to the work flow.

While one-sided printed circuts were used, the density ofcomponents was low and hand -soldering of components seemedto be standard. There were a number of strange unidentifiablecomponents; others were cleaned up versions of old technology.One was a three -gang variometer made with windings of whatlooked like silver-plated wire wound on white ceramic rotors. Theworkmanship looked good. There were a number of discretetransistors on the boards we saw.

Next we moved past the working group into an area where anumber of transmitters were set up for testing. We were shown a40kW tv transmitter and a 400 kW high -frequency ssb telecom-munications transmitter. Few of the transmitters had the finalamplifier tubes in them. Only one, a 400 -kW transmitter, wasbeing tested but not at anything near full power. The metersindicated about 4 -kW. We were told that water vapor cooling wasused for the final amplifier tubes; the high -power hf transmitterused "line tuning". This area had none of the normal work clutterthat one would expect to see in a test operation.

Following this, we were taken to another part of the plant througha machine shop where many fittings were being made on oldsimple lathes and milling machines. We were taken into a roomthat was no more than 25 X 25 feet. This was drab and dimly lit. Ithad two rows of work tables at which about 15 men and womenwere doing various tasks. Again, there was no pattern to the tasks.One pair of workers was trimming excess gasket material fromaround a window used in the front of a transmitter cabinet.Another was cleaning glass covers and mounting them onpressure scats. Still another was assembling some type ofthermostatic device in a small thermal vacuum bottle. In themiddle of this, a pair was threading small holes in some aluminum

29

cooling vane assemblies and deburring the fins with a hand file.All work pieces in the room were on the tables. There was noindication of any work flow. I he room was completely devoid ofany sign of work clutter.

It was obvious that everything had been staged for our benefit,except the test area. At first, 1 was confused and annoyed.However, after several visits my reaction become much morecharitable-and 1 now believe what we saw was an honestrepresentation of the operation in the plant. However, for reasonsof pride and efficiency, they had chosen to set up thedemonstrations so we could see them quickly without wanderingall over a 4,000 -man plant, scattered in many small and oldbuildings.

When we returned to the conference room for the concludingsession, there were a number of questions from both sides. TheChinese were anxious to know the size of the highest powertransmitter made in the United States. They were informed thatthe companies represented in our group do not manufacture hftransmitters for telecommunications purposes. We also explainedthat the trend was to achieve high power by combining smallerunits to achieve economy and because it provided redundancyand continuity of service at lower power when one unit failed.This was not fully understood. We learned that tv studioequipment and tv receivers are made in other and separate plantsin Peking but were given no details. The plant we visited wasreported to be the largest of three or four similar plants in China.

After the visit to Shanghai where I had a working shortwave radioin my room, it occurred to me that the very -high -power hftransmitters were made for jamming purposes. While the Voice ofAmerica, the BBC and Australian stations were not jammed (fewpeople in China understand English), a large number of stationsin the shortwave bands were being jammed with extremely strongsignals.

Visit to Qinghua University

Qinghua University is a large installation having several largebuildings in northern outskirts of Peking. We entered a verylarge room containing two long rows of overstuffed chairs andsofas facing each other. We were introduced to a Mr. Ma who wasidentified as the "Responsible Member"; we presumed him to bethe member of the Revolutionary Committee who had been giventhe responsibilities of managing the University.

Mr. Ma introduced Professors Wu and Lee of the Department ofElectronics, a Mr. Doc of the Revolutionary Committee, TeacherLee of the Industrial Automation Department (female) andStudent Fu (also female). The University was founded in 1911 andwas a school of science and technology. It included, amongothers, departments of Mechanical Engineering, IndustrialAutomation, Electronics, Electric Power and Precision In-struments. There are I 1 departments in all with 54 specialties. Thefaculty consisted of 3,000 teachers and 8,000 students. Theenrollment increases to 11,000 in the fall.

Educational system in China

At this point Mr. Ma launched into an ideological description of

the educational system in China and at this University inparticular. This was the most lucid and complete description wereceived. There were many further allusions to this system in thecourse of our trip. Quoting from Mr. Ma:

Before the Cultural Revolution the aim of the education system inChina was, like in the Soviet Union, not to serve the workers butto train people to be officials "and to be rich". The students, alsolike the Soviet Union, did not like to go to the factories or thefarms for fear of retrogressing. Now, there has been aneducational revolution in which the educational system is tryingto serve all the people. The educational system must open the doorfor the whole society to be able to cultivate proletariat successors.Successors must rely on workers and peasants as well as teachers.

They have "improved" enrollment and teaching. The studentscome not from middle school only but from peasants, workersand soldiers, and all have work experience. There is a process bywhich peasants, workers and soldiers participate in the selectionof students for the University. The curriculum must combinetheory and practice. It involves three years of study and isestablished to be uniform all over China. There are now fourstandards that must be met before a student is accepted:

I) The student must be "oriented" to serving the people.

2) Must have two years of work experience.

3) Must have a middle school education.

4) Must be healthy.

Before the Cultural Revolution, the University did not payattention to service to the workers and therefore tended todevelop a new bourgeoisie. During their studies, students now goto farms where they work in the fields. An organization wasestablished in the University that combines teaching, science,research and production. Twenty-five small factories are part ofthe University. They produce fifty products. The University alsohas 80 research projects in which the students participate as theyacquire scientific book knowledge. [Mr. Ma used as an examplethe development of a step -and -repeat camera and severalcomputers.] The students with this knowledge are expected to gointo work situations where they will diffuse their knowledge asrapidly as possible.

Examination procedures have also been restructured. The ex-amination questions are derived from current research andproduction problems rather than from texts. The examinationsare open book and open discussion sessions. The student is ratedon the basis of his analysis of the problem.

The University has other operating modes in place. There is abranch that provides practical training. There is also a spare timeschool attended by factory workers of Peking. Students will also"do research" for local plants. Some become teachers in thecountryside. He pointed out that one basic concept in this systemis to make sure that there is no difference in status of mental andmanual work (In several subsequent visits I began to believe thatthe last sentence might be a simplistic translation. It was clear thattechnical competence is valued and results in relatively highsalaries-even more than many general managers. On the otherhand there seems to be a real effort to prevent the formation of anintelligentsia or any other elitist group).

30

Semiconductor research

On the tour we visited the semiconductor research actiitiesdedicated to large scale integration. We first went into a smallroom where four or five girls were bonding IC's (with aluminumwire). Here, there was some staging; each girl seemed to have onlyone or two circuits and they were all different. There was a 128 -stage shift register, a circuit that was claimed to have 1000transistors. However, we were not shown where or how they wereproduced. The circuits shown were all P-MOS. In response to aquestion we were told that Qinghua was also working with TT L,12L, and DTL. However, much to my surprise none of the groupappeared to have heard of charge -coupled devices (CCD).

We were shown a homemade ion implantation machine pur-ported to be 100 kV. We could hardly believe this; the insulationwas not sufficient and moreover the voltmeter had a full scalereading of 30 kV.

Machine shop

From the semiconductor area we went to a machine shopdescribed as one of the small factories. In was an area of about10,000 square feet in which there were many old-fashionedmanually operated lathes and milling machines. There were alsoseveral types of gear -hogging machines. This was obviously anoperating shop. We were told that parts of milling machines werebeing made there. From the work pieces and new parts lyingaround I believe it would be more complete and accurate to sa)that they were refurbishing the spindle heads of conventionalmilling machines. There were large numbers of obviously usedheads piled around. The production parts appearedprimarily stepped spindles and several types of gears. In the backof this shop we were shown two milling machines, one large, onesmall, that had electronic consoles; these were described asresearch projects on numerically controlled machines.

Library

The final visit at the University was to the library where we met adelightful stereotype of an old-fashioned Chinese librarian. Heproudly showed us two enormous reading rooms, empty at thispoint because of vacations, and a number of small ones in whichseveral people were working. In addition to the ubiquitous racksof Mao teaching and current propaganda, he showed us that hehad most of the technical journals from the rest of the world.However, wherever 1 was able to see what was being read by thepeople in the reading rooms, it was always in Chinese.

The librarian then took us back of the book shelves where some ofhis 3000 -year -old special treasures were laid out on a row oftables. Outstanding among these was a set of bone inscribings-the oldest evidence of Chinese civilization I have ever seen. Therewere also examples of calligraphy from a mammoth encyclopediaof Chinese knowledge that was written hundreds of years ago.

Research programs

After the library visit, we returned to the large conference roomfor final discussion (and more tea). The only new information thatcame out at this session concerned the method of determining theresearch program; the program seemed to have three parts:

1) A part determined by the State.

2) A part that is research done jointly with factories (also presumablycontrolled by the State).

3)A part that was discretionary on the part of the University.

The program is definitely all applied research. Related basicresearch is done by the Scientific Academy. However, basic orpure research is also combined with relevant production. Unfor-tunately, this last comment was not explained further. Theprograms in Qinghua University and in the factories associatedwith it were considered as trial production; if successful, theproduction would go into regular Peking factories.

Mr. Ma made the point that his goal is in accordance with theteachings of Mao to increase output from the factories so that theUniversity would be completely self-supporting.

Seminar at MACHIMPEX

We were taken to MACHIMPEX (to an outbuilding labeled"Meeting Rooms" in English) where 33 Chinese were assembled.The plan was for the whole group to meet for introduction andwelcome and then to separate into two groups-telephone andtelecommunications (when the group used the term "telecom-munications" 1 learned that they used it strictly for com-munications other than the telephone, telex or data com-munications; on the other hand, earth stations appeared to beconsidered as telecommunications). The groups were separatedwith the telephone group going to another room with about 15 ofthe Chinese including Mr. Hsu. Before he left, Mr. Hsu made apoint of greeting me as though to apologize for not being able tobe in two places at once. I took the opportuntiy to give him acomplete set of the brochures on broadcast equipment. Cam-pobasso, Gifford, Johnson, and I stayed with Sodolski who actedas moderator for the following talks:

Campobasso covered microwave systems for use in private systems,railroads, and pipelines, and earth stations.

Gifford covered power line carrier systems and teletype equipment.

Johnson covered the mobile line of his company, including Citizen'sband and their applications.

I covered the broadcast equipment area and the RCA domestic satellite.

The official delegation of EIA executives (I to r): J.T. West of Western Electric; Dr. Hillier ofRCA: R. Sanders of GTE; E.F. ,ohnson of E.F. Johnson Co.: J. Redmond of GeneralDynamics; R.P Gifford of General Elect,c; John Sodolski representing EIA: and T.A.Campobasso o' Collins Radio. Rockwell International; Glen Solomon (IBM) of EIA Boardof Governors and head of the delegation is taking the picture. Lynn Ellis of ITT was notpresent.

31

Semiconductor Research Institute

We went a short distance north of the hotel to an old area ofPeking where the Semiconductor Research Institute is located.We were welcomed with tea and cigarettes by a Mr. Liu who toldus that the Semiconductor Research Institute was under theAcademy of Sciences and was established in 1960. The total staffis 900 of which 400 are research workers; others are staff andworkers. The work is distributed in several 4- and 5 -storybuildings, some of which were converted dormitories.

The Director of the organization heads a Party Committee butexperimental work is determined by department heads withapprovals. The Institute is divided organizationally into sevendepartments:

I) Materials Science-studying silicon. gallium -arsenide, gallium-hosphide.

2) Integrated Circuits and Hybrids -emphasis on LSI processing-(theydo not make masks for factories but do for tests).

3) Reliability of IC's.

4) MOS.

5) Microwave Semiconductors.

6) Instrumentation for all Departments.7) Semiconductor Laser.

In addition there is a small factory (or model shop) to processmechanical parts for all departments. For the past 15 years,emphasis has been on self-reliance (or self-sufficiency). Therefore.most instruments are homemade: a few instruments are imported(we noticed one or two from Hungary).

Mr. Chen was to guide us through the laboratories. We were toldthat we would be shown work on Impact diodes, mixer diodes,Gunn effect, S/ C laser measurement, and materials.

We were taken first to the microwave solid-state deviceslaboratory where we were shown some apparatus supposed to betesting the power conversion loss and noise characteristics of aShottky mixer diode. The microwave plumbing was quiteconventional-I learned later that it was Japanese. There was alocal oscillator with an attenuator and a noise source feeding tothe diode. The noise source was a homemade apparatus with amicrowave switch feeding a stub line or detector. There were twochart recorders on which previously made charts were left. Thepens were lifted on both recorders and were moving in some waythat was not explained. One was supposed to be power and theother a noise measurement. The power handling capability of thediode was stated to be 50 mW.

\\.._URRE NT

VOLTAGE

CURRENT

1---wj'--VOLTAGC

Traces were observed (center) on a sampling scope of an impatt or Gunn -effect diode(left). An onset of avalanche effect (right) was to occur with increased voltage.

The second laboratory had what was described as an lmpatt diodebut later seemed to be Gunn -effect diode. We were shown thedevice under a microscope and observed traces on a samplingscope. When the voltage was raised there was supposed to be the

onset of an avalanche effect. 1 ney claimed that it was working at200 MHz but could go to 2 GHz.

The next stop was the Materials Research where we were shownapparatus for making "dislocation free" GaAs. The apparatusconsisted of a tube (quartz) of two diameters. The larger diametercontained a boat (of quartz) in which there was some gallium. Thesmaller tube contained broken pieces of arsenic. The materialswere claimed to have purities of 0.01 ppm. These were heated to450°C and 610°C respectively to react on a seed crystalintroduced; the dislocation free GaAs was produced by movingthe oven along the tube. They claimed electron densities of 1 X10's and electron mobilities of 7000 at 300K and 30,000 at 80K.

Vxwizzail6a

am)

The smaller tube contained broken pieces of arsenic.

As

We were shown an epitaxial reactor using dry (-70°C dewpoint)Hydrogen AsCI, under controlled temperatures (not indicated).The apparatus looked crude and actually not used. They did notexplain the Ga source or show us how the GaAs substrate wasplaced. Samples of the epitaxial layers seemed to be put on thenatural surface of the original crystal. We were shown what wasclaimed to be a diffusion source that gave greater uniformity ofdiffusion for GaAs. The translation became impossible at thispoint so 1 am not really sure of what we were being shown. Themain point appeared to be the use of an "arsenosilica" film.

We were shown a computer said to be homemade and used forcomputer aided design. The computer had boards on which wereplaced a number of ceramic packaged 1C's-but judging from thenumber of leads these could only have been the simplest ofcircuits. The computer had a speed of 100,000 operations persecond and an 18-k memory (24 bits per word). Its outputs were S-and 7 -hole paper tapes and a numerical printer (about 20 -column). It became evident later that this machine was producingpaper tape to control a mask -making machine based on what theycalled the "building block system". The machine was a reducingprojector in which a selected "building block" on a custom masterwas moved into a specified position in the projector. The maskbeing exposed was on an x -y table also positioned accuratelybefore the exposure was made by flashing a discharge type oflamp. All functions in this machine were under control of thecomputer -generated paper tape.

Although we were shown mask -making apparatus, we did not seean actual picture, mask or other indication of the type ofintegrated circuits being developed.

National Meteorological Observatory

We were taken to the National Meteorological Observatorywhere a Mr. Lee, the Director of the Observatory, welcomed usand briefed us on the operation. The task of the central office ofthe National Meteorological Service is the large-scale forecastingfor all China with special emphasis on disaster warnings. There isan emphasis on communications since global meteorological

32

information is collected, plotted and redistributed through thisoffice. Since 1970, there has been an automatic picture transmis-sion system in operation for the facsimile transmission of weathermaps. In addition, this office has equipment to receive picturesfrom the ESSA 3 and 4, and the NOAA-8 weather satellites. Thesecloud pictures were used as one of -The tools in forecasting. High -frequency radio and wire lines still provide the backbone of thecommuncations system.

Mr. Lee pointed out that China is a developing country-amember of the Third World and that, in spite of their manydevelopments in their field, China does not meet the needs of theirnational defense and economy. He feels, in comparison with otherparts of the world, their facility is "backward". (By now I wasexpecting that we were to be impressed-and we were.)

We were taken first to some large rooms where a large battery ofold and slow teletype machines (with paper tape punches) werebringing in weather data from some 200 reporting stations inChina and from a number of other countries by radio or wire. Theoperators were retransmitting the data to forecasting centersaround the country. The same information was used to plotweather maps for all of China in the central office (we were notshown this operation). Maps were distributed in China byfacsimile. Frequency shift keying, approximately 1000Hz to2000Hz, was used to transmit a map in 21 minutes. In the facsimileroom, we were shown some new single-sideband receivers (allsolid state) that gave four channels of frequency shift keying ateach frequency. We were shown also a new facsimile printer madein China-all solid state. (We were not shown this equipment inoperation).

In another room, we were shown the equipment for receivingcloud pictures from weather satellites. This consisted of threereceivers and three drum photo -recorders, all of Chinesemanufacture. These looked very workmanlike and compact.

The final visit was to the computer room where we saw a typicalcomputer made in China. This machine was described as doing70,000 operations/ s and having a memory of 32-k with 48 -bitwords. They claimed to require 30 programmers and to useALGOL -60 as the language. The room had a row of five standard

Consumer Electronics Exhibit at the Shanghai Permanent Industrial Exhibition. A mixtureof monochrome television sets, transistor radios and multiband tube radios wee shown.

racks that were kept closed. There were four smaller cabinets thathad large drums associated with them. Two of the drums wereconnected and apparently operating. The other two drums werenot connected; the drums were about 18 inches in diameter by 24inches high. This was the only type of drum we saw on any of ourvisits in China. As far as I could tell, the computer was usedexclusively for massaging the weather data into a pressure isobarmap.

Shanghai permanent industrial exhibition

The Shanghai permanent industrial exhibition contained a rathercomplete representation of all the types of products made in theShanghai region. It was in a complex of fairly modern largebuildings near the downtown area of Shanghai. The buildingscovered about 100,000 square feet and contained two more or lessindependent exhibits-industrial and consumer products. Thebuildings were on a grand scale but austere and plain. Evendiscounting the fact that this was a showplace exhibiting productsat the leading edge of the PRC's capabilities, it was still veryimpressive. In the machine tool area-size, precision, andautomation were being stressed. Selected exhibits that impressedme were:

A precision -gear grinding machine for spur gears up to twometers in diameter.

Several numerically controlled machines including an 8 -bitlathe, a milling machine with 53 cutters in a lazy susan that wereremoved, placed in the spindle and returned on command.

Several large presses - a 250 -ton press for large, powdermetallurgy parts, a 30 -ton punch press, and a I2,000 -tonforging press.

A pantograph -controlled, precision -spark -etching machine forcutting complex patterns in thick steel plate.

Part of the exhibit was dedicated to indicating their ac-complishments in shipbuilding area. By models, they made thepoint that 38 ships-cargo and tankers-of over 10,000 tons loadcapacity had been built in Shanghai since Liberation. Includedwere samples of electromechanical and electronic marine gearsuch as automatic helms and radar (40 -mile range). In anothersection there were large numbers of examples of textile machines

The computer exhibit at the Shanghai Permanent Industrial Exhibition.

r SOL--

....... ........

I

Final test of racks of telephone communication equipment at afactory in Shanghai (photo by courtesty of J. West).

Tasting modules for telephone carrier equipment at a factory In Shanghai (photo by courtesy of J.T. West.Western Electric).

including several automatic loom and knitting machines. Anotherexhibit had many working models of different types ofautomotive and diesel powered equipment. Hand farming trac-tors, automatic rice -planting equipment, diggers and front-endloaders, large dump trucks, large and small trucks, busses andautomobiles are examples.

There was an electronics section with computers, measuringequipment, oscilloscopes and medical electronics (including alarge isotope scanner that plotted the results in color). Theelectronic home instruments consisted of transistor radios, black -and -white television, large shortwave radios like the one in myhotel room and some audio components. Some "see yourself"black -and -white cameras and monitors were set up in variousparts of this exhibit.

All in all, this exhibition was very worthwhile. Even afterappropriate discounting it still is clear that the P RC is developinga solid and diversified industrial base that must be factored intoany consideration of future trends in world trade.

Telephone carrier equipment plant

A Mr. Cheng, who was Deputy Director of the Carrier Plant (Isuspect a subdivision of the total plant), was the chairman. Theplant had 300,00 square feet with 2,000 workers. There was anadditional 100,000 square feet for dormitories. Mr. Chen was ineffect the Chief Engineer and a Mr. Tu, the ManufacturingManager. Most of the other people in the room seemed to betechnical, including several women.

Our first visit was to a shop making parts and materialsubassemblies for teletype machines. Without question, this was areal operating plant. There was a great variety of lathes, millingmachines and grinders in operation. In the quick walk through, itwas not possible to get any feel for the organization of the shop orthe work flow (the impression, however, was that there was littleorganization to the work flow). The final product was a slowteletype machine of about 1940 vintage. At the end of this tour wewere shown an operating unit of a more modern machine that was

described as semielectronic and operated at variable speeds from50 to 100 baud. It had a paper tape punch attachment.

We were then taken through the carrier equipment factory. Thisappeared to be primarily the assembly and alignment of modulesfor carrier -type telephone systems. These were assembled in largeracks as repeater or terminal stations. There appeared to be a 12 -channel system for open -wire lines and 60 -channel systems forcable. The final test location was emphasized.

Superfically, this all seemed very similar to many plants I havevisited in RCA in Broadcast Equipment. However, there wereseveral indications that things were not well in this plant (thethroughput was very low).

At the concluding tea ceremony, we learned that all thecomponents in the carrier systems were made in China.Capacitors, coils, printed boards and ferrites were made in thisplant. Transistors, resistors and relays were made in other plants.The Chinese asked if our more modern designs of teletypewriterscould have less mechanical complexity. They were told thatwhenever possible mechanical functions were replaced by elec-tronics and where mechanics were necessary there were manymethods of both lowering cost and increasing reproducibility andreliability. The use of die-casting and of powder metallurgytechnology instead of hand machining were given as examples. Inanother part of this discussion the telephone experts in our groupindicated that carrier systems-particularly the 60 -channelvariety-were no longer made in the U.S. This was largely due tothe excessive labor required to balance the circuits at installation.

In the course of this visit and discussion, I began to wonder if themanufacturing process in China was limited by the excessivenumber of hand operations, the concomitant human errors thatoccur in series throughout production and the apparent absenceof testing according to reliability statistics. In other words, cancomplex carrier systems or large-scale computers be effectivelyput into serial production from a "cottage industry" base, evenwhen the development engineering is very good.

34

Computer manufacturing plant

A Mr. Hsu, who was the Deputy Director of the factory, gave usthe initial briefing. The computer manufacturing plant we visitedmade desk -top calculators, medium and small general purposedigital computers, and some industrial computers for numericallycontrolled machines. The plant had 900 workers. It includeddining rooms, clinics and nurseries, a technical school and a part-time college. Mr. Hsu made a point that this factory was aparticipant in Mao's diffusion of technology principle. Hedescribed what he called 3 -in -1 combination groups. For instance,he described collaboration with high schools, universities andresearch institutes by the leaders, technicians (technical per-sonnel) and workers. All the collaboration was with institutions inShanghai. Some university students were working in the plant atall times.

The main targets of the factory were design, assembly, testing anddebugging. All the components were designed and manufacturedby component plants in Shanghai. The principal users of thecomputer products were research institutes, factories, and uni-versities.

Our first stop on this tour was a small room containing a numberof automatic magnetic -memory -core testing machines. The coreswere 0.8 -mm 0.D., 0.6 -mm I.D. and were fed to the magnetictesting head by a conventional vibrating feeder. The specific testswere not described. The tested cores were automatically dis-charged to either of "go or no-go" boxes. We then went to a largerroom where a number of workers, mostly women, were stringingthe cores in memory planes. This was a truly hand operation withno mechanical or visual aids. The basic memory module was asingle plane with 72,000 cores strung in groups of 4,000. Thememory module was exercised and tested by what was apparentlyconventional hardware but did not seem to have automatic faultindication (the translation fell apart at this point).

The next stop was at an assembly area for the desk -topcalculators. This was a conventional, simple four -functioncalculator with no memory. It was floating point with a large, (1cm high) bright 12 -digit fluorescent display. The display wasmade of 12 individual tubes plugged into sockets. The electronicsconsisted of seven printed circuit boards (about 3 inches x 6inches) that plugged into an array of sockets in the base of the rearof the calculator. Each of the boards had up to 18 IC's in small -ceramic, dual in -line packages mounted in two rows along the topedge. We were told they were Chinese made P-MOS IC's. Laterwe were told that some of these calculators were sold in the regularstores but, surprisingly the price varied from $500 to $1000, theonly example of a variable price that we encountered in the PRC.

We were then taken through the component insertion andsoldering operation for the computer boards. Again, the compo-nent leads were hand -soldered. In this area there was a new flow -soldering machine (imported from Canada). It had obviouslybeen set up very recently and was in the debugging phase. Theresults obtained from it in the demonstration for us were verypoor. From there we went to the final test of their type TQ- I 6small -size digital computer. It was described as being a 110,000operations/ s machine with a 32-k word memory with 48 bits perword (40 -bit mantissa and 6 -bit exponent) and a 2 microsecond

access time. It was a floating point machine and used 24 -bitinstructions Its peripherals consisted of a double paper -tapereader with photo -sensing, a teletypewriter, a drum -type lineprinter with speed of 120 lines/ min, magnetic tape units thatlooked like copies of IBM tapes of 15 years ago (they claimed 20bits/ mm) and the same magnetic drums we had seen at theMetereological Observatory in Peking.

In still another room we were shown their largest TQ-6 machine.It was claimed to have 106 operation/ s, a 128-k word memorywith 48 -bit words (40 -bit mantissa and 8 -bit exponent). Theperipherals were the same except that the printer was supposed tobe 1200 lints/ min. There was also an x -y plotter that wasdemonstrated by drawing the outline of a ballet dancer. The plotwas about two feet square and the figure was completed in about75 seconds. The printer was demonstrated by the common trick ofmaking it draw a pattern. This time it was a Mao saying in Chinesecharacters. The printer did not seem to be going anything like1200 lines/ min.

After this we returned to the conference room for the concludingdiscussion. We learned that the IC's in the computers were TTL'sand had one to several gates per package. The TQ-16's were beingproduced in dozens of units per year. The larger TQ- I 6 was firstproduced in 1974 and a few per year were being made. Futuredesigns would use LS l's but these were in the experimental stage.

Here, as in other visits, one could not escape being impressed withthe basic technical progress. On :he other hand, there seemed tobe unarticulated concerns regarding the manufacturingtechnologies, processes and procedures. The apparent lack ofinspection and quality control seemed in contradiction with thehigh content of hand labor in the total manufacturing process.The compounding of undetected human errors in their produc-tion systems cannot help but reduce the throughput of the finishedproduct.

Conclusion

Let me conclude with a few pointers concerning thetreatment that first-time U.S. visitors to the PRC mightexpect:

Important business will not be discussed with the newlyarrived first-time visitor to the PRC. He must spend sometime beginning to understand the PRC.

The visitor must show some Indications that he can thinkobjectively about the PRC.

The visitor must establish credibility with his hosts withregard to his competence in his position.

The visitor must be sensitive to the fact that the con-ventional wisdom of U.S. business with its accompanyingjargon is an anathema to the Chinese.

Great importance is put on personal relationships... It must be recognized that it takes patience to do business

in the PRC. Finally, the visitor should learn something of the

structure of the language. The objective is not to learn toconverse (that takes a lifet me) but to learn how toformulate his English so that it can be translated easilyand unambiguously.

35

Glass passivation of IC'sby chemical vapor depositionWerner Kern

Operating principles, performance capabilities, and limitations of chemical vapordeposition (CVD) reactor systems are described for producing glassy overcoats of silicondioxide and phosphosilicate passivation layers for metallized silicon integrated circuits.Reactor systems now available include horizontal tube types, rotation units, andcontinuous processing systems. The design of an in -house -built rotary reactor isdescribed in detail, and specific recommendations are made for the construction ofassociated gas flow equipment. In addition, the functions and advantages of CVDpassivation overcoats are discussed, and critical factors in their deposition are pointedout.

GLASSING and glass passivation arecommonly used terms to denote theprocess in which a glass -like, amorphous,inorganic dielectric layer is formed overthe surface of a completed microcircuitwafer for environmental protection. Thesequence for glass passivation consists ofdeposition of the dielectric layer over theentire surface of the device wafer withcompleted metallization patterns,followed by photolithographic delinea-tion to remove glass from the centralregion of bonding pads and from scribe -line areas. Typical deposited films are 0.5 -to 2-µm thick.

Most modern integrated circuits (lCs) aremetallized with aluminum. A compatibleglass passivation process must thereforebe performed under conditions where themaximum processing temperature is

below the Al -Si eutectic temperature(577°C) to avoid alloying or metalliza-tion melting problems. Similar con-siderations hold for metallization systemsinvolving gold. Chemical vapor deposi-tion (CVD) of dielectric films at lowtemperature (300 to 500°C) is ideallysuited to fulfill these requirements.Reactive sputtering, rf sputtering, andplasma deposition techniques can also beused for depositing dielectric layers incertain applications where the IC devicesare not degraded by such treatments.

Glassing of microcircuit wafers wasoriginally used to provide mechanicalprotection against scratches of the softaluminum interconnect lines. Vitreoussilicon dioxide (Si02) prepared by CVDwas applied first as the passivating glass,and is still being used by a number of ICmanufacturers. However, to provide

effective protection, a Si02 film thicknessincompatible with the aluminummetallization would be required, andcracking of the oxide film would result,with consequent problems of devicereliability. Many device manufacturers,recognizing these shortcomings, havesubstituted more compatible lower -stressfilms of binary silicate glasses, especiallyphosphosilicate glass (PSG) films, for themore highly stressed Si02 layers.

The main objective of this paper is tosurvey the CVD reactor systems andequipment now available for low -

temperature deposition of passivationovercoats, and to examine the principleof operation and the capabilities andlimitations of these systems. In addition,the types, functions, and benefits ofpassivation coatings are briefly reviewed.Important aspects of film deposition arenoted indicating the relationship of CVDparameters, film growth and properties.

Types, functions, and benefitsof passivation coatings

Types of passivation coatings

Passivation coatings are widely used toimprove the performance and reliabilityof silicon devices of various types, rang-ing from discrete mesa -type diodes andtransistors to complex planar integratedcircuits - and including both hermeticand plastic -encapsulated devices.

Passivation coatings may be classified asprimary when in direct contact with thesingle -crystal silicon from which thedevice is fabricated, and as secondarywhen separated from the device by an

underlying dielectric layer. The primarypassivation layer provides good dielectricproperties, low -surface recombinationvelocity, controlled immobile -chargedensity, and device stability at elevatedtemperatures under bias or operatingconditions. The secondary passivationlayer provides additional stability invarious ambients and serves as getter,impurity barrier, or mechanical shield. Acomprehensive survey paper' presentedelsewhere reviews the entire field ofsilicon device passivation and may serveas a source of general backgroundmaterial in the present paper.

This paper deals specifically with theapplication of CVD for producingpassivating and protective layers suitablefor overcoating aluminum -metallizedsilicon semiconductor devices in finishedwafer form. The two materials of majorimportance in overcoat passivation areSi02 and PSG deposited by thermallyactivated oxidation of the hydrides.These materials can be deposited in highquality at the low temperature pre-requisite with aluminum -metallizedsilicon devices; functional requirementsare summarized in Table I. A schematiccross-section of a typical IC with overcoatpassivation is shown in Fig. I.

SECONDARY

EMITTER METALBONDING PAD

PASSIVATION GLASS

THERMAL SiOaBASE METALBONDING MD

SECONDARY PASSIVATION OVER METALON BIPOLAR CHIP

Fig. 1 - Schematic cross-section of a typical bipolar ICwith CVD overcoat passivation (not to scale).

Uses and benefits ofglass passivation overcoats

Primarily, passivating overcoats provideprotection against scratching of thevulnerable interconnect metallizationduring chip handling, ensure immunity ofeffects of loose conductive particles inhermetic packages, improve devicestability in various ambients, lowersusceptibility to metal corrosion andelectromigration, and reduce effects ofion motion on the device surface. PSGhas, in addition, alkali gettering capabili-ty which is of great importance in passiva-tion and stabilization of devices, andexhibits less stress (tensile) than layers ofCVD Si02. The effectiveness of PSG for

36

Table I - Functional requirements of IC passIvatIon Table II - Benefits of IC glass passivation by CVDovercoat layers. overcoat layers.

No. Functional Requirements

I Mechanical and chemical protection formetallization interconnects

2 Diffusion barrier or gettering agentfor ionic contaminants and other exter-nal impurities

3 Good adherence to metallization andprimary passivating layers

4 Low stress from all causes

5 Good chemical and physical stability

6 Low defect density

7 Inertness toward metallization andother structural device components

8 High dielectric strength and electricalresistance

9 Low dissipation factor

10 Low mobile or trapped charge density

I I Isolation of electrical charge effectsexternal to semiconductor

12 Sufficiently matching thermal expansionwith device component materials

13 Reduce or maintain semiconductor sur-face state density

14 Moderately high dielectric constant tocontain junction fringing field

15 Ease of preparation and subsequentprocessing

16 Ease of formability into patterns byphotolithography and etching

17 Compatibility with plastic encapsulatingmaterials.

gettering alkali ions is particularly advan-tageous for MOS ICs because of theirincreased surface sensitivity as comparedwith that of digital bipolar ICs.

Alkali gettering capability is also of greatimportance for devices enclosed inceramic packages that are sealed t)

fusion of glass frits. These frits usuallycontain large amounts of sodium which isemitted during fusion into the ambient ofthe package enclosure. Encapsulatingplastic formulations are also a source ofions and other impurities that must beprevented from penetrating into thesensitive device regions. A summary ofbenefits and advantages achieved by IC

Hi'tirhl

I Gettermg or immobilizing of harmfulalkali ions (in the case of PSG)

2 Decreasing instabilities due to surfaceionic drifts or horizontal surface -ionmigration

3 Protection against metal corrosion

4 Suppression of electromigrationsusceptibility

5 Suppression of defect formationin aluminum metallization

6 Reducing penetration of moisture, gases,and chemical species from the outside,including components from the plasticencapsulating material forming thepackage

7 Quenching of fast states to lower leakagecurrents

8 Improved protection against high-energyelectromagnetic radiation

9 Mechanical scratch protection duringwafer and dice processing

10 Prevention of shorts from loose con-ducting particles in hermetic packages

12 Compatibility with many types of thin-film resistors used on ICs

13 Other benefits including general increasein device reliability and yield due todecreased failure rate and susceptibilityto electrical instabilities.

glass passivation by CVD overcoat layersfor the vast majority of IC types is

presented in Table II. It is clear from theseconsiderations that passivating overcoatlayers constitute an important step inmanufacturing ICs of high reliability andimproved performance.

Preparation ofpassivation overcoats

Chemical vapor deposition

Chemical vapor deposition is a process bywhich gases or vapors are chemicallyreacted, leading to formation of a solid -phase reaction product on a substratesurface. Activation of the chemical reac-

Werner Ken. Member of the Technical Staff. RCALaboratories, Princeton, N. J., received his education inchemistry at schools in Switzerland and the U.S.A.,including the Jntversity of Basle and Rutgers University.His thesis. published in 1947, was on thechromatographic isolation and characterization offluorescing polynuclear hydrocarbons which he dis-covered in soil. He was analytical research chemist withHoffmann -LaRoche in Switzerland until 1948 when hetransferred to their research division in New Jersey todevelop radioactive tracer methods for biochemicalapplications In 1958 he joined Nuclear Corporation ofAmerica where he became chief chemist directingapplied research in nuclear radiation chemistry. Hejoined RCA Electronic Components and Devices Divi-sion in 1959 primarily to investigate semiconductorprocesses by radiochemical methods. He was projectscientist and consultant on several research projects.and was in charge of radiological safety. Since 1964 hehas been at RCA Laboratories as a Member of theTechnical Staff specializing in semiconductor processresearch, chemical vapor deposition technology. and thedevelopmerrient of new analytical methods forcharacterizing dielectric films. In 1974-1975 he has beenproject scientist for two governmer t -sponsoredresearch contracts on integrated circuit glass passive -lion. Mr. Kern is a member of the American ChemicalSociety. the Electrochemical Society, the ResearchHonorary Society of Sigma Xi. and Geological Society ofNew Jersey, and is listed in American Men and Women ofScience. He is author or co-author of over 50 scientificpublications and holds four U.S patents. He received anRCA Achievement Award in 1966 for his work inintegrated circuit process research. Mr. Kern is therecipient of the T D. Callinen Award for 1971 of theDielectrics and Insulation Division of the Elec-trochemical Society, in recognition of his p oneenngwork in chemical vapor deposition research. He receiveda 1973 RCA Laboratories Outstanding AchievementAward for his team contributions to glass passivation ofsilicon device structures.

Reprint RE -21-5-7Final manuscript received August 11. 1975

This work was in part supported by the Air ForceMaterials Laboratory, Wright -Patterson Air Force Base.Ohio. under Contract No. F33615 -74-C-5146. Portions ofthis paper were presented verbally at NAECON '75, andan extended summary on CVD reactors was included inthe Proceedings volume.

37

Table III - Advantages and disadvantages of CVDtechniques for preparing passivation overcoat layers.

Vo. Advantages

I Low deposition temperature

2 High chemical purity of deposited layers

3 Wide choice of compositions

4 Ease of preparing a variety of layeredstructures

5 Desirable physical andchemical film properties

6 Ease of thickness control

7 Good uniformity of film thickness andcomposition

8 Good adherence to oxidesand aluminum

9 High -resolution patterns can be readilyformed in layers by photolithography

10 Economical and practicable on a pro-duction scale

II Process can be automated

Disadvantages

I Particulate impurities from the reactionarc formed and must be minimized

2 Toxicity of reactants requires safetymeasures

3 Accurate control of gas flowsand deposition temperature is needed

tion is most commonly effected byheating of the substrate. CVD is

employed extensively in semiconductordevice technology, for example, in theepitaxial deposition of single -crystalsemiconductors, silicon nitride impurity -barrier films, oxide and nitride dielectricfilms, doped oxide layers as diffusionsources, silicate glasses for passivation,and, in some instances, metal films asconductor elements.

Theoretical and practical aspects ofchemical vapor deposition processes havebeen treated extensively in several recentreviews- ". The advantages and disad-vantages of the CV D technique for deviceglassing are listed in Table Ill.

Basic CVD hydride reactions

The basic process for depositing Si02films from silane and oxygen at low

temperatures (250° to 550°C) wasreported in 1967.1' The exact details ofthis thermally activated, surface -catalyzed, heterogeneous branching -chain reaction are complex," but theoverall reaction can be expressed as:

si H.,(g) + 202(g) - Si02(s) + 2H2 0(g),

but under some circumstances it mayproceed as

Si th(g) + 02(g) - Si02(s) + MAR).

The reaction favored depends strongly ondeposition temperature and silaneconcentration:2-16 and probably also onthe oxygen -to -hydride ratio andvariations in reactor geometry.

Phosphorus can be incorporated into theoxide layers as an oxide of phosphorus bythe reaction of phosphine with oxygen:

2+ 402(g) - P2 05(5) + 3112 C)(g)

Cooxidation of PH; with Silo formsPSG.16-"'

Effects of CVD parameters on film growthand properties

The exact conditions used in the CVDprocess for Si02 and PSG films cancritically affect important film properties.Primary CVD parameters that affectdeposited film properties must be careful-ly optimized and controlled; suchparameters are listed below in their ap-proximate order of importance:

1)Substrate temperature of deposition

2) Oxygen -to -hydride ratio

3) Hydride flow rate

4) Silane-to-phosphine ratio

5)Nitrogen flow rate

6)Geometry of reaction chamber and gasinlet/ outlet configurations

7) Wall temperature of reaction chamber orgas disperser

8)Cleanliness of CVD system and purity ofgases

9) Nature and cleanliness of substrates sur-face

10)Surface topography of substi ate

We have investigated in considerabledetail the effects of these factors on filmdeposition rate, film uniformity, and filmcomposition (in the case of PSG). Insummary, it can be generalized that eventhough the mode of operation may differgreatly for various types of CVD reactorsystems, the basic CVD parameters un-derlying oxide and glass film formationby gas phase oxidation of the hydrides attemperatures below 500°C are, in princi-ple and often semiquantitatively,applicable to all systems. The mostcritical factors determining PSG com-position and film quality are substratetemperature of deposition, oxygen -to -hydride ratio, silane-to-phosphine ratio,and hydride flow rate. These factors mustbe controlled and optimized for a givenCVD system by analysis of the filmproduct obtained.

The CVD process should be directed in afashion conducive to heterogeneous gas -phase nucleation to produce clear, glassyfilms free of defects. Homogeneous gas -phase reactions produce particulate con -

Table IV - Effects of CVD key parameters for preparing P80 films.

DIRECTION OF ARROWS INDICATES RELATIVE INCREASE OR DECREASE

/ \ STRONG; CSLIGHT; -. NONE

CVD PARAMETERSEFFECTS ON FILM

DEPOSITIONRATE

PHOSPHORUS

CONTENT

INTRINSICSTRESS

HYDRIDE SI H4 PH3 1

FLOW RATETI me

HYDRIDE P K3

RATIOSI H4 -..../ / \

OXYGEN 02 r-\\ ./....../ *----...RATIO

SI 114 P H3 1

DEPOSITIONT

TEMPERATURE1

H

V. \ \DILUENT

GAS FLAWN2RATE

1 /--\ -H HIGH. L LOW OXYGEN RATIO

38

taminants and must be suppressed byapplication of suitable techniques, suchas cooling of the reactor walls,22 use ofhigh diluent gas flow rates,22 addition ofselective inhibiting agents," or utilizationof close -space reactor designs.

The essential effects of CVD keyparameters on PSG film deposition rate,phosphorus content, and intrinsic filmstress are shown in Table IV. Thisschematic also allows one to predict, at aglance, what results can be expected oncombining the various parameters, orhow to compensate an effect by in-creasing or decreasing the relativemagnitude of some of the other factors.Quantitative results have been presentedelsewhere,32 and will be discussed in detailin a forthcoming paper." Even thoughthe exact reaction mechanism of filmformation is quite complex and is in-fluenced by many variables, theoptimized preparation of high -qualitySiO2 and PSG films for IC overcoatpassivation is a CVD process wellsuitable for large-scale production.

Compatibility of deposition temperaturewith metallization

The theoretical maximum temperaturefor processing devices metallized withaluminum is limited by the melting pointof the eutectic formed between silicon andaluminum (577°C). In practice, however,the maximum temperature during glassdeposition is held below 500°C becausesolid-state reactions between aluminumand silicon begin to exert degradingeffects in many sensitive devices at ap-proximately 500°C. Some reaction ofaluminum with SiO2 or PSG can bedetected in certain devices attemperatures as low as 400°C, forming athin intermediate layer of alumino-silicate, but normally this presents noserious problems when proper etchingtechniques are used in delineation of theglass layer to expose the bonding padsand scribe lines.

Problems associated with corrosion, edgeeffects, stress, and expansion mismatchesbetween aluminum and the passivationovercoats have been discussed'. An up-to-date biliography of many other relatedaspects of aluminum and other metalliza-tion for silicon devices became availablerecently. 34

Layer combinations

Combinations of PSG and SiO2 layers

can offer distinct advantages over any onesingle layer. Structures consisting of SiO2over PSG, or of Si02/ PSG/ SiO2, can bereadily prepared by CVD techniques,often in one continuous operation, simplyby regulating the hydride input in the gasstream .22 A thin (1000-A) CVD SiO2 toplayer over the phosphosilicate main layerof 0.6- to l.5 -µm thickness is desirable forimproved photoresist adherence and con-sequently improved pattern etchingdefinition, unless organo-silaneadhesion -promoting agents are used. Thecomposition of PSG layers used in in-dustry is generally in the range of 2 to 5 wt% P (or 2 to 5 mol % P2O5).

Survey of CVD reactorsystems and equipment

Requirements for reactor systems

A CVD system for depositing passivatingfilms of the type described must provideequipment that accomplishes the follow-ing functions" 4 10.33

I) transport, meter, and time the diluent andreactant gases entering the reactor;

2) provide heat to the site of reaction, namely,the substrate material being coated, andcontrol this temperature by automatic feed-back to the heat source; and

3) remove the by-product exhaust gases fromthe deposition zone and safely dispose ofthem.

The reactor system should be designed tofulfill these three primary functions withmaximum effectiveness and simplicity ofconstruction. It must consistently yieldglass films of high quality; good thicknessand compositional uniformity from waferto wafer and from run to run; and highpurity with a minimum of structuralimperfections such as pinholes, cracks,and particulate contaminants. Forresearch applications, these requirementsare usually quite sufficient, but for large-scale production applications the systemshould, in addition, perform with highthroughput, be simple and safe tooperate, and be easy to maintain routine-ly. Labor-saving automation36 and com-puterization are attractive additionaloptions that can be well worth the in-creased capital cost in reducing laborcosts, resulting in net savings. Similarconsiderations concern the capital cost ofthe equipment, which ranges from lessthan $3,000 for an in-house built researchor pilot production unit to over $70,000for a high -capacity production system.

The capital cost must be carefullyevaluated in terms of the system's actualproduct output, operator andmaintenance labor required, consumedgases, power consumption, and partsreplacement cost.

A considerable degree of sophisticationand refinement has been engineered intopresent-day commercial reactor systems,emphasizing maximum productthroughput, improved process control,and convenience and economy in process-ing and equipment maintenance. Varioustechniques for process automation havebeen developed, including full computerautomation with separate digital con-trol of each of the various parameterssuch as temperature control, gas com-position, and timing sequence. Heating isusually effected by resistance -heatingtechniques, and operation is at nominallyatmospheric pressure. Provisions havebeen made in several of the commercialreactors for cooling of the reactor walls orthe gas distributor to suppresshomogeneous gas phase nucleation.

Basic types of reactors

Reactors can be classified in essentiallyfour main categories, according to theirgas flow characteristics and principle ofoperation:

I) Horizontal tube displacement flow reac-tors.

2) Rotary vertical batch -type reactors.

3) Continuous reactors employing premixedgas flow fed through an extended areaslotted disperser plate.

4) Continuous reactors employing separate,nitrogen -diluted oxygen and hydridestreams that are directed toward the sub-strate by laminar flow nozzles.

The schematic diagrams presented in Fig.2 show the basic features of these fourtypes. Each system and its variants arebriefly discussed and exemplified. Dataon commercially available equipment tobe described were derived from informa-tion supplied by the manufacturer ratherthan by experimental studies by theauthor other than inspection of the equip-ment.

Horizontal displacement flow reactors

Horizontal tube reactors consisting of anelongated quartz tube of circular or

39

0 HH

E!

'_Ca

0

E-4.

ATUBE DISPLACEMENT FLOW

fM=1_}E(000000p000000)

0IIROTARY B

0

1 1

E .

111111

of i

M

re. E

EXTENDED AREA DISPERSERC

LAMINAR FLOW NOZZLESD

0. 02 112 H SIH4 N2 I P115 E EXHAUST GASES

RESISTANCE HEATER SUBSTRATE WAFER

DIRECTION OF GAS FLOW

-'W DIRECTION OF TRAVEL

Fig. 2 - Operating principles of the basic types of CVD reactors forpreparing passivation overcoat layers: a) horizontal tube displace-ment flow reactors, b) rotary vertical batch -type reactors, c) con-tinuous reactors employing premixed gas flow fed through anextended area slotted disperser plate, and d) continuous reactorsemploying separate, nitrogen -diluted oxygen and hydride streamsthat are directed toward the substrate by laminar flow nozzles.

rectangular cross section, as used in high -temperature semiconductor processing,were the first types used in CVD of oxidesat low temperature. The diluent andreactive gases are continually beingsupplied, and are usually mixed anddispersed at the point of entry into thetube. The mixed gas stream passes overthe heated substrates which rest on atilted wafer carrier. Heat may be suppliedresistively from outside the tube or fromwithin the wafer carrier. Displacement -type forced laminar gas flow of this typefeatures a low degree of gas mixing,requiring careful optimization of the gasdynamics to obtain good film uniformityin thickness and composition. Criticalfactors affecting film uniformity are sub-strate positioning, temperature profile ofthe deposition zone, geometry of reactor,and exhaust configuration. Several typesof CVD systems based on horizontaldisplacement flow reactors are com-mercially available. A schematic of thebasic system is shown in Fig. 2a.

A horizontal, internally resistance -heatedflow reactor manufactured by AppliedMaterials Technology, Inc.° has a capaci-ty for ten 5 -cm -diameter wafers per batch.Cycle time for a 5000 -A -thick Si02deposition is approximately 15 minutes.

atAMS-2600 Silox System, Applied Materials Technology.Inc.. 222 San Ysidro Was. Santa Clara. CA 95051.

Uniformity of film thickness is specifiedas typically ±7.5% within a wafer, ±10%from wafer to wafer within a run.

A tubeless horizontal production reactorof entirely different design is manufac-tured by Thermcoh. It utilizes aresistance -heating block with the gasesintroduced across its width to minimizedepletion of the gas reactants movingacross the heating platen. The gasdelivery system has provision for bi-directional gas entry/ exhaust. The unithas a loading capacity for 27 five -cmwafers and a deposition rate capability of1000 A/ min. The film uniformity within awafer is ±5%, and ±10% from wafer towafer.

A developmental close -spaced horizontalcold -wall reactor has also been des-cribed." A high -velocity stream of thepremixed gases passes over the resistance -heated substrate wafers parallel to theirsurface. The reactor can accommodate 28wafers of 5 -cm diameter.

Rotary vertical batch reactors

In rotary vertical reactors for batchprocessing, the substrate wafers areplaced on a plate rotating above aresistance heater (Fig. 2b). The reaction

bl Thermco Thor II Oxide Reactor, Thermco Products Corp..Box 1875. Orange, CA 92667.

chamber consists of a cylindrical, conical,or hemispherical bell jar that may haveprovisions for cooling. The gases enterthrough single or multiple inlets from thetop or side of the belljar, pass over thesubstrates, and exit at the bottom of thechamber below the rotating plate. Theincoming gases mix gradually with thepartially reacted gas before passing overthe substrates. Intentional changes in gascomposition therefore proceed gradually(rather than abruptly as in the case ofhorizontal flow reactors); this behaviorfor the present application is advan-tageous, particularly when double layers(PSG + Si02) with a graded interface aredesired. The geometry of the reactionchamber and the configurations of the gasinlet and exit openings are critical inattaining the most effective gas flowdynamic conditions required for produc-ing uniform deposits.

Construction and performance of a reac-tor of this type designed for researchapplications was described previously".The unit consists of gear -driven discsheated by conduction from a hotplate.Planetary motion of the substratespromotes maximum uniformity of thefilm deposits by thermal and gas dynamicaveraging effects over a wide range ofoperating conditions.

A modified single -rotation reactor wehave built for research and pilot produc-tion is described below in some detail,since it exemplifies a relatively simple andinexpensive in-house built unit. Fivewafers of 5 -cm diameter can be coatedsimultaneously with a uniformity that isindistinguishable by visual inspection(±2% at 1-1.1m film thickness) from waferto wafer, if the nitrogen diluent flow isproperly balanced. Nine wafers can becoated simultaneously if thickness uni-formity is less critical. Very high deposi-tion rates (up to I µm/ min) can beobtained, although these are not normal-ly used.

A photograph of this unit with a glass belljar as reaction chamber is presented inFig. 3. A commercial high -temperaturehotplate (or electric stove element) servesas a heat source. A 1.4 -cm -thick, 20 -cm -diameter disc of aluminum alloy 6061 isused as sample stage. A small depressionin the center of the disc aids in positioningit freely on a support pedestal machinedof a thermally insulating ceramic(AlSiMag 222). The pedestal fits in arotating steel shaft which extends

40

Fig. 3 - Single -rotation vertical CVD reactor (diameterof the glass bell jar is 22 cm).

through the center of the hotplate directlyto an instrument motor. A mica -insulatedbypass in the heater ribbon can be madeto provide sufficient clearance for theshaft, as indicated in Fig. 4. The discrotates at 6 rpm above the hotplate top,with a clearance of 2 to 3 mm. Construc-tion details are defined and shownschematically in Fig. 4. A chromel-alumelthermocouple serves as temperature sen-sor for the rotating substrate plate. Thethermocouple wires are insulated by aceramic tube and are inserted through thehotplate from below, extending through asmall opening in the Pyroceram toptoward the rotating metal disc, butwithout touching it. The thermocouplesignal is relayed to an automatic tem-perature controller which regulates thetemperature of the hotplate. The rotarydisc temperature is also monitored in-dependently, prior to deposition work,using calibrated bimetallic surface ther-mometers.

The deposition chamber shown consistsof a Pyrex bell jar of 25 -cm height and hasan outside diameter of 22 cm. Thegeometry of the single gas inlet on top ofthe bell jar is clearly visible in Fig. 3.Alternatively, a water-cooled, double -walled, stainless -steel chamber of similardimension has been used for cold -wallCVD studies. The deposition chamber

rests on a Transite plate that has foursymmetrically located slots (13 X 38 mm)serving as exit openings for the exhaustgases. The entire apparatus is placed in achemical exhaust hood whose air intakeis regulated to avoid excessive air drafts,so as not to cause thermal disturbance ofthe hotplate's temperature equilibrium.

The oxide and glass deposits ac-cumulating on the aluminum substratedisc should be removed periodically byimmersion in diluted HF, followed byflushing with deionized water. Occasionalmechanical cleaning with steel wool anddetergent followed by immersion in con-centrated HNO1 for three to five minutesand thorough rinsing with deionizedwater is recommended. To avoid warpingof the symmetrically machined plate onextended use at 450° C, it is simplyinverted after cleaning when reposition-ing it on the support pedestal. This willeffectively nullify the slight sagging thatmight otherwise occur.

Several important considerations shouldbe noted for designing and constructingthe gas -handling equipment associatedwith in-house built units. Heliarc-weldedand leak -checked stainless -steel tubing isrecommended for hydrides of high con -

1. Heater assembly with cut out clearance hole for shaft.

2. Aluminum alloy rotating disc (1.5 cm x 27cm).

3. Centering insert of ceramic support pedestal.

4. Centering recess on both sides of disc.

5. Bushing.

6. Ceramic support pedestal and sleeve drive.

7. Surface of rotating disc for substrates.

8. Transite reaction chamber support plate.

9. Pyroceram hotplate top.

10. Backing plate.

11. Output for heating element.

12. Steel tubing centering shaft.

13. Boston sleeve coupling (eGR4).

14. Hexagonal socket heat cap screw.

15. Thrust bearing (Aetna F-1).

16. Stainless steel sleeve -drive to motor shaft

17. Mica insulation with cut out clearance hole for shaft.

Fig. 4 - Sectional view of essential construction detailsfor single -rotation CVD reactor shown in Fig. 3.

centration (greater than five percent).Leak -tested polyethylene tubing is ad-vantageous for oxygen, nitrogen, anddiluted hydrides, particularly in theterminal parts of the installation whereflexibility and quick removability forclean-up is important. Metering of thehydride gases is best done with calibratedmass flowmeters and controllers. Float -type flowmeters are perfectly adequatefor oxygen and nitrogen. It is very impor-tant to have provision for nitrogen purg-ing of the entire hydride line system,including the pressure regulators back tothe source tanks, to be able to removeresidual hydrides after deposition work isterminated at the end of the day, or to usenitrogen flow during setup to conserveexpensive hydride gases. A schematic of atypical gas flow system is shown in Fig. 5.Additional recommendations, includingvalves, temperature measurement, andsafety considerations were noted inprevious publications.12'24'" A varietyof gas flow control panels is nowavailable commercially' or can be ob-tained custom built for any desiredrequirements."

A commercial rotary, resistance -heatedvertical reactor system for semi -

continuous operation is sold by Unicorp,Inc.' It features premixed gas inputthrough four tubes inside the conicaldeposition chamber for uniform distribu-tion, and has separate stations on acarrier disc for wafer loading/ unloading,preheating, deposition, and cooldown.Up to seven 5 -cm wafers can be accom-modated per station. The maximum ox-ide deposition rate is 1700 A/ min, with aconservatively estimated overall un-iformity of ±5%.

An oxide reactorf featuring two planetarysystems arranged concentrically wasdescribed in 19714°. In this complexreactor, the driver element is a cylinderrotating within an annular spacingbetween a circular inner resistance -heated block and an outer heater ring.The gases are dispensed in a counter -rotating array to ensure good mixing of

c) For example. Matheson Gas Products. P.O. Box 85. 932Paterson Plank Rd.. East Rutherford. NJ 07073.

dl For example. 1 ylan Corp., 4203 Spencer St.. 1 orra nce. CA90503.

dl For example. I ylan Corp., 4203 Spencer St.. Torrance. CA90503.

el Rotox-61:1. Unicorp Inc., 625 N. Pastoria Dr., Sunnyvale,CA 94086.

I) Oxide Reactor 315 Ino longer manufactured). MaterialsResearch Corp., Orangeburg. NY 10962.

41

the gas flows, resulting in a thicknessuniformity of ±1.5%.

Vertical CVD reactor systems withrotating gas feed over a stationary,resistance -heated plate are manufacturedby Phoenix Materials Corporation'. Un-its for processing either 26 or 70 five -cmwafers are available.

A vertical cold wall reactor system byHLS Industriesh, featuring a periphasegas injection arrangement that introducesthe gases into the reactor through twoconcentric wall sections, was describedrecently. Low gas depletion losses andelimination of static boundary conditionsresult in uniform deposits'''.

Continuous reactor systems

Application of continuous processingconcepts to CVD reactor systems makeslarge-scale production of oxide and glassfilms possible at a lower operating costthan that of batch -type reactors, especial-ly if combined with automation that isparticularly suitable for such systems. ACVD processor of this type is manufac-tured by Applied Materials Technology,Inc."' The substrate wafers are moved

g) Phoenix Materials Corp. Vapor Deposition Systems. 500and 1500 Series, M.R. 10. Kittatinny, PA 16202.

h) Univax Reactor Model 100-60-30. HLS Industries, 762Palomar Ave., Sunnyvak, CA 94086.

i) AMS-2000 Continuous Silox Reactor, Applied MaterialsTechnology, Inc., 2999 San Ysidro Way, Santa Clara.CA 95051.

ROTAMETER ORGAS MASSFLOW CONTROLLER

conveyor -fashion on lnconel traysthrough the reactor at constant speed.Heating is by radiation from a resistanceheater. The nitrogen, oxygen, andhydride gas streams are combined beforeentering the manifolds to the slotteddisperser plate from which the gas mix-ture passes by laminar flow over thewafers. The cross -flow gas dispersion andthe relatively close spacing (approximate-ly 1 cm) of the substrate to the water-cooled, extended area disperser plateminimize undesirable homogeneous gasphase nucleation. Wafer -to -wafer un-iformity of film thickness is better than±5%, with unusually low densities ofparticulate inclusions and pinholes. Themachine has a throughput capacity of 400five -cm wafers/ h with a 1 -µm -thick Si02deposit. A schematic is shown in Fig. 2c.

In nozzle reactors, separate streams ofnitrogen -diluted hydrides and oxygen im-pinge on the substrate surface where theymix and react, forming the oxide or glassfilms. The effects of numerous nozzlegeometries on the resulting oxide deposi-tion pattern has been examined in detailby several investigators'''". An openhorizontal circular compound nozzle wasempirically found most desirable, withthe deposition occurring where the heatedmoving wafer intersected the unconfinedjet of the reacting gases, while the exhaustgases were removed by a pump. Uni-formity of ±5% was obtained across afive -cm wafer'''.

A commercially available continuous

- -- PURGE LINE

- - VACUUM LINE® CHECK VALVE

REGULAR AREAFILTER

SOLENOIDVALVE

O ONE -WAYNORMALLY CLOSED

e TWO-WAYNORMALLY OPEN

PH3+ N2

PH3

SiH4 +N2 (I)

I I IVACUUM 0J

SiH4 0

TO EXHAUST

(.1)

r -

TO REACTIONCHAMBER

Fig. 5 - Schematic of an automated gas flow and metering system for CVD of silicon dioxide and PSG films. Lessexpensive mass flow meters or inexpensive rotameters with high -accuracy flow control needle valves can be substitutedfor the electronic mass flow controllers, especially in the case of the nitrogen and oxygen lines. The vacuum lines can beomitted it two -stage gas regulators with provisions for purging are used at the hydride source cylinders.

reactor system based on nozzle -type gasdispersion is manufactured by PacificWestern Systems, Inc!. In this unit thenitrogen -diluted oxygen and hydrides aredirected to the substrate surface asseparate streams flowing through thelaminar flow slots forming the nozzles.The water-cooled nozzle array unit isfour -cm wide and 25 -cm long, extendingover the width of the heater blockassembly. The space between the nozzlearray and the wafer surface duringdeposition is only about 2.5 mm anddefines the reaction zone where the gasesmix and react forming the oxide or glassfilms. Exhaust gases are removed throughseparate slots at the periphery of thenozzle array. The wafers traverse underthe nozzle unit on a resistance -heatedconveyor plate at a variable speed. Thewafer plate (18 cm X 41 cm) accepts 21five -cm wafers. Unusually high filmdeposition rates are obtainable with thissystem, and the density of particulateimpurities in the film deposits is low. Thethickness uniformity for Si02 films is±5%, with a deposition of liam at a rate of6.3 cm/ min travel speed. Figure 2ddepicts schematically the construction ofthis type of nozzle reactor.

A continuous nozzle reactor system isalso marketed by Chemical ReactorEquipment, lnc.446 A high -velocity noz-zle is used, through which the gases aremixed and impinged over a relativelysmall area of deposition surface. A con-tinuous mode of operation results fromthe reciprocating motion of the substrateunder the nozzle. A five -nozzle system iscapable of depositing 0.5-µm Si02 filmson up to 350 five -cm wafers per hour,with excellent uniformity.

Summary and conclusions

Chemical vapor deposition has becomethe most widely used technique forproducing glassy passivating layers overaluminum -metallized semiconductordevices, particularly complex planarsilicon bipolar and MOS integrated cir-cuits for both hermetic or plastic encap-sulation. The most commonly usedpassivating overcoat layers are Si02 andphosphosilicate glass, singly or in com-bination, deposited by oxidation of the

j) PWS Model 2000 Vapor Deposition System. PacificWestern Systems. Inc., 855 Maude Avenue, MountainView, CA 94040. It is now available from Applied Materials,Inc. (AMS-1000 Silox Reactor System).

k)CRE 1101, 1103, 1105, Chemical Reactor Equipment, Inc.,1080 HE Duane Ave., Sunnyvale. CA 94086.

42

nitrogen -diluted hydrides at a substratetemperature ranging from 325 to 450°C.

The functions of passivation coatingshave been defined, and the uses, benefits,and advantages of CVD overcoats forimproving device reliability have beendiscussed and shown to be highlyeffective.

The preparation of passivation overcoatsby CVD techniques has been reviewedand referenced. The most critical factorsdetermining film deposition rate, PSGcomposition, and overall film quality aresubstrate temperature of deposition,oxygen -to -hydride ratio, silane-to-phosphine ratio, and total hydride flowrate. These factors must be controlledand optimized for a given CVD system byanalysis of the film product obtained. TheCVD process should be directed in afashion conducive to heterogeneous gasphase nucleation to produce clear, glassyfilms free of defects. Homogeneous gasphase reactions produce particulate con-taminants and must be suppressed. Eventhough the exact reaction mechanism offilm formation is quite complex, and isinfluenced by many variables, theoptimized preapration of high -qualitySi02 and PSG films for IC overcoatpassivation is a CVD process well suitedto large-scale production.

A considerable variety of CVD reactorsystems for preparing S102 and PSGpassivating overcoats is now available,including horizontal tube -types, rotationreactors, and continuous processing un-its. We have categorized and describedthe design, operation, and capability ofthese systems, which encompass designsfrom relatively simple to sophisticatedautomated concepts.

An example of an in-house built rotaryvertical batch -type reactor we designedand have used successfully for severalyears has been described in detail, andspecific recommendations for the con-struction of associated gas -handling andflow control equipment has been made.

It was concluded that even though theprinciple of operation differs greatly forvarious types of systems, the basic CVDparameters underlying oxide and glassfilm formation by gas phase oxidation ofthe hydrides at temperatures below500°C is in principle, and oftensemiquantitatively, applicable to a widevariety of systems.

Acknowledgments

I would like to thank Dr. G. L. Schnablefor many valuable comments and forcritically reviewing the manuscript. A. W.Fisher contributed significantly to thedesign of the rotary CVD reactor, and R.D. Vibronek proficiently carried outmuch of the experimental CVD work.

ReferencesI. sehnablc. (if.... Kern. W.. and Comizzoli. R.B.. "Passiva-

tion Coatings on Silicon Devices." J. Electrochern. Soc.122. 1092 (1975).

2. Powell, C.F., Oxley. J.H., and Blocher. J.M.. Jr.. Eds..lisp°, Deposition. (John Wiley & Sons. New York. 1966).

3. Feist, W.M., Steele. S.R.. and Ready, D.w.. "ThePreparation of Films by Chemical Vapor Deposition." inPhysic., of Thin Films. Vol. 5. Hass. G. and Thun. R.E..Eds.. (Academic Press. New York and London. 1969), pp.237-314.

4. Campbell. D. S.. "The Deposition of Thin Films byChemical Methods." in Handbook of Thin FilmTechnology. Maissel, L. I., and Glang, R.. Eds.,(McGraw-Hill Book Company. New York. 1970). pp. 5-1to 5-25.

5. Amick. J. A., and Kern. W.. "Chemical Vapor Depositiontechniques (or the Fabrication of SemiconductorDevices." in C lir.lllll a! Vapor Deposition. Blocher. J.M..Jr. and Withers, J.C., Eds., Second Intril Conf.. TheElectrochem. Soc.. Inc.. New York (1970). pp. 551-570.

6. Chu, T. L.. "Dielectric Materials in SemiconductorDevices." J. Vac. Sci. Technol. 6. 25 (1970)

7. RCA Review. Special Issue: "Chemical Vapor PhaseIkposition 01 Electronic Materials." RCA Review 31, No.4 (1970).

8. MacKenna. E. L. -Chemical Vapor Deposited DielectricsProperties and Deposition Methods." Proc. 1971

Semicond.I IC Proc. and Prod. Conf., pp. 71-83. Ind.and Sci. Conf. Management. Chicago. III. (1971).

9. Chu, T. L.. and Schmehter, R. K., "Recent Advances inthe Chemical Vapor Growth of Electronic Materials," J.Vac. Sci. Technol. 10. 1 (1973).

10. Blocher. J.M.. Jr. and Withers. J.C.. Eds.. "ChemicalVapor Deposition" (International Conferences): SecondIntn'l Conf. The Electrochem. Soc., Inc., New York11970): Glaski, F. A., Ed. - Third Intril Conf.. TheAmerican Nuclear Society, Hinsdale. III.. (1972):Wakefield. G. F.. and Blocher. J. M., Jr.. Eds. - FourthIntn'l Conf., I he Electrochem. Soc., Inc., Princeton, N.J.(1973); Blocher. J. M.. Jr.. Hintermann. H. E.. and Hall.I.. H.. Eds. Fifth 111011 Conf., The Electrochem. Soc..Inc.. Princeton. N.J. (1975).

II. Tietjen, J. J.. "Chemical Vapor Deposition of ElectronicMaterials." in Ann. Rev. Marl. Sci., Vol. 3 (1973).Huggins. R. A.. Bube, R. H.. and Roberts. W., Eds..published by Annual Reviews. pp. 317-326.

12. Goldsmith. N.. and Kern. W.. "The Deposition of VitreousSilicon Dioxide Films from Silane." RCA Review 29, 153(1967).

13. Fmeleus. H.J.. and Stewart. K., "Oxidation of the SiliconHydrides." J. (hem. Sc,.. (London). Part 1. 118211935).

14. Chu, T. T.. Szedon. J. R., and Gruber, G. A.."Silica Filmsby the Oxidation of Silane," Trans. Met. Soc. A /ME242.532 (1968).

15. Strater. K.. "Controlled Oxidation of Silane," RCAReview 29. 618 (1968).

16. Baliga, B. J.. and Ghandi. S. K.. "Growth of Silica andPhosphosilicate App!. Phys. 44. 990 (1973).

17. Kern. W.. and Heim. R. C.. "Chemical Vapor Depositionid Silicate Glasses for Semiconductor Devices."Electrochem. Soc. Ext. Abstr. 5, No. I. pp. 234-235(Spring 1968).

IN. Fisher, A.W., Amick. J.A., Hyman, H.. and Scott. J.H..Jr., "Diffusion Characteristics and Applications ofDoped Silicon Dioxide Layers Deposited from Silane(Sill..)" RCA Review 29. 533 (1968).

19. Fisher, A. W.. and Amick. J. A.. "Dif fusion Characteristicsof Doped Silicon Dioxide Layers Deposited from Pre -Mixed Hydrides," RCA Review 29, 549 (1968).

20. Strater. K.. and Mayer. A.. "Oxidation of Silane.Phosphine and Diborane During Deposition of Doped -Oxide Diffusion Sources." in Semiconductor Silicon.Haberecht. R. R.. and Kern. E. L., Eds..(The Electrochem.Soc.. Inc.. New York. 1969). pp. 469-480.

21. 1 okuyama, T.. Miyazaki. 1.. and Horiuchi. M.. "Low -Temperature -Deposited Glass for Passivation of Semicon-

ductor Devices." in Thin Film Dielectrics. Vratny. F.. Ed..11 he Electrochem. Soc.. Inc.. New York. 1969). pp. 297-326.

22. Kern. W . and Heim. R. C.. "Chemical Vapor Depositionof Silicate Glasses for Use with Silicon DeviCes - I.

Deposition Techniques." J. Electrochem. Sc.,. 117. 562

23. Kern. W., and Heim. R. C.. "Chemical Vapor Depositionof Silicate Glasses for Use with Silicon Devices - II. FilmProperties." J. Electrochem. Soc. 117. 568 (1970).

24. Kern. W.. and Fisher. A. W.. "Deposition and Propertiesof Silicon Dioxide and Silicate Films Prepared by Low -Temperature Oxidation of Hydrides." RCA Review 31,715 (1970).

25. Schlacter. M. M.. Schlegel. E. S.. Keen. R. S.. I.athlaen. R.A.. and Schnable. G. L.. "Advantages of Vapor -PlatedPhosphosilicate Films in large -Scale Integrated CircuitArrays." IEEE Trans. Electron Devices ED -17. 1077

(

26.S1c9h7n0a).bk. G.L. "Glass Passivation of Integrated Circuitsby Chemical Vapor Deposition of Oxide Films." IEEE1971 Intl Conv. Digest. pp. 586-587 (1971).

27. Shibata. M. and Sugawara. K.. "Deposition Rate andPhosphorus Concentration of Phosphosilicate GlassFilms in Relation to 0, Siff. + PH. Mole Fraction." J.Electroc hem. Soc. 122. 155 (1975).

28. Shibata, M.. Yoshimi. 1. and Sugawara. K.. "DepositionRate and Phosphorus Concentration of PhosphosilicateGlass Films in Relation to PH., Si fit + PH. MoleE raction." J. Electrochern. Soy. 122. 157 (1975).

29. Wong. J. and Ghezzo, M.,"Vapor Deposition of High P:O.Content Binary Phosphosilicate Films on Silicon." J.Ekctrochem. Soc. 122, 1268 (1975).

30. Wahl, G. "Chemical Vapor Deposition of SiO: and Boro-and Phospho-silicate Glasses by the Reaction of Siff,11:11. or PM with Oxygen." in Chemical VaporDeposition. Blocher, Jr., J.M., Hintermann, H.E., andHall, L.H., Eds. -Fifth Inel Conf.. The Electrochem.Soc.. Inc.. Princeton, N.J. (1975) pp. 391406.

31. Middlehoek. J. and Klinkhamer, A.J., sac), DepositionUsing Mixtures of Sill. -0: and A." in Chemical VaporDeposition, Blocher, J.M., Jr.. Hintermann, H.E.. andHall. L.H.. Eds., Fifth Intl Conf., The Ekctrochem.Soc.. Inc.. Princeton. N.J. (1975), pp. 30-42.

32. Kern. W. "Chemical Vapor Deposition Techniques forGlass Passivation of Semiconductor Devices." Pro, IEEE1975 National Aerospace and Electronics Conf.;NAECON 75 Record, pp. 93-100 (IEEE Cat. No.75CH0956-3-N A ECON).

33. Kern. W., et. al.. to be published.

34. Vossen. .1.1., Stephens, A.W., and Scnnable. G.L.."Bibliography on Metallization Materials and Techniquesfor Silicon Devices." Monograph '2, Thin Film Division.American Vacuum Society (1974).

35. Hammond. M.L. "Thin Film Preparation by ChemicalVapor Deposition." (Abstract). J. Vac. Sri, Technol. 10,268 (1973).

36. Hewing. W.C. and Fisk. R.. "Automation in CVD WaferProcessing.' Solid State Technology 18. No. I. 30(January 1975).

37. Mayer. A.. Strater. K.. and Puotinen. D.A. "Manufac-turing Techniques for Controlled Deposition and Applica-tion of Doped Oxides." Tech. Rm. A FML-70-TR-191(Septemher 1970).

38. Kern. W. "Apparatus for Chemical Vapor Deposition ofOxide and Glass Films." RCA Review 39, 525 (1968).

39. DeNicola, J.A. and Mastropiero. R.D.. "Design andMaintenance of High Purity Gas -Handling Systems,"Solid State Technol, 15. No. 2. 51 (February 1972).

40. Wollam, J.. "Equipment for the Chemical Vapor Deposi-tiim of Ultra -Uniform Silicon Dioxide Films." Solid StateTechnol.. 14. No. 12, 72 (December 1971).

41."Cold Wall Semicon Reactor." Circuits Manufacturing.14. 47 (November 1974).

42. Benzing, W.C. Rosier. R.S. and East. R.W."A ProductionReactor for Continuous Deposition of Silicon Dioxide,"Solid State Technol. 16. No. I I, 37 (November 1973).

43. Haschen. S.S., Evitts, H.C.. and Black. J.R.. "Large AreaSemiconductor Surface Passivation Production Refine-ment Program," DA -36 -039 -AMC -02280(E). Final Tech.Rrp. 11965). (Motorola. Inc.).

44. Barry. M.L. "A Continuous Process for the Deposition ofSilicon Oxide from the Oxidation of Silane." in ChemicalI also, Deposition, Blocher. J.M., Jr. and Withers. J.C.,Eds.. Second Ina Conf., The Electrochem. Soc.. NewYork (1970). pp. 595-617.

45. "Vapor Deposition Unit is Modular," Electronics. 47. 138(September 5. 1974).

46."Danish Firm offers Vapor System," Electronic- News.page 60. (June 5. 1972).

43

Structure andproperties ofthin metal filmsD.M. Hoffman

The properties of vacuum -evaporated thinmetal films are directly attributable to theconditions of preparation. Among theseparameters are residual atmosphere, ratesof evaporation and deposition, purity ofsource material and evaporant, substratetemperature, and cleanliness. Some of thereasons for variation of the properties offilms from those of bulk metals are dis-cussed as well as ways of minimizing theseeffects. Illustrations focus on metals usedin crystal coating, e.g., gold, chromium,and sliver.

THE SUBJECT of the structure andproperties of thin films fills manyvolumes" of learned papers. There havebeen a number of excellent reviews on thesubject. This paper presents a very briefoverview of the field as it pertains to thinmetal films of up to I micrometerthickness and specifically to those metalscommonly used in quartz crystal coating,e.g., gold, chromium, and silver. Bynoting the effects of depositionparameters on the structure andproperties of these films, the reader maydevelop some appreciation of the reason-ing behind the procedures developed forthe manufacture of thin-film devices.

Some of the desired properties of thesefilms are:

Good adhesion to the substrate undervarious conditions of temperature, at-mosphere, and stress

Stable electrical conductivity or resistivity

Bulk density-low porosity

Principal problem areas affecting theproperties of thin films are:

The substrate and interaction with it

The structure of the film

The density of the filmStresses in the film

The deposition parameters affecting all ofthe above areas are:

Substrate material

Film material

Substrate temperature

Deposition rate

Gas pressure

Gas composition

Adhesion

Adhesion involves the interactionbetween substrate and film interface. D.Mattox' and B. Chapman' have recentlywritten excellent discussions of the sub-ject which are briefly summarized here.The four types of adhesion are: (Fig. I):

Mechanical

I nterfacial

Intermediate layer

Interdiffusion

Mechanical adhesion occurs wheremacroscopic substrate roughnessprovides a mechanical locking of the filmand a larger surface area for bonding. Itcan be enhanced by mechanical abrasionsuch as polishing, lapping, or sand orglass blasting of the substrate. Chemicaletching or ion bombardment can be usedas well. Excessive roughness and selfshadowing' can, however, cause loss ofadhesion by producing uncoated areas orvoids and vacancies in the film.

In interfacial adhesion, the film andsubstrate meet at a well-defined interface.Obviously, any interference with thisinterface, or any contaminating layerwhether superficial dirt or an undesiredoxide layer will degrade the adhesion.Proper cleaning and maintenance of theclean surface until deposition is ofparamount importance. Many treatiseshave been written on the subject ofsubstrate cleaning, where there seem to beas many "correct" procedures as authors.Holland's6 excellent discussion of thesubject points out the use of solventcleaning and/ or chemical cleaning toremove organics; low power ultrasonicswith solvent or detergent; final vapordegreasing in iso-propyl alcohol which,interestingly, offers very nearly as cleansurfaces as glow -discharge cleaning.

It should be recalled that at 10-6 Torr onlyone second is required to form a

monomolecular layer of residual gas onthe substrate. The elapsed time betweencessation of cleaning and the onset ofdeposition must be very short in order toensure a clean surface. One obvioussolution is to work under ultra -highvacuum conditions where longer workingtimes are possible. A more generallypracticable method is to vacuum bake oruse a glow -discharge cleaning immediate -

////1/1////////1/ // I/ , /1/

/,//SOURCE

Mechanical

FILM

DIFFUSEDLAYER

7 /0/ /a. Wino/ ////// /

SUBSTRATE

Interdiffusional

FILM

Interfacial

FINAL LAYER

INTERMEDIATE LAYER

SUBSTRATE

Intermediate Layer

Fig. 1 - Types of adhesion: mechanical, Interdiffusional,Interfacial, and Intermediate layer.

ly prior to depositions. Neither of thelatter methods guarantees an "absolute-ly" clean surface because of the timelimitation described. A system may bebaked and purged in an inert gas tominimize residual gas components suchas 1120 or 02 which can cause undesirablereactions.

Since the properties of thin metal filmsare a function of the amount and kind ofcontamination present, an "absolutely"clean surface is not always desirable.Intermediate layer adhesion requires abonding layer between the film and sub-strate. This consists of compounds of thematerials with each other and/ or theresidual gases. A natural intermediateoxide forms between metals such as Al or

44

Cr and the oxygen in glass. Colbert' hasreported that metal films adheringstrongly to silica -containing substratescan be produced by deposition of verythin intermediate layers of metallic ox-ides, sulphides, selenates and phosphates,silver chloride, and magnesium fluoride.The oxides were obtained from an oxygenglow discharge of an evaporated material.A reactive evaporation, evaporation in a"high" partial pressure of a desired gas,could also be used.'' Intermediate oxidelayers can be attained by depositingoxygen active metals such as Ti, Cr, Mo,and Ta. These can then be coated withmetals adherent to them. This is the basisof the widely used Cr-Au, Ti -Au, and Ta-Au systems.

lnterdiffusion of the film and substrate iswidely used in the semiconductor in-dustry to produce contacts on silicon forvarious devices. All of the transitionmetals such as Ti, Zr, Hf, Nb, Ta, Cr, Mo,Fe, Ni, Co, etc., can be used, as can thenoble metals Pt, Rh, Pd, etc., other thanAu and Ag. The latter form undesirableeutectics with silicon. Diffusion duringdeposition can occur with proper choiceof substrate temperature. In mostoperations, however, a diffusion step isused after deposition.

Structure

Substrate temperature

At low temperatures, film structure

20

3000

2000

1000

depends on the bond character of the filmmaterial. Typical metals, e.g., Cu, Al,form crystalline layers even at liquidhelium temperature.

With increasing substrate temperatures,the size of the crystallites increases andnumber of lattice faults decreases. Theeffect of substrate temperature oncrystallite size has been measured byFleet" for nickel on glass or fused silica.Fig. 2 illustrates the change in averagesize from 60 A at 25 °C to 1000A at350° C.

Condensation rate

I hun ' has demonstrated, for 1000 A Crfilms, that grain diameter at 25`'C wasindependent of rate from 10 to 500 A/ sec(Fig. 3). At higher substrate temperature,grain diameter remained constant up to athreshold beyond which it began todecrease with increasing rate. Hepostulates that the high mobility surfaceatoms were buried before they couldreach ordered lattice sites. Similar resultswere reported by Campbellm for Au andSennett and Scott" for Ag.

For a fixed condensation rate and sub-strate temperature, the crystal size isdependent upon the film thickness. In-itially, an increasing crystal size is ob-tained followed by a gradual decrease toconstant average diameter of about 100 Afor Ni and Au (Fig. 4). Blakely,I2 shows

250 500 750 '000

SUBSTRATE TEMPERATURE C

Fig. 2 (abcve) - Crystallite size in 500 -Angstrom -thick nickel films as a functionof substrate temperature." Fig. 3 (right) - Variation of specific resistivity withsubstrate temperature and condensation rate.

400

Dorothy Hoffman, Thin Film Technology, RCALaboratories. Princeton, N.J., received the BS in

Chemical Engineering from Rensselaer Polytechnic In-stitute and the MS in Chemical Engineering fromBucknell University. After a year with the GeneralElectric Research Laboratory in Schenectady, she joinedthe International Resistance Company in Ph ladelphia,where as research engineer and later as head of processdevelopment she played a major part in the developmentof the evaoorated metal film resistor. In 1962 she joinedRCA as a member of the Technical Staff in the Astro-Electronics Affiliated Laboratory at the David SarnoffResearch Center. Since 1967 she has beer in charge ofthe Engineering Service Thin -Film Technology Group.working with evaporated films of many materials. Mrs.Hoffman, oast President of the American Vacuum Socie-ty, is a member of the Society of Women Engineers. and amember of the Engineers' Club of Philadelphia. She ispast President of the Engineering and TechnicalSocieties Council of Delaware Valley. She has publishedmany papers and holds several patents on evaporatedfilms.

Reprint RE -21-5-9Final manuscript received August 15. 1974.

100 -

100 200

10( OA CHROMIUM FILMS

Ct4aMBER PRESSLRE, I X106 MM I -g

UM'S OF 6/88 CONSTANT

300

CONDENSATION FATE IN A /SEC

400 500

that Au films deposited at 40°C consist ofgrains that extend from one surface to theother. This has also been observed in Alfilms.'

Annealing

Annealing may cause large changes incrystallite size. Blakely,'` observingcarbon -coated Au films at 600°C in anelectron microscope, indicated that twostages of aggregation occurred. The firstwas recrystallization to a stable grainstructure and then separation of grainsalong stationary grain boundaries. Thismay leave voids between somecrystallites. Grimp114 has observed anamorphous Au film between the islands.

If a film is annealed at a temperaturehigher than the deposition temperature,resulting grain sizes are much smallerthan those found in films deposited at thehigher temperature.

Residual gas atmosphere

Adsorbed gas atoms lead to decreasedsurface mobility and small grain size.Trapping of gas atoms results in porous,highly disordered films. Formation ofchemical compounds may occur at highersubstrate temperatures where gases ofgreater chemical affinity are present.

The effect of residual gas on highly activemetals may be illustrated by chromiumfilms. Reaction with 02 or H2O in the

300

X0et

zOl

200

N

ta

7.7

4co) - 100

4.3

04CC

4

residual atmosphere causes chromiumoxides to form. These segregate at thegrain boundaries and result in a hard filmthat has a lower electrical conductivity,more negative temperature coefficient ofresistivity, and a brownish opticaltransmission rather than the blue foundin "pure" Cr filMS.22.23

Density

Where, because of low substratetemperature, low mobility exists andresidual gases are present, a density lowerthan bulk is often obtained. The greatmajority of structural faults are concen-trated at grain boundaries. Many metalsexhibit a density maximum at substratetemperatures between 150° and 300°C.

A density maximum is observed forvariation in condensation rate.9 At verylow rates, the low nucleation rate andlong mean -free time of the surface atomspermit preferential growth and therebyrough films with relatively loose packing.With increasing rates the film density andsmoothness increase up to the thresholddescribed previously.

X-ray diffraction indicates that the densi-ty within crystallites is almost that of thebulk metal. The low density of films mustthen arise from voids in discontinuousfilms or high concentrations of im-purities, e.g., oxides, as well as vacanciesat the grain boundaries in continuousfilms.

o

/0A./44GArro'c

250 500 750

FILM THICKNESS ANGSTROMS1000

Fig. 4 - Crystallite size in nickel films as a function of film thickness."

Stress

Stress in thin films consists of two com-ponents:

I) Thermal Stress. Thermal stresses are basedon differences in thermal coefficients ofexpansion between film and substrate atdeposition temperature and temperature ofinterest. Actual measurements are difficultbecause of problems in measuring filmtemperature. Thermal stress varies widelywith substrate temperature and materialsinvolved. The effects can be large and eithertensile or compressive. They can be con-trolled in some cases by choice of film andsubstrate material. For metals on quartz, thethermal stresses in the film are tensile atroom temperature.

2) Intrinsic Stress, Intrinsic stress is believed toarise from film contamination and in-complete ordering of the structure duringgrowth. We can expect, therefore, a

dependence upon thickness, condensationrate, deposition temperature, residual gas.etc.

Average stress in thick films is relativelyindependent of film thickness.Measurements's on some 1000-A thickfilms have shown that stress does notappear to depend strongly on substratematerial.

Measurement of internal stresses in Aufilms on room -temperature quartz,measured by Kinbara16, indicates that thefilms consist of a 1000-A thick base layerof inhomogeneous stress and a super-posed region of uniform stress. Fordeposition temperatures between -150°Cand +200° C metals commonly deposit intension. With increasing substratetemperature, the deposition stress

decreases, often linearly, and goes tocompression at higher temperatures"(Fig. 5).

Some influence of the deposition rate onaverage stress has been reported forpermalloy" and copper's where the stressincreases with rate up to some saturationat rates of -70 A/ sec. No strong effecthas been observed by others over a 20%change.

Large stress changes have been observedin films prepared in vacuum systemswithout cold trap especially with anneal-ing, but the stresses in films deposited inlow -water -content systems are not verysensitive to moderate changes in gas

pressure. There is some evidence that thedeposition atmosphere affects film stress.Compression stress was found in Al filmsdeposited at over 104 Ton.° and in

46

IS -

2U

-O

cn0

to

cr

z0Z -O 1 0z

O

z05,

CLCL 5a.20

-10

- Iron

75 A Sec -I

Nickel

Copper6 A Sec -I

0 4 A Sec -I

Permolloy (WI

4 A Sec -I

Permolloy IP)

0 100 200 300 400

SUBSTRATE TEMPERATURE

copper films aged in air at roomtemperature." In both cases this wasattributed to oxide formation within oron the surface of the films.

Irreversible changes in intrinsic stressoccur upon an extended anneal. Latticeeffects are reduced and recrystallizationoccurs. Annealing data for goldI6 is

shown in Fig. 6. Widely differing resultsbetween x-ray and mechanical bendingdeflection measurements indicate a stressrelief with the grains but an increase instress in the highly disordered grainboundary areas.

Conclusion andrecommendations

We have seen that the following con-ditions are necessary for obtainingsatisfactory metal films:

I) Cleaning of substrate to provide properconditions for adhesion.

2) Maintenance of surface cleanliness untilfilm is deposited.

3. Provision of intermediate layer wherenecessary. This can be an oxide, a metal, orother nucleating agent.

The effect of deposition parameters onthe film structure are summarized below:

1) Crystallite size is independent of depositionrate up to 500 A/ sec. but increases markedlywith increasing substrate temperature.Grains generally are the thickness of thefilm.

2) Film density and smoothness increase withcondensation rate up to the thresholddescribed.

3) For many metals, a density maximum maybe obtained at substrate temperatures

e C

00

Gold (Pt

Said IX)

C o sor r

50 100 150

ANNUL INC 1 011PERATURE , C

700

Fig. 5 (left) - Average stress as a function of substratetemperature. Permalloy (P) is total stress measured al roomtemperature. Other data are intrinsic stress values.', Fig. 6

(above) - Stress annealing in gold and copper.' Shown aboveare X-ray (X) and circular -plate (P) experiments.

between I50°C and 300°C.4. The presence of adsorbed gases generally

causes smaller grain size and porous, highlydisorganized films. It also leads to films oflower density than that of bulk.

5) Annealing may cause grain growth whickisnot as large as that obtained by deposition atthe same substrate temperature.

6. Thermal stresses may be controlled byjudicious choice of film and substratematerial.

7. Intrinsic stresses appear to be independentof substrate material. They arise from filmcontamination and structural disorder dur-ing growth. Compressive stresses may occurdue to oxide formation at the surface orwithin the film.

8)Stress tends to decrease with increasingsubstrate temperature and increase with rateup to about 70 A/ sec. Stress relief within thegrains occurs with extended anneal while anincrease in stress occurs within the grainboundary areas.

In order to obtain films withcharacteristics of the bulk metal, it isnecessary to deposit onto a "clean" sub-strate held between 150°C to 300°C atrates of the order of 70 A/ sec. Smooth,maximum density film should result.These can be annealed to provide somestress relief.

On the other hand, a vast array of usefulfilms can be made by carefully controllingthe position, amount, and chemicalnature of the "dirt" or contaminationincorporated into the film.

References

I. Hass. G. (ed.); Physics of Thin Films, Vol. 1-7, (AcademicPress, N.Y.).

2. Maissel, LI; and Glang, R. (eds.); Handbook of Thin FilmTechnology, (McGraw-Hill; 1970).

750

3. Mattox, D.M.; "Thin him metallization of oxides inmicroelectronics," Thin Solid Films, Vol. 18, (1973) pp.176-186.

4. Chapman, B.N.; "Thin film adhesion," J. Vac. Sci.Technol.. Vol. I I (1974) p. 106.

5. Kern, W.; Vossen, J.L; Schnable. G.L.; "Improvedreliability of electronic devices through optimized coverageof surface topography." II th Annual Proceedings onReliability Physics (IEEE, New York, 1973).

6. Holland. L.; Properties of Glass Surfaces (Wiley; 1964).

7. Colbert. W.H.; Weinrich. A.R.; and Morgan, W.L.; BritishPatents Nos. 605 871.605 872, 605 873, 605474 (1948).

8. Fleet, S.G.; Mallard Res. Lab. Rep. 466 (1963).

9. Thun, R E.; "Structure of Thin Films," Physics of thinfilms (G. Hass, ed.). Vol. I. (Academic Press. New York1963).

10. Campbell, DS.; Stirland. DJ.; and Blackburn, H.; "Astudy of the structure of evaporated lithium fluoride," Phil.Mag. Vol. 7, (1962) p. 1099.

1 I. Sennett, R.S.; and Scott, G.D.; "The structure ofevaporated metal films and their optical properties," J.Opt. Soc. Am. Vol. 4011950) P. 203.

12. Blakely, J.M.; "Mechanical properties of vacuum -deposited gold films," J.Appl. Phys. Vol. 35, (1964) p.1756.

13. Palatnicl, L.S.; YaFuks, M.; Boiko, B.T.; and Pariiskii.V.B.; "Electron diffraction investigation of the substruc-ture of fine aluminum condensates with the Inicrobeam*method," Friz. Metal, i Metalloved. Vol. 11 (1961) p. 864;Phys. Metal Metallog. (USSR) English Trans!. Vol. II (6)(1964) p. 44.

14. Grimpl.M.L.; McMaster, A.D.; and Fuschillo, N.;"Amorphous oxide layers on gold and nickel filmsobserved by electron microscopy," J. App!. Phys. Vol. 35(1964) p. 3572.

15. Hoffman, R.W.; "Mechanical propenies of thin condensedfilms," Physics of Thin Films (G. Hass and R.E. Thun,eds.; Vol. 3; Academic Press. New York; 1966).

16. Kinbara, A.; Oyo Butsuri Vol. 30, (1961) p. 496.

17. Weiss, G.P., and Smith, D.O.; "Isotropic stressmeasurements in permalloy films," J. Appl. Phys. Vol. 33(1962) p I166S.

18. Horikoshi, H.; Ozawa, Y.; and Hasunuma, H.; "Stress inevaporated copper films." Japanese J. of Applied Phys..Vol. 1 (1962) p. 304.

19. Murback, H.P.; and Wilman H.; "The origin of stress inmetal layers condensed from the vapour in high vacuum."Proc. Phys. Soc. (London) B66 (1953) p. 905.

20. Halliday. J.S.; Rymer, T.B.; and Wright. K.H.R.; "Anelectron -diffraction examination of thin films of lithiumfluoride and copper prepared by vacuum evaporation."Proc. Roy. Soc. (London) A225 (1954) p. 548.

21. Hoffman. D.M.; and Cows, M.; J. Vac. Sci. Technol., tobe published.

22. Hoffman, D.M.; and Rieman, J.; "Resistance -temperature characteristics of evaporated chromiumfilms." Trans. of 4th National Symposium on l'UCLIUMTech. (1957; Pergamon Press).

23. Hoffman. D.M.;and Riseman. J.; "Evaporated chromiumfilms on hot substrates." Trans. of 6th .Valium! Sym-posium on Vacuum frch.. (1960; Pergamon Press).

47

Engineer and the Corporation

Technicaleducationin RCA:current statusfuture directionsDr. W.J. Underwood

A series of studies conducted in 1973 and1974 has shown that RCA's internaltechnical education program reaches aneffective cross-section of engineers and isuseful to them in their jobs. In the future,the engineers would like the education tobe more applied, shorter, and more self -regulated. The direction of the internaleducation program has already beenaltered to be responsive, and possible newfuture directions involving a larger per-sonal development concept for engineersare taking shape.

2

RCAhas long been active in offeringin-house training to its technical com-munity. During -hours programs, after-hours programs, formal courses,seminars, colloquia, "film theaters,"workshops-all have played a part in theeffort to maintain a viable technicalcompetence. The primary organizationalresponsibility for such activity rests withthe Training group in IndustrialRelations.

In 1967, an adjunct unit was establishedon RCA's Corporate Engineering staff toprepare and disseminate formal technicalcourses in video tape format. Because ofits corporate position, its exclusivetechnical focus, and its video tape format,the new unit could attract and makeavailable outstanding teaching talent tonumerous geographically separatedengineering departments. It functions,therefore, as a cooperative effort betweenthe technical and Industrial Relationscommunities. Descriptions of theeducational technology of this unit havebeen published elsewhere."

Figures 1, 2, and 3 summarize the unit'sactivity in terms of number of courses,enrollments, and number of classes.

18_

16

14 _

12 _

2

0

68-69 69-70 70-71 71-72 72-73 73-74 74-75

YEA

Fig. 1 - Number of new courses added to the video tape inventory by year. (Remakes andrevisions are not indicated.)

1800

1600

1400

1200 _

1000 _

- A

O -800 _

CC

w 600

400

200 _

68-69 69-70 70-71 71-72 72-73 73-74 74-78

YEARS

Fig. 2 - Number of enrollments by academic year.

110_

100

90 _

80

70

UJ 60

a50

40 -

30.

0 I III68-69 71-72 73-74 74-7569-70 70-71 72-73

YEARS

Fig. 3 - Number of separate classes by academic year.

48

The general impression regarding thework of the Corporate EngineeringEducation unit over the eight years of itsexistence is favorable. However, in 1973and 1974 a series of studies were com-pleted to probe in more detail the unit'srelevance and effectiveness. Severalfindings from those studies have hadconsiderable impact internally and mayhave importance elsewhere. They aredescribed in this paper, and theirprobable effect on the future is discussed.

Job application

An initial study sought to discover theextent to which engineers applied theknowledge and skills gained from thevideo tape education program. Question-naire responses were obtained from 135scientists, engineers, and technologists.These respondents had completed a videotape course six to ten months prior toreceiving the questionnaire, allowing thatlength of time for job application. Theywere asked to indicate the extent of jobapplication and to provide an example ofit or to explain why no or very littleapplication had occurred.

A very conservative interpretation of theresponses indicated that 64% had madesubstantial use of the course material intheir work. Typical of examples givenare:

"the Fourier Transform was used topredict the output purity of amicrowave tube. Substantial circuitsimplification and cost reductionswere achieved."

"1 design power supplies for air-borne and space applications. Thecourse helped me to utilize digitalcontrol circuits in switchingregulators."

Another 14% indicated that they ex-pected to use the course material in theirwork in the near future, but had not doneso in the time interval surveyed.

The number reporting job application isgratifying. It becomes even more signifi-cant when the following are taken intoaccount:

Job relevance is not a matriculation require-ment.

The reasons most frequently given for lack ofjob application were:

- Took the course for broadening into anew field.

ENROLLMENTS POPULATION40).. 2988

64

61

58

55

52

49

46

43w

40

LEGEND

Rang*

Middle 3:4

- Mean

37

34

31

28

25

22

Fig. 4 - Age distribution of course enrollments and a relevant engineering population.

100

90

80

70

60

50

40

30

20

10

0

BACHELOR

ASTER

POPULATION (2.750)

ENROLLMENT (400)

roR

Fig. 5 - Distribution of educational levels of course enrollments and a relevant engineering population.

Table I - Percentage of course enrollment avid a relevant engineeringpopulation by performance ratings.

0111/17

Lower I/3 Middle Vj tipper 'A

Enrollments (70) 4

Population (400) 5

60

6.2

- Took the course as prerequisite foranother which is applicable.

- Was applicable, but I subsequentlychanged jobs.

- Not enough time has elapsed.

It is concluded from the survey that mostparticipants select courses in the expecta-tion of job relevance and are able to findsignificant application to their jobs.

Enrollment distribution

A second study was undertaken todetermine what segment of the engineer-ing population at RCA enrolls in thevideo tape courses in respect to age,degree, and performance. The results aresummarized in Figs. 4 and 5 and Table 1.These comparisons show that dis-

49

tributions of age, degree, and perfor-mance rating among enrollments are verysimilar to those in the population. Weregard this as the optimum condition.

General study

A third and rather extensive study soughtanswers to several questions amongwhich were these: What methods do engineers rely on to

remain technically viable and which of theseto they believe are most effective?

What factors make it difficult for engineersto remain technically viable?

What changes in our in-house educationprogram would be helpful?

About 300 engineers and engineeringsupervisors from seven differentengineering departments were in-terviewed for two hours in groups of fourto six. Respondents were judgmentallyselected for a population of 2,000 as beingrepresentative.

Table II answers the question concerningupdating methods used by engineers.Preliminary interviews identified ten

Table II - Updating methods.

methods as most prevalent. These tenmethods were then ranked by the respon-dent groups-first in terms of most fre-quently relied on and then in terms ofmost effective.

Technically challenging work, informaldiscussions with colleagues both insideand outside the company, and readingwere ranked one, two, three, respectively,against both criteria. Education in theform of seminars, lecturers, meetings, andformal courses was next in effectiveness.However, learning from vendor!eustomer contact was reported as usedmore frequently than education.

Table HI shows a rank ordering of factorswhich make is difficult to maintaintechnical viability.

Lists of inhibiting factors have been theobject of several studies. Factors locatedin the job, in the organization, and in theperson have been identified.' The factorsranked by respondents in this study placejob pressure highest. They define thisprimarily as a lot to do in a given timewhich results in a "rushed" feeling. It is

Most relied on

Challenging workInformal discussionReadingVendor/ customer contactFormal coursesSeminar/ lecture/ meetingCompetitor analysisDesign reviewTrade show/conventionPlanned job rotation

MOST

LEAST

Most effective

Challenging workInformal discussionReadingSeminar/ lecture/ meetingFormal coursesPlanned job rotationVendor/customer contactDesign reviewCompetitor AnalysisTrade show/convention

c;oe6,0\500-ee`.

62,0 4k.s pe 0otc..,`Vase ,4,5\°

.,,%kes ,;,c, for ke el -AA:, vizr* oe -5 ce x.c.

...-P'' ON 02' Ile' el\ 11G t

in 2' 4.12'....keZik1 C) 2N5 ,a erg

' va. A 1 - et° ek 0.

-o0.9 0.\e"no0160 0\5 Pkkke

-0 op.lt- t,tc- ti.e c,,(c xose..5,0o

.,5 oVve coke oot oi? el'. Goct,

,s \ Go020 \,4- zoicsk112fiziak2:so,%\',- 0

k20(9.:,0e1-e6 Y1/42:1c,71.0kcP\cOlk'vec

lk Go"09V. .,eC*kse .S at< \ P CaSe04`:-

s, 4\'' c"> l'O- CS e e

.. r .S - 0e A sO _ tIO kSk NIC' )Ns' cAZ , Oe ,,,?_ot.oqe

ko ,ocycok."-

014 co

QS, cO, .cxe sg,V,

e%bkket",

\c, Pkojcoeco'(ikli.ee.2012,se_cvaosvc,,,oi

.\\c, <1.vo e\I:\tezelfep\ec.::,:v`c.

cre sc5°P.: Se5k2Pit,iS\c'

<4 eSY

dell\ 2N1V%e° zc*1-peCtcc*ee(5%)

\e.112#

e\e'k2P ,c6cc°-\\se

Table III - Inhibiting factors

1(>ti I Job pressure

Lack of reward

Outside interests

Lack of organizational importance

Technically non -challenging work

Lack of adequate formal courses

Personality

Lack of adequate resources

Complexity of the technology

LEAST Age

inhibiting due to "no time to study"and/ or "too tired to study." Job pressurewould probably be classified as anorganizational factor as would lack ofreward which placed second. Reward wasdefined as both tangible (salary increase,promotion) and intangible (compliment).Ranked third is a more personal factor, apreference to devote time to family,community, hobby, etc., rather than totechnical updating activities. Fourth isclearly an organizational factor, lack oforganizational importance. This factorrefers to tangible signs of importance orlack of importance. It was aptly put byGray,6, "...the engineer gets a toughhassle if he...asks for an hour or two aweek off to take a course...gets his headchopped off if he can't travel right in themiddle of exams...is castrated if he triesto take a day off to study for finals..." Infifth position is the degree of technicalchallenge in the job. Respondents were innear unanimous agreement that jobpressure is the most inhibiting factor andthat age is the least inhibiting factor.

Responses concerning how our in-houseeducational program could be more help-ful were especially enlightening. The mostfrequent recommendation and the onemost emphatically expressed concernedtheory versus application. Therespondents urged us to

"stop competing with the colleges andmake your courses highly applied."

They explained further,"Use just enough theory to makeapplication intelligent, but concentrateon practical examples of actualapplications-focus on how -to -use -it."

We were accused

"your courses now emphasize theorywith application used to help unders-tand the theory. Turn it around; in thatway, you would fill a gap not alreadyfilled by the colleges. Courses should betailored specifically to RCA'sparticular needs and practices."

Another strong recommendation wasmade concerning the length of courses.Respondents expressed considerabledifficulty attending a class schedule oftwelve to fourteen weeks. Much moreacceptable would be shorter courses ofsomething between one and five weeklysessions. A related recommendation wasfor split time. Most respondents did notprefer either all company time or allpersonal time.

There were several requests for someform of self -study material. The possibili-ty of an individual being able to choosehis own time and place for study wasappealing.

Future directions

Several conclusions drawn from theabove studies will guide our direction forat least the next five years.

We will continue the video-tape format. It isan acceptable, cost effective, and flexiblemedium. At this time, we use black -and -white format on one inch, open reels. We seeno need for color. We do expect to convertto three -quarter -inch cassette.

We will give much greater emphasis toapplication. This process has already com-menced, and we have discovered thatemphasizing application requires a verydifferent and much more difficultprocedure.-An advance field study is made among

representative samples of potentialstudents and of informed people workingin the subject. This effort is to determinehow a subject should be focused, struc-tured, slanted, etc., to make it mostrelevant and useful to the design engineer.Also, this search locates examples ofapplications and potential instructors.

-Syllabus preparation is more often acommittee project than an individual one.

-Team teaching is used more often.-More time is devoted to student exercises,

especially "hands on" experiences wherepossible.

Instead of concentrating on just the "bigwinners", we are searching for four- to eight -session "minicourses." It has been possibleto pull sections from long courses, and withsome filling, produce useful educationalmodules.

A new experimental project, termedEngineering Monographs, is beingdeveloped as a series of short tutorials onsingle concepts (e.g., Fourier Transform) foruse as a refresher or as a reference. TheMonographs, which consist of an audiocassette with printed visuals and a writtensupplement will be stocked in companylibraries or at other central points inengineering departments. An engineer canlisten to the audio while following thevisuals by which the essential characteristicsof the subject will be introduced. The writtensupplement can then provide additionaldepth or breadth. At present, this is ouranswer to requests for self -study material.More elaborate self -study is under con-sideration.

Engineering departments are being en-couraged to establish a regular program ofseminars. These are envisioned as groupdiscussions stimulated by an expertwho can be either from within RCA or theoutside. The object is to enhance colleaguediscussion-one of the more effective up-dating methods.

Consideration is being given to a programfor granting Continuing Education Units(CEUs). The primary purpose would be togive continuing education more stature inthe engineering organizations.

These directions are a change from ourpast. They are responsive, we believe, towhat we learned from our series ofstudies. It there's a message here foreducators, it must be that educators canbecome obsolete. Our constituents will letus know about it, if we ask.

Engineering management have been mostinterested in the results of the studies. Insome aspects, their expectations wereconfirmed; in others, they were surprised.Overall, the results have tended to in-crease management's commitment toprovide opportunities sufficient to insuretechnical depth, breadth, andcurrentness. In some instances, newprograms were commenced (e.g., aregular technical seminar series). Inothers, more extensive use was made ofinternal education courses. One divisionappointed an engineering manager ateach of its several locations to beresponsible for insuring effective use oftechnical education, both internal andexternal.

This increased commitment has openedthe door for consideration of a largerpersonal development concept than justeducation. An attempt is being made tointegrate a variety of activities that nowoperate rather independently, e.g.,technical libraries, professional societyactivities, internal recognition andachievement award programs, publica-

Dr. William J. Underwood, Director. EngineeringProfessional Programs. Corporate Engineering Staff.Cherry Hill, N.J., manages a group of activities havingcorporate responsibility for technical education.technical publication, and technical informationservices. His background combines engineering andbehavioral science. He received the BSEE from WestVirginia University and the MEd in group dynamics andthe EdD in organizational studies from Temple Universi-ty. His previous position in RCA concerned managementand organization development in corporate IndustrialRelations. He hold membership in IEEE, ASEE, ASTD,American Psychological Association, and is a chartermember in the International Association of AppliedSocial Scient sts.

Reprint RE -21-5-20Final manuscript received June 24, 1975.

Copyright 1976 by the Institute of Electrical and Elec-tronics Engineers. Inc. Reprinted, with permission, fromIEEE Trans. on Education, Vol. E-19, No. 1, February1976.

tion and presentation activities, andeducation programs. These share thecommon aspect of being useful to main-taining technical viability. It appears thata gain in synergy is possible if eachactivity can be effectively related to theothers. Experimentation will be requiredand management support will make thispossible. Such an integration, if it can beachieved, could be the most importantfuture direction of all.

References

I. Biedenbach. J.M.; "Continuing engineering education byvideo tape," Journal of Continuing Education and Training(May 1971) pp. 23-25.

2. Biedenbach. J.M.; "Continuing engineering educationutilizing video tape," Engineering Education (May June1971) pp. 886-7.

3. Biedenbach, J.M.; "Industrial video tape applications tocontinuing engineering studies programs," IEEE Trans-actions on Education (Nov 1970) pp. 186-9.

4. Biedenbach. J.M.; "The RCA continuing engineeringeducation program," RCA Engineer (Oct/ Nov 1969) pp.186-9.

5. Kaufman, H.G.; "Obsolescence and professional careerdevelopment." AMACOM, 1974.

6. Gray. I.; "Book review: Obsolescence and professionalcareer development," IEEE Engineering ManagementSociety Newsletter, (May, June 1974) p. 29.

51

02950 50 kW A2950 500 kW 02975 100 kW

Tungsten -matrix cermolox tubesR.E. Byram J.T. Mark

The development of long-range deep -search radar has created a need for power tubes capable of operation at pulselengths up to several milliseconds. To meet this need, RCA is developing a series of long -pulse gridded tubes usingtungsten -matrix cathodes for improved efficiency and ruggedness, as well as reduced power demands. One application forthese tubes is in phased -array radars where many tubes are operated at moderate power ratings: the necessary phaseshifting is accomplished at a lower power level.

TO MEET THE NEED for long -pulsetubes, RCA is developing a series ofgridded tubes using tungsten -matrix (ortungsten -dispenser) cathodes.' Com-pared with the nickel -matrix -oxide andthoriated-tungsten types now in use, thenew tubes represent improvements inthree areas: better efficiency, reducedpower requirements, and increasedruggedness. The new tubes can producepulses from I to 5 ms in duration atnormal duty, and from 10 to 20 ms atreduced duty.

Three tube types are presently in develop-ment for pulse -modulator service:

The RCA A2950 is rated for 50 kW of peakpower input,

The RCA A2975 for 100 kW, andThe RCA A2960 for 500 kW.

Advantages oftungsten -matrix cathode

The key to long -pulse operation in thesetubes is the cylindrical tungsten -matrixcathode impregnated with bariumaluminate. This construction offersseveral advantages, particularly inmilitary application:

1)Arc-resistant operation at high pulsecurrents,

2) No pulse breakup

3)No pulse droop,

4) Very long life at moderate temperatures,and

5) Rugged construction.

Arc resistance andno pulse breakup

The tungsten -matrix cathode (Figs. I and2) has a smooth, machined metallic sur-face. This smooth surface reduces high -voltage gradients at the cathode surfaceand provides for excellent resistance toarcs.

The metallic emitting surface has a verylow internal resistance. It is free of self -heating effects and does not form aninterface layer. These factors preventinternal arcing (and possible gas evolu-tion) at high pulse currents; theyalso eliminate the pulse breakupsometimes experienced in the con-ventional nickel -matrix -oxide cathodetubes with age. Furthermore, unlike thenickel -matrix -oxide cathode, looselysintered particles cannot break off andcause interelectrode shorts. Also, thecathode coating cannot peel, which is aproblem with sprayed cathodes.

No pulse droop

The pulse -emission and dc -emissioncapabilities of the tungsten -matrixcathode are essentially the same. Thepulse length and emission level are thusnot limited by the cathode, but by theallowable dissipation of the other elec-trodes in the tube. For this reason, nodroop in emission occurs with long -pulseoperation.

Reprint RE -21-5-11Final manuscript received December 2. 1975.

Long life

The tungsten -matrix cathode is notdamaged by gas -ion bombardment, nor isit affected by electrons returned to thecathode due to transit -time effects at thehigher frequencies. Provided thetungsten -matrix cathode is not operatedat too low a temperature, it is less prone topoisoning by residual gas than oxidecathodes because barium and bariumoxide are constantly replenished from thebarium aluminate impregnate which fillsthe entire volume of the tungsten -matrix.

A cathode brightness temperature of1050°C is high enough to prevent poison-ing by residual gas, and yet is a moderatetemperature for a tungsten -matrixcathode. At this temperature, cathodeemission remains constant and providesfor a very long tube life, even underconditions of high cathode loading. Usingtest diodes at cathode current densities ashigh as 10 A/ cm2, constant emission forthousands of hours has been reported,and verified, at RCA.

In the new series of RCA tubes, the dccathode -current density is held to 200 to250 mA/cm2 of cathode area. This figureis double the usual rating for a nickel -matrix -oxide cathode, but it is moderateindeed for a tungsten -matrix cathode. Atthis level of dc emission, and at thecathode operating temperature beingused, the life expectancy of the cathodeitself is 50,000 to 100,000 hrs. RCAtraveling -wave tubes using tungsten -matrix cathodes are exbiting 10,000 to

52

Weld

Tungsten MatrixBarium Alummate

Moly RutheniumBraze

Molybdenum

Sintered_Molybdenum

Molybdenum -

Cu Clad Kovar -

Tantalum

Tungsten/Ni Bra:e

A1203

Tungsten Rhenium Heater

Molybdenum

Fig. 1 - Cross section of the heater -cathode assembly used in the RCA A2960. Thecathode is a cylindrical tungsten matrix impregnated with barium aluminate.The heatdam, which supports the tungsten -matrix cathode, is made of molybdenum. Theheater spool and the heater -spool lid are of tantalum. These are the parts that providemechanical support for the tungsten -rhenium heater and the associated high -alumina heater retainer ceramic. Tantalum and molybdenum parts are used ratherthan the more conventional nickel parts because of the operating temperature of thetungsten -matrix cathode (about 1050°C brightness temperature).

15,000 hrs of life, and one Europeanmanufacturer of gridded tubes has

reported cw dynamic life 'in excess of16,000 hrs-and the tubes are stilloperating.

Rugged construction

Many of the advantages that have beenoutlined for the tungsten -matrix cathodealso apply to the thoriated-tungstenfilamentary cathode. The tungsten -matrix cathode is much more efficient,however, as only one-fourth to one-halfthe filament power is needed forequivalent emission levels. The tungsten -matrix cathode is also much more ruggedthan the thoriated-tungsten filament, andis not susceptible to breakage underanticipated environmental conditions orin shipment. Tubes of the A2950 typehave been vibrated to 5- and 10-g levels,with random inputs to 5000 Hz, forperiods up to 4.5 hours with no shorts andno damage, and with very low noise levelsindicated.

Grid -screen construction

All three of the new RCA tetrodes are ofthe proven RCA Cermolox construction,with the grids and screens simultaneouslyproduced by electrical discharge cutting.The result is perfect alignment of the gridand screen wires. The RCA A2960 makesuse of molybdenum for the grid andscreen wires, while the A2950 and A2975have grids and screens of a copper alloy.

Moly RutheniumBraze

Tungsten MatrixBarium Nominate

Moly RatheniumBraze

Molybdenum ---Rhenium

Weld -

Nickel -

Molybdenum

LCu Clad Koyar

- Tungsten/Ni Bran

A1203

- Molybdenum

Fig. 2 - Cross section of the PCA A2975 cathode: the RCA A2950cathode is of similar construction. In both of these types, because of thesequerce of manufacturing operations. which requires the entireheater -cathode assembly to be exposed to a hydrogen -atmospherebrazing furnace, no tantalum is used: all parts associated with thecathode are of molybdenum or a molybdenum -rhenium alloy.

Robert E. Byram, Senior Eng.. Regular Power ProductsDesign and Applications Engineering. Solid State Divi-sion. Lancaster. Pa . received the BSEE from M I T in1949 and joined RCA in Lancaster that year He receivedthe MS in Physics from Franklin and Marshall College in1973 Mr Byram's in teal engineering assignment was thedesign and development of a beam -power tetrode for usein mobile communications systems He was alsoresponsible for the design of a very -high -mu triode usedin a shunt regulating circuit to regulate the 27.000-Vsupply in early color tv receivers A special anode designused in this tube was later granted a U S Patent MrByram has had several years experience in applicationsengineering of both power tubes and gas tubes Atpresent. he is engaged in design and developmentengineering of a series of gridded power tubes usingimpregnated tungsten matrix cathodes Mr Byram is amember of Sigma Pi Sigma and is a registeredprofessional engineer in the State of Pennsylvania

Authors Byram (left) and Mark.

John T. Mark, Eng Ldr , Regular Power Products Designand Applications Engineering. Solid State Division.Lancaster. Pa received the BS in Radio Engineeringfrom Valparaiso Technical Institute in 1950 and joinedRCA in Lancaster that year Since joining RCA Lancaster25 years ago Mr Mark has developed a vast amount ofexperience in the design. fabrication, and testing oftransmitter and test equipment, regular- and super-power tubes, and ultra -high -vacuum equipment, and inlaser tube design. and systems and applicationstechnology for gas lasers and power tubes In 1958 MrMark was assigned the responsibility for the MatterhornProject As. the Project Engineer. he was responsible forthe direction and control of the C-Stellerator VacuumSystem. a thermonuclear fusion research machineLater, he served in the same capacity for the Mark I

Aerospace Systems Environmental High VacuumChamber. the world's largest simulator. Mr Mark haseleven U S patents and has written numerous papers. Heis a member of the Standards Committee and theprogram committee for the American Institute ofPhysics. the Committee for Establishing a Glossary ofTerms for the American Vacuum Society, the IEEE, andthe International Association of Vacuum Scientists andTechnicians

53

Table I - Typical operation of A2950 tube in pulse modulatorservice: rectangular waveshape pulses, duty factor of 0.01, and apulse width of 1000 microseconds.

DC plate voltageDC grid -No. 2 voltageDC grid -No. 1 voltagePeak positive grid -No. 1 VoltagePeak plate currentDC plate currentDC grid -No. 2 currentDC grid -No. I currentLoad resistanceUseful dc peak power output at

peak of pulse

3,000 V800 V

-120 V25 V10 A

100 mA10 mA50 mA

180 ohms

18,000 W

Table 11 - Typical operation of A2975 tube in pulse modulatorservice: rectangular waveshape pulses. duty factor of 0.01, and apulse width of 1000 microseconds.

DC plate voltageDC grid -No. 2 voltageDC grid -No. I voltagePeak positive grid -No. I voltagePeak plate currentDC plate currentDC grid -No. 2 currentDC grid -No. I currentLoad resistanceUseful dc peak power output at

peak of pulse

4,000 V1,000 V- 250 V

320 V20 A

200 mA25 mA35 mA

100 ohms

40 kW

Table III - Typical operation of A2960 tube in pulse modulatorservice: rectangular waveshape pulses, duty factor of 0.01, and apulse width of 1000 microseconds.

DC plate voltageDC grid -No. 2 voltageDC grid -No. I voltagePeak positive grid No. I voltagePeak plate currentDC plate currentDC grid -No. 2 currentDC grid -No. 1 currentLoad resistanceUseful dc peak power output at

peak of pulse

20,000 V1,500 V- 400 V

50 V25 A

250 mA15 mA20 mA

680 ohms

425 kW

Table IV - Typical operation of A2960 tube used in a pulsed rfamplifier.

Frequency 4(K) MHzPulse width 2 msDuty factor 5 %DC plate voltage 11,000 VPulse grid -No. 2 voltage 1,500 VDC grid -No. I voltage -250 VDC plate current during pulse 7.7 ADC plate current 0.39 ADC grid -No. 2 current 7 mADC grid -No. I current 19 mAOutput circuit efficiency 85 %Drive power at peak of pulse 2.3 kWUseful power output at

peak of pulse 40 kW

Typical operationas hard -tube modulatorIn pulse -modulator service, the A2950,the smallest of the three tubes, is rated for5000 V of plate voltage and 10 A of peakplate current at 0.01 duty and a 1 -ms

pulse width. Table I summarizes thetypical operating characteristics for thistube used as a pulse modulator switching18 kW of peak power in the load.

The A2975 is rated for 5000 V of platevoltage and 20 A of peak plate currentunder the same conditions. Table II liststhe typical pulse -modulator servicecharacteristics with the tube switching 40kW peak power in the load.

The largest tube, the A2960, is rated for20,000 V of plate voltage and 25 A of peakplate current, again at 0.01 duty and 1 -mspulse width.

Table 111 gives typical pulse -modulatorservice characteristics for the A2960 . Inthis case, a peak power of 425 kW is beingswitched in the load.

Life resultsVaried life experience has been ac-cumulated for the A2950 and A2975 inpulse -modulator service and in rf service,and for the A2960 in pulse -modulatorservice.

As a hard -tube modulator, the A2950 hasbeen run for 3000 hrs at a peak platecurrent of 13 A with 1 -ms pulses at 0.01duty. In rf service, the A2950 has beenoperated for 18,000 hrs in 30 -MHz cwservice; dc plate current was 1 A andpower output was 1 kW, both of whichare double the values obtained from anickel -matrix -oxide tube of equivalentsize. The A2950 has also been operatedover 5000 hrs in the visual IPA stage of acommercial tv transmitter in actual tvbroadcasting use. The A2950 heater hasbeen cycled on and off for 20,000 cycleswith no heater failure.

The A2975 has been tested at 2.5 -kWpower output at 100 MHz for 5000 hrs.

The A2960 has been operated in pulse -modulator service for 4500 hrs under avariety of conditions that simulated ran-dom pulse -width radar. Pulse widthvaried from 1 to 4 ms and peak currentsfrom 10 to 25 A. The A2960 heater wascycled 12,000 times without failure.

Life tests of all three tubes are continuing.

1300 MHz50 MHz Bandwidth Id 1 dB Double TunedScreen Pulsed Cathode Drive2000 Microsecond Pulse, 5% Duty

80 VI/

A2950500 W kW

A2950

DC Plate Voltage 1600 2200 VPulsed Grid No. 2 Voltage 800 800 V

DC Grid No. 1 Voltage -30 -26 V

DC Plate Current During Pulse 2.94 4.33 ADC Plate Current .147 .217 A

DC Gr No. 2 Current 003 005 A

DC Grid No. 1 Current 001 033 \Fig. 3 - Two A2950 tubes used in pulsed broadband -r1 -amplifier application.

RF operation

In addition to pulse -modulator service,which has been stressed up to this point,this new series of tubes is also useful inpulsed-rf-amplifier service. Fig. 3 shows achain of two A2950 tubes providingbroadband operation similar to that re-quired in one section of a radar systemthat has been under consideration.

Table IV shows typical operation of theA2960 in pulsed-rf amplifier servicewhere it provides 40 kW of pulsed poweroutput at up to 400 MHz with a 2 -mspulse width.

other operating conditions arepossible, of course; however, limits onallowable inputs and plate currents atvarious duty factors and pulse widthsmust be observed.3'4'5

Conclusion

Tungsten -matrix cathodes can be

successfully utilized in gridded tubes.Tubes with this type of cathode providemany advantages in both long -pulse andhigh -voltage service for military radarsystems.

References

I. A general description tit the tungsten -matrix cathode maybe found in the following references:Levi, R.; "New Dispenser Type Thermonic Cathode," J.Appl. Phi's. Vol. 24, (Feb. 1952) p. 233; and "ImprovedImpregnated Cathode," J. App! Phys. Vol. 26 (May 1955)p. 639.Kohl, W.H.; Handbook of Materials and Techniques forVacuum Devjces (Reinhold Publishing Corp.; 1967)Haas, G.A. and Thomas, R.E.; "Work Function Distribu-tion Studies of Pressed Matrix Cathodes," J. Appl. Phys.Vol. 38 (Sept. 19671 p. 3969.Additional information is contained in an earlier version ofthis paper presented at the I Ith Modulator Symposiumsponsored by the IEEE Advisory Group on ElectronicDevices, New York, September 18, 1973.

2. RCA Application Note AN -4887 "Tungsten MatrixCathodes for RCA Cermolox Tuber.

3. Developmental Data Sheet-RCA Type A2950.4. Developmental Data Sheet -RCA Type A2960.

5. Developmental Data Sheet-RCA Type A2975.

54

Adapting structuredprogramming toexisting systems and languagesDr. P.G. Anderson

Structured programming has been suggested as one of the most promising techniques forimproving computer program reliability. Its disciplined approach to coding control -flowstructure produces clear, understandable program elements that can be combined toconstruct large programs. This approach contrasts sharply with the standard practice ofemploying open-ended control figures that permit program flow complexity to growunchecked. The Structured Programming Adaptor (SPA) introduces a workable approachto applying the advantages of structured programming to a wide variety of existingsystems and languages.

TRADITIONAL computer program-ming techniques are becoming less andless effective for coping with the in-tricacies of very large, complex systemrequirements. Too often, the softwareproduced is full of errors and impossiblefor anyone but the original author tocomprehend (and then only barely). Eachpiece of these complex programs is in-timately tied to every other piece so thatchanges have unforeseen secondaryeffects, thereby causing the need for otherchanges, with more unforeseen secondaryeffects..., ad infinitum. Thus, instead ofyielding to continual refinement and im-provement, these programs simply growin complexity until they reach a point atwhich there is no alternative but completereplacement.

A primary cause of the tendency towardcomplexity is the use of ad hoc controlsequencing in which the logic flow jumpseverywhere, almost randomly, making itnearly impossible to understand the pur-pose of any one section of the program.'Recently, software researchers haveidentified a key distinction between suchcomplicated software and easy -to -understand, high quality software. Theyhave found that better programs areconstructed of nested program sections,with each section having one entry point,one exit point, and a very simple controlflow. This is called structured program-ming.

There are several ways to take advantageof the benefits of structured program-ming. One approach is to use a modern,high-level programming language (e.g.,ALGOL and PL/ I), with the control

structures of structured programmingbuilt in. However, the real-time process-ing requirements of many system projectscan make such languages an impossibleluxury. Then too, the computer selectedmay not have these languages available,e.g., a modified first- or second -generation computer. Some systems canbe programmed only in a dialect ofFortran or an assembly language whichdoes not have the desired control struc-tures. Some assembly languages don'teven permit the stylistic freedom neededto produce readable programs.

Another way of achieving the benefits ofstructured programming is to impose adiscipline for coding conventions tosimulate the structured programmingfigures; structured programming is, afterall, only concerned with programtopology, not implementation details.However, this approach has the dangersinherent in all externally imposed con-ventions. In addition, the simulation usedmay often obscure those very aspects ofthe program that are to be clarified.

The best way to achieve proper language(coding) utilization, obtain lucidprograms, and relieve some program-ming drudgery is to use a preprocessorthat will incorporate structuredprogramming control features. ThisStructured Programming Adaptor, orSPA concept, provides users with thepower and expressivity of structuredprogramming without the loss of com-puting efficiency usually attributed tohigher -level languages. This paperdescribes the advantages of structured

programming and the implementationtechniques involved in its application.The Structured Programming Adaptor isthen introduced as a working tool forbroad application of these techniques toall levels of programming.

Elements ofstructured programming

The basic control structures usuallychosen for structured programming arethe alternative, given by I F -THEN -ELSE(Fig. I), and the loop, given by WHILE -DO (Fig. 2). These two control structures

Fig. 1 - IF - THEN - ELSE.

Fig. 2 - WHILE - DO.

are sufficient to write any algorithm thatcan be expressed using conditional GOTO statements.2 However, since they aresometimes felt to be too restrictive, somemore general structures are added, suchas DO -UNTIL loops (Fig. 3) and CASEstatements (Fig. 4). One investigator has

Fig. 3 - DO - UNTIL.

Flg. 4 - CASE.

55

even suggested using a SEARCH loopstructure (Fig. 5) which is reminiscent of

Fig. 5 - SEARCH.

an electrical bridge circuit, unlike theothers which are more like series -parallelcircuits. SEARCH gives a loop with twoexits. It is used, for instance, to look up anentry in a table and take exit I when theentry is found, and exit 2 when the table isexhausted without the entry being found.

With the structured programming dis-cipline, one constructs large interestingprograms by nesting the control struc-tures within each other. That is, any of theclosed control structures may be placedwherever a rectangle is shown in thediagrams (Figs. I through 5). Also, thesestructures may be strung together in alinear sequence of program steps. In fact,the SEQUENCE (Fig. 6) is often treated

Fig. 6 - SEQUENCE.

as one of the basics for structuredprogramming. This nesting is in contrastto the method of producing large, in-teresting, complicated (and unreliable)programs by inserting conditional jumpinstructions in a program from anywhereto anywhere. Structured programminghas no requirement for such jumps or thecorresponding statement labels at theconceptual level; of course, at the mostdetailed level, the computer instructionsare conditional jumps.

Reprint RE -21-5-12Final manuscript received September 16. 1974

Reliability of structuredprograms

Programs produced within the structuredprogramming discipline are more reliablethan those developed with ad hoc controlsequencing. There are several reasons forthis improved reliability.

First structured programs require moreextensive design effort before they can becoded. So the purpose and the workingsof programs are better understood beforedetails are decided. Experienced pro-grammers using structured techniques forthe first time find that the programmingprocess takes more time than do standardmethods. But because of the early invest-ment in making the design more com-plete, there are fewer inconsistencies topatch up at the end of the project. Theentire process actually takes less time.

In addition to designing the major datastructures, the programmer must alsodesign the control flow and not just codein conditional jumps to somewhere to bedetermined later. Complete control struc-tures are not made for such loose ends,and postponed coding decisions take adifferent form in structured programs. Ina structured program, we may find a"stub" such as the following:

IF condition THENBEGIN

COMMENT: "worry about this later"END

In this case, the program flow is un-derstood and the details can easily befilled in later. Contrast this with thefollowing:

IF condition THEN GO TO Labe1999

Labe1999: COMMENT: "worry aboutthis later"

In the second example, the flow ofcontrol is disturbed, the code to besupplied is out of the context of itspurpose, and clarity (and thereforereliability) suffers. The same idea appliesto flow charts, for which program detailstake up several pages. Instead of off -pageconnectors, flow chart process boxes aresubstituted that contain short narrativedescriptions and a reference to anotherpage. Each page of flow charting containsone entering arrow and one exiting arrow(see Figs. 7 and 8).

Fig. 7 - Off-pag confider,.

Fig. 8 - Oft -page connectors replaced by off -pagemodules. (The details for the rectangles are givenelsewhere.)

The staging of events in which one firstsketches out an entire system and thenfills in the details to an ever finer degree isknown as the "top -down" approach. Thisis the approach generally seen in problemsolving-we recognize it as "divide andconquer." Structured programmingoffers a similar approach to computerprogramming.' Top -down programmingpermits communication betweenprogram implementers and their manage-ment or their customers. The highestlevel-the forest-is constructed first andcan be discussed independently from thepostponed details-the trees. Since aprogram that is constructed top -down isalways a whole entity that is beinggradually fleshed out, there is never aconglomeration of unrelated parts thatneeds to be integrated. And it is wellrecognized that the integration phase oflarge software systems is the longest and

56

most unpleasant part of the entireproduction process.

Because they are built by nesting, struc-tured programs have a natural hierarchy.A program is composed of componentpieces; each piece is composed of itssubpieces; and so on. A small set ofprimitive control structures is used tomake larger components out of severalsmall ones. The number of subcom-ponents of each component need notexceed "the magical number seven plus orminus two".4 This gives the simplicitypeople generally require to understand acomplex system. Whether naturally oc-curring systems are naturally hierarchicalis an issue for specialist metaphysicians.What counts for us is, that's how they aredescribed.

Structured programming gives programsall the advantages of hierarchicalsystems. This includes the possibility of abottom -up approach as well as a top -down approach. One can construct anduse stable subassemblies-pieces thathave an identity of their own. One canmodularize and use modules that aremodifiable and replaceable. One can alsoinfer characteristics of the whole fromcharacteristics of its pieces, such as cost,resource utilization, and proof ofcorrectness.

Using structured programm-ingStructured programming techniques canbe used in any programming language;the design discipline remains unchanged.All that is needed is to encode the basiccontrol structures (Figs. 1 and 2) usingthe conditional branch instructions of thelanguage. In other words, structuredprogramming is language -independent.It has to do only with the underlyingtopology of a program.

It is cerainly preferable to write programsin a modern higher -level language withthe control statements of, say, ALGOL orPL./ I. But we cannot always take advan-tage of such systems. Often the computersystem we are working with is notequipped with such language support(e.g., very old or very new or very special-purpose computers). The applicationmay also demand programming at thedetailed machine-langauge level, sincethe stylized code produced by most com-pilers is considered unwieldy or unable toperform the desired function (e.g., bit oraddress manipulation). Or one may be

obliged to maintain an existing systemwritten in a lower -level language.

In those cases, even though the abstractdiscipline of structured programming canstill be followed, a great deal of thebenefit is lost. Code in high-levellanguages can be indented to show thenesting of control structures, so theprogram flow can be clear to readers.Top -down programming can be done,but it is not very easy to block out a largeprogram in terms of program stubs in alower -level language. And it is a totallyfrustrating experience to attempt tosimulate the desired control structures,making unique but meaningless state-ment labels, and guaranteeing that thestructures are all properly nested.

A solution to"unstructured languages"

Instead of cursing the darkness-or morelikely, ignoring it altogether-we canapply a form of a preprocessor to im-prove our lower -level languages. Thepreprocessor developed at MSRD iscalled "SPA" for Structured Program-ming Adaptor. In developing SPA, weidentified the constructions missing fromour programming languages (which wereassembly languages and Fortran IV) andalso some "syntactic sugar" deemedworthwhile. The language PL360provided some suggestions here, as willbe seen.6

Elementary assembly langauge in-structions are specified in a functionalnotation, with a function name and anargument list enclosed in parentheses, asin the following sequence of three

operations:

LE( F4,X); AE(F4,DELTA); STE( F4,X)

As this indicates, a sequence of suchoperations can be written on a single cardseparated by semicolons. As in SNOBOL4, a card boundary is also treated as asemicolon. In this way, a short sequenceof simple steps that is thought of as agroup can be written on a single line toconvey that notion.

Statement labels can be written as usual;they start in the first column of a state-ment, that is, either in card column one orimmediately following a semicolon.Blank cards may be freely used to im-prove program readability; they break aprogram listing up into "paragraphs."

Comment cards follow the rule of thetarget language for the preprocessor; theyare usually denoted by a special characterin card column one.

Structured programmingprimitives in SPA

The control structures are provided bycompound statements such as the follow-ing:

IF condition THENstatement

ELSEstatement

andWHILE condition DO

statement

The statement is either a simple function -type operation, or an IF or a WHILEcompound statement, or a sequence ofstatements preceded by a BEGIN state-ment and followed by an END statement.Comments may be placed after the wordsTHEN, ELSE, DO, BEGIN, and END;

Dr. Peter G. Anderson, Software Design and Develop-ment. Missile and Surface Radar Division. Moorestown,N.J., received the BS in Mathematics in 1962 and the PhDin Mathematics in 1964, both from MIT. He served on thefaculty at Princeton University and in 1965 joined theRCA Computer Division's Systems ProgrammingDepartment to work on the development of the Fortransystem for Pie Spectra 70 computer line. His specialinterest in that and subsequent compiler developmentactivities was object code optimization. In September1971, Dr. Anderson joined the Computer Science Facultyof the New Jersey Institute of Technology as anAssociate Professor. His research interests include thesoftware engineering areas of higher level programminglanguages, software reliability, and programmingmethodology: computerized conferencing; and com-binatorial mathematics. especially graph theory andtopology. He is currently working with MSRD's SoftwareDesign and Development Department in addition to hiswork with NJIT.

57

such a comment is terminated bysemicolon or the end of the card. Thecondition used in both control structurestatements depends on the target machinearchitecture. If the conditional transferinstructions of the machine are based on acondition code setting or the sign of anaccumulator, then some very simple con-ditions are chosen. For example, "IF =THEN" or "IF < THEN". For machineswith single accumulators used for com-parisons, the condition may take the form

IF [left -symbol] relation [right -symbol]THEN

If the left -symbol is present, the variablenamed is loaded into the accumulator; ifthe left -symbol is omitted, the ac-cumulator is left unchanged. If the right -symbol is omitted, the test is based on thesign of the accumulator; if the right -symbol is given, then a comparison(possibly in the form of a subtraction) isgenerated. There is a wide range ofpossibilities here. For machines such asthe 360 with many types of comparisons,the first letters of the left -symbol andright -symbol can be used to direct whichtype of comparison to generate: RR, R X,SS, half -word, full -word, double or singleprecision floating point, etc.

Many computers are equipped withspecial instructions for incrementing anindex register and conditionally jumpingbased on the result. To take advantage ofthese "DO -loop" instructions, the condi-tion in a WHILE statement has someother possibilities. The "DO -loop" is

specified by building the condition with akeyword followed by the necessaryparameters to set up and step through theloop. For example,

WHILE UPS, [n [,m]]DO

is used on the DDP-124 computer toemploy a loop using the JXI (jump indexincremented) and JIX (jump on index). Ifn is present, the index register is loadedwith -n; if m is present, the looping isaccomplished by adding m to the indexregister and jumping on negative via theJIX instruction; if m is absent, the JXIinstruction is used, which adds one to theindex register and then jumps onnegative.

An example of the SPA assemblylanguage coding style is shown in Fig. 9.This is a portion of a Fast FourierTransform program written to run on theHoneywell DDP-124 computer.

LDX(32,1)1 STX(DX.1)LDX(o.i), STx(ANGL.1)

WHILE UPS.32.2 DOBEGIN

LDA(L,1), JST(C)04P), STA(L.)), STO(L.i,i)LDA(ANGL). ADD(.2). STD(ANGL ANGL ANGL 2)

ADXI-32.1), JST(BFLY), ADx(32,1)END

SUM THE ABSOLUTE VALUES OF THE REAL AND IMAGINARYPARTS TO FORM THE FINAL ANSWER. PUT THE FINAL ANSWERIN THE FIRST 32 WORDS OF THE INPUT ARRAY.

LDx(-64,1), STx(ST4.1)WHILE UPS...1..2 DO

BEGINLDA(L.I), ANA(..3777777?), STA(L,1)LDA(1.4.1.1)t ANA0.113227222211 ADDCL,11STX(ST3.1),STA LDX(...1)) STA(L,IIADX(1,1)1 STX(STA,1)1ST3 LDx(",1)

END

JHP*(FFT EXIT EXIT EXIT EXIT EXIT EXIT EXIT EXIT)

Fig. 9 - The last stage of a Fast Fourier Transform (on 32 points) for theHoneywell DDP-124 computer.

SUBROUTINE FFT ( LIST, NFFT )

COMPLEX LIST I NFFT), FAC, MOD. CIS

INTEGER STEP, WIDTH. SKIP

CALL BITREV ( LIST, NFFT )

STEP NFFT / 2, WIDTH ). SKIP 2

WHILE STEP ,O DO

BEGIN

NUB 1

WHILE N 0. NFFT-1, STEP DO

BEGIN

FAC CIS ( N. NFFT )

WHILE I NUB. NFFT, SKIP DO

BEGIN

J I WIDTH

MOD FAC LIST ( J I

LIST t J LIST ( I ) - MOD

LIST ( I I LIST ( I ) MOD

END

NUB = NUB 1

END

STEP STEP / 2 WIDTH . WIDTH 2 I SKIP SKIP 2

END

RETURN

Fig. 10-- A Fast Fourier Transform written in SPA -Fortran style.

WHILE NULCYC DROPS REMAIN 00

TRY MANY TIMES TO FIND A READY TASK.EACH TRY WHICH FAILS DECREMENTS THE NULL COUNTER BY ONE.

3EGIN

WHILE TCBREG < ITBLLEN DO

SEARCH TCB TABLE FOR A READY TASK.

BEGIN

IF A FOREGROUND TASK IS READY. ENTER IT.

BEGIN

SET UP ALL WORK REGISTERS AND ENTER TASK.

ENDEND

BEGIN

A BACKGROUND TASK IS READY, ENTER IT.

ENDEND

HAVE FOUND NO READY TASK, INCREMENT THE NULL COUNTER.

Fig. 11 - An example of logical -level program documentation producedautomatically from a SPA -written program.

58

The SPA idea applies to languages suchas Fortran IV as well as to assemblylanguages. In this case, the SPA ideaprovides a test bed for proposals forlinguistic reform. (Since complainingabout the inadequacies of Fortran IV is afavorite sport, one may as well do itright.) The SPA additions to Fortran IVare shown by example in Fig. 10, anotherFast Fourier Transform program. Toappreciate the enhanced reliability ofsuch a programming style, notice howeasily one can prove that the programdoes not have an infinite loop.The input program to SPA can be writtenin highly structured programming stylewith indentation to indicate the programnesting structure. SPA takes care ofproducing correctly aligned input tosatisfy the rules of the target language.

Because SPA -written programs are struc-tured programs whose composition is bymeans of nesting, these programs can beedited automatically to show their struc-ture: namely, every statement of theprogram is at a certain depth of nesting;that statement should appear on a listingbeginning at the tab position cor-responding to the depth. For instance, astatement at depth n should begin inposition 3n on a listing. In this way, muchof the cosmetics of a program can bemaintained automatically.

SPA -written programs also provide aform of automatic documentation aid. Ithas been noticed that programdescriptions written in a pseudoprogramming language in structuredprogramming style are just as suitable asflow charts for documentation. Thispseudo flow charting can be achieved byselectively listing a SPA -writtenprogram. The lines listed for documenta-tion are the program comment lines andthe SPA keyword lines (IF, WHILE,BEGIN, and END). Documentation canbe done completely by printing all thekeyword lines, or selectively by printingonly the lines under a predetermineddepth of nesting. Of course, any suchdocument can be edited after it is

automatically produced. The notoriouslyubiquitous attitude of programmerstoward documentation is partially ex-plained by the observation that theprogram itself is their most carefullydescribed algorithm-so why should theydescribe the whole thing all over again inanother language?

Higher level programming languages

have not been successful in fulfilling theirpromises as self -documenting programs;the level of detail in program listings ismuch too high. Proper documentation issupposed to guide the reader from theforest to the trees. The SPA autcmaticdocumentation idea does just that. Fig.11 shows an example of such documenta-tion. This is taken from a task dispatcherfor a real-time operating system writtenfor the Univac 1616 (AN/ UYK-I 5). Theperiods at the beginnings of some linesindicate comment lines; lines withoutperiods at the beginning are SPAkeywork lines.

Implementation of SPA

The first version of the SPA processorwas implemented using SNOBOL 4 run-ning on an interactive timesharingnetwork. SNOBOL 4 was chosen, ofcourse, because of the ease with whichcharacter string information ismanipulated. The facility of recursivesubroutines was also heavily utilized inorder to process the nested structures. Byconstructing different versions of initialvalues, we were able to use essentially thesame SPA processor for differentlanguages enhanced by structuredprogramming constructions. Using thistechnique, we had both a Fortran -IV andan ULTRA -16 version (the ULTRA -16 isthe assembler for the Univac 1616 whichis also known as the AN/ UYK-15).

Because SNOBOL 4 permits characterstring data to be used as code we wereable to include a macro definitionalfacility as part of the SPA. This gaveprogrammers a macro language with thesame power as that of the host SNOBOL4 system.

A later version of SPA has been written inPL/ I. PL/ 1 has the necessary recursivesubroutines and enough string handlingmachinery to do the job. Of course, wedid sacrifice the macro feature as well asthe expressive ease of the SNOBOL 4version, but in the exchange we gainedenormously in speed.

Because of the expressivity of SNOBOL4, we were able to implement the initialversions of SPA with little coding effort,giving us an experimental program rapid-ly. Because of the experimental nature ofthe effort (initially we wanted "somethinglike PL360"), we changed our mindsseveral times before we were pleased withwhat we had. SNOBOL 4 facilitated this

experimentation more than any otherlanguage would have. When the PL/ 1version was coded to do what theSNOBOL 4 version did, the broad orlogical outline of the new SPA was thesame as the previous one. The concepthad been debugged before the difficultprogramming was even started.

SNOBOL 4 has a reputation of being a"fun" language with no application in aproduction environment. It is easy onprogrammers and hard on computerresources (time and core space). But sinceit is so easy on programmers, so ex-pressive, and so powerful, it can be usedto check out concepts and global designdecisions for production programs thatare to be eventually coded in moreefficient 'anguages.

Through breadboarding, electricalengineers can discover properties of theirdesigns that can be applied to give themgreater reliability. Programmers can alsoprofit from similar "quick and dirty"implementations of their designs.

Conclusions

The benefits of structured programmingcan be fully realized even byprogrammers who need to work inassembly languages. Moreover, the struc-tured programming style can be appliedon a wide variety of languages by using apreprocessor. Programs written in such astyle are easy to document by automaticmeans.

It is hoped-and expected-that someform of the language techniques ex-hibited herein will be used by theassemblers of the future. This will bringclarity, and therefore high -reliabilityprograms, to the broad spectrum ofcomputer users.

References

I. Dijkstra, E.W.; "GOTO Statement Considered Harmful,"CACM. Vol. II, No. 3 (March 1968) pp. 147-148.

2. Bohm, C. and Jacopini, G.; "Flow Diagrams. TuringMachines, and Languages with only two FormationRules,"C4CM. Vol. 9. No. 5 (May 1966). pp. 366-371.

3. Wirth. Niklaus; "Program Development by StepwiseRefinement." CACM, Vol. 14, No. 4 (April 1971) pp. 221-227.

4. Miller, George A.; 'The Magical Number Seven, Plus orMinus Two; Some Limits on our Capacity for ProcessingInformation," The Psychological Review. Vol. 63. No. 2I March 1956) pp. 81-97.

5. Simon. H.A.; The Sciences of the Artificial, Cambridge,M.I.T. Press (1969) pp. 87-107.

6. Wirth, Niklaus; "PL360, A Programming Langauge for the360 Coirruters," Journal of the ACM. Vol. 15. No. I (Jan1968) pp. 37-74.

59

xo

xl -X2

METAL

INSULATOR 1K01

DEPLETION REGIONn TYPE tri, KNI

K+ TYPE 112,Kol

METAL

Fig. 1 - Metal -insulator -semiconductor one-dimensional cross section.

CI

Cd

Rd 1VI

Rs

CI

CdIVI

Rd IVI

Rs

INSULATOR CAPACITANCE

DEPLETION CAPACITANCE

DEPLETION RESISTANCE

SUBSTRATE RESISTANCE

Ko E0Xy

$(sE° xdlvl

[X1-X0(XI)A

FP2 X2

Fig. 2 - MIS approximate equivalent circuit

vLvH

CL

cat

hg. 3 - MIS varactor characteristics.

V

Fig 4- Junction varactor characteristics.

Microwave metal -insulator-semiconductor varactorparametric amplifierDr. R.L. Camisa B.F. Hitch S. Yuan

Parametric amplifiers using junction or Schottky barrier varactors have excellent noisefigure performance, but have proved quite unstable in practice unless elaborate and costlystabilizing circuits were employed. The unique properties of metal -insulator -semiconductor varactors make possible paramp designs with inherent gain stability withrespect to pump power so that pump stabilizing circuits may be reduced or eliminated. Inthis paper, S -Band parametric amplifiers using MIS varactors are described.

THE metal -insulator -semiconductor(MIS) structure was first proposed in1959 by J.L. Moll' for a "Variable Reac-tance with Large Capacity Change." Thisdevice was called a surface varactorbecause when reversed biased, a deple-tion region forms at the insulator -semiconductor interface. In 1972,Midler' presented a microwave MISupper-sideband up -converter whose per-formance compared favorably with steprecovery diodes. In 1973, Camisa, Hitch,and Yuan' developed a 4 -bit phase shifterat 3.5 G Hz using MIS varactors. Mar-quardt and Schiek4 have presented resultsof a uhf paramp using MIS varactors,although no noise figure or bandwidthinformation was given. In this paper,physical mechanism of MIS varactorswill be described. Theoretical and ex-perimental behavior of a microwave MISvaractor parametric amplifier are alsopresented.

MIS theory

Figs. I and 2, respectively, show a crosssection of an MIS varactor and itsequivalent circuit. Cr is the capacitance ofthe insulating layer and It is thespreading resistance of the N+ layer. Cdand Re are associated with the depletionregion in the N layer.

For a large negative bias, the capacitanceis approximately constant at a minimumvalue, Cc. At large positive voltages, thereis no depletion and the capacitance equalsthe insulator capacitance Ci(or CH). Forintermediate bias voltages, thecapacitance exhibits a sloping curvebetween these two limits as in Fig. 3. Thisis in contrast to the junction (or Schottky)

varactor C- V curve of Fig. 4 whose C- Vcurve continues to display a sharp up-ward slope with increasing bias. Also, theforward bias region must be avoided in ajunction varactor since the dc currentflow would cause noise figure degrada-tion. This is avoided in an MIS varactorsince the insulator prevents dc currentflow. Hence, the full range of capacitancecan be used. It is these two distinctions,the C- Vcurve shape and the absence of dccurrent, that make the MIS varactorattractive in parametric amplifierapplications.

Device design and fabrication

The MIS devices were fabricated on an Non N+ silicon (e, = 11.8) wafer. The N -layer resistivity was 0.36 ohm -cm, and thethickness was 0.6 p.m. After surfacepreparation, 800A of SO2, (e,= 3.9) wasgrown on the surface in a dry oxygenatmosphere. Aluminum metallizationwas deposited on the oxide and photo -etched to produce an array of 2 -mil -diameter dots for the top electrodes. Thewafer was then scribed, and individualchips were mounted in coaxial pinpackages. A paramp was constructed ona 25 -mil alumina substrate (e, = 9.6). Thedesign is similar to that of Bura, et al.5

Parametric amplifier analysis

In an analysis similar to that of Penfieldand Rafuse,6 the gain of a parametricamplifier can be defined as a function ofthe ratio CI / Co, where Co is the dccomponent at the fundamental pumpfrequency. Fig. 5 shows how the com-puted value of Cr/ Co varies with pumpvoltage swing at various dc bias voltages.

Reprint RE -21-5-4Final manuscript received February 15, 1975.

03

02

O

0.1

VB r OV

-V6- 2V8

6

4 LL

1.2

1.0

0.8

6 e100 6

PEAK PUMP VOLTAGE

Fig. 5 - Cl/CO and CO. as functions of pump voltage swing for various biaslevels. in an MIS varactor with 2:1 capacitance ratio.

An idealized piecewise linear curve, likethe one shown was assumed for a varac-tor with approximately 2:1 capacitancevariation (CH CO. A sinusoidal pumpingvoltage was also assumed. Due to theunique C-1; curve of the MIS device,C, Co (and hence, gain) eventuallybecomes constant with pump power. Thegraph shows that for a bias voltagemidway between the knees of the C- V

Dr. Raymond L. Camisa, Microwave Technology Center.RCA Laboratories, Princeton, New Jersey received theBEE, MEE. and PhD degrees from the City College of theCity University of New York in 1965. 1969. and 1975.respectively. From 1966 to 1967 he was a member of thetechnical staff at the RCA Advanced CommunicationsLaboratory, New York. While there he worked onmicrowave filters, low -noise parametric amplifiers. andmicrowave integrated circuit techniques. From 1967 to1970 he was with Wheeler Laboratories, Inc., Great Neck,L.I., where he was part of a group developing amicrowave integrated circuit receiver for IFF

applications. Specifically, he developed various MICcomponents including low -noise transistor amplifiers,frequency multipliers and filters. From 1970 to 1974 hewas a part-time lecturer at the City University of NewYork teaching courses in electromagnetic theory, elec-tronics, and microwave measurements. At the Universityhe also worked as a graduate research assistant in-vestigating the use of MIS varactors in microwavenetworks. During the same period, he was a consultant toWheeler laboratories and the RCA Advanced Com-munications Laboratory. In 1974 Dr. Camisa joined theMicrowave Technology Center at RCA Laboratories.

Camisa

0.5

0.4

a 0.3

02

0.1-

I I I I 1 I I 1..

0 I 2 3 4 5 6

CH/CL

Fig. 6 - Cl/CO as a function of MIS varactor capacitance ratioCH/CL.

curve (-2 V in the case shown), the ratioC. C. approaches a constant value atlower pump voltages than for othervalues of bias. Hence, the midway point isthe optimum bias point for minimumpump power. Fig. 5 also shows that C,varies with jump voltage for bias

voltages other than optimum. This varia-tion causes the amplifier response to shiftin frequency, even in the pump voltage

Princeton, N.J.His responsibilities include research onGaAs field -effect -transistor device technology andlinear amplifier development.

Benjamin F. Hitch, Advanced CommJnicationsLaboratory. GCASD, Somerville, New Jersey, receivedthe BSEE degree from the University of Tennessee in1967, and joined RCA's Communication Systems Divi-sion (CSD) the same year. From 1967 to 1970. Mr. Hitchworked primarily on HF and power amplifiers and relatedequipment. Since 1970 his work has been in themicrowave integrated circuits area. In 1972 he was co -recipient of a CSD technical excellence award for hiswork on the SH. Small Satellite Terminal Program. Forthe last two years 'te has worked on the applications ofMIS devices, especially to phase shifters.

S. Yuan, Advanced Communications Laboratory,GCASD. Somerville. New Jersey, received the BSEEfrom the University of California at Berkeley in 1952 and

Hitch (left) and Yuan.

7 8 9 10

range where gain is fairly constant. Sucha frequency shift is also present inparametric amplifiers using junction orSchottky varactors. Thus at the optimumbias point, two important advantages arepresent: gain is constant as a function ofpump power, and frequency shift withpump power is eliminated. Fig. 6 showshow the limiting value of CI. C, varieswith CH. Ci.

the MSEESom Columbia University in 1956. From 1953tc 1955. he worked for RCA on the deve opment ofbroadcast equipment, and from 1957 to 1963 he was asenior engineer with the electronics division of CurtisWright Corporation and a member of the electronicresearch laooratory staff at Columbia University. From1963, when he rejoined RCA, to 1966, his work was in thearea of solid-state rf techniques. He became a Leader,Advanced Communications Laboratory. GovernmentCommunications and Automated Systems Division,Somerville, N.J., in 1966, with responsibility for thedevelopment of circuits such as uhf amplifiers, uhf/vhfvoltage -controlled oscillators, phase comparators, andcrosspoints using beam -lead diodes for switchingmatrices. lie is the co -inventor of a balanced mixercrrcuit that can alleviate high -order intermodulation andcrossmodulation distortion. His work on frequencymultipliers has resulted in a theoretical analysis forpredicting the efficiency of an XB multiplier as a functionof the nonlinear coefficient and in an expenmental modelof the multiplier which yields a 10% efficiency "octupler"(500 MHz to 4 GHz) Mr. Yuan is a member of the ChineseInstitute of Engineers. Tau Beta Pi. Eta Kappa Nu. andPhi Tau Ph.

+16 +18 +20 +22PUMP POWER- dBm

Fig. 7 - Measured gain vs. pump power characteristics for an MISparamp.

ac

5

17.0

16.0

15.0

14.0

13.0

12.0

11.0+4

16

14

12

10

02.50

PUMP POWER+17.5d8m156m911+18 dBm 163mW1+18.5 d8m170.8

+24

2.54 2.58 2.62

FREQUENCY- GHz

2.66 2.70

Flg. 8 - Passband characteristics of the 1.5% bandwidth paramp atvarious pump levels.

2.08 2.16 2.24

FREQUENCY - GM:

2.32

Fig. 9 - Passband characteristics of the 4% bandwidth paramp.

Table I - Performance characteristics of MIS paramps.

Experimental results

The experimental paramp operated at 2.6GHz with an 8.6-GHz pump. Fig. 7 showsthat the gain remains constant within 0.5dB as the pump power is varied over a 5 -dB range. Note that as a rule, the gain of ajunction varactor parametric amplifierincreases by 0.5 dB for a 0.1 -dB increasein pump power. The passbandcharacteristic of Fig. 8 illustrates thefrequency shift with varying pump powerwhich was mentioned earlier, indicatingbias was not at optimum for this case. The3 -dB bandwidth was 1.5%, or 40 MHz.The noise figure, measured at a pumplevel of +18 dBm was 2.0 dB.

Another paramp was developed with asignal frequency of 2.2 GHz and thepump was 9.3 GHz. The 3 -dB bandwidthof this paramp was approximately 100MHz (or 4%). The passbandcharacteristics of this paramp are shownin Fig. 9.

Summary

The performance characteristics of thetwo MIS paramps are shown in Table I.MIS diodes haveperformance in parametric amplifierdesigns. Their unique properties providean inherent gain stability with respect topump power which offers potentialreduction in the cost and complexity ofparamp systems. These properties arealso applicable to up -converters.

Acknowledgment

The authors wish to thank A. Mikelsonsfor his help in constructing and testing thecircuits.

References

I. Moll. J.L.; "Variable Capacitance with Large CapacityChange," 1959 IRE WESCON Convention Record, PT -3.pp. 32-36.

Paramp No. 1 Paromp No. 2

Signal frequency 2.6 GHz 2.2 GHz 2. Muller, J.; "An Upper-Sideband Up -Converter Using MISVendor,." IEEE Journal of Solid State Circuits, Vol. SC -

Pump frequency 8.6 GHz 9.3 GHz 7. No. 1 (Feb. 1972) pp. 43-50.

Pump power +18 dB +24.3 dBm 3. Casima. R.L., Hitch, B.F., and Yuan. S.; "Microwave

Gain at 25°C 16 dB 15 dB Metal -Insulator -Semiconductor (MIS) Varactor PhaseShifter." RCA Review,. Vol. 36, No. 2 (June 1975) pp. 296-

Bandwidth (3 dB) 40 MHz 100 MHz 315.

Noise figure 2 dB 4. Marquardt. J., and Schick, B.; "An Even C -V Characteristic

I -dB compression at output -29 dBm Parametric Amplifier Using MIS Varactors." Proceedingsof the IEEE (Nov. 1969) pp. 2076-7.

Gain at 0°C 16 dB 5. Bun, P.. Camisa, R.. Pan. W.Y., Yuan. S., and Block. A.;Gain at +65°C II dB "Design Considerations for an Integrated I.8-GHz

Parametric Amplifier." IEEE Transactions on ElectronicDevices (July 1968) pp. 486-490.

6. Penfield and Rafuse; Varactor Applications, MIT Press,1962.

62

Automatic test equipmentsoftware cost performanceH.L. Fischer

Automatic Test Equipment (ATE) software is an example of the improving trend insoftware productivity. ATE programming costs have decreased as much as 100:1. Thedecrease is a result of experience, improved hardware, and utility and operating systemsoftware. Utility and operating software has become a major factor in new ATE systemdevelopment. It is demanding an advanced level of software understanding by ATEengineers.

H ISTORICALLY, software has beenunderrated, treated as a necessary evil,and considered of secondary importance.The result: Software projects were late,some were never delivered, and many thatwere delivered did not perform accordingto specification.' System developmentand support costs were higher than ex-pected and rose rapidly as software in-creased as a percent of the total system.Available management capability waslong on experience with hardwaresystems while short on experience withsoftware systems. By 1961, however,papers began appearing on the subject ofsoftware management.'

Software, it was recognized, is not unlikehardware in that it represents thedeliverable end item of a process whichcan be well defined.' By the mid sixties,leading universities were teaching in anew field, computer science; andengineers were using the computer in thedaily routine of solving classroom andlaboratory problems. By the late sixties,people began to analyze the science ofprogramming in order to organize anddiscipline the implementation phases ofsoftware development.'" Softwaretranscended from witchcraft to anengineering discipline in just over adecade. The new methods of softwaregeneration have proven to be moremanageable, more productive and theresults more easily maintained. In addi-tion there are experienced, skilled per-sonnel available to manage the im-plementation. Software costs havedecreased from ten and fifteen years agoand will continue to decrease as the newmethods become more widely im-plemented.

The purpose of this paper is to identifyfactors that have increased software

productivity (decreased programmingcosts). Types of ATE software and thestate of their development are in-troduced, followed by a review ofapplication program cost performanceduring the past 15 years. Programmingcost factors are identified. The impact ofutility and operating system software onATE is highlighted. Emphasis is oncurrent developments in utility andoperating system software, an area thathas presented problems to many ATEdevelopers in the past.

Types of ATE software

There are three generic, interrelated typesof ATE software: utility software,operating system software, and applica-tion program software.

-- Utility software includes the tools forgenerating operational and applicationprograms, and includes compilers,assemblers, file maintenance programs,debugging aids, update aids, andprogramming automation software.Many of these programs come bundledwith the ATE computer and are animportant cost factor in the selection ofcomputer hardware.

- The operating system softwaremanages the allocation of resources andservices such as memory, processors,stimuli, measurement, and control. Theoperating system correspondingly in-cludes the programs that manage theseresources such as monitor, schedule, filemanagement, and supervisory controlprograms.

- The application programs or testprograms are unique to the units undertest ( U UTs). They exercise the inputs and

Henry L. Richer, Mgr., Product Design, AutomatedSystems Division. Burlington, Mass.. received the BS inElectrical Engineering from Tufts University in 1952. In1983, he assumed responsibility for coordinating anddirecting the entire programming effort on Multi -systemTest Equipment programs for the U.S. Army. In thisregard, he played a key role in the development of allsoftware and checkout aids, procedures, and techniquesfor automatic test and system self -test for the MTE,DIMATE, and LCSS Systems at RCA. He co-authored theProgram Design Handbook for Automatic Test Equip-ment, 1988, and the Avionic Design Guide for VASTCompatibility, 1970. He has supervised activit.esencom-pasing compilers, time-shared systems, interpreters andinteractive real-time systems incorporating newprogramming technologies. Recent developmentprograms include: real-time, multi -programmed testoperating systems for jet engine accessory testing,ATLAS Compiler and Run Time Operating System forthird generation ATE systems, and hardware andsoftware for the Army's automotive vehicular testprograms. He is a Registered Professional Ergineer inMassachusetts and member of the IEEE and Sigma Xi.

Reprint RE -21-5-1Final manuscript received September 8, 1974.

63

outputs of the UUT to check perfor-mance, align, adjust, calibrate, and faultisolate.

How ATE software developed

First generation ATE had card or papertape input with little or no utilitysoftware. The hardware was anassemblage of individual functional box-es. Test programs were coded in machineor assembly language. Programmers withpredominantly mathematicalbackgrounds, lacked the training to copewith the testing interface. Asa result, firstgeneration ATE required a very complexengineering/ programming process in-volving many interdisciplinary functions,as shown in Fig. I. Confusion and con-flict existed between the differenttechnical disciplines, each with its uniqueterminology, biases, and misun-derstandings. Costs were understandablyhigh and management difficult.

Second -generation systems containedcomputers that performed simple controland data processing tasks. High -ordertest language compilers and simpleoperating systems were introduced,simulators were tried, mostly un-successfully, and test programs werecompiled on large off-line computers.Small test systems appeared with on-lineinterpreters. Integrated measurementdevices and wide band synthesizers wereused. The high -order language made itpossible to eliminate the specialized com-puter programmer function from the testprogram preparation process (see Fig. 1);and the newly developing skills of theATE test engineer included test analysisand design. The test engineer oftendeveloped a deeper understanding of theUUT than an individual on the originaldesign team. Test program quality im-proved greatly; and faults were beingdetected correctly 100% of the time andisolated correctly 95% of the time. A

PERTEST

(1974DOLLARS)

600

400

1200

1000

800

600

400

1ST

GENERATION

60

YEAR

65

2NDGENERATION

70

3RD

GENERATION

75

Fig. 2 - ATE test programming cost trend for high -quality software (100°,0 fault detection; 95., fault isolation).

handbook was written s to disseminatethe breadth of information on the testprogram generation process.

Third generation systems are heavilysupported by utility and operatingsystems software to synthesize stimulus,analyze responses, control the system,and support test program preparation.RCA's test systems are also compatiblewith automatic programming systems fordigital U UTs. The ATE test engineer astest analyzer has become obsolete. Thedesign engineer models the unit under testfor auto -test design. D-LASARautomatically programs the test design.(D-LASAR is an automated test designprogram for digital logic.) The ATEprocessor is used to translate the D -LASER output to executable code. Novalidation is required. However,automatic programming for analogU UTs is less available and widespread useis still several years off.

Cost experience in applicationsoftware

The largest software investment is inapplication programming. Applicationsoftware costs often exceed hardwarecosts on large support programs. It is

unfortunate that these costs are identifiedwith the unit cost of ATE, when in

1ST

GENERATION

2NDGENERATION

GENERATION

PERFORMANCE TEST

SPECIFICATIONS REQUIREMENTS

LAJ T

DESIGNER

Ulf)DESIGNER

UlfTDESIGNER

UUT

ANALYST

TEST

DESIGN

TEST

DESIGNER

PROGRAM

COMPUTERPROGRAMMER

VALIDATE

TEST DES'R

/PROGRAMMER

ATE TEST ENGINEER

Fig. 1 - Evolution of the engineering programming process.

actuality they are part of the non-recurring costs associated with each unitunder test. Understandably the majorcost reduction in ATE software has beenmade in the application programmingarea.

Overall, a 100:1 cost improvement hasbeen achieved in digital UUT testsoftware and a 20:1 improvement inanalog U UT software over the 15 years ofATE development. This trend is il-lustrated in Fig. 2.

These cost figures were derived fromactual RCA ATE program experienceover a I5 -year span. The figures havebeen normalized to a 1974 cost base toaccount for inflationary influences onengineering salaries, services, andmaterials. A test is a single completelydesigned and validated operational teststep which includes signal connection,measurement, evaluation, decision, andresulting action (i.e., operator com-munication and/or program branch).This measurement criteria has notchanged over the years. The data wasextrapolated from five major program-ming efforts representing hundreds ofUUT test programs on all types of elec-tronic equipments, and it closelyresembles the industry trend.

Obviously the more mature and ex-perienced ATE manufacturers havelower cost experience over a given timeperiod than relative newcomers to theindustry.

A large increase in productivity occured(A) when an accurate standard of com-munication between the test designer andthe programmer was developed. Thisstandard in one case was a combinationfunctional flowchart and timing diagramthat ensured test functions were un-

64

100

so

PERCENT 60TOTALSYSTEMDEVELOPMENTCOST 1.1 r

1905

HARDWARE

YEAR

1970 A15

Fig. 3 - ATE hardware/software development cost trend (excludes application software).

ambiguously described. Its rigid, simpleformat resulted in a structured designthat was easily reviewed and controlled.

Further cost improvement (B) resultedwhen the test analysis function was com-bined with test design. A major costefficiency (C) was experience when theHigh -Order Test language compiler wasintroduced. Lesser improvementsfollowed mostly as a result of increasedengineer experience, interactive terminalsand other production efficiencies. Com-puter aided circuit analysis (D) providedminor cost improvement in the early 70'sand finally automated programming (E)made a major contribution to lowersoftware costs.

Hardware/software costrelationship

The improvements in applicationprogramming cost are principally theresult of improved utility and operatingsystem software. Utility and operatingsoftware has become a major factor inmodern ATE development. Already ac-counting for 50% of RCA's third genera-tion system costs, (see Fig. 3), the percen-tage of software cost will increase furtheras hardware becomes cheaper andsoftware labor more expensive. Overall,however, system costs have droppedsignificantly, and will continue to drop ashardware size and complexity decreases,and as our ability to manage and producesoftware increases.

RCA's third generation ATE (System 3)is an example of the significant departurefrom previous ATE architecture. System3 makes extensive use of recentdevelopments in minicomputer hardwareand software to make reductions inhardware size and complexity and toincrease software productivity.

System 3 software features include:

ATLAS language

Virtual memoryAutomatic calibrationSampled -data measurement techniques

Computer generated stimulus

Automatic test generation for digital logic

"Universal pin" programmable interlace

On -station programming and editing

Complete arithmetic package

All math functions

16 -decimal -digit precision

- Boolean operators

Software cost factors

Various factors influence the speed andeffectiveness of producing software. Dr.Boehm of Rand Corporation" identifiedthose shown in Table I.

Computer system response is the ease inwhich a user can make software cor-rections. Variations in individualperformance varies as a function of train-ing, experience, job -knowledge, andproficiency. Programming languages are

designed for different applications andshould be selected carefully. Smallercomputers with less powerful instructionsets tend to gain more productivity fromhigh -order programming languages thanlarge computers. Sqliware developmentcriteria refers to the design ground ruleswhich programmers follow, e.g., the re-quirement for efficient code, memoryconservation, prompt completion of task,etc. Learning curve is the experiencegained on the job or with the problem.Structured programming is the develop-ment of a simple, logical structurepartitioned into self-contained modularHocks. Computer saturation occurswhen 85% of the CPU or memory capaci-ty is reached. Programming costs in-crease exponentially after this point. Therecommended computer capacity is 50%more than absolutely necessary. Softwareresponsiveness is being aware of userneeds and considering them in the designprocess.

Two factors that effect softwareproductivity that have not yet been

mentioned are: ( I ) Top -down programm-ing, which is the integration ofprogramming while it is being developedstarting with the executive and controlmodules at the top and expanding to thefunctional modules at the bottom, and (2)a chief programmer team which is a smallfunctionally specialized and skilledsoftware engineering team.`.

Cost experience in utility andoperating system software

Management's control over the cost fac-tors described in the previous section hasa high degree of influence on software

Table I - Factors affecting software cost.

Facwrs

Computer system responseVariations between individualsProgramming languagesSoftware development criteriaLearning curveStructured programmingComputer saturationSoftware responsiveness

llf

21)%

26:1

3.5:1

f (Criteria)2:1 (single experience)3:1

65

cost. Fig. 4 is a prediction of a likelyfuture trend in software productivity.9Overlayed are productivity figures fromrecent RCA software experience,(circles), and from model softwareprograms outside the realm of ATE,(triangles).

The upper and lower bounds represent agenerally accepted range of coding com-plexity. Each point on the plot has a letterinserted indicating the relative complexi-ty of the programming task, i.e., e, s, andd. Easy coding (e) generally has fewinteractions with other system elements(e.g., support software). Standard coding(s) has some interactions with the othersystem elements (e.g., functionalsoftware), and difficult coding (d) hasmany interactions with other systemselements (e.g., control elements).'Software productivity generally changesby a factor of two with each step incomplexity.

The plotted points in Fig. 4 indicate thatATE software has kept pace withproductivity predictions and with theperformance of better-known softwareefforts outside ATE. The paragraphs thatfollow discuss these programs andidentify cost factors relating the resultsindicated.

System H is RCA's real time, multi -station, multi -programmed test systemfor jet engine fuel controls and hydraulicaccessories. The 10 000 words of SystemH operating system software fall in thedifficult category. When System H com-puters were selected, the bundled realtime executive software had not yetmatured and mastery of the CPU and itsreal time executive (not only what it did,but how) was essential. The effort meantadded work, and an understanding and

SO

MACHINEWORDS PER 3°MAN DAY

5

Ins

PRIMARILYMACHINELANGUAGE

appreciation of executive system designconcepts. System H software productivi-ty of 10 words of code per man day wasrealtively high considering that conceptsof structured programming were onlybeginning to be used, and a satisfactoryhigh order language was not available.Software productivity factors were:

Positive

High computer system response

Selected key individuals

Software responsiveness

Negative

Machine language

Computer hardware/software maturity

Software learning curve

Computer saturation

The use of FORTRAN as the base for theSystem H test programming language isof passing interest. FORTRAN was ex-tended to include test -oriented System Hstatements. These English -like statementsparalleled FORTRAN syntax and resem-ble the terminology of hydraulic test.Using an existing compiler proved to behighly successful, and at an investment ofseveral man -months of engineering, wasvery cost effective."

System 3 is a general-purpose electronictest system. The coding is standard com-plexity consisting primarily of utilitysoftware operating under the computersexecutive system. Eight man years wereapplied to develop the 150 000 words ofexecutable code. Two of the eight manyears were devoted to the development ofan adapted ATLAS language and anATLAS compiler/ interpreter. This is incontrast to twelve man years expended ona second generation test language com-piler, or a six -fold improvement. Theimprovement is attributed to several fac-

1970

PRIMARILYHIGH ORDERLANGUAGE (HOUR

Fig. 4 - Milestones in software productivity

STRUCTUREDPROGRAMAPPIROACMFS

1955

tors: (1) The compiler was developed bytwo skilled ATE engineers with state-of-the-art programming experience.Programmer support was not required,and the problems of communication andcontrol were virtually non-existent; (2)the use of ALGOL facilitated structureddesign and high -order language efficien-cy; and (3) the design was implementedinteractively on a video terminal whichfacilitated editing and quick turn -around.

One man year covered all of the otherutility software, e.g., text editor, panelinterface handler, D-LASAR translator,etc. The remaining five man years werespent on run time software for signalsynthesis, signal analysis (Fast FourierTransform) and device handlers in-cluding digital, LF, RF, pulse, universalpin programmable interface, and com-mon equipment interfaces.

System 3 overall software productivityaveraged 70 words of code per man day.Software productivity factors were:

Positive

H igh-order language

Top engineering personnel

Computer system response

Negative

Disc saturation

Cassette response

Using an all engineering staff instead of aprogramming staff for ATE utility andoperating system software has proven tobe a cost effective practice.

IBM's OS/360 and Honeywell's MUL-TICS (Multiplexed Information andComputing Service, developed jointly byBell Labs, G.E., and Project MAC ofMIT) are large computer utilities in thevery difficult coding category. They areshown on the plot in Fig. 4 as points ofreference. OS/ 360 took 2 000 man yearsto build in assembly language accordingto Computerworld. MULTICS took 200.Prof. John Donovan of MIT stated thatMULTICS was first used by MITstudents after 50 man years of develop-ment; however, 200 man years wasapplied to the current system containing

1 000 000 words of procedure code(300 000 lines of Pl./ On. MULTICSwas referred to as an operating system ofthe future because of the techniques thatwere used in its. development. Softwareproductivity factors included:

66

High -order language

Computer system responseStructured design

System I is a portable automatic testsystem for testing internal combustionengines. It contains a task orientedoperating system and interactivesoftware''. The coding level is difficult.The system demonstrated the feasibilityof computer based diagnosis of Armyvehicles with built-in diagnostic connec-tors. An average assembly levelprogramming efficiency of 25 in-structions per man day was achieved by afour -man chief programmer team in onlytwo months. This software system is ahighly structured hierarchy of modules,as shown in Fig. 5. The modules average50 instructions. Each module is self-contained with its own data structure,internal linkages, and accessing andmodifying procedures. Calling sequencesare part of the module, and designdecisions are inherent in the moduleinstead of distributed throughout thesystem. The system is self -integrating.Each module executes within the contextof the total system sharing commonprocedures and stack facilities existinghigher in the hierarchy.

System I software developed is an ex-cellent example of top -down structuredprogramming. The effort was highlyproductive even though programmingwas in machine language. Softwareproductivity factors are:

Positive

Computer system responseChief programmer teamStructured top -down programming

Fig. 5 - System I software hierarchy

All engineering staff

Negative

Assembly language

Learning curve

Phase 2 of the System I programdeveloped vehicle performance and faultdiagnosis software. This standard level ofcoding is being produced at a rate of 50words per man day.

NYTIB is IBM's New York Times Infor-mation Bank system.' It is a chiefprogrammer team experiment with morerigorous control of "GO -TO free"programming. NYTIB is standard levelcoding and gave results similar to thoseexperienced in System I. Dr. Mills, who isresponsible for many of the proceduresIBM used, stated at the IEEE Collo-quium on "Software for the Engineer" inLexington, Mass., Feb. 7, 1974, "struc-tured programming is the mental processof programming itself. Using structuredprogramming will result in: 3:1 moreproduction; 10:1 less errors; and 2:1 lesssize."

Software productivity factors included:

Chief programmer teamStructured "GO -TO free" top -down

programming

Software responsiveness

Conclusion

The fundamental methodology ofdeveloping cost effective software is notnew. The principles have been used byengineering management for years. Yet itwas not obvious, not even to the engineer-ing manager, when his project washeading into software trouble. Recently(1969-1973) Dijkstra, Mills, Liskov,Baker, and others have been effective intheir analysis and solution of theproblem. Fundamentally the problem isone of relating program text (design) withexecution (code). The solution is a struc-tured, step -wise development ofprograms, expanding functionalspecifications into text. Software shouldbe conceptually simple and the text easilycoded. The result is a design that helpskeep track of what is being done, reducesthe problem of mental verification, andimproves management control.

Selecting the right people for the job is themost important factor in increasing

Fig. 6 - ATE engineering team oganization.

software productivity. Establishing theproper environment for them to work in,including the tools and procedures towork with, is the other.

The ATE System is best engineered bythree interdisciplinary teams. Each team,as illustrated in Fig. 6, must haveoverlapping knowledge of the other'sengineering skills. Team members areproven performers, experts in their field,and usually include selected newengineers. Management provides thecatalyst, lets people know what is ex-pected of them, stages the environmentthat will allow it to happen, and accountsfor their performance.

Ref ere -ices

I. Van Roekens. P.N.."Probkms Associated with Implemen-tation of Software Systems.- IEEE Colloquium onSoftware for :he Engineer. Lexington, Mass., Feb. 7.1974.

2. Hosier. W.A.. "Pitfalls and Safeguards in Real -TimeDigital Systems with Emphasis on Programming". IRETransactiorn on Engineering Management. Vol. EM -8

(June 19611 pp. 99-115.

3. liellcomm. Inc.. Management Procedures in ComputerProgramming for Apollo -Interim Report. FR -64-222-1.Nov. 30. 1964.

4. Dijkstra. E.W.; "Structured Programming.- SolocareEngineering Techniques. Buxton. J.N. and Bandell, B. Ed.(Oct. 1969) p. 5448.

5. Baker. F. -Tr. "Chief Programmer Team Management ofProduction Programming." IBM Systems Journal.Vol.No. I. 1972.

6. Mills. HD.. 'How to Write Correct Programs and KnowIC IBM. Federal Systems Division. FSC 73-1008. Jan.1973.

7. Liskov. "Guidelines for the Design and Implementa-tion of Reliable Software Systems," AD -757 905. MitreCorp.. Feb. 1973.

M. -Program Design Handbook for Automatic Test Equip-ment". RCA Corporation Report CR-68-588-30,Burlington. Mass.. Junc 1968.

9. Boehm. B.W., "Software and Its Impact: A QuantitativeAssessment." Datamation May 1973) pp. 48-59.

Ilk Fischer. H.L. and Valiance. A.F.;"A FORTRAN Solutionto a Higher Order -lest language Requirement." IEEEProceedings. on Automatic Support Systems S.{.1111,0341111,Nov. 1972.

I I.Corbato, (Amgen. C.T.. and Saltier. J.H.."MULTICS: The First Seven Years."HoneywellComputerJournal. Val. 6. No. 1 (1972) pp. 3-14.

12. Teixeira. N.A.. l'radko. F.. Sarna, D.. and Garland. P.:"Diagnostic Instrumentation for Military Vehicles."Society of Automotive Engineers. 730658, June 1973.

67

Bernard M. Pietrucha, Associate Member TechnicalStaff. High Reliability Engineering. Solid State Division,Somerville,.N.J., received the BSEE in 1967 and theMSEE in 1973, both from the Newark College ofEngineering. From 1967 to 1971. he served with the U.S.Air Force as a project engineer with the Satellite ControlFacility, and as a launch controller on both Air Force andNASA space programs at Vandenberg Air Force Base. In1971-72 he was awarded both a teaching fellowship anda research assistantship at Newark College of Engineer-ing where he investigated deposition of silicon nitridefilms using a carbon dioxide laser as an energy sourcefor evaporation. He joined the RCA Solid State Divisionin 1973 as a member of the High Reliability IntegratedCircuits Technical Staff. Mr. Pietrucha is a member of theIEEE.

Eugene M. Reiss. Manager. High Reliability IC Engineer-ing. Solid State Division, Somerville, N.J.. received hisBS in Chemistry and MBA in Management from FairleighDickinson University in 1957 and 1971, respectively. In1959 he joined RCA where he worked on the develop-ment of high-speed switching transistors. In 1965, Mr.Reiss joined the newly formed integrated -circuitsdepartment and engaged in the manufacture of linearand digital devices. His assignments included processimprovements in the diffusion and photochemistryareas. In 1966 he was named Leader for the Spectra 70circuits. Since 1972 he has directed the High ReliabilityIC Engineering department and has been heavily in-volved in space applications of C/MOS devices.

Reprint RE -21-5-6Final manuscript received January 15, 1975.

The reliability of epoxyas a die attach in digitaland linear integrated circuitsB.M. Pietrucha E.M. Reiss

The implementation and advantaaes of epoxy die -attach are described with specialemphasis given to areas of mechanical integrity and product stability. A recent special testprogram indicates that the epoxy die -attach process results in a reliable product.

D I RING the past fifteen years theIncreased use of epoxy (Ref. I has

resulted in the growth of the number ofavailable sources. This study is confinedto the evaluation of a one -part -conductive epoxy which as been usedsince 1966 in the manufacture of bothlinear and digital I.C.'s, by both thebipolar and MOS technologies.

It is easy to understand the growingacceptance by electronic componentmanufacturers of this die -attach system,since when it is compared to the con-ventional eutectic mount it offers thefollowing advantages.

I. The device temperature does not exceed200°C ± 10°C during the die -attach process.

2. The integrity of the die -attach is not afunction of the back surface of the semicon-ductor, thereby eliminating the need forspecial operations; i.e.. the evaporation ofgold onto wafers.

3. 1 he efficiency of the die -attach operation.

Authors Pietrucha (left) and Reiss.

higher die -attach yield (approaching 100percent). and the adaptability of the methodto mechanization result in a lower cost thaneutectic mounting.

Despite these advantages, the use of anepoxy die -attach system in high -reliability applications caused the follow-ing concerns:

I. Integrity of the die -epoxy package system.For example, would the effect of hundredsor thousands of temperature cycles result inthe die separating from the package?

2. The effect of by-products or side effects ofthe epoxy on the reliability of the finished

If the device experiencedtemperatures, would the epoxy continue todecompose. and would the outgassingproducts result in latent failures?

3. Arc the process controls well enough definedto prevent defective product from beingmanufactured? (process controls are in-cluded as Appendix A).

As a result of these concerns, users ofhigh -reliability devices in both industry

68

and the military have been reluctant touse parts with an epoxy die -attachsystem. This view is reflected in militarydocuments, such as the MIL -M-38510general specification, which specificallyprohibit the use of an epoxy die -attach.This study evaluates each of these con-cerns. Both the mechanical integrity andlifetime of die -epoxy package systems areevaluated under varying levels of stress.

Description of epoxy used formounting

The epoxy evaluated in this paper is a onepart. thermosetting, conductive paste; thefiller is silver. This epoxy was chosenbecause of the good thermal conductivityof the silver. Thermal conductivity wasconsidered since the epoxy is used inmounting both bipolar and MOS devices.

The mechanical andof the epoxy are:

physical properties

Thermal Conductivity 140X10 4Cal/scm°CElectrical Resistivity 5X10 ohm cmPercent Resin 14.5 - 15.5%,

Percent Silver 69 - 71%Total Solids 84 - 86%

The epoxy is a proprietary resin with aphthalic anhydride catalyst. The solventis a butyl cellosolve compound.

Test program

The test program for evaluation of deviceintegrity and product stability was

designed to address two major concerns:

I. Does the epoxy provide a stable die -to -package interlace capable of withstandingthermal and mechanical stress?

2. Does epoxy contribute any detrimental ele-ment to the environment within a

hermetically sea led integrated -circuitpackage?

Both questions require specific attentionsince uncertainties in either area couldseverely affect reliability of the finishedproduct.

Mechanical evaluation

A random sample of 200 type 4001CMOS devices composed of productmanufactured over a ten -week period wasselected for study (Ref. 2). These partswere subjected to the test program out -

Table I - Mechanical screen sequence.

Constant Acceleration 60.000 GS V1 Direction 1 Minute

Temperature Cycle 65°C to 200°C. 10 Minutes atE xtremes. 100 Cycles

Constant Acceleration 60.000 G's `11 Direction. 1 Mmute

Temperature Aging No Bias 1000 Hours 200°C

Temperature Cycle 65°C to 200°C 10 Minutes atExtremes. 100 Cycles

Constant Acceleration 60.000 (T's Y1 Direction 1 Wrote

lined in Table I. At the conclusion of thetest sequence, 20 devices were opened andexamined for mechanical integrity. Avisual examination was performed toevaluate die orientation, corrosion of themetallization system, and silver migra-tion. A shear test was then performed byapplying a force of 1000 grams to the sideof the chip. No corrosion of the metalliza-tion or silver migration was noted and alldevices passed the shear test. (Whileseveral packages were broken duringdecapping, the breaking also fracturedthe chip and the chip pieces were stillattached to the ceramic substrate). "restresults are summarized in Table II.

Table II - Mechanical screen results.

Inspect For

Die Orrentatron

Mrallitation Corrosion

Silver Migration

Shear Test

Results

Correct

None Observed

None Observed

No Die Removed Intact

Determination of outgassingproducts

Outgassing of epoxy has been a majorconcern of both manufacturers and usersof epoxy -mounted integrated circuits andhybrid systems. An improperly ventedcuring oven could prevent solvent frombeing carried away from the devicesduring cure, and this might result insurface contamination of the mountedcomponents. There is also concern thatoutgassing of cured epoxy within a sealeddevice could present a reliability hazard.

To determine the time required to proper-ly release the organic solvents and allowthe epoxy to stabilize, a ther-mogravimetric analysis was performedon uncured epoxy at temperatures ofI 75°C, 200°C, 225°C, and 250°C. This

analysis was performed using a thermalanalyzer which, in isothermal operation,measures the weight of a material as afunction of elapsed time. The systemutilizes a semi -micro balance to preciselymeasure the weight loss of a small sample(30 to 100 milligrams) to minimize theproblems associated with heat transfer ina larger mass. The results of this analysisindicate that after 20 minutes attemperatures below 200°C the epoxystabilizes at a figure approximating itstotal percent solids, and outgassingceases. At temperatures above 200°C theepoxy continues to outgas, as shown inFig. I. The normal curing cycle for theone -part epoxy during production is twohours at 200°C.

100

B 90

85

80

75

175°C200°C7

225°C250°C

I

0 20 30 40 50 60 70 80Time Minutes

Fig. 1 - Thermogravimetric analysis (isothermal).

Based on the results of this analysis, testcells were prepared to determine theoutgassing products of the epoxy, againat the four temperatures of I 75°C,200°C, 225°C, and 250°C. The test -cellmatrix is shown in Table III. Twelve testdevices were prepared for each

temperature. Four packages containeddummy chips mounted in epoxy (no wirebonding); these packages were cured fortwo hours, tack welded, pre -sealed bakedfor four hours, and sealed in a sealing boxwith dry N2 atmosphere. Four morepackages prepared with dummy chipsmounted on uncured epoxy were sealedin the sealing box with no cure or bake.Four additional packages were sealedempty to provide a control cell. Each cellof twelve devices was then baked at itsrespective temperature for four hours. Atthe completion of the four-hour bake,

Table Ill - Oulgassing tint calla (N = 48).

175°C 200°C 225°C 250°C

Cured Eposs Mount 4 4 4 4

Uncured Epcs y Mount 4 4 4 4

Empty Pack. ge 4 4 4 4

69

three devices from each cell were sub-mitted for analysis by mass spectrograph.

The gas ambient in a device package canbe analyzed accuractely by mass spec-trometer if the device is sealed in ahermetic unit attached directly to the inletsystem of the mass spectrometer andpierced invacuo after a suitable evacua-tion of the entire system. The sensitivityof the mass spectrometer used in the testsdescribed is high enough to permitreporting of test results accurate to 5parts per million in most devices.

Table IV - Spectrographic analysis (175°C to 25°C) ofgaseous content of sealed packages in PPM.

175°C 200°CCured Uncured Control Cured Uncured Control

Hydrogen 830 775 795 617 870 870

Oxygen 210 175 140 7625 45 200

Argon 90 133 122 515 118 275

Carbon Moo"(le 810 900 1700 2141 950 5000

Methane 30 125 280 119 1600 475

Ethane N D 36 75 95 145 410

225°C 250°CCured Uncured Control Cured Uncured Control

Hydrogen 650 670 739 900 780 690

Oxygen 45 290 9157 70 105 300

Argon 125 320 736 140 180 85

Carbon Droxrde 3200 4000 2399 4200 8500 800

Methane 200 610 112 400 900 22

Ethane 106 455 N.D.150 700 N.D.

The analysis indicated no significantdifference in the quantities of hydrogen,oxygen, argon, carbon dioxide, methane,and ethane between the cured epoxy anduncured epoxy, and the control (emptypackages) for each temperature range,Table IV. A mixture of butanes, butenes,propanes, and propenes was detectedbut could not be quantified. These samespectral lines were noted during analysisof a sample of epoxy heated to 250°C.The analysis did not reveal any concen-tration of ionizable compounds emittedfrom the epoxy that would contribute tosurface contamination.

The remaining twelve devices (one fromeach cell) were submitted for analysis of

Table V - Water vapor content in PPM.

Bake Temperature

175°C 200°C 225°C 250°C

Cured N.D. N.D. N.D. N.D.

Uncured 100 100 200 4900

Empty Package N.D. N.D. N.D. N.D.

N.D. = None Detected

water content so that in the event offailure of any of the cells during stresstesting some correlation between watercontent and failure rate might be drawn.Water vapor was only detected in the cellcontaining uncured epoxy, Table V.

Effects of outgassing ondevice stability

Following the determination of thenature of the environment within thehermetic package, a program was under-taken to evaluate the effects of theseelements on device performance. Threetest cells of 4007 CMOS integratedcircuits were chosen because these devicesconsist of 3 PMOS and 3 N MOStransistors on a single chip - two com-plementary pairs with the third pairconnected as an inverter. The presence ofonly six transistors greatly simplifiesparametric evaluation. Additionally, theCMOS devices are surface sensitive andrespond readily to contamination of thesurface.

The first cell consisted of units whichwere epoxy mounted, cured, and sealed.A second cell was mounted eutecticallyand processed in an identical fashion. Thefinal cell was epoxy mounted with no cureor bake of any kind and was used to assessthe effects of uncured epoxy on theperformance of the device. This third cellprovides a "worst -case baseline" forevaluation; however, devices would notnormally be mounted in this manner.

An initial test group of 24 devices (eightdevices per cell) was placed in a 250°Coven with bias voltages applied by meansof the circuit shown in Fig. 2. Parametricdata was recorded at times of 4, 8, 12, 54,124, and 196 hours. A second group of 24devices was then placed in the chamberand tested identically for 600 hours. Thecombined test results show zero failures

ALL RESISTORS10 K OHMS

vim 10 PIN 41YgsGM) lel% 71

*C4101 CONNECTED

Fig. 2 - Bias stress -test schematic diagram.

0 HourCured Epoxy

20

in both the cured- and uncured- epoxytest cells. The eutectic -test cells showedone failure. Histograms of leakagecurrent for both zero -hour and 600 -hourdown periods indicated no significantparameters shifts in any of the three cells,Fig. 3.

Eutectic

20 20

O 025 05 aA 0 025 0.5 PA600 Hours

20

O 0.25 0.5 9A 0 0.25 0.5 aA0 Hour 600 Hours

Uncured Epoxy20 20

O 0.25 0 5 atk 0 025 0 5 0A0 Hour 600 Hours

Fig. 3 - Leakage -current histograms.

Since these tests were time terminated,acceleration factors (Ref. 3) werecalculated on the assumption that thefirst failure was about to occur in theepoxy cell. An activation energy of I . I eV(Ref. 4) for silicon integrated circuits wasused in the Arrhenius equation, based ondata generated for bipolar devices andsupporting data for CMOS (Ref. 5).

R(T)= A exp (- El kT) (I)

where R(T) is the failure rate at a giventemperature T, A is the scale factor, E isthe activation energy in electron -volts, kis Boltzman's constant (8.63 X 10-3eVI° K), and T is temperature (° K).

To determine the acceleration factor, F,let

F= R(T2)I (2)

where R(T2) is the failure rate attemperature T2 and R( Ti) is the failurerate at temperature Ti.

From Eqs. (1) and (2);

A exp (- El kT2)F - (3)

A exp (-Elk TI)

But F is also inversely proportional to theratio of the times to first failure, so that

F= 102 = exp [El k (II T1- II T2)]OF (4)

tl = exp [El k (11 T - IT2)]

where 12 is the time to first failure at T2and ti is the time to first failure at T.

70

Table VI - Time to first failure using calculatedacceleration factors.

Temperature

150°C 1423.10

125°C 1398°81

100°C 1373010

85°C 1358.10

55°C 1328°C)

25°C 1298°81

Time (Approximate)

180,000 Hours

1.2 a 106 Hourt

1 08 x 107 Hours

4.5 x 107 Hours

1.2x109 Hours

6 x 1010 Hours

Using Eq. (4), the calculated time to firstfailure at 55°C would be 1.2 x 10" hours.The time to first failure at other values oftemperature is given in Table VI, andplotted in Fig. 4.

li

350

300

250200i150

-100 .--50

25 1 III 1 1 11 1 111 1 111 1111 1111 I 1112 466 4. I 466 2 4611 2 464 2 466 2 446

102 1C11 104 106 106 101 101 100Hours

Fig. 4 - Predicted life distribution of epoxy mounteddevices.

To estimate the median life of the sampleat 250C, the distribution of failure wasassumed to be log -normal with a = 1.3(Refs. 3,4). Again, the first failure wasassumed imminent at 600 hours. A lineconstructed for a sample size of 16 isshown in Fig. 4. The constructed lineindicates a time to median failure of atleast 4000 hours at the 250°C level.

The fact that there were no failuresattributable to the epoxy after 600 hoursindicates that epoxy (even uncured) doesnot affect the stability of the device at the250°C temperature. The small sample sizeand the number of failures in the eutecticcell are not sufficient evidence to provethe superiority of the epoxy mount.However, under the test conditionsdescribed, the epoxy provided an

environment which did not degrade thereliability of the device.

Table VII - Reliability estimates at varioustemperatures.

Temperature

125°C

Failure Rate (%/1000 Mrs.)

0 011

100°C 0 0013

55°C 0 000012

Failure -rate estimates at 125°C, 110°C,and 55°C at 60 percent confidence,assuming one failure and using theacceleration factors calculated, are

shown in Table VII. Field data (Ref. 6)supporting these estimates is shown inTable VIII: the data was taken from foursatellites using epoxy -mounted devicesand presently in earth orbit. Thedemonstrated failure rate of 0.013% per1000 hours compares favorably with thecalculated values in Table VII.

package invacuo. There is some questionwhether vapors desorbed during analysisinvacuo would be present in the vaporphase while the package was stillhermetic. Another unexplored area is

analysis of solids which might condensewithin the package after stress testing.This technique could provide a moredefinitive indication of the environmentwithin the package.

Acknowledgments

Table VIII - Field usage data on epoxy mountedintegrated circuits.

The authors gratefully acknowledge thecontributions of D. Masakowski, J.

Name of Satellote OSCAR -6 ITOS 0/F AE Zuber, C. Whelan, H. Foxman, F. Reiss,Time in Orbit 14 Months 20 Months 3 Months M. Vincoff, J. Schoen, and V. Fowler ofNumber of Units 90 42 2400 the Somerville, N.J. Technical Staff, S.A.Test Hours 985 500 613.200 5,256,000 Decker of the Harrison, N.J. TechnicalNumb.. of Failures 0 0 0 Staff, and High Reliability TechnicalFailure Rate 1%/1000 Hrs.1

MTTF lefoursl

0 092 0 145

1.080.000 680 000

00.6

6.010.000 Staff at Findlay, Ohio.

Total Failure Rate 0 013140000 Hit

MTTF 7.400.000

Conclusions

Mechanical screening has shown thatepoxy provides a positive die -to -

substrate interface capable ofwithstanding a high degree of thermaland physical stress. Silver migration andmetallization corrosion were notobserved during the visual inspection.

Histograms of leakage parameters showthat outgassing of epoxy does not affectthe stability of an integrated circuit, evenafter 600 hours of exposure totemperature stress. There was no signifi-cant degradation in the leakage

parameters during test.

Analysis of outgassing constituentsreveals no substantial concentrations ofionizable compounds which might con-tribute to contamination -type failuremechanisms.

Recommendations for furtherwork

Variations in gas analysis techniquesutilizing mass spectrometry can produceconflicting results (Ref. 7) when analyz-ing the ambient gases present in anintegrated circuit package.

Refinements are required in this areawhich would analyze only the gasespresent and exclude the products whichare desorbed from the mount and the

References

I kelp. N., ti,nderman. N.; Epoxy Die Bonding Catches On.Electronic News. December 18. 1972; p. 23.

2. Vincolt. M N.; Reliability Chip Silver fawn. SS..stent.RCA COS - MOS Integrated Circuit Reliability Data, 1970.

3. Reiss. F.; RCA Private Communication.

4. Zicrdt. C.H.: Priwurement Specification techniques liarHigh Reliability Transistors. Proceedings 01 the thirteenthAnnual Symposium of Reliability. Washington. D.C.:1967.p. 3118.

5. Whelan, C.; RCA Private Communica t i ,,,, .

6.5 incotl. M.N.; RC/1 Private Communication.

7.1 homas, II W. and Moore. B.A.; Unique Reliability.4otalvsis Capability, Rome Air Development Center. NewYork; 1973.

Appendix A: Process controls

Vigorous incoming inspections are per-formed on incoming lots of epoxy inaddition to stringent process controls.The vendor supplies a Certificate ofCompliance with each lot certifying thatthe requirements for adhesion, thermalconductivity, and electrical conductivityare met. The incoming inspection verifiesthe silver, resin and total solids content inaddition to the viscosity. Each lot is

identified by the vendor by a uniquenumber and expiration date: specialhandling procedures have been es-

tablished to assure that the epoxy is notused beyond this date.

The in -process controls consist of themonitoring of oven temperature andpurging -gas flow rate by both manufac-turing and quality -control personnel.Control charts showing both action andabsolute limits are posted. After die -attach, a sample of units is subjected to ashear -test force of forty grams. Eachdevice is visually inspected for properapplication of the epoxy.

71

Television applicationsof PLZT ceramicsB.M. Soltoff

The electrically controllable properties of PLZT ferroelectric ceramic dev ces can beexploited for use in television systems as variable neutral density filters, as selectivespectral filters, and as light gates in a system for stereoscopic viewing.

EXISTING spacecraft televisionsystems use a rotating color -filter wheelfor spectral separation and a motor -driven iris for exposure control. Thesemechanical elements compromisereliability and limit lifetime. Under aNASA contract, " the Astro-ElectronicsDivision has investigated a solid-stateelectro-optic filter as a possible replace-ment for these mechanical functions.

'The work described was performed by the Astro-Electronics Division under contract NAS 9-13549 toNASA. Johnson Space Center.

The birefringent properties of ferroelec-tric ceramics offer a viable electricallycontrollable alternative to the mechanicalfunctions. PLZT (polycrystallinelanthanum -modified lead zirconatetitanate) ceramic provides significantlyincreased transparency compared topreviously available materials,' and alsoovercomes the traditional limitations ofsingle -crystal ferroelectric materials (i.e.,limited size, non -uniformity, and smallelectro-optic effect). Ceramic composi-tion 9065 (9/65/35, La/ Zr/ Ti ratio) was

Bed M. Soltoff, Project Manager, Astro-ElectronicsDivision, Princeton, N.J., received the BSEE from DrexelInstitute of Technology in 1956 and the MSEE from theUniversity of Pennsylvania in 1964. Mr. Soltoff has beenassociated with television and related equipment for 18years. He is presently the Project Manager forERTS/Shuttle/SEOS sensor programs. Mr. Soltoff joinedRCA in 1962 and transferred to AED in 1963. In 1968, hewas assigned as the Manager. Camera Development.where he initiated the design and development of thehigh resolution Return Beam Vidicon Cameras used forsurveillance on the Earth Resources Satellites In 1969.

he directed the development of a minature color tvcamera for space applications. During 'his period, healso directed the development of the vioeo multiplexerand sequence generator for the RBV system. From 1956to 1962. he worked for the Philco Corporation ontelevision systems, first for the Researcn Division andlater for the Commercial Engineering Section. Mr.Soltoff is a member of the IEEE. the Technical Group onBroadcast and Television Receivers, and the planningcommittee of the Aerospace and Electonic SystemsSociety.

Author. Bert Soltof I. demonstrating the stereo goggles used for the stereoscopic television application described inthis paper.

V POLARIZER

ELECTRODEGRID PATTERN

PLZTPOLARAXIS

VOLTAGESOURCE

PLZTCERAMICPLATE

ANALYZER

LIGHTPROPAGATION

Fig. 1 - Solid-state electro-optical I ilter configuration.With no voltage applied, the PLZT plate is isotropic. andthe polarizers are crossed at 90 . effectively blocking thelight path. Voltage applied to the PLZT plate causeslinear rotation of the incident light. and thus controlledlight transmission.

used for the experiments. The testsamples are 1.5 -inch in diameter and0.010 -inch thick.

The improved characteristics, in conjunc-tion with suitable polarizers, enable theassembly of a neutral variable light -control gate of wide dynamic range.Operation as a spectrally selective filtercan be achieved by extending the controlrange into one of the higher orders ofNewton's series,' or by designing a com-pound optical network for a better pass -band shape factor. Finally, the rapidresponse time of the PLZT permits theassembly of a stereoscopic televisionviewing system which presents very lowinterference to normal vision and permitsobservation over a large viewing area.

Variable density filter

The elementary filter configuration isshown in Fig. I. The polished PLZTplate, interposed between two crossedpolarizers, is operated in the transversemode by voltage applied to narrow elec-trodes deposited as an interdigitatedarray on the ceramic wafer. [In thetransverse mode, the optical axis of thePLZT plate is perpendicular to the direc-tion of light propagation.] Unpolarizedlight at the system input is selectivelyfiltered by the polarizer to provide linear-ly polarized light at 45° to the X -X axis.This light passes through the PLZT andimpinges on the analyzer whose optical

Reprint RE -21-5-13

A version of this paper. coauthored by William E Perry ofNASA-JSC. was presented at the IEEE Symposium onApplications of Ferroelectrics at Albuquerque, NewMexico. on June 11. 1975

axis is at 90° to the polarizer (135° to theX -X axis). In the absence of an electricfield, the PLZT plate is isotropic, and thepolarizer and analyzer act as 90° crossedpolarizers to effectively block transmis-sion of light (off state). Application of anelectric field causes the PLZT ceramicplate to become anisotropic in a con-trolled manner. When the potential isadjusted to provide half -wave retarda-tion, the plate provides 90° linear rota-tion of the incident light and thus max-imum light transmission (on state)through the analyzer.

For lower potentials, elliptic polarizedlight is obtained from the PLZT plateoutput and the analyzer selects a vectorportion. Controlling the operatingpotential between the on and off statespermits use of the PLZT plate as avoltage -dependent variable -density filter.

The electrode pattern contains fingers0.003 -in. wide, spaced 0.040 -in. apart.This provides a reasonable tradeoffbetween ease of fabrication, transmissionloss due to electrode size, and requiredoperating potential. Best results wereobtained from direct evaporation of goldon chrome.

300r- -7 T T 1 T

00 100 200 300 400 500 600 700APPLIED VOLTAGE IVI

800 900

Fig. 2 - Birefringent retarcation as a function of applieddc voltage.

Fig. 2 shows the retardation range as afunction of voltage, measured on sampleplates. The breakdown potential for airdielectric (as exists on the PLZT surface)is on the order of 25 V / mm; halfwaveretardation (750 V) can be achieved wellbefore surface breakdown occurs with the0.040 -inch spacing.

Light -control for tv camera

In an automatic light control system (Fig.3), the amplified output of the camerasensor is detected and compared to afixed reference to develop an error signal.The amplified error signal is used to

LENS

SSEF/

SENSOR VIDEOAMP AMP.

VOLT-AGE AMP.

°VREF.

JALC - ALCGET AMP.

.VIDEOOUT

Fig. 3 - Automatic light control scheme for tv cameraapplication.

control the transmission of the filter, andthus the exposure of the camera sensor.This technique is particularly suitable forsensors which have no inherent gain -control mechanism, such as a siliconvidicon.

1.0,COLOR THRESHOLD

CC0 -100 200 300 400 500 600 700 800 11130

BIAS VOLTAGE IVI

Fig. 4 - Filter density vs. voltage.

Measured control range is shown in Fig.4. A range in excess of 1000:1 has beenobtained (change in optical densitygreater than 3). Maximum attenuation isestablished by the quality of thepolarizers (type HN-32 were used for themeasurements). Of greater importance isthe minimum attenuation which is es-tablished by the effective transmission ofthe polarizers in a parallel mode, coupledwith the insertion loss of the electrodedPLZT ceramic plate. The index of refrac-tion of PLZT is relatively high (2.5),thus resulting in significant reflectionlosses (=30%) at the air/ PLZT interface.

mor-

60,

40-z:4 20-

0

UNCOATED DISC .40

SAMPLE 420. LR COAT ON ELEC-TRODED SURFACE 7/45 IP 520 NM.

LR COAT ON OPPOSITE SURFACE1/43 @ 520 NM.

ANGLE OF INCIDENCE NORMAL 1_L I -10 0

350 400 530 600 700 803WAVELENGTH (nm)

Fig. 5 - Transmission loss of PLZT wafer coated toreduce reflection losses.

Anti -reflection coatings may be appliedto minimize the surface reflection loss.Preliminary tests conducted using Thcoatings indicated significant reductionof losses can be achieved. Fig. 5 shows thetransmission results achieved on a sample

plate with one side coated 7A/ 4 at 520 nm,and the other side coated A/4 at 520 nm.Based on this data, and the capabilities ofcurrent polarizers, an effective ontransmission of 20% should beachievable.

Spectral filter operation

The intensity I transmitted by abirefringent plate between parallel polarsis given by:

I = IB cos2ir (ni-n2) t v ( I)

where ./o is the incident intensity, B is afactor that is essentially independent of v,m and n2 are the refractive indices for thetwo orthogonal polarizations in themedium, v is the wave number (thereciprocal of the vacuum wavelength ofthe light, proportional to the opticalfrequency), and t is the thickness of theplate. The intensity, I, plotted as a func-tion of v, represents the spectral responsefunction of the array, and can be made tohave a maximum at any desired wavenumber vo if the thickness and or thebirefringence of the plate are adjusted sothat:

(ni = m/ vo

where m is any integer. The value of mused to make an array with a transmis-sion maximum at v0 is called the order ofthe filter. Filters with large values of mhave the desirable property that theirtransmission decreases rapidly as v

departs from v0. However, they sufferfrom the drawback that the adjacentmaxima are more closely spaced thanwith filters with lower orders. The colorsproduced by simple filters of this sort areactually quite different from pure spectralcolors. When attempting to approximatethe visible spectral hues with a simplefilter whose value of v0 can be "tuned"over the range of visible wavelenghts, thebest subjective impression is obtainedwith a filter having parallel polars and m= 2. For m = 1, the transmission peak istoo broad; for m = 3 and higher, theadjacent transmission peaks in the visibleproduce unsaturated colors. The colorsproduced by an m = 2 filter represent thevisible spectrum fairly well, except that asatisfactory green is not obtained.

It is possible to make filters whosetransmission decreases rapidly as v

departs from vo, without degradationfrom closely -spaced adjacent maxima, if

73

more than one birefringent element isemployed. Arrays of many parallelpolarizers, interspersed by birefringentplates whose retardations increase in apowers -of -two series, have been used asmonochromatizing filters with a verynarrow spectral passband.3 The presenceof many polarizers will introduce ex-cessive losses. Narrow passband filtershave been described° that utilize a mul-tiplicity of birefringent plates and as fewas two polars.

A filter that closely approaches anyspecified spectral response can beproduced by incorporating a sufficientnumber of birefringent plates betweentwo polars. Optical network synthesistechniques can provide a wide variety ofspectral responses using a number ofidentical birefringent plates.5 If thebirefringence of each plate in the networkis changed by the same amount, then theshape of the spectral response function isretained but the periodicity is changed, sothat the positions of the transmissionpeaks are shifted. Provided the spectralbandwidth and insertion loss can bereasonably low, such a tunable filter canreplace the rotating color wheel in a field -sequential television system by shiftingthe birefringence during the verticalblanking interval. Review of a number ofsynthesized networks shows that a systemwith two identical active plates provides aclose approximation to the desiredresponse, and has a response given as:

I = (L/9) [3 + 4cos2r(ni-n2)1v + 2cos4r (ni-n2)1 v] (3)

This function has the narrowest principalpeak that is allowed for a two -plate filter,but is not as free of subsidiary maxima asare other allowed functions.

RELATIVETRANSMISSION

1.0

0.8

0.6

0.4

0.2

0400 450 500 550 600 650 700

WAVELENGTH, Inm)

Fig. 6 - Spectral response, 2 plate vs 1 plate.

750

Fig. 6 shows the spectral response for thesynthesized two -plate network, com-pared to a single -stage second -orderfilter. The two -plate configuration andoptic angles are shown in Fig. 7. Each

OPTIC AXISOPTIC AXISViSRATION VIBRATION

DIRECTION 42e36 I 27°22* DIRECTION

POLARIZER SIREFRINGENT PLATES POLARIZER

Fig. 7 - Disassembled representation of two -plateoptical network.

plate is required to shift 520 nm, with aninitial bias (which can be provided by apassive retarder) of 830 nm. Tests on amodel of this filter verified the improvedspectral purity available.

Stereoscopic television

Exising stereoscope television systemshave used various means for displayingthe third dimension. Differences lieprimarily in the viewing concept; all relyon similar binocular camera techniqueswhich may use dual cameras or a singlecamera with an optical splitter. Displayshave generally taken the form of adjacentimages, with image separation main-tained through polarizers, or anaglyphfilters worn by the viewer.

These viewing systems distort the viewer'svision when he looks away from thetelevison display to perform otherfunctions. Reflective hood systems sufferfrom image fusion time and view posi-tion constraints. A field -sequential stereotelevision system using a pair of PLZTlight gates for solid-state switching of thevisual path overcomes these problems.6Standard television equipment may beused with little additional equipment toproduce the stereoscopic system. Thesystem may be used in space explorationand habitation operations such as rendez-vous and docking of space vehicles,remote instrumentation control, andremote gathering of visual data.

Fig. 8 shows the system configuration.Two cameras are shown for convenienceof illustration; a single camera can also beemployed with a electro-optic shutteringsystem. The scene is viewed by the left andright cameras, separated by the normalinterocular distance and yawed toprovide the desired convergence angle.

The video output signals from the left andright cameras represent the left -eye andright -eye view of the object, respectively,as sensed in stereoscopic perspective.These signals are time -division multiplex-ed on a field basis by a synchronized

'SCENE]

lirl

LEFT RIGHTCAMERA CAMERA

POLARIZERPLZT PLATES

VIEWINGSYSTEM

Fig. 8 -F eld sequential stereo television system.

multiplexer with, for example, the odd -field signal of the left camera sampledalternately with the even -field signal ofthe right camera. The video output signalfrom the multiplexer is a sequence of odd -field signals from the left camera alter-nating with even -field signals from theright. These correspond to a sequence ofalternating left -eye and right -eyestereoscopic views of the object. In allother respects, the multiplexed videooutput signal is similar to the compositesignal of either camera, and is fullycompatible with other television re-quirements. The multiplexer istransmitted to a standard televisionmonitor with attached polarizing screen,where the left and right images of theobject are alternately displayed at a 60 -Hz broadcast rate.

The viewing system uses a pair of PLZTplates and associated analyzers as on offfilters in a pair of goggles worn by theviewer (Fig. 9). The filters are actuatedalternately, in synchronism with the dis-played video, by the power control cir-cuit. In this manner, first one eye of theviewer sees the screen of the monitorwhile the other eye is blocked, then theother eye sees the screen while the first eyeis blocked.

Fig. 9 - Stereo television goggles.

74

The control circuit is triggered by the syncseparator which detects the odd/ evenfield information and accordinglytriggers the power control to enable theleft -eye optical path on when the dis-played image is from the left camera.Similar triggering of the power controloccurs for the signals from the rightcamera. As a result, the viewer's left eyesees, in rapid sequence, left -camera im-ages, and the right eye sees right -cameraimages. The switching of the views occursrapidly enough, with otherwise -standardtelevision components, so that normalpersistence of vision leaves the viewerwith the impression of continuous screenimage exposure for each eye. Thus theviewer experiences a true stereoscopicreproduction of the object. Since thetransmission of the PLZT filter is optical-ly neutral, the system may be used toreproduce both color and monochromestereoscopic images.

Since the viewer's goggles include onlyelectro-optic elements and analyzers, heexperiences only minor visual degrada-tion when looking away from themonitor. The analyzer, in this case,blocks that part of the ambient light thatis naturally polarized perpendicular tothe polarization axis of the analyzer. Thiseffect will be the same as a light shade ofpolarized sun glasses. and adequate lightwill always pass through to the viewer'seyes in these situations.

The viewing system is particularly advan-tageous in cases where the viewer mustobserve both the monitor and othercontrols in performance of some task.For example, an astronaut piloting aspacecraft which is to rendezvous anddock with another space vehicle may viewa three-dimensional televised view of thedocking structure and also be able toobserve internal gauges and controlsrequired in piloting the craft.

The multiplexer function can be im-plemented using a conventional special -effects generator operating in a verticalsplit -field mode. By using a trigger rateequal to one-half the vertical sync fre-quency, the vertical split will alternatelyselect full vertical fields of video.

The power switch block diagram is shownin Fig. 10. The PLZT wafer represents areactive load with a capacity of about0.015 µF. A ringing circuit using LI (I H)in conjunction with the PLZTcapacitance generates the control

VIDEO N_ SYNCIN' SEPARATOR

ODD FIELDPULSE "II

rSI LI

"IL -CR1

g/REF LEFTPLZT

I ILJ L _ _ JLOGIC RIGHT

TIM NG

L L __J

rsz -

L

Fig. 10 - Power switching for steno tv application.

voltage. Transistor switches S I throughS4 are enabled during the vertical blank-ing interval. S4 discharges the PLZTelement during the first half of the inter-val. Charging is accomplished by closingSI in conjunction with S2 or S3 (for leftor right channel selection) during thesecond half of the interval. CR1 dis-connects the charging path after fullpotential is reached. The logic timing(implemented with CMOS logic)generates the required control pulses,synchronized by the sync generator out-put. The entire power control systemrequires less than one-half watt of power,and is designed for battery operation.

Ruggedized filter holder

The basic PLZT wafer is delicate, andsubject to damage by handling. Toprovide protection and ruggedness, aholder assembly was designed which en-capsulates the wafer in Dow -Corning 93-500 elastomer. This material is opticallytransparent and compliant so as not tomechanically strain the PLZT. Theholder is made in three pieces, as shown inFig. I1. Front and rear polarizers (orclear glass plates) are bonded into theouter pieces. The middle section is

recessed to hold the PLZT element.Electrical connection to the interdigitatedelectrodes is accomplished by ultrasonicbonding of several 0.001 -in. lead wiresbetween the wafer finger rings and exter-nal contacts, to provide a solidmechanical connection. After the three

I I-1; -..

O

SECTION A

F 11 - Ruggedized bonded holder assembly.

parts are assembled and optical axesaligned, the interior cavities are vacuumencapsulated with the 93-500 material.This provides a moisture barrier andmechanical protection. In addition, theabsence of an air path provides signifi-cant improvement in high -voltagebreakdown ability. Units of this construc-tion have been successfully operated at1800V.

Conclusion

Electro-optic properties of PLZTceramics provide a versatile means forlight control where low power, wide -dynamic range, and rapid response arerequired. The maximum available lighttransmission in the on state (typicallyabout 20%) is the only significant restric-tion on the device application. Im-provements anticipated in associatedpolarizers, together with optimum anti -reflection coding selection, show promiseof 30 to 35% transmission.

Incorporation of the device as a lightcontrol unit should be considered inconjunction with sensors such as siliconvidicons, and charge -coupled imagerswhich have no inherent gain controlmechanisms. The other applicationsdescribed in this paper should provideseveral imaginative usages for this ver-satile device.

Acknow edgment

The author acknowledges the contribu-tion of M. Kravitz in the area of circuitdesign; F. Hubit, who performed most ofthe required fabrication and testmeasurements; and Dr. A. Miller, of theDSRC, for analytic studies of theapplications.

References

I. Haeriling, GM.. and Land. C.E.; "Recent Improvements inthe Optical and Electro-optic Properties of PLZTCeramics", IEEE Transactions Sonic., and Ultrasonics Vol.SU-I9 (1972) pp. 269-280. Also published in Ferroelectrics.Vol. 3 (19721 pp. 269-280.

2. Deily. J.G.;,"Microsonics, Color Key" Industrial Research.Vol. IS. No. I I (Oct. 1973) pp. 44-30.

3. Lyot, B., Arm Astrophys, Vol. VII (1944) p 31.3. Evans. J.W ; "Lyot Filters", Journal of the Optical Society

of America Vol. 39 (1949) pp. 229-242.

5. Chang. I.C., et al. "Optical Network Synthesis UsingBirefringent Crystals. Pt I." Journal of the Optical Societyof America, vol. 54. No. 10. (October 1964) pp. 1267-1279.Pt II, ibid. Vol. 55, No. 7 (July 1965) pp. 835-841. Pt III,ibid. Vol. 56. No. 7 (July 1966) pp. 943-951. Pt IV. ibid. Vol.56, No. 7 (July 1966) pp. 952-955. Pt V. ibid. Vol. 56. No. 12(Dec. 19661 pp. 1746-1754.

6. NASA Technical Brief No. B74-10223.

75

Opticalprocessingof widebanddataG. T. Burton' R. F. Croce

This paper describes optical equipmentfor wideband data recorded at frequenciesup to 100 MHz. Bulk processing at highresolutions - and Fourier plane filteringand processing concepts - are described.

ELECTRON -BEAM and laser -beamrecorders are presently used to recordwideband signals on film at base -bandbandwidths as wide as 100 MHz. Theprocessor described here produces a two-dimensional Fourier transform of therecorded data in a system having a timebandwidth product of 106 to produce inthe Fourier plane a resolution < 100The processor employs a hybrid conceptin that it uses the parallel input and pre-processing capability of an optical systemto perform operator assisted pre-screening and data reduction while allow-ing additional data manipulations anddecision making functions to be handledby a computer operating on the reduceddata.

The optical processor with its ability toinstantaneously form a two-dimensionalFourier transform of input data - whichmay contain on the order of 106 resolu-tion elements - is better suited than acomputer system for performance of thepreliminary data reduction operation; acomputer would require prohibitivelylong processing times and large memorybanks to perform the same transformoperation. On the other hand, once thedata is reduced, the computer becomesefficient in performing the complex dataanalysis, interpretations which might in-volve several data iterations using non-linear, as well as linear algorithms.

The optical processing system used in theconfiguration of Fig. I is a spectrumanalyzer presenting (in the frequencyoutput plane) the two-dimensionalFourier transform of the recorded time -

BEAMCONDITIONER

ARGON LASER

FILMTRANSPORT

LIQUIDGATE

BEAMEXPANDER

TRANSFORMLENS

INPUTPLANE

TRANSFORMLENS

TIME OUTPUTPLANE

RELAY LENS

TPOT

VIDICON

/ FPOT 7=rVIDICON

RELAY

FREQUENCY OUTPUT LENS

PLANE

Fig. 1 - Processor configuration.

amplitude information presented at theinput plane - while also presenting afiltered replica of the input information atthe time output plane.

The processor system implemented con-sists of an air cooled argon ion laserwhich produces 44mW at 4880A, beamconditioner, variable diameter beam ex-pander, variable speed capstan drive filmtransport, liquid gate, 70mm and 35mmtransform lenses, knife edge filters andoutput transducers. The outputs from thetransducers are displayed on an isometricand raster display. Fig. 2 pictures thisequipment. Fig. 3 depicts the rack con-sole containing the isometric and rasterdisplay, all the control electronicsassociated with the laser, film transport,sensor transducers and displays.

The system is capable of accepting forprocessing, information recorded on 16-,35- and 70 -mm film strips produced usingeither an electron beam recorder (EBR),or a laser beam recorded ( LBR). Variousreal-time input systems such as recentlyimproved PROM devices, deformabletheromplastic, oil films and bubbledevices have also been considered forincorporation into the processor and canbe implemented within the present con-figuration. Methods for using thesemedia - especially a photoconductivethermoplastic device - in the Fourierplane to perform the filtering and imageenhancement are also being evaluated.

Processor concepts

The distribution of information in thefrequency or Fourier plane is directlydependent on the organization of the datain the input plane. In the illustration of

ISOMETRIC

DISPLAY

RASTER

DISPLAY

Fig. 4, the time amplitude information isrecorded in a raster format having a lineperiod, T. If the recording is ac-complished with the film moving atvelocity v1, the spacing of the raster linesis d, where d = v1X T. For recorded datato be analyzed over a baseband extendingfrom 0 Hz to a frequency, f the numberof information cycles per raster line is n =

f, X T,. Defining w as the active width ofthe film, the number of resolutionelements per unit length along a rasterline becomes n/ w. The optical processor,with an aperture length, a, has an inputplane acceptance aperture having an areaequal to a X w, the total number of cyclescontained in the aperture or the timebandwidth product of the system is N =(al d) n. Note that as the acceptanceaperture of the Fourier transform lensincreases, so does the time bandwidthproduct.

The transform of the data recorded in theformat of Fig. 4 to a Fourier planerepresentation is as shown in Fig. 5. Onlythe upper right-hand quadrant of theFourier plane is represented in the figure,although in the asembled processingsystem the complete plane is formed. Forthe orientation presented, coarse fre-quency loci are developed with coarsefrequency gradations extending along thevertical axis and fine frequencygradations extending along loci lines thatare nearly parallel to the horizontal axis.If a single tone were recorded at afrequency between 0 and f = 11 T thatfrequency would be transformed to apoint on the first locus line above the finefrequency axis. Frequencies higher than1/ T, are transformed to higher order locilines which are equally spaced in thevertical direction in the Fourier plane.

76

Fig. 2 - Optical assembly.

The coarse frequency axis in the Fourierplane has a limiting resolvability equal tothe extent of the fine frequency axisThe fine frequency axis can in turn beresolved to fa/ N, where N is the timebandwidth product and is dependent onthe aperture width. An overallresolvability at the Fourier plane greaterthan one part in 106 is possible in thepresent implementation under theassumption that the recorder andrecording material do not limit resolu-tion. The resolution at the display isdetermined by the magnification of therelay optics from the Fourier plane to theoptical sensor plane-and the resolutionof the "Frequency Plane OutputTransducer" (FPOT); the latter convertsfrom the optical signal at the transformplane to an electrical signal suitable fordriving this display. The sensor in thissystem is a 1225 -line vidicon camera.

Reprint RE -21-5-8Final manuscript received April 28, 1975

This work was sponsored by the Rome Air DevelopmentCenter, Griffiss Air Force Base, New York (ContractF30602 -73-C-0235) under the supervision of Mr. AlJambardino and Mr. W. Czyeycki.

INPUT PLANEACCEPTANCEAPOITUTE

RECORDEDRASTER

LINES

RASTER DISPLAY

ISOMETRICDISPLAY

CAMERA CON-ROLS.

ISOMETRIC DI5P_AYCONTROLS

TRANSPORTCONTROLS

LASER POWEFSUPPLY

Fig .3 - Electr,nms rack assembly.

N TIME BANDWIDTH MODUCT

"

n 1 :YCLEVRASTER LINE) - IT Tr

BANDWIDTH OF RECORDEDDATA 0 TO IT

T RASTER LINE PERIOD

INPUT DATA FORMAT

Fig. 4 - Input data format.

FINE FREQUENCY

FOURIER PLANE FORMAT

FREQUENCYLOCUS LINES

I

Af = RESOLVABLE FREQUENCYSEPARATION

N

Fig. 5 - Fourier plane format.

77

AMP

FRED.iDEAG

BEACON

AMBEACON

AM

1 111111 11111 11111111

LOG FREQUENCY

RADAR

FSK

TV

FINE FREQUENCY

b) RASTER DISPLAY

FS, RADAR

..42V+,, I 1 I 111111111

a) CONVENTIONALANALYZER

AMP

P..i11111

.40

SPECTRUM

RADAR

.v.411111111111111111111101111111

FINE FREQUENCY

C) ISOMETRIC DISPLAY

Fig. 6 - Display concept.

BEACON

Table I - Performance specifications for wideband optical processor.

TV

F5K

Data format Recorderbandwidth

Raster rule C'oarse freq. Fine .freq.resolution resolution

16mm 100 10 MHz 15.7 kHz

35 100 MHz 75 kHz

4f< 5 Hz for 96mm aperture length**4f< 20 Hz for 96 aperture length

15.7 kHz 25 Hz*

75 kHz 54 Hz**

Representative signals

A pictorial representation of how sometime domain signals of known frequencycharacteristics appear on the raster andisometric display is shown in Fig. 6. InFig. 7, the actual displays are shown inoperation. Table 1 lists the system perfor-mance specifications.

During the course of the program,various radar signatures were analyzedwhich had been recorded on 35 -mm filmusing an EBR. The 35 -mm film had thephysical data formats indicated in Fig. 8aand the EBR recorded electrical spectrumindicated in Fig. 8b. Note that therecorded format contains two separaterecording channels, each having a30M Hz bandwidth. During recording thereceived radar signals were passedthrough a mixer stage where they wereconverted to one of the baseband EBRrecording channels.

the Fourier transform plane is obtaineoby converting the input electrical fre-quencies to spatial frequencies, fi. Indoing this, R, the raster line rate, is givenas 75 kHz; and the scan width, W, is 28mm. The expression for radial distance, 1= A Ff is then used; A of the argon laserequals 4880 A- and the focal length Fofthe transform lens is 864 mm for thisprogram.

In the expression above, / is defined as theradial distance in the transform planefrom the optical axis. A vertical displace-ment (shown in Fig. 8c) corresponds to amodulating frequency along the width ofthe scan lines in Fig. 8a-while a horizon-tal displacement in the transform planecorresponds to the spatial occurrence offrequencies across the scan line. Theresulting spatial extent of the informationin the Fourier plane (only the upper righthalf plane is shown for simplicity) isshown in Fig. 8c.

The conversion from the input plane to The frequency characteristics of the

Fig. 7 - Raster and isometric display In operation.

signal as they appeared in the transformplane were displayed and screened.Selected signals were subsequentlyfiltered using the variable slit. The filteredspectrum was transformed by the secondFourier transform lens, imagedonto the vidicon camera and displayed.In this way, the original time input planewas reconstructed but only for those timesignatures corresponding to the spatialfrequency passed by the filter function.This resulted in greatly increased signal-to-noise ratio for the signals of interestand allowed the extraction of signal PR Fand pulse width by direct measurementsin the time domain. Note that theappearance of raster scan lines, filmtransport scratches or unwanted signalssuch as recorder pilot tones areeliminated in the filtering process.

ACTIVE LINEWIDTH

a)

b)

e)

18.8 rnn,(87 MHO

12 men

157 MH.)

6 tn..(30 MN,)

111111 1111

-04 I.- RASTER SPACING

CHANNEL A CHANNEL II

30 MHz 57 MHz 87 MHz

CHANNEL I

CHANNEL A

18.8 min75 kHz

Fig. 8 - EBR recorder format and Fourier plane distribu-tion.

78

The types of signals which are of interestinclude spread spectrum, frequencyhopping (e.g., frequency shift keyedsystems or frequency agile radars),amplitude -modulated and frequency -modulated signals. Examples of varioussignal types processed are shown in Fig.9. Two signals are shown, a single spreadspectrum pulse in Channel B and asequence of spread spectrum pulsesoperating at various carrier frequencies inChannel A.

While the requirement that prompted thedevelopment of the hardware of Fig. 2was one which required the processing of100 MHz signals, the system can readilybe adapted to the processing of low -frequency signals (such as sonar informa-tion) by adjusting the recorder line periodTR so as to maintain the number ofinformation cycles in the input apertureand hence the time bandwidth product. Itis also possible to multiplex and display anumber of smaller bandwidth channelssimultaneously.

The equipment described in this paperwill ultimately allow not only the displayof the frequency and filter time planes,but also conversion of the reduced data totime amplitude signals which may bedigitized for further computer process-ing. A system for doing this is shown inFig. 10. The image dissector in the timeplane allows for accurate raster linetracking while the storage tube operateson the Fourier plane in a line snatchingmode. Other devices such as large formatvidicon cameras and photodetectorarrays can also be used with some advan-tages, but this method offers a com-promise between size, complexity, costand processing rate while maintainingfull system resolution.

Summary

The optical processor display presen-tations produced during this programdemonstrated: high time -bandwidthproduct of the order of 106; bulk process-ing of wideband data at high resolution;Fourier plane filtering to enhance timedomain signatures; and direct spectrumanalysis in the Fourier plane.

Acknowledgments

The authors wish to thank Mr. R. Tetrevof the Advanced TechnologyLaboratories for his contributions to theisometric display.This work was sponsored oy the Rome Air DevelopmentCenter, Grit f iss Air Force Base, New York (ContractF30602 -73-C-0235) under the supervision of Mr. AlJambardino and Mr. W Czycycki

87 MH:

57 MHz

30 MHz

J

ARGON LASER

BEAMCONDITIONER

BEAM EXPANDER

DISIOSZATYR I C I I__

RASTER

DISPLAY

TIMEPLANE RELAY

\ LENS

I 0

IMAGEINTENSIFIER

75 kHz

FREQUENCY DOMAIN IIME DOMAIN

Fig. 9 - Sample radar signatures.

FILMFOURIER

TRANSPORT PLANE

FT LENS FT LENS

SPLITTER

FREQUENCY, Ft> RELAY LENSPLANE Ilk

VIDICON

Fig. 10 - Extended processor configuration.

Gardner T. Burton. Mgr., Coherent Optics and DisplaySystem Development, ASD, Burlington, Mass., has beenresponsible for the direction of the development oftechniques for holographically storing and retrievingdigital and image information and for the development ofoptical signal processing and pattern recognitionsystems. He has been active in this area since 1969. From1967 until 1969 he was a Leader of an Electro-OpticsGroup concerned with the development of laser beamimage and video recorder systems. Prior to 1967 Mr.Burton was responsible for the development of theelectronics and overall systems integration of the RCAIdeographic Composing Machine used for the composi-tion of Chinese tee: on photographic film. He developedan automatic exposure control system for night aerialcameras using image intensifier tubes. and has extensiveexperience in the application of aperture responsetechniques to the analysis and design of electro-opticsystems. He has evaluated methods of synchronizingbinary data transm ssion systems in noisy environments.modulation methods 'or transmitting pictorial informa-tion at high data rates, and techniques for readingprinted documents automatically. He has developedcircuitry for digital logic operations, IF and Video SignalProcessing in noisy environments and FM modulationand demonstration systems for magnetic video taperecorders.

TIME PLANE A/DCONVERTER

FREQUENCY PLANE WDCONVERTER

Richard F. Croce, ASD, Burlington, Mass., received theBEE from Manhattan College in 1962. He continuedhis studies at New York University while serving as aSenior Electronics Lab instructor at Manhattan Collegeand received his MEE degree in 1963. Since joining RCA,Mr. Croce has been involved with the design anddevelopment of an ICW radar range tracker anc a three -tone radar tracker associated with the larding anddocking radars, respectively, in the Lunar Module (LM)on the APCLLO Program. Mr. Croce has also in-vestigated the 0 -switch properties of singly and doublydoped neodymium YAG laser rods which resulted in thedelivery of a ranging and guidance laser system. Heparticipated in the development of a light -weight GaAsdiode laser rangefinder and has investigated the energytransfer time from the chromium to the neocymium ionsin double doped YAG lasers. Recently, he has madecontributions to programs involving a holographic multi-color moving -map display, methods for holographicallystoring and retrieving a miniaturized data base, andpattern recognition systems using parallel processinglogic techniques. He is presently involved in programs toapply coherent processing techniques to the problem oftwo dimension pattern recognition and image -enhancement systems.

79

Pulse doppler radar performancein clutter environmentDr. H.H. Behling

Performance of a pulse doppler radar system, designed and developed as an airborne aft -warning system against airborne targets, can be seriously degraded by ground clutterentering the same resolution cell occupied by the target echo. The design of receiverthreshold detection circuits under these conditions becomes a critical item, sincethreshold levels must be high enough to avoid false alarms, but kept low enough to becompatible with the requirement of a low cross-section target -detection capability at longranges. A computerized program is used to determine the necessary threshold increase asa function of aircraft and radar parameters for each resolution cell of interest: furthermore.the amount of detection degradation. wherever applicable. is computed. The usefulness ofthe clutter computation program was evidenced by flight tests performed with tailwarningradars to which this program was applied.

H. H. Behling, Senior Engineering Scientist, AutomatedSystem Division, Burlington, received his PhD in Physicsin 1937 from the University of Jena. Dr. Behling wasemployed at the Telefunken Company in Berlin and Ulm,Germany, where he served as microwave group leaderduring and after the war. During this period, hesupervised design of receivers, radar equipment, andmultiplex communications systems. From 1954 to 1958,he worked in the broadcast studio section of RCA on thedesign and fabrication of test equipment for colortelevision transmission. In 1958 he transferred to theAirborne Systems Division, where he worked on thedesign and development of equipment used in advancedradar, satellite, and ECM -systems. Most recently, he wasresponsible for systems and error analysis on the LEM-Rendezvous radar. He was engaged in analytical workconcerning multifunction radar systems. He alsodesigned and developed an experimental VHF -Radar forFoliage Penetration and performed a clutter computa-tion on a digital computer for the Pulse -Doppler Radar.Further work included computer aided design of circuitsin the Expendable Jammer. He was presented papers onthe computation of noise figures, the performance of lownoise input circuits, broad band amplifiers, filternetworks, and FM -modulators. He also is co-author ofthe Handbook of High -Frequency Techniques edited byH. Meinke and F.W. Gundlach, Springer Verlag Berlin,1968. Dr. Behling is a member of Sigma Xi and IEEE.

A PULSE DOPPLER radar has beendesigned and built by the ASD-Burlington activity as an effectivecountermeasure defense for special air-borne applications. Installed in the tailsection of an aircraft, the tail warning set(TWS) must be capable of providing fastresponse detection of closing threat mis-siles or aircraft. Mission requirementspostulate an extremely low false alarmrate, so that special attention must bepaid to the critical design area of thethreshold detection circuits to satisfy therequirements imposed in the latter, torespond to relatively weak returns fromgenuine targets and, on the other hand, tobe nonresponsive to signals from falsetargets.

Ground areas reflecting energy in thesame resolution cell occupied by agenuine target represent false targets, andthe echoes returned are called clutterechoes or simply clutter. The objective ofthe work described in this paper is to

Reprint RE -21-5-14Final manuscript received April 28, 1975

To - PR F

determine the amount of clutter enteringthe resolution cells, so that a thresholddesensitization program can be devised tosatisfy the requirement of low false -alarmrate, and simultaneously permit only theabsolute minimum degradation in targetdetection capability.

Typical clutter spectrum

Fig. I illustrates the typical appearance ofthe clutter spectrum in an airbornetailmounted pulse-doppler radar. Theclutter spectrum is distributed about eachPR F -line and consists of three majorportions: the mainbeam clutter, thesidelobe clutter, and the altitude return.Generically, the latter belongs in thesidelobe clutter category, but differs fromit significantly in peak power level andspectral shape. Both the relatively highpower level and the narrow width of thealtitude return are due to specularreflections from ground areas beneath theaircraft with essentially zero dopplershifts. In contrast, the sidelobe clutter isconsiderably lower in power level due tothe non-specular scattering mechanismthat produces it and occupies a bandranging from f, - 2V/A to fo + 2 V/ Aabout each P R F -line. In general, thesidelobe spectrum is non -flat, but shapedby the antenna sidelobe pattern .

The mainbeam clutter spectrum is es-sentially shaped by the mainbeampattern; its spectral location dependsupon the antenna axis orientationrelative to that of the aircraft velocityvector. The mainbeam clutter spectrumproduced in a tailmounted radar consistsof components that are doppler shifted inthe negative direction relative to thecarrier and P R F -lines.

Since targets producing negative dopplershifts (targets with opening velocities andalso those with essentially zero relativevelocities) do not constitute a potential

PR

MAIN BEAM CLUTTER

ro ),

ALTITUDE TARGET ECHORE TURN

SID EL OBE

CLUTTER

2r

A

To . PRF

Flg. 1 - Typical ground clutter spectrum In an airborne tallmounted radar.

80

10

-20

CON/MINCE LIPS.

Mal/4 =3woman issss3

LOW 035221

Z 10 LOO 11;41

NOTE. APPLICAILE FOR P 10 TI -BAND, RR ANDW POLARIZATIONS.

A:ING ANGLE (DEGREE*

Fig. 2 -Grazing angle dependence of "farmland"

threat to the aircraft-signals in frequen-cy bands with zero or negative dopplershifts, including the main beam clutterand the altitude return, are suppressed inthe receiver by a crystal filter. Therefore,the only remaining clutter competingwith the target return is the sidelobiclutter spectrum ranging fromf,+ji to/.+ 2 V/ A, where IL = 2 VLJA and VLrepresents the lower limit of targetvelocities covered by the radar.

When target closing velocity exceeds theaircraft velocity, the target return mustcompete with thermal receiver noise only.Since the thermal noise power density in awell designed radar receiver is essentiallyconstant and gain variations in thereceiver can be held at a tolerableminimum, the design of threshold detec-tors (circuits distinguishing small targetreturns from random noise andsimultaneously keeping false alarm ratesbelow a certain limit), is relatively un-complicated compared to that of clutter -limited detection techniques. Therefore,only the latter will be given furtherconsideration.

Subclutter visibilityand threshold increase

To provide sufficient subclutter visibility,i.e., permit detection of low -speed targetswith a closing velocity less than theground velocity of the radar platform,special attention must be paid to theproblem of minimizing the gain of theantenna sidelobes.'

The minimizing process, however, is

limited by theoretical and practical con-siderations, so that even after a thoroughoptimization process, the clutter powercan exceed the thermal noise power in thelower doppler channels by several orders

10 90

clutter (10°101r).

10 -

Al -

SR

-50 1 I

10

CONRDINCE

SMOTE =1MODERAlt Essma

SOW CI=

1 1

TO

NOR, CURVET. APPLICABLE FOR MODERATE -TO ROUGH SEA AND LAKE CONDITIONS.

I 1 I I 1 1 1

70 50 90

GRAZING ANGII (DEGREES)

Fig. 3 - Grazing angle dependence of "sea/lake" clutter (10° to 90 1.

of magnitude. Therefore, the thresholdlevel in these channels must be increasedjust enough to avoid false alarms, butshould be as low as possible to maintainadequate subclutter visibility.

In general, the clutter power received bythe radar in a particular doppler channelis a function of various parameters, suchas antenna sidelobe gain in a certainangular region, terrain reflectionproperties, antenna axis orientation, air-craft velocity, altitude, and range. Clutterpower can vary drastically when one ormore of these parameters are changed.For this reason, clutter computations areperformed to determine the necessarythreshold increase under various flightconditions; when properly implemented,such threshold increases result in nearoptimum performance relating to anadequate subclutter visibility combinedwith a low false alarm rate.

The clutter model

A vast amount of data collection andprocessing was performed by the lITResearch Institute, Chicago, to develop aclutter model2 by using experimental datataken over the years from 1945 to 1968 bymore than 20 different investigators andagencies and by consulting numerousreports dealing with this matter. Themodel is briefly described below. TheParameter Dependence Model definesthe backscattering cross section per unitarea of terrain as a function of basicallythree parameters:

a.= kl T,4), A)where

T = terrain category= grazing angle= radar wavelength

Four broad terrain categories have been

identified: 1) sea, lake, 2) farmland, 3)woodland, and 4) desert. It has beenfound that the first three categories arepractically independent of the radarband; furthermore, there is no differencewhether horizontal -horizontal orvertical -vertical polarization is used.

Mountains and cities cannot be modeledusing any of the four categories, due totheir structural ruggedness which doesnot permit the kind of scattering thatproduces distributed clutter.

The clutter cross sections of the first threecategories (the most significant for amajority of applications) can be repre-sented as a function of the grazing angleonly. Figs. 2 and 3 show the models forthe categories farmland and sea/ lake,respectively, which have been used fre-quently in the computations.

The parameters Mo and So, appearing onthe labels of the curves, are defined asstatistical parameters associated with theunderlying probability law briefly dis-cussed below.

Distributed clutter can be looked upon asbeing produced by a spatial continuum ofindependent scatterers, and the resultingradar return can be modeled as a narrow -band Gaussian process. The mean powerof this process depends only upon thephysical size of the scattering area andcan be characterized by the backscatter-ing cross section per unit area of il-luminated ground, ao.

However, irregular variations of theterrain reflectivity cause ch, to vary ac-cordingly when the radar platform movesthrough greater distances. Therefore, themodel must reflect the spatial variations

81

of ao in terms of statistical parameters,such as mean value and standard devia-tion.

The spatial variations of a. suggest theintroduction of an additional degree ofrandomness. First -level randomness isassociated with the scatteringphenomenon of an area involving a largenumber of small and independentscatterers, resulting in a a narrowbandGaussian process of the clutter return;second -level randomness treats thestatistical parameters themselves as ran-dom variables, so that the overall processcan be considered as being non -stationary.

From experimental data (used to es-tablish the model), it was found that theprobability density function of theprocess could be expressed by the log-normal distribution

P (E°)= (2r 502) 112 exp [- (E0 - Mo)2/ 2502]

whereE. = 10 logio a.Mo = mean value of E.So = standard deviation of E.

The cumulative probability that thevariable E. does not exceed a particularvalue E.., is therefore

rEoP(EO<E)=J P (E0) d E.

- 00

= (2 r So2)

f exp [-(E0- M0)2/ 2501- 00

For example, the probability that E.exceed M. - S., M., Mo + So is 15.87%,50%, and 84.13%, respectively.

FROm

.001E12

ALA

GATE PULSE

RGATE X- TALFILTER

N DOPPLER FILTERS

2

Application of the cluttermodel in the computations

From the previous discussion it wouldappear that the curves labeled M. + S. oreven hypothetical curves with M. + aS..(a>1) should be applied in thesimulations to achieve the required lowfalse alarm rate. However, the TWS-resolution cell is several orders ofmagnitude larger than the cell sizeapplicable during the experiments thatled to the development of the cluttermodel in its present form.' Therefore, theTWS-resolution cell involves a largenumber of smaller cells with differentE.' s, whereby the latter are normallydistributed. The standard deviation ofE. involving n samples is inverselyproportional to re' [n = number ofelementary cells within the TWS-resolution cell] and approaches zerowhen n is large. Therefore, the meanclutter characteristic was selected andapplied in the computations; see curves inFigs. 2 and 3 labeled M.

Clutter computations

The basic signal processing chain con-figuration of the TWS is shown in Fig. 4.The received r.f. signal is converted to i.f.,amplified, range gated and passedthrough a crystal filter which suppressesaltitude and main lobe clutter. The filterbank, following the crystal filter, involvesN filters of individual bandwidths B andcovers the target velocity range of in-terest. The lower doppler channels maycontain clutter, the upper channels onlythermal noise. The output of each filter isapplied to a detector/ integrator/threshold circuit. An alarm signal isinitiated when the signal level exceeds thethreshold.

DET

TOPES -HOLD

CIRCUITS

ALARLA

Fig. 4 - Simplified receiver and signal process block diagram.

The computation of the threshold in-crease involves the determination of theclutter/ threshold (C/ T/ 2) ratio:

T = (Gal GEXaLl ar)(Rrl Ro)

GsL. = gain of the sidelobe that il-luminates the ground area andproduces clutter

GE = antenna gain at edge of requiredcoverage

ac = (ao)(A) = effective clutter crosssection

A = illuminated area that reflectsclutter energy

ar = target cross sectionRr = threshold range = range at which

target of cross section ar at beamedge can be detected in thermalnoise with a given detectionprobability and false alarm rate

RG = range corresponding to the rangegate position (Ra< Rr).

Gsc and ac are complex functions ofaircraft and system parameters, includingvelocity, altitude, range, antenna orienta-tion, etc. The geometry to compute ac orA is shown in Fig. 5.

The areas A. and A.' (n = 1, 2, 3, ...) thatreflect clutter energy in the same dopplerfilter and range gate as the target arelimited by the so called isodops,representing loci of constant volocity (ordoppler shift) and, range circles.

An individual area A. or Ao' lies betweentwo isodops, corresponding to thebandwidth of an individual doppler filter,and two range circles, essentially cor-responding to the leading and trailingedges of the range gate.

zDIRECTION OF A/C VELOCITY

C

RANGE GATE WIDTH

VELOCITY GATE WIDTH

Flip. 5 -'sedans and range circles.

TOM I - C/T ratio, threshoid Inereaaa and threshold difference as functions of resolution cell index (doppler channel Table II - Detection probability (in percent) as aand ange), function of resolution cell index.

T (dB) TI (dB) A = TI - 10log (RTI RG)4Range (ft) Range (ft) Range (ft) Doppler Range (It.)

channel 2500 4500 6500 2500 4500 6500 2500 4500 6500 channel 2500 4500 6500

I 24.2 8.0 0.7 29.7 13.7 7.1 11.5 5.7 5.5 I -0 9 10

2 15.3 0.53 -13.2 20.8 6.2 0.7 2.6 -1.8 -0.9 2 61 99+ 99+

3 12.0 0.1 - 6.6 17.6 6.7 2.5 - 0.6 -1.3 0.9 3 99+ 99+ 95

4 13.8 1.1 - 6.0 19.4 7.5 2.8 1.2 -0.5 1.2 4 93 99+ 93

5 15.7 8.3 1.0 21.2 14.0 7.4 3.0 6.0 5.8 5 51 8 9

A2,A2',A1,A1', represent areas that returnsecond -time, third -time, -aroundechoes, but their contributions in thepresent situation are negligible comparedwith the energy reflected by A I and A 1'.

To preserve a specified fasle alarm rate inthe presence of clutter, the receiver detec-tion threshold must be increased by anamount TI, which is computed as follows:The ratio T/(C+N) must be raised by TIsuch that

[TI(C+N)](TI)= N

TI = (TI N)[(C + N)1 T1 = [(C + N)I N)]=[(C/T)+(N/T)]l(N/T)

where C/ T is the result of the computerevaluation and TI N is the pre -integrationsignal-to-noise ratio required to detect atarget at maximum range (thresholdrange) at beam edge with a specifiedprobability of detection and false alarmtime.

Typical computations resultsof clutter computationsAs a typical example, Table 1 shows theresults of computer runs to determineCI T and the required threshold increaseTI for the conditions listed below.

Antenna:

Aircraft parameters:

Range gate centers:

Doppler channels:

Terrain model:

dielectric rod,pointingdown by 16 deg.

Velocity 850 ft/ s.Altitude 1000 ft

2500 ft,4500 ft,

6500 ft.

1)135 - 270 (ft/s)2) 270 - 4053)405 - 5404) 540 - 6755) 675 - 810

farmland

The TI N ratio for a 99% probability ofdetection and a false alarm time of 30hours was +5.5 dB and the thresholdrange 7144 ft. The last column lists thethreshold difference A in dB between TIand a threshold increase that can bepermitted without loss in target detectioncapability at a particular range. Forranges of 2500 ft, 4500 ft, 6500 ft, and athreshold range of 7144 ft the permissiblethreshold increases are 18.2 dB, 8.0 dB,and 1.6 dB respectively.

Flight tests performed with tailwarningsets installed in stragetic aircraft andutilizing threshold increase programs ofthe kind just described do indicate theusefulness of the simulation program.

For resolution cells in Table I where A isnegative, the detection probability ex-ceeds 99%, and conversely for those witha positive A - value, the detectionprobability is lower than 99%. The loss indetection probability can be estimatedfrom curves°, which are graphs of theprobability of detection versus signal-to-noise ratio for fixed values of the false -alarm probability.

Table 11 shows the equivalent detectionprobability in percent for each resolutioncell of the above example.

Correlations of computationsand flight test resultsand other applications

Clutter computations have been per-formed pertaining to tailwarning sets tobe installed in strategic and tactical air-craft with a different antenna system ineach radar. Matrices of threshold in-crease and target detectability as

functions of resolution cell index havebeen computed in a manner similar tothat shown in the above example fordifferent aircraft velocities, ranges,altitudes, antenna pointing directionsand aircraft attitudes (roll).

Flight tests performed with tailwarningsets installed in stragetic aircraft andutilizing threshold increase programs ofthe kind just described do indicate theusefulness of the computations program.Further applications of the computationsprogram include the following majorareas:

Comparison of different antennas systemsconcerning subclutter visibility anddetermination of critical areas of thesidelobe patterns that need improvement.

Computation of the mainbeam clutter returnto determine the spectral purity requirementof the transmitter in the spectral regionoccupied by the lower doppler channels.

Spectral analysis of the clutter return in aparticular channel to determine the effect offoldover on the spectral density in the postdetection filters as function of the signalfrequency relative to the center of thedoppler filter.

Acknowledgments

[hanks to K.E. Utley for his very helpfulsuggestions and discussions on the sub-ject matter, for the preparation of thetables in this paper, and for all his effortsinvolved in the reduction process andtabulation of a vast amount of clutterdata, which contributed significantly tothe success of the program.

References

I. Honda. H. -Low Backlobe Antenna for Airborne Missile -Warning System". RCA Engineer. Vol. 18, No. 5 (Feb -Mar 1973).

2. Greenstein. L.J. et. al. "A Comprehensive Ground ClutterModel for Airborn Radars". IIT Research Institute.Chicago, III, Sept. 15, 1969, AD NO. 8619I3L.

3. Greenstein, L.J. et. al.: "ORT Clutter Model". OverlandRadar Teclmology Program Contract F33615 -67 -C -I387.IIT Research Institute. March 22, 1968.

4. Barton. D.k.. Hanson, V.G.; "Radar Signal DetectabilityCalculator". Lecture II of IEEE Boston Series on ModernRadar Theory. Oct 1972.

Author's Note

Details of the configurations metioned inthis paper and the clutter computationsare available from the author upon re-quest.

83

Engineering andResearchNotes

Microstrip discriminator

Alfred SchwarzmannMissile and Surface Radar DivisionMoorestown, N.J.

The microstrip discriminator shown in Fig. I includes a microstrip-coupled structure of the odd -mode -dominant type. The coupledstructure has three narrow strip -like conductors ( I, 2 and 3) closelyspaced from one another. The strip -like conductors are spaced by adielectric substrate (4) from a ground conductor (5) covering the bottomsurface of the substrate. Frequency -modulated rf signals at microwavefrequencies are applied at one end (Port I) of the center striplikeconductor. The remaining two strip -like conductors (2 and 3) are closelyspaced from strip -like conductor (I) near the opposite end.

X21 PORT 2

'DETECTEDR2

RT

Fig. 1 - Microstrip discriminator.

Strip -like conductor (2) is a quarter wavelength long at a frequencylower than the center carrier frequency of the rf signals applied at Port I.Strip -like conductor (3) is a quarter wavelength long at a frequencyabove the center carrier frequency of the rf signals applied at Port I.Strip -like conductors (2 and 3) are coupled to ground at their ends nearend I b of strip -like conductor (1).

A first detector diode DI has its cathode terminal coupled to strip -likeconductor (2), and its anode terminal coupled to the output terminal(Port 2). A second detector diode D2 has its anode terminal coupled tostrip -like conductor (3), and its cathode terminal coupled to the outputterminal (Port 2). A resistor R2 and capacitor C2 form a low-passnetwork. Similarly, a resistor RI and capacitor CI form a second low-pass network. The low-pass network may be selected to pass signalslower than about twice the deviation frequency.

If the frequency of the applied signal at Port I is lower than the centercarrier frequency, more of the rf power is coupled to the longer strip -likeconductor (2) than to conductor (3), providing more rf power fordetection to diode D2. The resultant output at Port 2 is negative. If thefrequency of the applied signal at Port I is higher than the center carrier

frequency, more of the rf power is coupled to the shorter strip -likeconductor (3), providing more rf power for detection to diode Dz. Theresultant output at Port 2 is positive.

To increase the amount of voltage change for deviation in frequencyfrom the carrier frequency (sensitivity), the tap points along the strip -like conductors (2 and 3) can be moved away from the grounded end.Further, this sensitivity can be altered by adjusting the value of resistorsRI and R2 and the coefficients of coupling between the strip -likeconductors ( I, 2, and 3).

Reprint RE -21-5-51 Final manuscript received April 11. 1975

High reliability COS/MOS IC's -qualified to MIL -M-38510, Class A

E. ReissSolid State DivisionSomerville, N.J.

The term high reliability as applied to integrated circuits covers a broadspectrum of reliability classes. These classes were originally defined bydevice manufacturers (based on test methods described in MIL -STD -883, a compendium of quality and conformance tests) and later werestandardized by the government. Today, there are both in-house- andgovernment -defined reliabilty classes. The contents of in-houseprograms vary with the manufacturer. The government program,referred to as MIL -M-38510, consists of a general specification for adetailed specification. The general specification applies to alltechnologies, such as TEL, ECL, linear and CMOS; the detailedspecification delineates the requirements for a circuit function within atechnology.

The purpose of the MIL -M-38510 program is to achieve standardizationamong integrated -circuit suppliers and to assure delivery of deviceswhose long-term life will satisfy the requirements of the system for whichthey are intended. Three reliability classes -A, B, and C- are describedin MIL -M-38510. Class A devices are of the highest reliability level andare intended for critical applications where replacement of componentsis not practical. RCA COS/ MOS devices were the first devices of anttechnology to be approved as Class A. Class B and C devices are thoseintended for applications in which reliability is important but in whichcomponents can be replaced.

In summary, then, devices are manufactured in accordance with theMIL -M-38510 (general and detailed) specification, which includesportions of the MIL -STD -883 test method. The device class isdetermined by the level of inspections, screening, and conformancetesting to which the devices are subjected.

Qualification and conformance testing

The accelerated stress tests delineated in MIL -STD -883. Method 5005(tests that subject devices to stress levels greater than those normallyexperienced in a typical application) consist of: Group A, Electrical;Group B, Package and Internal Mechanical Strength; and Group C,Indicators of Long Term Reliability for Package and Die (Table I). [Ed.note: Class A, B, C, refers to reliability level; Group A, B, C, refers to testtype. Do not confuse the two.] Both qualification devices and a sampleof devices from the production line are subjected to this series ofaccelerated tests. The tests performed on the qualification devices arecalled qualification tests; the tests performed on production -line devicesafter a specific type has been qualified are called conformance tests.Electrical end -point limits for the tests are defined by MIL -M-38510 andare more demanding of CMOS than of TEL. DC parameters aremeasured at -55°C, +25°C, and +125°C and, depending on circuit

84

Table I - MIL -STD -883 test. Table II - Failure rats - 60'o confidence level.

Group A or,up B Group C Total devices tested 2.693

Total inoperable failures 3

Total device hours 2,963.000

Static tests 25, -55, 125°C Physical dimensions Thermal shockTemp. Failure rate

Dynamic tests 25, -55 I25°C Resistance to solvents Temperature cycling 125°C 0.143'i 1000 hrs.

55°C 0.020', 1000 hrs.

Functional tests 25°C Internal visual &mechanical

Moisture resistance 25°C 0.006':( 1000 hrs.

Functional tests --55. I 25°C Bond strength Seal

Solderability Mechanical shockLead integrity Vibration variable frequency

Constant accelerationSeal

Salt atmosphereHigh -temperature storageOperating -life test

complexity, require between 80 and 120 individual tests at eachtemperature. Additionally, parametric drift is limited to 10 nA for /,(gates); 40 mV for low-level output voltage, Voi; and 80 mV for high-level output voltage. Vou. Five device types have been tested to date andshown to be qualified as MIL -M-38510 Class -A devices; twentyadditional device types are now under test.

Qualification test data

Qualification data has been generated on COS/ MOS types CD4019A.CD401IA, CD40 I 3A. CD4027A and the CD400I A. The data generatedfrom the Group B qualification tests indicates excellent packageintegrity. There was one failure in a total of 729 devices tested.

Group C data indicates only two failures of a total of 1504 devices tested.Considering that the end -point measurements require that minimumdrifts be maintained over the entire military operating range (-55°C to+125°C) during 80 to 120 individual measurements, this data indicatesexcellent device stability.

Subgroup C5 defines a I 25°C accelerated operating -life stress test and isconsidered to be one of the better indicators of long-term stability. Datagathered during the qualification runs indicates no degradational orinoperable failures during 645.000 device hours of test. A degradationalfailure occurs when a device drifts outside the parametric limit definedabove. An inoperable device is a device that does not function accordingto its truth table.

Conformance test data

In addition to the qualification devices tested, Group A. B, and C testsare periodically performed on production -line samples. Again, thistesting is referred to as conformance testing. In the first half of 1975,over 2.3 million device hours of I25°C accelerated -stress testing werecompleted on a wide variety of circuit functions, from gates to MSI's;there were only five degradational failures and three inoperable failures.The test results indicate that devices produced on the production linecertified to MIL -M-38510 standards are reliabile. It should be notedthat the certified COS/ MOS wafer -fabrication area is jointly used forthe production of high -reliability and standard wafers.

Failure rate

I he combination of qualification data and conformance data results in asignificant number of device hours from which a failure rate can hecalculated. Table II. Since the data was generated at an acceleratedtemperature of I 25°C. the failure rate is derated to 55°C and 25°C. amore typical operating range for satellites, the intended application areafor the des ices tested.

Field -usage data

Data generated horn four satellites has ing user 2500 CD4000A-seriesdevices on board corroborates the results achieved through acceleratedstress testing. As of June IS. 1975. over 34 -million device hours haveelapsed without a single inoperable failure. 'Fable 111.1

Conclusions

The qualification of RCA COS/ MOS devices to MIL -M-38510, ClassA. demonstrates the capability of these devices to be used in criticalapplications. Field -use data from four satellites supports the excellentqualification and conformance test results. Additionally. the periodicauditing of the manufacturing facility by two government agenciesprovides assurance that procedures and practices will be reviewed with

in mind.

References

I and Schnahle. b.: "Reliability 61 Complementary MOS Integrated Circuti.."IEEE Trans. un Reliabilitt. (Oct. 1973/.

2.6allace. LJ. and Whelan. C.D.:-Accelerated Resting of COS MOS Integrated Circuits".SI-6379. Solid State Disision (April 1975).

Table III - Field -usage data CD4000A series.

Name of satelliteA tmosherit.

OSCAR 6 ITOS Di F Explorer

Time in orbit (months) 32 43 18

Number of units 90 42 2.400

Device hours 2,073,600 1,300,320 31.104,000

Number of inoperable failures 0 0 0

Failure rate -60', confidence level 0.044 0.07 0.0031

Total device hoursFailure rate -

60% confidence level

34,477,9200.0026%/ 1000 hrs.

85

Penand Podium

RecentRCAtechnical papersand presentations

Both published papers and verbal presentations are indexed. To obtain a published paperborrow the journal in which it appears from your library or write or call the author for areprint For information on unpublished verbal presentations write or call the author (Theauthor's RCA Division appears parenthetically alter his name in the subject -index entry )For additional assistance in locating RCA technical literature contact RCA TechnicalCommunications, Bldg. 204-2, Cherry HNI, N.J. (Ext. PC -4254).

This index is prepared from listings provided bimonthly by RCA Division TechnicalPublications Administrators and Editorial Representatives-who should be contactedconcerning errors or omissions (see inside back Cover)

Subject index categories are based upon standard announcement categories used byTechnical Information Systems. Bldg. 204-2. Cherry Hill. N.J.

Subject Index

Titles of papers are permuted wherenecessary to bring significant keyword(s) tothe left for easier scanning. Authors' divisionappears parenthetically after his name.

SERIES 100BASIC THEORY &METHODOLOGY

105 Chemistryorganic, inorganic. and physical

DC GLOW DISCHARGE, The deposition oftin oxide films from a - D.E. Carlson(Labs.Pr) Journal of Electrochemical Society,Vol 122. No 10 (10/75) pp. 1334-37

DIELECTRICS, Chemical etching of -WKern (Labs.Pr) Symposium on Etching. Elec-trochemical Society Meeting. Washington.DC. (5/2-7/76) Extended abstracts. ECS. Vol76-1 (5/76)

125 Physicselectromagnetic field theory. quan-tum mechanics. basic particles.plasmas, solid state. optics, thermo-dynamics. solid mechanics. fluidmechanics, acoustics.

INTERFACE between an insulator and anelectrolyte. Structure and energy of the -RWilliams (Labs.Pr) RCA Review. Vol 36, No 3(9/75) pp 542-550

130 Mathematicsbasic and applied mathematicalmethods

ADLER's generalized equation for interactionof oscillators with injected signals, - Com-puter simulation of - H H Behling(ASD.Burl) IEEE Proceedings (10/75)pp. 15-22

FINITE ELEMENT analysis, State of the art of- R Pschunder (MSRD.Mrstn) PrincetonUniv , Princeton. NJ (12/11/75)

135 Information Theory andOperations Research

codes. modulation methods, signalprocessing, communication theory.game theory. queueing theory, com-putational linguistics. decisiontheory.

SIGNAL PROCESSING applications,CMOS/SOS developments for - W FGehweiler. J I Pridgen (ATL,Cam) 9th An-nual Asilomar Conference, Pacific Grove. CA(11/11-13/75) Symposium Proceedings.

140 Circuit and Network Theorymethods for circuit synthesis andanalysis. network topology.equivalent circuit methods.

SYNTHESIZER technique, A new highresolution, single -loop - R J Bosslaers(ASD.Burl) Con) on Automatic SupportSystems for Adv. Maintainability, Westbury,New York (10/28-30/75)

150 Environmental and HumanFactors

influence of physical environmentand/or human users on engineeringdesign, life support in hostileenviornments

EMP-INDUCED TRANSIENTS, on integratedcircuits. Effect of - E VanKeuren(GCSD,Cam) IEEE Proceedings on Electro-magnetic Compatibility. (10/7/75) p 3A II.

NUCLEAR SURVIVABILITY, A self -triggeredcircumvention technique for - R J Hen-drickson. E VanKeuren (GCSD.Cam) Sym-posium on Vulnerability and Survivability ofSurface 8 Aerial Targets. Proceedings. Vol

1

(10/21/75) pp 6-23

160 Laboratory Techniques andEquipment

experimental methods and equip-ment, lab facilities, testing, datameasurement. spectroscopy. elec-tron microscopy. dosimeters.

CALORIMETER for VHF -UHF powermeasurements - J H Bowen (MSRD.Mrstn)O.S. T. Journal of American Radio RelayLeague. Vol 59. No. 12. (12/75) pp. 11-13.

PHOSPHOR COATINGS by luminescencethermal degradation, Analysis of -S Larx,h,J Gerber (Labs.Pr) Analytica Chimica Acta(1975) pp 329-331

170 Manufacturing and Fabrica-tion

production techniques for materials,devices and equipment

CHEMICAL VAPOR DEPOSITION oftransparent electrically conducting layers ofindium oxide doped with tin - J. Kane. H.P.Schweizer. W Kern (Labs.Pr) Thin SolidFilms. Vol 29 (1975) pp 155-163

SEPARATING COMPLETED DIODES fromsemiconductor wafers, Simple method for -J Klatskin. A Rosen (Labs.Pr) Journal of theElectro-chemical Society. Vol 122. No. 11(11/75) pp 1565-66

VAPOR DEPOSITION. Chemical processesin -V S Ban. S L Gilbert (Labs, Pr) Journalof the Electrochemical Society, Vol 12. No. 10(10,75) pp 1384-91

175 Reliability, Quality Controland Standardization

value analysis, reliability analysis.standards for design and production.

RELIABILITY Predictions -M L Johnson(ASD.Burl) State of the Art of SystemAssurance Technology (Fall Lecture Series)Boston IEEE Reliability Chapter, Bedford.MA

180 Management and BusinessOperations

organization. scheduling, marketing.personnel

CONTRACTS MAN, Profit and the - gettingthe contract -R S Miller (GCSD,Cam) Profitand the Contracts Man (A Limited Ed. CourseManual. Fed. Pubis., Inc.) Sect. A (10/8/75)pp A 1-65

INVENTORY RETURNS, Computerizedmanagement of -D Mishra. J C. Pearson, ADogar (Record Div.Ind) AIEE 1975 SystemsEngineering Conference. Stardust Hotel, LasVegas. Nev (10/75)

LOGISTICS management, Industrial - J WHurley (MSRD.Mrstn) Logistics Systems(Threepart course) Temple University(10/20/75 - 11'75 - 12/8/75,

PROJECT MANAGEMENT -M W Buckley(MSRD.Mrstn ) Only of Delaware. Neward,DE (12/4-6/75).

SERIES 200MATERIALS, DEVICES, &COMPONENTS

205 Materials (Electronic)

preparation and properties of con-ductors. semi -conductors, dielec-trics, magnetic. electro-optical,recording, and electro-magneticmaterials.

AG-SiO films, Surface Plasmons In granular- E B Priestley, B Abeles, R W Cohen(Labs.Pr) Physical Review B., Vol 12. No. 6(9/15/75) pp. 2121-2124

CdSe, Speed of response of photocurrents in- H Kiess, B Binggeli (Labs.Pr) RCAReview. Vol 36, No 3 (9/75) pp. 485-98

Mu by Brillouin scattering from thermalmagnons, Investigation of -W Jantz. JSandercock, W Wettling (Labs.Pr) 21st An-nual Conference on Magnetism and MagneticMaterials. Philadelphia PA (12/9-12/75)

GRANULAR METAL FILMS, Tunneling con-ductivity in - B Abeles (Labs.Pr) RCAReview. Vol 36. No 3 (9/75) pp 594-620

(In, Ga)P, Dislocations in vapor -grown Com-pesitionally graded -M S Abrahams, C JBuiocchi, G H Olsen (Labs.Pr) Journal ofApplied Physics Vol 40, No 10 (10/75) pp4259-70

MIXED -OXIDE WO. Moo electrochromicfilms and the nature of the color center -B W Faughnan (Labs.Pr) ElectrochemicalSociety Meeting, Washington. DC 15,76)

OPTICAL AND ELECTRICAL effects - DRedfield (Labs.Pr) Journal of ElectronicMaterials. Vol 4. No 5 (1975)

SIMPLICITY IN THEORY: an anecdotal ac-count of current injection in solids - intribute to Albert Rose - M.A. Lampert(Labs.Pr) RCA Review, Vol 36, No 3 (9,75)pp 444-68

Si() films, Hole photocurrents and electrontunnel injection induced by trapped holes In- R J Powell (Labs.Pr) Journal of AppliedPhysics Vol 40. No 10 (10/75) pp 4557-63

III -V COMPOUND materials and devices,Directions in -C.J.Nuese (Labs.Pr)Seminar. Wright -Patterson Air Force Base.Dayton. OH (12/16/76).

210 Circuit Devices andMicrocircuits

electron tubes and solid-statedevices (active and passive), in-tegrated. array and hybrid micro-circuits, field-effect devices,resistors and capacitors, modularand printed circuits. circuit inter-connection, waveguides andtransmission lines.

CHARGE -COUPLED DEVICES. Basic con-cepts of -W F Kosonocky. J E Carnes(Labs.Pr) RCA Review, Vol. 36, No. 3 (9/75)pp 566-93

CHARGE -COUPLED DEVICES for gigabitrecording applications - D.A. Gandolfo, Agoornard. L J Nicastro (ATL,Cam.)Proceedings (11/11-13/75) Electro-OpticalSystems Design Conference. Anaheim. CA

CHEMICALLY VAPOR -DEPOSITED FILMSfor microelectronics -W Kern (Labs.Pr)National Bureau of Standards, Gaithersburg,MD (1 8 761 Invited Lecture

GUARD LAYER DESIGN in COS/MOS struc-tures, Surface doping using ion implantationfor optimum - E.0 Douglas. A.G.F. Dingwall11-abs.Pri IEEE Transactions on ElectronDevices. Vol 22. No, 10 (10/75) pp 840-857

86

LINEAR AND DIGITAL ICs. Introduction toopeation, fabrication, application of - M.P.Rosenthal (Global,NY) Understanding In-tegrated Circuits. Hayden Book Co., RochellePark, NJ (11/75).

MICROSTRIP CIRCUITS and enclosures,Mechanical design of - D Olivier!(MSRD,Mrstn) Microwave Journal (11/75) pp.32. 33. 34. 36. 37 and 72.

RESISTORS, High -power density, thin-film- T. Faith (AED,Pr) IEEE Electrical(Elec-Ironic Insulation Conf. (Also IEEE Trans. onPart. Hybrids. and Packaging), Boston, MA.

TRANSISTOR for 14/12 GHz communicationsatellite transponders, A low -noisemicrowave -frequency - A B Bell, G. Lo(RCA Ltd Ste Anne) Canadian IEEE Con-ference, Toronto, Canada (10/75)

215 Circuit and NetworkDesigns

analog and digital functions in elec-tronic equipment: amplifiers, filters,modulators. microwave circuits. A -Dconverters, encoders, oscillators,switches, masers, logic networks,timing and control functions, fluidiccircuits.

MULTIPLIER. CMOS/SOS Serial -Parallel -D.Hampel, K McGuire, K. Prose(GCSD,Cam) IEEE Journal of Solid -StateCircuits. Vol SC 10, No. 5 (10/75) pp. 307-313.

RF PULSE MEASUREMENT techniques andpicture quality - Dr K. Praba (CCSD,Cam)IEEE Broadcast Symposium (2/25/75) IEEETransactions.

TRANSVERSAL FILTER design and applica-tion in satellie communications - Dr. K.Feher (RCA Ltd., Ste Anne) Midwest Sym-posium on Circuits & Systems, ConcordiaUniversity. Montreal. Canada (8/75).

TRAPATT diode arrays, Integrated - A.Rosen. H Kawamoto, J. Klatskin, E.L. Allen,Jr. (Labs,Pr) Journal MIT (10/75) pp. 841-843.

220 Energy and Power Sources

batteries. solar cells, generators,reactors, power supplies.

EPITAXIAL SILICON SOLAR CELLS -D'Aiello. P.H. Robinson, and H. Kressel(Labs.Pr) 1975 IEDM. Washington, DC (12/1-3/75).

EPITAXIAL SOLAR CELLS on silicon EFGribbon substrates - H. Kressel, R.V.D'Aielloand P.H. Robinson (Labs.Pr) Applied PhysicsLetters, Vol 28, No 3 (2/1/76) pp. 157-159.

SOLAR -CELL capacitance - A.R. Moore(Labs..Pr) RCA Review, Vol. 36, No. 3 (9/75)pp. 551-65.

225 Antennas and Propagation

antenna design and performance,feeds and couplers. phased arrays,randomes and antenna structures.electromagnetic wave propagation.scatter, effects of noise.

ANTENNA ARRAYS, Size and performancetrade-off characteristics in multiple - Dr.M.S. Siukola (CCSD,Cam) IEEE BroadcastSymposium (9/25/75) IEEE Transactions

ATMOSPHERIC PROPAGATION EFFECTSon a laser ship mast detector - H.C. Curtis(AED,Pr) Telemetering Conference,Silver Springs, MD

240 Lasers, Electro-Optical andOptical Devices

design and characteristics of lasers,components used with lasers,electro-optical systems, lenses, etc.(excludes: masers).

(AlGa) As heterolunction LED for fiber-opticcommunications, A new edge -emitting - H.Kressel, M Ettenberg (Labs, Pr) Proceedingsof the IEEE (9/75) pp. 1360-61

CHARGE -COUPLED IMAGERS, Method forvarying gamma In - P.A. Levine (Labs.Pr)ISSCC 76, Philadelphia, PA (2/19/76)Conference Digest.

GRATING COUPLING between low indexfibers and high Index film waveguides -J.M.Hammer, R A Bartolini, A Miller and C.C.Neil (Labs.Pr) 1976 Integrated Optics Con-ference. Salt Lake City, UT (1/12/76.).

HETEROEPITAXIAL SEMICONDUCTINGFILMS, The electrical characterization of -W.E. Ham (Labs.Pr) HeteroepitaxialSemiconductors for Electronic DevicesChapter VII, Edited by G.W. Cullenand, C.C.Wang (Springer-Verlag).

HETEROJUNCTION STRUCTURES tooptical devices, The application of - HKressel (Labs.Pr) Journal of ElectronicMaterials, Vol 4, No 5 (1975).

HOT ELECTRON transport, A criterion for theonset of -R.S.Crandall (Labs, Pr) RCAReview, Vol. 36. No. 3 (9/75) pp 531-41

INFRARED IMAGING with monolithic, CCD-addressed Schottky -barrier detectors:theoretical and experimental results - E.S.Kohn (Labs.Pr), S.A. Roosild, F.D. Shepherd,A.C. Yang (AF Labs) 1975 Int. Conf. on theApplication of ChargeCoupled Devices. SanDiego, Calif. (Oct. 1975): Conf. Proceedings(1975) pp. 59-69.

IRICON: a passive infrared camera tube -S.V. Forgue, L.D.Miller, J.O. Schroeder(Labs.Pr) IEEE Trans. on Electron DevicesVol. Ed -22, No. 10, (10/75) pp. 904-910.

MODERN OPTICS - the impact of the laser- R A Bartolini (Labs. Pr) Seminar, EastStroudsburg State College, EastStroudsburg, PA; (12/75).

MULTIPLEXED TWISTED NEMATICDEVICES, The effect of surface orientation onthe operation of - L.A. Goodman, D.Meyerhofer. S. DiGiovanni (Labs.Pr) S.I.D.Symposium. Washington, DC (4/75) IEEETransactions on Electron Devices.

PbO VIDICON, An approximate model of thebeam -blocking contact in a - A.M. Goodman(Labs.Pr) RCA Review, Vol 36, No. 3 (9/75)pp 408-24

PHOTOMULTIPLIERS for high temperatureapplications - D.E. Persyk, J.L. Ibaugh, A.F.McDonie. and R.D. Faulkner (PTD,Lanc) 1975Nuclear Science Symposium, SheratonPalace Hotel. San Francisco. CA (11/19-21/75)

SEMICONDUCTOR LIGHT SOURCES forfiber optical communication, Reliable - HKressel, I. Ladany, M. Ettenberg and H.F.Lockwood (Labs.Pr) 1975 IEDM, Washington,DC (12/1-3/75).

SENSORS, From camera tubes to solid-state- P K Weimer (Labs.Pr) RCA Review. Vol.36. No 3 (9/75) pp. 385-407

STRIPE WAVEGUIDES and couplingmodulators oflithium niobate-tantalate- W.Phillips. J.M. Hammer (Labs.Pr) 1976 In-tegrated Optics Conf.. Salt Lake City. UT(1/12-14/76)

VIDICONS. Semiconductor heterolunction -C R Wronski (Labs.Pr) RCA Review, Vol. 36,No. 3 (9/75) pp. 425-43.

245 Displays

equipment for the display of graphic.alphanumeric. and other data incommunications, computer military,and other systems. CRT devices,solid state displays, holographic dis-plays, etc.

FIBER-OPTIC INTENSIFIER SCREEN fortopography, An X-ray sensitive - R.W Smith(Labs.Pr) RCA Review Vol. 36, No. 3 (9/75) pp.632-37

PATTERN meaningfulness and patterndelectability - R.J. Mezrich (Labs.Pr) RCAReview, Vol 36, No. 3 (9/75) pp. 621-631.

250 Recording Components andEquipment

disk, drum, tape, film, holographicand other assemblies for &Act. im-age, and data systems.

LASER LINE PRINTER. Design con-siderations for a dye transfer - R.D. Scott(ATL,Cam) Electro-Optical Systems DesignConference, Anaheim, CA (11/11-13/75)Symposium Proceedings.

LASER RECORDER, How to specify a - R.F.Kenville (ATL,Cam) Electro-Optical SystemsDesign Conference, Anaheim. CA (11/11-13/75).

VIDEO TAPE RECORDER, Design andmanufacturing considerations for a newquadruples - J R West, G.S. Moskovitz(CCSD.Cam) University of SouthernCalifornia (10/8/75).

ANNULAR ARRAY IMAGING, Progress in -D. Vilkomerson, B. Hurley (Labs,Pr)Acoustical Holography, Vol. 6.

ULTRASONIC wavelronts, System forvisualizing and measuring - R S Meznch.K F Etzold, D H R Vilkomerson _abs,P0Acoustical Holography, Vol. 6.

280 Thermal Components andEquipment

heating and cooling componentsand equipment, thermal measure-ment devices, heat sinks.

THERMAL CONTROL LOUVER system bycomputer simulation, Analysis of theresponse of a bimetal -actuated -R Mancuso(AED.Pr) 8th AIAA/NASA/ASTM/IES SpaceSimulation Conf., Washington, D.C.

SERIES 300SYSTEMS, EQUIPMENT,& APPLICATIONS

310 Spacecraft and GroundSupport

spacecraft and satellite design,launch vehicles, payloacs, spacemissions, space navigation.

AEROSAT Problems, Present - C.R. Hume,J Chaumeron (AED,Pr) 6th U.S. - EuropeanConf., Monte -Carlo. Monaco.

TELEVISION SYSTEMS for remotemanipulation - L Freedman, W H Crooks.P P Coan (AEC/J.0 Human Factors Soc Ann.Meeting. Dallas. TX

315 Military Weapons andLogistics

missiles, command and control.

REMOTELY PILOTED VEHICLE - D. Shore(MSRD,Mrstn.) Greater Omaha AFCEAChapter, Omaha. NE (10(21/75).

TRANSFERRING USABLE INFORMATIONto the military maintenance man - M LJohnson, C F Ma:thews (ASD.Burl ) BostonChapter, Society of Logistics Engineers.Hanscom Field Officer's Club (2/4/75)

TRANSFERRING USABLE INFORMATIONto the military maintenance man -Johnson, C.F. Matthews (ASD,Burl) 11th Intl.Logistics Symposium, Valley Forge, PA (8/15-19/76).

320 Radar, Sonar, and Tracking

microwave, optical, and othersystems for detection, acquisition.tracking, and position indication.

RADAR multipath and resolution techniques- S.M. Sherman (MSRD,Mrstn) (Complex -Angle Monopulse (Lecture 7): MultipathApplication of Complex -Angle Monopulse(Lecture 8) Microwave Journal, Boston, MA(10/22-24/75).

325 Checkout, Maintenance,and User Support

automatic test equipment, (ATE).maintenance and repair methods.

BUS MAINTENANCE, Application ofautomatic test equipment to -J.M.Laskey.R Barry (ASD.Burl) SAE Fleet Week 9/75

DIAGNOSIS of internal combustion enginefaults through remote sensing, Non -contact- S.C. Hadden, L R Hulls. E M Sutphin(ASD,Burl) 1976 Automotive EngineeringCongress and Exposition, SAE, Detroit MI(2/23-27/75).

EQUATE Atlas implementation and FutureExpectations - R E Turkington (ASD.Burl)IEEE 1975 Automatic Support Systems Sym-posium, Westbury, New York (10/28-30/75)

340 Communications

industrial, military, commercialsystems. telephony, telegraphy andtelemetry, (excludes: television, andbroadcast radio).

TIMING RECOVERY TECHNIQUE for burstmodem application, A new symbol - Dr. K.Feher (RCA Ltd., Ste Anne) Canadian IEEEConference, Toronto, Canada (10/75).

TRANSMISSION TRENDS, Simultaneousdigital/analog radio and cable - Dr. K Feher(RCA Ltd., Ste Anne) World Telecommunica-tion Forum, Geneva, Switzerland (10/75).

345 Television and Broadcast

television and radio broadcasting.receivers, transmitters, and systems,television cameras, recorders, studioequipment.

TELEVISION TEST WAVEFORMS, Measur-ing and interpreting transmitted - T MGluyas (CCSD.Cam) AFCCE, Washington.DC (12/18/75)

87

360 Computer Equipment

processors. memories, andperipherals.

MICROPROCESSORS in consumer markets- Ft 0 Winder (Labs,Pr) Computer. Vol 9,No 1 i1 76) pp 39-40.

370 Computer Programs(Scientific)

specific programs and techniquesfor scientific use, computation.simulation, computer aided design.

MOS design automation -W F Heagerty(ATL.Cam) Hardened Guidance & WeaponDelivery System Conference. Baltimore. MD.(10,23/75)

STANDARD CELL TECHNIOUES forcomputer -aided design of LSI arrays - ASmith (ATL.Cam) IEEE SOS TechnologyWorkshop. Lake Tahoe, CA (9/17-19/75)

Author Index

Subject listed opposite each author's nameindicates where complete citation to his papermay be found in the subject index. An authormay have more than one paper for eachsubject category.

RCA Laboratories

Abeles. B., 205Abrahams. M.S., 205Allen. Jr., E.L., 215Ban. V.S., 170Bartolini, R.A., 240Binggeli, B.. 205Buiocchi. C.J., 205Carlson. D.E.. 105Carnes. J.E., 210Cohen. R.W., 205Crandall, R.S., 240DAiello, R.V., 220

DiGiovanni, S., 240Dingwall, A.G.F., 210Douglas, E.C., 210Ettenberg, M., 240Etzold, K.F., 255Faughnan, B.W., 205Forgue. S.V.. 240Gerber. J.. 160Gilbert, S.L., 170Goodman. A.M., 240Goodman, L.A., 240Ham, W.E., 240Hammer. J.M., 240Hurley, B., 255Jantz, W., 205Kane. J., 170Kawamoto, H., 215Kern. W., 105. 170, 210Kiess. H., 205Klatskin, J., 170. 215Kohn. E.S., 240Kosonocky, W.F., 210Kressel, H., 220, 240Ladany, I.. 240Lampert. M.A., 205Larach, S., 160Levine, PA., 240Lockwood. H.F., 240Meyerhofer, D., 240Mezrich, R.J., 245.255Miller, A., 240Miller, L.D., 240Moore, A.R., 220Neil, C.C.. 240Nuese, C.J., 205Olsen. G.H., 205Phillips, W., 240Powell. R.J., 205Priestley, E.B., 205Redfield, D., 205Robinson. P.H., 220Rose, A., 205Roosild, S.A., 240Rosen. A., 170Sandercock, J., 205Schroeder, J.0., 240Schweizer. H.P., 170Shepherd, F.D., 240Smith, R.W., 245Vilkomerson, D., 255Weimer, P.K., 240

Wettling, W.. 205Williams, R., 125Winder, R.O., 360Wronski, C.R., 240Yang. A.C.. 240

Solid State Division

Faulkner, R.D., 240lbaugh. J.L., 240McDonie, A.F., 240Persyk, D.E., 240

Global Communications, Inc.

Rosenthal, M.P., 210

Record Division

Dogar, A., 180Mishra, D., 180Pearson. J.C., 180

Missile and SurfaceRadar Division

Bowen. J.H., 160Buckley, M.W., 180Hurley, J.W., 180Olivieri, D., 210Pschunder, R., 130Sherman. S.M.. 320Shore, D., 315

RCA Limited

Bell. A.B.. 210Feher, Dr., K.. 215,340Lo. G., 210

Automated Systems Division

Barry. R.. 325Behling, H.H., 130Bosselaers, R.J., 140Hadden, S.C., 325Hulls. L.R., 325Johnson. M.L., 175

Laskey. J.M., 325Matthews, C.F., 315Sutphin, E.M., 325Turkington, R.E., 325

Astro-Electronics Division

Chaumeron, J., 310Curtis. H.C., 225Coen, P.P., 310Crooks. W.H., 310Faith. T., 210Freedman, L., 310Hume, C.R., 310Mancuso, R., 280

Government and CommercialSystems Division

Hempel, D., 215Hendrickson, R.J., 150McGuire, K., 215Miller, R.S., 180Prost, K., 215VanKeuren, E.. 150

Commercial CommunicationsSystems Division

Gluyas, T.M., 345Moskovitz. G.S., 250Probe, Dr., K., 215Siukola. Dr., M.S., 225West, J.R., 250

Advanced TechnologyLaboratories

Boornard, A.. 240Gandolfo. D.A., 210Gehweiler, W.F., 135Heagerty, W.F., 370Kenville, R.F.. 250Nicastro, L.J., 210Pridgen, J.I., 135Scott, R.D., 250Smith, A., 370

Patents

Grantedto RCA Engineers

Laboratories

Method for vapor -phase growth of thin filmsof lithium niobate- B J. Curtis, H R. Brunner(Labs., Zurich, Switz.) U.S. Pat. 3911176,October 7 1975.

Two color information record - R.E. Flory(Labs . Pr.) U S Pat 3911453. October 7,1975

Interlaced readout of charge stored In

charge -coupled image sensing array - P.A.Levine, J.E. Carnes (Labs.. Pr.) U.S. Pat3911467, October 7, 1975.

Color picture/sound record - E. 0. Keizer(Labs. Pr) U S Pat 3911476, October 7,1975

Optical fiber to planar waveguide coupler -J M Hammer (Labs.. Pr ) U S Pat 3912363,October 14, 1975

AC deformable mirror light valve -W RRoach (Labs Pr.) U.S Pat 3912370. October14. 1975

Color image intensification and projectionusing deformable mirror light valve - IGorog, A H Firester (Labs. Pr ) U.S. Pat3912386, October 14, 1975

Microstrip carrier for high frequencysemiconductor devices - A Rosen, J FReynolds (Labs., Pr.) U.S.Pat. 3913040. Oc-tober 14, 1975

Method of making a transmissionphotocathode device - H Kressel (LabsPr.) U S Pat 3914136, October 21, 1975

Process for forming transition metal oxidefilms on a substrate and photomaskstherefrom - J Kane, H Schweizer (Labs.,Zurich, Switz.) U.S Pat 3914515, October 21,1975

Voltage regulator for deflection circuit -GForster. H Keller (Labs., Zurich Switz ) U.S.Pat. 3914653, October 21. 1975

Gated oscillator having constant ge DCoutput voltage during on and off times - C.R.Carlson. I Gorog. P.M.Heyman (Labs., Pr.)U.S.Pat 3914711. October 21. 1975

Crystals for recording phase holograms -J J. Amodei. W. Phillips (Labs.. Pr.) U.S. Pat.3915549, October 28. 1975

Method for determining an accurate align-ment of a laser beam -K G. Herngvist (Labs.,Pr.) U S Pat 3915574, October 28, 1975

Stabilized autocatalytic metal depositionbaths -N Feldstein, J A Weiner (Labs., Pr.)U S Pat 3915717, October 28, 1975.

Photomask bearing a pattern of metal platedareas - N Feldstein (Labs.. Pr.) U.S.Pat3916056. October 38, 1975

Asymmetrically excited semiconductor in-jection laser - I Ladany, D.P. Marinelli. HKressel, & V M Cannuli (Labs ,Pr.) U.S Pat3915339. October 28. 1975.

Temeprature-compsensated surface -propagated wave energy device -M Toda,

S. Osaka (Res. Lab.. Tokyo, Japan) U.S. Pat.3916348. October 28. 1975.

Frequency -encoded focussed imagehologram record - M.T. Gale (Labs., Zurich,Switz.) U S Pat 3917378, November 4, 1975

Protection Circuit - S. Weisbrod (Labs.,Prd ) U.S Pat 3917980, November 4, 1975.

Charge transfer circuits - P.K. Weimer(Labs . Pr) U.S Pat. 3919468, November 11,1975

Automatic target control system for a televi-sion camera tube -D P Dorsey. W E Rodda,R.S. Filson (Labs.. Pr.) U S Pat 3919472,November 11, 1975.

Loop controller for a loop data com-munications system - N F MAxemchuk(Labs , Pr.) U.S.Pat 3919484. November 11,1975

Electroluminescent cell with a currentlimiting layer of high resistivity - J.J. Hanak(Labs., Pr.) U.S. Pat. 3919589. November 11,1975.

Automatic peaking apparaturs -Bingham (Labs.. Pr.) U.S. Pat, 3919714,November 11, 1975.

88

Method of making a semiconductor device -R H Dean (Labs., Pr.) U.S Pat. 3920861,November 18. 1975

Field emitting device and method of makingsame - J.D. Levine (Labs.. Pr.) U.S. Pat.3921022. November 18. 1975.

Frequency multiplier circuit - D.E. Mahoney(Labs.. Pr.) U.S.Pat 3921056. November 18,1975

Method of bonding two bodies together -J D. Knox (Labs.. Pr.) U S Pat. 3921885.November 25. 1975

Method for coating ferrous -metal mask forcathode-ray tube - G.S. Lozier (Labs., Pr )U.S. Pat. 3922394, November 25, 1975.

Method of making a semi -transparentphotomask -G.L.Schnable, N Feldstein(Labs., Pr.) U.S. Pat. 3922420. November 25,1975

Circuit for producing odd frequency multipleof an input signal - W.G. McGuff in, R.W.Burgen (Labs.. Pr.) U.S. Pat 3922593,November 25, 1975.

Electroluminescent semiconductor device -J.I. Pankove (Labs., Pr.) U.S. Pat. 3922703.November 25, 1975.

Liquid crystal compositions - C.S. Oh, E.F.Pasierb (Labs., Pr.) U.S. Pat. 3923685.December 2. 1975.

Method of making an optical waveguide -M.T. Duffy. D.J. Channin, J. M. Hammer(Labs.. Pr.) U.S. Pat. 3924929, December 2.1975.

Noise cancelling magnetic antenna for usewith watercraft - K.H. Powers, L.L. Stetz, Jr.(Labs., Pr.) U.S. Pat. 3913107, October 14,1975; Assigned to the U.S. Government.

Heat sinks for microstrip circuit - HKawamoto (Labs.. Pr.) U.S. Pat. 3908188.September 23, 1975: Assigned to the U.SGovernment

Noise suppression arrangement for com-munication receivers -H.G. Schwarz (Labs..Pr.) U.S. Pat 3671867; Assigned to the U.S.Governemnt.

Method of fabricating an apertured mask for acathode-ray tube -H B Law 1Lalos, Pr U.S.

Pat 3923566. December 2. 1975

Solid State Division

Relaxation oscillator having stable pulsewidth - P. Goodheart, A.G. Sheng (SSD,Som.) U.S Pat. 3911377. October 7, 1975.

Signal comparison circuits - R.H. !sham(SSD.Som ) U S Pat 3012942, October 14,1975.

Direct current protection circuit - R.P.Fillmore (SSD. Som.) U.S. Pat. 3912977,October 14. 1975

Voltage regulator circuit with relatively low -power consumption - R.P. Fillmore,(SSD,Som ) U S Pat 3913006, October 14.1975.

Current proportioning circuit -A.J. Leidich(SSD,Som ) U.S Pat 3914684. October 21,1975

Feedback amplifier - J.Craf. (SSD,Som.)U.S. Pat. 3914704, October 21, 1975.

Differential amplifier - B. Zuk (SSD, Som.)U.S. Pat 3916333. October 28, 1975.

Modulator - H.A. Wittlinger (SSD, Som.)U.S Pat. 3916346, October 28, 1975.

System for eliminating substrate bias effect infield effect transistor circuits - R.C. Heuner,D K Morgan, G W Steudel (SSD,Som.) U.S.Pat 3916430, October 28, 1975.

Bipolar integrated circuit transistor withlightly doped subcollector core - H Kha-lezadeh (SSD.Som.) U.S. Pat. 3916431, Oc-tober 28, 1975.

Black ceramic body - L.F. Hart, R.L. Buttle(SSD.Som.) U.S. Pat. 3918982, November 11,1975.

Alternating current line voltage supply isola-tion using deflection system outputtransformer - W.F Dietz (SSD Som) U SPat. 3920892, November 18, 1975

Biasing current attenuator - A.A. Ahmed(SSD,Som.) U.S. Pat. 3921013, November 18.1975.

Operational transconductance amplifier -C.F. Wheatley, Jr.. H.A. Wittlinger(SSD,Som.) U.S. Pat. 3921090 November 18,1975.

Amplifier protection system - M.V. Hoover(SSD.Som ) U.S Pat 3924159. December 2,1975.

Target mounting structure for use in cameratube - S.A. Ochs (SSD.Lanc.) U.S. Pat.1919582, November 11, 1975

Radiant energy device mount - D.B. Kaiser(SSD/EOD,Lanc U S Pat 3916336. October28. 1975

Quartz to glass seal- R.D. Faulkner(SSD/E0D.Lanc.) U S Pat. 3923189,December 2. 1975

Consumer Electronics

Composite printed circuit board - L.P.Thomas (CE.Indpls.) U.S Pat. 3912849. Oc-tober 14. 1975

Stylus arm pivot - M.E. Miller (CE,Indpls.)U.S.Pat 3912857, October 14, 1975.

Electromagnetic radiation filter for coaxiallyfed hot chassis television receiver - G.W.Carter, S.E. Hilliker (CE.Indpls.) U.S Pat.

3913038. October 14, 1975.

Timing signals for velocity error correction -C D Boltz, Jr (CE. Indpls ) U.S Pat 3914542,October 21. 1975

Controllable reference supply for televisiontuners -W W Evans (CE.Indpls.) U.S.Pat.3914696, October 21, 1975.

Detachable pickup arm magnetic coupling -B.K. Thylor, W.H. Garlin (CE,Indpls ) U.S.Pat. 3917903, November 4, 1975.

Signal translation using the substrate of aninsulated gate field effect transistor - J.B.George (CE. Indpls.) U.S. Pat. 3924192,December 2, 1975.

Electronic oscillator - J. Craft (CE,Som.)U.S. Pat. 3924202, December 2. 1975.

Automatic luminance channel bandwidthcontrol apparatus responsive to theamplitude of color image informabon -L AHarwood (CE,Som.) US. Pat. 3924266,December 2, 1975

Commercial CommunicationsSystems Division

Four -wise conference circuit - L _F Goeller,Jr. (CCSD.Cam ) U S Pat. 3912867, October14. 1975.

Signal sensing for film projectors - A EJackson, G A Singer (CCSD,Cam U.S.Pat3915566, October 28. 1975

Multinomial processor system - 3. Hampel.R.W. Blasco (CCSD.Cam.) U S. Pat 3922536.November 25. 1975.

Automated Systems Division

Phase locked loop including an arithmeticunit - R.J. Bosselaers (ASD,Burl ) U.S. Pat.3913026, October 14, 1975.

Staircase waveform generator - HLogemann. Jr (ASD,Burl ) U.S.Pat 3919649,November 11, 1975

Picture Tube Division

Plural -beam color picture tube with improvedmagnetic convergence structure - J Evans,Jr (PTD,Lanc.) U.S. Pat 3916244, October28. 1975.

Purity adjustment for color display system -R.L. Barbin (PTD,Lanc.) U.S. Pat. 3916437,Octobe- 28, 1975.

Densitometer for measuring average aper-ture diameter - G.S. Gadbois (PTD,Lanc.)U.S. Pat. 3918815, November 11, 1975

Method of degassing a cathode-ray tube priorto sealing - F.S. Swaicki (PTD.Scranton)U.S Pat 3922049, November 25, 1975

Method for applying organic polymericcoating composition to ferrous metal sur-faces - B K Smith (PTD. Lanc ) U S. Pat3922395, November 25. 1975.

Special Contract Inv.

Dyeing station in an apparel'us for con-tinuously dyeing fibrous material - B.M.Childers. C. Fesperman, Jr. (Cor. Ird., Dalton.GA) U.S. Pat. 3913359, October 21, 1975.

Graphic Systems

Electronic identification system - R JKlenscn (IS.Dayton. NJ) U.S.Pat 3914762.October 21. 1975

Electronic Components

Device and method for restoring cathodeemission in a thermionic electron tube -HKozicki (EC. Atlanta, GA) U.S.Pat 3915533,October 28. 1975

Advanced TechnologyLaboratories

Complementary field-effect transistoramplifier - W.F Gehweiler (ATL.Cam.) U SPat 3914702. October 21, 1975.

Missile and SurfaceRadar Division

Antenna scanning - R.W. Howery(MSRD.Mrstn.) J.S. Pat. 3916415, October28, 1975.

Approximator for square root of sums ofsquares - J A Lansford (MSRD.Mrstn ) U.S.Pat. 3922540. November 25. 1975.

Insertion phase trim method - L.J. Lavedan.Jr (MSRD.Mrstr, ) U.S Pat. 3911380.October7, 1975. Assigned to the U.S. Government.

Accurate normalization for a monopulseradar - D.N. Thomson (MSRD.Mrstn.) U.S.Pat. 3921173, November 18, 1975; Assigned tothe U.S. Government.

Government CommunicationsSystems Division

Frequency measurement by coincidencedetection with standard frequency - E JNossen, E. R. Starner (GCSD.Cam.) U S Pat.3924183, December 2. 1975.

Palm Beach Division

Embossed character reader - E.H. Del Rio(PBD,Palm Beach Gardens) U.S. Pat.

3917925, November 4. 1975.

Web advance mechanism - K.D. Peters(PBD. Palm Beach Gardens) U.S. Pat.

3921879, November 25, 1975.

Avionics Systems Division

Digital feedback relay controller - G.A.Jones (ASD,Van Nuys) U.S Pat. 3922587,November 25, 1975.

Low voltage bus -operated overvoltageprotection system - F.C. Easter (ASD.VanNuys) U.S Pat 3916262, October 28. 1975;Assigned to the U.S Government.

APC Corp.

Thermoplastic heat responsive fire vent ap-paratus - E.S. Naidus (APC Corp..Hawthorne, NJ) U.S. Pat. 3918226, November11. 1975.

Astro-ElectronicsDivision

Satellite propellant management system -R J. Lake, Jr. ,AED,Pr.) U.S. Pat 3923188.December 2. 1975

RCA Limited

Fabry-Perot polarization laser beammodulator -A L. Waksberg (RCA Ltd , Mon-treal) U S Pat. 3918007. November 4, 1975

High -resolution digital generator of graphicsymbols with edging - R.J. Clark (RCS Ltd..Montreal) U S Pat. 3918039. November 4,1975

Peak voltage detector circuits - R.T. Griffin(RCA Ltd. England) U S Pat. 3921010.November 18. 1975

89

Datesand

Deadlines

Calls for papers-be sure deadlines are met

Ed. Note: Calls are listed chronologically by meetingdate. Listed after the meeting title (in bold type) are thesponsor(s), the location, and deadline information forsubmittals.

JAN 30 - FEB 4. 1977 - Power Engineering SocietyWinter Meeting (PE) Statler Hilton Hotel. New York. NYDeadline Into: (paper) 9/1/76 to IEEE, 345 East 47thStreet. New York. NY 10017

FEB 1977 - Low -Noise Technology (Special Issue ofthe) IEEE Transactions on Microwave Theory andTechniques Deadline info: (paper) 4 copies 6/1/76 toJesse Taub. AIL Division of Cutler -Hammer, Walt Whit-man Road Melville, NY 11746

MARCH 22-25, 1977 - ELECTRO (EEE InternationalConvention) IEEE. ERA) Americana Hotel. Coliseum.New York, NY Deadline infor: (Sessions) 7/22/7610W CWeber, Jr . ELECTRO. Suite 1108, 3600 Wilshire BlveLos Angeles, CA 90010.

MAY 25-27. 1977 - 1977 Spring Meeting 8 Star Sym-posium - SNAME "Energy Research in the Oceans"Deadline Info: (abst) 400-500 words - 15 copies toRADM N. Soneshein, USN (Ret.), Assistant to thePresident. Global Marine Development. Inc 4100MacArthur Blvd Newport Beach. CA 92660

AUG 24-26, 1976 - Product Liability Prevention Con-ference (R et al) Newark College of Engineering,Newark. NJ Deadline Info: (paper) 11/1/76 to JohnMihalasky. Newark College of Engrg , Newark, NJ 07102.

SEPT 13-15, 1976 - The Spirit of Standardization (SES)Chalfonte-Haddon Hall. Atlantic City. NJ Deadline Info:(abst) 200-500 words 3/1/76 to J.J. O'Donnell. ProgramChairman. Sperry Univac - M.S. 1A -E4-109. P 0 Box500, Blue Bell, PA 19422.

SEPT. 29 - OCT. 1, 1976 - Ultrasonics Symposium (SU)Annapolis Hilton Hotel, Annapolis. MD Deadline into:(ms) 7/1/76 to P.H. Carr, AFCRL, LG Hanscom Field.Bedford, MA 01730.

OCT 11-14. 1976 - Industry Applications SocietyAnnual Meeting (IA. Chicago Section) Regency HyattO'Hare. Chicago, IL Deadline info: (abst) 3/8/76 to R VWacher, Aluminum Co of Am . 1501 Alcoa BldgPittsburg. PA 15219

OCT. 13-15, 1976 - Software Engineering Conference(C) Jack Tar Hotel, San Francisco. CA Deadline infor:(ms) 4/15/76 to C.V. Ra. Dept of of EE & Computer Sci

OCT. 13-15. 1976 - Software Engineering Conference(C) Jack Tar Hotel, San Francisco, CA Deadline Into:(ms) 4/15/76 to C V. Ramamoorthy. Dept of EE &Computer Sci.. Univ. of Calif.. Berkeley. CA 94720.

OCT 19-21, 1976 - System Designs Driven by Sensors(AIAA) Pasadena, CA Deadline Info: (abst) 4 copies3/1/76 to Dr. Robert Greenberg. Department of DefenseResearch and Engineering, Washington. DC 20301

OCT 21-22.1976 - Canadian Communications 8 PowerConference (Canadian Region, Montreal Section)Queen Elizabeth Hotel, Montreal. Quebec, CanadaDeadline Info (A&S) 3 copies 3/1/76 to J.J. Archam bau it,CP/PO 958. Succ. "A". Montreal, Que. H3C 2W3 Canada.

OCT. 25-26. 1976 - Joint Engineering ManagementConference (EM. EIC et al) Hyatt Regency Hotel,Toronto, Ontario, Canada Deadline Info: (paper) 6/30/76to K.L. Coupland. Ministry of Colleges & Univ., 60 SirWilliam's Lane. Islington, Ont. Canada M9A 1V3.

NOV 6-10. 1976 - Engineering in Medicine 8 BiologyCont. (EMB, AEMB) Sheraton Hotel, Boston, MADeadline Info: (abst) 4/15/76 to AEMB Suite 1350, 5454Wisconsin Ave., Chevy Chase. MD 20015

NOV. 28 - DEC 3, 1976 - Joint Cryogenic Meeting at1976 ASME Winter Annual Meeting Statler Hilton Hotel,New York, NY Deadline Info: (abst) 300 words 1/30/76(draft ms) 4/20/76 (final drafts) 6/30/76 to Prof. LeonardWendel, Dept of Chemical Engr., Lehigh University,Bethlehem, PA 18015 and Prof. Geoffrey G. Haselden,Dept. of Chemical Engr., University of Leeds. Leeds LS29JT, England.

DEC. 1-3. 1976 - Decision and Control - AdaptivePr ICS. SIAM) Sheraton Sand Key, ClearwaterBeach, FL Deadline Info: (A or MS) 4/1/76 to Earl Barnes.IBM TJ Watson Res. Ctr., POB 218, Yorktown Heights,NY 10598.

DEC 6 - 10, 1976 - Submillimeter Waves and TheirApplications (IEEE) San Juan, Puerto Rico Deadlineinfo: (ms) 8/1/76 to K J Button. MIT, Nat'I Magnet Lab ,

Cambridge. MA 02139

Dates of upcoming meetings-plan ahead

Ed. Note: Meetings are listed chronologically. Listedafter the meeting title (in bold type) are the sponsor(s).the location, and the person to contact for more informa-tion

APRIL 12-14, 1976 - Acoustics, Speech 8 SignalProcessing Intl Conference (ASSP. Phila. Section)Marriott Hotel, Phila., PA Prog Info: Thomas Martin.Threshold Tech. Route 130, Union Landing Rd., Cin-naminson, NJ 08077.

APRIL 20-22, 1976 - Computer Software Engrg.Reliability, Management Design (ED. R) Caesars Palace.Las Vegas. NV Prog Info: J H Martin, SS Draper Lab.. 75Cambridge Parkway. Cambridge. MA 02142

APRIL 22-24. 1976 - Milwaukee Symposium onAutomatic Computation 8 Control (SMC. MilwaukeeSection, Univ of Wisc ) Milwaukee. Wisc. Prog info:Warren Koontz. Rm 2A-208. Bell Labs., Whippany, NJ07981

APRIL 26-27, 1976 - Southeastern Symposium onSystem Theory (C. et al) Univ. of Tenn.. Knoxville, TennProg info: T W Reddoch, Div. of EE, ERDA, Washington.DC 20545.

APRIL 26-27. 1976 - Pittsburgh Conference on Model-ing and Simulation (SMC, Univ. of Pitts , ISA) Prog Info:W G Vogt -M. H Mickle, 348 Benedum Engrg. Hall, Univ.of Pittsburgh. Pittsburgh, PA 15261

APRIL 26-28. 1976 - Electronic Components Con-ference (PHP, EIA) Jack Tar Hotel, San Francisco, CAProg Info: J H. Powers. Jr., IBM -System Pdts. Div. Dept .104. Bldg 414-700, Hopewell Junction. NY 12533.

APRIL 27-29, 1976 - Optical Computing Intl Con-ference (C. et al) Capri, Italy Prog info: D P. Casasent.CMU. Dept. of EE. Pittsburgh, PA 15213

APRIL 27-29, 1976 - Circuits and Systems Intl Sym-posium (CAS. VDE) Technical Univ . Munich, GermanyProg info: Alfred Fettweis, Lehrstuhl furNachnctentechnik, Univ. of Bockhum. PO 2148, 0-4630Bochum. F.R. Germany.

APRIL 29-30, 1976 - Appliance Technical Conference(IA et al) Purdue Univ., W. Lafayette. IN Prog info: R GLeonard, R.W. Herrick Labs.. Purdue Univ.. W Lafayette,IN 47907.

MAY 1-6. 1976 - 78th Annual Meeting and Exposition(ACS) Convention -Exposition Center, Cincinnati. OHProg Info: The American Ceramic Society. Inc.. 65Ceramic Drive. Columbus, OH 43214.

MAY 3-6. 1976 - Offshore Technology Conference(OEC et al) Astrohall. Houston, TX Prog Info: OTC, 6200N Central Expressway. Dallas, TX 75206

MAY 3-7, 1976 - Subscriber Loops & Services Inter-national Symposium (COM. Region 8, IEE, UKRI Sectionet al) London. England Prog info: A.G. Hare. Head ofTSS8 PO Telecommunications Hdos., Rm 702.Cheapside House, London EC2 England.

MAY 5-7. 1976 - Carnahan Conference on CrimeCountermeasures (AES, Univ. of Kentucky) Univ. ofKentucky, Lexington. NY Prog Info: J.S. Jackson, Dept.of EE, Univ. of Kentucky, Lexington, KY 40506

MAY 11-14, 1976 - ELECTRO (IEEE InternationalC lion) (IEEE, ERA) Boston Sheraton Hotel &Hynes Audit Boston. MA Prog Info: W C Weber, JrELECTRO, 31 Charming St.. Newton, MA

MAY 16-19, 1976 - Industrial Power Conference(ASME) Memphis, TN Prog Info: Maurice Jones, Direc-tor. Public Relations, ASME, 345 E 47th St New York,NY 10017

MAY 17-19, 1976 - General Engineering Conference(ASME) St. Louis. MO Prog info: Maurice Jones, Direc-tor. Public Relations, ASME, 345 E. 47th ST., New York.NY 10017

MAY 18-20, 1976 - Aerospace & Electronics Conference(NAECON) (AES et al) Dayton Cony. Ctr Dayton, OHProg Info: J.E. Singer, NAECON 76.140 Monument Ave..Dayton, OH 45402

MAY 24-26, 1976 - Plasma Science Intl Conference(NPS et al) Univ. of Texas, Austin, TX Prog Info: E.J.Powers, Dept. of EE, Univ of Texas at Austin, Austin, TX78712.

MAY 24-26, 1976 - Lubrication Symposium (ASME)Atlanta. GA Prog Info: Maurice Jones, Director, PublicRelations. ASME. 345 E 47th St., New York. NY 10017

MAY 25-27. 1976 - Laser 11 Electro-Optical Systems(CLEOS) (OE Council, OSA) Town & Country, SanDiego, CA Prog info: M.E.Rabedeau. IBM -F 55/015. SanJose, CA 95193.

MAY 25-28, 1976 - Intl Symposium on Multi -ValuedLogic (C, ACM/ICS, Utah State Univ et al) Utah StateUniv., Logan. UT Prog info: Z Vranesic, Dept of EE.Univ. of Toronto, Toronto. Ontario. Canada M5S 1A4

MAY 27. 1976 - Trends and Applications: Micro andMini Systems (C. Washington Section. NBS) Nat'l.Bureau of Standards. Gaithersburg. MD Prog Info: J WBenoit. MITRE Corp., Westgate Res Park, McLean , VA22101

JUNE 7-10, 1976 - National Computer Conference (C,AFIPS) New York, NY Prog info: Stanley Winkler, IBMCorp., 18100 Fredrick Pike. Gaithersburg. MD 20760

JUNE 8-10, 1976 - Power Electronics Specialists Con-ference (AES) NASA Lewis Res. Ctr . Cleveland, OHProg Info: Forest Golden. Gent Elec Co . Box 30, WGenesee St.. Auburn, NY 12021

JUNE 14-16, 1976 - Electrical Insulation Intl. Sym-posium (El) Queen Eliz. Hotel. Montreal, PO, CanadaProg info: E.0. Forster. Exxon Res & Engrg. Co., POB45. Linden. NJ 07036

JUNE 14-17, 1976 - Applied Mechanics Conference(ASME) Salt Lake City, UT Prog info: Maurice Jones.Director, Public Relations, ASME, 345 E. 47th ST., NewYork, NY 10017.

JUNE 14-16, 1976 - International Microwave Sym-posium (MTT) Cherry Hill, NJ Prog Info: Martin Caulton,RCA, Princeton, NJ 08540

JUNE 14-16, 1976 - Intl Conference on Com-munications (CSCB) Marriott Motor Hotel, Philadelphia,PA Prog Info: Ralph Wyndrum, Bell Labs Whippany Rd.,1B306, Whippany, NJ 07981

JUNE 14-18. 1976 - Quantum Electronics Intl Con-ference (ED, MTT, AIP et al) RAI Congress Ctr., Amster-dam. The Netherlands Prog Info: 0 Svelto. c/o H.J.Frankena, Dept. Applied Physics. Lorentzweg 1 DelftNetherlands.

JUNE 15-18, 1976 - Jt. Intl. Magnetics 8 Magnetism 8Magnetic Materials Cont. ( INT ERMAG & M') (MAG, A I P)Pittsburgh Hilton Hotel. Pittsburgh. PA Prog info: H CWolfe, AIP, 335 E 45th St New York, NY 10017

JUNE 21-24, 1976 - Summer Annual Meeting (ASME)Quebec City. Oue. Prog Info: Maurice Jones. Director,Public Relations, ASME. 345 E 47th St., New York, NY10017

90

Two RCA men elected IEEE FellowsThe two RCA men cited herein have been honored for their professional achievements by being elected Fellows of theInstitute of Electrical and Electronics Engineers. This recognition is extended each year by the IEEE to those who navemade outstanding contributions to the field of electronics.

Walter F. KosonockyFor contributions to solid-statelogic, memory, and imaging.

Dr. Kosonocky, received the BSEE and MSEE from NewarkCollege of Engineering in 1955 and 1957, respectively, and theScD in Engineering from Columbia University in 1965. FromJune 1955, he has been employed at RCA Laboratories, whereafter one year as a Research Trainee, he became a Member ofthe Technical Staff and since that time he has been doingresearch on new solid-state devices and circuits. Since thespring of 1970, Dr. Kosonocky has been working on the study ofperformance limitations of CCD's and the development ofcharge -coupled devices for image sensing, memories, andsignal processing applications. This work has included studiesof device performance limitations (interface trapping, fringingfield drift, free -charge transfer, noise, and low light level

imaging) and the development of processing technology forsurface and buried channel CCD's, input and output tech-niques, and new structures for CCD image sensors and d gitalmemories. Dr. Kosonocky received two RCA LaboratoriesAchievement awards in 1959 and 1963, and was awarded a DavidSarnoff Fellowship for the academic year 1958-1959. He haspublished 24 technical papers, and has been issued 23 U.S.patents. Dr. Kosonocky is a member of Tau Beta Pi, Eta KappaNu, Sigma Xi, the American Ordnance Association, a SeniorMember of IEEE, and a member of the IEEE Solid State CircuitCommittee.

Dr. Sherman received the BA and the MA in Physics in 1934 and1939, respectively, and the PhD in Electrical Engineering in1965, all from the University of Pennsylvania. He served in theUS Army Signal Corps and Air Forces from 1942 to 1946 wherehe led the development of two airborne fire control radarsdeployed during and after World War II. Since 1955 Dr. Shermanhas been directly associated with the Systems Engineeringorganization of RCA Missile and Surface Radar Division insupervisory and staff positions. His assignments and res-ponsibilities have been devoted principally to system analysisand system concepts for sensor development and application.He was responsible for major portions of the TALOS LandBased Weapon System and the Ballistic Missile Early WarningSystem. He also proposed basic concepts from which the theoryand practice of radar signature analysis (space object identifica-tion) has evolved. In recent years he has devoted considerableeffort to analysis and concept formulation for advanced radarsystems. Two of the areas in which he has gained recognitionare theory and application of wideband radar, and low -angleradar propagation problems. His "Complex -Angle Monopulse"technique has gained wide acceptance in the radar communityas a useful approach to correcting low -angle multipath errors.Dr. Sherman is a member of Phi Beta Kappa, Sigma Xi, and Pi MuEpsilon.

Samuel M. ShermanFor contributions to radar systemsengineering and signal processing.

91

Engineering News and Highlights

American Ingenuity: 200 years of engineering

National Engineer's Week (Feb. 22 - 28, 1976)

The National Society of Professional Engineers is sponsoring National Engineer's Week(Feb 22 - 28, 1976) using a bicentennial theme: "American Ingenuity: 200 years ofengineering". The NSPE began sponsoring National Engineers Week in February, 1951.The purpose of the Week is to familiarize the public with the work of engineers and tohonor outstanding members of the profession. The week of George Washington's birthdayis traditionally observed because the first President was a trained land surveyor and adesigner of roads, fortification, and other structures. He also had the educationalbackground of a civil engineer in the 18th century.

Awards

RCA Limited

The first award winners of RCA Limited'snew Technical Excellence Program wereannounced recently in Montreal.

Two groups of engineers in Aerospaceand Government Systems were grantedteam awards. They are:

M.W. Altman, B. Dinsdale, G. Dziub, E.George and C.M. Kudsia "for thedevelopment of the GFEC MultiplexAssemblies for the SATCOM 24Channel Transponder, and

R. C. Baxter, P.Y. Chan, J.M. Keelty, P.Oldfield and N. Whittaker "for thedevelopment of the 2GHz RangingTransponder for CTS".

Two awards were also granted in Com-munications Systems. They are:

Dr. S.C. Acharyya, A. Grosswindhager,J. Labelle and L.A. Mora "for thedevelopment of Modular Design EarthStation Ground CommunicationsEquipment utilizing MIC Techniques".

J. Barkwith "for Project Engineering ofthe Haiti Small Earth Station."

Service Company

Adolph E. Hoffman-Heyden, Manager,Radar Systems Analysis, RCA ServiceCompany, Patrick Air Force Base, Florida,

was recently honored by the Florida In-stitute of Technology at a testimonialdinner.

The award was for his significant ac-complishments in the field of pulse radartechnology. He was presented a plaqueappointing him a lifetime faculty memberof the Institute.

Throughout his more than 20 -yearassociation with RCA Service Company,Mr. Hoffman-Heyden has been engaged inefforts to improve the trackingpulse radar instruments in support ofmissile and space vehicle testing.

Astro-Electronics Division

Dr. Alvin W. Sheffler, of Astro-ElectronicsDivision's Mechanical EngineeringDepartment received an Engineering Ex-cellence Award in recognition of his out-standing contributions to the structuraldesign analysis of the RCA Satcomspacecraft which AED has built for RCAGlobal Communications Inc. and RCAAlascom Inc.

Dr. Sheffler was cited by Dr. WarrenManger, Chief Engineer, for"demonstrating outstanding leadership inthe solution of some extremely difficulttechnical problems. As a consequence,his efforts have advanced RCA's technicalreputation in the area of spacecraftdesign."

Degrees granted

Pang -Ting Ho, Microwave TechnologyCenter, RCA Laboratories, Princeton,N.J., recently received the PhD in Elec-trical Engineering at Rutgers University.

Donald C. McCarthy, Process and AppliedMaterials Laboratory, RCA Laboratories,Princeton, N.J., received the BS inChemistry at Rutgers University.

Missile and SurfaceRadar Division

Chief Engineer, J.C. Volpe recently an-nounced the Technical Excellence Awardwinners for the second half of 1975:

R. Lieber and G.M. Sparks individualawards for systems effort on the AN/TPQ-27.

J.A. Lunsford and L.W. Martinsonindividual awards for the devleopment ofan advanced signal processing system forvideo bandwidth reduction.

J. Liston for technical expertise andcreativity in developing and applyingspecial analytical tools for solving radarsystem problems.

F.E. Ogle for engineering, software designand principal implementation of theAABNCP antenna system operationalcomputer.

E.W. Veitch for creativity and competencein developing the AEGIS Tactical Ex-ecutive System.

T.E. Anderson and W.I. Smith a jointaward for design and development of aklystron anode modulator for CCSD'sUHF tv transmitters.

D.C. Drumheller for technical ac-complishments and leadership as seniorsoftware engineer on the APPLICON pro-ject.

W.A. Harmening for concepts, develop-ment and design of a new approach toprecision steering mechanisms for lasersystem applications.

A.G. Chressanthis, C.J. Hughes and G.S.Oakes, Jr. individual awards for develop-ment and validation of a new approach tomoving target indication for the EDM-3CAN/SPY-1A radar.

92

A.G. Chressanthis receives1975 excellence awardA.G. Chressanthis received MSRD's an-nual technical excellence award for 1975for his consistently high level of technicalperformance on the AEGIS EDM-3C MTIexperiments conducted in USS NortonSound, and in particular for his perfor-mance leading up to and during the finalat -sea data gathering mission in earlyNovember 1975. His contributions to theevaluation of the clutter lock conceptualdesign by translating test results intorequired corrective equipment modi-fications were outstanding. His technicalefforts yielded a clutter lock loop conceptthat successfully demonstrated satisfac-tory performance in land, sea, andtransitional clutter under a multitude ofsea states and land topological con-figurations.

R. Lieber

J.A. Lunstord

J. Liston

G.M. Sparks

L.W. Martinson

F.E. Ogle

T.E. Anderson

1.0

W.I. Smith

W.A. Harmening

C.J. Hughes

11.41ti6"11*D.C. Drumheller

A G

fal

*:112.4Chressanthis

G.S. Oakes

On sitegraduate programat Mountaintop

In the fall of 1974, a new concept in thearea of Graduate Education was institutedat Mountaintop With the cooperation ofWilkes College Graduate School, Wilkes-Barre, Pennsylvania, a Graduate Programin Physics was established for technicalstaff, with the classes being presented atthe plant for the convenience of theemployees enrolling in the program. Nowin its second semester, this programappears to be a success.

Twice each week, Dr. John Orehotsky,Professor of Physics at Wilkes College,travels to the Mountaintop Plant to teachthis semester's course offering "SolidState Devices," to the eleven staffmembers currently enrolled. Lastsemester the course offering was "Topicsin Mathematical Physics."

The employees enrolled in this programhave the added flexibility of using the in -plant classes to supplement additionalcourses taken at Wilkes College, or theymay choose to obtain all the requirementsfor an advanced degree from the coursesoffered at the Plant each semester.

Zappasodi and Collinswin first place inpackagirg competition

G.G. Zappesodi and F.P. Collins ofPackaging Activity in Commercial Com-munications Systems Division, Camden,N.J., won first place in the "plastics(cushioning) packages" competition ofthe 1975 Packaging and Handling Com-petition, sponsored by the Society ofPackaging and Handling Engineers.

The entry was a molded polystyrene con-tainer for shipping 2 -way radio systems.The simple two-piece package designedby Messrs. Zappasodi and Collinsreplaces some 30 corrugated containers ofvarious sizes used in the past.

Zap Zappasodi (left) and Skip Collins holding their SPHEaward placgues. On the table is the molded polystyrenepackage that ,,,on the award. This ingenious designpermits packing any size radio and any combination ofaccessories in She same molded styrene pack.

G.G. (Zap) Zappasodi is Manager of Pack-ing Design, Broadcast Systems, CCSD,Camden, N.J. He received the BSIE in 1950from Penn State. He joined RCA in 1957 inthe Radio & 'Victrola" Engineering depart-ment. In 1958 he was appointed Manager,Packing Design for the newly formedIndustrial Electronic Products Division.Since that time he has been responsiblefor packaging and materials handling of allCommercial products in the BroadcastDivision. Mr. Zappasodi is active in theSociety et Packaging and HandlingEngineers and is presently NationalMaterials Handling Chairman for thatorganization. He is also active in ANSI andASME committee work on Standards inPackaging and Materials Handling.

F.P. (Skip) Collins is a Packing Designer inBroadcast Systems, CCSD, Camden, N.J.He began with RCA in 1951 upon gradua-tion from Camden County VocationalSchool wt -ere he specialized in the sheetmetal trade. He completed night coursesin drafting while employed in the factory.In 1954 he became a draftsman andprogressed to a junior packing designer in1958 and firially a senior packing designerin 1961. Mr. Collins was awarded a "Best ofShow" trcphy at the National Packagingcompetition of the Society of Packagingand Handling Engineers in 1965.

93

ilk

Jenny namedManager, TechnicalInformation Programs

Hans K. Jenny has been appointed Mgr.Technical Information Programs. His res-ponsibilities include those areas con-cerned with RCA's Technical Com-munication Programs, TechnicalPublications, Technical InformationSystems and the Minorities in EngineeringProgram (MEP). Technical InformationPrograms is part of Corporate Engineer-ing, Cherry Hill, N.J.

Hans K. Jenny, received the MSEE fromthe Swiss Federal Institute of Technologyin 1943 where he was subsequently activeas assistant professor doing research onlow -noise behavior of microwave tubes.He joined the RCA Tube Division in 1946as a microwave tube design engineer.Starting in 1950 he held various engineer-ing management positions concernedwith microwave tubes and solid statedevices. In 1966 he became OperationsManager, Microwave Solid StateOperations in Harrison N.J. From 1972 -1975 he was Senior Technical Advisor tothe Industrial Tube Division in Lancaster,Pa. and joined the Engineering Profes-sional Programs activity in Cherry Hill in1975. Mr. Jenny is a Fellow of the IEEE.

Professional activities

Dr. Gerard A. Alphonse, InformationSciences, RCA Laboratories, Princeton,N.J., has been named to the NationalScience Foundation Advisory Committeefor the Electrical Sciences and Analysissection of the NSF Engineering Division.The Committee advises the NSF on theimpact of its research support programs,reviews and evaluates proposals or pro-jects, and undertakes special studies.

Harold G. Behl, Missile and Surface RadarDivision, AEGIS SM-2 Test & Evaluation,West Coast Test Site, has been appointedto the National Technical Committee onSystem Effectiveness and Safety,American Institute of Aeronautics andAstronautics.

Phillips namedEditor, RCA Engineer

John C. Phillips has been appointedAdministrator, Technical CommunicationPrograms, and Editor, RCA Engineer.

John Phillips received the BA inMathematics from Rutgers in 1970 and hasdone graduate work in Communications.He has sixteen years of technical com-munication experience, fourteen withRCA. For the past seven years, he wasAssociate Editor, RCA Engineer. Beforethat, he was an Engineering Editor at theAstro-Electronics Division. Mr. Phillips is aMember of the IEEE and Past President ofthe Group on Professional Com-munications. He is a member of theRutgers Honor Society.

RCA Review, December, 1975Volume 36. Number 4

Contents

Spontaneous -Emission -Rate Alteration by Dielec-tric and Other Waveguiding Structures

J.P. Wittke

The Physics of Electrical Charging and Dischargingof Semiconductors

H. Kiess

Measurement of the Effective Photoelectror Emis-sion Energy

L A Ezard

MTF Measurements of an Image Tube Electron Lens

L.A. Ezard

Ultra -Thin RF Silicon Transistors with a Copper -Plated Heat Sink

H.S. Veloric, A. Presser, and F.J. Wozniak

Wideband Signal Recording on Film Using (AlGa)As CW Injection Lasers

J.E. Roddy

The Chirp Z-Transform-A CCD ImplementationG J Mayer

DOMESTIC FOREIGN1 -year 56.00 56 402 -year 10.50 11.303 -year 13.50 14.70

W.O. Hadlock retires

W.O. (Bill) Hadlock retired in February,culminating 41 years of service in theradio -electronics business, 29 of themwith RCA. Mr. Hadlock is probably bestknown for his work as Editor of the RCAEngineer. His service in that capacityspans 20 years, to the start of the journal in1955. During that time, he was alsoManager of Technical Communications,with responsibilities for the Corporatepapers and reports programs, and RCATrend, as well as the RCA Engineer. From1947 to 1955, he was with the BroadcastSystems Advertising Department wherehe worked on TV station plans,promotional programs for Broadcastproducts, and industrial advertising. Hewas also Editor of Broadcast News. Priorto 1947, he was with the General ElectricCompany. Mr. Hadlock received the BSEEfrom Clarkson Institute of Technology in1934. He is a Senior Member of IEEE and isactive in the Group on Professional Com-munications.

Globcom hosts DraftingCoordination CommitteeMeeting

RCA Globcom played host for the semi-annual meetingof the RCA Corporate Drafting Coordination Committeein Noventber. Those attending included (clockwise)John Feeley. Manager of Facilities and Maintenance.James Hepburn. Vice President of Operations. andFrank Micara. Supervisor of Drafting, all from RCAGlobcom. Also Al Skavicus. Engineering ServicesManager. Automated Systems Division: John W. Geier.Administrator. Corporate Standards. Engineering,Cherry Hill: William Wilson. Drafting Manager. Astro-Electronics Division. Jim Woodward, StandardsEngineer. RCA Limited; Clifford Hoffer. Drafting Leader.Commercial Communications Systems; and RaymondFitzmaurice, Standards Engineer. Corporate Standards.Engineering, Cherry Hill.

94

Staff announcements

RCA Communications

Howard R. Hawkins, President, RCA Com-munications, announced the election ofBen W. Agee as President and a Director ofRCA Alaska Communications, Inc.,Anchorage.

RCA Americom

The Board of Directors of the newly es-tablished subsidiary company, RCAAmerican Communications, Inc., (RCAAmericom), elected the following officers,to become effective when RCA Americombecomes operational: Philip Schneider,President, and Harold W. Rice Vice Presi-dent, Operations.

RCA GlobalCommunications, Inc.

Valentine Arbogast, Vice President,Leased Services has announced theorganization as follows: Thomas J. Buico,Manager, Leased Services Sales; RobertD. Mollehnauer, Manager, Major Systemsand Press Accounts; Lawrence P. Cam-borde, Manager, Leased ServicesAdministration; Paul B. Silverman,Manager, Data Services Development;Notis M. Kotsolios, Manager, LeasedServices Engineering; Notis M. Kotsolios,(Acting) Manager, Leased ServicesDesign and Implementation; WarrenSchulhafer, Manager, Technical &Administrative Services; Dan A. Patrisso,Manager, Technical Support; and JohnTerry, Manager, United Kingdom.

Gil A. Beltran, General Manager, Com-puter Switching Systems, has announcedthe organization as follows: Russel E.Blackwell as Manager, ComputerProgramming; Samuel A. Thompson,Manager, Computer Projects: Gil A.Beltran, (Acting) Manager, ComputerSystems Marketing; and Patrick C. Keilty,Manager, Mini -Plus Marketing.

Industrial Relations

George H. Fuchs, Executive Vice Presi-dent, Industrial Relations has announcedthat George A. Fadler will assumeresponsibility for the Materials Activity,and Donald 0. Corvey, Staff Vice Presi-dent, Materials, will report to Mr. Fadler.Mr. Fadler will continue to report to theExecutive Vice President, IndustrialRelations.

Corporate Engineering

Dr. William J. Underwood, Director,Engineering Professional Programs, hasappointed Hans K. Jenny as Manager,Technical Information Programs.

Hans K. Jenny has announced the staff ofTechnical Information Programs asfollows: Doris E. Hutchison,Administrator, Technical InformationSystems; John C. Phillips, Administrator,Technical Communication Programs andFrank J. Strobl, Administrator, TechnicalPublications.

Picture Tube Division

Joseph H. Colgrove, Division Vice Presi-dent and General Manager, Picture TubeDivision, has announced the organizationas follows: Wellesley J. Dodds, Director,Quality, Product Safety and Reliability: D.Joseph Donahue, Division Vice President,Engineering; William G. Hartzell, DivisionVice President, International; Donald L.Gilles, Division Vice President, IndustrialRelations; Robert B. Means, Division VicePresident, Domestic Marketing; Stanley N.Roseberry, Division Vice President,Finance; Stanley S. Stefanski, DivisionVice President, Manufacturing; andCharles W. Thierfelder, Division VicePresident, Technical Programs.

Stanley S. Stefanski, Division Vice Presi-dent, Manufacturing, has announced theorganization as follows: John M. Fanale,Director, Glass Operations; Richard H.Hynicka, Director, Mask and MountOperations and Lancaster Manufacturing;David 0. Price, Director, Materials;Douglas H. Spa-ks, Manager, OperationsPlanning and Industrial Engineering; andStanley S. Stefanski, Acting, Manufac-turing.

Charles W. Thierfelder, Division VicePresident, Technical Programs, has an-nounced the organization as follows:Richard C. Allen, Director, Materials andIndustrial Engineering; John W. Hensley,Manager, Plant Engineering; Arthur W.Hoeck, Director, Fianance; Lawrence A.Kameen, Director, Industrial Relations:Clifford H. Lane, On -Site Director; JosephJ. McHugh, Director, Tube Manufac-turing; Richard L. Spalding, Director,Manufacturing Engineering: John F.Stewart, Director, Equipment Engineer-ing; Frederick C. Weisbach, Director,Design and Construction; and R. HowardZachariason, Director, ComponentsManufacturing.

Clifford E. Shedd, Manager, EquipmentDevelopment, has announced theorganization as follows: Leonard M.Clutter, Manager, Resident EquipmentEngineering - Marion; Edward A.Gronka, Manager, Resident EquipmentEngineering-Scranton; William R. Miller,Manager, Electrical Design and Develop-ment; James W. Axmacher, EngineeringLeader, Test Equipment; Theodore C.Loser, Jr., Engineering Leader, ProcessControl; Keith D. Scearce, Administrator,Turnkey Project; N. Roger Thornton,Manager, Project Engineering; John W.Palmer, Administrator, Facilities andPlanning; C. Norman Yunginger,Administrator, Facilities and Planning;

Morris R. Weingarten, Manager,Mechanical Design and Development;Raymond A. Alleman, EngineeringLeader, Process; Frank D'Augustine,Engineering Leader, Cap Preparation;Richard A. Lambert, Engineering Leader,Materials and Process; and MarinusVanRenssen, Engineering Leader, Preci-sion Design and Gauging.

David D. VanOrmer, Manager, ProductDevelopment, has announced theorganizaion as follows: Albert M. Morrell,Manager, Tube Development; Ralph J.D'Amato, Engineering Leader, AdvancedDesign; Richard H. Godfrey, EngineeringLeader, Product Design;Gerald J. Mc-Cauley, Manager, Engineering Standards;Lewis I. Mengle, Engineering Leader,Laboratory Test Equipment; Albert M.Morrell, Acting, Prototype Equipment;Richard A. Nolan, Manager, Pilot Develop-ment Shop; John E. Eberle, Manager, PilotProduction Center; Glenn R. Fadner,Engineering Leader, Tube Fabrication;Francis P. Farkas, Manager, MachineShop; David D. VanOrmer, Acting, MountDevelopment; Paul R. Andorn,engineering Leader, Equipment; Martin K.Brown, Engineering LEader, Fabricaiton;and Frans VanHekken, EngineeringLeader, Design.

Leonard F. Hopen, Manager, Process andMaterials Development, has announcedthe organization as follows: Austin E.Hardy, Manager, Panel Process Develop-ment and Laboratories; David T.Copenhafer, Engineering Leader,Analytical Laboratory; Alan C. Porath,Engineering Leader, Screening Develop-ment; Louis C. R ledlinger, Jr., EngineeringLeader, Colorimetry Laboratory; Martin R.Royce, Engineering Leader, PhosphorDevelopment; Leonard F. Hopen, Acting,Masks, Materials and Finishing Develop-ment; Charles T. Lattimer, EngineeringLeader, Materials Development; CharlesT. Lattimer, Acting, Finishing Develop-ment; John J. Moscony, EngineeringLeader, Mask Development; Robert E.Salveter, Manager, Marion ResidentEngineering; *Robert E. Benway,Manager, Marion Resident ProductDevelopment; "Richard W. Osborne,Manager, Marion Resident Applications.Reliability and Safety; James A. Stankey,Manager, Materials and Process Develop-ment; Yonichi Uyeda, Manager, ScrantonResident Engineering; 'George T.Ashenbrenner, Engineering Leader,Scranton Resident Applications, Reliabil-ity and Safety; 'Louis J. DiMattio,Engineering Leader, Scranton ResidentProduct Development; William J. Kerzic,Manager, Special Services; and Robert J.Murray, Engineering Leader, Process andMaterials Development. ['Messrs. Benwayand DiMattio receive technical directionfrom David D. VanOrmer, Manager,Product Development. "Messrs.Ashenbrenner and Osborne receivetechnical direction from Robert L. Barbin,Manager, Applications Reliability, andSafety.]

Robert L. Barbin, Manager, Applications,

95

(Stall announcements - cont'd)

Reliability and Safety, has announced theorganization as follows: David C. Ballard,Engineering Leader, Reliability and Safe-ty; Robert L. Barbin, Acting, CustomerCoordination; William R. Menges,Engineering Leader, Specification, Test,and Evaluation; and William A. Sonntag,Engineering Leader, Receivers, Com-ponents and Services.

Consumer Electronics

Harry Anderson, Division Vice President,Manufacturing Operations, has an-nounced the appointment of Cortland P.Hill as Director, ManufacturingTechnology and Services.

John M. Wright, Chief Product DesignEngineer, Television Engineering has an-nounced the appointment of Perry C.Olsen as Manager, Vendor Contacts andMeasurements.

RCA Laboratories

William M. Webster, Vice President, RCALaboratories, has appointed Nathan L.Gordon Staff Vice President, SystemsResearch Laboratory.

Nathan L. Gordon, has announced theappointment of Jack Avins as StaffScientist in the Systems ResearchLaboratory.

Richard E. Guinn, Manager, Mechanicaland Instrumentation Technology, has ap-pointed Marvin A. Leedom, Manager,Mechanical and InstrumentationTechnology.

Paul Rappaport, Director, Process andApplied Materials Research Laboratoryhas announced the appointment of Jon K.Clemens as Head, Signal SystemsResearch, Process and Applied MaterialsResearch Laboratory.

Emil V. Fitzke, Manager, TechnologicalServices, has appointed Austin J. Kelley asManager, Instrument Center

Patent Operations

Glenn H. Bruestle, Director, ElectronicComponents and Materials, PatentOperations has announced the appoint-ment of Birgit E. Morris as ManagingPatent Attorny and Arthur E. Wilfond asPatent Counsel.

Promotions

Astro-Electronics Division

B. Cohen from Jr. Engr. to Mgr., SystemsEffectiveness Safety (H.L. Wuerffel)

J. Hoffman from Mgr., Graphic ArtsProduction to Mgr. Information Svcs. (J.Mackin)

Alaska Communications, Inc.

K.L. Anderson from Engrg. Assoc. to Sup.,Outlying Area (K.L. Laing, Mgr.,Anchorage Outlying Areas)

E.L. Braunston from Sr. Engrg. Assoc. toMgr., Trans. Engineering (E.C. Hall, Mgr.,Sys. Engrg., Anchorage)

J.R. Martin from Logistics SupportSpecialist to Mgr., Pipeline Adm. &Logistics (A.F Perkins, Mgr., PipelineEngrg., Anchorage)

P.A. Schilling from Sr. Engineer to Mgr.,Plant Ext. Engrg. (E.C. Hall, Mgr., Sys.Engrg., Anchorage)

R.B. Sharp from Sr Engrg. Assoc. toSupervisor, Circuit Provision (B.M.Snodgrass, Mgr., Plant Services,Anchorage)

F.E. Whiton from Sr. Engr. to Mgr.,Pipeline Radio Systems Engrg. (A.F.Perkins, Mgr., Pipeline Engrg.,Anchorage)

Global Communications, Inc.

L. Kamins from Sr. Mbr. to Engrg. Ldr.,Technical and Administrative Services (W.Schulhafer, Mgr. Leased Svcs)

N. Fatseas from Princ. Mbr. to Ldr., RadioEngrg. (E. Doherty, Mgr., Radio Engrg.)

Commercial CommunicationsSystems DivisionR. Marks from Mbr. Engrg. Staff to Ldr.,Engrg. Staff (B. Autry, Meadow Lands)

Solid State Division

E.J. Chabak from Mbr., Tech. Staff to Ldr.Tech. Staff (L. Zampetti, Mountaintop)

RCA Service Company

M.E. Mattheiesen from Field SupportEngrg. to Ldr., Instrumentation Systems(G. Berube, Range MeasurementSystems, Antigua)

T.V. Tolman from Field Support Engrg. toLdr., Instrumentation Systems (G. Berube,Range Measurement Systems, GrandBahama Islands).

Missile & SurfaceRadar DivisionW. Atwood from Ldr., Engrg. Sys. Proj. toMgr., Projects (R.W. Howery, WCS/FCSProj. & ORTS C/P, Moorestown)

A. Barresi from Time Study Engr. to Ldr.,Mfg. Methods (J. Ervin, Mfg. Tech.)

D. Fuerele from Sr. Mbr. Engrg. Staff toLdr., Engrg. Sys Proj. (J. Haake, WeaponSys. Software Develop.)

S. Paskow from Mbr. Prog. Mgmt. to Ldr.,Engrg. Sys. Proj. (W.E. Scull, Test Ping.)

Madenfordis TPA forPicture TubeDivision

Dr. D.J. Donahue, Division V.P., Engineer-ing, Picture Tube Division, has appointedEdward K. Madenford TechnicalPublications Administrator for PictureTube Division. In this capacity, Mr.Madenford is responsible for the reviewand approval of technical papers; forcoordinating the technical reportingprogram; and for promoting the prepara-tion of technical papers and presen-tations.

Mr. Madenford is presently Administrator,Picture Tube Engineering Administrationat Lancaster. He received the BS inBusiness Administration from Lehigh Uni-versity, Bethlehem, Pa.He joined RCA Lancaster in 1954 andserved as Cost Estimator, (1954-56);Administrator, Standard Cost Data-Super Power Tubes (1956-64); andAdministrator, Picture Tube EngineeringAdministration (1964 to present). Mr.Madenford has been an RCA EngineerEditorial Representative since 1964.

NubaninamedScrantonEd Rep

John S. Ignar, Plant Manager, ScrantonPlant, has appointed Jack I. Nubani asEditorial Representative for the ScrantonPicture Tube Plant. Editorial Represen-tatives are responsible for planning andprocessing articles for the RCA Engineer.

After receiving the BS in Engineering fromMichigan State University in 1952, Mr.Nubani joined RCA in January of 1953 as atrainee and became area engineer thesame year at the Marion plant. During hisyears at Marion, he served in variousengineering capacities. In 1966, he wastransferred to the Scranton plant as anEngineering Leader. He assisted in thestart ups of the Mexican, Thorn,Videocolor, and Circleville Glass plants. In1971, Mr. Nubani was promoted toProcess and Production EngineeringManager, which position he presentlyholds.

96

Editorial RepresentativesThe Editorial Representative in your group is the one you should contact inscheduling technical papers and announcements of your professional activities.

Government and Commercial System:,Astro-Electronics Division I.M. SEIDEMAN' Engineering, Princeton, N.J.

Automated Systems Division K.E. PALM Engineering, Burlington, Mass.A.J. SKAVICL- Engineering , Burlington, Mass.L.B. SMITH Engineering, Burlington, Mass.

Commercial CommunicationsSystems DivisionBroadcast Systems W.S. SEPICH Broadcast Systems Engineering Camden, N.J.

R.E. WIN -I' Broadcast Systems Antenna Equip. Eng., Gibbsboro, N.J.A.C. BILL iE Broadcast Engineering, Meadow Lands, Pa.

Mobile Communications Systems F.A. BARTON Advanced Development, Meadow Lands. Pa.

Avionics Systems C.S. METCHET I Engineering, Van Nuys. Calif.J. McDONOUGI-, Equipment Engineering, Van Nuys, Calif.

Government Communications A. LIGUORI' Engineering, Camden, N.J.Systems Division H.R. KETCHAM Engineering, Camden, N.J.

Government Engineering M.G. PIETZ' Advanced Technology Laboratories. Camden. N.J.

Missile and Surface Radar Division D.R. HIGGS' Engineering, Moorestown, N.J.

Ret,-Corporate Engineering H.K. JENNY Technical Information Programs, Cherry Hill, N.J.

Laboratories C.W. SALL' Research, Princeton, N.J.

Solid State Division

Consumer Electronics

RCA Service Company

Distributor andSpecial Products Division

Picture Tube Division

J.E. SCHOEN Engineering Publications, Somerville, N.J.H.R. RONAN Power Devices, Mountaintop, Pa.S. SILV=_RST Power Transistors, Somerville, N.J.A.J. BIANCU Integrated Circuits and Special Devices, Somerville. N.J.J.D. YOUNG IC Manufacturing, Findlay, OhioR.W. ENGSTRON Electro-Optics and Devices, Lancaster, Pa.

C.W. HOYT' Engineering, Indianapolis, Ind.R.J. BUTH Engineering, Indianapolis, Ind.P.E. CPOOK". Television Engineering, Indianapolis. Ind.F.R. HOl.' Advanced Products, Indianapolis, Ind

J.E. STEOGER'. Consumer Services Engineering, Cherry Hill, N.J.R. MacWILLIAMS Marketing Services, Government Services Division. Cherry Hill. N.J.R.M. D DMBROSKY Technical Support, Cherry Hill, N.J.

C.C. READ - Product Development Engineer'ng, Deptford, N.J.J.N. KOFF Receiving Tube Operations, Harrison, N.J.

E. K. WADENFORD Engineering, Lancaster, Pa.J.H. LIDSCOMBE Television Picture Tube Operations, Marion, Ind.N. MEENe Glass Operations, Circleville, Oh.J.I. NUB, Television Picture Tube Operations, Scranton, Pa.

RCA Global Communications, Inc. W.S. LEI'. RCA Global Communications, Inc.,New York, N.Y.P. WEST RCA Alaska Communications, Inc., Anchorage, Alaska

NBC, Inc. W.A. HOWARD' Staff Eng., Technical Development, New York, N.Y.

RCA Records J.F. WELLS Record Eng., Indianapolis, Ind.

RCA Ltd W.A. CHISHOLM Research & Eng. Montreal, Canada

Patent Operations J.S. TRIPOLI Patent Plans and Services, Princeton, N.J.

Electronic Industrial Engineering J. OVNICK* Engineering, N. Hollywood, Calif.

'Technical Publications Administrators (asterisked above) areresponsible for review and approval of papers and presentations

RCA EngineerA TECHNICAL JOURNAL PUBLISHED BY CORPORATE TECHNICAL COMMUNICATIONS"BY AND FOR THE RCA ENGINEER"

Pr,nted U S A Form No. RE -21-5