mcsa Engineer Vol 21 No 3Oct l Nov

1975

The next step inmanufacturing technology

Inflation can be defined as an uncontrolled imbalance between changingproductivity on the one hand and spiraling labor and material costs on the other.In our efforts to control inflation, we strive simultaneously to increaseproductivity and limit costs. In a sense, we live in a dichotomous environment. inthat while limiting costs, we improve our standard of living through increasedwages. That improved standard, of course, should derive from increased laboroutputs. Solutions to such a complex problem exist, and our managerialresponsibilities dictate that we constantly seek those solutions.

One such solution has been the development of custom equipment forspecialized production applications. The demand for custom machinery, or atleast customized adaptations of standard machines, led to another breed ofengineers, the equipment technologists Some of the equipment technologists'work is reflected in several papers in this issue of the RCA Engineer, which isthemed around manufacturing productivity through automation. I'm pleasedthat we recognize these technologists' contributions in this issue, for we oftenshroud their work to preserve the competitive edge we realize from tneirinnovativeness.

But we have another tool at hand, a tool that both supplements and expandscustom -equipment approaches. The microprocessor is that tool, and its impacthas two bases. First, it is a large-scale integration (LSI) device-and LSI meansthat more functions are available per unit of labor, whether that unit involvespeople or machines. Second, standard commercially available LSI designs suchas the microprocessor have a multitude of applications, ideally suited to processcontrol. The manpower efficiency required to develop one complex device istherefore applied to the solution of many problems and, in effect, improves theproductivity of those highly skilled people who design and develop thesedevices.

Thus, we find an editorial subtlety in this issue by way of the inclusion of anarticle on COSMAC. Implications for the equipment technologist are immense,as indeed they are for all equipment and system design engineers.

Aadottee4.0a..:2,4._B.V. Vonderschmitt,Division Vice Presidentand General Manager,Solid State Division,Somerville, N.J.

W.O. Hadlock

J.C. PhillipsJ.P. Dunn

P.A. Gibson

Juli CliftonDottie Snyder

C.A. Meyer

C.W. Sall

F.J. Strobl

Dr. J.J. Brandinger

R.M. Cohen

F.L. Flemming

C.C. Foster

C.H. Lane

H. Rosenthal

P. Schneider

RCA Engineer Staff

Editor

Associate EditorArt EditorCompositionEditorial SecretaryEditorial Secretary

Consulting Editors

Technical Publications Adm.,Picture Tube Division

Technical Publications Adm.,Laboratories

Technical Publications Adm.,Technical Communications

Editorial Advisory Board

Div. VP, TelevisionEngineering, Consumer ElectronicsDir., Quality and ReliabilityAssurance, Solid State Div.VP, Engineering, NBCTelevision Network Div.Mgr., Scientific PublicationsRCA Laboratories

Div. VP, Technical PlanningPicture Tube DivisionStaff VP, EngineeringResearch and EngineeringExec. VP, SATCOM Systems

Dr. W.J. Underwood Director, EngineeringProfessional Programs

Our Cover

is a before -and -after look at part of the solid-statediffusion area at the Mountaintop plant. To accomplishthis striking transition without curtailing production, adedicated crew virtually had to perform "open heartsurgery on a living factory" (p.39). In the "before- photo.Tom Ruskey (left) and Bob Guerin discuss the planneddiffusion area rearrangement task (DART). The "after"photo shows the culmination of this planning and effort.In the photo. Carol Ann Piestrak and Tom Lyden areworking at the left side of the aisle. Norah Smith isattending the furnaces at the far end, Helene Dobranski.Tom Ruskey. and Mildred Cerrito are at the right. MMes.Cerrito, Dobranski. Piestrak. and Smith are productionworkers: Messrs. Guerin. Lyden and Ruskey aremembers of the DART team.

Photo credit: John Semonish, Solid State Division, Clark,NJ

Vol. 21INo. 3Octl Nov1975

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

RCA Engineer articles are indexedannually in the April -May issue and inthe Index to RCA Technical Papers.

Contents

IIICBLU Engineer To disseminate to RCA engineers technical informa-tion of professional value To publish in an appropri-ate manner important technical developments at RCA.and the role of the engineer To serve as a medium ofinterchange of technical information between variousgroups at RCA To create a commucity of engineer-ing interest within the company by stressing the in-terrmated nature of all technical contributions To

help publicize engineenng achievements in a mannerthat will promote the interests and reputation of RCA inthe engineering field 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 - Automated process control

Editorial input an invitation 2

Automatedprocesscontroland manufacturingsupport

Productivity and manufacturing at RCA R.T. Vaughan 3

Programmable controllers, minicomputers, andmicrocomputers in manufacturing J.L. Miller 7

The computer in manufacturing operations R.M. Carrell 11

Merit of direct digital control M.H. Lazar 16

Trend toward standardization and automation M.J. Gallagher 20

Twisted -pair wire dispenser-automaticand programmable L.D. Ciarrocchil J.J. Colgan' H.F. Schellack 26

18,000 -point automatic test interface L.D. Ciarrocchil H.F. Schellack 28

Semi -automatic wiring equipment R.Piscatellil S. Patrakis 32

A concept of factory automation R Walter 36

Cover feature Open heart surgeryon a living factory

W.R. Guerin' M.J. Martin' T.J. RuskeyT.B. Lydenl G.D. McLaughlin' S.J. Blaskowski 39

Interview The business end of engineering J. Volpe 48

Conservation Chemical conservation in SSD R. Epifanol J. Zuberl Dr. J. Amick 55

Microprocessors Leased channel control A. Longo' Dr. P.M. Russo' M.D. Lippman 60

COSMAC N.P. Swalesl J. A. Weisbecker 64

Generalinterestpapers

Comparison of command and control systems J.R. LoPresto 70

SARA-a research planning tool C. Havemeyerl R.R. Lorentzen 74

Mechanical design of mobile data processingsystems for worldwide use J.S. Furnstahll J. Herzlinger 78

Refractive index determination M.M. Hopkins' Dr. A. Miller 83

Engineeringandresearch notes

Separating silicon chips E.P. Herrmann 86

Thyristor trigger circuits M.A. Kalfus 86

High -dynamic -range, low -noise preamplifier L.A. Kaplan 87

Automatic pedestal compensation for tv data processing H. Logemann 87

Departments Pen and Podium 89

Patents Granted 90

Dates and Deadlines 91

News and Highlights 92

Copyright 1975 RCA CorporationAll rights reserved

1

editorialinput an invitation

Throughout RCA's 56 -year history, efficient cost-effective productionhas been an important ingredient to success. But never before hasproductivity been as critical as it is today.

We are up against quality competitors in a marketplace characterizedby decreased consumer spending and demands for lower costs andhigher reliability. At the same time, many product lines have maturedto the point where no one company can sustain a technological edgefor any length of time. Obviously then, RCA must design andmanufacture better products at lower cost.

Automated factory operations represent one way to meet thischallenge. Some automated systems and techniques are described inthis special issue of the RCA Engineer. Other ways will be examined infuture issues, when the RCA Engineer addresses product design(next), manufacturing engineering, industrial engineering, residentengineering, and many other areas that represent the solid applica-tion of engineering know-how in the factory.

In this way, we hope to improve the professional dialog among factoryoperations as well as between the manufacturing and the design anddevelopment groups. To make such dialog as productive as possible,we need more input from the technical staff in all our manufacturingoperations.

We invite your comments, your ideas, and your professional papers.Contact us directly, or through your Editorial Representative.

-J.0 P.

Address correspondence to:

Editor, RCA EngineerBldg. 204-2Cherry Hill, New Jersey 08101

or call (609) 779-4254.

Editorial Representatives are listed on the inside back cover of eachissue.

Right: Automatic component Insertion equipment In action atJuarez. Consumer Electronic's mechanization program encom-passes domestic and offshore operations with the objective ofimproved quality and cost performance.

Below: Process control monitoring at Coronet in carpet manufac-turing. Sensing equipment measures bonding material thicknessto assure quality and minimize power consumption.

Far right: A recent development indiffusion furnace technology inoperation in the Solid State Divi-sion. This equipment incorporatesmicroprocessor control andprovides a major improvement inmanufacturing efficiency. J Kau.Design Engineer. is loading a

process card into the reader. Thisview shows one half of an 8 -tubemodule.

Right: Automatic record produc-tion in Indianapolis. Thismechanized press system is beingevaluated along with other equip-ment approaches as the RecordDivision improves Its productivity.

Below: Automatic wire wrap hasprovided a major productivity Im-provement in GliCS operations.The unit shown is presently Inoperation in the Camden plant.

2

-x=imulimmiNIMMINEMPOIEMPlellik%NRigt P. These two ph Nos show integrated cirouit assemblymechanization now in operation in the So id Stale Dimon.

Ilk

Right: Chickens are processed through clever mechanization athigh rates in Banquet facilities. This simple unit removes heads.

B flow. Densely-pacxed conveyors carry chickens at rates ex-ceeding 5.000 an hour through a carefully monitored series ofautomatic and manual steps.

Productivityandmanufacturingat RCAR.T. Vauc han

Manufacturing is. and will continue to b. amajor .:ontrit utor to product cost for themajority of FCA s products. Productivityimprovement are therefore essential tobusiness su :cess. This paper reviewssome recent )roductivity improvements inseveral ope.ating units and providesgeneral guicelines for across-the-boardimproemen s.

THE WORD productivitt has manymeanings that generally match the in-terests or specialty of the particular user.Although the term seems only partiallydefined in trie dictionary as the procuc-tion of goods or services having exchangevalue its meaning has become acceptedgenerally as the ratio of output comparedwith input. For example, the develop-ment and use of automatic screwmachines as compared with the earliertechrique of an individual operatorpainitakingly turning out pieces one byone on a simple lathe clearly improvedproductivity. There seems to be littledisagreement on the general meanir.g ofthe term, but there are definite differencesof opinion as to the specific measurementto use ani the various factors to beinclided is the definition, such as infla-tion and o:hers.

3

On a large scale, economists think interms of gross national product as a total,including all industries and manyactivities. At the other end of the range, asupervisor is concerned about the outputof his employee group and the absenteeproblem he has to contend with. At yetanother level, a production superinten-dent looks at his department, includingthe personnel and facilities, and considersindividual efforts and output as well asmachine performance and maintenanceproblems with their total impact onproduction. He compares his currentoutput with last year or last month

A manager responsible for a particularproduct must continually evaluate hisoutput with respect to the people andequipment resources utilized. He is in-terested in his own trends and progressand also his performance within theindustry. His product must be designedfor competitive performance andproducibility. He must equip his opera-tion with efficient facilities and controlsystems; maintain practical labor stan-

Reprint RE -21-3-1Final manuscript received July 11, 1975

Robert T. Vaughn, Staff Vice President, Manufacturing.Cherry Hill, N.J., joined RCA in 1973 from the post ofDirector of Manufacturing for the International Divisionof Philco-Ford Corporation. Mr. Vaughan received theBSME from the University of Pennsylvania in 1949 andreceived the MS in 1952. After several years of electroniccomponent and process development experience, Mr.Vaughan joined Philco in 1952 as a Senior Engineer andwas appointed Chief Engineer. Equipment Operations in1959 and General Manager in 1962. Two years later hewas named Manager of Manufacturing for TelevisiorPictures Tubes, and in 1969 was appointed ManagerTube Operations. He is a member of Tau Beta Pi and ha:been awarded eight U.S. Patents.

dards; and keep scrap, inventory, andvariances to a minimum. In addition, hemust maintain high employee morale forits important effect on productivity.

Thus, the application of the termproductivity to our business activities as apractical measurement is desirable andimportant as one of a number ofcalibrating means for operating manage-ment.

In practice, productivity is used tocompare product value produced withthe labor, or possibly total personnel,involved in the operations. This is a usefulratio, but if too vigorously used as ameasurement of managementeffectiveness, could theoretically leadlong-term manufacturing planning in thewrong direction. For example, in the caseof television, if productivity is defined asproduct value compared with labor cost,the ratio would be improved by reducingcomponent integration and its labor costcontribution by increasing purchasedcontent. Conversely, the manufacturingintegration of additional components in-volves adding labor and thereforeadversely affects a productivity indexthat simply compares product value withhead count. In spite of thesetechnicalities, one practical and ob-tainable index is the ratio of manufac-tured product value in constant dollarsdivided by the total personnel involved inthe operation or division to which it isapplied.

Productivity in televisionmanufacturing

Any review of productivity in televisionmanufacturing would be incompletewithout considering the effects of thecolor tube, solid state, and other majorcomponents at the same time - althoughsome of these products will also bediscussed individually. While the chassisassembly process has been undergoingsimplification and mechanization viaprinted circuit boards and automaticinsertion equipment, the color tube hassimultaneously become more difficult toproduce as brightness performance andother design features have reducedprocess yields and increased labor. Thereal task in television manufacturing hasbeen to find ways of improving efficiencyto compensate for the additionalfunctional complexity that has developedas a result of competitive pressure. Thedevelopment of integrated circuits has

already simplified the television assemblytask by incorporating functions andreducing the total component count.Additional assembly benefits can beanticipated as further IC developmentsare utilized.

Since the emergence of television in the40's, industry manufacturing and productdesign trends have followed generallyparallel paths among the many manufac-turers. Some have naturally moved fasterthan others into new manufacturing tech-niques and have chosen to use low-costlabor resources in various parts of theworld. It should be recognized that thereis probably no single best way and theoutlook five or ten years ago was sub-stantially different than that of thecurrent period.

Today, in television manufacturing as inmost other areas, the name of the game isefficiency or productivity. Manufac-turing must do a thorough job of plan-ning, controlling, and economizing in theuse of all its resources - includingfacilities, personnel, materials andutilities. It must be producing a productdesigned not only for excellent perfor-mance but also for manufacturing costeffectiveness and reliability.

Mechanization

To review some specifics, televisionassembly technology adopted theprinted -circuit board quite a few yearsago with a surge toward mechanization ofcomponent insertion. This included thedevelopment of the long, many -stationedautomatic insertion machines of the Unit-ed Shoe Machinery Company. Theseappeared to be the answer to one of themajor needs - getting the vast number ofdiscrete components into the circuitreliably and inexpensively. Thesemachines were reasonably effective in thisapplication but required relatively longmodel runs for effective utilization. Thelarge variety of chassis and relativelyfrequent new developments the industryexperienced since their acquisition putthem in the role of handling the "commondenominator" assemblies anddemonstrated the need for somethingmore versatile and flexible. This gap hasbeen filled by a related development,covered thoroughly by previous issues ofthis publication: the variable -center -distance, pre -sequenced component in-serter with its accompanying sequencingequipment.

4

There has been a tendency to consider thetelevision assembly mechanization taskas very well along via the automaticinserters with their capability of handlingaxial -leaded components. However,quite a few other components, including alarge number of disc capacitors, haveonly recently begun to become insertablefrom a practical viewpoint. Interestingly,this component had already undergonean excellent job of productivity develop-ment, but its manufacturing process anddesign did not fit readily into the dual -taped configuration of the newer inser-tion machinery. However, this problem isbeing solved and the industry can counton mechanized insertion of a largeproportion of the individual componentsinvolved.

All this progress might lead one to feelthat the television assembly mechaniza-tion job must be relatively complete andproductivity very high as a consequenceof the negligible remaining labor. Anexamination of the labor content in atypical television receiver shows that thisis far from the case. Component assemblylabor, even without assuming the use ofmechanization, is only one of a number oflabor -intense areas. Some of the othersare instrument assembly, chassis

assembly, and testing.

Testing - more automatic

Testing is recognized in the televisionindustry as important both from a qualityand efficiency standpoint. Obviously it isimportant initially to assure that goodcomponents are being fed to the manufac-turing system. It then becomes essentialthat the various assemblies be controlledthoroughly and reproducibly to minimizerepair, which can become a large laborelement in itself, and improve productuniformity and throughput. Test equip-ment developments have reducedoperator dependence and simplified thetraining task as well as improved quality.The trend in test equipment is to have thetest system make the determination on ago/ no-go basis or point to the problemarea. Tests and adjustments have becomemore automatic.

Component manufacturing improvements- yokes

In .addition to the very prominent

developments in assembly mechanizationand test methods, substantial progresshas also been made over the years inimproving the manufacturing productivi-ty of other major components. One of themore interesting developments has beenthe toroidal yoke winding system and itsbonding along with other neck com-ponents to the color tube prior to instru-ment assembly, significantly reducing thefinal instrument assembly task.

Solid state devices

The solid state business has made almostunbelievable strides in productivity if themeasurement were to consider numbersof individual components combined onan IC chip. This progress made in ICmanufacturing technology has alsodramatically impacted the televisionbusiness as well as many other industries.

The trend to LSI

Progress from point contact and alloyedjunction technology to high -densityphotolithography, diffusion, epitaxialand ion implantation techniques has beenvery fast. Many of the important processsteps are carried out on a large number ofchips simultaneously and some of theingredients for high productivity are es-sentially built into the process. But thename of the game in this business, as inthe past, is yield, which has become thevital ingredient of LSI productivity.

A number of recent and currentdevelopments that will continue to im-prove manufacturing productivity in-clude the transition to larger wafers andmechanized bonding approaches. Otherimportant factors long recognized in thesolid state business have become evenmore important with the growth of LSIdevelopments. The larger the chip, thegreater the need for very clean, defect -freeprocessing. Defects can be introducedthroughout the process and are not justfrom airborne or solution contamination.

The very high resolution necessary duringthe photographic steps requires thatphotomasks be in close contact with thewafer during exposure. This process steptends to result in damage to the mask.Developments in alignment and exposuresystems have reduced this problem byallowing a gap to exist between the mask

and wafer with a reduction in abrasiondamage.

Microprocessors in process control

Another important trend has been theimprovement in process control madefeasible as a result of the availability ofmini and microprocessors. Process infor-mation feedback for control purposes willcontinue to have a major impact onmanufacturing efficiency and productivi-ty in the solid state business.

Picture tubes

As mentioned above, picture tubes have,by a number of performance im-provements over the years, become moredifficult to produce - although con-currently those in the business have donean excellent job in improving yields.Picture tube manufacturing economicsand productivity are highly dependent onattaining high yield and low materialscrap, starting at the glass plant.

A number of the currently used methodsand equipment approaches are naturalcarryovers from monochrome days.Relatively high -volume, efficientmechanization had been developed formount sealing and a few other of theprocess steps, but color manufacturingadded a substantial screening task thathas been further complicated by the"matrix" development. Other productmodifications such as 110° deflection andthe slotted mask have also madeproductivity improvement even more of achallenge.

Yield in this process -oriented business isextremely sensitive to control at everypoint in the process, being similar to solidstate manufacturing in this respect.Optical and photographic procedures re-quiring relatively precise control are in-volved, and improved yield and qualitywill require additional use of processcontrol instrumentation and computercapabilities.

Government and CommercialSystems

Manufacturing developments in theGovernment and Commercial Systemsdivisions are exploiting automatic testing

5

techniques to reduce the otherwise un-manageable testing task associated with asubstantial volume of complex boardsand assembiies often containing manyLS1's.

Automatic component insertion is in-creasing as a proportion of total com-ponents involved and automatic wirewrap machinery is used extensively. As aspecific example, the benefits that canresult from Manufacturing Engineeringand Design Engineering teamwork in"producibility engineering" isdemonstrated vividly by the TR-600Broadcast Video Tape Recorder. Thisproduct, which evolved from the earlierTR-70, was designed with maximumutilization of automatic test and assemblytechniques in mind.

Boards were designed to a standardphysical configuration for most efficientcomputer -controlled testing. They werealso designed for automatic componentinsertion. A major portion of the wiringhas been incorporated in a back paneldesigned for automatic wire wrapoperations. The wire -wrapped back planeand connector terminated cable harnessare also automatically circuit checkedprior to assembly.

The combined effect of thesemanufacturing -oriented developmentsyields a similar product in terms ofperformance but requires only 30% of theprevious assembly labor to put it together

a prime example of productivity im-provement.

The record industry

The record business involves a variety ofmanufacturing technologies. As viewedfrom the outside, record pressing appearsto be the primary production activity.Actually, pressing is a major activity ofRecord Division manufacturing, butmany other specialized methods andtechniques are involved. The RecordDivision is improving its efficiency andproductivity through the use of moderninjection -compression presses and is alsotesting other approaches to automaticpressing.

A good deal of fine work had been doneyears ago in developing a practical com-pression press that had remained virtual-

ly a standard for the industry for quitesome time. The industry is now in theprocess of gradually replacing this equip-ment with improved presses of semi-automatic and automatic variety. Alsoinvolved in the operations are veryprecisely controlled plating operations,compound mixing, and printing. Theneed for improved process control andmonitoring in all operations is clearlyrecognized and undergoing rapid im-provement.

Substantial development work has

already been accomplished in tapeoperations where tape duplicating speedsare being increased and assembly andpacking operations mechanized.

Carpet manufacturing

The Coronet carpet business was es-tablished as a direct result of a basicproductivity improvement in carpetmanufacturing known as "tufting." Thistechnique allowed an order of magnitudeincrease in the speed of assembly of yarninto the backing.

Since it began operations in the late 50's,Coronet has been a leader in productivityimprovement. Recently, innovations inprocess machinery operation have pro-vided a substantial increase in throughputcapability.

Food processing

The Banquet facilities include many high-ly mechanized operations, typicallyoperating at cycle rates well in excess of4,000 an hour.

Chickens are cut up and eviscerated atimpressive speeds while traveling on con-tinuous conveyors. Chicken parts areprepared for batter coating and fried via acombination of operator -controlled andautomatic process steps.

Food processing machinery has beendeveloped and modified to extendoperating speed capabilities substantiallyfor increased productivity. Currentdevelopments are aimed at approaches toreduce equipment downtime, increaseoperating speed and improve quality.

Banquet's operations involve a

stimulating combination of automaticprocessing and handling machinery run-

ning smoothly in conjunction with skilledoperators.

Conclusions

The comments and background aboveare basically related to the operationsinvolved. However, there are somegeneral productivity improvementpractices that apply across the board:

Better planning

More effective management

Improved processes and procedures

Better communications

Effective manpower and personnel policies

At first glance, these factors which havebeen commonly cited by U.S. manufac-turing executives' appear to be in thecategory of truisms or "motherhood"statements. But, as you think about them,you realize there are no substitutes. Thereare specific solutions to many specialmanufacturing problems such as newmachinery, test equipment and processdevelopments, but for long-rangeproductivity performance the depth andresourcefulness of planning for manufac-turing and manufacturing -orientedengineering is critical.

Better management is an obvious require-ment. But how is it achieved? Manufac-turing management is strengthened bytraining, selection, and experience in avariety of assignments. Mobility must beimproved for talented individuals. Divi-sion management is recognizing the im-portance of training and is establishingtraining programs for Manufacturingpersonnel.

Communications is clearly a two-wayneed. Employees, both salaried and hour-ly, need to know what is going on andhow well they are doing. In addition, ifcommunications are good enough,employees can be an excellent source ofideas for methods to improve productivi-ty and reduce waste. Some Divisions haverecognized the value of this source ofideas and installed employee -basedprograms that have been major con-tributors to profit improvement. Thisresource has been a real asset.

Reference

I. Kat/ell. N.E.: Productivity: The Measure and the Myth.American Management Association Survey Report.

6

Programmable controllers,minicomputers, andmicrocomputersin manufacturingJ. L. Miller

The need for new techniques in manufacturing has become increasingly critical during thepast several years. Factors that contribute to this need are 1) increasing complexity ofproducts. 2) severe pressures of domestic and foreign competition. 3) increasingconsumer demand for higher quality. 4) influence of governmental and other agencies forgreater safety and reliability, and 5) inability to increase prices to keep pace withincreasing production costs. The ele;tronic and other businesses of RCA demandmanufacturing development to keep pace with product development so that RCA canretain its competitive position.

James L. Miller, Director, Manufacturing Systems andTechnology, Corporate Staff Manufacturing, Cherry Hill.N.J. graduated from Iowa State University in 1918 andjoined RCA as an engineer in the Component PartsDepartment of the Tube Division. During that time hedesigned printed circuit i.f. amplifiers and rt tuners. In1953 he transferred to the Home Instruments Division.Color TV, designing i.f. amplifiers, color demodulatorsand monochrome video circuits. In 1960 Mr. Millermoved to the Computer Division where he worked onhigh-speed tunnel -diode circuits and later became amanager of peripheral equipment design responsible forcard readers and optical character readers. In 1966 hejoined the International Licensing organization as theManaging Director of RCA Engineering Laboratories inTokyo, Japan, and, in 1969. joined Corporate StaffManufacturing as a manager and later as the Director ofManufacturing Systems and Technology.

Reprint RE -21-3-2Final manuscript received June 23. 1975.

COMPUTERS have been available formany years to provide the speed andcapabilities which made many complexand, at times, impossible tasks practicalfor the first time. Most of these earlysystems were quite large and expensiveand were used in such applications asfinancial control, scientific and engineer-ing calculations, communications, etc.Because of their complexities and cost,only limited use was seen in manufac-turing. Introduction of programmablecontrollers, minicomputers, andmicroprocessors during the past fewyears is making it possible for the firsttime to develop practical manufacturingsystems for automatic test, data collec-tion, and process control in manufac-turing.

Programmable controllers

Although not as flexible and powerful asa minicomputer or microprocessor,programmable controllers can replacecomplex relay banks with logic switchingthat conforms to Boolean expressions.Programmable controllers offer severaladvantages. They are more flexible andeasily changed than static logic or relays.As the complexity of control increases,they offer a cost advantage at about 40 to100 relays over standard switching logic.Reliability and life of solid-stateprogrammable controllers generally farexceed the expectancy of relay logic. Theability to easily change logic statements

7

10000

9000

8000

7000

6000

5000

4000

3000

2000

1000

0 ill I

68 69 70 71 72 73 74 75 76 77 78 79 80

YEAR

Fig. 1 - Minicomputer prices.

gives the controller greater flexibilitywhen compared with the need to modifywiring in a relay panel. This is especiallyuseful when designing and testing a newsystem. Since most suppliers providestraightforward methods to program thecontroller, it can be easily used bypersonnel familiar with relay switchinglogic without the need for computerprogramming skills. Early users were inthe automotive and machine tool in-dustries but installations are now beingimplemented in process control ofchemical, petro-chemical, food process-ing, paper and pulp industries. In someapplications, programmable controllersare used as primary control devices withminicomputer backup for the purpose ofloading programs, monitoring processconditions and providing more completecontrol.

Minicomputers

In addition to programmable controllers,minicomputers are being used more ex-tensively in manufacturing applications.One contributing factor has been thedrastic reduction in minicomputer pricesduring the past five years as seen in Fig. I.In addition, both hardware and softwarecapabilities of minicomputers have beenexpanding during this same period.Many of the functions previouslyprovided only in large computer systemsare now available in these devices. Largecore memory at decreasing prices alongwith such features as hardware mul-tiply/divide, memory protect, automaticrestart, etc., have become common. Insoftware, such things as real timeoperating systems, file management, and

process monitor -direct digital controlpackages are available from one or moreof the minicomputer suppliers.' In addi-tion, high level languages such as For-tran, COBOL, BASIC, etc., are availableon most equipment.

Multiplexing capability permits sensingprocess conditions, inputting of testresults, or entering data manually frommany points around a manufacturingfloor. The ability to add memory andperipheral devices, as needed, makes itpossible to tailor systems to meet specificrequirements. This added capability overprogrammable controllers brings with itthe requirement that programming per-sonnel are usually required foreffective use of the minicomputer. Themost important ingredient of a successfulinstallation is careful planning. Athorough study of the application and itsrequirements should be completed beforeconsideration of a vendor and the specificsoftware and hardware approach to beused.

Microprocessors

The most recent device to make itsappearance in the control scene is themicroprocessor. This device has most ofthe Central Processing Unit (CPU)capabilities of the minicomputer on oneor two chips. Added chips are providedfor memory, input/output control, etc.

Microprocessors, in general, are lowercost with smaller memories and lowerspeeds than minicomputers. Early

applications appear to be as controllersfor testers, process control, and machinetools. Again, with more capability thanthe programmable controller, they re-quire a higher technical competence touse and maintain. The use of program-mable controllers, minicomputers andmicroprocessors in manufacturing variesover a wide gamut including machine toolcontrol, testing, facility monitoring,process control, automatic assembly,data collection, injection molding, andwarehousing.

Applications

New applications in manufacturing andproduct control are being introduced at arapidly accelerating rate. It is expectedthat this trend will continue in theforeseeable future to meet the pressuresof cost, reliability, and safety. Twogeneral areas and one specific examplefrom the thousands of specificapplications will illustrate the breadth ofuse.

Automatic warehousing

Automatic warehousing has exploited theminicomputer extensively and is offeredby several companies, such as SI Hand-ling, FMC, and GE. These systems in-clude control of trucks for placement andretrieval of product, and range in com-plexity from operator -assisted to com-pletely automated warehousing.

Injection mold control

Several packaged systems are availableusing minicomputers from such com-panies as IBM and Harrel, Inc. Theseprovide monitoring and/or control ofcavity pressure, melt temperature, dwelltime, etc.

A system developed by the Rochesterproduct division of General Motors Cor-poration uses General Automation'sSPC-16 minicomputers for carburetoradjustment and testing with a group of 31minicomputers reporting to an IBMSystem 7 for supervisory control. TheSystem 7 acts as a communication con-centrator providing information to a370/ 145 at the management level. TheSystem 7s have disc storage in case the370 goes down.'

RCA has been active in the application of

8

minicomputers to manufacturing duringthe past few years in the areas ofautomatic test, data collection, andprocess control.

Automatic test

Some of the earliest computerapplications in manufacturing have beenfor automatic test. The need for thoroughand sophisticated testing of components,subassemblies, and complete systems byGovernment and Commercial Systemshas made it necessary to develop ratherextensive capabilities in this area. TheGovernment Communications andAutomated Systems Division inBurlington, Mass. has extensive ex-perience in this area, and additionalcapabilities have been developed in thevarious divisions to meet specific needs.

The test center concept is being usedeffectively in Camden. A complex ofthree RCA 1600 computers is used fordata input to control testers for logicsystems, analog video equipment, andcommunication equipment; the complexalso includes diagnostic capability.Additionally, a General AutomationSPC- I 6 is used for system testing and aDEC PDP-8 for magnetic headwheelbalancing. Most modules produced in theCamden Plant are tested at this center.

The semiconductor industry has beenanother early user of automatic test. Theneed for high speed, accurate testing ofdevices has become increasingly impor-tant as complexity has progressed fromdiscrete transistors to integrated circuits(lCs) to medium scale integration (MSI)and finally to large scale integration(LS1). One of the first in-house developedtest systems was the Computer Con-trolled Automatic Test equipment(CCAT) for probing of wafers.' Thissystem uses an RCA 1600 computer andspecial equipment controller with threeextenders, each of which controls threewafer probe stations. Data from these testsystems is logged and taken to an off-linecomputer for analysis includinghistograms and wafer mapping.

Hewlett-Packard equipment has beenused extensively for testing of powerdevices. This testing started with thedevelopment of a system using theHewlett-Packard 2114 minicomputercontrolling three test stations to qualify

components. Since that time, continuous progressing to further improve utilizationdevelopment has been in progress.' through the use of resulting data output.

One of the first linear testers developed inRCA was the Automatic Dynamic Un-iversal Linear Tester. (ADULT), whichuses General Automation's S PC -12 com-puter as the controller.' This system isaimed at specific circuit configurationswith separate test stands for each IC type.The SPC-I 2 provides switching ofstimulus and measurement equipmentand pass or fail decisions.

General Automation SPC- I 6 minicom-puters are also being used to measureglass faceplates and funnels for colorpicture tubes. Initially started as a meansto measure faceplate contour, it has sincebeen extended to include funnel andother measurements.

Systomation in -circuit testers are beingused successfully for testing componentvalues after assembly and dip solderinginto printed circuit modules at ConsumerElectronics. These machines measure thevalue of most components to within ±1%.Since the average test time using thisequipment for a normal size board is inthe order of 5 to 10 seconds, it provides avery powerful tool for testing thosedevices which do not require adjustment.Since the system provides printout ofcomponents not meeting specification,the complicated and sometimes inac-curate troubleshooting steps can beeliminated. Additional work is progress-ing to use computer control for testingadjustable modules and subassemblies.

The use of an Intel 4004 microprocessorfor the testing of horizontal outputtransformers has been developed by theCE division. Tests performed includecontrol of an indexing table and measure-ment of inductance ratio, phase and highvoltage breakdown with the advantagesof reduced test time and increased testaccuracy. In addition, the micro-processor keeps an account of good andreject transformers. In all of these testsystems, the inherent possibility exists togather product information and enter itinto defect reporting or manufacturingdata systems. This provides manufac-turing and engineering with a tool toidentify repetitive product faults andimplement corrective action to furtherimprove product performance andreliability. In most areas where automatictest equipment is being used, plans are

Data collection

An early system developed for collectinginformation from the manufacturingfloor is the Defect Reporting System(DRS) developed in Consumer Elec-tronics. This application provides inputterminals located at troubleshootpositions on the manufacturing floor tieddirectly into an RCA1600 computer. Thedata is compiled in report format andprintout is provided by teletypewriterslocated at strategic points. These earlysystems were aimed at collecting defectinformation as it occurred and presentingit to management in a form making rapidcorrective action possible.

Another defect reporting approach beingused is Automatic Defect DiagnosisSystem (ADDS), with manual defect logsused by troubleshooters and inspectors asthe source of information. This data iskeypunched and entered into a 70/45computer for analysis. The system is quiteflexible and can be configured by eachuser to fit his own needs. Another featurepermits the user to select that portion ofdata in which he is interested such asspecific line number, model, type ofdefects, etc., by the use of delimiters whenrequesting reports. These delimiterspermit examination of the total data baseor specific portions of interest. TheADDS system is very flexible in itsapplication and, therefore, has been or isbeing used in such places as ConsumerElectronics in Taiwan, Government and -Commercial Systems, Consumer Elec-tronics and Appliances in Canada, andElectronic Components in Canada.Applications include analysis of fielddefect information as well as in -plant use.

Another development providing not onlydefect reporting capability, but alsoproduct tracking through the manufac-turing process is the Manufacturing DataSystem (M DS). This uses a GeneralAutomation S PC -I 6 computer and inputterminals similar to those used by theDefect Reporting System (DRS). Greaterflexibility is provided as the user canconfigure the entire system by filling intables which define data validationroutines, input terminal locations, datameaning and data names to be printed insubsequent reports. The user can requestreports at any time, giving product status,

9

product flow, in -process inventory, anddefect analysis. A unique feature of thissystem is the X -Y report which permitsthe user to list any combination of twoitems, one on the X axis and the other onthe Y axis, with the body of the reportshowing coincidence of the number ofcounts or quantities summed for thesetwo parameters.'

Data collection systems for specificdivisional needs have also been

developed. Video terminals as inputdevices are being used by Solid StateDivision in conjunction with an IBMSystem 7 processor. The terminals areused in a community mode in which datais entered by clerical personnel as productenters and leaves a functional area such asfinal test. In addition, data concerningthe results of tests can be entered for lateranalysis. As in most computer controlleddata collection systems, the user has thecapability of on-line examination of datain a real-time manner.

Although not completely manufacturingoriented, Electronic Componentsdeveloped an Order Control System(OCS) using a General Automation S PC -16 computer which provides warehousingcontrol throughout the country atvarious locations. Customer orders areentered by clerical personnel throughvideo terminals at each location and anexamination of the data base as toproduct availability and inventory statusis provided. The system printout permitswarehouse personnel to select productand also provides billing information toinclude with the packaging.

Process control

Process control is many times thought ofas the complete closed -loop control of atotal manufacturing plant such as thosesystems which have been installed in thepetro-chemical industries. Although thistype of complete control is usually im-practical in most of our manufacturingprocesses at RCA, it is many timespossible to identify small process stepswhich will yield very effectively to tighterand better control. For some years, cer-tain components have been automaticallyassembled into printed circuit boards atConsumer Electronics. This approachuses the United Shoe Machine equipmentin a continuous line process, assemblingone component at each station. Althoughvery effective for high volume runs with

few changes, line changeover due tomixtures of products is a time consumingand expensive procedure. Recently, bothUnited Shoe Machine and UniversalInstruments Corporation have madeavailable stand-alone component inser-tion equipment which utilizes X -Y tablesto position the workpiece under theinsertion head. The insertion head theninserts each component in turn until allinsertable components for that boardhave been put into place. The controlmechanism for this device is a GeneralAutomation SPC-12 computer which isfed from a paper tape prepared off-lineand is used to instruct the machine in theproper sequence of X, Y and Z positions.The computer also adjusts the insertionhead to accommodate the size of thecomponent to be inserted. Sequencingmachines are used ahead of the insertionequipment to place components on tapedreels in the proper sequence so that partswill go into their proper location.

Process control systems using the DECPDP-8 computer have been developed toadjust to the thickness of latex applied totufted carpeting and the speed of carpetthrough drying ovens. Microwavetransmitters and receivers are used tosense the amount of moisture in the latexmaterial, since the microwave signals areattenuated as a function of the amount ofmoisture in the carpet. With this systemmeasuring the moisture content at theoutput of the drying oven, it is alsopossible to adjust carpet speedautomatically for various carpet stylessince the oven temperature is considereda long-term constant. In addition, gasusage monitoring has been added whichhas proven to be effective in adjustingoven efficiency.

Process control has been developed forthe production of photomultiplier tubesat the Electro-Optics and Devices activityin Lancaster. Initially using a PDP-9 andlater upgrading to a PDP- I I. this systemcontrols the rate of deposition ofphotomultiplier material on the anodesof the tube and makes specific tests of theproduct as it is being processed.'

In the production of integrated circuits,many critical steps need to be monitoredand controlled. Included are the furnacediffusion steps which require tight con-trol of the variables as well as verificationthat all process steps have been satisfac-torily completed. To make these morepositive and reliable, many furnace

suppliers are developing control systemsbased on micro- and mini -computers.Solid State Division has recently orderedfurnace control systems, one using amicroprocessor with card reader input tocontrol insertion and withdrawal rates,setpoints on temperatures, and gases. Thesecond system will include a minicom-puter which feeds recipes to themicroprocessor and provides a certainamount of supervisory and reportingcapability.'

Extreme care and careful planning mustbe exercised when applying process con-trol. People with intimate knowledge ofthe process and the control system whocan effectively communicate and worktoward a common goal should be

available. In most cases, process variablesare interactive; therefore, a period ofprocess monitoring is usually needed toestablish correlations to verify intuitiverelations established during manualoperation. Special attention must also begiven to equipment reliability and backupprocedures in process control.

Conclusion

RCA is actively involved in many areas ofmanufacturing technology. The tech-niques necessary to provide acceptableyield when producing large scale in-tegrated circuits, color picture tubes toextremely tight tolerances, competitivelypriced tv sets with high reliability, andother quality products have been underdevelopment for some time. Only a few ofthe projects involved with automatic test,data collection, and process control havebeen outlined in this paper.

Since manufacturing is a major con-tributor to product cost, performance,reliability, and customer acceptance, in-creasing attention must be centered onthis part of our business.

References

1. 1 a/ar. M.11.. "Mee of direct digital control, this issue.

2. Production (April 1974)

3. Lambert. H.R.; "Computer controlled automatic testsystem, RCA Engineer. Vol. 21, No. 2 (Aug - Sept 1975) p.20

4. Guerin. W.R.; Lyden, T.B.; Martin. M.J.; McLaughlin. G.:and Ruskey. Ti.:. -DART: Diffusion Area RearrangementTask open heart surgery on a living factory.- this issue.

5. Fowler, V. and Roden. M.; "Adult II test system" RCAEngineer. Vol. 21. No. 2 (Aug - Sept 1975) p. 30.

6. Carrell. R.M.; "The computer -its use in managingmanufacturing operations." this issue.

7. Halknneier. S.; "Computer controlled processing ofphototubes." to be published.

S. Dillon, L.; Van Woeart. W.: and Shambelan, R.; "Productcontrol system for semiconductor wafer processing." thisissue.

10

7 1-7 R I 1-7 -7 Z-.7 I 121_-77

Computers have served as a useful tool in the management of manufacturing operat.ons.Such activity has involved big computers. microcomputers - and now the minicomputergathers data and produces (on demand) the reports that management personnel requires.Such an idealized manufacturing data system provides great management flexibility.reduces defect rates. and provides a convenient easy -to -use. man -machine interface byusing a personal input terminal (PIT). A special program called the "XY report generator'.can access any file, extract specific data, and provide a "quick look- or a more detailedreport for management.

R. Michael Carrell, Administrator. Manufacturing Sys-tems and Technology, RCA Staff, Cherry Hill, N.J.received a BSEE from Iowa State University in 1949 andjoined RCA Engineering Products the same year. Hebecame active in the design and development ofacoustical transducers, magnetic recording devices, andmilitary keyboards and printers. Mr. Carrell has alsoparticipated in system studies of electromagnetic in-terference problems. He joined the Graphic SystemsDivision in 1965 as a member of the Product Planningstaff. At GSD, he was also active in Systems Engineering.In 1973 he joined the RCA Manufacturing Staff at CherryHill where he is working with minicomputers in thecollection and retrieval of data concerning manufac-turing process -control systems related to production ofsolid state devices and color picture tubes. Most recent-ly, he has been involved in the process -control studiesfor the manufacture of the video disc. Mr. Carrell holdssix patents and has co-authored and authored twenty-seven technical papers. He was a chairman of NMA's"Computer on Microfilm Quality Standards Committee."Mr. Carrell is a member of the IEEE.

Reprint RE -21-3-4Final manuscript received April 15. 1975

Thecomputerits use inmanagingmanufacturingoperationsR.M. Carrell

S OM EON E once suggested that signifi-cant inventions occur not so muchbecause of the singular genius of thecredited inventor, but rather that asevents unfold, these inventions are readyto appear. Thus we have steamboats notbecause of Fulton, but because it wassteamboat time, and later, telephone timeor moonshot time.

One could say that it is now time for usingminicomputers in managing manufac-turing activities. What follows may seemobvious; hopefully it is the obvious quali-ty that results from good design.Emphasis in this paper is placed on designprinciples implemented within RCA;thus, specific applications are not dis-cussed.

Computer universes

Fig. I shows three nesting computeruniverses (big, mini, and micro). We arefamiliar with the big computer universe,which comes under the control offinancial operations in most companies.The computer operation thus tends tooperate from a viewpoint which un-derstands a manufacturing operationonly in terms of financial statements.

11

BIG COMPUTER UNIVERSE

PAYROLL

RAW MINICOMPUTER UNIVERSE FINISHED

MATERIAL MANUFACTURING PROCESS GOODS

INVENTORY INVENTORY2 3 4

CONTROL CONTROLMICRO MICRO MICRO MICRO

SCHEDULING

Fig. 1 - Three computer universes.

Actually, manufacture per se has nothingto do with money and if money is used asthe only criterion the operation will getbent out of shape one way or another.

Most financial computer systems operatein a batch mode using high levellanguages such as COBOL. Batchoperations are frustrating to manufac-turing because information isn't availablewhen needed - like now.

Time share business systems areoperating at Indianapolis and Informa-tion Services at Cherry Hill. Powerfullanguages and real-time entry/responseare possible, but at a cost too high formanufacturing process control involvinga continuous flow of data.

Now that it is minicomputer time, it isfeasible for a manufacturing operation toown a free-standing computer systemthat gathers real-time data and producesreports that manufacturing managementpersonnel want on demand. Such systems

UPSTREAMPROCESSSTAGE

FROMPARTS

INVENTORY

MOVE I

MOVE

FROMFLOOR

INVENTORY

PARTSINVENTORY

MOVE

are usually not supported by corporateinformation systems management. Incontrol applications, it is often necessaryto write programs in assembly languageswhich are unfamiliar to financialprogramming staffs.

In declaring independence of a big com-puter operation we are cut off fromsupport as well and must design free-standing turnkey systems. Now thedanger is excessive specialization.

Minicomputers were originally used indedicated, specialized control situations,a function now being taken over themicrocomputers. As minicomputercapability rivals the "big computers" of10 or 15 years ago, there emerges a newapplication tradeoff.

The idealized manufacturingdata system

\ manufacturing data system should fit

SALVAGE

SALVAGE

SCRAP

GOOD

Fig. 2 - MDS generalized manufacturing model.

TOSALES

MOVEvEDOWNSTREAMPROCESSSTAGE

MOVE

TOFLOOR

INVENTORY

MESSAGE ENTRYPOINT

like a glove, not a cast. To tailor a newsystem for each application isprohibitively costly. Nor can one alwaysafford or use the power of a high-levellanguage in a big computer system togenerate specialized applicationprograms to run in a small computer.

A middle path taken in the RCAManufacturing Data Systems has been tobuild on an idealized model of a stage ofproduction as shown in Fig. 2.

Here are the elemental actions of produc-tion. A message format is provided foreach action taken. The terminology mayvary, according to local custom, but Ithink all will recognize these elements.Just as giraffes and elephants have thesame number of noses, necks, ears, legsand tails that serve different purposes -so may manufacturing operations differwhile containing the same fundamentalelements.

If great flexibility is provided in themessage structure and nomenclature, wecan capture the correct data from anykind of manufacturing operation. This isdone by designing a table-driven inputsystem, which I will outline subsequently.

We are approaching a system with real-time data entry through personal inputterminals used by inspectors,troubleshooters, movers. etc. instead ofpaper tallies and counters. It is vital that amanufacturing data system enable peoplein all activities to describe fully theiractions. If in any way the system abridgestheir communication, forces them to lie,or go outside of it, sooner or later thepersonnel will cause the system to fail.

The most important part of the installa-tion of a data system of this type is thepreparation of the messages andnomenclature. Such preparation imposesa salutory discipline on all concerned -auser may for the first time have to thinkcarefully and logically about his opera-tion. Nomenclature must be defined withgreat care. What, precisely, do we do withscrap, rework, reclaim and salvage?

I have seen large reductions in materialvariances result from the careful prepara-tion for a computer system, before thehardware was installed. The preparationtook hard work and dedication - but itwas done. There is no magic in a com-puter system. It is a tool which works as

12

Fig. 3 - What the manufacturing data system does.

ghat MDS dues:

M DS collects messagesM DS compiles data tablesMDS prepares reports

AIDS messages MDS messages describe the manufacturing processin terms of these basic factors:

a)Count of good productb) Disposition of productc) Movement of productd) Location of producte) Description of defective product

MDS messages consist of elements and values of elements.

Example:

I pass one gorwi type XYZI reject one type XYZ for bent widgetI move 50 type PQR from soldering to location 9

Elements

1 (operator)DispositionQuantityTypeDefectFromTo

Values

Operator No.Pass, Reject, MoveI. 50XYZ PQRBent WidgetSolderingStorage Location 9

Message Formats - Messages are organized into a number of standardformats, beginning with the disposition element, which serves as a title.Examples are:

Message titles

(Pass) typeRejectReclaimReworkSalvageScrapMoveChange

description

Count good product by type and stageRecord initial defect, type and stageProduct sent downstream to next stageProduct sent upstream to previous stageProduct sent to salvageProduct scrappedMoved to or from floor storage locationChange product identity

well as it is well designed and wellunderstood.

What a manufacturing datasystem does

What a manufacturing data system doesis shown in Fig. 3. We see that eachoperator action or decision when statedas an English text message will be foundto contain elements and values which canbe represented by numerical codes. Givena well organized nomenclature, adequate

Fig. 4 - MDS message elements.

MDS message elements

Element

TypeDispositionQuantity

Definition or example

Unit type no.Pass, reject. rework, etc.Number of items

Initial defectFinal defectDefective itemDefect classResponsible machineDefect zone

Inspector's defect codeTroubleshooter's defect codeCode for defective partCode for defect causeNumber of machine causing defectLocation of defect unit

Inspectors, 1-5 Inspectors handling unit

Terminal No.StageOperator No.TimeDateShift

FromToLocation

Number of entering terminalStage of manufactureOperator entering dataHour of message entryDate of message entryShift of message entry

Starting point of a MOVEDestination of a MOVECode for a floor storage location,

or in special cases, a category

library capacity, and a table-driven inputsystem, a natural and flexibleman/ machine interface is possible.

An illustrative list' of elements is shown inFig. 4. Each of these is associated with alist of values, which for unit types anddefects may run into the hundreds.

Real-time data entry is provided by inputterminals widely distributed on the fac-tory floor. Available terminals fall intothree classes. There is the very primitive

turn -a -knob or small pushbuttonterminal. These are slow to operate orvery limited in information capacity. Thenext level of complexity usually includesa badge or card reader. Finally, there is aCRT terminal with a keyboard. Thesefunctions all miss the electronics industryas an applications area. They are reallydesigned for heavy industry operations orinventory operations where typing skillsare available. Those terminals that areinexpensive enough for wide use on aproduction line are too slow. CRTterminals are too expensive for wide use,

13

-44,/ e"/

Fig. 5 - Typical layout of a personal input terminal (PIT).

and require typing skills not available inproduction workers.

Our answer to this problem was thedesign of a Personal Input Terminal(PIT) which has a free -form button panelinterfaced with standard electronics (Fig.5 shows a typical layout).

Here the buttons are grouped by messageelements, with each button representing avalue of that element. Lights operated b}the computer guide the operator in thecorrect message -entry sequence. Theelectronics pacKage has a serial ASCII 20mA current loop interface. It is intendedfor use with a minicomputer equippedwith a communications multiplexer.

There is extensive use of canned messageswhere a button represents a part identityor a defect descriptor. The most commonmessages can be sent by a few keystrokes.Special programs were written for con-trolling the PITs. A user fills out a seriesof tables, working from his knowledge ofhis process. The table formats effectivelylead the user into programming -typedecisions and translate easily into actualprograms.

The PIT control programs validate in-coming messages and operate the lamps.Close or loose control of incomingmessages is'available at the user's option.

Data flow and status reports

Fig. 6 illustrates the data flow through thesystem. Incoming message elements aredistributed into' five major data tables,

from which five reports are extracted.Fig. 7 shows the current status report.This is a quick look at the operation as itis now. It is available on demand but isprinted automatically at the end of theshift. It shows the good output for eachtype with a breakdown of defects foundby inspectors, and the dispositions madeby troubleshooters or analysts.

A quick look at manufacturing perfor-mance is given by the stage performancereport shown in Fig. 8. This report can besummed over a selected recent periodretained in the computer data files.

In some operations, material salvage isimportant and a special report is

available to show the source and condi-tion of material flowing into the salvageoperation. This too, can be summed overa convenient time period.

A material balance sheet is available for

PERSONAL INPUTTERMINALS

MESSAGEVALIDATION

financial analysis. Fig. 9 shows a

representative situation where material isbeing moved about in a complex pattern.Fig. 10 shows a corresponding materialinput/output report from which a

financial analysis can be made.

So far we have discussed conveniencefeatures of a manufacturing data systemwhich makes life easier for foremen andreduces clerical burdens.

Savings by defect reductions

Large savings can flow from smallreductions in defect rate. The systemaccumulates detailed descriptions ofdefective product. The report of Fig. I I isthe output of a powerful data analysisprogram called the XY report generator.

Any of the data files can be accessed bythis report. In any single file any element

TRANSLATEDVALIDATEDMESSAGE

USER WRIT TEN

MESSAGE FORMATVALIDATION ANDCODE TRANSLATIONTABLES

(DATA DISTRIBUTED BY MESSAGE TYPE)

GOOD (TYPE)LOT

GOODCOUNTTABLE

REJECT

REJECTSUMMARYTABLE

MOVE MOVE RECLAIMRECLAIM REWORKREWORK SALVAGESALVAGE SCRAP

INVENTORY MATERIAL DEFECTLOCATION TRANSFER DATATABLE TABLE TABLE

CURRENTSTATUSREPORT

STAGEPERFORMANCEREPORT

Fig

XYREPORT

1

MATERIALINPUT /OUTPUT

SALVAGEBUY/SELLREPORT

6 - Data flow in the manulactuirng data system.

DATAINPUT

DATAFILES

REPORTS

REPORT OPTIONSSTAGE,DETAIL,RODE.

REPORT TIMETIME XX.10(

SH 1 PT,P ROD . HAS.DATE XX/XX

SSSSSSSSSS-- -APPLICATIONS INITIALIDENTITY CURR SHIFT DEFECT OTY. % RECL REWK SPUN SCR,II IIIIIII 0000 0000 DOODOODODD 000 00.00 0000 0000 0000 0000

000DDDDDDD 000 00.00 0000 0000 0000 0000nITITTITT 0000 0000 DDDDDDDDDD 000 00.00 0000 0000 0000 0000

TOTAL -0000---0000 -000 00.00 0000 0000 0000 0000

NOTE, S denotes StageI denote. IdentityO denotes QuantityD denotes Defect Nameo denotes Unspecified chatecter

Fig. 7 - Current status report format.

14

can be X and any other can be Y. A fileextract for a stated period is done,delimited by other elements in the file.The data is arranged in rows of X andcolumns of Y. Rows and columns aresummed, then rank ordered by columntotals.

When applied to defect data, the mostserious problem rises to the upper lefthand corner of the report. Because it isavailable on demand from the computer,an engineer can easily request a series ofreports, slicing the defect data file indifferent ways to isolate causes of defects.

Computers are not needed to isolatedisaters which shut down a line. But innormal operation, defect levels becometolerated which are expensive over a longterm. The XY report provides the meansto flexibly analyze a large body of data tofind what to work on and where to applycorrective effort.

System advantages

I he system outlined here has somethingfor everyone. It places no more burden onoperators and inspectors than existingpaper and counter systems do. It lifts apaperwork burden from foremen andgives them time to supervise, which istheir function. Supervision can get aquick look and study summary reports.Scheduling operations has a current lookat the location of work in process inven-tory. Financial operations can get acomprehensive month -end materialbalance sheet. Engineering and Qualityget a powerful defect analysis system.

These report formats may not matchlocal custom for any plant. They dologically present the principal data re-quired. Other special reports can, ofcourse, be written as needed.

Conclusion

In designing a system of this kind there isalways a danger of painting oneself into acorner. It is prudent to have enough paintleft over to paint a door in the cornerthrough which one can escape. In thepresent case this is the XY reportgenerator. Because it can access any file,all data collected can be extracted, andthere is no risk of some data being lostbecause it doesn't fit into a preconceivedreport format.

STAGE PERFORMANCE REPORTSTAGES, 101-166

STAGE WIDGET FORMHALF

--AVERAGE PRDAY A

SHIFTDUCTION

9

RATE-c

PAGE 2DATE APR 26

REPORT PERIODPROM APR 22-CTO APR 22-S

..,»PERCENT REJECT ----

DAT A

XYZ 486 61.3 606 2438 7.77 0.00 6.61 7.93

POR 110 180 150 660 21.52 0.00 17.23 26.67

DGF 256 169 600 1539 11.1/ 0.00 27.14 6.84

APE 185 167 150 634 10.26 0.00 10.18 10.34

TOTAL 878- 0.00--11.91-- 9.68

Fig. 8 - Stage performance report !ample showing two shifts of production on same date.

A315 GOOD

RECLAIM 15

REWORK 10

297 GOOD

REWORK 3 RECLAIM Ii

SALVAGE D8

C

RETURN 44 RETURN 8

STORE 4

STORAGE SALVAGE

500 GOOD SCSCREENFINAL

Fig 9 - Product flow example (see material input/output report sample)

MATERIAL INPUT/OOTPOT PEP RI'

STAGES 121-123

STAGE AINPUTS

TYPE FROM QTY

XYZ E 10

MATERIAL INPUT'OUTPUT REPORTSTAGES 121-123

STAGE B

PAGE 1

DATE MAY :9REPORT PERF,D

PROM MAY 18-ATO MAY 18-A

TYPE-OUTPUTS

TO OTY315

15

PAGE 2DATE MAY 19

REPORT PERIODPROM MAY 18-ATO MAY 18-A

INPUTS -OUTPUTSTYPE FROM CTY TYPE TO OTTXYZ A 315

STORAGE 44

a 3

MATERIAL rsiTirr,ouTruy REPORT

STAGES 121-123

STAGE C

XYZ C 297

STORAGE 14

A 103

C 11BB 12

PAGE 3DATE HAY 19

REPORT PERIODMs MAY 18-ATO MAY 18-A

INPUTS -OUTPUTS

TYPE FROM QTY TYPE TO OTY

XYZ IS 297

STORAGE 8

XYZ 0 300

Fig. 10 - Input output report sawple

X -Y REPORTFILEPERIOD FROM S DIVVY TO S DD'YYNO.

NO. YDELIMITERSFUNCTIONS

Y TITLEMY!YYYYY

rnrrtYYYYY

TYYTYWM!

YYYYYYYYYY

YYYYYYYYYY

YYYYYWm

YYYYYnewt

YYYYYYYYYY

OTHER10001

YYYYYMITT

TOTALTITLE

)00oopo0coc 00000 00000 00000 00000 00000 00000 00000 00000 00000 00000 00000

00000 OGOGQ 00000 00000 00000 00000 000.170 00000 00000 00000 00000MC0030000(

OTHER(000) 00000 00000 00000 ,30000 00000 00000 00000 00000 00000 00000 00000

Fig. 11 -X -Y report formal

15

CO3:) Merit of direct digital controlM. H. Lazar

The industrial use of computers for all aspects of manufacturing control is advancingrapidly. The economics of people versus machines and the competition for better productquality with mass production dictate it. At the present time there is much investigationtaking place at various stages within the RCA divisions on how computers can best beapplied to manufacturing. This article discusses one technique called Direct DigitalControl (DDC) that has been developed and applied in the continuous flow industries.

Max H. Lour, Manager, Systems Programming forCorporate Staff, Manufacturing Systems andTechnology, Cherry Hill, N.J., received his un-dergraduate training in Chemical Engineering fromDrexel Institute of Technology and did graduate work inComputer Technology at the University of PennsylvaniaMoore School of Electrical Engineering. As Manager ofIndustrial Systems Programming at Leeds and Northrup,he designed and developed real-time process controlsoftware for power and cement companies. Thesedevelopments included Direct Digital Control andDigitally Directed Analog Control. He joined RCA Cor-porate Staff in 1968 and has been involved in thedevelopment of software systems and tools for manufac-turing control applications throughout the corporatedivisions.

M AN U FACTU RE of electronicproducts is, for the most part, a discreteprocess and does not lend itself easily todigital (computer) process control.Although some parts of the process canbe automated, such as testing and compo-nent insertion, the entire process ofcreating an electronic product using acomputer has not been a reasonable goal.

The opposite has been true for thecontinuous -flow industries such as

petroleum refining where the computerhas been applied extensively. As a result,there is a

software and hardware technologyavailable today for using the computerfor controlling continuous -flowprocesses. Coupled with this technologyis the fact that computer hardware priceshave dropped to the point where com-puter automation in many areas hasbecome economically feasible.

Beyond process technology andeconomics, several developments are re-quired before any industry can takeadvantage of automation. One is theinterface between the computer and theproduct that transduces the computersignals to regulate or make the product.In electronics, robotics become impor-tant because of the difficulty of replacingthe dexterity of the human hand. Anotherrequired development is the un-derstanding of the process steps and theirinterrelation to the extent that the com-puter can be applied with economicadvantage over other methods. In somecases, the process may require someredesign to be compatible with computercontrol.

Reprint RE -21-3-3Final manuscript received April 15. 1975

16

Types of computer control

Direct digital control

DDC is defined as the use of a digitalcomputer for the closed -loop control ofprocess variables. This means that, atregular intervals, the computer reads inthe analog signal such as a temperature orpressure, converts the signal to a binarynumber representing the engineeringunits, compares the value to a setpoint,calculates a correction factor and outputsthe correction to a device that is tht.control mechanism for that variable. Aflowchart of this process is represented byFig. I.

The rate at which signals are read into thecomputer can vary from once per secondto once per minute, depending on thefrequency response characteristics of theprocess.

Digitally directed analog control

DDAC is defined as the use of a digitalcomputer to calculate the corrections tothe setpoints of a process. The DDACflow chart of Fig. 2 shows the setpointchanges.

The setpoints themselves are inputs toanalog controllers, one for each point inthe process, and these controllers operateindependently of the computer. Thecalculation for the setpoint correction iscalled a second level control program andis based on some higher level decision -making program related either to productvolume or product quality. Examples ofsuch calculations are:

I) A power plant where total demand haschanged. The second level programcalculates the optimum power output foreach generator based on its characteristics tomeet the total load.

2) A cement plant where an analysis of productfeed from the crushing section shows anincorrect ratio of elements. This means thatthe raw feed content has changed, and thecomponent feed setpoints must be altered tooffset this.

DDC second -level control

The control programs that determine theoutputs to each process variable andmaintain setpoints are generally calledthe first level of control. The change to thesetpoints through stored program

PROCESS

TEMPERATURE

OF

PROCESS

INPUT

COMPUTER

SE"POINT

ALARM

-A CONVERTE- -0 ENG.

UNITSPIO

OUTPUT

Fig. 1 - Direct digital control (DDC) process flow.

calculations rather than through humanintervention is generally called thesecond level of control. These setpointchanges are made to a list stored in

computer memory. The purpose ofsecond -level control is to determine thesetpoint values for the best overall opera-tion of the process. This determination ismade by reading input variables anddeciding by comparison with some per-formance index whether the process per-formance is satisfactory. If it is not, thennew setpoints must be derived that willcorrect the present condition and possiblycompensate for the length of time the less -than -optimum performance existed. Insome cases, a straightforward second -level control program will not be possibleand a trial -and -error situation must beadopted where the setpoints are alteredcontinuously for some period untiloptimum performance is again attained.

Advantages of DDC

A minor advantage of DDC over regularanalog control, if there are sufficientcontrol points in the process, is the costsaving of eliminating each analog con-troller.

SETPOINTCHANGES

A more important advantage, however, isthe capability to customize each controlloop with special programming, if re-quired. This feature cannot easily beconsidered with analog control. But witha computer, the engineers can redesignthe control because the possibilitiesbecome greater. Thus, DDC offers con-trol flexibility that is superior to analogcontrol.

Feed forward control is more flexible andeasier to implement with a computer.Feed forward control is where the ex-amination of product conditions in anearly stage of the process dictates acorrection in a later stage of the process tooffset the possibility of product degrada-tion or even total loss. This can be donewith a higher degree of sophistication inDDC than with analog control due tostored program capabilities.

Supervisory control can easily be addedto a DDC system with some additionalprogramming. This type of control isused where the computer periodicallymust interrogate the states of switchesthrough digital inputs, make alarm out-puts or printed statements providing arecord of change, and generate outputs

ANALOGCONTROLLERMI

TO CONTROL VALVES

TO CONTROL VALVESCOMPUTER ANALOG-SPPROGRAM CONTROLLER

{ CALCULATIONS #2

PROCESS ANALOGVARIABLES CONTROLLER TO CONTROL VALVES.3

Fig. 2 - Digitally directed analog control (DDAC) process flow.

17

for changing switches based

schedule.

Disadvantages of DDC

on a Elements of control action

With one computer controlling a process,backup in case of computer failurebecomes an important consideration.One solution is to have complete manualor analog backup. In the case of analogbackup, the setpoints that are changed inthe computer must also changedthrough motorized setpoints as in digital-ly directed analog control (DDAC). Thismethod effects a bumpless transfer to theprocess when the computer fails and thesystem automatically switches to analogcontrol. The drawback for analog controlis that the cost of the controllers is notsaved. A second method is to have abackup computer that would take over inthe case of failure. Mainframes today arelow enough in cost that this solutioncould be attractive. A backup computerrequires a link between the two or morecomputers with a checking system so thatif one computer fails, the other can takeover the control action after the properalarms have been sounded.

Microprocessors forautomatic backup

A more sophisticated version of DDCwith automatic backup can be achievedwith the use of microprocessors. Thesampling and control functions are per-formed by one microprocessor for one ormore process points. Data is fed back tothe central process unit (CPU). if theCPU fails, then the microprocessor main-tains the control action. Similarly, if themicroprocessor fails, the CPU takes overcontrol. In this way, adequate redundan-cy is achieved. This type of controldistribution using microprocessorsbecomes advantageous or even requiredwhen sampling frequencies are so highthat one CPU becomes saturated at peakperiods, and when operators must enterinformation conveniently for part of theprocess under their supervision. In thelatter case, a local intelligent controllercan provide communications in place ofthe CPU and thus share more of the load.This last approach is the most difficultone and requires considerable systemsanalysis to achieve success. The rule isthat the greater the flexibility, the moredifficult the design becomes because thepossibilities ate greater.

For the purpose of understanding DDC,a brief and simplified explanation of theelements of control action are describedhere. There are essentially three separatecontrol calculations performed toproduce a control action output. Theseare called proportional, integral andderivative. Mathematically, they are ex-pressed as follows:

ProportionalM = K,(SP - I)

Integral (reset)

M = Ki f (SP - 1)dtor

dMIdt = K,(SP - I)

Derivative (rate)M = Ku dIldt

Or

M = KD del dtwhere:

M is the total movement of the valve orcontrol element from its stable positionwith the process at setpoint.

SP is the setpoint.

/ is the measurement input.

e =SP - I. is the measurement control errorK is the gain factor that relates the amount of

control action to be taken to the type ofcontrol action, proportional (Kr). integralti,), or derivative (K0). This is sometimes

called the tuning factor because, in analogcontrollers, there is one knob or dial foreach factor. The process operator uses eachdial to tune the controller to the processand achieve the correct amount of controlcustomized to the process idiosyncrasies.Thus, without any mathematical analysis,given a three -mode analog controller(PID), the controller circuitry equationscan be fitted to meet the requirements of aprocess. For DDC, these factors are usual-ly entered into the computer via a terminal.

Proportional control

As seen from the equation, such controlprovides a corrective action that is

proportional to the control setpointerror. The amount of action taken isdetermined by the size of Kp. When Kr istoo large it can cause an overcorrection inthe valve position and an overshoot of thesetpoint. The process will oscillatearound the setpoint and appear unstable.If Kr is too small, valve corrections areproduced that require too much time forthe process to reach the setpoint after adisturbance. The proper Kr, therefore, is

one that will be tuned to the responsecapability of the process. Such tuningbrings the process back to setpointsmoothly with perhaps some minorovershoot and minimal oscillation.

Proportional control is rarely used byitself because it is insufficient for bringingthe process back to the setpoint whenthere is a sustained large load change.Since the proportional action is takenwith reference to a given amount of load,and a valve position to maintain setpointat that load, a large sustained loaddisturbance can cause a permanent con-trol error, or offset. The control equationsees the error but cannot position thevalve properly to correct it. Thus, theproportional control alone is not capableof stabilizing a process at setpoint underthis condition.

Integral control

In practice, integral control, or resetcontrol, action is used to supplementproportional control in order to adjustthe process to the setpoint whereproportional alone is not sufficient. Asseen from the equation, the amount ofvalve motion contributed by integralcontrol depends upon the accumulatederror over some period called the resetinterval or period. If the algebraic sum-mation or integration is zero, then nocontrol action results. As long as a neterror exists, however, there will be con-trol action. Thus, where proportionalcontrol will permit an offset due to asustained overlaod or large demand, thereset control will continue to move thevalve until the setpoint is reached. Ineffect, this action tends to 'reset' or 'shift'the proportional control to balance thenew demand load with the valve travel.

To see this, it must be remembered that M= Mu, where Mr is the final valveposition, and Mo is the initial valveposition at setpoint conditions for theprevious demand situation. Another in-sight to integral control may be gained byconsidering the derivative form of theequation, dM 1dt = K, (SP - /). Thisshows that the rate of motion of thecontrol valve is proportional to themeasurement error.

Derivative control

Derivative or rate control is sometimescalled booster or anticipating control. It

18

is used where large fluctuations in loadare expected and where quick correctiveaction is needed. Looking at either formof the rate equation, the amount of rateaction is determined by the rate of changeof either the measured variable or thecalculated error. By sensing a suddenlarge change, rate action is coupled withproportional control to stabilize theprocess. Rate control will not recognizean error, no matter how large, if that erroris steady state. Thus, rate is never used byitself but it is always coupled withproportional, or proportional and in-tegral, control.

Control analogy

To visualize the effects of the differentparts of the control equation, it might behelpful to think of a man driving anautomobile. Starting out from zerovelocity to a setpoint of 50 mi/ h,proportional control causes the pedal tobe depressed an amount in proportion tothe error. Corrections are applied to thepedal in the form of both proportionaland reset action to find the correct pedalposition for 50 mi/ h. As the car ap-proaches a very steep grade, it rapidlybegins to lose speed due to the addedgravitational pull. The rate action takeseffect here, seeing the change in loadthrough the large change in velocity andcauses a corrective depression of thepedal. This will be supplemented by bothproportional and integral control to re-adjust velocity to the 50 mi/ h setpoint.

Position vs velocity

The preceding discussion for DDC usedthe position form of the controlequations. This means that the finalposition of the control valve wascalculated. The equations normally usedin DDC are called the velocity form. Thedifference is that instead of M, AM iscalculated. Thus,

AM = AP+ AI + ADwhere

AM = M,-

This means that the position of the valveat setpoint is not considered. Only theincremental valve movement is calculatedbased on data saved from previousreadings of variables and the presentreading. From this calculation, an outputis made to the valve disregarding even itspresent position, since all that is con-

sidered is the change, or AM. ming in order to use the system efficiently.

Programming DDC

In practice, designing a program to per-form DDC in its most elegant form is acomplex task because there are manyproblems to consider. One of them iscontroller windup. Windup is defined asthe improper storage or loss of controlinformation (outputs) due to constraintson the controller output. Here, the com-puter can attempt to output controlaction at a rate beyond the ability for theactuator to respond. Thus, part of thecontrol information must be saved andmerged with the results of the nextcontrol calculation.

Bumpless transfer

When a process is going on DDC or goingof[DDC to analog or manual control, theprocess may suffer a 'bump' or suddenchange in the variable which could beupsetting to the process dynamics. .1 hiscondition must be managed carefully inthe program design of DDC.

The strategies for solving these problemshave been developed by the major ven-dors of process instrumentation.Algorithms for the conditions of windupand bumpless transfer are not trivialconsiderations. These algorithms havebeen incorporated in sophisticated DDCsoftware systems by such companies asLeeds & Northrup, Honeywell, BaileyMeter and Foxboro. More recently,General Automation, a minicomputervendor, has developed such a full-scalesoftware package called PMS, DDC.This package is currently being evaluatedby Corporate Staff, ManufacturingSystems and Technology.

A good package should have all of theprograms required for doing straight-forward DDC. These include analoginput and conversion and inputsmoothing, digital inputting, controlcalculations, analog and digital out-putting. It should be implemented, for themost part, through table filling for allparameters required to define the systemfor first level control. Further, allparameters should be easily modifiableon-line via a teletype, CRT, or specialcommunications device. Most importantis that the user engineer should not berequired to have a programmingbackground or knowledge of program -

Applications within RCA

I here is potential for the application ofDDC in areas within RCA where themanufacturing processes are complexand sensitive to minute variations inprocess conditions.

In some cases, the process steps and thevalues of the controlling parameters havebeen derived empirically, and theoperator controls the overall process byexperience and intuition. When processupsets occur that cause significant yieldreductions, it is sometimes very difficultto restore the process to equilibrium.Then too, even when the process is

running smoothly, yields may still beunpredictable or beat such low levels thata 10, i increase -could drastically alter theprocess economics. With DDC, betterinterrelated control might be achievedand at the same time other importanttasks could be performed. Some of theseare:

I. Evaluation of yields as linked to processparameters.

2. Analysis of process dynamics through theon-line collection of data.

3. Determination of defect sources throughcorrelation with parametric and mechanicalattributes.

4. Special reports, alarms and trending.

The evaluations and analyses could leadto the formulation of second -level controlprograms that would take the 'blind'spots out of the process those eventsthat cause upsets with no clues as to why.Thus, there could be many advantagesand byproducts to the application ofDDC to just these areas.

Conclusion

Process control in many forms probablywill become a major part of RCAmanufacturing in the years to come.DDC is one type of control that will, nodoubt, be applied at the proper time inthese areas where it makes goodeconomic sense. Because of the cost andcomplexity, it is advisable to use DDCsoftware supplied from a vendor ratherthan attempt an internal development.DDC is not a panacea; rather, it is asophisticated use of a digital computerand must be approached cautiously, un-derstanding what improvements it canbring to manufacturing.

19

Trend toward standardizationand automationin manufacturingM.J. Gallagher

The design of electronic products has been revolutionized. in recent years. by theintroduction of various "state-of-the-art" concepts and technology changes. Thiscontinually accelerating trend of technical innovation has, perhaps, had a tendency tocloud the significance and relative importance of some of the fundamental conceptsinvolved in the expression "producibility of design." "Producibility." specifically withrespect to the cost connotations implied, is necessarily associated with standardization ofmaterials, components. and operations which, in turn, become the basis for extendedmechanization and automation of the operations. Corollary to this process is the necessityfor improved data management at the design/manufacturing interface and, in a sense, thestandardization and automation of the data and data handling techniques involved.

Maurice J. Gallagher, Mgr.. Industrial Engineering. Communications Systems Division. Camden,New Jersey received the BS degree from Temple University in 1958. His experience in the field ofIndustrial Engineering is varied. Dating from 1946. he has been employed in progressive positionswhile pursuing his formal degree. In 1952, Mr. Gallagher was promoted to Process Engineer and in1962 assumed responsibilities of Leader. Manufacturing Engineering. In 1969 he assumed likeresponsibilities in solving divisional multi -layer printed circuit problems. In May 1970, Mr.Gallagher assumed his present responsibilities of Manager, Industrial Engineering, responsiblefor all assembly processing, standards and measurement techniques, facilitation, and manpowerplanning for effective implementation of program goals. Mr. Gallagher is presently a member of theEIA Automated Component Processing and Insertion (ACPI) JC-11 Committee representing RCA.

20

P RODUCT COST has always been animportant consideration in industrialoperations, but in the present climate ofspiraling material and labor costs, shor-tages in basic materials and aggressivecompetitive practices, we can no longerafford the luxury of "back -burner" ap-proaches to such fundamental conceptsas standardization and automation. Thispaper presents the programs undertakenwithin RCA Camden in recognition ofthis situation and details some of theaccomplishments obtained as a result.

The Camden facility is composed of twoseparate operating divisions; one broadlychartered for the development and designof government communications systems,and the other with correspondingdevelopment and design responsibilitiesfor commercial broadcasting equipment.Both divisions are serviced by a singlemanufacturing organization.

The product lines of the two divisions arebroad and diverse, with unique problemsin each product area. Product sizes andcomplexity cover a spectrum rangingfrom "manual portability" to major fixedinstallations. The one point of com-monality which exists, however, is thatnone of the products of either divisioncan be described as a mass production orhigh -volume product. Because of thisbasic fact, automation concepts cannotbe based on product volumes, but ratheron operation volumes across productsand product lines. And this, in turn,promotes the desirability of standardiza-tion concepts across products andproduct lines to build and provide thenecessary operation volumes (see Fig. I ).

Interdiscipline communica-tion

Development and implementation ofprograms involving both design andmanufacturing disciplines obviously can-not be successfully accomplished byeither discipline alone. Thus, it is critical-ly important to establish a working basiswithin which mutually agreeable ap-proaches to cooperative efforts can befound. Fortunately, at Camden, this factwas appreciated by management levelswithin both disciplines, and a workingrelationship, founded upon the necessityfor interdiscipline communication andaction was established.

The working committees formed to fosterthis concept were, of course, composed of

Reprint RE -21-3-9Final manuscript received April 15. 1975.

MECHANICALWIRE

COLOR

CODING

AUTOMATIC

CUTTINGSTR PPI 6

WIRE

AUTOMATICCUTTING&MARKINGSLEEKING

CABLEFABRICATIO

WIREFABRICATIONAREA

rAUTOMATIC

WIRE

WRAP

SEMI-AUTOMATICWIRE

WRAP

WIREWRAP

"FACILITY.

INCOMING

MATERIAL

FINALBACKPLANEASSEMBLY

SUBASSEMBLY

FINAL

ASSEMBLY

SYSTEMS

r --AUTOMATIC

SEQUENCING

MACHINE

AUTOMATIC

COMPONENTINSERTIONMACHINE

AUTOMATICINSERTION

FINALMODUL

ASSEMBLY

AUTOMATICWAVE

SOLDERING

Fig. 1 - Genetral production flow.

representatives from both Engineeringand Manufacturing as well as fromvarious Engineering and Manufacturingsupport activities. The committees wereall oriented toward developing and main-taining sets of compatible standardsacross program lines.

A considerable standardization effort hasevolved from these committees includingthe development and publication of anEngineering Standards Manualcontaining lists of preferred componentsand related engineering and manufac-turing technical data. Significantly,however, it has been found that thecommittees represent a pooling ofknowledge and experience which can betapped as an information source for theresolution of many "non-standard"problems.

A direct outgrowth of this standardiza-tion effort has been the development andpublication of an Industrial EngineeringManual. This manual serves to docu-ment, for continuing implementationcontrol, the specific standardized techni-ques for transmitting and handling dataacross the Engineering/ Manufacturinginterface, and serves as a continuing

AUTOMATICCUTTINGFORMING

COMPONENTS

AUTOMATIC

TRIMMING -INTEGRATEDCIRCUITS

CUT TINGFORMING

MISCCOMPONENTS

COMPONENTPal! PAA R AT INIC

L -- --I

guide to the Industrial Engineers in theselection of appropriate standardoperations and all pertinent data relatedto them. Information covering all aspectsof the Industrial Engineers' function in-cluding standard processes, toolingidentification, related labor measurementstandards, and similar data are all in-cluded in the manual.

The Industrial Engineers Manual is by nomeans complete - nor is it ever intendedto be completed. It will continue to growand be updated as additional progress ismade. Some of the accomplishmentsmade to date will be detailed in theremainder of this paper.

Data handling

The handling of data is by no means asmall part of the overall task. In somecases, simplification and uniformity ofdata and data handling is a goal in itself;in other cases, successful data handling ismandatory to properly implement andcontrol automated operations.

One of our long-range standardizationprojects is the development of a format

for engineering documentation that willallow the data to be more readilyavailable to other, using activities. Todate, with the cooperation of Engineer-ing, material lists have been redesigned toinclude information that aids in schedul-ing, purchasing, and manufacturingdecisions.

Additionally, by analyzing thedocuments used by such diverse activitiesas material ordering, stock disbursement,and manufacturing, certain elementswere found that were sufficiently alike asto encourage standardizing efforts inattemps to provide data in a formatcommon to all areas. One of the formsdeveloped, and now computeroriginated, is capable of serving as ashortage notice for material ordering, a"pull card" for the Stockroom, adimensional and operational order forthe component preparation activity and abin card for manufacturing areas.

In the realm of manufacturingoperations, some functions, essentiallybecause of generated volume con-siderations, are more conducive to stand-ardization and automation than areothers. Wire fabrication operations haveprovided a fertile field for such efforts -to the point, in fact, that computeroperations are utilized to accumulaterequirements and to generate appropriateinstruction documents. Computer inputsare, of course, standardized and con-trolled.

Wire fabrication

Wire is cut to length and stripped onautomatic equipment. Although theequipment has been in operation for anumber of years, recent innovations haveimproved the overall operation. Due tothe automatic nature of the equipment,the restricting dimension affectingproduction economics is operationquantity and resultant setup time. Byutilizing an operations method based oncomputerized generation of bulk fabrica-tion quantities (by shop order rather thanby reference number), and thereby assur-ing maximum operation output from theequipment, the use of automatic cuttingand stripping equipment (see Fig. 2) hasnot only aided in obtaining dollar savingsbut has offered the assurance of positivelycontrolled measurements.

An added benefit of this computer

21

program is that each wire, with its in-dividual dimensions, identification, andfabrication needs, is not only listed, but isalso accompanied by a computer card.Once the wire has been cut and stripped,the card functions as a travel tag as thewire progresses through variousoperations to the point where it ultimate-

ly loses its identity by being included inanother assembly.

After an analysis of wire color -codingrequirements, it was determined thatstandardization in this area could lead toadditional benefits. Equipment has beeninstalled to automatically stripe white

Fig. 2 - Automatic wire cutting and stripping equipment can cut and strip wires up to 30 gauge to predetermined lengths.

Fig. 3 - Automatic wire color -coding equipment.

wire, using the new standard, in anycombination of four colors desired (seeFig. 3). The massive inventory ofdifferent colored wire that had to bemaintained has now been minimized; yet,the desires of the Design Engineer can bemet, often on a more timely basis thanbefore.

The wire -fabricating area also has theresponsibility of providing assembledcoaxial cables to the other assemblyareas. It has been the practice of In-dustrial Engineering to process eachprogram and cable on its own; however, astudy of past requirements revealed that,in many cases, connectors and cablesused on one program were also used onothers. The present method relies onstandardized processes for each connec-tor and cable assembly. These processesare compatible to many programs assur-ing that all connector assemblies are nowidentical and alleviating the need toprepare processes for new equipment.

Prior to the advent of standardizationactivities, sleeving was purchased in shortlengths of various diameters, thicknesses,material, and colors. This, of necessity,created a large inventory and the cuttingand marking by time-consuming manualoperations. Now, using automatic equip-ment, the activity has the ability to cutand mark sleeving in a fraction of the timeon the sleeving machine shown in Fig. 4.

As the old three-foot lengths becameobsolete and the sleeving machine couldbe fed from a reel, a program was startedto standardize sleeving. Sizes and colorshave standardized, allowing for bulkpurchasing and processing plus a

minimum inventory. As a result, thedemands of production are kept at aminimum and savings can be realized.

Wire -wrapping facility

If standardization and automation are tobe successfully achieved the utmostcooperation between Engineering andManufacturing cannot be over-emphasized; both activities must realizethat the benefits are mutual. An extreme-ly good example of beneficial interfacebetween these two disciplines can befound in the progress made within theCamden wire -wrap facility (see Fig. 5).

The Camden manufacturing activity hasa full capability for performing wire -

Fig. 4 - Sleeving machine

Fig. 5 - Automatic wire -wrapping equipment.

wrapping services utilizing either fulllyautomatic or semi -automatic equipment.

The automatic wire -wrap machine iscompletely automatic and its sequence ofoperation, as well as the actual electronicimpetus, is controlled solely by the tabcard deck. Upon command, the machinewill strip and wrap both ends of a wire,cut it to length, and dress it in accordancewith a pre -planned program. Themachine has the capability of wrapping

three levels of wire on panels as large as 22by 42 in. Although limited to single, solidconductor wires, the machine can handlefrom 22- to 30 -gauge wire. Allowing forsuch variables as setup time and loadingtime, the machine has the ability to wrapup to 500 wires/ hour.

The automatic wire -wrap machine con-sists of movable carriages containingwrapping -tool assemblies and dressingfingers that are positioned on modular

points to form a desired wire pattern. Thecommands activating the variousmachine movements originate with com-puter inputs. The data utilized to preparethese tapes are the Design Engineeringblueprints and specifications which canbe easily traced to Design Engineeringdocuments. When Design Engineeringprepares their wiring information, anextract of the information is generatedwhich, with minor manufacturing inputs,ultimately becomes the card deck thatoperates the equipment.

The computer data is supplied on aprintout and a deck of punched cards;each card of the deck contains informa-tion for one wire; a card reader, in turn,translates the input into mechanical mo-tion. Fig. 6 shows a backplane preparedby the automatic wire -wrap machine.

The semi -automatic machine is alsonumerically controlled. This equipmentis utilized to wrap wires that are beyondthe scope of the automatic wire -wrapequipment. For example, twisted pairsare laid in by an operator in accordancewith equipment -controlled locationpatterns. The machine is controlled by apaper -tape input that directs the machinethrough programmed operations. Theseoperations locate the wrapping head overthe particular terminal to be wrapped.The operator then loads the head with thewire and completes the termination. Themachine has the capacity for a 30 by 72 in.panel, can be programmed to do point-to-point or path wiring, and can also beconverted to two heads (permittingsimultaneous assembly of two smaller

Fig. 6 - lhis wire -wrap panel is an example of thecapabilities of the automatic wire -wrap equipment

23

panels). An average of 170 wireterminations/ hour can be produced bythis equipment.

The key to progress in this area has beenthe development of feasible, standardizedtechniques for handling the engineeringdata, controlling it to prevent significantwiring errors, and development of thecomputer program which automaticallygenerates the equipment operating in-structions in the form of card decks. Thewire -wrap facility has proven sosuccessful that additional equipment hasbeen purchased to handle the increasingload. The expansion will include bothautomatic and semi -automatic equip-ment. Due to improvements in design, thenew equipment will more than double thecanabilities of the activity.

Automatic insertion activity

Another facet of manufacturing recentlyautomated in the Camden plant is theassembly of components to printed cir-cuit boards. The automatic insertionequipment (see Fig. 7) contained in thearea consists of taping, sequencing andinsertion equipment; modules may beassembled in a fraction of the time, withless assembly personnel than previouslypossible.

Although the automatic insertion techni-que is not new, standardization conceptshave allowed for greatly expandedutilization of this economical manufac-turing facility.

In the past, components were assembledto printed circuit boards by hand - atime-consuming operation exposed tooperator error. Today, due to the stan-dardization of components, mountingdimensions, and other design con-siderations, a greater proportion of axiallead components are mounted byautomatic equipment. In the Camdenactivity, the percentage of automaticallyinserted components has risen from 25%to approximately 75% of the componentsin recently designed equipments.

For automatic insertion, the normalmethod of purchasing components is torequest that they be on taped reels;however, this is not always possible. Theautomatic taping machine has been in-stalled to cover this contingency. Loosecomponents are placed in a hopper andfed through the machine, dropping at

preset intervals onto tapes that enclosethe lead ends; they are then gatheredupon reels. The machine has the capabili-ty of taping 5,000 components/ per hour.

The taped reels of components (eitherpurchased or taped in-house) are placedin position on the sequencing equipment.In operation, the machine is whollyautomatic - its commands being dic-tated by punched tape. This machine hasa capacity of 39 reels. The tape, punched(in code) to a pre -determined componentassembly sequence, is fed over a scannerwhich dictates release of one or moreparts from one of the 39 reels. Thereleased components travel along a trackto a taping unit and are taped and fedonto a reel in the proper insertion order.

Although the machine has the capabilityof sequencing 5,000 components/ hour, itis the limiting piece of equipment in thearea. In order to minimize down time forsetup, each module utilizing automatical-ly inserted components is scanned forcommon usage of parts. Those com-ponents with the most common usage areassigned to specific locations on themachine. In this manner, only aminimum number of reels need bechanged when preparing the sequencerfor a new assembly.

The reel of sequenced components isplaced on the automatic insertionmachine. This machine, also operated bytape, removes the component from thetape and inserts it in its proper positionon the printed circuit board. The machine

selects the part, inserts it, and clinches itbelow the board automatically. At thesame time, the machine will move its tableso that the various locations on the boardare correctly oriented to receive the com-ponent.

The automatic insertion machinepresently in use can insert axial leadcomponents on mounting centers of 0.4to 0.7 in. It has the ability to insert up to3,000 components/ hour.

Although the present equipment haslimitations, such as the necessity toprogram separately for each mountingcenter, the automatic insertion area, likethe automatic wire -wrap activity, is alsoundergoing expansion. Moresophisticated equipment, increasing thepotential of the facility, is being added.The new variable -head insertion machinecan insert components on mountingcenters ranging from 0.3 to 0.8 in. It canalso insert components to more than onemounting center automatically. This, initself, may be considered a major asset -as it minimizes setup time. Although theequipment has not yet been exposed toproduction operation, it is estimated that,with setup time, it will insert up to 6,000components/ hour.

A computer program is presently underdevelopment that will ultimately controlall data needed to operate this equipment.Once the path or sequence of insertionhas been plotted on a drawing, the datawill be fed into the computer. The com-puter, using data stored in its library, will

Fig. 7 - Automatic insertion activity.

24

Fig. B - Simi -automatic component cutting and formingmidden.

then supply all the necessary informationfor the insertion machines. At the sametime, a punched tape will be supplied thatwill operate the equipment. In this way,the conversion of the Design Engineer'sdata to production data can be ac-complished in a minimum time span.

A new sequencing machine will relievethe capacity restrictions presently ex-perienced and a new variable -headautomatic insertion machine has beenpurchased. Eventually, this facility will beable to process an expanded number ofmounting dimensions, components ofgreater body size and, in addition, in-tegrated circuits.

Component assembly

Almost all printed circuit modules con-tain components that do not lendthemselves to automatic insertion. Usual-ly, this is due to the configuration of thepart, although there are otherrestrictions. Components of this naturemust be assembled by hand. To minimizethe time needed in actual assembly, thesecomponents are pre -formed and cut tolength, where possible, on automatic andsemi -automatic equipment. Here, too,standardization of components, as well asrelated design and manufacturing con-siderations, has allowed automatic ap-proaches which minimize operatingcosts. Tooling has been designed to pre-form these components to meet thedemands of the Design Engineer.Whether the component be socket -mounted, inserted through a spacer, orraised above the printed circuit board,tooling is available to assure compatibili-

ty with the design and guarantee con-sistency of form.

A semi -automatic machine (see Fig. 8)that can cut and form components topreset dimensions, is capable of process-ing up to 8,000 axial lead com-ponents/ hour. The machine can be ad-justed to form components to mountingcenters of 0.4 to 3.00 in. When desired,this equipment can also form com-ponents with loops (to facilitate assemblyto terminals) on the ends of the leads.

Other semi -automatic equipment canalso process components of various con-figurations and can be used not only tocut and form axial lead components forhorizontal or vertical mounting but toprocess parallel lead components such astransistors and capacitors.

Another of the semi -automatic machines(see Fig. 9) in the area trims to length theleads of 14- and 16 -lead integrated cir-cuits. It is gravity fed; the components fallfrom one magazine into another, passingover the trimming blades enroute. Themachine can trim the leads of 1,000integrated circuits/ hour.

Wave soldering

Upon completion of assembly, whether

Fig. 9 - Integrated circuit Nod trimmer.

by machine, hand, or both, the module isthen soldered in wave -soldering equip-ment. The modules are mounted inholding fixtures that are, in turn,suspended between endless conveyorspassing through the machine where acoating of flux is applied to the bottom ofthe module. After the flux has beenapplied, the module then passes over acascade of molten solder. In this way, thebottom of the board, just touching thesolder, receives its coating. After solder-ing, the module passes through an area inwhich it is subjected to a spray of solutionthat dissolves the residue of flux. Fromthe bath it goes through a drying area,and thence back to the loading station tobe removed. The soldering machine iscapable of processing 100 modules/ hour.

Like the previous operations, wave -soldering is also influenced by stan-dardization considerations. Componentselection, component density, copper lineand space width, even board size andconfiguration, can influence the basicfeasibility and general quality of thisoperation.

Conclusion

Standardization efforts and the automa-tion of operations made possible throughstandardization efforts are never-endingjobs. Currently, the Engineering,Manufacturing, and ComputerProgramming activities are engaged in aninvestigation into the feasibility of con-verting raw engineering data into multi-use documents. Working fromschematics and logic diagrams, data hasbeen processed that has been beneficial toDrafting and Test Process as well as thetwo Engineering groups involved. "Fromand to" lists. cross-reference lists, andDITMCO data have been compiled alongwith the various numerical control inputsnecessary for machine operation. Effortis underway to develop a library of data inthe computer that will enable it to furnishadditional information such as wirecolor, type, and part number. A formulathat will permit the computer to furnishrouting information for use duringdevelopment of a cable harness, and alsofurnish the wire lengths needed tofabricate it, is being devised. Additional-ly, research is presently being conductedinto the methods of assembly that willreduce assembly time and yet assure thatthe design is faithfully being reproducedon a production basis.

25

Twisted -pair wire dispenserautomatic and programmableL.D. Ciarrocchi J.J. Colgan, H.F. Schellack

There has long been a need for twisted -wire dispensing machines that are programmable.To fulfill this need at RCA, the Equipment Development Engineering Section in Camdenmore than three years ago introduced a prototype machine design. More recently. asecond machine (also RCA design) was placed in production. This second machineincorporates many updated innovations. Solid-state reliability. ease of maintenance andproduct flexibility are the key features of this design. The design provides tangible costsavings in terms of wire prep and production scheduling; also. it provides an alternate wireprepping capability under the direct control of semi -automatic wire -wrapping machines.

THERE ARE a number of reasons whyit is not possible to wire wrap certainprinted circuit backplanes on currentlyavailable sophisticated automatic wire -wrap machines. These include the size ofthe backplane (some of which fill astandard floor -mounted cabinet), someunusual connector terminations to wire -wrap pins, and the increasing use oftwisted wire pairs for most runs. Ouroriginal prototype, even though limited

L.D. Ciarrocchi's and H.F. Schellack's biographies andphotographs appear with their other article in this issue.

J. J. Colgan, Equipment Development Engineering,Communications Systems Division, Camden. NewJersey has had 33 years of industrial experience, in-cluding 30 years in tooling, machine design, and produc-tion engineering. He came to RCA in 1942 as a ToolDesigner. He left in 1943 to serve three years in the U.S.Air Force. After returning, he attended Drexel InstituteEvening School. He joined the Equipment DevleopmentEngineering group in 1949 and had a prime responsibili-ty in facilitating RCA's early introduction to printedcircuit fabrication and assembly, tv yoke coil equipment,.and the ferrite plant.

Reprint RE -21-3-6Final manuscript received July 8, 1975.

to use with a single semi -automatic wire -wrap machine and limited in its flexibili-ty, paid for itself in the first six months ofoperation.

The 'Twister'

The Twisted -Pair Wire Dispenser, il-

lustrated in Fig. 1, is the new productionversion currently in service. The Twistercan be pre-set to dispense a single wire,two single wires or a twisted pair ondemand of the semi -automatic wire -wrapmachine. The Twister will feed, twist, cut,and strip to the specified length, asprogrammed on the wire -wrap machinetape. The machine will handle lengthsfrom 5 to 80 in. with turns/ in. variablefrom 1.5 to 5. On a shared time basis (withits built in first -come, first -servedmemory), the twister will supply wires totwo semi -automatic wire -wrap machineswhich may be operating from entire!)

different tape programs. The Twister mayalso be pre-set to deliver two twisted pairsfor each wire -wrap machine should therebe a requirement to wrap two identicalpanels on a given program on each wire -wrap machine.

It is an oversimplification to merely statethat we feed, twist, cut, and strip the wire.From past experiences in a number of coilwinding applications (tv deflection yokecoil winding, etc.) we have been exposedto some of the difficulties in handlingwire. Straightening of the wire, as well asmaintaining proper tension, is important.In addition, it was necessary to compen-sate for the "twist" memory and the"spool" memory of the wire.

Twister design

The wire tension and straightening deviceis designed to operate with all wire sizeswith little or no adjustment. It consistsprincipally of clothes lines twisted aboutthe wire to which weights are added forproper tension. A large change can bemade by altering the turns; or, a smallerchange is made by varying the weights(see Fig. 2).

Threading of the wire is accommodatedby removing the clothes line clamp andpermitting the lines to unravel, which is aconvenient way to reset. The tension isrelatively constant and the straighteningeffects are adequate for the application.There are no complex moving parts, andmaintenance will consist of replacing theline possibly in a year or more. The wiredereeling device will handle all knownspool sizes up to 18 in. in diameter

1 - Tvoleled-poir wk. dispenser

26

Fig. 2 - Wire -handling mechanics.

without the use of special adapters.

Careful attention was paid also to theother mechanical aspects. The carriageclamp, which holds the wire as thecarriage moves to its programmed length,is self adjusting and compensates forvariations in the O.D. of the wire.Telescoping tubes are used to cover thepath for the wire from the pre -feed beltdrive directly to the carriage clamp toprevent possible erratic feeding of thewire. The strip and cut blades areassembled for a particular wire gage and afixed stripping length. A changeoverfrom one wire gage to another takes onlyseveral minutes. The stripping lengthsmay be altered from 0.100 in. to 1.500 in.through use of a mix of spacer blocks.Since strip and cut blades are assembledas a set, there is no difficulty in alignmentwhen inserting them in the machine (seeFig. 2).

The selection of a stepping motor for themain carriage drive provided the flexibili-ty to obtain the varied functions desiredin wire handling. To compensate for thebuilt-in wire memory, which withoutcorrection could cause the wire to unravelif released, the stepping motor is

programmed to provide varying amountsof stretch, proportional to the wirelength. The stepping motor is alsoprogrammed for pulsed reversing whiletwisting is in progress to return thecarriage a fixed distance in proportion tothe turns/ in./ unit length of wire.

A vacuum system to draw off the strippedinsulation into a collector tank keeps themachine clean and the blades un-cluttered. The exhaust from the vacuum

Fig. 3 - Operator control coneolette.

system also provides the cooling air forthe program controller. The compart-ment is insulated to reduce the noise toacceptable levels.

Control consolette

The controls provide three modes ofoperation: manual, semi -automatic andautomatic (see Fig. 3).

Each basic motion of the machine may beactivated in the manual mode from anarray of seven separate pushbuttons toprovide capability for setup andcheckout.

The semi -automatic mode permits opera-tion as a batch wire -prep unit. Wireparameters can be pre-set for a maximumbatch of 9999 wires. The counter will stopthe machine at the end of its setting.

The automatic mode is used in conjunc-tion with remote control lines to interactwith one or two wire -wrapping machines.The automatic mode will also permit aninterruption within the current dispensercycle without going out of phase. Thisfeature was required to provide for make-up wires and proper sequencing whenplacing one of two wrapping machineson-line or off-line.

Programmable controller

The sensing devices and the program-ming controls, as well as, control of themain carriage drive stepping motor aresolid state. Mechanical relays are usedprimarily for isolation of power outputs.

There are twenty-nine independent mov-ing functions performed during eachcycle, all of which are pre-programmed ina sequence compatible with the conditionstatus of over 30 sensing devices.

As an aid to altering the program, as wellas assistance in maintenance, a visualdisplay on the controller indicates thestatus of the sensing devices. A perma-nent memory system retains the se-quence, once the program is established.The program may be altered in thememory board through other availablechannels or through use of other pre-programmed memory boards, as shownin Fig. 4.

Conclusion

The new Twister has met the basic designgoals in filling the void for wire -wrapsupport equipment. OSHA safety stan-dards were observed, a fast "pay foritself' was realized, and reliability wasenhanced by the unique mechanicalfeatures which were incorporated and theuse of solid-state controls predominately.

Fig. 4 - Solid-state programmable controller.

27

18,000 -point automatic testinginterface machineL.D. Ciarrocchi H.F. Schellack

An 18.000 -point automatic testing interface machine designed and developed in Camdenpermits testing of large backplane interconnect systems by interfacing with DITMCOAutomatic Test Equipment (DATE). Manual insertion of hundreds of connectors andmanual test probing are circumvented with the DATE machine.

THE MOST RECENT TRENDS forvery large electronic systems are to makeuse of cabinet sized backplane wire-wrapped interconnects. Typically, thenumber of interconnects are in excess of10,000 points per backplane. The testing

of completed wiring is an astronomicaltask for manual testing. One concept thatprovides a quick and reliable hookup for

Reprint RE -21-3-10

Final manuscript received October 10, 1974.

Louts D. Clarrocchl, Mgr.. Production Engineering Com-munications Systems Divisions, Camden, New Jersey.graduated from Drexel University in 1937 with a dipolmain electrical engineering. He served in the USAF from1941 to 1946; his last assignment was Staff Ordnance andArmament Officer for the 9th Air Force. From 1946 to1956 he was employed as a project engineer by theBaldwin Lima Hamilton Corporation responsible forDiesel -Electric and Electric Motive Power Controls. Hislatest design covered in -motion automatic switching oflocomotive power from third -rail power to diesel electricpower and back again to meet environmental pollutionlimitations in the New York City area. Prior to joiningRCA in 1959. he was employed by Selas CorporationasSenior Design Engineer with responsibility for in-strumentation and controls in heat treating and fluidprocessing systems In his present assignment, in addi-tion to production engineering support for Broadcastand Government Communications Systems Products.he has the responsibility for the direction of the Equip-ment Design. Development, and Tool Design Engineer-ing Group.

Harry F. Schellack, Equipment Development andProduction Engineering. G&CS. Camden. New Jerseyreceived the BSME in 1955 and the MSEE in 1961 fromNewark College of Engineering. Newark, New Jersey. In1955 he joined the Tube Division, Harrison. N.J., as aDesign and Development Engineer and worked onmanufacturing research in the Methods and ProcessesLab. He was granted foreign aria domestic patents ontube components manufacturing an delectornic packag-ing configurations. In 1961. he transferred to the DefenseElectronics Products Division, Camden. N.J. His produc-tion engineering and manufacturing equipment designassignments were associated with various RCADivisions in the United States and foreign countries. Hehas written several magazine articles on manufacturingequipment for printed circuits and in 1969 presented apaper to NEPCON on fabrication controls for manufac-turing printed circuits. For several years he taughtevening courses in computerized N/C programming atthe Camden County Vocational and Technical School.Mr. Schellack is a member of Pi Tau Sigma.

an 18,000 -point system was designed andbuilt by Camden's Equipment Develop-ment Engineering department (see Fig.1). The concept interfaces with DITMCOAutomatic Test Equipment (DATE).

With the system, automatic testing ofcontinuity and shorts, copper -path -networks and hi -potting is accomplishedquickly, safely, and accurately.

Need for an interface machine

From past experience, the manual "plug-in" on large systems was found to con-sume a disproportionate share of theactual testing time. In addition, repeatedinsertion and removal of wired recep-tacles introduced errors which requiredanalysis to determine whether the errorwas in the unit under test (U UT) or theinterface cables and connections. Theeconomics of automated testing withDITMCO Automatic Test Equipmentdictated development of methods toreduce an operator's time in connectingto a U UT and to improve reliability.

Interface equipment was developedwhich reduces the operator load -unloadtime from approximately one-half hourto less than one minute. The machinedesign utilizes a spring pin contact foreach of the 18,000 points (see Fig. 2). Thisfeature keeps all the interconnect wiringfrom the interface contact pin to theDATE stationary at all times, and, for allpractical purposes, eliminates wiring andequipment errors. As such, there are nowear problems as with repeated pluginsertion. Spring pin assemblies arerhodium or gold plated for long-termcorrosion and wear resistance.

The fact that a plug-in connection of a"card -edge" is not being used to simulatethe "final usage" board connection hasnot been a significant factor. The bulk offaults are concentrated in the wire wrap-ping operation. The spring pin backsidecontact to the terminal pins of the U UTshas been extremely effective for DATEinterface testing.

Machine description

The machine frame panel A (see Fig. 3) ispermanently wired to DATE. WithinPanel A, a spring pin is pressed in place ateach wire termination and in a spacing

28

Fig. 1 - Interface machine. in foreground, which connects with DITMCO Automatic Test Equipment(DATE). partially visible on platform in background

array to match the UUTs (e.g., a gridspacing of 0.125 in., 0.100 in., etc.). Thetest panel is positioned in a T cart onwheels. The machine has a moveableframe member (B) which travels from theretracted load -unload position to the testposition. The T cart, with the UUTbackplane accurately prepositioned andsecured, is wheeled in and out of theinterface machine. Two T carts werefurnished with the machine; one with theU UTin the machine undergoing test anda second to permit the operator to loadand unload during machine running time.

The second T cart is also used as repair

TYPICAL PLUG -RICONNECTOR

BACRPLANESUESTRATE

TYPICALWIRE.WRAP

SACKPLANE PANELUNIT UNDER TEST

station for the backplane to correct errorsrecorded while under test.

The operator's function is to place thepanel on register pins and clamp. Theoperator performs this function with themounting platform in a horizontal posi-tion. After clamping the UUT in place,the platform is pivoted and locked in thevertical position for insertion into themachine (see Figs. 4 and 5).

The rear of the backplane mounting plateon the T cart is fitted with T -slots, similarto that found on surface plates of milling

TYPICALAMINO -LOADEDINTERFACE CONTACTPIN

FIXED MACHINEINTERFACE PANEL

TYPICAL

: AWIRING TO1 DATE"

Fig 2 - Spring pin contact design used in interfaceequipment.

machines, that engage T -bars in themachine during loading and unloading.This engagement serves a dual purpose;first, it places the T cart in position forpositive alignment with the pins; andsecond, it prevents damage to the springpins in the loading -unloading process.

Since the machine is capable of handlinga number of types of U UT's the 1 artsare equipped with sensing actuatorswhich enable the machine to detect thetype of UUT and control the closure onthe pins accordingly.

During the operating cycle, the move-ment of frame B also lifts the T cart toaccurately align the UUT in relation tothe spring contact pins and lock theassembly in place against the B framebefore closure with the A frame springpins is complete. The same closure motiveforce provides the cam action to align andlock. A permanent stop preventsoverstressing of spring pins whilepermitting positive contact. The travel ofeach spring is thereby controlled to assurea minimum of 3 oz of pressure contact.

The B frame drive motor is directlyto a gear reducer which in turn is

connected to a four -post jack screwarrangement that provides positive mo-tion without distortion.

Upon achieving test position, themachine locks up the UUT and dis-connects the power drive. OSHA -conforming operating controls and in-terlocks are provided for operator safety.The typical interface cycle takes about 20seconds to move the 18,000 points from

NOYASLE

FRAME

Fig 3 - Machine interconnection systems.

18000 WIRESTO

'TATE"

29

- - T-et grivirialt& wa ..c-AMW1111111. 11116111LM:_

mrao

Fig. 4 -A typical backplane UUT is shown with the T cart mounting surface in the horizontal "load -unload" and "repair" position.

the load -unload position to the test posi-tion. The same time is used to reverse thesequence. Wheeling the T cart into or outof the machine takes 30 seconds.

Tons of force

The frame work of the machine is sub-stantially rugged to cope with the forcesrequired to compress all 18,000 springpins simultaneously. As can be seen fromFig. 6, the uniformly distributed field of"ounces" of spring pin pressures in-troduces a possible net force of over 2tons (up to 4,500 lbs.) in the testingposition. As such, the T cart includesbackup support for the U UT backplane

"HUT"RETRACTED

4ozs.MAX.

NETDISTRIBUTED

FORCE

IN

POSITION

P=I18.000PINS i4ozs per pin mu.)P=72.000ont OR 2 25TONS

panel to prevent buckling. Aircraft -typehoneycomb sections (see Fig. 7) are usedfor support of the UUT and to transmitthe load through the T cart onto themoveable B frame.

Manual vs. machine reliability

Hand insertion of 18,000 points with 360connectors (assuming 50 points per con-nector), at about 10 seconds per plug-in,would take approximately one hour tointerconnect to the plug-in side of U UTs.

Conversely, it is never necessary to dis-connect the wires that permanently tie-inthe interface machine with DATE. Thus,

SPRING - PINS

MACH.

FRAME

Qi

I

I

I

I

Fig. 5 -A typical T cart with a backplane UUT registeredand locked onto the mounting surface which has beenflipped and secured into the vertical position preparatoryfor wheeling into the interface machine.

many hours of time to connect DATE,shown in Fig. 8, with the UUT areobviated.

Once the permanent DATE hook-up tothe interface machine is made, it is

reasonable to expect to approach 100%reliability. Possible operator errors areeliminated by virtue of not insertinghundreds of mating connectors intoreceptacles.

Most commercially available multipointinterface units are arranged with springpins in a horizontal plane and usually ona hinged upper panel to provide access forloading and unloading of a backplane.The connecting wires are in motion for

Prei Apr1/141-t I ilifittlr?

Fig. 7 - Aircraft -type honeycomb structure for prevention of buckling of UUT backplaneFig. 6 - Net force distribution with UUT in test position. panel.

30

each loading and unloading cycle withthis design concept. Alignment andmaintenance are also more difficult.

The vertical concept provides clear accessto both the spring pins and the wiring inthe rear and, also, keeps the wiringstationary.

Machine versatility

The machine was designed and built toconform to a fixed grid spacing of pins ina configuration of connectors peculiar toa specific project in process at the time.For other later projects, now in process.conversion units have been developedand furnished for automatic alignmentand insertion in the machine, utilizing thesame T cart technique for loading. Theconversion unit consists of a rigid pinconfiguration on one surface that is

arranged to mate with the basic array ofprimary spring pins in the A panel of theinterface machine.

The opposite surface contains a springcontact pin array to mate with the newUUT spacing (see Figs. 9 and 10).

Installation and removal of a conversionunit is accomplished by using a T cart.The conversion unit is driven underpower and automatically aligns with andengages the spring pins in the A panel.The operator locks the conversion unit inposition and actuates the drivemechanism to extract the empty T cart inthe normal manner as for a UUT. Thesame procedure is used to remove aconversion unit.

The machine is now ready to accept thenew UUT on its T cart, the same asdescribed previously. Again, no wiring isplugged in and out when converting todifferent UUT requirements. Reliabilityof making a changeover utilizing thesame philosophy approaches that of thebasic U UT testing system.

The interface machine has been providedwith a long-term stand-alone capabilityby virtue of its versatility to accom-modate different connection point arrayspacings. It is also evident that the DATEmachine presently wired into the inter-face can be replaced with differentautomatic testing equipments. Long-term utilization is, therefore, limited onlyby the creation of new types of "inter -interfaces" (conversion units) that can

Fig 8 - DITMCO Automatic Test Equipment and cabling.

adapt to the basic framework of themachine.

Results and conclusions

As of this writing, the machine has beenin use for a period of three years and hasrun an approximate total of 1095backplanes produced at the Camdenfacility. An average of 2.5 testings perpanel has been typical.

The 2.5 testings per panel is governed bythe yield factor on the wire -wrapmachines which is constantly improving.It is often necessary to check each level of

Fig. 9 - Conversion unit on T cart shows bowloading -unloading is similar to handling of a

backplane UUT.

wiring on the backplane beforeproceeding with the next level. Thiswould be particularly true on a newproject as a means of checking the wire -wrap programs and changes.

For the past 8 months 50 backplanetesting per week were run. In themachine's history, there have been a fewinstances where bent spring pins werecaused by distorted wire -wrap pins onbackplanes (see Fig. 2). These bent springpins are replaceable in less than 30seconds per connection point. There hasbeen an average of less than one suchinstance per month in the life of themachine.

Fig. 10 - Interface machine with conversion unit in place.

31

Robert N. Piscatelli', Government Communications andAutomated Systems Division. Burlington.Massachusetts. received the BS in Physics IronProvidence College in 1965, the MEd in SecondaryEducation from State College at Boston in 1966. and theMS in Electrical Engineering from Northeastern Un-iversity in 1968. He joined RCA in 1969 and since thenhas performed as a Systems Programmer researching,designing, and implementing methods for improving theefficiency of the MIS Computation Center as well asproviding technical assistance to the programming andoperation staff with respect to the computer and itssoftware. He also worked on development of a similarprogram for the generation of integrated circuit andhybrid array masks.

'Since writing this article. Mr. Piscatelli has left RCA.

Programming for semi-automatic wiring equipmentR. Piscatelli S. Patrakis

Semi -automatic wiring with its relatively low-cost, highly flexible, quick -reaction, aideasily changeable characteristics has opened new dimensions in the field of backplanepackaging. Backpanel packaging left to the individual engineer's imagination can presentwiring problems for which there is no simple answer. The availability and use of a well -planned, disciplined computer program will allow the engineer to realize all the benefits ofsemi -automatic wiring.

Stanley P. Patrakis, Mgr., Design Engineering, ProductsEngineering. Government Communications andAutomated Systems Divisors. Burlington,Massachusetts. received the BS in Mechanical Engineer-ing from Northeastern University and spent two years inthe Army as a member of the Scientific and ProfessionalPersonnel program participating in the Army's SpecialWeapons Program. He joined RCA in 1958 and worked asa mechanical design engineer on the ASTRA and F-108Program. He provided the engineering -manufacturingliaison and configuration control on the criticallyscheduled APCHE and Mobile APCHE program (ATLASmissile automatic checkout equipment). Mr. Patrakiswas responsible for the electronic packaging design ofthe Variable Instruction Computer. He was responsiblefor the electronic packaging design of the RCA -builtcomponents on the ADA system and for the aircraftinstallation design of the complete ADA system. He wasactive in developing electronic packaging concepts forairborne equipment using medium -scale integrated cir-cuits and in developing the electronic packaging con-cepts for GCASD's 200 Series Computer. More recently.Mr. Patrakis supervised the Materials ApplicationLaboratory and the Mechanical Design Standards activi-ty.

COMPLEX ELECTRONIC SYS-TEMS, primarily but not exclusivelydigital, require a variety of wire sizes andtypes (single conductor, twisted pair,triplets, etc.) to be placed down in a point-to-point manner. The characteristics ofMS1% LSI circuits and the competitiveelectronic market require highly flexible,quick reaction, easily changeable wiringtechniques. To meet these needs, semi-automatic wiring machines are used.These machines combine the bestproperties of an automatically directedmachine with the versatility of anoperator.

The advantages are readily apparent.Semi -automatic wiring machines weredeveloped to fill the gap between fullyautomatic wiring methods, and manualoperations. The machines are in fact postlocators that remove this task from theoperator and reduce the interconnectionerrors. The size of panels or boards to bewired is not a problem since the machine'sdesign usually allows panel shifting ordual -head wiring. The various postspacings found in connectors can also behandled. The movement of the machinehead does not restrict any grid patternsfrom being wired. Twisted -pair wires canalso be terminated on semi -automaticwiring machines by connecting the endsof paired wires in succession.

Semi -automatic wiring does have somedisadvantages. The production rate issolely dependent on the operator.Operator efficiency, attitude, speed, andperformance will have a direct effect on

Reprint RE -2t-3-8Final manuscript received August 15. 1975.

32

the output. Some machines require theoperator to route wires during the wiringoperation which could result in excessivebuildup and inconsistent wire routing.

WWRAP programing

At Burlington Operations of Govern-ment Communications and AutomatedSystems Division a computer program-ming system called WWRAP has beendeveloped which will generate both of theperforated tapes to drive the in -plantnumerically controlled wiring machines,provide the engineering informationnecessary to document the wiring, andtest the assembled panels. The typical jobuses signal names to denote each string ofwires. The actual connection pattern isdetermined by the computer program.The program is used for a variety ofequipment and is therefore general innature. Provisions are made to inputcritical dimensions and nomenclature re-quirements for the various types ofbackplanes.

WWRAP generates information re-quired for wire -wrapping assemblies onthe Hughes Simi -Automatic Wire Wrapmachine, available at Burlington. Theassemblies can be plug-in cards,backplanes, platters, or any otherassembly which will physically fit on thetable of the wire -wrap machine. In addi-tion to the numerical control (N j C) papertapes, the system produces the necessaryerror wire connection and wire routinglistings to insure proper assembly fabrica-tion.

Wire -wrap machines

Before detailing the WWRAP system, abrief description of the Hughes machineand its operation is necessary. Themachine is a tandem one consisting oftwo x -y coordinate tables for holding theassemblies, above which two numericallycontrolled wrapping heads move. Thisallows two identical assemblies to befabricated simultaneously. An operatingcycle starts with the reading of two x -tcoordinate pairs and a wire size codefrom a paper tape. This causes the wire -wrapping head to move to the first x -ycoordinate pair. In addition, theapplicable wire size code appears on theoperator console indicating from whichwire bin the operator is to take the wire.

OPTIONS

MODULE DEFINITIONS

TWISTED PAIR INFO.

W WRAP

DATA INTEGRITY INFO.

0ENGINEERING INFO.

MACHINE OPERATING INFO.

Fig. 1 - WWRAP system overview.

(The Hughes machine requires that all ofthe wire for the assembly be pre-cut, endstripped, and stored by size in wire binsbefore fabrication time). The operatorinserts the wire into the wire -wrappinghead, positions the head down onto thewire -wrap pin, and allows the headmechanism to wrap the wire onto the pin.The wire -wrapping head, using the se-cond x -y coordinate pair, moves to theposition of the other end of the wirewhere the operator wraps it on theindicated pin. This cycle is repeated untilall of the wires called for by the paper tapehave been wrapped.

Since the wires are manually placed intothe wire wrapping heads, wire shorterthan 1.75 in. cannot be handled. All wiressmaller than this limit are automaticallyrounded up with one exception. There areassemblies containing a significantnumber of wires which connect adjacentpins. These wires, termed 'bus' wires, arewrapped using a numerically controlledPratt -Whitney drill machine with a

custom -designed wire -wrap fixture. If anassembly contains bus wires, they areattached before the assembly is placed onthe Hughes machine. The use of bus wiresprovides for a neater and less congestedwire -wrapping package than one whichused the I.75 -in. wires in place of buswires.

WWRAP program

The WW RAP program provides theinformation necessary for the operationof both of the above machines. Fig. I

gives an overview of inputs and outputsof WW RAP. The system is controlled bythe user via the option parameters, shown

below:

All listings but no NiC tapes. NiC tapes only.Specific listings only. Input error listings only.All listings and N C tapes.Option of one or two output tapes. (When

this option specifies two output tapes. i.e..two output formats, tape format No. I willbe used for all wires connecting adjacentpins 0.100 mil spacing. However, thesewires must be included in the overall :-levelcomputations.)

Selection of bin No. I size and incrementalincrease.

Option to select order of wiring as long toshort or visa -versa.

These parameters allow a great deal offlexibility since they control all of theoutputs of the system. The user cansuppress unnecessary outputs when theyare not required, i.e., when inputing datafor an assembly for the first time, suppres-sion of the machine operating informa-tion is generally done since the validity ofthe data has not been established. Themodule -type definitions and placementinformation determine the geometric pinmatrix for a given assembly while theinterconnection and twisted -pair infor-mation determine which pins must bewired together.

The outputs of the WWRAP system canbe generalized into three areas, dataintegrity, engineering information, andmachine operating information. Thefollowing gives a detailed breakdown ofthe various outputs for each area.

Data integrity informationModule type listModule placement listInput net list

33

INPUT 'IT LISTI11D/F0401S0002 01, 0 Li SFIJC,X402 040 0 WON^

nor T 3 71,3 -P -----X402 043 0 114721107 EI-J1-053015- - " rit - XA12 ONO 0 NtO

0 ,TvI^F pit

5402 051 0 7VrigOrE6

x4A2 055 0 0-Cf FF-11E01 u'7 _.. _46.72 050 0 /04T Fr

SI:NAL OASF .. 786----TIM, TT- 1I4TANr0 2.1.2021.

5693-060 0403.060 1.75n I

xR,i3=060 TcA-03--1163 ---1;3ir-----2--x4n4-0P4 X403-060 7.25) 2

----x404.4,14 - iA137-e711 1.501 1

,- SPINAL NAME - 6OTTO-15-FIF-pow, 7u IIPTANcF 2 -LEVEL

- - - SA .11-1W6- " - AA -ZIT -1%i- - - - -I 7,117,- - --1-- ' - - -

S I niis L "RA ,fr -; -PCS I ii ai--------- ------row. To tteToire 7.1EVEL

06767:033 x403 -055 1

x -Fr-aso ir-Tr7frbe iinNxi: 4:174F:-it.s-ifeiF--- -----__ __.xAo: 071 0 AOLL FAO. TO ri!TAPei.__Ztklq_k_____----- iiii -Off -- 67,-Ewit --Tii't - - - -- - - ---TV6"--014 '-in-1590133 ---- - - 7:7`b, 1X402 078 0 GO'

--if-kill-010 -- 0 - i"0-i- - - - - - - -)(An? 001 0 1,401r r 1r. 3Xltn3 002 0 naJP C Gv0

A. INPUT NET LISTCFNIER NIXES

TY--- tet PIN rst

SIr.NAL 1A0iF - 0C417 .4"FRO. To u1fTANU Z.LtvEl__

7A-4;051 x606-63, 7.751 1

B. SIGNAL ROUTING LIST

N xv6a---777VFL Four SIGNAL

337 TO00 -32571- 16076- IT -11(Zief -ftr;--- -1- - -1 60- tire- OR4-02 325o 16079 051 7125 27 1/1.CO IN. 17.14 IN, 1 ..A009.002 X406 -061____041_72 _______3 PEI 1625 3500- -16875- -111..00 Ia. 17;00 iN XAO6Z-013- 400-0.001 111/1.44 25, 1375 3500 1%0(0 2, 1N ,CO TN. 17.30 OA

1 XAQ7-059 A009.015 D06.17 A70d -17673- 470 1210 77 14:60 IN. 17.02 IN. 1 60064021 5606-076 DP2466 375' 15075 750 1175 27 10.00 IN. 17.75 IN. 1 A009018 5407..081 084-4

- --7- 375. TOS-T - * -_85N- - TV:0- - Tv -Ir a 0" Pr..17;3V Tic.- - - r - -1 fiblv-1510- - -WRY6=675- - -60 I-4 - - -- -II 75 1750 3500 15875 27 18.00 IN. 17.3* IN. 1 56074071 6006-n17 084-2

- 9 75 1250- 2,5-M -13675--34- --17 a IT TPr. --is.-esi.To:- --r- ii01;61-9-- "ATicre-ki-2-2- --DON - - - - --10 375 15125 454 1375 26 17.00 IN. 16.15 IN. 1 4009-030 5606-077 082.4- ----rr--n0J 15305 730 1375 70. 17.00 IN. 16.75 IN. 1 A0084025 x607-077 DP312 325J 15010 PSI 10'..o 24 17.0 IN. 16.40 IN. 1 4009-032 5406-003 083.8

- - TT - -3 SOY ---17s754 - - 754--123-0 -TA- -I T.-E`O-TN. -1 e . s i TN. 1 *009Tile X006.063 DRI-114 05.. 2375 325, 14750 24 16.00 IN. 15.77 IN. 1 X006-061 A004-020 082-1

- -IS--- g5.-,- 20')0 - -130n- - 1512-5--n- -15.20-1-14;.1577rN; ----r- WA -56=067 - - -Ao-05zon---bIE-2-10 3,103 1512L_25.5.86,.1A0011..02, X407..067 OPZ-IT-- "TY0V--14-TsoolP-041 xe06-081 16113.4II 3501 150m0 851 1875 7! 16.00 1h 15.77 IN. 1 4009-031 5406-069 083-1_

- ---Tv -3237" -13.-5F73--.an--173-0---/r" -IF:CT TI4-1-5:13-1-4-.- -- r - -Ah-si---6 at- --xro-szYri- - -613--TPO 40v 60.0 11751 7125 24 16.00 II.. 15.07 IN. 1 X406.004 5004-016 051-01.0

--- f 3-- -75. *' -7773 -360N-107S- - TN- -16-X0- "I Ft; -15;2 i" TN: - - -r--kirrzon- - -A5 oil =n11- - -ols I- -___-

C. HUGHES WIRE LIST

Sparc pin listError lists

Engineering informationSignal routing listWire connection listSignal listTwisted pair vs return list

Machine operating infinmationHughes paper tapeHughes wire listBus Paper tapeBus wire listTwisted -pair paper tapeTwisted -pair wire listTotal wire list

The data integrity information is used tovalidate the data input to the system.Obvious errors, i.e., two signals con-nected to the same pin, are identified onthe error lists below, while conceptualerrors, i.e., an incorrect module place-ment, can only be identified by the userfrom the other lists.

Error listings

I. Duplicate input,2.0nly one pin in net,3. Pin already assigned to4. Pin not within wiring matrix, or5. Pin not defined.

*I lie engineering nil (inflation is used dur-

Fig.2--sampleoutpuiws.

ing the test and debug cycle for theassembly as well as for final documenta-tion after design acceptance.

The machine operating information isused by the personnel responsible for thefabrication of the assembly. It consists ofthe numerical control paper tapes used onthe wire -wrap machines and the variouslistings required by the machineoperators. Typical output lists for eacharea can be seen in Fig. 2.

WWRAP program functions

[he W WRAP program performs fourmajor functions (see Fig. 3). The first, asstated earlier, is definition of thegeometric pin matrix from module -typeand placement information. For theWWRAP system a module can be anycollection of wire -wrap pins. The userdetermines the module types for a

particular assembly based on its physicallayout. If the assembly is a plug-in cardon which integrated cirucit modules aremounted, the 14- or 16- pin IC socket pinconfiguration should be defined as amodule type. If an assemlby is a

backplane, the card mounting socketsshould be given a module type. When

classifying a group of pins as a moduletype, one pin is chosen as the referencepin for the module and the orthogonaldistances of all the other pins from thisreference are given. In addition each pinis given a name which is unique within themodule. This method of module typedefinition allows for not only standardcomponent mounting but also for specialcomponents or pin configurations (see

DETERMINE ASSEMBLYGEOMETRY

VALIDATE INTERCONNECTAND TWISTED PAIR INPUT

GENERAT ROUTINGPATHS FOR NETS

GENERAT MACHINEOPERATING INFORMATION

Fig. 3 - Major system functions.

34

1

2

16

10015o -1-

3 14

4 0 13

5 0 12

6 0 10

7 o 11

VCC 8 o 9 GND

A) 16 PIN IC SOCKET

04

0,

03

81 IRREGULAR COMPONENT

Fig 4 - Typical module types.

A) PIN PAIR DEFINITION

11) PASS I

CI PASS 2

DI PASS 3

A0

A0

D

1

C

C

Fig. 5 - Minimum spanning tree function.

Fig. 4). Once a set of module types arechosen for an assembly, the pin geometrycan be fixed by the placement of thesetypes. The user accomplished this bygiving a unique name to a module, givingthe x -y location of its reference pin, andgiving its type. The WWRAP system nowconstructs a pin matrix for the modules.Each pin is identified via its module namefrom the placement information and itspin name from the type information. Atthis time, module conflict errors aredetected and reported.

After the geometric pin matrix is es-tablished, the processing of the inter-connections between these pins is done. Itis necessary to differentiate between twokinds of connections - standard andtwisted pair. Standard interconnects aresingle wires connecting two pins whiletwisted -pair wires are two wires twistedtogether connecting two pairs of pins.For twisted pair wires one wire is calledthe signal and the second is termed thesignal return. When wrapping twisted -pair wires, the machine operator wrapsone end of the signal wire, then the signalreturn closest to that signal, followed bythe other end of the signal wire, andfinally the other signal return. It shouldalso be noted that twisted pair wires arewrapped after the standard wires havebeen placed on the assembly. For themost part standard and twisted pairinterconnects are identically processed.As the interconnections are read by thesystem, the mod -pin in the geometric pinmatrix is given the signal name of theinterconnection. Cross -connected netsand identified module or pin errors arereported. For twisted -pair wires thesignal return is also input and is identifiedin the geometric pin matrix along with thesignal name.

At the conclusion of the interconnectprocessing, the mod -pins for each signalare collected in preparation for the wire -routing phase. The signal or net routing isperformed on each net independent of theothers. For the net a minimum spanning'tree' is found. This is the shortest wirepath which will connect all of the pins inthe net together. In practice the 'tree' pathis not necessarily the minimum due tosystem constraints. The most significantconstraints are the fact that bus wirescould be used in the net and that amaximum of two wires per wire -wrap pinare all that are allowed. Thus theminimum spanning tree can be treated asa local rather than a global optimum. The

algorithim tor the tree follows.

First all possible pairs of wires betweenall of the mod -pins of the net are foundalong with their length (Fig. 5a). Thepaths are now selected starting with thesmallest, subject to the conditions that I)the path does not connect two pinsalready connected, and 2) there are nomore than two wires per pin (Figs. 5b, 5c,5d). It should be noted in Fig. 5 that wireBC is shorter than DC but could not beused since pin B already has two wires onit.

After the minimum spanning tree isfound for a net, the wire paths are definedand saved by the system. The wire -routing information is now generated.This consists of mod -pin pairs and theirlengths. If the wire -routing report is

requested, the information is outputed.The process continues until all of the netshave their routing determined. It shouldbe noted that routine information doesnot refer to the actual layout of the wireon the assembly but rather to thedetermination of which pin pairs in thenet are to be connected. The lengths of thepin -pair wires are the straight-line lengthwhich are not necessarily the actual wirelengths.

The final system function is the genera-tion of the machine operation informa-tion. From the minimum spanning treeinformation, an orthogonal length foreach wire is found. This is used as thelength of the wires to be placed on theassembly. Now the records required forpaper tape and report generation on theHughes and bus wire machines can beproduced if called for by the user. Forease of wire wrapping and minimumcongestion, the wires are sorted bylengths and wire -wrap level. Since eachpin can have two wires, one higher thanthe other, level sorting provides for twoseparate planes of wires. This produces aneater assembly which is easier to main-tain.

Conclusion

The W W RAP programming systemprovides control of the wiring fromdesign through fabrication and assembly.The programming system represents theapplication of a computer -aided designtechnique to improve the efficiency andperformartce of complex but basicallyroutine tasks required in the wiring ofelectronics equipment.

35

A concept of factory automationR. Walter

Currently. many manufacturing functions are automated through the use of computersand associated equipment. The support of administrative functions such as material andlabor control to regulate costs and measure results is aided by the use of medium to largecomputers at the plant site or through communications to a central computer site. Thecontrol of processes, such as testing and assembly, are improved through the use of miniand microprocessors as well as programmable controllers. The engineering problemsolving and design now involve the use of hand-held and desk top calculators as well astime-shared and remote processing on a central computer. All of these things have butonebasic goal: cost effective manufacturing. This article outlines one possible path foradvancement toward that goal, based on existing applications and computer systemscapabilities. As always, the final challenge is related in our ability to cost-effectively designand implement the processes which use the tools we have or can acquire.

C ONTROL of the manufacturingprocess can be compared, in many ways,with the organizational structure of suchan operation, i.e., the plant manager atthe top and the individual worker at thelowest level. From this viewpoint, aneffective automation support systemmust be capable of providing informationfor all elements of the manufacturingorganization and be available whereneeded. Therefore, one objective ofautomation support is to ensure thatcomputer power be widely available, andusable. For example, a test system, inaddition to applying inputs and measur-ing outputs in varying sequences, shouldprovide data for statistical summation,for engineering analysis, and for materialcontrol. Thus, the information collectedis useful to people serving in differentcapacities and can be used at different

Ronald G. Walter, Mgr Manufacturing and EngineeringSystems, Picture Tube Division. Edison. N.J.. receivedthe BSME from the University of Maryland in 1954. Heentered the RCA Engineering Trainee Program in 1954In 1955 he entered the US Navy as a radar andcommunications officer. He returned to RCA in 1958 as atest engineer in the Solid State Rating Laboratory wherehe established and was responsible for the environmen-tal test laboratory and the development of test equip-ment In 1961 he became Engineering Leader responsi-ble for all test systems and development. In 1962 he wasassigned Manager. Finished Goods Quality at Mountain-top in the Solid State Division. In 1965, he wastranetorred to Electronics Components M.I.S. as asystems analyst for technical and manufacturingsystems. Currently, he is Manager of Manufacturing andEngineering Systerils for the Picture Tube Division.

Reprint RE -21-3-12Final manuscript received August 18. 1975

levels in the organization for better un-derstanding and control. If this principleis carried throughout the manufacturingorganization, then a network of informa-tion is developed to support the manufac-turing organizational structure. In turn,automated assists for all major functionsare distributed among these operationsfor improved performance.

Existing applications

Minicomputers, microprocessors, andprogrammable controllers have beenused in many major process controlapplications; this issue highlights severalexamples.L In the Picture Tube Divi-sion, an Intel microprocessor has beendesigned into a critical television -picture -

tube alignment procedure to alleviateproblems resulting from manual ad-justments. Such classes of operationscould be considered dedicated or worker -level assists which make processesrepeatable, accurate, and fast.

Support of Engineering in the design,applications, and production areas hastaken some new forms. Time-shared andhand-held calculators have been in usefor some time with many applicationsfrom artwork design, process control,and problem solving leading the list ofapplications. Time-shared services havebeen widely available, accessible, andheavily used. Recently, however,programmable desktop calculators,which can also act as a remote station to alarge computer as an intelligent terminaland can interface with laboratory in-struments, have placed easilyprogrammed and economical capabilityinto the laboratory. This equipment hasreplaced certain time-sharedapplications, while providing readilyprogrammable instrumentation for usewith laboratory instruments.

Manufacturing, admisistrative, and con-trol operations-including financial,materials control, warehousing, purchas-ing, and quality control-are now usingsome of the capabilities mentioned aboveplus mini and general class computers.The bulk of these systems use the stan-dard business computer language(COBOL) and are typically batch(scheduled and serially processed)operations. Most of these are run on alocal computer or over communicationslines to a remote computer. In a smallernumber of cases, the users themselvesinteract directly with the computerthrough remote terminals to either largeor small (mini) computers to request datafrom a file or input data to a file.

All the above applications are in everydayuse but not necessarily at the same factorysite. Practically speaking, the need for allof these applications does not appear inthe same physical site, but systems in usecan often be applied At other sites withreduced development costs.

The benefits of computer aids tomanufacturing operations generally in-clude increased productivity, higherquality, faster turnaround of problems,faster machine throughput, and timelyinformation to management. Thetoughest problem is to apply our

36

resources successfully to accomplishthese tasks either through automated ormanual methods. Future automatedsystems enhancements are aimed towardsbetter use of our resources-capital andhuman-to solve our operatingproblems.

Future

In the near future, primary gains shouldbe in productivity through improvedsystems development and programmingtechniques as well as the concomitantgains that accrue to systems users. Thebulk of these gains will be throughsoftware improvements which accountfor 80% of 1975 systems costs. This pacewill continue over the next decade.'

Another future improvement will be

reliable and easier communicationsbetween humans and the computer andamong computers. This will lead totransfer of information to a wider com-munity. For example, dedicatedminicomputers regularly and easily ac-quire data from operations on the factoryfloor; this information can be passed to alarge computer for storage, retrieval, andanalysis. Production control personnelwould, in turn, be able to monitor factoryschedules in real time. The ordering ofmaterials and the status of inventory thencould be adjusted to correspond withplant needs.

The near future brings these capabilitiescloser to practicality because of reducedcosts. The following material describes afew of the major advancements whichshould impact the future of automatedoperations.

Large computers

Large computers with remote accessingcapability, cheaper hardware, and bettersoftware will be more widely available tousers. The processor hardware will befaster and cost less per unit of storage.Also, mechanical peripheral devices willbegin to disappear; thus there will befewer breakdowns, faster access of data,and lower dollar rates per unit of time.High level software, i.e., fewerprogramming steps to perform an equaltask, will become more widely used. Inthe case of mathematical programming,software systems like APL, which mightbe considered to be a shorthand notation,

will reduce programming costs for thesolution of complex mathematicalproblems. For the end user, simple in-quiry programming for the retrieval andmanipulation of stored data will also beavailable for remote use. For example, itwould be possible for someone in Per-sonnel to say literally "SELECTENGINEERS EQUAL TO AGE.> 25 or< 34" resulting in a composed listpresented on a video screen; no need forprogrammer intervention. The filemanagement of data stored for manufac-turing operations is another area wheresoftware support is improving. Themanagement of files by systems andprogramming people will be enhancedthrough the use of data base managementsoftware systems such as I DMS, I MS andTOTAL, which have been developedspecifically to simplify retrieval of datafrom files and reduce the complexities of

PLANT

Program developmentData base inquiryRemote operationsData inputProgrammable calculators

maintaining files. The net effect of theseadvances is that within the next two tothree years, there will be widespreadimproved turnaround of information tothe user, and a reduction in cost.

Communications

Communications between humans andcomputers have been primarily time-shared. To date, the cost of reliable andaccurate transmission from remote areashas been prohibitive with any significantvolume of data. As remote capabilities

LARGE COMPUTER

Information on:Product structureStandard costInventorySchedulesMaterial planningPayrollPersonnelGeneral ledgerAccounts payable

I9

PROCESS CONTROL/FACTORY FLOOR

Perform:Process monitoringDirest digital controlAnalysisNumerical control

Measure:SpeedTemperatureVoltageCurrentAirflow

Report:DefectsProduct inventoryLaborMachine status

continue to be enhanced, through bothminicomptuers and general-purposecomputers, greater communciationvolumes can be handled; this in turn, willbreed better communication accuracyand reliability.

Other remote capabilities, such as theability to initiate programs which are nottime shared, i.e., the entry of remoteprocessing jobs, will expand significantly.At remote sites people with no moreknowledge than is required to operate atime-shared terminal will be able to in-itiate jobs at will. Normally, these jobs arenot suitable for terminal operationbecause the volume of input data and thevolume of output data (printing) are toohigh.

A communication network inside theplant will also become more actively used.Many times dedicated minicomputerscollect data and control machines on thefactory floor leaving no capacity for otheruses of acquired data. However, enoughtime can be "stolen" from the work periodto transmit data for processing on othermachines for other than process -controlpurposes. Using communications for thispurpose will reduce the manual effort,hence the cost for identifying, collecting,and transporting such data.

Documentation

The practical aspects of managing thewritten word in a technical business are atbest a difficult, relatively expensiveproposition. There are many cases wherethe paper version of reports,specifications, and instructions cannotkeep up with the activities they support.Because the creation, updating, and com-position of text on paper is mainly amanual operation, machine assists

become a real possibility not only forbetter productivity, but for a betterproduct. In the case of composedtechnical publications, such as manualsand technical data sheets, use ofcomputer -aided file management,editing, and exposed -film output hasreduced turnaround times and cost ofpreparation up to 5:1 per page of finalpublication. Also, in the realm of thestandard typewritten word, equipmentspecially designed to input, edit, and printis in use for medium to large typing jobs.These so-called word processors enhancetypists' productivity by reducing editingand repetitive effort. This capability,

coupled to microfilm or microfiche viamagnetic tape, and paperless documenta-tion, looks practical. The net effect ofthese systems will be adequate, timely,and reasonably priced documents alongwith their maintenance.

Mini/micro computers

The range of electronic calculatingdevices runs from fixed programmeddevices such as hardwired controllers andpre-programmed calculators to largeprogrammable processors. Between theseextremes, microprocessors and minicom-puters have significant impact onautomation. These two types of machinescan effectively perform similar overlap-ping functions. Yet, there are applicationswhere microprocessors are much moreinexpenseive to use than minicomputersand applications where the minicomputercan effectively perform tasks where themicrocomputer cannot. In any case, thereare a wide variety of applications, rangingfrom machine control with a

microprocessor imbedded in the equip-ment to minicomputers acting as businesscomputers or process control systems.Both categories of processor hardwarewill continue to decrease in cost. Themain expense associated with their usewill be in the peripherals - magnetictape, disks, terminals, interface hardware- and programming.

Microprocessors will fall into twocategories based on user capabilities. Aclass of microprocessors whose end useris skilled in programming (such as equip-ment manufacturers) would use devicessuch as the RCA COS M AC for whichassembly (low level) software support isprovided. On the other hand, peoplewhose prime skill is not programmingwould use devices such as the INTEL8080 which use high level (less

detailed/ somewhat less efficient)programming. These microprocessordevelopments could affect manufacturingin several ways. Equipment we purchase,such as programmable calculators orremote computer terminals, will providemore pre-programmed and program-mable capability to a variety of users,including engineers, data processing, andbusiness personnel. Pre-programmedfunctions will become more widespread,and simpler programming capabilitieswill permit more non -skilledprogrammers to service themselves formore of their needs.

Minicomputers - having evolved to ahigh degree of sophistication in hardwareand software - will play an increasingrole in process control and dedicatedbusiness operations. Under a distributedcomputer concept, the application ofthese machines will permit us to placeindependent computer capability at thesite required. These machines will controlprocesses and provide support foradministrative functions such as materialand inventory control, word processing,and financial control at the normal placeof work. An independent machine is notat the mercy of a single processor at acenter site used for other purposes, whereconflict of interest could cause delaysbecause of system load. Also, outage ofthe central computer will not putoperations at the remote site out ofbusiness.

Minicomputers will continue the trendtowards lower hardware costs and moreeffective software. Minicomputer ven-dors now are providing the COBOLlanguage on their machines, which iscommonly used to support businesssystems. The machines will also be usedfor communications switching and assupport systems for large computers.°

Conclusion

Up until now, the cost of hardware andsoftware has been too high for all butsome compelling manufacturingapplications. In the next one to threeyears, there will be substantial reductionin the cost of both human and machineresources to implement computer -supported operations.

In turn, the effectiveness of our manufac-turing activities, through more efficientuse of our manufacturing people andfacilities, will increase. Lower cost andhigher quality product will result, therebyadding to the probability that our role inindustry will still be dominant.

References

I. Miller. J.L.; -Programmable controllers, minicomputers,and microcomputers in manufacturing," this issue.

2. Dillon, L.; Van Woeart, W.; and Shambelan, R.; -Productcontrol system for semiconductor wafer processing,"to bepublished.

3. Five Year Technology Forecast. Part II. AuerbachPublishers, Inc. (1974).

4. Frankel, H.D. and Martin, R.R.; -Electronic Disks in the1980's" Computer. IEEE Computer Society, Vol. 8, No. 2,(Feb 1975), pp. 24-30.

38

W HY PROJECT DART? The pro-jected five-year plan (1973-1978)demanded that the diffusion area increaseits capability 58%. However, the presentrooms were jammed with equipmentlocated haphazardly-a malady incurredthroughout the years of acceleratedgrowth. Since wafer processing evolvedas a batch process, no conscious thoughthad been given to the possible benefits ofproduction -line manufacturing.Preliminary study showed that somewafer types traveled between two andthree miles during wafer fabrication. Thiswas a profit leak that had to be plugged.

Furthermore, particulate matter hadalways been a problem at Mountaintop;particulates contaminate the surface ofwafers and, as a result, lower diffusionyields. When the existing diffusion roomswere constructed in 1966, it was felt thatremovable ceiling -high partitions weresufficient to avoid cross contaminationbetween the n- and p -diffusionoperations. These rooms were con-structed without drop ceilings and with

DARTopen heart surgeryon a living factoryW.R. Guerin, Jr. M.J. Martin T.J. RuskeyT.B. Lyden G.D. McLaughlin S.J. Blaskowski

In August 1973. there was clear evidence that the Solid State Divisionpower business was going to expand to the point where existingmanufacturing facilities would not be adequate. The diffusion area, inparticular. limited the factory capability, and was in dire need ofimprovement. Thus, a plan was formulated-the Diffusion AreaRearrangement Task, or DART-to rearrange or replace every piece ofequipment in the existing diffusion area of the Mountaintop factory.This task was to be accomplished within the confines of the existing

diffusion area and without production interruption. The risk ofundertaking the DART program was awesome, but its suc-

cess represented a projected cost savings of S1 millionper year. This describes the planning, design, and

execution which culminated in the success ofproject DART.

all the service piping and exhaust ductsexposed, thus contributing to theparticulate problem. In addition, theunderside of the roof, which is insulatingboard, caused particulates to be shed intothe diffusion rooms whenever the roofloading was changed by snow, rain, or bythe movement of people and equipmentacross the roof. Typical dust counts asdetermined by Quality Assurance, were70,000 particles per cubic foot of 0.5micron size and smaller material, and 500particles per cubic foot of 5.0 micron sizeand larger.

The photoresist rooms were designedwith similar wall partitions, but did havedrop ceilings to aid in the control oftemperature and humidity. However,each time Plant Engineering was called tomake an emergency repair or a newinstallation, it was necessary to removethe ceiling tiles to gain access to thepiping. This caused the entire room toexceed humidity specifications, and madeit possible for particulates to enter the

room from above. In summation, whenrepairs had to be made during productionhours, the photoresist rooms operated inan atypical manner or not at all.

Obviously some drastic changes wererequired if the five-year plan were to berealized. What an opportunity! To con-ceive an equipment layout to optimizeproduct flow, environment,superviseability, floor space re-quirements, and equipment design allwith the factory in full production.Several alternatives were considered:

I )0btain additional floor space near theexisting Diffusion Area and install therequired equipment. This would only curethe capacity problem.

2) Acquire a new building and move andexpand the entire area. This would consumetwo years (an eternity in our business) andexpend precious capital.

3) Rearrange and replace the equipment withinthe existing area-DART.

Reprint RE -21-3-13Final manuscript received August 7, 1975

39

The third alternative, DART, was clearlythe practical way to accommodateadditional capacity within a reasonabletime and treat the deformity of theexisting area. We had sold ourselves, nowwe had to sell management. We made thebold commitment.

I) Production capabilit could he increased by58% in the same floor space.

2) The job could be accomplished in twelvemonths.

We made our proposal and managementbet the entire output of the factory andover a million dollars in capital that wecould do it!

The objectives

An eight -man task force was organizedand a detailed set of objectives wasformulated.

I) Increase production capability 58% in thesame floor space.

2) Locate 70% of the "new area" within the

Walter R. Guerin, Jr., Mgr., Project DART, and Ldr.,Technical Staff, Mechanical Equipment Technology.Solid State Division. Mountaintop. Pa.. attended NewarkCollege of Engineering from 1951 to 1955 and joinedRCA Solid State. Somerville in 1963 as an equipmentdesigner. Prior to his transfer to Mountaintop in 1972, Mr.Guerin was engaged in the design of wafer fabricationand assembly equipment-specifically, equipment forchemical vapor deposition, evaporators for metalliza-tion, and chemical processing stations. He collaboratedin the design of the first cartridge "assembly system" inSomerville and acted as project coordinator for installingand debugging this equipment. Since 1973, he hasserved in his present position in Mountaintop and has theresponsibility for all water fabrication equipment used insolid state power -device manufacturing.

confines of the present area. Approximate-ly 7000 square feet of empty space wasrequired to start the moves. This spacewould be returned to the factory at thecompletion of DART, but in a differentlocation.

3) Provide a clean environment for waferfabrication.

4) Design an area which would allow fororderly future expansion.

5) Design new chemical processing stationswith maximum flexibility.

6) Pursue, to the fullest, process standardiza-tion.

7) Pursue production -line philosophy tooptimize product flow, decrease spacerequirements, increase operator efficiencyand improve supervisory control.

8) Provide a safe working area (to O.S.H.A.standards) for employees, and one thatwould motivate through attractiveness.

9) Improve the effectiveness of the existing airconditioning system.

10)Complete the project with no productioninterruption.

Open heart surgery

It was essential that a master plan bedeveloped to negotiate the requiredrearrangement with minimum in-convenience to the operating factory.

Milton J. Martin, Plant Services, Plant Engineering. SolidState Division, Mountaintop, Pa., received the BSMEfrom Newark College of Engineering in 1961. Prior tojoining RCA in August 1966, he held various positions inplant engineering, building construction, and fieldservice engineering. One of his active projects at thepresent time includes the design and installation of abulk chemical storage facility at Mountaintop. Some ofhis completed projects include the refurbishing andpreparing for occupancy of two buildings in the in-dustrial park along with design and installation of airconditioning and exhaust systems.

This plan detailed and sequenced theprojected start and finish dates for everyelement of construction, new equipmentinstallation, equipment debugging,process testing, and movement ofoperational equipment and departmentalinventory.

The plan

The plan called for dividing the projectconstruction into five phases. Phase Iprovided for the simo (simultaneous)diffusion room, a portion of the boronroom and about one half the photoresistroom. Phase 11 governed completion ofthe boron room while Phase II detailedthe construction of one half of thephosphorous room. Phase IV providedfor completion of the phosphorous roomand Phase V the photoresist room and theauxiliary room. With the exception ofPhase 1, all construction took place in anarea vacated by the completion of the

Thomas J. Rusk's, Project Coordinator, DART, SolidState Division, Mountaintop. Pa., graduated from PennState University in 1961 and has attended RutgersUniversity and Wilkes College. Mr. Ruskey's early ex-perience with Bell Telephone Laboratories, from 1961through 1965. was primarily in the design of telephoneapparatus. In 1965. he joined the Bendix Corporationwhere he became involved in the packaging and designof aircraft instrumentation systems. In 1966, he joinedthe RCA Solid State Division at Mountaintop where hehas served in various capacities. His initial assignmentwas in the Equipment Technology Group where hedesigned and developed numerous automatic and semi-automatic chemical processing and assembly machines.From 1969 to 1972, he served as Facility Planner for theLinear Power Transistor organization where he wasresponsible for preparing and implementing capitalequipment plans. From 1972 to his appointment asProject Coordinator, he has been engaged in thesupervision of production departments and the in -process quality control organization.

40

previous phase; i.e., in Phase I, construc-tion was completed and equipment in-stalled from the Phase -11 area; construc-tion relative to Phase II was then com-pleted and the equipment moved into thearea from the Phase -Ill area, etc. Newequipment was delivered, installed, anddebugged before moving the equipmentwhich was operational, such as silanesystems, diffusion furnaces, etc. Alloperating equipment was moved duringthe second shift on Friday and wasoperating again during the first shift onMonday. Fig. la shows the diffusion arealayout at the beginning of DART, Fig. I bdelineates the active areas for the phasesof the program, and Fig. Ic shows thediffusion area at the completion ofDART.

The schedule for equipment delivery,which was adopted at the outset of theproject, required the selection of anequipment vendor with a minimum ofdelay. To inform all potential suppliers as

Thomas B. Lydon. Member. Technical Staff, Solid StateDivision. Mourtaintop, Pa.. received the BSME fromYoungstown University in 1961. Since joining RCA inAugust. 1973. he has spent nearly all his time designing,inspecting, installing. and debugging the wafer process-ing equipment related to the "DART" project. From 1961to 1965 he was a Test Engineer for the Mobile HydraulicsDepartment of Commercial Shearing and Stamping Co.From 1965 to 1969 he was a Value Analysis Engineer anda Senior Design Engineer responsible for developmentof precision custody -transfer turbine flow meters withA.O. Smith. Meter Systems Division. From 1969 to 1971he was a Project Engineer involved in book -bindingequipment design for the Sheridan Corporation. Divisionof Harris Intertype. Prior to joining RCA. he was ChiefEngineer of L.A. Fish Engineering Company responsiblefor contract machine -design projects, and was in-troduced to the electronics business as a DesignEngineer, with Plessey, Mechanization Division. He .s aprofessional engineer in Pennsylvania

11411111111111%Nie

quickly as possible of the magnitude,importance, and timing of the project, avendor conference was called at Moun-taintop to which all potential vendorswere invited. At this meeting,specifications of typical chemical stationswere presented for review, discussion,and quotation. This day -long vendorconference provided the opportunity togive all the interested bidders the same setof information and permitted them to askquestions as the specifications werepresented. The vendor conference savedvaluable time.

On the basis of the quotations received,some bidders were eliminated; a visit wasmade to the remaining potential suppliersto evaluate their capability, capacity, andexperience in the laminar -flow aspect ofthe electronics business.

In an effort to assure that the bid priceswould be an accurate indication of the

Glenn D. McLaughlin, Buyer, Equipment and Com-ponents, Solid State Division, Mountaintop. Pa. receivedthe BA in Business Administration from Eastern Ken-tucky University in 1973. Prior to completing his degreerequirements he served in the U.S. Air Force's AirTraining Council and Strategic Air Council where he wasschooled in electronics. During the period from late 1968through 1970 he remained at the Air Force ElectronicTraining Center and served as an instructor in solid stateelectronic theory. Upon receiving his business degree in1973 he joined RCA's materials training program, andlater the same year became a buyer at the Solid StateDivision. MountaiMop facility. Mr. McLaugh in initiallywas responsible for purchasing capitalized equipmentand related eqi.iipment items. During the DART project,he shared with the project director the responsibility forthe selection of vendors to supply the clean -roomequipment and the selection of the contractor for theroom construction His current assignment is to supportthe factory raw material requirements by purchasing theglass -to -metal sealed transistor bases for the variouspower devices manufactured at Mountaintop.

ultimate project cost, a more detailed setof modular specifications were drawn upand submitted to the remaining bidders.An advantage of this modular conceptwas in the resultant pricing structure.Because each unit was priced separately,it was possible to order all stations duringthe project from one quote and therebyeliminate a time-consuming and detailedquotation procedure when additionalstations were ordered.

The final vendor selection was basedupon the factors of cost, delivery, quality,proximity of the vendor to the plant,capability and experience.

Phase I

On Monday, May 13, 1974, Phase 1began. Eight weeks were scheduled for itscompletion. A plastic cocoon was builtaround the area designated as the Phase1area, much as a doctor covers the patientwith sheets prior to an operation. The

first incision was a three -foot -wide cutinto the existing concrete floor for a newtrench for acid -drain piping. Un-fortunately, the plastic cocoon did not

Sy Blaskowskl, Designer Draftsman, Solid State Divi-sion. Mountaintop. Pa.. graduated Penn State University(evening division) with an Associate Degree in Engineer-ing. Early experience with Vulcan Iron Works. Wilkes-Barre. Pa included design of cement kilns and sugarprocessing equipment from 1956 to 1959. In 1960, hejoined A.0 F. Industries. Berwick. Pa. in design of missileteam for guided missiles. In 1961. he ;oined FosterWheeler Corporation, Mountaintop. Pa. where hedesigned f xtunng and special tooling for processing ofsteam generators and boilers. He joined RCA EquipmentTechnology group in 1965 where he has been responsi-ble for the design of equipment in the Wafer ProcessingArea.

41

GOLDDiFFOFFICES

jr

CHEM ETCH &PHOTO RESIST REMOVE

r1 r`i

PHOTO RESIST PHOTO ALIGNAPPLICATION & RESIST , &

ALIGN & E xPo..,t APPLICATION EXPOSI

-1 11

Fig. la - Diffusion area before DART-DART now occupies the crosshatched area.

SIMOROOM

PHOTO RESISTROOM

PHASE 1 PHASE I

BORON PHASE V PHOSPHORUSROOM ROOM

PHASE II PHASE IN

AuxiLIARY R(10,1

Fig. lb - The active areas for each program phase.

=.ICE AREAL T1IJ

=I=I

IMMINIME.1.11

II I I

WORK AREA 1TyP

1 1 1 IIiti1

BORON

PT WORK PASS THROUGHS = SUPERVISOR STATION

SERVICE CHASE

ItI IIII

I 111

1 1 11111 III,vrr I 1 1 1 I 1 1 1

511 WORK I 1 vP

vri I I I I I III1 1 1 1 1 1 1 1

PHOTO RESIST

I i11Ico

11

i:4

NOMMEN=

11II 1

II111.11 III 11 )74

1 III co4_1

1 it t

OlI 1 III.111 1

lIII U

II ET,11111111

Fig. lc - Diffusion area al completion of DART.

contain the concrete dust being generatedby the jack hammers and an additionaltent had to be erected directly over thetrench. Upon completion of the newtrench, the floor tilers arrived and beganinstalling the new floor. However, onlytwo days after flooring had begun, theBrotherhood of Floor Tile Workers wenton strike. Fortunately, the strike lastedonly six days, but we were behindschedule already. By working overtime.for several weeks, Phase I was quicklyback on schedule.

Keeping the project on schedule wasimperative since a major air-conditioningrearrangement was required for the newphotoresist room, and the July vacationshutdown was the only time that it couldbe accomplished.

After the floor was complete, theperimeter walls and ceiling steelworkwere erected, thus separating the Phase (area from the remainder of the factory.

Room construction

As part of the DART proposal, PlantEngineering seized the opportunity toprovide the type of room constructionthat approximates a clean room, isaccessible for maintenance, but yetutilizes the existing air conditioning,main exhaust ducts, and main overheadservice piping (see Fig. 2)

For serviceability, the proposal includeda structural steel (Unistrut) frame with a3 4 -in. fire -retardant plywood roof. Theunderside of the roof is lined with analuminized Mylar vapor barrier to pre-vent moisture migration. This "walk-on"deck provides accessibility to theoverhead area of the diffusion area forservice while preventing particulates fromentering the room from the dust -generating ducts, pipes, and the in-sulating boards previously used in thefactory ceiling.

The walls of the diffusion area are double -SUPT wall constructed using steel -faced gyp -OFFICE: sum board which has a baked enamel

finish on the exposed side. These wallpanels are snapped onto steel studs bymeans of an aluminum filler strip whichserves as a fastener as well as a decorativestripe. This non -progressive design makesit possible to remove panels in the centerof a wall without starting in a corner.

42

A suspended ceiling was attached fromthe steelwork supporting the plywoodroof. This ceiling is a drop -in metal -pantype which is located approximately 14in. below the plywood. This I4 -in. areaserves as an air -distribution plenum forsupplying air-conditioned filtered air tothe various rooms in the diffusion area.The ceiling tiles are readily removable,easily cleaned, and non -shedding. Someof the tiles are active to permit theexpulsion of air and others are inactiveand contain a sealed sound -absorbingbatting which prevents the expulsion ofair. This system permits an infinite ad-justment of air distribution to the areaand is readily adjustable for changes inheat loads within the operating areas.

The floor covering selected was a

homogeneous PVC tile approximatelytwo -feet square. The seams are beveledand heat welded, which results in a seam -free monolithic floor which is waterproofand chemical resistant.

Area lighting was improved from 50 fc to80 fc and cool white bulbs were replacedwith warm white bulbs. The new lightingsystem has vastly improved working con-ditions in these areas.

Because of the proximity of the equip-ment to the ceiling, a new type of flush -mounted sprinkler head was installedthroughout the DART area. This headnegated the possibility of sprinkler dis-charge as a result of mechanical damageduring the equipment moves.

With the exception of the drain lines, allservice piping, electrical conduit, andexhaust ducts enter the rooms fromoverhead through sealed service chaseslocated between the back-to-back equip-ment. These "service aisles" make repairsconvenient during working hours sincethe maintenance mechanics do not haveto disturb the operators in the productionaisles.

Like the pieces of a puzzle, everythingbegan to fall into place-electricalservices for lights, service piping.sprinkler systems, air-conditioning ducts.exhaust ducts, new laminar flowequipment-each segment went togetheras planned. Particular attention wasgiven to the new gas service lines to thearea since contamination is a deadlyenemy in diffusion. Dust counts wererecorded and analyzed periodically todetermine contamination both

The Incision-start of the trench used for the acid drainpiping.

UNISTRUTSTRUCTURALSTEEL

Trench prepared for the pouring of concrete

Underside of ceiling showing vapor barrier.

PIPE CHASE

COLUMN

ROOF

WALK ON PLYWOOD DECK

VAPOR BARRIER

PLENUMCEILING

PERFORATEDCEILING TILE

OUTSIDE WALL

Fig. 2 - Typical room cross section showing construction.

43

quantitatively and qualitatively. Allopen lines were covered with plastic eachevening and purged with nitrogen whencompleted.

The construction of the furnace room wascompleted within two weeks. By con-solidating the furnaces, it was possible toincorporate a temperature scanner tomeasure and to printout furnace -tubetemperatures, thus saving hours of profil-ing by operators. It made the heatrecovery system (described later) possi-ble, and avoided adding the furnace heatto the operating portion of the factory,thus reducing the air-conditioning load.It decreased the safety hazard of havingfurnace pullers extending haphazardlyinto work aisles, as was previously thecase, by positioning them so that anoperator always approaches them headon. Since the furnaces were purchasedthroughout the past ten years, they con-sisted of several breeds and many vin-tages. To be able to "hide them" behind awall also made for a much improvedappearance.

It was obvious at this point that, at least,the furnace wall would look impressive.All service piping to the furnaces wasterminated in such a way as to minimizethe time to reconnect thefurnaces in theirnew location. Now the eight weeks ofPhase I were over, and all systems were"go". The last major question was soon tobe answered. Can we move an entiredepartment over a weekend and havethem operating properly by MondayA. M .?

This question had been answered in theaffirmative, to some extent, by the carefulplanning of equipment design and layout.

Equipment design

In the design of the chemical processingstations for Project DART, the challengefacing the Equipment Technology activi-ty was to blend equipment standardiza-tion and equipment flexibility. In addi-tion, it was imperative to achieve highthroughput, clean environment, chemicalconservation, ease of maintenance, andoperator safety.

The first step was to meet with theproduction department engineers andforemen to ascertain what processes werebeing used in the factory. A review of thisinformation yielded many opportunities

Furnace wall during construction.

Fig. 3 - Typical DART chemical processing station contains five modular processing units with coreapondingfront-mounted control panel.

to standardize these processes. This stan-dardization was to take many additionalmeetings and many hours of testing. Theresult was the modular design conceptshown in Fig. 3.

The modular design concept developedfor the chemical processing stationsmakes use of a basic laminar -flow struc-ture lined with acid -resistant plastic andcontaining a plastic drain/ exhaustplenum (bath tube) under the work sur-face. This structure was fitted with thework -top functions required by the

processes involved. Each work -top func-tion was designed using common -sizedwork top "modules" (equal width), whichresulted in module interchangeability,and flexibility. To accommodate thismodular concept, lot sizes of 2 -in., 2 1/4 -in., and 3 -in. wafers were reduced to acommon dimensional denominator andthe optimum processing tank sizedetermined. The result is that in a six-foot -long structure, five processingmodules and one centrifugal -dryermodule can be installed. The rinser -dryermodule by necessity is wider than the

44

processing modules, but this combinationof five and one makes up a standardstation.

Each module consists of a perforated,stainless -steel, Kynar-coated work top,and a processing tank with its associatedcontrols. These controls are mounted ona removable Kynar-coated, stainless -steelpanel mounted on the station front. Thesepanels are the same width as the worktops; therefore, work -top modules andelectrical control panels can be movedanywhere along the station.

The perforated work tops allow spillageto drain into the plenum, and the laminarflow air, laced with acid fumes, to passthrough to the exhaust ports located inthe plenum. When a particular processrequires something less than five process-ing modules, or does not require a rinser -dryer, a blank perforated work surfaceand electrical panel is substituted in itsplace.

This modular concept was of major valuein reducing the number of detailedengineering drawings required duringDART. A general equipment specifica-tion which covered both generalchemical -station structure and individualprocess modules was generated. Thisspecification, combined with a drawing ofthe station or module, made it possible toorder any chemical station much as onewould select dinner from a menu.

Equipment layout

The objectives of the equipment layoutplan are optimum product flow, stan-dardized wafer handling systems (jigs,fixtures, etc.), functional space utiliza-tion, operator efficiency, and operatorsafety. It was ascertained that 7000 squarefeet of factory space would be requiredadjacent to the diffusion area to beginconstruction. All new equipment wouldbe uniform in length for completeinterchangeability between product lines.The area was oriented in the factory insuch a manner that it could be expandedany time in the future. Pass-throughswere provided for work flow between thediffusion rooms (phosphorous andboron) and the centrally locatedphotoresist room to eliminate operatoregress from their respective work places.

Space allocations for each room wereapproximated according to the ground

rules to fit within 22,000 square feet (onehalf of a football field) and preliminaryequipment layouts were begun. Theproduct recipes had been prepared byManufacturing Engineering and wereused for these layouts. Meetings wereheld with the engineers and foreman ineach operating area, and the layout wasrevised on a weekly basis.

The contractors had removed their toolsto an assigned area, the clean-up crewshad gone over the new area with finetooth combs (would you believe mopsand vacuums?), the manufacturing peo-ple were packing their inventory andmiscellaneous equipment. The momentof truth was at hand. What a weekend?!!Plumbers, electricians, foremen,technicians, engineers, machine atten-dants, porters, designers-they were allthere doing what they do best. Manyhundreds of employees could be affectedif this turned into a "long weekend". ByMonday morning, everything was in acomplete state of confusion-notchaos-confusion . But the furnaces weremoved and at temperature by Mondaya.m. as anticipated! As Monday woreon, the heartbeat of the factory gotstronger and stronger. The area was alivewith the sounds of unfamiliar buzzers andalarms. By Tuesday morning, all the vitalsigns were back to normal.

Phase I was complete, and Phase II hadalready begun. The process engineers andformen now were fine tuning the newarea. The remaining construction phaseswere much the same as the first; however,each one had its individual crisis. such asa carpenters' strike during Phase II.Project DART was completed on MarchI, 1975. The operation was a success, andthe patient lives.

Sixth -month checkup

The health of the patient can only beconsidered improved if this improvementcan be measured, in this case, in dollarsand cents. The savings realized aresensitive to the production levels in thearea; however, below are estimatedsavings based on the June 1974 produc-tion levels.

Material savings -The savings which areattributable to increases in yields anddecreased breakages have been estimated tobe in excess of $300,000 annually.Labor savings-The savings directly relatedto increased labor rates and operator

efficiencies will approximate $250,000 an-nually.

Reduced chemical consumption --Thisbenefit will effect a savings of $250,000annually despite the fact that chemical costshave been soaring.Improved distribution of tested units-Improved distribution results in morepremium product to be sold at a higherprice.

Hea recover system-This system willsave approximately $20,000 annually.

But how are these savings made? Theanswer is through innovation. A few ofthe innovative methods that make DARTsuccessful are described below.

Particulate control

For particulate contamination control,all processing, loading, and unloadingstations are Class -100 (Federal Standard209A) vertical -laminar -flow stations, andthe furnace walls are protected by Class -100 laminar -flow bonnets. The filtered airfrom loading and unloading sections onthe furnace wall is recirculated back intothe room, whereas the air which enters thechemical processing stations is exhaustedfrom the rooms. Since the chemicalstations comprise about 25% of thenumber of stations, a typical air mass inthe room is H EPA filtered three timesprior to leaving the room. This factor hasmade a significant improvement in theparticulate levels in the operating areas.The dust counts in the wafer processingareas under the laminar flow hoods areless than one-third of the level allowed inthe class -I00 standard. Typical dustcounts in the working aisles are averaging300 particles per cubic foot of the 0.5micron (and smaller) size (down from70,000 particles per cubic foot) and 7particles per cubic foot of the 5 micron(and larger) size (down from 500 particlesper cubic foot).

Air conditioning

To reduce the air conditioning re-quirements in the diffusion area, all thediffusion furnaces were concentrated infurnace rooms, one phosphorous, oneboron, and one for simultaneous diffu-sion. Because of their isolation from theoperating portion of the factory, theserooms can be operated at an elevatedtemperature (90° F). The 145 furnacetubes are located in these rooms and areoperating at temperatures averaging1100'C. Prior to DART, some of theexpended heat was absorbed in water

45

AIROLTS1DF

THERMOSTAT

INo mail

em sDIFFUSION FURNACES ]

HOT WATER

MILb

XRAuST

CHILLED WATER COIL

C TO FURNACE ROOM

HEAT RECOVERY COIL

OUTSIDEAIR TOBUILDINGAIR CONDSYSTEMS

Fig. 4 - Heat recovery system. The outside air is introduced into the space at point A and mixed with return air from thefurnace room at point B. Mixing is accomplished by the positioning of the dampers, which are controlled by the spacethermostat. In summer, the outside air is modulated by passing it over chilled water coils; the coils are controlled byanother thermostat in the space. The air is then distributed to space via ductwork at point C. The air that is removed fromthe furnace room is exhausted or returned after the heat has been extracted at the water coil, point D. The heated water isthen circulated to the main building air-conditioning systems at point E, where it tempers (preheats) the outside air.

cooling coils which were piped to a drainand released to an outside stream. Itappeared prudent to utilize this wastedheat to temper the outside air used forexhaust make-up, as shown in Fig. 4.

Based on a recoverable heat load of 8 kWfrom each furnace, it is estimated that60,000 gallons of number -six fuel oil canbe saved in one year. At S0.30/ gallon, thissavings amounts to $20,000 a year in fuelcost.

Temperature-controlled etch systems

Several processes require the use ofvarious combinations of nitric, acetic,and hydrofluoric acids. The processes arevigorously exothermic when etchingsilicon, and the etch rates are extremelytemperature sensitive. For this reason, anacid cooling system was installed in aplastic, fume -control, safety cabinetlocated adjacent to the processing station.The processing tanks are equipped withdiffused bottom inlets and an overflowstandpipe. This overflow is piped to theacid sump in the cooling system.

A submersible centrifugal pump con-tinuously pumps the acid through a coilof plastic tubing submerged in a

controlled -temperature cooling -waterjacket surrounding the sump. Overflowsare provided to allow over -fills of etchantto run into the cooling -water jacket whichoverflows into the plant chemical drainline. The pump is constantly recirculating

cooled acid to the two process tankswhich are used alternately to iinprovecooling.

When the etchant needs replacement, it isdrained into the sump and aspirated tothe plant chemical drain using city water.The sump is then filled by dumpingetchant into the processing tanks, whichdrain by gravity into the sump.

These cooling systems have improvedproduct quality by virtue of theirrepeatable etch times, and have extendedthe life of the etchants two and one halftimes. Other process modules weredesigned for standard organic and in-organic cleanups, cold -water rinsing, hot-water rinsing, ambient -temperature aciddips, and several unique processes in-volving ultrasonics and heated tanks withreflux condensers.

Process control was emphasized in eachmodule and station layout. All timersused are digital for repeatable setting andall temperature controlled systems have afeedback sensor in the processing liquidto control temperatures.

All pumps and solenoid valves used in theprocessing modules are fitted with athree -wire pigtail and male plug. Thesevalves are plugged into mating recep-tacles mounted on the rear of the station,thus allowing a valve change to be madeby a plumber only. By supplying oneutility receptacle, which is not used in theprocess, valves can be checked for elec-

trical failure without removing them fromthe system.

Modules can be prefabricated and addedto an existing station in hours, andwithout the associated construction in theoperating area that is usually requiredwhen modifying a chemical station.

The uniformity of a given type of modulepermitted effective training to be given toplumbers, electricians, and machineattendants. All spare parts are stockedand standardized. This measure hasgreatly reduced the number of compo-nent varieties required.

All acid modules are designed withseparate drain lines that are run into acommon drain manifold in the serviceaisle. This enables these lines to besegregated in the future if acids can besold to chemical processers for recycling.This, along with lessening the load in theplant neutralization facility, may provecost effective. Also, with acid segregation,it may be cost effective to recycle the useddeionized water from these processstations. These matters bear further in-vestigation.

This modular construction in conjunc-tion with process standardization hasalready proven its value in two distinctways: changes in process sequence can behandled during lunch hours (i.e., throughthe rearrangement of modules within astation), and operator retraining is

simplified when curtailed or increasedproduction causes operator reassign-ment.

Although, perhaps, the major savings arethose listed above and counted in dollars,many other savings and benefits are beingrealized from DART. These are theintangible benefits, in morale for exam-ple, which, while incalculable, affect thetotal operation, and which are testimonyto the communication, detailed investiga-tion, and cooperation shown by all whowere involved. The attitude and com-ments of the people charged with opera-tion of the diffusion area perhaps bestreflects the merits Of the program:

"The impact of Project DART on the powertransistor manufacturing capabilities is inone word-GIGANTIC". Wafer processingcapacity of my product line has beentripled. Recent dust counts have shownthat the rooms approach Class 100.

Because of this condition and better

46

process control, Epitaxial processing lineyields have improved significantly.

The ability to standardize on processingwith quick feedback of data are big pluses.The labor content has been greatlyreduced-by better than $.50 a wafer. TheDART program allowed for reorganizationleading to reduced manpower and im-proved efficiency."

L.V. ZampettiManager

Power Transistor Manufacturing

"The standardization, flexibility, and cleanconditions offered by the new chemicalcleaning stations and clean room havegiven me much greater control of theproduction process. We realized a

dramatic yield increase when we startedprocessing in the new area."

C. JobstProcess Engineer

"The design of the chemical cleaningstations is the single biggest benefit ofDART. It allows for pre -diffusion chemicalcleaning standardization across allproduct lines with increased throughput."

E. VancavageProduction Superintendent

"The new diffusion area has created aprofessional working atmosphere; safe,clean, orderly, in -line processing, betterproduct flow, and reduced cycle times.

The design and layout of the chemicalcleaning stations have had a tremendousimpact. Increased volume capability withreduced labor are a reality. The mostimportant accomplishment, however, hasbeen the ability to standardize thechemical cleaning processes. We nowprocess more than 80 types across fiveproduct lines in the same shared facilties.The results are improved process controland lower labor and chemical costs."

W. CubaBoron Room

Production Foreman

"The advantages of the DART program aremany. The room design and the use oflaminar flow stations have greatly reducedcontamination levels. Standardization ofequipment has led to standardization ofprocessing. The order and manner inwhich the diffusion furnaces are locatedhas enabled us to implement some verysophisticated monitoring systems. We

now have the ability to continuously scanall our diffusion furnaces to assure that

critical parameters are being met. Theresult-better control of the actual diffu-sion process and quicker response toproblems."

J. MurrayProcess Engineer

Diffusion Furnaces

"The new DART area has given us con-trolled ambient and atmospheric con-ditions which are necessary forphotoresist processing. The standardizedprocessing stations give us wide flexibilityin scheduling work through the area. Thesolvent disposal systems have eliminatedspills and improved safety conditions forthe operators."

J. FaddenProcess Engineer

Photoresist Application

"The standardized wafer handling systemis responsible for decreasing waferbreakage 5%. Improved cleanliness con-ditions are directly responsible for in-crease in yields and distribution. Averageyields are running 5% above standard. Inaddition, a much higher percentage ofunits show blocking capability of 500 volts,thus improving distribution."

Pat RomanEngineering LeaderThy. Manufacturing

"Rate studies of the new diffusion areaequipment now underway show dramaticincreased rates, thus, lower labor costs.For example, of the ten cleaning ratesstudied this month, in the linear powerarea, the average rate increased 250%

effecting a $22,000 annual labor savings.Also, the asthetics of the area play heavilyon the workers' pride in what they do.

G. DiehlIndustrial Engineer

"Wafer Processing is much simpler tomonitor and control with the in -lineprocessing equipment. Product qualityhas improved and output has increased."

R. GlahnProduction Foreman

'Easy accessibility of standardized com-ponents has greatly reduced repair time.The order of the equipment layout putseverything within reach. The atmosphereis pleasant to work in."

R. DraughtMachine Attendant

"Rejection of product due to airbornecontamination has dropped to zero. In -process quality checks have beensimplified due to process standardization,thus allowing increased frequency ofchecks with no increase in labor. Waferbreakage and damage has significantlydecreased. The order, neatness, andcleanliness of the area have increased thepride of the hourly personnel. The degreeof technical competence and sophistica-tion displayed in the DART area can onlyleave customer survey teams with theimpression that RCA is an authority onsemiconductors."

H. RichardsonManager

In -Process Quality

47

The business end of engineeringan interview with Joe Volpe

Quite a few people were surprised in July1974 when Max Lehrer, Division VicePresident and General Manager of theMissile and Surface Radar Division. an-nounced that Joe Volpe would be the newChief Engineer.

Few were more surprised than Joe Volpehimself: -I never thought of myself as aChief Engineer... and I certainly neverthought of myself as a replacement forDudley Cottler as a Chief Engineer."

With a refreshing mixture of self con-fidence and humility, Joe realizes now thathe was being asked to bring a different setof talents to the Chief Engineer's job.

Ever since his youth. Joe Volpe has hadthe desire to "run things his way." Thisreflected itself in his aptitude at managingthe family business - a retail food store -while he was attending high school inPhiladelphia. Later. while he was workingpart-time testing paper and puttinghimself through St. Joseph's College. hebecame supervisor of a group of older andmore experienced people. Again, workingfor the Brown Instrument Division of the

Minneapolis Honeywell Regulator Com-pany, he moved naturally into a leadingengineering position.

Through almost 17 years with RCA, he hasmoved through assignments in design andproject engineering, and into projectmanagement. After almost ten years inproject management, and at a point wherehis career in project management hit itszenith (as AN/SPY-1 Development ProjectManager and later as System ProjectManager on the staff of RCA's largestGovernment program - AEGIS) JoeVolpe returned to engineering as ChiefEngineer - bringing with him a somewhatdifferent perspective from that of someonewho had "grown up" within the Engineer-ing Department.

He shares these perspectives with us inthis RCA Engineer interview.

The questions were asked by JohnPhillips. Associate Editor: Bill Underwood,Director. Engineering ProfessionalPrograms: and Don Higgs. Manager,Technical Communication. EngineeringOperations. MSRD.

You started with MSRD's EngineeringDepartment, then went into projectmanagement. Now you're back again inEngineering. The relative perspectivesfrom the various jobs might be in-teresting. Can you tell us aboutthem?

It was very interesting. I hope I don't gettoo wordy but this is a favorite subject ofmine. And it's been very educational.

When I was in design work I probablyhad far less appreciation of the totalproject picture-or even theorganizational structure-than I

developed later on when I was in Projectslooking at Engineering. So I had a littledifferent perspective. I was rather myopicand "Projects" to me meant the in-dividual project engineer, or leader, ormanager that we were interfacing withdirectly; not at all an entity. It was moreor less a situation in which if you had adecision to be made that was outside ofyour scope of work, you just went toProjects. It was the answer to a designengineer's prayer, And it made a design

Reprint RE -21-3-25Interview conducted March 19. 1975.

engineer feel very comfortable. He had ajob to do that was very technicallyoriented and not too responsible. Heknew that there was a cushion betweenhim and the real world-the real worldbeing the customer and all those nastythings such as overruns. But that was along time ago.

Didn't the project engineer or managerimpose a lot of constraints?

No, not too much. He was the guy youwent to for the answers. He would putsome constraints on you, but I don'tremember any undue pressures. But youhave to consider the era. It was differentthen. The problems were not withoverruns, they were the schedules. Get thejob done. It wasn't so much a question ofhow much does it cost; rather, do youneed resources here? Then get moreresources. Get it done-and get it doneright. We didn't have the fixed -priceenvironment and the extremely com-petitive situation that we have now.

Then when I moved into Projects, I wentthrough a transitional period: it seemed Ialways picked up the jobs that were introuble. I would get the tail end of every

one of them, either because the guy whohad it before me moved off to anotherjob, or -as we did things around herethen-if the job was in trouble, the way tofix it was to change the guy who wasrunning it. Sometimes that helps,sometimes it doesn't.

It was in a troubled environment; a lot ofthe programs were fixed price and hadtight schedules, and the relationship withEngineering could have very easily turnedinto an adversary environment. I don'tthink it did, in general, but I had to play avery strong directing role as a projectmanager and I guess I developed muchmore critical attitudes toward Engineer-ing people, because if they did somethingwrong, it would reflect back on mysuccess. Then a little further along, in thelater 60's on the MPS -36 radar days andthe AEGIS days, I started to think a littlemore ... I'll call it ... cooperatively. As amatter of fact, I made it a very basiccharacteristic of the way in which weoperated the MPS -36 program: on a we,not a you and they situation. I tried toestablish that there was no such thing asan Engineering problem, it was a

program problem and the modusoperandi was not to prepare for meetingsor review so you could throw spears at theother guy, but to try to work it out. Andthis was a way of life that I tried todevelop from there on out; to try to workwith people instead of trying to push andshove them.

L

I tried to establish that therewas no such thing as anEngineering problem, it was aprogram problem and themodus operandi was not toprepare for meetings or reviewso you could throw spears atthe other guy, but to try to workit out.

48

What +as your reaction to the first moveinto Projects?

When they asked me to make thetransition-even though I had been

working very closely with the projectgroup for maybe a year or two-all Icould think of was that Projects is a paperorganization. All they're doing is dealingwith paper and I could see all thesedirectives and policies and 1 said to myselfI don't want a paper job, I want to stayclose to hardware.

But I remember going around talking to afew people to get some advice; I've donethat a number of times when I've had tomake decisions. And the advice I got ingeneral was to look at it from theviewpoint that the people making thedecisions think you're qualified and thatit's the right place for you to go. I

remember the advice because I had op-portunities for jobs being offered andsometimes I didn't know what it allmeant. But I said, okay, let's see how itworks out. And it's worked out well.

Were you pleasantly surprised when youdid move into Projects full time?

Yes, I really was. First of all it was entirelydifferent and as I stayed with Projects andwent into each of the succeeding projects,I became so enamored with the operationI thought I'd never want to get out.Really, I felt this was the best job in thewhole Company for two reasons: one,you get involved in everything; you'reworking with a customer contact or withContract Administration, or withEngineering or Materials, and you havesuch a broad spectrum-if you ever gohome thinking you've had a dull day, it'smost unusual. Second, as projectmanager you are able to control, you canplan how you want to run a job, and youcan influence it and control it. Not allproject managers do, but you have thewherewithal. And I enjoyed that veryvery much.

When I went into AEGIS, I had a verylarge organization, and I enjoyed thatvery much. Certain aspects of that jobwere very frustrating, but not the projectmanagement aspect. So having had thatas a background, to me the next obviousstepping stone was (if you can ever getthere) to become a General Manager. Inever envisioned going back toEngineering-no way.

How did you view the Chief Engineer'sjob?

Everytime there's a reorganization and

We need new business to not only sury ve but to grow, and thechallenge to me is to do something different.

the rumor mill goes around as to who'sgoing to what job, I've been uniquelyunsuccessful in predicting the results.And prior to this job, if anyone had askedme, I'd be the last one I would havepicked for the Chief Engineer's job.

Do you remember the exact momentsomeone mentioned the Chief Engineer'sjob?

Oh yes, I can remember that. It was MaxLehrer. We had been having discussionsprior to this on my wanting to changepositions from the job I had in AEGIS. Ihad been on the program for five yearsand I felt I was getting stale. And that wasa problem to me in the sense that AEGISis such a large job, and I couldn't seetaking another job that would be a stepdown, at least in responsibility andproblems.

I didn't know it, but an opportunity forDudley [Cottler] to make a change hadcome up at the same time. So Max toldme to sit tight for a couple of weeks.Okay, so I didn't feel like sitting tight verylong but things developed within a coupleof weeks and he called me up and said"How about having lunch with me'?" Onthe way into the lunch room he said, "Ihave a job opening and you're one ofseveral candidates I'm considering"... andI think he implied that if we were mutallyagreeable I might be his first choice. Buthe didn't totally make the commitment.He's a good negotiator. And all the whileI was wondering, what the hell job is hegoing to offer me. He said, "I'd like you toconsider taking the job of ChiefEngineer." I stopped in the corridor-cold.

It was a traumatic thing for me, and Iguess it was sort of a lack of confidence inmyself more than anything. Because Inever thought of myself as a ChiefEngineer, number one, and I certainlynever thought of myself as a replacement

for Dudley Cottler as a Chief Engineer.And that's probably the most difficultthing I was faced with at that time.Fortunately I didn't try to replace Dudleybecause I never can. And Max wasn'tlooking for that. He was apparently asinterested in the management andadministrative aspects of the job as thetechnical part. But I couldn't see thatimmediately, and it was my biggest con-cern.

Within a few weeks after that, or no morethan a couple of months, I got over thefeeling of walking in Dudley's shadow (hehelped me a lot in this area) and started todevelop my own sense of security, and Istarted to carve out my own niche. And Ican truthfully say that of all the jobs Ihave had, I've never enjoyed a job asmuch as rve enjoyed this one. What do Ienjoy about it? I feel comfortable in thejob. I know the people, and that's been abig help. I also know the operation, Iknow how the people think and work, so Ifelt comfortable about doing somethingabout it. And I am very pleased with thereactions of the guys who have all pitchedin and made me feel comfortable. Theycould have all gone their private littleways and left me out in the cold. Youcan't force people to be cooperative, theyhave to want to do it. They have and it'sworked out very well. So I'm enjoying itthoroughly.

Are you enjoying the problems?

We have problems and I thrive on them.

What are they?

I can spell that out very simply: gettingnew business in the house. We have thesituation where we need new business tonot only survive but to grow, and thechallenge to me is to do somethingdifferent. That's not a criticism of the waythings have been done in the past, but it'sa responsibility I feel. I'm trying to do

49

things different but better.

This leaves several questions to beanswered. How are you going to generateproposals? What kind of an engineeringattitude, or changes in attitudes do youhave to get into the organization togenerate winning proposals? How areyou going to be able to do that so youdon't spend all your resources or youspend less of the resources? You haveonly so much money and it's the old story:the more business you need, the fewerresources you have to go out and get it.It's almost a positive feedback situation.What are you going to do so that yourproposals, from the cost standpoint, aremore competitive? How are you going toplan and run the jobs differently? Whatwould allow me to put in a cost that'slower and therefore raise my chances ofwinning? What are the different ways ofdoing things?

When I came over here I just felt thatresponsibility to do something better-not just different! I was very anxious notto be different just for the sake of beingdifferent. The fact that I came over veryopen minded surprised a lot of people,because I had been pretty opinionated inmy earlier dealings with them. There wereprobably a few of them who felt theyshould be writing their resumes rightaway, based on my prior dealings withthem. But it was good that I had an openmind, because in a few months I

developed some new ideas and appraisalsof the people and the organization. Wemade some operating adjustments, wemade an organization change which wasreasonably major, and I started todevelop a feeling for how I should run theEngineering Department and how theChief Engineer's office should operate. Ithink we're going through another transi-tion now, and we're going to have to startthinking about things in a different lightthan we have in the past.

What's the transition?

Driving home last night, anticipatingperhaps a question like this, I thought ofan analogy that might be helpful. Thesituation we're in now is a little like amartini. In the past, there was quite a bitof emphasis on building up technicalskills and abilities, advanced techniques,and advanced technologies; I liken theamount of time and energy and invest-ment that you can put into advancedtechniques and technology now to thevermouth in our martini. And right nowwe're living in an environment of very drymartinis. What can you do for thiscontract today to get business tomorrow,

versus planning the systems of the futureand building your skills, your techniques,and your know how to keep advancingthe state of the art.

In our present business climate, we havevery little to set aside to try the things thatdon't have an immediate payoff. Thatmakes my job very difficult in trying tokeep a balance between keeping ourtechnical skills up and running a business.And right now the business pressures areso heavy that you have to be careful thatyou don't cut the technical side down tothe point where-when you try to reviveit-there's nothing left. There is a storymy father used to tell about how in the oldcountry (he was born in Italy) they wereteaching the donkey to eat less and lessand they kept cutting the donkey's food inhalf every day until they finally had thedonkey trained so it wouldn't eatanything at all. And the donkey died.

How do you keep the donkey from dying?

I have ... I won't say pressures ... but Ihave guidance and inputs that help thebalance. On the one hand, I have contactsthat keep the business pressure on. We'rebidding new jobs, reviewing contracts,and looking at investment plans andbookings reviews; also Max's staff helpskeep the business pressure on.

Then there's a different kind ofpressure-more like reminders and thelike-from G&CS staff, and from theEngineering Department itself: IR&Dreviews and things like discussions onwhat kind of training we should do. Itcomes up every day in different ways. For

...right now the businesspressures are so heavy thatyou have to be careful that youdon't cut the technical sidedown to the point where-when you try to revive it-there's nothing left.

example, every time somebody wants togo to a technical seminar, you're facedwith making the decision of balancingyour overhead budget against your needs.Where do you draw the line? It's not aneasy problem in today's environment.

Given this environment for at least thenear. future, how do you think you canmotivate the technical staff to do a betterjob than they did before?

One of the things that I set as my goal-that I'm trying to have permeate theorganization-is critical to a lot of thethings we've discussed so far.

Engineering assignments have to bebroadened substantially over what theyhave been in the past. For whateverreason, we've gotten into an age ofspecialization and when you have 5000employees, such as this plant once did,specialization was not a big problem.Because you had 50 of every kind ofspecialty, you had a variety of resourcesto handle your assignments. As yourresources are reduced, specializationstarts to be a real problem. It's a dilemmabecause, along with the competition, youmust be technically innovative. At thesame time, you must be very cost com-petitive, so you can't afford to use all themost senior engineers as your totalresource. That might satisfy the first needbut not the second.

So to accomplish a major reduction in theengineering content that it takes to do ajob, we have to broaden the assignmentsthat we give the individuals. And-as ageneral statement -1 have a great deal ofconfidence in the ability of individualengineers to take on far broaderassignments than they have in the past.They may not do all assignments equallywell but they will do them satisfactorily,at least average or above average. Areceiver engineer, you know, doesn't haveto work on just i.f.'s.

In my own experience, not as a red-hotdesign engineer, because I'm not (I con-sider myself an average engineer), I'veoften had assignments in which I've hadno prior experience-and I found I couldadapt very readily. You do enough in-tensive homework and pretty soon you'redoing the job very nicely and are able tohandle it efficiently. If we can spread outand break down some of the technicalcubbyholes, its going to be a big help outof this obsolescence problem we've had.And at the same time from a businessstandpoint, it's going to have a goodimpact.

50

As your staff becomes higher priced andyou don't have new business coming in,you can't hire the younger, lower -scaleengineer. What kind of a problem is that?

Not easily solved, and I don't have anyready solutions. It's a double problem: Itbrings your average dollar rates up and itprevents you from getting the fresh bloodin, something 1 think is essential. Theyoung fellows coming out of school noware trained differently. They have freshideas, they have energy, they're not fixedin their ways as are some of the olderfellows. You need that constant flow intoany living, growing, vibrant organiza-tion. It's a kind of intangible thing, butyou can feel it. You can really feel it. I'mbeginning to sound like an old man, now,but I am getting older. Some of theseyoung fellows move a little faster and theyhave different ideas and it stimulates therest of us to think a little differently.

Now if you don't have that you're justgoing to continue into what I would call adecadent way of life. In today's businessclimate the stimulus must come fromshining stars within your organization-the good guys-these high priced but top-notch performers; they are yourmainstays. But if you get unbalanced, it'sa very difficult situation because in-variably these are key people that youcould not very casually put out on thestreet without it having an impact on yourorganization, your business, and on youpersonally.

If you remain unbalanced and you'reforced to cut back, you run the high riskthat you can never grow again becauseyou're going to be low on critical mastertalent. So more business is the answer. It'slike a gambler. You give him cash to workwith, he can do something with it, but yousend the best gambler into a gamewithout a bankroll, you've really tied hishands and his knowledge of the laws ofchance won't help a bit.

How do you feel you relate to your staff?How available is any engineer in theorganization to you? How available areyour managers?

I've developed a 90/ 10 philosophy ofoperation in this job. 1 soon found thatany one of maybe a dozen major aspectsof the job would bury the Chief Engineerimmediately if he decided to pursue it indepth. So looking at the resources I haveavailable, I got into this 90/ 10 conceptwhere I will delegate authority andresponsibility for say 90% of each job andtry to arrange to have the remaining 10%come to my attention.

I've developed a 90/10philosophy of operation in thisjob.

Now invariably you find yourself beingdrawn into a specific assignment thattakes a lot of your time. That probably isalso a way of life. But with this 90/ 10system, I have a lot more time available tospecialize, it's not like taking 100% of mytime and spending it on the 10-percenters.With that as a background, I try tooperate through my own first -levelmanagers if possible, and I say trybecause it's not a very natural thing forme to do. I know so many of the lower -level people, it would be much easier tooperate at the engineer level and bypassthe management team.

What kind qfproblems do you think yourstaff should bring to you?

I'm just going through appraisals andreviews, so we have had a chance todiscuss that question, at the first levelanyway. The feedback I'm getting-andI'm taking it at face value-is that theylike the idea of more responsibliiity andauthority being delegated to them. Thisgives them an opportunity, and they areexercising it. of using judgment as to whatthey think is impoi tali to bring to me.We have a very open relationship as far asinformation flow.

I make special mention of that becauseone of the constructive criticisms given tome before I took this job was-and I gotthis from Max and others-"you tend tokeep things to yourself and you don't letpeople know what's going on. People area little bit cautious about dealing withyou because they're not always surewhat's in your mind." So I took this toheart and I tried to get this "let's get it outon the table" attitude with the Engineer-ing Department and the same

relationship going oetween Engineeringand Projects, and there's a very positiveend result I'm aiming for.

There should be less need for havingproject engineers looking to see what'shidden in the design engineer's shoporder. "I That's the kind of relationship thathad existed for ten or fifteen years. That'show I thought when I was a projectmanager. But I know how they feel, and Ican direct assurances to the heart of theproblem. And say, "okay, now this is

what we are going to do different. I knowhow you're feeling, I was there a fewmonths ago. Trust me. If I don't do whatI'm saying ... if it turns out to be justwords, then we'll go back to the old wayof doing things." And the Projects peoplehave been very cooperative and it's work-ing out so far.

When you were discussing your earlybackground you mentioned joining RCAhere in Moorestown as a B engineer, yousaid you really didn't know who the ChiefEngineer was. In a sense you said theChief Engineer wasn't even part of yourworld. I wonder if that's how you feelabout the B engineer today? Would it beokay if he said that?

No, it wouldn't, although I havewondered whether they have anawareness of the Chief Engineer-not asan individual, but what he's doing andwhat he's trying to do. I don't know theanswer, but engineers today aren't nearlyas far removed from the Chief Engineer,the Marketing Manager, and all the restof them as I was. One major reason is thatthere are fewer people here now; it's easierto see through the crowd. We have lessthan 2000 people now; we had over 5000when I joined the Company. Also back inthose days there wasn't the opportunityfor visibility or to meet the ChiefEngineer. Now we have, and have had forseveral years, weekly reviews, so-calledChief Engineer's Line ManagementReviews. The principal engineer presentsthe progress and problems, and he usual-ly has his supporting cast in the audienceand they get a chance, not only to getvisibility, they get to meet the guy. Theyknow who he is, what he looks like, andtalk with him. Other mechanisms forcontact have also been built up: awards,patent disclosure honoraria-there'smore communication now, so that theguy would really have to be out of it not toknow who the Chief Engineer is

nowadays.

How accessible do you want them to feelyou are? In other words, could a B

51

engineer walk into your office? For whatpurpose?

I hat one is a little more difficult becausethe stock answer is motherhood and I'veheard a number of managers say it overthe years-"my door's always open." Andunless you put that in context, it may ormay not happen. One of the things I

wanted to do when I came to this job wasto make sure the managers manage. I'msensitive about going around them,because I've had some experiences overthe past years where some of my bossestended to go around me and I didn't likeit. What happens is that the managementpeople stop managing. Many times theycan't manage because they don't knowwhat's going on. And the lower group isworking without management.

Okay, what really happens when you slipinto that way of operation? You find thebusiness is being operated by the lowestlevel. This doesn't degrade them by call-ing them the lowest level, but it's thelowest responsibility level. With thisduality ladder at Moorestown, we havesome very strong, sharp people at this so-called lower management level, or non-management level. But one of the thingswe look for in upper management is abroader perspective and balancedemphasis on cost, schedule andadministrative aspects. So I am going toexpect my managers and managementpeople generally to manage.

To that point, the thing that's comingacross very consistently is the sort ofnaturalness that you feel about authori-ty. In this interview you haven't flauntedit nor have you been humble about it.You just seem to discuss it very naturallyas if it's just an ordinary life style. Is thatthe way you feel about authority?

Yes, I feel an urge to run things or take aposition of authority. I guess the mostfrustrating experience I have is everySunday at Mass; I watch the inept usherstrying to regulate the flow up to thecommunion rail. I suppress an urge to getup there and tell them how to do it right. Icontrol myself because I know if I go anddo that, the guy would say 'well why don'tyou join the Holy Name Society and theushers' committee?' I am not willing topay that price. But it just seems tohappen. When they started a father's clubat the local high school my girls went to,somehow or other I got elected president.I didn't know most of the people there,but I just naturally felt the need to takecharge. And, of course, I've used

52

meetings to accomplish my goals, but thisis a natural thing. I like to participate ingroups. I enjoy it. As a matter of fact, I'mmore comfortable working within groupsthan making an after -dinner speech. I feelless comfortable there than if I were in theaudience heckling.

What is the role of awards, publishing,and other professional motivationalprograms?

Well I think they are very important. In asense, they serve as catalysts to keep upnot only the technical image but thecapability. It forces some technical dis-cipline. It gives people, if you will, theexcuse, the incentive to do that little extrato keep technically sharp. It's not a bigdeal to get an award, at least certainly notfinancially, but it's tremendous to the egoand to an individual's well being, and Ithink it's very important to give recogni-tion, whether it be through publishing apaper or a Technical Excellence Award.

How do you think you relate with the restof Max's staff. How do you relate to himand his other department managers?

I guess I'll start with Max. I was ap-prehensive of that relationship. I thoughtthere might just be a little bit of a conflictof philosophies because I consider Maxto be a strong guy and I like to think ofmyself as a strong individual. To beperfectly honest with you (and we canalways edit this) it's worked out very well.I think there has been a mutual respectthat's been building. He provides aperspective for the business side of theoperation. However, you wouldn't knowhe didn't have a technical backgroundunless I told you because he assimilatestechnical material very quickly. I haven'tfound myself overly controlled in time,resource allocation, or in what I'm doing.I've been given reasonable freedom ofoperation and I've got enough time to domy thing over here. So that's good.

With the other members of the staff I gotoff on the right foot. Before this job couldbe announced, I started working with JimForan, our comptroller, very closely andhe started briefing me in anticipation ofthe job. So we got a good start in that wayand that's an important relationship for -the Chief Engineer.

Typical of the relationship is one I've hadwith one of the project managers who hada way and plan for running his primeprogram. It was almost 180° out of phasewith the way I'm trying to run the

Engineering operation. We sat down andworked this thing out. I think I convincedhim of 90% of what we wanted to do, andhe's willing to go along with it. I think hefeels confident that it has the prospects ofbeing a better way of doing the job. AndI've given in on a lot of points.

The important thing is that we've done itin a friendly way and not in an adversaryatmosphere-even though there weresome strong feelings on both sides to getin there and start shouting at one another.That's a different way of life with me-instead of being an arm -waving shouter,I'm trying to be more relaxed and give theother guy a chance to say something.

That's a different way of lifewith me-instead of being anarm -waving shouter, I'm tryingto be more relaxed and give theother guy a chance to saysomething.

Does arm waving and shouting have itsplace?

Yes, I think so. I probably wouldn't havegotten this job if I hadn't done some ofthat earlier in my career.

I still have no compunctions about clear-ing my throat, but I've tried to do that in acareful, calculated way. A case in pointcame up a few months back in a kickoff'meeting for a proposal; there were 50 to55 people there-at a time when I wastrying to implement a philosophy ofcutting costs of doing proposals. It wasprearranged for me to set the stage with alittle pep talk and ground rules. I sat therestewing and when I got up I don'tremember exactly what I said but I reallylaid it out that this was not the way wewere going to run this place with 50people attending a prop kickoff, and thatI was very disappointed and dissatisfiedwith this as a start. And then I used thatopportunity to express some of my

philosophy of operating and wound upsaying very few words about what I reallyhad intended. And then I left, and themessage got through. So you have tocome across strong sometimes.

On the other hand, do you think you canbend too much?

Sure. You can become wishy washy andbe almost apologetic. The importantthing is to be able to take the time andeffort to listen to other people's ideas.And then try to make a judgment aboutwhich are the best ways of doing it, and bewilling to change your mind. Numberone, I think you should develop your ownideas and try to improve them or changethem. But, if the other guy has a betteridea, you must be man enough to say,hey, that's a better idea, let's do it yourway.

What are some of your basic personalvalues? What are the things that you'llstand up and be counted for?

I find it very difficult if someone calls mea liar. And for me to be a liar is evenworse. That to me is something that isuncompromisable. My father used tomake a big issue of this. "If you ever lie,"he said, "just once, then I don't knowwhen you're not lying. That's the worstthing you can do and the hardest thing torepair."

So I guess the first word that really comesout in my own mind, is integrity. I like tobe extremely frank and above -board withpeople, but I have no compunctionsabout trying to strategize how to dothings. When you're out to win aproposal, you lay out a strategy; that'snot being untruthful, it's not reducingyour integrity. I mean, if I want toaccomplish something I may have to talkto the right people, make sure theyunderstand what I'm trying to do and viceversa, and so get them on my side. Youhave to do some of that or things justdon't happen. But I have to feel that I'mthe guy that people can believe; when Isay something I'll stand behind it. That ismost important.

Do you think the business is less fun now?

No, surprisingly enough I would havethought that would happen as I keptgoing up in the ranks and the responsibili-ty became greater. Definitely, if you hadasked me that question five or ten yearsago, I'd have said there would be less timefor fun. My present experience is just the

A good engineer, as far as I'mconcerned, is not someonewho designs a device thatworks. He has to fit the rest ofthe criteria.

opposite. I think this is more fun than anyother job I've ever had. Maybe ten yearsor two years or two months from now I'llfeel differently, but right now it's good.

There has been a definite trend in yourbackground-of being concerned withmonetary aspects of the business. Do youthink it's important for all engineers tohave such a concern and do you thinkthere is something there that a lot ofengineers lack?

Yes, you might say it could be a good or abad characteristic for a Chief Engineer,depending on what one says the ChiefEngineer should be, much less any othertype of engineer. As a Chief Engineer, yourun the risk of being termed a non-believer or a non-belonger to theengineering cult when you mention costand schedule. But I am a firm believerthat every engineer, not just the ChiefEngineer, should have a heck of a lotmore cost consciousness. A goodengineer, as far as I'm concerned, is notsomeone who designs a device that works.He has to fit the rest of the criteria.

Engineers often feel that they are to bemeasured by whether they can make a jobwork technically-to the exclusion ofother parameters. That approach may beall right for the research engineer. But inthe environment we work in, that's a veryunreal world. And to me, a good engineeris the one who gets the job done technical-ly within the cost and schedule con-straints. And that does not mean,necessarily, meeting all those, but doingthe best that can be done within thoseconstraints.

You've been in the defense business nowfor about seventeen years. How has itchanged? What does ii look like for thefuture?

When I look back and think of thedefense business-the front end of thatseventeen years-it was a big amorphousmass to me at that time and it seemed itwas the era of big contracts in which I wasa very small cog. We're constantly goingthrough cycles in the defense business. Iguess a lot of them are driven by whoeveris the current Secretary of Defense and byhis unique policies, whether they be "flybefore buy" or "total procurement," orwhatever.

Does the Secretary of Defense himselfhave that much of an impact?

Oh yes, I think so. It starts usually as asimple little memo or a concept thatsomehow or other gets propagated in aspeech or a very driving directive. Andfrom there it just filters down through thehuge community and pretty soon we havenew policies. (The current variety is

design to cost.) And from there we getdirectives and seminars and procedures,and you just see it everywhere you turn.Some of these things are just given lipservice - something that has to beincluded in the contract or the contracthas to appear to have it; others areintricately involved and you see theminfiltrating every aspect of a job.

Where are we now? Now we havedifferent pressures. The economy, or thecombination of the economy and thepeople's will or the reaction to theGovernment, is influencing defensepolicy. This impacts in two ways: first,because of inflation we have less realmoney available for defense or any otherthings out of the budget. Second, becauseof the reaction against involvement, peo-ple are unsympathetic to defensespending, and this influences con-gressmen who serve the people.

How can you compete in tha4

environment?

Well, with the shortening of funds thatthis environment brings out, the servicesstart to reflect this in their procurements.They want off -the -shelf equipment thatdoesn't require development. They wantvery competitive price bidding. Then theystart to get troubled after they'reburned with overruns. Now I wouldn't besurprised to see an era of penalty clauses.They may try to put constraints on toprotect themselves. This is going to cause

53

industry to be reluctant to bid in thatenvironment. And then the pressures willmount; you have to bid in order to stay inbusiness and yet you are afraid of beingforced into a situation that is unaccep-table. I think we're going into a wholenew set of challenges in industry. In-dustry will have to take risks-technicaland cost risks-in which the odds are infavor of having financial reverses. Andyet the alternative to doing this is to goout of business.

Do you foresee a negative industryreaction? For example, no oneresponding to a particular invitation tohid?

I say yes quickly, and I guess my second,slower answer would be no. It seems thatthere are always people who will bid. I

can't recall an example where it (a com-plete industry no -bid response) has

happened in the past, no matter howunrealistic the situation. There have beena few not too distant experiences in whichI felt that the procurement and theamount of money available for theprocurement (this is fairly commonknowledge) were just completelyunmatched-yet they became closelycompetitive situations. 1 think the realend result is that more companies willleave the defense business because it's notprofitable. When you add that to therelationship between the defense businessand the basic economy, companies willlook very critically at each of theirdefense -oriented entities and peel off theones that are not good performers.

Do you think that survival is going todepend on the technology or on businesspractice?

Well, 1 have an intuitive feeling thatsurvival is going to depend on somecombination of innovative technical con-cepts and implementation, combinedwith a lot of solid good business ratherthan on advanced techniques andtechnology. By innovative, I don'tnecessarily mean a new type of solid-statedevice, but a better way of implementingwhat is available in a very cost effective orperformance -producing way that willgive you the edge-so that you havesomething to sell that is better than thenext guy's product. I think you will see atremendous amount of cost vs perform-ance pressure. We are working in a

seemingly impossible cost and scheduleenvironment, with usually very difficulttechnical performance constraints. Somecompanies are going to fall by thewayside.

And possibly we'll go into some othercycle that I can't foresee-hopefully withthe economy straightened out, inflationnot too bad, and not everybody hatinganything associated with defense. Wemay be able to spend money again, andwe'll get back to increasing our technicalcapabilities. Maybe it'll be something likeanother moonshot that will be thecatalyst. It seems that we need a catalyst.I've never seen people come up with ideaswhen there's no need for them. There area few guys who just sit there and think ofthem, but they're rare. When the pressureis on. though, we make advances.

Given this environment, then, andpossible future environment of Govern-ment Contracts, where is MSRD today,what are the trends and where are wegoing to go with those trends?

First of all, we've gotten down to thepoint where we are lean and hungry. Sothere's motivation as a starter. It's achallenge getting enough new business into keep our people employed, but theorganization has been pruned down overthe years so we're getting close to that,critical mass.

Given that as the starting point, what dowe have to build on? We've got a prettygood repertoire of product experience.We've built many different kinds ofradars and systems; our experience hasbeen broad enough. Basically, our line isradars and big systems. The hooker is,what do we do with all that?

The first thing we are trying to do, is todecide very selectively where we're goingto go. We're in that process now. That'sprobably the biggest challenge we have:looking at the business areas we are in orwill pursue and then selecting those onwhich we can build.

What are the areas we want to pursue?Well, our makeup and structure here inMoorestown help us decide that, at leastpartially. We have all these skill-andthe laboratories and support facilities togo with them. And that means wt shouldprobably avoid most of the basic, simplelittle jobs, because we can't compete incost with somebody who isn't carrying theexpense of our big pool of talent andfacilities.

So we have to go after the jobs that arecomplicated-programs that need us andcan afford us. Also, we're going to findourselves changing our business aperturefrom one -of -a -kind targets to those thathave repetitive payoffs, those that haveproduction followup. And production

I think we enjoy a far bettersituation than a lot of the otherGovernment -business -basedfirms.

might be five or ten, or thirty-notthousands. And that business base mayswing over to be predominantly manufac-turing, but there's always an engineeringcontent.

So if we're successful and build ourselvesup a certain level of business base, we canafford to think a little more calmly andsanely, to build on that base and go aftersome new businesses.

Okay, that's one aspect of it. Now, theother aspect is that we're going to have tofind a way of doing the jobs with half theengineering content that we've used in thepast. Everybody says I'm out of my mindto say that. And maybe I don't reallymean that, but it does emphasize thepoint. It has to be a dramatic change, notjust 5 or 10%. To do that dramatically,we've got to be constantly thinking ofways-reorganizing, accomplishing thejob-all the things we kind of touched onin various places, so that each one of themmay be a 5 or 10%. We have the samesituation in manufacturing and all theother parts of a job. But addressingmyself just to engineering, whichsometimes is a major part of a business,that's one thing we have to do-a bigchange.

So where does that leave us? Well, wehave a reasonable business base that letsus hold our people and gives us a chanceto build. 1 think we enjoy a far bettersituation than a lot of the otherGovernment -business -based firms. Now,I think we're going to have to turn thecorner and build up our base. We haven'tbeen uniquely successful up to now, butI'm confident we'll do it. This may be theyear.

54

Chemical conservationin the Solid State DivisionR.N. Epifano J.R. Zuber Dr. J.A. Amick

In 1973. the Materials and Processes activ ty of the Solid State Division began a programaimed at conserving chemicals used in solid state device manufacturing. Throughextended use, substitution. and recycling of chemical materials. the Solid State Divisionhas been able to reduce costs. alleviate shortages of critical materials and 'educecontaminants in plant effluents. The conservation methods include extended use ofbuffered etch, continuous addition of hydrogen peroxide to sulphuric acid. andreconstitution of aluminum etch.

I NCREASING COSTS, scarcity ofcritical materials, and ecological con-siderations necessitate an activechemical -conservation program at SolidState Division installations. Risingproduction levels require ever-growingvolumes of chemicals. This additionaldemand, coupled with the increasedprices of chemical materials will, if noconservation action is taken, have a

Reprint RE -21-3-17 (ST -6323)Final manuscript received December 11, 1974.

Robert Epltano, Ldr., Materials and Process, Solid StateDivision, Somerville, N.J., graduated from Lehigh Uni-versity in 1959 with the BSChE. He completed severalgraduate level courses in materials technology atNewark College of Engineering. Mr. Epifano joined theBendix Corporation in 1959 as a materials engineer, andcontributed to the development and evaluation of insula-tion and encapsulation materials for airborne elec-tronics. such as the solid-state inverter used in all Geminimissions. He joined RCA in 1965 and worked in the areaof encapsulation of semiconductor devices. Currently heis responsibile for two Technology Groups in theMaterials and Processing Department, the Polymer andthe Analytical Technology Groups. Mr. Epifano hasgiven many in -plant lectures on p astic encapsulation.and has presented papers at technical seminars onplastic encapsulation of integrated circuits and theevaluation of mask -to -wafer aligners. He holds a U.S.patent.

significant cost impact on Solid StateDivision operations in 1975 and beyond.The unprecedented and unpredictedgrowth of the need for chemicals coupledwith petroleum shortages and the lack ofsufficient new capacity in the refining andchemical -manufacturing industries havecombined to produce shortages inmaterials of critical importance to theSolid State Division. To further com-

John R. Zuber, Advanced Materials anc ProcessLaboratory. Solid State Division, Somerville, N.J.,received the BS in Chemistry from Fairleigh DickinsonUniversity. He has completed several post -graduatecourses at Rutgers University and numerous RCA spon-sored courses. Mr. Zuber was employed by the HeydenDivision of Tenneco Chemical Corporation in 1952,initially in Research and Development, and then in theQuality Control Laboratory. He joined RCA in 1959 as ananalytical chemist in the Materials Laboratory. In 1965 hewas assigned to the Advanced Materials and ProcessLaboratory as a Member of the Technical Staff. He hasworked in various areas of process control includingclean processing, chemical etching, and pollutionidentification and control. Most recently, he has beenresponsible for the initiation. development, testing, andimplementation of many chemical conservat on techni-ques. Mr. Zuber has co-authored two papers publishedin Analytical Chemistry, and has delivered lectures onanisotropic etching and analytical techniques for solid-state process control. He has one patent granted, withothers pending, and is a member of the Electro-ChemicalSociety.

plicate the situation, limitations on thecomposition of plant air and watereffluents place additional restrictions onboth the manufacturers and users ofchemical materials. Chemical conserva-tion through extended usage, substitu-tion, and recycling of materials is impor-tant as a means of reducing costs,alleviating the critical materials shortage,and controlling air and water pollutants.As a means of effecting these conserva-tion goals, committees were establishedin 1973 at the Findlay, Mountaintop, andSomerville locations. Members of theMaterials and Processes activity ofthe Solid State Division, Somerville,were assigned to work in the area ofchemical conservation. This articlereviews the status of the chemical -conservation program and detailsselected areas of achievement.

Dr. JamesA. Amick, Mgr. Materials and Processes, SolidState Diy.sion, Somerville, N.J. received the AB inChemistry from Princeton University in 1949. He thenspent a year as a predoctoral fellow at BrookhavenNational Laboratory, before returning to Princeton forthe MA and PhD in Physical Chemistry in1951 and 1952. Dr. Amick joined the staff of the RCALaboratories in 1953. where his starting assignment wasto the physical analysis laboratory. In 1955 he beganworking on Electrofax, and during 1956-57 he wastransferred temporarily to the Zurich, Laboratories tocontinue this research. On returning to Princeton. hewas assigned to the Materials Research Laboratorywhere he carried out research on the stabilization ofsemiconck.ctor surfaces, and the epitaxial growth ofsemiconductor elements and III -V compounds. For thislatter work he shared in the 1966 David Sarnoff Award forOutstanding Team Performance in Science. In 1963 hejoined the Process Research and DevelopmentLaboratory where he was head of the Process ResearchGroup. In 971 he was named Manager of the Materials &Processes: Department of the Solid State TechnologyCenter in Somerville. N.J. In 1974 this department wastransferred from the Technology Center to the SolidState Power Operations of the Solid State Division. Dr.Amick has three issued patents, has contributedchapters to two published books, and is author or co-author of 26 papers. He is a fellow of the AmericanInstitute of Chemists, and a member of the AmericanChemical Society. the Electrochemical Society. AAAS,and Sigma Xi. He is listed in American Men of Science.

Ady e.f.31

\NO::

55

CHEMICAL CONSERVATION PLAN

RETURN TO PROCESS

NO CHANGEEXTENDUSAGE

CHEMICALFROM AFTER ANALYZE CHANGE

PROCESS NORMAL USE CHEMICAL

DISPOSAL

USECHEMICAL

-01 DEPLETION

_...1PARTICULATE

CONTAMINATION

3

IONIC

CONTAMINATION

C

0

o

_.]ORGANIC

CONTAMINATION

_....1 COMBINATIONOF

ABOVE

STEPS IN RECYCLING MATERIALS

ANALVIEEXTEND

USETREAT AND'ORRECONSTITUTE

INCREASING COST

Fig 1 - An approach to chamIcal conaarvatior..

Program initiation

The Mountaintop location, which usesmore chemical materials than any otherSolid State Division location, was thelogical choice for initial conservationefforts. As the first step in organizingthese efforts, the chemicals used atMountaintop were grouped as shown inTable I. The percentages are six-monthaverages for September 1973 throughFebruary 1974. A listing of the top ninechemicals on an expense basis gives abetter representation of specific materials

Table I - Chemical usage at Mountaintop.

Material al. total costfur chemicals

Group AOrganic Solvents 36

Group BAcids (e.g., Nitric, Sulfuric,

Hydrochloric)Hydrogen PeroxideAmmonium Fluoride

43

Group CSilicon Chlorides 11

Group DMiscellaneous Chemicals 10

RECOVER

involved. The materials in Table II ac-count for 75% of total consumed -chemical cost.

Clearly, chemical conservation effortsshould be directed at those materialsfrom which the greatest impact on costreduction and pollution control can berealized. In addition, certain materialsnot representing a large dollar outlay, butwhich are in short supply, also requiredimmediate attention.

The steps in recycling chemicals, in orderof increasing cost, are extended usage,reconstitution, and recovery of con-stituents. The extent to which thisprogression is followed depends on theextent of change in the material, the easeof reconstitution, relative cost of newversus recycled reagents, value of theconstituents, cost of waste treatment,availability of the chemical, andenvironmental considerations. One ap-

Table II - Mountaintop chemical usage-top ninechemicals ranked in terms of decreasing expense.

hluonnatea solventsSilicon TetrachlorideChlorinated solventsMixed acid silicon etchHydrogen peroxideAcetoneBuffered oxide etchHydrofluoric acid

preach to chemical conservation is shownschematically in Fig. 1.

The success of the chemical -conservationeffort depends on an ability to analyze the"spent" chemical materials for the pur-pose of specifying appropriate recyclingaction. Analysis has shown that certainprocessing chemicals (e.g., buffered etch)remain essentially unchanged during longperiods of use, and can be conservedsimply through extended use. However,many of the other reagents employed inmanufacturing become depleted or arecontaminated with ionic materials,organics, and particulates during use.After analysis to determine the nature ofthe change, recycling of these materialsrequires treatment such as reconstitution,distillation, absorption, or filtration.After treatment, materials are re-analyzed by techniques such as emissionspectroscopy, atomic absorption spec-trometry, and wet analysis to ascertainthe effectiveness of the recovery treat-ment. In addition, the effects of extendedusage or of reclaimed material on siliconwafers is monitored by analyzing thesurface test wafers.

Buffered etch

Buffered etch, prepared by mixing am-monium fluoride and hydrofluoric acid,was one of the first chemicals to bestudied in the conservation program sincethe allowed period of use is determinedempirically. A laboratory test was per-formed to determine the effect of ex-tended use of a buffered etch on its abilityto dissolve the oxide layer on thermallyoxidized silicon wafers. This etch, alongwith samples of new and discardedbuffered etch obtained from manufac-turing locations, was subjected toanalysis for assay, ionic contamination,and etch rate. Also, calculations wereperformed to determine how much of thecapacity of the etch to dissolve SiO2 wasbeing used up. These calculations in-dicated that 5.5% was lost, while atmanufacturing locations, the percentageloss ranged from 1.5 to 8.5

These low consumption factors offeredconsiderable encouragement that a

reduction in buffered etch volumes couldbe achieved through extended usage.Further encouragement came in the formof analytical data on the etches from themanufacturing locations. The results of achemical assay and emission spec -

56

Elements Level (PPM)Cu <0.005 0.005 to 0.05 0.005 to 0.05 <0.005B ND' ND' ND' 3.0'Ca 0.005 to 0.05 0.005 to 0.05 0.05 to 0.5 0.05 to 0.5Al ND 0.05 to 0.5 0.05 to 0.5 0.05 to 0.5Ni ND ND ND NDNa ND ND 0.5 to 5 NDFe 0.005 to 0.05 0.05 to 0.5 0.05 to 0.5 0.05 to 0.5Ph ND 0.005 to 0.5 0.005 to 0.05 0.05 to 0.5

Table III - Buffered etch - normal factory usage before conservation. trographic analysis of new and usedbuffered etch from one of the Mountain -

ASSAY top manufacturing areas is presented inChemical New Used Table III. The data show that the

(%) (%) characteristics of the used buffered etchHF 7.4 7.4 were virtually unchanged at the time itNH4F 33.9 33.8

was discarded (after normal usage).

IMPURITIES (by emission spectrographic analysis) Prompted by this favorable information,one manufacturing area began triple -

Level (PPM) Impurity usage (three shifts instead of one). At the0.05 to 0.5 None None end of hte triple -usage period, the "spent"0.005 to 0.05 AI,Mn,Si,Fe,Ti,Ca,Mg AI,Mn,Si.Fe,Ti.Ca.Mg etches were submitted to the Materials<0.005 Cu Cu and Processes Activity for analysis and

characterization. In addition to chemicalETCH RATE UNCHANGED assay and emission spectrographic

analysis of the new and triple -used etches,polished silicon slices were exposed to the"spent" etches and examined with a

Table IV - Chemical analysis of etch solutions used for three times normal usage. Reichert Differential InterferenceMicroscope in an effort to detectparticulate material. Samples were also

ASSAY analyzed at the Princeton LaboratoriesChemical New Used #1 Used #2 Used #3 by spark -source mass spectrometryHF 8.2 7.5 6.5 6.9NIL,F 35.5 37.4 32.5 35.0

(SSMS) to determine the presence ofsurface contamination. Results of the

IMPURITIES (by emission spectrographic analysis) etch analysis are depicted in Table IV.Table V shows the results of SSMSanalysis of test wafers exposed to the newand triple -used ("spent") buffered etches.

Comparison of this data with SSMS datafor polished silicon wafer controls thathad not been exposed to buffered etchindicated that all test and control wafershad similar impurity levels. Based on this

Other elements not detected or at very low levels information, buffered -etch usage hasETCH RATE UNCHANGED been extended by a factor of three in

production. More recent experiments aredirected toward extending the life of

I. Not detected - detection limit 0.3 PI'M buffered etch by an additional factor of2. Not detected detection limit 0.1 PPM three. Initial data indicates that this3. Used to etch boron -doped oxide further extension is feasible.

Table V - Spark -source mass -spectrometer analysis on wafers exposed to new and"spent etches.

ElementNew etch(PPMA)

Used III(PPMA)

Used #2(PPMA)

Used #3(PPMA)

z

4.

0

W

21 22 12 6

v

iB 0.4 2.5 0.6 2.4

Ca 1.2 <0.9 <0.7 <1.1Al 1.1 <0.8 <0.5 <0.8Ni <5 <6 10 <9

Not detected P, Na, Fe, Ph, Sh, Sm, Zn, Cr, K. Mg. Ag. Pt

The successful conservation of thischemical material obviously provides acost reduction. More importantly, it alsoreduces material handling and removes

....oak \%

... .soN.,,,,- COST

...so.

,..e. ,....

APR MAY AUG SEPT OCT NOV DEC JAN FEB MAR APR MAY

L 973

MONTH

19 74

Fig. 2 - Cost of buttered etch in dollars and thepercentage of the total cost of the top ten chemicalsplotted as a I unction of the month of use.

57

part of the load on the plant efiluen t -treatment facilities. The cost reductionachieved is illustrated in Fig. 2 iii whichthe cost of buffered etch in dollars and thepercentage of the total cost of the top tenchemicals is plotted versus the month ofusage. Note that both yardsticks show adowntrend. The decrease in cost is,however, partially offset by 'increasingproduction requirements. Similar infor-mation is plotted for chlorinated solventsin Fig. 3. In this case, both dollar cost andpercent of the top ten is rising as conser-vation efforts have not yet been applied inthis area.

The cost reduction achieved withbuffered etch can be dramatically il-lustrated by plotting the percentage costsfor chlorinated solvents and buffered etchon the same graph with the total cost forthe top ten chemicals, Fig. 4. It is readilyapparent that the percent cost which thechlorinated solvents represent remainsrelatively constant while the percent costfor buffered etch drops markedly becauseof extended usage.

Caro's acid

In mixtures of sulfuric acid and hydrogenperoxide, commonly referred to as

"Caro's"' acid, the sulfuric acid andhydrogen peroxide are mixed in thedesired ratio and, because of the ex-othermic nature of the reaction, a

temperature of approximately 100 toI 30°C is obtained. The silicon wafers tobe cleaned are immersed in the hotsolution for 10 to 20 minutes and are thenquenched with deionized water. The"spent" acid, now cool and substantiallydepleted of hydrogen peroxide, is dis-carded. Sulfuric acid and hydrogenperoxide are widely used in both power

#1\

- COST

APR MAY AUG SEPT OCT NOV DEC JAN FEB MAR APR MAYMONTH

141--- -1973-- 1974 -

Fig. 3 - Cost of chlorinated solvents in dollars and thepercentage of the total cost of the top ten chemicalsplotted as a function of the month of use.

Table VI - Standard impurity mixture 1000 Mg/literof the ions listed.

Cu'AI' Fe"As" Ni

P*2

Ca'' Ph"Cr''

and integrated circuit operations and areon the list of the top ten chemicalsdiscussed earlier. Therefore, they rank asprime candidates for chemical -conservation efforts. Experiments per-formed in the Materials and Processeslaboratory have demonstrated that muchof the hydrogen peroxide is exhausted insupplying the initial temperature rise,amount required for the cleaning func-tion is small. To further test this conclu-sion, sulfuric acid (without any hydrogenperoxide added) was electrically heatedto 125°C. A batch of wafers coated withphotoresist was then immersed in the hotsulfuric acid. It was found that the hotacid alone was capable of removing thephotoresist film; however, carbonizedorganic residue from the resist turned thesolution black, and it would be expectedthat, without the addition of thehydrogen peroxide, the carbon residue onthe surface of the wafer would be high.With the acid still hot, hydrogen peroxidewas added slowly until the solutionchanged from black to clear, indicatingcomplete oxidation of the carbon. Theamount of peroxide required for thisoxidation was only ten milliliters per literof sulfuric acid - far less than thatrequired in the normal method in whichthe hydrogen peroxide also provides theheating action. In addition, it was foundthat the same bath of sulfuric acid, if kept

COST TOP TEN CHEMICALS% CHLORINATED SOLVENTS% SUFFERED ETCH

1 I 1 I

APR MAY AUG SEPT OCT NOV DEC JAN FEB MAR APR MAY JUNEMONTH

1973 i - -1974 - --1

Fig. 4 - Plot of the percentage cost for chlorinatedsolvents and buffered etch with the total cost for the topten chemicals.

0

hot and provided with a continuousinfusion of peroxide, could be used formultiple batches of wafers. To determineto what extent the bath life could beextended, a test was devised in whichwafers cleaned in standard "Caro's" acidand in the modified mixture preparedfrom hot sulfuric acid by continuousaddition of hydrogen peroxide were com-pared by analyzing the surfaces of wafersexposed to these solutions. To simulateextended usage in the laboratory, a stan-dard impurity mixture was prepared andadded to test solutions at two levels. Thestandard impurity mixture is described inTable VI.

An experiment was performed com-paring wafers cleaned in standard"Caro's" acid with wafers cleaned in three"continuous addition" Caro's mixtures.

The first of these mixtures was unused;the second contained elevenmilligrams/ liter; and the third containedsixty-six milligrams/ liter of impurities.The third sample should be consideredhighly contaminated. Following clean-ing, the wafers were sent to theLaboratories at Princeton for spark -

results of this analysis are given in TableVII.

From this experiment and others, it canbe concluded that the life of sulfuric-acid/hydrogen-peroxide mixtures can besignificantly extended by the continuousaddition of peroxide to heated sulfuricacid. The resist removal and cleaningeffectiveness of these mixtures does notdecrease with time, and even largeamounts of impurity ions added to thesolutions do not adsorb on the siliconsurfaces. Equipment capable of con-tinuous addition of hydrogen peroxide tosulfuric acid is commercially available forwafer cleaning and photoresist stripping.

Aluminum etch

A hot mixture of phosphoric and nitricacids is used in semiconductor manufac-turing to etch the aluminum metallizationpatterns on device wafers. The perfor-mance of the etch degrades rapidly withtime, and it is discarded frequently. As inthe case of buffered etch, samples of newand used aluminum etch were analyzed todetermine the changes responsible for therapid degradation of etch performance.The results of the chemical assay are

58

shown in Table VIII. These data weretaken from a factory location and fromtests performed in the analytical' facilitiesof the Materials and Processeslaboratory. The latter tests involve thecontrolled addition of aluminum powderto phosphoric -nitric aluminum etch andindicate that the basic degradationmechanism in this etch is the loss ofvolatile nitric acid and water. Since onlyvery small amounts of aluminum aredissolved in the etch, and virtually nophosphoric acid is consumed, this etchbath can be reconstituted by periodicaddition of a one-to-one nitric -acid / water mixture. Such reconstitutionpermits extended usage of the etch withno perceptible change in etchcharacteristics.

Solvents

Organic solvents are consumed in largevolumes at Solid State manufacturinglocations and represent a considerablepercentage of the total dollars spent onchemicals. At the Mountaintop plant, forexample, fluorinated hydrocarbons as agroup represent the single largest ex-penditure for a chemical. When applyingchemical conservation methods tosolvents, several additional constraintsare introduced, one of which is the largevariety of materials utilized; therefore, amore diversified approach to conserva-tion is required. Fluorinated andchlorinated hydrocarbons, aliphatic andaromatic hydrocarbons, ketones, esters,and ethers are all used at various points inthe manufacturing process and possessvarying levels of flammability and toxici-ty. Nevertheless, the high cost and scarci-ty of many solvents required in devicemanufacturing makes it imperative thatconservation be practiced.

Solvents are lost in many operations bysimple evaporation, or they are selective-ly contaminated depending on theapplication for the solvent; therefore,simple extended usage, as employed forbuffered etch, is not an effective means ofconserving solvents. Numerous alter-natives are being explored to provide costsavings and to alleviate critical materialshortages.

Fluorinated hydrocarbon solvents areused in large quantities as drying agentsfor semiconductor devices. Thesematerials, while not in particularly shortsupply, are nevertheless costly. It hasbeen found that substantial differences in

Table VII - Results of SMSS analysis of silicon wafersexposed to 'Caro's" acid samples.

Impurity Std

"Caro's"(control)

Continuous addition "Caro's"

New II Mg' L 66 Mg/ LImpurities Impurities

CCCuAgB

43.

1.1

0.60.5

13

0.8

< 0.i0.8

87 25

2.0 0.9

<1.0 <0.30.8 1.4

Not detected - Al. 4s, Cu. Cr. Fe. Ni. P. Ph

Table VIII - Change of composition of nitric phosphoricacid aluminum etch with LAP

SampleHNOGram% / iter

New 74

After one hour 30.8

After eight hours 3.1

Ht PO,Grams/ liter

1295

1444

1620

efficiency exist in drying equipmentsupplied by competing manufacturers.Equipment now being installed at onefactory location will, by reducingvolatilization of the fluorinatedhydrocarbon, reduce consumption of thismaterial by a substantial margin over thatof the equipment being replaced.

In the case of a triple -distilled proprietarychlorinated solvent used for post -mountcleaning of pellets, a cost reduction ofsixty percent is achieved by converting toan alternative high purity solvent ofslightly different composition having amuch lower toxicity. Methods of recover-ing acetone and photoresist developingsolvents (materials which are in shortsupply) are currently being developed tomake possible conservation of thesechemicals.

Conclusion

Chemical conservation through extendeduse, substitution, and recycling ofchemical materials is an effective meansfor reducing costs, alleviating shortagesin critical materials, and reducing con-taminants in plant effluents. The savingsachieved to date attest to the fruitfulnessof chemical conservation efforts. It is

expected that the level of savings willcontinue to rise as the chemical -conservation effort gains momentum andas more conservation practices are im-

plemented in the manufacturing plants.In addition to the direct dollar savings,other benefits accrue from the chemicalconservation program. Shipping costs,which previously were included in thechemical prices, are now an added cost;thus these costs will also be reduced assmaller quantities of chemicals arepurchased. The size of chemical inven-tories and storage areas may be reduced.Plant effluent -treatment facilities willoperate more effectively with the lowervolumes of waste materials, perhaps cir-cumventing the need to increase the sizeof these facilities. Because of the currentbusiness environment, chemical conser-vation warrants the attention of all RCAmanagement as an effective means ofcontributing to RCA's competitive posi-tion through more efficient manufac-turing.

Acknowledgment

The authors thank Carlo Grilletto andAlex Gaska of Somerville for their effortsin emission -spectrographic and wetanalysis of chemicals in support of thiswork, and also Efram Botnick of thePrinceton Laboratories for spark sourcemass spectrographic analysis of siliconwafer surfaces. The authors also thankthe members of the Mountaintop andFindlay chemical conservation com-mittee who contributed information usedin the preparation of this article.

59

Michael D. Lippman, Communications ResearchLaboratory. RCA Laboratories. Princeton, N.J., receivedthe BEE in 1966 and the MSEE in 1968, both from NewYork University. He joined RCA Laboratories in July1966. as a participant in the Research Training Program.In 1967 he joined the Communications ResearchLaboratory where he was initially involved in researchaimed at finding efficient coding techniques for digitalimage storage and transmission. He has also served as aconsultant on facsimile and image data compression to anumber of corporate divisions. Since the fall of 1972. hehas been working on the application of microcomputersin data communications systems. As part of this ef fort henas designed communications interfaces and writtencommunications applications programs for the RCACOSMAC microprocessor. Mr. Lippman has publishedsix technical papers. He is a member of the Tau Beta Pi,Eta Kappa Nu. and the IEEE.

Anthony Longo, Eng. Ldr.. Special Equipment DesignEngineering, RCA Global Communications, Inc., NewYork, N.Y. received the BSEE (magna cum laude) fromNew York Institute of Technology in 1969 and the MBAfrom New York Institute this year. Mr. Longo joined RCALaboratories in 1958 wh4re he was engaged in acommunications system evaluation program. In 1961 hetransferred to the RCA Advanced CommunicationsLaboratories where he gained extensive experience inadvanced military systems. Mr. Longo joined RCAGlobal Communications in 1968 and was promoted toEngineering Leader in 1970. At Globcom he has beenengaged in the analysis of customer data com-munications requirements and is responsible for thedesign and development of specialized systemhardware. Mr. Longo is a member of the Delta Mu DeltaNational Honor Business Society.

Dr. Paul M. Russo, Systems Research Laboratory, RCALabortories. Princeton. N.J. received the B Eng inEngineering Physics from McGill University in 1965, andthe MS and PhD in Electrical Engineering from theUniversity of California, Berkeley, in 1966 and 1970respectively. During the 1969-70 academic year he heldthe position of Acting Assistant Professor with theDepartment of Electrical Engineering and ComputerScience at the University of California, where he taughtcourses in Circuit Theory and Circuit Optimization. Dr.Russo joined the David Sarnoff Research Center inSeptember 1970 where he has done research in com-puter architecture, program behavior, computer systemperformance evaluation, microprocessors andmicroprocessor applications. He is a member of theIEEE, ACM, Eta Kappa Nu and Sigma Xi, and is currentlychairman of the IEEE. Committee on Social Implicationsof Technology's Working Group on Energy/Environ-ment.

Leased channelwith microprocessor controlA. Longo Dr. P.M. Russo M.D. Lippman

The use of international teleprinter leased channels has provided the Americanbusinessman with the tools he needs to communicate quickly, privately, and economicallywith offices and correspondents located virtually any place in the world. However, the everincreasing complexity of today's communications systems and the explosive growth intraffic volume coupled with spiraling implementation costs, present a challenge to RCA toeffectively meet the goals of efficient. reliable and flexible leased channel service. A storedprogram approach to system design appears to be the long-term answer to this challenge.This paper describes a stored program approach to leased channel design, based on theRCA COSMAC microprocessor. which satisfies present applications at minimum cost yetcan accommodate anticipated future performance requirements.

THROUGH the use of teleprinterleased channels. RCA Globcom providesthe American businessman with the toolshe needs to communicate quickly,privately, and economically with officesand correspondents located in other partsof the world. Teleprinter leased channels(Fig. I) arc private communications cir-cuits which are interfaced at an RCAGlobcom Central Telegraph Office, andare designed to accommodate a sub-scriber's specific communications trafficrequirements. Availability, speed,reliability, and privacy at low subscriberrates are a few of the features which have

Reprint RE -21-3-21Final manuscript received January 30, 1975.

contributed to the extensive growth of theleased channel service.

Changing systemrequirements

In the past. implementation of leasedchannel interfaces to suit subscriber re-quirements presented no significantproblem. However, due to the growinguse of sophisticated communicationsterminals and computer controlled datasystems, interfacing requirements havebecome increasingly complex. Today'sleased channel traffic varies among usersfrom simple teletypewriter messages tocomputer -to -computer communications.

Authors (from left) Lippman. Russo, and Longo discussing the U.S. - overseas link using a typical leasedchannel system.

60

SUBSCRIBER'S

DOMESTIC

OFFICES

TTY

WASHINGTON

LOCAL

LINESRCA

CENTRAL

TELEGRAPHOFFICE

GPDC

NEW YORK

TTY

BOSTON

CHANNEL

CONTROL

ANDINTERFACE

EQUIPMENT

CABLEOR

SATELLITE

RCA LEASED CHANNEL

OVERSEAS I LOCAL

CENTRAL I LINESTELEGRAPH

OFFICE

T T Y TELETYPE TERMINALSGPDC GENERAL PURPOSE DIGITAL COMPUTER

CHANNEL

CONTROL

AND

INTERFACE

EQUIPMENT

SUBSCRIBER'S

OVERSEASOFFICES

ITT

LONDON

GPDC

GENE VA

Fig. 1 - Typical RCA talaprintar lama channel configuration.

involving intricate handshakingprocedures and line protocols. In addi-tion, functional demands such as codeconversions, speed changing, messagewithholding, line switching, message

retrieval and playback have become morecommonplace. Code conversion becomesnecessary to convert subscriber networksusing the domestically common 8 -bitASCII code to the 5 -bit internationaltelegraph code used almost exclusively oninternational links. To reconcile timingdifferences in data or baud rates within agiven network, speed changing factorsmust be considered. Messagewithholding techniques, typically thestore and forward approach, are used toensure the receipt of complete cohesivemessages. In order to address messages toa specific link within a multiple linksubscriber network, line switchingcapabilities must be included. Innetworks where circuit path verificationis essential, playback circuits must beemployed.

Presently, subscriber requirements aremet by adapting and integrating availablefunctional "blocks" into a custom -designed system. Often a system com-prises as many as five different "blocks".This approach is costly and usuallyresults in long implementation lead times.When subscriber requirements exceed thecapabilities offered by the availablefunctional blocks, an intense engineeringprogram is required involving systemdefinition and design, partsprocurements, breadboarding, fabrica-tion and debugging, installation, and

TTY

AMSTERDAM

finally, on-line testing. Each task con-sumes a substantial amount of time andmoney.

Projections of future communicationdemands, even by the most conservativeestimates, indicate a continuation of ex-plosive growth both in volume andsophistication. This presents a challengeto RCA Globcom, to effectively meet thegoals of efficient, reliable and flexibleleased channel subscriber service, in acost-effective manner. A stored -programapproach to system design appears to bethe long-term answer to this challenge.

Stored -program approach

The stored -program approach to datacommunications system design is notnew. The past several years have

witnessed a large and ever-increasingnumber of minicomputers and largerprocessors dedicated to the implementa-tion of a variety of data communicationfunctions. To date, however, the use ofcomputers has been relegated primarilyto medium size and larger systems wherecomplex data communication re-

quirements justify reasonably large in-vestments in hardware and software.However, in many low -end applicationsthe high cost of minicomputers and theirassociated peripherals cannot be

justified.

The advent of low cost LSImicroprocessors and mass storagedevices (e.g., floppy discs) is having asignificant impact on the design of new

low -end data communications systems. Amultitude of systems, that until recentlywould have required a hard -wired logicimplementation, can now realize themany advantages of the stored programapproach. These advantages includelower cost, flexibility, improved reliabili-ty, ease of maintenance, and the additionof many new system functions which, dueto rigid time schedules and cost factors,are impractical or impossible to im-plement via hardwired logic.

A feasibility model of a leased channelsystem with RCA COSMACmicroprocessor control has been im-plemented and exhibited at the 1974International Communications Associa-tion show in New Orleans, La.

Microprocessor -systemconcept

I he microprocessor -based systemenables designers to provide software thatis best suited for the particular functionsto be performed using standard LSIchips. The heart of the microprocessorsystem is the central processing unit(CPU) and its addressing logic.

rhe microprocessor is capable of per-forming logical functions under the con-trol of software instructions. Closely tiedto the microprocessor is the memory unit,capable of storing data and programmedinstructions. Memory contents can beused to identify inputs on multiplexers,enable counters, select ALU (arithmeticlogic unit) functions or perform othercontrol functions associated with thedesign. The rest of the system is made upof peripheral units. Devices such askeyboards, teletypewriters, tape readers,CRT displays, disc memories, and evencommunications links, are all consideredto be peripherals, when they are con-nected to the processor. Data flowsbetween the processor and theperipherals over a data bus. Individualbinary data bits travel on this bus ingroups called bytes.

Peripheral interfaces are necessary toconvert data from the processor formatto one that is acceptable to the peripheraldevice, and also to perform the requiredconversion from peripheral to processordata formats. The interface alsoreconciles timing differences and relaysprocessor instructions in the form ofcontrol signals to the peripheral. Flags

61

INTERRUPT LINE

ADDRESS BUS

COSMAC RANDOM ACCESS

PROCESSOR MEMORY (RAM)

MIMPIIMIrAII AIMPAIMPDIMiiIIMPAMPIEPIPDATA OAS

CONTROL BUS

COMMUNICATIONSINTERFACE(DOMESTIC)

MODEM

DOMESTICCOMMUNICATICMS

LINK

JCYCA_E STEALLINES

AMIMPAIMPINIP

A 4 ILINE

COMMUNICATIONS FLOPPY DISKINTERFACE INTERFACE

(INTERNATIONAL)

INTERNATIONALCOMMUNICATIONS

TERMINALEQUIPMENT

TVINTERFACE

TV NE Y BOA RD

Fig. 2 - Microprocessor -based international leased line communications system.

(usually flip-flops) in the interface are setto inform the processor of significantcurrent peripheral conditions. Interruptsignals generated by the interface forcethe processor to take immediate actionwhen the peripheral must have priorityservice.

Microprocessor -basedleased channel system

The RCA COSMAC leased channelmicroprocessor system, as illustrated inFig. 2, was developed at RCALaboratories and consists of a two -chipmicroprocessor, 4096 -byte RAM, andperipheral interfaces. Each of theperipheral interfaces serves to connectdifferent devices to the system. Incomingmessages enter the system through com-munications interfaces. Here themessages are converted from a stream ofbits into characters, each contained in an8 -bit data byte. These bytes aretransferred, one at a time, into thesystem's random access memory ( RA M).When 232 bytes accumulate, they form ablock of data. The data block is thenmoved into the floppy disc memory,where it is held until needed forretransmission. Outgoing data blocksmove from the disc to the appropriatecommunications line via the RAM. TheRCA COSMAC microprocessor con-trols this sequence of events withprograms written to satisfy subscriberrequirements of the overall store -and -forward' communication system.

The microprocessor makes both a databus and a control bus available to theperipherals. These buses carry most of theinformation that flows between the

processor and the peripherals. Sinceseveral different interfaces are connectedto these buses, there must be a clear wayto indicate which interface is permitted tobe active at any given moment. ASELECT instruction performs thisassignment function. Each interface hasits own unique selection number. Forexample, a SELECT instruction togetherwith the number 08 on the data bus, willactivate the floppy disc interface. Once aninterface is selected, it is free to act onfurther processor instructions.

Three special lines allow the peripheralinterfaces to initiate system actions,without first getting permission from theprocessor. By using the INTERRUPTline, the communication interfaces de-mand immediate handling of incomingdata as it arrives on the communicationslinks, and an immediate supply of outgo-ing data from the RAM, as it is needed fortransmission over the links. With theCYCLE STEAL lines, the floppy discmemory and tv displays gain direct accessto the RAM so they can write into thememory, or read from it, withoutsoftware instruction. Finally, via theLOAD line, the system can be reset andrestarted using a disc -stored programafter a catastrophic failure or loss ofpower.

Communications interface

When a communications interface hasreceived an incoming character from itscommunications link, it raises themicroprocessor interrupt line. At thesame time, this unit raises an external flagto indicate that a received character is

available. At the microprocessor, theinterrupt causes the ongoing program tobranch to a special software routinedesigned to service interrupts.

The system allows up to four externalflags (El -s) for each peripheral interface.Each of these flags or combinations offlags, when activated, indicates a requestfor attention by a specific interface. Forthe communications interfaces, EF4 is setin conjunction with EF I, if the receivedcharacter is erroneous (bad parity). Whenan interface is transmitting data, EF2 isset to indicate that the next character canbe transferred. Finally, EF3 is used it'indicate special conditions, such as ab-normal communication linecharacteristics.

While a given interrupt is being serviced,all other interrupts must wait their turn.Priorities for servicing interrupts areestablished in the processor's softwareinterrupt routine. For example, thedomestic communications interface is

always selected first by the routine.Domestic data rates are usually higherthan international rates, and therefore,the penalty for keeping the domestic linewaiting is greater. Likewise, read in-terrupts are always given priority overwrite interrupts, because failure to readmay result in loss of data, but the worstpenalty for failure to write is time lost onan idle transmission line.

Cycle stealingfor more efficient operation

The key feature of the floppy disc inter-face is its direct access to RAM memory,

62

without need for detailed microcomputerprogram control. Using this directmemory access ( DMA) feature (built intothe COSMAC processor) the disc can putdata bytes into the RAM or take themout, without receiving even a SELECTcommand. In fact, it can transfer this datawhile the processor is occupied with othertasks, such as talking to a com-munications interface. The directmemory access mechanism used here iscalled cycle stealing. There are normallytwo microprocessor cycles for each

program instruction; a fetch cycle,followed by an execute cycle. When thecycle steal line comes up during a fetchcycle, the processor will complete thatfetch, and the corresponding executecycle, and then hold its breath for a one -cycle interval before moving on with thenext instruction, fetch cycle. It is duringthese stolen one -cycle intervals that databytes are moved between the RAM andthe disc.

Before the cycle stealing can begin, a.DMA address register must be loadedwith the first RAM memory address ofthe data block to be read from, or writtenon, the disc. Then the register is

automatically incremented at eachsucceeding stolen cycle, until an entireblock of 232 data bytes has been

transferred. The processor sees only aslight slowdown, usually less than a 1 -

percent reduction in the time available forongoing program activities.

As a convenience feature, the disc stores abootstrap program that can restart theentire system after a power loss, or whensome other condition puts the system outof service. Any other program residing onthe disc can then be loaded into the RAMusing this bootstrap program, thuseliminating the need for auxiliaryprogram load devices such as cassettes orpaper tape, and greatly simplifyingsystem recovery.

Less vital to system operation than thefloppy disc, but still significant from amaintenance viewpoint, is the tv display.

TV display diagnostics

The tv can display text indicatingcommunication -link fault conditions,and other system status conditions.However, experience has shown that itsmost useful function is to display memory

patterns; bit patterns on the data bus; andother information for diagnosis, test, andmaintenance of the system. A standard tvset is used for the display. The display iscontinuously refreshed from 128 bytes ofRAM memory. Like the floppy disc, thetv display uses the DMA (cycle -stealing)capability of the system. Every 60th of asecond, the tv interface interrupts theprocessor and asks for new information.The interrupt routine then points to thebeginning of the 128 bytes of RAMmemory that contain the tv display infor-mation. These data are then sent to the tvinterface on a cycle -stealing basis.

The lowest -priority peripheral in the

system is the manual keyboard. Thisdevice is used to enter data bytes (in thehexadecimal code internally used by thesystem) into the RAM memory thusfacilitating debugging and programmodifications when required.

Summary of advantages

In addition to its low hardware cost, amicroprocessor -based leased channelsystem will provide the followingfunctions and features:

Rapid implementation of subscriber re-

quirements

Message priority queuing

Message retrieval capability

Increased traffic volume capability Programmable system parameters

data rates

character formats & codes

character expansion

playback /answerback

call directory codes

transmit select codes

Maintenance

single unit backup

fault isolation

- rapid system recovery

Ongoing development

At RCA Laboratories, a new COSMACbased leased channel control package iscurrently being developed. In this system,the approach taken will allow minimumcost implementation of present low speedleased channel applications and flexibili-ty to accommodate the anticipated highspeed performance requirements of thefuture. For low speed applications, thebulk of the floppy disc control functionsare relegated to software, resulting in

reduced controller hardware and relatedcosts. The extended high speed capabilityis provided by the addition of a secondCPU chip which is dedicated to disccontrol. The addition of a second CPUresults in a powerful parallel processingarrangement which relieves the excessivedisc control burden from the main CPUthus providing additional processingtime. This additional time may be used toimplement sophisticated disc manage-ment schemes, preprocess data, or tocontrol additional peripheral devices.

The future

The advent of low cost LSImicroprocessors will bring about a

profound change in the architecture ofthe next generation of data communica-tions systems. Microprocessor -baseddata communications, typified by theleased channel system described in thisarticle, will begin to emerge within thenext few years.

Many other microprocessor -based datacommunications systems are now beingdeveloped. Intelligent multiplexers buf-fers, concentrators, interface messageprocessors, and/or any combinations ofthe above are but a partial list. Thesesystems should stimulate the develop-ment of new service offerings to a com-munications dependent world com-munity.

In addition to the references andbibliographic information cited below,detailed COSMAC information is

available from R.O. Winder, Head, LSISystem Design Research, RCA SolidState Technology Center, Somerville, N.J. 08876 (telephone 201-722-3200, exten-sion 6005).

COSMAC related references

I. Winder. R.O.; -The COS MOS Microprocessor". 1974WESCON, Los Angeles, CA (Sept 1974).

2. Welshes:1km. J.; -A Simplified Microprocessor Architec-ture". IEEE Computer ( March 1974) pp. 41-47.

3. Swaks. N. and Weisbecker. J.; -COSMAC AMicroprocessor for Minimum Cost Systems". IEEEINTERCON (March 1974).

4. Russo. P.M. and Lippman, M.D.; "A MicroprocessorImpknr.:ntation of a Dedicated Store -and -Forward Data

Communications Systems". 1974 NCC, Chicago. IL (May1974).

S. Lippman. M.D. and Russo. P.M.; -A MicroprocessorController for International Leased Data Channels". /CC74. Minneapolis. MN

6. Russo. P.M. and Lippman. M.D.."Case History: Store andForward". IEEE Spectrum (Sept 1974) pp. 60-67.

7. Welshes:ker. J.. -A Practical, Low Cost. Home, SchoolMicroprocessor System". IEEE Computer. ( Aug. 1974) pp.20-31.

63

COSMAC a microprocessorfor minimum cost systemsN.P. Swales J.A. Weisbecker

COSMAC. a microprocessor developed by RCA. is a COS/MOS LSI processor designedfor use as a general purpose computing element. The processor architecture described inthis paper has been developed to provide maximum flexibility for low-cost computer -based systems. The COSMAC instruction set and input/ouput interface have beendesigned to minimize memory requirements and system complexity. Experience with thisarchitecture has verified its usefulness over a wide range of potential applications.

M ICROPROCESSORSI are becom-ing an important tool for logic andsystems designers. Although they willhave some impact on the low perfomanceend of the minicomputer market,' theirmajor successes are being scored in con-trol function applications wheremicroprocessors are being used to replacecomplicated switching functions whichwere previously realized with discretedigital logic.' The low cost, flexibility,

and fast design cycle time which thesedevices provide are making them in-creasingly popular with designers. A

Norman P. Swales, Systems Planning, Palm BeachDivision, Palm Beach Gardens. Florida, received the BSfrom Cornell University in 1970. Prior to graduation, heworked at General Radio co. as a co-op student. Upongraduation. Mr. Swales joined RCA at Palm BeachGardens. With RCA he was responsible for RCA Series400 Miniprocessor logic design and COSMAC LSI logicdesign. He is presently responsible for the systemsplanning of the Dataway Hotel/Motel Data ManagementSystem.

Reprint RE -21-2-14Final manuscript received June 24. 1974.A version of this paper was published in the 1974 IEEEINTERCON Convention Record.

microprocessor, coupled with a smallamount of semiconductor memory and afew inexpensive peripheral devices, canprovide a cost effective system forapplications where the use of computertechnology was previously unthinkable-- home entertainment, automobile con-trol, and educational and businesssystems to name a few.

Microprocessor design

Basically, a microprocessor is a devicecapable of performing arithmetic, logical,

Joseph A. Weisbecker, Member, Technical Staff, LSISystems Design, Solid State Technology Center. RCALaboratories, Princeton, New Jersey, received the BSEEfrom Drexel University in 1956. He has been developingnew computers for almost 20 years together with com-munications terminals, input/output devices, and com-puter related commercial games. He has held engineer-ing, advanced development, and product planningpositions within RCA's Computer Systems Division priorto joining RCA Laboratories in 1970. Mr. Weisbecker haswritten several articles, received two RCA LaboratoriesAchievement Awards, and holds 22 patents with otherspending. He is a Senior Member of IEEE and a member ofEta Kappa Nu.

and decision -making operations underthe control of a set of instructions stored,either temporarily or permanently, insome memory device. It is capable ofcommunicating with a set of peripheraldevices via some defined input/ output(I/O) structure. Its operation is slowwhen compared to large computingdevices, such as miniprocessors, but it isimplemented on one or a few monolithicintegrated circuit chips, and it is notexpensive.

The above provides a geneial frameworkinto which a microprocessor should fit.There are, however, other stringent con-straints on the design of a usefulmicroprocessor. The basic designproblem is to develop a simple, butflexible, processor architecture which canbe used to realize inexpensive systems. Itis important to minimize the complexityof the processor itself, with respect toboth internal logic and the requirednumber of external connections, so thatthe device is easy to produce and package.The architecture should possess anefficient I/O structure to help reduce thenumber of circuits required to interfacewith external devices and to help increasesystem performan& In addition, thearchitecture should provide for efficientuse of main memory storage.

It is these latter considerations whichhave led to the development of theCOSMAC microprocessor.

COSMAC architecture

An eight -bit, parallel, register -orientedarchitecture was chosen for COS M AC asbeing best suited to the requirements ofoptimizing performance, memory usage,and processor complexity. An eight -bitmachine provides sufficient width toeffectively manipulate the standard codeand data units of a majority of thecommunications and informationprocessing fields while providing asignificant performance advantage overbit serial and four -bit machines. A 12 -bitmachine would have provided more per-formance but it would also have causeddifficulties when addressing more than4K words of memory, and it would havecaused inefficiencies when manipulating8 -bit data. Also, both I2 -and I6 -bit

machines suffer from the disadvantagesof requiring more logic and more I / 0pins both of which are inconsistent

64

with the desire to minimize the chip arearequired by the processor. A register -oriented structure was chosen to provideconvenience in implementing programsutilizing interpretive subroutine codingtechniques, for macro programming, aswell as the ability to efficientlymanipulate data when programming inthe machine language and to effectivelyimplement foreground/ backgroundprocessing using the processor's interruptfacility.

A simple two-step fetch and executesequence was selected for the basic

machine cycle and considerable emphasiswas placed on the I/O interface. A totalof twenty-three lines, including an eight -bit bi-directional data bus are used tocontrol the I/O. In addition to a datatransfer capability, these twenty-threelines provide internal processor stateinformation, an interrupt capability, adevice sensing capability, an 1/O com-mand code modifier, and controls for abuilt-in, cycle stealing, direct access I/Ofacility.

The block diagram of Fig. I shows thegeneral architecture of themicroprocessor. The register matrix is anarray of sixteen I6 -bit registers whichmay be addressed by the P, X, or Nregisters. The I, P, X, and N registers areall four bits in width. The 1 and Nregisters are used to hold the instructionfetched from main memory; the contentsof the 1 register determine the genericinstruction type to be executed, and.depending upon the contents of the I

register, the contents of the N register areused to select one of the matrix registers,to control the I/O devices, or to providefurther definition of the instruction to beexecuted. The contents of the P registerdetermine which of the 16 matrixregisters is being used as the currentprogram counter. The X register is usedto address the register matrix to fetch theaddress of memory operands for certainmemory reference instructions. The Tregister is an eight -bit register used tostore the contents of the P and X registers-whenever a program state change occursin response to an interrupt. The A registeris a I6 -bit register used to temporarilyhold the data fetched from the registermatrix. A 16 -bit increment/decrementnetwork is used to update informationfetched from t' e register matrix. Oneeight -bit multiplexer is used to gate thecontents of the A register to the eight -bit

INCREMENT/DECREMENT

'A" BUSMUX

MEMORYSYSTEM

MAR MUX

REGISTER

REGISTERMATRIX

RI RO

1 tit o

8 BIT INTERNAL DATA BUS

INPUT/OUTPUTINTERFACE

Flu. 1 - COSMAC Internal architecture.

memory address bus and a second mul-tiplexer is used to gate the contents of theA register to the eight -bit, bi-directionaldata bus.

The I) register is an eight -bit accumulatorwith associated zero decode and carry. orlink, indicator which may be interrogatedwith the branch instruction. I he AI.11 isan eight -hit parallel arithmetic and logicunit capable of performing binary add.and subtract, logical and, or. exclusive or.

and shift operations. One operand is

TIMINGANDCONTROL

D

4

00

4O

U

contained in. the I) register and the otheris contained in memory and present onthe data bus. I he add, subtract, and shilloperations may modify the carry in-dicator.

As shown in Fig. I, the COSMACmemory system shares themicroprocessor I/0 interface with thesystem peripheral devices. Although thememory address is sent to the memorysystem separately, data is transferredbetween the processor and memory viathe I; 0 data bus.

As we go to press...

On October 3, the Solid State Division announced the commercial availability of theCDP1800 microprocessor family, including the CDP1801 CMOS 8 -bit microprocessor,the "Microkit" hardware support kit, microprocessor manuals, and software develop-ment packages. This commercial version of COSMAC is presently available from stock.with the CDP1801 full -voltage (15V) chips available at $56 and the CDP1801C 5-V chipsat $40 (quantities of 1-99).

The "Microkit" hardware support kit contains the CPU, 1-k words of RAM, 512 words ofROM, space for additional memory and user -designed interface cards, input/outputdecoders, an input/output interface for a teletype or other terminal, and power supply. A"Microkit" with a resident editor, assembler and debug board option with a user -suppliedterminal provides a complete, independent system for producing debugged programs.

More powerful software development aids are available that offer assembly, editingsimulation, and debugging. This program is available either on the General Electrictimesharing network or as a Fortran IV tape for installation on an interactive computer. Amicroprocessor manual, describing the CDP1801's architecture, instruction set, I/Ointerfacing, and programming techniques is available as well as manuals on the varioussoftware design aids.

Futher information may be obtained by writing RCA Solid State Division, Box 3200,Somerville, New Jersey 08876, or by calling William J. Dennehy, MicroprocessorMarketing Manager, (201) 685-6713.

65

COSMAC instruction set

A notation convention has beendeveloped to describe the operation of theCOS MAC micropocessor and will nowbe presented in order to abbreviate thedescription of its instruction set. R will beused to designate one of the matrixregisters; R I will designate the mostsignificant byte (MSB) of R, and RO willdesignate the least significant byte (LSB)of that same register. R(N), R(X) andR(P) will be used to designate the matrixregister specified by the N,X, and Pregisters, respectively. For example,R I (N ) represents the MSB of the matrixregister specified by the N register andRO(N) represents the LSB. Similarly, Mwill designate the contents of a memorylocation, and therefore, M(R(X)) willdesignate the contents of the memorylocation addressed by the matrix registerspecified by the X register. As an exam-ple, M(R(N)) D; R(N) + I, describesthe memory transfer to D instruction(instruction code (I) = 410. The contentsof the memory location addressed by thematrix register specified by the N registerare transferred to the D register and thecontents of the matrix register are in-cremented by I. Using similar notation,the COSMAC instruction set is describedin the table of Fig. 2.

All the COSMAC instructions use thesame fetch and execute cycle sequence.During the fetch cycle the four -bit ad-dress contained in the P register is used toselect the matrix register which has beendesignated as the current programcounter. The contents of the selectedmatrix register are gated into the Aregister and are then sent to the memorysystem via the memory address mul-tiplexer. The contents of the A registerare incremented by one in theincrement/decrement network, and theresult is stored in the matrix registerspecified by the P register. Finally, thecontents of the addressed memory loca-tion are gated into the I and N registersvia the eight -bit bi-directional data bus.In the notation defined above, this opera-tion would be written asM(R(P)) -I, N; R(P) +I. During theexecution cycle of the instruction, thedigit contained in the I register is decodedand the instruction is executed asdescribed in Fig. 2.

Certain of the microprocessor instruc-tions require further explanation otherthan that given in Fig. 2.

INSTRUCTIONINSTCODE)HEX)

FUNCTION

inINCREMENT

..1

REGISTER 1 RINI I

DECREMENTREGISTER 2 RINI - 1

RO to D II ROINI +O

RI to D 11 RIM) -.D0 to RO A I) -. NOON

0 to RI 11 D -.4111N)

00 to ROO C 00+ ROOM

IDLE 0 1011; MINIM)) -.BUS

MEMORY TO D 4 MOURN -.DANN 1

D 70 MEMORY S 0 -.MININII

LOAD P 0 14 -e.P

LOAD x E NI -. X

CHANGE STATEAND RESETINTERRUPTMASK 70 MIRIXII -. X.P;RIX11:FIESET IM

CHANGE STATEAND SETINTERRUPTMASK 71 MIRIXII .-. R.P1201BET IM

SAVE PREINTERRUPTPROGRAM STATE 76 T -4. MIRIXII

INDEXEDMEMORYTRANSFERTOO FO MIRIXII +0

OR Fl BMX)) 0 - DAND F2 MIRIXII - 0 -4.0

EXCLUSIVE OR F3 MIAOW O-4. D

ADD F4 MIRIXII PLUS 0 -.). CI

SUBTRACT FS PAM XII MINUS 0 -0- CI

SHIFT RIGHT FS SHIFT 0. 1BR --. DF

REVERSESUBTRACT F7 D MINUS MINI/01 -.0

DATA IMMEDIATETRANSFER TO D Fll MIRIPII -.17RIPI 1

OR IMMEDIATE F9 MIRIPII 13 -.0 ORIPI 1

AND IMMEDIATE ' FA M IRIPII D -..D.RIPI 1

EXCLUSIVE ORIMMEDIATE FB MI RIP)) (II D - 0.111P) 1

ADD IMMEDIATE FC MIRIP)) PLUS 13 -1. DRIP) 1

SUBTRACTIMMEDIATE FD MIRIPII MINUS D -.D.RIPI 1

REVERSESUBTRACTIMMEDIATE FF D MINUS MIRIPII - DRIP) 1

COSM AC

INSTRUCTIONINST.CODEIHEXI

TESTFIELD.IHEXI

FUNCTION

TEST ANDBRANCH 0

2

3

4

5

6

7

C

MIAMI) - ROW)

MIRIPI I - ROIPI IF D d 0/111111

-.ROIPI IF D 0/111111

MIRIPII -00 RCMP) IF CIF 1/RIP).1

IF EF1 1/RIPI1

PAIRIPII - ROIPI IF EF2 1/1111.1.1

TAIRIPII - RO)P) IF EF3 1/111111.1

TAIRIPII -..ROIPI IF EF4 1/RIPI1

RCMP, I (SKIP)

MIAMI! --.R131P) IF CIF 0/R1140

MI RIP)) .-.R131P1 IF EFT. 0/RIPI1

MIRIPII -.ROW) IF EF2 0/111P11

401101)) -a. ROM) IF EF3 0/RIPI1

MIMI)) -.ROIPI IF EF4 0/1111.1.1

Unwed TEST CONDITION SHOULD BE CONSIDERED ILLEGAL.

1/0 TRANSFER 6

2

3

S

7

A

C

S

MIRIXII - I/0;RIX) I

MIRIXII -41/0:RIX) 1

MIRIXII -. I/0411X) 1

MIRIXII - 1/0.81 XI 1

MIRIXII -.1/0.1%1X) 1

MIRIXII - 1/0711Xl I

MIRIXII - I/0;RIX) I

MIRIXII - I/O;RIXI 1

I/O -4MIRIXI/

I/O MIMI)I/O -.MIRIXII

I/O MIRIXII

I/O - MIRIXII

I/0 MIRIXII

I/O -.MIME)

I/0 MIRIXII

Fig. 2 - Instruction summary.

8 SIT B1 DIRECTIONAL DATA BUS

4 BIT FO COMMAND CODE MODIFIER

4 BIT DEVICE FLAG BUS

2 EXTERNAL TIMING PULSES

2 EXTERNAL STATE CODE INDICATORS

44

INTERRUPT REQUEST

DIRECT ACCESS INPUT REQUEST

DIRECT ACCESS OUTPUT REQUEST

Fig. 3 - COSMAC input,ouput Interlace.

INPUT/OUTPUT

DEVICES

66

The IDLE instruction (instruction code(I) = 016) is used as an instruction halt.Whenever this instruction is encounteredin the program flow, the contents of thememory location specified by the matrixregister specified by the N field aredisplayed on the I/O bus. The system willremain in the IDLE state until the receiptof an interrupt or direct access input oroutput request.

The DO to ROO instruction (1=C16) placesthe four least significant bits of the Dregister into the four least significant bitsof the matrix register specified by the Nfield and therefore may be used to imple-ment single -digit table look -upoperations.

The load P instruction (1=1316) causes thecontents of the N register to be

transferred to the P register, providing avery simple branch and link capability.

The save state instruction (I, N = 716, 8,6)helps provide the capability to store themachine's pre -interrupt state after aninterrupt initiated program state changehas occurred. The return instructions(I,N = 7,0 and I, N = 7, I) allow programcontrol of the interrupt mask bit as well asthe ability to change the contents of the Pand the X registers simultaneously. Theselast three instructions enable theprocessor to implementforeground/ background programmingin an interrupt driven system.

The data immediate instructions (I, N =F, 8 through F, F) provide the ability toeasily inject constants from the mainprogram flow into data and addressmanipulations. These instructions use thecontents of the memory location im-mediately following the instruction codelocation as one of the operands in thespecified operation.

The test and branch instruction' (1=316)tests the condition specified in the N field.If the condition is met, then tht contentsof the memory location immediatelyfollowing the instruction are placed intothe least significant byte of the matrixregister specified by the P register;otherwise, the next instruction in se-quence is executed.

Input/output

One area of major concern in anyprocessing system is the computer I 0interface. All of the peripheral devices in

a system must use this interface whencommunicating with the processor, and.therefore, the level of complexity and theefficiency of this interface have a greateffect on the overall cost and per-formance of any given system. This isespecially true in systems usingmicroprocessors where the cost of themicroprocessor represents a very smallportion of the overall system cost andwhere system performance is limited bythe speed of the processor.

In order to extend the useful operatingrange of the COSMAC microprocessor,considerable emphasis was placed on itsI/O structure. The processor interface,physically composed of twenty-threesignal lines (see Fig. 3), is capable ofsupporting devices operating in polled,interrupt driven, and direct access modes.The processor is equipped with a set ofvery flexible 1/O instructions, a built-indirect access I/O capability, and I/Ointerrupt line, four external flag in-dicators, a set of external timing pulses,and an eight -bit, bi-directional data bus.

The I/O instruction (instruction code (I)= 6,6) is used to control the I/O devicesoperating in the programmed mode. Ascan be seen in Fig. 2, there are actuallysixteen sub -instructions incorporatedinto the I/O instruction; eight of theseprovide for information transfer from theI/O devices to the processor memory;and, and the other eight provide forinformation transfer from the processormemory to the I/O devices.

During the execution cycle of the I/Oinstruction, the external state code (ESC)lines of the I/O interface asst.me aparticular state, indicating to the I/Odevices that a programmed mode datatransfer is to take place. The I/O devicewhich was last selected, using the deviceselect command (one of the eight dataoutput 1/OI/Osubinstructions), responds byeither placing data on the I/O bus or bytaking data from the I/O bus dependingupon the state of the four I/O "N"(I/ 0command code modifier) lines.

To avoid confusion on the I/O bus, onlyone device at a time should communicatewith the processor via the data bus. Toensure this condition, a device selectionconvention has been adopted for use inlarge systems. The 16 instruction with theN field (and therefore the I/O "N" lines)equal to 1,6 has been designated the selectinstruction. All devices in a system are

assigned a unique address. Whenever adevice detects the 16 condition on theESC lines and a value of 116 on the I/O"N" lines, it compares the informationpresented to the data bus by the processorwith its assigned address; if these twobytes are the same, the device becomesselected, and, if the two bytes differ,thedevice becomes or remains de -selected. Adevice may communicate with theprocessor in the program mode onlywhile it is selected.

The sixteen 16 instructions provide a verypowerful tool when designing controlelectronics units (CE's) to interfacebetween the processor and its I/Odevices. The action taken by any givenCE in response to any of the 16 in-structions is defined by the CE designerand may vary from CE to CE. Theseinstructions may be used to replace se-quencing logic in the CE's, to distinguishbetween command, status, and datatransfer requests, or to control multipledevices through a single CE. In short,they provide a flexible method of im-plementing simple I/O controlprocedures for small, dedicated systems.Large systems may be required to use allof the features provided on the I/Ointerface, but smaller systems can becreated using any subset of the I/Osignals and conventions.

Four external flag signals are provided onthe COSMAC interface to enable theCE's to quickly transfer status informa-tion to the processor. These signals maybe tested directly by the test and branchinstruction.

A single interrupt line is provided toenable any control electronics unit todemand immediate program service fromthe processor. 1 his line may be treated asa common interrupt bus or as a hardwirepriority daisy -chain interrupt facilitydepending upon the requirements of thesystem in which it is used. Whenever theprocessor detects an interrupt condition.assuming interrupts are not masked, itenters an interrupt response state at theend of the instruction which was beingexecuted when the interrupt was received.The ESC lines at the 1/ 0 interface assumethe interrupt state conditions indicatingto the I/ 0 devices that an interrupt isbeing honored. The contents of the P andthe X registers of the processor aretransferred to the T register so that thepre -interrupt state of the machine may besaved. Finally, a value of 11. is placed in

67

the P register and 21,, is placed in the Xregister. Normal instruction fetch andexecution is then resumed using R I as thenew program counter, effectively causinga hardwired branch and link to thesubroutine addressed by the matrixregister R I. The machine state instruc-tion (instruction code = 71,,, see Fig. 2)may be used to control the interrupt maskas well as to save and alter the state of theP and X registers.

A cycle -stealing direct access I/O facilitywas incorporated into the COSMACprocessor to provide a high-speed datapath between the I/O devices and theprocessor. Two of the I/O signals, InputRequest and Output Request, maybe used by the I/O devices to initiate adata transfer via this direct access

channel. Only one device at a time mayoperate in the direct access mode.

A direct access device must be selectedand activated in the programmed mode.Once activated, the device may initiate adata transfer by signaling the request tothe processor via the Input Request or theOutput Request lines. The processorresponds to the request by entering theDirect Access State after finishing theinstruction which was in progress whenthe request was received. The processorforces the ESC lines to assume the DirectAccess State condition to indicate to theI/O device that is processing the transferrequest. The CE places data onto the databus if an input request has been initiatedor removes data from the bus if an outputrequest has been initiated. The data isplaced into or removed from the memorylocation specified by the RO register of theregister matrix. At the end of the directaccess transfer, RO is incremented by onebyte so that the processor is ready to actupon the next transfer request. A CEneed not be in the selected state in orderto issue direct access transfer requests.The use of this channel, therefore, doesnot interfere significantly with programexecution or with the simultaneous use ofother programmed mode devices. Thischannel may be employed to com-municate with devices which have hightransfer rates.

A program load facility using the directaccess channel is provided to enable usersto enter programs into the COSMACmemory. This facility provides a simple,one step means for initially entering

programs into the microprocessor systemand eliminates the requirement forspecialized ROM's in main memroy tobootstrap user programs into the system.

Chip technology

COSM A( is presently implemented ontwo chips employing RCA's standardCOS/ MOS technology. Both chips werelayed out manually using standard celltechniques with computer -aided maskgeneration and checking. One chip con-tains the register matrix, theincrement/decrement network, the Aregister, and the A register multiplexersshown in Fig. I; the second chip containsthe remainder of the processor elementsshown in Fig. I. The register matrix chipis 236 X 246 mils and the arithmetic andcontrol chip is 256 X 254 mils. Both chipscontain approximately 3000 transistors.The COS/ MOS technology was chosenbecause it provides many features whichare advantageous in the design of inex-pensive systems. The two -chip processoris capable of operating with any supplyvoltage from 5 to 12 volts; this wideoperating voltage range enables directconnection to a variety of circuit types.Inexpensive, unregulated power suppliescan be used. The current drain on thepower supply is negligible - each chipdissipates only about 100 µW.

The operating temperature range of thedevices extends from -55 to +125°C.Most important, the inherent high noiseimmunity of COS/ MOS providesreliable operation even in hostileenvironments.

Considerable care was taken in the chipcircuit design to ensure that the finalproduct would be easy to use and tointerface. Only a single phase clock isrequired. The voltage required to drivethe inputs and outputs is not dependentupon the main supply voltage so that theprocessor can take advantage of the speedbenefits of operating at a high voltage,while the inputs and outputs may beoperated at lower, TTL compatiblelevels. Also, all registers in the machineare static, providing the ability to stop theclock generator for indefinite periodswithout losing information in theprocessor.

Although the COSMAC devices are new,future enhancements are already being

developed. It is anticipated that theprocessor will soon be implemented on asingle chip, and, the implementation of ahigh-speed version of COS MAC using asilicon on sapphire technology is present-ly under investigation.

Software and software support

No matter how convenient a computersystem is to implement in hardware, itcannot be considered easy to use unlesssome facility is provided to ensure thatthe system is easy to program. In order toprovide this facility, a complete machinelanguage assembler and simul-tor/debugger system were created andmade available on RCA's corporate time-sharing service. This interactiveassembler system provides the ability toeasily program the microprocessor usingthe COSMAC machine language or therepertoire of macro instruction sub-routines which were created to simplifythe programming of large softwaresystems. The capability is provided foron-line editing of source programs.

A standard Fortran version of the abovementioned assembler/ simul-ator/debugger is being made availablefor batch processing as well as for use onany IBM time-sharing operating system.

To prove the utility of the COSMACinstruction set, a number of experimentalsystems have been designed andprogrammed. The applications whichhave been studied include word process-ing, educational and calculator functions,entertainment systems, and com-munications and real-time device controlsystems. All of the applicationsprogramming done to date, requiringmemory sizes ranging from I K to I6Kbytes, have indicated that the COS MACinstruction set does make efficient use ofmemory and that its processing speed issufficient to handle a wide variety ofprocessor applications.

Hardware and typical systems

I o facilitate the breadboarding ofpotential systems, a number of standardbuilding block devices and control elec-tronics units have been designed.Processor boards providing a full TTLinterface and up to 24K bytes of memoryhave been implemented. I/O devices and

68

COW :C

COSMAC

\PUT (IllTPCI iNTEHTACT

SHIFT1REGISTER

CE

TWO AUDIO CASSETTEDRIVES

MEMORYSYSTEM

FLOPPY DISCCE

KEYBOARDCE

HEXKEYBOARD

COMMUNICATIONSCE I

TYPEWRITERCE

I/C)

TYPEWRITER

COMMUNICATIONSCE II

COMMUNICATIONS LINES

Fig. 4 - Two typical COSMAC systems.

their associated control electronics whichhave been built include I/O typewriters,tape cassettes, floppy discs, dot matrix tvdisplays, video data terminals,keyboards, and various communicationscontrollers for teletypwriter equipmentand acoustic coupled data terminals.

Fig. 4 illustrate two typicalmicroprocessor systems using some of theabove mentioned hardware. Fig. 4a

shows a word processing system employ-ing the COSMAC processor and 4Kbytes of main memory storage. Thesystem uses inexpensive audio cassettetape recorders as mass storage units aswell as for voice system operating in-structions. The shift register CE is used asan intermediate storage device for on linedata manipulation. The hexidecimalkeyboard is used for entering initializa-tion parameters into the system, and theI/O typewriter is used as a hard copy,manual input/ output device. The systemhas been programmed to generate andedit form letters for storage on the tapedrives as well as to process the formletters using a recorded mailing list.Programs have been generated to processpayroll information and to printpaychecks. And, an inventory control

and accounts receivable processingsystem has been investigated.

Fig. 4b illustrates a leased channel com-munications control system which is

currently being developed. The system,consisting of a COSMAC processor withmemory, a floppy disc, and two com-munciations controller, was designed for"turnkey" operation. Both com-munications controllers are capable ofoperating in half or full duplex modes.The system is capable of performing theanswerback and playback operations re-quired by the telecommunicationsnetwork line procedures as well as codeand speed conversion. The disc unit isused to provide non-volatile storagespace for a message store and forwardfeature and for the storage of allprograms. A message forwarding priorityweight may be assigned to all messages sothat the sequencing of the forwardedmessages is independent of the messageinput sequence.

All of the experimental systems whichhave been developed to date using theCOSMAC microprocessor have shownthat it can be used to effectively imple-ment low-cost data processing systems.

Conclusion

The COSMAC microprocessor is aneight -bit, parallel, general purpose com-puting element designed for use in theimplementation of low-cost digitalsystems. Every effort hai been made tomake it easy to program and inexpensiveto interface.

The COS/ MOS technology with whichthe IS1 processor is implementedprovides a number of features which areimportant in the design of low-costsystems. COS/ MOS provides a highnoise immunity, so, the processor canoperate in electrically hostileenvironments and can be powered byunregulated power supplies. Theprocessor has a wide operating voltagerange and the internal voltage supply isseparated from the I/O voltage supply sothat the processor may operate at max-imum speed while interfacing to variousexternal circuit technologies, includingTTL. Only a single phase system clock isrequired; and, the processor power con-sumption is minimal.

COSMAC posses a built-in matrix ofsixteen I6 -bit registers and a uniqueinstruction set chosen to make efficientuse of main memory. The register matrixmay be used to provide multiple programcounters as well as address and datastorage. Unlimited subroutine nesting ispossible. and the instruction set facilitatesthe use of interpretive subroutine macroinstructions. A large amount of supportsoftware has been generated to aid theuser in programmer and debugging hissystem software.

The COSMAC I/O interface wasdesigned to provide intimate control ofI/O devices so that overall system com-plexity and cost can be reduced. A directaccess I/O capability is included in theprocessor structure to enable the high-speed transfer of blocks of data withoutprogram monitoring.

In short, COS MAC has been designed tohelp minimize the cost of intelligentdigital systems.

ReferencesI. Lapidus. t, "MOS LSI Launches the Low -Cost

Processor.- // / Spec/ruin (Nov. 1972) pp. 33-40.

2. Jackson, B.F..; -Designing A Microprocessor -BasedSystem.- SplentS Engineering hiday(Dgc. 1973) pp. 62-67.

3. Reyling. G.. Jr.; -Microprocessors: The Next Generation inDigital Design?' SysietnA Engineering Today (Nov. 1973)pp. $6.90.

69

Algorithm for comparison ofcommand and control systemsJ.R. LoPresto

In the analysis of Command and Control (C-) systems, a comparative analysis of theperformance of various configurations is important for obtaining cost effective. reliable.and technically adequate systems. This article presents one algorithm that can beemployed for such an analysis. Using one of the candidate system configurations as astandard. the result of the algorithm is to obtain relative quantitative indices whichmeasure C' performance with respect to the execution of given mission tasks. The articleincludes a non -mathematical description of the rationale used in the algorithm derivationas well as a sample calculation to illustrate its use.

A COMPARATIVE configurationanalysis is important for obtaining costeffective, reliable, and technically ade-quate system configurations. Thealgorithm described in this paper can beemployed for such an analysis; in its usethere are no claims of perfection, totalobjectivity or infallibility. Although thisalgorithm was originally conceived tocompare military C- systems. there is ananalogy to commercial process control ormanagement information systems. Theexpansion of the model to these areas iseasily implemented.

J.R. LoPresto, Software Design and DevelopmentEngineering. MSRD, Moorestown. N.J.. received the BSin Mathematics and Physics from St. BonaventureUniversity and the MS in Applied Statistics fromVillanova University. He had been associated with RCAfor 14 years following prior association with FordInstrument Company and Martin -Orlando. His principalresponsibilities at RCA involved the proposal and ad-vanced study and development phases of many largeCommand and Control systems including BMEWS.Sam -D. AABNCP, Cobra Dane, AEGIS. and DCDRS.

Since this paper was written, Mr. LoPresto has left RCA.

Reprint RE -21-3-28Final manuscript received November 5, 1973

The algorithm compares the performanceof mission execution for various con-figurations, where performance includesassociated factors such as reliability,availability, etc. The algorithm does notcompare technical risk, schedule risk, lifecycle costs, or other factors which, inaddition to performance, are also impor-tant in the analysis of C- systems.

The result of the algorithm is a set ofquantitative indices which measures theC2 performs nce with respect to the execu-tion of given mission tasks. By selectingthe best system as a reference standard.viz., with an index of 1.0, then as a resultof the algorithm computation the systemswhich are compared produce indices lessthan or equal to 1.0.

The assumptions under which thealgorithm can be used are:

l) The mission/ mission phases are clearINdefined.

2) The functions to b.. executed in each missionphase are well defined. e.g.. as a result offormal military standard functional flowdiagrams and documentation (F'D')analysis.

3) No political or customer bias is involved inthe algorithm computation.

The algorithmic model is based on crosscorrelation involving the following fac-tors:

I) Importance of each functional task duringeach mission phase.

2)Contribution of the various subsystemstoward the execution of each functionaltask.

3) Degradation of contributing elements, interms of:

a) loss of external communication systems.

h) failure or degradation of subsystems.

c) smaller allowable operational status in thereduced configuration, or

d) loss of external information due to theloss of external data sources or degradedreporting procedures.

4) Mutual degradation effects between theabove contributing elements.

The weights and values attributed to thesefactors are, in many cases, valuejudgments by the analyst and not basedon hard and fast known information. Thesensitivity of the algorithm to some ofthese value judgements varies. For exam-ple, the weight given to each functionwithin a phase does not have as great aneffect as changing the form of thealgorithm expression. Special re-quirements may require specialprocedures in the definition of tasks forthe algorithm use. For example, specialreaction time to an extraordinary threatmight require the redefinition of afunctional task into two functions, onewhich measures reaction against a "nor-mal" threat, and one which measuresreaction against this extraordinarythreat.

The remainder of this paper contains abrief functional description of ageneralized command and control systemthat will be used to clarify the algorithmicmodel description, a non -mathematicaldescription of the rationale of thealgorithmic form, and an example toshow how the algorithm may becalculated.

Composition of acommand and control systemA command and control system is

described as a system which senses thedata that reflects an environment or atactical situation, evaluates this data withrespect to a specified mission, formulatesa reaction to this environmental situa-tion, and initiates and controls this reac-tion. It then repeats this process until themission has been accomplished or it canno longer operate in the tactical environ-ment.

As such, the system is composed of acollection of subsystems, e.g., dataprocessing, communications, operatingpersonnel. etc., which must performdesignated functional tasks duringspecific phases of the mission. Ina generalsense, any system may be called a com-mand and control system. Normally,however, it is defined as that portion of a

complete system which receives sensodata, formulates the reaction to thtenvironment, and initiates this reactionby interfacing with a subsystem which canchange that environment. e.g., weaponssystems, process control mechanisms, etc.There is a significant similarity betweencommand and control systems andmanagement information systems, theformer being the title when applied tomilitary systems while the latter is usedfor commercial systems.

Assume that a command and controlsystem has been formulated, e.g., throughthe use of P132 analysis, to execute amission which can be subdivided into mmission phases. Further, assume that nfunctional tasks must be performed toexecute the total mission, but notnecessarily each task within each missionphase. To execute the mission, the C2system must be composed of M sub-systems, e.g., data processing, et al, whichperform these functional tasks. It is

important in this algorithm analysis toinclude the operating personnel as asubsystem since the personnel interactwith every subsystem and provide backupfor these subsystems when degraded.

Rationale of algorithm

The Mission Execution Tasks (METs)vary in importance with respect to thephases of the mission, e.g., an MET maybe more important during strike thanduring alert. For any particular missionphase, the importance of a MET can beassigned a numerical value or weight. Themore important the MET, the higher thenumerical value of the weight. The weightfor the rth MET during the jth missionphase is denoted as B,,. Each majorsubsystem of the C2 configuration (e.g.,communications, data processing, staff)contributes toward the execution of theMETs. The importance of any subsystemvaries with the MET being executed; e.g.,communications may be more importantfor MET 2 than it is for MET 9. If a

subsystem is at full capability, then it cancontribute toward a MET at full value ofits importance. However, if it is not at fullcapability, then it can only contribute aproportionate value. Therefore, the basiccontribution of any subsystem toward theexecution of an MET is a product of thesubsystem importance in that MET andthe subsystem efficiency compared to fullcapability.

Up to now, only the effects of the internal

capabilities of the C2 system have beenconsidered, but the performance of theSystem also depends on the informationit receives from external sources. If thisinformation is of poor quality then thewhole system is degraded somewhat bythis fact. Furthermore, if com-munications or staff capability is alsodegraded, then the degradation of exter-nal information is amplified in the sensethat the system is not as capable ofobtaining or using the information whichcould theoretically be obtained.Therefore the total capability of ex-ecuting a particular MET can be ex-pressed as:

Degradation factor of subsystem AX Degradation factor of subsystem Bx Basic ...intribution of subsystem C

toward MET= Effective contribution of subsystem C

toward MET

The sum of these effective subsystemcontributions for all subsystems is thetotal internal contribution toward theMET execution.

Any subsystem in a command and con-trol center is not independent of the othersubsystems. A severe degradation in anyone subsystem effectively degrades theothers, e.g., if communications is

degraded sharply, the staff must spend agreat deal more time trying variousmeans of obtaining and evaluating infor-mation which normally would have beenautomatically stored in the data base. Forthis reason, subsystem degradation fac-tors are used to modify the other basiciubsystem contributions. For example,the data processing degradation factorwould modify the basic subsystem con-tributions of communications and staff.When a subsystem has full capability oronly slight degradation, then thedegradation factor associated with it is1.0; otherwise it is <1.0. In other words, ifdata processing were only slightlydegraded, a modifying degradation fac-tor of 1.0 would multiply the basicsubsystem contributions of com-munications and staff. If it were severelydegraded, a modifying factor of less than1.0 would reduce the other basic sub-system contributions. In a system com-posed of three subsystems. A, B, and C,then

Evaluation index of particular MET= effective subsystem contributions

toward the MET- Degradation of external information

with respect to METDegradation factor for stall

X Degradation factor for communications

Notice that for this last term the divisionby degradation factors of less than 1.0would tend to increase the negativemagnitude of the information quality andthereby increase the total degradation.

The expression above would yield anumerical value for the evaluation indexof any MET. However, as stated initiallyin this section. the METs vary in impor-tance. Hence, to obtain an evaluationindex for a total mission phase, a

weighted average of these terms, based onMET importance, must be calculatedover all the METs. Thereture, the modelfor the evaluation index for missionphase j (/,) is generally a weighted averageusing MET importance, subsystemcapabilities and quality of external infor-mation all modified by factors that repre-sent the cross -degradation effects of theother subsystems.

Stated mathematically, the model has theform:

8,,l,= E [sr, 4" El,r-1 7;

where k is the evaluation index for phase( 1.0 . /, 0); r is the MET number (1.2,...,n);j is the mission phase number (1,2, ...,m); B,, respresents the weight for the rthMET during phase j; and T, is the totalweight of METs (LB,,) during phase j,which is used to normalize the /, suchthat /, 1.0. Also,

S =k=1

E PirjWictakrjiritsk

where pn., is a degradation factor forsubystem i for rth MET during phase]; (i =I, 2, ...,M); M is the number of sub-systems which constitute total system; a kjis subsystem k efficiency for rm METduring phase]; k = 1, 2, ...,M; wk, is thecontribution of subsystem k toward ex-ecution of rth MET; and

M

EWit = 1.0k=1

i.e., numerical values of the subsystemcontributions are distributed such thattheir sum equals 1.0.

-I= max p(conini),, p(stqlf),,

71

Table I - Sample calculations

\11 I

rl

1.0

2.0

3.0

4.05.0

11 ,u,,, Wdradr4 a Sr4= Er.,= max l ) (P," P,")fi B.4 (S,4 sk4)

I0 0.3 0.5 0 10 1.0 0.9 -0.3 6.001 0.4 0.2 0.16 2,3 2/3 [0.4 + 0.2] = 0.56 -0.9/2 0.11

l0 0.3 0.3 0 0, 2/3 2/2 [0.3 + 0.3] + 0.16 = 0.56 -0.9/2 1.1010 0.4 0.4 0.00 1/3 1 3 [0.4 + 0.4] + 0.06 = 0.3267 max 1-0.3276, -0.9) = -0.3267 0.00

1 0.2 0.3 25 1.0 0.75 -0.3 0.45

_ i /3,4 = 32

obtained inlahle II. col 4

Table IIIcol. I. row r)

p, = pa = 1.0 for this example

Table III(col. 2 row r)

Table VI(min. of E row r

or 1.0)

p,,,Obtained byusing Fig. 1

x (1,4 x O,,4

Table IV Table III Table V(min. of E row r (col. 3, row r) (col. 4, row r)

or 1.0)

where 0 is the quality index of externalinformation for Ith MET during phase j;

Note: en, is defined as the maximum of twonon -positive values, i.e., the leastnegative of two values. If either p(),, issmall, then the term (Q, - 1)1(p,,, p,,,)becomes a negative value of highmagnitude and would overpower theother terms in the expression for I,.Therefore, by taking the maximum ofthis term and - So, this definition for ei,bounds the execution index of anyparticular MET to a value no lower thanzero, i.e., 0 4'4 1.0.

Illustrative example

As a means of further clarification, thissection includes a sample calculation ofan index for a reduced configuration forone phase of a mission when the externalinformation is poor (Q, = 0.7). Thecalculation is based on the followingassumptions:

1)The C' system is composed of three sub-systems - data processing. com-munications. and staff personnel.

2) The system must execute five mission execu-tion tasks (METs) for mission phase #4.

3) There is no degradation of the data process-ing or communications subsystems;therefore p,,, = pd, = 1.0. The values of p,,,are derived from Fig. I.

4)The values of the terms of the algorithm arederived from the tables referenced in TableI. The table entries are arbitrarily selectedfor this example and no further supporting

rationale is given for selecting thesenumerical values.

Under these assumptions, the algorithmhas the form:

14 =

5

r-1 T4

8,4[ elr4]

For the sample calculation, the expres-sion is computed in its equivalent form:

T41,= E8r4 [S,4 1,4]

.

The, /4 is found by dividing both sides ofthe equation by

T4= E 8.4 = 7.66.r=1

/4 is numerically solved in the box at thelower right of the Table I and equals0.239.

The calculation is generally self-explanatory with remarks beneath eachcolumn as necessary. The formal evalua-tion expression is as given above.

For the system illustrated in the example,the /, could be called "percent effective-ness" for higher values, say 0.6</, <1.0;but below the lower limit this connotationwould not hold. The histogram in Fig. 2

= 0.7 for this example

= /4 T4 = 7.66

7.66 7.66- = = 0.239T4 32

gives a qualitative indication of how theindex, 1,, might be interpreted. Theassignment of these levels was based onqualitative interpretation of perform-ance of METs for which quantitativevalues could be interpreted with a higherlevel of confidence. Then these qualitativevalues were extrapolated to hold for thecumulative indices.

A study of the model parameterdefinitions and constraints will show thatfor the given model possible numericalvariations in the input parameters willtend to compensate each other to someextent so that the resultant cumulativeevaluation indices will not substantiallychange any conclusions. As an example,in the illustrative computation let thenumerical weight given to the most im-portant M ETs during the phase be halvedwhile the weights of the lesser METsremain the same; then the /4 will vary inabout the third decimal place.

Although reasonable variance ofnumerical input values will probably nothave a great effect on the results, anyvariance in the form of the model expres-sion may have significant effects.

For example, if the cross degradationeffects are eliminated, or if the effect ofexternal information quality is neglected,then the results will probably vary con-siderably. A cursory explanation of this is

72

Fable II ME I. weighting matrix (11,4). Table Ill - Assignment of is values ( W,4).

Mission phases (I) Subsystem WeightMET # MET # Comm. Data Staff

(r) I3 4 5 6 (r) proc.

1.0 10 10 10 10 10 10 1.0 0.3 0.5 0.22.0 10 10 I I I 10 2.0 0.4 0.2 0.43.0 3 10 10 10 10 3 3.0 0.3 0.3 0.44.0 I 3 10 10 10 3 4.0 0.4 0.4 0.2

5.0 I I I I I I 5.0 0.2 0.3 0.5

Table V - Staff efficiency/ MET (a,m) 1 .1 hie \ I I)1,trihuti,on of <r \I1 1.

MET #Mission phases (j)

MET # Main I/O Mass(r) I 2 3 4 5 6 (r) CPU memory controller memory

1.0 1.0 0.7 0.6 0.5 0.5 1.0 1.0 0.250 0.3002.0 1.0 0.7 0.6 0.4 0.4 0.5 2.0 0.200 0.1503.0 1.0 1.0 0.5 0.4 0.4 0.7 3.0 0.250 0.2254.0 1.0 1.0 0.4 0.3 0.3 0.8 4.0 0.325 0.1505.0 1.0 1.0 0.5 0.5 0.5 0.6 5.0 0.250 0.175

as follows: Changing the form of themodel alters the effect of the parameters,(when, how, etc.) and therefore can besignificant. Any reasonable variation inthe numerical values of the parameters,however, affects both the reference stan-dard and the candidate configurationwith comparable intensity; hence theindex, being a relative value, is notsignificantly affected.

Conclusions

System performance is one of the mostimportant of the factors that constitute

090807

Psi, 05e

04050201

Psrj.0

02 03 04 05

0srj

Table IV - Contribution of communicationsto METs(a.,4).

M ET# Comm. Systems(r) 905 903 904 901 908 181

1.0 1.0 0.1 0.2 0.1 --

2.0 1.0 0.1 0.2 0.33.0 1.0 0.1 0.2 0.24.0 1.0 0.1 0.2 - 0.1 -5.0 1.0 0.1 0.2 - 0.1 0.1

DisplayControl Mag.unit tape

Printer/TabDisplaycontrol

Comm.dataControlconsole

0.125 0.050 0.075 0.2000.200 0.050 0.100 0.3000.200 0.050 0.075 0.2000.150 0.050 0.075 0.2500.250 0.075 0.100 0.150

total analysis of Command and Controlsystems. A quantitative measure of per-formance is an extremely desirableparameter when a comparison of varioussystem configurations is required. Thisalgorithm establishes a basis whereby aquantitative measure is obtainable.

This model may not be perfectly suitedfor a given mission - it may be toosophisticated, or conversely, it may ig-nore pertinent constraints. Nevertheless,the model described in this paper doesestablish a general rationale which can be

Fig. 1 - Degradation factors due to staff personnel related to subsystem efficiency.

applied in tailoring to a given systemmodel. As shown by the illustrative com-putation, the form of the algorithm ismore important than the actualnumerical values (within reasonablebounds) entered into the computation.

The implementation of the model may beprogrammed on a digital computer ortime-sharing terminal so that variationsand parameter perturbations can be easi-ly assessed. Several variations of modelswere assessed before the one presentedhere was selected.

EXCELLENT

52

GOOD

MARGINAL

SUB -MARGINAL

UNACCEPTABLE X45c ,

0 05 0S 07

Fig. 2 - System ettectivenns.

0B 09 1.0

73

rNTSS

SARAa research planning tool

C. Havemeyer R.R. Lorentzen

SARA (System for Analyzing Research project Attributes) was developed to aid themanagement of the Laboratories in the research planning process: it is an amalgamationand extension of systems previously used at the Laboratories.' The purpose of thisinteractive tool is recording and subsequently retrieving, analyzing. and reporting theallocation of resources to projects and the characterization of these projects by means ofuser -defined attributes. SARA is designed for use throughout the planning process: torecord the characteristics and resources required for proposed projects: to aid in theassigning priorities and in the selection procedure: to help ensure a balance in theportfolio of planned projects: and to assist in the monitoring of resource utilization.

PLANNING of research at theI.ahoratories is a continuing process. Thenature of the planning process has beenevolving over the past several years andwill continue to change in the future. Theprocess is structured so the Laboratoriescan he responsive to change bothinternal and external.

The development of the plan for aparticular year begins with projectproposals that are recommended by theLaboratory Directors based on theirevaluation and assessment of variousforces which are at work; these includethe technological and business needs ofRCA .divisions, inputs from RCA's cor-

Craig Havemeyer. Mgr., Management Information Systems, RCA Laboratories, Princeton. N.J., received the BS inIndustrial Engineering in 1966 and Master of Engineering in Operations Research in 1967, both from Cornell UniversityUpon graduation, he pined RCA as a member of the Operations Research Career Program, which consisted ofassignments with several RCA divisions and corporate staff activities. Mr. Havemeyer then accepted a position with theComptuer Systems Division where he was involved in implementing a large business planning system and other O.R.projects. He was promoted to Manager. Operations Research in 1970. In late 1971, he assumed his present position at theLaboratories. directing the development and use of administrative computer systems and management decision aidsRecently his responsibilities were broadened to include the Computer Services organization.

Robert R. Lorentzen. Mgr.. Computer Services, Management Information Systems. RCA Laboratories. Princeton. N J..received the BSEE in 1961 and the MSEE in 1965. both from the Newark College of Engineering. he received the MS inComputer Science in May 1975 from the N.J. Institute of Technology. After working at Western Electric. Hewitt -Robinsand Federal Telephone and Radio, he pined RCA's Microwave Operations in Harrison. N.J in 1956. where ne heldseveral positions in Equipment Development. Applications Engineering. Staff Planning. and Marketing. In 1972 hebecame responsible for the computer support and scientific programming for RCA's Thermoelectric Operations InJanuary 1974, Mr. Lorentzen transferred to RCA Laboratories in Princeton as Manager of the Operations Analysisactivity in Management Information Systems. In February 1975. he was given the responsibility for the ComputerServices organization at the Laboratories. Mr. Lorentzen has written papers on Microwave and Thermoelectric Devicesand Business Planning, and is a member of Eta Kappa nu. Tau Beta Pi. the IEEE. the ACM. and the Institute ofManagement Science.

Authors Lorentzen (left) and Havemeyer.

74

CHANGESERMINAL

GGfsgl

04).4,

SARADMPROGRAM

SARARPPROGRAM

REPORTS

SARADATABASE

HIGHSPEEDPRINTER

Flg. 1 - SARA system structure.

porate staff, the progress and status of thecurrent research program, guidance fromthe technical staff within theLaboratories, and evaluation of com-petitive organizations, businesses, andtechnologies.

The next step is an interactive process inwhich projects are assigned prioritiesbased on their estimated return to RCA.Required resources are compared to theresources likely to be available for theyear; resources are diverted from thelower priority projects to those of higherpriority. At the same time, it is necessaryto ensure that the list of projects is

balanced in several dimensions: withrespect to the divisons supported, withrespect to the time interval until payoff,and with respect to the portion of effortplanned for research of various types(e.g.. close -in divisional support, ex-ploratory research. etc.).

The research plan at this stage forms abasis for the formal Business Plan sub-mitted to RCA Corporate Management.

The final portion of the research planningprocess consists of continuous monitor-ing of actual resource use compared toplan. In addition, the Plan itself is up-dated during the year in response tochanging conditions.

System structure

SARA is a special-purpose data -basesystem. It consists of:

A data base for storage of the resource andattribute information for the projects: thisincludes the allocation of people to projects;

A data management program (SARADM)for creation and retrieval of the data baseand for input, deletion and update of theresource. attribute and other information:and

An anal sis and report generator program(SARARP). Several reports are available.As will be shown, a great deal of flexibility isavailable to the user in specifying the struc-ture of reports.

'Earlier systems were PROPP (Project PlanningProgram) and MAP (Manpower Allocation to Projects).They were used extensively in 1973

As shown in Fig. I, both programs areaccessed conversationally from a remoteterminal. Reports may be directed toeither the terminal or a high speed offlineprinter. Special provisions are made forconfidential or sensitive data.

Data base structure

I SARA data base contains four majorsegments:

general datadictionaryproject datapeople data.

Fig. 2 shows these four segments whichare linked together through a nicknamehierarchy.

The project data segment contains infor-mation about each current and plannedresearcn project. In addition to identifica-tion data (project nickname and title), theresources required for the project and theattributes of the project are stored.

There are twenty predefined resourceswhich can be input for each project.Resources include people, materials.computer costs, and other resourceswhich may be directly allocated to pro-jects. Provision is made for enteringoverhead (as a percent) and credits.

01 the twenty resources, ten of themrepresent different people skill and costcategories; the number of people requiredin each category is input and stored in

DICTIONARYATTRIBUTES

GENERAL DATAFIRST YE R:1973RATES:

*MT' SOK/YR.01AT2 : GISOK/YR4111ATS : STOK/YR

the data base. Each of these peoplecategories is characterized by a dollar ratewhich is used to translate the number ofpeople into a dollar cost for the project.The other ten resources are entered eitherdirectly as dollars or as a percentage.

Unlike the resources, attributes are

defined by the user. The definition of eachattribute is contained in the dictionarysegment of the data base. Two types ofattributes may be specified: value orrange. A value attribute has specificallowable values. For example, the usermay wish to characterize the projects interms of Payoff. The attribute nicknamemight be defined as PAYE and theallowable values defined as: low,medium, and high. A range type attributeallows the user to pre -assign and labelvarious contiguous ranges. For examplean attribute, 'lime to Payoff, might bedefined as:

range0 to <3 years3 to < 8 years<8 years

labelshortmediumlong

In this case, the attribute value for aspecific project would be a number ofyears (e.g., 4). At report generation time,this project would be identified as havinga medium time to payoff.

The fourth segment of the data base dealswith individual people and theirassignments to projects. The projectassignments cover a two-year period: thecurrent year and the following year. Ascan be seen in Fig. 2, the assignment listconsists of project nickname, date of

PROJECT NICKNAME:AA APROJECT TITLE:SAMPLE PROJ.

ALLOWABLE %SLUES COMMENTS:RESOURCES REO'D :oisoaciao MMMMMMMMMM m

41-1.1T2 3.0*MT5 2.0

ocuscoomom MMMMMMMMM

ATTRIBUTESSECN 4DIVA SSD

---M.PROJECT DATA

Fig. 2 - Simplified data base structure.

J DOE MTS SECN 4

PROJECT ASSIGNMENTS'

PROJ. DATE %EFF

MOTH 1.0 100AAAA 3.0 1005055 13.0 50UNAS 13.0 50

PEOPLE DATA

PROJECT iNFO4mAT,ON

LAST TEAR CURRENT YEAR NEXT YEAR1975 IWO

PLANS-.

CURRENT PLAN -4. -

YEAR -TO -DATEACTuALS

140sT RECENT MONTHSAC' uALS

MANPOWER PROJECT ASSIGNMENTS

PLANNINGEMPHASIS

1..1..1..1..1..1..1 MMMMM 1CURRENT NEXT

YEAR YEAR

Fig. 3 - Treatment of time in SARA.

beginning of assignment and percent oftime spent on this project.

The special nickname NOTH indicatesthat the person is not here during thatperiod (e.g.. leave of absence, not yethired). A second nickname U N AS is usedfor individuals who are not yet assignedto a project.

Another dimension to the structure of theSARA data base is the treatment of time.As shown on Fig. 3, project data covers athree-year period: last year, current year.next year. In addition, depending onwhich of these years is being considered.one may also be concerned with businessplan data, current plan data, or actualresults. Each of the blocks in the matrixshown is called a frame and contains theresource and attribute information forthe indicated year/data-status.

The availability of the six frames enablesmanagement to make meaningful com-parisons of research activity across timeand between planned activity and actualresults.

SARA reports

At the present time, SARA can produceseven report types grouped in the follow-ing categories:

Detail reports,Resource utilization summaries, andResearch program' analysis.

Two detail reports are available -- onedealing with project information and theother providing an alphabetical list ofpeople showing project assignments.

Reprint RE -21-3-26Final manuscript received February 25. 1975. 75

*A D aaaaA I*A P pppp

*C D aaaa*C G gggg*C I*C P pppp

*D D aaaa*D I*D P pppp

*F:

*G N*G 0

*I PPPP*0 PPPP

rrrrAl,

:taaa

1.1-11

itila

*P D aaaa*P G*P I*P P pppp

*U

Fig. 4 - SARADM activity commands.

add dictionary (attribute)add individualadd project

change dictionary (attribute)change general itemchange individualchange project

delete dictionary (attribute)delete individualdelete project

frame

generate new filegenerate old file

in projectout project

list resource or attribute values

Mass (input)

print dictionary (attribute)print general informationprint individualprint project

update frames

Fig. 5 - SARARP activity commands.

:C F:C M

:G

:0

M:RAT

0

change factorschange mode (terminal/ offline)

generate (old) file

offline

report - alphabetic listing of people

: R P[aaaa][aaaa]report - listing of projects

M:RMT

0report - matrix (personnel)

: R P[aaaa][aaaa]report - project descriptions

M:RST

0report - summary (personnel)

: R S[aaaa]kaaalreport - totals

where: aaaa

TM

= an attribute nickname= MTS= Technical Support Personnel= Other People

Two of the resource utilization summaryreports show the mix of the portfolio ofresearch projects by grouping projects byattribute values. An example of this willbe shown later. A third report in thiscategory displays the variances betweenproject effort required and effortassigned. Another report shows thedegree to which projects are staffed byout -of -section people.

The final report shows the results of somespecial computations which prorate theresearch effort across RCA businesses.

Each of the seven report types providesthe user with a high degree of flexibility asto the nature of the information dis-played. For example, the report whichgroups projects into attribute categoriescan be generated:

- with 0. I or 2 attributes or grouping- with I 2, 4 or 6 frames of information

with I or 2 data items shown.

Using SARA

\ key to the use of SARA is the activity

:r p

PROJ-ENERFROJ.ATTRIBUTE LIST? (Y,N)y

command structure. Each of thefunctions of the two SARA programs isaccessed by the user issuing commands.Fig. 4 shows the activity commands forthe data management program. Note thatin addition to the particular function tohe performed, the activity command isused to indicate to the program thespecific nickname of the project, at-tribute, etc. to be added, changed, deletedor displayed.

Fig. 5 shows the activity commands forthe report program.

Example

A sample data base consisting of fifteenprojects was developed for demonstra-tion purposes.

Fig. 6 shows the terminal prompting andresulting project description report. Thedegree of flexibility available to the user isevident from the prompting sequence. Inthis case, information on just one project,nicknamed COLR, is requested. Threeframes are shown: I975P is this year'soriginal plan; I975C represents the

F. 75pF. 75cF. 76pp4

DIVISION SUPPORT (DOUBLE COUNTING) DATA? (Y,N)nDATA ITEMS? ((C)OLUNNS.(T)ITLED,(CR)-NONE)cENTER DATA ITEMSITEM.IsteITEM -lame

ITEM-St:us

ITEM.

LIST PEOPLE? (M,T.0.(A)LL.(CR)-NONE).(A)VG.(S)NAP, OR (E)VENTS? e

START DATE- 1

END DATE. 25

INCLUDE START DATE ASSIGNMENTS?(Y,N) y

ADVANCE THE PAPER ...H1T RETURN

RESEARCH PROJECT DESCRIPTIONSARA - ISSUE 2FILE 842

LtER-ENERGY RESEARCH

ATTRIBLIES:

SECTIONTIME TO PAYOFFPAYOFF

DATA ITEMS/STS -TOTAL NUMBER OF YTS/AMA -AVG I MIS ASSIGNEDSTUN -GROSS COST (SE)

1975P 1975C I976P

71 71 71

LONG LONG LONGMEDIUM MEDIUM MEDIUM

5.00 5.50 7.000.00 4.25 5.50

475.00 522.50 693.00

MIS ASSIGNED - ON 1/01/75SEC SAME

1 (72) McKellar, James 100

2 Bachmenn, Frank 50

Puccini, Louise 100

6 Jennines, Robert 100

MTS ASSIGNED - CHANGES DURING THEI SEC NAME

DATE: 03/20/75VERSION I 2.008PAGE 1

PERIOD FROM 1/01/75 TO 1/01/77DATL

4 Dickson, Jams 7/01/75 100

5 Falrarano, John 10/01/75 100

Fig. 6 - Project description report.

76

current plan for the year; and I976P is the Acknowledgmentlatest iteration of next year's plan.

The data items section of the report firstshows the planned level of effort on theproject, then the average number ofpeople assigned to the project, and finallyan estimate of total project cost.

The last section deals with individualsassigned to the project. First, the roster atthe beginning of the current year; thenchanges during the course of the year. Inthis case, the two additional individualsare scheduled to start full time work onthe project on the dates indicated.

A second example, Fig. 7, illustrates howSARA can be used to aid in the analysisof the project portfolio through the use ofattributes. The example shows projectsgrouped by two attributes, time to payoffand payoff. For each project and eachgrouping, the planned manning level isshown for 1975P and I976P.

Another version of the same report type isshown in Fig. 8. In this case, projects aregrouped by section. This report format isused to help analyze the degree to whichactual effort expended on projects is

consistent with this year's current plan.The first two columns compare theoriginal plan with the current plan. Thenthe current plan, factored to representavailable effort level, is compared to theactual average effort applied year to date.The last two columns are similar, butfocus on last month's effort. The year-to-date and monthly actuals are available toSARA from the Laboratories' ProjectCost System.

Conclusion

SARA was first used for the 1975 plan-ning cycle, starting in the spring of 1974.It was used heavily during the prepara-tion of the 1975 Laboratories' BusinessPlan submission. Since that time, actualresource utilization has been comparedto the plan on a monthly basis and theplan has been modified in response tochanging conditions.

The system itself is viewed as an evolvingtool. Several additional capabilities havebeen added during the past year andothers will be added as the participants inthe planning process gain experience inusing the system.

For the successful development and con-tinued use of SARA, we are happy to

acknowledge the support and guidance ofthe Laboratories' Management. In addi-tion, we are grateful Barbara Banko andVince Boccanfuso for their valuable con-tributions to this project.

.r 1 tin. payfF. 75pF. 76pITEM -/.tITEM.

DISPLAY (C)HANGE. (P)ERCEN7 CHANGE. OR (CR) -NOTHING c

ADVANCE THE PAPER ...HIT RETURN

RESEARCH PROJECT LISTINC

SARA - ISSUE 2PILE #42

TIME PAY? PROJECTS

DATE. 03/20/75

VERSION 0 2.008

PACE I

PMTS -TOTAL NUMBER OF MIS1975P 1976P COG

SHORT HIGH COLR-COLOR TV RECEIVERS 10.00 8.00 -2.00SHORT HIGH CCDA-CCD APPLICATIONS 2.00 2.00 0.00SHORT HIGH ELOP -ELECTRON OPTICS 1.00 2.00 1.00

3 13.00 12.00 -1.00

SHORT MEDIUM MINI -MINICOMPUTER APPLICATIONS 6.00 8.00 2.00SHORT MEDIUM 1001 -ION IMPLANTATION 4.00 3.00 -1.00

2 10.00 11.00 1.005 23.00 23.00 0.00

MEDIUM HIGH IRODBROADCAST SIGNAL SYSTEMS 9.00 7.00 -2.00MEDIUM HIGH STRC-STRUCTURAL TECHNOLOGY 5.00 5.00 0.00MEDIUM HIGH HOME -NOME ELECTRONICS SYSTEMS 4.00 4.00

3 14.00 16.00 2.00

MEDIUM MEDIUM AUTO -AUTOMOTIVE ELECTRONICS 5.00 -5.00

MEDIUM LOW RADR.COMMERCLU. DOPPLER SYSTEMS 2.00 2.00 9.00MEDIUM LOW AVEC-AVIONICS ELECTRONICS CTR 1.00 1.00 0.00

2 3.00 3.00 0.006 22.00 19.00 -3.00

LONG HIGH K1NE-SOLID STATE KINESCOPES 10.00 12.00 2.00

LONG MEDIUM ENER -ENERGY RESEARCH 5.00 7.00 2.00LONG MEDIUM El:AC-ELIXIR° ACOUSTIC DEVICES 1.00 1.00 0.00

2 6.00 8.00 2.00

LONG LOW ACOU -ACOUSTIC SYSTEMS RESEARCH 1.00 1.00 0.0017.00 21.00 4.00

TOTAL 62.00 63.00 1.00

Fig 7 - Analysis of protect portfolio through the use of attributes.

r I @rnF. cyrITEHImtsYTD FACTOR- .85LAST MONTH'S FACTOR- .9ADVANCE THE PAPER ...HIT RETURN

RESEARCH PROJECT LISTING DATE: 03F20/75SARA - ISSUEFILE P42

SEC:.

2

PROJECTS

VERSION I 2.008

PAGE I

AITS-TOTALFULL TEAR

NUMBER OFTEAR TO DATE

mrsLAST MONTH

.............1975P 1975C 19750 1975Y

...........m.19750 19751

FACTOR. 0.850 0.900

71 KINE-SOLID STATE KINESCOPES 10.00 10.00 8.50 9.00 9.00 9.50

71 STRC-STRUCTURAL TECHNU_DGY 5.00 5.00 4.25 3.on 4.50 2.0071 EVER -ENERGY RESEARCH 5.00 5.50 4.67 4.00 4.95 5.51571 IONI-107 IMPLANTATION 4.00 4.00 3.40 4.10 3.60 4.70

4 24.00 24.50 20.82 20.10 22.05 21.70

72 COLR -COLOR TV RECEIVERS 10.00 10.00 8.50 5.00 9.00 5.10

72 AUTO -AUTOMOTIVE ELECTRONICS 5.00 3.00 2.55 2.50 2.70 3.00

72 HOME -HOME ELECTRONIC: SYSTEMS 3.00 2.55 2.60 2.70 2.0072 ACOU -ACOUSTIC SYSTEMS RESEARCH 1.00 1.00 0.85 0.75 0.90 0.:5

72 ELAC-ELECTRO ACOUSTIC DEVICES 1.00 1.00 0.85 0.80 0.90 0.7517.00 18.00 15.30 11.65 16.20 12.30

73 PROD -BROADCAST SIGNAL SYSTEMS 9.00 8.00 6.80 7.50 7.20 6.1073 MINI -MINICOMPUTER APPLICATIONS 6.00 7.00 5.95 10.00 6.30 12.5073 CCDA-CCD APPLICATIONS 2.00 3.00 2.55 2.80 2.70 2..0

73 RADR-GOMMERCIAL DCPPLER SYSTEMS 2.00 2.00 1.70 1.50 1.80 1.20

73 AVEC-AVIONICS ELECTRONICS CTR 1.00 1.00 0.85 0.90 0.90 0.10

73 ELOP -ELECTRON OPTICS 1.00 1.50 1.27 1.00 1.35 1./0

6 21.00 22.50 19.12 23.70 20.25 24.30

TOTAL 62.00 65.00 55.25 55.45 58.50 58.70

Fig. 8 - Project summary of Fig. 7 arranged by section.

77

Mechanical design of mobiledata processing systemsfor worldwide useJ. Furnstahl, J. Herzlinger

Modern tactical command and control systems must be automated to a high degree. betransportable anywhere in the world, and operate continuously after arrival. Themechanical design implications are considerable. Specific solutions to mechanical designencountered in the design of the Tactical Information Processing and Interpretation (TIPI)System are described. TIPI is a mobile land -based system designed to apply automaticdata processing techniques to tactical intelligence processing. RCA implemented theDisplay Control/ Storage and Retrieval (DC SR) Segment.

John S. Furnstahl. Manager. Preliminary Design.Government Communications and Automated SystemsDivision. Burlington. Massachusetts received hisBachelor's degree in Mechanical Engineering from Mar-quette University in 1950 and his BS degree in ElectricalEngineering from Drexel University in 1961 Sincejoining RCA in 1952. Mr Furnstahl has been engaged indesign and development of light -weight airborne firecontrol equipment. airborne TV, airborne and spacecommunications, modification of aircraft for equipmentflight tests. high -resolution radar, high-speed tapetransport. thermo-electric. special electromechanicaldevices and environmental testing. From 1959 to 1963.his responsibilities included reliability, maintainability.value engineering, drafting coordination. MeasurementEngineering Laboratory and design support MrFurnstahl was Manager of Engineering Controls andSupport and later was Manager of Computer ProductDesign As Manager of Product Design and Installation.he was responsible for design and development ofcomputer. monitoring and control equipment His mostrecent assignment consists of equipment product andinstallation design of the display and control/storageand retrieval system for tactical information and in-telligence support for military operations. Mr Furnstahlis a member of Pi Tau Sigma, ASME. IEEE and AIAA. Heis a registered Professional Engineer in the State of NewJersey (No 11029) and the Commonwealth ofMassachusetts (No. 25315)

Joseph I. Herzlinger. Manager. Product Design, Govern-ment Communications and Automated Systems Divi-sion. Burlington. Massachusetts received his BS inMechanical Engineering from Newark College ofEngineering and his MS in Mechanical Engineering fromDrexel Institute of Technology in 1961 Mr Herzlingerjoined RCA following his graduation in 1940 He wasengaged in the mechanical design of shipboard radarequipment and television transmitter and antenna equip-ment until 1946 As Leader of the advanced electro-mechanical design group, he assumed responsibility forthe mechanical design of the Vernier II High ResolutionRadar System. overall responsibility for an advancedhigh-speed tape transport, and responsibility for thedevelopment of the Ranger Shutter At the AerospaceSystems Division, Burlington, Massachusetts. MrHerzlinger has had mechanical design leaderresponsibility for projects involving camera shutters. IRenergy detection using fiber optics. and laserapplications He was responsible for the mechanicaldesign of the electronic assemblies of the LM ApolloRendezvous Radar and Transponder and the mechanicaldesign of the Pulse Doppler Radar System. He wasresponsible for the mechanical design of the Aegis OrtsEquipment and the Walt Disney World AutomaticMonitoring and Control System. In his most recentassignment. he had mechanical design responsibility forthe configured shelters. bare shelters, environmentalcontrol system. passageways, and pallets for the TIPIDC,SR equipment He has been granted two U S patentsand isa Professional Engineer in the State of New Jersey

78

RANDOM ACCESS MEMORY

AUTOMATIC DATA PROCESSING

COMMUNICATIONS

THE NEED for ease of transport andrapid deployment severely limit size andweight and pose other problems such ashow to protect equipment frommechanical shock. Worldwideenvironmental extremes require that per-sonnel and hardware be protected fromheat, cold, and humidity. Reliable andeffective use of these complex systemsunder such conditions demand carefulhuman engineering and maintainabilitydesign.

The Display Control/Storage andRetrieval system (DC/SR) assists theintelligence analysts in their tasks ofkeeping the tactical intelligence data baseup to date. The DC/SR segment consistsof shelter -mounted equipment, computersoftware, trained personnel, information.intelligence data and facilities. TheDC/ SR segment is configured for use bythe United States Marine Corps( US MC)and the United States Air Force( USAF),and will interface with other segments ofthe TI PI System and other existing fieldsystems. The DC/SR equipment is

functionally divided into the areas ofcommunications, non -digital materialsand automatic data processing. Theprime contractor for the developmentmodels of the DC/SR segment is theSystem Development Corporation,Hampton, Virginia.

Physically the TIPI DC/SR segment isconfigured in a set of 20 ft long by 8 fthigh by 8 ft wide shelters. The USAFconfiguration consists of I I shelters (Fig.

c:ftfis 3 - 150KW 60 HZ

2 - 15OKW 400 HZ

LOG JOURNALIZING

NON -DIGITAL STORAGE 8RETRIEVAL CENTER

q) ANALYSTS

Fig. 1 - Display and control/storage retrieval segment (TIPI)

I ). Three basic shelters are the automaticdata processing (ADP), random accessmemory (RAM), and the com-munications type. There are six in-telligence analyst station (IAS) sheltersand two non -digital shelters. The USMCconfiguration consists of 7 shelters, usingthe same three basic shelters as the USAF(ADP, RAM, and Communications).Three intelligence analyst station (IAS)shelters and one non -digital shelter makeup the remaining US MC complement. Inaddition to the I I or 7 shelters in a TIP!DC/ SR segment there is a set ofenvironmental control units, auxiliarypower generation units, and pallets fortransporting cables and other ancillaryequipment.

A major consideration in selection ofequipment was to employ available on -the -shelf designs for integration into theTI PI DC/SR segment with virtually norisk. Some of the equipment alreadyqualified. Other equipment was anenhancement, expansion, or modifica-tion of equipment already qualified. Insome cases the design had been proven incommercial systems and repackaging wasrequired to meet the TIP! requirements.The size, weight and volume of theselected equipment are consistent with itscapacity, performance and environmen-tal capability. Shelter equipments arcconfigured to keep the center of gravitywithin two feet in the longitudinal direc-tion, within one foot in the lateral direc-tion, and below a distance of four feetabove the bottom of the skids to prevent

unbalanced loading and upset duringhandling.

Transportability

The DC/SR segment was designed to be transported by rail, ground, air or sea.The USMC shelters arc transported byend -mounted trailers while the USAFshelters are transported by undercarriagetrailers. All pallets. including theenvironmental control system ( ECS) andpower distribution and cable pallets aretransported by a smaller, end -mountedtrailer.

The Marine Corps DC SR segment canhe loaded on or off ships, includingamphibious vehicles, and C130 -type air-craft, with or without the use of the 4631loading system. These shelters and palletassemblies can use railroad transporta-tion, including humping at 9 mi/ h whilethe shelter is resting on its skids. Eachshelter and pallet assembly with trailerattached can be transported byhelicopter. The Air Force shelters arecapable of the same transportation modeas the Marine Corps shelters except thatthe trailer must be detached for C' -I30type aircraft and railroads.

1 he shelters have a gross weight of 10,000pounds. including equipment, in thetransport mode. The shelter trailersweigh approximately 4500 pounds. Theweights and centers of gravity weremonitored constantly during the designcycle; each configured shelter and loaded

Reprint RE -21-3-19Final manuscript received February 12, 1975. 79

pallet in the transport mode was wellwithin the specified limits.

Special attention was given to main-taining commonality between theDC SR Marine Corps and Air Forcesegments as well as to the TIP! ImageryInterpretation (II) segment to facilitatedeployment and logistics. The commonitems include shelter trailers, pallets forthe EC'S, shelter jacks, shelter and pallethoist slings. The basic hare sheltermodule structure and overall dimensionsare similar including interface shelterhardware items such as fittings for trailerattachements, lifting, and towing rings.

The DC: SR segment was designed sothat no dangerous or sensitive equipmentor material must be transported. Noenviornemntal control during transit orintransit storage of any equipment is

required to retain stability or precludecompromise of reliability. [he shelterscontain pressure relief valve for airtransport.

Without special material handling eq tii p -men t or packaging design, the DC SRsegment can he deployed from a

transport to an operational mode or froman operational to transport in two hoursor less.

1)uring shelter transport, certainperipheral and ancillary equipment areremoved from their normal operatingstation and secured in transit cases. Theseitems include commercial microfilmreader printer, paper -tape punch, andcommunication buffer. I he transit casesare secured to tie -down attachments inthe shelter floor. Provisions have beenmade to easily remove any classifiedequipment and documents during transitand or storage. Peripheral and ancillaryequipment which are not put into transitcases are tied down to the floor or securedfor transit. These items include chairs,telephones, telephone switchboards,magnetic tapes, disc packs, disc drives,and plotter. The detailed steps forpreparation of shelters for transport areas follows:

Remove heavy documents from cabinets andpack in transit cases.

Place certain commercial peripheral equip-ment items in transit cases.

Secure all transit cases and loose I urnishingsin shelter.

Disconnect and stow all external power andsignal cables.

Disconnect and stow ECS ducts and controlcables.

Disassemble and stow passageway.

Disconnect and stow ground anchors andelectrical grounds on ECS pallet.

Jack shelter and install undercarriage trailerset (Air Force only).

Rk:inoc and stow shelter and 1(5 palletjacks aboard FCS pallet.

I -or ADP and RAM shelters. stow digitalcables on ancillary equipment pallet.

Lmplace dust covers over shelters ECS ductports.

Purge generator fuel systems.

Attach prime movers.

Note that, in general, items normallyoutside the shelter in the operationalmode arc stowed outside of the shelter onpallets, whereas items normally withinthe shelter are stowed within the shelterduring transport mode. This avoids theintroduction of dirty cables, jacks, etc.into the shelter, retains documents andequipment specific to a given shelterwithin that shelter - and avoids therequirement to open a shelter before it isjacked and leveled. Items stowed on theECS pallets are common to all shelters.permitting ECS pallets to be in-terchangeable within the DC/ SRSegments.

Deployment

The site planning and preparation fordeployment are functions of thegeographical location, terrain, season ofthe year. weather and time duration of thesegment employment. See Fig. 2, TIPISegment deployment functional flowdiagram.

Prior to the arrival of the DC/ SR seg-ment at the site, an area must be selected,surveyed, staked, and planned requisiteto emplacement of equipment. Theshelters. ECS and passageways aredesigned to be capable of being jacked toa level altitude for operation on a terrainhaving a slope up to 10 percent and soilcapable of supporting 12 psi. Any in-dividual shelter shall be convertible froma transportable status and from anoperational status to a transportablestatus in 2 hours or less. Fig. 3 shows theemplacement time scheule common to allmodules.

The deployed configurations representessentially the minimum area compatiblewith all other requirements. Specialattention was given to the layouts tominimize lengths of critical signal cables,walking distances for personnel betweenshelters, and passageway length, weight,and volume. The power generators arclocated remotely from the shelters tominimize acoustic noise level, fuel odor,and fire hazard. Each ECS is locatedapproximately 10 ft from its shelter andconnected to the shelter by flexible ductsto minimize vibration and acoustic noiseinside the shelters.

The passageway system is composed ofdetachable modules to facilitate stowageon the pallets for transport and forerection during deployment. Each majorpassageway component is capable ofbeing handled by two people. Allhardware and tools are captive. Groundanchors are used to secure thepassageway against wind.

I ATTACH MODULE JACKS AND 4 MENRAISE MODULE

2 REMOVE TRAILER 4 MEN

3 LOWER AND LEVEL MODULE 4 MEN

4 POSITION ECS PALLET USING 2 MENPRIME MOVER

5 UNSTOW AND CONNECT POWER 2-4 MENCABLES I ECS, MODULE ANDGENERATOR)

6 CONNECT ECS TO MODULE 2 MENDUCTS, CONTROL CABLE ANDTRANSITION

7 ENTER MODULE ACTIVATE 2 MENPOWER AND ECS

8 UNSTOW AND POSITION SIGNAL 2 MENCABLES

9 UNSTOW, POSITION AND 2.4 MENROUGH LEVEL WALKWAYS

0 20 30 40MINUTES

Fig. 2 - Emplacement lime schedule common to all modules.

SO 60

80

ADVANCE SITESURVEY

OFF-LOAD SEGMENT ATDEBARKATION POINT ANDESTABLISH DEPLOYMENT SEQUENCE

ON -SITE SURVEY IF REQUIREDTO PRECEDE DEPLOYMENT

HELICOPTER OR GROUNDTRANSPORT TO SITE

REMOVE POWER ITEMSFROM POWER PALLETS

EMPLACE POWERGENERATORS

CONNECT POWERLINKS

CHECK FUELSTATUS

ENERGIZE AND CHECKOUT SYSTEM

OPERATE SYSTEM

EMPLACE MODULES ANDASSOCIATED ECS

INSTALL STOWED EQUIPMENTIN MODULES AND ON SITE

CONNECT SIGNAL ANDCOMMUNICATIONS LINKS

CHECK SIGNAL ANDCOMMUNICATIONS LINKS

Fig. 3 - TIPI segment deployment functional flow.

the TIP! DC/ SR segments can bedeployed and put into operation withoutthe pas.,.:geways, permitting randomorientation of the shelters for such pur-poses as utilization of natural terrain andfoliage as camouflage.

Environmental requirements

RCA. as the system integrator, specifiedthe complement of TI PI DC/SR equip-ment to assure that when installed withinthe shelter, the configuration would sur-vive the complete range of militarymobility hazards. An auxiliaryenvironmental control system maintainsthe shelter environment within a rangeenabling operators to work full shifts on a24 -hour basis. .The control sytem alsoassures the proper operating environ-ment for storing magnetic equipmentwhich may be adversely affected byhumidity; a minimum of 20% (±5',j)relative humidity is maintained.

I he Environmental Control System con-nected to each configured shelter main-tains the temperature and maximumrelative humidity within human comfortconditions. This temperature and

humidity range is also within the safeoperating limits of the equipment insidethe shelters. The ducts within the sheltersdistribute the conditioned air at a lowuniform velocity, and with smalltemperature variations at all operatorstations.

Humidification is a unique requirementfor the TIN DC'/ SR system. 'tree of theconfigured shelters contain magneticrecording equipment which is sensitive tothe extemely low relative humidity whichoccurs in desert areas and arctic areas inwinter. A humidifier is installed in each ofthese three shelters to maintain theminimim relative humidity at 15%.

The addition of humidity at lowtemperature causes a moisture condensa-tion problem; thus, all shelter surfaces, incontact with humidifed areas, are eitherheated or insulated so that the surfacetemperature is above the dew point. Aninsulated blanket is provided over the airducts connecting the shelter to the ECS toavoid condensation.

Structural requirements for the con-figured shelter design are derived from

the rail impact test at 9 mi/ h. A pair ofelastomeric skids (provided at the base ofthe shelters) attenuate the shock loadsdeveloped during the 12 -inch flat andcorner drop tests. Elastomeric skids arealso provided under the environmentalcontrol system pallet to attenuate shock,in addition to acoustic noise attenuation.

The equipment has been designed towithstand steady-state loads of 25g forsmall components and 15g for largecomponents in the axis of rail impact(both directions) and 25g and I 5g in thevertical axis to withstand the drop tests.For lateral loads the design criteria are Kand 5g. I o meet the requirements of theroad test, and vibration test, all looseitems such as transport cases are fasteneddown secur,:ly with straps.

All doors and openings in the shelter areEMI tight including the rectangularopenings which interface with the ECSducts. Ground stakes and lightning rods.are also provided for the configuredshelters. A Scotch .1 -read insulatingmaterial is cemented over the shelter floorto provide electrical isolation. Tie -downdevices in shelters and the passagewayprevent overturning in high winds.

Human engineering

The equipment was designed in confor-mance with human engineering designcriteria. Human factors design re-quirements are employed to determineoperator interfaces which areanthropometrically correct for height.arrangement, illumination, legibility andaccessibility. Environments conducive togood operator performance and reducedfatigue are maintained. Functionaloperator areas provide low acoustic noiselevels to enable operators to maintainefficiency.

Safety is designed into the system suchthat no chassis or assembly withpotentials in excess of 70 volts tins isexposed. and appropriate lightningprotection is provided. There are nohazards from implosion, gases, acids,alkalis, or fumes. Equipments aredesigned for two -man carry whereverpracticable.

Legibility of displays, printed pages, andcontrol panels is designed to be compati-ble with normal vision and normal view-ing distances in the ambient illumination

81

provided within the shelter. Non -reflective CR-I. shields arc used to reduceunwanted reflections. Control colors aredesigned to simplify operation andmaintenance.

Accessiblility is designed into the shelterconfigurations to assure operation andmaintenance procedures. Equipmentcontrols and indicators arc located on thefront of the equipments. External cableconnection terminal boxes are located onthe outside of the shelters to enable rapidset-up and dismantling operations.

The Environmental Contol System(ECS) controls ventilation, air flowvelocity, temperature and relativehumidity including filtration of theventilation air. [he static airspace peroccupant (based on four men per shelter)is approximately 200 ft' per man. Theminimim ventilation rates of 30 ft' / m to45 ft'/ m per man prevent the build up ofstale air; i.e., as the effective static air-space decreases, the ventilation flowshould he increased. The static airspacewas determined as follows: the overallshelter volume is 1280 ft' and the staticairspace for each of four men is 320 fe. Itis assumed that the shelter structure,equipment and furnishings displace ap-proximately 120 ft'/ man; therefore, thenet allocated airspace per man is 200 ft'.This is considered the extreme manloading for DC/SR shelter habitability.

The air flow velocities for either coolingor heating is maintained between 20 and25 It m. The discharge louvers arelocated so that airflow is not directed atthe personnel; air flow velocity is con-trolled by using large area dischargelouvers with orifice plates.

The effective temperature is a compositeindex of dry and wet bulb temperatureswhich relate humidity and ambienttemperature to air velocity. The comfort/one for a sedentary work environmentwith an air velocity of 20 to 25 It in andrelative humidity of 30 to 70' is main-tained by controlling the dry bulb(ambient) temperature between 68° and85"F with control of vapor pressure to 7and 8 mm Hg. The temperature uniformi-ty within the shelters is maintained within10"F between the floor level and headlevel and within 5°F at any operator'sstation.

The general illumination for the shelters

is provided by fluorescent luminariesalong the center line of the longitudinalaxis of the shelter ceiling. Fluorescentluminaries are also provided at the situa-tion maps in the analyst shelters and atthe switchboard in the communicationsshelters. Incandescent lamps are providedat the map stations and work tables in theanalyst shelters and at the map tables inthe non -digital shelters. Portable lightsare provided for maintenance tasks in allof the shelters.

The maximum permissible acoustic noiselevel in the interior of the shelter whenmeasured at any operators position dur-ing self-contained integrated operationwith the ECS operating shall not exceed anoise criteria (NC) level of 60. Theacoustic noise was minimized by applica-tion of one -inch thick acoustical materialon the shelter walls and ceiling, locationof the ECS up to 10 feet from the shelter,use of soft flexible air ducts, and locationof the diesel power generators up to 100feet from the closest shelter. Noise baffl-ing and acoustic material were used in allperipheral equipments which has a noisesource such as fans and mechanical move-ment of subassemblies. Acoustic noiselevel of cooling fans and blowers was amajor factor in their selection.

Maintainability

The equipment was designed formaintenance by military personnel in thefield. Front and/ or top access is providedand modular construction enables rapidrepair by replacement of plug-inassemblies for 95% of the repair actions.Automatic error detection and faultisolation is provided by either hardwareor software. Maintenance is enhanced byilluminated operational status panels,equipment status indicators, andauxiliary maintenance controls.

Installation of the equipment in theshelters, ECS, and power distributionpallet affords maximum access for opera-tion and maintenance. The equipmentdrawers arc mounted on slides providedwith locks to secure in open position.Easy and ready access to all assembliesand their interior parts or components foradjustments, checkout, repair andremoval, or replacement of parts withoutdamage to adjacent units and withoutextensive disassembly was a major designconsideration. Equipment design reliedon the use of standard parts, modular

replaceable assemblies, avoidance ofspecial tools, use of captive hardware.adequate physical access, and lifting aideson assemblies.

Interconnections and cabling

Interconnections and cabling areparticularly important in the TIPIDC SR segment because of EMI/ EMC.TEMPEST, and the need for secureconnections. As a result, the configuredshelter internal cables associated with thetransmission of control and data signalsare considered in the red (secure)category. These cables are separated fromthe black (unsecure) telephone cables inaccordance with the minimum separationand crossover criteria.

Power wiring between each equipmentwithin the shelter and the power entrypanel is filtered at the power entry panel.The filtering eliminates the red cable datafrom being transmitted outside theshelter via coupling to power cables.

Cables within each configured shelter arerouted through ferrous metal racewayslocated near the floor level or overheadabove the dropped ceiling. Irhe signal andpower cables are run in separate raceways(separated by distances dictated by elec-trical requirements).

I he signal and power cables enter theshelter via separate stepped entrancepanels. Weatherproof connectors areprovided for all external connections.

Each different type of cable is keyed toprevent incorrect connections. Thenumbers and types of cables were kept toa minimum to enhance deployment andlogistics. All signal cables except fortelephone cables connect to the shelterentry panel. l'he telephone, power andECS control cables connect to the shelterpower entry panel. This separationprovides isolation between secure linesand unsecure power and telephone lines.The signal (digital) cables are routedwithin the deployment pattern whileblack telephone cables are routed outsidethe deployment pattern. This providesphysical separation for EMI purposes.Additional EMI security is provided byuse of shielded twisted pairs and EMIfilters at the communication shelter cableentry panel on telephone lines interfacingwith equipments outside the DC/SRsegment.

82

Convenient interferometric method forrefractive index determinationM.M. Hopkins) Dr. A. Miller

A Mach-Zehnder interferometer is used for the determination of refractive indices bymeasuring the change in optical path length through a sample as it is rotated about avertical axis. The apparatus works equally well with low index as well as high indextransparent optical materials. Interference orders are counted electronically. Therefractive index of routinely processed samples that are moderately wedge shaped withnon -flat surfaces can be measured to better than 1% accuracy.

Maxwell M. Hopkins. Physical Electronics Laboratory,RCA Laboratories, Princeton. New Jersey. joined RCALaboratories in 1957 as a research technician after threeyears of schooling at Hanover College. Indiana. Heattended night school at Polytechnic Institute ofBrooklyn and completed the requirements for the BS inPhysics. with the degree being granted in 1961 fromHanover College. Subsequently, he was appointed aTechnical Staff Associate. Presently. he is engaged inthe study of new materials useful for modulation andprocessing of light by electro-Optic and related effects.Previously he had worked in the field of ferroelectrics. Hethen helped design and construct a high -temperaturecarbon -arc imaging system for the growth of singlecrystals of refractory oxides namely zirconium oxide andhafnium oxide. Later. he was engaged in the spec-troscopy of laser materials and designed apparatus forhigh -temperature differential analysis to determinephysical properties relevant to crystal growth from themelt. phase transition temperatures, and the magnitudeof energies involved in phase changes. His work on aBismuth Titanate Display Panel. involved a polingtechnique to eliminate unwanted domain structures thatprevented a uniform optical response in bismuthtitanate. The work on non-linear optical materials led tothe introduction of optical waveguide in bariumsodium niobate. Phase -matched frequency doubling ofthe cw output of a Nd-YAG laser was demonstrated insuch a waveguide. Mr. Hopkins has authored and co-authored five papers and a chapter in a book. One invitedpaper has been presented at a meeting of the Electro-chemical Society. He has one issued U.S. patent and onepending. Also. he is listed in American Men of Science.

Authors Hopkins (left) and Miller.

Dr. Arthur Miller. Physical Electronics Laboratory. RCALaboratories Princeton, New Jersey. received the BS inChemistry. Summa Cum Laude, from BrooklynPolytechnic Institute in 1951 He attended the CaliforniaInstitute of Technology. where he was granted tie PhDin Chemistry in 1957 His thesis research there was onthe determination of the structures of intermetaltic andorganic compounds by x-ray diffraction methods. Dr.Miller spent two summers at Brookhaven NationalLaboratory developing radio -chemical procedures andone summer at Los Alamos Scientific Laboratory per-forming research in actinide chemistry. Upon joiningRCA Laboratories in 1956. he worked mainly in the areaof magnetic materials reserach. His studies includedspectrophotometric investigations of transition metalcations, elucidating the valence behavior in spinets.formulation of a theory explaining the distribution ofcations in spinets. and development of methods for thepreparation of new materials for magnetic recordingmedia. In 1964, he initiated a research program in thearea of electro-optic materials. He has extended thisprogram to include non-linear optical materials andoptical devices. Dr. Miller is the author of over two dozenpapers dealing with crystallography, radiationchemistry. the crystal chemistry of spinets. and electro-optics. He has been issued eight U.S.patents and has anumber of others pending. He is listed in American Menof Science. and is a member of the AmericanCrystallographic Association. the American PhysicalSociety. Phi Lambda Upsilon. and Sigma Xi. He is therecipient of two RCA Laboratories Outstanding Achieve-ment Awards

GENE:R.11.1.Y, interferometry is

thought of when accuracies better than apart in 104 or 10' in the determination ofthe refractive index are required. Whenrefractive indices > 1.8 are measured, twoof the mo7e common methods used arethe Duc de Chaulnes' method and theminimum de\ iation method. I he I )uc de('haulnes method is a microscope focusdisplacement method and has about thesame aCCULIC\ as the one we are about todescribe hut requires much more effortfur a refracti\ e index determination. I he

minimum deviation method is the mostaccurate but requires the fabrication of alarge aperture prism.

I he met hod described here is a modil ica-tion of the technique described by

Proctor, yon tiardrotf.' and later refinedsomewha: by Shumate.' I hey used adouble pass interferometer 01 theMichelson and I w yman-tireen type. Weuse a single pass Mach-tehnder in-terferometer to determine the change in(optical path length through a plane -parallel plate of material as it is rotatedIrons normal incidence through an angle0. Hie change in interference order(fringes) caused by the rotation is noted.From a separate determination of thesample thickness and a knov.ledge of thewavelength of the incident radiation v,ehase all the information necessary tocalculate the refractive index ()I the sam-ple. Experimentally. we base found themethod accurate to better than I' i ortypically t = 0.007(n = 2). I he limit()I the accuracy is determined by theerrors in the Iringe count and thicknessmeasurement.

Description and analysis of themethod

I he increase in optical path lengththrough the sample is measured by usinga Mach-lehnder interferometer (Fig. I ).

As the sample is rotated away Irom

Reprint RE -21-3-20Final manuscript received November 15. 1975

83

normal incidence there is a resultingchange in interference order caused bythe increase in optical path lengththrough the sample as a function ofrotation angle. The fringes are countedelectronically. Others have mainlycounted the fringes manually. However,using thicker materials and higher indexmaterials, the counting of possibly 1000fringes manually would be much toocumbersome.

Fig. 2 shows the sample rotated an angle 0from normal incidence. Assuming thatthe index of refraction of the referencemedium rt = I, we can readily show interms of the rotation angle 0, thethickness of the sample t and therefractive index of the sample n, that thenet optical path length P(0) is

P(0) = /kn. - sin 'B)' - cos 0]. ( I )

l'he change in optical path length AP is

= I(0)- P(0)= 4[(n- - sin -0)'2- cos 0 - n+ It (2)

where P(0) is the optical path length atnormal incidence.

1 he change in interference order in for aMach-Lehnder interferometer is

In = AP/A, (3)

where A is the wavelength of the incidentradiation. Substituting Eq. 2 into Eq. 3and solving for refractive index n of thesample we have

-x2 - 2xcosO + 2x + 2cos0 - 2n - (4)

2x + 2cos0 - 2

where x = 1.

1441-N LASER

BEAM EXPANDERWITH SPATIAL FILTER

Si BEAM SPLITTER

OPTIC AXIS OFINTERFEROMETER

SAMPLE ROTATED ANL B TO INCIDENTRADIATION

I

-SAMPLE AT NORMALINCIDENCE

Fig. 2 - Sample shown rotated an angle e to incidentradiation. Smaller samples would require smaller rota-tion angles.

The interferometer is illuminated with632.8-nm radiation from a He-Ne laser,as shown in Fig. I. Prior to entering theinterferometer the laser beam is ex-panded to 10 -mm diameter by a laserbeam expander with spatial filtering. Thesample is mounted on a goniometer head'attached to a smoothly rotating table thatrotates about a vertical axis. The tablerests upon a thrust bearing that is

responsible for the necessary smooth,vibration -free rotation. Fringes arecounted electronically by a

photomultiplier tube (PMT) connectedto an electronic counter. A 0.43 -mmaperture and narrow -band interferencefilter for 632.8-nm radiation are placed infront of the PMT that allows workingin normal room light. A fixed rotationangle of about 30° is used. Since ourequipment does not allow us to measure 0with any reasonable accuracy wecalibrated the angle by determining n, ofcrystal quartz and using Eq. 4 to find 0.The refractive index n and cos 0 areexactly interchangeable in Eq. 4.

The sample is aligned with the front facenormal to the optic axis of the in-terferometer by placing a small aperturebetween beam splitter SI and mirror MIand retroreflecting the beam from thefront face of the sample upon the aper-

M2 MIRROR

NV POWERSUPPLY

APERTURE

MI S2

MIRROR \- OPTICAL STOPSAMPLE

ELECTRONICCOUNTE R

OSCILLOSCOPE

000 9 cT.1

Ism(

NARROW -BANDINTERFERENCE FILTER

ENLARGED FIELD SHOWNECLIPSED BY OPTICAL STOP

Fig. 1 - The experimental set-up showing the Mach-Zehnder interomeler with the location of the sample and optical stop.The inexpensive mirror type beam splitters have intone) coatings. They are flat to approximately one -tenth the wavelengthof light and are parallel to 20 seconds.

ture. If the sample is not wedged, thisposition is the extremum in the fringemovement and can also be found byobserving the fringe movement, but withless accuracy. Any lack of parallelism inthe sample faces can be easily detectedbecause the beam is also reflected fromthe back face of the sample producing asecond spot. Shumate' made an analysisof a wedge-shaped sample and found tnatthe error in the refractive index did notexceed l0-4 even for a wedge angle of0.25°.

The angular range over which the sampleis rotated is determined by a mechanicalstop at one end and a metal vane thateclipses the interferometer beam of lightpassing through the sample at the otherend. The optical stop is used to eliminatespurious fringe counts caused byvibrations from a mechanical stop. Whenthe sample is positioned normal to theoptic axis of the interferometer, theoptical stop is adjusted to eclipse one-halfof the field. The PMT (mounted on an x -y -z translator) with the 0.43 -mm diameteraperture is then very accuratelypositioned just into the eclipsed field at apoint where a tap on the laser table willnot cause a fringe count. I- his position issensitive to 0.001 in. this techniqueallows us to reliably position the PMTaperture from sample to sample andmeans all samples are rotated through thesame 0. We also found it was particularlyimportant for the electronic counter weused to adjust the PMT voltage fromsample to sample in order to keep thephotovoltage excursions between fringemaxima and minima about the same forall runs.

Experimentally it was most convenient toinitiate sample rotation from themechanical stop at the extreme angle ofincidence and rotate through 0 to normalincidence where the optical stopterminates the fringe count by eclipsingthe PMT aperture. The sample is rotatedin this manner because the fringe density

84

Table I - Typical values and errors for a plane parallel sample having a refractive index Table II - Crystal orientations and the refractive index determined using polarizedof 2. light.

I unable IVim al value Lrror

A 632.8 nm Negligible0' 30" ± 3 minni 950 + 1

i 0.115 cm ± 13 pm

Negligible4.7 X 104.6 X 106.6 X 10'

II is a fixed rotation angle and has to he calibrated on our apparatus.Measured with a micrometer.

Orientation Refractive index determined

Isotropic Anyllniaxial [.Optic axis parallel Electric vector parallel to

to rotation axis optic axis determines n(preferred orientation Electric vector, perpendicular

to optic axis determines n2.0ptic axis perpendicular Electric vector perpendicular to

to rotation axis optic axis determines it.Biaxial The principal axis must coincide with both the rotation axis and

polarisation axis to determine n for that axis.

is greatest at the maximum angle ofincidence and the system would be moresusceptable to errors in the fringe count ifthe optical stop were placed there. Weactually rotate the sample through 0 byhand in approximately one second.

A degree of optical imperfection of thesample causing poor interferencepatterns does not appreciably degrade theaccuracy of the method if handledproperly. When a highly curved orbullseye-type interference pattern is ob-tained, the center must be positioned overthe PMT aperture and remain therethroughout the sample rotation.Otherwise, any movement of the center of

Fig. 3 - Interterograms of lithium mobate. Both crystalsare good optical quality but Fig. 3a has been processedto laser quality. Fig. 3b is a typical fringe pattern causedby poor processing.

the interference pattern relative to thePMT aperture as the sample is rotatedwill cause an error in the fringe count.Two fringe patterns are shown in Fig. 3.Fig. 3a is an interference pattern of a laserquality LiNba crystal. Fig. 3b is a

respresentative poor fringe pattern of agood optical quality LiNba crystal thathas been poorly processed. However,there was no problem counting thefringes in Fig. 3b. The error in determin-ing n1., and n, for both samples of LiNbawas 0.6%. Experimentally we find if aninterference pattern can be obtained witha fringe spacing greater than the PMTaperture the fringes can be counted. Thebasic limitation of the accuracy of theexperiment is determined by the measure-ment of the interference order in and thesample thickness 1; A and 0 should remainquite constant. Errors in in show up asvariation in n for successivemeasurements on the same sample, anderrors in the measurement oft show up asa systematic offset from the real value ofthe refractive index. Table 1 is a list of thevariables in Eq. 4, their typical values,and the errors we experience in each andtheir effect upon n.

The method can be used to measure theprincipal refractive indices of suitablyoriented crystals using polarized light.The crystal must be cut with the axis to bemeasured lying in the plane of the plate.Table II lists different orientations forisotropic, uniaxial and biaxial crystals,and the refractive index determined usingpolarized light.

The refractive indices of several liquidswere determined by using a fused quartzspectrometric cell. Generally, the fringepatterns are very good because of thequality and alignment of the spec-

trometric cell windows. the procedure isto measure the change in interferenceorder of the empty cell, then measure thechange in interference order of the cellcontaining the liquid. The interferenceorder caused by the liquid is?i1ix = Mho,

mill.

The overall accuracy of the method wasdetermined by measuring the refractiveindices of several transparent opticalmaterials and liquids. We covered therefractive index range from 2.3 to 1.33using crystals of LiNba, CdS, fusedquartz, crystal quartz, and two liquids,methylene iodide, and HBO. Errors indetermining n were always less than 1%.

Conclusion

The method described for refractive in-dex determinations using a Mach-Zehnder interferometer appears to workequally well for low index as well as highindex transparent optical materials. Thinsamples as well as thick samples can beaccommodated. The interference ordersare counted electronically. A particularadvantage of the apparatus is that it canhandle routinely processed samples thatare moderately wedge shaped and stilldetermine the refractive indices to betterthan I( ( accuracy with very little effort.

References

I . I.eDuc de Chau Ines. "Sur qualques experiences relatives a ladiplititie," Huron, de f ilademie Royal des Sciences. 1767.162.

2. Proctor. C.A.: "Index of Refraction and Dispersion with theInterferometer," Phys. Rev., 24 (1907) p.195.

3. von Nardroff, E.R.; "New Interferometer Method forRefractive Indices; Science, 17 (1903) p.861.

4. Shumate, M.S.; "Interferometric Measurement of LargeIndices of Refraction." Appl. Optics, 5 (1966) p.327.

5. Available from Charles Supper Company, Natick. Mass..01760.

85

Engineering andResearchNotes

Separating silicon chips to permit chip butting

E.P. HerrmannAdvanced Technology LaboratoriesSomerville. N.J.

Certain electronic systems require integrated circuits that aredimensionally longer than are practical, or possible, to fabricate.Conceptually, it is a simple matter to design submodules on which thecircuitry extends to the chip peripheries and then to butt the chips end toend. There is a serious problem, however, in separating the chips (inwafer form) so as to allow them to be butted with fine geometricalprecision. Separation of chips on a wafer by scribing and fracture notonly leaves ragged edges on the chips but also tends to mechanicallystress (diamond scribe) or thermally stress (laser scribe) the circuitry atthe edge of each chip. In addition, it is extremely difficult to perform themechanical alignment (for accuracies and tolerances less than I mil)required to accomplish these procedures. This leaves chemical etching asthe only practical separation technique where subsequent butting isrequired.

Chemical etching is routinely used to define integrated circuit patternson silicon. The etches generally undercut the etch masks to some extent.but the depth of etching is small and the undercutting is negligiblecompared to the gross definition. In etching through a standard 6 to 10mil wafer, however, the undercutting would be so severe as to eithercause mask failure or at best not he sufficiently reproducible to yieldfractional -mil accuracy.

The present technique involves a shallow front side (circuit side) etchwherein the required accuracy is achieved, followed by an oversizeddeep back -side etch or wire saw cut for which the required tolerance isnot so great. Front -side mask alignment is done by conventional means.while back -side mask alignment can be accomplished with an infraredaligner which permits one to see through silicon.

To achieve the front -side etch with fractional -mil accuracy (e.g., controlof mask undercutting), one can take advantage of an anisotropic siliconetch. In particular, if a (110) wafer is etched with any of the reportedanisotropic silicon etches, and the etch mask is aligned to the vertical(III) crystallographic planes, it is possible to etch long slots with verticalsides. Undercutting is approximately 10% of the etch depth, and it ispredictable and reprodUcible. The slot depth should be deep enough (1to 2 mil) so that subsequent processing on the back side will not affect orstress the active -side silicon. The slot definition is as good as the ICdefinition. The circuit should be aligned to the direction of the vertical

planes (at least the street area of the circuit). These planes are easilylocated by etching a shallow slot, with an anisotropic etch, in the generaldirection of the planes (determined from the wafer flat) prior to circuitfabrication. The etch will undercut the mask along the vertical plane andthus define the exact direction to which the circuit must be aligned. It isexpeditious to define both front and back side etch masks to perform theshallow etch on both sides of the wafer. Following the shallow etch, aprotective coating is applied on the front surface and the walei i.mounted on a substrate (wax on glass) with the circuit side down. I hewafer must be mounted to protect the circuitry from the long etchthrough the back side.

Either an isotropic or an anisotropic etch may be used on the back side.The back -side slots should be wider than the front slots to case maskalignments and to assure that the chip edge is defined by the front -sideslots.

Alternatively, the back etch may be replaced by a wire sawingtechnique. The shallow etch is performed on the back to provide wire-saw alignment. The wire diameter of the saw should be larger than thefront slot to produce an overhanging or cantilever structure with theedges to be butted defined by the front etch.

Reprint RE -21-3-71 Final manuscript received May 19. 1975.

Thyristor trigger circuits employing capacitor-loaded bridge rectifiers

M.A. KalfusSolid State DivisionSomerville. N.J.

.1.he circuits of Figs. I and 2 provide time -dependent alternating polaritytrigger pulses phased to trigger a thyristor in its most sensitive modes (III ). The circuits, differing only in the connections of a bridge rectifierand current -sensing resistor, are easily adaptable to provide either oftwo desired operating modes. The circuit of Fig. I. for example.provides a momentary turn -on load current upon application of acpower and is useful for energizing the starting winding of an electricmotor to provide increased starting torque. The load current can becontrolled to vary between an initial lull power value and a final halfpower value which is desirable when driving load devices has ingvariable reluctance magnetic circuits (relays, solenoids. etc.) to provide

86

10-0

12

F

01 D2

W1t laRI

D3 04

R3

R2

[WADIDIAL

TRiAC T-

9

Fig 1 - Momentary turn -on

Di 02 R264 1 ; 1.3

RI

R DIAC [LOAD]

TRIAC

6

T 2CIT4

,-P Dt

03 D4 Ti

Fig 2 - Gradual turn -on.

Fig. 3 - Alternativebridge arrangement

most economic use of available energy. The circuit of Fig. 2 provides agradual turn -on of load current as needed, for example, to pros idc a"soft start" for motors used in electric hoists or elevators. In each circuit.the final value of thyristor conduction angle is easily controllable byselection of component values.

In Fig. I, a current sensing resistor R: connected in series with acapacitor -loaded full -wave bridge rectifier i(Di -134,C1) across ac inputterminals (10, 12) produces trigger signals proportional to the bridgecurrent. The trigger signals, conditioned by a conventional pulse -shaping network (12:,C:, diac) are applied to the gate of a triac forcontrolling current flow through a load. The connections are such thatthe trigger signals arc of a sense relative to the triac main terminals(1.1,T2) to trigger the triac in its most sensitive modes

In operation (assuming C, is initially uncharged) application of acpower initially causes a relatively large current flow through currentsensing resistor R: since the bridge impedance is relatively low due to theuncharged condition of capacitor C1. As C1 charges, the potentialproduced acts as a "bucking" voltage which effectively increases thebridge impedance thus lowering current flow through 12:. Eventually. ifR, were not present, C1 would charge to substantially the peak value ofthe ac input signal and thus reduce the current through 12: tosubstantially zero. Since the thyristor is triggered in proportion to theft:current, its conduction angle and hence the load current varies from aninitially high value to substantially zero as the capacitor C1 charges(assuming R1 absent).

A discharge path for capacitor Cu is needed to return the capacitor to itsinitial uncharged condition when ac power is removed. This could beprovided by a relay (not shown) having its coil connected across the acterminals and the normally closed contacts across the capacitor. A moreeconomic approach is provided by resistor R, which, in addition toproviding the power -off discharge path, determines ( in ratio with R thefinal value of the triac conduction angle and thus the final value of loadcurrent. If R, is selected to be very large compared to R: the final valueof potential developed across R: can be made to be less than the diactrigger voltage so that the final value of load current is zero (for motorstarting -winding control applications). Choosing a smaller R 1 K: ratiocan result in non -zero final load -current value (for solenoid or relay -driver applications).

The circuit of Fig. 2 is identical to that of Fig. I except that the bridgeand current -sensing resistor connections to the ac power terminals arcreversed. The operation of these two elements is the same as previouslydescribed but since the trigger signals are proportional to the bridgevoltage (rather than the R: voltage), the triac action is complementary tothat of Fig. I. Here, upon application of ac power. no voltage isproduced across the bridge (capacitor Cu uncharged) so the triac is

initially not triggered. As C, charges, the potential across the bridgeeventually is sufficient to begin triggering the triac at a minimumconduction angle which gradually increases to provide lull conductionof the load current. The previous remarks regarding resistor R , apply. tothis circuit as well except that the R 12: ratio now controls themaximum value that the load current can achieve. By making this ratiorelatively large, essentially full power can be delivered to the load. Asmaller ratio can be selected to limit the linal value of load current tosomething lower, if desired, or R1 may be replaced by a relay aspreviously noted.

The rate of change of conduction angle is controlled in both circuits byselection of the 12:Ci time constant. The alternative bridge arrangementof Fig. 3 may be substituted for the bridge in either figure to achieve areduction in the number of diodes 4D1D4 eliminated) at the cost ofrequiring an additional capacitor and resistor (C1'.121').

Reprint RE -21-3-71 Final manuscript received November 26. 1974.

High -dynamic -range, low -noise preamplifier

Av. di..

L.A. KaplanSolid State DivisionSomerville, N.J.

4.

A general arrangement for a volume -controlled amplifier having a highinput impedance, high dynamic range and low noise is shown in Fig. I. Aspecific configuration utilizing an RCA type CA3094 integrated circuitoperational amplifier is shown in Fig. 2.

Fig 1 -- High -input -impedance. high -dynamic -range. low -noise amplifier

Fig 2 - Amplifier arrangement of Fig. 1 with CA3094 Op Amp

87

This type of amplifier is particularly useful. for example. as amicrophone or musical instrument input amplifier where a highdynamic range of input signals is to be expected. The overloadcharacteristics of the amplifier improve as the volume control isdecreased.

Specifically, referring to Fig. I. the variable resistors RI and 124 areganged together so as to reduce the gain of the operational amplifier atthe same time that the portion of the output signal selected from acrossR4 is reduced. At the maximum volume setting of R4. the full resistanceR, is included in the feedback path of the amplifier so that the voltagegain may be expressed as

E,,, Es = + (111 + ROI K1

At the minimum volume setting, the portion of R; included in thefeedback is also at a minimum (e.g., zero). Thus, the output signal level isdecreased due to the potentiometer action of R4 while the gain is reducedbecause the value R: + R, also is decreased.

At low volume settings, therefore, the overload point of the amplifier isincreased while noise output is reduced. The input impedance of theamplifier will be relatively high with the signals applied to thenoninverting (+) terminal. Where the potentiometers are linear, thevoltage gain may be expressed as

&,j Es= 11012 + A (R: R1) + A

where A is the fractional rotation of the control. This results in aparabolic variation of signal as a function of the position of the control.

Reprint RE -21-3-7) Final manuscript received February 10, 1975.

Automatic pedestal compensation fortv data processing applications

Hugo Logemann, Jr.Government Communicationsand Automated Systems DivisionBurlington, Mass.

Television camera systems normally clamp video during horizontal fly-back, thereby preserving the dc signal component due to ambient lightlevels. In commercial television, the camera operator maintainsstandard signal levels and the consumer (viewer) has complete freedomto manually adjust his received brightness and contrast control to alterthe dc and ac signal components to suit his ambient -light viewingconditions. Further, camera lighting in the television studio is adjustedto be uniform over the field of view and to control dynamic range. Anyvariances which occur are readily manually accommodated by theviewer.

There are many systems, other than commercial television, whichemploy tv sensors. For instance, in such systems as optical characterreaders, pattern recognition systems, or as sensors for surveillancesystems, tv sensors are frequently used under conditions where it isnecessary to process video signals automatically. In such systems, it isgenerally desirable to process video signals before display and withoutthe aid of an operator making manual adjustments. Further, the lowfrequency signal component, such as that relating to ambient lightconditions, usually contains no useful information and if not removed,imposes a larger dynamic range requirement on signal processing

circuits than would otherwise be needed.

A simplified idealized camera video signal is shown in Fig. I. Althoughsimplified, this waveform is representative of a video signal from a tvsensor used in satellite detection systems where essentially all signals ofinterest result from point sources. In this case, the ambient lightbackground depends on where the telescope is pointing, being relativelylow in looking at the celestial pole and relatively high when looking neara full moon.

CLAMPLEVEL)- -

SYNCPULSE

ACTIVE

LINE TIME

- - -

LBLANKTIME

A "SET UP PEDESTAL LEVEL WITH LENSCAPPED (NO LIGHT)

B PEDESTAL LEVEL, LENS CAP OFF

C VIDEO.SIGNAL COMPONENT WE TOBACKGROUND LIGHT

D USEFUL VIDEO SIGNAL COMPONENT

(OBJECT)

Fig. 1 - Simplified idealized camera video signal.

The object of the circuit shown in Fig. 2 is to remove automatically thisvariable background, due to variations in ambient light which are of nointerest. In so doing, it is important that signal levels are assessed onlyduring the active line time of each tv scan; and that signal levels duringthe blank interval be excluded. These requirements are met with thecircuit shown in Fig. 2. in which a semiconductor switch is closed onlyduring the active line time of each tv. scan.

BLANK

OPEN SWITCHDURING BLA

^4t. 11%-RC± Ai

El RI R2

E2 R3R4

A2 DIFFERENTIAL AMP

EeuyPE2-Ei

LW_FR243,R4

Fig. 2 - Pedestal compensation circuit.

I o -

Eour

Specifically. if the semiconductor switch has an open resistance ot

in the order of 10' ohms and has a closed resistance R4. ,brd only in theorder of 10- ohms, then the charging time constant of capacitance Cthrough the semiconductor switch and resistance R can be made suchthat the time constant with the semiconductor switch open is muchgreater than the active line time r, while the time constant with thesemiconductor switch closed is short with respect to the active line time rbut is still much greater than the duration S of a video signal component:i.e.,

(R,,, ,,, + R)C >> r( R,. + R)C >> S.

Therefore, while the semiconductor switch is closed. capacitance C willchange to the potential El. shown in Fig. 2 as the ambient backgroundpedestal level. Further, the ambient background pedestal level Ei will bemaintained from line to line as the tv raster is scanned. The high inputimpedance of buffer amplifier Ai assures that capacitance C will not bedischarged through this path. The inputs to the differential amplifier Alare E1, the ambient background pedestal level, and E2, the clampedcamera video signal. The differential amplifier Al output is E1 - El. sothat the ambient background component (object) is now alwaysreferenced to ground. (0 volts), independent of the background level.

Reprint RE -21-3-7) Final manuscript received November 26. 1974

88

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 after his name in the subject -index entryFor additional assistance in locating RCA technical literature contact RCA TechnicalCommunications, Bldg. 204-2. Cherry Hill, N.J. (Ext. PC -4258).

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 hack coven

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 where

necessary to bring significant keywords(s) to

the left to easier scanning. Author's division

appears parenthetically after his name.

SERIES 100BASIC THEORY &METHODOLOGY

105 Chemistryorganic inorganic. and physical

INTERFACIAL FREE ENERGIES betweenpolymers and aqueous electrolyte solutions- R Williams (Labs,Pr) J. of PhysicalChemistry, Vol. 79, No. 13 (1975) pp. 1274-76.

125 Physicselectromagnetic field theory, quan-tum mechanics, basic particles,plasmas. solid state. optics, thermo-dynamics, solid mechanics, fluidmechanics, acoustics.

CHEMICAL VAPOR DEPOSITION, Thegrowth of thin films of lithium niobate by -B.J Curtis, H.R. Brunner (Labs,Pr) Mat ResBull., Vol 10 (1975) pp 515-520.

ELASTIC CONTINUM THEORY cutoffs andorder in nematics and solids. Comments on -P.Sheng, E B Priestley. P.J.Woitowicz(Labs,Pr) J. of Chemical Physics. Vol. 63, No2 (July 15. 1975)

HOT ELECTRONS - R.S. Crandall (Labs,Pr)Physical Review B. Vol. 12, No. 1 (July 1.1975).

PARTICLE VELOCITY probability densitiesmeasured in turbulent gas -particle ductflows. One-dimensional -C E4 C.I, E L

Peskin (Labs,Pr) Intl. J. of Multiphase Flow,Vol 2 pp 67-78.

130 Mathematicsbasic and applied mathematicalmethods.

POISSON'S equation, Computer solution ofone-dimensional -R.W.Klopfenstem, C.P.Wu (Labs,Pr) IEEE Trans. on ElectronDevices Vol. Ed. 22, No.6 (June 1975) pp. 329-333.

160 Laboratory Techniques andEquipment

experimental methods and equip-ment. lab facilities. testing, datameasurement, spectroscopy, elec-tron microscopy. dosimeters.

CV TECHNIQUE for ion -implanted profiles,Limitations of the -C.P.Wu, E.G. Douglas.C W Mueller (Labs,Pr) IEEE Transaction onElectron Devices. Vol. ED -22, No. 6 (June1975) pp. 319-329.

170 Manufacturing and Fabrica-tion

production techniques for materials,devices and equipment

CHEMICAL VAPOR DEPOSITION techni-ques for glass passivation of semiconductordevices -W Kern (Labs,Pr) NAECON '75Record, pp. 93-100.

EVAPORATOR FACILITY for deposition ofmultieiement thin-film patterns -L H Meray(Labs.Pr) J of Vacuum Science &Technology. Vol 12. No 3 (May/June 1975)pp. 709-712.

180 Management and BusinessOperations

organization, scheduling, marketing,personnel.

MANUFACTURING, Canada's future In -GD Clark (Laid .Mont) The Conference BoardRecord. Vol. XII, No. 8 (August 1975)

NEW -PRODUCT PLANNING, Dynamic pricemodels for - B Robinson. C Lakhani(Labs,Pr) Management Science, Vol. 21, No10 (June 1975).

SERIES 200MATERIALS. DEVICES. &COMPONENTS

205 Materials (Electronic]

preparation and properlie, of con-ductors. semi -conductors. dielec-trics. magnetic, electro-optical,recording, and electro-magneticmaterials.

CdSe FILM grown under excess Se flux,Vacuum -deposited - R M. Moore, J.T.Fischer, F. Kozielec. Jr. (Labs,Pr) Thin SolidFilms, 26 (1975) pp 363-370.

In from Ga,In. , Preferential euaportaionof - B Goldstein, D J Szostak (Labs.Pr)Applied Physics Letters, Vol. 26. Np. 12 (June15. 1975).

InGa, ,A,:rie and GaAs:SI prepared byliquid phase epitaxy, A comparison of -MEttenberg, H.F.Lockwood (Labs.Pr) InstPhys. Conf. Ser. No. 24, 1975: Chapter 3.

IRON SULFIDE grown by normal freezing.Composition of single crystal of - KAmetani, T Takahashi (Labs.Pr) Bull ofChem. Soc. of Jap., Vol. 48, No. 21Feb. 1975)pp. 721-722.

LIQUID CRYSTAL with various surfacealignments, Field induced distortions of a -D. Meyerhofer (Labs,Pr) Physics Letters Vol.51A No 7. (April 21, 1975) pp 407-408

LITHIUM NIOBATE and other phworeflectivemedia. Time -dependent characteristics ofphoto -induced space -charge field and phaseholograms in -G A Alphonse. R C A119.0 LStaebler, W Phillips (Labs.Pr) RCA Review.Vol 36, No 2 (June 1975) pp 213-229.

SEMICONDUCTOR PASSIVATIONGLASSES. The relative dielectric constant ofIwo - A M Goodman. K.W. 'fang. J.WBreece (Labs,Pr) J. of the Electro-chemicalSociety Vol 122, No 6 (June 1975) pp. 823-824.

SILICON ON SAPPHIRE (SOS)Mms, Mobili-ty of current carriers in - S. T. Hsu. J.H. Scott.Jr (Labs.Pr) RCA Review, Vol. 36 No. 2 (June1975) pp. 231-239.

SODALITE. A paraelectric defect In - R.C.Alig, A.T.Flory (Labs.Pr) J. of Physics andChemistry of Solids, Vol. 36. pp. 695-698.

ZnS:Mn.:Cu, rf-sputtered films,Electroluminescence in - J.J. Hanak(Labs,Pr) Proc. 6th Intern!. Vacuum Congr.1074. Japan. J. Appl. Phys. Suppl. 2, Part 1(1974) pp. 809-812.

210 Circuit Devices andMicrocircuits

devices (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.

DUAL MODE FILTER for terrestrial and spaceapplications - C.M. Kudsia (Ltd.. Mont) Mid -West Symposium on Circuits and Systems,Concordia University, Montreal (August1975)

FET, Submicrometer self -aligned dual -gateGaAS - R.H. Dean. R.J.Matarese (Labs,Pr)IEEE Trans. on Electron Devices.

MICROSTRIP RESONATORS. Loss con-siderations for -E Belohoubek, E Denlinger(Labs,Pr) IEEE Transactions on MicrowaveTheory and Technique (June 1975) pp. 522-526

PHASE SHIFTER. Microwave Integrated -circuit MIS varactor -R.L.Gamma, B.F.Hitch, S. Yuan, M Ettenberg (Labs,Pr) RCAReview. Vol. 36, No 2 (June 1975) pp. 296-315

PHOTOCATHODES grown onti,iP In (J.I I' substrates. The effect oflattice parameter mismatch in NEA GaAs -R.E. Enstrom, D.G. Fisher (Labs,Pr) J. ofApplied Physics. Vol. 46. No. 5 (May 1975) pp.1976-1982.

PICTURE TUBES, Three -filter colonmetry ofcolor -television -G.M.Ehemann, Jr.(Labs,Pr) RCA Review. Vol. 36, No. 2 (June1975) pp. 254-270

THICK -FILM RESISTORS. Laser -inducedmicrocrack in -a problem and solutions -K.R. Bube (Labs,Pr) The American CeramicSociety. Vol 54. No 5 (May 1975)

TRANSFERRED -ELECTRON LOGICDEVICES by proton bombardment for deviceisolation. Fabrication of 3 -terminal - L CUpadhyayula. S.Y.Narayan. E C Douglas(Labs,Pr) Electronics Letters, Vol. 11, No. 10(May 15. 1975).

VIDICON, Granular metal -semiconductor -C.R. Wronski. B Abeles, A. Rose (Labs,Pr)Applied Physics Letters, Vol. 27, No. 2 (July15, 1975).

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.

ANALOG building blocks - W.A. Clapp(ATL.Cam) Standard Electronics Workshop.WPAFB. Dayton. Ohio (7/21/75).

ELECTRONICALLY PROGRAMMABLEARRAYS, Application trends for- D Hempel(GCA.Cam) NAECON. Dayton, Ohio.NAECON Proceedings (June 1975) pp. 470-478

TRAPATT AMPLIFIERS for X -band opera-tion, A study of second -harmonic -extraction- W.R. Curtice. J.J. Risko, P.T. Ho (Labs,Pr)RCA Review, Vol. 36, No. 2 (June 1975) pp274-295.

220 Energy and Power Sources

batteries, solar cells, generators.reactors, power supplies.

89

PHOTOVOLTAIC CONVERTERS in concen-trated sunlight, The conversion efficiency ofideal Shockley P -N junction - F. Sterzer(Labs.Pr) RCA Review, Vol. 36, No. 2, (June1975) pp. 316-323.

SILICON SOLAR CELLS for highly concen-trated sunlight- R.H. Dean, L.S. Napoli. S. G.Liu (Labs.Pr) RCA Review. Vol. 36, No. 2(June 1975) pp 324-335.

SOLAR CELLS, Short-circuit capacitance ofilluminated - A.R. Moore (Labs,Pr) AppliedPhysics Letters. Vol. 27, No. 1 (July 1975) pp.26-29.

240 Lasers, Electro-Optical andOptical Devices

design and characteristics of lasers,components used with lasers,electro-optical systems, lenses, etc.(excludes: masers).

ELECTROLUMINESCENT DIODES,Reduced degradation in In,(ia - M.Ettenberg. C.J. Nuese (Labs.Pr) J. of AppliedPhysics, Vol. 46, No. 5 (May 1975) pp. 2137-2142.

INJECTION LASERS, Coupled lateral modesin narrow strip - H.S. Sommers, Jr., J.K.Butler (Labs.Pr) J. &Applied Physics, Vol. 46.No 5 (May 1975).

LASER BEAMS Ill, Review of industrialapplications of high -power - I.P. Shka rot sky(Ltd ,Mont) RCA Review, Vol. 36. No. 2 (June1975) pp 336-368

LASER DIODES at room temperature,Reliability aspects and facet damage in high -power emission from (AIGa)As CW - H.Kressel, I Ladany (Labs,Pr) RCA Review. Vol.36. No. 2 (June 1975) pp 231-239.

LASER DIODES, Internal optical losses invery thin CW heterojunction - J.K. Butler, H.Kressel, I. Ladany (Labs.Pr) J. of QuantumElectronics. Vol OE-11.No 7 (July 1975)

LASERS, (AIGa)As double heterojunctionCW: the effect of device fabricationparameters on reliability - I. Ladany, H.Kressel (Labs.Pr) Inst. Phys. Con!. Ser.. No.24 (1975) Chapter 4

PHOTODIODES, Planar germanium - JConradi (Ltd., Mont) Journal of AppliedOptics (August 1975)

SILICON VIDICON, High light level bloomingin the -E.C.Douglas (Labs,Pr) IEEE Trans.on Electron Devices. Vol. ED -22. No. 5 (May1975) pp 224-234

STRIPED LASERS, Internal dynamics ofnarrow - H. Sommers. Jr. (Labs,Pr) J. ofApplied Physics, Vol. 46, No. 4 (April 1975) pp.1844-1846

TRIODE OPTICAL GATE: a new liquid crystalelectro-optic device - D.J. Channin(Labs.Pr) Applied Physics Letters, Vol. 26. No11 (June 1. 1975) pp. 603-605.

SERIES 300SYSTEMS, EQUIPMENT,& APPLICATIONS

310 Spacecraft and GroundSupport

spacecraft and satellite design.launch vehicles, payloads, spacemissions, space navigation.

SPACECRAFT, Nutational stability of an

asymmetric dual -spin - G. T. Tseng(AED,Pr) AAS/AIAA Astrodynamics Cont.,Nassau, Bahamas (July 28, 1975).

320 Radar, Sonar, and Tracking

microwave. optical, and othersystems for detection, acquisition,tracking, and position indication.

RADAR, Advanced tracking logic for com-puter directed (on -axis) - R.J. Pepple, A.E.Hoff mann-Heyden (Svc.Co.,PAFB) Twenty-first Annual Tri-Service Radar Symposium,Naval Post Graduate School, Monterey.Calif., 7/16/75; The 21st Annual Tri-ServiceSymposium Record. (Classified Secret) (July1975) pp. 197-206.

340 Communications

industrial. military, commercialsystems, telephony, telegraphy andtelemetry, (excludes: television, andbroadcast radio).

LASER COMMUNICATION SYSTEM usingelectro-optic modulation, Optimization of a -A.Waksberg (Ltd ,Mont) IEEE Journal ofQuantum Electronics (July 1975)

VFR TERMINAL air -traffic -control facilities,Alternatives for fixed - J.N. Ostis (ASD.Burl)Defense Magazine, England. Vol. 6, No. 3(March 1975)

VIDEO DELTA MODULATION CHANNELS.A technique for correcting transmissionerrors in -A Ali (GCS.Cam) InternationalCommunications Confrence Proceedings.San Francisco, Calif Session 270, Paper 27E.(June 16-18, 1975)

VIDEO TRANSMISSION, Analog vs. digitalanti -jam - C. Haber. E. Nossen (GCS.Cam)National Association for Remotely PilotedVehicles, Dayton. Ohio. Proceedings, Vol. 1

(June 3, 1975) pp. 1-3.

VIDEO DELTA MODULATION CHANNELS,A technique for correcting transmissionerrors in - Z Ali (GCS.Cam) InternationalCommunications Conference Proceedings.San Francisco, Calif., Session 270, Paper 27E,(June 16-18, 1975)

VIDEO TRANSMISSION, Analog vs. digitalanti -jam - C. Haber, E. Nossen (GCS.Cam)National Associates for Remotely PilotedVehicles, Dayton, Ohip; Proceedings, Vol. 1

(June 3, 1975) pp. 1-3.

345 Television and Broadcast

television and radio broadcasting,receivers, transmitters, and systems,television cameras. recorders, studioequipment.

ELECTRONIC JOURNALISM, An NBC viewof - R Mausler (NBC.NY) NAB Workshop.Las Vegas. Nevada (April 6, 1975)

FRAME SYNCHRONIZER, Operational im-plementation of a broadcast television - R.J.Butler (NBC.NY) Journal of the SMPTE Vol.84, No. 3. (March 1975).

360 Computer Equipment

processors. memories, andperipherals.

MNOS/PMOS/SOS MEMORY in space radia-tion environments, Properties of fullydecoded 258 -bit - G. Brucker (AED,Pr) IEEEConf. on Nuclear and Space RadiationEffects, Eureka, Calif., (Jyly 15, 1975.

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.

Advanced Technology Laboratories

Clapp, W.A., 215

Astro-Electronics Division

Brucker, G.. 360Tseng, G.T.. 310

Automated Systems Division

Ostia, J.N., 340

Government CommunicationsSystems Division

AIL Z., 340Haber, C., 340Hempel, D., 215Nossen, E., 340

RCA Laboratories

Abeies, B., 210Alig, R.C.. 205Alphonse, GA., 205Ametani, K., 205Belohoubek, E., 210Breece, J. W., 205Brunner, H.R., 125Bube. K.R., 210Butler, J.K. 240Camisa, R.L., 210Carlson, C.R., 125Channin, D.J., 240Crandall. R.S., 125Curtice. W.R., 215Curtis, B.J., 125Dean, R.H., 210,220Denlinger, E., 210Douglas. E.C., 160, 210, 240Ehemann, G.M., Jr., 210Enstrom, R.E., 210Ettenberg, M., 205, 210, 240Flory, A.T., 205Fischer, J.T.. 205Goldstein, B., 205Goodman, A.M., 205Hanak, J.J., 205Hang, K.W.. 205Hitch, B.F., 210Ho, P.T., 215Hsu, S.T., 205Kern, W., 170Klopfenstein, R.W.. 130Kozielec, F.. Jr., 205Kressel. H., 240Ladany, I., 240Lakhani, C., 180Liu, S.G., 220Lockwood, H.F., 205Matarese, R.J., 210Meray, L.H., 170Meyerhofer, D.. 205Moore. A.R., 220Moore, R.M. 205Mueller, C.W., 160Napoli. L.S., 220Narayan, S.Y., 210Nuese. C.J., 240Peskin, E.L., 125Phillips. W., 205Priestley, E.B., 125Risko, J.J.. 215Robinson, B., 180Rose, A., 210Scott, J.H., Jr., 205Sheng, P., 125

Sommers, H., Jr., 240Staebler, D.L., 205Sterzer, F.. 220Szostak, D.J., 205Takahashi, T., 205Upadhyayula, L.C., 210Williams, R.. 105Wojtowicz, P.J., 125Wronski, C.R., 210Wu, C.P., 130, 160Yuan. S., 210

RCA Limited

Clark, G.D., 180Conradi, J., 240Kudsia. C.M., 210Shkarofsky, I.P., 240Waksberg, A.. 340

NationalBroadcasting Company

Butler. R.J.. 345Mausier, R.. 345

RCA Service Company

Hoffmann-Heyden, A.E., 320Pepple. R.J., 320

Patents

Grantedto RCA Engineers

Consumer Electronics

System for dynamic and static muting -G.D.Pyles (CE,Indpls.) U.S. Pat. 3894201, July 8,1975

High voltage protection circuit - P.R. Ahrens(CE. Indpls.) U.S. Pat. 3894269, July 8, 1975.

Electric circuit for providing temperaturecompensated current - S.A. Steckler(CE,Som ) U S. Pat. 3895286, July 15, 1975.

Hysteresis voltage supply for deflection syn-chronizing waveform generator - S.A.STeckler (CE,Som I U.S Pat. 3898525,August 5, 1975.

Crystal -lock tuning system for tuning regular-ly and irregularly spaced channel frequencies- J. B. George (CE.Indpls.) U.S. Pat.3898567, August 5, 1975.

Dual mode deflection synchronizing system- S.A. Steckler, A.L. Limberg (CE,Som.) U.SPat. 3899635, August 12. 1975.

Commercial CommunicationsSystems Division

Spectrum differential noise squelch system -L.0 Schaeperkoetter (CCSD.Meadowlands)U.S. Pat. 3894285, July 8. 1975.

90

Notch rejection filter - W.J. Hannan, R.S.Singleton (CCSD,Palm Beach) U.S. Pat.

3898375, August 5, 1975.

Government Engineering

Control circuits - G.T. Burton (ATL,Cam.)U.S. Pat. 3895317, July 15, 1975.

Missile and Surface Radar Divi-sion

Radar system with a very short pulse length -N.I. Korman (MSRD,Mrstn.) U.S. Pat.

3895383, July 15, 1975.

Integral correlation and transverse equaliza-tion method and apparatus - T.V. Bolger(MSRD.Mrstn.) U.S. Pat. 3899666. August 12,1975

Insertion and differential phase -trim method- L.J. Lavendan, Jr. (MSRD.Mrstn.) U.S. Pat.3886501, May 27, 1975; Assigned to U.S.Government.

Average voltage detector for nonlinearwaveforms - R.B. Webb (MSRD,Mrstn.) U.S.Pat. 3898551, August 5, 1975.

RCA Limited

Circuit for supplying power to a load - U.M.Van Bogget (Ltd., Belgium) U.S Pat. 3898553.August 5. 1975.

Compact frequency reuse antenna - A R.Rabb (Ltd., Montreal) U.S Pat. 3898667.August 5, 1975.

Semiconductor radiation detector - J. Con-rad' (Ltd., Montreal) U.S. Pat. 3898686,August 5, 1975.

Coronet industriesAdjustable bed - R. Goodman (Cor.lnd.,Dalton,Ga.) U.S. Pat 3898702, August 12,1975.

Laboratories

Provision of reproducible thin layers of silicon

dioxide - A.M. Goodman, J.M. Breece(Labs..Pr.) U.S. Pat. 3892891. July 1, 1975.

Electron beam recording media - M. Kaplan.E.B. Davidson (Labs Pr) U S. Pat. 3893127,July 1, 1975.

Chemical vapor deposition of luminescentfilms - J.P. Dismukes, J.Kane (Labs..Pr)U.SPat 3894164, July 8. 1975.

Technique for fabricating high 0 MIMcapacitors - J. Mitchell, Jr., L.A. Carr, Jr.(Labs..Pr.) U.S. Pat. 3894872. July 15. 1975.

Method of selectively depositing glass onsemiconductor devices - R.B. Comizzoli(Labs..Pr.) U.S. Pat. 3895127. July 15, 1975.

Method of making a semiconductor device -H. Huang, A. San -Paolo (Labs..Pr.) U.S. Pat.3895429. July 22, 1975.

Fabrication of liquid crystal devices - L.A.Goodman. D.E. Carlson (Labs..Pr.) U.S. Pat.3896016, July 22. 1975.

Method for epitaxially growing a semicon-ductor material on a substrate from the liquidphase - S.L. Gilbert. M. Ettenberg (Labs., Pr)U.S. Pat. 3897281, July 29. 1975.

Method for manufacturing semiconductordevices -J.B. Klatskin (Labs.. Pr.) U.S Pat3897627. August 5, 1975.

Method of forming a thin piezoelectric bodymetallically bonded to a propagation mediumcrystal - J J. Hanak, D.M. Stevenson(Labs..Pr.) U.S. Pat. 3897628, August 5, 1975.

Method of making a transparency of a coloredimage in a magneto -electric printing system- E.G. Giaimo, Jr. (Labs., Pr.) U.S. Pat.3898082. August 5,1975.

Holographic recording media - R. J. Ryan.H.G. Scheible (Labs..Pr.) U.S. Pat. 3898358.August 5. 1975.

Method of forming pn junctions by liquidphase epitaxy - I Ladany, W.M. Cannuli

(Labs..Pr.) U.S. Pat. 3899371. August 12,1975.

Method of electrolets plating - N.Feidstein(Labs..Pr.) U.S. Pat. 3900599. August 19,1975.

Apparatus for scanning raised indicia - C.T.Wu (Labs..Pr.) U.S. Pat. 3900717, August 19,1975

Apparatus for scanning raised indicia - C.T.Wu (Labs..Pr.) U.S. Pat. 3900717. August 19.1975

Charge amplifier - P.K. Weimer (Labs, Pr.)U.S.Pat 3900743, August 19, 1975.

Picture Tube Division

Method of setting the 4.5 MHz trap In atelevision receiver -J.H.Owens (PTD.Dallas) U.S. Pat. 3892913, July 1, 1975.

Self -converging color television displaysystem -A.M. Morrell, W.H. Barkow, J. Gross(PTD,Lanc.) U.S. Pat. 3892996, July 1. 1975.

Cathode-ray tube having lithium silicate glarereducing coating with reduced light transmis-sion and method of fabrication - M. G.Brown, Jr. D.W. Bartch (PTD.Lanc.) J.S. Pat.3898509. August 5, 1975.

Automated Systems Division

Circuit for determining the time of transitionsin an alternating signal - G.B. Dodson, Ill(ASD.Burl.) U.S. Pat 3892949. July 1, 1975.

Average value crossover detector - G.B.Dodson. III (ASD.Burl.) U.S. Pat. 3892950,July 1, 1975.

Rapidly accessible optically stored informa-tion - B.R Clay. D A. Gore (ASD.Burl.) U.S.Pat. 3887276, June 3. 1975. Assigned to U.S.Government.

Valve position indicator - D.O. Blake(ASD.Burl.) U.S. Pat. 3896280, July 22. 1975;Assigned to U.S. Government.

Avionics Systems Division

Computer for threshold of TAU - J.E. Miller(ASD.Van Nuys) U.S. Pat. 3893112, July 1.1975.

Solid State Division

Multiple cell high -frequency power semicon-ductor device having bond wires of differinginductance from cell to cell - I. E Martin(SSD.Som.) U.S.Pat 3893159, July 1, 1975

Cathode-ray tube screening correction lenswith a non -solarizing material - R.J. D'Amato(SSD,Lanc ) U.S. Pat. 3893750. July 8. 1975.

Megasonic cleaning sysem - A. Mayer. S.Schwartzman (SSD,Som.) U.S. Pat. 3893869.July 8. 1975.

Proximity image tube with bellows focussingstructure -G.N. Butterwick (SSO,Lanc.) U.S.Pat. 3894258. July 8.1975.

Apparatus for non-destructively testing aforwardly biased transistor for secondbreadkown - R.B Jarl (SSD.Som.) U.S Pat2895297. July 15. 1975

Power transistor having good thermal fatiguecapabilities - J.E. Wright (SSD.Som.) U.SPat 3896486, July 22, 1975.

Solid State Division

Method of making a junction -isolatedsemiconductor Integrated circuit device -M.A Poiinsky (SSD.Som.) U.S. Pat. 3898107.August 5. 1975.

Government CommunicationsSystems DivisionTechnique forminimizing interference in

video recorder reproducer systems - F.L.Bechly. J.J. Younc (GCSD,Cam ) U.S Pat.3893168, July 1, 1975.

Radar recording and reproducing systems -P.F Muraco. J.S. Griffin (GCSD.Cam.) U.S.Pat 8993193, June 10. 1975; Assigned to U.S.Government.

DatesandDeadlines

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.

FEB. 18-2J, 1976 - IEEE International Solid -StateCircuits Conference (IEEE Solid -State Circuits Council,

IEEE Phila. Sec. & Univ. of Pa.) Sheraton Hotel,Philadelphia. PA. Deadline info: (abst & sum) 10/1/75 toD. Koehler. Bell Laboratories/Room 2A131. 60C Moun-tain Ave., Murray Hill. NJ 07974

APR 20-22. 1976 - Computer Software Engineering:Reliability. Management Design (IEEE, C&R. PINY)Deadline info: (abst) 12/1/75 to M. L. Shooman,Polytechnic Inst. of N.Y., 333 Jay St., Brooklyn, NY11201

JUNE 14-16. 1976 - International Microwave Sym-posium (MIT) Cherry Hill, NJ. Deadline Info: (abst)1/2/76 to Martin Caulton, RCA. Princeton. NJ 08540.

JUNE 21-24. 1976 - Information Theory Intl Sym-posium (IEEE, IT) Ronneby Brunn, Ronneby. Sweden.Deadline Info: (ms) 11/15/75 to R.W.Lucky, Bell Labs.,Room 1F-532, Holmdel, NJ 07733.

JULY 18-23. 1976 - Power Engineering Society SummerMeeting (IEEE. PE) Portland Hilton Hotel, Protland. OR.Deadline info: (papers) 2/1/76 W S Greer,Westinghouse Elec. Corp.. 1414 N E Grand Ave..Portland, OR 97212.

AUG. 1976 - Special issue of the IEEE Transactions onElectron Devices Deadline Info: (papers) 11/1/75 to Mr.

Richard A. Kokosa. General Electric Company. Elec-tronics Park. Box 41. Building 7, Syracuse, NY 13201and/or Dr. Daniel R. Muss, Westinghouse ResearchLabs., Building 801, Third Floor, 1310 Beulah Road,Pittsburgh, PA 15235.

AUG. 25-27. 1976 - Product Liability Prevention Con-ference (IEEE. R et al) Newark College of Engrg.,Newark. NJ. Deadline Info: (paper) 11/1/75 to JohnMihalasky. Newark College of Engrg., 323 High St.,Newark, NJ 07102.

AUG 30 - SEPT. 1, 1976 - Petroleum & Chemical Ind.Tech. Conference (IEEE. IA) Marriott Hotel, Phila., PA.Deadline info: (me) 3/1/76 to J A Stewart. FMC Corp..633 Third Ave.. New York. NY 10017.

SEPT. 19-23, 1976 - Jt. Power Generation Tech.Conference (IEEE, PE, ASME. ASCE) Buffalo HiltonHotel. Buffalo, New York, NY. Deadline info: (abet)1/29/76 to E.F. Chelotti, Gibbs & Hill. 393 Seventh Ave..New York, NY 10001.

SEPT. 20-24, 1976 - Intl Broadcasting Conference(EEA. IEE, IEEE UKRI Section) Grosvenor House. ParkLane, London. England. Deadline Info: (syn) 11/3/75 toIEE, Savoy Place. London WX 2R OBL England.

SEPT. 29 - OCT. 1. 1976 - Ultrasonics Symposium

91

(IEEE. SU) Annapolis Hilton Hotel. Annapolis, MD.Deadline info: (ms) 7/1/76 to P.H. Carr, AFCRL, L.G.Hanscom Field, Bedford. MA 01730.

OCT. 25-27. 1976 - Frontiers in Education (IEEE and theEd. Res. & Methods Div. of the ASEE. College of Engrg.Univ. of Ariz.) Ramada Inn, Tucson. Ariz. Deadline info:(syn) 1/15/76 (final drafts) 6/15/76 to Dr. E.R. Owen,Program Coordinator. FIE '76, Electrical and ElectronicEngineering, California Polytechnic State University,San Luis Obispo, CA 93401.

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.

NOV 3-5, 1975 - 8th Space Simulation Conference(IES-NASA AIAA/ASTM) Sheraton -Silver Spring MotorInn, Silver Spring. MD. Prog info: Institute ofEnvironmental Sciences. 940 Eut Northwest Highwa1Mt. Prospect. IL 60056

NOV. 19-20. 1975 - Computer Arithmetic (IEEE. C)Southern Methodist Univ., DaIls, TX. Prog Info: DaveMatula. Southern Methodist Univ., Dallas. TX.

NOV. 19-21, 1975 - Nuclear Science Symp. (IEEE. NPS)Sheraton Palace, San Francisco, CA Prog Info: D.A.Mack, Lawrence Berkeley Lab., Univ. of Calif . Berkeley.CA 94720.

NOV. 30 - DEC. 3. 1975 - Intl. Electron Devices Meeting(IEEE, ED) Washington. DC. Prog Info: C.N. Berglund,Bell -Northern Res. Ltd., POB 3511, Sta. C, Ottawa,Ontario. Can. K1 Y4H7.

NOV. 30 - DEC 5, 1975 L 96th Winter Annual Meeting(ASME) Hyatt Regency Houston 8 Sheraton HoustonHotels, Houston, TX. Prog Into: Paul Drummond. Direc-tor. Meetings and Conferences. ASME. 345 E. 47thStreet, New York, NY 10017

DEC. 1-3, 1975 - National Tele-Communications Conf.

(AES, COMM, GE, New Orleans Section) FairmontRoosevelt Hotel. New Orleans, LA. Prog Info: I.N. Howell,Jr.. S. Central Bell 1 elepnone Co.. POB 771,Birmingham, AL 35201.

DEC 1-4, 1975 - Signal Filtering (IEE. IEEE UKRISection. Inst. of Physics. IERE) IEE, London. EnglandProg info: IEE. Savoy Place, London, WC2R OBLEngland.

DEC 8-9, 1975 - Chicago Fall Conference on Con-sumer Electronics (CE. Chicago Section) O'Hare Inn.Rosemont, IL. Prog info: Bob Podowski. Zenith RadioCorp., 6101 Dickens Ave.. Chicago, IL 60639.

DEC. 9-11, 1975 - Electrical Safety in HazardousEnvironments )IEE, IEEE UKRI Section, Inst ofPetroleum, Inst. of Physics) London, England. Prog info:IEE, Savoy Place. London W.C. 2 R OBL England.

DEC. 9-12. 1975 - Magnetism and Magnetic MaterialsConference (IEEE, MAG. also AIP et al) BenjaminFranklin Hotel, Philadelphia, PA. Prog info: B. Stein,Univac Div.. Sperry Rand. POB 500, Blue Bell. PA 19422.

DEC 10-12. 1975 - Decision and Control - AdaptivePr (IEEE, CS) Hyatt Regency Houston. Houston,TX. Prog info: J. B. Pearson. Dept. of EE. Rice Univ..Houston, TX 77001.

JAN. 12-14. 1976 - Integrated Optics Topical Meeting(IEEE, 0 -EC. ACM) Salt Lake Hilton Hotel, Salt LakeCity. UT. Prog info: Optical Society of America, In-tegrated Optics Meeting, Suite 620. 2000 L St.. N.W.,Washington. DC 20036.

JAN. 19-21, 1976 - Computer Architecture (IEEE,C&ACM) Clearwater, FL Prog info: Oscar Garcia,College of Engrg., Univ. of South Florida, Tampa, FL33620.

JAN 20-22, 1976 - Reliability 8 Maintainability Sym-posium (IEEE, R, et al) MGM Grand Hotel. Las Vegas,NV. Prog info: H.L.Wuerffel, RCA Astro-Electronics.POB 800. MS 55, Princeton, NJ 08540

JAN. 25-30. 1976 - Power Engineering Society WinterMeeting (IEEE. PE) Statler Hilton Hotel. New York, NY.Prog Info: J.W.Bean. American Elec. Pwr. Svc.. 2Broadway, New York, NY 10004.

FEB. 4-5, 1976 - Modulator Sympolslum (IEEE. EDAGED) Statler Hilton Hotel, New York. NY. Prog Ingo:L.H. Klein, Palisades Inst for Res. Svcs., Inc. 201 VanckSt., New York, NY 10014.

FEB. 17-20, 1976 - Aerospace 8 Electronic SystemsWinter Convention (WINCON) (IEEE, AES) Los Angeles.CA. Prog info: Dick Harmon, ITT Cannon, 666 E. DyerRd.. Santa Ana. CA 92702.

FEB. 10-20, 1976 - Solid State Circuits Conference(IEEE, SSC Council, Phila Section, Univ. of Penna.)Marriott Hotel, Phila., PA. Prop info: IEEE, 345 East 47thStreet, New York, NY 10017

FEB. 24-26, 1976 - COMPCON SPRING (IEEE, C) JackTar Hotel, San Francisco. CA Prog info: Signey Fern-bach, Computer Dept.. L-61, Lawrence Livermore Lab.,POB 808, Livermore. CA 94550.

MAR. 8-10, 1976 - Industrial Electronics & ControlInstrumentation (I ECI) Sheraton Hotel, Phila.. PA. PropInfo: S.J. Vahaviolos, Engrg. Res. Ctr., Western Elec. Co.,POB 900. Princeton, NJ 08540.

MAR. 9-11, 1976 - Intl Zurich Seminar on DigitalCommunications (Switzerland Section. ASSP, COMM,C. CAS) Fed. Inst of Tech.. Zurich. Switzerland. Prog.info: Albert Kundig. PTT Res. Lab., TZV 907, CH -3000.Bern 29, Switzerland.

MAR. 10-12, 1976 - Region V Control of Power SystemsConference (IEEE Region V. Oklahoma City Sec.) Holi-day Inn. N.W. Oklahoma City. OK. Prog info: M.E.Council. Sch. of EE. Univ. of Oklahoma. 202 W. Boyd.Normal, OK 73069.

MAR 30-31, APR 1. 1976 - The Personal Com-munications Two -Way Radio Show (EIA) Las VegasHilton. Las Vegas NV. Prog Info: Mr. Robert Black. TheShow Company International. 1605 Cahuenga Blvd.. LosAngeles, CA 90028.

APR. 1976 - Carnahan Conference on Electronic CrimeCountermeasures (IEEE. AES) Univ. of Kentucky. Proginfo: Office of Continuing Education Engrg.. Rm. 779.Anderson Hall. Univ. of Ky.. Lexington, KY 40506.

APR 5-7.1976 - SOUTHEASTCON (IEEE Region 3, S.C.Affiliation of Sections) Clemson Univ., Clemion, SC.Prog info: J.T. Long. EE Dept., Clemson Univ.. Clemson.SC 29631.

Engineering

New RCA managementorganization

A new RCA management organizationwas announced by Chairman Robert W.Sarnoff. At its apex is the newly createdOffice of the Chairman. This consists ofMr. Sarnoff, who is Chief ExecutiveOfficer; Anthony L. Conrad, President andChief Operating Officer; and thepresidents of three major businessgroups. They are:

RCA Electronics: This consists of theConsumer Electronics Division, the SolidState Division, the Picture Tube Division,the Distributor and Special Products Divi-sion, Government and Commercial

News and Highlights

Systems, RCA Service Company and the"SelectaVision" VideoDisc Project. EdgarH. Griffiths becomes President, RCA Elec-tronics. He continues as an Executive VicePresident of the Corporation.

RCA Communications: This consists ofRCA Global Communications, Inc., Ran-dom House, Inc., and the RCA RecordsDivision. Howard R. Hawkins becomesPresident, RCA Communications. He con-tinues as an Executive Vice President ofthe Corporation.

RCA Diversified Businesses: This consists

of Banquet Foods, Coronet Industries,Cushman & Wakefield, Inc., The HertzCorporation and the Oriel Foods Group.Mr. Griffiths becomes Acting President,RCA Diversified Businesses, pending theappointment of a successor. He hasheaded this group of subsidiaries since1971.

The National Broadcasting Company con-tinues to report directly to Mr. Sarnoff.Corporate staff functions that previouslyreported either to Mr. Sarnoff or Mr.Conrad will now report to the Office of theChairman.

92

Mr. Sarnoff said the new, broadenedmanagement structure was in response tochanges that have occurred in RCA overthe past decade.

"During that period our sales have nearlydoubled and our assets have nearlytripled," he said. "The combination of ouracquisitions and divestitures has changedRCA from a company almost exclusivelyinvolved in electronics to a diversified,world-wide enterprise in which the newlyacquired businesses have played a vitaland growing role."

Mr. Sarnoff said the three major businessgroups, as well as NBC, will function withmaximum operational autonomy. TheOffice of the Chairman will concern itselfwith broad policy guidance, the establish-ment of corporate objectives, the alloca-tion of corporate resources and thecoordination of activities among thebusiness groups. The group Presidents,by virtue of their membership in the Officeof the Chairman, will participate in top-level decision making for the entire cor-poration.

Mr. Sarnoff said the primary structuralchange in the business groups involvesthe consolidation of all the company'straditional electronic product and serviceoperations into RCA Electronics.

William C. Hittinger, RCA Executive VicePresident, continues to be responsible forConsumer Electronics, Solid State, Pic-ture Tubes, Distributor and SpecialProducts, and the "SelectaVision"VideoDisc project. He now reports to Mr.Griffiths.

Irving K. Kessler, RCA Executive VicePresident, Government and CommercialSystems, and Julius Koppelman,President, RCA Service Company, con-tinue their present responsibilities, alsoreporting to Mr. Griffiths.

The organization of RCA Com-munications under Hawkins is: Robert L.Bernstein, Chairman of the Board, Presi-dent and Chief Executive Officer, RandomHouse, Inc.; Kenneth D. Glancy,President, RCA Records Division; andHoward R. Hawkins, Chairman of theBoard and Chief Executive Officer, RCAGlobal Communications, Inc.

The organization of RCA DiversifiedBusinesses under Griffith is: Kenneth E.Guebert, President and Chief ExecutiveOfficer, Banquet Foods Corporation;James G. Gulliver, Chairman of the Boardand Chief Executive Officer, Oriel FoodsGroup; Martin B. Seretean, Chairman ofthe Board and Chief Executive Officer,Coronet Industries, Inc.; Leone J. Peters,Chairman of the Board and Chief Ex-ecutive Officer, Cushman & Wakefield,Inc.; and Robert L. Stone, Chairman of theBoard and President, The Hertz Corpora-tion.

"In the years ahead," Mr. Sarnoff said,

Neumann named Chief Engineer Erickson named Manager,for Mobile Communications Manufacturing Engineering

Karl L. Neumann has been appointedChief Engineer for Mobile Com-munications Systems, Meadow Lands, Pa.In the newly created position, Mr.Neumann is responsible for all productand development engineering activitiesfor RCA's line of two-way radio com-munications systems.

"RCA, like other companies, will operatein an environment of increasing technicalcomplexity, intensified competition fordomestic and international markets, andmounting governmental regulatory re-quirements. I believe the neworganizational structure responds mosteffectively to the management needs of alarge and diversified international com-pany. It will give us the depth and flexibilityto react swiftly to new business oppor-tunities. It is a logical and evolutionaryresponse for the management of a chang-ing RCA that faces the future with hopeand confidence."

Mr. Griffiths was elected an Executive VicePresident of the Corporation in June,1972, at which time he also was elected tothe RCA Board of Directors.

Prior to becoming RCA Executive VicePresident, Services, in 1971, Mr. Griffithswas President of the RCA Service Com-pany for three years. He joined RCA at itsCamden, N.J., facility in 1948, and a yearlater he was assigned to the RCA ServiceCompany where he held varied financialpositions. In 1963, he was appointed Divi-sion Vice President, International Finance,RCA International Division, and threeyears later was named Division Vice Presi-dent, Commercial Services, RCA ServiceCompany.

Mr. Griffiths received the BS in BusinessAdministration from St. Joseph's College.

Mr. Hawkins also was elected an ExecutiveVice President and Director of RCA inJune, 1972. Prior to that he served asPresident of RCA Global Communicationssince 1966. While an RCA Executive VicePresident he continued his leadership rolein international communications by serv-ing also as RCA Globcom's Chairman andChief Executive Officer.

Mr. Hawkins joined RCA Global Com-munications in 1946 as Assistant GeneralAttorney. He became General Attorney in1949 and Vice President in 1951. He waselected Executive Vice President and aDirector in 1964 and President on June 3,1966.

A graduate of Indiana University with theBS in Public Business Administration, hereceived the degree of Doctor ofJurisprudehce with Distinction in 1941. Heserved with the Federal Bureau of In-vestigation from 1941 to 1946.

Samuel P. Yrankovich, Plant Manager,Meadow Lands plant, announced the ap-pointment of Dale 0. Erickson asManager, Manufacturing Engineering, forRCA's Meadow Lands facility. Mr.Erickson is responsible for the planning,operation, and supervision of theManufacturing Methods Engineering,Test Methods Engineering, and TestMaintenance Activities.

Donahue heads Picture TubeEngineering

Dr. D. Joseph Donahue has beer ap-pointed Division Vice President, Engineer-ing, Picture Tube Divison. Dr. Donahue isresponsible for all engineering activities inthe Picture Tube Divison, includingresearch and development. He is locatedin the Lancaster, Pa., headquarters of thePicture Tube Division.

E.O. Johnson namedDirector of Japanese Laboratories

Appointment of Edward 0. Johnson asDirector of RCA Research Laboratories,Inc. (Tokyo), has been announced by Dr.Jan A. Rajchman, Vice President of theRCA subsidiary. Located on the outskirtsof Tokyo, RCA Research Laboratories wasestablished in 1961 to foster closerrelations oetween the Japanese andAmerican scientific communities. (SeeRCA Engineer, Vol. 20, No. 3, Oct -Nov.1974 which contained a profile of theTokyo Laboratories.) Mr. Johnson isreplacing Dr. Bernard Hershenov,Director cf the Japanese Laboratoriessince 1972. Dr. Hershenov is returning toRCA in the United States.

Laschever appointed Manager, Elec-tronic Warfare and Radar Programs

Dr. Harry J. Woll, Division Vice Presidentand General Manager, AutomatedSystems Division, has appointed NormanL. Laschever as Manager, Electronic War -face and Radar Programs. In his newcapacity, Mr. Laschever will be responsi-ble for all electronic warfare and radarproducts developed and produced by theRCA facility.

Bachynski elected V.P.

G. Denton Clark, President, RCA _imited(Canada). has announced that Dr. MorrelP. Bachynski was elected Vice President,Research and Engineering. He has beenDirector of Research since 1965.

(See RCA Engineer, Vol. 19, No. 6. April -May 1974 which contains a paper by Dr.Bachynski on "Research and Develop-ment in Canada").

Neumann

Donahue

Johnson

Bachynskl

93

Organizational Changes at G&CS

Government and Commercial Systemshas made an organizational realignmentthat includes the re-establishment of itsBurlington, Mass., facility as a Divisionand the consolidation of certain com-mercial activities.

Irving K. Kessler, Executive Vice Presi-dent, G&CS, said the Automated SystemsDivision in Burlington will be headed byDr. Harry J. Woll as Division Vice Presi-dent and General Manager. Dr. Woll,formerly Division Vice President, Govern-ment Engineering, has had actingresponsibility for the Burlington operationfor the past several months. TheBurlington facility was known as theAerospace Systems Division until March1974 when it became part of the Govern-ment Communications and AutomatedSystems Division in Camden, N.J.

RCA government activities in Camdenhave been reconstituted as the Govern-ment Communications Systems Divisionand Dr. James Vollmer has been namedDivision Vice President and GeneralManager. He succeeds James M. Osbornewho has resigned.

Dr. Vollmer was formerly the Vice Presi-dent and General Manager of the PalmBeach Division, Palm Beach Gardens,

organizational responsibilityhas been transferred to Commercial Com-

munications Systems Division andrenamed Palm Beach Operations.

Commercial Communications SystemsDivision headed by Andrew F. Inglis,Division Vice President and GeneralManager, now encompasses RCAbusiness in Avionics Systems (Van Nuys,California), Broadcast Systems(Camden), Mobile CommunicationsSystems (Meadow Lands, Pa.), and DataManagement Systems (Palm Beach).

Other organizational changes in Com-mercial Communications Systems Divi-sion include the transfer of the BroadcastSystems international business to NeilVander Dussen, Division Vice President,Broadcast Systems. Mr. Vander Dussenalso has acquired responsibility for theRCA Film Recording Systems activity inBurbank, California. The internationalbusiness of Mobile Communications istransferred to Jack E. Underwood,Division Vice President, Mobile Com-munications.

The two international activities formerlywere the responsiblity of Joseph P.Ulasewicz, a Division Vice President, whohas been appointed to a newly createdposition of Division Vice President,Product Operations. He will be responsi-ble for product management and productengineering activities in BroadcastSystems.

Awards

Government Communications SystemsDivision

The Technical Excellence Committee ofGovernment Communications SystemsDivision in Camden, N.J., has made a teamaward to Jim Devlin, Al Foster, JayHoover, Dick Orr, and Dave Sapp ofRecording Systems for their outstandingwork in the development of the RCA HighDensity Multi -Track Recorder (HDMR).Their efforts have resulted in a recent solesource contract from NASA for anengineering model of a 240 -megabit persecond spacecraft recorder for EOSProgram.

RCA Laboratories

Robert A. Gange and Carl C. Steinmetz,Materials Research Laboratory, andEugene M. Nagle, Systems ResearchLaboratory, have received NASA NewTechnology commendation certificatesfor their work on a laser actuatedholographic storage device.

Law and Schade receive SID awards

Dr. Harold B. Law Director, Materials andDisplay Devices Laboratory, Picture TubeDivision, Princeton, N.J., received theFrancis Rice Darne Memorial Award of theSociety for Information Display "for out-standing contributions to color picturetube development resulting in practicalcolor television."

New equipment at Lancaster

During the plant shutdown at Lancaster, anew Romicon Ultra Filtration water systemwas installed. This will solve a long-standing problem of providing quantitiesof high quality water for the CCD opera-tion. The unit supplies 60 gal/minute. It itanticipated that long lifetime operation

will be obtained using 0.2 -micrometerpoint -of -use filters.

A Macrodata MD154 large scale IC testerhas been installed in the CCD productionarea in Lancaster. This equipment is oneof the most sophisticated pieces of test

Tom Edwards (left) and Bob Blazek inspecting the new Ultra Filtration water system atLancaster

equipment in the Corporation. It iscapable of very wide adaptability in testingall manner of devices incorporatingmemory, shift registers, and logic circuits.The Macrodata equipment has now beenprogrammed for complete testing ofCCD's.

John Schollenberger using the new Macrodata IC tester in the CCD area at Lancaster.

94

Dr. Harold Law (right) accepts SID award from RobertKlien. President of SID.

Otto H. Schade, Sr. (retired) of RCAElectronic Components, Harrison, N.J.,received a special recognition award ofthe Society for Information Display "forthe pioneering applications of frequencyresponse concepts to the analysis andoptimization of electrooptic systems."

Hanak receives best paper award

Dr. Joseph J. Hanak, Fellow of theTechnical Staff, Materials ResearchLaboratory, RCA Laboratories, Princeton,N.J. recently received a special Certificateof Recognition from the Mid -AtlanticChapter of the Society for InformationDisplays. Dr. Hanak was cited for hispresentation, "Electroluminescent ThinFilm Displays," as the best paper of the1974-75 seminar program.

Goodwin, Petrillo win highest Navyaward

The Navy's Distinguished Public ServiceAward, the highest recognition that can beextended to an individual not employed bythe Navy, was presented by Rear AdmiralWayne E. Meyer to William V. Goodwinand Edward W. Petrillo "foroutstanding...contributions" to thedevelopment of AEGIS, the Navy's newestand most advanced fleet air defensesystem.

The ceremony was held at the company'sMoorestown Plant. Also attending wereAnthony L. Conrad, President and ChiefOperating Officer, RCA, and I. K. Kessler,Executive Vice President, RCA Govern-ment and Commercial Systems.

Mr. Goodwin was cited for his"...dedicated professional participationand innovative engineering leadership" indirecting AEGIS "toward its role as theSurface Ship Weapon System of thefuture."

Mr. Petrillo was commended for"professional and innovative engineering

Otto Schade. Sr., (right) accepts special SID award fromPhilip Damon, SID awards/honors chairman.

participation" that substantially assistedin the development of AEGIS. As AEGISProgram Manager, he was responsible fordelivering a fully integrated AEGISEngineering Development Model fortesting.

Professional activities

Commercial Communications SystemsDivision

A.C. Luther, Chief Engineer of BroadcastSystems, has been named a Member of theBoard of Managers of the newly formedPhiladelphia Section of the Society ofMotion Picture and Television Engineers(SMPTE).

J.R. West, Engineering Leader in Broad-cast Systems has been named as the TVprogram co-chairman of the PhiladelphiaSection of SMPTE.

Missile and Surface Radar Division

Dr. J.C. Williams was re -appointed to theScientific Advisory Committee of theDefense Intelligence Agency for theperiod July 1975 - July 1976 (see Julie -July 1974 issue which announced theinitial appointment).

D.L. Pruitt has been selected to serve onthe committee for the Twelfth ModulatorSymposium.

Degrees

Government Communications andAutomated Systems Division-Burlington

Doug Gore of Electro-Optical SystemsEngineering received the MSEE with amajor in Electro-Optics from NortheasternUniversity in June.

Errata

To: Editor

Subject: "MOS Array design: universal,APAR, or custom" by R. Bergman, M.Aguilera, G. Skorup

I recently co-authored the MOS arraydesign article which appeared in the June -July issue of the RCA Engineer. Due to anoversight on our parts, no acknowledg-ment appeared in this article.

I strongly feel that professional con-tributions should be recognized and amrequesting that you include the attachedin a future issue along with our personalregrets for the oversight.

M. Aguilera

Acknowledgment: Development of theprograms referenced in this paper resultsfrom the efforts of many RCA personnel.The Advanced Technology Laboratoriesteam under the leadership of A. Felleroriginally wrote and evaluated the PR2DAand PR2DAE programs and evaluated thesupporting standard cell family. The SolidState Technology Center Design Automa-tion group under the direction of Dr. L.French developed the ART, MAP CRITICand other minor supporting programsnecessary for processing custom, APARand Universal Arrays.

The authors gratefully acknowledge theefforts both of these teams put forth tosupply the basic Computer Aided Designprograms. The authors especially wish torecognize the efforts of Mr. R. Noto for hiswriting of PR2DA and PR2DAE thebackbone of our present Automatic Place-ment and Routing system.

New Receiving Tube Manual

A new edition of the widely used RCAReceiving Tube Manual is now available.The manual - RC -30 - like itspredecessors, has been written forengineers, service technicians, educators,experimenters, and others interested inelectron tubes and their applications.

The RG-30 can be purchased from localRCA Family Stores at a special employeeprice of $2.25 per copy, plus applicablestate and local taxes. Orders from the fieldshould be addressed to the attention ofFamily Stores, RCA Distributor andSpecial Products Division, Deptford, N.J.08096.

95

Oliver is new Ed Rep forIndianapolis plant

C. Wayne Hamilton, Plant Manager, In-dianapolis Component Plant, Manufac-turing Operations. Consumer Electronicshas appointed James S. Oliver RCAEngineer Editorial Representative forManufacturing Operations at In-dianapolis. Editorial representatives areresponsible for planning and processingarticles for the RCA Engineer and forworking with the local TechnicalPublications Administrator (TPA) to sup-port the corporate wide technical papersand reports program. Clyde Hoyt is theTPA for Consumer Electronics.

James S. Oliver received the BSIE fromGeorgia Institute of Technology andjoined RCA at the former Memphis,Tennessee Facility in 1966. Initially hiredas a Timestudy Engineer, he later

Obituaries

Ray Guy

Raymond F. Guy, retired NBC engineeringexecutive and radio-tv pioneer, died onJuly 12. He was 76. Mr. Guy, whoseassociation with radio dates back to the"spark gap" transmitter and "QuakerOats" receiver, retired from NBC in 1961and became a consulting engineer. Hestarted with the Marconi WirelessTelegraph Company in 1916; served in theU.S. Army in WWI; and received the BSEEfrom Pratt Institute in 1921. That year, hebecame Engineer, Announcer, et. al. forWJZ - the second radio broadcastingstation in the country. In 1924, hetransferred to RCA researcn laboratoriesat Van Courtland Park, N.Y. He joinedNBC in 1929 as Staff Radio Engineer andthen became Senior Executive Engineerresponsible for all NBC networks,transmitters, and frequency allocations.

Based on his sixty-three years of opera-tion, Ray Guy was one of the oldest activeHAMS. He began in 1912 (station 2ANC).His first calls were W20, W2AK, and, onretiring to Florida, the well-known W4AZ.His professional affiliations included

Raymond F. Guy

96

accepted assignment as a ManufacturingMethods Engineer responsible for thegeneration and maintenance of assemblyprocesses. In August, 1968, he becameManager, Chassis Assembly Process. Thisposition he held until the closing of theMemphis Facility in January, 1971.Transferring to the Bloomington, IndianaFacility, as a Manufacturing MethodsEngineer, he assumed processing res-ponsibilities until assigned to the ForeignOperations Support Group in January,1974. In March, 1974, he was transferred toRCA Electronica, Rio de Janeiro, Brazil.Following the October, 1974, decision todiscontinue efforts to develop the BrazilFacility, he was transferred to the In-dianapolis, Indiana Facility. He is nowengaged in make -versus -buy activities asa Manufacturing Methods Engineer.

Fellow and Life Member of IRE (President,1950); Member of FCC ConsultingEngineers Assn.; Radio and TV ExecutiveSociety; Radio Pioneers; IEEE (Fellow andLife Member); TV Broadcasting Assn.;Radio Club of America (Fellow); VeteranWireless Operations Assn. (Life Member,President 1939); Radio Pioneers (VicePresident 1951-53); NSPE. He was aProfessional Engineer in New Jersey andNew York; co-author of two books; andauthor of 150 papers.

Gil Lang

Gilman A. Lang, Manager of PowerEngineering at Solid State TechnologyCenter in Somerville, N.J. died on July 20.He was 46. A 1974 recipient of the DavidSarnoff Outstanding Achievement Award,Mr. Lang received the BS in Chemistryfrom the University of New Hampshire in1951, and served the next four years withthe U.S. Air Force. From 1955 to 1958 hewas employed as an analytical chemist atthe Electronic Tube Division ofWestinghouse Electric Corporation inElmira, New York. He joined the Semicon-ductor and Materials Division of RCA atSomerville, N.J. in 1958 as an analyticalchemist. In 1963 he joined the Industrial

Gilman A. Lang

Aerospace Projects Department as a

process engineer. He transferred to theIndustrial Power Transistor DesignDepartment in 1965 as a transistor designengineer with responsibility for thedevelopment of high power silicon"overlay" transistors. He was appointedManager of Power Engineering on Oc-tober 1, 1974.

Hal Schwartzberg

Harold L. Schwartzberg, Staff SystemScientist and a key member of the Astro-.Electronics Division organization for 13years, died on August 23. He was 52. Mr.Schwartzberg received the BSEE fromLehigh Univeristy in 1950 and worked formore than 21 years in engineering andengineering management of spacecraftand communication systems. He was pro-ject manager for several satellite programsat AED-most recently the TIROS/ITOS/NOAA Weather Satellite. Prior toRCA, Mr. Schwartzberg was associatedwith Philco and General Dynamics. Hewas a Professional Engineer in Penn-sylvania; senior member of IEEE; theAssociation of Old Crows; Space andRange Pioneers; and a member of theAmerican Meteorological Society.

Harold L. Schwartzberg

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 SystemsAstro-Electronics Division I. M. SEIDEMAN* Engineering, Princeton. N.J.

Automated Systems Division K.E. PALM Engineering, Burlington. Mass.A.J. SKAVICUS Engineering , Burlington. Mass.L.B. SMITF, Engineering, Burlington, Mass.

Commercial CommunicationsSystems DivisionBroadcast Systems

Mobile Communications Systems

Avionics Systems

Palm Beach Operations

Government CommunicationsSystems Division

Government Engineering

Missile and Surface Radar Division

Research and EngineeringLaboratories

Solid State Division

Consumer Electronics

RCA Service Company

Distributor andSpecial Products Division

Picture Tube Division

RCA Global Communications. Inc.

NBC, Inc.

RCA Records

RCA Ltd

Patent Operations

Electronic Industrial Engineering

W.S. SEPICH Broadcast Systems Engineering Camden. N.JR. E. WIND Broadcast Systems Antenna Equip. Eng., Gibbsboro. N.JA. C. BILLIE Broadcast Engineering, Meadow Lands. Pa.

F. A. BARTON Advanced Development. Meadow Lands. Pa.

C. S. METCHETTE Engineering, Van Nuys. Calif.J. McDONOUGH Equipment Engineering, Van Nuys. Calif.

B.B. BALLARDPalm Beach Gardens, Fla.

A. LIGUORI* Engineering, Camden, N.J.H.R. KETCHAM Engineering, Camden, N.J.

M. G. PIETZ Advanced Technology Laboratories, Camden. N J

D. R. HIGGS Engineering, Moorestown, N.J.

C. W. SALL* Research, Princeton. N.J.

E.M. McE'_WEE Engineering Publications. Somerville, N.JJ.E. SCHOEN Engineering Publications, Somerville. N.J.H.R. RONAN Power Devices, Mountaintop, Pa.S.SILVERSTEIN Power Transistors, Somerville. N.JA.J. BIANCULLI Integrated Circuits and Special Devices. Somerville. N.JJ.D. YOUNG IC Manufacturing, Findlay, OhioR.W. ENGSTROM Electro-Optics and Devices, Lancaster. Pa.

C.W. HOYT Engineering, Indianapolis, Ind.R. J. BUTH Engineering, Indianapolis. Ind.P. E. CROOKSHANKS Television Engineering, Indianapolis. IndF. R. HOLT Advanced Products, Indianapolis, Ind.J. S. OLIVER Consumer Electronics, Indianapolis, Ind.

J.E. STEOGER Consumer Services Engineering, Cherry Hill. N.J.R. MacWILLIAMS Marketing Services, Government Services Division, Cherry Hill. N.JR.M. DOMBROSKY Technical Support, Cherry Hi". N.J.

C. C. REARICK Product Development Engineering, Deptford. N.J.J. N. KOFF Receiving Tube Operations, Harrison. N.J.

C.A. MEYER Commercial Engineering, Harrison. N.J.J.H. LIPSCOMBE Television Picture Tube Operations. Marion. Ind.E.K. MADENFORD Engineering, Lancaster, Pa.

W. S. LEIS RCA Global Communications. Inc.. New York, N.Y.P. WEST RCA Alaska Communications, Inc., Anchorage, Alaska

W. A. HOWARD Staff Eng., Technical Development, New York, N.Y

J.F. WELLS Record Eng., Indianapolis, Ind.

W. A. CHISHOLM Research & Eng. Montreal, Canada

J.S. TRIPOLI Patent Plans and Services, Princeton, N.J.

J. OVNICK Engineering, N. Hollywood, Calif.

Technical Publications Administrators (asterisked above) areresponsible for review and approval of papers and presentations

alma EngineerA TECHNICAL JOURNAL PUBLISHED BY CORPORATE TECHNICAL COMMUNICATIONS"BY AND FOR THE RCA ENGINEER"

Printed in U S A Form No. RE -21-3