americanradiohistory.comamericanradiohistory.com/ARCHIVE-RCA/RCA-Engineer/RCA-Enginee… · mcoaEngineer A technical journal published by the RCA Technical Excellence Center 0 13

RCA EngineerVol.No .2 Mar. /Apr. 986

N

ellftw,

[co Solt Zat.r book)oarspbxs F -,4.r irdox.

1. lotOribil-compu4o:. Orate/00oLies 0..16e. 19111111kileGIAL.

2. MAcrocompu bee. b4a booblbo

oypt000. I. Ehaw.0.

11. Title

JUL R5 1u4.33391CM'-*

mcoa EngineerA technical journal published by theRCA Technical Excellence Center 0 13 Roszel Road 0 P. 0. Box 432 0 Princeton, NJ 08543-0432 0 Tacnet: 226-3090

RCA Engineer Staff

Tom King Editor

Mike Lynch Associate Editor

Louise Carr Art Editor

Rosemarie Cruz Editorial Assistant/Composition Specialist

Maria Pragliola Secretary

Editorial Advisory Board

Tony Bianculli Manager, Engineering Information, TechnicalExcellence Center Jay Brandinger Staff Vice -President,Engineering, Electronic Products and Technology 0 FrankBurris Director, Technical Excellence Center 0 JimCarnes Division Vice -President, Engineering, ConsumerElectronics 0 John Christopher Vice -President, Technical

Operations, RCA Americom 0 Jim Feller Division Vice -President, Technology, Aerospace and Defense 0 MahlonFisher Division Vice -President, Engineering, Video Component &Display Division 0 Larry Gallace Division Vice -President,Product Assurance, MIS and Materials, Solid StateDivision 0 Arch Luther Senior Staff Scientist, RCALaboratories 0 Bill Webster Vice -President and SeniorTechnical Advisor, Electronic Products and Technology

Consulting Editors

Ed Burke Administrator, Marketing Information and Commun-ications, Aerospace and Defense 0 Walt Dennen Manager,Naval Systems Department Communications and Information,Missile and Surface Radar 0 John Phillips Director, BusinessDevelopment and Services, RCA Service Company

Cover design by the RCA Engineer staff.

Cover photo by James Pryor, RCA Solid State Division

About our cover the pervasive PC has become very much apart of the engineering workplace. With the proper software,PCs have successfully performed the functions of manytraditional engineering tools.

In this issue we see some of the possible uses inengineering for the PC and its more sophisticated cousin, theengineering workstation. Of course, with each new applicationcomes a new set of problems to solve, and this issuediscusses a number of those, too.

0 To serve as a medium of interchange of technicalinformation among various groups at RCA 0 To create acommunity of engineering interest within the company bystressing the interrelated nature of all contributions 0 Todisseminate to RCA engineers technical information ofprofessional value 0 To publish in an appropriate mannerimportant technical developments at RCA, and the role of theengineer 0 To help publicize engineering achievements in amanner that will promote the interests and reputation of RCAin the engineering field 0 To provide a convenient means bywhich the RCA engineer may review professional work beforeassociates and engineering management 0 To announceoutstanding and unusual achievements of RCA engineers ina manner most likely to enhance their prestige andprofessional status.

computer model includes the effects of dis-tributed parasitics. The model enabled Mr.Wignot to predict a parasitic oscillationmode and have a fix ready when the prob-lem showed up during the factory pilot run.It also was used to solve a radiation prob-lem. The work done is usable on all MTTunits, and will be useful to future rf designersin reducing factory problems with tuners.

Astro Employee of the QuarterAwardsFive engineers at RCA Astro-ElectronicsDivision have been named Employee of the

Quarter for their exceptionally high level ofpersonal commitment, sustained perfor-mances, and leadership.

Mesut Ciceker for providing the guidance,encouragement and initiative for his teamduring the integration and test of AmericanSatellite Company's communications satel-lite, ASC-1.

Michael Mihavetz and a team from Astro'sManagement Information Systems (MIS), formaking it possible for RCA to meet thecontractual requirements of the Air Forceon the DMSP 11-14 series of meteorologicalsatellites.

Roger B. Hogan, Michael Cummings, andKenneth Ward were selected by the Amer-ican Ihstitute of Aeronautics and Astro-nautic:3 (AIAA) to receive the Haley SpaceFlight Award for their outstanding contribu-tions to the advancement of astronautictechnology. All three Astro employees werea part of the NOAA/NASA/RCA RecoveryTeam ()f the NOAA-8 satellite.

Professional activities

Hittinger receives IEEE award

William C. Hittinger, Executive Vice Presi-dent (retired), has been named by the IEEEBoard of Directors as the recipient of the1986 Frederik Philips Award "for leadershipin electronics research and development."Mr. Hittinger received the award at the Inter-national Solid State Circuits Conference inFebruary. The Philips Award recognizesoutstanding accomplishments in the man-agement of research and development. Dr.William M. Webster received the sameaward in 1980.

Nelson is ACS Chairman

J. Roy Nelson, RCA Laboratories, has beenelected 1986 Chairman of the Central JerseySection of the American Chemical Society.

Bauer and Perlman electedIEEE Fellows

The Board of Directors of the Institute ofElectrical and Electronic Engineers (IEEE)recently announced the election of John A.Bauer, Missile and Surface Radar Division,and Dr. Barry S. Perlman, RCA Laboratories,to the membership grade of Fellow. Thegrade of Fellow recognizes unusual distinc-tion in the engineering profession and isconferred only by invitation of the Board of

Directors upon persons of outstanding andextraordinary qualifications and experiencein IEEE -designated fields who have madeimportant individual contributions to one ormore of these fields.

John Bauer's citation is "for outstandingtechnical leadership in establishing leadlesschip carrier technology and surface mount-ing packaging techniques as a viable ap-proach to high -density, low-cost militaryand commercial packaging."

Dr. Perlman was honored "for contribu-tions to microwave solid-state device andcircuit design, and leadership in computer -aided methods for microwave engineering."

Brown receives NAB award

Dr. George H. Brown, retired RCA engineer-ing executive, has received the NationalAssociation of Broadcaster's 1986 Engineer-ing Achievement Award. It was presentedat the 40th annual Broadcast EngineeringConference luncheon on Tuesday, April 15during NAB's annual convention at theDallas Convention Center.

Communication andInformationSystems Division

Donald Bussard of RF and MicrowaveEngineering was chosen the Engineer of

the Mohth for the month of November. Thisaward is recognition of Don's outstandingcontribuition on the GWEN and AF-FSPprograms.

On the AF-FSP, Don was responsible forthe successful design and development ofa state--of-the-art solid-state power con-verter unit (PCU). He was instrumental insolving two major problems on the GWENprogram. On the interface of the LFPA to theATU, Dc4n quickly diagnosed a problem andimPlemnted a retrofitable fix to protectthe PA and was able to diagnose reliabilityproblerris on the UPS manufactured byNOVA as a faulty control circuit design. Donredesigned the control circuit using the samecomponents and furnished the design tothe vendor for implementation. His perfor-mance on these two programs had a tremen-dous imbact on cost and schedule.

Marianne Nolan was chosen Engineer ofthe Month for the month of December forher outstanding contribution on the GWENprogram and the successful acceptancetesting of the TRW HMI Computer program.

Her at)ility to analyze and solve problemswas quite apparent in her efforts toward thefinalizati7n of the acceptance test and pre-liminary qualification test procedures. Thisdedicaticn led to the successful acceptancetest and POT of HMI.

84 RCA Engineer 31-2 Mar. / Apr. 1986

Engineering Excellence Award for Ills out-standing performance on the STC-%/DBSAttitude Control Subsystem test program.Without the benefit of experienced personneland sufficient staffing, this very complexattitude control test was successfullY com-pleted as a result of his dedication, broadinherent capabilities, and long hours Of work.Mr. Tracy performed as director and coordin-ator as well as customer liaison.

The Engineering Recognition Award hasbeen presented to Patrick Burkle for hisefforts in completing critical acceleratedthermal balance testing, analysis, and designmodifications for the thermal control systemdesign for the ASCI spacecraft.

Circleville presents first TECAward

Thomas H. Kirchgessner of the VCDD Circle-ville plant has been awarded a TechnicalExcellence Award for the design and instal-lation of a hot funnel glass code date proc-ess that has proved to be a virtually main-tenance -free system since its installation.

Fourth Quarter, 1985 MSRDTEC Awards

The following are winners of the Missileand Surface Radar Division's fourth-quarterTechnical Excellence Awards:

Geoffrey S. Edelson, for major contributionsto RCA's growing capabilities in advancedairborne lightweight surveillance radar tech-nology. Mr. Edelson's contributions to theground -based test bed proposal effort in-cluded a set of unique processirtg algo-rithms required for operation of trte long -dwell radar system. His technical iirectionof the test bed effort resulted in an advancedprocessor that meets or exceed:4 perfor-mance requirements, delivered on time andunder budget.

Miles C. Johnson, for outstanding analyticalskills and technical leadership, demOnstratedin the design implementation of the AEGISantenna power supply. In additicn to hisdirect responsibility for the input power sec-tion design, Mr. Johnson applied his circuitand mathematical skills to assist others inovercoming magnetic and stability rroblems.His efforts contributed significantly to theimplementation of the technically complexmulti -output switching power supply.

Abraham E. Ruvin, for innovative systemdesign, performance analysis, and technicalleadership on the lightweight surveillanceradar program. His broad radar systemsexpertise was particularly effective in select-ing waveforms and processing approachesfor detecting stealth targets in heaVY clutter.Mr. Ruvin established the feasibility of achiev-

ing the very high clutter cancellation re-quired, and performed the antenna configu-ration tradeoff studies that resulted in anovel low-cost approach for the NavalAirship Radar.

Thomas J. Sikos, for significant contributionsto the software system development for thelightweight surveillance radar test bed. Asresponsible software systems engineer andlead designer of the radar control computer,Mr. Sikos developed a control architecturefully compatible with both the array pro-cessor and the host computer operatingsystems. His modular top-level design andprogramming direction were major factorsin the on -schedule, below -budget imple-mentation.

Ruvin named 1985 AnnualTEC Award winner at MSRD

The Missile and Surface Radar Division's1985 Annual Technical Excellence Awardhas been given to Abe Ruvin for specialtechnical achievements directed towardestablishing MSRD's credentials as a bonafide competitor in the field of lightweightairborne radar.

Ten receive CE fourth-quarterawards

The fourth-quarter 1985 Consumer Elec-tronics Operations Technical ExcellenceAwards were announced on Feb. uary 20,1986. They were:

Jacob Sosniak and Leslie Avery, for thedevelopment of structures on integrated cir-cuits that protect the device from damagedue to electrostatic discharges (ESD). Mr.Avery designed and developed structuresfor both bipolar and analog integrated cir-cuits that protect each pad and the internalcircuits from ESD damage. Mr. Sosniakdesigned and developed laboratory appara-tus and techniques to evaluate and quantifythe performance of ESD protection struc-tures. He also coordinated the implementa-tion of Mr. Avery's structures in several ICdesigns at CE. The structures, which provideprotection to discharge levels of 4KV, offerESD immunity levels two to three timesgreater than the competition.John Furrey, for the initiation and design ofa module (PWCPA) to provide cost-effectivestereo in a leader model television receiver.The module combines a volume/tone con-trol module and stereo power amplifier. Mr.Furrey worked to minimize the parts countand power requirements, and designed asystem that resulted in a net savings ofnearly $8.00 at the instrument level. Althoughthe design was started at GR-1, it wascompleted in time to meet the original instru-ment introduction date.

Philip Morris and Thomas Sipher, for theextensive and thorough evaluation of severalcandidate CAD/CAM systems, leading toidentification and RCA corporate approvalof a multi -million dollar, second -generationmechanical CAM system for CE. This evalua-tion focused on several key attributes (geo-metric modeling, programming environment,hardware platform, application software, andimaging capability) that provide a foundationfor improved design quality and significantdesign cycle reductions. In addition to per-forming thorough evaluation and benchmarktesting of several vendor's mechanical CADsystems, Mr. Morris and Mr. Sipher havedeveloped an innovative parametic approachto remote transmitter and television cabinet/mask design.

William Fitzgerald, for the development of anew VIPUR (Variable Interval Pulse Regu-lator) power supply system. Utilizing a high -voltage power MOSFET operating in a non -synchronous mode, the VIPUR provides notonly increased average power capability,but also higher peak power and transienthandling capabilities. The unique use of thepower MOSFET operating at increasedfrequency has permitted considerable reduc-tion in ferrite size, space, and cost whencompared to the chopper system as used inthe CTC 131.

Bill Testin and Juri Tults, for the develop-ment of the FS -III tuning system and its firstsuccessful application in the MTT1used in the CTC 120 chassis. The systemincludes improved testability, automaticchannel programming, extended cable tun-ing range, a new channel change mutingalgorithm, and on -screen display stabiliza-tion. The FS -III IC combines the functionalityof four ICs from previous tuning systems. Anew five -band op/switch replaces the pre-vious system, which used a three -bandswitch and a discrete circuit. The FS -III wasdesigned to be used across the line ofdigitally -tuned systems.

G. Rene VanHoorde, for the research anddevelopment of a reliability model for colortelevision. Although established reliabilitymodels are well known, they are primarilydesigned for military applications and donot reflect the home television environment.Mr. VanHoorde's model predicts reliabilityusing data from stress tests, past failurerates, vendor ratings, and component rat-ings. It is considered unique in the televisionindustry. Field data has shown the resultsof the new model to be statistically twiceas precise as the previous method. Themodel is used to evaluate each chassisgo -round and highlight areas for improve-ment. It has been instrumental in the reduc-tion of field failure rates since 1981.

Leroy Wignot, for the development and useof a computer model to design the oscillatorcircuits in the MTT-1 tuning system. The

RCA Engineer 31-2 mar. / Apr. 1986 83

Technical excellence

VCDD TEC Awards

Cooper

Thro

Deal

Adams

Opresko

Lausman

RCA Video Component and Display Division(VCDD) recently honored thirteen employeesfor their outstanding contributions andachievements in engineering during 1985.Charles A. Quinn, Vice President and Gen-eral Manager of the Video Component &Display Division, presented the awards toeach of the quarterly winners and concludedby presenting the 1985 Engineer of the Yearaward to John C. Cooper. Mr. Cooper, whohas five patents to his credit, was recog-nized for his design of the degaussing sys-tems and evaluation of the magnetic charac-

Krufka

Baker

MulhollandMulholland

Michielutti

teristics of RCA's complete family of COTYcolor picture tubes. Others honered were:

Stephen T. Opresko, for the implementationof a thermal/mechanical modeling capabilityfor analyzing electron guns for color picturetubes.

Frank S. Krufka, for the design of concepts,circuits, and algorithms to measure openingsize and other parameters of unphosphoreddisplay panels.

Denis G. Mulholland, for the developmentand implementation of vibratory finishingtechniques for electron gun parts resultingin improved color picture tube performance.

Dennis E. Thro, for the development of auniversal yoke holder to support deflectionyokes in test sets.

Thomas C. Lausman, for the design anddevelopment of an electron gun for use inhigh -resolution data display tubes havingstate-of-the-art resolution and convergenceperformance.

Paul D. Baker, for the evaluation of theperformance and establishment of potentialproductivity gains of iron oxide workingplates in making 19V100 aperture masks inthe Barceloneta production environment.

Thomas J. Michielutti, for the evaluation ofthe performance and establishment of poten-tial productivity gains of iron oxide workingplates in making 19V100 aperture masks inthe Barceloneta production environment.

Samuel B. Deal, for the successful coordina-tion, development, and implementation ofthe automated ADP materials and processesfor the 27V SP color picture tube.

Hsing-Yao Chen, for the design and develop-ment of an electron gun for use in high -resolution data display tubes having state-of-the-art resolution and convergence per-formance.

Samuel P. Benigni, for the development andimplementation of vibratory finishing tech-niques for electron gun parts resulting inimproved color picture tube performance.

Michael E. Mitchell, for his contributions tothe design and installation of automatic loadand unload equipment for color picturetubes in the Subjective Color Automatic TestSystem.

John L. Adams, for this contributions to thedesign and installation of automatic loadand unload equipment for color picturetubes in the Subjective Color Automatic TestSystem.

Astro TEC Awards

P. Bruschi, D. Corrington, D. Sable, A. Spitz,and T. Young received the Astro-ElectronicsEngineering Excellence Award for the Mes-sage Authentication Device concept anddesign implemented on the ASC Communi-cation Satellite.

The RCA Astro Engineering Excellence Com-mittee presented the Engineering ExcellenceAward to Chirasak Thasanatorn for hisdevelopment of an in-house analysis capa-bility to perform Coupled Transient LoadsAnalysis for payloads launched by the Shut-tle. This new in-house analysis capabilityis essential for accurately predicting thedynamic loads for all Shuttle launchedpayloads.

Terry Tracy, member of the Attitude ControlSystems group, has been awarded the

82RCA Engineer 31-2 Mar. / Apr. 1986

Istvan Gorog, Director, Manufacturing Tech-nology Research Laboratory, announceshis organization as follows: David P. Bort-eld, Head, Control Systems Research;Phillip M. Heyman, Head, Advanced Tech-nology Research; and Saman Hong, Head,Modeling and Analysis Research.

Fred Sterzer, Director, Microwave Technol-ogy Center, announces his organizationas follows: Erwin F. Belohoubek, Head,Microwave Circuit Technology; RaymondL. Camisa, Head Microwave ComponentsTechnology, S. Yegna Narayan, Head,Microwave Device Technology; MarkusNowogrodzki, Head, Subsystems and Spe-cial Projects; Barry S. Perlman, Head,Design Automation Technology; Herbert J.Wolkstein, Head, Space and Counter Mea-sures Program.

Communication andInformation Systems Division

Charles A. Schmidt, Acting Division VicePresident and General Manager, Communi-cation and Information Systems Division,announces the appointment of Donald J.Parker as Division Vice President, DigitalCommunications and Recording Systems;Joseph Pane as Director, Special Programs;and Joseph B. Christopher as Director,Plant Operations.

Microelectronics CenterDavid S. Jacobson, Director, IC ProductOperations, announces his organizationas follows: Richard H. Bergman, Manager,Test Operations; Robert A. Donnelly, Man-ager, Product Engineering and ProductionControl; Peter P. !dell, Manager, WaferFabrication Operations; and Richard H.Zeien, Manager, Package and AssemblyOperations.

Missile and Surface RadarDivision

William V. Goodwin, Division Vice Presi-dent and General Manager, Missile andSurface Radar Division, announces theappointment of Bryce D. Inman as Director,Submarine Combat Systems.

NBC

Michael Sherlock, Executive Vice President,Operations & Technical Services, an-

nounces his organization as follows: JamesM. Overman, Director, Broadcast Controland Communications; Joseph M. Rohr,Director, Program Production, Sharon R.Johnston, Manager, Capital and ProjectImplementation; Theodore O'Karma, Direc-tor, Communications and Support Services;Thomas C. Nollie, Manager, ScreeningRooms; Joseph A. Ralston, Jr., TechnicalManager, Studio Technical Operations;Warren Allgyer, Managing Director, LongRange Space Plan, Engineering; S. MerrillWeiss, Managing Director, Systems Engi-neering; Charles Jablonski, Director, Tech-nical Strategic Planning; David Rabinowitz,Director, Computer Imaging & AutomationSoftware; Peter Haas, Manager, Engineer-ing Operations and Services; Louis Libin,Manager, Allocations Engineering; and AlanJ. Wechsler, Project Manager, BSE.

Solid State Division

RCA and Sharp have entered into a strategicalliance with WaferScale Integration, Inc.The agreements provide for licensing andjoint development of WSI's CMOS macrocelllibrary, non-volatile memories and bit -sliceproducts. WSI is a privately held Californiabased corporation.

Carl R. Turner, Division Vice President andGeneral Manager, Solid State Division,announces the appointment of Larry J.Gallace as Division Vice President, ProductAssurance, MIS and Materials.

Carl R. Turner, Division Vice President andGeneral Manager, Solid State Division,announces the appointment of Richard A.Santilli as Division Vice President, StrategicPlanning.

As part of the agreement, RCA/SharpMicroelectronics, Inc. has acquired 7 per-cent of WSI's outstanding stock. RichardSantilli, Division Vice President, StrategicPlanning will represent RCA, Sharp andRSM on the WSI board of directors.

Heshmat Khajezadeh, Division Vice Pres-ident, Integrated Circuit Products, an-nounces his organization as follows:Stephen C. Ahrens, Director, ProductMarketing-Standard IC Products; John R.Kowalak, Administrator, IC Administration;James L. Magos, Director, Government &High Reliability-IC Products; Charles J.Nuese, Division Vice President, Manufac-turing-ICs; H. Gene Patterson, Director,Application Specific IC Products; and BrunoJ. Walmsley, Director, Engineering-Stan-dard IC Products.

Bruno J. Walmsley, Director, Engineering-Standard IC Products, announces his orga-nization as follows: Kenneth W. Brizel,Manager, Engineering-Logic, Memory &Microprocessor Products; Charles Engel -berg, Manager, Test Engineering; Lewis A.Jacobus, Jr., Manager, Engineering-Auto-motive & Consumer Products; Sterling H.Middings, Section Manager, Layout Ser-vices; and Bruno J. Walmsley, Acting Man-ager, Engineering-Computer, Telecom-munications & Industrial Products.

H. Gene Patterson, Director, ApplicationSpecific IC Products, announces his orga-nization as follows: R. Adrian Bishop,Manager, Product Marketing-ApplicationSpecific ICs; James E. Gillberg, Director,Engineering-Application Specific ICs; andHenry S. Miller, Manager, Design Centers.

Charles J. Nuese, Division Vice President,Manufacturing-ICs, announces his orga-nization as follows: Frank F. Cocozza,Manager, Plant Operations-Palm BeachGardens; Raymond T. Ford, Manager, WaferFabrication-Palm Beach Gardens; CharlesJ. Nuese, Acting Manager, Operations Sup-port; and John R. Steiner, Director, Environ-mental & Plant Engineering-SSD.

James L. Magos, Director, Government &High Reliability-IC Products, announceshis organization as follows: Joseph A.Augusta, Administrator, Government andHigh Reliability IC Administration; WilliamF. Allen, Manager, Design EngineeringGovernment and High Reliability IC Prod-ucts; Jack Handen, Manager, Product Mar-keting-Government and High ReliabilityIC Products; Donald R. Carley, Manager,Applications & Product Engineering-Gov-ernment and High Reliability IC Products;and James L. Magos, Acting Manager,Operations Support.

Stephen C. Ahrens, Director, Product Mar-keting-Standard IC Products, announceshis organization as follows: Donald L.Beech, Manager, Product Marketing-Tele-communications & High Speed Logic Prod-ucts; Edmund J. Blickle, Manager, ProductMarketing-Microprocessor Products; Rich-ard E. Funk, Manager, Applications Engi-neering-Linear & Logic Products; RalphS. Hartz, Manager, Memory & Microproc-essor Applications Engineering and RSMProduct Definition; Thomas J. McInerney,Manager, Product Marketing-Distribution& Memory Products; and Jack Yellin, Man-ager, Product Marketing-Linear Products.

RCA Engineer 31-2 Mar. / Apr. 1986 81

Manager, PRICE Model Operations; andWilliam W. Kuhn, Manager, PRICE SystemProjects.

Mark H. Burmeister, Director, PRICE Sys-tems, announces his organization as fol-lows: Mark H. Burmeister, Acting, PRICEMarketing; John J. Procaccino, Manager,PRICE Business Operations; William E.Rapp, Staff Engineer, PRICE Systems; andRoy M. Summers, Manager, PRICE Opera-tions and System Projects.

Americom

John Christopher, Vice President, TechnicalOperations, announces his organizationas follows: Joseph F. Elko, Manager, Space-craft Engineering and Operations; MurrayFruchter, Director, Terrestrial Systems,Technical Operations; Joseph J. Schwarze,Director, Quality Management and Ad-vanced Space Programs; and Carlton N.Thomas, Manager, Business Operations.

Joseph F. Elko, Manager, Spacecraft Engi-neering and Operations, announces hisorganization as follows: Jacob Levin,Manager, Mission Operations; Archie T.Miller, Manager, Spacecraft Operations;and Joseph F. Elko, Acting Manager, Space-craft Engineering.

Joseph J. Schwarze, Director, Quality Man-agement and Advanced Space Programs,announces his organization as follows:Lindy W. Morgan, Manager, ProcurementQuality Assurance; Albert E. Schmidt, Man-ager, Reliability and Quality Engineering;William R. Palme, Manager, AdvancedSpace Programs; and Rodney A. Stevens,Manager, Configuration Management, Prod-uct Safety, and Drafting.

Automated Systems Division

Eugene M. Stockton, Division Vice Pres-ident and General Manager, AutomatedSystems Division, announces his orga-nization as follows: Duane M. Belden,Director, Plant Operations; Glenn W. Cole-man, Division Vice President, Market-ing and Advanced Planning; Emilio W.Dusio, Division Vice President, Engineer-ing; Thomas E. Fitzpatrick, Division VicePresident, Vehicle Test Systems; James E.Kupec, Director, Employee Relations; Ken-neth I. Pressman, Director, Finance; DavidM. Priestley, Division Vice President, Auto-matic Test Systems; Murray D. Radio,Director, Materials; and Eugene M. Stock-ton, Acting, C31 Systems.

David M. Priestley, Division Vice President,Automatic Test Systems, announces hisorganization as follows: Donald R. Bartlett,Manager, Program Operations; John H.

Groff, Manager, Program Operations; JohnHallal, Manager, Program Operations; andWalter R. Wadden, Staff Scientist, SoftwareSystems.

Eugene M. Stockton, Acting, C31 Systems,announces his organization as follows:John J. Morris, Manager, Program Opera-tions; Gerard T. Ross, Manager, ProgramOperations; Allan R. Stern, Manager, Pro-gram Operations; Chris A. Wargo, Manager,Program Operations, and James C. Wil-liams, Manager, Program Operations.

Astro-Electronics DivisionJack A. Frohbieter, Division Vice Pres-ident and General Manager, Astro-Elec-tronics Division, announces the appoint-ment of Frank A. Boyer as Director, PlantOperations.

Consumer ElectronicsOperations

Jack K. Sauter, Group Vice President,Consumer Electronics and Video Com-ponents, announces that effective imme-diately, the Video Display Monitor ProductsDivision name will be changed to DisplaySystems Division. The organization willcontinue to report to the Group Vice Pres-ident, Consumer Electronics and VideoComponents.

Richard A. Sunshine, Director MechanicalDesign Engineering, announces the ap-pointment of Roger D. Sandefer, as Man-ager, Product Protection Engineering.

Bruce M. Allan, Vice President, StrategicPlanning, announces his organization asfollows: Theodore L. Allen, Manager, Prod-uct Operations; James 0. Early, Manger,Quality Systems and Procedures; R. EugeneEddy, Vice President, Warranty Programsand Training; Reginald J. McKinstry, Jr.,Director, Entertainment Tubes and PartsProduct Management, Distributor and Spe-cial Products Division; Denzil R. Miller,Director, Business Analysis; William H.Needham, Manager, Finished Goods Pro-curement; Gilbert M. Ravelette, Manager,Video Cassette Recorder Merchandising;Alfred J. Schick, Manager, Video ProductsTechnical Support; and Brian K. Smith,Manager, Television Product SystemsDevelopment.

Jack K. Sauter, Group Vice President,announces that Bruce M. Allan will assumeresponsibility for the Direct Broadcast Sat-ellite Program. These responsibilities in-clude the interdivisional coordination ofproduct design, development and marketintroduction. Mr. Allen will continue to reportadministratively to his functional superior

in his current position, Vice President,Strategic Planning.

Larry J. Byers, Manager, Test Technology,announces the appointment of James M.Keeth, Manager, Direct Broadcast SatelliteTest Systems.

Earth Observation SatelliteCompany

An Executive Committee has been namedfor the Earth Observation Satellite Company(EOSAT), which has taken over operationof the Landsat satellite program. Membersof the committee from RCA are Charles A.Schmidt, RCA Group Vice President, Gov-ernment Communications Systems, andRobert A. Amadio, Director, Finance, forRCA Astro-Electronics Division, East Wind-sor, N.J. Names from Hughes Aircraft Com-pany's Santa Barbara Research Center,Goleta, Calif., were Dr. Robert Talley, Presi-dent, and Warren Nichols, Vice President.The executive committee is responsiblefor EOSAT policy and executive manage-ment and overview. EOSAT, a joint venture/partnership formed by Hughes and RCACorporation, is headed by C. P. Williams,President and Chief Operating Officer. Thefederal government, which formerly oper-ated the Landsat satellites, transferred theprogram to EOSAT last September. Landsatinvolves the remote sensing of the Earth'sresources by satellite and the disseminationof the collected data to users around theworld.

Government VolumeProductionCharles A. Schmidt, Group Vice President,Government Communications Systems,announces the appointment of Joseph L.Mackin as Division Vice President, Govern-ment Volume Production.

The Division Vice President and GeneralManager, Government Volume Production,announces the appointment of Victor C.Scarpello as Manager, Manufacturing.

RCA Laboratories

James J. Tietjen, Vice President, RCALaboratories, announced that effectiveMarch 1, 1986, Jay J. Brandinger wasappointed Staff Vice President, Manufac-turing and Materials Research, replacingJames L. Miller who retired on April 1,

1986. Additionally, Dr. Brandinger will con-tinue his present responsibility as StaffVice President, Engineering, and in thiscapacity will report to the Executive VicePresident, Electronic Products andTechnology.


Engineering News and Highlights

Frohbieter is headAstro-Electronics

John D. Rittenhouse, Executive Vice Pres-ident, Aerospace and Defense, has an-nounced the appointment of Jack A.Frohbieter as Division Vice President andGeneral Manager, RCA Astro-ElectronicsDivision. He succeeds Charles A. Schmidt,who has been named an RCA Group VicePresident.

Prior to his current appointment, Mr.Frohbieter served as Division Vice President,Communication Satellite Programs. His re-sponsibilities included Video DistributionSatellites, General Purpose CommunicationsSatellites, and the NASA Advanced Com-munications Technology Satellite (ACTS)program. Previously, as Director, SatellitePrograms, he was responsible for manage-ment of Astro's fixed -service communica-tions satellite programs.

Mr. Frohbieter received the BS in Elec-trical Engineering from the University ofEvansville. He also attended Purdue Uni-versity and is a graduate of the HarvardBusiness School Advanced ManagementProgram.

deBastos named Division VPat Astro

Jack A. Frohbieter, Division Vice Presidentand General Manager, RCA Astro-Elec-tronics has announced the appointment ofRicardo deBastos as Division Vice Pres-ident, Communications Satellite Programs.Mr. deBastos is responsible for managementof all Astro communications satellite activ-ities, including Video Distribution Satellites

and General Purpose Communications Sat-ellites. He was previously Manager, GeneralPurpose Communications Satellites, respon-sible for all aspects of each program'sdelivery, performance, and cost. In 1971,he became a Systems Manager for theAtmosphere Explorer Program and laterserved as the Program Manager for otherAstro programs such as the DynamicsExplorer, TIROS and GSTAR satellitesystems.

Mr. deBastos received the BS and MSdegrees in Mechanical Engineering fromOklahoma State University. He is a memberof Pi Tau Sigma, Theta Chi and the Amer-ican Geophysical Union, and is an AssociateFellow of the AIAA. He is also an assitantprofessor, adjunct staff and Chairman ofthe Mechanical Advisory Commission atMercer County College in Trenton, N.J.

Bedrosian is named NPD TPAand Ed Rep

Peter G. Bedrosian has been named theTechnical Publications Administrator andthe RCA Engineer Editorial Representativefor the Lancaster plant of the New ProductsDivision. He is a Member, Technical Staffin the Communications Department of theClosed -Circuit Video Equipment activity, andis responsible for the establishment of datasheets, brochures, and product guides forall products manufactured or marketed bythe New Products Division in Lancaster,Pa. and Vaudreuil, Quebec.

Mr. Bedrosian joined RCA in 1953 as aspecialized trainee in the Color KinescopeProduction Development group of the Lan-caster plant. After a leave of absence to

serve in the Armed Forces, he returned toLancaster as a production engineer in PowerTube manufacturing. He has held variousengineering positions including Manager,Engineering Standards. He has also servedon several military ad hoc and JEDEC com-mittees. Mr. Bedrosian received the BS inElectrical Engineering from the Universityof New Hampshire in 1953.

Staff announcements

Aerospace and Defense

Leonard V. Fox, Staff Vice President, Fi-nance, announces his organization as fol-lows: Arthur J. Barrett, Manager, Facilitiesand Special Programs; Mark H. Burmeister,Director, PRICE Systems; John R. Lee,Director, Operations/Contract Review andOperations Analysis; and Peter J. Pernice,Director, Financial Planning and Operations.

John D. Procaccino, Manager, PRICE Busi-ness Operations, announces his organiza-tion as follows: Benjamin Chackman, Man-ager, PRICE Timeshare Services; GeraldineC. Devlin, Administrator, Contracts; RichardE. Moyer, Administrator, PRICE BusinessOperations; and Susan Steinberg, Admin-istrator, PRICE Systems.

Roy M. Summers, Manager, PRICE Opera-tions and System Projects, announces hisorganization as follows: Francis B. Barnett,Manager, PRICE operations; Peter B. Korda.

RCA Engineer 31-2 Mar. / Apr. 1986

Communication andInformation Systems Division

T. Anastasia I D.G. Herzog I D.W. DonzeHardware Implementation of a Two -Axis,Three -Tapped, Finite -Impulse Response(FIR) Digital Filter -Presented at theSociety of Photographic Scientists &Engineers (SPSE), 25th Fall Symposium -Imaging, Arlington, Va. Nov. 17-22, 1985and published in Symposium Brochure

H.R. Barton, Jr.Unit Production Cost Tracking ModelUPCT6-Presented at the ElectronicIndustries Association (EIA) G-47Meeting, Washington, D.C., October 2,1985 and published in EIA EngineeringDesign Integration Bulletin No. 1-B

O.E. BessetteHigh Data Rate, High Capacity, SevereEnvironment Optical Disk RecordingEquipment -Presented at the 21st AnnualInternational Telemetering Conference,Las Vegas, Nevada, Oct. 28-31, 1985 andpublished in ITC -85 Proceedings

S.L. Corsover W.L. VaughanFrequency Response of an ImageRecorder Using Repeat Scan -Presentedat the Society of Photographic Scientistsand Engineers (SPSE), 25th FallSymposium -Imaging, Arlington, Va., Nov.17-22, 1985 and published in theSymposium Brochure

D.A. Ellengold I M.H. Merkert Oddo J.J.SalernoSimplified Voice Tracking ModelSimulation -Presented at the IEEEGlobcom '85 New Orleans, La., Dec. 2-5,1985, and published in the Proceedings

H. HolcombeProject Development Using Design andDevelopment Tools -Presented at the1985 Fall Digital Equipment ComputerUsers Society (DECUS) U.S. Symposium,Anaheim, Cal., Dec. 9-13, 1985

H.P. MillerA Comparison of the Flux -CorrectedTransport Method with Other Low -Dispersive Schemes for 2-DIncompressible Convection DominatedFlows with Steep Gradients -Presentedat the American Society of MechanicalEngineers (ASME) Winter AnnualMeeting, Miami, Fla., Nov. 17-21, 1985and published in ASME Journal of FluidsEngineering

H.P. MillerA Comparison of High ResolutionSchemes for the Two -Dimensional LinearAdvection Equation -Presented at the38th Annual Meeting, American PhysicalSociety, Division of Fluid Dynamics,Tucson, Ariz., Nov. 24-27, 1985

D.P. O'Rourke I R.O. Yeager I J.I. PridgenA.L. BrambleAgile, Low Power CMOS/SOSFrequency Synthesizer -Presented at theGOMAC '85 Conference, Orlando, Fla.,Nov. 7, 1985, and published in theGOMAC Conference Proceedings

M.H. Riddle I P.F. MuracoRotary Digital Recording -Addressingthe Storage Needs of the Future -Presented at the 21st Annual 1985International Telemetering Conference,Las Vegas, Nev., Oct. 28-31, 1985, andpublished in the ITC -85 Proceedings

T.E. SasenaApplications Specific Chip Set forMicroprocessor I/O -Presented at theGOMAS '85 Conference, Orlando, Fla.,Nov. 6, 1985, and published in theGOMAC Conference Digest

J.D. WhalenTrends in Personal ComputingTechnology -Auerbach Data ProcessingManagement Periodical, December 1985

RCA Laboratories

W.F. Kosonocky F.V. Shallcross T.S.

Villani J.V. Groppe160X224 Element PtSi Schottky -BarrierIR-CCD Image Sensor-lEE Transactionson Electron Devices, Vol. ED -32, No. 8,August 1985

P.A. LevineLow -Noise CCD Signal Recovery -IEEETransactions on Electron Devices, Vol.ED -32, No. 8, August 1985

E.S. Nartowitz I A.M. GoodmanEvaluation of Silicon Optical AbsorptionData for use in Minority -Carrier -Diffusion -Length Measurements by the SPVMethod -Journal of the ElectrochemicalSociety, Vol. 132, No. 12, December 1985

J.R. NelsonElectromagnetic Interference Shielding ofPlastics -Presented at the University ofToledo (11/7/85)

D. RaychaudhuriStability and Optimal RetransmissionControl of Announced RetransmissionRandom Access Systems -IEEECommunications Society, IEEEInternational Conference onCommunications, June 23-26, 1985,Chicago, Ill.

L.K. WhiteA Modeling Study of SuperficialTopography for Improved Lithography -Journal of the Electrochemical Society,Vol. 132, No. 12, December 1985

Missile and Surface RadarDivision

F.J. BuckleyThe Search for Software Quality, or OneMore Trip Down the Yellow Brick Road -ACM Software Engineering Notes, Vol. 11,No. 1, January 1986

T.A. DorsayCommand -Activated Diagnosis and Test(CADET) on the DDG and CG 65 ClassShip-AN/UYK-44 Users' TechnicalConference, Monterey, Cal., November 6-7, 1985

R.J. Felbinger H.G. RoulandImproved system engineering support ofinstruction modes, technical manuals,and training courses -RCA Engineer, Vol.30, No. 6 (Nov/Dec 1985)

D.R. HiggsEngineering Documentation in theComputer Age -Are We Using the ToolsWisely? -IEEE Philadelphia Section, JointPCS/EM Meeting, Philadelphia, Pa.,January 27, 1986

W.K. HoldenMechanical design problems associatedwith glass -reinforced polytetra-fluorethylene printed -wiring boards -RCA Engineer, Vol. 30, No. 6 (Nov/Dec1985)

F.E. OlivetoReliability for the '80s -Productivity andTechnological Innovation Symposium,The Engineers' Club of Philadelphia,January 23, 1986

H.R. WeissA new peripheral computer interface forthe AEGIS weapon system -RCAEngineer, Vol. 30, No. 6 (Nov/Dec 1985)


Pen and Podium Recent RCA technical papers and presentations

To obtain copies of papers, check your library or contact the author or divisionalTechnical Publications Administrator (listed on back cover) for a reprint.

Advanced TechnologyLaboratories

J.A. Adams I J.W. Dempsey I MG.W. Kaizar N. StraguzziArtificial intelligence applications-RCAEngineer, Vol. 31, No. 1 (Jan/Feb 1986)

Gale

E. Alexander R. McClain C. Tsikos I B.

Siryj R.M. CarrellAdvanced technology for mail processingin the 1990s-RCA Engineer, Vol. 31, No.1 (Jan/Feb 1986)

W. P. Altman G. M. Claffie M. L. LeveneOptical storage for high performanceapplications in the late 1980s andbeyond-RCA Engineer, Vol. 31, No. 1(Jan/Feb 1986)

D.E. BrittonVerlangen: formally verifying systemdesigns-RCA Engineer, Vol. 31, No. 1(Jan/Feb 1986)

E.J. ConklinLeonardo: design environment of the1990s-RCA Engineer, Vol. 31, No. 1(Jan/Feb 1986)

R.J. Dick W.A. Helbig W.D. MoyersA High -Level Systolic Architecture forGaAs-Presented at the 19th AnnualHICSS, January 8-10, 1986 and publishedin the Proceedings, Vol. 1.

A. FellerA New design methodology reducesdesign time for very large applications -specific integrated circuits (ASICs)-RCAEngineer, Vol. 31, No. 1 (Jan/Feb 1986)

W.A. Helbig W.F. Heagerty R.M. ZiegerDesign Issues in GaAs ComputerSystems for Efficient Execution of HLLPrograms-Presented at the 19th AnnualHICSS, January 8-10, 1986 and publishedin the Proceedings, Vol. 1, pp. 238-245.

W.A. Helbig R.H. Schellack R.M ZiegerThe design and construction of a GaAstechnology demonstrationmicroprocessor-RCA Engineer, Vol. 31,No. 1 (Jan/Feb 1986)

W.A. HelbigRISC vs. CISC, GaAs vs. Silicon, andHardware vs. Software-Presented at the

joint meeting of the IEEE MTT/EDSocieties and the Computer Society,January 23, 1986

F.C. Luce I W.B. SchamingAutomatic classification of ship targets-RCA Engineer, Vol. 31, No. 1 (Jan/Feb1986)

T.E. Lynch D.E. GraffAcoustic phonetic techniques forautomatic speech recognition systems-RCA Engineer, Vol. 31, No. 1 (Jan/Feb1986)

G.W. Mebus J.B. Shields I L.T.Armstrong I H. RosenthalAda software development tools-RCAEngineer, Vol. 31, No. 1 (Jan/Feb 1986)

P.B. PiersonAdvanced robotics projects at ATL for

servicing-RCA Engineer, Vol. 31,No. 1 (Jan/Feb 1986)

F.B. WarrenAdvanced technology for future imagingsensors-RCA Engineer, Vol. 31, No. 1(Jan/Feb 1986)

D.A. Wille A. RosenLaser development and applications-RCA Engineer, Vol. 31, No. 1 (Jan/Feb1986)

Automated Systems DivisionJ.E. Greenup S.H. Ward I E.G. ZablockiNon -Contact Sensors for AutomotiveTesting-SAE International Congress andExposition, Detroit, Mich., February 26,1986

S.C. HaddenA Design Manager's View at RCA-Corporate Window Session, CornellUniversity, Ithaca, N.Y., February 6, 1986

S.C. HaddenEngineering and Physics Roles inIndustry-Aeronautics and EngineeringPhysics Address at Cornell University,Ithaca, N.Y., February 26, 1986

M.E. Petty I K.J. St. Pierre J.R. WilkinsonMultiple Fault Diagnosis with PortableATE Utilizing a Knowledge -Based System

Architecture-SAE InternationalCongress and Exposition, Detroit, Mich.,February 26, 1986

A.J. SilvaReconstruction of UndersampledPeriodic Signals-Thesis TowardMasters Degree in EE and CS at MIT,Department of EE and CS, Cambridge,Mass., January 1986

Astro-Electronics DivisionD.R. Chalmers I J.J. PustayAdvanced Communication SatelliteThermal Design Using VariableEast/West Radiators -24th AIAAAerospace Sciences Conference, Reno,Nev., January 6-8, 1986.

G.F. Chilson, Jr. D.F. Chu C. TrundleComparison of Predicted and MeasuredAcoustic Response of Components -Mounted Sandwich Panel -4th IntlModal Analysis Conference, Los Angeles,Cal., Feb. 3-6, 1986

D.R. Chu B.P. WangOn Changing Boundary Conditions inStructural Dynamics -4th InternationalModal Analysis Conference, Los Angeles,Cal., Feb. 3-6, 1986

D. Chu I B.P. Wang T.Y. ChenModel Refinement Using Test Data -4thInternational Modal Analysis Conference,Los Angeles, Cal., Feb. 3-6, 1986

J.P. LaytonMulti -Mode Nuclear Space PowerSystems -3rd Symposium on SpaceNuclear Power Systems, Albuquerque,N.M., Jan. 13-16, 1986

V.J. Mancino V.R. Monshaw I W.J.SlusarkReliability Considerations forCommunications Satellites-ReliabilitySymposium, Las Vegas, Nev., Jan. 29,1986

T. Tracy I R. Hogan K. WardThe Reactivation of Attitude Control onNOAA-8-Rocky Mountain Guidance andControl Conf., Keystone, Col., Feb. 1-5,1986


Patents

RCA Laboratories

Knop, K.H.Solid-state color -encoding televisioncamera -4546380

Kosonocky, W.F. I Elabd, H.Schottky -barrier diode radiant energydetector with extended longerwavelength response -4544939

Kosonocky, W.F. I Shallcross, F. V.Method of making a charge -coupleddevice imager which includes an array ofSchottky -barrier detectors -4548671

Kumar, M.Broadband adjustable phase modulationcircuit -4549152

Levine, P.A.Low noise black -level reference for CCDimagers -4549215

Levine, P.A.Dark -current level regulation in solid-state devices -4551762

Lewis, H.G., Jr. I Stiller, T.M.Logarithmic color matrix for a digitaltelevision receiver -4544945

Lewis, H.G., Jr. I Stiller, T.M.Digital signal processing systememploying logarithms to multiply anddivide -4555768

Longeway, P.A.Apparatus for mounting crystal -4547648

Maa, J. I Halon, B.Anisotropic etching ofaluminum -4547261

Martinez, M.R. I Mitchell, W.J., Jr.Press workpiece thickness measuringsystem and method -4548066

Nicholson, W.H. I Pritchard, D.H.Apparatus for frame -to -frame combfiltering composite TV signal -4550340

Palmer, R.C. I Plotnick, M.A.Video disc encoding and decodingsystem providing intra-field track errorcorrection -4551770

Pankov, J.I.Method of making p -type hydrogenatedamorphous silicon -4551352

Powers, K.H.Compatible and hierarchical digitaltelevision system standard -4550335

Pritchard, D.H.Apparatus for reducing motion induceddistortion in a frame combedchrominance signal -4555723

Rao, S.T. I Trager, L.L.Bath and process for theelectrodeposition of micromachinablecopper and additive for saidbath -4551212

Reichert, W.F.Method of making a gallium arsenidefield effect transistor -4545109

Russell, J.P. I Goodman, A.M.Power switching circuitry

Schiff, L.N.Noice reduction system for a singlesideband multiplex signal -4554679

Sepp, W.E.Progressive scan speed-upprocessor -4550336

Southgate, P.D.Thermally isolated imager -4547809

Wagner, T.M.Complementary color splitting filters usedin a color camera -4555163

Whartenby, J.C.Kumar, M.Predistortion circuit withfeedback -4554514

White, A.E.Panel seal and support structure for fiberoptic cable -4548465

Wine, C.M.Apparatus for synchronizing a source ofcomputer controlled video to anothervideo source -4554582

Woodward, O.M.Cross -polarization corrector for circularwaveguide-4549310

Microelectronics CenterDingwall, A.G. I Zazzu, V.Apparatus for matching FET switches asfor a video digital -to -analogconverter -4553132

Patent Operations

Meise, W.H. I Dischert, R.A.Compatible wide-screen color televisionsystem -4551754

Solid State Division

Chandrasekhar, H.S.Method of making integrated PNP andNPN bipolar transistors and junction fieldeffect transistor -4553318

Dackow, P.N.Auto ranging counter -4546487

Faulkner, R.D. I Rhoads, J.L.Electron discharge device having aceramic member with means for reducingluminescence therein -4554481

Harwood, L.A.1 Shanley, R.L., 2ndTri-level sandcastle pulsedecoder -4555722

Rodgers, R.L., 3rdFeedback kinescope driver -4547799

Savoye, E.D. I Tomasetti, C.M.Imaging device having improved hightemperature performance -4547957

Video Component andDisplay Division

Opresko, S.T.Cathode-ray tube having a low powercathode assembly -4554479

Schlack, R.E.Cathode-ray tube having an electron gunassembly with emissivity modifyingmeans -4554480


The Engineer's Notebook

license fee. Traditional methods of pricing relate to product cost,are based on historical data, or are established through marketsampling and testing. The nature of the product combinedwith the by-product approach to marketing makes it impracticalto apply these traditional approaches. Our method is to establisha fee in accordance with the utility value of the software to thecustomer. Through accurate quantification of this value is notalways possible, a reasonable approximation can be made.

Promotional methods depend upon the size of the market andthe method of distribution. When third party distribution is used,the prime licensee assumes the responsibility for distribution. Forvery small markets and to locate a suitable third party, promotionconsists of direct contact with the potential licensee. In the caseof large markets, the established business -to -business methods ofdirect mail, promotional articles in trade journals and papers,advertising, presentations at technical meetings, and trade showsare used.

All software is released under a licensing agreement thatprotects RCA's rights to the software and limits RCA's obliga-tions and liabilities. Preparation and administration of the agree-ments is performed by the Contracts Administration activity.

The Software Marketing activity, though part of RCA Lab-oratories, markets software for any RCA organization havingneed of its service. In such cases, an agreement is reached

between the MOU originating the software and RCA Labora-tories in which RCA Laboratories takes responsibility for allmarketing, licensing, and administrative functions. A significantaspect of the relationship is that all revenues are shared by theMOU and RCA Laboratories.

In addition to the marketing functions just described, SoftwareMarketing prepares and distributes the RCA Software Catalog,which contains a listing of RCA proprietary software. Thecatalog is distributed to all libraries and updated periodically topromote greater use of existing software and to prevent dupli-cation of effort.

Summary

The increased use of computers in the performance of engineer-ing and manufacturing tasks, coupled with a deficiency ofexisting software, has resulted in greater amounts of engineeringtime to develop software. Much of this software is unique andhas value to others both inside and outside RCA. This situationpresents RCA with an excellent opportunity to realize additionalgain from the engineering dollars spent in developing software.Addressing that opportunity is the task of the Software Marketingactivity.



Software, a valuable by-product at RCAA.J. Stranix

RCA LaboratoriesPrinceton, N.J.

Software is the new driving force in the world of computing. Asrecently as 1981, it was still considered a cottage industry, inspite of annual sales of 2.7 billion dollars. By 1984, annual salesexceeded 10 billion dollars, and are expected to more than triplein the next five years. To put this in perspective, 1984 televisionsales totaled six billion dollars.

Prospects of a rapidly expanding market have enticed not onlydozens of small entrepreneurs, but also have prompted thetraditional hardware manufacturers (IBM, Hewlett-Packard,DEC, etc.) to establish new software units or to enter into jointventures with existing software suppliers. Even unlikely players

such as communication companies and publishers-includingCBS, Dow Jones, Dun & Bradstreet, McGraw Hill and Simon& Schuster-are licensing programs that are then marketedthrough their distribution channels.

RCA develops more software than many companies in thesoftware business. Our programs are usually of the applicationstype and scientific in nature. As our engineers and scientistsperform their tasks, they continually advance the leading edge oftechnology. Ever increasing percentages of technology advance-ments are being accomplished with the use of computers and

software. The necessary software, in all too manycases, does not exist and is developed as one would create a toolneeded to perform an engineering task. The expense of thisdevelopment is considered part of the cost of designing ormanufacturing the product, and is recovered through the sale ofthat product.

Software, however, is an intellectual property that has valuebeyond its direct utility to RCA. As such, it can be copyrightedand licensed to others for a fee. Thus through the marketing ofsoftware, RCA has the opportunity to realize a greater return onits software investment dollars. It was to address this businessopportunity that the Software Marketing activity was established.

The software marketing concept is to take what as developedto solve RCA's software needs and market this "by-product" toothers with similar needs. RCA offers a license to interestedparties for an appropriate license fee. Our licensees receive acopy of the RCA software, appropriate documentation, and theright to use the software. Being a by-product, the software isusually licensed "as is" with no warranties. Since the marketingof software is beyond the objectives of the RCA activity thatgenerated it, involvement of the designers is kept to a minimum.

The tasks performed by the Software Marketing activityinclude:

Determining product availability Determining product marketability Establishing a distribution method Establishing appropriate licensing fees

01986 RCA Corporation.Final manuscript received November 25, 1986Reprint RE -31-1-13

Developing promotion programs Negotiating licenses Administration of productThese tasks, described briefly below, are those faced by a smallentrepreneur.

Candidate programs are located from several sources, includ-ing RCA Technical Reports, programs submitted to the PatentDepartment for copyright, or programs that are submitteddirectly to Software Marketing by the software authors. Inaddition, incentive programs have been proposed that shouldaid in increasing software author participation.

Determining marketability includes the obvious decisionregarding market size, value, expected return, competition, etc.But in addition, it entails:

Determining the existence of and the resolution of any legal

obstructions. Establishing whether it is in RCA's best interest to license or

to retain the software as company -private. Ensuring that RCA has copyright ownership.

is responsible for assuring thatthe marketing decisions are made in compliance with RCACorporate Policy, and that all affected MOUs have the oppor-tunity to evaluate the decision to license a particular programand to comment on any effects a decision to license would haveon their business activity. The decision to market a softwarepackage is made jointly by the Patent Department, LegalDepartment, the MOU originating the software, any user MOU,and the Marketing Department. In special instances, the reviewwill also include the Licensing Department.

Methods of distribution include direct licensing to the end useras well as marketing through a third party. In keeping with thecriteria to minimize RCA manpower involvement, the mostdesirable distribution method is through a third party, preferablyone whose product/service is complemented by the licensedsoftware package. An ideal third -party candidate is the manu-facturer of the equipment controlled by the software, because hissales force is in place and his customers are potential licensees.The software package enhances the value of his product, thereforemaking it more appealing to his customers. In such a scenario,the third party becomes a prime licensee with the right to grantsublicenses to his customers. He takes full responsibility for thesales, installation, and servicing of the software. In return, hereceives a greater market acceptance for his product as well as aportion of the license fee, with*the balance of the fee going toRCA.

When the third party approach is not possible, direct licensingto the end user is considered. This approach is used when thecustomer base is small and the required support is minimal and

available.One of the more difficult tasks is establishing an appropriate



slowed down by the need to move data in and out eight bits ata time, instead of 16 bits at a time. A related processor, theIntel 8086, is a true 16 -bit machine, and has the full 16 -bit datapath to the outside world. Unfortunately, the two processors arenot pin -compatible, so you just can't pull out your 8088 andreplace it with an 8086. There are, however, some measuresyou can take to speed up your IBM PC by adding a co-processor. The co -processor is a device that takes over certainfunctions that the main processor would otherwise perform, anddoes them more rapidly. There are two types of co-processorsavailable for the PC. One of these is specifically designed totake over the floating point mathematics operations from the8088, and the other takes over the bulk of the computing tasksfrom the 8088 and leaves the 8088 just doing input-output andmemory management.

The first of these, the math co -processor, is an Intel 8087processor. It was designed by Intel specifically to performfloating point math functions, and works along with the 8088processor, taking over when certain machine instructions directit to do so. Since the 8087 has 80 -bit -wide registers, as comparedto the 16 -bit registers of the 8088, it can perform arithmeticoperations with much greater precision and much faster thanthe 8088. Typically, the addition of an 8087 co -processor to asystem may result in as much as a five -hundred -fold increase inprocessing speed for certain operations. A typical overall averagespeed increase is from five to seven times for a computationaltask that performs a mix of floating point operations. It shouldalso be noted, however, that NO speed increase will be realizedunless the software being run is written in such a manner as toutilize the co -processor, because it relies entirely on the executionof a special extension of the processor instruction set in order toperform its work. Thus, you won't see a speed increase for taskssuch as running an interpreted BASIC program, which doesn'tuse the co -processor.

A second type of co -processor is an auxiliary processor boardthat fits in one of the expansion slots in the machine, and plugsinto the processor socket on the motherboard. This type ofprocessor essentially replaces the 8088 processor with an 8086,80186, or 80286 processor, which are newer, more sophisticatedmembers of the same processor family as the 8088. These arecapable of higher performance levels than the 8088, but asnoted earlier, are not plug -compatible with the 8088, and thusrequire the extra hardware on the expansion board to operate.Examples of this type of board are the PC -Turbo from OrchidTechnologies, and the Quad Sprint board from Quadram. Theseboards will give the PC some of the operating properties of thenew PC AT, which uses the 80286 processor, but they arereally no substitute for the PC AT.

Also perhaps worth mentioning is the QuadLink board fromQuadram Corporation. This is a co -processor board that allowsthe use of Apple II software on the IBM PC, by taking overcontrol of the disk drives and display. This might be useful ifyou have a heavy investment in Apple software, but the usershould be aware that it will not run all Apple software, becausesome is copy protected.

There is one other type of processor that is available for theIBM PC. Nippon Electric Corporation makes a processor thatis designed to replace the 8088 processor in the IBM PC

entirely. This chip, called the V20, is designed to implementmany of the functions of the 8088 in hardware instead ofsoftware. What this means is that operations which might takeseveral machine cycles for the 8088 processor to complete areperformed in fewer cycles by the V20 processor. This results ingreater speed, for although the clock time is not changed, thenumber of clock cycles needed to perform a task may bereduced, depending on the task. Although some advertisementsclaim as much as a 50 percent increaw in speed, a morerealistic number is 10 to 12 percent, based on a mixture of tasksto be performed. This chip also does not require special softwareto effect a speed increase, as does the 8087. While this chip issaid to be 100 percent .hardware and software upward com-patible, I have not tested it, and thus cannot vouch for thatclaim.

Printer buffers

Generally, it is true that a machine such as the IBM PC canonly do one thing at a time. However, by offloading some ofthe work to another processor, you can gain an effective speedincrease. One of the ways to do this is with a printer buffer.Essentially, the way a printer buffer works is to provide thecomputer with a place to dump the data to be printed, acceptthe data as fast as the computer can output it, then store thedata and feed it to the printer at the rate that the printer canaccept it. Typically, an IBM PC can output data from theparallel port at a speed several times greater than even thefastest printer can accept it. Thus, if you put a buffer on the lineto the printer, the IBM PC will be done "printing" the datalong before the printer is actually finished, and thus will be freeto move on to the next task. The PC does its printing as abackground prompts anyway, stealing CPU cycles from whateverprocess is going on when they won't be noticed. By using thebuffer, the PC is making optimum use of the machine cyclesthat it dedicates to the printing. It may be noted that there aresoftware programs that dedicate a portion of memory to be setaside as a printer buffer. While these are better than having noprinter buffer at all, they suffer from two drawbacks. First ofall, they take up part of the random access memory space thatwould otherwise be usable as program or data storage space.This is not too much of a problem if you have the machineconfigured with 640 kbytes of memory, because very fewprograms can make use of all that memory anyway. Thesecond problem is that these programs also take up machinecycles, and can therefore actually slow down the execution ofprograms, since they are essentially another task for the processorto manage.

What if it doesn't say IBM on the front?Many of the upgrades mentioned here also will work on IBMclones as well. Generally, if the machine to be upgraded hasfull-length expansion slots, is slot interface -compatible with theIBM PC, and uses the same processor and operating system, atleast some of the upgrades mentioned in this article shouldwork.



floppy drives to be mounted. Larger drives are available as full -height units. If an expansion chassis is used, the fixed -diskdrives can be put in it, and four half -height floppy drives can bemounted in the main chassis.

The addition of a fixed -disk drive also requires the use of afixed -disk controller card, which will use up one of the expansionslots in the PC. The standard fixed -disk controller used by IBMsupports drives up to only 10 Mbytes in size, but there aremany aftermarket controllers produced by companies such asXebec, Western Digital, and Data Technologies Corporationthat may be outfitted with firmware to support the larger drives.If you are short of expansion slots within the PC, there is evena controller card available from Xebec with the capacity tohandle two hard drives and two floppy drives at the same time.

Another option available for updating an IBM PC is to addexternal mass storage. One of the most popular choices is theIOMEGA Bernoulli Box. The Bernoulli' Box uses removablecartridges as the storage medium. A single Bernoulli cartridgemay contain up to 20 Mbytes of data, 'and any number ofcartridges may be swapped in and out of the drives in a similarfashion to the way diskettes may be swapped in and out of adiskette drive. In addition to allowing virtually unlimited offlinestorage, this has the additional advantage of allowing a user toremove the cartridge for the purpose of locking up sensitivedata in a secure area if so desired. With the installation of theproper ROM in the PC, you can even start-up the machinefrom the Bernoulli Box instead of a floppy disk, in a similarmanner to the way you would start up from a fixed disk. TheBernoulli Box, in addition to being used as an alternative massstorage device, is also often used as a backup device for fixed -disk drives.

Backup for the massesFixed disks, because they can hold so much data, hold withinthem the potential for disaster as well. Anyone who has everhad a fixed disk crash, and has lost valuable data because of it,can describe in great detail the value of backing up your disk.

Fixed -disk backups may be performed in either of two basicways; either file -by -file, or image. In a file -by-file process,individual files are copied to another medium, one at a time. Inan image backup, an image of the entire disk, including unusedsectors, is copied to the backup medium. There are advantagesto both methods; file -by -file backup offers a convenient meansof restoring damaged files one by one. Image backup, on theother hand, is usually faster and, since it backs up unusedsectors as well, can usually back up a fixed disk containingcopy -protected software, and later restore it without making are -installation of the software necessary.

In Versions 2.1 and later of DOS, IBM provides a systemutility to allow a file -by -file backup of a fixed disk, using floppydisks as a backup medium. The capacity of a floppy disk is 360kbytes, while the capacity of the IBM fixed disk is 10.6 Mbytes.A little arithmetic shows that it takes 31 diskettes to back upone fixed disk. In addition to taking a fair-sized pile of diskettesto back up even a small fixed disk, it also takes a fair amount oftime. Both the expenditure of time and the number of diskettesrequired rises linearly with the size of the hard drive to be

backed up. Thus, a pair of 32-Mbyte hard drives could easilytake 200 diskettes and several hours to back up. This lack of aconvenient means of backing up the data on a fixed disk hasled many users to ignore this vital function until it is too late. Inorder to remedy this situation, an easy to use means of backup(that would also be inexpensive to use) was needed. Theanswer to this requirement came first in the form of the BernoulliBox. The original 10-Mbyte Bernoulli Box has a capacity ofslightly less than that of a 10-Mbyte (actually 10.6-Mbyte) harddrive; thus, in order to do an image backup of the disk, twocartridges are needed. This unfortunately wastes about 94 percentof the capacity of the second cartridge. In order to back up a10-Mbyte drive on a single Bernoulli cartridge, a file -by -filebackup can be done.

Another backup medium that is gaining popularity is thequarter -inch tape drive. A drive of this type can back up from10 to 60 Mbytes on a single tape, can do either image or file -by -file backups, and can often restore a single file from animage backup. Some of the tape drives on the market, such asthe Archive Scorpion, use their own controllers, while others,such as the Irwin Magnetics, can use the connector on the backof the floppy drive controller, which is normally used for theconnection of a third and fourth drive. The obvious advantageto the type that uses the floppy drive controller is that it doesnot take up a slot for the tape controller. However, if you areusing more than two floppy drives, or if you have a diskcontroller that can only handle two floppy drives, such as theXebec 1220 combination controller or the controller in theCompaq or Corona IBM lookalikes, you can't use this type ofdrive. There are usually separate controllers available for oper-ating this type of drive, but they are often designed only for thePC AT, and may or may not work with the PC XT and regularPC.

Power to run it allThe standard IBM PC has a power supply capable of providing63.5 watts of power to run the motherboard and all theexpansion cards and disk drives within the machine. This powersupply is insufficient to power a fixed disk, and may also beinsufficient to power all of the expansion memory and otherdevices that may be stuffed into the chassis by using expansioncards. Both the IBM PC XT and the IBM Expansion Chassis,on the other hand, are equipped with 135 -watt power supplies,which are sufficient to power one or two fixed disks and justabout anything else one might wish to put in the machine. Sucha supply, which will fit in the space occupied by the originalPC power supply, is now available from many sources as areplacement part, for those who might wish to upgrade the PCwith a fixed disk. For those who fear overloading even such aheavy duty supply, 150- and 175 -watt supplies are available asaftermarket add-ons from a few suppliers, as well.

More processing powerThe PC family, as noted, is based on the Intel 8088 micro-processor. While it uses 16 data bits in all its calculations, it hasonly an 8 -bit data path to the outside world. Thus, it often is



in either the enhanced color mode or the monochrome graphicsmode.

There are other display options to consider as well. IBM alsoproduces a Professional Graphics Adapter, which provides evennicer text and higher -resolution graphics display. PrincetonGraphics Systems has produced a high -resolution monitor, calledthe SR -12, which has finer horizontal resolution, and twice thenumber of scan lines for increased vertical resolution. Thismonitor is normally used with a card that doubles the scan rateof the signal going to the monitor, thus eliminating the "dot-matrix" look of the text characters displayed. Cards such as theTseng Laboratories ColorPak and UltraPak allow such thingsas more color flexibility in graphics and 132 -column displayson the monochrome display. Special software drivers are neededto take advantage of these enhancements, however.

Thanks for the memories

As mentioned earlier, the original IBM PC was limited to 64kof memory on the motherboard, and the later models of thePC, including the PC XT, are limited to 256k of memory onthe motherboard. This limitation in memory size was intendedby IBM to be overcome by adding additional memory on plug-in boards placed in expansion slots on the motherboard. Theoriginal IBM memory expansion boards held only 256k, so thatin order to increase the memory of the machine past 512k, twoof the available expansion slots would be used. This couldpresent a problem for the owner of a five -slot PC, or even aneight -slot PC XT. You could run out of slots rather quickly thatway. Companies such as AST research and Quadram soonsolved this problem by providing multifunction cards that pro-vide more than one function on a card, thus vastly increasingthe array of options you can put in one machine. Typically, amultifunction card might contain up to 512k of memory, oneor two serial ports, an on -board clock, a parallel port, andperhaps even a game port. This is as much equipment as couldbe contained on seven of IBM's boards; two for the memory,two more for the serial ports, and one each for the clock, theparallel port, and game port. When you count the fact that theIBM PC requires a monitor interface card and a disk drivecontroller card, you might need a nine -slot motherboard toconfigure a system that could be built using a multifunctionboard and only three slots. Of course, IBM also thought of away to get around the lack of slots in the PC. They haveproduced an expansion chassis for the PC and the PC XT. Theexpansion chassis has eight expansion slots, one of which istaken up by a transceiver card. A matching card takes up oneslot in the main chassis, providing a total of four slots in themain chassis, and seven in the expansion chassis.

The expansion chassis also contains a full -height, 10-Mbytehard disk. While this is fine if you want to add a 10-Mbytehard disk, as you will see later in this article, the 10-Mbytefull -height disk is no longer exactly the state of the art. In myopinion, it would have been much more useful had IBM soldthe expansion chassis without the hard disk, although as wehave seen, the need for an expansion chassis can often beeliminated altogether with multifunction cards. There are othercompanies that sell an expansion chassis without any disk

drives at all, so if that is really what is required, it should beeasy enough to find.

Mass storage devicesThe IBM PC, like any other computer, loses the contents of itsmemory when the power is turned off. Therefore, in order tosave programs and data, some sort of non-volatile storage isneeded. The original IBM PC could be equipped with a cassetterecorder for program and data storage, but this was very slowand inefficient, and did not allow for much flexibility in storage.IBM also equipped the original PC with full -height, single -sideddiskette drives, each of which held up to 160k of storage.Beginning with the introduction of Version 1.1 of the DiskOperating System (DOS), double -sided drives, which held 320k,were supported. Version 2.0 of DOS provided two moreadvances in disk storage. One of these was an increase in theamount of data one could store on a drive by increasing thenumber of sectors in a disk track from eight to nine. Thisprovided a capacity of 180k for single -sided drives, and 360kfor double -sided drives. A far more significant advance, however,was the introduction of a 10-Mbyte fixed -disk drive. This drivestores as much data as 31 floppy disks, but the storage mediumis not removable. In order to equip the PC -1 and some veryearly models of the PC -2 with a fixed -disk drive, the BIOSROM in the machine must be updated to the one that supportsthe hard disk drive. The 10-Mbyte hard disk was originallyoffered as standard equipment on the PC XT, but it has sincebeen made an option. The same disk drive was also availableinstalled in an expansion chassis, should one desire to add it tothe standard PC.

The IBM PC can accomodate up to two full -height floppydisk drives within its main chassis. There are, however, anynumber of half -height disk drives available that may be installedin the PC or PC XT; thus it is possible to install up to four diskdrives in the PC or PC XT. As a matter of fact, the standardIBM diskette drive controller is capable of supporting fourfloppy disk drives through an the external 37 -pin connector onthe back of the drive. Those interested in the technical details ofthis connection are referred to an article called "The ThirdDrive" in the June 1895 issue of PC Tech Journal (Vol. 3, No.6), written by Jack Wright and David Zarodnansky of RCALaboratories.

Of course, the disk drive mounting bays in the PC can just aseasily be used to mount fixed -disk drives. The original PC XTcontained one full -height floppy disk drive and one full -height10-Mbyte fixed -disk drive. It is just as possible to outfit the PCor PC XT with a pair of half -height floppy drives and one ortwo half -height fixed -disk drives as well. Machines using version2.0 or 2.1 of DOS support fixed -disk drives up to 16 Mbytes insize. Versions 3.0 and 3.1 of DOS support drives of up to 32Mbytes each without special software device drivers, and it ispossible to install drives as large as 140 Mbytes each by usingspecial software device drivers supplied by the vendors of thedrives. Ten, twenty, and even forty-Mbyte fixed -disk drives arenow readily available in half -height form factors. This makes itpossible to fit an IBM PC with up to 80 Mbytes of onlinestorage within the chassis, while still allowing one or two

RCA Engineer 31-2 Mr..r. / \pr. 198671


Dressing your PC for successDavid ZarodnanskyRCA Laboratories

Princeton, N.J.

It's been five years since the IBM PC first appeared on thescene, and there has been a host of newer and shinier machinesintroduced, but just because there are some fancier machinesaround doesn't mean that you should relegate your PC to thescrap heap yet. There are plenty of items that can be added tothe PC to make it perform nearly as well as the newer models,often at a fraction of the cost of trading up to a newer machine.

The IBM PC series includes the original IBM PC, which willbe referred to herein as the PC -1, the later models of the IBMPC, which will be referred to as the PC -2, and the IBM PC XT.The series is based on an Intel 8088 microprocessor. The PC -1had up to 64 kilobytes (kbytes) of memory on the motherboard,and used 4116 memory chips. The PC -2 and PC XT have upto 256 kbytes of memory on the motherboard, and use 4164memory chips. The two types of memory chips are not inter-changeable. Additional memory is available with add-on mem-ory boards, which are inserted into one of five expansion slotswithin the machine. Other slots can be used for a floppy diskdrive interface board, a color or monochrome video interfaceboard, serial ports, parallel ports, and so on. The original PC -1was equipped with BIOS (Basic Input Output System) firmware(the BIOS ROM), which allowed a maximum of 544 kbytes ofmemory to be installed in the system. This BIOS ROM madeno provision for a hard disk. The PC -2 and the PC XT areequipped with a BIOS ROM that supports up to 640 kbytes ofmemory and other devices that require a BIOS -callable firmware,such as a hard disk.

The PC -1 and the PC -2 each have five expansion slots. ThePC XT has eight expansion slots, although two of these areso-called "short" slots; that is, they will not accommodate afull-length expansion card. This is usually not a problem, how-ever, because many manufacturers produce short cards to go inthese slots.

Monitor optionsIBM's original idea with regard to monitors was to allow twooptions; either a monochrome monitor or a color monitorcould be purchased and used with the appropriate monitorinterface card. The advantage to the monochrome monitor andinterface is that the text quality is excellent-however, therewas no graphics capability. A color monitor system could bepurchased, which allowed for graphics capability, but the textquality was not nearly as good, and this was felt by many to bea definite disadvantage. If one desired both high -quality textdisplay and graphics, the only option open was to equip thesystem with two monitors and two monitor interface cards.

One of the first aftermarket vendors to attempt to remedythis situation was Hercules Computer Technology, which pro-duced a monochrome monitor interface card that also had the

01986 RCA Corporation.Final manuscript received January 28. 1986Reprint RE -31-1-12

Illustration by Joseph McGarrity

capability of displaying bit -mapped graphics. The Hercules cardis supported by most, but not all, major software packages thatuse graphics, so you have to be careful to note whether thesoftware you plan to run supports the Hercules card. Due tomemory addressing conflicts, the earliest Hercules cards couldnot co -exist with a color graphics card in the same system. Thenewer versions of the Hercules Monochrome Card may beturned off and on at will to allow the use of a color card in thesame system, although they will not give acri-ss to both graphicspages if the color card in the system is not also made byHercules.

A disadvantage of the IBM color card was the lack of aprinter port on the card. The monochrome card from IBM, inaddition to serving as a monitor interface, also contained aparallel port for use with a printer. If one used a color cardinstead, a printer interface card was also required if a printerwas to be used. Recognizing the need for a color interface cardwith a printer port built in, Hercules also has produced a colorinterface card that provides one. In addition, as mentionedabove, the Hercules Color Card has a software switch thatallows it to be used with a Hercules Monochrome Card. Anotheradvantage is the fact that the Hercules Color Card is a shortcard, so it can be placed in one of the short slots in a PC XTwith no problems.

IBM's monochrome monitor comes in only one color, green.Other companies, such as Princeton Graphics Systems (PGS)

and Amdek, have produced monochrome monitors in amberand white, which many users feel are much easier on the eyesfor long term use. PGS and Sony have also produced colormonitors with a finer dot -pitch than that of the IBM colormonitor, thus providing a sharper image, although with nofurther increase in resolution due to the fact that they are stillused with the standard color monitor interfaces. Thus the textdisplay on even these higher -quality color monitors is relativelypoor when compared to that on the monochrome monitor.

If you really want to go first class with respect to displayoptions, you can outfit your PC with IBM's Enhanced GraphicsAdapter and Display. This adapter is the only unit presentlysold by IBM that will allow graphics display on the monochromemonitor. When used with the Enhanced Color Display, the textquality is superior to that of the standard IBM color monitor,and higher graphics resolution is possible. Unfortunately, thereis also a scarcity of software that supports this display and card


professional -looking results. We present the included examplesas evidence.

AcknowledgmentsI wish to thank all those with included examples for their time,patience and help. Because of time limitations, I could not beginto interview everyone from Staff Vice -Presidents down whohave made significant productive (and often elegant) use of theMacintosh. A stack of functional and aesthetically pleasingMacDrawings would simply be overkill. I am sure that I havemissed packages and application areas that are significant, buthope that what is included will lead to more exploitation of themachine. In addition to the tools used by the people creatingexamples, I used MacPaint to edit icons and captured screenimages, MacDraw to integrate and create figures, Mac Write towrite the text (MacSpell+ to check it), and PageMaker to layout and integrate the entire document.

Table II. Comparison of software for the Macintosh.

1 111-1- 0 F-.

1g,I

cnun

-'-'

-aeo nom

,-.2

p.,qg-.§ to

....-g

.5-ggiE._sg.m

'n iCommentsProgram

MacTerrninal x 85 VT -100 emulatorTekalike x x 250 Supports plotters

VersaTenn X x 100. Better VT 100 & 4014

Multiplan x 110 Replaced by Excel

Excel x x X 225 More power than JazzChart x 75 Replaced by Excel

OverVue x x 150 Fast, easyHelix x 220 Stores pictures

MacWrite It 85 Easy

Word x 115 Mote power than WriteJars a x x x x 350 Does everything easily

MacDraw x 140 Easy, powerful

MacDraft x 140 Rotates 1 degree

MacPaint x 85 Easy

PageMaker x 360 Mom power than RSGReady Set Go x 65 Inexpensive

ThinkTank x 125 l'o organize projects

Statview a 100 With ANOVA SSO for TekPrint to pan, zoom, and print graphics on the LaserWriter.

Prior to receiving his MS in Operations Research (with a minorin systems design) from Rutgers University, Don Barton workedfor nine years as a field engineer on U.S. Air Force flightsimulators, mostly in Japan. After receiving his degree, heworked as a hardware design engineer in data communicationsand computer peripherals, and later he worked for RCA Camdenon a distributed computing system. He joined RCA Laboratoriesin 1980, where he developed a testing information system forVideoDisc research using a database and graphics on an IBMmainframe. He joined the Information Systems Planning andComputer Services Group in 1983, and now supports scientificcomputing for RCA Laboratories. His primary interests arecomputer graphics and the use of the computer as a tool for theresearcher. Don is a member of ACM and a cofounder andChairman of Grafuse.Contact him at:RCA LaboratoriesPrinceton, N.J.Tacnet: 226-2995

Barton: Using the Macintosh as an engineering workstation 69

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

Fig. 5. Printed -circuit board laid out using MacCad.

Gate Parasitic Capacitance

(Compute) ( Quit ) ( Help )

Overlap Capacitance= 3.450E-04

Edge Capacitance= 8.677E-05

Top Capacitance= 1.295E-05

0.2

0.5

d

0.071

o Capacitance per micronC) 3dimensional capacitance

Gate Width=

Fig. 6. Output of a program designed to calculate capacitivefringing capacitance.

A second example of Dr. Amantea's work is a graphicalprocess editor under development. In Fig. 7, the program givesusers palettes for objects (process steps), and actions (erase,select, etc.). With the mouse, users can place process step iconsin order in the process window. If they double-click on a stepicon, the program displays a window showing the defaultparameters for the step. Users can edit these with the mouse.

Figure 8 shows a parameter window for a step that is

graphical. The user draws the "curve" by positioning the mouseat the first data point on the graph and pressing the button,repeating the process for each subsequent point, double-clicking

Ede Edit Wear Process

Sarre I 1300RevertClear 1200

1100

1000

900

T

e

mp

Temp 700Time 0

800

700

( Get Ambient )

a File Edit Blew Process terirort

HSE

LE !`F

FEND

CERT

c,IMP LANTSIFF USEPHO TO

DIFFt

I

I

WRF

Fig. 7. Screen from a custom-made graphical processeditor.

ElsE

SELECT

CIFENC

CWER CLEH

0 Time max

Fig. 8. Parameter window for a step in the graphical processeditor.

to set the last one. The program creates an EDIF (standardizedformat for design data) file, which users can send to a mainframerunning on CMS by means of the Kermit file transfer protocol.Users will then be able to select regions for simulation, capturingdata back into the Macintosh for manipulation and/or plottinglater, including the results in a report. Dr. Amantea has pro-grammed the same function on an Apollo workstation usingApollo's Dialog Manager. Since the Dialog Manager functionsat a more primitive level than the Macintosh's QuickDraw, theprogram is larger on the Apollo and is much simpler graphically.The Macintosh Resource Editor allows one to modify objectssuch as windows, dialog boxes, and icons directly with themouse, while the user must specify the parameters of suchobjects in a text file for the Apollo Dialog Manager. Expect apaper by Dr. Amantea on his programs.

SummaryWe believe that the Macintosh can serve as a welcome alternativeto traditional personal computers as an engineering workstation.Its user-friendly graphics interface (developed first for the Lisaat a cost of 200 man-years of programming), combined withinternal, publication -quality laser printing, make it a tool thatcan get many an engineering job done easily and quickly with


'Ink Replenishment System Connection Diagram

plArrrnnimControl

for TFD and P T array

circuit det cts when inkis no longe covering theimage plate-s ignaling the

replenistc cle.

SolenoidActuator

Refill

Air Line

Air Line

3 Vent

j Vent

10 psi.

11 Vent

Solenoid

it Line

1199Met!

ElastonmrMe eringPump

SolenoidA to torDispense

Vent

Vent

Ink

Valve*

?neut.,

Ink

Valve #2

Pneuma,L

LedsPhotoTransisto

Det ctorArray

of Ink tray

Note: During initial setup both valves Cartridgeare open and ink is fed through the Linertubing, using a special switch. During for

operation valves #1 and #2 are operated Ink

with XOR logic. Valve #2 is normally open, C'ntaine

50 psi.

Fig. 2. An example of documentation used for the transferof bar code technology from the Labs to Video Componentand Display Division.

m

H

30

25

20

15

10

Line Chart of mR/Hour

Observations

Fig. 3. A sample plot of experimental data using Multiplan,Chart, and Statview.

Dr. Roy Nelson of the Display Systems Materials and ProcessingResearch Laboratory uses a Macintosh with MacTenninal tocollect data from a porisimeter, an instrument that measuresparticle sizes by mercury intrusion (this is the only installation inNorth America with such a device). The instrument is useful formeasuring the size distribution of fine particles in the dry state,and can measure particle size and aggregate size. After hecaptures the data with MacTenninal (the porisimeter thinks it issending data to a serial printer), he pastes it into Multiplan for asignificant amount of manipulation and then to Chart for plotting(see Fig. 4). He expects to replace Multiplan and Chart withExcel for this purpose. He also uses StatView for statisticalmanipulation, MacDraw for data presentations and model work,and Mac Write augmented with MacSpell+ for word processing.He says that since he acquired his Macintosh, his output of

100

90

80

70

60

Percent 50

40

30

20

10

0

0 5 10 15 20

Particle sizeFig. 4. Statistical data in graph form. The data was capturedwith MacTerminal, pasted into Multiplan for manipulation,and plotted by Chart.

25 30

technical papers has climbed dramatically (he formerly did hiscomputation on a mainframe).

John T. Zurich of the Television Research Laboratory usesMacCad (Vamp, Inc.) to lay out non -dense printed circuits. Heoperates the package as if he were taping the board on a lighttable, but instead of tape he uses the mouse, instead of the table,the Macintosh screen. While the operation is manual, the ease ofchange makes it easy to use for small prototype boards. Vamphas announced additional packages for autorouting, schematiccapture, surface -mount technology, and circuit simulation. Figure5 is the only case in which we have been forced to manually cutand paste, since MacCad is designed to produce circuit -boardlayouts and not figures in papers.

Dr. Robert Amantea of the Integrated Circuit Research Labora-tory is one of the few people, if not the only person, at RCALaboratories developing custom programs for the Macintosh. Heuses a compiled Pascal. He is interested in using the Macintoshfor:

1. Development of programs that are not CPU -or I/O -intensive,and that can benefit from a Mac -like user interface.

2. A user-friendly frontend to batch programs.3. A terminal connected to a more expensive workstation.

He also uses the Macintosh in the more usual way, for wordprocessing, diagraming, etc.

A first example of Dr. Amantea's work is a program designedto calculate capacitive fringing in an MOS transistor. By filling inthe blanks for a textbook -like example diagram, the user cancalculate fringing capacitance (see Fig. 6). Help dialogue boxesexplain such things as the theory underlying the calculations.This is an example of taking advantage of the Macintosh's userinterface facilities to create a program tailored to the problemand that speaks in the Mac's language.This program makes theMacintosh appear to be a custom calculator designed to solvejust this type of problem.


enough for ram disk operation in a 512k Macintosh. MacMemoryDisk lists at only $29.95.

Switcher (Apple Computer, Inc.) allows you to partition randomaccess memory to run more than one program at a time,switching between them with a mouse click. It is especiallyconvenient for repeated copying and pasting operations betweendifferent kinds of documents (terminal emulation and wordprocessing or spreadsheet, for instance), or for programs designedto be used together ( VersaTerm and TekPrint, for instance). Itallows what appears to the user as a limited amount of multi-tasking.

Dubl-Click (Dubl-Click Software) is a calculator constructionkit that allows you to create your own calculator desk accessory(desk accessories are available from within Macintosh programs),with choices from a wide variety of mathematical and simplestatistical functions (as well as business math functions). Youselect the keys you want and position them as you want (as ifyou were drawing the calculator). The tape can go to the printeror to a disk file if you want. Dr. Robert Amantea has evaluatedthis product and found that it provides functions missing fromthe Macintosh's built-in calculator.

MacSpell+ (Creighton Development, Inc.) is a spelling checkerthat installs in the menu of a word processor (Mac Write orWord) as a desk accessory (so that you can check spelling in themiddle of writing a document), and features a thesaurus, wordhyphenation, and the ability to add to the dictionary. It searchesfor the right spelling phonetically. We used it to spell -check thispaper. Spelling errors are now a thing of the passed.

LanguagesAt this point, most of the major programming languages areavailable for the Macintosh. Two are in use at RCA Laboratories.A number of users have used Microsoft BASIC to port simpleprograms from other machines to the Macintosh, and all havereported success. One user uses Softech Microsystems Pascal toprogram engineering applications in such a way as to takeadvantage of the user interface. While the results can be remark-able, you should be warned that for the latter, learning time toget started can be months. The Macintosh user interface includesabout 480 ROM calls. We have, and are evaluating MicrosoftFORTRAN, Aztec C, ExperLisp, and ExperOps-5 (this rule -based expert systems tool is really somewhere between anapplications package and a language). Our natural inclination isto suggest that before you write anything for the Macintosh, youshould see if there is an applications package that will do the jobfor you. (This is wise for any computer, but doubly so for theMacintosh). If you want to develop a specialized applicationwhose user interface is of primary importance, the Macintoshis a fine tool, but be prepared to commit a considerable amountof effort to learning how to use it.

Macintosh usersDr. Alvin Goodman of the Integrated Circuit Research Labor-atory uses the Macintosh in his research on MOSFETs (MetalOxide Silicon Field -Effect Transistors) and COMFETs (Comple-mentary Field -Effect Transistors): Mac Write for word processing,MacDraw for viewgraphs, Multiplan and Chart to calculate

statistics (mean and standard deviation, for instance) and plotexperimental data.

Dr. Marvin Leedom, Director of the Manufacturing SystemsResearch Laboratory uses MacDraw for presentations, etc. Hehas also programmed the RCA salary guidelines into Excel anduses it to track the salaries of approximately 150 people. Hehad previously done this on an IBM PC using Lotus 1-2-3, andon the Macintosh using Jazz. The changes in spreadsheet usedhave provided faster programming and execution time as wellas much greater ease of use. He also tracks capital requirementsand manpower planning for Manufacturing and MaterialsResearch.

Walter Marz of the Advanced Video Systems Research Labora-tory has been investigating the possibilities of using the Macintoshas a workstation for the electrical engineer, concentrating ondocumentation (he field tested MacDraft), and on CAD/CAE.He has used and recommends the printed circuit layout packageMacCad as well as the schematic capture package ParaGenesis(Advanced Engineering Solutions). He plans to link schematiccapture by the Macintosh into the more powerful Cadnetixworkstations. Expect a paper from him on the results of hisinvestigations.

L. Dimarco of the Optoelectronics Research Laboratory usesthe Macintosh to collect data from a die attachment evaluator,which measures thermal resistance characteristics of laser diodemounts. He modified a BASIC program written for the HP -85A to run on the Macintosh, collecting the data over an RS -232c data communications link. He (via theclipboard) on to the Microsoft Excel spreadsheet for manip-ulation and graphing. He also uses the Macintosh to producereports.

Kenneth Hang of the Optical Systems and Display MaterialsResearch Laboratory uses Mac Write, MacPaint, and MacDraw.Figure 2 is an example from transferring bar code technology tothe VCDD for processes developed at RCA Laboratories. Hefinds himself updating such documentation frequently, and theMacintosh makes this easy. He also does research in otherareas, including thick film multi -layer copper circuits. He findsthat the Macintosh can speed his preparation for meetings tothe point that it is faster to use it than to delegate the job. He isusing MacPascal to learn programming.

Dietrich Meyerhofer of the Materials and Processing ResearchLaboratory uses the Macintosh in his research in resists formicrolithography. He uses Mac Terminal and Tekalike to makethe Macintosh emulate text and graphics terminals for his CMSaccount, where he is able to run mainframe simulations, viewingtext and graphical output on his Macintosh screen. He also uses itfor word processing for report preparation, printing the reportson the LaserWriter. He has transferred CP/M BASIC programsfrom an older Apple II into the Macintosh.

Dr. James R. Matey of the Materials and Processing ResearchLaboratory uses Mac Write for Purchase Order templates, letters,and patent disclosures. He uses MacDraw to sketch experimentalapparatus, and for viewgraph presentations. He uses Mac Terminaland VersaTerm to communicate with other computers. He usesMultiplan, Chart and Statview (see Fig. 3) to plot experimentaldata. He also uses Kermit to exchange files with an IBM PC AT.


We recommend Ready.SeL Go as an inexpensive way to over-come the layout limitations of Mac Write. Carol Irwin of Infor-mation Systems Planning and Computer Services at RCA Lab-oratories uses it to lay out a monthly newsletter. It does notfollow the Apple interface guidelines as exactly as some otherprograms do, but once you master moving and resizing blockswithout the Apple top bars and resizing boxes, it is easy to use.You can move and size blocks with the mouse or type actualpositions and sizes (to 1/10,000th inch) into a dialog window.Ready.SeL Go's major limitation is speed (you must leave it andenter other programs, such as Mac Write, to copy information tobe pasted). The latest version can import text files directly, butnot graphic ones. Switcher (see Utilities) could provide somehelp, provided the memory is available to run all those programssimultaneously. Starting and stopping programs does take time.Ready. SeL Go should be fine for the small newsletter or occa-sional technical report.

PageMaker (Aldus Corp.), a layout program costing severaltimes as much as Ready.SeLGo, provides comprehensive facilitiesfor laying out documents. I used PageMaker to lay out the draftof this paper, and any deficiencies were a result of humanlimitations and the wish to conform to technical report formatrequirements. From within PageMaker you can bring in textand graphics from external files and place them arbitrarily (thisis a speed feature as much as a convenience one). You cancreate master pages for numbering, running heads, etc. You cancrop graphics, edit text with full control over attributes, etc. Itconforms well to the Apple interface guidelines.

Spreadsheets /graphicsExcel (Microsoft Corp.) includes the features of Multiplan andChart, and builds on them, adding the ability to format thespreadsheet for printing or inclusion in another (bold, etc.),database functions to sort, select, and report material stored inthe spreadsheet, to graph directly from the spreadsheet, and tocreate macros. While you create macros on a conventionalpersonal computer by typing in command sequences, in Excelyou turn on a macro recorder and then perform the actions.These are recorded, and placed in a window of macro commands(which you may edit if you want to learn the commandlanguage), and you name the macro. Any time you select thenamed macro (with the mouse) for execution, Excel will repeatthe steps involved. It is as if by simply turning on and off acassette recorder, you could request a computer to watch whatyou did and to write a program to repeat it. We recommendExcel for sophisticated spreadsheet users. Much of what engineersused to program in BASIC or FORTRAN (with graphicsenhancements) can now be done with Excel.

Integrated office

Jazz (Lotus Development Corp.) is a program combining textterminal emulation, spreadsheet, database (including forms andmail merge), graphics, and word processing. Since it is a singleprogram, moving from one function to another is fast and easy.It does as good a job of adhering to the Apple interfaceguidelines as any program we have seen, and as a result, isexceptionally easy to use, even more so than most Macintoshprograms. Its limitations are a result of having to cram all thosefunctions into a single program capable of running on a 512kMacintosh. It does not have macros. You cannot create plots

with logarithmic axes (you could take logarithms of the datafirst using the spreadsheet, but it is not the same thing). It haslimited capabilities for annotating plots (which could, of course,be pasted into MacDraw). It does have more word processingpower than Mac Write. For someone whose requirements donot exceed its capabilities, it is a remarkably easy way to makeuse of a Macintosh. I captured the information for the Grafuse(RCA Corporate Graphics Users' Group) mailing list (over 250people) from a mainframe database (terminal emulation), createda Jazz database, loaded it, created a form for updating informa-tion (database), and created mailing labels (database and wordprocessing) in half a day, all with no previous experience withJazz's database functions.

Outline programs

ThinkTank 512k (Living Videotext, Inc.) is an outline programthat allows you to create an outline, indenting and un-indentingwith the mouse, expanding to view information at lower levels,contracting to hide it. Moving a contracted topic moves indentedinformation under it. At first, ThinkTank 512k might seempleasant, but perhaps not worth its cost. It can, however, storean entire Mac Write or MacDraw document, for instance, as atopic. With those capabilities, it starts to look more like an adhoc mixed text and graphics data management system. You canprint a table of contents created by ThinkTank 512k and/orthe actual complete contents themselves. You can create docu-ments of this sort in Mac Write (mixed text and graphics,outlined and indented), but you cannot collapse them to examineonly high-level information. ThinkTank 512k can be of use inorganizing projects.

Statistics

StatView (BrainPower Inc.) is a statistics program with spread-sheet -like tables for data to be analyzed, and graphical as well astabular output. In late 1985, Joseph Blanc and Dr. James R.Matey were kind enough to compare a number of statisticspackages for the Macintosh, and StatView proved to be the firstchoice. It can calculate the usual descriptive statistics and tests,one and two-way analysis of variance, nonparametric statisticsand regressions. For diagnostics, it will plot residuals as well asthe regression model. Tables may be pasted from terminalemulators, spreadsheets and databases, and graphic output maybe copied and pasted to MacDraw. We recommend StatView foranalyzing experimental data in a Macintosh.

Utilities

MacMemory Disk (Assimilation, Inc.) allows you to createwhat appears to be a disk in random access memory. RAM canbe accessed much faster than any disk drive, so programs storedin ram disk can run faster. Whatever is stored in the ram diskdisappears when the computer is turned off (or suffers a powerinterruption), and data must be copied to a real disk to preserveit, but you can control when disk updates take place rather thanhaving the computer do it while you are kept waiting. Mac -Terminal, MacDraw, Mac Write and Chart are examples ofprograms that run much faster in ram disk. RAM dedicated toram disk is not available for program execution, and programsthat use much of the 512k RAM (Jazz, for example), don't leave


provides easy to use tabs, it will not allow multiple columns oftext, as in a newspaper. Along with this, although it will allowyou to paste a MacDraw or MacPaint graphic in the middle oftext (how many word processors do you know of that permitintegration of text and graphics?), it will not allow you to placetext alongside the figure (only above and below it). It does, ofcourse, allow you to select blocks of text with the mouse forcutting and pasting, allow multiple fonts with attributes, andallow searches and replacements of whole or partial words withor without case sensitivity. Many Macintosh users will need toovercome Mac Write 's limitations, either through the use of amore powerful word processor or by means of a page layoutprogram. We have taken the latter approach with this paper.

Word (Microsoft Corp.) is the first word processor providingfeatures omitted from Mac Write. It provides more keyboard -oriented commands. It lets you print closer to the margins thanMac Write. It will allow multiple columns of text, as in anewspaper. Along with this, it will not allow you to place textalongside a figure. For the accomplished typist with a heavyword processing load, concerned with speed, this is probablythe solution of choice to MacWrite's limitations. You could alsoadd a page layout program for even greater capability.

Spreadsheets

Multiplan (Microsoft Corp.) was the first spreadsheet availablefor the Macintosh. Users of spreadsheets on other computersmight be surprised by its ability to adjust the width of a columnby grabbing the dividing line with the mouse and dragging it tothe desired location (who wants to count characters: I want itthat wide!). Many of the things that engineers used to do withBASIC or FORTRAN programs can now be done more pro-ductively with a spreadsheet! You can construct formulae withtranscendental functions, conditions, etc. You can paste informa-tion captured from a remote computer as a table by a terminalemulator into Multiplan. For our purposes, Multiplan has beenreplaced (along with Chart) by Microsoft's Excel (which combinestheir functions and adds more in one package).

Databases

OverVUE(ProVUE Development Corp.) is a database programfor the Macintosh. You can paste in tables or text from aterminal emulator or a spreadsheet, for instance (capturinginformation from other computers, when it is available, ratherthan typing it in). Known for its speed and ease of learning,OverVUE lets you create tabular and graphical reports.

Helix (Odesta Corp.) is a relational database program for theMacintosh. You can paste tables, text, objects or bit -maps intoHelix. A graphic or even a Mac Write document can be a datafield to be retrieved by a query. (Consider a relational databaseof plots of experimental data, which lets you retrieve all the runsmade before lunch with a query, or of integrated circuit basediagrams, which lets you retrieve the pin -outs of all the 16kdynamic RAM chips with an access time less than 50 nanosecondswith a query). Helix uses icons for everything, including queries(multi-user Helix has been announced). Helix appears to requiremore learning than OverVue, but to provide more features.

Graphics

Chart (Microsoft Corp.) was the first graphics package availablefor the Macintosh. It plots tables (typed in or brought in fromspreadsheets, terminal emulators, databases, etc.) as line, scatter,bar, etc. It provides logarithmic axes. For our purposes, it hasbeen replaced (along with Multiplan) by Microsoft's Excel.

Drafting programs

The drafting programs we describe below have the familiarMacintosh interface.

MacDraw (Apple Computer, Inc.) was the first drafting programsupplied for the Macintosh. You can draw lines (with or withoutarrowheads), boxes (with or without rounded corners), circlesand ellipses, arcs, polygons and free -form figures, all with select-able line widths, line patterns, and (where appropriate) fillpatterns. You can determined dimensions of objects to within ahundredth of an inch. The maximum multi -page drawing size is48X96 inches. MacDraw provides a complete variety of textplacement and insertion. With the exception of the icons included(distinguished by their bit -mapped jagged appearance), I couldhave produced this entire paper in MacDraw. Since it producesobjects that are represented internally as a line of width w andpattern p between points (x0, yO) and (xl, yl), for example, thequality of output produced is limited only by the hardware (300dots per inch for the Apple LaserWriter, 2500+ dots per inch foran Allied Linotype phototypesetter). I produced the figures forthis paper using MacDraw, electronically pasting in bit -mapsfrom MacPaint for icons and screen examples.

MacDraft (Innovative Data Design, Inc.) is similar to MacDraw,but provides additional features such as automatic dimensioninglines, the ability to rotate objects at 1° intervals (MacDraw canonly rotate objects at 90° intervals), and others aimed at makingit a more professional two-dimensional drafting program. Ourusers have reported problems using MacDraft with Switcher (seeUtilities), and MacTutor magazine (November 1985) reportedproblems using MacDraft with PageMaker.

Painting

The painting program we describe below has the familiarMacintosh interface.

MacPaint (Apple Computer, Inc.) was the first painting programwritten for the Macintosh. While it can draw lines, boxes, etc.,the 72 dots per inch bit -map limits the quality severely if you areprinting on a LaserWriter. We recommend MacPaint only forthose tasks not possible with MacDraw (editing saved examplescreens, creating bit -map objects, spray painting, etc.), at whichpoint we recommend copying and pasting those bit -maps intoMacDraw for the addition of objects created in MacDraw.

Page layout

The layout programs we describe below can place text andgraphics in arbitrary positions on a page using the mouse.

Ready.Set.Go (Manhattan Graphics Corp.) is a page layoutprogram for the Macintosh. You can copy blocks of text andgraphics into Ready.Set.Go. You can save page layouts (asopposed to the contents) so that you can establish a given typeof format for a document, and repeat the look page after page.


ulation with programs such as MacWrite as well as tables fortransfer to Macintosh spreadsheet, database, or graphics pro-grams. We have sent text to other computers using MacTerminalbut not to the IBM mainframe. MacTerminal provides for filetransfer to and using the Xmodem protocol, and from othercomputers in ASCII (American Standard Code for InformationInterchange) text (files can be pasted in text to other computers,but only a clipboard -full at a time. Its latest version allows theBACKSPACE key to be programmed to be a DELETE key (amenu selection), a convenience for DEC users.

Tekalike (Mesa Graphics) is the earliest graphics terminalemulator we used on the Macintosh. It emulates the Tektronix4010, 4012, 4014 and 4016 graphics terminals (a later versionwe have not yet tested adds VT -100 and VT -640). It alsoprovides the ability to pan and zoom on areas selected by themouse and to save plots for MacPaint on MacDraw for full 300dots -per -inch LaserWriter resolution. It was this program thatfirst enabled us to consider the Macintosh as a replacement forgraphics terminals. The Macintosh could not only view plots atits 512X342 resolution, but thanks to the pan and zoom, couldview parts of plots at full 4014 (4096X3072) resolution, afeature then available only on much more expensive graphicsterminals (some other personal computers now have programsfor Tektronix 4010/4014 emulation, but the best known to meat this time zooms in 4X and only provides 1024X768 resolutionat zoom; 4010, not 4014 resolution). The plots could also, ofcourse, be printed and stored in on a Macintosh disk for laterviewing, avoiding the cost of recreating them. Tekalike alsoenables emulation of the Tektronix 4662 8 -pen color plotter,connecting to low-cost Apple, Hewlett-Packard, and HoustonInstrument plotters for color output (plotting graphics stored ondisk, enabling you to have panned and zoomed plots). Theability to paste into MacPaint on MacDraw enables you toinclude all sorts of plots from other computers into technicalreports such as this one, and to edit the plots.

VersaTerm (Peripherals Computers and Supplies, Inc.) providestext (more complete DEC VT -100 emulation than Tekalike,DG Dasher) and graphics (Tektronix 4014) emulation. It allowsthe BACKSPACE key to be reprogrammed to be a DELETEkey (a menu selection), a convenience for DEC users. It alsoallows the mouse -controlled cursor to transmit appropriate se-quences to the remote for either EMA CS or vi editors. It allowswhatever appears on the screen to be simultaneously directedinto a disk file or to the printer, convenient for capturing longpassages (text or graphics) from a remote computer. Screensmay be printed at any time. Its Tektronix 4014 emulationprovides simulation of the four sizes of 4014 hardware characters.Remote computer -issued commands to print the Tektronix 4014screen do so (on either ImageWriter, the Apple dot-matrixprinter, or LaserWriter). It provides file transfer as plain ASCIItext, Xmodem and Kermit (the public -domain file transfer pro-tocol provided on most RCA Laboratories computers). Kermitis available for the Macintosh as a separate public -domainprogram, but there are obvious advantages to having it builtinto your terminal emulation program. A companion programTekPrint (also from Peripherals Computers and Supplies, Inc.)displays plots saved by VersaTerm and provides panning andzooming, rotation (vertical or horizonal page), direct printing(ImageWriter or LaserWriter), and saving (MacPaint). Laser-

Start of waveform, magnified

Writer printing, either from TekPrint or MacDraw, provides afull 300 dots -per -inch, much finer quality than that provided byan actual 4014 with attached copier. We captured the followingwave form plot from Issco's Tellagraf running under CMS on aCISS mainframe (data courtesy of Dr. James Matey of RCALaboratories), passed it to TekPrint, saved it for MacDraw,where we annotated it, zoomed in with the mouse, saved itagain, annotating in MacDraw, naturally, we could annotatemore, adding arrows to points of interest, with boxed ex-planations, etc.

Word processors

The word processors we describe below have the familiarMacintosh interface.

Mac Write (Apple Computer, Inc.) was the first word processorwritten for the Macintosh (This paragraph was written withMac Write). Mac Write is limited in its capabilities, but not in itsease of use. It appears that Apple concluded they had made amistake in providing a comprehensive selection of integratedsoftware for the Lisa (virtually no one wrote software for theLisa, since the major functions appeared in the original release ofthe Lisa Office System). They did not repeat the mistake with theMacintosh-features apparently omitted from Apple packagessuch as Mac Write, MacPaint, etc. appeared in third -party pack-ages. Mac Write permits both mouse -oriented operation andkeyboard -oriented operation (COMMAND -b = bold, etc.).More keyboard operation could speed things up for accomplishedtypists oriented toward more traditional word processors. Mac-

Write refuses to let you print farther to the left than the 1 -inchmark or farther to the right than the 8 -inch mark, even if you areprinting a horizonal page rather than a vertical one. While it

Barton: Using the Macintosh as an engineering workstation63

PostScript-words and pictures

When Apple Computer introduced the Lisa andMacintosh computers, they not only were sellingnew computing hardware, they were selling aradically different environment in which text andgraphics could be created and edited via agraphical icon -based user interface. With all thisemphasis placed on graphics, Apple needed a wayof generating high -quality printed output. One yearafter the introduction of the Macintosh computer,the Apple LaserWriter was introduced.

Apple was one of the first companies to adopt agraphic output protocol named PostScript (AdobeSystems, Inc.). PostScript is a programminglanguage that generates text and graphics on aprinted page. Its programs only work on"PostScript -equipped" printers.

PostScript is a complete programming language.In addition to text and graphic operators, thelanguage offers mathematical functions, stringmanipulation, and file I/O operations. PostScript issimilar to the FORTH language in that it uses thenotion of a "stack" to pass arguments and data. The"stack" is a reserved portion of memory whereitems are placed and manipulated. Hewlett-Packardcalculators operate on a similar principal.

PostScript offers many data types such asintegers, reals numbers, arrays, and strings."Objects" such as dictionaries can also be createdand manipulated on the stack.

PostScript programs are readable 7 -bit ASCII textfiles. Usually, these programs are generated by agiven application program running on a computer,such as Mac Write on the Macintosh. However, mostPostScript -equipped printers can operate in a modewhere PostScript commands are fed to the printerinteractively by any host that can create ASCII files.

Variables and procedures can be defined andplaced into dictionaries. For example, the followingprocedure would define and create a rectangle:

/RECTANGLE10 0 rlineto0 5 rlineto-10 0 rlinetoclosepath def

5 units t/ 10 units X

current XY location

This procedure will generate a rectangle that is 10default units in X and 5 default units in Y.

The operator "rlineto" is short for relative line to,indicating that the rectangle will be placed relativeto the current X,Y location. The "def" operatorplaces the procedure into a dictionary so that futureinstances of rectangle can be created by simplycalling the procedure name.

Much of PostScript's power is in its ability tohandle textual operations. Text is denoted by anysequence of characters enclosed in parentheses.The following program creates the text string "Text

Sample" in 24 -point Helvetica typeface, 50 defaultunits up and over on the page.

/Helvetica findfont24 scalefontsetfont50 50 moveto(Text Sample) showshowpage

Text SampleFrequently, other systems store type fonts in bit -map form. A separate bit map is required for everycharacter in every size. Because different printershave varying degrees of resolution, bit -mapencoding of characters is usually optimized for aspecific output device.

In place of bit -map encoding, PostScript "builds"a one -point typefont "template" from a set ofmathematical instructions stored in the printer'smemory. When a program invokes a particular fontand size, PostScript scales the original one -pointtemplate and then stores this scaled image in bit -map form in cache memory. This technique allowsthe resolution to be independent of the point size.Memory resource use is kept to a minimum as wellbecause only a single template is required for allsizes.

At the present time, the majority of PostScript -equipped printers are 300 -dot -per -inch (dpi) laserprinters. Printers that offer resolution between 200and 1000 dpi are considered "near -typeset" quality.However, for many printing tasks this resolution ismore than adequate. For those jobs requiring higherresolution there are PostScript typesetters with 2540dpi resolution.

Manufacturers supporting the PostScript protocolare Apple Computer, Inc., QMS, Inc., DataproductsCorporation, Prime Computer, and the AlliedLinotype Company. Recently, Digital EquipmentCorporation announced that they would be usingthe PostScript protocol for certain models of theirlaser printer line.

PostScript has already had an significant impacton the printing industry. One of the new "buzz-words" in the computing industry is "desktoppublishing." This whble new industry is due mostlyto the availablity of low-cost laser printing and thesuccess of PostScript.

The promise of PostScript is that unlike otherproprietary graphic protocols, PostScript is trulydevice -independent.

-James K. AndersMissile and Surface Radar Division


Word processing

Word processing programs for the Macintosh create text filesand can accept text, objects, or bit -maps from the clipboard.They can thus include graphics in a text document, although theywill not place text alongside a figure (only above or below it).They allow you to change fonts and their attributes and displaythe effect on the Macintosh's bit -mapped screen.

Text

Tables

:0::

Objects

word processing

Text

Tables==L,

:

:0::

Objects

d,Bitmap Bitmap

Layout

Page layout programs for the Macintosh permit the arbitraryplacement of graphics and text on a page. They permit multiplecolumns, text alongside figures, etc. This is true, completeintegration of text and graphics. It is generally possible to storepage formats as templates, thus creating standard documenttypes.

Text

Tables

: 401:

0:

Objects

Bitmap

layout

Text

Tables

0..Objects

Bitmap

Specific programsThe table on the next page shows Macintosh programs we havetried and found useful. It does not include languages andutilities, described in later sections, but shows applications thatcommunicate with each other well and are useful for getting ajob done. It shows the kinds of inputs and outputs possible.Arrows pointing to the left show inputs to the program, whileones to the right show outputs. Also shown are generic functions(as covered previously) supported by the particular program.

Terminal emulators

All of the terminal emulators we describe here have the familiarMacintosh interface with radio buttons, and check boxes toselect communications settings such as baud rate, parity, etc.We have successfully used them with our Ungermann-Basslocal -area network, and with modems.

MacTertninal (Apple Computer, Inc.) is the earliest text terminalemulator we used on the Macintosh. Since it emulates the DEC(Digital Equipment Corp.) VT -100 terminal, we have found itcompatible with our Decmail electronic mail system. It workssuccessfully with the DEC full screen applications we havetried, as well as with line mode applications on a variety ofcomputers. We have also used it successfully with our Data -stream protocol converters for full screen operation on an RCACISS (Computer Information Systems and Services) IBM main-frame running the CMS operating system in Cherry Hill. Wecan capture text from other computers for Macintosh manip-

Table I. Communications among Macintosh software pack-ages.

pi Fri] 1-n 0Ca!Pow,/ Remote

labRemotegraph.. Tear Tables Objects earflap Fucc.ir,Ptogrem

terminal emulation

MacTerminal 4-, 4-, 4-, text DEC VT 1 OCI terminal

Tekalike .4-, 4- --4. -111'text ,DEC VT.100) & graph.t Tek 4010'14) terminal

Versaterm 4-4. 4_ .4_0. .4_, _f. text MEC VT -100. DG 0200) &graphics iTek 4010/14) Heroinel

__4.

word orccessing

MacWrite 4-1 4-1 4- 4- word processing

Word 4-, 4-1 .4- 4- word processing

spreadsheet

MufhplanI I

I 4-0. II

I soreadsheet

databaseOverVue 4-1 4-1, -, -1 deteoase & grata,.

Helix 4-, 4-, 4-, 4-11. database & graph.

graph.,

Chan1 I

1 411-1110. I .4-11. 1 -0.1 -110.1 grata..s r g i , c

MacDraw 4-, 4-1, 4- drafting

MecDratl 4-, 4-1 4-_ drafting

I...MIN

MacPamt II I 4- I 1 .4-- I 4-, I painting

!amts

PageMaker .4- 4- 4- page layout

Ready.SelGo 4-- 4- 4- page layout

Sbreadst'set ' a, once

ExcelI I I

I 44. 1 -4.1 -0. I spreadsheet. graohbs, react,

^trictred c....C5

Jazz 4- 4- 4-1, -1111. -- text terminal. spreadsheet.database. graphics, wordprocessing

2,,,2.,

Thinktank 512k II

I 4-, 1 41-1, I 4-0-I 4-1 'outline, including tab & graphic-I&IistrhStareiew

I Ij 41-410. 1 .4-111. I -III. i -III. Isoreadsneergraonics statistic'


Text

Tables

database:0::

Objects

Bitmap

Text

Tables

: co::0::

Objects

A:1

Bitmap

SpreadsheetsSpreadsheet programs allow you to manipulate tables of infor-mation, calculating parts from other parts. A surprising amountof technical computation can be programmed into spreadsheets.

Tables

spreadsheet

graphics

Tables

=

Text

ITables

ElText

I.....Tables

:0::Objects

0 AN

Bitmap

GraphicsGraphics programs make plots from tables of information.

PaintingPaint programs create bit -maps and can accept text, objects, orbit -maps from the clipboard (which they then turn into bit -maps). They are extraordinarily flexible in manipulating graphicsinformation, but very limited in the amount of detail (72 dots perinch) that they can display.

Text

AN

Bitmap

Drafting

Drafting programs create objects and can accept text, objects, orbit -maps from the clipboard (which they then turn into objects).They are less flexible in manipulating graphics information, butlimited only by hardware in the amount of detail (300 dots perinch on the LaserWriter) that they can display. To eliminate partof an object (you cannot erase part of an object in a draftingprogram), you cover the part with an appropriately -shaped whiteobject with a white border.

Text.t,

: !b::0::

Objects

1/A

Bitmap

drafting

Text

.b

:0::Objects

III

AM

Bitmap


Text, tables, objects and bit -maps

There are four primary types of information passed around andmanipulated within a Macintosh:

1. Text: letters, numbers, spaces, punctuation, carriage returns;may include attributes-bold, underline, italics, outline,shadow, superscript, subscript, font, size; attributes may bepassed from one program to another or may not; text isone-dimensional-first character, next character, . . . , lastcharacter.

2. Tables: two-dimensional (rows and columns) of text (usuallynumbers and labels).

3. Objects: straight lines, circles and ellipses, rectangles, polygons,free -form lines, arcs, etc., with attributes including width andpattern for boundary, pattern for interior, and so on. Someprograms may stretch, squash, distort, or rotate. An object isan entity, so part of it may not be altered or erased separately.

4. Bit -maps: patterns of dots (at 72 dots per inch in the Macintosh)to produce the appearance of objects. Some programs(MacPaint, for instance) may manipulate individual dots.

Clipboard, notepad, scrapbookThere are three primary paths for passing information aroundwithin a Macintosh:

1. The clipboard: you cut or copy to and paste from theclipboard; it may contain text, tables, objects, or bit -maps; ifthey fit, the operating system keeps them in random accessmemory (RAM), otherwise it keeps them in a clipboard fileon disk; each copy or cut replaces the previous contents of theclipboard.

2. The notepad: available at any time, even from within aprogram; will accept up to eight pages of pasted or typed text;the operating system keeps the notepad contents in a disk file;also useful for jotting down notes.

3. The scrapbook: available at any time, even from within aprogram; will accept pasted objects, bit -maps, and text; theoperating system keeps its contents in a disk file.

Software integration

Figure 1 shows the most commonly used paths among differenttypes of application software. Note that some programs includemultiple functions (Lotus Jazz includes terminal emulation,word processing, spreadsheet, database, and graphing, for in-stance), and others are generically equivalent to one or more ofthese functions (StatView uses spreadsheet and graphics functionsto provide statistical computations and graphs).

The following diagrams each show an application in themiddle, the types of inputs common on the left, and the types ofoutputs common on the right. Except for terminal emulation, theinputs on the left can be pasted in from other programs, and theoutputs on the right copied to other programs.

Terminal emulationTerminal emulation allows the Macintosh to appear to be a textand/or graphics terminal. Although this allows a Macintosh tobe used in place of a terminal, its real value lies in its ability tocapture information from other computers (and paste it to otherprograms) without entering it into the Mac by hand. Terminal

RemoteComputer

painting

terminalemulation

graphics

,database

layout

word processingspreadsheet

Fig. 1. The most commonly used paths among differenttypes of application software.

RemoteGraphics

terminal emulation

RemoteText

RemoteText

Text

ITables

=Th.

:0::Objects

!i

0 Am

Bitmap

emulators may allow text and tables to be sent to a remotecomputer, depending upon the emulator and the computer.

Databases

Database programs allow you to store information, retrievingparts of it based on some selection criteria. The primary methodof storage is the table, although some databases will even storepictures.


D.P. Barton

Using the Macintosh asan engineering workstation

It may not be all things to all people, but at RCA Laboratoriesthe Macintosh is an Apple with quite a shine.

Since November 1983, when the first Apple Lisas arrived atRCA Laboratories, Information Systems Planning and Com-

puter Services has had an ongoing evaluation of the window andmouse technology for personal workstations. After the intro-duction of the less expensive and more popular Macintosh, focusshifted in its direction. For some time we have been recom-mending the Macintosh as a low -end graphics terminal. This

paper describes the Macintosh, its hardware, software, some ofits applications at RCA Laboratories, and additional possibilities

for taking advantage of its capabilities.

Hardware featuresThe Macintosh has a number of hardware features not commonamong microcomputers in its price class:

Abstract: Before automation, engineers and scientistsdescribed their work graphically, stressing the use of diagrams,graphs and formulae (with these considered the "meat" of both

technical reports and presentations). While they would include

text (spoken or written) for guidance, they used blackboardsand drawing boards far more than typewriters. Engineeringworkstations with high -resolution, bit -mapped displays, graphic

input devices and direct manipulation software automated manyengineering tasks, but at a cost too high to provide every engi-neer with one. When Apple introduced the Macintosh, manyengineers and scientists noticed the presence of workstationattributes, as well as consistent, friendly user interfaces and easeof integration of programs, all this at prices comparable with

graphics terminals. This article describes examples of howengineers and scientists at RCA Laboratories use Macintoshesin their work, as well as how they might use them.

'01986 RCA Corporation.Final manuscript received February 3. 1986Reprint RE -31-2-11

1. A high -resolution (for microcomputers) 512X342 bit -mapped

screen;2. A graphic input device (the mouse);3. Operating system commands to allow applications programs

to communicate with both;4. A high -quality (300 dots per inch) black -and -white graphics

laser printer (LaserWriter), available separately.

Software featuresThe Macintosh software has a style unlike that of conventional

microcomputers.It is not command -driven (it does not display a prompt and

wait for you to tell it what to do).It is not menu -driven, although it has menus (it does not

display a list of choices and wait for you to tell it which you wantit to do).

It works by direct manipulation. You may select an object atany time by pointing at it (using the mouse to move an arrowaround the screen) and pressing the mouse button (if you did notwant to manipulate that object, point elsewhere and click thebutton). Any time an object is selected, you may point at one ofseveral action categories at the top of the screen and press the

mouse button to see a menu of actions for the category. If youkeep the button pressed, point at an action, and release thebutton, it will take that action against the object selected.

Since this type of operation is built into the machines, andApple has provided guidelines for how the man-machine inter-face should appear to the Macintosh user, applications writtenfor the Macintosh tend to be quite consistent, and as a result

require less time to learn. Common operations become second

nature, and the user can concentrate on the problem at handrather than the tool.

In addition, Apple has provided standard formats for developersto use for data, making it easy for data to be passed from oneprogram to another. They have also provided built-in facilitiesfor passing information around.


Lightyear SUBJECT:PROCESSOR EVALVERSION:7 JANUARY 1986

DETAIL EVALUATION: CRITERIA AND RULES SCORE WEIGHT

II AN/ZZZ-77 839 92011

II PROCESSOR BBB 737

SIZE

NUCL HARD.

SECURE OS

WEIGHT

OS LEVEL

ADA SUPPORT

7580

100100

100100

9090

7515

7575

AVAIL DEV SUPT 7575

S/W SUPPORT75 75175

MAX. KOPS

NUMBER OF PROC

LANGUAGE

OS LEVEL

MEMORY CAP.

I/O CHANNELS

SECURE OS PannedPassed

1518

100

100

1001

100

75

75

75 I

5011

50 5050

50 5011

50

50 5011

0

8 1008

11

1 1011

11I II

II

Fig. 6. The major difference between Processor BBB and the AN/ZZZ-77 is the OS level(criteria and rule). Processor BBB is ranked over 100 points lower based on this single factor.

Help is at hand

Whether you need a few facts or a literaturesearch, or you want to know how otherengineers hayed solved the same problemsyou face, your RCA Libraries are ready to help.

Ready access to statistical data or researcharticles is available online at many of RCA'slibraries. Online literature searching of hundredsof databases and access to the collections ofRCA, public, and academic libraries expandyour information gathering possibilities.

Check with the librarian or informationspecialist at your location for assistance. If youdon't have a library, contact TechnicalInformation Systems, Tacnet: 226-2410, forassistance and information.

IiIlllllll

00

irri-Trnrr rr

Kjellquist: Lightyear-an engineering analysis tool 57

Lightyear SUBJECT:PROCESSOR EVALVERSION:7 JANUARY 1986

DETAIL EVALUATION: CRITERIA AND RULES

11 PROCESSOR BBB

SCORE WEIGHT

7371 9201ill

SIZE

NUCL HARD.

- - 80

100

100

100

SECURE OS 100 100

WEIGHT 90 100__---

OS LEVEL 15 75

ADA SUPPORT 75 75

AVAIL DEV SUPT 75 75

S/W SUPPORT 75 75

MAX. MOPS 18 50

NUMBER OF PROC 50 50

LANGUAGE 50 50

OS LEVEL 0 50

MEMORY CAP. 8 10

I/O CHANNELS 1 101

I

SECURE OS ,Pammed

HARD PassedINULL ----------r----------1-- Jr--- -I

r r 1

Fig. 5. The detailed evaluation of any alternative can be displayed with a single keystroke.Processor BBB scored 737 out of a possible 920 points.

D.T. Kjellquist is Manager, Command SupportSystems at Communication and InformationSystems Division. Mr. Kjellquist joined RCA in1983, and has been technical director onseveral proposal and study efforts. Mostrecently, he has been program manager on theElectronic Warfare Coordination Module(EWCM) program. Mr. Kjellquist has beeninvolved in PC applications for several years,and is currently supporting a pilot project foroffice automation in CISD. He is a graduate ofthe University of Connecticut with BSEE andMSEE degrees.Contact him at:Communication and InformationSystems DivisionCamden, N.J.Tacnet: 222-6359

shines. Figure 1 shows the summaryoutput of an evaluation, but considernow that a detailed evaluation of Proc-essor BBB is wanted. In a single keystroke a detailed evaluation is displayed(Fig. 5). In this figure the horizontalbars represent the maximum score forany criterion. The darkened portionshows the score for this particular candi-date. A second question is often askedas to why a particular alternative wasranked higher than another. Lightyearaddresses this question by "locking" analternative and selecting a second for adetailed analysis. Any pair of alterna-tives can be analyzed in this way (Fig. 6).

But what if the score for a particularcriterion was changed to a new value-would that have changed that results?With a few simple key strokes any cri-terion for any alternative can bechanged and the resulting ranking canbe recalculated.

SummaryThe Lightyear decision support pro-gram provides a key analysis tool thatbelongs on the software shelf of anyengineering PC workstation. It providesa more timely and consistent approachto tradeoff analyses that are either notdone or involve tedious pencil and paperefforts. Though the example discussedaddresses an engineering application,Lightyear has proven value in manybusiness situations.

ReferencesI. Bulkeley, D., "The Deciding Factor," PC Prod-

ucts (December 1985).2. Hogan, T, "Are the Biggest Sellers the Best?" Busi-

ness Software (July 1985).3. Taylor, J., "Lightyear's Ahead of Paper," PC Mag-

azine (April 16, 1985).


Lightyear hardwarerequirements:

192k of random access memoryMS/DOS 2.00 or higherAt least one double -sided floppydisk driveA printer is desirable

is expressed in values from Al to ClThe verbal criteria are used for includ-ing this factor in the model. Lightyearcomes preconfigured with several verbalcriteria categories such as quality, yes/no, frequency, and risk. Each of thesecan be used to describe a particular cri-terion. In addition, a vocabulary may beeasily customized for such factors assecurity. In any of these cases theweights assigned to each verbal criterionare under control of the user and may berapidly tailored.

Once the weights are assigned, how-ever, the user of the model is unaware oftheir values. Only the appropriatevocabulary selections are shown duringan evaluation session. Figure 3 illus-trates the verbal criteria.

The third type of criterion uses agraphical representation in Lightyearand provides an effective method todeal with subject criteria that cannoteasily be expressed in words or withnumerical values. Often an evaluatorcan determine that one alternative isbetter than another, but has difficulty intranslating these feelings into concretevalues. For graphical criteria Lightyearpresents a bar with an "X" located onthe line. The evaluator simply has tomove the X along the line from mostdesirable to least. Lightyear automati-cally computes the relative position ofthe X, translates the position to a score,and computes a weighted result. Theexample analysis did not use any graph-ical criteria.

The ability to create and evaluaterules in a weighted -criteria model is thefinal step in the model creation, and oneof the most powerful and useful capa-bilities of Lightyear. In general, a rule isa statement of the relationship of thevalues of one or two criteria in a model.There are two types of rules that may bewritten for a Lightyear model; elimina-tion or weighted. The elimination rulesare those rules that represent pass/failcriteria for an alternative under consid-

Lightyear

CRITERIA: OS LEVEL

ALTERNATIVE WORD ALTERNATIVE

SUBJECT:PROCESSOR EVALVERSION:7 JANUARY 1986

SECURE OS

AlB3B2BIC2CINONE

WORD

AN/XXX-99 NONE

PROCESSOR AAA 83

PROCESSOR CCC Al

AN/ZZZ-77 Al

PROCESSOR FFF NONE

PROCESSOR HHH B3

AN/YYY-88 NONE

PROCESSOR BBB I C2

PROCESSOR DDD NONE

PROCESSOR EEE I Al

PROCESSOR GGG i NONE

PROCESSOR III NONE

Fig. 3. The operating system security level is a custom verbal criterion based on DoDsecurity standards. The most desirable value (Al) is indicated with an asterisk in the upperright corner.

Lightyeer

RULE NAME RULE


II

II

II

II

II

SECURE OS

NUCL HARD

LANGUAGE

OS LEVEL

S/W SUPPORT

SECURE OS MUST BE EQUAL TO Yes(ELIMINATION RULE)

NUCL HARD. MUST BE EQUAL TO Yes(ELIMINATION RULE)

ADA SUPPORT SHOULD BE EQUAL TO Yea(WEIGHT 50)

OS LEVEL SHOULD BE GREATER THAN BI(WEIGHT 50)

IF ADA SUPPORT EQUAL TO YeeTHEN AVAIL DEV SUPT SHOULD BE EQUAL TO Yee

(WEIGHT . 75)

Fig. 4. Design considerations concerning the value or relationship of design criteria may beexpressed as rules and evaluated in a Lightyear model.

eration. A weighted rule is one that has anumerical score or weight associatedwith it. If an alternative meets aweighted rule requirement, additionalpoints are added to its respective score.In addition, these two types of rules maybe either simple or compound (if -then).A simple rule (elimination or weighted)addresses the requirement for a singlecriterion. A compound rule addresses arequirement for a criterion "if" a sec-ond criterion meets a specific condition.

Examples of each of these types ofrules are shown in Fig. 4. Of particularnote in the use of Lightyear is the sim-plicity built into the program for the

construction of these rules. The user isnot required to remember any of thesyntax for the rules. All the possibilitiesare displayed as menu options duringeach stage of the rule building process,and the user simply makes selections.

Once the alternatives, criteria, andrules have been entered into the Light-year program the model is ready forevaluation. Models can be evaluated inthree ways; criteria only, rules only, orcriteria and rules. The first two providea method for evaluating portions of thetotal model. The last (criteria and rules)is the final goal. It is in this final stage ofmodel evaluation that Lightyear really

Kjellquist: Lightyear-an engineering analysis tool55

Table I. Evaluation criteria are listed with their modes (numerical, verbal) and their assignedweights (0-100).

Lightyear

CRITERIA MODE WEIGHT CRITERIA


MODE WEIGHT

MAX. KOPS 50 MEMORY CAP. N i 10li

II

I/O CHANNELS 10 SIZE N 100il

NUCL HARD. V 100 SECURE OS V 100 ;1

II

OS LEVEL V 75 NUMBER OF PROC N 50 II

ADA SUPPORT V 75 WEIGHT V 100 ii

!!

AVAIL DEV SUPT IV I

75II

II

L

Lightyear

CRITERIA: SIZE


II

li MOST DESIRABLE: 0n LEAST DESIRABLE: 100

VALUE RANGE0

1

ALTERNATIVE VALUE ALTERNATIVE VALUE

AN/XXX-99 50 AN/YYY-88 80

PROCESSOR AAA 20 PROCESSOR BBB 20

PROCESSOR CCC 50 PROCESSOR DDD 40

AN/ZZZ-77 25 PROCESSOR EEE 80

PROCESSOR FFF 20 GGG 20

PROCESSOR HHH 25 III 12

li

II I

Fig. 2. Size is a numerical criterion expressed in vertical inches of rack space, from 0 to 100.

Ada programming support was desir-able, and light weight and small sizewere important because this was an air-borne application. As with most com-puter applications, maximum through-put and memory capacity were anadvantage. Finally, the equipment hadto operate in a military airborne envi-ronment, with consideration given tonuclear effects. In addition to determin-ing requirements, engineering collecteddata on potential equipment candidates.The candidates included a large varietyof computers ranging from multipleparallel microprocessors to large mini-computers. Computers in governmentinventory as well as those commerciallyavailable were included. Even withsome culling of the potential candidates,more than a dozen remained for analy-sis.

With these basic requirements (andothers) and the list of candidates in

hand, systems engineering turned toLightyear for assistance in organizingthe analysis and selection of the bestcandidate.

Creating a model starts with definingthe alternatives to be analyzed. For ourexample of processor evaluation, a largenumber of computer manufacturerswere solicited for prospective hardware.The field was initially narrowed toapproximately 12 models. Theseremaining alternatives were entered intothe Lightyear program by selectingalternatives from the main menu andlisting the name of each model in turn.The names may be up to 15 characters inlength. The next step in using theLightyear program is the selection of thecriteria to be used. This portion of themodel construction is the most time-consuming; not because of the Light-year program but because of thethought that must go into a well -

constructed, accurate model. In addi-tion to simply deciding which criteria touse, their relative importance (weight)must be estimated and their type(numerical, verbal, etc.) must be speci-fied.

The processor evaluation used 11numerical, verbal, and graphical criteriaas a basis. They are listed in Table I.

The most common criterion for engi-neering evaluations is numerical data.Such factors as size, weight, processorspeed, memory speed, and availablegrowth are expressed in quantitativeterms. Lightyear accepts numerical cri-teria by defining the most desirablevalue and least desirable value and lin-early scaling values in between. Themost desirable value is given a maxi-mum score (maximum weight assignedto the criteria) and least desirable theminimum (0).

Figure 2 shows the numerical criteriaused for processor size in inches of rackheight for each alternative.

The ability of Lightyear to use verbalcriteria is one of its most powerful fea-tures. One particularly useful aspect ofthe verbal criteria is the ability to createnonlinear evaluation factors. For exam-ple, consider the situation where such afactor as equipment weight is being usedas a criterion. Normally this type of fac-tor would be expressed numerically,with values ranging from heaviest tolightest in linear progression. However,there are cases when this approach dis-torts the appropriate model. This kindof case can occur if only the extremes ofweight are important. In our example,computer weights between 100 and 200lb are considered equally acceptable.Weights above 200 lb are increasinglyundesirable. Weights below 100 lb arenice but should not control the model.With a verbal criterion this conditioncan be handled quite easily, as shownbelow:

Verbal criteriaUnder 100 lb100 to 200 lb200 to 350 lb350 to 500 lb

Over 500 lb

Weight1009050200

The more common use of a verbal crite-rion is for the scoring of design factorsthat can not readily be expressed interms of numerical quantities. Com-mon examples of this are quality, effec-tiveness, and other factors involving aspecialized vocabulary. In our examplethe security level of the operating system


D.T. Kjellquist

Lightyear an engineering analysis tool

Lightyear is a tool for the most common type of systems engineeringanalysis-the weighted -criteria alternative -selection analysis.

All systems engineering involves theanalysis of design alternatives and selec-tion of the optimum design solution.Examples of such analyses include theselection of computers, peripheralequipment, and software. The list isendless. In a large system, the numberof analyses and alternatives requiringevaluation may number in the hun-dreds. A PC workstation provides anexcellent basis for supporting the myr-iad system tradeoffs, when coupled withappropriate software.

Lightyear, a product of ThoughtwareInc., is a software package that providesan effective tool to aid engineers in anal-ysis of design alternatives. With asuperbly engineered man -machine inter-face and a singular focus on a weighted-criteria analysis, Lightyear is a uniqueengineering tool. It is clear that Light-year was developed with a rare insightinto user needs. Figure 1 illustrates the

Abstract: One of the newest uses forthe PC is in the area of decisionsupport. Lightyear is a program basedon the weighted -criteria method forranking decision alternatives. Theprogram has wide engineeringapplication for tradeoff analyses. Itsadvantages include ease of operationand the ability to use non -numericevaluation criteria.

Lightyear is a trademark of LIGHTYEAR, Inc.

n1986 RCA Corporation.

Final manuscript received January 21, 1986

Reprint RE -31-2-10

key output from a Lightyear analysis ofseveral candidate computers.

An example

A recent contract will illustrate thepower and usefulness ofLightyear. Thecontract required that the RCA Com-munication and Information SystemsDivision (CISD) study design alterna-tives for an advanced airborne com-munications system. Little guidance wasprovided by the customer except foroverall requirements for the system. Of

primary importance to the customerwas the selection of comprehensiveanalyses and clear and complete sup-porting rationale. A team of approxi-mately ten engineers performed theanalyses. An example of the type oftradeoff performed during the contractwas the selection of the primary com-puter used for controlling the system.From the guidance provided by the cus-tomer and standard military specifica-tions, basic requirements were deter-mined. For example, the system had tosupport a B1 computer security level,

Lightyear

RANK SUMMARY EVALUATION: CRITERIA

SUBJECT:PROCESSORVERSION:7 JANUARY

AND RULES

EVAL1986

SCORE,

1II

2

;1

3II

4II

!!5

6II

7

8

9it

10

11

12

AN/222-77

PROCESSOR

PROCESSOR

PROCESSOR

PROCESSOR

PROCESSOR

PROCESSOR

AN/XXX-99

PROCESSOR

PROCESSOR

PROCESSOR

AN/YYY-88

839 1

1

8381

7431

610!

587

4611

3891

3761

3741

370 1

1919

AAA

HHH

BBB

CCC

III

EEE

DDD

GGG

FFF

r---ii

POSSIBLE . 920

Fig. 1. Even alternatives that fail elimination rules (lighter bars) are ranked in the summaryevaluation output.


Box (a 10 -Mb cartridge data storage sys-tem), a 360 kb floppy disk drive, and amodem. The software is PMT, dBASEIII, and Relay communications soft-ware (published by VM Personal Com-puting). The processing sequence(shown in Fig. 1) is:

1. Create an ASCII export file for eachrelated PMW project file (PMWoperation).

2. Process PMW export file(s) withcustomized dBASE III program.This creates a data file of informa-tion required to set up a PAC IIIpro-ject (dBASE III operation).

3. Generate local reports using thecombined capability of PMW anddBASE III for audit prior to uploadto the mainframe (dBASE III opera-tion).

4. Transfer processed dBASE III/PMW file(s) to a mainframe file(Relay/modem operation).

5. Run PAC III (IBM mainframe oper-ation using either a terminal or a PCas a terminal).

The input requirements of PAC IIIinclude resource (primarily people rates)and actual cost data. Both types of datareside in routine CISD operating sys-tems. These sources are tapped to pro-vide continually -updated data to PACIII on a periodic (monthly) basis.

In generalConsiderations of the man/machineinterface are probably the single mostimportant parts of a system. They can

F.F. Lazarus joined RCA in 1955. Sincethen, he has participated in many RCAoperations. Currently, as a staff engineer,he is designing computer applications toenhance the productivity of CISD ProductEngineering. He holds a BSEE from CUNYand a MSEE from FIT. His technicalinterests are in the field of computerapplication engineering.Contact him at:Communication and Information SystemsDivisionCamden, N.J.Tacnet: 222-2079

Wiliam T. Kelley is an Administrator,Engineering Project Management at CISD.He joined RCA in 1950. He has held avariety of manufacturing, engineering,program management, business analysis,and administrative positions. He iscurrently focusing on improving projectplanning.Contact him at:Communication and Information SystemsDivisionCamden, N.J.Tacnet: 222-2421

make or break any new systemsapproach to project planning and man-agement. An assessment of the generallevel of expertise of the planned usercommunity in both basic project man-agement theory and computer literacy isessential before setting up such pro-grams. The more experienced membersof a user group may have a good practi-cal grasp of project management tech-niques but only minimal acquaintancewith computer -assisted planning. Entry-level technical personnel, on the otherhand, will characteristically be strong inthe computer area and have little back-ground in project management. Thisdiversity makes a user training programan essential part of an automated pro-ject management implementation plan.

A major training investment wasmade by CISD to bring all technicalmanagers up to a level of understandingthat is essential for successful introduc-tion of automated project management.

A combination of seminars by an out-side consultant, in-house hands-ontraining with PC software, and lecturesessions and workshops aimed at themainframe system were used. Thecourses were open to all technical man-agers. Easy access to terminals and PCsis another key factor in the acceptanceof automated project management.

The era of automated project man-agement is here. Those individuals whoare equipped to use automated systemsas part of their management techniqueswill have a competitive edge. The key isattitude and adaptability-the managerwho is waiting for the perfect system willstill be waiting years from now. Thoseresourceful individuals who can takeadvantage of the best elements of a com-bination of systems to attain a marriageof strengths will make significant contri-butions to successful project manage-ment at RCA.


quickly evaluated. PMW has become anaccepted standard for planning andcontrolling engineering projects withinCISD.

The ABT attitude toward customerrelations and philosophy of keepingtheir programs abreast of PC technol-ogy has strengthened user confidence intheir product. They have also been veryreceptive to requests for PMW featureenhancement and minor operationalmodifications. ABT's future plans forexpanding the utility of PMW track wellwith RCA needs for continuallyimproving resource and project man-agement efficiency.

Following a survey of mainframe sys-tems, PAC III by AGS ManagementSystems, Inc. was chosen for evaluationof a mainframe project managementsoftware program for overall projectand resource integration. It is a typical,versatile mainframe program that iscapable of processing and analyzing alarge database, and it generates bothtabular and graphic managementreports. This includes calculation andpresentation of financial data and thecapability of "earned value" reporting.

CISD's experimentation has been lim-ited to two specific program manage-ment software packages. However, thegeneral scheme of uploading data froma PC system could be adapted to manycombinations of PC and mainframesoftware. The prime requirement is thatthe PC system have the capability ofcreating "export" files. Then a com-mercially available, programmabledatabase system such as dBASE III(published by Ashton-Tate) can be usedto translate the data to a form compati-ble with the mainframe project manage-ment program requirements.

A translation program has been writ-ten that facilitates electronic transfer ofdata from PMW to PAC III. This is amajor breakthrough that will allowexploitation of the strongest elements ofboth systems in preparing integratedproject plans. Figure 1 shows the rela-tionship between the PC and mainframesystems.

Establishing the degree of compati-bility between the PC and mainframe iskey to preparing an interface program.

System differencesDetailed analysis of the specific formand content of the PC export file andabsolute knowledge of the mainframe

Project Plan ModulesCreated with

PMW and IBM PC

Prepare Auditusing PMW

and dBase III

Project

Performance

Reports

TranslateInternal PMW Data

Files to Fixed LengthASCII Records usingPMW Export Routine

Process Files withPAC III Project

Management Systemon the Main Frame

Process Files intoPAC III Format usingdBase III Database

Management Software

Transfer FormattedFiles to Main Frame

using RelayCommunications Software

Fig. 1. Relationship between the PC and mainframe systems.

system data input requirements is essen-tial. A few examples of differences thathad to be resolved to accomplish thePMW to PAC III transfer were:

1. Task ID: PMW uses a five -digitnumerical scheme assigned inter-nally by the program, outside of usercontrol. PAC III uses a four-

character alphanumeric ID, userentered.

Solution: Translate the PMW fivenumeric digits to three alphanumericcharacters, using a translationalgorithm. The fourth PAC III char-acter position was reserved for theassignment of an alphanumeric pre-fix, unique to each file transferred.This avoids the danger of duplicatetask IDs.

2. Logic network connections: PMWallows greater than 60 dependencies.PAC III has a limit of 10 dependen-cies per input record, and an abso-lute limit of 14 per task.Solution: Edit PMW files by puttingin dummy tasks as collection pointsfor every multiple of 10 dependen-cies. Put error trapping code in thedBASE III translation program tostop processing if the limit of 10dependencies is violated.

3. Privileged characters: PAC III usesthe left parenthesis at at the begin-ning of a task description to indicatethat the record is a dependencyrecord.

Solution: Code dBASE III to test fora leading left parenthesis in the task

description field, and if present, sub-stitute an asterisk.

4. Work days: PMW allows three digitsfor work days. PAC III allows twodigits for either work days or workweeks.

Solution: Code dBASE III to trans-late any work day durations of morethan 99 to work weeks.

5. Task description: PMW allows 60characters as a task description. PACIII has a maximum of 22 characters.Solution: Edit PMW files to abbrevi-ate task descriptions. Code dBASEIII transfer program to truncate anydescription longer than 22 charac-ters.

6. Task/phase hierarchy: PMW has afour -tiered hierarchy-project/phase/activity/task. PAC III usesthree tiers-project/phase/activity.Solution: Edit PMW files to transfermeaningful phase information to theactivity level. Code dBASE III trans-fer program to eliminate phase data.

System operation

The system for transfer of project plan-ning data from PMW to PAC III is amultiphase process. An IBM PC is usedwith well -established, commercially-

available software. Use of available PCdatabase management and communica-tions programs as building blocks mini-mizes the custom programming task.The PC is configured with 640 kb ofmemory, an IOmega Corp. Bernoulli

Lazarus/Kelley: Automated project management with PCs and a mainframe 51

F.F. Lazarus W.T. Kelley

Automated project managementwith PCs and a mainframe

PCs can't do the job alone, and mainframes can be unfriendly. Putthe two together with the right software, and project managementstops being the chore it once was.

RCA's Communication and Informa-tion Systems Division (CISD) designsand builds communications and infor-mation processing systems for myriadgovernment agencies. Typically, hun-dreds of contracts, ranging in valuefrom $100,000 to over $100,000,000,are in process concurrently. The desirefor improved scheduling and resourcemanagement focused attention on theneed for state-of-the-art automated pro-ject management systems.

A system specification was writtenfor an automated planning system thatwould improve the management effi-ciency while meeting the stringentreporting and control requirements ofRCA/CISD customers. Among otherrequirements, the ease of data entry,interactive on -screen planning capabil-

Abstract: CISD wrote a systemspecification for an automatedplanning system that would improvethe management efficiency whilemeeting the stringent reporting andcontrol requirements of customers.Among other requirements, the ease ofdata entry, interactive on -screenplanning capability, and large-scalemultiproject integration were primerequirements. This article discusses thesystem tested at CISD. It uses PCs asfrontend devices and a mainframe fordata "crunching." The software is alsodiscussed.

©1986 RCA Corporation.Final manuscript received December 12, 1985Reprint RE -31-2-9

ity, and large-scale multiproject integra-tion were prime requirements. The facil-ity to allow for personal and privateanalysis during the conceptual phases ofplanning a project was also high on thepriority list of desired features. Theoften overlooked importance of theman/machine interface was givenmajor consideration in the ratingscheme devised for evaluation of projectmanagement systems.

Review of the capabilities of both per-sonal computer -(PC) -based and largemainframe project management soft-ware systems showed that, when usedalone, neither the PC nor themainframe -type system could meet thetotal requirements for operation withina large and diverse business and culturalenvironment.

The better PC packages offer excel-lent screen interactive facilities, withvarying degrees of reporting and analy-sis capability. However, PC systems arelimited when it comes to handling inte-gration of resource data and subsequentrescheduling of 100,000 -plus tasksacross multiple projects. Repeatedattempts to stretch PC -based systems toplan multiple large projects were onlypartially successful.

Mainframe planning systems are wellsuited to "crunching" masses of dataand providing good multiproject analy-sis. However, the data input require-ments are characteristically "unfriend-ly" to the point of hostility to the casualcomputer user.

The concept of using a PC system as afrontend planning and data entry device

and a mainframe system for data inte-gration appeared to be the right answer.An implementation team was estab-lished to design and install such a hybridsystem. PC software was used for initialplanning and optimization of projectmodules, a PC database system wasused to customize and upload data, andthe mainframe system was used for finalprocessing and reporting. The resultingcombination of resources embodies thestrength of both PC and mainframe sys-tems while avoiding the weaknesses thatlead to non -acceptance by the user com-munity.

After a literature search for suitablePC systems, several copies of ProgramManager Workbench (PMW) byApplied Business Technology Corpora-tion (ABT) were purchased for evalua-tion. The ease of use, innovative screengraphics, and the live interaction fea-tures of the program made it a goodcandidate for consideration. In particu-lar, the speed of the in -memory proces-sing keeps pace with the user's thoughtprocess without the delays characteristicof "batch" -oriented systems.

During the first year of availability,many first -level engineering managershave adopted PMW as a prime manage-ment tool. The use of PMW has spreadthrough the engineering organizationdue to user enthusiasm as opposed tomanagement edict. During a PMW ses-sion, the manager is in intimate anddynamic contact with his project. Thedrudgery of using paper, pencil, anderaser is truly eliminated, and theimpact of alternate scenarios can be


expertise yielding the foundation of theDigicube.

References

1. Oppenheim, Alan V. and Schafer, Ronald W., Dig-ital Signal Processing, Prentice -Hall Inc., EnglewoodCliffs, New Jersey, 1975.

2. Strolle, C.H., Smith, T.R., Reitmeier, G.A.,Borchardt, R.P., "Digital Video Pror"sing Facilitywith Motion Sequence Capability," InternationalConference on Consumer Electronics, Digest ofTechnical Papers, pp. 178-179, June 1985.

3. Grogono, Peter, Programming in PASCAL (RevisedEdition), Addison-Wesley Publishing Company,Reading, Massachusetts, 1980.

4. Apollo Computer General Purpose Input/Output(GPIO) User's Guide, Apollo Computer Inc., 1983.

5. Archer, David E. and McLean, Richard H., "DigitalTime -Base Correction for Video Signal Processing,"SMPTE Digital Video, March 1984, pp. 69-73.

6. Lowry, John D. and Kupnicki, Richard, "DigitalProcessing in the DPS-1," SMPTE Digital Video,March 1979, pp. 43-63.

7. Grob, Bernard Basic Television Principles andServicing, McGraw Hill Book Company, New York,New York, 1975.

8. "The Logic Simulator," IDEA System User's Man-ual, Mentor Graphics Corporation, 1984.

9. Pritchard, D.H., "A CCD Comb Filter for ColorReceiver Picture Enhancement," RCA Review, 41,No. 1, p. 3, March 1980.

10. Law, K.A. and Bernard, F.S., "An EngineeringWorkstation and Video Frame -Storage Peripheralfor Simulating Digital TV Signal Processing," RCAReview, Vol. 47, No. 1, March 1986.

Multibus is a registered trademark of INTELCorporation.

DOMAIN is a trademark of Apollo Computer Inc.

AEGIS is a trademark of Apollo Computer Inc.

UNIX is a trademark of AT&T Bell Laboratories.

Ethernet is a trademark of Digital EquipmentCorporation.

VAX is a trademark of Digital EquipmentCorporation.

VT100 is a trademark of Digital EquipmentCorporation.

Digilink is a product of Fortel Corporation.

OIGICUBE GRAPHICAL PROGRAMMING ENVIRONMENT

Cursor position: It" 1023, Y= 0

INDOW MONITOR LIVE FREEZE LEVEL ADJ

INPUT 1 FIELD I I 1 1 VIDEO 1111 lllllll 1111

16 . 1 .1101111 2 FIELDS 2 FIELDS CHROMA

64 : 1 SPLIT 4 F IFI tr- 4 FIELDS I Hi

HUE

DATA TRANSFER COMMANDS

NTSC_2WRITE '1111111111111 READ

NTSC_3

NTSC_1

RED _1

r r GRE_1 FILE

PATTERN BLU_1

Fig. 6. Illustration of the graphical control environment for the Digicube.

Kirk A. Law graduated from PurdueUniversity, earning a BS in ElectricalEngineering in 1981 and an MS inElectrical Engineering in 1983.

During the summer of 1981, at RCAConsumer Electronics Division, Mr. Lawworked on redesigning preamplifiercircuitry for infrared remote controlreceiving. After receiving his MS degree,he joined RCA's rotation program, whichinvolved an assignment at AdvancedTechnology Laboratories. There heworked on the definition of processingthroughput benchmarking criteria for aDSP parallel processing simulationprogram. He was then assigned to theDigital Signal Processing Research groupat RCA Laboratories, where he helped todevelop simulation models for digital TV.In 1983, Mr. Law joined this group as anAssociate Member of the Technical Staff.He was promoted to Member, TechnicalStaff, in 1984. He has been working on thedevelopment of digital TV features byusing software simulation techniques. Mr.Law has five joint patents pending and is amember of IEEE and Eta Kappa Nu.Contact him at:RCA LaboratoriesPrinceton, N.J.Tacnet: 226-2475

Francis S. Bernard earned a BS degree inElectrical Engineering from RutgersUniversity in 1982. He was then awardedan RCA Graduate Engineering forMinorities (GEM) fellowship and earnedan MS degree in Electrical Engineering

Law (seated) and Bernard.

from the Georgia Institute of Technology in1984.

Mr. Bernard joined RCA Laboratories in1984 as an Associate Member of theTechnical Staff in the Digital SignalProcessing Research group. Prior tojoining the Laboratories, Mr. Bernard alsoworked on the initial details of interfacingthe Fortel Digicube system to theMENTOR engineering workstation. Mr.Bernard holds one U.S. patent, and is amember of IEEE and Audio EngineeringSociety.Contact him at:RCA LaboratoriesPrinceton, N.J.Tacnet: 226-2304

Law/Bernard: Simulation of digital TV signal process vith an engineering workstation 49

A brief description of a digital television receiver

The block diagram for a typicalTV system is shown in theaccompanying figure. Thedashed line represents the blockfor which the main article isconcerned.

This block diagram would berepresentative of an analog ordigital TV. In the case of DTV, thelocation of the analog -to -digital(A/D) converter might be at theoutput of the picture IF amplifier.The input to the TV scanningsynchronization circuits could beanalog or digital. The digital -to -analog (D/A) conversion isassumed to be located at theoutput of the luminance (luma)and chrominance (chroma)separation circuit. Therefore,there will be a D/A for eachoutput R, G, and B. For thepurpose of this article it isassumed that the samplingfrequency of the A/D converter is14.32 MHz, or four times the3.58 -MHz color subcarrier, andthe sample phasing of the A/Dconverter is such that thequadrature components I and Qof the chroma signal areextracted from the compositevideo signal (CVS).

The Fortel Digicube performsthe A/D conversion, initialstorage of video, and D/Aconversion operations. TheMultibus interface between theDigicube and an EWS or DSPunit allows for data transfer intoand out of the Apollo DOMAINSystem. The Digicube employsthe use of video frame

ANTENNA

RF TUNINGCIRCUITRY

PICTUREIF

AMPLIFIERAND VIDEODETECTOR

COMPOSITEVIDEOSIGNAL

-LUMINANCE ANDCHROMINANCE

SEPARATION TORED, GREEN, AND

BLUE PRIMARYCOLORS

i TV SCANNINGSYNCHRONIZING

CIRCUITS

Block diagram of a typical TV system.

synchronizing circuitry.Therefore, input video to theDigicube is analyzed andstripped of its horizontalsynchronizing (sync) signal andits color burst reference signal of3.58 MHz so that new generatedhorizontal sync and color burstsignals can be put in theirplaces. These new signalsgenerated by the Digicube will befree from the nonstandarddeviations in horizontal linefrequency experienced in suchsignals as video output from avideo tape recorder (VTR).Because of this stripping of thehorizontal sync and burst, thevideo data we can access fromthe memory of the Digicube is

RED

GREEN

I BLUE

DEFLECTION SIGNALS

PICTURETUBE

that of the active video sceneonly. This consists of the part ofthe video line after the sync andburst and between theoccurrences of the verticalsynchronization (vertical sync)periods. The simulation programcan take this into account.

In a DTV, the actual separationof the luma and chroma signalscan be performed using variousprocessing schemes. Theliteratures discusses the visualimprovements noted through theuse of line -comb filtering, whichis filtering of the CVS in thevertical resolution direction, overthe standard practice ofluminance and chrominanceseparation.

with the Digicube using various user-friendly features. These communicationprograms have included control throughmenus consisting of textual prompting andeven high -resolution graphics symbols thatrepresent command choices. These graphicssymbols can be selected via a graphiccursor (using a mouse, for instance). Figure6 is an EWS screen hardcopy illustratingthe Digicube's graphical programmingenvironment. Shown are the symbols forthe various function choices: "WINDOW,"the digitizing window size; "MONITOR,"the viewing choice at the monitor; "LIVE,"

which allows the user to view the inputvideo according to the chosen time basecorrector/frame synchronizer mode;"FREEZE," the number of fields to dig-itize; "LEVEL?ADJ," which allows variouslevel adjustments for the incoming videoto the Digicube; and "DATA TRANSFERCOMMANDS," the data transfer choicesbetween the EWS and the Digicube.

AcknowledgmentsWe would like to acknowledge those whohave given us their support and given

their guidance in the development of thevideo functional modeling system. Specialthanks are given to Stanley Knight, JonClemens, David Holmes, Jack Fuhrer,Thomas Bolger, Robert Dawson, and LarryKratzer for their continued support andenthusiasm. Others lending their time,efforts, and expertise include HerbertTaylor, John Lee, and Eric Batterman.Also, we would like to thank Danny Chinfor providing the basis for the graphicalprogramming environment.

Appreciation is extended to Fortel Cor-poration for providing their engineering


Mentor Quicksim

Mentor Quicksim allows theverification of the functionality ofdesigns produced by the otherMentor applications programsthat yield schematic capture ofcircuits as well as symbolcreation for these circuits. Thecreation of symbols for thesecircuits allows for the inclusionof previous designs in othercircuit schematics. Quicksim is anine -state, event -drivensimulator designed specificallyfor MOS, TTL, and ECL logic.There are several features of theMentor Quicksim that tend tomaximize its performance. Threeof these are: (1) Quicksim'soscilloscope -type logic leveltracing of network data line (net)activity; (2) its interactive controlallowing the user the choice ofsignals to monitor, list or tracehow the circuit is to bestimulated, and any probes orbreakpoints to be set; and (3) thepossibility of saving the circuit'sinput stimuli, history, and theparticular state of the simulation.

Quicksim is a fairly efficientsimulator because it onlyexamines components with

changing input states. Thisapplication program is not,however, a worst -case timinganalyzer. This is performed byanother Mentor Graphicsapplications program calledTVER, or timing verifier. Theapplications program Quicksim,is driven by events that are madeup of transitions or changes inlogic state on a net that occurthrough time. At the invocation ofthe Quicksim program, no valuesexist on any of the circuit's nets.As the circuit is simulated, eventsare scheduled and Quicksimthen propagates the statechanges through all the netsconnected to this circuit. Anychanged inputs on what aretermed "primitives" or behaviorlanguage models (BLMs) areevaluated to see if any changesare occurring at their outputs. Ifany outputs of the elements arechanging, then more events arescheduled.

The term primitive, used above,refers to Mentor's use of"modeling primitives" or built-insimulation models whose func-tional behavior Quicksim

understands. Examples of someMentor primitives include:

Simple logic gates: INVERTOR,AND, OR, NOT -OR,EXCLUSIVE -OR, andEXCLUSIVE -NOT -OR gates;Memory devices: RAM, ROM,and PLA;Registers and latches.When Quicksim attempts to

evaluate the inputs to an elementin a circuit schematic so as todetermine if an output is to begenerated, it references aproperty of the element known asMODEL. If the MODEL propertyof this element corresponds to aMentor -defined primitive, like oneof those listed above, it refers toits own built-in functionaldescription when determining anoutput value from the inputstimuli. If Quicksim does not findcorrespondence between theMODEL property found and itslist of possible primitives, it looksat the underlying schematicsheet or sheets of that elementuntil it finds a set of modelingprimitives by which it candetermine the part's functionalbehavior.

In the functional modeling of a videoprocessing system, such as the DTV, it isdesirable to observe the effects of motionon the signal processing methods chosen.The Digicube's windowing mode providesa means to store long sequential sequencesof video. In windowing mode the Digicubeis able to store a portion of each field ofthe incoming video in the NTSC mainmemory block. This main memory blockis capable of storing up to four full videofields (each full video field being 768 pixelsby 240 video lines). The window sizesavailable allow the storage of up to 16fields (384 pixels by 120 video lines perfield) or 64 fields (192 pixels by 60 videolines per field) of incoming video. Forexample, a reduced picture area (window),which contains 1/4 of the maximum num-ber of video lines (height) and 1/4 of themaximum number of video pixels (width)of each field is stored rather than a fullfield of video.

An illustration of the various windowsizes (1:1, 4:1, and 16:1) is given in Fig.

5, along with the partitioning of the incom-ing reduced active video area fields withinthe main NTSC memory block of theDigicube.

In the display of TV video image data,the conventional method is a 525 -videoline 2:1 interlacing format.' Anothermethod suggested for the display of TVsignal data uses a 525 -line non -interlaceformat. This reduces the visual artifactknown as line -flicker and raster linestructure.

The Digicube allows one to display ineither the 525 -video line 2:1 interlaceformat or to create the impression of a525 -video line non -interlaced format. Thisfeature allows the subjective evaluation ofboth display formats. The latter displayformat is achievable through the use ofvarious processing techniques such as line-or field -progressive scanning.

Non -interlaced displays can be achievedthrough the display of a single field ofvideo data where the number of horizontalline synchronization pulses is forced to be

262 instead of 262.5. This causes the videoto be rewritten over the same spacial loca-tions on the display monitor. The monitor'svertical scanning height is then decreasedto half its standard height. This methodimplies being able to simulate the displayof a noninterlaced format for only half ofthe original image. Through software, forexample, to generate a field -progressivescan simulation, one would write a line ofvideo data from field 1 followed by a lineof video data from field 2 into the outputfield to be displayed. Obviously, the outputfield becomes full at the point halfwaythrough fields 1 and 2. The resulting imageis half the size of the original. Pertinentinformation characteristic of the displayformat chosen can be discerned withoutadditional hardware.

In summary, the Digicube has becomea very useful and versatile peripheral forthe investigation of video signal process-ing problems. Also, because of the periph-eral interface and graphics capabilities ofthe EWS, we are able to communicate

Law/Bernard: Simulation of digital TV signal processing with an engineering workstation 47

VIDEO IN

GENLOCK IN )

VIDEO OUT

FROMDOMAIN -OwNETWORK

TODOMAINNETWORK

INTERFACE/REAR PANEL

BOARD

MULTIBUS MULTIBUSADAPTER

BOARD

APOLLO CABINET

Fig. 4. Digicube block diagram.

A. Illustration of the possible window sizes:

4

3

2

NORMAL

2

16

FRONT PANEL

INPUT/OUTPUTBOARD

(R/D & D/A)

DIGILINK

4:1

PARALLELDATA INTERFACE

DIGITIZED VIDEO

PIXEL CLK & SYNC SIGNAL

B. Illustration of the field sequence and partitioning for single channel's four field memoryfor the different window sizes:

2

2

3

4

I 4

3

2 6

5

9 13

1515

14

16

Fig. 5. Relationship of window size and memory storage partitioning.

and routes this information to the Digi-cube's control board for subsequent execu-tion. The read/write adapter board permitsdirect memory accessing of the Digicube'svideo storage memory to allow transfer ofdata to and from the EWS. The Multibusadapter board is located in one of theEWS Multibus card cage slots. As shownin the figure, the input/output board con -

CONTROLBOARD

READ ,WRITEADAPTER

16:1

tains the A/D and the D/A convertersalong with circuitry that performs videosignal conditioning. The front panel offersmanual control of various functions anddisplays the Digicube's status.

The video data is stored in a memorycomposed of an array of 64k dynamicRAMs (DRAMs). There are actually fourblocks of main memory, each one capable

ADDRESSESAND CLOCKS

DIGILINK

AUXDATA IN

MEMORYBOARD

rDATA OUT

SERIAL COMMUNICATION LINK

DIGICUBE CABINET

of storing four video fields. The four blocksare denoted as NTSC, R, G, and B. Theyare used as follows. When the Digicube isused as a viewing device for RGB data,the R, G, and B main memory blocks areused. An incoming video sequence, whichis in an NTSC composite video signal

into the NTSCmemory when requested. If the user hasan NTSC CVS stored on the EWS anddesires to view the video data on theDigicube, he or she would transfer thisvideo data into the NTSC main memoryblock. When the video data is read fromthe NTSC video memory block a colorburst reference signal is added to the signalalong with the horizontal sync signal, andother associated chroma processing is per-formed before the data is displayed. Notethat the designation of the video memoryblocks as NTSC, R, G, and B is to signifythat the R, G, and B memory blocks donot perform any insertion of color burstreference or associated chroma processing,but do insert the required horizontal synctiming signals.

The Digicube samples the incomingCVS at the sample positions correspondingto the chroma quadrature positions of Iand Q. This convention is assumed forthe display of composite video placed inthe NTSC memory block also. Thus, ifone loads the NTSC memory block withcomposite video, it will assume that thesamples are located at the quadrature posi-tions of I and Q. If the data to be viewedwas sampled differently from this conven-tion, one will see a chroma hue error inthe displayed video signal.


per color. This yields over 16 millionpossible colors.

This latter feature is used by the authorsand allows the viewing of decodedcomposite video pictures (RGB) usingdata digitized by the Fortel Digicubeand subsequent processing by the func-tional modeling software emulating adigital TV.

3. The DOMAIN network-The DO-MAIN network is a high-speed, token-

passing local -area network that giveseach user a dedicated computationalnode based on a 32 -bit computer archi-tecture. This network uses a 12 mega-bit -per -second token passing ring imple-mented with coaxial cable. Apolloclaims that a network is maintainableeven with nodes separated by as muchas 1000 meters (over 1/2 mile). Thistoken passing protocol insures that allworkstations and DOMAIN server pro-cessor units (basically workstations withno display hardware that can be usedto run non -viewable processes to serviceperipherals) on the network have guar-anteed access to the network, and allowsa large number of workstations to sharea single network without collision detec-tion schemes.

By the end of 1986 there will be approxi-mately 50 nodes (EWSs and DOMAINserver processing units) at RCA Labora-tories, and about 50 elsewhere in the cor-poration. The total amount of disk spaceis expected to be around ten gigabytes.Since the DOMAIN system appears to allusers as a single file structure, and eventhough the actual data may be physicallylocated on various system nodes, the diskresources are available to all users in thenetwork.

Gateways allow access to other DO-MAIN local -area networks, to remotemainframes and superminicomputers, andto public long -haul packet -switching (X.25)networks. Communication with IBM sys-tems is supported by gateway softwarethat emulates several RJE (Remote JobEntry) subsystems-IBM's HASP and thestandard 2780 and 3780 environments.An Ethernet gateway provides DOMAINsystem users shared access to Ethernettransport facilities. The TCP/IP (Trans-mission Control Protocol/InternetworkingProtocol) provides easy access to DECVAXs and other systems.

All DOMAIN system workstations andservers have RS -232C ports and providefor VT100 terminal emulation. In addition,some models of the EWS and all DO-

GATEWAYEWS

0

DOMAIN SYSTEMLOCAL AREA NETWORK

EWS

EWS

MAINFRAME

GATEWAY

Fig. 3. Illustration of a possible DOMAIN

MAIN server processor units contain Mul-tibus Interfacing hardware. Apollo givesthe user the capability of defining theperipheral interfacing environment throughthe use of general-purpose input/output(GPIO) software.'

Figure 3 is an illustration of two DO-MAIN local -area networks, one with threeand the other with four engineering work-stations. These two local -area networksare linked by a communications gateway.The drawing also shows a gateway to amainframe.

It must be pointed out that one of theadvantages that we saw in using this par-ticular workstation was that in addition tothe offerings discussed above, it allows forstraightforward interfacing of user -definedperipherals. In our case this was the FortelDigicube digital frame store device.

Video frame storage deviceWe chose the Fortel Digicube as our work-station video multiframe storage device.This system is useful for subjectively testingdigital signal processing algorithms thatoperate on NTSC video pictures. Thisframe store is built around a Fortel framesynchronizer. It is capable of capturing upto 64 fields of video, and the user canchoose from three digitizing window sizeformats. Once this video data has beencaptured the digitized video can be accessedby any node on the DOMAIN networkvia the Multibus interface. Software run-

EWS

EWS

DOMAIN SYSTEMLOCAL AREA NETWORK

EWS

networking scheme.

ning on an EWS can now access andmanipulate this digitized NTSC video data.For example, the results could be decodedusing a DTV simulation program, and theresulting RGB components could be trans-ferred back into the RGB memories ofthe Digicube for viewing. The video storedin the RGB memories will then be readout at a rate of 14.32 MHz, converted toan analog signal, and displayed in one ofthe window size formats on a displaymonitor. Because the Digicube is a framesynchronizer/time base corrector (FS/TBC),54 non-standard signals from a video-tape recorder (VTR) or videodisk playercould be captured for analysis also.

The Digicube's windowing option allowsa reduced area of the active video field tobe digitized and stored. This enables longertime sequences of video storage for thesame amount of the Digicube's memory.Thus, one can digitize a video sequence ofup to one second in duration. NTSC de-coding algorithms can use this feature toshow (and allow the study of) motionartifacts due to temporal type processingsuch as frame combing and frame recursivefiltering.

The Digicube allows freeze -field, freeze-frame, hue, chroma, setup, and video levelsto be altered under remote control via theFortel Digilink serial communication link.This is shown in Fig. 4 in an illustrationof the main blocks of the Digicube. TheDigilink interface interprets commandsfrom the EWS (via the Multibus interface)


A functional software simulation scheme for digital television

The simulation of a digital signalprocessing (DSP) system cantake many forms. Mentorsoftware packages allow adesigner to simulate digitalcircuits in such a way as toextract the timing and logicalcorrectness of the circuit design.With a good understanding of theDSP algorithm being implementedin simulated hardware compo-nents, a designer can relate thelogical correctness to theparticulars of the algorithm. Forcomplex algorithms, however,the implementation in logiccircuitry becomes so compli-cated that extracting algorithmparticulars becomes a burden-some task at the logic gate level.Simulating the DSP algorithmwhile still at the mathematicalexpression level offers easiermodification and parameter

variation testing. Themathematical expression level ofsimulation can be done usinghigher -level programminglanguages such as Pascal,Fortran, or C. At this level, blocksof programming code thatcorrespond to blocks of logiccircuitry can be denoted andoptimized with special attentiongiven to such effects as finiteword length (FWL) processing.

Mentor allows functionalrepresentations of logic-gate-interfaceable circuit elementsusing Pascal. These elementsare referred to as behaviorlanguage models (BLMs).Through the use of BLMs adesigner can model variousfunctions in Pascal and havethese functions be executable inan event -controlled environmentknown as Mentor Quicksim6

(interactive simulator-see thesidebar on Mentor Quicksim)applications program.

BLMs can be used with otherelements whose functionaldescription is defined by a set oflogical primitives. Because theBLM also operates in a timing -controlled environment, timingand logical verification of thesystem can be obtained usingQuicksim. This BLM capabilitybecomes advantageous whenalgorithms require complex logiccircuitry to implement. Oneadvantage is the reduction of thetime required for total systemsimulation. Another is thecapability to provide designerswith a means of testing thealgorithmic equivalence of theirlogical primitive -derived circuitsagainst a verified functionaldescription, the BLM.

r

000000

000000

I F-100000 0 00 0'

oorionoop00000l

U U

DISPLAY

MONITOR

(NTSC & RGB)

NTSC VECTORSCOPE

WAVEFORM MONITOR

VIDEO SIGNALGENERATOR

OFF -THE -AIR TUNER

VIDEO SIGNAL SWITCHER

Fig. 2. Illustration of VFMS componentrack used by the authors.

features in actual televisions. The off -the -air tuner allows for the digitizing of livevideo signals, and the video signal switcheris used for routing the choice of inputvideo to the Digicube.

The Mentor engineeringworkstationIn the early 1960s, data processing meantdealing with large mainframes in batch

processing terminology. In the mid -1960smore alternatives became available as com-puter technology grew. These alternativesincluded time-sharing and dedicated mini-computers. Some pitfalls of the two com-puting schemes were limited performancedue to shared computing power, delaysdue to time-sharing, and lack of readilyavailable program, file, and peripheralsharing in minicomputers.

Apollo Computer (the computing hard-ware used by Mentor) developed theDOMAIN operating system, which com-bines the favorable aspects of the comput-ing schemes mentioned above. There arethree basic principles behind the DOMAINsystem architecture:

1. Mainframe -style dedicated computing,which entails virtual memory, largeaddress space, and an assortment ofsoftware.

2. Tightly integrated high -resolution graph-ics features.

3. A distributed system linked by a high-speed local -area network. This offersusers the responsiveness of a single -user system with the centralized fileaccess of a time-sharing system.

These principles are illustrated by Apolloin the following manner:

1. Operating system-The DOMAIN sys-

tem architecture provides the user witha computer workstation (node) inte-grated into a transparent local areanetwork. Each EWS features a 32 -bitprocessor. There are two operating en-vironments: AEGIS and AUX (AUXis Apollo's implementation of UNIXsoftware).The operating system of the Apollocomputer workstation used by theauthors was AEGIS. Apollo makes thepoint that the software engineeringresults and/or application packages ofone operating system are compatiblewith the other. All Apollo system soft-ware runs under either operating system,and in fact AEGIS and AUX could berun concurrently in different processes.

2. High -resolution graphics-The DO-MAIN workstations employ high -res-olution, bit -mapped, raster graphicstechnology. Resolution ranges from1024 X800 (monochromatic or color)to 1024 X 1024 (color). The presenceof a dedicated graphics processor allowsfor the display of graphical informationsuch as fast vector generation, area fills,and bit block transfers. On some Apolloworkstations, one can purchase extradisplay memory that allows viewing ofpictures using red, green, and blue(RGB) data with eight bits of resolution


on a monitor, or it can be routed totypical video analysis tools such as a wave-form monitor and/or vectorscope.

One could use the high -resolution dis-play capabilities of the EWS to view theprocessed video data. The EWS allows 8 -bit -resolution red, green, and blue signalcomponents to be viewed on its high -resolution monitor. Also, the digital datacan be plotted in the form of a waveformanalyzer curve. Because of the high -resolu-tion graphics available on the EWS, auser could generate other graphical repre-sentations of the video data (multidimen-sional plotting).

Via the EWS network gateways, aninterface to other computer systems ispossible. One such system offering somesimilar video signal processing modelingcapabilities is the Digital Video Facility(DVF)2 at RCA Laboratories. The abilityto transfer video data between the DVFand VFMS implies that one can processor view the information that has beendigitized on either system. Since both sys-tems have the ability to digitize a videosignal, the versatility of each is enhancedby their interaction.

The VFMS is different from the DVFin the sense that the former offers a designera dedicated desktop computer, while thelatter is composed of a time-shared com-puter system with a user's display room.The DVF does allow for processing of thevideo data from remote terminals and inbatch mode, but the viewing must bedone in the display room. The high -resolu-tion graphics features of the workstationare more flexible than those offered bythe DVF. One advantage of the DVF isits video storage capacity. It can store alonger video time sequence than the VFMS.Since data transfer between the two systemsexists, it is possible to digitize a long videotime sequence on the DVF and transfer itto the VFMS for processing. The VFMScould be used to view parts of this digitizedsequence or parts of the processed se-quence. For viewing the total sequencethe transfer of this data back to the DVFwould be required.

The EWS that was used as the hub forthe development of this functional simula-tion environment is the Mentor GraphicsWorkstation. The Mentor Graphics EWSis composed of an Apollo Computer engi-neering workstation combined with MentorGraphics software.

Note that the majority of illustrationsand original text entry of this paper wheredone using Mentor Graphics software tools.Highlights of the workstation will be dis-

VIDEO TRANSFER

TO/FROM FORTELDIGICUBE

FORTEL

DIGICUBE

MULTI -

FRAME

STORE

U U

DISPLAYMONITOR

L J

[17 0 1 1

0 000 0

E 00000000

EQUIPMENT RACK

INPUT

VIDEO

SIGNALS

COMPUTER

DISPLAY

MENTOR -

GRAPHICS/

APOLLOENGINEERING

WORKSTATION

IEEE MULTIBUS INTERFACE

Fig.1. Configuration of a video functional modeling system (VFMS).

cussed later, and will center around theactual features of the Apollo ComputerEWS.

The sidebar contains a brief discussionof Behavior Language Models (BLMs).BLMs are a feature of the Mentor Graphicssoftware logic simulator which gives auser the ability to model complex logiccircuitry through a descriptive algorithmwritten in Pascal,. These BLMs can beused in circuits containing other BLMs orformal logic elements. An example of aBLM's value becomes evident when onedesires to include intensive logic devicessuch as microprocessors in the circuit de-sign. If this logic element is not present inthe library of components available to thedesigner, he/she can model the functionaldescription of the device in a Pascal struc-ture instead of in its logic gate equivalent.

The basic simulation system to be dis-cussed herein is illustrated in Fig. 1. In itthere are references to an Apollo Network.The Apollo Computer System allows fordatabase sharing among other nodes (orEWS) by placing the nodes in a ringsarchitecture referred to as the ApolloDOMAIN (Distributed Operating Multi -Access Interactive Network) System. TheApollo is able to communicate with periph-eral devices such as high -resolution printersand other plotting devices, tape drives,

INPUT FROMAPOLLO

NETWORK

OUTPUT TO

APOLLO

NETWORK

and user -defined peripherals (such as FortelDigicube) through its Multibus Interface.

Component options for a VFMSHere we will briefly describe the compo-nents that made up our video functionalmodeling system (VFMS). Figure 1 showsa basic VFMS, and Fig. 2 shows someadditional components in a rack configu-ration. These components were found tobe useful in the analysis of both the videoinput and output signals used in functionalsimulations. The monitor is capable ofdisplaying both NTSC composite as wellas RGB-format signals. For the purposeof analyzing DTV results as if from anactual TV, one might also want a monitorwith a picture tube characteristic of apossible DTV product. The Fortel Digicubeis capable of outputting data in NTSCcomposite or in the RGB format.

The vectorscope and waveform monitorshown in the figure are standard videomeasurement tools and have been quiteuseful in video signal analysis and char-acterization. The video signal generatorallows standard test signals to be digitizedby the Digicube for use as input to asimulation. This has proven effective forcomparing the performance of simulatedDTV features with the performance of


K.A. Law F.S. Bernard

Simulation of digital TV signal processingwith an engineering workstation

In this article, which is a shortened version of one that willappear in the March issue of RCA Review, the authors proposea technique for modeling and evaluating digital TV signalprocessing schemes at the algorithmic level, rather than at thecomplex logic circuit level.

Abstract: This paper describes a videofunctional modeling system (VFMS)developed to simulate digital video signalprocessing. The VFMS comprises aMentor Graphics engineering workstation,a video data multiframe storage device,and software generated by the authors.The VFMS functions as a node in anetworking system that permits communi-cation with other engineering work-stations and mainframes. Softwaredeveloped by the authors for use with thissystem is briefly described A digitaltelevision chrominance/luminanceprocessing system is presented as anexample of a typical modeling system. Thestorage/playback device in the systempermits digital frame or multiframestorage. It includes means to synchronizethe incoming video signal, digitally samplethis information, and store the data on adisk file located in the engineeringworkstation. The data is processed with auser -developed functional modelingprogram of a digital video processor, andthe results are sent back to the framestore for viewing and subjectiveevaluation. This system concept couldeasily be used in the development of aspeech and/or music processing system,as well as for a digital image processingsystem.

©1986 RCA Corporation.Final manuscript received February 4, 1986Reprint RE -31-2-8

Much attention has been focused on theapplications of engineering workstations(EWS) to integrated circuit (IC) and/orhardware logic design. In some applicationsoftware packages the intent is to givedesigners a way to enter schematics, verifylogic, verify timing, and document results.This paper will discuss a scenario for mod-eling and testing digital video processingsystems at the algorithmic level as opposedto dealing with the details of such a systemin the form of very complex logic circuitry.This paper will also describe how inputdata for an algorithm is obtained throughthe use of hardware interfaced to the EWS.

The need for video processing functionalmodeling was seen in the development ofdigital television (DTV). In particular, thenumber of digital signal processing featuresthat the design engineers want to evaluatefor this product makes a functional simu-lation tool a necessity. The functionalmodels of many of the DTV features havebeen or are being simulated, and havegreatly reduced the amount of time thatwould have been needed for initial eval-uation of the feature. Note that once anactual feature algorithm has been identified,without the functional simulation capabil-ity, the building of these features on bread-boards would have been required. It couldtake days or months instead of hours ordays to implement such features as lumi-nance/chrominance decoding schemes,automatic horizontal luminance peakingschemes, luminance vertical detail peaking

schemes, finite word length effects in digitalfiltering,' etc. As one might expect, be-cause of time deadlines on product designs,some algorithms may be overlooked be-cause of hardware breadboarding time andcomplexity. These overlooked algorithmsmay actually offer better performance forthe implementation of a specific feature.Suffice it to say that software functionalmodeling has allowed the evaluation ofmany approaches and their perturbationsfor DTV and its many features.

In order to evaluate the effects of DTVfeatures on true video information, it wasnecessary to capture and display videosignals. This need to perform data capturethrough digitization from signal sourcessuch as signal generators and off -the -airprograms and eventual display of processedresults was met by a multiframe storagedevice-the Fortel Digicube. The combina-tion of the EWS, multiframe storage de-vice, and associated software will be re-ferred to as the video functional modelingsystem, or VFMS.

The multiframe storage device will dig-itize video at the rate of 14.32 MHz (fourtimes the NTSC color subcarrier frequency)and store it internally until transfer to theEWS. After the data has been processedby the functional modeling software, itcan be analyzed by the user in one ofseveral ways. The processed video couldbe transferred back to the Digicube, whichcan convert the digital data back to ananalog signal and allow it to be viewed


As a graphics emulator Reflection 3 hasallowed almost all of our two -terminalworkstations to be converted to singleIBM PC workstations. When the 2623 isused to preview graphics for final out-put on a plotter, the terminal can bereplaced with this emulator; however, ifa 2623 is being used for final outputfrom a remote mainframe, the resolu-tion of the PC monitor or of the print-out may not be high enough to suit theneeds of the application. Figure 6 showsa graphic image displayed on a standardHP 2623A terminal. The quality of thisimage can be compared with that of thesame image produced by Reflection 3 onvarious IBM PC displays. In Fig. 7 theimage is reproduced on a standard IBMPC color display. In Fig. 8 the image isshown on an IBM monochrome displaywith a Hercules driver board, and inFig. 9 the image is shown on anEnhanced Graphics Display. Both themonochrome and the EGAD displaysproduce image quality similar to the HPterminal. The resolution of these moni-tors when used with Reflection 3 is listedin Table I.

Advantages:

1. Ability to display HP graphics on anIBM personal computer.

2. Much more terminal memory than2623.

Disadvantages:

1. None of the supported monitors ispixel -for -pixel compatible with the2623, which causes some aliasing inthe display.

2. Printout is not high resolution.

Reflection 3 requires an IBM PC, PCXT, or PC AT with at least 256k ofRAM. At least one floppy disk drive isrequired as well as a serial port for asyn-chronous communication. Version 2.0or higher of DOS is needed. There arealso versions of Reflection 3 availablefor the AT&T personal computer andthe HP Vectra personal computer,although we have not tested these.Reflection 3 is a new product and is inthe process of being enhanced. Themain enhancements being pursued aresupport of color graphics on theEnhanced Display, support of plottersattached to the PC, and higher resolu-tion for graphics printouts.

Authors, left to right: Miller Renfro, Kunz.

Peter J. Kunz received the BSME with a minorin Materials Science from the University ofConnecticut in 1979. He joined RCA in 1979 andhis initial work centered on machine design ofproduction equipment for the electron gun,including cathode problems and the quick -heatbimetal cathode process. He received his firstpatent in 1983 and has several patents pending.In 1983, Mr. Kunz was promoted to Member,Technical Staff and joined the VCAD project. Inthis capacity, he was responsible forimplementation and support of the CADAMcomputer -aided design system includingtraining, database management, finite -elementmodeling, productivity measurement, andsystems support. In March 1986, Mr. Kunzbecame leader of the VCAD group.Contact him at:Video Component and Display DivisionLancaster, Pa.Tacnet: 227-6492

Richard H. Miller received a BSEE from LehighUniversity in 1978. He joined the RCAEngineering Rotational Program in 1978 andaccepted a permanent assignment withLancaster Applications Engineering. His initialwork involved automated testing equipment forpicture tubes and equipment support forYAMMED entertainment product for theEuropean market. In 1982 he was promoted toLeader, Technical Staff of Test Methods andSystems in Applications. His currentresponsibilities center around picture tube

testing and include development of display tubeYAMMING systems, laboratory test equipmentand the associated data analysis systems,picture tube setup equipment for the customer,and support for production testing. He hasrepresented Applications Engineering on theVCAD Task Force since its inception in 1983.Contact him at:Video Component and Display DivisionLancaster, Pa.Tacnet: 227-6450

G. Mike Renfro is the Manager of ApplicationsEngineering at VCDD in Lancaster, Pa. Heearned the BS, MS and PhD in Physics from theUniversity of Oklahoma. He is a member ofSigma Pi Sigma and Phi Beta Kappa. He cameto RCA in 1980 as a Member of the TechnicalStaff in the Tube Development department ofVCDD. His initial work involved electrontrajectory studies for the COTY picture tubeproject. Between 1983 and 1985 he served asleader of the deflection components group ofProduct Design. During that time he alsorepresented Product Design on the VCAD taskforce. His current interests include thedevelopment and implementation of acomputer -integrated manufacturing strategy forthe Division.Contact him at:Video Component and Display DivisionLancaster, Pa.Tacnet: 227.6131

Conclusion

Despite the fact that IBM entered intothe engineering graphics arena ratherlate by most standards, VCDD has suc-cessfully achieved a marked increase indesign and drafting productivitythrough the use of IBM products. Wehave avoided some of the potentialproblems with this decision by choosingnot to write our own graphics software.We have made great strides in engineer-ing automation by simply using com-mercially available software. In its 24months of operation, the VCAD engi-neering computer system has become anintegral part of our Division.

References1. J.C. Miller, "VCAD, an engineering productivity

strategy," RCA Engineer, 29-3, pp. 4-9 (May/June1984).

2. J.C. Miller, W.T. Kelley, "Personal computers, anengineering productivity tool," RCA Engineer, 29-3, pp. 10-14 (May/June 1984).

3. J.C. Miller, P.J. Kunz, "CADAM, an engineeringproductivity tool," RCA Engineer, 29-3, pp. 15-19(May/June 1984).

4. T.V. Hoffmann, "Graphic Enhancement," PCTech Journal, 3, pp. 58-71 (April 1985).

5. K. Koessel, D. DiNucci, "Clearly Resolved," PCWorld, pp. 256-261 (June 1985).

6. D. McCune, "The Myth of the Virtual Device," PCWorld, pp. 178-187 (April 1985).

7. Reflection 3, copyright 1985 by Walker, Richer andQuinn, Inc., Seattle, Wash.

Kunz/M i I ler/Ren f ro: IBM graphics workstations at Video Component and Display Division 41

1111.1111111111111111111111_Table I. Monitor resolution.

GraphicsPixel Count

Columns Rows

Text

HorizontalLines Char

VerticalLines Char

HP 2623 512 x 390 39 73 51 55

Standard Color 640 x 200 33 74 53 40Graphics 4Enhanced 640 x 350 43 74 74 43Graphics Adapter

Hercules Board 720 x 348 38 74 53 58

4. Preset function keys for initial val-ues.

5. File transfer with HP 3000 system.6. Optional hard disk backup program

to HP 3000 system.

Disadvantages:

1. Not all screen attributes are sup-ported (e.g., underline and inversefeatures simultaneously)

2. Edit functions must be relearned(e.g. insert line, delete line, etc.).

3. Only one line of user key labels avail-able with 24- line screen (two linesavailable with enhanced graphicsadaptor).

A CADAM lookalike from ADAGEIt would be desirable to have a Table II. Feature comparison of IBM and ADAGE workstations.standalone graphics workstationcapable of running CAE/CAD/CAMsoftware without the need for a hostcomputer. Such a workstationshould be fast enough to providesub -second response time for CAD/CAM software. To conform to theVCAD philosophy of integration, atrained CADAM user should be ableto use the software on this worksta-tion without additional training.Such a device could be used toprovide CAD/CAM capabilities at aremote site without the added costof high-speed data communicationlines. It could also be used locally toincrease the number of users with-out increasing the load on the host.

For this purpose we evaluated aCADAM look -alike product fromADAGE.. The name of this productis CADRA, and we found it to bealmost exactly like 2-D CADAM inspeed and function. To the user,CADRA presents menu screensthat are very similar to CADAM andso fulfills the requirement for noadditional user training. Despite thisfact, however, the internal softwareand file formats are not CADAM.The unit is able to perform datatransfers with a host -based CADAMsystem via CADAM's geometryinterface module.

Since CADRA has the ability toperform data transfer over lowspeed phone lines, it is potentially akey product for the support ofremote sites. It also has local E -sizepen plotter capability and can store

FEATURE IBM5080

ADAGE6080

Refresh RateNon -Interlaced

Resolution

Pixel Writing Rate

Lightpen

Data Tablet

Transformation Dials

Element Highlighting

Number of Terminalsper Display Controller

Data TransferRates

PRANCE Compatible

CBDS Compatible

IBM 327X Compatible

Coax/Microwave/Fiber-optics/Telephone

60 HzNon -Interlaced

1024 x 1024

100 nanosec

No

Yes

Yes

Changes color

1

2 MBPS

Yes

Yes

Yes

60 Hz

1024 x 1024

50 nanosec

No

Yes

Yes

Intensifies element

1

3 MBPS

Yes

Yes

Yes

Yes/Yes/Yes/Yes Yes/Yes/Yes/Yes

This information was compiled from vendor data sheets and from CADCORRESPONDENCE (CADCOR, Inc.) Vol 1.2, p. 8, Nov./Dec. 1984.

data locally on diskettes. Our onlyreservation is our uncertainty aboutwhether the third party CADRAsoftware will remain compatible

with all future releases of CADAM.Table II gives a feature comparisonbetween the IBM and ADAGE work-stations.

40 RCA Engineer 31-2 Mar / Apr 1986

Apparently, few software vendorshave elected to follow the Graphics Ker-nel Standard in supporting the high -resolution displays. None of the soft-ware tested at VCDD would run on theProfessional Display after being confi-gured for the Enhanced Display. Thisprobably means that these programsaccess the EGAD adapter directly ratherthan as a virtual device through calls tothe GKS. The high -resolution featuresof the Enhanced and the Professionaldisplays are not accessible from BASIC3.0, which is supplied with the personalcomputer; however, they are accessiblethrough IBM's Professional Fortran.

At the fall meeting of the CADAMUser's Exchange in Chicago, CADAM,Inc. announced that early in 1986 theywould market a 2-D version ofCADAM that will run on the IBM PCAT. This CAD/CAM software will beable to take advantage of the high reso-lution and multiple color capabilities ofthe Professional Graphics Display. Aversion of this software-called Micro-CADAM-has been available in Japanfor some time. In VCDD we see thisproduct as a potential solution for ourneed to support CADAM at remotesites.

HP graphics terminal emulation on theIBM PC

Not all users of IBM PCs work strictlywith IBM PC graphics or IBM main-frame graphics. Within VCDD there area number of IBM PC users who need tobe able to display graphics intended forHewlett-Packard terminals. Untilrecently, these users had workstationsconsisting of two terminals; an IBM PCand an HP 2623 or 2393 graphics termi-nal.

During the Fall of 1985 the solution tothis graphics problem became available.Walker, Richer, and Quinn, a softwarefirm in Seattle that had been selling aPC software package that emulates theHP 2622/4 alphanumeric terminal,introduced Reflection 3, an HP 2623graphics terminal emulator for the PC.'This package will permit the IBM PC,PC XT, or PC AT to display graphicsthat were intended for the HP 2623 ter-minal.

As an alphanumerics emulator,Reflection 3 is highly compatible withsoftware written for Hewlett-Packardterminals. No bugs have been uncov-ered in the emulator to date at VCDD.

9

6

~6s

0 4

E3L.J

2

1

Emission Worm u p204 90 CM MOM 91A MOUNT

Ef - 6.40vG1 - -151v02 455v

: COE2 R 4&&.- C.0E2 - El 462v

COE2 G 470v

010 15 20

Time (seconds)

11111 I 1.41.414 11111111

*KG 2Z

30

Fig. 8. A graphics image produced by Reflection 3 on an IBM PC with a monochrome displayand Hercules card.

Ef -C1 = -gO2 - 45

: CCE2 = 456v7 -GOE2 = fl 4Pi2v

COE2 = G 475v

E

worlds)2

Fig. 9. A graphics image produced by ReflectionAdapter and Display.

The disadvantages listed can causeproblems in certain applications; how-ever, in general, they do not cause anyserious limitations.

Advantages:1. Terminal memory is limited only by

3 on an IBM PC with Enhanced Graphics

PC memory. 640k PC memory givesmore than 2000 lines of terminalmemory.

2. Provides spooled localusing the PC printer.

3. Easy storage of ASCII data to PCdisk.

printouts

Kunz/Miller/Renfro: IBM graphics workstations at Video Component and Display Division 39

Ef = - 5.46vG1 - -151vC2 455vCOE2 R 466vC0E2 13 462vCOE2 C 476v

Fig. 6. A sample graphics image on a Hewlett-Packard 2623A terminal.

Fig. 7. A graphics image produced by Reflection 3 on an IBM PC with standard color display

quality graphics requiring a wider rangeof colors, Video Show was selected. Italso supports the dual monitor worksta-tions, but provides a special piece ofhardware to drive the color monitor.This driver allows up to 1000 differentcolors to be displayed on a standardcolor monitor.

Color is also needed to distinguishinformation contained in very denseengineering graphics. Engineers quicklydiscovered that both Lotus and Stat-

graphics support this dual monitor con-figuration. The tables of numeric infor-mation being analyzed are presented onthe monochrome display, and thegraphic summary of this information issimultaneously presented on the colordisplay.

Enhanced graphics display

Late in 1984 IBM announced theEnhanced Graphics Adapter and Dis-

play (EGAD). This display producesalphanumeric characters in an 8 x 14dot matrix rather than the 8 x 8 dotmatrix of the standard color display. Itsresolution is sufficient to present rea-sonably crisp characters in 80 -columnby 25 -line format. In addition, it cansupport 16 colors with a graphics resolu-tion of 640 x 350 pixels compared toonly two colors at 640 x 200 pixel reso-lution for the standard color display. Inorder to support the full 16 colors theEGAD must be fitted with an optionalGraphics Memory Expansion Card andhave a total of 128k of display mem-ory.°' In 1985 we decided that all IBMPC ATs purchased would have en-hanced displays installed.

Most third party software vendorswere somewhat slow in supporting theEGAD; however, by about the thirdquarter of 1985 many of the major ven-dors had drivers available. Softwarethat does not explicitly support theEGAD board can still be run becausethe board is capable of emulating thestandard color display driver. DIPswitches accessible through an openingat the rear of the card are used to selectthe desired display mode.

Professional graphics display

Also in late 1984, the ProfessionalGraphics Adapter and Display wereannounced. It can support 245 simulta-neous colors from a palette of 4096. Ithas a graphics resolution of 640 x 480pixels and uses an 8 x 16 dot matrix forcharacters. The Professional GraphicsAdapter automatically emulates thestandard color display without the needto set DIP switches. It is switched intohigh resolution mode via software. Fewapplication programs are available todate that take advantage of the capabili-ties of this display, despite the fact that ithas been on the market for over a year.

Both the Enhanced and the Profes-sional displays were intended to beaddressed as "virtual devices" by appli-cations software. IBM markets a devel-opment package for programmers wish-ing to write applications for thesedisplays. The Graphics Kernel System(GKS) provides a large set of graphicssubroutines callable from high-level lan-guages such as Pascal and Fortran. GKStakes care of accessing the device driversrequired to execute the desired graphicscommands on most popular periph-erals.6


little firsthand experience with writinghost graphics programs. However, forthose who must write their own graphicsapplications, IBM has designed their5080 workstation to be compatible withthe Programmer's Hierarchical Interac-tive Graphics Standard (PHIGS).PHIGS is a proposed standard for 3-Dgraphics programming. It allows theapplications programmer to interfacewith the peripheral graphics devicethrough high-level commands withoutthe need to provide a customized devicedriver.

It is only necessary to list the widerange of application software that issupported to see the impact that this sin-gle workstation has on the engineeringcommunity. Of course, the main designtool used at VCDD is CADAM. Thevalue of CADAM has been dramati-cally demonstrated by the fact that inonly 24 months of operation we haveachieved a 5:1 increase in draftingproductivity-surpassing our originalgoal. In addition to this productivitygain, CADAM has enabled us to modelour production equipment to a level notpreviously attainable. Figure 3 shows anisometric view of a picture tube, con-veyor, and carrier assembly. It repre-sents a section of a large assembly draw-ing showing an automated work cell.The CADAM information panel is visi-ble at the top of the screen, and the taskmenu is visible at the bottom. Isometricsof this complexity could not be feasiblyattempted by manual methods.

We have extended our use of the sys-tem to include the area of finite elementmodeling with other software productssuch as CAEDS and ANSYS. Figure 4shows a 3-D wire frame model of a 26VCOTY picture tube. This model wasconstructed using CAEDS. Models suchas this can be overlayed with informa-tion about material properties andexternal forces, and then submitted to afinite element solver-such asANSYS-for analysis. Figure 5 is aCAEDS display of the results of such ananalysis. It shows a false -color contourplot of a picture tube mask -frameassembly. The colors represent theamount of displacement along the z-axisthat this mask will experience under agiven set of thermal conditions.

PRANCE, a printed circuit boarddesign module, is being installed toreplace our Applicon system. PRANCEwill be used for schematic capture ofmachine control ladder diagrams as

CADS 2.S: OUTPUT 01$PLAN VCAD

/11, mt. OvIIIISC/III S oill.114.O I

11,1L.A.1111.0 If It.Y

10-70t4-86 16,05,00

"rs III ?. 9041-e1

Y=

Fig. 5. False -color contour plot of a picture tube mask- frame assembly. Colors represent z-axis displacements of the assembly resulting from a given thermal distribution. See the colorphotograph on the back cover

well. Other applications covered byCADAM are numerical controlmachining, 3-D wire frame models, sur-faces, solids, 3-D piping, and facilitiesdesign and management. The StandardLibrary has proven to increase drawingspeed, accuracy and uniformity. We aremoving toward the increased use ofmacros for parametric design. Otheradvanced features available include hid-den line removal, silhouettes, kine-matics, and shading.

Personal computer graphics

Monochrome and color displays

The VCAD philosophy of purchasingrather than writing graphics softwareapplies to personal computers as well asto the mainframe. PC graphics pack-ages in use in VCDD include Lotus 1-2-3,Statgraphics, Graph Writer; Chart Mas-ter; Sign Master, Reflection 3 and VideoShow. Before our initial purchases wecame to the conclusion that most of our

PCs would be heavily used for manipu-lation of alphanumeric data, and thatthe standard color monitors available in1983 did not have adequate resolutionto justify their cost for this purpose.Therefore, we decided to purchasemostly monochrome displays and toadd Hercules cards for those users need-ing monochrome graphics capabilities.

As the number of PCs increased,however, it became clear that users needa facility for producing high quality pre-sentation graphics on their PCs and thatcolor capabilities were necessary. Eachengineering department was providedwith one personal computer havingboth a monochrome and a color moni-tor as well as a multi -pen plotter. Copiesof Graph Writer were provided with thisworkstation. Graph Writer has the abil-ity to support two monitors simultane-ously. As the user constructs his graphicimage, the alphanumeric informationand system commands are displayed onthe monochrome screen while thegraphic image being constructed can beviewed on the color screen. For high


Fig. 3. Complex CADAM isometric. This is a portion of a large assembly drawing of anautomated work cell. Isometrics of this complexity cannot feasibly be constructed by manualmethods. See the color photograph on the back cover.

GAMS 2.S. HOWL calualom ACAS le-TATiqa I5I32t20

OBUICT NEMO

clOISPLAY_OPTIOAS

L-LIMILOMAIIIMG.MOSEC-coorromus_roult..ma

.S.ATITy_SuS

LA-LASSILIPSsv-smaP -Immarm_or_couoilST -STATUS

SSA-SMAGi_mOSPASE

SGA-SGSTCH_GPOUP.AME

i-SACKSPOCt,-A AIM MIGAu

GLOSAL NEM

z x

Fig. 4. 3-D wire frame model of a 26V COTYpicture tube. This model was constructed usingCAEDS software. CAEDS serves as a pre- and post -processor for finite element solvers suchas ANSYS. See the color photograph on the back cover.

Another major difference betweenthe 3250 and 5080 workstations is in theterminal controller. In the 5080 system,each display has its own controller witha 32k buffer, so flicker is eliminated. Inaddition, the 5085 display controller canbe configured in a variety of ways tomeet the needs of the user and the appli-cations. Installable features includeexpansion video pixel memory for up to256 colors, local memory of up to 1.1megabytes, clipping and transforma-tion cards for fast local display proces-sing, 3270 emulation to allow the graph-ics terminal to operate as an alphanu-meric terminal, RS- 232C to supportlocally attached pen plotters, and dialsto control local zoom, pan, and rotate.The device can also perform as a remotenode with a V.35 attachment; however,due to the high information content ofvideo images, rapid data communica-tion speeds are required. The minimumcommunication rate recommended forremote operation is 56 kbps.

This wide range of options wouldseem to complicate the system designproblems, but instead it is the key to theworkstation's versatility and futuregrowth. A diskette drive in the 5085 unitprovides a means to configure the work-station and to collect error statistics.The terminal controller is attached bycoax to the IBM 5088 channel controllerup to three miles away, or alternatively,the 5085s can be daisy -chained to thatsame length with up to sixteen terminalsin the string.

The 5088 is channel -attached to theCPU and comes in three models. Thereare two locally attached models capableof supporting 16 terminals or 32 termi-nals and a remote model with the V.35communication option. The remoteoption allows the configuration of a sin-gle remote 5085 or a remote cluster of upto 16 terminals. The controller has adual address range to accommodate thegraphics devices, and it acts like a 3274to handle the 3270 addresses. Thisallows a single controller to contain 64system addresses. I/O contention canoccur on a heavily populated 5088 eventhough the CPU -to -5088 data exchangerate is 2.5 Mbps and the 5088 -to -5085rate is 2.0 Mbps.

Graphics software

As a result of our decision to purchaseas much software as possible, we have


chrome, local vs. host transformations,and data transfer rates are all importantconsiderations. However, we were will-ing to sacrifice, if necessary, some spe-cific features in order to achieve a highlevel of integration. A key factor inbuilding an integrated system, one thatis easily overlooked in the quest for evermore powerful features, is whether thegraphics software driving the device willpresent a consistent user interface for awide variety of solver algorithms. Theuser should not have to learn many dif-ferent sets of commands and menus inorder to use different tools such as 2-Ddrafting, 3-D drafting, FEM, electricaldesign, etc. It should be easy to use anyof the devices on the system to retrievegraphics data without learning a com-pletely new set of commands and inter-faces. This allows the managers, engi-neers, and designers to concentrate ontheir task rather than on operating thecomputer. By adhering to these princi-ples, VCAD has been responsible for amarked improvement in the communi-cation of engineering information andfor an 80 -percent decrease in the timerequired to produce an engineeringdrawing.

Our selection of graphics worksta-tions was artificially limited by our deci-sion to purchase an integrated system.Although several IBM plug -compatibledevices existed on the market, we ini-tially chose the IBM 3250 as our graph-ics terminal. This decision gave us a sin-gle vendor to rely on for both softwareand hardware, and opened the door toother goals as well. These workstationsfilled our needs for fast response time,significant increases in productivity, andthe ability to access a wide range of soft-ware tools from a single terminal. As iscommon in the rapidly changing com-puter market, the 3250 terminal becameobsolete within the first year of theVCAD project with the announcementof the IBM 5080 workstations. All sub-sequent purchases consisted of 5080devices.

IBM 3250 terminals

Currently, two families of terminals aresupported by CADAM. The originalsystem was configured with six IBM3251 terminals, which are monochromeand use vector display technology.Three terminals are attached to a singleIBM 3255 terminal controller. These ter-minals must be within fifty feet of the

controller. Two such clusters of three areconnected to a single IBM 3258 channelcontroller by coaxial cable at a distanceof 1000 feet. Finally, the 3258 controlleris channel- attached to the CPU.

At each 3251 workstation the user hasa program function keyboard, an alpha-numeric keyboard, and a light pen tointeract with the display. The display hasexcellent resolution but is prone toflicker due to the I/O contention causedby having three displays on the samecontroller and a 16k limitation on dis-play buffer size. Despite the flickerproblem, many users prefer to use thesedevices because they believe the interac-tive response achievable from a light penis faster than that of the 5080 tablet.

IBM 5080 terminals

The remaining 29 terminals are 5081 dis-plays. The configuration is similar to the3250 model, but with many additionalfeatures (see Figs. 1 and 2.). TheCADAM software is capable of han-dling any of the many hardware config-urations available, even interchangingthe devices across controllers if neces-sary. The major difference is that the5081 is a non -interlaced raster refreshdevice rather than a vector device. It is

available in a 50 -Hz monochromemodel or a 60 -Hz color version. We havefound that the majority of our userswho perform standard 2-D draftingtasks prefer the monochrome monitorfor this purpose. This choice allows usto pay less for the display, use less CPUresource per terminal, and supportmore devices.

Fig. 1. IBM 5080 graphics workstationconfiguration.

Fig. 2. IBM 5080 graphics workstation. The workstation consists of a 5081 CRT display, atypewriter keyboard (left), a cursor tablet (center), a lighted function key pad (right), and a5085 graphics processor (right, on the floor).


P.J. Kunz R.H. Miller G.M. Renfro

IBM graphics workstations at VideoComponent and Display Division

In 1983 there was a graphics revolution at VCDD. They decided theywanted to encourage engineers and designers to concentrate onthe product rather than on the computer, so they developed theVCAD system.

In 1983 a graphics revolution started inthe Video Component and Display Divi-sion. An IBM computer system runningCADAM software was installed to pro-vide CAE/CAD/CAM graphics toolsto the engineering community. Thiscomputer system became known by theacronym VCAD-video computer -aided design. The primary goals ofVCAD were to increase engineering anddesign productivity and to integrateexisting engineering hardware and soft-ware within the division. To achievethese goals the VCAD system included

Abstract: There are many types ofgraphics workstations available to theengineering community, and becausethere are so many choices, mostengineering organizations eventuallyfind themselves with several differentsystems that are not compatible. In1983 the Video Component andDisplay Division adopted a divisionalstrategy for future purchases ofCAE/CAD/CAM tools. This articledescribes our experience with IBMworkstations in engineering. Twogenerations of graphics are discussed,and four PC displays available fromIBM are compared. We are successfullyusing the IBM PC as a graphicsterminal communicating with aHewlett-Packard host.

©1986 RCA Corporation.Final manuscript received January 20, 1986Reprint RE -31-2-7

not only host -based graphics, but also alarge number of personal computerslinked by a local area network. The net-work serves the role of hardware inte-grator, allowing existing minicomputersystems as well as different types ofmicrocomputers to successfully com-municate. Details of the VCAD systemarchitecture and implementation philos-ophy were published in a series of artic-les in the May/June 1984 issue of RCAEngineer.i."

From the beginning IBM was selectedas the prime vendor of hardware andsoftware. At the time many engineersdid not consider IBM to be a leader inCAD/CAM graphics. Nevertheless,they offered a proven mainframe -basedsystem, a long-term strategy for engi-neering automation, and the range ofproducts necessary to achieve thedesired integration. This integratedapproach allowed us to build a systemthat is reliable, has fast response times,supports a large number of users, and iseasily taught to new users.

The initial target for design automa-tion was the equipment developmentorganization. This department containsour largest number of designers anddraftsmen. The ease with which this sys-tem could be expanded allowed us tostart simple and to rapidly expand intoproduct design, adding tools such asfinite -element modeling (FEM), 3-Ddesign, and printed circuit board design.Solids and NC machining modules arealso available from CADAM, Inc. Anorganized and intense training program

brought to each user the graphics skillsnecessary to successfully improve pro-ductivity.

Host graphics

System configuration

The VCAD host is an IBM 4381-02 with16 megabytes of real memory, 12 I/Ochannels, and a processing speed of 2.7MIPS. This very stable CPU runs 24hours a day, seven days a week. Periph-erals consist of 10 gigabytes of fixed diskstorage, two tape drives, two 3274 ter-minal controllers, HYDRA protocolconverters, and 5088 graphic worksta-tion controllers. As many as 166 userscan be locally attached to the system. Ofthis 166, up to 35 can be 3250 or 5080graphics workstations. An IBM 3725communications controller was recentlyadded to allow data communications tothe corporate computer center.

From the beginning our intention wasto purchase as much engineering soft-ware as possible and to avoid writingcustom routines. We wanted to encour-age our engineers and designers to con-centrate on the product rather than onthe computer. We chose CADAM fordesign and drafting, CAEDS as a pre -and post -processor for finite elementanalysis, and PRANCE for printed cir-cuit board design.

There are several key elements inselecting a graphics workstation. Dis-play size, resolution, color vs. mono-


the analytical equations to accommo-date changing hardware configurations.By running on our own PC instead of amainframe, we have kept costs low.Along the way, we have certainly dem-onstrated that there is a place for the PCin hardware reliability calculations. Wealso found we had to sacrifice speed forlower costs. A simulation run of 10,000simulated missions used to take a fewminutes (elapsed time, not CPU time)on the mainframe; on the PC, it cantake eight hours. However, once the PChas been purchased, whatever use youmake of it is free. The combination ofan owned PC, the APL programminglanguage, and graphics hardware andsoftware has proved to be extreme-ly cost-effective for RCA Astro-Electronics.

References

1. Solomon, J.G., "APL and engineering productiv-ity," RCA Engineer, Vol. 28, No. I (Jan./Feb.1983).

Table I. Current library of simulation programs.

Program

MCS/MO/

MCSIM02MCSIM03MCSIM04MCSIM05MCSIM06

MCSIM07

MCSIM08

MCSIM09

MCSIM10MCSIM11MCSIM12MCSIM13

MCSIM14

MCSIM15

MCSIM16

Configuration for which designed

Identical groups of identical units-e.g., 2G8X6. Any spare can beswitched to replace any active in the same group.2G11X8 with special switching restriction: all 3 units at the same end.2G11X8 with special switching restriction: any 3 of 5 at the same end.Groups of identical units, but the groups need not be of the same length.Two sets of different units, connected in series and switchable.All the units in any one group have the same failure rate, but differentgroups may have different rates, and the groups may be of differentlengths.Two sets of units, connected in series and switchable. The units in eachset need not be identical.Two sets of units, connected in series, and switchable according to adefined pattern. Each set consists of identical units.Identical groups of identical units, but some channels are premium, andsome are pre-emptable.One ring -redundant group of identical units.Identical groups of identical units, but the failure rate varies with time.One broken -ring -redundant group of identical units.Identical groups of identical units, but with a spare satellite to belaunched when any group runs out of spares.Identical groups of identical units, but with a spare satellite to belaunched at a preset time.Identical groups of identical units. Any spare can be switched to replaceany active in the same group, but there is a prescribed sequence forusing the spares, and the program keeps track of the switching history ofeach channel.Groups of identical units, but the groups need not be of the same length.Under specified conditions, a spare can be switched to replace a failedunit in another group.

Solomon: Spacecraft reliability predictions by computer simulation 33

The Statgraphics package

1.1

0.9

0.7

0.5

0.3

Fig. 1

0.02

-0.02

-0.06

-0.1

-0.14

-0.18

-0.22

0

Fig. 3

Plot of 1286X4

Plot of 5266X4M6

40111000.41111"ftlimay -----0.4.-14132F414.1111.1004rirdr

Avarantawair-.--

\AIL77 AthAilikItiftt

41111 411-0.1iLIL-

2

ilkar#441 irlr-r. 8

4,

JO12 I° e6

8

YEARS

In probing the "anomaly" men-tioned in the accompanying article,I generated an enormous amount ofdata. Many people have noted thatit is difficult to read tables of num-bers and comprehend theirsignificance -and it is. On the otherhand, I was most reluctant to plotseveral dozen graphs, each ofwhich had 120 points. It so hap-pened, however, that ReliabilityEngineering bought a graphicspackage and a color monitor -and,later, a five -pen plotter. The graphicspackage is called Statgraphics, andwas written by Dr. Neil Polhemus,then of Princeton University. As itsname implies, it is actually a statisti-cal and graphics package -and avery powerful one. It is alsoinexpensive -even though we haveused only about ten percent of itscapability, we have more than got-ten our money's worth from it. Stat-

4

(X 1000)

0

(X 1000)

0

0.14

0.11

0.08

0.05

0.02

-0.01

-0.04

Fig. 2

0.11

0.08

0.05

0.02

-0.01

-0.04

Fig. 4

graphics is written in APL, whichmakes it even more convenient touse with our library of programs.Statgraphics is menu -driven, is veryeasy to use, and does not requireany knowledge of APL. It is availa-ble from Statistical Graphics Com-pany, 2 Wall Street, Princeton, NJ08540, or from STSC Inc., 2115 E.Jefferson St., Rockville, MD 20852.In order to run Statgraphics, youmust have the APL interpreterAPL*PLUS, which is available fromSTSC at the above address.

The accompanying contourgraphs, drawn by Statgraphics, useas data the probability of survival ofeight groups, generated byMCSIM01 under different assump-tions. Figure 1 shows the probabilityof survival of eight groups in a2G8X4 configuration. This probabil-ity is denoted P6(2G8X4), and isshown over each year of a twelve -

Plot of Z268X4116

YExas

Plot of 72880416

YE,4ao

(X 1000)

(0 1000)

year mission. SFAIL, the failure ratefor standbys, is 0.1 x AFAIL, thefailure -rate for actives. AFAILranges from 1000 to 10,000 FITS(failures in 1070 hours). The otherthree graphs show the difference insurvival probability P8(2G8X4)-P6(2G8X6), for each year of atwelve-year mission, and the samerange of AFAIL. Figure 2 shows thedifference when SFAIL is0.1 x AFAIL. Figure 3 shows thedifference when SFAIL= AFAI L.Figure 4 shows the difference whenSFAIL = 0. As explained in thearticle, when SFAIL = AFAIL, thereis no penalty for the "pseudo -spares" in 2G8X6, and P8(2G8X6)remains higher than P8(2G8X4) forall values of AFAIL. WhenSFAIL = 0, the penalty ismaximized -over the course of themission, P8(2G8X6) drops belowP8(2G8x4), and remains below.


2G8X4 configuration to have eightamplifiers operating. Each group musthave four amplifiers operating. If onegroup loses five of its eight amplifiers,it doesn't matter how many amplifiersin the other group are capable ofoperating-the total number of amplifi-ers actually operating for both groupswill be less than eight. On the otherhand, the 2G8X6 configuration has theflexibility that one group can lose asmany as six of its eight amplifiers, andthe configuration will still have eightamplifiers operating until the othergroup has lost three of its eight. This wasnot exactly obvious-although it isnow.

One way of looking at the 2G8X6configuration is that it has two extra"pseudo -spares" in each group, as com-pared with the 2G8X4 configuration.However, these "pseudo -spares" arereally in use, and therefore they will failat the active rate, rather than at the(lower) standby rate. One would expect,then, that 2G8X4 would "catch up"with 2G8X6 over the course of a longmission-especially if the failure rate ishigh. And so it was: If the failure -ratewas as high as 7000 FITS (failures in onebillion hours), the cross -over point wasreached about eight years into a mis-sion. After that, P8(2G8X6)- the prob-ability of survival of 8 channels in a2G8X6 configuration-was less thanP8(2G8X4), and decreased more rapidlywith time than did P8(2G8X4). Thiseffect was even more pronounced withhigher failure rates.

In our simulation runs, we usuallyassume that the failure rate for standbyamplifiers is 0.1 times the failure rate foroperating amplifiers; that is, the shelflife is ten times as long as the operatinglife. If the explanation suggested aboveis correct, changing this ratio shouldsignificantly change the relationshipbetween the calculated reliabilities. Inparticular, if the failure -rate is the samefor standbys as for actives, then there isno penalty for having the pseudo -spares, and 2G8X6 should maintain thehigher reliability for the entire mission.On the other hand, if the failure rate forstandbys is 0, then the penalty for hav-ing the pseudo -spares is maximized.One can predict that the crossover effectwould be even more pronounced. As thegraphs in the sidebar show, all these pre-dictions were verified.

As the design groups at Astro-Electronics requested reliability predic-

RFINPUT

SMACONN

DROP -INCIRCULATOR

OuADRATUREPOWER

SPLITTER

FLC-02WF FLC- 08MA

POWERSPLITTER

FLC-I5MA FLC- 30MA

HFET- 2201 i CIRCULATOR

DROP -INCIRCULATOR

DC-81ASINTERFACE

PLUG

Solid state power amplifier.

tions for different kinds of configura-tions, it became necessary for us to writemore simulation programs. These pro-grams had the same basic structure, butreflected the different selection rules ordifferent linkages of the different con-figurations. Some of the changes wereminor, but some were drastic. Even so,making the changes usually requiredonly a minimal effort, because all theprograms are designed and imple-mented in APL. To give an idea of the

THICK -FILMBIAS

NETWORK

FLC -15MA

4-IPOWER

COMBINER

RFOUTPUT

SMACONN

OUTPUTCIRCULATOR

SUPPORTLUGS

scope of the current library of simula-tion programs, Table I shows the list ofprograms and the configurations theywere designed to handle.

ConclusionOur experience has shown that reliabil-ity calculations can be done usingMonte Carlo simulations instead ofanalytical solutions. Also, it is easier tochange the simulation programs than

Josef G. Solomon was awarded a BS degreein Mathematics from the University ofOklahoma while on active duty with the United

States Air Force. He first joined RCA in 1959,working on BMEWS at the Missile and SurfaceRadar Division. He was awarded the MA degreein Physics from Temple University in 1963. Aftertwo years of research in high-energy physics atthe University of Pennsylvania, he returned toRCA in 1966, joining the Systems Programminggroup of the Information Systems Division inCherry Hill, where he participated in theimplementation of the FORTRAN compiler forRCAs Spectra series of computers.

Mr. Solomon became an independentsoftware consultant in 1971. He joined theSystems Engineering group of GovernmentCommunications Systems in March 1982, andworked in digital signal processing, propagationof measurement errors, and networkvulnerability. In 1983 he transferred toAstro-Electronics, where he has worked inhardware reliability predictions, orbitalrendezvous, satellite eclipses, and space -basedradar.Contact him at:RCA Astro-Electronics DivisonPrinceton, N.J.Tacnet: 229.2185

Solomon: Spacecraft reliability predictions by computer simulation 31

Traveling wave tube amplifier.

replace the failed unit. The programkept a cumulative total of how manyamplifiers were still operating at the endof each year of each mission. Dividingthat set of numbers by the total numberof missions gave the fraction of missionsin which that many amplifiers sur-vived-and, hence, the probability ofsurvival. These probabilities were thencompared with probabilities generatedby solving the coupled differential equa-tions.

This comparison showed that theMonte Carlo simulation method gaveessentially the same answers as the exactanalytical solution. Our experience hasshown that simulation is indeed fasterand cheaper than the analytical solu-tion. In addition, using simulationmakes it much easier to accomodatechanges in the hardware configuration.

Rewriting the program in APIThe original program was written forthe Reliability Engineering group by anoutside consultant who had departedbefore my arrival on the scene. It wasdesigned for a configuration in whichany spare could be switched to replaceany failed active amplifier in the samegroup; it was not possible to switchamplifiers from one group to another.Any configuration with those character-istics could be simulated by that pro-gram. The program was written in

Apple BASIC, and simulation runs hadbeen made on the consultant's com-puter. The size of the configuration wasbuilt into the program. This meant thatseveral lines of the program had to bechanged each time the program was run.Also, the consultant had not suppliedany program documentation. Finally,the program did not work on the Applecomputers at Astro-Electronics, andmaking it work was my first assign-ment. In order to be able to debug theprogram, it was first necessary to under-stand it. Because APL is the language Iprefer, my explanatory notes (to myself)on the program were written in APL.The interesting result of this effort wasthat, by the time I understood theBASIC program, a replacement APLprogram had thus been brought intoexistence. Results obtained from thenew APL program agreed closely withresults previously obtained from theBASIC program-keeping in mind thatthe results will never agree precisely,because of the statistical nature of theMonte Carlo method.

The fact that the operational programwas in APL, rather than BASIC orsome other conventional language, con-ferred unanticipated benefits almostimmediately. Very soon after the dem-onstrated success of that program, wewere confronted with configurationsthat the program could not handle: dif-ferent switching restrictions within a

group, groups with different numbersof amplifiers, groups composed of dif-ferent types of amplifiers, etc. Torespond to these new challenges, newsimulation programs were designed andimplemented. At present, the simula-tion library contains over a dozen pro-grams, each designed to handle a differ-ent type of hardware configuration. Asthe "traffic" became quite heavy, thecost of running these programs on amainframe became prohibitive. Wetherefore purchased a PC (and then sev-eral more), and have been running thelibrary on PCs for the past two and ahalf years. In addition, we have pur-chased a graphics package, and we cannow plot graphs of the results from indi-vidual simulation runs, or contourgraphs from a collection of runs.

PC or mainframe?

The specific event that triggered thedecison to purchase a PC instead of con-tinuing to rent time on the in-housemainframe arose from an apparent"anomaly" in our simulation results.Comparing the results for two differenthardware configurations showed that

program predicted alower reliability for the configurationthat seemingly should have had a higherreliability. Attempts to resolve theanomaly consumed a great amount ofmoney on the mainframe, and yet it wasessential to resolve it, because the relia-bility numbers generated by the simula-tion programs are a critically importantinput in the design process. If thosenumbers are wrong, we have to know itnow, and correct them. So we continuedon our new PC, and resolved the prob-lem. Here is a description of the prob-lem and its resolution.

If you compare reliabilities for a con-figuration described as 2G8X6 and onedescribed as 2G8X4, which configura-tion has the higher probability that eightamplifiers will still be operating after,say, 10 years? It certainly seems that the2G8X4 configuration, which has fourspares in each group, should have ahigher reliability than the 2G8X6, whichhas only two spares in each group-andyet 2G8X6 consistently came out higherthan 2G8X4. Does one increase reliabil-ity by decreasing the number of spares?Varying the parameters at our disposalbegan to show a pattern, and the answeremerged. The apparent anomaly arisesbecause there is only one way for the


J.G. Solomon

Spacecraft reliability predictionsby computer simulation

Satellite hardware reliability simulations are being run on PCs at

Astro-Electronics, and the payoffs are reduced costs and increased

flexibility.

The Reliability Engineering Depart-ment at Astro-Electronics is chargedwith the responsibility for predicting thereliability and survival probability, ofcomplex satellite hardware. It is essen-tial to know the probability of survivalof electronic circuits over the entire life-time of a satellite. This information isthen used, for example, in deciding howmuch redundancy to build into a partic-ular section of the circuitry. Because of

Abstract: The equations used fordetermining the reliabililty of satelliteelectronic circuits are a set of coupleddifferential equations, and solvingthem can be very expensive in bothtime and money. Accordingly,engineers at Astro-Electronics wrote aMonte Carlo simulation program inAPL for a particular electronic circuitand switching configuration. Very soonafter the demonstrated success of thatprogram, configurations were designedthat the program could not handle. Torespond to these new challenges, newsimulation programs were designed andimplemented. As the "traffic" becamequite heavy, the cost of running theseprograms on a mainframe becameprohibitive. Therefore, several PCswere purchased, and the library ofprograms has been running on themfor two and a half years.


the great cost of a failed mission, RCAwants to build in "enough" redun-dancy. But because of the great cost ofputting any weight into space, it is alsoimportant not to build in more redun-dancy than is necessary to meet the mis-sion reliability specifications. For agiven set of performance criteria, then,the problem is to calculate how muchredundancy should be employed. In thecontext of this paper, however, the prob-lem is a different one. Solving the set ofcoupled differential equations used todetermine the reliability of electroniccircuits can be very expensive in bothtime and money. Accordingly, we had tofind an alternative method of determin-ing the reliability of complex systemsone that is faster and cheaper than directsolution of the differential equations,and acceptably accurate. Inasmuch asour attention had already been drawn tocomputer simulation methods, ourproblem was even more specific: Cancomputer -simulation satisfy the abovecriteria?

Monte Carlo simulation

A Monte Carlo simulation program waswritten for a particular-and particu-larly simple-electronic circuit andswitching configuration. The resultsagreed closely with results obtained by"brute -force" methods. The configura-tion was an array of identical traveling -wave -tube amplifiers, arranged intogroups that were physically and elec-

Solid state power amplifier.

tronically separated. Each group con-tained the same total number of amplifi-ers, the same number of active ampli-fiers, and (of course) the same numberof spares. A typical configuration mightbe described as "2G11X8", meaningthat there are two groups in the configu-ration, and that each group has 11amplifiers, of which eight are active.Given the failure rate for active amplifi-ers and the failure rate for spares, theprogram simulated a large number ofmissions in which amplifiers failed atrandom times in accordance with thosefailure rates. When an active amplifierfailed, the program attempted to find anavailable spare in the same group; ifthere was one, it was switched in to


RCA Astro-Electronics must determinethe reliability and survival probability ofcomplex hardware to be installed onsatellites. The equations used to predictthe reliability of electronic circuits are aset of coupled differential equations, andsolving them can be very expensive inboth time and money. Accordingly, aMonte Carlo simulation program waswritten for a particular electronic circuitand switching configuration. Theprogram was written in APL, of course.'The results agreed closely with resultsobtained by "brute -force" methods. Verysoon after the demonstrated success ofthat program, we were confronted withconfigurations that the program couldnot handle. To respond to these newchallenges, new simulation programswere designed and implemented. Atpresent, the simulation library containsover a dozen programs, each designed tohandle a different type of hardwareconfiguration. As the "traffic" became

quite heavy, the cost of running theseprograms on a mainframe becameprohibitive. We therefore purchased a PC(and then several more) and have beenrunning the library on PCs for the pasttwo and a half years. In addition, we havepurchased a graphics package, and wecan now plot graphs of the results ofindividual simulation runs, or contourgraphs from a collection of runs.

Our experience has shown thatreliability predictions can be done usingMonte Carlo simulations instead ofanalytical solutions. Also, it is easier tochange the simulation programs thanthe analytical equations toaccommodate changing hardwareconfigurations. By running the programson our own PC instead of a mainframe,we have kept costs low. Along the way,we have certainly demonstrated thatthere is a place for APL and the PC inpredictions of hardware reliability.

28

designer the capability to place androute circuit components using a rela-tively inexpensive personal computer.The ability to use any path width andpolygon shape, to combine paths andpolygons to make cells, and then tocombine cells to make circuits reducesthe layout time and the time it takes toredo circuits. Once you find that thefirst board isn't quite what you wanted,the task of changing the circuit is simple.Figure 3 was drawn by Robert M. Evansof the Signal Conversion Group at RCALaboratories. Bob has had no real pre-vious experience with circuit board lay-out systems other than a graphics rout-ing he wrote for our HP 1000. He wasable to pick up the system after a fewhours work with me, and over the pastthree months developed this workingboard.

Acknowledgments1 wish to acknowledge the assistance ofSue Handman of CISS in Cherry Hillfor the help she has given our group inevaluating CAD packages, and for find-ing this one. I would also like to thankBernie Wiegand of the Design Automa-

TrademarksThe following is a list of trademarks used inthis issue of the RCA Engineer. Not shown arethose trademarks that are properly cited withinarticles.

VAX is a trademark of Digital EquipmentCorporation

MicroVAX I is a trademark of Digital Equip-ment Corporation

DECmate is a trademark of Digital EquipmentCorporation

DATAIRIEVE is a trademark of Digital Equip-ment Corporation

IBM PC is a trademark of International BusinessMachines Corporation

IBM PC AT is a trademark of InternationalBusiness Machines Corporation

IBM PC XT is a trademark of InternationalBusiness Machines Corporation

Apple is a registered trademark of AppleComputer, Inc.

Macintosh is a trademark of Apple Computer,Inc.

Lisa is a trademark of Apple Computer, Inc.

MacDraw is a trademark of Apple Computer,Inc.

Tom Marett an Associate Member, TechnicalStaff in the Television Research Laboratory atRCA Laboratories. He previously worked in thePicture Tube Systems Research Laboratory atRCA Laboratories, and in the TechnologyTransfer Laboratory, Magnetic DeflectionGroup, Lancaster, Pa. His professional interestsinclude rf, magnets, CAD and CAM. He has aBS from Spring Garden College, Philadelphia,Pa.Contact him at:RCA LaboratoriesPrinceton, N.J.Tacnet: 226-2278

\

tion group at Missile and Surface RadarDivision for giving us a demonstrationof PERSONAL870, and for his help inordering the hardware needed; PhilDeem of the New Product Laboratoriesfor his assistance in getting us startedwith the software; Bill Kimble of RCA

MacProject is a trademark of Apple Computer,Inc.

LaserWriter is a trademark of Apple Computer,Inc.

MacTerminal is a trademark of Apple Computer,

Inc.

MacWrite is a trademark of Apple Computer,Inc.

MacPaint is a trademark of Apple Computer,Inc.

Imagewriter is a trademark of Apple Computer,Inc.

Intel is a registered trademark of IntelCorporation

Supercomp-Twenty is a registered trademark ofAccess Technology, Inc.

Jazz is a Trademark of Lotus DevelopmentCorporation

Lotus 1-2-3 is a registered trademark of LotusDevelopment Corporation

Microsoft is a registered trademark of MicrosoftCorporation

ANSYS is a trademark of Swanson AnalysisSystems, Inc.

Ethernet is a trademark of the Calma Company

Laboratories for first teaching me theApplicon 4000 system, and then inter-facing it with the PERSONAL870; andEd Cuomo and Greg Zak, Process Tech-nology, RCA Laboratories for their helpwith PREP on the Cherry Hill CMSsystem.

Wordperfect is a trademark of SSISoftware

Harvard Total Project Manager is a Trademarkof Harvard Software, Inc.

dBase III is a trademark of Ashton-Tate

PDP-11 is a trademark of the Digital Equipment

Corporation

PDP-8 is a trademark of the Digital EquipmentCorporation

CADAM is a trademark of International Busi-ness Machines, Inc.

CAEDS is a trademark of International BusinessMachines, Inc.

PRANCE is a trademark of Automates Systems,Inc.

Statgraphics is a trademark of the StatisticalGraphics Corp.

Graphwriter is a trademark of Graphic Com-munications, Inc.

Chart -Master is a trademark of DecisionResources, Inc.

Sign -Master is a trademark of DecisionResources, Inc.2DOMAIN is a trademark of Apollo Computer,Inc.

Marett: Analog printed -circuit board design using an IBM PC AT 27

Fig. 1. Example of a component cell.

to another. This allows many designersto work on individual circuits and thenassemble the pieces into one workingfile for a final board.

Board design

Our group is using two sets of the soft-ware, one on an IBM PC AT and one onan IBM PC XT. The PC is not perma-nently changed, so it can still be used forother tasks (writing this paper, forexample). The sidebar on this page liststhe hardware requirements for PER

Three methods are used for creatingPC Boards from the drawing files in ourIBM PC AT The first two methods usethe IBM mainframe in Cherry Hill(CMS). The first of these gives us fastturnaround time because it produces animmediate print on a thermal printer.Resolution is low, 300 dots -per -inch,and we have a size limit of 3.5 inches by3.5 inches with this method.

Fig. 2. Developmental circuit drawn using cells.

IaJ

1#1:r! 41:54. ;

I

Iltl

eaVid.747..iltis.;01

Mee

Fig. 3. Completed circuit.

The second method uses a Gerberplotter driven by software on CMS, andgives us 1000 dots -per -inch accuracy.This software, called PREP, creates astandard Gerber -formatted tape fromplot files created by PERSONAL870software.

The third method is to transfer thedrawing file from the IBM PC AT to anApplicon system. We use the Applicon4000 system in the Microwave Technol-ogy Center at RCA Laboratories, whichalso has a resolution of 1000 dots -per -inch.

Software has been written to checkfor design problems in drawing filessuch as unclosed polygons. It highlightsproblem areas so they can be moved offto the side of the page, corrected, andmoved back in place.

Most components used in our groupare surface -mount, so we create cellsconsisting of polygons or pads forattachment to the board. One such cellis shown in Fig. 1. Active devices used at40 MHz and up do not normally havestandard packages, so we create what-ever pad configuration is needed for theindividual device. These cells are puttogether to make developmental circuitslike the one in Fig. 2. Figure 3 is anexample of a complete circuit that wasrecently developed.

Conclusion

The powerful (yet inexpensive)computer -aided design software used bythe Signal Conversion Group at RCALaboratories gives the analog circuit

The following is a list of thehardware necessary to runPERSONAL870:

1.1BM PC AT or XT (or compatible)

2. At least 640k of memory

3. Two serial ports

4. Printer port

5. Display controller

6. Display (color or monochrome,high -resolution preferable)

7. Pen tablet


T Marett

Analog printed -circuit board designusing an IBM PC AT

A group at RCA Laboratories has found a software package that isspecifically intended for the computer -aided design of analogprinted -circuit boards.

Digital printed -circuit board designsystems, with their elaborate schematiccapture, placement, and routing rou-tines, are of no real use to the analogcircuit designer without heavy modifica-tion of the algorithms controlling theplacement of components.

Analog circuits used in television tun-ers, for example, operate at frequenciesabove 40 MHz with low-level signals.These circuits amplify and change thefrequency of the incoming signals sothat they may be displayed by the TV

Abstract: Analog circuits operating atfrequencies above 40 MHz cannot takeadvantage of the elaborate placementand routing routines that have beendeveloped for low -frequency circuitry.The powerful and expensive

printed -circuit board design packagesthat need a medium to large computerare therefore wasted, since the designdemands all of the circuit designer'sattention. The placement of eachcomponent and the routing of eachpath between components is critical tothe performance of the circuit. TheComputer -aided design software usedby the Signal Conversion Group atRCA Laboratories allows the designerthe freedom of individually placing thecomponents while operating with anIBM PC A7; keeping the cost of thesystem down below $20,000. Thissoftware will also run in an IBM PCXT and many IBM compatibles.


set. This signal amplification and con-version must be done in a small area freefrom interaction with the rest of the TVset. The placement of the individualcomponents of the tuner is critical. Leadlength of components such as capaci-tors, inductors, resistors, and activedevices will determine the frequencyresponse of the overall tuner. Extra leadlength will also act as an antenna, radi-ating unwanted signals throughout theTV set. The exact placement of eachcomponent so as to minimize leadlength and unwanted radiated signalsbetween circuits is left up to the designer.The actual width and length of eachtrace, the shape of the pads connectingthe components to the board, and anyground plane used in and around thecircuit will determine the characteristicsof the tuner as much as the componentsthemselves.

The Computer -aided design softwareused by the Signal Conversion Group atRCA Laboratories is a simple two-dimensional system capable of describ-ing multilayer printed -circuit boardswith any shape polygons and paths oneach layer. It also can output to severaltypes of high -resolution printer/plottersfor fabrication of the board.

The software used by the Signal Con-version Group at RCA Laboratories iscalled PERSONAL870, and can be pur-chased from Automated Images Inc.,Woburn, Ma. PERSONA L870 emu-lates the Applicon 870 and the Applicon4000 2-D printed -circuit board layoutsystems. The limits for the physical sizeof the boards are the same as those forthe Applicon systems, and you can work

in mils or millimeters. This software isalso used by New Product Laboratoriesin Indianapolis and Missile and SurfaceRadar Division in Moorestown, whichmakes it possible to transfer drawingfiles to and from these locations. If youare familiar with these systems you willbe able to use this software with littleassistance. I am familiar with the Appli-con 4000 system, and was able to usePERSONAL870 system after a fewhours of reviewing documentationobtained with the software. Three otherpeople in the group have picked up theuse of the system in a short period oftime.

Most of the commands of PER-SONAL870 are the same as those usedby the Applicon software. Commandsare communicated to the computer bypen strokes or with a pen table. We haveadded the IBM PC AT function keys tosupplement the pen strokes and tabletmenus. The user sets up an account witha start file consisting of personalizedpen stroke commands, menu, and com-mand set. There is also a definitions sec-tion that has a group of cells used tocreate a new board. These cells arecapacitors, resistors, inductors, activedevices, and combinations of cells thathave been drawn to the correct physicalsize. A group of cells making up a work-ing circuit can also be set up as a cell.This is how most of our circuits aredeveloped-small sections are made,and the cells are then connectedtogether. Individual cells, wholelibraries, whole drawings, pen strokelibraries, and individual or whole com-mands can be copied from one drawing

Marett: Analog printed -circuit board design using an IBM PC AT 25

5. RADC Thermal Guide for Reliability Engineers, RADC TR-82-122, June1982.

6. Yeh, L.T., "CHIC-Convective Heat Transfer," RCA Corporation, Govern-ment Communications Systems, Camden, N.J. September 1984.

Authors, left to right: Piazza, Miller, Butt, Hoke.

Stephen Butt, Manager, Mechanical Computer -Aided Design (MCAD),joined RCA in April, 1984. He worked in Hardware Integration and DesignAutomation before joining the MCAD group. He received his BSE in 1976from the University of Pennsylvania (Bio-Engineering), a BSME fromDrexel University in 1982, and an MSME from Drexel University in 1984.Contact him at:Communication & Systems DivisionCamden, N.J.Tacnet: 222-2415

Harvey R Miller received his BS in Mechanical Engineering fromRensselaer Polytechnic Institute, an MS in Mechanical Engineering fromMassachusetts Institute of Technology, a M. Phil in Plasma Physics fromColumbia University, and will complete his PhD in Applied Physics fromColumbia University in the spring of 1986. He joined RCA GovernmentCommunications Systems in 1984 as a Senior Member of theEngineering Staff. Mr. Miller has been responsible for computer -aidedthermal design involving computational heat transfer and fluid dynamics.

He has developed a front -ended computer -aided design general purposeheat -transfer program for use on the RCA CISD Auto-trol engineeringworkstation to allow interactive graphics for 3-D model generation andtransient thermal analysis.Mr. Miller is a member of the American Society of Mechanical Engineers,American Physical Society, Society for Industrial and AppliedMathematics, American Nuclear Society, and Tau Beta Pi and Pi TauSigma Honor Societies. Mr. Miller was a Raytheon Fellow from 1966 to1967 while attending MIT and has been elected to Who's Who inFrontiers of Science and Technology for the past two years. He hasnumerous technical publications in Computational Fluid Dynamics andHeat Transfer, and is a registered Professional Engineer in New Jerseyand Pennsylvania.Contact him at:Communication & Systems DivisionCamden, N.J.Tacnet: 222-4896

Frank M. Piazza, Software Engineer, Design Automation, is currentlypursuing a BS in Computer Science at Rutgers University. Mr. Piazzacame to RCA CISD in 1982, joining Engineering Computer -Aided Design.Mr. Piazza is responsible for CAD/CAM integration and softwaredevelopment supporting automation.Contact him at:Communication & Systems DivisionCamden, N.J.Tacnet: 222-4851

David N. Hoke is Senior Member, Engineering Staff, Design Automation.He received a BS in Industrial Management from Georgia Institute ofTechnology. He joined RCA CISD in 1984, and has been responsible forthe CAD/CAM integration effort within engineering and specifically forinitiating the cable harness automation effort. Mr. Hoke is also anassociate member of the IEEE.Contact him at:Communication & Systems DivisionCamden, N.J.Tacnet: 222-4610


(a)

(b)

(c)

Fig. 5. The designer constructs cable harness routesthrough the 3-D model (a), then forms and displays thecable harness (b and c).

locally on a floppy disc and transferred to the VAX 11/750where it will be accessible to those users specified and con-trolled by Engineering Documentation. In addition, softwarewas completed for the transfer and reformatting of thesesame files for manufacturing engineering, where editing maytake place. This information may then be used by the factoryin the manufacture of the cable harness.

Design

SpecificationConceptual

Design

Previous

Concepts

Hardware

Design 8 Detail

Drawings

1 1

Engineering

Finite Element

Analysis

Historical

Database4

Released

Drawings

NC Programs &

Cut Metal

Cable Harness

Design

Assembly

Instructions

Tool 8. Fixture

Design

Fig.6. The information flow in the mechanical designprocess. Each transition represented different utilizationof the CAD database.

Software to aid in the automation of cable harness designhas been developed and implemented on the Auto-Trol CADsystem. This software interfaces with Auto-Trol's CAD sys-tem, which allows the design of a 3-D assembly. The model ofthe assembly sets the stage for the automation of cable har-ness manufacture. The designer can readily see possible pathsor positive routes for the main trunk of the cable harness, andthen constructs these paths via the Auto-Trol CAD systemwithin the 3-D model illustrated in Fig. 5a. At this point, thedesigner executes the option for "visual inspection," and thecable harness bundle is formed and displayed as shown inFigs. 5b and 5c. Here each signal is routed between connectorpins as well as the wire radii accumulated for cable bundling.Next, the "report generator" option may be selected in orderto calculate lengths, resistances, and weight for each wire.Further development is required for the implementation ofthe flattening of the 3-D model for manufacturing and forthe assessment of connector and tooling data.

ConclusionTo realize the benefits of a CAD system, information must beavailable from all the areas that would normally interfacewith the using activity (Fig. 6). At CISD, the original pur-chase of mechanical design CAD equipment was imple-mented in a Design Engineering/Drafting group. As oursystem evolved, we soon realized that in addition to creatingproduction hardware drawings in an efficient manner, thesystem enhances communication between various depart-ments involved in the mechanical design process. By using ashared 3-D database, we are eliminating the regeneration ofgeometric data, which reduces the possibility of error andtherefore improves the design cycle. This approach willimprove RCA's competitive position in the communicationsand information systems field.

References1. Miller, H.P. and Wolfe, D.B., "Computer Aided Thermal Analysis and Design

of Complex Electronics Equipment," ASME paper 85 -DE -E-1.2. MIL-HDBK-217, "The Reliability Stress and Failure Data for Electronic

Equipment," U.S. Department of Defense.3. "SINDA-Systems Improved Numerical Differencing Analyzer," National

Aeronautics and Space Administration, Houston, Tex., 1980.4. Weiss, M.E, MOTHA-"Moorestown Flow and Thermal Analyzer," RCA

Corporation, Missile and Surface Radar Division, Moorestown, N.J., October1982.

Butt/Miller/Hoke/Piazza: Comprehensive application of a CAD/CAM system to mechanical design 23

Fig. 4. An iso-stress plot from a finite -element model of astructure.

another graphics input system. There is also a wide range ofautomatic mesh generation routines embedded in the soft-ware that provide such features as linear or controlled loga-rithmic volume fills of elements between defined surfaces.All of these features provide the ability to create a complexfinite -element model for input to ANSYS.

With the model complete in Series 7000, a menu is selectedto produce a finite -element file. This file is converted intoANSYS format using an ANSYS module (AUX15), and theengineer enters ANSYS with an intact geometry model. Oncein ANSYS, the stress -behavior environment must be speci-fied. The type of analysis is chosen (static, modal, lineardynamic, etc.); the applied forces, pressures or constraintsare entered; and any physical or material properties associ-ated with the elements are specified. The ANSYS solutionphase is then executed and produces superimposed displace-ment plots, color stress contour plots (Fig. 4), and tabulateddisplacements and stresses (at numbered nodes and ele-ments). Sorts can be done to list critical values exceeded in themodel. With this tool an engineer can "tweak" a design toachieve optimum behavior and raise the level of confidenceby subjecting more designs to a greater detail level of analysisin a shorter period of time.

Manufacturing applicationsThe next phase in the development of mechanical designs isto have the parts fabricated and assembled. Again, the 3-Ddatabase can be used by a number of manufacturing engi-neering functions. For fabricated parts, the CAD softwarecontains modules for flat pattern and punching as well as NCmachining.

Numerical control

The majority of the hardware designed and manufactured atCISD is for the packaging of electronic communicationsequipment. The types of packaging fall into two categories;sheet metal chassis and machined casting housings. The sheet

metal chassis designs are generally folded box structures withnumerous cut-outs for connectors, purchased electrical com-ponents, and assorted mechanical mounting hardware. Man-ufacturing engineers can use the flat pattern software togenerate a flat pattern layout of a part simply by indicatingwhere the bend lines are and entering the bend radius andmaterial thickness. This layout is generated with all of thecut-outs in their proper locations. The punching software canthen be used to create a command file to drive an NC punchto "knock out" the part. Time is saved by eliminating redun-dancy (the manual entry of geometry from drawings) andreducing chances for error.

In the "cast chassis" area of design, tooling changes areexpensive, and extensive checking is needed to eliminatedesign mistakes. With the NC software, there is a relativelyshort turnaround time for mechanical fit verification modelproduction. Here, an NC programmer can generatetoolpaths in an interactive session on the CAD workstation(using the part that is already in the database) in one tenth thetime of manual toolpath programming methods. This proce-dure can also be applied when schedule crunches on a pro-gram cannot tolerate the long lead times involved in thepurchase of castings.

Process and assembly engineers also benefit from thedesign information in a CAD database. In our assemblyarea, process engineers routinely make copies of productiondrawings, cut and paste parts of these drawings and add textto direct the assembly operation. Redisplay and formattingfunctions that exist in the CAD software are already tailoredto accomplish this easily.

Cable harness

The cable harness design process is labor intensive, complex,and detailed. It is for this reason that CISD has pursued thisCAD application.

In the past the electrical engineer, with the assistance of amechanical engineer, produced a sketch that defined criticalpaths for the cable bundles and connector locations. Thissketch was then delivered to drafting, which produced a wirelist for the pin connections and a detailed schematic. All cableharness pin connections and cable routing information werethen manually typed and reviewed by engineering for accu-racy. The schematics were distributed to manufacturing engi-neering along with a wire connection list. Here the flat pat-tern is developed. In some cases this is accomplished byfitting the cable harness to the rack cabinet. This process islabor intensive and would best be solved through engineeringvia a CAD system rather than through manufacturing byhand.

After manually reformatting all of the above information,it was then sent to MIS to be entered into their materialsinventory system and a final hardcopy was returned for useby manufacturing engineering. The process described waslengthy and, because of human intervention, the error ratewas high.

We first concentrated our efforts on establishing a linkbetween engineering and manufacturing. This became possi-ble with the installation of the Ungermann-Bass local areanetwork. Also, software was designed to assist the draftsmanin entering the wire connection list and formatting the infor-mation for the customer. Upon completion, the list is stored


(c)

Fig. 3. (a) Shows the wireframe outline of a fiber-optictransceiver; (b) is a preliminary RC thermal model of 3-Dinternal grid conduction; (c) is an evolved RC thermalmodel of 3-D conduction including the transceiver lid andbase coldplate; and (d) is the complete RC thermal modelof a fiber-optic transceiver.

CHTC convection resistors (for convection in fluids thatare not presently in PROPTY);

6. "Nodal" calculation of free and forced convection fromrespective grid nodes to ambient nodes, and internal con-vection;

7. "Nodal" calculation of direct radiation from respectivegrid nodes to ambient nodes, and re -radiation to adjacentgrid nodes.

The operation of GCSMOTHA involves:

1. Construction of a physical model in 3-D (so-called "wire -frame");

2. Execution of the Finite -Element Modeler for creatingnodes and elements;

3. Response to menu -driven screen prompts for creating theinput data file in Auto-Trol's graphics system Series 7000;

4. Running GCSMOTHA using the graphics -generated datafile;

5. Providing transient and steady-state temperature outputin matrix format.

A thermal problem recently solved by GCSMOTHAinvolved a high -resolution discretization of a fiber-optictransceiver. Figure 3a illustrates the wireframe of the pro-posed package design. A wireframe is first constructed sothat all surfaces are visible for executing the Finite -ElementModeler. The thermal model is constructed employing thevitally important techniques of "color coding" and levelmanagement, so that resistor elements corresponding to vari-ous modes of heat transfer can be effectively differentiated.Figure 3b shows the FEM-constructed thermal model thataccounts for 3-D internal conduction on the grid surfaces.The linear conduction coupling in this high -resolution ther-

(b)

(d)

mal model can be readily observed. Note that heat -dissipating electronic components were carefully distributedin three separate circuit modules. Figure 3c illustrates thethermal model evolved one step further, accounting for con-duction to the transceiver lid and to a base coldplate. Figure3d shows the complete RC thermal model including theeffects of internal and external convection, ambient radia-tion, and the use of special thermal resistors accounting forperpendicular conduction (through several different materi-als) and lateral boundary conduction from the coldplate tothe chassis.

The fiber-optic transceiver model was successfully runwith GCSMOTHA, and a design change was modeled andanalyzed within hours using this CISD CAD tool on an Auto-Trol engineering workstation. It should be noted that thecomplete thermal model used over 200 nodes and approxi-mately 700 thermal resistors. The GCSMOTHA input datafile was created by direct console prompting within one hour.It is expected that this CISD CAD tool will yield both muchhigher resolution RC thermal models, when deemed neces-sary, and significantly reduce thermal design turnaroundtime.

Structural analysis-ANSYSThe other area of engineering concern in mechanical design isthat of structural integrity. At CISD we have ANYSYS (awidely used finite element and model solver) input capabilityfrom any of the designs created on the CAD workstation. Inan interactive session, an engineer can create a mesh modelfrom existing geometry using a wide variety of ANSYS-supported elements (2 -node beams, 4 -node plates, 8 -nodeisoparametric solids, etc.). By working in the environment ofthe Series 7000 software, the engineer has a greater number ofgraphics tools at his disposal and doesn't need to learn


(a)

(b)

(c)Fig. 2. Detailed design progression: (a) shows the de-signer's layout (note the utility of color part definition); (b)shows a part "cut" from the layout; and (c) shows a pageof the fully dimensioned product drawing.

electronic components increases approximately exponen-tially as a function of increasing temperature. The designer ofan electronics assembly is typically faced with the problem ofmaintaining a semiconductor junction temperature below125°C. In fiber optics applications, light emitting diodesshould not exceed 64°C for proper operation. Hence, theneed for CAD for thermal modeling and analysis becomesapparent.

Many computer programs such as NASA's SINDA (Sys-tems Improved Numerical Differencing Analyzer)' andindustry codes such as MOTHA (Moorestown Flow andThermal Analyzer)° are presently available for general-purpose heat transfer analysis. Although these programs arefast and accurate, they do require extensive thermal modelingand very carefully prepared data input files for the effectiveuse of the respective computer codes. The RADC ThermalGuide for Reliability Engineers' estimates that a heat transferspecialist with at least five years of experience would need oneweek to perform one 20 -node model of a printed board, andtwo weeks for a 70 -node model, using a general-purpose heattransfer program.

CISD has developed GCSMOTHA, a FORTRAN pro-gram with a user-friendly, menu -driven front end, which hasa 3-D graphics modeling capability for a complex electronicspackage thermal analysis. The Auto-Trol mechanical designCAD system, on which GCSMOTHA is currently operating,first discretizes the physical model using the Series 7000Finite -Element Modeler (FEM). It then performs the neces-sary data extraction by utilizing Auto-Trol's ApplicationsInterface (AI) fortran programs for generating user-friendly

command instructions. Data input is fast andreliable, and is done by responding to screen prompts and byinterrogating the FEM-constructed thermal mode. The inputdata file is automatically created by the CISD-developed AIprograms for running GCSMOTHA in the operating system.Both transient and steady-state heat transfer solutions areused for temperature predictions. GCSMOTHA uses thelumped -parameter explicit finite -difference network solutionin MOTHA for thermal predictions. In addition, GCSMO-THA dynamically couples MOTHA to CHTC (convectiveheat transfer),6 a CISD FORTRAN program for performingconvective heat transfer calculations; PROPTY, a CISDFORTRAN program that is a resident library of 12 commonfluid coolant properties; MATPROP, a resident library ofmaterials properties; VIEW, a resident library of radiationview factors; and applications interface (AI) FORTRANprograms for creating user-friendly, menu -driven data entryand thermal model network data extraction.

GCSMOTHA features include:

1. 3-D thermal modeling using the Auto-trol Finite -ElementModeler for creating nodes (temperatures) and elements(thermal resistors);

2. Transient thermal analysis using MOTHA's explicit finitedifference equation algorithm;

3. Multiple grids (or meshes), currently up to 25;

4. Linear conduction coupling between adjacent grids;

5. Special thermal resistors accounting for contact resist-ance, perpendicular conduction (in depth), and non -


(a)

(c)

Fig.1. An example of conceptual model evolution: (a)represents a wire frame outline of the design; (b) shows

pulled from the layout. Here again, the detailer has a headstart in preparing a fully dimensioned drawing.

It is in this detailing phase that many tailored time -savingapplications programs can be developed. The tedious, repeti-tious tasks such as dimensioning, adding fabrication, mate-rial and finish notes, generating schedules and formatting adrawing can be automated. This automation not only savestime in the design process, but allows the designer to concen-trate his effort on fit and functionality problems. This resultsin a more productive, challenging work atmosphere.

For example, we have developed a drawing note and mate-rial note program that has frequently -used embedded notes.If a designer wishes to choose a material, he can "page"through a list of standard materials by their commercial des-ignations, make a selection, and the proper Federal Specifi-cation for the material will be placed on the drawing. Thisensures that the manufacturing community will receive draw-ings that call out a standard set of materials across diverseprograms (a purchasing advantage), and that typographicaland numerical transposition errors will be virtually elimi-nated. Applying this philosophy to numerous in-house pro-gram tools has benefited our design process.

Another area where these savings have been demonstratedis in parts list documentation. At CISD we employ a CRP(Configuration Records Processing) system that is an accu-mulated parts list database on an IBM mainframe. This sys-tem allows users to sort, by a variety of methods, the partsand documents that make-up any particular program. Thestandard method of input to such a system is to manually

(b)

(d)

surfaces defined for shading; and (c) and (d) show differentviews of the shaped solid.

generate a parts list, and then have keypunch personnel pre-pare the information with the proper database coding. This isa rigorous, redundant process prone to error. We have devel-oped an applications program that allows our designers toenter their parts list information into a text file while they areworking on their drawing. With this file on the system, it ispossible to convert it to a form that is compatible with ourparts list system on the IBM and transfer it directly to thatsystem. This represents increased efficiency in the flow ofinformation.

Engineering analysis

Concurrent with the design of hardware, engineers are con-cerned about the structural integrity and the thermal impactof their designs. Again, with the hardware in a 3-D database,engineers can spend their time performing analyses of partsand systems of parts rather than entering the geometry. TheCAD software that we use has a finite -element mesh genera-tion module. With this capability, an engineer can quicklyand easily prepare a finite -element model of the existingdesign and pass this model into one of the finite -difference orfinite -element solution programs.

Thermal analysis-GCSMOTHAOne of the most frequent causes of electronic componentfailures in complex electronics equipment is heat.' An exami-nation of MIL-HDBK-2172 indicates that the failure rate for


S.W. Butt H.P. Miller D.N. Hoke F Piazza

Comprehensive application of a CAD/CAMsystem to mechanical design

RCA integrates CAD/CAM into all phases of the mechanical designprocess.

The benefitsof CAD/CAM in the mechanical design proc-ess can best be seen when the system is tailored to the designflow process in a particular industry. At RCA's Communica-tion and Information Systems Division an abundance ofprograms are developmental, short production, or large-scale systems -integration projects requiring design of a widevariety of hardware. Whereas many industries look at CAD/CAM as an efficiency tool in the design of a family of high -

volume production products, at CISD we are attempting toautomate the process flow of mechanical design to minimizeduplicated efforts, improve overall productivity, and reducedevelopment schedules.

This comprehensive approach includes:

Preliminary marketing modeling using surface -shadedwire frames

Design layout of a product assembly from the preliminarydesign

Structural and thermal analysis when deemed necessary onthis design

Production drawings of parts "cut" from the layout

Abstract: At RCA's Communication and InformationSystems Division, a CAD/CAM system is being integratedinto many phases of the mechanical design process. Startingwith conceptual design, the 3-D database created can beused to start design layouts, generate detail drawings,perform engineering analysis, pass parts list information toa centralized database, and generate machining, fabrication,and assembly information for manufacturing. This is madepossible by tailoring a versatile graphics software package(Auto-trol Technology's Series 7000) to meet the particularneeds of the communications industry. The benefits of thisapproach are twofold: much redundancy of effort iseliminated, and better communications are establishedbetween various design disciplines.


Parts lists transferred to a centralized database Machining, fabrication, and assembly data Libraries of repetitive or evolving hardware for use in

future designs.

This requires the use of a functionally versatile 3-D graphicspackage (Auto-trol Technology's Series 7000), and the devel-opment of custom software tailored to the design flow.

Conceptual design

The first step in mechanical product design is to consider thedesign parameters as determined from a specification of per-formance and to generate a conceptual design. This design isgenerally the joint effort of engineers and marketing peoplewho must communicate to the customer the proposed aes-thetics of the product in addition to its functionality. Thisprocess can be quite evolutionary and often many designconcepts are generated. In a non -CAD environment the rateat which alternative concepts can be generated is dependenton the rate at which artists and designers can manually pro-duce sketches.

Figure 1 shows a case where 3-D modeling can beemployed to develop a geometric model and easily showdifferent views and color combinations of the proposedproduct. This allows the proposal team to visualize alterna-tives and make decisions on form and aesthetics prior tomodel making.

Hardware design

Once the appropriate model has been chosen and it is time tostart the actual design, the engineer or designer already has aconfiguration with which to work. The detailed design has a"head -start." The design layout may now progress with thedesigner working in a 3-D frame of reference to best see therelationship and fit of the assembled parts. Figure 2 showsone such layout that was generated from a conceptual sketch.This shows the ease with which individual part details may be


Acknowledgments

The authors would like to thank the fol-lowing people for their help with thisarticle: J.E. Greenup, T.A. Arsenault,M.F. Greene, W.B. Bradbury, R.L. Pre-noveau, R.S. Joress, and P. Berrett.

VAX is a trademark of Digital EquipmentCorporation

MicroVAX I is a trademark of Digital Equip-ment Corporation

DECmate is a trademark of Digital EquipmentCorporation

DATATRIEVE is a trademark of Digital Equip-ment Corporation

IBM PC is a trademark of International BusinessMachines Corporation

IBM PC AT is a trademark of InternationalBusiness Machines Corporation

IBM PC XT is a trademark of InternationalBusiness Machines Corporation

Apple is a registered trademark of AppleComputer, Inc.

Macintosh is a trademark of Apple Computer,Inc.

Lisa is a trademark of Apple Computer, Inc.

MacDraw is a trademark of Apple Computer,Inc.

MacProject is a trademark of Apple Computer,Inc.

LaserWriter is a trademark of Apple Computer,Inc.

MacTerminal is a trademark of Apple Computer,Inc.

MacWrite is a trademark of Apple Computer,Inc.

MacPaint is a trademark of Apple Computer,Inc.

Imagewriter is a trademark of Apple Computer,Inc.

Intel is a registered trademark of IntelCorporation

Supercomp-Twenty is a registered trademark ofArms Technology, Inc.

Jazz is a Trademark of Lotus DevelopmentCorporation

Lotus 1-2-3 is a registered trademark of LotusDevelopment Corporation

Microsoft is a registered trademark of MicrosoftCorporation

Fred Voto is a Senior Administrator in ASD'sOperations Control Department, where he isresponsible for cost control and cost proposalpreparation. He received a BS degree inFinance from Boston College, and an MBA fromthe Boston College School of Management.Since joining RCA in 1975, Mr. Voto haspioneered the use of interactive computers forcost estimating, using Data General and VAXBasic. Over the next couple of years, Mr. Votoused Data General's C-350 to eliminatekeypunched cards for DEPICT applications bywriting various data entry screens andprograms in Cobol to create electronic cardimages. He has also initiated other businessapplications in Supercomp-Twenty andDATATRIEVE. In his current assignment insupporting the Avionics group, Fred is using theMacintosh PC for a variety of cost reporting andcost analysis applications.Contact him at:Automated Systems DivisionBurlington, Mass.Tacnet: 326.3305

Morton D. Stern joined RCA AutomatedSystems Division in 1966. As Manager, ProgramControl and Pricing, Mr. Stern is responsible forthe management of two departments at ASD:Program Administration and Control, and CostAnalysis and Pricing. He was a charter memberof the Committee that developed the ProgramManager's Guide. He received his BS degree inEngineering Management from NorwichUniversity in 1961, and did his MBA studies atthe University of Rochester. After college, hewas on active duty with the U.S. Army SignalCorps Procurement Agency, where he was aProject Manager. He spent several years withGeneral Dynamics/Electronics Division asBusiness Supervisor for the F-111 ATE program,and was a Senior Value Engineer for BellAerosystems prior to joining RCA. Mr. Stern hasbeen an active instructor in the RCA After HoursTraining Program, and was a member of theEvening Division Faculty at NortheasternUniversity for several years.Contact him at:Automated Systems DivisionBurlington, Mass.Tacnet: 326.5050

17Stern/Voto: Personal computers in the engineering workplace

RCAATS/AV PLANNING 35FT. VAN PRODUCTION SCHEDULE

AS OF 21 OCT 85

ACTIVITYMUG JUL 1985 AUG SEP OCT NOV DEC 1986 JAN FEB

WEEKENDING

28 4 II 18 25 I 8 15 22 29 6 13 20 27 3 10 17 24 I 8 15 22 29 5 12 19 26 2 9

INSTALL FOLD ROOF

INSTALL DOORS

INSTALL SEALS

INSTALL A/C'S

INSTALL POWER ENTRY PANELS

RAIN TEST

BUILD ELEC. PANELS

WIRE/CHECK-OUT ELEC. PANEL

FOAM VAN

INSTALL AIR DUCT

INSTALL INTERIOR PANELS

INSTALL WIRING/LIGHTS

INSTALL FLOORS

PAINT VAN INTERIOR

FINISH VAN INTERIOR

INSTALL FLOOR MATERIAL

INSTALL ELEC. PANELS

"HOT- CHECKOUT VAN ELEC.

FINISH VAN EXTERIOR

PAINT VAN EXTERIOR

FINISH WORK

DRY - RUN ELEC. ATP

CONDUCT ATP

DELIVER TO RCA

Wir

Yar7

V

.--

V

7V

I

VV

S7

I

IL.',,,/ t

W4

7 V

V 1

I

I

I

I

I V

V7

7

Fig. 5. With PCs, production schedules can be produced quickly and inexpensively. Thisproduction schedule sample was developed on a Macintosh PC.

has also advanced, allowing efficientcommunication between dissimilarhardware, and the creation of softwarefor all types of hardware. Two factorshave contributed most to improvingoffice automation at ASD: the efficientdistribution of equipment and software,and extensive hands-on training withlive applications.

Management use of PCs

Division management has also bene-fited from the use of PCs. For example,during the infancy stage of productdevelopment, various designs are iter-ated and reiterated as product require-ments are refined. Because engineeringdata is stored on the PC in a graphicformat, presentation material, such asviewgraphs, can be quickly and inex-pensively produced for managementreviews. Functional block diagrams,conceptual schematics, organizationalcharts, software module breakouts,

flowcharts, and work breakdown struc-tures are just some of the outputs of this"electronic pencil." Managers can nowlook at various product alternatives andexamine each until the most suitablecandidate is developed.

Examining alternate approaches toan evolving system can be very time-consuming and costly. For this reason,design tradeoffs must be measured andlimited in scope. With the ease and fea-tures of a system like the Macintosh, thesystem designer has more freedom to becreative and can readily examine "whatif" solutions to a problem. This abilityto examine more alternatives will ulti-mately result in the definition of a bet-ter, higher -quality system and a higherdegree of confidence that the evolvedsystem is the best one.

Looking ahead

In looking to the future, we find uses ofthe PC -like system as a laboratory aid.

One such system is the Intel 310, a multi-user system based on the Intel 80286processor. The Intel 310 can be confi-gured with the only currently availableEthernet transport control softwarethat is approved by the InternationalStandards Organization (ISO). AnIR&D project, looking forward to theATE and test architectures of the future,is making use of networking technologyto create systems that are "infinitely"expandable, and that can communicatewith a wide variety of other communica-tions systems. The 310 System will beused, like the IBM PC, as a surrogatesystem controller. Since its network soft-ware will be known working, it willanchor the process of bringing up allother systems and attaching them suc-cessfully to the network. And, havingthe only ISO -standard software pack-age available, it will give ASD a leg up inthe race for the ATE architecture of thenext decade.


with an impressive and meaningfuldemonstration.

Preparing illustrations with the PC

A further PC application is in engineer-ing design. The design process involvesthe creation of many technical illustra-tions that may be changed several timesas the design evolves.

In the past, all illustrations were pro-duced by the Technical PublicationsDepartment. The results often exceededthe level of quality required for internalengineering use, and the normal proces-sing time required to produce these illus-trations tended to slow down the itera-tive design process. Now, engineershave begun using an Apple MacintoshPC 512 for their design tasks. Analysisof results showed that by using the Mac-intosh in engineering, the cost per illus-tration was considerably reduced. FromMarch through September 1985, VehicleElectronic Systems Engineering pro-duced 1600 engineering drawings on theMacintosh, resulting in a well -illustratedsystem. By facilitating the iterativedesign process, a higher quality designhas been possible.

Program administration and control

The Apple Macintosh has helped ASDstandardize scheduling in all of its Pro-gram Management Offices (PMOs).This has improved the quality of outputand reduced the cost of monthly sched-ule updates. Figure 5 is a typical exam-ple of a production schedule developedon a Macintosh by the Avionics Plan-ning Group.

In the past, schedules were hand -generated in draft. Once agreed upon,the schedules were manually convertedto master vellum, a labor-intensive proc-ess that involved hand taping letters,symbols and bars. This was followed bya quality check prior to reproductionand distribution.

Today, standard templates and sym-bology are stored in the Macintosh'smemory. A project planner can nowcreate the schedule by moving a mouse,using either MacDraw and/or MacPro-ject software. Additional savings occurduring the program evolution since allchanges can be entered directly onto thescreen. The Macintosh system allowsfor rapid resequencing and redrawing,thus eliminating the tedious and time -

BATTERY

ENGINE

IN EACH CYLINDER

GRAPHICS

GRAPHICS

GRAPHICS

PRINTER

ST /

6 CYLINDER

ENGINE

PC

BATTERY SIGNAL FROM

WAVEFORM PRESSURE TRANSDUCER

ICEDISPLAY

TRIGGER

SIGNAL

Fig. 4. Personal computers are appropriate as portable, low-cost field data aquisition sys-tems. In this system, all components were purchased commercially, including the dataacquisition software. This significantly reduced costs for the STE-ICE Data Collection Sys-tem.

consuming manual schedule updates. Inaddition to reducing overall labor, thistechnique has also lowered the cost ofupdates by over one third. Another bigplus has been the exceptional quality ofoutput. With a standard printer, sched-ules are suitable for viewgraphs; with alaser printer, schedules are nearly bookquality.

Other business applications using thePC are increasing daily. The Cost Anal-ysis and Pricing Group uses IBM PCs torapidly update cost proposals. By usingelectronic spreadsheets that facilitaterevisions and updates for such commonthings as rate changes, this group hasbeen able to handle an increase in work-load of forty percent with essentially thesame resources.

The administrative community reliesheavily upon the need to produce accu-rate, constantly revised and updatedspreadsheets for internal and customerconsumption. Because of the dynamicnature of these spreadsheets, their utilityis greatly enhanced in a PC environ-ment. Frequent and rapid spreadsheetmodification can be accomplished withextreme ease using the Macintosh sys-tem and its mouse and pull -down menufeatures. Final output can be rapidlyproduced on hard copy, using either theImageWriter or LaserWriter printer,depending upon output quality require-ments.

The Macintosh has become a truebusiness machine at ASD. With over500 pieces of retail software, eightmajor hard disk systems, telecommuni-cations, high quality printing, half a

megabyte of standard memory, and thesupport of respected names in the soft-ware and peripheral industry, the Mac-intosh has grown into a system thatoffers untold dividends in even its sim-plest configuration.

Office automation with PCs

The evolution of office automation hasstemmed largely from the desire tobecome less dependent on the main-frame while still having it available fordata access. The minicomputer wasthought to be a good solution to thisproblem, and quickly became a way oflife here at Automated Systems Divi-sion. However, new problems soonemerged in using this equipment. Mini-computers became overloaded withusers and various business softwareapplications suffered from poorresponse times. This pointed clearly tothe need for smaller, self-containedunits. Personal computers became avery attractive option for this particularapplication.

For most of our day-to-day businessapplications, PCs provide virtually thesame results as software loaded on amini or mainframe for a lot less powerand far fewer dollars. The PC can accessa distant mainframe via modem, accessand download the required source data,and reconfigure the data as desiredwithout fear of destroying it.

Over the past few years, office auto-mation has accelerated, with usersexchanging ideas, resources and, mostespecially, learning skills. Technology

Stern/Voto: Personal computers in the engineering workplace 15

The personal side of the PC

1985 was a banner year for theacquisition and use of PCs in Engi-neering at ASD. The currently mostpopular PC choice of ASD engi-neers is the Apple Macintosh (Mac).Engineers prefer the Mac over itsclosest competitor, the IBM PC, by aratio greater than 2:1. Our mostnumerous PC in Engineering is stillthe DECmate, which is used forword processing by most of thesecretaries.

It is not difficult to understandwhy engineers prefer the Mac. It iseasy to learn to use in a short timedue to such features as icons, pulldown menus, and the mouse. Thus,an engineer who has no prior expe-rience on the Mac can becomeproductive in a matter of hours-not days, weeks, or months. Notraining is necessary on the Mac;just turn it on and start getting famil-iar with the interaction of the mouseand the user. Before long, the userhas the hang of using the mouseand clicking on the menu bar andthe fun of using the Mac begins. Tosome, this may appear to be acontradiction-implying that workcan be fun-but this is the feed-back from Mac users.

PCs are used to create PERT

diagrams for schedule visibility andcritical path; high -quality presenta-tion material (in a short amount oftime); technical illustrations (use ofthe Mac was found to be about fourtimes less expensive than previousmanual means); specifications, testplans, and status reports (usingMacWrite); viewgraphs, block dia-grams, and work breakdown struc-tures (MacDraw); free -form graph-ics (MacPaint); and project andbudget planning (using MacProject,Jazz and Microsoft's Excel).

PCs are used in both a stand-alone mode and interfaced to ourEthernet/DECnet system in order tocommunicate with our installedbase of engineering VAX systems.We have used both terminal emula-tors (such as Mac Terminal) and thecommunications feature of LotusJazz to transfer files from the Macsto our VAXs and vice versa.

Engineers I have talked to areexcited about the Mac's capabili-ties, so much so that demand forMacs far exceeds current supply.Engineering managers report thatMacs already acquired are receiv-ing heavy use. They facilitate thecreative process such that an engi-neer can quickly see and improve

upon the results of his or her ownthinking. This fast responsiveness isan important reason why the Mac isfun to use and a morale boosteramong engineers.

The bottom line for engineeringmanagers is a good one- qualityimprovements, cost savings, andcost avoidance as a result of usingthe Macs. I've been told by severalmanagers that on a recent proposaleffort, the Mac in combination witha LaserWriter printer was indispen-sable; without it, timely response tothe customer would have beenimpossible.

At ASD, a number of engineershave bought Macs (mostly the "FatMac" with 512K of memory) forhome use. With a Mac of one's own,there is a tendency to take morework home than would otherwise bethe case.

Based upon our experience withPCs so far-quality improvements,and hours and dollars saved/avoided-I would have to say thatthe PC is turning out to be a goodinvestment for ASD.

-Dr Vaughn Grace

filtering routines. The commandprocessor/expert system is written inPascal, which includes the detectionalgorithms. Basic expert system technol-ogy is used to determine the rules forfault detection.

Field data collection with the PCIn a unique engineering application, anIBM PC is used in conjunction with aninternal combustion engine test set(STE- ICE) for field data collection anddevelopment. The PC collects dynamiccylinder pressure data and electricalwaveforms from an operating internalcombustion engine. The cylinder pres-sure is collected from each cylinder in

the engine simultaneously in real timeand stored on disk, along with the bat-tery cranking waveform and severalparameters from the STE/ICE. Thedata is then analyzed to determine thecompression balance of the engine andto calculate several internal constants tobe placed in STE/ICE memory. The sys-tem implementation is shown in Fig. 4.

PCs and Vetronics testingAnother application of the IBM PC inengineering addressed a crucial compo-nent of the Vetronics crew station.Vetronics involves heavy automation ofcomplex military vehicles using a broad-band bus for interstation audio/video

and high-speed data communication.One of the many tasks of the Vetronicscrew station core processor is to controlthis communication, allowing a crewmember to select from the various chan-nels.

During the initial bus testing and inte-gration, in preparation for the prelimi-nary design review, not enough coreprocessor boards existed to support thebroadband bus demonstration. In theinterest of time, IBM PCs were usedto simulate the Vetronics ComputerResources Subsystem. The I/O andgraphic capabilities of the IBM PChardware and software allowed for aneasy implementation of a high -qualitysimulation and provided our customers


bearing that has a small spall on one ofits races, as shown in Fig. 1. During theearly stages of failure, the spall is smalland will not immediately jeopardize thebearing's functionality. If undetected,the spall will grow in size until the bear-ing fails completely, possibly endanger-ing the operation of the entire machine.Even during the very early stages of fail-ure, the bearing assembly will emitbursts of energy at a repetition rateequal to that at which the balls encoun-ter the small spall. These bursts ofenergy cause the machine elements tovibrate at specific frequencies and, ifisolated from the other concurrentmachine vibrations, the failing bearingcan be detected and identified at a veryearly stage.

The vibration signal levels associatedwith the failing bearing at this earlystage of failure are, however, typically30 to 50 dB below the backgroundmachine vibration levels. Until recently,this has prevented the early detection ofmost minor bearing faults. However,the advent of small, powerful com-puters such as the IBM PC AT and thecommercial availability of signal proces-sing software have greatly simplified theimplementation of complex vibrationalanalysis techniques.

Many of the processing techniquesused in detecting bearing faults are alsocommon to modern spread -spectrumradar signal processing. As in radar pro-cessing, the fault signature is masked innoise from various sources includingclutter and synchronous and asynchron-ous noise from mechanical sources.Unfortunately, unlike radar signal pro-cessing, no transmitted reference signalis available. Many classical digital signalprocessing techniques similar to fre-quency filtering, pulse width identifica-tion, demodulation, and time -to -frequency domain transformation areused.

All of these special signal processingtechniques require considerable sophis-ticated computational capabilitieswhich, until recently, were not availablein portable equipment.

Fault detection using the PCEngineers at ASD have found the PC tobe a helpful tool for fault detection intruck transmission and transfer assem-blies. In this system, a laser velocimeteris used to detect vibration signals with-

LAISER

BEAM

2 1/2 TON TRANSMISSION

ON/OFF SWITCH

'CIRCUIT BREAKER'

11011. 15 A

REMOTE

CONTROL

SLIDING SHELF

FOR KEYBOARD

WHEN USED

LASER ELECTRONICS

VIDEO

DISPLAY

PRINTER

MICRO

COMPUTER

POWER SUPPLY

SPARES/STORAGE

Fig. 2. A laser velocimeter works with a PC to detect transmission vibration signals. The PCprovides a user-friendly interface to the test system while handling system control, dataacquisition, and display functions.

MOTOR (_ TRANSMISSION

SPEEO

SENSOR

LASER BEAM

LASER

VELOCIMETER

HYDRAULIC

PUMP

LOAD

SPEED

SENSOR

GRAPHICS

DISPLAY

C 1

a.

A-0CONVERTER

DIGITAL

CIRCUITRY

IBM PC AT

REMOTE

CONTROL

- PRINTER/PLOTTER

KEYBOARD

Fig. 3. Fault detection hardware includes an IBM PC AT. The PC's architecture allows easyintegration of custom peripherals, a powerful microcomputer and a large existing softwarebase.

out physically contacting the transmis-sion (see Fig. 2).

The vibration signal, along with inputand output shaft rotational positioninformation from the transmission, issent to the PC. Functionally the hard-ware consists of an IBM PC AT withgraphics display, printer/plotter, andanalog to digital (A/D) converter andsensors, as shown in the block diagramin Fig. 3. System capabilities make fulluse of the PC's features including dis-

play, storage, and hardcopy output ofsignals, variables, and test parameters.The fault detection software is run ascommands under the standard diskoperating system. The signal processingcore programs are purchased from anoutside vendor and modified for thespecific application, thus allowing a sig-nificantly reduced programming task.Processing features that require opti-mized coding are written in MacroAssembler, which includes A/D and

Stern/Voto: Personal computers in the engineering workplace 13

M.D. Stern F.A. Voto

Personal computers in the engineering workplaceThe innovator amongst us

At Automated Systems Division, nobody questions the -why" ofPCs-they want to know when they will have more.

As the PC becomes more and more ofan "external brain," a subtle change isbeginning to emerge, with greateremphasis being placed on quality. Wecan now differentiate between what wehave to learn and what we simply wantto access. As the pocket calculatorreplaced the slide rule, the PC is chang-ing how we perform engineering andrun the office.

RCA Automated Systems Division(ASD) has, over the past few years, beena part of this evolutionary process in theareas of office automation and engi-neering. This transformation has beenenhanced through the accumulation of21 VAX computers, ranging in size fromthe MicroVAX I to the VAX 11/780.These are supplemented by over 300 ter-

Abstract: Automated SystemsDivision has found innovativeapplications for PCs in engineering,business and management. Inengineering, PCs are useful forvibration analysis, fault detection, fielddata collection, and Vetronics testing.Business applications of the PC includeprogram schedules and proposalupdates. In management, PCs are usedfor producing engineering data ingraphic format for managementreviews. In the future, the PC will serveas a laboratory aid, and will play an.integral role in automatic testequipment and test architectures.

01986 RCA Corporation.Final manuscript received January 3, 1986Reprint RE -31-2-3

1 I

minals, 10 IBM PCs, and 28 AppleMacintosh PCs.

We are all familiar with the burgeon-ing use of digital logic and circuitry inimplementing algorithms in operationalequipment. Special signal processingtechniques are incorporated into reliablecompact hardware through VLSIdesigns. These designs are the end resultof extensive engineering using small,powerful computers like the IBM PCAT.

PCs and vibration analysisAn example of engineering design usingPCs is the effort in vibration analysis.

The PC has been used to help develop atechnique for non -intrusively assessingthe condition of rotating machinerycomponents through vibrational energysignal processing. This technique is notdependent upon trending analysis, andcan be implemented in inexpensive port-able multipurpose test equipment usingpersonal computers. The introductionof commercially available digital signalprocessing software for an IBM PC hasmade these advancements economicallyfeasible.

To illustrate the difficulty of detectingfaults through vibration analysis, con-sider a machine incorporating among itsmany components a rolling element

I 1 l I 1 IIII 1 1 I I I I 1 II 1 I I

83 264

16

12 16 20 2.4 28

24 3,2 0 48 56

Fig.1. This magnified photograph shows a bearing outer race that contains a Class C fault asgraded by the Army. This class of fault does not immediately impair the bearing's functionality,but warrants replacement. With PCs and signal processing software, minor bearing faultssuch as this one can now be detected earlier


Wang PC or DECmate word proces-sors. The DECmates and Wangs canaccess VAX as terminals or transfer textfiles, eliminating the most notoriousform of pseudo office automation-managers typing memos into a termi-nal, printing it and handing it to theirsecretaries for retyping. We are alsoemploying IBM personal computers forspreadsheet -intensive applications andto access specialized PC applicationssoftware. All IBM PCs are connected tothe VAXcluster and can double as termi-nals. We have recently installed DEC'sAll -In -One package, which replacesDECmail, as well as adding an applica-tions integration environment and someenhanced features, such as calendarmanagement. The migration of users toAll -In -One will be a continuing processover the next year.

User supportNo investment in computer facilitiesmakes sense unless people know how touse them. We have made substantialefforts to provide supporting services toour user community. A weekly scheduleof classes provides instruction in DEC -mail, DECalc, and DECgraph. Specialclasses can be arranged for busy man-agers. Computer -based instruction cov-ering office automation, VAX/VMSand touch typing can be accessed at anytime at any terminal. We have estab-lished an information center adjacent toour office area that includes top -of -the -line DEC VT241 terminals and IBM PCATs. It is used for training classes as wellas individual use and evaluation. Wepublish a bimonthly newsletter, Mail-box, which includes hardware and soft-ware news, user tips, and system sched-ules. To insure the integrity of oursystems, we maintain a rigorous databackup policy, offsite data storage, andwe have installed sophisticated securitysystems to protect our facilities. Mostimportant of all, we provide a telephonehelp line- Tacnet: 325-7273-which isthe central point for reporting systemand performance problems, as well asasking for help or advice on any of ourproducts and services.

FutureIf current trends continue, we expect tosee an expansion of the overall user

The MEC Information Center is equipped with DEC VT241 graphics terminalsand IBM PC AT PCs. In addition to serving as a site for classroom and individualinstruction, the information center can be used by employees who do nototherwise have access to PCs.

base, as well as increasing emphasis onengineering workstations and personalcomputers. To meet these needs, ECSwill be providing supporting services onnew engineering workstations in 1986,as well as expanding its support of per-sonal computers. We expect a majorshift of manufacturing processing toVAX systems in the next few years asobsolete versions of IBM software arephased out and the HP -3000 systems are

retired. Probably the most importanttrend will be towards computer -integrated manufacturing (CIM),wherein all design, administration, andmanufacturing is completely integratedthrough common databases and com-munication links. We will be workingwith Somerville's CAD, CAM, and MISorganizations over the next few years tobring the CIM evolution to Somerville.

Louis Judice joined the RCA Solid StateTechnology Center (now the MicroelectronicsCenter) in 1978 after receiving the BE degree inElectrical Engineering from Manhattan College

in New York City. As a Member of the TechnicalStaff in the Design Automation department heworked on computer -aided design softwaredevelopment, including automatic VLSI routingalgorithms and the porting of the productionART system to superminicomputers. In 1983 hebecame Project Leader of CAD TechnicalSupport and was responsible for expandingSomerville's engineering computer capabilitiesand commercial software support. Between1983 and 1985, VAX -based computing capacitywas increased seven fold, as the subscriberbase swelled from 20 to over 400, and overallusage increased 1600 percent. In late 1985 hewas appointed Manager of EngineeringComputer Services, and in that capacity he isnow responsible for the Engineering ComputerCenter and VLSI Design Center, as well as officeautomation systems for the MicroelectronicsCenter and Solid State Engineering.Contact him at:

Microelectronics CenterSomerville, N.J.Tacnet: 325.6425DECmail: JUDICE SMDCDM/SMDC::LJJ

Judice: Engineering Computer Services keeps Somerville online 11

Electronic mail links most RCA engineers ...Why aren't you using it?

One of the least known benefits of tying together most ofRCAs VAX and IBM computers is the ability of virtually anyuser to exchange electronic messages with another. Onestumbling block is the existance of several electronic mailprograms and some slightly obscure rules for addressingmail to remote sites. The current status of networked elec-tronic mail systems is as follows:

Somerville (Solid State Division, MicroelectronicsCenter)-DECmail, VMSmail, HPmail, ISMEFPrinceton (RCA Laboratories)-DECmail, VMSmail, ISMEFCherry Hill (Corporate Information Systems andServices)- PROFS, ISMEFMoorestown (Advanced Technology Labs)-VMSmailCherry Hill (A&D Staff)-VMSmailEast Windsor (Astro-Electronics)-VMSmail

Essentially, all systems with DECmail can talk to other sys-tems with DECmail, and all systems with VMSmail can talkto other systems with VMSmail. ISMEF is a special com-munications system developed by Information SystemsPlanning in Princeton Labs. It translates one mail protocolto another mail protocol, thus allowing a PROFS user at 30Rockefeller Plaza to send mail to a DECmail user in Somer-ville.

Almost anyone in the above sites with access to VAX orCMS can send mail to almost anyone else in the same oranother site. Of course, you should contact your local VAXsystems administrator or MIS department for additionalinformation on using the systems at your site.

Here's an example of an electronic message that startsin Somerville, and will be read by Wilson in Somerville,Roosevelt in Princeton, Cleveland in New York, and Chur-chill in Brussels:

Command> CREATE (This is the DECmail command to"create" a memo)To: WILSON (Wilson is in Somerville)

To:Cc: ROOSEVELT @COPSDM (Roosevelt is in Princeton,

and @COPSDM is the address of the DECmail VAX inPrinceton)

Cc: MAILMAN (MAILMAN is the local mailbox for theISMEF message switch)

Cc: CHURCHILL @MRGATE@RCABRU (Churchill is onVMSmail, which requires routing through the VAX's Mes-sage Gateway)

Cc:Topic: SOS Device Characteristics Proposal

Enter Message Text@PROFSCH:GCLEVELAND (This is the PROFS addressthat ISMEF will use to route from DEC to IBM)

Type in your message text . . .

Command> SENDThis may look very complex. How on earth does oneremember so many names and addresses? Well, Engineer-ing Computer Services in Somerville has published anElectronic Mail Directory, with assistance from CorporateInformation Systems and Services and Princeton Informa-tion Systems. Although we print the directory, we can alsoprovide a VAXNMS command procedure that searchesthrough the directory online. VAX groups interested in con-tributing to the directory should contact me at Tacnet: 325-6425 for further information, or better yet, send me mail atSMDC::LJJ on VMSmail, JUDICE @SMDCDM on DEC -mail, or @SMDC:JUDICE through ISMEF.

Of course, another excellent use of electronic messag-ing is transmitting your RCA Engineer manuscript to yourEditorial Representative or to the RCA Engineer staff inPrinceton. For example, this article was mailed electroni-cally to:

Somerville: Sue Suchy (SUCHY @SMDCDM orSMDC::SPS)andRCA Engineer Staff: Tom King (COPS::TEK)

A final note: my electronic mail address is shown after mybiography at the end of this article. It is typeset in the formthat we are suggesting that people at Somerville use onbusiness cards, in articles, on memos, etc.

the process of acquiring a 36 -inch colorVersatec plotter.

Office automation

In 1984, our organization was given theresponsibility of developing a compre-hensive office automation system forthe Microelectronics Center and SolidState Engineering. The philosophy ofour approach to office automation is tomake maximum use of the synergy thatwill exist between engineering, manu-facturing, management, and adminis-tration using one computing environ-ment. Therefore, we chose to center our

office systems around the VAXcluster,choosing specialized workstations forusers with very specialized needs. Tointegrate communications we installedDECmail, a comprehensive electronicmail and filing system. We wrote afrontend menu system, called Inte-grated Office Services, to allow noviceand casual users to log in and selectfrom DECmail, DECalc spreadsheets,DECgraph graphics, bulletin boards,corporate mail directories, and otherfunctions. During the past six months,we have solidified our system by install-ing links to Princeton's ISMEF elec-tronic mail exchange and direct DECnet

links to the Princeton "LINX" DEC -mail system. These allow users inSomerville to communicate with usersof Princeton DECmail, as well as usersof the corporate PROFS office automa-tion package on IBM. We are workingwith Princeton and Somerville MIS toextend the network to include the SSDHP -3000 -based manufacturing systems.

In terms of workstation choices, wehave taken a very conservativeapproach. For most users, inexpensiveDEC VT220-type terminals provideadequate access to text and data proces-sing applications. Text -intensive users,such as secretaries, are served with either


modems on the Micom to allow users toaccess our systems from home terminalsor personal computers.

A major design goal of our terminalnetwork was to let any user access anycomputer resource in Somerville fromone desktop workstation. Amazingly,until the introduction of the Micom,there were offices that contained two ormore terminals for accessing differentservices. Though VAX and HP com-puters were easily supported on theMicom, special-purpose emulators wererequired for IBM access. We employ aLocal Data Datalynx 3274 protocolemulator, which allows DEC terminalsto enter the RCA SNA network andaccess Somerville and Cherry Hill IBMcomputers as if they were IBM 3278 ter-minals. We have also provided autodialmodems on the Micom to allow a user athis or her office terminal to easily con-nect to outside databases such as DowJones News Retrieval or the Source. TheMicom also supports a network ofabout 20 local printers, including laserprinters and letter -quality impactprinters, which are scattered throughoutthe Somerville facility.

Though the Micom provides an excel-lent medium for terminal communica-tions, we support many other local- andwide -area networks. Engineering Com-puter Services is a part of the Aerospaceand Defense DECnet network, linkingVAX systems in Moorestown,Burlington, Camden, Cherry Hill, EastWindsor, and Somerville to each other,and to the CISS Engineering ComputerFacility (EC'). Somerville and RCALaboratories established DECnet com-munications in late 1985 to support elec-tronic mail file transfer and remote ter-minal access. We are also networked tothe SSD Semicustom DECnet, whichlinks Somerville to the Brussels andSanta Clara design centers. Closer tohome, DECnet, over Ethernet, con-nects the Engineering Computer Centerto departmental VAX and PDP-11 com-puters within the Somerville Plant. Allgroups with DEC hardware are beingencouraged to join the network to takeadvantage of the VAXcluster's im-mensely greater computing and storagecapabilities, and to help integrate vari-ous types of design, test, and manufac-turing databases. The same Ethernetthat allows DEC computers to talk toeach other also allows DEC and Mentorengineering workstations to communi-cate, allowing remote terminal emula-

SOMERVILLEDATA

CENTER

CHERRY HILLDATA

CENTEREC3

SNA

A& DDECNET

°SRCDEC NET

SSDSEMICUSTOM

DECNET

DECNET

VAX CLUSTER

MICOMPBX

PERSONALCOMPUTERS TERMINALS AT&T.

NIPPE RN ET

ETHERNET

APOLLO.MENTOR

DEPARTMENTALVAX ANDPDP-1 1

CALMA

At the center of the ECS network is the VAXcluster. The Micom, DECnet, Ethernet and SNAeach play a role in the overall communications architecture.

tion and file transfer. This has beenaccomplished by providing specialUNIX-to-VAX/VMS communicationsemulation software on the VAX andApollo systems.

Although the majority of engineeringcomputation is VAX -based, it is impor-tant to maintain tight communicationswith the Corporate mainframe com-puters in Somerville and Cherry Hill,since there are certain classes of prob-lems that lend themselves well to solu-tion on IBM mainframe hardware. Weare implementing a DEC SNA gatewayon the Somerville Ethernet to link oursystems to IBM. This network willreplace an aging and unreliable remotejob entry (RJE) system based upon out-dated 3780 protocols. The SNA gatewaywill also allow new functionality, includ-ing the ability to route IBM printouts tothe VAX lineprinters, maintain multipleVAX -IBM connections, and permitmore flexible file transfers. The SNAgateway will operate at 56 kbps, asopposed to 9.6 kbps for the existing RJElink.

VLSI Design Center

Of principle importance to the Somer-ville community is access to sophisti-cated computer graphics systems thatallow custom integrated circuits to belaid out and edited in the quickest andmost productive manner possible. The

VLSI Design Center, located in theMicroelectronics Center, has been ful-filling this function for four years. Orig-inally formed from the merger of SSDand MEC design centers, the facilitytoday consists of both workstation andshared -logic graphics systems. FiveCalma GDS-II/32 and two MentorGraphics workstations are employed,along with three Calma shared -logic sys-tems and one Applicon shared -logic sys-tem. In addition, a Calma S/280 mini-computer provides background pro-cessing capabilities for the Calma work-stations. Over the next two years weexpect to replace the remaining shared -logic terminals with either Calma orMentor workstations. An interestingbenefit of our new workstation equip-ment is the ability to network into theVAXcluster. Both Calma and Mentorusers can now transfer files to and fromthe VAX systems, thus eliminating thetedious and very error -prone task ofwriting and reading magnetic tapes.

The VLSI Design Center also sup-ports several specialized pieces of equip-ment. The Calma Fast Mask Engine is adedicated CPU that is optimized to exe-cute design -rule verification algorithms.It is networked to the Calma worksta-tions to allow high-speed local verifica-tion of circuit designs. Three black -and-white Versatec plotters, including two72 -inch units, provide high -resolutionplots of circuit designs. We are also in

Judice: Engineering Computer Services keeps Somerville online 9

9.6 Kb/s CONNECTION

LOCAL TERMINALS AND PC'S

M478MULTIPLEXER

AT&T2212D

AUTOANSWERMODEMS

138 Kb/sTRUNK

X.25PAD

M

M

M600

P

B

SMDAVAX -11/785

SMOBVAX -11/780

SMDCVAX -11/780

SMDEVAX -8600

RCASOMVAX -I 1/782

HP 3000

DATALYNX3274

UDS 212DAUTODIALMODEMS

X.25PAD

ADMINISTRATIONCONSOLE

DIAGNOSTICCONSOLE

STATISTICSLOG

The Micom M600/2 Data PBX is the central switching system for terminals inthe ECS communications network. Over 300 terminals are connected to theMicom-every office cubicle in some 90 percent of our service area is providedwith at least one terminal.

Calma GDS-II/32 VLSI design workstations provide a significant improvementin layout designer productivity since they are based around dedicated 32 -bitsupermicrocomputer technology. Each station is connected to Ethernet, allowingdesign files to be transfered to the VAXcluster.

neering Computer Services was formedto manage what had grown to become afour -node VAXcluster consisting of twoVAX -11/780s, one VAX -11/785 andone VAX -8600. These four systemsshare a pool of 52 megabytes of mainmemory and over eight billion bytes ofdisk storage. An important feature ofVAXcluster is the ability of each proces-sing node to share disks, common batchqueues, and print spools. This lets usersview the cluster as one giant computercomplex. Since each machine operatesin an independent (but cooperative)manner, there is built-in redundancy. Innearly two years of VAXcluster opera-tion, we have experienced completeclusterwide failures only a handful oftimes. Our overall uptime consistantlyaverages over 99 percent month aftermonth. The VAXcluster supports over500 subscribers-mostly based inSomerville, though many Princeton,Moorestown, and Mountaintop usersalso access our systems. Somerville'sVAXcluster is the largest such installa-tion in RCA, and supports one of thelargest user communities of any divi-sional computer center in the corpora-tion.

Networking

To support this user population, aMicom M600 Data PBX is used toswitch some 350 terminals and tele-phone lines between the VAXs, as wellas to provide access to Somerville's HPand IBM computers. We selected theMicom approach to terminal accessover other approaches, such asEthernet, due to its very low cost perconnection, open architecture, andmaintainability. In some two years ofservice, our Micom switch has displayed100 -percent reliability, an impressivestatistic. Most office areas are con-nected to the Micom switch by time -division multiplexed trunk lines operat-ing at 138 kbps over twisted -paircabling. The vast majority of our termi-nal population consists of DEC VT220,VT240, and VT102 video display termi-nals. A growing population of IBM per-sonal computers, including XT and ATmodels, are supported as VT100 look -a -likes. A number of Wang and HP per-sonal computers, as well as DECmate IIand III word processors, are also con-nected to our terminal network. Wehave provided a large number of thelatest AT&T 2212D autoanswer


L.J. Judice

Engineering Computer Serviceskeeps Somerville online

Integration is the key to building an information system that servesengineering, manufacturing, and administrative users.

Somerville's Engineering ComputerCenter has its roots in a 1972 DigitalEquipment Corporation (DEC) PDP-11/20 minicomputer. Purchased by theDesign Automation department, thePDP-11 was a great step forward fromthe PDP-8 computers then in use. Thesystem was used for the initial develop-ment of the ART program, the produc-tion software system still used to convertmask artwork into numerical controltapes for optical and electron -beam pat-tern generation equipment. ART grewand grew, and was eventually trans -

Abstract: RCA's Somerville plant isthe home of the Solid State Divisionand the Aerospace and DefenseGroup's Microelectronics Center.Somerville is the focal point of RCA'sLSI technology and integrated circuitproduct development. Providinginformation services to this diversecommunity is the responsibility of theEngineering Computer Services (ECS)organization. ECS is a part of theComputer and Design Servicesfunction of the RCA MicroelectronicsCenter. Its responsibilities include theoperation of Somerville's EngineeringComputer Center, the VLSI DesignCenter and office automation systemsfor the Microelectronics Center andSSD Engineering. This article is a brieflook at Engineering Computer Servicesand the ways we serve our usercommunity.

©1986 RCA Corporation.Final manuscript received February 28, 1986Reprint RE -31-2-2

Technicians in the Engineering Computer Services Operations Center canpinpoint and correct problems anywhere in the computer and communicationscomplex. A sophisticated telephone and paging system can be used to instantlycontact other staff members anywhere in the plant or at home.

ported to the Solid State Division's IBM370/168 mainframe. Many otherinternally developed computer -aidedengineering tools were developed inSomerville over the years, including theRCAP circuit simulator, MIMIC logicsimulator, and the AUTOROM ROMpattern generator, just to name a few.Few changes to the computing environ-ment occurred until the late 1970s, whena new phenomenon occurred in thecomputing industry-the super mini-computer-epitomized by machinessuch as the DEC VAX -11/780 and thePrime 750. The Design Automationdepartment purchased a VAX -11/780 in

1981 to help speed the development ofnew automated layout software sys-tems. The VAX quickly became the pre-ferred computer system for softwaredevelopers, layout designers, and engi-neers, due to wide software availability,good response time, and numerous userfriendly tools and utilities. Most ofRCA's IBM -based CAD tools were con-verted to run on VAX in the early 1980s.Concurrently, the phase -out of severalvenerable, but hard -to -support pro-grams such as CRITIC and SMARToccurred during this period.

By 1985, the majority of engineeringcomputation was VAX -based. Engi-


ties installed. The actual programsreside on the network system volumesand look for this "key diskette" in drive"A" when loaded. Other packages, likeWordPerfect, are available with net-work licenses and no awkward copy-

protection schemes.The standard issue of software

includes WordPerfect (wordprocessor),Lotus 1-2-3 (integrated spreadsheet),and PC -VT Additional software pack-ages being used are Harvard Total Pro-ject Manager and Timeline (projectmanagement), dBase III (relationaldatabase), and Dr. Halo II (high resolu-tion graphics). At this time, pointingdevices such as a mouse, light pen, orgraphics tablet that might furtherenhance the PC graphics capability arebeing evaluated.

All PC users have received formaltraining via three separate courses:Introduction to PC -DOS and the ATL/LAN; Lotus 1-2-3; and WordPerfect.

Plans for the future

A new version of the PC -DOS mail soft-ware that will allow the interchange ofmail messages between VAX/VMS andPC -DOS is expected to be released in1986 from Syntax Systems. This capa-bility will allow us to further integratethe two environments and offload addi-tional VAX processing to the PCs.Additional mechanisms to integrate theVAX/PC installations will be investi-gated.

The new version (2.0) of Lotus 1-2-3will be installed as soon as it arrives.Some of the new features are file levelpassword protection, linear regressionanalysis, an increased size of 8192 rowsby 256 columns, 8087 math co-processor support, and improved mem-ory management.

The latest release of WordPerfect(4.1) will also be installed on the net-work. A few of its new features are abuilt-in thesaurus, split-screen editing ofmultiple files, newspaper (snaking)columns side by side on the screen,memory caching, and cutting and past-ing of rectangular blocks of text.

Authors, left to right: Rosenthal, Peters, and Abrams.

Gary Abrams has been a computer systemsanalyst at ATL since March 1985. His firstachievement was the creation of ATL'scomprehensive PC network. Currently, he isresponsible for all phases of the operation andenhancement of this network. His other effortsinvolve ATL's VAXs and parallel Ethernets.

Before coming to ATL, Mr. Abrams was incharge of all PC hardware and software supportat MicroAge. During this time, he becamecertified to service and support innumerablekinds of PCs: IBM, ATT, Compaq, and Altos, forexample. Earlier, at UNIVAC, he was responsiblefor the repair of minicomputer systems.Contact him at:Advanced Technology LaboratoriesMoorestown, N.J.Tacnet: 253-6676

Jack Peters is Manager of ComputerOperations and Administration in ATL's SoftwareEngineering Laboratory. The computer centerhe manages contains a cluster of VAXcomputers. This center services over 300 useraccounts. Currently, he is in charge of a systemof networks connecting all nonsecurecomputers in ATL. This networking extends,also, to other divisions of Aerospace andDefense and an IBM 3083 mainframe. His

experience includes systems and softwareprogramming, operations research, andnumerical control programming. Mr. Petersstudied Computer Science at Carnegie-MellonUniversity.Contact him at:Advanced Technology LaboratoriesMoorestown, N.J.Tacnet: 253-6644

As Manager of ATL's Software EngineeringLaboratory, Harry Rosenthal is responsible forthe development of advanced softwareengineering tools, for transfer of softwaretechnology, and for computer operations andadministration. Areas of technical focus includevalidation/verification (both formal and informal),modeling/simulation, compilers, operatingsystems, project/product management, database management, and systems analysis.

Before joining RCA, he spent over twentyyears in DoD consulting, primarily in theresearch and development sector, and alsoworked for two computer manufacturers. Mr.Rosenthal holds MA and BS degrees inMathematics from New York UniversityContact him at:Advanced Technology LaboratoriesMoorestown, N.J.Tacnet: 253.6451


Xerox disk server

A Xerox disk server was also consid-ered. In addition to a disk server, aninterface board (with software) wasneeded for each PC. The server wouldprovide moderate performance viaEthernet connection, scheduledbackup, some file security, and limitedshared resources. Among the disadvan-tages were high initial and incrementalcosts. Additionally, non -IBM personalcomputers would not be supported, thedisk server required operator attention,and very little commercial software wassupported.

Corvus Omninet

Corvus Omninet was another candi-date. A stand-alone disk server, inter-face board (with software) for each PC,and twisted -pair cabling was required.This network supported both IBM andDEC PCs, had moderate initial andincremental costs, and an electronicmail utility was available. However, at 1Mbs the transfer rate was relativelyslow, the total distance of the networkwas only 1000 feet, and no scheduledbackup was available. Furthermore, nobridge or gateway to the VAXcluster viaEthernet was available.

3Com 3Server

The 3Com stand-alone 3Server wasevaluated. When used with the 3ComEtherlink board and software, it pro-vided each PC with a virtual hard disk.The Etherlink board supported directEthernet connection via a DEC DELNIor transceiver, or through PC -to -PCchaining with Thin Ethernet cable.Either way, the transfer rate of 10Mbs provided high throughput. Astreaming -tape backup unit, permittingscheduled backups, was available.3Com's 3Server provided good file secu-rity, required little end -user retraining,and posed few software compatibilityissues. However, the cost/Mbyte wascomparatively high, and no bridge orgateway to the VAXcluster was availa-ble. Maintenance and lack of failoverprotection (explained below) alsocaused a concern. The server, along withoperating software, only supported 60users; after that, another server wouldhave to be added. Only IBM PCs andcompatibles were supported by theEtherlink product, and an additional

board had to be installed in the 3Serverto provide a bridge to the AppleTalkNetwork.

Syntax VAX Interface Manager (VIM)

The last alternative, VAX InterfaceManager (VIM) from Syntax Systems,used in conjunction with the 3Comproducts, was the design of choice. VIMwas found to be a software product thatruns as four detached processes underVAX/VMS. To a PC connected to thenetwork using 3Com's Etherlink prod-uct, the VAXcluster emulates a large3Server. The initial cost is low, and up to200 users can be supported. Virtual diskcapability is provided, as well as sharedaccess to two printer queues. A mail util-ity is included, although the IBM PC -DOS mail software from 3Com isneeded to use it. A number of utilities tofacilitate file transfers between VMSand DOS are included, but require VMSprivileges to use. If the VAX that thesoftware is running on goes down, thesoftware can be started on another VAXin the cluster, thus providing failoverprotection. Backup is provided throughnormally scheduled VAX/VMS back-ups. An extra software package, calledRemote File Manager, allows file trans-fers between DOS and VAX/VMS to beinitiated from the PC. To connect theApple Macintosh computers to the restof the network, a 3Server, with anAppleTalk interface board in it, is usedas a bridge between the VIM softwareon the VAX and AppleTalk.

Current networkThe current PC network allows PCusers to share VAX disk and printerresources. System volumes containingvarious utilities and application pro-grams can be linked to by any user onthe network. Each non -secretarial PChas its own local dot-matrix printer anda link to a letter -quality printer throughthe network. A small laser printer isincluded as part of the standard PC sec-retarial system. Acting as local printservers in addition to being worksta-tions, these systems allow localization ofhigh -quality printing resources for eachlaboratory.

Each PC has two connections-onenetwork connection and one asynchron-ous serial line for terminal emulation tothe VAXcluster. All Ethernet connec-tions are made through standard trans-

ceiver cables from either DELNIs ortransceivers. No Thin Ethernet cable isused because it greatly limits the overalldistance of the network and requires theuse of repeaters.

Each user has a five- or ten -megabytedata volume on the VAXcluster server,depending on needs. This volume is theuser's own private area, and cannot beaccessed by another user unless theowner allows it. There are two systemvolumes that support PC -DOS 3.1 forthe IBMs, and MS-DOS 2.11 for theDEC Rainbows. Other than having dif-ferent versions of DOS utilities on them,these system volumes contain basicallythe same application programs wher-ever possible. The system volumes areread-only so users cannot alter theinstallation parameters of the individualpackages once they have been estab-lished.

The 3Com 3Server not only bridgesthe gap between the Apple Macintoshesand the rest of the network, it providessecure storage for a management data-base. Although VIM stores the DOSfiles in a translated manner (from aVMS standpoint), additional securitywas desired. Users who need an extralayer of security for their data store it onthe 3Server, which is kept inside alimited -access computer center. At theend of each day the 3Server is backed upand powered down, making the datastored on it unaccessible.

The mail utility allows users to sendmessages as well as binary files. When auser receives a message with an attachedbinary file, both can be stored for lateruse with another application. This fea-ture is especially useful for spreadsheetfiles that must be updated periodicallyby various people, or for the productionof large documents by combining anumber of smaller documents that havebeen mailed from different sources.

A site license was obtained for a DECVT100 emulation package (PCVT). Itallows the IBM and compatible PCs tocommunicate with the VAXclusterthrough an asynchronous serial connec-tion. Some software, such as 1-2-3 fromLotus Development Corporation, is notavailable with a network license, butemploys a copy -protection scheme thatrequires a "key diskette" to run. Thisdifficulty has been overcome withoutviolating the license agreement by eras-ing (not formatting) the files on thediskette and making it a DOS systemdiskette with network drivers and utili-

Abrams/Peters/Rosenthal: Expansion of the RCA engineering network 5

G.O. Abrams J.K. Peters H. Rosenthal

Expansion of the RCA Engineering Networkby means of PCs

ATL's personal computer local area network is designed to takemaximum advantage of, integrate with, and build upon the existingATL VAXcluster and RCA Engineering Network.

Enhancement of the data handling andprocessing capabilities for personnel inthe Advanced Technology Laboratories(ATL) required the addition of com-puter processing power for manage-ment, support and technical engineeringuse. Additional power was obtained viaestablishment of a personal computer(PC) local area network (LAN).

The resultant implementation allowsPCs to network directly and to integratewith the ATL local VAXcluster, com-prised of a combination of VAX 8600/780/750 equipment running VAX/VMS. In addition, a variety of othercomputers are connected to the VAX -cluster. As the hub of the RCA Engi-neering Network, the VAXcluster inter-faces with over 40 VAXs and theATL-resident IBM 3083-J of the Engi-neering Corporate Computer Center.

The ATL network of IBM, Macin-tosh, and Rainbow PCs provides bothshared hardware and software

Abstract: Enhancement of the datahandling and processing capabilities formanagement, support, and engineeringpersonnel at the Advanced TechnologyLaboratories (ATL) required theaddition of specialized computerprocessing power. This was achievedthrough the establishment of a personalcomputer network integrated with theATL VAXcluster.

@1986 RCA Corporation.Final manuscript received November 20, 1985Reprint RE -31-2-1

resources. System -wide backup is easilyprovided. The PC network performshigh-speed (10 Mbs) data transfers viaextensions to an existing ATL Ethernet.In addition, a separate managementdatabase is provided by an independentserver whose files cannot be accessed byunauthorized users.

The system will be discussed from thestandpoint of design requirements,alternatives investigated, and utility.

Requirements

The goal was to take maximum advan-tage of, integrate with, and build uponthe existing ATL VAXcluster and RCAEngineering Network. Asynchronoustransmission of ASCII and binary fileswas one requirement-others includedsupport of IBM, Apple Macintosh, andDEC PCs; ease of use; high speed/performance; scheduled backup; trans-parent operation; reliability; versatility;data security; expandability; minimalincremental costs of adding users andincreasing disk storage; PC -to -PCdirect transmission and electronic mail;shared resources such as high -qualityprinters and large -capacity disk drives;the ability to move applications fromthe VAXcluster to the PCs in order tofree VAX resources for large jobs; andeasy maintenance. Additionally, man-agement wanted to avoid completelyrewiring the building since two Ethernetcoaxial cables were already in place.

A formal study was done to deter-mine the level of functionality that the

general ATL userbase expected from aPC. After the results were tallied, stan-dard PC configurations were defined.The base system consisted of a dual -drive IBM-PC XT with a 12 -inch (diag-onal) amber display, a 3Com Etherlinkboard, a Tseng Laboratories multifunc-tion graphics board providing720 x 348 -pixel resolution, a real-timeclock with battery backup, RAMexpansion up to 640 kbytes, a parallelprinter port, and an asynchronous serialport. Okidata Microline 193 printerswere chosen for local non -secretarialoutput. Their characteristics includea draft -mode print speed of 160 char-acters per second and compatibilitywith standard IBM PC graphics. Theselected DEC Rainbows have 256 kbytesof RAM, an amber display, and a DECLA100 printer. The Apple Macintosheshave 512 kbytes of RAM, an externaldisk drive, and a numeric keypad.

Alternatives investigated

"Walk -LAN"

The first design alternative investigatedwas "Walk -LAN" (i.e., non -networkedPCs). Although this would be very flexi-ble in terms of relocation of equipmentand low initial cost, it provided noshared resources, scheduled backups,VAXcluster integration, or PC -to -PCmail utility. In addition, having to"walk" floppies throughout ATL wasconsidered too time consuming and,therefore, not very cost-effective or effi-cient.


in this issue . . .

PCs & engineering workstations

Abrams/Peters/Rosenthal: "The ATL network of IBM, Macintosh,and Rainbow personal computers provides both shared hardware andsoftware resources."

Judice: "Somerville's VAXcluster is the largest such installation inRCA, and supports one of the largest user communities of anydivisional computer center in the corporation."

4

VAXcluster

4.17

9.6 Kbis CONNECTIONM

C0M

Stern/Voto: "In a unique engineering application, an IBM PC is usedin conjunction with an internal combustion engine test set (STE-ICE) forfield data collection and development."

Butt/Miller/Hoke/Piazza: "We have developed an applicationsprogram that allows our designers to enter their parts list informationinto a text file while they are working on their drawing."

Marett: "Most of the commands of PERSONAL870 are the same asthose used by the Applicon software."

Solomon: "Our experience has shown that reliability predictions canbe done using Monte Carlo simulations instead of analytical solutions."

Kunz/Miller/Renfro: "This integrated approach allowed us to build asystem that is reliable, has fast response times, supports a largenumber of users, and is easily taught to new users."

Law/Bernard: "Note that once an actual feature algorithm has beenidentified, without the functional simulation capability, the building ofthese features on breadboards would have been required."

Lazarus/Kelley: "Repeated attempts to stretch PC -based systems toplan multiple large projects were only partially successful."

Kjellquist: "With a superbly engineered man -machine interface and asingular focus on a weighted -criteria analysis, Lightyear is a uniqueengineering tool."

Barton: "For some time we have been recommending the Macintoshas a low -end graphics terminal."

A 0

CONVERTER'

DIGITALIBM PC -AT

CIRCUITRY

12

29

Monte Carlo

34 IBM4381 - 02

58

-I SMD

-20 PIS

25

531 AM/EZZ-77

2 PROCESSOR AAA

PROCESSOR RHM

4 PROCESSOR ROB

5 PROCESSOR CCC

6 PROCESSOR III

7 PROCESSOR SEE

AN/XXX-99

9 PROCESSOR DDD

10 PROCESSOR OGG

11 PROCESSOR FFF

12 AN/Y7Y-88

o tgEgigig I I

kipp00000000

loomommon

42

terminal emulation

Zarodnansky: "... just because there are some fancier machinesaround doesn't mean that you should relegate your PC to the scrapheap yet."

Stranix: "Software . . . is an intellectual property that has valuebeyond its direct utility to RCA."

in future issues . . .

laser technologymanufacturingquality

LASER

OPERATING

alCBM EngineerVol. 31 No. 2 Mar. Apr. 1986

networks 4 Expansion of the RCA engineering network by means of PCsG.O. Abrams J.K.1 Peters 1H. Rosenthal

7 Engineering computer services keeps Somerville onlineL.J. Judice

applications 12 Personal computers in the engineering workplaceM.D. Stern F.A. Voto

18 Comprehensive application of a CAD/CAM system to mechanicaldesignS.W. Butt H.P. Miller D.N. Hoke IF. Piazza

25 Analog printed -circuit board design using an IBM PC ATT. Marett

29 Spacecraft reliability predictions by computer simulationJ.G. Solomon

34 IBM graphics workstations at Video Component and Display DivisionP.J. Kunz R.N. Miller 1 G.M. Renfro

42 Simulation of digital TV signal processing with an engineeringworkstationK.A. Law IFS. Bernard

50 Automated project management with PCs and a mainframeF.F. Lazarus iW.T. Kelley

53 Lightyear-an engineering analysis toolD.T. Kjellquist

58 Using the Macintosh as an engineering workstationD.P. Barton

The Engineer's Notebook 70 Dressing your PC for successD. Zarodnansky

74 Software, a valuable by-product at RCAA.J. Stranix

departments Patents, 761Pen and Podium, 77 1News and Highlights, 79

Copyright 1986 RCA CorporationAll rights reserved

Personal computing workstations in engineeringLower cost and increased functionality have madePCs and Workstations (PCWs) the ubiquitoussliderule of the decade. PCWs have greatly changedthe engineering environment. A continuum ofcomputing power is now available-from PCWsthrough micro and minicomputers to mainframesand supercomputers. The distinctions amongcomputing machines are becoming increasinglyblurred.

What are the opportunities and problems arisingfrom this technical cornucopia? First, it is clear thatsignificant administrative and technical tasks will bedone on PCWs. Strong productivity gains are alreadydocumented. The market is brimming with usefultools aimed at the engineer/manager, and these willincrease in power. Second, as change accelerates,the need to integrate the work product of individualswith the broader purpose of organizational groupswill place a high premium on sharing andcooperation. More engineers will work acrossmultiple computer systems to solve their technicaland administrative problems. Better tools merely letus solve bigger problems.

Many business units have invested in PCWs, butmostly on an ad hoc basis with limited access. Toooften, progress is driven by departmental technicalactivists without senior management involvement.Several Divisions have already achieved significantresults from the coordinated use of PCWs on a near -

saturation basis and are at the beginning stages ofbroader system integration.

What do we need? Helter-skelter individualismmust yield to shared approaches that are driven fromlarge, sophisticated computer systems, strong inintegration and communications, and from a variety

of PCWs and stand-alone, dedicated machines.Unfortunately, today's technology cannot combine allapproaches into a seamless whole; the crunchcomes at the interfaces between dissimilar systems.

The technical community requires a communalvision or purpose: an evolving "single technicalcomputing system"-not one vendor or one way, butrather an overall view of how best to assemble andmesh smaller discrete systems into a larger,cooperative whole. This is a bold step, truly beyond acomprehensive solution today, but a requirement forour future business.

There are several principles that can be followedincrementally: a guiding architectural view,integrated telecommunications, gentle but firmsystem standards, a cautious narrowing of thevendor base, and use of common systems forcommon tasks. The benefits of following theseprinciples are enormous, but we must use care to doit right the first time. Tip -toeing into sharedcomputing on a massive basis is like staging a balletin a mine field.

Looking forward to our new relationship withGeneral Electric Company, we can anticipate evenmore support for this vital area.

James C. MillerDirector, Engineering ComputingCorporate Information Systems and Services

Editorial Representativesand Technical Publications Administrators

Contact your Editorial Representative at the Tacnetnumbers listed below to plan your RCA Engineer articleand to announce your professional activities.

*Technical Publications Administrators, responsiblefor review and approval of papers and presentations,are indicated here with asterisks before their names.

Advanced Technology Laboratories Tacnet Network Services Tacnet*Merle Pietz Moorestown, New Jersey 253-6429 *Richard Spencer Princeton, New Jersey 272-7750

Ed Master Moorestown, New Jersey 253-6436New Products Division

Aerospace & Defense Staff*Pete Bedrosian Lancaster, Pennsylvania 227-6880

*Theresa Law Cherry Hill, New Jersey 222-5833 Bob McIntyre Ste Anne de Bellevue 514-457-9000(Approvals only)

Patent OperationsAmerican Communications George Haas Princeton, New Jersey 226-2888*Anita Labib Princeton, New Jersey 258-4346

(Approvals only) RCA CylixCarolyn Powell Princeton, New Jersey 272-4194 Paul Dixon Memphis, Tenn. 739-0214

Astro-Electronics Division RCA Laboratories*Frank Yannotti Princeton, New Jersey

Carol Coleman Princeton, New Jersey229-2544229-2919

*Julie Dann Princeton, New Jersey 226-2061(Approvals only)

Automated Systems Division Eva Dukes Princeton, New Jersey 226-2882

*Gene Galton Burlington, Massachusetts 326-3775 RCA Records DivisionLinda Kriesman Burlington, Massachusetts 326-3001 *Greg Bogantz Indianapolis, Indiana 424-6141

Communication and Information Systems Division RCA Service Company*Dan Tannenbaum Camden, New Jersey 222-3081 *Murray Kaminsky Cherry Hill, New Jersey 269-5418Harry Green Camden, New Jersey 222-2423 Dick Dombrosky Cherry Hill, New Jersey 269-5618Bruce White Camden, New Jersey 222-3908 Ray MacWilliams Cherry Hill, New Jersey 269-5005

Consumer Electronics Operations RCA Technical Excellence Center*Eugene Janson Indianapolis, Indiana 422-5208 *Tony Bianculli Princeton, New Jersey 226-2111John Hopkins Indianapolis, Indiana 422-5217 (For corporate approvals)Larry Olson Indianapolis, Indiana 422-5117

Corporate Information Systems & Services Solid State Division*John Schoen Somerville, New Jersey 325-6467*Sue Handman Cherry Hill, New Jersey 222-6242 Sy Silverstein Somerville, New Jersey 325-6168

Global Communications Virginia PateHarold Ronan

Somerville, New JerseyMountaintop, Pennsylvania

325-6704327-1473

*Dorothy Unger Piscataway, New Jersey 335-4358 or 327-1264John Young Findlay, Ohio 425-1307

Microelectronics Center*Susan Suchy Somerville, New Jersey 325-7492 Video Component and Display Division

*Ed Madenford Lancaster, Pennsylvania 227-6444Missile and Surface Radar Division Lou DiMattio Scranton, Pennsylvania 329-1435

*Don Higgs Moorestown, New JerseyGraham Boose Moorestown, New Jersey

224-2836253-6062

Nick MeenaJ.R. Reece

Circleville, OhioMarion, Indiana

432-1228427-5566

Jack Friedman Moorestown, New Jersey 224-2112

National Broadcasting Company*Bob Mausler New York, New York 324-4869

March 26. 1986

RCA EngineerA technical journal published bythe RCA Technical Excellence Center"by and for the RCA engineer"

13 Roszel Road, P.O. Box 432, Princeton, NJ 08543-0432

Address correction requested

Printed in the U.S.A.Form No. Re -31-2

BULK RATEUS Postage

PAID

Phila., Pa.Permit No. 2906

47E080 rNIIIH A?J2i31-13

Ob SNIN13 EOZNOSSON r3

Color is often used in computer graphics toenhance the display of selected information,especially when the data set is complex orbusy. These three examples are from theKunz/Miller/Renfro article on page 34.

At the left is a 3-D wireframe model of a26V COTY picture tube. At the bottom left isa complex CADAM isometric. At the bottomright is a false -color plot of a picture tubemask -frame assembly. The colors in this lastphoto represent z-axis displacements of theassembly resulting from a given thermaldistribution.

ColDS ! 0,11,141 10./.-45 .4

l0.0 04,141 1

a Rt.. 2,9.as ]DIS,,EMIxf!L OK. r.1114 ..1114 , r,Il Ins

Documents

americanradiohistory.comamericanradiohistory.com/ARCHIVE-RCA/RCA-Engineer/RCA-Enginee… · mcoaEngineer A technical journal published by the RCA Technical Excellence Center 0 13