aiMa Engineer

C I 1rCAA.idiv 111/4)t asJ

Vol 23 No 4Dec 1977Jan 1978

advanced communications for the nineteen -eighties

Our cover is an interpretation ofworldwide communications byBob Canary, who is a graphicdesigner at ATL in Camden, N J

alma EngineerA technical journal published byRCA Research and EngineeringBldg. 204-2Cherry Hill. N.J. 08101Tel. PY-4254 (609-779-4254)Indexed annually in the Apr/May issue

EditorAssistant Editor

Art EditorContributing EditorCompositionEditorial Secretary

Div. VP. Engineering,Consumer ElectronicsVP. Tech. OperationsRCA AmericomDiv. VP, EngineeringPicture Tube Division

VP and Technical Director.Globcom

Manager, TechnicalInformation ProgramsChief Engineer,Broadcast SystemsStaff VP. EngineeringDiv. VP, Solid StatePower Devices

Director. EngineeringProfessional Programs

Chief Engineer,Missile and Surface RadarVP. Laboratories

Ldr , Presentation Services,Missile and Surface RadarMgr., News and Information.Solid State DivisionMgr . Scientific Publications.Laboratories

To disseminate to RCA engineers technical information of

professional value To publish in an appropriate manner important technicaldevelopments at RCA. and the role of the engineer To serve as a medium of interchange of technical informationbetween various groups at RCA To create a community of engineering interest within the companyby Stressing the interrelated nature of all technical contributions To help publicize engineering achievements in a manner that willpromote the interests and reputation of RCA in the engineering field To provide a convenient means by which the RCA engineer mayreview his professional work before associates and engineeringmanagement To announce outstanding and unusual achievements of RCAengineers in a manner most likely to enhance their prestige andprofessional status

Professional satisfactionAlthough my viewpoint may be biased by my own experience as a design engineer, Ibelieve that every engineer, in every field, has chosen and practices daily a highlycreative profession.

As individuals, oJr approaches to daily work are influenced by many complexfactors, but one of the most important job -related motivations for engineers is thesatisfaction they derive from seeing the results of what they have done put to usefuland successful purpose. I believe that the responsibility for achieving this creativesatisfaction lies in at least three areas:

Engineers must first know and then continually develop their own unique talents,and understand and grow in their own disciplines. They must involve themselvesin developing something new, something better, something useful.

Engineering managers, for their part, must understand the capabilities of theirengineers, and give them the tools, experience, and resources necessary to trulysucceed and to make genuine contributions. Engineers' creative instincts mustbe nourished through challenging assignments.

Those of us in business and product management must select, for theengineering team, those viable projects that will result in successful, salableproducts. In this way, the engineers will be able to see, in actual constructive use,the end result of their labor and creativity.

When all three groups fulfill their responsibilities in these areas, we all benefit.Engineers enjoy the rewards of creative satisfaction; management can measure thepositive tangible results of their own skills; and the company succeeds in themarketplace.

J. E. HillDivision Vice Presidentand General Manager,Broadcast Systems,Camden, N.J.

advanced communications for the 1980s

- tv broadcasting-camera to antennaThe industry is changing. Here are some examples:

electronic -journalism camerasmicroprocessor -based editingautomatic tv transmissioncircularly polarized antennas /

5, 8, 15, 20, 29 --I

N' -communication in Alaska

Climate, terrain, and low population density used tohamper efficient communication in Alaska. Now,though, they're pushing the state to the forefront ofsatellite communication.

test yourself

coming up

Take two minutes to find out if you're a high or lowachiever relative to the rest of RCA's engineers. High andlow achievers have different work habits and attitudes; theEngineering Information Survey results tell you what theyare.

Our next issue (Feb/Mar) looks at radar. See what the new developments are for radar inmilitary systems, weather radar, and even radar for blast furnaces.

Later issues will have software, space technology, and manufacturing themes.

WHERE D2 You 1

FIT IA) ?

imr28a EngineerVol 231 No. 4 Dec 771 Jan 78

editorial inputC.W. Sall Life in the communication factory

broadcast communications

A.C. Luther Broadcast communications-review and surveyS.L. Bendell 3 Electronic journalism with the TK-76 camera

K.J. Hamalainen 15 Video tape editing systems using microprocessors

R.W. Zborowski 20 Automatic transmission systems for television

L.L. Ourslerl D.A. Sauer 24 The BTA-5SS-RCA's all -solid-state 5 -kW a.m. broadcast transmitterM.S. Siukola 29 Circular polarization for better tv reception

other advanced communications

J.L. Rivard 35 Long -line communication in Alaska-then and now

M.E. Logiadis 42 Statistical coding methods speed up image transmission

J.R. Richards1W.F. Meeker 46 Linear predictive coding to reduce speech bandwidth

professionalism

H.K. Jenny) W.J. Underwood 50 Engineering Information Survey results-Part 2

general interest papersC.A. Berard 5b A high-performance switching regulator for advanced

spacecraft power systems

H.R. Ronan 61 The Universal Second -breakdown Tester:electronic cage for the power dragon

H.R. Barton 68 Spares allocation for cost-effective availability

departments71 Dates and Deadlines72 Patents74 Directory of licensed engineers at RCA76 Awards to RCA engineers78 News and Highlights

Copyr ght c 1978 RCA CorporationAll rights reserved

editorialinput

Life in the communication factory

It's been a long summer. Thirty-five years ago, on June 18, 1943,when I was still in the school -teaching business, RCA gave me aJuly -August summer job at the Industry Service Laboratory inNew York. Now that long summer is about to end, ironically incold January rather than in August. Makes me want to sneak off toFlorida for a couple of months.

Anyway, my thirty-five year stint with RCA in the technicalcommunication field has given me many a window to look inthrough and out of-and the scenes in either direction have beeninteresting, stimulating, and, on occasion, fascinating. Admitted-ly, most of the windows through which I have viewed theemanations of the paper age have been in the electronicsresearch sector, but in traveling around a bit I've also peered intoquite a few product -division windows here and there. Now, as Icome to the end of the line, let me share with you some of thesights and insights stemming from my journey.

Have you ever seen a young father gazing in rapture at his firstoffspring through the window of the newborn -baby nursery? Hislook of pride and joy is akin to the expression I've often observedon a young engineer's face as he views his first technical article inprint. It is a sight to behold, and it runs counter to the moth-eatennotion that scientists and engineers dislike to write or speak. Ithas been my experience that given the proper vehicle or forumfrom which to address their peers in the technical world,engineers are usually cooperative and willing to participate.Naturally, oral or written expression comes easier to some than toothers, but with a bit of coaxing, encouragement, and incentive,even the shy engineer will try his hand at preparing an article for ajournal or a conference (particularly if the conference is in Miami,San Francisco, or Europe). Once he's fallen in love with that firstbaby, he's ready to repeat the performance, and a seasoned writeris often in the making. One of my great joys has been to witnessthis metamorphosis from neophyte to pro by many members ofour staff, often getting their start in the pages of the RCAEngineer. This same experience must also have been shared bymy colleague TPAs and Editorial Representatives throughoutRCA.

In similar vein, I have also seen the more prolific writer -engineersclimb the ladder of success to sometimes spectacular heights.How about these RCA engineers as examples: Zworykin,Engstrom, Brown, Donahue, Hillier, Herold, Williams, Webster,Schade, Goldsmith, Luther, Rose, Rajchman, Sonnenfeldt,Olson, Vollmer, Vonderschmitt, Heilmeier, Powers, and Tietjen!These men not only had something to say, but more important,they said it. And from their saying it, they, and the world, benefitedbeyond measure.

In the earlier years of my time with RCA, the exciting subjectswere transistors, television from black -and -white to full color, 33 -and 45 -rpm phonograph records, electronic computers, andradar. Now that those areas have become an accepted part of thesocial scene, new and even more exciting items are emerging-lasers, microprocessors, integrated circuits, fiber optics, CCDs,satellite communications, home video-tape recording,

automotive electronics, medical electronics-and still more.What splendid opportunities lie ahead for aspiring engineerauthors!

Luckily, this business of technical communications is a two-waydeal, looking in and looking out. On looking in, we observe thepulse of our own operations; on looking out, we see what othersare doing. There we become part of the IEEE, the AIP, the ACS,the ECS, and the several other science -oriented organizationsthat look to us for support even as we share in their offerings. Bysuch association, whether through their literature or via thepersonal relationships that stem from membership and participa-tion, the mutual rewards are great. From my window I've seenmuch proof of this.

My personal reward over the past third of a century has been thepleasure of my association with so many of RCA's dedicatedscientists and engineers, and as TPA, of having had some smallpart in helping to get their fruits of tongue and pen distributed to awaiting society. Corollary to this activity has been a mostrewarding association with the professional communicationspeople of RCA who, for the past 20 -odd years, have met regularlysix or more times each year to review and plan, and often tomeditate on, the outpourings that will bring yet new scientificinformation to man, and possible fame to the authors.

Yes, indeed, it's been a long, great summer!

-Chet Sall

Ed note: Chet Sall is retiring after thirty-five years with RCA. Most ofthat time he's been helping engineers get their names in print. He'sdone it well, but he's had to wear several hats to do it-planner,psychologist, editor, ghost-writer, mediator, wet nurse,administrator, whipping boy, and confidant. This experience hasgiven Chet some unique insights on professional technical com-munication that we've asked him to share in this guest editorial.

4

Broadcast communications review and survey

A.C. Luther Broadcasting is a mature industry, but technologicaladvances and new operational needs keep it changing.

Television and radio broadcasting is so much a part of thepresent-day scene in the United States that one could easilyoverlook that it is still developing and growing. Yet, in thelast five years, major changes in techniques and servicesprovided have occurred, and still more changes are visiblein the near future. Several of the articles that follow in thisissue of the RCA Engineer go into detail about specificpresent and near -future technological advances. Thispaper gives an overview of broadcasting and its equipmentneeds to help put those technological subjects intoperspective.

Broadcasting worldwideBy visiting various places around the world today, one cansee radio and television broadcasting systems at variousstages of evolution. In the United States, Western Europe,and Japan, for example, complete broadcasting systemsare fully in place and developed to an advanced level oftechnological sophistication. However, these countrieshave different system standards (NTSC-pAL-SECAM-AM-FM, etc.) that affect equipment considerations. Also, broad-casting facilities have wide differences in the ways theyserve the public-education, entertainment, promotion,sports, politics, advertising, news, and so on. These varioususes, which are differently distributed around the world,place a differing emphasis on certain equipmentcharacteristics.

Lesser -developed areas of the world are still building theirinitial broadcasting systems. These countries may not needthe same technological sophistication of service that in-dustrialized nations require; instead, advanced technologyshould be applied to make it easy for them to acquire,install, and operate their broadcasting systems. Since amajor problem in less -developed areas is the availability oftrained technical workers, these countries should havegreat interest in automated equipment that can simplifyskilled manpower requirements.

Broadcast equipmentThe broadcasting -equipment industry today is quite mature.

Numerous suppliers around the world compete in most ofthe world markets. Systems standards are published on aworldwide basis by the CCIR (Consultative Committee onInternational Radio).* Most equipment designs are second -or third -generation solid state, and most basic marketneeds can be fulfilled at a price. Present equipment -designeffort is strongly directed to improving the cost ofownership of future broadcast installations and to ad -

Arch Luther has been Chief Engineer of Broadcast Systems since1973. In his 27 years with RCA, he has worked on color-tv studioequipment, many video tape recorders (including the Emmy-winning TCR-100 video cartridge recorder), and other studio andtransmitting equipment for radio and television.Contact him at:Broadcast SystemsCommercial Communications Systems DivisionCamden, N.J.Ext. PC -4312

vanced features for certain special applications. Newbroadcast products feature all of the state-of-the-arttechnologies-microprocessors, LSI, CCDs, etc.

Broadcasting equipment must be designed to deliver high -qualityservice over an operating life of at least ten years.

In many cases broadcast installations run 24 hours a day,seven days a week, so reliability and maintainability areparamount considerations. The basic structure of broad-casting (one transmitter and many receivers) means thatreceivers should cost as little as possible. Therefore, thetransmission performance should substantially exceedoverall system performance needs so that the performancerequirements on the receivers can be minimized. All of thismeans that broadcasting equipment becomes expensive

The CCIR gives technical advice on radio matters to the International Telecom-munications Union, which acts, in a sense, like an international FCC.

Reprint RE -23-4-151 Final manuscript received December 2, 1977.

5

when compared to similar functions in mass-producedreceivers; costs one or two orders of magnitude higher thanmass-produced items are not unusual. One of the broadcastengineer's greatest challenges is to keep equipment costunder control.

Broadcasting operationsThe functions of a broadcaster can be divided into two parts-origination and continuity.

Using cameras, recorders, and switching or mixing equip-ment, programs are put together from live input material. Inmost cases today, this process is done ahead of time and thetotal program is recorded on tape or film. The process ofpreparing recorded program material is called productionor teleproduction (for television). Most broadcasters dosome of this, but many organizations do onlyteleproduction-they are not broadcasters. Teleproductionoperations may be in business to produce total programpackages, or they may just offer technical services to otherswho provide the program content material.

Every broadcaster performs the second function of broad-casting. Called a continuity operation, it is the process ofassembling program material from various sources into acontinuous stream that feeds a transmitter. Fig. 1 showshow these two functions combine in a typical United Statestelevision broadcasting environment, including ournetwork system. Notice that news is shown as a completelyseparate operation at both the network and local levels.News is a specialized teleproduction and continuity opera-tion that should be explained further.

111111111=1

Recent developments at RCAThe need for immediacy in news programming is overwhelming.

Broadcasters compete actively to be first with a news break,and so want to minimize the time delay from shootingpictures at a news scene until finished material is availablefor airing. Much news programming is done with 16mmmovie film, and most broadcast stations have sophisticatedfilm operations for this purpose. Recently, however, therehas been a trend to doing news "electronically," usingtelevision cameras and video tape recorders. Called "elec-tronic news gathering" or "electronic journalism," it isgrowing because it provides greater immediacy, betterquality on -air, and lower cost of operation than filmsystems. Electronic journalism is leading to a specialcategory of equipment optimized for portability, flexibility,ease of operation, and reasonable performance. RCA'selectronic journalism camera, the TK-76, has been verysuccessfJI, with over 700 units sold by the end of 1977. SidBendell's article in this issue describes the camera and itscrash design program.

Video recording is the essential technology in the process ofteleproduction today.

Live program material from either studio or locationshooting is recorded on video tape. Then the material goesthrough the "post -production" processes, where it is cor-rected, adjusted, mixed, and assembled into the finishedprogram sequence. In these processes, where artisticfactors dominate, it is essential to have the greatest possibleflexibility for manipulating the recorded materials. This hasled to extremely sophisticated systems for controlling thevideo recorders and special peripheral equipment forediting the recorded material. RCA's microprocessor -based AE -600 time -code editing system is the subject of apaper in this issue by Jukka Hamalainen.

INDEPENDENT

TELEPRODUCER

TELEPR000CTIONOPERATION

TYPICAL BROADCASTERTELE PRODUCTION

OPERATION

NEWS

CONTINUITYOPERATION

TRANSMITTERPLANT

Yx

x

TELEPROOUCTIONOPERATION

TYPICALTV NETWORK

NEWS

CONTINUITYOPERATION

> TO OTHERBROADCASTERS

VIA LAND LINES.MICROWAVE. OR

SATELLITE

> FLOW OF RECORDED MEDIA

- ELECTRICAL SIGNAL

Fig. 1Broadcasting structure for U.S. television. Note division intoteleproduction and continuity operations at both the station andnetwork levels. News requires special treatment.

multiplexers

Digital techniques are becoming more important.

Although the signals for radio or television broadcasting arebasically analog, the use of digital technology has in-creased significantly for certain signal -handling or signal -processing functions. This is occurring because the low-cost digital circuit techniques that have been developed forcomputers can provide some features that are extremelydifficult for analog techniques to achieve. Probably themost significant one of these features is the memoryfunction. With digital LSI memory devices, even a completetelevision frame of information can be stored at reasonablecost. This makes possible numerous signal -processingfunctions for synchronization, noise reduction, or pictureenhancement. Bob Hurst will describe how far digitaltelevision has progressed in an article to be published in theRCA Engineer in the near future.

All -solid-state transmitters are replacing tube types.

Broadcast equipment today is all solid-state except forpickup devices, display tubes, and the higher -power stagesof transmitters. However, solid-state devices are becoming

What equipment does RCA make for thebroadcasting industry?Just about everything, to put it briefly. This abbreviatedlist and the photos on these two pages show the breadthand completeness of RCA's broadcast line.

Video tape recordersTV camerasTelecine cameras, projectors andTV slide projectorsTerminal and switching equipmentRadio and tv antennasRadio and tv transmittersTransmission lines and related equipmentOutside broadcast vansSound -on -film recording equipmentBroadcast audio equipment

rons- -1

ti IN

more capable and even the high -power circuits are going tosolid state. RCA's new all solid-state 5 -kW and 10 -kW a.m.radio transmitters, described by Len Oursler and DaveSauer in this issue, are a good example. Here the solid-stateapproach is providing higher power efficiency, better on -airreliability, and better performance than previous tube -typetransmitters.

Another new development that has received a go-ahead in theUnited States is circularly polarized broadcasting for television.

Circular polarization promises improved reception andreduced "ghosting" in areas with difficult multi -pathproblems and where portable antennas such as "rabbitears" are used. The FCC authorized circular polarization fortv broadcasting in the late spring of 1977, and four stationshad ordered the new type of antenna by the fall. Circularpolarization is the subject of Matti Siukola's article in thisissue.

ConclusionBroadcasting today is very much a worldwide industry. It isgrowing dynamically, with new features and services beingdeveloped and offered. Likewise, broadcast equipment hasembraced the latest technological developments to providemore and more capability at lower cost to the broadcaster.With this kind of environment, we can expect the broadcastindustry to continue its growth and success pattern for theforeseeable future.

2 3 4 5

1 TK-760 studio camera. 2 TP-55 telecinemultiplexer. 3 Antenna installation atChicago's John Hancock building. 4 TT-25FL transmitter. 5 TR-600 video taperecorder with AE -600 time -code editingsystem and moiitor bridge. 6 Mobilebroadcast van. 7 TCR-100 video casetterecorder. 8 Traveling -wave antenna forhigh -band uhf stations.

7

Electronic journalism with the TK-76 camera

The TK-76 camerastarted a revolution

in tv newsgathering.

S.L. Bendell

"Electronic journalism" has become, during the past fewyears, the most powerful way to cover news and livegeneral -interest events for television. In large part, thisrevolutionary newsgathering technique has been closelykeyed to technological advances in the field of televisionpickup and recording equipment.

Forces behind "the revolution"Up to perhaps five years ago, news pickups were done almostentirely on 16mm motion -picture film.

The film, following coverage of the event, would have to bedeveloped, edited, and then converted to the requiredelectronic format by a special "telecine" tv camera.

The number of steps in this total analog process and theopportunity for cumulative picture degradations to occurwas great indeed. Only a small part of the original film(estimates are from 5% to 100/o) was ever useful for finalairing, for reasons of technical quality as well as content.Although stations would have liked routine direct pickup ofunprogrammed news events on live color-tv cameras, it wasonly a theoretical possibility prior to 1975 because of thelarge size, complication, and logistics of tv-camera equip-ment. Fast -breaking news required the mobility of smallman -carried 16mm movie cameras, but the developmentand editing time lag of the film process prevented newsteams from filming events occu'ring later than about twohours before air time.

What brcadcasters wantedSteadily as the importance of tv news coverage grew, a realneed also grew for a better system which would have thefollowing features:

1) The immediacy of live tv pickup with the opportunity ofreal-time news coverage.

2) Lightweight, easy -to -operate equipment to achievemobility with minimum on -site crew size.

3) A correspondingly small recording/storage systemwith the lexibility of immediate playback and reusablestorace medium.

In short, the broadcasting industry needed small color-tv camerasand video tape recorders capable of being mancarried.

Fortunately, fast -paced developments in both tv camerasensors (a dominant component determining camera size)and small nelical-scan video tape recorders were beingrefined rapidly. Specifically, the sensor activity was beingdirected toward both solid-state charge -coupled devices(CCD) and 2/3" photoconductor tubes. These two ap-proaches were highly competitive by mid -1975, and RCAcarried out considerable testing and system evaluation todetermine the best sensor to use in a small color cameraweighirg less than 20 pounds.

As it turned out, solid-state sensors were then not yetsufficiently developed to meet tv broadcast performance

requirements. For example, an experimental 3 -sensor CCDcamera developed by RCA was demonstrated at the NABConvention in April 1975 in Las Vegas. While this ex-perimental camera generated considerable interest amo igthe broadcasters, it had some limitations that made it, at thattime, noncompetitive with the photoconductor -tube types.Its major limitation was sensitivity; when it was comparedwith a small tube -type camera, it fell sitort by a factor ofabout 50X. The CCD camera's cosmetic defects and limitedresolution also favored the tube approach.

Considerations and tradeoffsRCA demonstrated an advanced development model of anelectronic journalism camera (Fig. 1) at the 1975 NABConvention in April and at the Montreux TV Symposium inJune. This camera was a first pass at designing a systemthat would best fit the needs of electronic newsgatherirg.Many factors were considered in making this preliminarydesign, whicn formed the basis for the eventual productversion, named the TK-76. A partial listing of some of theseconsiderations follows:

Sensitivity. The tv camera would have to compete w th16mm news cameras, which usually use high-speed fi mhaving an ASA rating of 125. Such cameras were capableof taking pictures at light levels down to 40 foot-candleswith lenses having maximum apertures of f:2.0. Uncerpushed film development, the 16mm cameras coldoperate down to 12 to 15 foot-candles, but with noticeablepicture degradation. Films introduced more recently arerated at ASA 400 and can be pushed to over ASA 1000. (AtASA 1000, a film camera can operate at about 5 foot-candles.)

Fig. 1Advanced development modelof the TK-76 showed it coldbe done, but still neededrefinements (see Fig. 2 forcomparison with finishedproduct).

Light weight. By observing closely the form that most ofthe successful 16mm cameras had taken relative toweight, size, shape, and man -machine interface, thephysical characteristics of the TK-76 became concep-tualized. Many mockups were assembled for critiques bybroadcasters, 16mm news camermen, and many others,including our very enthusiastic top management whotook an intense interest in the TK-76 program. A weight of15 to 20 pcunds seemed most desirable. Lighter weightwould adversely affect camera holding stability; moreweight would be too bu -densome for the cameraman. Thecamera should be a single package with a separatelightweight battery belt.Low power. Most broadcasters indicated the need for aminimum battery -operating time of 90 minutes. Bylimiting the battery -belt weig it to 4 to 5 pounds and usingNiCad batteries (the most practical type), camera powerrequiremer ts should to sorething less than 35 watts.The power source should be a single -voltage 12-V supply,the most universally available source.Stable, reliable, and rugged These three requirementsare all important, but tie last one has become far moreimportant than we ever could have imagined; several TK-76s delivered since the first product shipment in April1976 have suffered such fates as being dropped out ofhelicopters and flung from speeding motorcycles. Opera-tion over a large range of environmental conditionsshould be expected in newsgathering.Simple no -hands operation. Traditionally, tv colorcameras have contained a very large number of controlsrequiring skilled attention over a wide range of intervals.However, operating personnel for the tv news camerawould quite often be film cameramen or other non-electronic types having little or no experience in tv-camera setip. Therefore, a pt mary concern for the newscamera was that it would require no technical attentionfor extended periods uo to several months; the rationalefor much of the mechanical and electronic designfollowed from this very rigorous requirement. Downtimefor broadcast equipment means lost dollars for thebroadcaster. Therefore, equipment packaging wouldneed to favor quick checkout and repair when necessary.Performance. At the very beginning of the electronic -journalism development, equipment makers and usersgenerally assumed and accepted that small cameraswould necessarily produce a compromised level ofperformance relative to a top -line studio camera. Thisview was very short-lived, however, and soon gave way tothe more realistic conclusior that picture quality from tvnews cameras would have to be practically indis-tinguishable from their big brothers in the studios.Manufacturers who initially offered limited -performancecameras fcr electronic journalism eventually fell by thewayside. Performance parameters of major importanceare colorimetry, resolution, signal-to-noise ratio, andregistration.International markets. A growing segment of the tvbroadcast business is in countries that have differenttelevision standards than the American NTSC standard.Packaging and partitioning of the new camera's elec-tronic circuitry should facilitate cost-effective conver-sion.

9

Fig. 3TK-76 system consists of the camera and video tape recorder, which aretied together with an "umbilical" cord. Battery belt weighs only fivepounds.

Fig. 2Production version of TK-76 was coming off the assembly lineonly twelve months after the advanced development modelwas shown.

Acceptable initial equipment cost/operating cost. The16mm film cameras most commonly used for tv news arerelatively inexpensive ($5,000 to $15,000) compared withhigh -quality color-tv cameras. This difference is partiallyoffset by the added cost of film development and therecurrent high cost of the film itself. To the user, the costequation is obviously very complicated. However, for tvnews cameras to be a viable business for RCA, it becamepainfully apparent that keeping the product cost to anacceptable low value would be an extremely challengingassignment for all elements of the program, includingmarketing, manufacturing, and design.

The camera sensor choiceThe advanced development version of the TK-76 alluded toearlier answered some of the questions posed by the abovelist of requirements and paved the way for the productversion of the TK-76 (Figs. 2 and 3). Our evaluation of manytypes of camera sensors showed that the 2/3" lead -oxidetube (the Plumbicon, manufactured by Matsushita andPhilips) was best suited for this application. The sensorparameters influencing this decision were:

Sensitivity. Using the 2/3" PbO, the camera can operate atexcellent S/N (51 dB) at an incident light level of 125 foot-candles (3200°K) with an aperture of f:2.8. A cameraequipped with such tubes and zoom lenses (availablewith apertures to /1.6) can operate at light levels down to40 foot-candles. The S/N margin is so great that a high -

sensitivity mode (increased video gain) is possible, whichfurther increases camera sensitivity by a factor of 3X andsti I maintains a very acceptable S/N.Resolution. At 400 tv lines, the 2/3" PbO has an apertureresponse of approximately 20% to 25%. This is about halfthat obtained with the larger 30mm lead -oxide tubes usedin the top -grade studio cameras. However, the lowex*.ernal circuit capacity associated with the smaller tube(2 to 3 pF) permits such a high S/Nat these frequencies (5MHz) that the loss in response at the high line numberscan be corrected by linear phase aperture compensationwith m nimal S/N degradation. The 2/3" Saticon tubes,which have recently been introduced for use in the TK-76as an alternative to 2/3" lead oxide, give a response at 400tv lines of 40% to 45%, thus making it even easier to matchthe resolution performance of the large studio cameras.(The Saticon is manufactured by Hitachi and is currentlybeing sold by RCA Electro-Optics and Devices in Lan-caster)

Lag. The image smear that occurs with picking up movingobjects is a type of fundamental picture impairment thathas high visibility in the final display. Lag performance inthe small -area 2/3" tube is significantly improved overthat of the larger tubes because of the formulation of thephoto layer, which provides lower storage capacitance inconjunction with the reduced scan area. Lag grows worseif the tube is forced to run at smaller signal currents (i.e.,higher sensitivity). However, using bias light canminimize lag. With the 5.0-nA bias light supplied in theprism optics of the TK-76, lag is held to quite acceptablevalues even when operating at high sensitivity.Many other characteristics must be considered in makinga cho ce of sensor, such as highlight handling,temperature stability, color response, transfercharacteristic, geometry, and cosmetic defects.

10

Fig. 4Optical system-lens, prism, yokes, tubes, and preamps-is mountedas a single unit and isolated from the rest of camera by three-pointshock -mount system. Method keeps system in precise registrationduring rugged use.

Optics and registrationPrism optics are now the most widely accepted color -separationmethod.

Prism optics provide, at highest efficiency, the properbandwidth color to each of the three (red, green, blue)pickup tubes. Very -high -index glass (n = 1.76), first used inthe TK-76, makes possible a very small prism having a highnumerical aperture and a relatively short total light path.This short light path is important because it lets the camerause small lightweight zoom lenses having high opticalspeed. A unique optical/mechanical system, used in thefirst developmental model and carried over to the productdesign, has simplified the manufacturing of the camera andprovided stability in operation.

The optical system must remain stable within one micron oftranslation.

In this system, the three tube/yoke assemblies are hardmounted to an optical bed plate that also rigidly supportsthe main picture -taking lens and the beam -splitting prism.The entire system (Fig. 4) is mechanically isolated from themain camera frame via a three-point shock -mount arrange-ment. This scheme protects the delicate and precisealignment of the various optical/mechanical elementsnecessary for preserving registration. To maintain 0.02%registration, the translational positional stability of thelisted components must be kept on the order of one micron.(Registration error between any two color -channel signalsis expressed as a percentage of picture height; 0.02% ismeasurable and 0.1% is quite useable.) The front surface ofthe optical plate is brought out to the front of the cameraand terminates in radiator fins, so the assembly effectivelysinks heat generated by the yoke assemblies andpreamplifiers.

Fig. 5Human factors were important in T K-76 design. Photos show howcamera is designed for easy balancing on cameraman's shou der.On/off/standby switch and other operator controls (kept to a bareminimum) are all located conveniently in one spot.

Human -factors considerationsCamera technology means nothing if the operator has troubleusing the camera.

As noted, equipment packaging and human engineeringplayed a vital part in shaping the mechanics of the camera.The eventual product version weighed 20 pounds, notincluding battery pack. (See Fig. 5.) The camera profile islow; and the electronic viewfinder (a 11/4" kinescope viewedvia a low -power magnifier and mirror system) position isadjustable. These two features allow the cameraman to seatthe camera directly on his shoulder in a balanced stance,which provides excellent camera stability, causes minimumfatigue, and permits comfortable viewing of the finder. Thesystem's center of gravity is approximately 5" above thecameraman's shoulder, with the camera lens axis at ap-proximately eye level. In normal picture -taking, thecameraman has an almost completely unobstructed view ofthe region to his left and right front sectors. Such anunobstructed view is very important to the safety of thecameraman who very often faces hazardous walkingconditions and even finds himself at times in hostile crowds.Normal operating controls are easily positioned so they canbe operated by "feel" with minimum chance of error

ElectronicsNew integrated circuits, printed -circuittechniques, and circuit designs sub-stantially reduced the physical size ofthe circuit boards.

Fig. 6 shows the camera's circuitryand Fig. 7 is a simplified blockdiagram. Here are some briefhighlights:

A miniaturized videopreamplifier housed within theyoke/tube assembly provides a51 -dB signal-to-noise ratio atnormal operating current levels.This is within 1.0 dB of thetheoretical limit.An integrated -circuit syncgenerator has a genlockcapability so that two or morecameras can be readily lockedtogether for multi -cameraoperation.The novel pulse -type powersupply has a constant conver-sion efficiency greater than 75%,even as the battery supplyranges from 10.5 to 15.0 volts.Pulsewidth control during thehorizontal blanking periodenables the necessary regula-tion mechanism to operate in atimely fashion that avoidspower -switching transients dur-ing active picture time.

RED

PICK-UPTUBE

RED

DEFLECTASSEMBLY

GREEN I GREEN

PICK-UP DEFLECTTUBE ASSEMBLY

L

PRE -AMP

-1.1PRE-ANP

BLUE BLUEPICK-UPS DEFLECTTUBE ASSEMBLY

IDEFLECT

LENS

Fig. 6Electronics depend on integrated circuits andnew printed -circuit techniques to keep the size ofthe boards as small as possible. Adjustmentsshown are all preset at the factory.

INPUTPROC

INPUTPROC

OUTPUTPROC

I IH/2HDELAY

IN AND 2H DELAYEDSIGNALS FOR CONTOURS

PRE-AMPI-:

AUTOBALANCE

ANDIRIS

INPUTPROC

OUTPUT

PROC

OUTPUTPROC

VIEW-FINDER

--111.

HORIZ.DEFLECT

COLORLUMIN- OUT I

ANCE COLOROUT 2

CHROMA

SYNC

PULSEAND

EXT LOCK

(GEN-LOCK)

IH DRIVE

POWER

SUPPLY

BATTERY OREXTERNALPOWER

Fig. 7Electronic system depends on pu.sed power supply to keep high efficiency over a widerange of battery voltages.

12

Automated control systemTo simplify camera operation, it is necessary to automate as muchcontrol as possible without unduly constraining the cameraman'sesthetic expression.

For example, white balance is automated on an "ondemand" basis; the cameraman points the camera at any"white" object (as small as 5% of picture width and one linehigh) and actuates the white -balance button. A cursor linethen shows up in the viewfinder display to guide him inaccurately placing the white object within the cursor -indicated sampling interval. Within a few milliseconds, thewhite -balance circuits electronically control (and store) thevideo gain of the red and blue channels and equalize themto the green signal level.

The automatic -iris mode is another control at thecameraman's option. It controls exposures by sensing peakvideo level and giving special weighting to the centralpicture area. For special lighting conditions, such as heavybacklighting, two other non -automatic exposure controlsare available. These operate by providing point -enhancedvisual indication of exposure in the viewfinder display.

A "zero -based" control philosophy made operation by technicallyunskilled people possible.

Traditionally, color cameras have required a very largenumber of operating and setup controls This could not betolerated in a small hand -carried camera operated underrugged field conditions by technically unskilled persons. A"zero -based" control philosophy was adopted early in theprogram-the system started with zero controls and neededones were provided only when they could be justified on arational basis.

Concept to production in thirty monthsChoosing the sensor came first.

The quest for a suitable sensor started in early 1973, duringwhich time a number of different types, including siliconvidicons and CCDs, were tested. Developmental samples ofthe first 2/3" Plumbicons were supplied to RCA by Philipsduring the 4th quarter of 1973. Analysis of these tubes andpreliminary evaluation of their performance in a colorsystem was carried out during the first quarter of 1974. Forthis purpose. a breadboard test camera was jerry-rigged,using an ancient developmental large -studio camera. Bythe spring of 1974 results using the Plumbicons lookedencouraging enough to warrant going ahead on the designof a developmental electronic journalism camera.

At this point, Mr. Joe Bulinkis, the RCA industrial designer,was called in, and a lively period of human engineeringactivity followed. Some five camera versions employingvarious packaging concepts were built in mock-up form andcritiqued by marketing, management, and some selectbroadcasters. By late spring 1974, the decision to go with asingle -package concept was made, and from then toFebruary 1975 the Advanced Development Camera Groupengaged in a frantic effort to design, fabricate, and test adevelopmental 20 -pound electronic journalism camera.

Everything going into the design literally had to start from scratch,including optics, yokes, circuits, power supply, and battery pack.

A successful landmark demonstration was given to RCA topmanagement in February 1975. The first advanced -development model worked well, but would have won noprizes for styling. This, coupled with management's deci-sion to show the unit in April at tne 1975 NAB Convention inLas Vegas, created a minor styling crisis. Again Mr. Bulinkiscame to the rescue and, with some artful cosmetic work,gave the camera a look of respectability for its debut in LasVegas.

As this effort continued, the Camera Product Design groupbegan work cn the TK-76 in the second quarter of 1975.Their schedule for the production version of the TK-76seemed well beyond reach at that time. The dedication andskill that the members of this group displayed in carryingout this seemingly impossible assignment won the entireteam a David Sarnoff Achievement Award. Enthusiasm onthe TK-76 program was contagious; all elements of theproduction team were caught up in the excitement of thisnew and promising product. In April 1976, the first produc-tion camera came off the line. passed all tests, and wasshipped to the first customer. The TK-76 program seemsdestined, for some years to come, to be used as a standardof comparison for judging other difficult programschedules and successes.

Subsystem inspection required special test equipment.

Many of the major components of the TK-76 were sub-systems that were purchased outside the company underclose purchasing specifications and then went throughrigorous incoming inspection and testing. Yoke assemblies,optics, and tubes are examples of such items. In somecases, special test equipment had to be developed to verifysubsystem performance and assure that the total systemwould meet its performance requirements.

For example, registration is a particularly difficult aspect ofcolor -camera performance. It was anticipated at the start ofthe program that it would be an even more severe require-ment for the TK-76 than for other cameras, because of thesmall scan size (0.26" X 0.35") of the 2/3" pickup tubes andtheir tiny associated deflection assemblies. RCA thereforehad to develop a geometry -testing apparatus for measuringyoke and tube geometry errors to a precision that had neverbeen attempted before. This equipment, including a com-puterized matching capability, became part of a ratherelaborate component acceptance process; RCA manufac-tured and supplied similar units to specific vendors.Computerized/automated testing of module boards hasbeen widely used throughout the program and is onereason we have met our production commitment of morethan 50 cameras per month.

Start of a new generationBy the end of 1977, RCA had delivered approximately 700 TK-76cameras.

Repeat orders by enthusiastic TK-76 customers, as well asnew sales, attest that all or most of the original goals for thiscamera have been met. Field experience has been very

13

gratifying, especially in the area of stability, operationalsimplicity, and reliability, which depended so crucially onmany new design concepts. The cameras are shipped to thecustomers fully operational and are often taken directlyfrom the packing crate and put into service. Many of theunits delivered at the very start of the program have, to thisdate, never required servicing or re -adjustment.

Picture quality from the TK-76 is subjectively equivalent to that ofthe top -line studio cameras, especially when the newer Saticontubes are used.

Broadcasters are finding that the camera performs quitewell for many types of field production-that is, pickup of ascheduled event, like a baseball game, opera, or com-mercial, that may be presented in real time or rebroadcastlater. This application is quite far afield from the originalintent of a limited -use news camera. However, camerasystems for production must include many specializedoperational features not required for news work (e.g.,chroma key special effects, color matching, largeviewfinder, and large lenses).

To fill this need, RCA has designed the TK-760 camera;deliveries will start in early 1978. Essentially a TK-76repackaged into a studio/production configuration, it has aremote operating panel, larger viewfinder, and the ability tooperate on long cables. The TK-76 contained within the TK-760 can be readily reconfigured back to a normal TK-76shoulder -carried version in a matter of a few minutes. It isbecoming clearer that the small -format (2/3" pickup tube)camera is having a revolutionary impact on the waybroadcasters are perceiving the trends in camera design.Many observers hold the view that introduction of the TK-76marked a dramatic turning point in camera design.

Electronic journalism will continue to demand improved equip-ment, particularly video tape recorders.

Small (28 to 30 pounds) U-matic-type helical -scanrecorders, most commonly used for news, pose a definiteperformance limitation. The majority of the tv news cameraspresently available produce pictures that are significantlybetter than those the portable recorders can produce. Themajor degradations that occur are in resolution, signal-to-noise ratio, and time -base stability. Unfortunately, theseimpairments become compounded as a result of multi -generation recording, a necessary step in the editingprocess.

The typical small U-matic video tape recorder is designed tooperate in a very restricted environmental window, whichprecludes operation, for example, on damp or very colddays. In addition to their limited performance and lack ofextended environmental ruggedness, the present recordersare too heavy and must be carried by a second person. Theumbilical relationship between the camera and recorderseriously limits the camerman's ability to react in fast-moving news situations. Wireless links between the cameraand a distant receiving point, through the use of either amicrowave or optical transmission system, have provenpractical only for specialized and limited applicationswhere the characteristics of the path are controllable.Ideally, the tape recorder and camera should be capable of

being readily carried by one person, which means amaxi mum weight of 20 pounds. The technology for mergingthe tape recorder and camera into the present weight limitfor the camera alone is not yet at hand. Accomplishing this,and at the same time satisfying all of the other performanceneeds. stands out as the challenge for the next generation ofelectronic -journalism cameras.

AcknowledgmentThe TK-76 camera was, above all, a team effort. The team ofBroadcast Systems engineers that worked on the projectand won the David Sarnoff Award for OutstandingTechiical Achievement included Lucas J. Bazin, John J.Clarke, Donald C. Herrmann, Cydney A. Johnson, AnthonyH. Lind, Mark R. Nelson, Dennis M. Schneider, Alexis G.Shukalski, Harry G. Wright, and the author.

Reprint RE -23-4-7Final manuscript received November 18. 1977

Sid Bendel' is the lead engineer for the advanced developmentgroup tnat works on new cameras and systems, covering opticalconfigurations and colorimetry plus pickup -tube performance,evaluation, and optimization. He has been a member of theBroadcast engineering staff since 1944, and has been instrumentalin the development of many monochrome and color-tv broadcastcameras.

Contact him at:New Products EngineeringBroadcast SystemsCamden, N J.Ext. PC -27C4

14

Video tape editing systems using microprocessors

Modern video tape editing systems demonstratethe degree of sophisticationthe broadcast industry has reachedduring the last five years.

K.J. Hamalainen

Early video tape editing was quite crude bytoday's standards-in the early sixties thetape was cut with a razor blade. But as therecorders improved, especially in the areaof servo stability, electronic editing equip-ment was developed. Basically, these earlyelectronic devices converted the physicaldistances between the erase and recordheads (Fig. I) to time differences andenergized the correct functions at the righttimes and in the correct sequence. Forexample, for the recording to coincide withthe erased portion of the tape, the videorecord head had to be turned on ap-proximately 0.7 s after the master erasehead was energized; only in this mannercould a disturbance -free edit be made.

Reprint RE -23-4-5Final manuscript received October 17, 1977.

ROTATING

VIDEO HEADS

MOTOR

MASTER ERASE AUDIO ERASE

AUDIO REC PB

The editing evolutionTiming and sequencing the recordingfunctions were only the first steps in theevolution of electronic editing.

The next requirement was to identify theedit point and to preview it before ex-ecuting the final edit. Initially, this wasdone by recording a bench mark (cue tone)on one of the two audio tracks on the videotape. By using this tone for initiating theedit, the edit could be simulated,previewed, and modified several timesbefore the actual recording was made. Thelatest improvement and basis for themodern editing systems was the introduc-tion of the time -code signal, a digital signalthat is usually recorded on the cue (secondaudio) track of the video tape. This signalidentifies every tv picture by hours,

TAPE MOVEMENT

Fig. 1Editing system must keep track of the video tape recorder'scharacteristics: the location of and separatior between the audioand video heads; separation between the erase and record heads;and system acceleration and deceleration.

minutes, seconds, and picture -framenumber (Fig. 2). By using the time -codesignal as reference, the edit points for bothaudio and video can be selected whileviewing the recorded material.

The first editing controllers were built withhardwired logic, but microprocessors soonbecame the key building block of modernvideo tape editing systems.

Since most of the functions required are, asalready indicated, direct control functionsor mathematical calculations, editing isideally suited for computer -type controloperations and decision making.

A good editing controller should be direct-ly interfaced with the individual editingmachine. It must "know" all the specialfeatures and characteristics of the machine

Fig. 2Time -code signal improved editing systems significantly when itwas added. It identifies each picture by hours, minutes, seconds,and frame number (1-30).

I5

to be able to perform the control functionscorrectly. At the same time, the controllermust provide the operator with a clearpicture of the status of the editing processwith easily understandable displays andhave a simple and functional control panel.

RCA's two microprocessor -based editing systemsDuring the last three years, RCA hasdeveloped two different microprocessor -controlled editing systems to operate withour TR-600 video tape recorder: the AE -600, a sophisticated time -code editingsystem using the Intel 8080 microprocessorwith 15 -kilobyte memory; and the SE -I"simple editing system" using the RCACDPI802 microprocessor with 2 -kilobytememory.

Jukka Hamalainen is a Unit Manager withthe Electronic Recording Engineering Dept.at Broadcast Systems. During the last twoyears, he has mostly been working with thedesign of microprocessor -controlled videotape recorder accessories. He has beenassociated with video tape recorders since1961, when he was a member of the TR-22development team. Between engineeringassignments, he has held sales and manage-ment positions in Europe. For two years, hewas manager of RCA's project managementteam for the state broadcast complex inVienna, Austria.

Contact him at:Electronic Recording EngineeringBroadcast SystemsCommercial Communications Systems Div.Camden, N.J.Ext. PC -4658

Editing video tape for broadcast-what's involvedEditing is the process of assembling a number of individually recordedsegments together in the proper sequence and time relationship to produce adesired total program effect. Regardless of the recording medium used(photographic film or magnetic tape), editing requires two functions:

1) decision making (where to cut one recorded segment off and beginanother, including the choice of the type of transition to use in the cut point);and2) the final editing (actually assembling the master recorded material into acontinuous program).

Decision making requires very versatile capabilities for viewing the recordedmaterial (searching the edit points, simulating the final transitions, etc.) Thefinal edits require sophisticated means of cutting and joining the actualrecordings. With magnetic -tape systems, we refer to all this as "electronicediting," since the entire process is accomplished electronically.

An editing device must recognize the technical features of the video taperecorder it works with (in this case, the TR-600 quadruplex machine). Theediting device must know the sequencing and timing of the VTR's record anderase functions for both the video and audio channels. (See figure below.) Itmust also synchronize both the record and playback machines so that a newsection can be added (recorded) precisely at the correct predeterminedlocation on the master tape. (See figure next page.)

All technical functions (searching, sequencing, simulating the editor, syn-chronizing the machines, and performing the final edits) must be done as fastas possible to reduce editing time, and personnel must be able to make theirediting decisions as easily as possible. The total emphasis must be given toartistic creativity. The complex technical system behind a simple, easilyunderstandable control panel should be inconspicuous and hidden.

The mechanics of videotape editing

TRANSVERSE VIDEO TRACKS-17 TV LINES/SCAN

,,/ 111'11

4 0

VIDEO ERASE HEAD

4 ROTATINGVIDEO HEADS

fLONGITUDINAL AGM TRACK

AUDIO REC/P13 HEAD

AUDIO ERASE HEAD

0.3"

TV PICTURE, 32 TRACKS

<TAPE MOTION I

The typical video tape recorder shown above has audio and video heads thatoperate independently. When a recording or "edit" is made, the following stepstake place:

First, the machine goes into the playback mode.Then the master erase head turns on.After the tape has traveled 10 inches (approximately 0.7 seconds later), theerased portion of the tape reaches the rotating video heads and these headsare switched from playback to record. The accuracy required for this switchis approximately 0.007 inch along the direction of tape travel, or ap-proximately one microsecond in rotational timing.

I t,

The audio erase head turns on approximately 1.5 inch or 0.1 second beforethe video record head starts so the audio and video record heads areenergized simultaneously.

All early editing devices could do these functions and simple calculations; themicroprocessor -based AE -600 editing controller does them with software andmemory rather than hardwired logic. The major advantage of the AE -600 is,however, its use as a controller-the operator uses it to select (mark) theportions of the tape that are going to be edited together. The AE -600 "marks"tape with the time code, an electronic signal that is recorded on the tape withthe picture (on a special audio track). This signal gives every picture on thetape an identification number-hours, minutes, seconds, and frame number(30 pictures per second).

When the edit points (start and end) are being selected during the previewsessions, their identification numbers are stored in the AE -600's memory.These edit points can be reviewed and modified several times during theediting process. In the final edit session, one machine (the record machine)records the finished version from several playback machines.

RECORD MACHINE

(TAPE)

PB MACHINE

(TAPE)

TIME CODE 'RACKCODE TIME CODE (PICTURE NUMBER) 01 22 23 15

HR MN SEC FRAME (PICTURE)

SECTION TO BE RECORDED ABOVE

A\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\1TINE

CODE "A"TIME C002 OS 1520.00ENDTIME CODE 02 16 20 29

SYNCHRONIZING AND RECORDING INTERVALI CYCLING OUT AND STOP

CLOCKING 5

In the typical edit shown above, the operator has already previewed the tapeand has decided to take roughly one minute of material from the playback tape(bottom) and put it at a specific location on the record tape (top). To performthe editing the operator puts the following identification times into the AE -600memory:

the location where the new material is to be added to the record tape(01:22:23:15);the beginning of the new material (02:15:20:00); andthe end of the new material (02:16:20:29).

The AE -600 controls both the record and playback machines. Upon an "edit"command, it has them go through the procedure below.

First, it cues both machines to a starting point approximately 5 secondsbefore the start of the edit.The AE -600 then starts both machines and synchronizes them so that whenthe record machine reaches picture 01:22:23:15, the playback machine isplaying back picture 02:15:20:00. At this moment the record machine startsrecording. (Actually, the AE -600 had given the record signal 0.7 secondearlier so that the video heads would start recording the new material on anerased portion of tape.)Recording stops when the end code (02:16:20:29) s reached.

The systems were designed with flexibility,time -saving, and simplified computer con-trol in mind

The primary objective in designing theseediting systems was to give the operatorflexibility for expanding artistic creativity.A second important objective was toenhance productivity by minimizing thetechnical noncreative delays associatedwith editing systems. The final objective forthe AE -600 was to develop a system so thatboth the basic video recorder and theediting sysem could be directly controlledby a central computer without complicatedmachine -oriented interface devices.

The AE-600-for sophisticated editingThe AE -600 editing controller, which isdiscussed here in more detail, is a goodexample of how these guidelines wereapplied.

Typical functions of a video tape editingdevice such as the AE -600 are:

store edit -in and edit -out points, usingthe time code.Provide simple ways to modify thesepoints (keyboard);search and cue the tape to a selected andstored point;preview or rehearse an edit;perform the final edit;animate (automatically record one pic-ture at a time); andsynchronize the machines (more thanone recorder can be handled by oneediting system).

In addition to performing these functions,the editing system must recognize and keeptrack of many special characteristics of thevideo tape recorder. These include:

location of and separation between theaudio and video heads (Fig. 1);separation between the erase and recordheads;

acceleration and deceleration character-istics of the tape -transport system;color framing, etc.

The AE -600 system allows editing machinesto interact with each other without an ex-pensive external computer.

All these features are available in mostediting systems. What makes the AE -600system unique is that each machine has itsown integrated editor, which can com-municate with the editors in other similarly

17

equipped TR-600 machines. This system(Fig. 3) has two distinct advantages. First,it increases versatility and shortens the timerequired for individual edits. Second, iteliminates the expense of the time-sharedcomputer sysem that is otherwise necessaryfor machine interaction.

The editing system can control one recordTR-600 and up to eight playback TR-600ssimultaneously. A thin cable of threetwisted -pair wires is looped through all themachines in the system for transferring therequired edit and transport controlfunctions. Any one of the TR-600s can bedesignated as the "edit" or the "record"machine by selecting the edit mode on thecontrol panel (Fig. 4). (Electronic in-terlocks insure that only one machine canbe the edit machine at any one time.)Playback machines are selected on the editmachine's control -panel keyboard.Machines connected in the system but notrequired by the edit VTR are available forother use.

Each machine has its own control panel, butthe system can be expanded into a centrallycontrolled editing system using a remote -control panel.

A standard remote -control panel includestwo playback -machine control panels anda master record panel with associatedkeyboards and displays. Communicationbetween each machine and its remote -control panel is through individual four -wire connections, but the main editingcommands between the machines are stilldistributed through the three twisted pairspreviously described. In other words, theplayback and source machines are stilldelegated to and controlled by the "record"machine.

The control panels have been designed tomaximize user effectiveness and minimizeoperator error. Several keys have upper-case functions very similar to those onscientific pocket calculators. Using thesemultiplexing techniques, the forty-sixswitches on the control panel provide sixty-four separate functions.

The microprocessor can directly interfacewith the machine functions as well as con-trol functions.

In the example shown in Fig. 5, themicroprocessor synchronizes the machineafter the "play" command and maintainsdirect control until lock -up by means of adata bus between the microprocessor andthe capstan servo. Within themicroprocessor, one register is preset with

MACHINE INTERCONNECT CABLE

Fig. 3Multi -machine operation is possible withcontrol up to eight playback machines.

Fig. 4Each machine has its own control panel and may act as the "edit" machine. Interlock systemassures that only one machine can be the edit machine at any one time.

the AE -600 system. One "edit" machine can

REFERENCEFRAMEPULSE

"MTTENABLEGATE

SYNC

PLAYIACXC011110110

VTRCONTROL

LOGIC

VTRTRUNNUS

CUE POINT

MR NMI SEC FRAMES

11111111"PERFECT MACHINE"

REGISTER

AE -600ELECTRONICS

PLAY SACKTIME CODE

MAGNITUDE

VTR ELECTRONICS

CAPSTAN

CUE CHANNEL

AMPLIFIER

COMPARATOR

DATABUS

CAPSTAN

SERVOSPEED *-

CONTROL

LOGIC

CAPSTANPOWER

AMP

TAO PULSES

Fig. 5Microprocessor synchronizes editing by working with AE -600 control electronics (top ofdiagram) and by direct interface with tape -recorder electronics (bottom). Microprocessorhas built-in no -inertia "perfect machine" register that starts instantaneously on edit cuepoint. VTR starts rolling at the same time, but has inertia and so takes time to catch up.Microprocessor compares "perfect" register with signal from VTR, then uses the resultingerror signal to drive the VTR until the two signals are locked together.

the cue point. This register is a model of a"perfect machine." It parks right on the cuepoint and has no inertia so that it can startinstantaneously. When a sync playbackcommand is received, this register is in-cremented with tv reference -frame pulses.At the same time, the VTR starts to rollfrom this one point. By comparing theVTR's playback time code with the in-crementing "perfect machine" register, anerror signal is produced and used to bringthe real machine in step with its perfectmodel. This error signal is in the form of adigital word that programs the capstanspeed in fractions of frame increments untilthe servo has properly locked.

This direct machine -microprocessor in-teraction is a good example of the advan-tages of the integrated approach. The built-in microprocessor system handles thespecial interface problems and individualmachine characteristics. The com-munications to the outside world can nowbe made on the "computer level" usingstandard codes.

To complement this advanced editingsystem, several other features weredesigned into the TR-600 video taperecorder. These features include:

A video character generator, which in-serts and displays the time code;

A time -code reader, also using a

microprocessor, which can read time attape speeds from I/ 10 to 100 timesnormal;

An edit servo, which automaticallymakes sure that the new signal will berecorded in correct phase on the tape;and

A CRT/ TTY interface module, whichprovides video display or printout of theediting log.

To achieve optimum video performance,the super high band deviation standard anda continuous pilot -tone error -correctiontechnique were added to the video and fmsubsystems.

The SE-1-for simp ified editingThe AE -600 system is one of the mostadvanced editing systems available and isan ideal base for interfacing with largecomputer -controlled editors or stationautomation. However, for everydayroutine operation, a much lesssophisticated system is appropriate. For

Fig. 6SE -1 system (control panel shown here) is simpler than AE -500 systemsignal on video tape, rather than a time -code signal.

this reason, RCA designed anothermicroprocessor -controlled editingaccessory, the SE -I "simple editor."

The SE -I system is built around the RCACDP1802 microprocessor.

This simplified system (Fig. 7) performsmost of the typical edit control functionslisted previously; however, its cueing ortiming reference is the control -track signalthat is always recorded on the video tape.Therefore, the time -code signal is notrequired.

In our estimation, even in this noticablyless sophisticated application, themicroprocessor had several advantagesover hardwired logic. These advantagesinclude less drafting work, fewer modules,a faster design cycle, and lower manufac-turing cost.

The best proof of successful designs andconcepts is customer acceptance. Ap-proximately 50% of the TR-600 machinessold now include the AE -600 editingaccessory, and a noticeable portion of theremaining 50% are delivered with the SE -Ieditor.

Future editing systemsThe AE -600 can be called a third -generationediting controller.

When the AE -600 is interfaced with acentral automation computer, it can per -

It uses control -track

form hundreds of successive edits. Whatdoes the fourth generation hold? The futureof editing systems is closely linked with thedevelopment of video tape recorders andother associated video and audio equip-ment. The computer technology that con-trols these devices is already available. Thepresent trend is in the Moviola (film -type)editing of video tape with slow, fast, andstop motion. Equipment that includesthese features is available, although it mustbe refined to achieve totally reliable high -quality results.

The video recorder is probably the weakestlink in the total editing chain.

Considerable time and effort is spent inadjusting and aligning machines foroptimum performance. One possibility forimproving video recorders, although it is along-range one, may be digital recording. Itcould provide us with multiple copies,tapes that have minimum degradation ofpicture quality, and machines that could beeasy to align and maintain.

Whatever the future holds, a micro-processor -controlled editor, like the AE -600, directly interfaced with the video taperecorder, will be a natural building blockfor the future editing systems.

ReferencesI. McKinley, R T.; "A new on-line quadruples VTR editing

system," SMPTE 1 18th Technical Conf., New York City (Oct17-22, 1976).

19

Automatic transmission systems for television

R.W. Zborowski

Automatic television transmission is

almost here. The Federal CommunicationsCommission recently authorizedautomatic operation of a.m. and f.m.broadcast transmitters and stated that itwill authorize similar operation for tvtransmitters in the immediate future. Con-temporary broadcast transmitters haveachieved such high levels of stability thatoperators, who normally monitor theirperformance at the required intervals,

R.W. "Sam" Zborowski joined BroadcastSystems in 1976 as a member of the Broad-cast Transmitter Design Group. He is activein Automatic Transmission System (ATS)design and has made contributions in theinterfacing of television transmitters forremote control. Sam in currently engaged inthe TTUE-44 uhf exciter program.Contact him atBroadcast Transmitter EngineeringBroadcast SystemsCommercial Communications Systems Div.Meadow Lands. Pa.Ext. 6241

Automatic systems may be able to monitor and control tvbroadcasting parameters better than manual operators can.The technology is available and waiting an FCC decision forthe go-ahead.

make significant operational adjustmentsinfrequently. Many transmitters arealready remotely located (on mountain-tops, for instance) and controlled fromproduction studios. These existing remote -control provisions make the transition toautomatic transmission systems (ATS)that much easier.

Aside from the obvious benefit of freeingthe operator to perform more productivetasks, a properly designed ATS canprovide more consistent broadcast quality.Automatic transmission is an extension ofthe trend that many broadcasters arefollowing-using a computer to selectprogram segments and commercial spots inthe correct sequence. In a few of thesesystems, the computer even keeps track ofthe billing for each commercial as it is

aired!

An examination of the FCC's notice ofproposed rulemaking, along with the ac-tual ruling on ATS for a.m. and f.m.broadcast services, leads to a more detailedprediction of the coming ATS regulationsfor tv broadcasting. Specifics are in the boxat the right; the major items expected are:

1) Automatic monitoring and control ofaural and visual rf output power, auralmodulation level, and several visualmodulation parameters.2) Automatic shutdown in the event ofequipment failure (failsafe).3) Absolute executive on/ off controlfrom an attended monitoring/ alarm site.

ATS system description

A microprocessor -based AutomaticTransmitter Controller (ATC) unit is theheart of the system.

Fig. 1 shows a system configured to satisfythe FCC's anticipated ATS regulations.

The ATC unit monitors transmitter opera-tion and makes supervisory decisions tomaintain optimum performance consistentwith the anticipated rules. It includes ananalog multiplexer that sequentiallydelivers a number of voltage samples to anA/ D converter for data acquistion, a bankof relay drivers for output commands, anda modem for communications with thestudio unit over a two-way audio link.Since the transmitter must have a blanking -level visual -power control loop, the ATCunit only monitors visual output power (viaa detector in the rf output line) to verifythat it remains within prescribed limits.

The remaining visual parameters are con-trolled by a separate, dedicated instrumentthat automatically corrects the video andshows how much correction is required.

A visual demodulator delivers a videosample representing the actual rf outputsignal to the automatic video corrector.The program video must include a verticalinterval reference (VIR) or vertical intervaltest (VIT) waveform. The automatic videocorrector takes either of these outputwaveforms, compares it against referencevalues, and pre -distorts the transmitter'svideo input to minimize the difference.

The video corrector scales all video samplelevels relative to blanking and zero -carrierlevels; blanking level is measured from the"back porch" of the video waveform. Acarrier blanking signal is sent to thedemodulator, which chops the rf inputsignal during zero -carrier -level measure-ment. The combination of automaticblanking -level control and the video cor-rector's ability to adjust sync amplitude (aswell as other parameters) relative to blank-ing and zero -carrier levels will hold peak ofsync power essentially constant.

Reprint RE -23-4-16Final manuscript received December 5. 1977.

20

OUTPUT VIDEO SAMPLETO

ANTENNA

PROGRAMVIDEOWITHVIR/VITREFERENCEWAVEFORM

PROGRAMAUDIO

CARRIERBLANKINGSIGNAL

AUTOMATICVIDEO

CORRECTOR

AURALMODULATION

CONTROL

LOOPERRORVOLTAGES

ATC

STUDIOUNIT

VISUALDEMODULATOR

RF SAMPLE

TV TRANSMITTERWITH

AUTOMATICBLANKING

LEVELCONTROL

VISUAL

ON/OFF FAILSAFE

2 -WAYAUDIOLINK

AURALPOWER

ATCTRANSMITTEF UNIT

OPTIONALCONTROLOUTPUTS

AURAL

OPTIONALVOLTAGESAMPLES

FILTERPLEXER

A

AURALMODULATION

MONITOR

TOWERLIGHT

MONITOR

Fig. 1Automatic transmitter system centers around the microprocessor -based automatic transmitter control (ATC) unit, whichmonitors output power levels and controls the transmitter for optimum performance. Separate instruments (presentlycommercially available) monitor and control program audio and visual parameters.

Fig. 2VIR waveform is frequently used in commercially availableautomatic video correctors. It facilitates reference measurementsof overall video gain, chroma gain and phase, burst amplitude andphase, sync amplitude, and black -level setup. For example,envelope at top of waveform provides chroma gain and phaseinformation.

Commercially available automatic videocorrectors can be improved.

Commonly available automatic video cor-rectors use the VIR waveform (Fig. 2),

Fig. 3VIT waveform may be a Defier approach. It facilitates referencemeasurements of all the VIR parameters (except black -level setup),then adds new ones: white -level reference, chrominance/luminance delay indication, low -frequency distortion indication,and differential phase anc gain indication. For example, staircasewaveform provides differential phase and gain information.

which includes no reference white level; themodulation depth must be extrapolatedfrom the 50 IRE level of the waveform.With this method, the transmitter coulddevelop a white -compression problem that

would continue only partially corrected asa result. A more suitable approach wouldbe to use a video corrector based on theVIT signal (Fig. 3), which does include areference white level.

21

In either case, the automatic video correc-tor provides analog voltages representingthe amount of correction required for eachof the important visual parameters listedearlier. The ATC samples each parameter(voltage) periodically to check if it is withinprescribed limits. Narrow limits on eachparameter would activate warning in-dicators at the ATC studio unit before thewider automatic shutdown limits arereached. This early "call for help" shouldhelp correct problems before shutdown inmany cases.

Audio measurement and control is donewith another piece of dedicated hardware.

The ATC similarly measures aural rf out-put power via a detector in the rf outputline and controls power by the usualremote -control interface (contact closuredriving a motorized pot or attenuator).Aural modulation levels must be

monitored continuously, indicating a re-quirement for a dedicated piece ofhardware for this function alone. As aresult of the adoption of ATS for f.m.broadcasting, dedicated modulationmonitoring/ control equipment is com-mercially available. In ATS for televisionbroadcasting, such a controller wouldmonitor and adjust aural modulation levelsconsistent with the ATS rules and providealarm signals to the ATC unit in case ofout -of -tolerance operation.

With parallel -transmitter operation, thevisual demodulator must include anautomatic gain control to permit continu-ing video correction during failure of onetransmitter. For a station havingmotorized output switching, the outputcombiner could be switched outautomatically for single -transmitter opera-tion. The ATC could additionally monitorthe outputs of the operational exciter toselect the alternate one in case of failure.During normal operation, the ATC unitcould adjust the relative phasing of the twotransmitters to minimize the power dis-sipated in the combiner loads.

With suitable current -monitoring devices,intruder detectors, and smoke/ fire detec-tors added, the ATC unit could evenmonitor the antenna tower lights andbuilding security.

The separate ATC studio unit would alsobe implemented in microprocessor form,primarily to simplify the communicationformat of the transmitter unit. Theoperator interface at the studio wouldconsist of a transmitter on/ off switch, a

Anticipated ATS provisionsThis listing, an educated guess at the forthcoming ATS regulations, was usedin preliminary system design.

1)Aural and visual rf output power must be monitored and controlled.

The transmitter must shut down automatically if output exceeds 110% oflicensed power for three consecutive minutes.

An alarm must sound if output power falls below 80% of licensed power forthree consecutive minutes.

An additional alarm must sound after carrier loss for over three minutes.

2) Aural modulation level must be monitored and controlled.

Modulation depth must be reduced if 10 bursts (a burst is understood to be asequence of over -modulation instances within a 5 -ms interval) of over 100%

aural modulation occur within any one -minute period.

The transmitter must shut down automatically if overmodulation is notcorrected within three minutes.

An audible alarm must be sounded following three minutes of no modula-tion.

All SCA (Subsidiary communications authorization) subcarrier generatorsmust have a deviation -limiting device.

3) Visual modulation levels must be monitored and controlled.

The following parameters must be monitored:

Color -burst amplitude

Color -burst phase relative to active video

Black -level set-up

Sync amplitude to pix ratio

Chrominance-to-luminance ratio

Depth of modulation

If any of the above parameters are out of tolerance for a specified time limit(tolerance and time not yet defined) an alarm shall sound.

A depth of modulation exceeding limits for three minutes must produceautomatic shutdown.

An alarm must sound following three minutes of no video modulation.

Note: The above video levels are measured relative to blanking and zero -carrier levels.

bank of visible status indicators, and anaudible alarm.

Failsafe andreliability features

The system must be failsafe againstcommunication -circuit failure that couldcause loss of control, and against failure ofthe supervisory / monitoring equipment. Tosatisfy the first requirement, the studio unitwould continually send a command toenergize a relay connected to thetransmitter's interlock circuits. If a

communication -circuit failure or ATC unitfailure occurred, the relay would drop outand shut off the transmitter; a three -minutedelay would preclude shutdown duringmomentary communication interruptions.The monitoring equipment dependsprimarily on the functioning of the analogmultiplexer and A/ D converter. Conse-quently, the second failsafe requirementcan be addressed by providing tworeference voltages within the ATC unit.These voltages would be periodicallyscanned and compared against prescribedlimits to check the measurement accuracy.

22

4) The ATS monitoring/alarm location must include:

Transmitter on/off control.

Off -air program and SCA receivers and monitors.

Automatic tower -light alarm indicator (unless visual observation is used).

All of the required alarms defined above must be both audible and visible.Once an alarm is triggered, the audible indicator may be shut off by theattendant, provided the visible indicator remains active until the fault iscorrected.

1

Any optional status/alarm indicators must be easily distinguishable from theFCC required alarms.

The monitoring/alarm location(s) must be reasonably secure from un-authorized access.

5) Miscellaneous ATS Provisions:

Automatic switching to an alternate main or auxiliary transmitter ispermitted, provided that such transmitters are fully equipped for ATSoperation.

Modulation must be monitored continuously.

All other required sampling of transmitting system parameters must be atintervals not exceeding one minute.

The ATS system must include a provision to test all automatic control andalarm devices.

More than one monitor/alarm location is permitted, provided each is undercontrol of the licensee.

6) Failsafe requirements:

The transmitter must shut down automatically after:

Loss of ATS sampling for three minutes.

Failure of the communications circuit between the transmitter site and themonitor/alarm location causing loss of manual on/off control for overthree minutes.

Failure of any required alarm system for over three minutes.

Overall ATS reliability can be enhanced byproviding alternate program audio/ videofeeds and an alternate communication linkwith the ATC studio unit. Uncorrectableoutput audio or video response could beprogrammed to have the alternate feedselected as a last resort before shutdown.The three -minute failsafe delay suggests thepossibility of using an automatic dialingtelephone accessory so that ordinary (asopposed to dedicated) phone lines couldbecome the secondary ATC communica-tion path. Relay switching can be providedto switch all transmitter control lines to an

existing remote -control system. This canprovide an additional measure of redun-dancy by enabling ordinary remote opera-tion in case of ATS failure.

What has RCA done?RCA has participated in long-term tests ofthe video correction scheme describedearlier at KDKA, Pittsburgh, Pa. (TT-30FL transmitter) and at WNJT, Trenton,N.J. (TTU-60B transmitter) in their actualoperating environments. The results ofthese tests were reported in Refs. I and 2.

The aural modulation control andautomatic transmitter controller equip-ment is currently being supplied by severalvendors for a.m. and f.m. ATS in-stallations. Although a complete system fortv transmitters has not yet been assembledand tested, the previous work indicates noserious difficulty in combining the existingportions of the system when the actual ATSrules are defined. Future designs mayincorporate part or all of the ancillaryequipment within the transmitter itself.

Future developments

So far, we nave seen ATS implementationsusing equipment that is available today.There seem to be two different schools ofthought regarding future ATS directions.One group believes that the transmitterstability is such that normal operatingdrifts produce no perceptable picture im-pairments. Indeed, as technology evolves,the transmitter stability should improve, sothe control interface with the rest of theATS system need not be further com-plicated. The other school of thoughtbelieves tr.at more controls should beprovided to allow automatic correction ofsuch parameters as low -frequency lineari-ty, subcarrier linearity, subcarrierdifferential phase, envelope delay, etc.Each of these parameters must be ad-dressed in terms of system reliability trade-offs for improved performance.

In any case, nearly everyone seems to agreethat ATS will monitor more functions topredict long-term drift -type failures. It willmonitor overall dc-to-rf efficiency alongwith a rtultitude of visual parametersobtainable from VIT signals. With uhftransmitters, because of the severe non -linearity of the klystron transfercharacteristic, automatic linearity correc-tion is likely to be added to compensate forprimary power line fluctuations. Astransmitters become more modular withgreater standardization, ATS may be usedto identify failed modules and so speed uprepair by module replacement. No doubt,ATS will improve overall transmissionquality as these systems evolve.

References

I. filuyas, 1.M.; "TV transmitting systems for unattendedoperation." Broadcast News. Vol. 154 (Oct 1974) p. 40. Alsopresented as the March 1974 NAB Engineering Conf.

2. Gluyas. T.M.. -Simulation of unattended operation of uhf-tstransmitters," Broadcast News, Vol. 158 (Jun 1976) pp. 40-47.

23

The BTA-5SSRCA's all -solid-state 5 -kW am. broadcast transmitter

L.L. Oursleri D.A. Sauer

The history of amplitude -modulatedbroadcast transmitters with electronicamplification began almost immediatelyfollowing the invention of the triodevacuum tube around 1906. A.M. broadcaststations progressed from operating powersof a few watts to fifty thousand watts andhigher within a few years as high powertubes were developed. Higher transmitteroutput power could be achieved byparalleling several power tubes. One exam-ple of this was an early RCA one -kilowatttransmitter which utilized four tubes, eachrated at two hundred and fifty watts, in apush-pull parallel circuit. Today, singletubes are capable of producing severalmillion watts of radio -frequency output.

The design technology necessary forproducing an all -solid-state broadcasttransmitter has been available since theearly 60s, but the required higher -poweredtransistors became available only recently.Now, large amounts of rf power can beproduced by combining these solid-statedevices into transistor arrays.

Solid state vs.tube transmitters

Much can be said about the differencesbetween all -solid-state broadcasttransmitters and vacuum tube transmitters.In general terms, the solid-statetransmitters are smaller, more reliable, andless costly to operate and maintain. Inaddition, the first -generation "all -silicon"transmitters have achieved signal qualitystandards comparable to the state-of-the-art tube designs.

Possibly, the best way to illustrate thesedifferences is in a side -by -side look at anall -solid-state broadcast transmitter and avacuum -tube -powered transmitter. The

Reprint RE -23-4-6Final manuscript received June 15, 1977.

RCA's first all -solid-state transmitter will save broadcastershundreds of dollars a year in electric bills alone.

RCA BTA-5 L2 is RCA's latest 5 -kW tube -type a.m. broadcast transmitter, this unit iscompletely solid-state at power levels

below the intermediate power amplifierand thus uses only four tubes. The RCABTA-5SS is RCA's new 5 -kW all -solid-state broadcast transmitter. These twotransmitters are shown in Fig. 1 and

compared in Table I.

Note that the signal quality of this first -generation all -solid-state transmittermatches that of the tube design. Theperformance may be improved further assolid-state design techniques mature. Thisis especially important now that moreemphasis is being placed on hi -fidelity a.m.broadcasting as a prelude to stereophonicbroadcasts by a.m. stations.

Table IComparison of the 5 -kW all -solid-state transmitter, BTA-5SS, and its predecessor, the tube -type BTA-5L2.

SpecificationsBTA-5L2Ampliphase

BTA-5SSall -solid-state

Size: an all -solid-state traismitter approaches one-half the size of a tubetransmitter of the same power level of recent design. (See Fig. 1.)Height 77 in. 77 in.Width 70 in. 38 in.Depth 32 in. 36 in.Weight 2500 lbs. 1000 lbs.

Reliability: Tubes become gassy, suffer from decreasing filament emission withage, and have lower overall power conversion efficiency. Transistor arrays canprovide a planned margin of power output capability in the event that a fewtransistors become inoperative. If the tube transmitter has but one final rf tube, thetransmitter has no output when the final tube fails, but the solid state transmittercan maintain full or reduced output if some of the active output devices fail.

Efficiency:Nominal power out 5000 WApprox. power consumption@ 0% modulation 10,000 W@ 100% modulation 17,000 WPower factor 90 %

5000 W

6920 W10,500 W

95 %

Signal quality: The solid-statedesign.AF response

@ 30 to 15,000 HzAF distortion

@ 95 % mod; 30-10,000 HzNoise

below 100% mod

transmitter meets the quality criteria of the tube

+1.5 dB

2% max

- 60 dB

+1.5 dB

2% max

- 60 dB

24

The solid-state transmitters will cost less tooperate and maintain.

The economy of a solid-state transmitter isachieved through higher efficiency, longer -life active elements, and smaller spacerequirements. Based on such operatingeconomies, a customer should pay for theincreased cost of the BTA-5SS in aboutfour years.

The solid -state -transmitter design can easi-ly implement power reduction without thecomplexity of high -power contactors andpower -wasting dropping resistors; a

vernier power output control can providean infinite number of reduced power out-put levels. By means of this feature, non-standard operating power levels can beeasily achieved, and instant on -the -airswitching with no program interruption ispossible.

Fig. 1The RCA BTA-5SS 5 -kW all -solid-state a.m.broadcast transmitter (top) is about half thesize of its predecessor, the tube -type BTA-5L2 5 -kW Ampliphase transmitter.

RCA in the broadcast radio transmittermarketplace

RCA has elected to concentrate its radio broadcast equipment sales effort instations from 5 kW to the maximum size permitted by the FCC, 50 kW.

We have long been a leading supplier of broadcast transmitters, though weformerly included 250-W and 1-,<W transmitters in our line. The history of ourinnovation ,ncludes the introduction of high-level modulation in 1 -kWtransmitters in 1933, in 5/10 -kW transmitters in 1936, and in 50 -kW transmittersin 1939. Simultaneous with the introduction of high-level modulation, RCAintroduced air cooling in place of water cooling In 1955, RCA introducedAmpliphase as a modulation system at the 50 -kW level. Ampliphase providesthe advantages of high-level modulation without the disadvantages of the highpeak voltages and large modulation components formerly required.

The broadcast transmitter business is highly competitive, including suchcompetitors as Harris, Collins, CSI, and Sparta. While some of thesecompanies have introduced solid-state 1 -kW transmitters, none so far hastackled the 5 -kW and larger field.

Within the US, approximately 14005/10 kW transmitters are in operation; manyare quite old and a large replacement market exists. The BTA-5/10 SS with itsease of installation, built-in high reliability, and minimum power consumptionshould have a large appeal to this market. Export sales tend more to the newstation business; and while the trend to build transmitters of this size locallyinhibits our sales opportunity, still there are areas where the inherentadvantages o' this transmitter make it an appealing product for new stationsales.

-D.S. Newborg

Dave Newborg s Manager of Radio Station Equipment Product Management. Contacthim at: Broadcast Systems; Camden, N.J.; Ext. PC -2146.

The transm:tter's push-pull bridge,saturated amplifier, commonly called aclass -D amplifier, has excellent efficiencyand provides a sink to either the powersupply or groend for any induced ortransient energy. Lightning and static dis-charge can be ,roblems to any transmitter,but careful design can help to minimizepossible damage and, or annoyingprogram interruptions. Effective lightningand static protection can take the form ofshunt static drain chokes, spark gaps, andreflectometer circuits. In a solid-statetransmitter, the type of rf amplifier usedcan also enhance the protection of theoverall system.

The basic class -D rf bridge amplifier isshown in Fig. 2. Each arm of the bridgecontains a transistor; the rf inputtransformer has a single primary windingand four independent secondary windings.

The polarity of each secondary winding issuch that transistors Q I and Q2 arc on andcompletely saturated for a given half cycleof rf, while transistors Q3 and Q4 areturned completely off during the same half

.v«

Fig. 2This class -D rf bridge amplifier circuitprotects :he system from lightning andstatic discharge; it ranges in efficiency from90 to 95%.

25

Fig. 3BTA-5SS-5-kW all -solid-state a.m. broadcast transmitter.

30

AC

INPUT

POWER SUPPLYMONITORS

VSWR

TEMPERATURE

PA BALANCE

PA DRIVE

LowVoltageSupplies

HighVoltageSupply

The transmitter control module(top) and the fault controlmodule (bottom) provide com-plete control and protection forthe transmitter. The moduleshave remote control capability,and a remote/local switch is

provided for the safety ofoperating personnel. The maincontrols are: transmitter on,transmitter off, rf on, and rf off.Two pushbuttons provide a

digital power increase/decreasecontrol, which gives eight stepsof power increase to 10% abovenominal and eight steps ofpower decrease to 10% belownominal.

AUDIOINPUT

26

rrt

The rf generator modulecontains a high -stability fre-quency synthesizer that allowsthe output frequency to beprogrammed in 1 -kHz steps tosatisfy both domestic andforeign frequency assignments.The heart of the synthesizer is aprecise 5 -MHz temperature -controlled crystal oscillator. Therf generator also has provisionsfor using an external frequencyreference for synchronousstations and for frequencymodulating the carrier for a.m.-stereo applications.

The rf pre -driver module is abuffer power amplifier betweenthe rf generator and the rf drivertray and is comprised ofsaturated class -0 rf amplifiers.

The final rf power amplifier stageconsists of six class -D bridgepower amplifier trays whose out-puts are summed by means of acombining transformer. Thismethod of combining allows thetransmitter to maintain its fullpower output even in the eventof an ocassional loss of an rfoutput transistor. The combin-ing transformer also provides astatic drain to ground and atwenty -to -one step down of in-duced voltage such as lightning.

The rf driver tray is aplug-in array oftransistors acting as aclass -D bridge amplifier.

SubcarrierFilter

The modulation generatormodule produces a pulse trainoutput with frequency equal tothe subcarrier frequency andpulse width variations propor-tional to the modulating audiosignal amplitude and frequency.In the absence of an audio inputsignal, the unmodulated dutycycle of the entire modulationsection of the transmittergenerates the required voltageacross the final rf stage toproduce the unmodulatedcarrier power output.

The modulation driver and the modulator poweramplifier consist of transistor arrays which turn onand off at the subcarrier frequency and in accor-dance with the modulated duty cycle. The subcarrierfilter removes the subcarrier frequency and applies avoltage, which varies at an audio rate, to the final rfamplifier.

OutputCombiningandImpedanceMatching

rf cycle. When the rf input reverses polarityduring the next half cycle, transistors Q3and Q4 are turned on and transistors Q Iand Q2 are turned off. The time required toturn one set of transistors on and the otherset off is extremely short-on the order of afew nanoseconds. During most of the rfcycle, the transistors are either turnedcompletely on in a saturated mode orcompletely cut off, and the only time asmall amount of power is lost in thetransistors is during the nanosecond transi-tion period and the saturation period.

The net result of the minimal power loss isexcellent rf power amplifier efficiency, inthe range of 90 to 95%. If transistors whichproduced zero transition time and zerosaturation voltage were available, the cir-cuit conversion efficiency would be 100%,but the above -mentioned circuit losses areever present in the real world and limit theobtainable efficiency. The RC network inthe base circuit of each of the transistorsproduces a small amount of bias to help

minimize the storage -time effect of thetransistors. The voltage produced acrossthe series -tuned -load network is a squarewave and the current through the loadresistor, RL, is sinusoidal due to the filter-ing effect of the series network. The loadresistor, therefore, has a sinewave ofvoltage across it and a sinewave of currentthrough it.

The BTA-5SS-RCA's firstThe RCA BTA-5SS is (Fig. 3) the firstmodel in the RCA line of all -solid-statea.m. broadcast transmitters. Thistransmitter is completely self-contained,produces 5 kW of carrier power, andfeatures low power consumption, highperformance, and high reliability.

The rf section of the BTA-5SS consists of thefollowing plug-in modules: rf generator, rfpre -driver, rf driver, and rf power amplifier.

Each power amplifier tray acts as a con-stant voltage source to its rf load, and all ofthe transistors on the tray share the outputcurrent demand. The trays are designed sothat at least 25% of the transistors wouldhave to fail before the tray could notmaintain its full current output. An in-operative transistor is automaticallyremoved from the circuit, and the remain-ing transistors continue to provide the fulloutput current.

The final link between the combiningtransformer and the output to the antennais the impedance matching and harmonicfilter rf network. A reflectometer is alsoincluded to monitor forward power andvswr and to provide protection by instantlyquenching the rf output whentransmission -line disturbances occur.

The modulator section consists of themodulation generator, modulation driver,modulator power amplifier, and subcarrierfilter.

The modulation section, including the sub -carrier filter, functions as a variable powersupply, and the transmitter's unmodulatedcarrier level can be adjusted by changingthe duty cycle of the modulator pulse train.After the required carrier level has been set,audio can be applied to the modulationgenerator to modulate the duty cycle at theaudio rate, resulting in a varying voltageacross the rf final.

In this highly refined pulsewidthmodulator, the subcarrier is derived direct-

ly from the frequency synthesizer in the rfgenerator module, and the resultingprecise control of subcarrier frequencyallows stable system performance.

In general terms, the modulation systemprovides low distortion, wide frequencyresponse, fast transient response, highmodulation levels, high efficiency, and aconvenient method of adjusting andregulating carrier output power.

This new transmitter reduces operating con-trols to a minimum, and has several self -monitoring and correcting features.

The basic transmitter operating controlsare on -and -off and power -level select. Aneight -step digital power control increasesor decreases the voltage on thetransmitter's automatic power controlcomparator, which then adjusts theamplitude of the subcarrier triangle wave.The resultant change of the triangleamplitude changes the duty cycle of thepulsewidth modulator. As describedpreviously the transmitter's output poweris adjusted by this change in duty cycle. Aswitch gives the operator the option ofeither automatic or manual overload recy-cle control, and a digital counter sets thenumber of overload steps allowed in theautomatic mode before the transmitter isshut down. The high -voltage supply is

protected from overcurrent and under -voltage conditions, such as the loss of asingle phase, and either condition shuts thetransmitter down. The front panel in-dicators show the reason for shutdown.

The low -voltage supplies are undervoltageprotected and current -limited. A reflec-tometer circuit sends a fault pulse to thecontrol logic when a high vswr conditionexists, and the transmitter's rf output isinstantly cut off. The drive level to the rfpower amplifier trays is monitored, and ifinadequate drive is present, the transmitterprotects itself by turning off. The rf outputlevel of each of the rf power -amplifier traysis detected by the tray balance circuit, and ifthe trays are not properly balanced inoutput, the transmitter does not allowoperation until the tuning on the trays is setproperly or the defective tray is repaired.Under normal operation, the tray balancecircuit provides a convenient check on trayperformance.

The temperature of each of the rf power -amplifier trays and the modulator power -amplifier trays is monitored, and if a tray

27

Fig 4Module nest in the BTA-5SS is designed to aid service of low -power -level module boards viaa module extender.

Leonard Ouraler has been a member of theBroadcast Transmitter Design Engineeringgroup since 1973 and has been active in thebroadcast industry for 16 years. While atRCA, his major programs have includedadvanced development of amplitudemodulation systems, design update of 100 -kW a.m. transmitters, and the design ofmedium and high -power all -solid-state a.m.broadcast transmitters.Contact him at:Aural BroadcastingBroadcast Transmitter EngineeringMeadow Lands, Pa.Ext. 6253

Dave Sauer has worked in broadcasttransmitter engineering since he started hiscareer with RCA in 1960. He has worked on5-, 10-, and 50 -kW broadcast transmittersand the 100 -kW Ampliphase transmitter.Contact him at:Aural BroadcastingBroadcast Transmitter EngineeringMeadow Lands, Pa.Ext. 6253

develops a higher than normal operatingtemperature because a malfunction, theprotection control circuitry turns off thesystem. In the event of a blower failure, theair -flow detector automatically reduces thetransmitter output power and keeps thetransmitter on the air. A front panelindicator is turned on when the transmitteris in this mode of operation. Thetransmitter has four illuminated meters tomonitor the rf power -amplifier voltage, rfpower -amplifier current, cic outputpower vswr, and 20 circuit parameters on amultimeter.

An automatic power control maintains thetransmitter's carrier output at a level presetby the broadcaster. In addition, anautomatic modulation control circuit willkeep the modulation depth at a level presetby the broadcaster as the transmitter'spower output is varied to eliminate theneed to readjust the modulation level whena switch is made from high to low power orlow to high power. The automatic featuresof the BTA-5SS are designed to make thetask of utilizing the newly authorizedautomatic transmission system extremelysimple.

In the design of a broadcast transmitter, itis very important to provide enough servicefeatures for routine inspection, cleaning,and in the event of a failure, easy repair.The BTA-5SS uses extensive modular con-struction, and the low -voltage -levelmodules shown in Fig. 4 are designed sothat they may be operated on a moduleextender. The transmitter cabinet wasmade large enough to give easy access to allcomponents. A multimeter on the frontpanel gives operating parameters in 20different circuits throughout thetransmitter, and several illuminated statusindicators provide instant evaluation ofoperational status or fault conditions.

The futureThe age of all -solid-state medium and highpower a.m. broadcast transmitters is hereand offers the broadcaster high perfor-mance, economy, and reliability.

As solid-state technology advances, powersgreater than 5 kW will be possible. Wewill see a rapid increase in power outputdensity in terms of watts per cubic foot ofcabinet space. More self -monitoring andcorrecting features will be introduced asextensions of the present AutomaticTransmission Systems authorized by theFederal Communications Commission.

Circular polarization for better tv receptionM.S. Siukola

Circular polarization is well known to allbranches of communications exceptbroadcasting, where it has been used onlyby fm broadcasters. Even for fm, not all ofthe potential benefits are being realized.The main thrust there has been to improveautomobile reception by simply adding avertically polarized component (for whipantennas) to the horizontal componentserving home receivers.' Nevertheless, thisuse of circular polarization, as restrained asit may seem, has increased the fm audiencesubstantially. During the 60s, it lifted fmbroadcasting out of a severe economicslump. But alas, there has been no furtherfm exploitation of circular polarizationeven though meaningful solutions to mul-tipath problems are certainly possible.

The television broadcaster's interest incircular polarization dates back almost asfar as that of the fm broadcaster. But atthat time, because of the predominance oftv as an entertainment medium, thereappeared to be no major unreachableaudience, and therefore television ex-periments were not conducted.

The first real effort at conclusive circularpolarization tests for tv was by theAmerican Broadcasting Company in 1973,using experimental antenna facilities builtby RCA for WLS Channel 7 in Chicago.'ABC then contracted with Smith and-Powstenko, consulting engineers, to per-form the experiments during 1974.3

Polarization modesThe benefits of circular polarization may beseen by studying the characteristics ofpolarized waves and comparing theirbehavior during transmission and recep-tion.

The polarization of an electromagneticwave is generally defined by the direction ofits electric field. Thus, the electric field of ahorizontally polarized wave is alwayshorizontal (Fig. I). Such a horizontallypolarized wave may be propagated by ahorizontal dipole. Accompanying the elec-tric field of the wave is, of course, a linearlyproportional magnetic field that is

orthogonal to the electric field (andtherefore vertical).

Television broadcasters, mindful of what circular polariza-tion has done for fm, are taking a closer look for possibleholes in their coverage.

RECEIVINGANTENNA

P.e

Fig. 1

Horizontally polarized wave, until 1974, has been the onlytransmission mode permitted for television transmission.

. - o, 40.,e014(

CYC1/4.4ktro

044.041.0

lil

Fig. 2Circularly polarized wave results from equalhorizontally- and vertically -polarized fields in phase quadrature.

A horizontal receiving dipole, perpen-dicular to the direction of propagation, willdevelop a voltage along its length as a resultof the vertical magnetic field crossing thedipole and inducing an emf. A verticaldipole would see no source of potentialsince the electric field is perpendicular tothe dipole and the magnetic field would beparallel to it. Thus, no signal would bereceived.

A circularly polarized field may beproduced by generating two linear fields ofequal magnitude, one horizonta: and theother vertical. Propagated in phasequadrature, the two fields add vectoriallyto produce a right-hand circularlypolarized wave with equal magnitude fieldsat various slant angles, resembling at anyparticular instant, a left-hand threadedscrew (Fig. 2). It is obvious from the

29

PfOn

1 s'

Fig. 3Propagating circularly polarized fieldprovides constant voltage at all slant angles. The wavedoes not actually rotate; instead, each field plane maintainsits special orientation and traverses along the axis ofpropagation.

TRANSMITTING 4ANTENNA

Fig. 4More solid coverage for tvis one result of circular polarizationsince reception is independent ofantenna orientation.

Fig. 5At reflection, the sense ofrotation of a circularlypolarized wave tends toreverse.

Fig. 6A signal with oppositesense of rotation isrejected by the circularlypolarized receiving antenna.

VERT.COMP.

0

NOR. COMP.

NOR. &VERT.COMP.

IvKE

to

o

R. H

A/4

RECEIVINGANTENNAS

CP CP

V-1

REFLECTOR

-180'

V 0

R. H.

X/4

TRANSMITTINGANTENNA

-Kr

horizontal and vertical components thatwhen the circularly polarized wavepropagates, it does not rotate; instead, eachfield plane maintains its special orientationand traverses along the axis of propaga-tion, in the manner of the lefthand screwbeing driven by a hammer (Fig. 3).

The wave illustrated is defined by IEEEStandards as a right-hand circularlypolarized signal since, if the passing wave isobserved in the direction of propagation atany particular point along the propagationpath, the field will appear to rotateclockwise.

Improved audiencepenetrationReception is improved with simple "rabbit -ears receiving antennas, but CP antennasprovide optimum reception.

Of course, the addition to the station'ssignal of an equal amount of energy in theform of a vertically polarized component,as permitted by FCC, should significantlyimprove coverage. Close -in, this doubledeffective radiated power (ERP) betterpenetrates difficult urban locations, such asinside large buildings, where tv signals arereceived only by whip or rabbit -ear anten-nas (Fig. 4). For fringe areas, again twicethe power provides higher signal-to-noiseratio for a given distance if circularlypolarized (CP) receiving antennas are usedto take advantage of all the power. Un-doubtedly, many fringe -area viewers and,especially, CATV operators will use CPantennas.

Better picture qualityA circularly polarized signal with CPreceiving antennas improves the receivedpicture in several other ways. For example,ghosting in tv reception caused by mul-tipath problems is reduced. With horizon-tally polarized signals, about the only wayto reduce ghosting is to use highly directivereceiving antennas, and orient them forminimum reflection and maximum signal.When a circularly polarized signal is

reflected, its sense of rotation tends toreverse. A right-hand polarized signal, forinstance, tends to become a left-handpolarized signal and vice versa. Thisphenomenon can be visualized by separate-ly considering the parallel and perpen-dicular components of the signal at thepoint of reflection (Fig. 5). Theperpendicular component of the fieldmaintains its direction, but the parallelcomponent reverses to meet the boundaryconditions at the surface.

30

Fig. 7An end -fire basket helix was used totransmit circularly polarized signal for thedemonstratior at NAB in 1976.

The other factor is that, with CP, thereceiving antenna has to be a right-handpolarized to receive a right-hand polarizedsignal (Fig. 6). A left-hand polarized receiv-ing antenna will not extract energy from aright-hand polarized field. These twocharacteristics provide ghost -reductioncapability for the complete system.

Another improvement in picture qualitystemming from circularly polarizedtransmission is a reduction in adjacent -channel interference. At present, the maintool for reducing this type interference is areceiving antenna with a high front -to -backratio. With circular polarization, a

directive righthand circularly polarizedreceiving antenna theoretically can beeither right-hand or left-hand polarized asseen from the back, and thus can bedesigned to reject an unwanted signal.

Circular polarization with some or all ofthese features-easier receiving antennaorientation, stronger signals in buildings,less noisy pictures in fringe areas, perhapsthree to five miles more coverage, ghostreduction and/or adjacent channel in-terference reduction-may in any partic-ular case improve service to the viewingpublic. It will, however, cost the broad-caster more for new transmitting equip-ment, and the viewer, who may also need aCP antenna to achieve the most from thecircularly polarized system, will also haveto pay for improved reception.

CP demonstration at NABAn on -air, 10-mW-ERP transmitter andcomplete receiving system dramatically

horizontally Solarized circularly polarized

Fig 8These comparison photos show that circular polarization reduces the ghosting in receivedpictures of the test pattern. Horizontally polarized transmission and reception at left;circularly polarized transmission and reception at right.

demonstrated the ghost -reductioncapabilities of circular polarization at the1976 National Association of Broadcastersconvention in Las Vegas. The output of aChannel -56 exciter, modulated by a testpattern and live studio signals, was fed intoeither of two transmitting antennas: abutterfly antenna to produce the horizon-tally polarized signals; or an end -fire helix"basket" antenna to generate the circularlypolarized signals (Fig. 7). The ghost signalwas caused by a 3x5 -ft metal reflectorplaced about ten feet from the antennas. Abuilt-in delay line added "distance" to thereflected signal path to provide betterseparation between ghost and main signal.Horizontally and circularly polarizedcorner -reflector antennas received thesignals some 25 ft away and displayed themon a color monitor. The dipole of thecircularly polarized antenna was slantedand positioned in the reflector for nearlyperfect circular polarization. Fig. 8 showsdifferences between the horizontal mode ofoperation and the circularly polarizedmode. The display demonstrated theprinciple; no attempt was made to quantizethe possible improvement.

The WLS experimentThese experiments were quite extensive anddesigned to provide adequate data to sup-port the FCC's rule -Making proceedings.

The transmitting antennas for the on -goingWLS tests consist of butterfly for horizon-tal polarization and a modified Model BFBfm panel for circular polarization. Theseantennas were designed, built, and tested atthe RCA Antenna Engineering Center inGibbsboro, N.J.' During the tests, the

signals were transmitted alternately in thehorizontally polarized mode and in thecircularly polarized mode, whilemeasurements were made or picturesobserved and rated.

Although many aspects remain to be

proven by practical experience, theprincipal conclusions are as follows:4

I) The coverage of the circularlypolarized signal, based on the horizontalcomponent only, remains substantiallythe same as for the conventionalhorizontally polarized transmissionmode. This is construed to mean thatdepolarization of the signal is minor and,therefore, any potential for additionalco -channel and adjacent -channel inter-ference does not exist. No change in theallocation table is indicated.

2) Pictures transmitted in the circularlypolarized mode and received with cir-cularly polarized receiving antennas suf-fer substantially less deterioration byghosts.

3) Adjustment of rabbit -ear antennas iseasier for the circularly polarized mode,but once the setting is optimized thepicture quality is about the same for eachmode of transmission.

4) Rabbit -ear antennas on existing in-stallations, randomly oriented, givemore ghost -free pictures when circularlypolarized transmission is received.

5) Circular polarization reception withrabbit -ear antennas appears to be lesssensitive to movement of people inproximity to the antenna.

31

hoixontally xuaii_ed

Fig. 9Circular polarization 'forgives" a receiving antenna misorientation of 60° in WLS field tests.Orientation' of 60° and 0° with horizontal polarization (left); orientation of 60° and 0° withcircular polarization (right). (The 0° -orientation is shown for reference.)

circtlarly polarized

Fig. 10Television receiver displays at two random locations (#4 and #1) during WLS testsdemonstrate the superiority of circularly polarized system. Reception of horizontallyPolarized transmission at #4 and #1 (left); reception of circularly polarized transmission at#4 and #1 (right).

horizontally polarized circularly polarized

horizontally poler.zedA

drooled/ polarized

hc.r.zontally polarized circularly polarized

Fig. 9 compares the reception of horizon-tally and circularly polarized signals with areceiving antenna misoriented by 60°. Fig.10 shows ghost reduction in two randomlocations when a switch was made to thecircularly polarized transmission modeand circularly polarized receiving antenna.Further data can be found in reports to theFCC by the American BroadcastingCompany.4

As stated, WLS continues to operatesuccessfully in the circularly polarizedtransmission mode. It should be noted thatcircular polarization is fully compatiblewith the present mode of tv operation, andCP signals can provide some improvementin reception with conventional receivingantennas. For best results, however, cir-cularly polarized receiving antennas arerequired.

Implementation of circularlypolarized systemsPlanning of a circularly polarized transmis-sion facility requires, of course, that varioustypes of receiving installations be served.

Presently, the best tv receiving installationsemploy horizontally polarized antennas.However, vertically polarized whip andrabbit -ear antennas are numerous.Therefore, the planned CP transmittingantennas must provide proper signals forexisting as well as future circularlypolarized receiving facilities. Thus, thehorizontal and vertical radiation patterns,gains, video responses and other para-meters must be designed for all types ofreceiving systems and to reduce ghostingand interference.

If the transmitting plant location is new,the design must provide proper horizontalfield intensities for present receiving anten-nas as well as a good circularly polarizedsignal for the new receiving installations. Inupgrading existing transmitting facilities,no compromises should be made thatwould deteriorate service to existing receiv-ing installations.

Since the FCC allows as much effectiveradiated power to be transmitted in thevertically polarized mode as in the horizon-tally polarized mode, either the antennaaperture (gain) or transmitter power couldbe doubled, or some lesser increase of bothcould be effected. In practice, however, anyincrease in aperture should be carefullyscrutinized, since a narrower verticalbeamwidth may deprive some nearby

viewers of a signal they are accustomed toreceiving. Doubling transmitter power maybe the only way to provide better service tothe public.

The uhf station operators have still furthertradeoffs to consider, since only a few of thestations are transmitting at the maximumallowable effective radiated power of 5M W. Therefore, if an increase in power iscontemplated, a judgment must be madewhether better service will be provided byincreasing the existing horizontallypolarized ERP or by adding a verticallypolarized component to produce the cir-cularly polarized transmitted signal. Thedecision will likely be guided by manyfactors, including local conditions and typeof service planned.

Available hardwareSince the early seventies, major manufac-turers have been planning for high powertransmitter components and circularlypolarized transmitting antennas, theprincipal items required for a circularlypolarized transmission. The necessarypower levels are easily obtained by parallel-ing transmitters. The major challenge intransmitter design was the development ofeconomical and efficient combining andfiltering networks needed for vhf channelsat power levels of up to 100 kW.

Circularly polarized service for fm beganwith the use of two separate antennas, onefor the horizontally polarized signal andanother for the vertically polarized signal.Except for the initial cost, the systemperformed well, since two essentially in-dependent services were provided. Today,of course, fm antennas employ specialelements developed to radiate circularlypolarized signals.

However, the early fm -type antenna willnot suffice for tv, since good axial ratio, acrucial characteristic in ghost and in-terference reduction, is difficult to obtainwith separate horizontally and verticallypolarized antennas. Therefore, a majoreffort of antenna manufacturers has beenthe development of circularly polarizedtransmitting antennas.

Following the experimental antennas forWLS by RCA, and for uhf station KLOCby Jampro, several new products have beenannounced. Jampro offers its helical anten-na design for vhf and uhf channels. TheHarris Corporation makes a panel -typeantenna for vhf channels.

The RCA circularly polarized antenna linenow comprises four standard antennas.

The Tetracoil antenna, TCLI6, forChannels 7-13, is a single -channel, top -mounted omnidirectional antenna (Fig. 11)consisting of three layers of pole -mountedhelical radiators. Each layer has four in-terlaced helices radiating in the secondmode; that is, the field rotates 720° aroundthe periphery. The feed system divides theinput power between the three layers. Italso determines their phase relationshipsand alters their relative phases verses fre-quency; the latter in such a way that aproper "null -filled" radiation pattern is

obtained and the video transfer responseis optimum for the service area. The endloadings terminate each of the helicesradiating the remaining power and main-tain low vswr along the helices. The resul-tant, in the nature of a traveling wave,assures proper radiation patterns and vswracross a channel. Transmitter power ofabout 50 kW is required to provide theallowable 316 kW ERP in both horizontal-ly and vertically polarized modes, or 632kW circularly polarized power.

The Fan-Vee, Type TFV (Fig. 12) providesan omnidirectional CP service for Channel2-6 top -mount applications.6 The antennauses four horizontally polarized "batwing"radiators and four vertically polarized "V -dipoles," each fed in turnstile fashion andoperating in mode one (360° aroundperiphery). A "branch -type" feed systemprovides the excellent stability required forlow channel operation. The seven -layer"dual -layer" model in combination with a50 kW transmitter is generally a cost-effective choice for an ERP of 100 kW. Afour -layer "dual -layer" model could bebuilt as a replacement for present six -layerhorizontally polarized superturnstile typeantennas, since windloadings are aboutequal. Some 60 kW of antenna input poweris required for an ERP of 100 kW.

The RCA Quatrefoil, Type TBK (Fig. 13) isa basket -type panel antenna, for Channel2-6 and can be side mounted for directionalpatterns or stacked for omnidirectionalradiation. For omnidirectional service,three panels around the tower are required.

The Channel 7-13 panel antenna, type TBJ,is designed for small-crossection towers toprovide directional or stacking capability.Panel antennas provide flexibility in radia-tion characteristics and gain, but generallywith slightly higher noncircularity,windload, and cost.

Fig. 11RCA top -mounted Tetracoll Ch. 7-13 anten-na, Model TCL16, for omnidirectionalservice.

Mgr

Fig. 12Upper three layers of seven -layer Fan Veeantenna, Model TFV7A4, for tv station WTTVundergoing tests at RCA's AntennaEngineenng Center in Gibbsboro, NJ.

Fig. 13RCA Quatrefoil panel antenna, Model TBK,mounts on tower to provide directionalservice or stacking capabil ty foromnidirectional service.

33

Some implications for equipment manufacturersFor more than two decades, horizontal polarization has been the onlypermissible mode of transmission for tv broadcasting. During thistime, many new tv antennas have been developed-for more gain, tohandle higher power, to sculpt special patterns-but always to radiatehorizontally polarized waves.

TV broadcasters are studying with great interest the brilliant perfor-mance records of television station WLS which has been operating onCP by special authorization since 1974. Some are applying forpermission to use circular polarization-use of which FCC hasapproved-to obtain the benefits in their particular areas.

Typically, conversion from horizontally polarized to circularlypolarized transmission will require doubling the station's transmitterpower. Some stations already have sufficient transmitter powercapability for CP by way of available unused transmitter power ratingor existing standby transmitter facilities that could be put intooperation. Other stations would need to replace or add to theirexisting transmitter facilities in order to achieve double transmitterpower. This requirement, plus the purchase of a CP antenna andpossibly the installation of new transmission line, represents asubstantial investment in new equipment.

On the other hand, many existing television transmitters and antennasare reaching normal retirement age and are already under considera-tion for replacement. This fact, plus the prospect of improved viewerreception, has caused the television broadcaster to take a careful lookat the potential benefits of circular polarization.

In addition to the standard productsmentioned, some manufacturers offercustom built transmitting antennas, andnew ones will undoubtedly emerge.

No circularly polarized receiving antennasare being marketed as yet, although somemanufacturers beside RCA are reported tobe developing antennas. Of course, RCAmanufactured the test receiving antennasfor its part in the WLS experiment.

Circularly polarized tv-present and futureThe experiments, though limited, indicatedthat the performance benefits closelyfollowed theoretical predictions. Therewere no major surprises, either positive ornegative; and FCC acted favorably in April1977 to the petition by American Broad-casting Co.

Broadcasters in general have shown a highdegree of interest despite the anticipatedsystem costs. WLS, Chicago, continues itsfourth year of eminently successful broad-casting in the circularly polarized mode.But this is only the beginning. WPBT,Channel 2, Miami, has just installed anRCA circularly polarized panel antennawhich is meeting expectations andproviding good reception on vertical and"rabbit -ears" antennas. KBYU is on -the -air with a Harris panel antenna. XETV,Channel 6, Tijuana, Mexico, is scheduledthis winter to install a Fan-Vee antenna. AFan-Vee antenna will be shipped toWTTV, Channel 4, Indianapolis, thiswinter. Further, RCA has under construc-tion another Fan-Vee antennas for WRAL,Channel 5, Raleigh, N.C.

While circular polarization may not be ofinterest in all locations, one or more of its

advantages should improve service enoughto warrant the cost expenditure.

A better penetration of signal may be veryimportant in some urban areas, or circularpolarization may offer a welcome relief inghost -ridden places. For the broadcasterinterested in covering every mile of fringearea, the circularly polarized mode oftransmission offers the most promisingopportunty available today. It is thereforeexpected that a great number of broad-casters will adopt circular polarization fortv broadcasting, and promote the system tothe public for mutual benefit.

ReferencesI. Siukola, M.S.; "Dual polarization fm broadcasting with a

single antenna," RCA Broadcast News, Vol. No. 134 (Jun1967).

2. IEEE Trans. on Broadcasting, Vol. BC -2I, No. I (Mar 1975).

3. Three papers in NAB Proc., 29th Annual Broadcast EngineersConf., Las Vegas, Nevada (Apr 6-9, 1975)

4. Four ABC Reports to FCC on Circular Polarization.

5. Paper presented at Broadcast Consultants Seminar,Washington, D.C. (Sep 1977) and to be published in the IEEEBroadcast Transactions this spring.

6. Paper to be published in the IEEE Broadcast Transactions thisspring.

Reprint RE -23-4-3Final manuscript received November 17. 1977.

Matti Slukola has more than 30 years ofbroadcast antenna design experience, 25 ofthem with RCA. As Unit Manager of Ad-vanced Development for RCA's BroadcastAntenna Engineering Center, Dr. Siukolahas been a primary contributor to thedevelopment, over the past five years, ofcircularly polarized television broadcastantennas.

Contact him at:Antenna EngineeringBroadcast SystemsGibbsboro, N.J.Ext. PC 4921

34

Long -line

communicationin Alaska

then and now

Climate, terrain, and population density,the factors that once made Alaska a

communication engineer's nightmare,are now making Alaska the forerunner in

effective satellite communications.

J.L. Rivard

Alaska's communication environmentThere seems to be a tradition that every article written aboutAlaska must overflow with superlatives. Among the onesusually mentioned are vast distances, a harsh climate, and asparse population. Undoubtedly, some of the statementsare truthful. It is true, for example, that the state's currentpopulation is around 400,000, which corresponds to apopulation density of 2/3 person per square mile. Peoplewho enjoy the wide open spaces may prefer this density, butit presents major problems to the communication engineer.

Typical articles on Alaska also include a frighteningreference to the climate. Permafrost does occur in ap-proximately 80% of the state, in some areas to a depth of2000 feet, and it presents a special challenge to theconstruction engineer. It can lift a telephone pole out of theground in a surprisingly short time, but an antenna properlyfrozen into the permafrost could remain fixed until Alaskaagain becomes tropical. Along the southern coastline of thestate, from Ketchikan to Attu, it rains-and rains-and -ains.At higher elevations it snows, and at intermediate levelsthere is almost continuous fog. The communicationengineer searching for a site to build a microwave repeatermust be especially cautious. Snowfall, which can exceed800 inches per year along the coastal range, could easilybury a repeater site, including the antenna. Foggy peaks areoften inaccessible, even by helicopter, for weeks at a time.

Every article on Alaska must also mention mosquitos. Thelegend that large Alaskan mosquitos are frequently refueledby mistake at remote Air Force Bases is not generally true-it has happened only twice.

The early daysIt once took a year for a message to travel both ways betweenAlaska and Washington, D.C.

Modern communication history in Alaska began in 1741when Vitus Bering, a Dane sailing under the Russian flag,discovered what is now southeast Alaska. Bering, himself,did not complete his voyage of discovery; his ship waswrecked as it neared home, and Bering and many of hiscrew are buried on an island in the Northern Pacific Oceannamed for the explorer. There is no doubt that little wouldhave been heard of the first Russian discoveries in Americafor some time, if ever, had it lot been for the beautiful fursbrought back from Bering Island. In 1790, AlexanderBaranof sa led from Siberia to firmly establish the Russianpresence in Alaska. Baranof presided over the colonies for30 years, a longer term than that of any of his successorsduring the 126 -year period of Russian rule.

Communications in Alaska had not changed much by 1867,when Russia sold its interest in Alaska to the U.S.A. for $7.2million. The U.S. Army formally took possession at Sitka onMay 30, 1867, and regular ccmmunication was established

35

between the new base at Sitka and headquarters in SanFrancisco. Ships could then sail that distance in about 15days. To send a message from interior Alaska toWashington, D.C. and back generally required a year's time.

The real beginnings ofAlaskan communicationsIf there is a single factor linked to the establishment of com-munications in Alaska, it is the discovery of gold in the KlondikeRegion in 1895.

The boundary between Alaska and Canada had never beendefined and both countries now realized its importance. Toprotect American interests, and until the matter was settled,the United States Army again occupied the Territory. TheArmy soon learned that it had a far bigger problem with goldhunters in a practically lawless Alaska than it had in aboundary dispute with Canada. Within the next few years,prospectors were finding gold in many sections of theTerritory; the Army followed them with a string of fortslocated primarily along the Yukon River. The Army was stillthe only form of government in the Territory and urgentlyneeded a long -lines communication network inter-connecting its new permanent bases with each other andwith higher headquarters.

The early telegraph system was for the military first, civilianssecond.

On May 26, 1900, Congress passed an Act that establishedthe Washington -Alaska Military Cable and TelegraphSystem, or WAMCATS. This Act expressly stated that thesystem would provide commercial service whenever thisfunction would not adversely affect military traffic. Con-gress provided $450,000 for telegraph and cable linesconnecting Department of Alaska Headquarters, at Fort St.Michael near the mouth of the Yukon, with other Army postsin the Territory. Alaska soon had its first long -line carrier,but did not have a circuit to the continental United States.

Reprint RE -23-4-8Final manuscript received December 8. 1977

John Rivard has been in-volved with com-munications at a numberof locations since hejoined RCA in 1958. Hestarted at the CocoaBeach Missile Test Pro-ject, then was com-munication manager atGrand Bahama I. andsystems engineer at theStadan Project at GilmoreCreek. Alaska: He joinedAlascom in 1972 as a

senior engineer in thePlant Extension Group.Contact him at:Plant Extension GroupRCA AlascomAnchorage, AlaskaExt. 7561

Elihu Root, Secretary of War during the McKinleyadministration, was determined not to make our com-munication dependent on another nation. (An interestingcontrast to the recent decision by President Carter totransport natural gas from Alaska fields to the 48 states via aCanadian pipeline.) Additional funds were made availablein 1903 and 1904 for submarine cables to connect Seattlewith Juneau and to extend service to Valdez. Commercialbusiness over these facilities was permitted insofar asdeemed equitable and in the public interest by the Secretaryof War.

The first wireless system had unusual maintenance requirements.

The year 1907 saw the beginning of the wireless system thateventually replaced the land lines in Alaska. The probabledriving factor behind this decision was the high cost ofmaintaining the land lines. During 1904, there were over twohundred outages on the circuit, divided almost equallyamong blizzards, sleet storms, high winds, forest fires, andvandalism.

These early wireless stations were of the spark type. The 60 -cycle, 500 -volt output of the alternator was stepped up in aheavy transformer to 20,000 volts. This high voltagecharged a Leyden -jar condenser, which discharged acrossa spark -gap formed by two metal rods. The radio energythus produced coursed through a massive helical coil whichhad the terminals connected to the aerial and a groundsystem The sending key was a heavy metal bar, a foot and ahalf long, to which was attached a hard rubber handle. Afterevery six or eight messages, the operator was forced to shutdown the transmitter and file the contacts, which becamepitted by the arc that formed each time the key was opened.

Cables became so deteriorated that no one could understand whythey continued to operate.

During the 17 years between the completion of the originalsubmarine cable installation in 1906 and the close of 1923,little changed with the WAMCATS cables. A few routealterations and short cable additions were made. However,the cables had deteriorated so during those years that,despite the efforts of maintenance personnel, interruptionsto cable service totaled 743 cable -days for the fiscal year1924 and 1139 cable -days during the next fiscal year. TheSeattle-Sitka cable, when laid in 1904, had an insulationresistance of four megohms. By 1924 it had dropped to 2000ohms including the copper resistance to the faults. Cabletheory does not readily explain why this cable had con-tinued to operate satisfactorily.

In this period, the United States Army Signal Corps was alsothe commercial communication system in Alaska.Seemingly the very life of the entire Territory dependedupon it-the economic development, the business. theadministration of government, and the dissemination ofnews. Published figures reveal that from its inception, thesystem handled an ever-increasing amount of business, sothat the value of traffic handled in 1924 amounted to over$400,000. This message traffic was about equally dividedbetween commercial traffic and that of the various govern-ment agencies operating in Alaska. So, one can readily see

if,

Just what does Alascom do?RCA Alaska Communications, Inc. is the certificatedlong -lines communication carrier for message service inAlaska. This item of news is still, unfortunately, notwidely known. The fact that some parts of this countryare not now, and have never been, served by AT&T, or"Ma Bell," can still win bets in many a corner establish-ment. Throughout the entire length and breadth of thestate of Alaska, all long -line (as opposed to local)message telephone service is provided by RCA Alascom.

Alascom, as the long -lines carrier in Alaska, is responsi-ble for the design and development of a network that willallow a small number of persons, living in each of severalhundred communities widely scattered about the state,to enjoy communication service similar to that which isavailable in the remainder of the country. The networkdesign should result in an overall system that can beinstalled and operated at the lowest user cost. Thisinteresting and exciting engineering challenge is un-derway today.

why Congress, in 1923, appropriated $1,500,000 fora newcable to connect Alaska with Seattle and replace the worn-out cables.

With funds now available, a study determined the routing ofthe terminus of the Alaska Railroad,

became the northern terminal instead of Valdez. The mainrelay point was established at Ketchikan, with one cablerunning south to Seattle and two branches running north toserve southeastern and central Alaska. The Seattle-Ketchikan-Seward cable was completed on October 10,1924. The cable era ended on December 1, 1931, when useof the Seattle-Ketchikan cable was discontinued, and alltraffic between the United States and Alaska and withinAlaska was handled by radio. The Seattle-Ketchikan andKetchikan-Seward cables, while not in use, were still inexcellent operating condition and tested daily. Even thesystem's name changed. The designation "Washington -Alaska Military Cable and Telegraph System" was changedby an Act of Congress, approved May 15, 1936, to "AlaskaCommunications System," or ACS.

The Second World WarWorld War 11 affected Alaska and its communications profoundly.

Alaska became a cornerstone of American defense and abridge to Russia during World War II. Alaska was the site ofthe only armed invasion by foreign troops on the NorthAmerican continent during the war. Communications wereso poor during the early days of the war that the Japanesewere able to bomb Dutch Harbor on the island of Unalaskawith unalerted American fighter aircraft only forty milesaway on Umnak Island. It was the Japanese attack on DutchHarbor and their occupation of the Aleutian Islands of Attuand Kiska that convinced the War Department that Alaska'scommunications had to be upgraded immediately.

On Decemoer 9, 1941, certain radiotelephone circuitsbetween Alaska and Seattle were withdrawn from com-mercial use and assigned to military traffic only. This wasthe first clear distinction to be drawn between military andpublic use of ACS.

As the enemy's grasp upon Alaska was pried loose and thepressure of defense operations lessened, ACS facilitieswere again made availaole for nonmilitary use. The first bigbreak came in July 1944, with the reopening of therehabilitated Seattle-Ketchikan and Juneau-KetchikanACS links to public and commercial messages.

The cold -war periodWhite Alice, DEW, and BMEWS were all part of the cold -warcommunications buildup.

As a result of the tremendoLs wartime expansion, civilianAlaskans had the use of a much improved communicationsystem linking them with the southern 48 states after WorldWar II. But then, the outbreak of the Korean War in 1950caused another general build-up for the military in Alaska,and so placed a severe strain on Alaska's communications.ACS embarked on a 6 -year, $10 -million constructionprogram of expansion and modernization. ACS built a newsubmarine cable system from Skagway to Ketchikan, whereit met a new AT&T cable coming from Port Angeles,Washington. The AT&T cable was initially equipped with 36channels, but was expanded to 48 in 1974. These two cablesare in use :oday.

System (WACS) was aproduct of the Cold Wa. The Air Force asked AT&T to studythe possibility of creating a fixed multichannel corn-municatiors system. From this grew the technicue firstreferred to as "Forward Propagation by TroposphericScatter." A network of tropo stations, coupled withmicrowave in the high -density areas, was submitted as asolution to the communication problems in Alaska. A seriesof Aircraft Control and Warning (AC&W) sites had beenbuilt and the communication system known as White Alicesoon followed. The WACS began service in 1956, makinghigh -capacity, high-qi..ality, and very -high -reliability com-munication available to the various users, including thegeneral public.

The Distant Early Warning (DEW) Line under the AlaskanAir Command was extended out the Aleutian chain toNikolski in 1959 and on to Shemya two years later. TheBallistic Missile Early Warning System (BMEWS), es-tablished by the Air Force at Clear, Alaska in 1961, requiredmore sophisticated and highly reliable communications.

Alascom enters the sceneSince 1904, when the telegraph line was completed, the militaryhad tried to withdraw from its involvement in Alaska's commercialcommunications.

But it was not until 1967 that Congress passed Public Law90-135, the Alaska Communications Disposal Act. The firstsystem to go on the block was the Alaska CommunicationsSystem. RCA Global Communications, Inc. submitted thewinning bid of nearly $28.5 million, and on Jan 10, 1971 the

37

1)10MIDIISLANDS,

/ST. LAWRENCEISLAND

NUNIVAKISLAND

PRIBILOF LSLANIN

raC2i

'

46C

T4S

BRISTOL BAY

0111:

BEAUFORT SEA

F00.114

(ALF OF ALASKA

1.\-( l X.\C

XIr..2E

LEGEND

Villages

- Major Cable,Microwave, and VHF Links

Fig. 1

In the beginning. This is the communication system that existed in 1971, when RCA took over the commercial outlets of themilitary system. Most, but not all, of the villages were connected to the system by high -frequency radio links.

ACS was transferred from the Air Force to the newly formedAlascom.

What RCA had purchased was the commercial outlets of themilitary system. Very little of the actual long -line systemwas sold. Alascom was established in Anchorage as theowner of four toll centers, a VHF marine radio network,open -wire and submarine -cable lines, and some isolatedVHF radio endlinks. The Air Force retained, and still owns,the major long -haul terrestrial transmission network withinthe state. Fig. 1 shows the network as it existed in January1971.

Alascom laced a multitude of problems. For one, it was a networkwithout a network.

Alascom was the certificated long -lines carrier for the state,but the ACS facilities that had been puchased were not, perse, a long -lines network. More than two-thirds of thecommunities without adequate communications wereremotely located and unable to even use WAGS circuits, ifany had been available. Key segments of the WACS wereoperating at capacity.

A preliminary evaluation of the communication needs of thestate indicated that: 1) the WACS was already operatingpast its design life; 2) the network could not provide therequired number of voice channels or accommodate wide -band circuits; 3) the network was grossly inefficient tooperate and maintain; 4) the network would have to beexcluded from long-range plans; and 5) a long-range planto prov de statewide communications would require acompletely new long -haul transmission system and anextensive capital outlay.

Alascom's initial action was to install temporary facilitieswhere the service demand was greatest and to begin theconstruction of permanent installations. The major tem-porary facilities that were added consisted of multiplexoverbuilds to relieve the bottlenecks in the WACS. Earlypermanent facilities included four new microwave systems:Lena Pcint to Sitka, Anchorage to Talkeetna, Fairbanks toTalkeetna, and Anchorage to Seward. A new tropo link wasadded between Barter Island and Frontier Camp in thePrudhoe Bay area. New toll -switching machines wereinstallec at Anchorage and Fairbanks and brought the firstdirect -d stance -dialing service to Alaska. Service was ex-

III1.1 \ \ `NI I

fall WIRSISIN

Peen 101

!HOMED/it 40Is1ANDS,/ worn

PPR

11

1.0104

Rk ..,r.

0100 kOSICSLY11

Kenn

/ST. 14114/1)NCEMAR 'Opts

BEAUFORT SEA

MI

ARC 11.1J414 It

111, kkti

.....x.s...

mavt ,o., ,....R trtMM

CM.. ,,,,,

kftnev. k

k AT * k UN." """"" /Ammo* ..(,,`"'"'",

t

k 1114.1st

owl SPASMS SIT Vol LAW k worm

100141. Sat

1011

NUNIVAKISLAND

sr worm

PRIBILO F ISLANDS

..,..r;:ISINIG 01 MOS

s0.01

[Art 04M1m40.111

BRISTOL BAYno. mor,

1,41.113111

1.1.0SIMMS t

k WRVS

MO,) "°""*" grft"" ......

ul.1., mo weowwwu

k ft

R r011.0.11s,

01111

k' MIS SINTAltes k

)taroatk ..1010

141.011111

4=10011Mill

GULF OF ALASKA

SISMIS

Ole7e,a

LEGEND

Toll CentersMajor Earth StationsSmall Earth Stations

kf.rage

qkr

POW .../.10

1 -- ALE( HA!. !SI ND,r==='"

e

.10110VDS.

Fig. 2The near future. Alascom's earth -station network as it will appear in 1980.

tended to additional villages in the Bush TelephoneProgram. The Class II -B Coastal Harbor Network wasupgraded. Also, in keeping with new FCC regulations, ClassIII -B marine vhf service was provided from 14 locations.This service uses international channel assignments and isavailable to vessels of all nations.

In its offer to purchase the Alaska Communications System, RCAhad proposed to proceed promptly with plans for a domesticsatellite system for Alaska.

In 1973 Alascom purchased the Bartlett earth station fromCOMSAT and constructed the Lena Point earth station.Together with RCA Globcom, the nation's first operationaldomestic satellite system was inaugurated on December 20,1973 using Telesat Canada's Anik satellite. Anik providedsatellite circuits to Alascom at about one-half the cost of theIntelsat system and paved the way for extended satellitecommunication service to Alaska.

In 1974 Alascom installed earth stations at Valdez, PrudhoeBay, Nome. and Bethel. A new class -5 switch was installedat Deadhorse for the development of the Prudhoe Bayoilfield, the first local service provided by Alascom.

In May 1975, the FCC ordered Alaska traffic to be switchedto Western Union's WESTAR Satellite. Alascom continuedto lease transponders on WESTAR until the FCC authorizedinterim transfer to the RCA Satcom System in May 1976.The FCC has not yet ruled on who will be permanentlyauthorized to provide satellite service to Alaska.

The first electronic toll -switching machine went into servicein 1975 at Lena Point. Later in 1975, the Angoon-Ketchikanmicrowave system was completed and direct cistancedialing was extended to the southeastern areas of tne statefrom the Lena Point switch. In February 1976, the newAnchorage Primary Switching Center was placed intoservice. Also in 1976, Alascom turned up earth stations atCordova, Yakutat, Gilmore Creek, and two pipeline pumpstations, and constructed five transportable earth stations,including one standby unit.

Alascom nowIn 1976 the Air Force agreed to lease the WACS to Alascom,ending a period of 76 years o4 service by the military to thestate. The lease requires Alascom to replace thetropospheric scatter portion of the WACS with a modern

39

gpaba

4.4

Wit :OM

L.

iM

system relying primarily on twenty-one new satellite earthstations within 27 months, commencing July 1, 1976.

A major gateway earth station, equipped with a pair of 51 -foot antennas, is under construction near Anchorage. Theearth station will be interconnected to the Anchorage tollcenter with a twelve -tube coaxial cable. Alascom will haveapproximately 45 gateway, mid -route, and transportableearth stations in service by 1979.

Alascom also has an agreement with the State of Alaska to install100 small earth stations in bush communities by the end of 1977.

Television -receive capabilities have been installed in 23 ofthe bush stations as a part of a state -funded experimentalprogram. The map in Fig. 2 shows the earth -station networkas it will appear in 1980. Present plans are to routeapproximately 70 percent of the interstate traffic via satelliteand 30 percent terrestrially. Traffic to rural Alaska will relyalmost exclusively on satellite, while all high -density in-trastate and interstate routes will be backed up by terrestrialalternatives to ensure uninterrupted communications.

Ways to improve the satellite communications system andreduce costs are constantly being sought. In addition to

.$

dramatic decreases in space -segment costs, recenthardware developments have lowered the price of ground -segment components, notably low -noise amplifiers. Unfor-tunately, these decreases are being offset by cost increasesfor other components, notably antennas. Alascom is active-ly investigating new developments such as demandassigned multiple access (DAMA) systems. An extensiveinvestigation has been conducted into single -channel -per -carrier systems and companders, and an investigation intofrequency -division -multiplexed companders is presently inprogress.

Tests are now being conducted on developmental hardware thatwill allow bush earth stations to be connected more easily with newlocal telephone exchanges.

The AlasKa Public Utilities Commission is currently holdingapplications from several telephone companies seekingpermission to install these local dial offices in most of thebush earth station locations. Formal hearings by theCommission are scheduled to start before the end of 1977.Installations will follow and Alascom is expected to be readywith a sufficient number of trunks. The intent is to remaincurrent in the state-of-the-art of satellite communicationsand also seek ways to improve service and reduce costs.

: 1

I

I

40

ConclusionIn the era of terrestrial communications, particularly land -line facilities, the density of Alaska's population was anextremely signi4icant economic factor; but today, withsatellite communications, the distance between populationcenters is of little consequence. Thus, the characteristics ofAlaska that in the past have worked against the develop-ment of communication systems are now working in favorof Alaska, making it the forerunner in the effective use ofsatellite communications.

In conclusion, the communications history of Alaska can besummarized in four chapters.

The first began with the gold rush at the turn of the century,which brought the Army and its need for communication.Commercial service was available to those persons whohappened to live near the military stations.

The second chapter is associated with World War II. Thetremendous military buildup throughout the state includedan accompanying communication network.

The cold war of the 50s brought the largest effort to date. Anextensive communication network criss-crossed the state,

but again the facilities were designed to satisfy militaryrequirements.

The fourth and final chapter is the story of RCA Alascom.Now, for the first time in Alaska's history, the communica-tion needs of the entire population are being considered.Everyone-both the military and the local communities-will be able to receive service, wherever they happen to belocated.

References1. Bailey, T.A.; The American Pageant (D.C. Heath and Co., Boston. 1961).2. Bancroft, H.H.; History of Alaska 1730-1885, (Hefner Publishing Co., Darien. Conn.1970).

3. Ford, C.; Where The Sea Breaks Its Back (Little, Brown and Co.. Boston -Toronto,1966).

4. Garfield. B.: The Thousand Mile War (Doubleday and Co., Inc.. 1969).5. Jenne, T.L.; Military Long -Line Communications in Alaska, IEEE 1976 NationalTelecommunications Conf.

6. Rivard, J.L.; The Domestic Satellite Program in Alaska, IEEE 1974 National Telecom-munications Conf.

7. The U.S. Army in Alaska, Headquarters, U.S. Army, Alaska, 1972.8. The World's Submarine Telephone Cable Systems, United States Department ofCommerce, 1975.

9. Warwick, J.C.; Providing Communications to Isolated Areas, IEEE EASCON, 1973.10. Wickersham, J.; Old Yukon, (West Publishing Company, St. Paul, Minn., 1938).11. Sellers. J.D.; "Alaska communications system,' RCA Engineer, Vol. 16 No. 4 (Dec

1970 -Jan 1971).

12. Beira, V. and McGill; E.F.; "Toll switching plan br Alaska," RCA Engineer, Vol. 20, No.4 (Jun -Jul 1974).

41

Statistical coding methodsspeed up image transmission

M.E. Logiadis

Quite often, cost/ performance con-siderations keep common -carriercommunication -channel bandwidths toonarrow for the message being transmitted.Thus, to successfully transmit a messagewith bandwidth requirements wider thanthose of the available communicationchannel, the information is "compressed."The techniques that produce this datacompression are frequently based on non -statistical properties of informationsources. Rather than exploit the statisticalproperties of the sources, these methodstake advantage of the receivers' ability totolerate some distortion and still correctlyinterpret the message. In this article,however, we will concentrate on data -compression techniques based on thestatistical nature of information.

Principles of encodingMost source symbols appear with unequalfrequency, and this is where the usefulnessof statistical coding lies.

Encoding capitalizes on the unequal fre-quencies of source symbols by assigningcodewords of the shortest possible lengthto symbols with the highest frequency. Thisis achieved by statistical encoding, a tech-nique that translates source symbols into anew set of codewords, according to thesymbols' probability distribution.

For example, Morse code uses a single dotfor e, the most commonly occurring letterin the English language. The letter q, on theother hand, occurs quite infrequently andso has the lengthy code of dash -dash -dot -dash.

Coding applicationsin image transmissionImages are inherently highly redundant.Eliminating the redundancy from thetransmission media produces a significantreduction in the amount of transmitted

Advanced encoding techniques allow documents to betransmitted over standard communication lines at ten toeleven times the uncoded rate.

Table laVariable -length relative address code (RAC)assigns the shortest lengths to the shortestdistances from the "standard elements."The code is a combination of unique codesand F(N) RLC codes of Table lb.

Distance Code

+0 0

+1 100

-1 101

N (N2-' I) III F(N)

+N (N>2) 1100 FIN -I)

-N (NZ2) 1101 F (N-1)

data. Consider, for example, a black -and -white document. We notice with tv-typescanning, the transitions from white toblack and vice versa are relatively few.Various techniques have been developed toremove redundancy. The following onesare the most important and have wideapplications in image transmission.

Run -length coding (RLC) takes advantageof the above -mentioned distribution oftransitions to remove redundancy.

As its name implies, this technique encodesblack/ white run -lengths betweentransitions. Codes are based on theprobability distribution of such runs forseveral documents. An optimum codeassigns codewords inversely proportionalin length to the probability of occurence ofrun lengths. However, implementing thiscode in image transmission requires a largememory, which makes the equipment quitebulky and costly. In practice, therefore, theaverage length of a code is usually longerthan in this optimum situation.

There are several RLC codes. An exampleof a variable -length RLC code, whichassigns the shortest code lengths to themost frequent runs, is shown in Table lb.Note that the runs 1-4 are coded with the

Table lbVariable -length code assigns the shortestcode lengths to the runs that occur mostfrequently. This code supplements somecodes in Table la; asterisks stand for all thebinary representations of each group ofruns. For example, F(20) = 101111.

N FIN)

I"4 0**5-.20 10****

21-84 110

85^340 I I 10********

341-s-1364 11110*** ***1365--5460 111110

+ displacement

RAC ----- -displacement30 - - run -length

z

a.a.

W 20

a.

0

r,

RLC

RAC

10 20

LENGTH TO BE CODED

Fig. 1Distance between white/black transitionshas this probability distribution for anEnglish -language document scanned at 200x 200 lines per inch. RLC is for run -lengthcoding, RAC is for relative address coding.

42

shortest code lengths, as they appear withthe highest frequencies.

The code in this table is based on theprobability distribution of black/ whiterun -lengths (Fig. 1) for the standardCCITT digital fax chart #1 (English text).This RLC code achieves a data -compression ratio of 9 at a resolution of100 X 100 lines per inch. At a data rate of4800 b/ s, the standard document will betransmitted in 25 seconds.

As stated earlier, the effectiveness of a run -length code depends on the probabilitydistribution of black -and -white run -lengths. Since no two documents are alike,the RLC code generally favors documentswith run -length probability distributionsclosely resembling that of the document forwhich the code was designed.

In reference encoding, a black/white transi-tion on the line being scanned is referencedto previous transitions on the preceding andpresent lines.

In the example shown in Fig. 2, thesereference entries are used as follows:

Left (L)-transition one position to theleft of the previous line transition.Right ( R)-transition one position tothe right of the previous line transition.Same(S)-transition one positionbelow the previous line transition.

The reference -encoding method also use!RLC to code transitions that cannot 13(coded by reference encoding. Take, forexample, the transition at position #20 inFig. 2. This transition cannot be coded as"S" because of an unused transition atposition 18 in the preceding line. To avoiderror, we have to use RLC to code the run -length between the transition at position 20and the previous transition at position 17.Therefore, RL = 3.

Transitions encoded by the referencetechnique are termed reference transitions;all others are termed as RLC transitions.Several reference transitions, such as "S -L- L - 5," can be grouped and encodedaccording to the probability distribution ofthese runs. The combinations appearingmost frequently are assigned short -lengthcodes, with the longer -length codesreserved for the least frequent com-binations. Generally, reference encodingoffers an average increase in data -compression ratio of approximately 25%

Position

Line (i)

Line(itl)

5 10 15 20 25

Reference L S L LCodewords

aR (RL.,3)

Fig. 2Reference encoding achieves shorter codes by referring to the transitions in the precedingscanned line. "L" indicates transition one position to the left of the transition on the lineabove; "R" indicates a transition one position to the right of the transition on the line above;and "S" indicates a transition directly below the one above. Run -length coding is used forsituations that do not fall into these categories, as with the transition at position 20 in thesecond line. If "S" were used, it would erroneously refer to a transition at position 18. An RL =3 code solves the problem.

PI

9

Relative address coding(RAC)

0

P2

Preceding line

Coding line

I square equals I pictureelement

Fig. 3Relative address coding (RAC) encodes transition at Q by determining its distance from the"standard element," which is either P1 or P2, according to the formulas given in Table II andtext. Method is more complex than run -length coding, b.it faster.

over the RLC technique for photograph -type images. However, for certain black -and -white documents, the increase ap-proaches 100% over the RLC technique.Reference encoding requires manyreference transitions and many high -orderrun lengths to be effective. The hardware ofreference encoding is therefore usuallymore complex than RLC and so is morecostly.

Relative -address coding (RAC) is similar toreference encoding, but achieves higherdata compression.

Relative -address coding (RAC) is used byKokusai Denshin Denwa Co., Ltd. ( KDD)of Japan in their "Quickfax" terminal. Fora brief description of relative -addresscoding, consider the two sample scanninglines of a black -and -white document in Fig.3. Suppose we want to code transitionelement Q. Two transition picture elementsare already encoded, P1 (last transitionelement on present line) and P2 (transitionpicture element on preceding line nearest toQ). If neither reference element exists, P1 istaken as the first picture element of thepresent line and P2 is taken as an

imaginary element next to the last pictureelement of the preceding line.

The coding for transition element Q isdetermined by its distance from the "stan-dard element" that is selected from the twotransition elements P1 and P2 as follows:

If the distance between PI and Q is morethan one picture element and less thanthe distance between P2 and Q. PI is

selected as the "standard element." Thedistance from P1 to Q is always ex-pressed in picture elements. In all othercases, P2 is selected as the "standardelement" and the distance from P2 to Q is

expressed in picture elements with a plus,a minus, or no sign. A plus sign indicatesP2 is above and to the left of Q. A minussign means P2 is above and to the right ofQ. No sign means that P2 is just above Q.

Table I gives the code; Table II shows howthe "standard element" is selected, alongwith the definition of signs for the distanceof Q from the "standard element."

Since the first line of a document cannot bereferenced to a preceding line, it is always

43

Table IIRelative address coding (RAC) requires determination of the "standard element" and sign to go with thedistance from the standard element. Distances are measured in picture elements.

RELATION BETWEEN Pi 0 AND P20 STANDARD ELEMENT

Pi Ci< P20 PI

PI CI> P20P2

PIO P20 I P2

Pi 0 P204 I Pi

STANDARDELEMENT

ADDRESS RELATIONS BETWEEN0 AND Pi /P2

SIGN(+)(-)

PI PI Q>0Pile on the lett of 0 onthe same line. No -sign

P2

P20;0P2 is lust upon or on theleft of 0 on the precedingline.

(4)

PT22is on the right of 0122

the preceding line.on(-)

Table IIICompression factor for RAC is superior to that for RLC by abouta factor of 2 for a standard English -language document, butcompression factor and transmission time vary significantlywith different document types.

RES. CODINGDOCUMENT

I 2 3 4 5 6 7 8

8x8200x200

L.PI.

RAC 28.0 48.9 18.2 7.1 I 6. I 30.0 7.3 27.0

RLC 12.9 14.9 7.6 4.3 7.2 9.6 4.3 8.6

4x4100x 100

L.P.I.

RAC 13.7 24.5 9.1 3.5 7.9 15.8 3.6 13.5

RLC 9.0 9.2 5.3 3.16 5.03 6.49 3.05 5.4

coded in RLC. One disadvantage of RACis that an error caused by the communica-tion media will propagate to all linesfollowing the line in which it occurred. In a4 -wire communication system, this can beavoided by retransmitting the line in error.The return path is used to notify thesending terminal that a line was received inerror and request its retransmission.

The protocol employed to control the linkbetween the terminals is the High-levelData Link Control (HDLC) adopted bythe International Standards Organization.In two -wire operation, a hybrid approachis being used to stop error propagation, i.e.,both RAC and RLC techniques are in-terleaved over K scanning lines as follows:

If K is chosen as I , then every line is RLCcoded. For K=4, the first line is RLCcoded, lines 2, 3, and 4 are RAC coded,the K + I = 5 line is RLC coded, etc.Thus an error in line 2 propagates to line4 and stops there. In a two -wire system,retransmission of a block in error ispossible, but time-consuming because ofthe turn -around delay encountered inreversing the direction of transmission.For this reason, the RAC technique ismostly used in four -wire systems withK=00. (Remember, the first line is alwaysRLC coded.)

Getting back to Fig. 1, we observe that thedistribution curves of RAC are more or lessexponential, with the probabilities cor-responding to displacements 0. -1, and +1equal to 0.5, 0.13, and 0.13, respectively.Hence, the displacement from the "stan-dard element" is not greater than onepicture element 76% of the time. Thisobservation suggests that the codesemployed for these most frequent dis-placements should have minimum lengths.The other, not -so -frequent displacements(24% of the time), are represented by codeswith longer lengths. This method of codingtherefore produces a significant data com-pression and reduces the transmission timefor certain types of documents.

Table Ill shows the transmission time andthe compression ratio for CCITT testcharts 1-8 using both the RAC and RLCtechniques. It is evident that the RACmethod of encoding is superior to the RLCmethod by a factor of approximately 2:1.However, as with reference encoding, theelectronics required to implement RACencoding are far more complex and costlythan for RLC. Another disadvantage isthat although the RAC technique is quiteeffective in four -wire systems, it is not soeffective in two -wire systems because of theproblems with reversing transmissiondirections if errors are transmitted.

CCITT TEST

DOCUMENT

TRANSMISSION TIME (SEC.)

RLC METHOD RAC METHOD

I. Letter 25 19

2. Circuit 23 1 1

3. Invoice 40 25

4. French 72 63

5 French 45 30

6. Graph 34 15

7. Japanese 73 62

8. Memo 45 I8

The data compression and transmissiontimes for RAC encoding shown in Table IIIwere confirmed by tests performed earlythis year by RCA Globcom and KDDbetweeen New York and Tokyo overtranspacific cable/ satellite facilities. As aresult of these tests, RCA Globcom isoffering international switchedstore/ forward and bureau -type facsimileservices in the near future. Present plansare to inaugurate Quickfax service toJapan (Fig. 4) in the first quarter of 1978and to other countries shortly thereafter.

Encoding techniquesof the futureConditional probability distribution is basedon the statistics of past symbols.

For example, in a black -and -white docu-ment we can predict if the next pictureelement will be black with a certainprobability if we know if its previouspicture element was black or white.

In statistics, this technique is called a"Markov process of then order." The orderis determined by the number, n, of theprevious symbols, whose statistics forecastthe next symbol. In RLC and RAC en-coding, the probability distribution of the

44

La.

Fig. 4"Quickies" system uses relative -address coding to improve transmission rates for facsimilecopies of documents. International service will begin in early 1978.

transitions is based on independentsamples. Using a Markov process, we takeinto account past symbols in determiningthe probability distribution that would bemore favorable to encoding. This couldproduce a code with codewords having anaverage length less than those currently inuse and so would improve transmissiontime per page substantially (perhaps 20-30%).

Adaptive encoding would automaticallyadapt its algorithm to many classes ofdocuments rather than favor certain types ofdocuments.

Obviously, such a method would require an"intelligent" terminal that could determinewhat code to employ for each class ofdocuments. This technique would haveapplications in point-to-point com-munications with fully compatibleterminals or in public data/ fax networkscapable of handling dissimilar terminals,provided the code employed is known tothe network before transmission. Amodified version of the HDLC protocol(CCITT recommendation T.30) could beused for signaling and control functionsbetween facsimile terminals.

Theoretically, the techniques outlinedearlier for black -and -white pictures couldbe applied to pictures with shades of gray. ifcombined with other techniques.

With gray -level encoding, the transmissiontime is a function not only of the resolution

rate, but also of the number of shadespresent. Since pictures with gray levels areloaded with redundant information, amore efficient approach would be todevelop an encoding technique that alsoaccounts for the psycho -physiology of thehuman eye. By exploiting the limitations ofthe eye with regard to fine detail, improveddata -compression ratios could result.

ConclusionComparing the encoding techniquesdescribed earlier, we conclude that theaverage data -compression ratios achievedfor the CCITT test documents at a resolu-

Reprint RE -23-4-4Final manuscript received September 20, 1977.

tion of 100 X 100 lines per inch areapproximately:

for run -length codingfor reference codingfor relative -address coding

6:1

10:1

11:1

The two last techniques produce almostidentical results because they are based onsimilar principles, differing only in thedetails of encoding.

All three techniques, as well as othervariations, are presently being used in high-speed facsimile terminals for transmittingdocuments, graphics, and newspaper/magazine pages to remote locations overregular telephone lines. These statisticalcoding developments have helped to openthe way to the electronic mail era.However, statistical coding applicationsare still in the formative stages of theirdevelopment, growing steadily with im-provements in computer software andhardware.

ReferencesI. Dood, P.D. and Wood, F.B.; "Image information, classifica-

tion, and coding," IEEE International Convention andExposition, Vol. I (1973).

2. Sakrison, DJ.; "The problem of applying information theoryto efficient image transmission," IEEE International Conven-tion and Exposition. Vol. I (1973).

3. Yamazaki. Y.; Wakahara, Y.; and Teramura. H.; "Digitalfacsimile equipment using a new redundancy reductiontechnique,'private communication.

4. Kim. H.K., private communication.5. Kazuhiko, N.; "Digital codings of multidimensional informa-

tion services and application in image coding: Electronics andCommunications in Japan. Vol. 55-A No. 11 (1972).

6. Arguello, R.J.; "Encoding. transmission, and decoding ofsampled images." Symposium on sampled images, Perkin-Elmer Corp. (1971).

7. Smith. S.A.; "Generalization of Hoffman coding for messageswith relative frequencies given by upper and lower bounds."IEEE Trans. on Information Theory. (Jan 1974).

Minas Logiads is Program Manager for thedesign and development of switched globalfacsimile and word processing services tobe inaugurated by RCA Globcom n the nearfuture.

Contact him at:EngineeringRCA GlobcomNew York, N.Y.Ext. 7841

45

Linear predictive codingto reduce speech bandwidthJ.R. Richards' W.F. Meeker

The speech waveform is highly redundant.Therefore, communication systemdesigners have expended a great deal ofeffort to reduce the bandwidth oftransmitted speech by removing some ofthis redundancy. In the past few years,interest has arisen in a technique foranalyzing and synthesizing speech bymeans of linear predictive coding.

Standard transmission bands are not wideenough for encrypted speech.

A transmission band extending from about300 Hz to 3000 Hz is generally adequate forspeech communication. However, presentmilitary systems require that the speechsignal be converted to digital form beforebeing encrypted. If straight -forward pulse -code modulation (PCM) is employed, asampling rate of 8 kHz and at least 6 bitsare required to represent the amplitude ofeach sample. Thus a data rate of 48 kb/ s isrequired. However, the practical datatransmission rate for normal telephonelines is 2.4 to 4.8 kb/ s. Hence, speechtransmission by PCM over telephone is notfeasible. Various modifications of deltamodulation allow the data rate to bereduced to about 20 kb/ s for good quality,and to about 10 kb /s for speech of sub-stantially reduced quality.

The channel vocoder is one solution.

The channel vocoder has been used ex-tensively to provide digital speechtransmission at data rates in the range of1.8 to 2.4 kb / s. This device has a bank of 16or 18 filters covering the speech range. Therectified output of these filters can belowpass filtered to provide a slowly varyingrepresentation of the speech spectrum. Thevocoder makes further use of the nature ofspeech by determining whether the speechat any given time is voiced or unvoiced(that is, whether it is produced by vocalcord vibration or by air turbulence); if thespeech is voiced, the vocoder alsodetermines the pitch (the fundamental

Reprint RE -23-4-9Final manuscript received November 11, 1977,

63 -10

The ATMAC microprocessor allowed us to apply linearpredictive coding theory to speech bandwidth compression.

frequency of vibration of the vocal cords).These parameters can be quantized,transmitted, and used to reconstitutespeech at the receiving end.

The channel vocoder performs satisfactori-ly for some voices but is objectionable forothers. Speech produced by a linear

0

-l0

2cr -30

-4 0

-50

predictive coding system is more naturalthan that produced by a channel vocoder.This improvement in quality of narrow -band digital speech, together with theadaptability of the linear predictive codingprocess to large-scale integrated -circuittechniques, is largely responsible for theinterest in such systems.

500 1000 1500 2000 2500

FREQUENCY, Hz

3000 3500 4000

Fig. 1Spectrum of a 20 -ms segment of the vowel i as in slid. The spectrum was obtained by a 256 -point Fourier transform with preemphasis for the high frequency portion.

0 20cc

-30a.

-40

-500 500

1 1

1000 1500 2000 2500

FREQUENCY, Hz

3000 3500 4000

Fig. 2Simplified spectrum envelope of the vowel i as in slid was produced by exciting a 10th -orderall -pole filter with repetitive impulses. The filter was determined by the linear predictivecoding process.

-40

-500 l030

rTl

MOO 2000 2500FREQUENCY, Hz

3500 4000

Fig. 3Comparison of the simplified spectral envelope (Fig. 2) and the original speech spectrum(Fig. 1) shows that the impulse response from the LPC-produced filter closely matches theenvelope of the original speech.

46

Linear predictive codingLinear predictive coding (LPC), as appliedto speech processing, derives from earlystatistical work on time -series analysis, inwhich the next term in a time series can bepredicted from linear combinations of pastterms. While the application of linearprediction methods to speech processing isrelatively recent, it has attracted the in-terest of many workers so that there arenow many papers and reports describingvarious aspects of the approach and themathematics involved." Consequently,we omit the mathematics of the process andconcentrate on the approach and a possibleimplementation.

The LPC system provides a simplifiedrepresentation of the short-term spectrumof speech.

In linear predictive coding, the spectrum isdescribed as the coefficients of a 10th -orderall -pole filter. Figs. I, 2, and 3 show howthis simplified spectrum approximates thespeech spectrum. Fig. I shows the spectrumof a 20 -ms segment of the vowel i in slid.The spectrum was obtained by means of a256 -point Fourier transform wit!'preemphasis for the high -frequency por-tion of the spectrum. Fig. 2 shows theimpulse response of a 10th -order all -polefilter determined by the linear predictivecoding process. If the filter were excited byrepetitive impulses, Fig. 2 would describethe envelope of the resulting spectrum. Fig.3 shows the correspondence between thetwo representations (Figs. I and 2). Thesimplified spectral envelope closely ap-proximates the envelope of the originalspeech.

The resonances of peaks in the spectralenvelope are called formants; their timevariation in amplitude and frequencycarries most of the intelligence in speech.Thus, the LPC process determines an all -pole filter whose impulse response ap-proximates the envelope of the short-termspeech spectrum. A 10th -order filter re-quires only 10 coefficients to specify theresponse, and these can be quantized for anefficient representation. Although digitalcomputers can do linear predictive coding,the process is so complex that until recentlydigital computers fast enough to imple-ment a system in real time were notavailable. When implemented in discrete orMSI components, such computers orprocessors are still too expensive to permitwidespread use. However, personnel ofRCA Advanced Technology Laboratories

have designed the ATM AC micro-processor,6 which permits implementationof a cost-effective, low -power, real-timeLPC system.

LPC speech bandwidthcompress on

A bandwidth -compression system employ-ing linear predictive coding is shown in Fig.4. At the transmitter, analog input speech(appropriately band limited) is convertedto digital samples. The digitized speech isfed to a pitch extractor, which determinesthree parameters: 1) the voiced/ unvoiceddecision; 2) pitch, if voiced; and 3) somemeasure of amplitude. The digitized speechis also fed to a scaler and a linear predictorto generate the coefficients of the filter tomodel the short-term speech spectrum. Thedigitized speech is processed in blocks of144 samples, each block representing 22.5ms of speech samples at 6.4Hz. Each blockof input speech thus produces a cor-responding output block of 12 parameters:pitch, voiced/ unvoiced amplitude, sync,and nine linear -predictive coefficients.These can be quantized so that they can bemultiplexed to a 2.4 kb/ s data stream. (SeeTable I.)

At the receiver, the parameters for eachblock are demultiplexed from the inputdata stream. The nine linear predictivecoefficients are fed to the linear predictor(all -pole filter) which establishes the shapeof the output spectrum. The voiced/un-

CLOCK

MODEM

H PITCH

EXTRACTOR

2400

BITS/S

Table IBit allocation for speech parameters ob-tained from 22.5 -ms segment of speech.

Parameter Bits Remark,

AmplitudePitchK1,1(2

K3,K4,KsK6,K7,K8K9

Sync

5

6

12

15

12

3

voiced/ unvoicedif voiced

PredictionParameters

54 Total bits/ 22.5 -ms segment

voiced decision then determines which oftwo excitation sources to use. The ex-citation function for voiced sounds con-sists of a stream of pulses from the pitchgenerator, with the repetition rate con-trolled by the pitch signal. For unvoicedsounds, the excitation function is a streamof pulses with random spacing. The outputfrom the linear predictor is then scaled torestore the original amplitude, converted toanalog form, and appropriately lowpassfiltered to produce the speech output.

We built a working model of the speechbandwidth compression system using LPCand modem techniques.

To demonstrate the quality of speech whenencoded to 2.4 kb/ s using linear predictive

PITCH

V/UV

AMPLITUDE

SCAL E

M

X

LINEARPREDICTOR

TRANSMITTER

PITCH

V/UV

PIT, 11

.1r1

9 COEFFICIENTS

UX

COEFFICIENTS

NOISECEN

INVERSELINEARPRE DIC TOR

AMPLITUDE

2400

HITS MIA,

SCALE

RECEIVER

D/A

SPE EC HOUT

Fig. 4Speech bandwidth compression system using linear predictive coding. The 2.4 k b/s outputcan be transmitted on normal telephone lines.

47

DMA

CONTROL

PROGRAMMEMORY

CPU

(ATMACMICROPROCESSOR(

I/0CONTROL

SPEECH

1-0

DATAMEMORY

MODEMTONES

SFU

INLPF A D DA

Ole

41-17-F74- D A DATA BUS A D LPF

SPEECH MODEM

OUT TONESIN

Fig. 5Laboratory model speech -bandwidth -compression system demonstrates that high -qualityspeech can be encoded to 2.4 kb/s using linear predictive coding. Note that a 16 -tonemodem was also incorporated, showing that it is feasible to process both LPC aid modemalgorithms with a single processor.

coding, ATL built a laboratory model (Fig.5) of a speech bandwidth compressionsystem incorporating an ATM AC micro-processor. A I6 -tone modem was alsoimplemented in the system to demonstratethat both the LPC and modem algorithmscould be performed with a single

Williard Meeker, a principal member of theengineering staff, has researched acousticaldevices and communications systems, andhas served as audio and acoustical consul-tant for many development and design con-tracts. Recently he has concentrated onspeaker authentication, word recognition,and bandwidth compression, includingdigital computer simulations of linearpredictive coding systems and digital con-ferencing schemes.

Contact him at:Signal Processing LaboratoryAdvanced Technology LaboratoriesCamden, N.J.Ext. PC -5186

48

microprocessor. The major componentsare described below.

ATMAC microprocessor consists of twodata chips and two control chips

In ATMAC, the basic chips are eight -bitmodular building blocks which allow

Jerry Richards is Unit Manager of theSpeech Processing Group in AdvancedTechnology Laboratories. Since 1959, hehas been involved with computer process-ing techniques for speech reccgnition, re-quiring the design of special computerprocessing systems and computermemories. His current responsibilities are inaudio communications, speech bandwidthcompression, and speech processing.Contact him at:Signal Processing LaboratoryAdvanced Technology LaboratoriesCamden, N.J.Ext. PC -4561

various microprocessor sizes, expandablein eight -bit increments.6 The processor ispartitioned into the data -execution unitand the instruction -and -operand -fetchunit. For a I6 -bit machine configuration, atotal of four chips are used: two data chipsin the data -execution unit and two controlchips in the instruction -and -operand -fetchunit.

Program memory is a ROM simulator fordesign flexibility.

In a final hardware design for a fixedapplication, the program would be storedin read-only memory. However, duringdevelopment it is necessary to be able toreadily modify program memory. AScientific Micro Systems, Inc. Model1000A ROM Simulator, used for theprogram memory, incorporates fifteen512X8 -bit modules. Since instructionwords are 24 bits long, this unit provides2560 words of alterable program memoryhaving an access time of 100 ns. The ROMsimulator is loaded by paper tape producedon an ATMAC cross -assembler, runningon a Univac Spectra 70/45 computer.Program memory contents can also bechanged by pushbuttons on the simulatorcontrol panel.

Data memory is more than adequate.

Data memory employs RCA M WS550 I Dmemory packages, 1024X1 random-accessstatic memories, whose C MOS / SOS con-struction is well suited for this application.These memories have an access time of lessthan 100 ns and operate at the same 10-Vsignal levels as the microprocessor, so thatno level shifting is needed. Their powerconsumption is low.

Data memory is implemented on twocircuit boards, which, when fully pop-ulated, provide 6k of data memory; opera-tion requires less than 3k of data memory.

The special function unit increases thecomputational speed of the microprocessor.

The ATMAC architecture provides forsimultaneous output of two 16 -bit

operands to a special function unit (SFU).In the narrowband speech demonstrationsystem, the special function unit consists ofa I 6X16 -bit two's complement multiplierwith a 32 -bit accumulate option. Operatingmodes are determined by instructions fromthe processor; 16 instructions are availableto the SFU, of which 8 are used in thepresent unit. Either the most significant

word or the least significant word may beplaced on the data bus, where it may bewritten into data memory or into aprocessor register as determined by in-structions. Multiplication (or a multiplyand accumulate operation) and subsequentwriting of the product to a processorgeneral register or to data memory areaccomplished in two instruction cycles.The multiply and accumulate operationitself is complete in 500 ns.

The speech and modem data input/outputdevices use the ATMAC interrupt facility.

Speech input and output are handled by theATMAC direct memory access (DMA)facility. Both input and output are ac-complished with a single DMA facility byinterleaving the speech input and outputdata in memory. One instruction cycle isrequired for each data word transferred,and one interrupt must be serviced for eachframe of speech data.

The I/O controller coordinates the transferof data between the microprocessor andexternal devices such as the A/ D and D/ Aconverters. It directs two-way communica-tion with the microprocessor via the I/Odata bus.

An interrupt request is raised by the I/Ocontroller when one of its devices requestsservice. While the microprocessor itself hasa single interrupt line, the I/O controllercan permit independent priority interruptson a priority basis determined by thesoftware.

Since the speech is processed in 22.5 -msblocks, full -duplex operation of both theLPC and modem can be achieved with the70-ns clock.

The ATMAC microprocessor executesshort instructions in four clock cycles andlong instructions in five cycles. ATMACwas expected to achieve a 70-ns clock cycleperiod, resulting in short and long instruc-tion times of 280 and 350 ns, respectively.Our initial samples of the ATM AC chipoperated with an 80-ns clock when notusing the LIFO stacks, and a 110-ns clockwhen running with the LIFO stack. Subse-quent examinations of the chip that con-tains the LIFO stock uncovered somelimitations in speed caused by design andlayout; these are being corrected.

Since revised chips were not yet available,the demonstration system used chips hav-ing a speed limitation. A 110-ns clock

Table IIProcessing time and memory needed for linear predictive coding and modem algorithms.

Algorithm

Processing time (ns) MemoryWith 110-ns clock cycle With 70-ns clock cycle RAM ROM

As presently With planned As presently With planned (16 -bit (24 -bitcoded improvements coded improvements words) words)

LPC analyzerLPC synthesizerModem transmitterModem receiver

11.73

4.49

9.70

8.53

9.91

4.49

3.984.47

7.46

2.85

6.18

5.43

6.302.85

2.532.84

1073

1021

6101212

1373

1441

458864

period was needed to demonstrate opera-tion of full -duplex linear -predictive codingand half -duplex modem processing. Theexecution times for both programs aregiven in Table II. The table includes theexecution times achieved on the presentsystem as well as the execution times thatwill be achieved with several planned im-porvements. The most significant improve-ment is in the modem, where a 50% speedincrease can be achieved by 1) using anFFT rather than a time series to synthesizethe modem tones and 2) by adding a FIFOregister to the modem interface to permitthe transfer of a block of data for eachinterrupt rather than a single word. Theprogram and data memory requirementsare also given in Table II.

Hardware projectionsfor the speech processor

The laboratory model of the speechbandwidth compression system is the firststep in the development of low-cost low -power narrowband systems. The ATMACmicroprocessor, which performed the ma-jority of the processing functions, con-sumes less than I watt. A six -chip set,including four 8X8 multiplier chips toimplement a I6X16 multiplier and two 16 -bit accumulators, can be built to operate inconjunction with the microprocessor andconsumes less than 600 mW. Hence, theentire processor subsystem can be

fabricated with ten CMOS/ SOS LSI chips,which would consume less than 2 W. Eightof the ten required chips have already beendeveloped by RCA. To build a completelow -power, half -duplex speech terminalcapable of both modem and linearpredictive speech processing, a total of 154standard off -the -shelf CMOS packages arerequired in addition to the ten LSI chips,with a power dissipation of 9 W. If 4kCMOS RAM and ROM memories with100-ns access time (presently under

development by RCA and other com-panies) replace the I k chip, and if the clockand interface chips are implemented withtwo LS1 chips, the entire system can bereduced to 48 chips housed in a 5X5X10-in.package, with a power dissipation of lessthan 5 W.

The progress of LSI, particularly inCMOS/ SOS, has made the developmentof a small cost-effective low -power LPCspeech processor possible, an accomplish-ment that five years ago was thought to beimpossible.

AcknowledgmentsThe authors gratefully acknowledge thesplendid efforts of Stan Ozga, whodeveloped the ATM AC microprocessor, aswell as the work of Jim Barger, Al Nelson,Dave Benima, and Dave Bryan, all ofwhom were key individuals in the develop-ment of the narrowband speech system.The authors also thank Richard Wien atthe U.S. Army Electronics Command(USAECOM) for his help and supportthroughout the entire program.

The work was sponsored in part byUSAECOM (Fort Monmouth) contractsDAAB07-76-C-0340 and DAAB07-75-C-1314.

References

I. N,lakl,,,! I ; "Linear prediction: a tutorial review," PrOC.ts3, No. 4 (Apr 1975) pp. 561-580.

2. Atal, H.S. and Schroeder. M.R.; "Adaptive predictive codingof speech signals." Sysi. Tech. J., Vol. 49, No. 6 (Oct 1970)pp. 1973-1966.

3. Mal. B.S. and Hanauer. S.L.; "Speech analysis and synthesisby linear prediction of the speech wave," J. Ammo. Soc. Am.,Vol. 50, No. 2 (1971) pp. 637-655.

4. Markel. .1.13.; "On autocorrelation equations as applied tospeech analysis." IEEE Trans. Audio Electrum -oust., Vol. AU -20 (Api 1W3) pp. 66-79.

5. Markel. .1.[S. and Gray. Jr.. A.H.; Linear preditrion of speech(Springer-% erlag. Berlin, 1976).

6. Feigenbaum. AD.; Helbig. W.A.; and Ozga. S.E.; "TheATMAC Microprocessor." RCA Engineer. Vol. 23, No. I

(Jun - Jul 1977) pp. 36-39.

49

Engineering InformationSurvey results

Part 2H.K. Jenny W.J. Underwood

High vs. low achievers how are they different?

This is the second article in a series that reports corporate -wide results of the Engineering Information Survey. Thefirst article in the October -November, 1977, issue of theRCA Engineer reported on:

Importance of keeping up-to-dateImportance of efficient access to informationUse and value of various sources of informationReading effort and obstacles

That article compared responses of engineers, leaders, andmanagers; this article compares high and low achievers,ignoring organizational level. By identifying and describingthe characteristics of high achievers, a kind of role model isdeveloped for use by all engineers.

For this article, an achievement irdex was developed basedon the answers to six specific questions in the survey. Theachievements and their survey questions are:

Technical currentness: How would you rate yourself interms of being up-to-date with the current state-of-the-artin your technical field?Awards: How many awards or recognitions (of a

technical -professional nature) have you received?Patents: How many patents (sole or with colleague) doyou have?Publications: How many papers have you as author or co-author published in the past five years?Presentations: How many formal paper presentationshave you made to engineering or scientific groups in thepast five years?Effectiveness as an information source: How frequentlydo other engineers seek you out to discuss technicalinformation?

The achievement index was constructed by translatinganswers to these six questions to a common scale and byjudgmentally weighting the factors. [At the end of this

article we give the details of how the achievement index wasconstructed; from this information, you can calculate yourown score.] The resultant achievement index yielded thedistribution shown in Fig. 1. High and low achievementgroups were defined as the upper and lower 300/o, placingapproximately 950 respondents in each group.

This achievement index is important in its own right. Butmore mpertant, such an index may be related to jobperformance.*

How important is keeping up to date?The previous article in this series established that engineersconsider keeping abreast of new technology quite impor-tant to both their present job and to their future goals. Thisanalysis asks, is it equally important to both high and lowachievers?

How important in your present job is keeping abreast of newtechnology in your field?

Importance Hi Lo

Extremely 54% 15%

Quite 38 43

Somewhat 7 30Slightly 1 10

Not at all 1 2

I n a separate study, four engineering organizations in different MOU's each identified sixhigh and six low performers and supplied data on patents, papers, and awards. Whenthese combined achievements were compared. we found that high performers hadproduced i7 times more achievements.

Other research has demonstrated a relationship between performance and several of theachievement variables. Three such studies are reported in

Managing the :low of technology by T J Allen (MIT, 1977) p 141-168Scientisrs in organizations by D.C. Pelz and F.M. Andrews (U of Mich., 1966, 2nd ed P

284-285

"Merit rating and productivity in an industrial research lab," by A Grasberg in the IEEETransactions on Engineering Management (Mar 1955)

50

How important to your future career goals is keeping abreastof new technology?

Importance Hi Lo

Extremely 51% 22%Quite 42 49Somewhat 6 25Slightly 1 4

Not at all 1 1

High achievers answered heavily in the two strongestdegrees of importance both for present job (92%) and future(93%). Only 58% and 71% of low achievers consideredkeeping up-to-date equally important. The highest scaleposition-extremely important-is also revealing. Nearlyfour times more high achievers assigned this degree ofimportance to the present job and more than twice as manyassigned it for future goals. Appreciably more low achieversconsider technical currentness more important for theirfuture than for their present jobs. Perhaps this implies thatlow achievers are in jobs that are not very technicallychallenging and either hope or expect the challenge willincrease in the future.

What's management's role?Most people in an organization respord to, or at least try torespond to, what they think is valued and expected of themby their management. It is important, therefore, to examinehow these groups perceive their management's emphasison technical currentness.

How much emphasis does your management put on theimportance of your staying abreast of new technology inyour field?

Emphasis Hi Lo

Very strong 16% 2%

Strong 30 12

Moderate 34 36Minor 15 32None 6 17

Three times more high achievers regard management'semphasis as strong. Perhaps more significant, nearly half ofthe low achievers believe that management's emphasis iseither minor or nonexistent. Low achievers woulc not findmuch encouragement from this influence to keepthemselves up-to-date technically.

What information is important?Efficient access to relevant informatioi is needed for theeffective performance of engineering. The survey exploredthe relative importance of five categories of information.

ACHIEVEMENT INDEX

DISTRIBUTION

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36

ACHEVEMENT INDEX

Fig. 1Achievement index was selected as the basis forcomparing groups cf engineers. This graph showsthe distribution of more than 3000 RCA engineersresponding to the survey as a function of achieve-ment index. The top and bottom "thirds" wereselected for this study.

Engineers generally start their careers with the same basic"tools."

"...high achieversuse appreciably more initiativein seeking out information andmake the effort requiredto be well informedbeyond the immediate job.

...These engineers produce more patents,papers, and presentations;are sought out more by colleagues;and are more up to date."

52

How important to you is having access to the variouscategories of information?

Category Hi LoTechnical-job related 96% 88%Technical-other 82 62Business-RCA 48 39Business-related industry 45 35Professional-engineering field 37 28

Numbers in the table are the percentages of respondentswho selected the two highest scale values of importance (ofa five -point scale). Both groups rank the categories ofinformation in the same order. However, all types ofinformation are more important to a larger number of highachievers. The largest difference between high and !owachievers is in the category Technical-other. The detailedresponse to this category is shown below.

How important is "Technical-other" information?

Importance Hi LoExtremely 35% 17%Quite 47 45Somewhat 14 32Slightly 3 5

Not at all 1 1

Twice as many high achievers assign the highest degree ofimportance to this category.

What are the important information sources?The survey also presented 32 sources of information thatcould be used by engineers. Respondents assigned a valueto each source from a four -point scale: very valuable,moderately valuable, somewhat valuable, little or no value.High achievers assigned more value than low achievers to25 of the 32 sources. The left column below lists the sourceshigh achievers valued appreciably more than did lowachievers; the right column lists the seven sources that lowachievers considered more valuable. These are also rankedin order from greater to lesser difference.

Valued more byhigh achievers

Conference proceedingsPapers external to RCAExternal conventions and

meetingsOutside technical journalsProfessional societiesRCA technical reports and

engineering memosRCA librariesRCA engineers at other locationsRCA -authored papersInternal meetings and seminarsEngineers at my locationRCA technical journals

Valued more bylow achievers

Military specs and standardsStandards handbooksRCA standardsEducational courses-internalEducational courses-externalHandbooks/manualsCatalogs

Thanks for taking that half hourDuring the f rst half of 1977, engineers and their supervisorsthroughout RCA responded to a questionnaire concerningthe sources that supply information needed in their jobs andcareers. This Engineering Information Survey wasdeveloped by the Technical Information Programs unit ofCorporate Engineering. The purpose of the survey was todetermine engineers' information needs for use in shapingprograms aimed at satisfying these needs.

The questionnaire was completed by over 3000 individuals,about 75% of RCA's engineers. Returns were sufficientlybalanced across such parar'eters as location, division,discipline, age, type of work, and job level to truly representthe RCA engineering population.

The major results of the survey are being reported in thisseries of articles in the RCA Engineer. This article con-cludes the general corporate -wide results; the next willexamine the specifics of some of the RCA informationservices, namely: the RCA Libraries, the RCA Engineer,RCA Technical Abstracts and TREND. Additional reportswill be made to local management, which will reviewlocation- and division -level results.

The staff of Corporate Engineering expresses sinceregratitude to all who responded to and administered thequestionnaire. Completing a detailed questionnaire caninterfere with business and personal pressures. The out-standing cooperation of RCA engineers resulted in anunusually high return rate and therefore credible data.

Except for educational courses, convenience or immediacyseems to be a difference. The sources valued more by lowachievers tend to be the available working tools ofengineers. Those valued more by high achievers have to besought out, requiring more initiative.

Note:The sources valued by low achievers should notbe interpreted as either unimportant or of lesserimportance. These are lists of differences between thetwo groups of engineers, not an importance ranking ofsources. An importance ranking was given in the firstpaper of this series (see the October -November, 1977,issue of the RCA Engineer).

A more specific inquiry was made regarding RCA -relatedinformation sources. Respondents were asked to indicatewhere they obtain the major amount of technical and non-technical information about RCA. There were no sub-stantial differences between the groups concerning non-technical information. Their responses to RCA technicalinformation sources, however, were:

53

Where do you obtain your technical information aboutRCA?

Source Hi Lo

Discussions with associates 84% 74%RCA Engineer 75 52RCA technical reports & engineering memos 73 39TREND 56 58Discussions with supervision 55 44RCA seminars and lectures 39 19RCA educational courses 30 20RCA Technical Abstracts 25 8

RCA Review 23 10

In every instance but TREND, high achievers made moreuse of the sources. The largest difference is for RCAtechnical reports and engineering memoranda, which drewnearly two times more high achievers' responses. Sub-stantial differences also exist for the RCA Engineer and forRCA seminars and lectures. In terms of each group's rankordering, low achievers assigned TREND a second posi-tion, whereas high achievers assigned it fourth.

Other points of difference between high and low achieversconcern use of the library and the RCA Engineer.

Library useI do not use the RCA technical library

at my location

Hi

4%

Lo

12%I use the RCA library more than once a month 70 45I use the RCA library for literature searchesI use the RCA library for journals and

proceedings

82

65

37

27I know how to use the library effectivelyRCA Engineer use

89 64

I receive the RCA Engineer at home 84 46I do not have access to the RCA EngineerWhat percent of the RCA Engineer do

you read?What percent of the RCA Engineer do

you scan?

5

21

60

33

16

37

How much time is devoted to reading?Since high achievers rather consistently value and use moreinformation and since reading is an important method, it isnot surprising to find the following difference in readingtime.

On the average, how much time during and after workinghours in a typical week do you spend reading?

Reading hours per weekHi Lo

Technical-job related 6.3 4.4Technical-other 3.3 2.6Business-RCA 1.5 1.1

Professional 1.7 1.3Business-other 1.3 1.2

Total 14.1 10.6

High achievers spend about one-third more time reading.The largest difference (43%) is in technical-job relatedmaterials. Both groups allocate about 50% of their readingto working hours.

The survey also inquired about obstacles to reading.

Which...are obstacles to you in keeping abreast of newdevelopments in technology by reading?

Obstacle Hi Lo

Insufficient time at work 71% 62%Insufficient time at home 52 41

Use of other methods to keep abreast 31% 23%

Management does not condone reading 18 22Need a better grasp of mathematics 10 19Need a better grasp of science 5 13Lack interest in reading 3 6

Reading wouldn't help in my job 1 8

The dotted line separates the responses by whether highachievers or low achievers responded more frequently.Numbers indicate percentage of response to the instruc-tion, "check all that apply." The two groups differ most onthe last four obstacles. Twice as many low achievers need abetter grasp of mathematics and science and lack interest inreading, and some believe that their jobs do not requirereading.

What is a high achiever?From these data, we can construct a profile of the highachievers. These engineers use appreciably more initiativein seeking out information. They do not limit themselves toinformation that is handy, which requires little effort toobtain. They place high value on being well informed andmake the effort that is required. They consider that beingwell informed includes their technical field beyond theimmediate job. As a result, high achievers participate morein professional societies as one means of maintainingbreadth, use a wide variety of sources of information, anddevelop contacts with engineers outside their work groupand even outside their locations. They study proceedings ofconferences, outside technical papers, and journals.

Internally, high achievers use the RCA libraries extensively,and read RCA technical reports and engineering memoran-da, RCA Technical Abstracts, and the RCA Engineer.

In general, high achievers keep their technical skills sharp,permitting them to comprehend information that oftencontains heavy mathematics and basic science. Theseengineers produce more patents, papers, and presen-tations; are sought out more by their colleagues; and aremore up-to-date. Their information gathering habits can berecommended as a model for all engineers.

Reprint RE -23-4-13Final manuscript received December 13. 1977.

54

To determine your achievement in-dex and see where you stand com-pared to the RCA engineers whoresponded to the survey, use thefollowing procedure:

Answer the following six questions.[These questions are taken directlyfrom the survey.]

1. How would you rate yourself in terms ofbeing up-to-date with the current state ofthe art in your technical field?

1 In the upper 10% of RCA engineers2 In the upper quarter3 About average4 In the lower quarter5 In the bottom 10%

2. How frequently do other engineers seekyou out to discuss technical information?

1 Very frequently2 Frequently3 Occasionally4 Rarely5 Never

3. How many papers have you as author orco-author published in past 5 years?

4. How many formal paper presentationshave you made to engineering or scientificgroups in the past 5 years?

5. How many patents (sole or withcolleague) do you have?

6. How many awards or recognitions haveyou received? (For example, David Sar-noff Achievement Award, Technical Ex-cellence Award, Status in ProfessionalSociety, etc.)

Use the table below to determine a subscore for each question. Then multiplyeach by its indicated weighting factor. Adding all six results gives you yourachievement index. Use this incex and Fig. 1 to determine your standing.

Q1

technicalcurrentness

Q2info. sourceeffectiveness

Q3No. ofpapers

Q4No. of

presentations

Q5No. ofpatents

Q6No. ofawards Subscore

5 5 0 0 0 0 04 4 1-2 1-2 1-2 1-2 1

3 3 3-5 3-5 3-5 3-5 2

2 2 6-9 6-9 6-9 6-9 31 1 9 9 9 9 4

Weighting2 1 1.5 1 1.5 2 factor

(multiplysubscoreby ...)

Example: Suppose you rate yourself as shown below:

From table above

Question Answer Subscore Weightingfactor

1 In Jpper quarter [2] 3 2

2 Occasionally [3] 2 1

3 1 paper 1 1.5

4 3 presentations 2 1

5 4 patents 2 1.5

6 no awards 0 2

Score

6

2

1.5

2

3

0

Achievement index (total score) =14.5

Your achievement index of 14.5 would place you with the upperquarter of RCA engineers (see Fig. 1).

Now compute your own standing. Are you satisfied with it?If not, consider the data in this paper carefully to gainsome insights on how to improve.

Biographies and photographs of the authors Hans Jenny and BillUnderwood, were published with their first article of this series inRCA Engineer, Vol. 23, No. 3 (Oct -Nov 1977) p. 30. To discuss thispaper or related issues, contact them at:Hans JennyTechnical Information ProgramsCorporate EngineeringCherry Hill, N.J.Ext. PY-4251

Bill UnderwoodEngineering Professional ProgramsCorporate EngineeringCherry Hill, N.J.Ext. PY-4383

-6

55

A high-performance switching regulatorfor advanced spacecraft power systems

C.A. Berard

1

SOLAHARRAY,UPPF

SO ARARRAY

OWERI

SHUNTRUM ATORIPAR TIAL I

POWER FI ON'---- CONTROL SIONAL S

Fig. 1

Direct energy transfer powerregulator.

Heavier power demands are making the design of spacecraftpower supplies and controls more challenging.

IIr CHAHGFREGuLATOR

I+28 VOLTS

MODECONTROL

BAT TEhv

-1 I

BOOSTVOLTACE

REGULATOR

SPACECRAFTLOAD.;

system centers around mode control and boost -voltage

Table IPerformance levels of the DET system are high; systems have operated four years on groundand sixteen months so far in space with no failures or degradations.

System parameters Performance level

Output interface

Regulated output voltage

Load rangeOutput ripple and noiseOutput impedance

+28.0 V + 0.56 V- 0.30 V

45 to 450 W; 530 W peak<50 mV p -p<200 mil and 0.3 µH

Boost regulator

Regulated output voltageLoad rangeConversion efficiency

+27.9 V ± 0.2 V0 to 450 W; 530 W peak>89% (110-280 W load)>87% (450 W load)

Source characteristics

Solar array currentSolar array voltageBattery voltage

0 to 32 A0 to 55 V (open circuit)16 to 23.5 V (discharge)21.0 to 26.1 V (charge)

Environment

Space operational lifeThermal rangeRadiationVibration (3 axes)

Shock (3 axes)

>3 years continuous-5°C to +50°C in vacuum400 to 500 nmi (polar orbit)22.5 g rms random15 g peak sine15 g peak

A new generation of high-performanceweather satellites developed by RCA uses aspacecraft power system that satisfiesdemanding performance requirements andis highly reliable, lightweight, and highlyefficient. Using a direct -energy transfer(DET) approach with a boost -modeswitching voltage regulator, this systemprovides optimal performance for earth -orbiting satellites above 400 nautical milesaltitude. At high altitudes where the effectof eclipse is less, system efficiency canapproach 100 percent.

The system features load power greaterthan 500 W at +28 V, line -load regulationunder 0.5%, output ripple and noise lessthan 50 mV peak -to -peak, and efficiencygreater than 90%. The design includesparallel paths with current sharing, hot -carrier rectifiers, and a modular, easy -access packaging approach.

Performance requirementsThe direct energy transfer power systemgenerates, stores, regulates, and controlsthe power used by all spacecraft systems.

The DET system accepts power from thesolar array and battery, both highly un-regulated sources, and delivers load powerfrom a well regulated, low -impedancepower bus; it also controls battery recharg-ing power from the solar array. (See Fig. 1.)

System operation is master -minded by themode control, which not only maintainsregulation but establishes the priority dis-tribution of available power (#1-loads,#2-battery, #3-shunt regulator). Theboost -voltage regulator operates in eclipseperiods and when the spacecraft's loaddemand exceeds solar -array power. Energyavailable from the solar array is returned tothe battery via the charge regulator; shuntregulators dissipate energy in excess of loadand battery requirements. Mode -controlsignals provide proper sequencing, guard -bands, and bus voltage control.

Table 1 summarizes performance for theDET system in its entirety and for its boost

56

voltage regulator. While an overall regula-tion of 2% is not in itself difficult, re-quirements on the control electronics areindeed stringent, considering that the rangeof source and load powers is greater than1300 W. All electrical performance is main-tained over the worst -case extremes of lineand load variation, temperature, radiationand aging drifts, and single -point failures.

Boost regulationtechniquesIn its simplified form, the boost -voltageregulator resembles the conventionalboost -configuration switcher.

Referring to Fig. 2, output voltage is sensedand compared to a voltage reference by acomparison amplifier. The error signal isamplified and loop stability is compensatedbefore pulsewidth modulation occurs. Aduty -cycle -limited, constant -frequencypulsewidth modulator circuit convertsvoltage error into a time -ratio error, whichcontrols the transistor power switches.When the transistor switch is on, theinductor current increases and storesenergy in its magnetic field. When thetransistor is off, energy is transferred viathe flyback diodes to the loads and energy -storage capacitors. The transistor switchesoperate at a frequency well above theresonance of the inductor -capacitornetwork.

The transfer function can be easily derivedby equating the net change of inductorcurrent to zero over a period of thetransistor switch. Assuming the inputsource voltage, E and output voltage, Eo,are essentially constant, this simplifies to:

Eo =

where a is the proportion of the period thatthe transistor switch is on.

This equation determines the maximumrequired duty cycle accurately and helpscalculate power transfer efficiency.

Circuit techniquesUsing switching regulators in high -currentapplications usually presents seriouscomponent -selection problems.

Paralleling transistors and diodes ,s un-desirable because of the difficulty and highcost of achieving and maintaining (over lifeand temperature) suitable matching. Very -high -current transistors exhibit relativelyslow switching times, which are incompati-ble with high -frequency and high -efficiencyoperation.

Parallel -path power sections can provide aunique way around some componentproblems.

A three -section approach (Fig. 3) had alower component weight than a single -

BATTERYBUS 0-rY-Y-Y-,--

INPUT

TRANSISTORPOWER

SWITCHES

Di

section design and also had no non-recoverable single -point failures. Com-ponents in each section are rated to carryone-half the load to further enhancereliability and ensure continued operationafter partial failure. All are drivensimultaneously from a commonpulsewidth-modulated error signal and areconnected to common input and outputpoints. Without selecting or matching com-ponents, tie sections share current withinrather close tolerance (measured data con-firms ±49 tolerance with full 450-W loadand 23-A total input current). Inductor L'and capacitors "3C" form a filter to removehigh -frequency noise caused by the non -ideal characteristics of the main filtercapacitor banks "6C." This circuit has beenspace qualified in I-, 2-, 3-, and 4 -channelversions; a 5 -channel regulator is currentlyunder development.

Reprint RE -23.4-11Final manuscript received May 19, 1977. A similar versionof this paper was presented at Powercom 3, June 24-26,1976, Los Angeles.

FLYBACKDIODE

_L

DUTYCYCLELIMITER

DRIVEGENERATOR

JUL

SUL

PULSEWIDTH

MODULATOR

17 5 kHzOSCILLATOR

ENERGY.1.s STORAGE

ERROR SICGAL

LOOPCOMPENSATION

NETWORKS

s'cLOADS

VOLTAGESENSE

RRORMP

VOL Al.EREF E HENCE

Fig. 2Boost regulator uses error signals to control transistor power switches, which regulateenergy flow to the loads from the flyback diode/storage capacitor network.

However, dealing with real switchingregulators requires the inclusion of (atleast) the first -order non -ideal parameters.

VOUT

VIN

The applicable equation is then

(Ed+LR.)

where

L

-1I +

I 3C

I

J_ -1 r -1 +

, I

6C

I rI ±__II6C

Ed is the flyback diode forward voltageL.- -.I L

drop A

E.,,, is the transistor switch voltage dropI, is the source current

A B INDICATES REDUNDANCYRL is the resistance of the inductor, andR. is the equivalent resistance in the Fig. 3flyback circuit. Parallel -path power sections have weight and reliability advantages.

57

POWERSWITCH

CONTROLSWITCH

DRIVECURRENT

SOURCE

.28 V BUS

DRIVESIGNALINPUT

Fig. 4Base -drive circuit uses transformer induc-tance in a resonant flyback network tbgenerate a reverse base current for rapidturn-off.

Identical base -drive circuits for each boostregulator section enhance high -efficiencyperformance.

Each section (Fig. 4) provides constant -current base -drive to the switchingtransistor; the novel feature is using thetransformer inductance in a resonantflyback network to generate a reversecurrent of about 400 mA for rapid turn-off.Efficiency and ground isolation areenhanced by the transformer couplingtechnique; 3 A of total base drive current isprovided by using less than 165 mA fromthe +28V output bus. Operation is (bydesign) controlled by the passive com-ponents; thus, performance is not in-fluenced significantly by transistorparameter variations.

Duty -cycle limiting is a positive means toprevent continuous application of base -drive to the boost -regulator switchingtransistors.

Duty -cycle limiting (Fig. 5) protects theregulator's switching transistors from over -current failure by using signals availablewithin the sawtooth oscillator. A diode -resistor network gates a saturating level tothe output to hold the pulsewidth-modulator output low (no base drive toswitches) while the sawtooth (integratedsquare wave) provides normal modulatoraction during the remainder of the period.By using an asymmetrical oscillator, anyduty -cycle limit may be generated; ourdesign incorporates a 65% limit.

TO BASEDRIVE

GENERATORS

LOWIPIPE DANCE BYPASS

ERRORINPUT

Fig. 5Duty -cycle limiting avoids continuousapplication of base -drive current to protectswitching transistors.

A mode -control circuit (Fig. 6) providesthe load priority, sequencing, and guard -bands between system modes; for voltageregulation, it provides the error -detectionand negative -feedback summing points foreach of the three regulators. A sharedreference -voltage source and resistivedivider maintain the tight tolerances foreach function and the tracking necessary toensure proper guardbands between modes.

Component selectionComponent selection for a switchingregulator is a complex process requiringtrade-offs of performance, weight, cost,and (for space applications) reliability andprevious applications history. Semicon-ductors are among the most difficult toselect and rely heavily on new componenttechnology.

Switching transistors were the most difficultparts to optimize.

Fast switching times, high current gain, lowsaturation voltages, broad safe -operating -area limits, high voltage and currentratings, and isolated case construction wereall desired. A 100-V, 30-A device waschosen for its performance characteristics.It is a double -diffused epitaxial type(similar to a 2N5330) with an acceptablyhigh second -breakdown resistance, thoughit is not as rugged as a single -diffused chip.Spacecraft loads are well defined, remov-ing much of the uncertainty of stress levels

28 V BUS SENSE

I llllllllllllll,11*

VOLTAGEREFERENCE

GROUND SENSE

BOOSTREGULATORERRORSIGNAL

CHARGEREGULATORERRORSIGNAL

SHUNTREGULATORERRORSIGNAL

Fig. 6Mode -control circuit establishes load priori-ty among the system modes, also provideserror -detection and feedback points for thethree regulation modes.

that would be present in the design of alaboratory- or OEM -type power supply.

Flyback diode selection was driven byseveral requirements, all indicating analmost -ideal diode.

Low forward drop was needed for efficien-cy, especially considering the desirability ofusing series -connected, redundant diodes.Fast switching and low recovery currentspiking are necessary to meet the low rippleand noise specifications without excessive-ly complex filtering. Hot -carrier rectifierssatisfied these stringent requirements best;a screened version of the TRW type SD -51was ultimately selected and used.

Inductors in the power -switching networkmust withstand high dc bias and be of low -loss design.

A gapped C -core of 3% silicon steel woundwith a 1.4" by 0.004" copper strip andKapton insulation provides suitablecharacteristics. Each 500-µH, 15-A chokeweighs 10 oz. and has 20 mil dc resistance.The core is rugged and needs no specialmounting provisions for vibration andshock resistance.

Capacitor selection is, like the semiconduc-tor choice, one of trade-offs.

Rectangular -case, hermetically sealed,tantalum -foil capacitors have been used inRCA satellites for over a decade withoutfailure or discernible performancedegradation. Our tradeoff evaluations

58

have consistently shown tantalum -foilcapacitors to have superior performance togelled -electrolyte types,* especially withrespect to low equivalent series resistance(ESR) and weight efficiency (and they arenot degraded by reasonable reversevoltage). Low ESR is essential to minimizeinternal heating (under high rms currentconditions) to prevent high temperature,which shortehs capacitor life; voltage stresslevels are much less significant. The rmsripple -current ratings for "identical" MIL -spec capacitors vary drastically amongparts from different manufacturers.Capacitors are one component where thetradeoff result may change as newdevelopments increase the performancecapabilities of competitive types.

Reliability techniquesBasic reliability techniques derived fromyears of spacecraft design experience wereapplied to all components used in thepower system electronics. These techniquesinclude extensive testing, screening, inspec-tion, and burn -in operations, all needing nofurther elaboration.

Configuration design and redundancy arethe primary means used to guarantee a highprobability of success for the boost -voltageregulator. Mode -control and low -powerboost -regulator circuits are independent,and back-up circuits in 'cold' (unpowered)redundancy are provided. Automatic +28-V bus failure -sensing and under- and over -voltage protections increase reliability.High -power portions of the booster haveparallel -path redundancy inherent in thethree -section configuration. The modecontrol and boost regulator have a

probability of success predicted to begreater than 0.9995 for three years ofoperation.

Packaging designThe power -supply package is designed forease of assembly; removal of plug-in circuitcards and simple unbolting of covers or sidepanels provides access to any circuit.

The package has two distinct com-partments. The EMI-gasketed end housesall components of the boost -regulator high -power switching networks, filters, andbase -drive generators (the major EMI -generating networks). Non -noisegenerating circuits, including low-level

Trade-offs consider worst -case effects relevant to space power -supply applications, including temperature extremes and agingdegradation in a radiation environment.

ISa -IR

25

20

15 -

10 -

5 -

-5

13-

15.

le - 15ALIMIT ',N.,

IR = 16A

CHARGE REG.

MODE

\11 1c 18=5A.LIMIT

REGULATED BUS VOLTAGE

SHUNT RE('-a MODE

1H 74

164

27.6 27.8

VB - 16 V -

28.0 28 2

BOOST RE G MODE

b. - Ve - 22 V

I- 80L. TEST IPSE --1.1

EOL W.C. SPEC. (PSE I

28 4

- SYSTEM SPEC.

Fig. 7Static regulation for the three modes of the direct -energy transfer is well within systemspecification; boost -regulator load regulation is 50 mV (0.18%) over normal load range.

portions of the boost regulator, final out-put filtering, and output power buses, arelocated in the opposite end, primarily onplug-in printed -circuit cards. Connectionsbetween the compartments use low -

capacitance feed -through terminals; dis-crete capacitors selectively decouple noiseand spikes where required.

Performance resultsEvery boost regulator made achieved thepredicted performance goals with adequatemargin from specification limits.Characteristics are identical within milli-volts from unit to unit, and retest of theprototype after almost four years of grounduse on several spacecraft showed no changein performance; on -orbit performance isidentical to the original test results.

Static regulation characteristics for thedirect -energy transfer power system arereproduced in Fig. 7. The vertical axis isscaled as a net current to make the resultsinterpretable for all load/ source com-binations. Line regulation for the boost

Clem Berard was lead design engineer forthe spacecraft switching regulatordescribed here. He is manager of a designand development group that is responsiblefor power supplies, high-speed motordrives, sevo electronics, and other analogfunctions.Contact him at:Advanced Analog TechniquesAstro ElectronicsPrinceton. N.J.Ext. 3231

S9

20

0.1

0.01

.001

SPEC. LIMIT

BOOSTREGULATOR

/VREGULATOR

CHARGE

10

-/SHUNTREGULATOR

1 I 1 11.1100

0.3 pH

0.2 II

\ ALL MODES\

1 1 11111 1 1 11111K 10K

F REGUENCY IHERTZI

1 1 1 11 111 1 , 1 1

100K

Fig. 8Output Impedance for each of the regulators remains below the 0.2 -ohm, 0.3 microhenryspecification.

96 - N%

95

94

93

92

91

90

as

ee

87

2

VIN - 22.5 VOL TS

VIN = 16 VOLTS

SPEC. LIMIT

6

VOUT 27 88 TO 27 94 V

I 1 1

10 12 14 16

REGULATED BUS CURRENT IAMPERESI

I IR

Fig. 9Power transfer efficiency is generally above 90%. Efficiency drop at higher loads isintentional, to reduce weight, because full -load operation is of short duration.

regulator is 20 mV (0.071%). Over thenormal load range of 2.5 to 16 A, loadregulation is 50 mV (0.18%).

Output impedance, measured for each ofthe three regulators in the DET powersystem, is shown in Fig. 8. Design of theboost -regulator filter network is stronglyinfluenced by the output -impedancespecification. High -frequency performanceis controlled by capacitor and wiringcharacteristics.

Boost -regulator efficiency is generally 90%or greater and exceeds the specificationrequirements (Fig. 9). Under orbital load -current profiles, full load is applied only fordata -readout transmissions (5-10 minutesduration) so the slightly lower full -load

efficiency, caused primarily by resistivelosses, is intentional to achieve lowerweight. At light loads, efficiency is reducedby the bias losses of the base -drive andcontrol circuits. Bias currents are 110 mAfor the entire DET power system.

Ripple and noise measurements were takenunder worst -case conditions of minimuminput voltage and maximum load. Actualperformance of 25 mV peak -to -peak(0.09%) as shown in Fig. 10 reflects theconservative worst -case design philosophynecessary for space applications. Less than10% variation is experienced over the fulloperating -temperature range.

Environmental exposures have had noeffect on the observed performance.

HORIZONTAL 20 msec cff,

Fig. 10Ripple and noise are only 25 mV p -p inworst -case conditions.

Thorough electrical testing has been per-formed over the operating -temperaturerange and after four years of ground usewith no discernible shifts. During vibrationand shock tests, the unit was powered at fullload with oscilloscope monitoring of theregulated output voltage; no variationswere observed. Performance is completelynominal for a unit with over sixteen monthson -orbit performance, even after exposureto a thermal environment exceeding thedesign range.

Summary andconclusionA high -current power system employing aboost -mode, pulsewidth-modulatedswitching voltage regulator has beendeveloped, qualified for space applications,and flown; stringent performance re-quirements have been achieved.

This design has been adapted for applica-tion to NASA and Air Forcemeteorological satellites with even higherload -power requirements and multiple -battery source arrangements. The basictechniques reported in this paper are easilyadapted to varying mission requirements.

A high -efficiency, high-performance,boost -voltage regulator extends the DETsystem usefulness and provides a well -regulated power bus over a wide range oforbits.

AcknowledgmentsThe boost regulator described in this paperwas developed for the DMSP satellite byRCA Astro Electronics under contractF04701 -72-C-022 l from the U.S. Air ForceSpace and Missiles Systems Organization(SA MSO). Personnel directly involved inthis development project were: Major O.E.Severo, Spacecraft Project Engineer(SA MS0); J.S. Kinsley, Electrical DesignEngineer; J.A. Streleckis, MechanicalPackaging Engineer; and R.A. Hoffmann,Electronic Technician.

60

The Universal Second -breakdown Tester:electronic cage for the power dragon

H.R. Ronan

If you never had to design and manufacturea power inverter or amplifier, you may notbe familiar with the second -breakdownphenomenon. Many of RCA's customersdo design power circuits and are familiarwith the problem it presents. Consider thefollowing hypothetical situation. Yourcompany has just experienced a rash offield failures of an audio power amplifierthat you had spent many months designingand evaluating in the lab. How is itpossible? After an engineering investiga-tion of the returned amplifiers, you findthat, in each case, the conservatively ratedpower -output transistor has a collector -to -emitter short. You're upset, so is your boss,and your company is out a considerableamount of money. What has happened?You have just been victimized by the powerdragon-second breakdown.

Second breakdown-what is it?When the energy absorbed by a transistorreaches a critical level, collector -currentfocusing takes place and produces localizedheating or "hot spots" in some areas of thesubstrate. If the current focusing, andhence the heating, is left unchecked, thecollector -to -emitter voltage will collapse,the junction will melt, and the transistorwill experience a collector -to -emittershort-second breakdown will have takenplace. Second breakdown can occur whenthe transistor is operating in either its cut-off or active region.1.2 It is this latter region,where the transistor is forward biased, thatis of concern in this paper.

Four variables are important in secondbreakdown:

I ) Case temperature2) Collector voltage3) Collector current4) Time

It is common practice, however, to speak offorward -biased second breakdown in terms

What do you do when testing costs too much and sometimesaestroys the power semiconductor? This solution

prcduced some unexpected ben9fits.

of the transistor collector current, with theother three variables held fixed. If youreview RCA Power Devices you will see thesymbol used for this set of conditions, 4b,listed under room -temperature electrical

100

characteristics. Normally, the valuespecified refers to a corner point on amaximum operating -area chart publishedfor the device (Fig. I).

MAX (PULSE(' - flood

2 " "

it

2

01

8 °

=1

I am

JJO

Is/b- LIMITED

.

-,- -4-TH7 11"4-VCED

: _

"X(2145583

I -a-1.(47r VCE0 ". 1 t

4-1. 114 12143584)

vc(0 MAX 300 V ---(253585 a 254240)

102 4 6 SOO 2

COLLEC113R-TO-EMITTE8 VOLTAGE (Vcc)-V

Ic ICOPETINL1005X2M38113 COLE)

a 55 C 135'50C (302/)

6 r5c(25vi00C 20 WI25E (15 W)450C (10 IV)

001

MAX..175(213603)%too MAX 250

4-1-""'"(21251:141 "

"cec, 544 300'255511.5 8 2542401

1 4 6 a 2 4 6100

COLLECTOR -TO -EMIT TER VOLTAGE OicE) -V

i

Fig 1

Second breakcown is a limiting corner pointmaximum operating area charts for devices.

003

on

61

Definitionsh second -breakdown

collector currentbase currentcollector currentcollector cutoff current,

emitter openVcE collector -to -emitter voltageVBE base -to -emitter voltagefr unity -gain cutoff frequencyQ -point static (dc) collector operating

point with no input signalcollector -emitter current gain

The maximum operating -area chart offersa concise profile of a power transistor, andis extremely useful for comparative andpreliminary application information, but itcannot guarantee the operational integrityof the device in a specific application.Therefore, cogent second -breakdown testsmust be performed on each device shippedto a customer. And therein lies the rub. Toprotect his customers, a transistormanufacturer would have to accept con-siderable profit loss as a routine risk andengage in testing brinkmanship by per-forming production second -breakdowntests. During tests, he would have to run therisk of bringing about second breakdownin, and consequently destruction of, hisproduct. RCA Solid State Power hasaccepted and met this challenge, perhapsthe only power transistor manufacturer todo so.'

Second -breakdown test equipment is not astandard product. Therefore, because ofthe importance of this test, and because of

PASSTRANSISTORVOLTAGESENSOR

L DIFCAMP.

PASSVOLTAGEADJUST

TRANSISTORUNDERTEST

IS/bDETECTOR

PASSSTRANSISTOR

TESTCURRENTDETECTOR

A

CUT-OUTLATCH

TESTCURRENTREAD OUT

DIFF.AMP.

TEST 4_1 VARIABLECURRENT PULSE -WIDTHADJUST GENERATOR

Fig. 2Conventional circuit configuration uses second -breakdown detector along with feedback circuitthat holds the transistor under test to a specifiedoperating point.

the risks and costs involved in a second -breakdown testing program. the RCApower activity studied the possibility ofproviding a "risk -less" nondestructive test.This paper chronicles the problems, com-plexities, and past experience with Lstesting and shows how, throughpersistence, RCA finally reduced second -breakdown testing to a routine virtuallywithout risk.

Early test circuitsFor many years, the basic circuit configura-tion used in testing second breakdownremained unchanged.

The test (Fig. 2) consists of three basicelements: Class A operation of thetransistor under test, use of an active seriesswitch, and an /, n failure -detection circuit.The most important element is the require-ment that the transistor under test beoperated Class A; i.e., be held to a specifiedVCE, and /c independent of current gain.Since transistor current gain varies widelyfrom family to family, unit to unit, andoperating point to operating point. a feed-back circuit must be used to regulate k.

This is done for two reasons. First, itminimizes operator error in setting testconditions. Second, it retains a reasonabletesting rate in the factory; the higher therate the lower the cost. But in regulatingcollector conditions to achieve circuitstability, it is all but impossible to divorcethe gain -frequency characteristics of thetransistor from a major role; and it is

difficult enough to test a transistor to near -

COLLECTOR CURRENT CONTROL

REFERENCE RPULSE

destruction without having to worry aboutstability.

Attempts at a practical solution to theseproblems have led to a plethora of manualtest equipments with limited applicability.

All require an extra insertion for testing/,1 and so increase testing costs. Ap-proximately five years ago, RCAdeveloped a second -breakdown test circuit,Fig. 3, that was a step closer to the ideal. Itallowed us to at least get a "feel" for thesecond -breakdown capability of most ofour volume types. It covered collectorconditions of 7 A, 150 V, and 200 Win bothn -p -n and p -n -p types. Because we hadsettled on a 50 -ms standard pulsewidth, wecould perform 1, b tests in our computer -controlled, automatic, dc -parameter(SCOPE) test sets.° This standardizationhelped eliminate the need for a separateinsertion in discrete second -breakdown testsets, but was, in reality, a tenuous solution.

The four transistor stages, including thetransistor under test (TUT), of the testcircuit's five -stage amplifier are unstablewhen the feedback loop is closed. This iswhere the operational amplifier enters thescene. In addition to providing a con-venient means of closing the control loop,the op amp acts as an integrator andstabilizes the amplifier.

This approach seriously limits thepulsewidths that can be handled by thiscircuit, but the circuit is perfectly adequateto handle our nominal standard of 50 ms

Fig. 3Computer -controlled tester made standardized testing possible and eliminated need for aseparate second -breakdown test set. Op amp closes the feedback loop and stabilizescircuit, but also limits the pulsewidths that the test circuit can handle.

62

for /sib pulse testing. This circuit is alsosensitive to changes in TUT characteristicsand becomes unstable when the TUT gainincreases from 30 to 200.

Historically, the failure -protection circuitsincorporated in second -breakdown testequipment have been collector -currentoriented.

These circuits detect an increase in collec-tor current, amplify the signal, and use it todrive a transistor in series with the TUT.Such a current -control circuit can in noway be expected to limit a sudden increasein /c because of its inherent slow speed, alimitation that places a premium on detec-tor and series -switch speed if ha, damage tothe TUT is to be prevented. The commonpractice of employing commercial powersupplies for VcE compounds the protectionproblem. These supplies invariably have ahuge output filter capacitor that is capableof delivering extremely high short-circuitcurrents, essentially limited in value onlyby wiring inductance. Devices under testthat have excessive /cb. (excessive to theextent that the protection circuit is trippedwithout the TUT going into second -breakdown) yield false test results andfurther complicate the matter.

Inaccuracy is an additional problemprecipitated by the current -detection ap-proach. The Class -A -operated seriesswitch in the test equipment makes preciseautomatic control of TUT VcE all butimpossible. Second breakdown is extreme-ly sensitive to collector voltage, i.e.:

!sib a licE4

COLLECTOR CURRENTPULSE REFERENCE

CONSTANTCURRENTGENERATOR

ADD

V8ECORRECTIONCIRCUIT

TUT

IB BASECORRECTION CURRENTCIRCUIT SAMPLE

where n ranges between 1.5 and 4,depending on device design. Therefore, ifcollector current is not carefully con-trolled, serious errors can occur. Fig. 4shows experimental data for a 2N3055transistor for which a 9% shift in `CE atconstant /c resulted in a 2000% change inenergy capability.

The solution-The UniversalSecond -breakdown Tester

In the face of these many problems,persistance, cooperative effort, and in-genuity have produced a nondestructive,accurate hzb tester, called the Universal hzeTester.

The key to the Universal Tester is inpreventing the dynamic characteristics ofthe TUT from having a detrimental impacton overall circuit stability.

Fig. 5 illustrates the approach and Fig. 6shows the collector -current control circuitin more detail. The TUT is connected in agrounded -base configuration. A constant -current generator in series with the emittercontrols circuit current; a simple voltagesource determines collector voltage.Regenerative feedback of base current andbase -emitter voltage regulates a specified Qpoint, /c, and VcE-independent of TUTgain variations. (The Q point is the static(dc) collector operating point with no inputsignal.) This feedback is necessary becausethe collector current is controlled throughthe emitter and VcE through VCI3.

The value of this circuit approach can beillustrated by the following statements:

ADD

COLLECTORVOLTAGE

4-

GENERATOR

COLLECTORCURRENTCROW MR

I S/bDETECTORa LATCH

COLLECTORVOLTAGEREFERENCE

±

Fig. 5The solution-the Universal Second -breakdown Tester. Testing method keeps thedynamic characteristics of the transistor under test from having a detrimentalimpact on overall circuit stability.

60

59

W

a 580

5 57

56 -02 55-- J

0U

0

2N3055 SINGLE UNIT

400 600TIME -MILLISECONDS

800

Fig 4Slight changes in collector -emitter voltage can leadto serious errors in testing. For 2N3055 deviceshown here, 9% voltage change produced 2000%change in energy capability. Such sensitivity makesthe current -detection approach potentially inac-curate.

1)The circuit is unconditionally stablewithout regard to the transistor testedand without the need of added compen-sation.

2) The bandwidth of the circuit is un-affected by the gain of the transistortested because the variations in theexpression 13/$+l, where (3 is transistorcurrent gain, can hardly be expected tobe more than 20%.3) The bandwidth is extremely broad andalways larger than the value of the unity -gain frequency cutoff, fr, of the TUT.For example, substituting standardvalues into the circuit of Fig. 6 produceda roll -off frequency 104 times fr. Thebandwidth of the test circuit is, then, fourpowers of ten better than that of the unittested!

Fig 6Collector -current control portion of the uni-versal tester. Feedback circuit keeps Q pointstable.

63

Making the "unfeasible" feasibleAn industry newspaper3 recently reported that

"...while it would be a great idea for manufacturers to test each powersemi to the limit of its safe operating area [the second -breakdown point],it doesn't seem feasible at present. ...[the manufacturers] don't test forthe parameter [Safe -Operating Area, SOA], so the SOA spec is only animplied guarantee-something they will not or cannot test. ...it'ssomething that most manufacturers are not skilled to handle....S0A can'tbe known with certainty...".

The Universal Second -breakdown Tester seems to obsolete this opinion.

4) The current risetime for the circuit isapproximately 1 µs, a truly impressivetest -circuit credential!

What has this circuit configuration done forsecond -breakdown testing?

1) High TUT current gain actuallyreduces the necessary feedback.Darlington transistors present noproblems.

2) The frequency response of both theemitter -current generator and collector -voltage drive can be maximized withoutregard to the TUT. The circuit can beeasily reproduced.

3) Control circuits can be checked outand repaired independently of the TUT,greatly easing maintenance tasks.

4) Test -set proliferation is drasticallyreduced.

5) The high-speed series -switch /',-shutdown transistor circuitry is no longernecessary. An increase in emitter current

VCEENABLE

SHUNT VcE REGULATOR

caused by TUT second -breakdowncollapse is instantly resisted by aneffective rise in the output impedance ofthe emitter constant -current generator.Without a series switch, VcE control isaccurate.

Circuit descriptionsand advantagesThe many limitations on collector -emittervoltage -supply design imposed by second -breakdown testing are effectively handledby the universal circuit.

The VcE control portion of the tester circuitis shown in Fig. 7. The maximum allowablesecond -breakdown fault current is definedfor all Vice, /c conditions by using a fixed"brute -force" voltage and series resistor. Astest -current ranges are changed (k = IA,2A, etc.) the series resistance is varied tomaintain the power available for testingapproximately constant. This techniquewastes power, but prevents test -set damagein the event of misprogramming.

CURRENTBOOSTER

'AbAsk CONSTANT

EMITTERCURRENT

TO

DRIVE

EE

OFF

Fig. 7Collector -emitter voltage control uses afixed "brute force" voltage and variableseries resistance to keep the testing powerconstant. Shunt regulator protects againstshort circuits, and thyristor switchterminates test current quickly if testtransistor goes into second breakdown.

TUTT

SERIESSWITCH

BRUTEBRUTEFORCESUPPLY

(11)BRUTEFORCE

ON

The shunt -regulator configuration protectsthe test set from short circuits and reducesthe dissipation absorbed by the regularcircuit when it is operating with test -pulsetimes of up to one second. The thyristorswitch in series with the brute -force supplyprevents excessive dissipation when thecircuit is in a standby (not testing) condi-tion, and helps to quickly terminate testcurrent if the TUT goes into secondbreakdown.

The shunt regulator is emitter -driven toenhance step response time. Circuit stabili-ty is not sensitive to pass -unit current gainin this configuration, and the only require-ment for the series string is to blockvoltage. The voltage capability of thecircuit is easily modified by adding passunits-up to 600 V have been handled inthis manner without adversely affecting thedynamic response of the circuit.

Second -breakdown testing offers a poorenvironment for digital logic because of thetransients inherent in switching largevoltages and currents.

The potential for self-destruction of theTUT, coupled with the potential loss oftester effectiveness, makes sound control ofsecond -breakdown test equipment rigidlyexacting. The four -to -one noise immunityadvantage that COS/ MOS devices main-tain over bipolar integrated circuits makesthem the first choice of logic family. Thesystem logic design is not so clear-cut.Monostable multivibrators (one -shots) aretempting because they are efficient,automatically reset themselves, and adaptwell to the wide range of timing required.Unfortunately, circuits using one -shots areextremely sensitive to noise.

The only recourse would appear to be asynchronous approach based on a clock.While a clocked system handles sequentialevents efficiently, it is cumbersome to applyto a wide time range and requires carefulbookkeeping of event flip-flop set/ resetlines, particularly when external sourcesinitiate some of the signal inputs. The usualmethod of handling these signal inputs,edge -setting flip-flops with RCdifferentiating networks, would have

negated the noise immunity gained byusing a clocked system. None of thesecompromises were made in the final designof the Universal I, ;b Tester. Not onecapacitor was used in the logic network.

The key to the success of the tester controlcircuit is the gating circuit illustrated in Fig.

64

8. This circuit issues a single completeoutput pulse, independent of the phaserelationship between the clock pulse andinput -level change. The only requirementneeded for an output is that the input betrue during one period of the clock. (If theinput goes false before the appearance ofthe first clock pulse, no output is issued.)This arrangement allows event flip-flops tobe triggered with a solid pulse of knownenergy and frees the designer from detailedset/ reset bookkeeping.

An unexpected dividend accrues from thistechnique. When an element in the logicnetwork fails to function for any reason,the system stops in that state-an in-valuable aid for debugging. A furtherrefinement of this circuit was reached byusing multiphased clock pulses developedfrom every other stage of a Johnsoncounter; the refinement eliminates thepossibility of a logic race anywhere in thecontrol circuit. Fig. 9 is a flowchart ofUniversal I, b Tester operation.

CLOCK

CLOCK

IN

OUT

OUT

Fig. 8Gating circuit solved timing problems without introducing noise. Circuit issues singleoutput pulse, independent of clock/input phase.

SELECTIC0-1.99A

SELECTVCE0-300 V

>300W2

YESALARM t

START(INSERT TUT)

TEST SETPASS UNITSCOOL

NODISPLAY"WAIT"(ROTATING)

YES

HAS SETLOOPED INWAIT MODE?

DISPLAY"READY"(BLINKING)

SELECTNPN/PNP

SELECT TIMERANGE B VALUE'0-999ms0- 99 9 ms0-999 ms

YES NO

WAIT 60 pSENABLE Is/bMONITOR

COUNT DOWNFROM TIMESELECTSETTING

OPERATORHANDREMOVED 2

ENTERCOOLINGCOUNT DOWN

DISPLAYAGAIN"

DISENGAGEBRUTE FORCEVCE . IC YES

UNIT ENTERIS/b

YESDISPLAY 'DISPLAY DISENGAGE ENABLETIME TO *-"FAIL" 4 -BRUTE FORCE CROWBAR

NOFAILURE VCE 'IC

FINISHEDCOUNTING ?

DISPLAYFULL SCALETIME SELECT

DISPLAY"PASS"

4- D I SENGA_GBRUTE MRCVCE.IC

YES NO

ENTER COOLING ...-COUNT DOWN

Fig. 9Operating for over a year on the productionline, tester gives devices second -breakdowntest and displays results as "pass" or "fail"(with time to failure). Notice protectionagainst overpowering, overheating, and test-ing open/short devices.

DISPLAY"SHORT"

DISPLAY"OPEN"

ENGAGE BRUTEFORCE SUPPLY

EN LE k1213 ms

Ic HIGH-SIc LOWWAIT I ms

ENABLE VC

10

0.1

TIME -MILLISECONDSFig. 10Typical second -breakdown curves (a magnification of a small section of the curves in Fig. 1)show uniform and repeatable readings, with no apparent error introduced by tester. Curvesalso show that second -breakdown capability is nearly a constant -energy characteristic, andthat readings at 50 ms can be related to readings at 1 s.

Factory experienceAfter a year of operation in the productionline, the Universal Second -breakdownTester has more than met expectations.

It has successfully tested every devicewithin its 300-W power range, or 90% ofSolid State Power's product line. Theequipment has performed with minimummaintenance and has been oscillation -free.More importantly, it has afforded tightcontrol of second -breakdown character-istics since exact and repeatable readingscan be made on individual devices withoutfear of unit degradation or destruction.

The new tester has eliminated much of thepast frustration associated with second -breakdown testing.

Without exception, every previous second -breakdown test set was go/no-go in con-cept, making every device test an attributeassessment only. Holding test sets incalibration had to be done indirectly.Establishing limit units for correlationpurposes was a tedious, unrewarding task.After starting with 100 devices and in-vesting countless hours "inching" each unitup to its failure point (destroying most ofthem along the way), one was left withperhaps four or five units, about whichonly one second -breakdown point wasknown. The old methods had meagerdividends for such prodigious effort.

New knowledgeSecond breakdown appears to be a nearlyconstant energy characteristic.

As a part of the Universal 1, b Testerevaluation, 100 2N3585 devices werestudied in detail. A family of second -breakdown curves was developed for eachunit; each unit was repeatedly driven intosecond -breakdown 40 to 50 times with 80%survival. The plotted results are mostinteresting; Fig. 10 is typical. The lack ofdispersion of the readings solidly supportsthe view that errors contributed by thetester are negligible and that thephenomena demonstrated are both uni-form and repeatable. Keep in mind thatthis plot represents a considerablemagnification of an extremely small por-tion of Fig. I, the maximum -operating -area chart. Fig. 10 also shows that for fixedVCE, second -breakdown capability is nearlya constant energy characteristic, and thatreadings at 50 ms are related to thoseobtained at Is. This relationship hadalways been clouded in the past because ofthe inability to generate large volumes ofdata on individual units.

Second breakdown may be non-destructively predictable.

During tester evaluation, a great manyoscilloscope photographs of typical

VCE

50V /C M

IC

IA /CM

10 ms/cm

a) Collector current and collector -emittervoltage for 2N5240 device;

50 V/CMVCE

IA /CMIC

10 ms/cm

b) same device as in a), but with vo'tageincreased to induce second breakdowr;

ICIA/CM

1000/cm

c) second -breakdown current;

VBE0.2V/DIV '14101111111aph.-

5 ms/ DIV

d) different device, 2N3585, before secondbreakdown; collector -emitter voltage is 100V and collector current is 700 mA.

0.2V/DIV

I ms/DIV

e) 2N3585 after damage; collector -emittervoltage is 17C V and collector current is 200mA.

Fig. 11Typical waveforms taken during testerevaluation demonstrate the second -breakdown process.

66

waveforms were taken (Fig. 1 I). Thesephotographs show that the average slope ofunit VBE as a function of time is inverselyproportional to the energy capability of thedevice-a very startling observation. Morephotographs were taken and the averageslope graphically determined. The result,plotted in Fig. 12, established therelationship of VBE to time beyond a doubt.Because second breakdown is a thermalfailure, one must conclude that the averagepellet temperature is a reasonable measureof hot -spot temperature, and that the VBEmeasurements actually indicate the changein hot -spot temperature, independently ofchanges in /c and 18. Further study revealsthat:

I) /c cannot add to variations in VBE as afunction of time because tester circuitsregulate it to a constant value; /c canonly contribute to the static VBE offset.2) When the collector operating point issuch that $ varies widely withtemperature, the h is small enough tocontribute only a few percent to thechange in VBE,

3) When the base current le is a largecontributor to VBE (low -fl Q point) itchanges very little with temperature andagain contributes but a few percent tothe change in V BE.

It is possible that the observedrelationships may be peculiar to the2N3585 device; however, it is difficult toavoid speculating on the possibilitiesshould the relationships prove to begenerally applicable. The first that comesto mind is using VBE slopes as a second -breakdown indicator without actually in-ducing second breakdown. A second, andpotentially more important, application isin using the relationships to investigate andeventually control transistor real-timethermal response without using samplingtechniques.

Second -breakdown damage may leave"fingerprints."

In obtaining the data for Fig. 12, a numberof units were damaged. (Damage wasconfirmed by subsequent data points thatwould not match the previously developedsecond -breakdown history for the units.)In each case, the device exhibited a dualV8E-characteristic slope as in Fig. 11c.

Should the dual slope after damage proveuniversal, it would firmly establish whetherany given circuit was responsible for thedamage of a device. The implication is thatwithout prior knowledge of device

2-

01

veE VOLTS/SECONDatFig. 12Second -breakdown prediction without in-ducing breakdown or damage may be possi-ble by developing curves showing base -emitter voltage vs. time to failure. Plot here isfor a 22 -unit sample for one device.

capability, observation of V BE character-istics in a Class -A circuit is sufficient toestablish damage. No more in-house dis-putes or factory/ customer confrontations.

How the testerwill help SSD businessThe new Universal /,/8 Tester will withoutdoubt improve Solid State Division'spower semiconductor business posture.Specifically, it will help us in these threeareas:

I) Distributor market-since our stand-ard products typically have 2 to 3 timesthe capability of published ratings, wewill be able to guarantee higher second -breakdown ratings and command thepremium price the higher ratingswarrant.2) Custom market-Marketing will havea competitive edge because we will knowour power capability accurately.3) High -reliability market-We will beable to supply data -logged readings ofsecond breakdown, a capability that upto now has not existed in the industry.

Acknowledgment

The author thanks Dr. R. Sunshine of theRCA Laboratories, Princeton, for his un-stinting support and encouragement dur-ing the pursuit of a viable test -circuitdesign. The assistance of Nicholas Magdaduring the difficult experimental anddebugging stages of the design is alsorecognized; his original ancillary circuitswere a valuable contribution to the finaldesign.

References andbibliographyI. Turner. C.R.. "Carl Turner of RCA explores selection of

second breakdown -resistant transistors," EEE, Vol. IS No. 7(Jul 1967).

2. RCA Corporation; Solid State Power Circuits, TechnicalSeries SP -52, pp. 113-145, 1971.

3. Sugarman, R ; "Bipolar power series getting more reliable, butno technology breakthrough in sight as designers juggle oldparameter trade-offs," Electronics Engineering Times, Nov 22,1976.

4. Cutshaw, E.; "Scope-the power testing paladin," RCAEngineer, Vol. 21, No. 5 (Feb Mar 1976) p. 22.

5. Moe, D.A.; "Testing for forward -bias second breakdown inpower transistors," RCA Application Note AN -4573 (Jul1971).

6. Fleming, D.J.; "Thermal breakdown delay time in silicon p -njunctions," IEEE Trans. on Electron Devices, Vol. ED -18 No.2 (Feb 1971).

7. Sunshine. R.A. and Lampert. M.A.; "Second breakdownphenomena in avalanching silicon -on -sapphire diodes."private communcation.

8. Sunshine, R.A.; "Infrared observation of current distributionin large -area power transistors," IEEE Trans. on IndustrialElectronics and Control Instrumentation, Vol. IECI-21, No. 3(Aug 1974).

9. Sunshine, R.A. and D'Aiello, R. X.;"Direct observation of theeffect of solder voids on the current uniformity of powertransistors," IEEE Trans. on Electron Devices (Feb 1975).

Reprint RE -23-4-10Final manuscript received October 6. 1977

Hal Ronan has designed test equipment forall of Solid State Division's product linesduring the last 18 years. Besides the testerdescribed here, he has also worked on thecomputer -controlled SCOPE testing systemand thyristor high -voltage testing.

Contact him at:Power DevicesSolid State DivisionMountaintop, Pa.Ext. 633

67

Spares allocation for cost-effective availabilityH.R. Barton

Initial spare -parts allotments for electronicsystems are normally computed item byitem, but this approach does not necessari-ly maximize total system availability foreach dollar invested in spares. Sparesquantities can, however, be determined in away that accomplishes that objective. Themethod described here, marginal utility,determines which spare will achieve thegreatest increase in system availability peradditional dollar invested. The marginal -utility model adds that item to the inven-tory list, determines the increasedavailability, and then iterates theprocedure until the availability objective isachieved.

Spares requirements

Reliability analysis of a system allows theengineer to assess the consequences ofsingle and multiple failures of eachspareable item. More than one failure maybe required for a system failure if thesystem has redundancy or if some degrada-tion of performance is allowable. Once thenumber of failures required for systemfailure is determined for each spareableitem, a spares protection criterion can thenbe established.

The number of failures required for systemfailure should be an independent numberfor each item. Independence simplifies thecomputation of spares requirements.

Independent failure criteria can representperfectly only those systems which havesimple redundancy relationships or com-pensating rules.

Fig. I gives an example of a system withsimple redundancy. Failure of the systemrequires failure of a critical number of anysubsystem boxes. Fig. 2, however,describes a system with dependent redun-dancy. Two failures of subsystem box typeC, D, or E are necessary for system failure,but only if failures have not occurred inothers of these box types. A failure of CIand D2 can produce system failure with orwithout failure of C2. Simultaneous failureof CI and DI cannot cause system failure,however, without a failure in the second

Which spare parts should you stock? This method ofselection has cut spares costs by as much as 75% andsimultaneously improved system availability.

CRITICAL FAILURE NUMBER

I I 2 I

Fig. 1Simple redundancy. Failure of this systemrequires the failure of the number of sub-systems listed below each subsystem box.

chain. Exact representation of dependentredundancy effects upon spares re-quirements is beyond the scope of thisdiscussion, which is restricted to indepen-dent failure and spares protection criteria.

However, it is possible to establishoperating rules to support independentspares protection criteria. For example, acannibalization rule would permit ex-changing DI for D2 if CI and D2 failed andno spares were available for either. Thisrule would establish the failure number attwo for boxes C, D and E. Without acannibalization rule, a failure number oftwo could be assigned to one of the boxtypes and a failure number of one to each ofthe others in the redundant chain.

The spares assignment should provide apredetermined probability that the systemwill not be unable to recover from a failedstate because of an unavailable spare.

This can be expressed as a probability ofspares sufficiency, spares confidence level,or spares availability. Since operationalavailability represents a probabilityaveraged over time, spares availabilityshould also be expressed as an average timevalue.

Spares protection is determined by theprobability of having sufficient spares for adesignated period, and also by the resupplymethod.

Some spares replenishment policies call forrelatively continual replenishment ofdepleted supplies; other policies requirereplacing a whole inventory's depletions inone batch. Most land -based systems havecontinual access via depot "pipeline" and

NEEDI OF 2 CHAINS

CRITICAL FAILURE NUMBER

I I 2 2 2 I

Fig. 2Dependent redundancy. With "cannibaliza-tion" allowed, these numbers of subsystemfailures are required for system failure.

so fall into the first category. Shipboard orairborne inventories are not generallyaccessible for replenishment except in port,so inventory deficiencies must wait untilthen for batch replenishment.

Because of the random nature of demandsunder a policy of continual replenishment,the expected spares -replenishment delaydoes not vary with time. Spares availabilitycan thus be computed in terms of theprobability that the number of failuresduring a replenishment period will notexceed the critical failure number plus thenumber of spares, for each type.

Batch replenishment creates a moredifficult computation problem because theexpected spares -replenishment delay doesvary with time. A stock failure near thebeginning of a mission may well abort themission because of an intolerablereplenishment delay. A stock failure at theend of a mission, on the other hand, wouldbe inconsequential. Computing the sparesavailability under batch replenishment re-quires time averaging to account for thevarying replenishment delay.

Computing probabilityof demandIf the operating population of an item islarge or the failure expectancy is very small,the spares demand rate is relatively con-stant for any likely failed state. This situa-tion is possible even with a population ofone, if the probability of failure is verysmall. In this circumstance, the probability

Reprint RE -23-4-1A similar version of this article appeared in LogisticsSpectrum, Winter 1976.

68

Harvey Barton has specialized in operationsanalysis, particularly logistics, at RCA. Hiswork has included the development ofanalytical methods for solving problems ofsystem design and support. He is presentlyManager, Product Effectiveness Analysis,for Government Communications Systems;earlier, he was Manager, Product Assuranceand Applied Research, RCA Service Co.Contact him at:Product Effectiveness AnalysisGovernment Communications SystemsGovernment Systems DivisionCamden, N.J.Ext. PC -2442

of a stock failure can be approximated bythe Poisson distribution.

Some inaccuracy is introduced by using thePoisson process to represent failureprobabilities of items that are in the systemin small numbers. This is especially truewhere expected use is significant, or whereactive redundancy exists. The true failuredistribution is binomial with no spares andapproaches the Poisson as the number ofspares approaches infinity. A reasonablerepresentation of the process is possiblewith the state probability model, which issimilar to the Poisson, but differs when thenumber of operating units decreases withsucceeding failures.

The Poisson model is limited to spareableitems with expected usage of somewhat lessthan 150 because of computer exponentlimitations. One way of overcoming thishandicap is by approximating the Poissonwith the normal distribution. When ex-pected usage is 100 or more, the ap-proximation is reasonably accurate.

Optimizing sparesallocationsWhen an inventory consists of a largenumber of items, a large number of alter-native inventory allocations can satisfyspares availability constraints.

In order to choose among the acceptablealternatives, a ranking criterion is

necessary. The most favored criterion iscost. Storage space could be used, butwould have an advantage over cost only ifspares costs are low relative to the spacethat spares occupy, an unusual situation.

The optimum allocation is achievediteratively by applying marginal utility.

At each step in the optimization process,the spareable items are compared with eachother to determine which spare should beadded to maximize the increase in protec-tion per unit cost. Mathematically, themodel determines Max APf/ C where

APf = increase in protection achieved byadding a spare, andC = cost of adding that spare.

The spare having maximum cost-effectiveprotection is added and the comparison ismade again. The process is continued untilthe spares protection criterion is met.

The marginal utility process is valid onlywith monotonically decreasing functions.Although a probability density function isnot generally monotonic throughout itsrange, a monomodal function has a max-imum at or near the mean value. Thevalidity of the optimization is assured byproviding an initial number of spares ofeach type to cover expected usage for thereplenishment delay.

To reduce computation requirements, thenumber of optimizing steps should be asfew as possible. To do this, Pm,,x is definedas one minus the minimum spares protec-tion requirement. Then, since the totalprobability of stock failure is constrainedto be less then P., the probability of stockfailure for each item must certainly be lessthan P.... Mathematically, since

Pp -.<... P...., then Pp Lc.

Upon this basis, spares of eadh type areadded when necessary to reduce the

COMPUTE SYSTEMFAILURE

CONTRIBUTIONP I 1

probability of stock failure per type to P..xor less, prior to optimization. Fig. 3 il-

lustrates how the optimization process isapplied.

Batch replenishment has two significantdifferences from continual replenishment.

First, there is a periodic opportunity torestore the inventory. This is not possibleunder continual replenishment because theconsumption of spares becomes negligiblein comparison with resupply rate duringthe replenishment action. Thus, 100%spares availability is not very probableunder continual replenishment becausethat system implies a continual state ofsystem operation and spares demand.

The second major difference between con-tinual and batch replenishment is thesensitivity of availability to the time offailure. In batch replenishment, if an item'sinventory is exhausted near the beginningof a mission, the system is likely to theinoperable for a major part of its missiontime. If such a depletion occurs near theend of the mission, the consequences aremuch less serious. In order to relate sparesprotection to operational availability,protection time must be considered as wellas protection probability. This can beaccomplished by integrating the sparesprotection probability over the missionduration. The result is defined as sparesavailabilit

Optimizing redundant -unitallocationRedundant -unit allocation can be optimizedas readily as spares allocation.

The techniques described here can establishthe optimum configuration of redundancyfor a system that meets the requirementsdescribed under the earlier section onredundancy. Applying this method toredundancy requires careful considerationof the similarities between the analysis ofspares protection and the analysis of redun-

SUM SYSTEMFAILURE

CONTRIBUTIONSPTOT C(PW

COMPUTE CHANGEM FAILURE

CONTRIBUTIONFOR NEXT SPARE

PRINT ITEM$ CU FT , P(I),S TOT, T VOL , PTOT

FIND MaxAPf/C

ADDONE

SPARE

Fig 3

Marginal utility- process determines the increase in protection that each additiona spare willadd. along with the cost of adding that spare.

69

dancy. Redundant units enable continuedsystem operation while a failed unit is beingrepaired or replaced. If the repair/ replace-ment period is made equal to the period forspares protection, inherent availability canbe analyzed by the same models used forspares availability. This is more evidentwhen inherent availability is considered tobe a probability, as in operationalavailability.

Some systems experience varying times forrestoration to active service, dependingupon characteristics of the failed unit. Amodified version of the spares modelmakes it more convenient to treat thisvariation by using individual item restora-tion times.

Analyzing redundancy derived from usingstandby units is relatively easy, since theredundant units are equivalent tocollocated spares. Analyzing redundancyobtained with active units is a more com-plex problem, since the model must ac-count for the increased operating failurerate with the added redundant units.

The model can analyze inherent availabili-ty and optimize the allocation of standby oroperating redundant units to meet acriterion of inherent availability. Opera-tion is like spares allocation, and is ad-dressed to a permissible level of un-availability, which is related to common orindividual repair/ restoration periods. Themodel may be used with either continual orbatch restoration.

Application experienceThe marginal utility method can increaseavailability or decrease cost.

Two different models have been used toallocate spares requirements for severalmajor systems. The batch replenishmentversion was used to estimate the cost ofshipboard spares. An initial review of oneshipboard system showed the possibility ofreducing spares cost by 75% while in-creasing spares availability several ordersof magnitude. In another case, the con-tinual replenishment version establishedthe required site spares for a land -basedsystem. In addition, the models have beenused, with some modification, to optimizeactive and passive redundancy in operatingsystems under design.

Fig. 4 shows three different sparesavailabilities that can be achieved for asystem by using three different sparesinventories. The area above any givencurve represents the probability that thesystem will be non -operational because ofthe lack of a spare. Curve A represents thetime -variant availability of a spares inven-

Table IConventional spares provisioning has very low aggregate availability when compared to theoptimized provisioning.

Conventionalprovisioning

(0.90 per item)

Optimizedprovisioning

(0.90 aggregate)

Average probability ofsufficient spares per item 0.9808 0.9996

Aggregate probability ofsufficient spares, 270 items 0.0053 0.9245

Cost $134,000 $167,000

t i.o

o.e

E 0.6

O

1- 0.4

OE. 0.2

0.010

A $136,000,100% SPARESAs: 0 90

B 557,000,865 SPARES,As: 0 90

C $13.e,000556SFARFS,LINE

ITEM CONFIDENCE = 090

20 30 40 50 60 70 80 90TIME (DAYS) -

Fig. 4Comparison of conventional and marginal -utility spares allocation methods. Area aboveeach curve represents the probability that the system will be inoperable because of the lackof a spare. Curve C is for the standard spares allocation, where each item on the spares listhas a 90% confidence level of availability at the end of the mission. Curves A and B, however,use the marginal utility method. B produces an aggregate system availability of 90% at lowercost; A produces an aggregate system availability of 99.1% at cost roughly equal to thestandard method's.

tory that provides an average missionspares availability of 0.991. Approximately270 inventory line items are involved. Anavailability of nearly 1.0 can be observedafter 10 days, decreasing to less than 0.80after 90 days. This is interpreted to meanthat the probability of sufficient spares tosupport system operation is about 80%after a 90 -day mission.

Curve B illustrates the effect of establishinga 90% average mission spares availabilityfor the same system. The spares investmentis cut in half, at the cost of a very low sparesavailability at mission's end, about 22%.Even at the mission's half -way point, sparesavailability is only 75%. Even this inferiorprotection, however, is better than whatcan be provided with the conventional 90%confidence level (spares availability) at themission end for each line item, representedby curve C. Curve C also has a cost nearlyas great as curve A, but far fewer items.

Table I compares conventionally derivedand optimized inventories for continualreplenishment. Note that the sparesavailability of the conventionally derivedinventory averages significantly better than

the 90% criterion. This is because each itemmust better the criterion, and only integernumbers of spares can be added. Whenthese numbers are small, their effects uponavailability can change it by several ordersof magnitude. Nevertheless, aggregateavailability is relatively small. In fact, noline -item availability criterion can providethe same protection as an aggregatecriterion with optimized allocation.

The model has been used for redundancyoptimization in the design of two com-munications systems. The first applicationwas to a seaborne communications receivercomprised of four major assemblies. Themodel allocated standby redundant unitsto meet an inherent availability require-ment and also allocated remote spares tomeet an operational availability limit. Bothallocations were achieved on the samecomputer run.

In the second system, redundancyoptimization was applied to a com-munications grid of more than 100

transmitters. The model allocated activelyredundant units to meet a high level ofinherent availability.

70

Dates and Deadlines

Upcoming meetings

Ed. Note: Meetings are listed chronological-ly. Listed after the meeting title (in bold type)are the sponsor(s), the location, and theperson to contact for more information.

FEB 15-17, 1978 - Intl. Solid State CircuitConf. (IEEE, U. of Penn.) Hilton, SanFrancisco, CA Prog Info: Mark R. Barber,Bell Labs., 600 Mountain Ave., Murray Hill,NJ 07974

MAR 21-28, 1978 - Industrial Applicationsof Microprocessors (IEEE) Sheraton,Philadelphia, PA Prog Info: W.W. Koepsel,Dept. of E.E., Seaton Hall, Kansas StateUniv., Manhattan, KS 66505

MAR 22-24, 1978 - Vehicular TechnologyConf. (IEEE) Regency Hotel, Denver, COProg Info: John J. Tary, U.S. Dept. ofCommerce, OT/ITS, 325 Broadway,Boulder, CO 80302

APR 3-7, 1978 - Design Engineering Conf.and Show McCormick Place, Chicago, ILProg Info: Tech. Affairs Dept., ASME, 345East 47th Street, United Engrg. Ctr., NewYork, NY 10017

APR 4-6, 1978 - Private ElectronicSwitching Systems Intl. (IEEE, IEE) IEE,London, England Prog Info: IEE Conf. Dept.,Savoy Place, London WC2R OBI, England.

APR 4-7, 1978 - Communications '78 (Intl.Exposition of Communications Equipment& Systems) Birmingham, England Prog Info:Exhibition Director, Tony Davies Com-munications, c/o Industrial & Trade FairsLtd., Radcliffe House, Blenheim Court,Solihull, West Midlands B91 2BG, England

APR 10-12, 1978 - Acoustics, Speech andSignal Processing (IEEE) Camelot Inn,Tulsa, OK Prog Info: Rao Yarlagadda,School of Elect. Engineering, OklahomaState Univ., Stillwater, OK 74074

APR 24-26, 1978 - Electronic Components(IEEE, EIA) Disneyland, Anaheim, CA ProgInfo: John Powers, Jr., IBM Corp. Hdqtrs.,Dept. 836 IB, 43 Old Orchard Rd., Armonk,NY 10504

MAY 6-11, 1978 -American Ceramic Soc.80th Annual Mtg. & Expo. (ACS) Cobo Hall,Detroit, MI Prog Info: Frank P. Reid, Exec.Director, The American Ceramic Society,Inc., 65 Ceramic Drive, Columbus, OH43214

MAY 10-12, 1978 - Conf. on SoftwareEngrg. (IEEE, NBS) Hyatt Regency Hotel,Atlanta, GA Prog Info: Harry Hayman, Conf.on Software Engrg., PO Box 639, SilverSpring, MD 20901

MAY 16-18, 1978 - NAECON (NationalAerospace & Electronics Conf.) (IEEE)Dayton Convention Ctr., Dayton, OH ProgInfo: NAECON, 140 E. Mounument Ave.,Dayton, OH 45402

MAY 17-19, 1978 - Circuits & Systems Intl.Symp. (IEEE) Roosevelt Hotel, New York,NY Prog Info: H E. Meadows, Dept. of Elec.Engineering & Computer Science,Columbia Univ., New York, NY 10027

MAY 23-25, 1978 - Electro/78 (IEEE)Boston -Sheraton, Hynes Auditorium,Boston, MA Prog Info: W.C. Weber, Jr., IEEEElectro, 31 Charming St., Newton, MA 02158

JUN 4-7, 1978 -Intl. Conf. on Com-munications (IEEE, et al) Sheraton Hotel,Toronto, Ont. Prog Info: F.J. Heath, PowerSystem Operation Dept., Ontario hydroElectric Power System, 700 Univ. Ave.,Toronto, Ont.

JUN 5-8, 1978-Natl. Computer Conf.(AFIPS, IEEE, ACM) Anaheim Cony. Ctr.,Disneyland Hotel Comp., Anaheim, CA ProgInfo: Stephen Miller: SRI Intl., ISE Division,Menlo Park, CA 94025

JUN 5-8, 1978 -13th Photovoltaic Spec.Conf. (IEEE) Shoreham Americana Hotel,Washington, DC Prog Info: John Goldsmith,M/S 169/422, JPL, 4800 Oak Grove Dr.,Pasadena, CA 91103

JUN 13-15, 1978 -Power ElectronicsSpecialist Conf. (IEEE) Syracuse, NY ProgInfo: F.B. Goldin, Mail Drop 30, GeneralElectric, Genessee Street, Auburn, NY13021

JUN 21-23, 1978 -Machine Processing ofRemotely Sensed Data (IEEE) WestLafayette, IN Prcg Info: D. Morrison, Pi. rd ueUniv. LARS, '220 Potter Drive, WestLafayette, IN 47906

JUN 26-28, 1978 -Design AutomationSymp. (IEEE) Las Vegas, NV Prog Info: POB639, Silver Sprirg, MD 20901

JUN 26-29, 1978-Conf. on Precision Elec-tromagnetic Measure (IEEE, NBS,URSI/USNC) Conf. Ctr., Ottawa, Ont. ProgInfo: Dr. Andrew F. Dunn, Natl. ResearchCouncil, Montreal Road, Ottawa, Ont.

JUN 27-29, 1978 -Intl. Microwave Symp.(IEEE, et al) Chateau Laurier, Ottawa, Ont.Prog Info: A.L. VanKoughnett, Com-munications Research Ctr., POB 11490,Station H, Ottawa, Ont. K2H 8S2

JUL 18-21, 1978 -Nuclear & Space Radia-tion Effects Conf. (IEEE, et al) Univ. of NewMexico, Albuquerque, NM

Calls for papers

Ed. Note: Calls are listed chronologically bymeeting date. Listed after the meeting (inbold type) are the sponsor(s), the location,and deadline information for submittals.

JUN 20-22. 1978 - Pulse Power ModulatorSymp (IEEE eta!) Statler Hilton, Buffalo, NYDeadline Info: (ab) 3/10/78 to LeonardKlein, Palisades Institute for ResearchServices, Inc., 201 Varick St., New York, NY10014

SEP 12-14, 1978 - AUTOTESTCON(Automatic Support Systems for AdvancedMaintainabi' ity) (IEEE) San Diego, CADeadline Info: 4/15/78 to Bob Aguais,General Dynamics, Electronic Div., M.S. 7-98, PO Box 81127, San Diego, CA 92138

OCT 1-5, 1978 - Industry ApplicationsSociety Conf. (IEEE) Royal York Hotel,Toronto, Ort. Deadline Info: 3/7/78 to HarryPrevey, 4141 Yonge Street, Willowdale, Ont.M2P 1N6

OCT 9-11, ' 978 - Semiconductor LaserConf. (6th) (IEEE) Hyatt Regency Hotel, SanFrancisco, CA Deadline Info: 6/15/78 to T.L.Paoli, Bell Laboratories, 600 Mountain Ave.,Murray Hill, NJ 07974

OCT 11-12, 1978 - 3rd Specialist Conf. onTech. of Electroluminescent Diodes (IEEE)Hyatt Rege-cy Hotel, San Francisco, CADeadline Into: 6/15/78 to R.N. Bhargava,Phillips Lab. Briarcliff Manor, NY 10510

NOV 7-9, 1978 PLANS '78 (Position Loca-tion & Navigation Symp.) (IEEE) San Diego,CA Deadline Info: 5/15/78 to Nelson Har-nois, Cubic PO Box 80787, San Diego, CA92138

DEC 4-6, 1978 - Natl. TelecommunicationsConf. (IEEE; Hyatt Hotel, Birmingham, ALDeadline Info: 5/78 to H.T. Uthlaut, Jr.,South Central Bell, PO Box 771,Birmingham. AL 35201

71

Patents

Astro-ElectronicsL. Muhltelderl G.E. Schmidt, Jr.Closed loop roll/control system forsatellites -4062509

Advanced TechnologyLaboratoriesW.A. BorgeseApparatus for measuring a dimension of anobject -4063820 (assigned to U.S. govern-ment)

Automated SystemsR.F. Croce' G.T. BurtonHolographic high resolution contactprinter -4043653 (assigned to U.S Govern-ment)

T.J. DudziakCylinder firing indicator -4055079(assigned to U.S. government)

H. HondaIn -line coax-to-waveguide transistor usingdipole -4011566 (assigned to U.S. govern-ment)

N. Hovagimyani J.M. LinkElastic buffer for serial data -4056851

E.M. Sutphin, Jr.Testing compression in engines from startermotor current waveform -4062232(assigned to U.S. government)

Avionics SystemsC.A. Clark, Jr.CRT display with truncated rho -thetapresentation -4059785

Broadcast SystemsL.L. Oursler, Jr.Radio frequency pulse width amplitudemodulation system -4063199

D.M. Schneider' L.J. BazinConstant pulse width sync regenerator -4064541

H.G. Seer, Jr.Apparatus for automatic color balancing ofcolor television signals -4064529

Consumer ElectronicsB.W. Beyers, Jr.Voltage storage circuit useful in televisionreceiver control applications -4065681

L.W. Nero' R.E. FernslerHorizontal deflection circuil with timingcorrection -4063133

LaboratoriesE.L. Allen, Jr.' H. KawamotoBias circuit for avalanche diodes -4058776

V.S. Banff S.L. GilbertApparatus for chemical vapor deposition -4062318

D.E. CarlsonSemiconductor device having a body ofamorphous silicon -4064521

D.E. Carlson' C.E. TracyDeposition of transparent amorphous car-bon films -4060660

J.M. Cartwright, Jr.Current mirror amplifiers with program-mable current gains -4064506

R. DestephanisStylus arm lifting/lowering apparatus for avideo disc player system -4059277

A.G. DingwallIntegrated circuit device including both N -channel and P -channel insulated gate fieldeffect transistors -4063274

R.E. EnstromStep graded photocathode -4053920(assigned to U.S. government)

M.T. GaleBlack -and -white diffractive subtractive lightfilter -4062628

A. Goldman P. DattaMethod for making etch -resistant stencilwith dichromate -sensitized caseincoating -4061529

A.G. IrpilJ.C. SaraceSimultaneous fabrication of CMOStransistors and bipolar device -4050965(assigned to U.S. government)

K. KnopSimplified and improved diffractive sub-tractive color filtering technique -4057326

A.R. MarcantonioPriority vector interrupt system -4056847

D.D. MawhinneyInjection -locked voltage controlledoscillators -4063188 (assigned to U.S.government)

K. Miyatanil I. SatoWater photolysis apparatus -4061555

L.S. OnyshkevychTransducer arrangement for a surfaceacoustic wave device to inhibit the genera-tion of multiple reflection signals -4060833

W. PhillipsLight modulation employing single crystaloptical waveguides of niobium -dopedlithium tantalate-4056304 (assigned to U.S.Government)

H.L. Pinch' H.I. MossComposite sputtering method -4060471

J.J. RiskoSheet metal waveguide horn antenna -4058813

T.F. Rosenkranz' R.J. HimicsPhotoresist containing a thiodipropionatecompound -4059449

D.L. Rossi L.A. BartonElectron beam recording media containing4,4-bis(3-diazo-3,4-dihydro-4-oxo-1-naph-thalene-sulfonyloxy) benzil-4065306

E.S. Sabiskyl C.H. AndersonGaussmeter-4063158

R.W. Smith' A.R. MooreMethod of filling apertures with crystallinematerial -4059707

R.G. StewartMemory cell and array -4063225

R.G. StewartCurrent mirror amplifier augmentation ofregulator transistor current flow -4061962

W.C. StewartVideodisc playback system -4065786

J.L. Vossen, Jr.Method of depositing low stress hafniumthin films -4056457

C.F. Wheatley, Jr.Current amplifiers -4057763

M.H. WoodsP+ silicon integrated circuit interconnectionlines -4057824 (assigned to U.S. Govern-ment)

C.T. WuMultiply -divide unit -4065666

72

Mobile Communications SelectaVision ProjectD.D. Harbert! R.G. FerrieReceived signal selecting system withpriority control -4057761

D.D. Harbert' G.R. KamererSignal quality evaluator -4063033

Missile andSurface RadarO.M. Woodwardi M.S. SiukolaLow cost linear/circularlyantenna -4062019

Patent OperationsA.L. LimbergCurrent sensing circuit -4057743

Picture Tube Division

polarized

R.H. HughesCathode support structure for color picturetube guns to equalize cutoff relation duringwarm -up -4063128

J.I. Nubanil W.R. RyszMethod of assembling a mask -panelassembly of a shadow -mask cathode-raytube -4058875

F.M. SohnColor picture tube having mask -frameassembly with reduced thickness -4056755

RCA Ltd., EnglandB. CrowleOver -current prevention circuitrytransistor amplifiers -4058775

B. CrowleCurrent regulating circuits -4063149

RCA Ltd., Canada

for

P. FoldesAntenna system with automatic depolariza-tion correction -4060808

A.L. BakerSync responsive systems for video discplayers -4057826

J.B. HalterWideband electromechanicalsystem -4060831

recording

Service CompanyE.L. Crosby, JrApparatus and method for measuringpermeability -4064740

Solid State DivisionA.A. AhmedReference potential generators -4058760

A.A. AhmedSemiconductor circuits for generatingreference potentials with predictabletemperature coefficients -4059793

A.A. AhmedVoltage standard based on semiconductorjunction offset potentials -4061959

A.F. ArnoldMethod of makirg gold -cobalt contact forsilicon devices -4065588

R.R. BrooksLight flasher circuit including GTO-4058751

R.D. FaulknerPhototube having domed mesh with non-uniform apertures -4060747

K.H. GooenSemiconductor thyristor devices havingbreakover protection -4063277

H. Khajezadehl S.C. AhrensControllably valued resistor -4057894

A.J. LeidichAmplifier circuit -4064506

J.M. NeilsonGate turn off semiconductor rectifiers -4062032

O.H. Schade, Jr.Differential amplifier -4060770

H.A. Stern' H.C. Schindler' H. SorkinMethod of filling dynamic scattering ikluldcrystal devices -4064919

73

Directory of RCA licensed professional engineersThis directory was updated in December 1977 by polling our Editorial Representativesthroughout RCA. We are publishing it for two reasons: first, to recognize those RCAengineers who have made the extra effort to receive their licenses; and second, to helpRCA engineers who need recommendations for obtaining new licenses. Omissionsmay exist in this list-have your license recognized in the RCA Engineer by writingyour name, division, location, and license number to Editor, RCA Engineer, Bldg. 204-2, Cherry Hill, N.J.

Advanced TechnologyLaboratory

Ammon, G.J.Calabria, J.A.Fedorka, R.T.Heagarty, W.F.Levene,

Litwak, A.A.Meeker, W.F.Scott, P.C.Scott, R.D.Siryj, B.W.Zadell, H.J.

Alascom

NJ -20515NJ -20061NJ -14248PA -17726ENJ -18348PA -6257ENJ -19913NY -23710EnglandNJ -18571NJ -13855ENJ -22374

Hall, S.A. AK -3915EHallett, E.R. AK -3215S, 1329E

CA -S1059, C7666GA -9883UT -1258, 2387

Hansen, P.G. AK -3710EHazel, W.M. AK -4334EKramer, R.J. AK -3127-E,

AZ -7547Talbot, G.A. AK -4069-E

UT -3788Weld, R.A. AK -4269-E

CA -4691 EIL -30688IN -14289

Americom

DeBaylo, P. CA-QU2047Derkach, L. NY -47798Ekeland, K.W. CA -7095Inglis, A.F. DC -168Kelly, L. NY -40769Lundgren, D. NJ -19440Mostafa, M. BrazilSolomon, S.M. NY -13053

NY -38663Walsh, J.M. NJ -24142

NY -24590

Astro Electronics

Aievoli, D.Bacher, J.Berko, L.Darcy, J.

MA -24941NJ -18719NJ -18983NJ -18908PA -10696E

Ganssle, E.R.Goldberg, E.A.Goldsmith, A.Hartshorne, F.Herrmann, J.Martz, A.F.McClanahan, J.M.

Nekrasov, P.

Strother, J.Welch, P.J.

NY -36323NJ -21027CA-QU-2379PA -2022ENJ -20608NJ -20257LA -5820ENJ -20815CA -E-9391NJ -19298NJ -17638PA -12493

Automated Systems

Anderson, J.M.Brazet, M.D.Fischer, H.L.Floridia, M.S.Frawley, P.T.Furnstahl, J.S.

Galton, E.B.Gibson, P.F.Gorman, R.K.Harrison, J.S.Herzlinger, J.I.Laschever, N.L.Meliones, N.O'Connell, J.H.Palm, K.E.Perra, S.S.Plaisted, R.C.Richter, E.W.Ross, G.T.Seeley, P.E.Shirak, F.R.Toscano, P.M.Wamsley, N.B.Woll, H.J.Zerfas, G.J.

MA -21285MA -25738MA -12791MA -24908MA -21429NJ -11029MA -25315MA -19751PA -8210MA -20701MA -21663NJ -6095OH -E17327MA -15536MA -24855MA -20720MA -25063MA -25720MA -16238NJ -11562MA -26873MA -20125MA -20714MA -20985MA -22261MA -20967

Avionics Systems

Davis, W.J.Lucchi, G.A.Vose, A.W.

TX -13332CA -E5662CA -E4844

Broadcast Systems

Sepich, W.S.Wentworth, J.W.Wolf, R.E.Wright, C.H.

NJ -14239NJ -10342NJ -12568NJ -11877

CommercialCommunicationsSystems Div. Staff

Daroff, S.ZDodd, J.A.

PA -14769NJ -12988

Consumer Electronics

Beyers, B.W.Cochran, L.A.Crick, R.W.George, J.3.Holt, F.R.Pollack, R.H.Riedwig, E W.

Secor, R.E.Wood, J.C.

KS -6005IN -14208IN -14209IN -14177EnglandPA -6191EIN -6470OH -19434IN -13342IN -12083

Corporate Staff

Lawson, K.D.

Ridley. P.SRoloff, E.A.

FL -10203OH-ME23182OH -20904MO -E10262NJ -17760

Distributor and SpecialProducts

Sterner, J.F.Stonaker, W.M.

Globcom

Correard, LP.Ginters, F.Hoffman, LMartin, J.Stackhouse, D.R.

NJ -15008NJ -11275

NY -41632NY -47848NY -39507NY -38519CA-, HI-

Electro Optics andDevices

Clayton, R.W. NJ -14841 Adams, B.B. PA -9293EFrank, D.J.H. NJ -22147 Byram, R.E. PA -10707EHymas, D.G. NJ-EE9830 Cox, R.W. PA-ME6212EMusson, C.H. NJ -11606 Eshelman, J.A. PA -21434Putnam, R.S. NJ -8839 Fanale, J.R. PA -79E

Fleckenstein, E.D. NJ -10197EForman, J.M. PA -11347Gote, J.T. PA -9748EHamilton, G.A. PA -7435EHammersand, F.G. PA -10163EHelvy, F.A. PA -11131EHensley, J.W.Jasinski, J.P.Licht, H.W.McDonie, A.F.Nekut, A.G.Nierenberg, M.J.Palmquist, D.W.Paul. W.H.Romero, E.L.

Seidman, N.Shannon, A.W.Smith, C.P.Tomcavage, J.R.Trout, D.R.Villanyi, S.T.Walton, H.B.Werntz, H.J.

IN -10912PA -21435EPA -7627PA -10693Pa -10276PA -12291EPA -7925EPA -10304MA -19979PA -10301EPA -10014EPA -13209PA -9714EPA -21428EPA -21431ECanadaPA -5063EPA -21436E

GovernmentCommunicationsSystemsAllen, R.W.Ames, M.E.Anzalone, P.Black, A.L.

Brill, H.A.Bucher, T.T.Buck, R.A.Chapman, H.H.Comninos, D.A.Daigle, E.J. Jr.Henter, C.Herman, S.H.Houck, R.D.Jellinek, E.Jobbe, I.Jones, A.G.Kaufman. J.W.

Knoll. J.Kozak, M.J.Livingston, H.N.Mack, A.Magasiny, I.P.McCauley, E.S.Meer, M.E.

Merson, L.N.Nahay, L.P.Nasto, S.N.Nossen, E.J.Parker. D.J.Risse, R.A.Rostrom, R.W.Sass, E.J.

NJ -7826PA -8695ENJ -11837NJ -13820PA -10368ENJ -24489NJ -14341NJ -13521OH-EE2079NJ -10857NJ -16322NJ -14241PA -1855ENJ -22536NY -22650NJ -17347PA -9735ENJ -16069PA -4572EPA -10430NJ -12050PA -1505ENJ -20598PA -10422EPA -11153ENJ -12191NY -36986NJ -12497NY -35259NJ -13493NJ -20601NJ -21110NY -37050NJ -19578NJ -17217

Sokolov, H. CT -3938PA -3507E

Steinberg, N. NJ -16531Tannenbaum, D.A. NJ -20154Thompson, A.C. NJ -18272Ubben, R.C. MD -5366Vallette, C.N. PA -2811EVallorani, A.A. PA -14955Wezner, F.S. NJ -20115

Governments SystemsDivision Staff

Gallager, J.B.Groman, E.M.Grossman, H.B.Hobson, J.F.Hopper, A.G.

Howery, R.W.Johnson, M.C.Karch, M.A.Kay, F.T.Kooperstein, R.Korsen, M.

NJ -14525NJ -11557PA -11840PA -2115EMA -21322NY -41722PA -1540ENJ -9518NJ -11645NJ -21862NJ -11053NY -3382NJ -16593

Mobile CommunicationsSystems

Bullock, J.B. NJ -8639Hanway, W.F. PA -8956ENeidlinger, J.R. OH -E40301Risko, A.J. PA -7334ESeymour, J.W. AZ -8340

NJ -14609PA -12479E

Springer, C.J. PA -61992EStewig, W.G. PA -17736-E

Records

Chang, B.J. IN -16479Devarajan, A. IN -15114Fuller, R.E. IN -16697Martin, M.K. IN -16685Mattson, G.A. PA -7155ENelson, G. IN -15343

NJ -19613Weaver, C.A. IN -17205

Research and Engineering

Coben, P.W. PA -9004E Kruger, I.D. NJ -6262 NBC Clark, J.F. NJ -24517

Hilibrand, J.Keller, M.A.Shore, D.Wilkinson, D.A.

NY -37383NJ -15007NJ -16176DC -2865

Lesser, D.Levi, P.Liston, J.Lurcott, E.G.

CA -0447NY -37438PA -12087ENJ -14641

Meany, Jr., M.H. NY -46233Polak, H.L. NY -42746Siebert, J.L. (Cons) CA-4732EE

Jenny, H.K.Lauffer, W.D.

SelectaVision

NJ -19775DE -5243

Wright, P.E. NJ -17027 Lyndon, D.L.Magun, J.

CA -9164NJ -13002

Patent Operations Liddle, S.W. IN -7671

McCord, M. OH -E20813 DeCamillis, M. MI -12131 Solid State DivisionInternational Licensing Melody, J.V. NJ -17589 NJ -11692

Metzger, G.V. NJ -14170 Emanuel, P.M. MA -25622 Ahmed, A.H. NJ -19596Romero, E.L. PA -10301E PA -01866 Bartlett, S.P. PA -26487

MA -19979 Moll, A.P. NJ -19338 Picture Tube Division Baugher, D.M. NJ -17921Shotliff, L.A. NY -40390 Molz, K.F. NJ -16906 Bennet, W.P. NJ -18910

MD -3001 Alleman, R.A. PA -2750E PA -9249ELaboratories Moncher, R. NY -35248 Blust, H.L. NJ - Blattner, D.J. NJ -16183

Haldane, H.R.Hellman, H.I.Lile, W.R.Mackey, D.

NJ -15158NY -22392ENJ -17467NJ -13627

Nessler, T.G.Nessmith, J.T.O'Brien, J.F.Paglee, M.R.

PA -009158ENJ -13187NJ -10285PA -4879ENJ -17679

Bumke, J.R.Chemelewski, D.Class, J.S.Clutter, L.M.D'Augustine, F.T.

I N-7550PA -17727EPA -6583EIN -11632PA -93'2E

Campbell, L.R.DiMassimo, D.V.DiMauro, J.

Greenberg, L.S.

NJ -12958NJ -14264NJ -A9989PA -18056MA -11061

Mawhinney, D.D. NJ -13246 Patterson, P. Ontario Davis, J.H. PA- Gubitose, N.F. PA -8511Rosen, A. Canada Petri, W.H. PA -8986E Dymock, L.P. PA -7417E Jetter, E. NY -36913Schell, R.E. NJ -13629 NJ -24156 Gadbois, G.S. PA -10538E Keller, J.P. NJ -19400Zollers, S.M. NJ -13695 Pschunder, R.J. NJ -13141 Goldberger, R.S. NJ -7943 Lindsley, C. NJ -13355

Missile and Surface RadarRaciti, S.A.Ray, P.

NY -39124FL -4922 Gruber, L.L.

PA -1696EPA -10316E

Meisel, H.R.Mendelson, R.

NJ -20307NY -27175

Abbott, S.L.Adams, F.G.

PA -10084NJ -13504

Robinson, A.S.NJ -14403NJ -14026NY -29222

Hall, L.B.Handel, R.R.

IN -7084PA -9317E

Moyer, J.H.Nash, T.E.

PA -8784EPA -6951

Aron, S.Bachinsky, R.D.

Baker, H.F.Baron, A.S.Beadle, P.R.

PA -9007ENJ -15447PA -11162ENJ -14872NJ -8113MD -3183Britain

Rogers, Jr., A.Rogers, G.J.Rubin, M.Russell, H.Scarpulla, G.Schnorr, D.P.

PA -01343NJ -15073NY -30763MA -21504NY -30772PA -12703E

Henderson, W.G.Hensel, V.B.Hughes, R.H.Hummer, A.P.Kimbrough, L.B.Konrad, R.J.Kuzminski, H.W.

PA -5323EPA-PA -12678IL -17848PA -19026IN -

PA -8736E

Puotinen, D.A.Scaran, E.D.Waas, G.J.Waltke, H.C.Williams, C.C. Jr.

Wilson, R.L.

NJ -14276PA -26505ENY -31494NJ -14483IN -14914OH -14667NJ -12968

Beckett, W.Berkowitz, H.

NJ -23495NY -23101

Scott, E.N.Scott, I.

NJ -8853PA-ME14398

Loser, T.C.Maddox, W.J.

PA -9246EPA -11173E

Bernard, W. NJ -8878 Scull, Jr., W.E. NJ -15283 Mengle, L.I. PA -3778EBoertzel, R.L. NJ -14398 Senior, G.A. NJ -9330 Miknis, W.D. PA-Bogner, B.F. NJ -19875 Simonetti, J.A. PA -15380E

NJ -18826 Ottos, J.G. PA -Breese, M.E.

Brown, H.W.Burke, T.J.Caplan, L.A.

NY -39719OH -E22759NJ -14672NJ -16533VA -00039

Smith, J.N.Stadnyk, P.J.Strip, J.Tillwick, F.H.

MA -20965NJ -10497NJ -14083NJ -12856OH -E22556

Pederson, W.E.Porath, A.C.Price, D.O.Royce, M.R.Savleter, R.E.

PA -10017EIN -11030PA-ME6203EPA -10005EI N-7373

Clarke, T.L. NJ -10274 Urkowitz, H. NJ -21564 Scearce, K.D. IN -10967Copestakes, J.E. NJ -13471 Volpe, J.C. NJ -22449 Swope, H.H PA -21426EDaut, Jr., S.E. DE -15569Y Weiss, H.R. PA -14637E Weingarten, M.R. PA-ME01206DeFelice, R.F. NJ -21673 MA -9920 Wolverton, P.W. IN -5548Dorman, M.L. NJ -24379 Weiss, M. PA -4796E

Duffin, J.D.WA -8949NJ -8653

Wells, W.D.Wick, R.H.

NJ -13185NJ -17490

RCA Ltd., England

Eble, F.A. DC-3664Str PA -18637E Pickering, J MIEEPA -4161E Widmann, F.W. NJ -8131

Felheimer, C. AL -1617 Wilsher, R.A. Britain RCA Service Co.Field, G.R. NJ -13595 CanadaFreedman, D.D. NJ -11641 Yanis, E.M. NJ -13476 Cox, H.C. FL -10778

Galbiati, L.J. OH -23069 VA -8947 Dombrosky, R.M. NJ -14495

75

Two RCA engineers elected IEEE Fellows

4.

The membership grade of Fellow is the highest attaintainable in theInstitute of Electrical and Electronic Engineers. The IEEE annuallyrecognizes as Fellows those members who have made outstandingcontributions to the field of electronics.

Anthony H. Lind

"for technical leadership in the design and product development of video taperecorders and color television cameras."

Tony Lind has been with Broadcast Systems since 1946, and has been anEngineering Manager since 1951. His engineering organizations have designedand developed many broadcast studio products-cameras, tv tape recorders,projectors, switchers, and terminal equipment. He has participated in manyindustry technical committees and is presently on the Board of Governors of theSMPTE.

Walter W. Weinstock

_utions to radar systems and for leadership in development of modernair defense systems."

Walt Weinstock is a Senior Staff Scientist at Missile and Surface Radar; he hasbeen with RCA since 1949. Most of his work has been in systems engineering anddefinition of air defense systems, such as BMEWS, ASMS, and AEGIS. His earlierwork in extending the area of radar target modeling brought him internationalrecognition through the "Weinstock Cases," now cited in a number of books onradar.

Henderson namedOutstanding Young Electrical Engineer of 1977Each year, Eta Kappa Nu, thenational electrical engineer-ing honor society, recognizesa young engineer who has"outstanding professionalachievements, civic andsocial activities, and culturalpursuits." Nominees must beno more than 35 years old,with a BSEE degree, orequivalent, held no more than10 years.

John G.N. Henderson

"for original contributions to the ad-vancement of television technologyand for his participation in civic andcultural affairs."

John Henderson has been with RCALaboratories since receiving his BSEEin 1967. His work has included projectsin i.f. filter design procedures, elec-tronic tuner control systems, and sur-face acoustic wave filters. He is also aninstructor in RCA's Minorities inEngineering Program, working withhigh-school students from minoritygroups to help them receivesupplemental technical education andexperience as preparation forengineering careers.

76

Good companyRCA's two new IEEE fellows join a select group-starting with DavidSarnoff, RCA has produced 147 Fellows over the years. The listincludes some famous names in RCA's past as well as prominent onesin RCA's present.

Name Awarded

Sarnoff, David 1917 Rays, Henry E. 1955 Jenny, H.K. 1966Batsel, Max C. 1927 Sinnett, Chester M 1955 Bradburd, E.M. 1967Beverage, Harold H. 1928 Weimer, Paul K. 1955 Mesner. M.H 1967Jolliffe, Charles B. 1930 Barton, Loy E. 1956 Muller, J.H. 1967Wilmotte, Raymond M. 1938 Glover, Alan M. 1956 Hernqvist, K.G. 1968Zworykin, Vladimir K. 1938 Korman. N.I 1956 Powers, K.H. 1969Engstrom, Elmer W. 1940 Leverenz, H.W. 1956 Sterzer, F. 1968Brown, George H. 1941 McElrath, G. 1956 Rappaport, P. 1969Hanson, O.B. 1941 Poch, Waldemar J. 1956 Robinson. A S. 1969Peterson, Harold 0. 1941 Spitzer, Edwin E. 1956 Woodward, J.G. 1969Wolff, Irving G. 1942 Tolson, W.A. 1956 Becken, E.D. 1970North, Dwight 0. 1943 Ankenbrandt, F.L. 1957 Gluyas, T.M. 1970Hansel!, Clarence W. 1945 Avins, Jack 1957 Jones, Loren F. 1970Beers, George L. 1947 Barco, Allen A. 1957 Parker, D.J. 1970Kell, Ray D. 1947 Flory, Leslie E. 1957 Shahbender. R.A. 1970Coleman, John B. 1948 Hillier, James 1957 Smith, C.P. 1970Herold, E.W. 1948 Schairer, Otto S. 1957 Vollmer, J. 1970Lindenblad, Nils E. 1948 Thompson, Leland E. 1957 Lohman, R.D. 1971Rose, Albert 1948 Tuska, Clarence D. 1957 Simon, R.E. 1971Carlson, Wendell L. 1949 Pan, Wen Y. 1958 Heilmeier, G.H. 1972Olson, Harry F. 1949 Peter, R.W. 1958 Kreuzer, Barton 1972Bedford, Alda V. 1950 Van Duesen, G.L. 1958 Sobol, Haro.d 1972Hirsch, Charles J. 1951 Brustman, J.A. 1959 Urkowitz, Harry 1972Knox, James B. 1951 Mueller, Charles W. 1959 Hershenov, Bernard 1973Landon, Vernon D. 1951 Simpson, LeRoy C. 1959 Kressel, Henry 1973Luck, David G.C. 1951 Curtiss, Arthur N. 1960 Luther, Arch C., Jr. 1973Morton, George A. 1951 Johnson, Edward 0. 1960 Behrend, William L. 1974Schade, Otto H. 1951 Johnson, Harwick 1960 Belohoubek, Erwin F. 1974Schmit, Dominic F. 1951 Moore, John B. 1960 Zaininger, Karl H. 1974Siling, Philip F. 1951 Rau, David S. 1960 Kosonocky, Walter F. 1975Epstein, David W. 1951 Smith, Philip T. 1960 Sherman, Samuel M. 1975Holmes, Ralph S. 1952 Sommer, Alfred H. 1960 Bachynski, Morrel P. 1976Law, Harold B. 1952 Spencer, Roy C. 1960 Lechner, Bernard J. 1976Nergaard, Leon S. 1952 Webster, Wm. M. 1960 Nessmith, Josh T., Jr. 1976Headrick, Lewis B. 1953 Nicoll, Frederick H. 1961 Winder, Robert 0. 1976Laport, Edmund A. 1953 Smith, Theodore A. 1961 Lind, Anthony H. 1977Rajchman, Jan A. 1953 Wickizer, Gilbert S. 1961 Weinstock, Walter W. 1977Trevor, Bertran A. 1953 Kihn, Harry 1962Young, Charles J. 1953 Kozanowski, Henry N. 1962Anderson, Earl I. 1954 Sonnenfeldt, R.W. 1962Bond, Donald S. 1954 Woll, Harry J. 1962Byrnes, Irving F. 1954 Cimorelli, J.T. 1963Corrington, Murlan S. 1954 Metzger, S. 1963Ewing, Douglas H. 1954 Kirkwood, L.R. 1964Gunther, Clarence A. 1954 Leyton, E. 1964Koch, Winfield R. 1954 Morrison, W.C. 1964Schrader, Harold J. 1954 Wallmark, J.T. 1964Schroeder, Alfred C. 1954 Pankove, J. 1965Shaw, George R. 1954 Forgue, S.V. 1965Speakman, Edwin A. 1954 Guenther, R. 1965Fredendall, Gordon L. 1955 Isom, W.R. 1965Harris, William A. 1955 Mason, W. 1965Janes, Robert B. 1955 Seelen, H.R. 1965Linder, Ernest G. 1955 Herzog, G.B. 1966Ramberg, Edward G. 1955 Wege, H.R. 1966

77

Engineering News and Highlights

New Technical Publications Administrators and RCA Engineer Representatives

Francis Holt at Consumer Electronics

Francis Holt is the Ed Rep at ConsumerElectronics in Indianapolis, Ind., where hereports to Clyde Hoyt, TPA and Manager ofProduct Safety and External TechnicalRelations. Francis has been with RCA fornineteen years and was previously the EdRep for the SelectaVision Project where hewas also a Senior Member of the Engineer-ing Staff. Prior to his work in SelectaVision,he was in Advanced Development at Con-sumer Electronics.

Maucie Miller at RCA Laboratories

Maucie Miller takes over as TPA andEditorial Representative for RCALaboratories, replacing Chester W. Sall whoretired. Maucie was previously tie TPA andEd Rep for RCA Americom at Piscataway,N.J. Throughout his career at RCA, Mauciehas been involved in technic& com-munications.

Bob Moore at SelectaVision

Robert Moore is the Ed Rep for the Selecta-Vision Project at Rockville Road, Ind.,replacing Francis Holt. Dr. Moore has beenwith RCA for twelve years with assignmentsat RCA Laboratories, Corporate Productand Market Planning, and SelectaVision. Hecurrently reports to the Staff Vice President,SelectaVision Videodisc Operations in In-dianapolis.

Promotions

Solid State DivisionRobert VanAsselt from Member TechnicalStaff to Leader Technical Staff.

Robert Nestel from Member Technical Staffto Leader Technical Staff.

Picture Tube DivisionDon M. Trobaugh from Member, TechnicalStaff, Manufacturing to Manager, TubeProcessing, Quality and ReliabilityAssurance (Marion Plant).

Astro-ElectronicsYvonne C. Brill from Senior Engineer toManager, (Spec.) Engineering.

William V. Fuldner from Staff SystemsScientist to Manager, (Spec.) Engineering.

Missile and Surface Radar

R. Morgan from Senior Member, Engineer-ing Staff, to Unit Manager, D&D Engineers.

N. Salzberg from Principal Member,Engineering Staff, to Unit Manager,Systems Engineering.

Staff Announcements

Solid State DivisionBernard V. Vonderschmitt, Vice Presidentand General Manager, appointed Edward M.Troy, Division Vice President, Solid StatePower Devices, and Carl R. Turner, DivisionVice President, Solid State Integrated Cir-cuits. In this newly established position, Mr.Turner will be responsible for MOS andBipolar Integrated Circuits, High -SpeedBipolar Integrated Circuits, and Offshoreand Integrated Circuits ManufacturingOperations.

Carl R. Turner, Division Vice President,Integrated Circuits, announced theorganization of Integrated Circuits asfollows: Marvin B. Alexander, ProjectManager, Business Systems; StanleyRosenberg, Director, High Speed Bipolar ICOperations, Richard L. Sanquini, Director,Bipolar IC Operations; Philip R. Thomas,Division Vice President, IC ManufacturingOperations; and Carl R. Turner, (Acting)Division Vice President, MOS IC Products.

Fred G. Block, Manager, Central Engineer-ing, appointed William B. Hall Manager,Manufacturing Systems Engineering.

Picture Tube DivisionJoseph H. Colgrove, Division Vice Presidentand General Manager, appointed CharlesW. Thierfelder Division Vice President,Product Safety, Quality and Reliability.

Charles W. Thierfelder, Division Vice Presi-dent, Product Safety, Quality and Reliabili-ty, appointed Wellesley J. Dodds Director,

78

Technical Publications Administrators (TPAs) are responsible for reviewing and approvingtechnical papers; for coordinating the technical reporting program; and for promoting thepreparation of technical papers and presentations.

Editorial Representatives (Ed Reps) assist authors by stimulating, planning, andcoordinating appropriate papers for the RCA Engineer. They also keep the editors informedof new developments as well as professional activities, awards, publications, andpromotions in their areas.

J.R. Reece at Picture Tube Division,Marion

J. R. Reece is the new Ed Rep for the PictureTube Division Plant at Marion, Ind. Mr.Reece has worked at RCA for nineteen yearsand is presently a Leader, Technical Staff, inthe Applications, Reliability, and SafetyLaboratory with responsibility for facilita-tion and maintenance. He also handlesapplications liaison with the ConsumerElectronics plant in Bloomington.

Murray Rosenthal at Americom

Murray Rosenthal becomes the TPA and EdRep at Americom in Piscataway, N.J. replac-ing Maucie Mille. Murray has been withRCA for eighteen years in administrativeand technical publications activities. He isnow Manager for administration of theTechnical Operations Department atAmericom.

Chet Sabi retires

Chester W. Sall, Technical PublicationsAdmiristrator, RCA Laboratories, retiredafter 35 years with RCA, most of that timewith RCA Laboratories. Chet has been anactive and prolific RCA Engineer EditorialRepresentative over the past twenty yearsand has also been a frequent contributor tothe pages of the journal.

Chet is a Senior Member of IEEE, a chartermember of the Group on Professional Com-munications, and a past chairman of thatgroup For the past twenty years, he has alsoserved as Editor of the IEEE Transactions onConsumer Electronics.

Kleinberg ends four-yearterm on ANSI Board

Quality and Reliability AssuranceOperations Analysis.

Richard H. Hynicka, Director, Mask andMount Operations and Lancaster Manufac-turing, appointed Richard L. SpaldingManager. Production Engineering, and JackJ. Spencer Manager. Quality and ReliabilityAssurance-Lancaster.

Clifford E. Shedd, Manager, EquipmentDevelopment, announced that the activity ofEquipment Engineering-PTC istransferred to the staff of the Manager,Equipment Development. Keith D. Scearceis appointed Manager, EquipmentEngineering-PTC, and will report to theManager, Equipment Development.

Mobile CommunicationsSystems

Lee F. Crowley, Manager, Engineering andTechnical Services, appointed Albert R.Allen, Manager, Technical Services.

Advanced TechnologyLaboratoriesFred E. Shashoue, Director, AdvancedTechnology Laboratories, has appointedHarold E. Haynes, Manager, Programs andPlanning, for ATL.

CommercialCommunicationsSystems Division

Irving K. Kessler, Group Vice President,announced that the Commercial Com-munications Systems Division will assumeresponsibility for the Electronic IndustrialEngineering organization. Henry Duszak,General Manager, Electronic IndustrialEngineering, will report to Neil VanderDussen, Division Vice President andGeneral Manager, Commercial Com-munications Systems Division.

Harry Kleinberg (right) RCA's Manager ofCorporate Standards Engineering, acceptsa certificate of appreciation from John W.Landis, President of the American NationalStandards Institute at the Board of Directorsmeeting in New York. Mr. Kleinberg servedtwo successive two-year terms as a Directoron the ANSI Board.

79

Awards

Four technical excellence award winners at Moorestown

Hatim Bharmal -for specialcontributions in the programdesign of the AEGIS ORTS DataBase Generation System,particularly for his pioneeringuse of special structured pro-gramming development tech-niques.

Andrew Moll-for definition anddesign of combined hardware -software configuration for a

complex signal/data processingsystem, applying cost-effectivemini- and micro -computertechniques.

Murray Rubin-for technicaland task -team leadership in thedesign and construction of theCombat System EngineeringDevelopment (CSED) facility.

First and second prizes awarded in COSMAC contest

John Kowalchik (second from left) of Solid State Division, Moun-taintop, Pa.. accepts his COSMAC Evaluation Kit andMicroterminal from Don Carley, Manager of Custom SystemDesign at Solid State Division. John was the first -prize winner in theCOSMAC Applications Contest for his COSMAC-based autopatchcontrol for amateur repeaters. Also in the photo are KeithLoofbourrow (left) Mountaintop Engineering Leader and JohnPhillips (right) Editor of the RCA Engineer.

Arthur Simons-for outstandingachievements in defining dataprocessing algorithms and man -machine interfaces, andtranslating complex systemoperational concepts intospecific computer software re-quirements.

Torr Lenihan (center) of RCA Laboratories, Princeton, N.J.,accepts his COSMAC Evaluation Kit from Don Carley, Manager ofCustom Systems Design at Solid State Division. Tom was thesecond -prize winner for his A/D-based burglar alarm system (seeRCA Engineer. Vol. 23, No. 3, p. 72). Also in the photo are AngeloMarcantcnio (left) and Paul Russo (second from right) of RCALaboratories and John Phillips (right), Editor of the RCA Engineer.

xu

Editorial RepresentativesContact your Editorial Representative, at the extensions listed here, to scheduletechnical papers and announce your professional activities.

Commercial CommunicationsSystems DivisionBroadcast SystemsBILL SEPICH Camden, N.J. Ext. PC -2156KRISHNA PRABA Gibbsboro, N.J. Ext. PC -3605ANDREW BILLIE Meadow Lands, Pa. Ext. 6231

Mobile Communications SystemsFRED BARTON' Meadow Lands, Pa. Ext. 6428

Avionics SystemsSTEWART METCHETTE Van Nuys, Cal. Ext. 3806JOHN McDONOUGH Van Nuys, Cal. Ext. 3353

3overnmt. Systems DivisionAstro-ElectronicsED GOLDBERG' Hightstown, N.J. Ext. 2544

Automated SystemsKEN PALM' Burlington. Mass. Ext. 3797AL SKAVICUS Burlington. Mass. Ext. 2582LARRY SMITH Burlington, Mass. Ext. 2010

Government Communications SystemsDAN TANNENBAUM Camden. N.J. Ext. PC -5410HARRY KETCHAM Camden, N.J. Ext. PC -3913

Government EngineeringMERLE PIETZ' Camden, N.J. Ext. PC -5857

Missile and Surface RadarDON HIGGS' Moorestown, N.J. Ext. PM -2836JACK FRIEDMAN Moorestown, N.J. Ext. PM -2112

Solid State DivisionJOHN SCHOEN Somerville. N.J. Ext. 6467

Power DevicesHAROLD RONAN Mountaintop. Pa. Ext. 633SY SILVERSTEIN Somerville, N.J Ext. 6168

Integrated CircuitsFRED FOERSTER Somerville, N.J. Ext. 7452JOHN YOUNG Findlay, Ohio Ext. 307

Electro-Optics and DevicesRALPH ENGSTROM Lancaster. Pa. Ext. 2503

Cc urns ElectronicsCLYDE HOYT'Indianapolis, Ind. Ext. VH-2462RON BUTH Indianapolis, Ind. Ext. VH-4393PAUL CROOKSHANKS Indianapolis, Ind. Ext. VH-2849

SetectaVision Project

RCA Service Company.

JOE STEOGER' Cherry Hill. N.J. Ext. PY-5547RAY MacWILLIAMS Cherry Hill, N.J. Ext. PY-5986DICK DOMBROSKY Cherry Hill, N.J. PY-4414

Distributor andSpecial Products DivisionCHARLES REARICK Deptford, N.J.Ext. PT -513

Picrurf:, rube DivfsmDn

ED MADENFORD Lancaster. Pa. Ext. 3657NICK MEENA Circleville, Oh o Ext. 228JACK NUBANI Scranton, Pa Ext. 499J.R. REECE Marion, Ind. Ext. 566

At.,scorn

PETE WEST' Anchorage, Alaska Ext. 0611

Ailiericc)rnMAUCIE MILLER' Kingsbridge Campus, N.J. Ext. 4122

Glop c.)m

WALT LEIS' New York. N.Y. Ext. 3089

RCA Records

JOSEPH WELLS' Indianapolis, Ind. Ext. VT -5507

NBC

BILL HOWARD' New York, N.Y. Ext. 4385

Patent OperationsJOSEPH TRIPOLI Princeton, N.J Ext. 2491

Electronic Industrial EngineeringJOHN OVNICK' N. Hollywood. Cal. Ext. 241

Research and EngiieeringCorporate EngineeringHANS JENNY' Cherry Hill, N.J.Ext. PY-4251

LaboratoriesCHET SALL' Princeton, N.J. Ext. 2321LESLIE ADAMS Somerville, N.J. Ext. 7357

'Technical Publications Administrator, responsible forFRANCIS HOLT Indianapolis, Ind. Ext. VR-3235 review and approval of papers and presentations.

alMa Engineer

A technical journal published by Corporate Technical Communications"by and for the RCA Engineer"

Panted in USA Form No RE -23-4