IteilE

ngineerV

ol 24 No. 1

JunlJul1978

23rd Anniversary Issue

Sapphire ribbon for S

OS

Our cover: Competitively priced silicon -on -sapphire circuits have been a goal of the solid-state industry for years. Now. RCA hasdeveloped a process for making sapphire ribbon,a major step toward that goal. Our cover photowas taken on the first day of ribbon production atthe Solid State Division's Mountaintop, Pa. plant.

The aluminum oxide crystals at the upper left-hand corner are the starting point for theprocess, and the ribbon running diagonallyacross the cover is a typical unpolished output.Rich Novak, (RCA Laboratories. Princeton),seated by the console, was responsible for thetechnology transfer of the puller concept fromRCA Laboratories. Nick Gubitose (SSD, Moun-taintop), who was responsible for the design ofthe ribbon puller, is handing down a sampleribbon to Ural Roundtree (SSD, Somerville), awafer user-he was responsible for developingthe epitaxial reactor that will be used in process-ing the SOS wafers. Read more about theadvantages of SOS technology in BernieVonderschmitt's article starting on page 4.

Photo credit: Tom Cook, RCA Laboratories,Princeton, N J

[EcBa EngineerA echnical journal published byRCA Research and EngineeringBldg. 204-2Cherry Hill, N.J. 08101Tel. 222-4254 (609-338-4254)Indexed annually in the Apr/May issue.

John Phillips Editor

Bill Lauffer Associate Editor

Joan Toothill Art Editor

Frank Strobl Contributing EditorBetty Gutchigian Composition

Joyce Davis Editorial Secretary

Harry Anderson

Jay Brandinge

John Christopher

Bill Hartzell

Jim Hepburn

Hans Jenny

Arch Luther

Howie Rosenthal

Carl Turner

Joe Volpe

Bill Underwoo,1

Bill Webster

Ed Burke

Walt Dennen

Charlie Foster

Div. VP, Mfg. OperationsConsumer Electronics Div.Div. VP, Engineering.Consumer Electronics Div.VP. Tech. Operations,RCA AmericomDiv. VP. EngineeringPicture Tube DivisionVP and Technical Director.RCA GlobcomManager, TechnicalInformation ProgramsChief Engineer, CommercialCommunications Systems Div.Staff VP, EngineeringDiv. VP. Integrated CircuitsSolid State DivisionChief Engineer,Missile and Surface RadarDirector, EngineeringProfessional Programs

VP, Laboratories

s_.onsulting Editors

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 anncunce outstanding and unusual achievements of RCAengineers in a manner most likely to enhance their prestige andprofessional status

The life cycle-it's a fact of industrial life

Today's RCA management is determined to ensure our corporate vitality by participating in electronic-based industries in their growth and development cycles.

Every successful product has a life cycle-development, followed by growth, maturity, and thendecline. During development, it is often true that expenses are high, income is low, and the productloses money. The picture reverses during the growth period, where income rises sharply and variablecosts drop as we gain production experience. Then comes maturity. Sound management, engineering,marketing, and manufacturing programs to improve the product can maintain or enhance profitabilityfor many, many years. It is income from such products that seed other products during their growthand development periods. Eventually, however, most products become obsolete, revenues drop, andthe decline is underway.

Industries too are bound by this cyclical pattern. Railroads are often used as an example of an industrythat has gone through development and maturity, and is now declining. We can apply this life -cyclemodel generally to any industry-if we use a narrow definition of "industry" (e.g., transportationobviously doesn't fit).

It follows, then, that if an industrial organization ties its future to a single product-or even a single(narrowly defined) industry-it is on its way to an inescapable decline.

These facts of industrial life are well understood at RCA. Our management recognizes the need toadequately support current mature businesses to maintain their vitality and profitability as long aspossible. And our management is determined to take the steps necessary to ensure growth andcorporate vitality by developing innovative new electronic products that will, in some cases, spawnnew industries. In both cases, management recognizes that, in an electronic -based company like RCA,a highly skilled and competent technical staff is the primary force that will carry our new products andindustries through development, growth, and product improvement to successful maturity.

I am confident that our engineering community will respond to lead this continued development andgrowth.

Howard RosenthalStaff Vice PresidentEngineeringPrinceton, N.J.

PHOTOCATHODE

LIGHT /DYNODES

Of fiNPHOTOELECTRONS

ANODE

I(

SIGNAL

23rd anniversary issue

SOS technology

SOS circuits used tc be "advantageous, but expensive,"but costs are dropping substantially.

Energy

Is this barrel half empty or half full? What's going toreplace the missing oil?

Photomultipliers

Starting as a very small research effort, they've becomea large and diverse business.

- RCA undersea

What are 700 RCA employees doing at Andros Island inthe Bahamas"

coming up

Our next issue (AugzSepj covers RCA space technology-weather, com-munications, and navigation Waite', and even amateur radio In space.

Future issues will cover RCA at the Missile Test Project, manufacturing, andenergy.

B.V. Vonderschmitt 4

B.F. Williams 10

R.W. Engstrom 18

R.F. Kenville 27

E.J. Schmitt 32

E. EricksonIR. Kennedy 40G. Virgin

W.S. SepichjN.C. Lund 46

R.W. Howery 52

J.H. Kalish 59

M.V. Hoover 62

N. Brooks 65

Staff

W.J. Underwood

76

7778

79

Copyright © 1978 RCA CorporationAll rights reserved

llama EngineerVol. 24INo. 1 JuniJul 1978

general interest articlesSilicon -on -sapphire (SOS): an LSI/VLSI technology

Energy-a review and survey

Photomultipliers-then and now

Optical video disc for very large digital memories

Government -related programs in the Solid State Division

AUTEC-technology at the tip of the tongueof the ocean

A computerized drawing control systemby and for the engineer

Developing and managing large computer programs

VLSI-tools, technology, and trends

Modern IC applications in communications systems

on the job/off the jobThe lure of electronic tishing

engineer and the corporationMandatory licensing of engineers: pros and cons

RCA's position on licensing engineers

departmentsDates and Deadlines

Pen and Podium

Patents

News and Highlights

Silicon -on -sapphire (SOS): an LSI/VLSI technology

B.V.Vonderschmitt

A cost-effective manufacturing techniquefor placing an epitaxial layer of silicon on apassive substrate has been the goal ofsemiconductor and, particularly, ICtechnologists for a number of years. Thereason for this interest is the potential thatisolated -device elements offer from thestandpoint of circuit performance, flex-ibility of design, and density. There isnearly uniform agreement among thoseengaged in research related to ICtechnology and product development thata system that virtually eliminates parasiticcomponents has significant merit. The

Bernie Vonderschmitt, Vice President andGeneral Manager of the Solid State Division,is responsible for the engineering, manufac-turing, and marketing of all the Division'sproducts.Contact him at:Solid State DivisionSomerville, N.J.Ext. 7051

Bernie Vonderschmitt holding a sapphireribbon and a substrate.

A new process for making sapphire wafers should bring thecost of SOS circuits down to a level competitive with bulksilicon circuits.

silicon -on -sapphire (SOS) technologymeets all of these requirements, but has notcaught the interest of IC manufacturers forone main reason-the cost of the sapphiresubstrate, currently six to eight times thatof bulk silicon material.

But the outlook is changing. Developmen-tal effort over the past five years hasbrought significant progress in theprocesses used in manufacturing sapphiresubstrates. Analysis indicates that, withinthe next several years, sapphire -substratecosts will be within 25 to 35 percent of bulksilicon wafer costs. Analysis further in-dicates that the increased densities andhigher yields resulting from the use of apassive substrate (SOS) rather than asemiconductor material substrate (bulksilicon) will result in system costs that arenearly equal. In some applications, SOScosts will be lower than those of alter-native, existing technologies. This reducedcost could well be the ingredient requiredto provide the impetus for an increasedeffort on the part of the IC industry todevelop the SOS technology into the mostsignificant new large-scale integration(LSI) technology of the future.

SOS advantagesThe basic difference between an SOScomponent and a standard bulk -siliconcomponent is that in the SOS componentthe active devices are constructed within asilicon island that is electrically isolated byeither oxide or sapphire from other suchislands within the same IC. Once theislands have been defined, standard silicongate processing is used to construct theactive devices. Fig. I is a simplified illustra-tion of the initial SOS processing steps.

The silicon -island concept in itself in-creases the component's density potentialthrough the elimination of the guard bandsor p -n junctions conventionally used forisolation. Silicon islands also increaseyields because a "killer defect," which

would normally cause component failureon a semiconductor material substrate, canoccur between silicon islands, and yet haveno effect on an SOS IC. For an extensivelist of SOS advantages, see the box at theright.

1-

2-

3-

4 -

Substrate

Sapphire

0.6 Micron Silicon Layer

Sapphire

Etch Silicon Islands

Implant p 8 n Wells

5- Standard Silicon GateProcessing

Channel Oxide Polysilicon Gate Source - Drain Implants Field Oxide - Reflow Aluminum Metal

Fig. 1

Electronically isolated silicon "islands"represent the basic difference between SOScomponents and standard bulk -siliconcomponents. In this simplified illustration,the nonconducting sapphire substrate (step1) is coated with a thin layer of silicon (step2). The islands are then etched out (step 3),and p and n wells are added (step 4) in

preparation for standard silicon gateprocessing (step 5). The result is a verydensely packed integrated circuit withoutthe parasitic devices obtained with bulksilicon processing.

4

SOS history

The development of SOS as a viabletechnology began in the mid -to -late 1960s.

RCA's effort began more than ten yearsago at RCA Laboratories in Princeton,N.J. Other companies, such as Rockwell,Westinghouse, General Electric, andHughes, have had, and still have, activeprograms in SOS aimed primarily atspecialty applications and focused on theadvantages of SOS, particularly inradiation -hardened circuits. Inselek, aventure-capital company that discontinuedoperations nearly three years ago,attempted to develop and manufactureSOS devices for the commercial market.The company could not make SOS com-petitive with bulk complementary MOS atthe SSI level, an area where the fundamen-tal advantages of SOS offered little or nocost -performance improvements.

Two U.S. companies are currently visiblyactive in SOS for commercial applications:Hewlett-Packard, an original equipmentmanufacturer of instrumentation and com-puters, and RCA, an IC supplier.

Hewlett-Packard's interest, according topublished information, lies in the perfor-mance and application advantages thataccrue to the end -equipment design.RCA's interest is in exploiting SOS com-mercially, entering it into competitiondirectly against other MOS technologies inboth performance and, ultimately, cost.

Many companies have demonstrated thesignificant technological and performanceadvantages that SOS has over bulkCMOS, PMOS, and NMOS, particularlyin static devices. However, the manufac-turing costs of SOS devices have remainedhigh, primarily because of the highsapphire substrate cost. To make it com-mercially viable, an innovation in SOSsubstrate manufacture was required; thegradual improvement of the technologywith time could never surpass the ex-perience or "learning curve" thatCzochralski silicon wafers have undergoneover the last fifteen years.

Although certainly not universallybelieved, there is general agreement that ifSOS could achieve cost/ price parity withother technologies, the fundamental at-tributes of SOS would make it an attractiveLSI / VLSI technology. It is, therefore,important to examine the cost fundamen-tals of sapphire substrates.

Why design in SOS?

SOS electrical isolation provides fundamental advantages in design andapplication development which, for the first time, free the circuit,product, and system engineer from limitations dictated by alternativetechnologies. Some of these advantages are:

1) Field inversion problems that can create extraneous parasiticdevices cannot occur in the SOS technology because the field region isnot a semiconductor material. Therefore, the designer can ignore thepresence of parasitics that his design may inadvertently include, andneed not expend development time in protecting his design from suchpossibilities.

2) The designer cannot encounter latch -up problems caused byparasitic SCR devices. An SOS component design is not susceptible tothe spike voltages that trigger parasitic devices.

3) The absence of parasitic devices enhances the reliability of SOSdevices.

4) Power dissipation, particularly static power, is low enough to relievethe device and system designer of the concern for the power -downtechniques currently required, for example, in many memoryapplications.

5) Once a proved -in set of s:andard "building block" circuits has beendeveloped, the potential for a successful first -pass design issignificantly improved. If the logic is correctly simulated and the circuit"blocks' are properly interconnected, the device is not subject tounexpected parasitic effects.

6) IC debugging is greatly simplified because SOS circuitry is com-pletely static and has no parasitic effects.

7) Operation over a wide temperature range is assured withoutsignificant changes in either device characteristics or power dissipa-tion.

8) The system designer can take greater liberties with his power -supply design since an order of magnitude less power is required.Complementary SOS also allows the system voltage supply to varyover a wider range, and requires less regulation circuitry.

9) The complementary SOS technology has high noise immunity,which reduces the amount of circuitry required to protect data linesfrom ripples and spikes.

Ribbon -pulled substrates:the cost breakthrough

There are two fundamental approaches tothe manufacture of sapphire substrates.

One approach is the familiar Czochralskiprocess used for fabricating bulk siliconwafers, except that high -purity A1203 is

1

used to grow sapphire crystals. Thisprocess results in a crystal three inches indiameter that is then sliced into individualwafers. The considerable cost and loss ofmaterial (approximately 45 percent)associated with the slicing operation fun-damental to this approach provided thestimulus for an alternate process forsapphire manufacture. The apparent

5

RCA's ribbon -growing process

RCA is using a new shaped crystal growth technique to producelow-cost sapphire substrates for CMOS/SOS processing. Thetechnique is called edge -defined film -fed growth, or EFG, and iscurrently used to grow three -inch -wide ribbons, three at a time,as shown in the diagram below. Grossly simplified, the processconsists of melting sapphire (aluminum oxide) crystals, con-tacting the melt with a seed crystal, and slowly pulling theresulting shaped crystal through a special die.

The die must be made of material that the melt will wet but notreact with too severely; either tungsten or molybdenum willfulfill this condition for an aluminum -oxide melt. A capillary iscreated by closely parallel spacing the two plates of the die. Thealuminum -oxide melt will climb to the top of the capillary if thetheoretical maximum capillary height is not exceeded, which istypically the case for all practical growth conditions. Surfacetension will hold the melt to a properly oriented seed crystaldipped into the capillary. Slow withdrawal of the seed will lift themelt onto the top surfaces of the die with the melt spreading tothe external edges of the die. The melt is constrained fromrunning over this external edge by surface -tension forces. Anisothermal temperature distribution is maintained across the dietop coupled to a very steep, vertical gradient. This will crystallizethe melt immediately above the die before surface -tensionforces can relax the shaped melt, thus producing a shapedsingle crystal. -Rich Novak, RCA Laboratories

SAPPHIRERI BBON

Ribbon is slowly pulled out ofdie. Surface tension keepsthe melt from running overthe edges. Crystallizationtakes place immediatelyabove the die.

AI.2 03MENISCUS

Ribbon -pulling machine. Note glow from heatedcrucible containing melted sapphire at bottom. Rib-bon is emerging at top of picture.

The product-three ribbonsat a time-will be used as thesubstrate for silicon -on -sapphire (SOS) integratedcircuits.

6

answer is a proprietary process called edge -defined film -fed growth (EFG), which in-volves pulling a continuous "ribbon" ofsapphire from a shaping die thatdetermines the ribbon's thickness andwidth.

Fig. 2 illustrates the EFG manufacturingtechnique. Capillary action pulls themolten Al203 up through the shaping die,where a seed crystal is brought into contactwith it and pulled to form the sapphireribbon. It is possible to pull more than asingle ribbon from the same crucible simp-ly by adding additional shaping dies andpullers, thereby increasing manufacturingvolume significantly and leading tobroader amoritization of equipment andlabor expenses. Growing the substrate inthe form of a ribbon eliminates the need tosaw the crystal into individual wafers andso saves labor and reduces material lossesor eliminates the expense of materialreprocessing for recycling.

EFG can be applied to SOS substratemanufacturing today.

However, additional development will berequired before it can be applied to silicon

Crystal Ribbon

Heat Shields

Shaping Die

Water CooledQuartz Envelope

RefractoryMetal Crucible

R F Coils

Fig. 2Ribbon -pulling technique for the edge -defined film -fed growth (EFG) process. RFcoils heat the sapphire into a molten state,and capillary action pulls it up through theshaping die. There, it contacts a seed crystaland is slowly pulled up to form the sapphireribbon. Wafers made in this manner do notrequire the expensive and material -wastingslicing operation otherwise required.

substrates. Because the sapphire substrateis only used as a mechanical support forsilicon islands, its crystal orientation doesnot need to be as exact as that required forthe silicon substrate, which is eventuallyused for the construction of active devices.

Once a ribbon has been pulled, it is scribedinto square wafers, as shown in Fig. 3.Currently, three-inch ribbons are pulled,and the resulting three -inch -square wafersare converted to mechanical replicas of therounded bulk -silicon wafers to permit theuse of existing wafer -handling fixtures.However, this method is only a temporaryexpedient; eventually the square wafer willbe processed directly, and manufacturingoutput will be increased even furtherbecause of the increased wafer area.

The rounded wafers are then polishedbefore the 0.6 -micron epitaxial siliconlayer is grown. Ultimately, the roundingwill be eliminated and, possibly, thepolishing. Continuing development hasdemonstrated the feasibility of growing theepitaxial layer directly on the scribedsquare wafers. Elimination of roundingand polishing will, of course, furtherreduce labor costs and material losses.

RF Heated

SOS substrate costs

Ribbon -pulled sapphire costs less thanstandard round sapphire wafers.

A comparison of the cost per square inch ofvarious sapphire substrates with a polishedsilicon substrate is plotted in Fig. 4. Costsfor Czochralski silicon and sapphire wereobtained from various U.S. suppliers; thecosts for sapphire ribbon were developedbased on the EFG manufacturing methodswithin RCA. All costs are stated on anormalized basis (with appropriatemarkups) and include the epitaxiallygrown silicon layer on the sapphire wafer.Based on information from suppliers, littleif any cost benefits can be realized fromincreases in silicon wafer volume; in-flationary labor and material costs willtend to offset any benefits derived fromthese increases. As a result, the cost/ priceper unit area of wafer is not expected tochange significantly.

The basic ribbon substrate represents lessthan half of the total cost of a givencomponent. Since a large percentage of theremaining cost is for surface preparation, asignificant cost improvement potential ex -

Edge -Defined Film -Fed Growth - Silicon On Sapphire

SapphireRibbon

MolyCrucible& Die

Pull Crystal

0.6 MicronEpitaxial Silicon

3 Square0.040 Thi9

(

Scribe

Epitaxial SiliconGrowth

ScribeCorners

DiamondGrind O.D.

% Circularize Wafer i

iI I

iI 0 0

1:?000Q0

%10h) QC'

U____y I( fi

Diamond Grind/Polish.

No.

Fig. 3Once a ribbon is pulled, it is scribed into square wafers. The silicon layer can then be grownon the wafer, but presently a number of steps (inside dashed line) are needed to make waferscompatible with existing wafer -handling equipment. Ultimately, these steps will beeliminated to further reduce materials losses and labor costs.

7

ists with any surface -condition improve-ment that would allow sapphire ribbons tobe processed "as grown", i.e., no roundingor polishing. This potential, plus thereduced cost of pulling multiple ribbons,has not been included in Fig. 4.

The future cost of a sapphire wafer isestimated to be 30 to 40 percent greater thanthat of a bulk silicon wafer.

This figure is for cumulative volumesshown in Fig. 4. The benefit of EFGmanufacturing of sapphire substrates isquantified by the difference between thecost of the three-inch round Czochralskisapphire wafer and the three-inch ribbon.The cost benefits of eliminating roundingand polishing are represented by the four -inch -square ribbon curve. Because of thenormal manufacturing learning curve andthe breakthrough in substrate manufac-turing represented by the EFG process, thecost of a sapphire wafer will begin toapproach that of a silicon wafer at relative-ly low accumulated volumes.

Die size for SOS is about half that of CMOS,and should get smaller.

An analysis has been made of the relativedie size and die cost for a CMOS eight -bitmicroprocessor and for a 16k static RAMusing various technologies. The followingground rules were used:

I) For existing devices, the die sizesrepresented the range of those availablein the marketplace.

10

9

a

7

5

4

3

2

1

0

3" Round Sapphire(Czochralski)

Sapphire Ribbon3" Square

Sapphire Ribbon4" Square

Polished Silicon3" 8 4" Round

Normalized Die Size Normalized Die Cost

Fig. 5Comparison of CMOS die sizes and costs, based on 8 -bit microprocessors. SOS die size ispresently about 50% of bulk -silicon die size, but should shrink to about 35%. Die cost is stillhigh at present, but will drop as ribbon process produces wafers in the future.

2) For scaled -up or extrapolated devices,critical dimensions were responsive toevolutionary improvements expected tooccur with time for each technologyanalyzed. The smallest critical dimen-sion assumed on any device is 2 microns.

3) The device costs were determinedfrom the basic material, labor, anddirectly associated expenses. The yields

Normalized to 1978 SiliconCost/Sq. In.. 1.0

100,000 300,000 1,000,000

Cumulative Volume - Sapphire Wafers

Fig. 4Cost comparison shows the advantage sapphire ribbon has over conventionally grownround sapphire. Eventually, sapphire wafers should cost only 30 to 40% more than siliconwafers. (This estimate does not include the possibilities of improved surface conditions ormultiple ribbon -pulling.)

were determined from a consideration ofmask levels and the potential for a defectto be fatal.

4) The probability of defects per unitarea were assumed to be the same for alltechnologies. Different defect rates wereassumed with time but were equitablyapplied to all technologies.

5) Material costs for sapphire comparedto bulk silicon were taken from Fig. 4.

For an eight -bit microprocessor, such asthe RCA CDP1802, the die size for thecurrent SOS design is slightly greater than50 percent of the bulk silicon size and, withsome process enhancements and tighterdesign rules (3.5 -micron critical dimen-sion), approximately 35 percent of bulksilicon size. Currently, the die cost for SOSis slightly greater than that of bulk silicon,primarily the result of the currentCzochralski sapphire cost. But this costwill diminish to 25 to 35 percent of the bulksilicon cost in the future, as illustrated inFig. 5.

SOS should eventually be cost -competitivewith NMOS in terms of normalized die costs.

The economic impact of the EFG process isdramatically illustrated in a plot of thenormalized die size and cost for 16k staticRAMs fabricated by means of varioustechnologies. Although not economicallyfeasible, four existing devices were scaled

8

Normalizes Die Size Normalized Die Cost

'With ReducedSapphire Substrate Cost

Fig. 6Comparison of die sizes and costs for competing technologies. (See text for description of scaling and normalizing processused.) SOS should eventually be cost -competitive with NMOS, but SOS will also have a number of system advantages.

up to 16k RAM products for the purposeof comparison; that is, their die size andcosts were extrapolated to reflect the sametechnology design capability and manufac-

fabricated. To complete the analysis, thefuture capabilities of NMOS and SOStechnologies, the two technologies capableof producing an economically feasible 16kRAM, were also compared. The itemsscaled up include the CMOS 5047, thecurrent NMOS design rules, and thecurrent- and enhanced -process SOS 4kdesign rules. The "NMOS Future" labelassumes an enhanced "new process"currently in development that uses a 3.5 -micron critical dimension. In determiningthe normalized die cost, sapphire sub-strates were assumed to cost 30 percentmore than bulk silicon substrates.

As illustrated in Fig. 6, the impact of SOSand EFG substrates on the "SOS Future"normalized die costs places the "SOSFuture" process in a competitive positionrelative to NMOS, even at the componentlevel. Other system advantages discussedearlier provide additional end -equipmentsavings. With the availability of a viablesolution to the fundamental sapphire -substrate cost problem, additional equip-ment manufacturers will begin to recognizethe advantages that accrue to their systemdesigns with the use of SOS, such asHewlett-Packard has already demon-strated. It is certain, therefore, that, with

time, other semiconductor manufacturerswill also demonstrate a marked accelera-tion of interest in SOS.

Market sizeAs noted above, SOS is a technology for LSIand VLSI.

The dynamic -memory market for largememory systems will, as a result of momen-tum and cost considerations, remain in theprovince of n -channel technology for theforeseeable future. The high-speed market(less than five nanoseconds propagationdelay) is the province of ECL and similartechnologies. However, much of the rest ofthe systems requirements could, from aperformance standpoint, be served bySOS.

The market size for digital signal -processing applications, excluding ECLand dynamic memories, is shown for theyear 1977, and projected for 1980, in thetable below:

1977

SM

1980

SM

Total MOS 1315 2100

Total bipolar(excluding ECL) 985 1200

2300 3300

Potential for SOS 1800 2500

The "potential" shown is that portion ofthe market that is identifiable as beingtechnically and economically served bySOS. Penetration into this commercialmarket willover the next few years as the costreductions discussed above are realized. Inaddition to the cost reductions in the SOStechnology, primarily resulting from theEFG process, the technical advantages ofSOS complementary static devices, alreadyrecognized in military and customapplications, will further expand theircommercial markets.

ConclusionIt now appears that SOS has anidentifiable, practical solution-the EFGprocess-to the last major obstacle block-ing its widespread commercial application.As a result, the SOS technology will nolonger be limited to only those applicationsthat can afford to pay a premium for itstechnological advantages, but will findbroad application in industrial, consumer,automotive, and other price -sensitive,commercial markets.

AcknowledgmentsNorman C. Turner, Eugene M. Reiss, andGerald K. Beckmann, all of the Solid StateDivis:on, helped prepare this article.

Reprint RE -24-1-8Final manuscript received April 12. 1978.

9

Energy a review and survey

B.F. Williams

Because the cost of energy has increased,the traditional techniques for energy useand management are no longer optimum.Further, uncommon sources of energy,from liquid fuels produced from coal toheat and electricity produced fromsunlight, have come much closer toeconomic viability because of the increasedcost of traditional fuels. What constitutesan appropriate RCA response to this newsituation depends on one's view of its long-term seriousness and on RCA's capabilitiesto address the problems raised. The situa-tion is serious and has its roots in world-wide industrial development, and RCAdoes have capabilities to address some ofthe problems raised. In this article, we willdiscuss the evolution of the present energysituation and describe the response of theU.S. Government (the National EnergyPlan). The specific RCA programs ad-dressing near- and long-term opportunitieswill be the subject of individual articles inthe Feb/ Mar 1979 issue of the RCAEngineer.

The world's changed energy situation has implications forRCA in terms of its existing businesses and new businessopportunities.

U.S. energysupply and demand

Fig. I shows the pattern of energy use in theU.S. for 1977.2 On the left of the figure arethe various fuel inputs. How those fuels aredistributed throughout the economy isshown by the pipes connecting the fuelsources with the end uses. As can be seenfrom the figure, in those cases where thefuels are used primarily for healing, such asin the industrial, residential, and com-mercial sectors, the efficiency of use isrelatively high. In the case of electricitygeneration and in the transportation sec-tor, the efficiency is low because of thefundamental inefficiency of the generatingplants and the gasoline engine.

We are running out of oil.

With the ready availability of low-cost,convenient oil and natural gas, essentiallyall of the growth of U.S. energy consump-tion in the last thirty years took place using

these fuels. As can be seen in Fig. 2, oil itselfgrew from 30% of the total energy use in1947 to 50% in 1977,3 while total energy usedoubled. This is an increase from 5.5million barrels of oil per day (MBD) in1947 to over 18 MBD today. At this rate ofoil consumption, the U.S. has enoughreserves for only ten years.° Including the"undiscovered recoverable resources"might give an additional seven to twentyyears worth of oil at present consumptionrates.°

Fig. 1 clearly shows that the U.S. is nowimporting half of the oil it uses. This is anew situation, as evidenced by Fig. 3. Thisrecent and explosive growth in the amountof oil we import reflects the state of ourown resources and present consumptionpatterns, and has created financial andpolitical problems. This situation isaggravated by the increase in the world'senergy consumption and the finite quantityof the world's conventional energyresources.

HYDROELECTRIC-C.

NUCLEAR

GAS

( DOMESTIC

COAL

OIL

(IMPORTS)

OIL

( DOMESTIC )

r.IMM.......101111111.11114

- ELECTRICAL

4ilee- ENERGY4.9GENERATION

a

7.0

8.6

9.5

10.6

REF. IDENTIAL

-

COANN)MERCIAL

8 3

111=111111W

7.8

EXPOR0 6

'4111111111111.-001 9 3

Nv25N

8.9

-41111kINDUSTRIAL

TRANSPOR-

TATION9.!

(UNITS: MILLION BUIS / DAY OIL EQUIVALENT)

2.3

2.6

6.9

19.4

6.5

16.07.2

2.3

LOST

ENERGY

USED

ENERGY

Fig. 1U.S. energy sources and sinks, 1977. (The format for this figure is borrowed from Ref. 1.) Numbers in the figure are in millions ofbarrels of oil per day (MBD) equivalents; e.g., for natural gas the numbers reflect the amount of oil that would give the same heatenergy as the amount of gas used. These data were taken from the January 1978 issue of the Monthy Energy Review (Ref. 2).Data for the full year of 1977 were estimated from the use through November 1977, and are accurate to about 5%.

10

40

30

111.

oaa

me 2

I0

TOTAL CONSUMPTION

NUCLEAR POWER

1975 ion

Fig. 2Total U.S. energy consumption, Ref. 3. Note that the reduction in consumptioncaused by the 973 oil embargo took almost 4 years to be eliminated.

130 -

120

110 -

100 -

90 -

00 -

TO -

60

50

40

30

20

10

0

1960

TOTAL PRODUCTION

INVENTORY CHANCES

PETROLEUM CONSUMPTION

196, 1962 1963 1964 1965 1966 1967 1960

t1969 1970 1971 1972 1973

Fig. 4World energy consumption, Ref. 3. As in the U.S., world energy use increaseshave been met by oil and natural gas.

For the past thirty years, U.S. energyconsumption has increased at a rate ofabout 3% per year, while the total worldenergy consumption has increased at a rateof 5% per year (Fig. 4). The U.S. share hasdropped from 34% in 1960 to just under30% today. The growth in the world energyuse took place using oil and natural gas, asit did in the U.S. As a result, much of thedeveloped world faces an energy situationsimilar to, or worse than, that in the U.S.Fig. 5 is a summary of who is using theworld's energy resources and who is

producing them. Present OPEC exportsare 32 MBD, practically all of the produc-tion shown in Fig. 5 labeled "rest of theworld."' Fig. 6 shows the present produc-tion capacity of the OPEC countries. Notethat because of excess capability (6 MBD),the Saudis are in a position to reduce oil -price inflation for the time being. Withinthe next several years, some analystsbelieve production growth from Alaska

20

18

16

14

12

10

88

6

4

2

NET ENERGY IMPORTSMBD OIL EQUIVALENT

o TOTAL PETROLEUM IMPORTS 1972-1977MBD

. .41-0'

-

moot0/ %if

-r-

1947 51 55 59 63 67 71 75 1977

Fig. 3Net U.S. energy imports and recent oil imports, Ref. 3.Importing such large amounts of oil is a recent phenomenon.

and the North Sea will be slowing and thatthe Soviet Union will switch from anexporter of oil to an importer.' The resultwill be a dramatic increase in demand forOPEC oil, with a concomitant increase inprices when excess OPEC capacity is ex-hausted. One estimate of the timing for thisdevelopment is shown in Fig. 7. Whether itis believed that the time for the nextdramatic increase in oil prices is three orten years away, we are running out of oil.While an increase in oil prices will makemore resources available through the use ofdifficult -to -recover oil, such increases willhave dampening economic effects anddestabilizing political effects world-wide.*

In the last two years our imports of oil have increased from 6 to9 MBD. To appreciate how difficult it is to bring on lineadditional capacity at any price, note that production from theAlaskan North Slope is presently running at 0.7 M BD, scheduledto increase to I MBD. We would have had to discover anddevelop three more Alaskan North Slopes in the last two years tostay even.

50-

40 -

30

20

1973

177 on.

EMI NAT CAS

f=1 COAL

C3 HYDRO l GEOTHERMAL

7;o

US W EUROPE

--to=JAPAN SINO- REST OF

SOVIET BLOC WORLD

Fig. 5World energy use and production, Ref. 4.The vulnerability of Japan and WesternEurope to curtailments in oil supply isapparent.

We do have alternatives.

From a world-wide energy point of view, apotential solution to this problem is a shiftto a more plentiful energy resource or to asource that is renewable. From a U.S.political point of view, a shift to a moreplentiful or renewable resources overwhich we have control is a desirablesolution. The U.S. is fortunate in that thisis possible.

11

MBD

12

ALGERIA

ECUADOR

CAION

INDONESIA

IRAN

IRAQ

MAU

III IA

NIGERIA

QATAR

SAUDI ARANIA

UN. ARAI ENIMATES

REIEZNALA

0I I I I I I 1 I I. I

2 4 6 8 10 12

Fig. 6OPEC production capacity as of March 1977, Ref. 5. The totalcapacity is about 38 MBD.

01976 77 78 79 60 81 82 83 84 85

Fig. 7OPEC oil will not supply the rest of the world forever. In thisOPEC supply and demand projection, Ref. 5, note that there isexcess capacity today, but a projected sharp rise in demandfor the immediate future. Crossover point is projected for1984, even with an increase in production capacity (which will,paradoxically, have to be made during a period of excesssupply.)

NORTHAMERICA

SOUTH

U.S.S.R.s...,R.

/AMERICA IA 1.9%1.5% \...---AFRICA

.---ASIA

...11. EUROPE

Fig. 8World's measured recoverable energy reserves, Ref. 4 SeeFig. 9 caption for a definition of reserves.

Half of the world's measured recoverable energy reserves are inNorth America, principally in the form of coal (Figs. 8 and 9). Coalmay be readily used to generate electricity, assuming that theenvironmental costs are acceptable and transportation is available.(A ton of coal has the same heat energy as 4 barrels of oil.)Eliminating oil for all uses except petrochemicals and transporta-tion, and substituting electric heating for oil heating would drop oilimports from 8.6 MBD to about 3 MBD. However, because thegeneration of electricity is only 30% efficient, and the direct use ofoil for heating is about 80% efficient, the substitution of electricityfrom coal for oil would require an increase of coal equivalent to 12MBD of oil, an increase twice as large as present coal production.More reasonable may be the direct substitution of coal for oil inheating. This requires an increase of only about 5 MBD equivalentof coal production (an 80% increase) and implies a return todomestic, commercial, and industrial coal-fired furnaces; also a

seemingly unattractive prospect.

A similar situation exists with natural gas, although not so serious,for while the supply is comparable to oil's, we use it at about half therate. Again, coal may be substituted for gas in electricity generationwith the same environmental and transportation problems, butswitching from gas to coal in domestic, commercial, and industrialheating and processing may be extremely difficult to achieve.

The reduction of our dependence on foreign energy resources forthe near term and development of renewable or essentiallyinexhaustible sources of energy for the long term is the subject ofthe National Energy Plan (NEP).6 (See the box on the next page.)Basically, the plan is designed to reduce consumption and in thenear term cause a shift to coal use. The strategy for accomplishingthese goals is to increase the price of oil and natural gas through acombination of increased price ceilings and taxes. The long-termgoals are to be obtained through the support of research programs.The NEP is presently being debated in Congress with progressimpeded by great interest in the section on oil and natural gaspricing.

What are our alternatives now?In the following discussion, the various energy sources arediscussed separately in the following categories: conservation andenergy efficiency; oil and natural gas; coal; nuclear electric energy;hydroelectric generators; and new sources of energy.

Conservation and energy efficiency"The cornerstone of the National Energy Plan is conservation, thecleanest and cheapest source of new energy supply."

Decreasing the growth rate of energy consumption can havesubstantial effects, as seen in Fig. 10. In this case, decreasing therate from 3.5 to 2.3% would save 5 MBD by 1985. As might beexpected from an inspection of Fig. 1, the transportation sectoroffers great opportunity for satisfying overall national goals, sincethe energy potentially conserved comes directly from oil. In 1975Congress enacted legislation which required that the sales -weightedaverage of new -car mileage be improved to 20 mpg in 1980 and 27.5

mpg in 1985. There has been progress in this direction, but the NEPhas further provisions to encourage this conservation. A proposedtax on new cars will penalize purchasers of cars that have mileagepoorer than that legislated in 1975 and reward purchasers of carswhich have better mileage. Further, the Plan provides for a standbygasoline tax to be imposed in the event that administration

20 U S RECOVERABLE RESERVES f, RESOURCES

7

1.5

05

03

02

=MAXIMUM

MINIMUM

a

iN ACOAL URANIUMOBITHERMAL NATURAL OIL OIL

SHALEGAS

Fig. 9Distribution of U.S. fuel supplies, Ref. 4. It isextremely important to differentiatebetween "reserves" and "resources." Whilecustom has it that these words meandifferent things for different fuels, ingeneral, reserves are fuel supplies that maybe tapped with existing technology withoutdramatic increase in cost. Resources, incontrast, are estimates of total existingsupplies.

consumption goals are not met. Removingthe excise tax on buses, increasing the taxon non-commercial aviation fuel, andeliminating the rebate of half the Federaltax on motorboat fuel are also proposed.

Tax credits for conservation investments inresidential, commercial, and industrialsituations are proposed. Reform of theutility rate structure is also proposed.Promotional rates would be eliminatedand rates reflecting costs implemented.Therefore, off-peak and interruptableservice would be encouraged.

Oil and natural gasBoth oil and natural gas are price controlledand cost less to use than their replacementcost.

In the case of gas, the intra-state prices inthe producing states are higher than theregulated interstate prices, leading to gasshortages in the nonproducing states. The

OPEC

OIL

r

80 A

70

60

50 -

40

30

20 -

10 -

I

1950 1956

SOURCE U.S. BUREAU OF MINES

HISTORY r FUTURE

2.3 %-wiGROWTH RATE,1968 -1976

3 5 cy._GROWTH RATE

1950-19'3

3.5% p.a. GROWTH

1968 1973 1976 1980

2.3 p.a.GROWTH

1990

Fig. 10Conservation can have a dramatic effect on energ? demand. Decreasingour annual energy growth rate from 3.5% to 2.3°/0 would save 5 MBD by1985.

2000

The national energy plan*"The U.S. las three overriding energy object ves. As an immediateobjective, which will become even more important in the future, the U.S.must reduce its dependence on 'oreign oil to limit its vulnerability to supplyinterruptions. In the medium term, the U.S. must weather the stringency inworld oil supply that will be caused by limitations on productive capacities.In the long term, the U.S. must have renewable andessentially inexhaustibleSources of energy for sustained economic growth. The strategy of the Plancontains three major components to achieve these objectives.

"First, by carrying out an effective conservation Drogram in all sectors ofenergy use, through reform of utility rate structures, and by making energyprices reflect true replacement costs, the nation should reduce the annualrate of growth of demand to less than 2 percent. That reduction would helpachieve both the immediate and the medium -term goals. It would reducevulnerability and prepare the nation's stock of capital goods for the :imewhen world oil production will approach capacity limitations.

"Second, industries and utilities using oil and natt. 'al gas should convert tocoal and other abundant fuels. Substitution of otter fuels for oil and gaswould reduce imports and mak a gas more widely available for householduse. An effective conversion program would thus contribute to meetingboth the immediate and the medium -term goals.

"Third, the nation should pursue a vigorous resaarch and developmentprogram to provide renewable and other resources to meet U.S. energyneeds in the next century. The Federal Government should support a varietyof energy alternatives in their early stages, and continue support throughthe development and demonstration stage to.- technologies that aretechnically, economically, and environmentally most promising."

This quotation and others appearing in this article are from the Naticnal Energy Plan. which. cont'ary towidespread belief, does exist This document (Ref.6) is available from that.. S. Government Printing Office, Stock11040-000-00380-1.

13

authors of the NEP recognize that new gasand oil are more expensive to find andproduce than old gas and oil. However,they do not feel that the U.S. producerswith oil in existing fields should benefitfrom the three -fold increase in oil pricesthat the foreign producers imposed.Presently, old oil has a price ceiling of$5.25/ barrel and new oil a price ceiling of$11.28/ barrel. The cost for an oil userdepends on the mix of old, new, andforeign oil he has available. It is proposedthat the price of new oil be allowed toincrease to the 1977 world price($13.50/ barrel) and thereafter increased toaccount for inflation. In order to curtail oiluse, it is proposed to tax old oil($5.25/ barrel) and previously discoverednew oil ($11.28/ barrel) an amount to bringthe cost to the consumer up to the worldprice. The revenues received would bedistributed back to consumers in general inthe form of an "energy payment," a creditagainst other taxes, or a general taxreform.

When the cost of oil increased, demand fornatural gas increased. Natural gas is pricecontrolled; it is now underpriced and thereis excess demand. It is proposed to increasethe price ceiling for new gas from itspresent $1.49/ 1000 ft3 so that it equals theaverage price of the equivalent amount ofall domestically produced oil, about$1.75/ 1000 ft3. Gas made available fromexisting intra-state production would alsoqualify as new gas. The new higher -pricedgas would be allocated to industrial users,since it is felt that they have the bestopportunity to convert to other fuels.Cheaper old gas would be conserved fordomestic use.

Industrial users of gas would be subjectedto a tax to bring the cost of gas to 1-1/4 timesthe world price of gas. A phased -in conser-vation tax, increasing from $0.9/ barrel to$3/ barrel, will be levied on industrial andutility oil users. The intent of these pricingstructures is described as making the con-sumer bear (something like) the replace-ment cost of the energy, while notproviding windfall profits for the energycompanies.

Coal"Expansion of U.S. coal production and useis essential if the nation is to maintaineconomic growth, reduce oil imports, andhave adequate supplies of gas for residentialuse."

The tax structures described above for oiland gas are designed to provide this shift to

coal use. Further, industry and utilitieswould be prohibited from burning oil andgas in large new boilers. Utilities burninggas will have to shift to other fuels entirelyby 1990 and would require a special permitto shift to oil instead of coal.

The environmental impact of this shift tocoal -burning facilities is the subject ofsome concern, and "the Administrationintends to achieve its energy goals withoutendangering the public health or degradingthe environment." 6 This may not be so easyto do. In areas with serious air -pollutionproblems, oil may continue to be burned.The pollution problem associated with coalburning is being addressed with researchand demonstration programs on flue -gasdesulfurization systems, fluidized -bedcombustion, coal grinding and washing,solvent -refining coal to remove sulfur, andmanufacturing low -BTU gas from coal forindustrial purposes and high -BTU gasfrom coal as a potential replacement fornatural gas. A research and demonstrationprogram to develop the technology of coal -to -oil conversion is also in progress.

While the coal is available in the ground, itis appropriate to note some of the problemsthe NEP for coal production and utiliza-tion poses.''' The Plan calls for a 90%increase in coal production by 1985, for atotal of 1.2 billion tons. This will require200 to 250 new mines, or almost three newmines opened per month; about 150 largemining shovels or drag lines, and theindustry can barely keep up with today'sdemands. The coal industry's long-rangeplans, which emphasize Western coal, callfor 80,000 new underground miners and45,000 new surface miners to be hired andtrained by 1985. The NEP intends toincrease the production of high -sulphurEastern coal because of the transportationproblem of bringing Western coal East tobe used. This proposal to expand Easternand contract Western production will in-crease the manpower numbers and add tothe capital investment because surface -mined Western coal productivity is tentimes greater than Eastern. The require-ment of comparable pollution controls forcombustion of low -sulfur Western coal asfor high -sulfur Eastern coal adds to thecosts, as does the additional investment forupgrading the poorer Eastern roadbeds.

Rapid expansion of coal use depends oncost and availability of pollution -controlequipment and industrial boilers. Theaverage cost of pollution -control equip-ment itself for large industrial boilers is $4

million per installation; the cost of con-verting to coal, including new boilers of6000 -MW generating capacity, is $4

billion. To burn the additional coal that thePlan assumes industry will use, 2500 newcoal-fired boilers will be needed by 1985 toreplace existing boilers. To meet thesegoals, increased production capacity mustbe on line soon. Faltering of demand orproduction will result in 1985 goals fallingshort.

Potential coal users must be assured ofadequate and reliable supplies. This im-plies about 100 new rail and ship or bargetransportation systems; several 500- to1000 -mile coal -slurry lines and about 8,000new locomotives and 150,000 new coalcars. This means about three newlocomotives and 60 new coal cars every dayfor the next seven years.

In the environmental area, let us assumethat the NEP goal to achieve coal conver-sion without sacrificing air -quality stan-dards is met. Nevertheless, the Plan's goalswill lead to other adverse environmentaland societal impacts which were not ad-dressed.

The best available control technology isnot adequate for nitrogen -oxide emissionsfrom coal burning. Radioactive materials,carbon monoxide, and heavy metals (gas,liquid, or solid state) from coal combustionare not regulated. Coal releases morecarbon dioxide during combustion than oilor natural gas. Since an increase in carbondioxide may contribute to global climaticchanges, accelerated use of coal mayaggravate long-term adverse effects onclimate.

Scrubbers produce about 3,000 tons ofsolid suspended in several thousand tons ofwater per day, per gigawatt day of energygenerated. Present nationwide electriccapacity using all fuels is 400 gigawatts.Disposal of all this sludge presents aserious land -use problem which includescontamination of ground and surfacewater by the sludge's moisture.

Western coal mining implies competitionwith the farmer for water for surfacereclamation and potential contaminationor loss of groundwater aquifers; Easternmines pose the threat of acid -water runoffs.

Longer and more frequent train trips willaffect rural towns in terms of sustainednoise levels, air pollution along rights -of -way, increased accidents, crossing delays,

14

and railroad congestion. Energy -relatedpopulation increases in small Westernlocalities will tax existing schools, medicalservices, water and sewer facilities. Loansfrom banks for homes and private facilitieswill be difficult to obtain because of theboom and bust nature of the development.This can lead to a degradation in thequality of life of these communities.

If the program to cause a shift to coal issuccessful, the impact of a coal strike suchas experienced this past winter could beenormous.

Nuclear electric energyThe most abundant energy source in theground in the U.S. after coal is uranium,about one -tenth of the coal reserve.

The uranium supply shown in Fig. 9 is thesupply of U-35. This rare isotope is used tofuel light -water reactors. In a plutoniumbreeder reactor, the abundant U238 is con-verted to plutonium, which can be used as afuel. In this case, the uranium fuel supplyshould be increased fifty times, to five timesthe coal reserves.

Although it may be possible to generate gasand oil from coal, nuclear energy simplyproduces heat, which may be used togenerate electricity. In the absence of theenvironmental impact of massive coalburning, there is not a clear domesticreason to develop nuclear electrictechnology, particularly since it has its ownwaste -disposal problems. However, not allcountries are blessed with our coal reservesand nuclear energy plays a major role in theplanning of many.

The common reactor in this country is thelight -water reactor, which uses enricheduranium as a fuel and produces smallamounts of plutonium as a by-product.There is particular interest overseas in thebreeder technology because it manufac-tures fuel and few countries have theuranium supplies of the U.S. The liquid -metal fast breeder reactors use plutoniumas fuel, however, and plutonium may beused to produce nuclear weapons. Becauseof the dangers with plutonium -basedtechnologies (plutonium breeder reactors),the U.S. has elected to delay thedemonstration phase of such powersystems. Further, the U.S. is asking foreigngovernments with nuclear capacity to dothe same. In order for this to be possible,the U.S. is offering to provide supplies ofenriched uranium for use in light -waterreactors so that foreign breeder reactors

RCA's energy -related programsAs described in the accompanying article, the U.S. government response tothe new energy situation is two -fold First, through conservation and aswitch to coal use, reduce consumption of oil and natural gas. Second, forthe long te'm, develop renewable sources of energy. RCA is addressingboth the near- and long-term situations with programs devoted tooptimizing energy -use practices in this new environment and exploringopportunities for new business activities in both the near and long term.

Briefly reviewed below, a description of RCA's programs will appear in theFeb/Mar 1979 issue of the RCA Engineer.

In 1973, RCA instituted a corporate -wide conservation program. Afteraccounting for RCA's increased production, energy use today is 30% lessthan it would have been using 1973 practices. Because the new energysituation is an escalating one, investments in optimized energy manage-ment, whether in building design and management, or industrial process-ing, have continuing and increasing returns. Last year, RCA Laboratoriesestablished the Energy Systems Analysis Group to assist RCA's conserva-tion program with software and hardware capability addressed to specificdivisional needs.

Several research activities evaluating the usefulness of recent RCAdevelopments for using solar energy are under way. A new optical coatingthat will improve the efficiency of solar thermal collectors is under study.Techniques for low-cost manufacture of single -crystal silicon solar cellsand the utility of an inexpensive RCA -developed, sun -tracking solar -cellarray are being evaluated. The practicality of amorphous -silicon solar cells,an inexpensive thin-film alternative to the single -crystal devices, is receiv-ing considerable attention. A technique for storing hydrogen as formic acidis being developed and its energy efficiency is being improved. This energy-storage technique could affect present energy use as well as the futuresolar -based technologies.

will not be necessary. In view of thedisruptions experienced by foreigngovernments caused by the increase in theprice of OPEC oil, it is not clear that thesesame governments will be willing to rely onanother external source of energy, this timein the form of enriched uranium.

In order for the U.S. to play a continuinginternational role in the control offissionable materials, it would seem

necessary to maintain an active effort in thedevelopment of nuclear technologies,particularly in the development of non -plutonium -based breeder reactors, (e.g.,thorium -based systems).

Hydroelectric generatorsWhen available, hydroelectric power coststen times less than that generated by ther-mal (nuclear, gas, coal, and oil) powerplants.

The year 1977 was the first one in whichmore energy was generated by nuclear

power than by hydroelectric power,principally because of a reduction inhydropower because of poor rainfall. In1975 and 1976 hydroelectric power ac-counted for 15% of the nation's electricitygenerated, while nuclear power increasedfrom 8.7% to 9.5%. In 1977 the hydroportion dropped to 10%. As it has beenestimated that capacity at existing sitesmight be increased by about one-third, theplan calls for a Department of Defense(Corps of Engineers) study to evaluate thispotentiaL The use of smaller -scale hydroplants in New England is also being con-sidered.

New sources of energy

SolarBy far the largest energy source available tothe U.S. is solar energy.

On the average, the sunlight deposits anamount of energy equivalent to our total

15

coal reserves in one or two months,depending on how big our coal supplyreally is. The problem with solar energy isthat it is difficult to collect, as it is

distributed over the whole country. Theyear -long average power density is 200W/ m' with a peak insolation (sunlightpower density) of I kW/ m2 at noon on aclear day.

The energy falling on the roof of a 1500 -ft"house has a yearly, country -wide averageof 2 million BTU / day, at a rate of about 75kW during the day. This energy can becollected and used for heating and coolingpurposes or for generating electricity. If theenergy is to be used as heat, it can becollected with an efficiency of between 30%and 60%, depending on the temperature ofthe delivered energy. If the sunshine is to beused to generate electricity, it can becollected and converted to electricity withan efficiency of about 5%. With theseconversion rates, the energy needs of such ahouse could be supplied by the sun if itshone in this average fashion.

Brown Williams became Director of theEnergy Systems Research Laboratory atRCA Laboratories in 1977. He had previous-ly been responsible for RCA's solar -energystudies.

Contact him at:Energy Systems ResearchRCA LaboratoriesPrinceton, N.J.Ext. 2535

Author Williams with a concentrating solararray.

Solar hot water and space heaters areavailable commercially and, compared toother solar technologies, are relativelycost-effective. The NEP proposes tax in-centives to make their cost to the consumerless. A tax credit of 40% of the first $1000 ofthe purchase price and 25% of the next$6400 (for a total of $2,000) is proposed.This credit would decline with time, butsome credit would be available for ex-penditures from April 20, 1977 toDecember 31, 1984. Some states havealready exempted solar equipment fromproperty -tax assessments. Solar equip-ment also qualifies for the proposed 10%tax credit for conservation in industrialuse.

Direct generation of electricity fromsunshine is cost-effective at the momentonly in very remote regions. The installedsystem costs must be reduced by a factor of50-100 for these technologies to represent acredible alternative to conventionalsources of electricity.

Two basic approaches are under develop-ment for direct electricity generation. First,photovoltaic devices; i.e., solar cells, whichare semiconductor devices that convertlight to electricity, are having their costsreduced. The cost of power supplies basedon these devices does not benefit from largeeconomies of scale, so that such systemsmight be installed wherever the energy isused. The deployment of large arraysduplicating the performance of centralpower stations is also an alternative. A1000-M W solar power supply, comparableto the new nuclear or coal plants, wouldoccupy about 20 square miles of land. Aproposal to locate such large arrays inspace where daily and seasonal variationsin the power output do not exist is understudy. Second, electricity can also begenerated from sunshine using the energyto heat a fluid and using the fluid to drive aheat engine. A 10-M W generating plant ofthis type is under construction in Barstow,California. In this facility, an array ofmirrors focuses the sunlight on a boiler,which drives a generator as in a con-ventional power station.

Technologies that use solar energy in-directly are also under development. Wind,ocean thermal, and biomass systems are invarious stages of implementation. Anumber of novel wind -driven electricitygenerators are under study and a federallysponsored 200 -kW machine is beingcoupled into the power grid in Clayton,NM. The ocean thermal devices are heat -engine electricity generators that operate

on the temperature differences betweensurface and subsurface water in the ocean.In this case the supply of energy is vast,although the efficiency is low because ofthe small temperature differences. Verylarge quantities of fluid must be moved,and warming the cold, nutrient -laden sub-surface water may result in some fouling.This type of electricity generator is in anactive study and research phase.

The use of agricultural and forest by-products for energy production is alreadywell known. The growth of productsspecifically for energy generation and asfeedstocks for liquid and gaseous fuels isunder development.

GeothermalGeothermal energy is in practical use now.

Geothermal energy in the form of steam ispresently used to generate 500 M We innorthern California, and there are

numerous sites in the West where hot wateris available. (If the temperatures are highenough, electricity may be generated;otherwise the heat energy may be useddirectly.) The NEP proposes to extend togeothermal drilling exploration the sametax benefits now available for oil and gasexploration. There is an active R&Dprogram to develop this resource.

Fusion

There are two approaches to generation ofelectricity using fusion: magnetic confine-ment and inertial confinement.

With magnetic confinement, a magneticfield is configured to form a cavity thatcontains the ionized gas fuel for a time longenough, and at a temperature and densityhigh enough, that the fusion reaction cantake place. In inertial confinement, a fuelpellet is compressed by a shock waveresulting from the ablation of the surface ofthe pellet. This compression shouldproduce the needed time, density, andtemperature to initiate fusion. In bothcases, the fuel is hydrogen or its isotopesand the fusion reaction results in theformation of helium and energy. So far,these systems have not fulfilled a criterionknown as "breakeven," (i.e., producingmore energy than they consume) by aconsiderable margin.

What is the government'semphasis on these programs?

Some insight may be gained on the relativeimportance the administration places on

16

$ BILLION

ENERGY SUPPLYRBD

ENERGYSUPPLY - PRODUCTION

DEMONSTRATION& DISTRIBUTION

CONSERVATION

STRATEGIC PETROLEUMRESERVE

ENVIRONMENT

BASIC SCIENCES

ATOMIC ENERGYDEFENSE

POLICY & MANAGEMENT

A

Fig. 11How the Department of Energy allocates Its money. Energy -supply R&Dis third, behind the strategic petroleum reserve and atomic -energydefense. Program areas are explained in text.

the programs outlined above by an inspec-tion of Fig. I I. The makeup of the R&Dprogram will be discussed in the paragraphbelow. The Energy Supply-ProductionDemonstration and DistributionPrograms include Naval oil reserves, solarheating and cooling demonstration,uranium enrichment and resource assess-ment, alternate fuels demonstration, andelectric power marketing administration.The Conservation area includes conserva-tion programs in all sectors of the con-suming environment, government andprivate. The strategic petroleum reserve, byfar the largest single item in the budget, is aprogram to stockpile 500 M BD (30 dayssupply at present consumption rates) byDecember 1980. The Environmentprogram is principally biomedical andenvironmental research. Basic Sciences is aprogram in life sciences, nuclear physics,and support for several high-energy parti-cle accelerators. Half of the Atomic EnergyDefense budget is for weapons activities.The rest is the inertial -confinement fusionprogram, Naval reactor development,materials production, and security andsafeguards.

Ignoring the strategic petroleum reserveand the defense activity, the R&D programis the largest by far. Fig. 12 shows how thevarious parts of the R&D program areallocated. Half is for nuclear electricityprograms and half for new energy sourcesor new uses of old energy forms. With theimportance that coal has in the overall

$ MILLIONN to A C.0

0 0 00 0 Q 8

COAL

OIL

GAS

SOLAR

HYDROELECTRIC

BIOMASS

GEOTHERMAL

BASIC SCIENCE

MAGNETIC FUSION

FUEL CYCLE R & D

BREEDER REACTOR

NUCLEAR RESEARCHLIGHT WATER REACTOR

FACILITIESINTERNATIONAL

SPENT FUELSTORAGE

§ OO

Fig. 12How DOE research is allocated. Coal research takes the lion'sshare. Next -largest is the fusion/fission area, followed bysolar -energy research, which is being expended mostly onsolar cells.

strategy, it properly is the largest program.Work on liquification and gasificationconstitutes half of the expenditures, anamount about equal to the solar program.

The solar program is almost all electricitygeneration, with solar -cell developmentand solar thermal -electric generation ac-counting for most. It is interesting thatwhile an inspection of Fig. I and mostnewspapers reveals that our massive con-sumption of oil, principally for gasoline, isour major problem, there is such a largeemphasis in the R&D budget on alternativesources for electricity generation.

SummaryAs in the U.S., the growth in energy usethroughout the world has been met almostexclusively with oil and natural gas. Theseresources are finite and, as we are ap-proaching the point where their finitenessis apparent, their costs are going up. TheU.S. response to this new and permanentsituation is two -fold. First, through conser-vation and a switch back to the use of coal,conserve natural gas to insure availabilityfor domestic use and conserve oil for use intransportation. Second, for the long term,develop renewable energy sources such assolar energy or sources with enormous fuelsupplies, such as fusion.

For RCA the challenge is also two -fold.First, the energy environment has changed

and is going to continue to change. WhileRCA is not a particularly energy -intensivecompany ($30 million/ year), we mustadapt to these changes to maximize profitopportunities in our existing businesses.Second, opportunities for exploiting elec-tronic device, materials, and systems ex-pertise in the enormous energy field doexist. Energy is one of the largest businessactivities in the nation and its changedeconomics means that companies that havenot traditionally been in the energy fieldmay now participate. Matching RCA'scapabilities to this opportunity presentsexciting prospects for the future.

ReferencesI. Understanding the National Dilemma. A Report of the Joint

Committee on Atomic Energy, published by The Center forStrategic and International Studies, Georgetown University.Washington, D.C., (1973). Library of Congress Catalog CardNo. 73-87420.

2. Monthly Energy Review, prepared in the Energy InformationAdministration. U.S. Department of Energy, (Jan 1978),NTISUB.'D/ 127.001.

3. Voluntary Industrial Energy Conservation. Progress ReportNo. 5. prepared by the U.S. Department of Commerce, Officeof Energy Programs, and the Federal Energy Administration.Office of National Energy Conservation Programs,Washington, D.C., (Jul 1977).

4. Energy Perspectives 2, prepared by the U.S. Department of theIntenor, Washington. D.C.. (Jun 1976).

5. The International Energy Situation: Outlook to 1985,prepared by the U.S. Central Intelligence Agency.Washington, BC., (Apr 1977).

6. The National Energy Plan, prepared by the Executive Office ofthe President. Energy Policy and Planning, Washington,D.C., (Apr 1977).

7. U.S. Department of Energy FY 1979 Budget to Congress-Budget Highlights, prepared by the U.S. Department ofEnergy. Washington, D.C., (Jan 1978).

Reprint RE -24-1-1Final manuacripl received April 25, 1978.

17

R.W. Engstrom

Today, RCA has a major share of theimportant photomultiplier market, but wegot there without really planning to. Tobegin with, the very limited market forphotomultiplier tubes was purely scientificin orientation. No mass market was fore-seen. Competition was minimal. Afavorable attitude of RCA's managementtoward high-technology products nurturedthe product line. The evolution of the widevariety of photomultiplier tubes was large-ly guided by our responses to severalunanticipated applications.

BeginningsThe origin of the photomultiplier is rooted inearly studies of secondary emission.

In 1902 Austin and Starke' reported thatmetal surfaces impacted by cathode raysemitted a larger number of electrons thanwere incident. A great deal of research wasdone in the 1920s and 1930s on theproperties of secondary emission.2 Thedevelopment of radio and television led toseveral practical applications of thisphenomenon; amplifier -power tubes ex-ploited secondary emission for additional

1000

80

WI-LL

OW 60

3%z040uo2utitit 20

CC

1.12.ilto then and now

An account of their development from a very specialized andminor business to a diverse but major enterprise.

gain as did television camera tubes (such asthe image dissector, the image orthicon,and the isocon) and, of course, the photo -multiplier.

The distribution of emission energies ofsecondary electrons from metals is general-ly reported to be as shown in Fig. I. Thenumber of secondary electrons emitted perincident primary electron is, of course, veryimportant in many practical applications.For the photomultiplier, it is also useful tohave a relatively high yield at modestprimary energy to minimize appliedvoltage. Typical secondary -emission yieldcurves are shown in Fig. 2 for a number ofsurfaces that have been used inphotomultiplier tubes. In the firstphotomultiplier tubes made, thephotocathodes were Ag-O-Cs, a red -

infrared -sensitive photoemitter, and it wasfound expedient to use the same materialfor the secondary -emission surfaces (ordynodes). Later, with the development ofthe very sensitive Cs3Sb photocathode itwas found that the same material alsomade a very good secondary emitter,although not as stable at moderate current

0 20 40 60 80 100 120 140 160

ELECTRON ENERGY-ELECTRON VOLTS

Fig. 1Typical energy distribution of secondary emissionof electrons; the peak at the right is caused byreflected primary electrons. In metals, secondaryelectron energies generally range from 0 to 10 eV.

100

levels as MgO of CuBe. Major use today ismade of the Cs3Sb and the CuBe emitters.

In recent years, a remarkable "negative -electron -affinity (NEA)" secondary -emission material (Cs -activated GaP) hasbeen reported by R.E. Simon.' Very highsecondary -emission yields are obtained(see Fig. 2) because of the reduction of thesurface energy barrier. This NEA materialis used primarily for special applicationswhere the very high gain results in areduction of the statistical noise associatedwith the secondary -emission amplificationprocess.

Secondary electrons are directed to the firstand succeeding dynodes by electron optics.

In 1934, Philo Farnsworth' reported on amultipactor device in which electrons weredirected back and forth between twosecondary -emitting surfaces. Energy wasapplied by means of an alternating electricfield whose frequency was timed to the timeof flight of the electrons. The multipactornever evolved into a practical photo -multiplier; Farnsworth's principal interestin this idea was in applying it to the image

6

6

GaP(CS)

2

Cs3Sb

4*a

6

., ..

Cs -Cs0 -Ap / .0.

../.."Cu 8.0 -Cs

2 f2 0 6 6 2 4 6 8 2 4 6 13

10

ACCELERATING100 000 101(

VOLTAGE OF PRIMARY ELECTRONS - VOLTS

Fig. 2Secondary emission yieldplotted as a function ofaccelerating potential ofprimary electrons forvarious emitters that havebeen used in photo -multiplier tubes. With thenewer negative -electron -affinity (NEA) materials-e.g., GaP(Cs)-second-ary electrons can becreated at greater depthsand still escape.

18

dissector for television pick-upapplications.

Also, in 1934, Georg Weiss in Germanywas working on a photomultiplier in whichthe secondary -emission surfaces werefabricated of mesh.' The mesh dynodeswere mounted in a parallel arrangement sothat primary electrons impinged on oneside of the mesh and secondaries wereextracted and accelerated to the next meshfrom the opposite side. No commercialdevices resulted.

In 1935, lams and Salzbere of RCAreported on a single -stage photomultiplier.The device, Fig. 3, consisted of a semicylin-drical photocathode, a secondary emittermounted on the axis, and a collector gridsurrounding the secondary emitter. Thetube was basically intended as an improve-ment over the gas -filled phototube whichwas used as a sound pickup for movies. Ithad a gain of about eight and a much betterfrequency response, but despite these ad-vantages saw only a brief developmentalsales activity before it became obsolete.This electron -optical arrangement for asingle stage of secondary emission wasrelatively unsophisticated, electron -opticalproblems become more difficult when anumber of stages of secondary emission arerequired.

In 1936, Zworykin, Morton, and Matter,all of RCA' reported on a multistagephotomultiplier. The principal applicationcontemplated was sound -on -film pickup.Their photomultiplier used a combinationof electrostatic and magnetic fields todirect electrons to repeated stages of secon-

INCIDENT LIGHT

Fig. 3Single -stage photomultiplier reported bylams and Salzberg in 1935. Photoelectronsleaving the concave photocathode strike thedynode strip and eject secondary electronsthat are collected by the flattened spiralcollector grid surrounding the dynode.

The photomultiplier tube: how it works

The photomultiplier is a very sensitive detector o' light.

LIGHT

-Ow

PHOTOCATHODE

DYNODES

PHOTOELECTRONS

ANODE

SIGNAL

When light strikes the photocathode, located inside the vacuum envelope,photoelectrons are emitted ano directed by an appropriate electric field to thenext electrode or dynode. For each primary photoelectron hitting this dynode,a number of secondary electrons are emitted. These secondary electrons, inturn, are directed to a second dynode to produce more secondary emission-and so on until a final gain of perhaps 106 is achieved. The electrons from thelast dynode are collected by an anode which provides the signal current that isread out. Because the photomultiplier's high gain is practically noise -free, tubeperformance is close to the ideal, limited only by the shot noise of the initialphotoelectron current.

In applications where speed of response is important, the photomultiplier alsodoes well because of the short electron emission and transit times.

dary emission, Fig. 4. Each successivesecondary emitter was operated at a highervoltage-perhaps 100 V -than theprevious emitter. Above each emitter was afield plate whose potential was set to beequal to that of the next emitter down theline. As a result of this electrostatic -magnetic configuration, electrons emittedfrom the photocathode, or from one of theemitters, were caused to follow ap-

ICATHODE

FIELD PLATES

proximately cycloidal paths to the nextelectrode.

The design of the Zworykin magneticphotomultiplier was based on relativelystraight -forward calculation of electrontrajectories. Developmental samples werebuilt and tested, Fig. 5. But although themagnetic -type photomultiplier did providehigh gain, it had several difficulties. The

EMITTERS COLLECTOR -2'

Fig. 4Magnetic -type multistagephotomultiplier reported byZworykin, Morton, and Malterin 1936. A magnetic field,perpendicular to the plane ofthe drawing, and an elec-trostatic field, produced by theupper field plates, combine to

bend the paths of electronsleaving the lower plates so

that they strike the nexttarget to produce sec-

ondary electronswhich are in turndeflected to an-other target and soon to the collector.

Fig. 5Developmental sample of the Zworykin magnetic photomultiplier described in Fig. 4.

19

adjustment of the magnetic field was verycritical, and to change the gain by reducingthe applied voltage, the magnetic field alsohad to be adjusted. Another problem wasthat the rather wide open structure resultedin high dark current because of feedbackfrom ions and light developed near theoutput end of the device. For these reasons,and because of the development ofelectrostatically -focused photomultipliers,commercialization did not follow.

The design of multistage electrostaticallyfocused photomultipliers required an

analysis of the equipotential surfacesbetween electrodes and of the electrontrajectories. Before the days of high-speedcomputers, this problem was solved bymechanical analogue. A very convenientand successful model was a stretchedrubber membrane. One such analoguemodel, as used by Jan Rajchman, is shownin Fig. 6. By placing mechanical models ofthe electrodes under the membrane, theheight of the membrane was controlled andcorresponded to the electrical potential ofthe electrode. The vertical position at anypoint on the membrane between electrodes

Fig. 6Rubber sheet mechanical analogues were used long before the computer to analyzeelectron trajectories and equipotential surfaces between electrodes. This photo from theFebruary 20, 1939, issue of Life, shows Jan Rajchman of RCA Laboratories using the rubber -dam analogue to analyze photomultiplier focusing potentials.

Fig. 7Equipotential lines and electron trajectories in Rajchman's linear dynode configuration(circa 1938). The linear dynode array is used today for high -gain wide -bandwidth photo -multipliers.

also corresponded to its electric potential.Small balls were then allowed to roll fromone electrode to the next. The trajectoriesof the balls were shown to correspond tothose of the electrons in the correspondingelectrostatic fields. With appropriate con-figurations, friction and depression of themembrane by the ball were made negligi-ble, and the analogue was essentially exact.The model, however, was only valid forgeometries in which the electrodes could beassumed to be cylindrical surfaces(generated by a line parallel to a fixeddirection and moving along a fixed curve)sufficiently long to be considered infinite inextent.

Working with the rubber -dam analogue,both J.A. Pierces of Bell Laboratories andJ.A. Rajchman9 of RCA devised lineararrays of electrodes that provided goodfocusing properties. Details of Rajchman'soriginal linear design are shown in Fig. 7.Although commercial designs did notresult immediately from the linear dynodearray, the Rajchman design, with somemodifications, eventually was, and still is,used in photomultipliers-particularly forhigh -gain wide -bandwidth requirements.

The first commercial high -gain photo -multiplier was the 931, now the 931A.

The 931 tube has a circular array of ninedynodes, which provides a very compactarrangement, Fig. 8. The first such circulararrangement was described by Zworykinand Rajchman.9 Modifications were laterreported by Rajchman and Snyder'° and

SHIELD

GRILL

INCIDENTLIGHT

0 PHOTOCATHODE1-9 DYNODES10 ANODE

Fig. 8Circular -dynode arrangement used in thefirst commercial high -gain photo-multiplier-the 931. This very compact arraywas first described by Zworykin and Rajch-man in 1939.

20

by Janes and Glover," all of RCA. Thebasic electron -optics of the circular cagewas thus well determined by 1941 and hasnot changed to the present time althoughimprovements have been made in process-ing, mounting, and performance of the93IA product.

The success of the 931 type also resultedfrom the development of a much improvedphotocathode, Cs3Sb, reported by Gorlichin 1936. The first experimentalphotomultipliers had used a Ag-O-Csphotocathode with a typical peak quantumefficiency of 0.4% at 800 nm. (The Ag-O-Cslayer was also used for the dynodes.) Thenew Cs3Sb photocathode had a quantumefficiency of 12% (higher today) at 400 nm.It was used in the first 931s, both as aphotocathode and as a secondary -emittingmaterial for the dynodes.

Applications and designsWhen Jan Rajchman began working forZworykin on January I, 1936 in Camden,(the Princeton Laboratories were not oc-cupied until 1942) his assignment was todesign a photomultiplier that would beuseful for sound -on -film pickup. Themagnetic focus tube described above wasalready on hand but was not consideredphysically or economically feasible for thesound -on -film application. At the time, the931 seemed a logical replacement for thephototube plus amplifier then being used.Although tests were made, the applicationof photomultipliers to sound movies wasnot adopted by the industry. Instead, theearly applications of photomultipliers wereprincipally in astronomy and spec-troscopy.

Because the effective quantum efficiency ofthe photomultiplier was at least ten timesthat of photographic film, the applicationto astronomy was obvious, and manyastronomers used the tube to considerableadvantage." The output of the photo -multiplier is linear with the incident radia-tion, and so the tube could be used directlyin photometric and spectrophotometricastronomy. Zwicky of Mt. Wilson wasprobably the first astronomer to usephotomultiplier tubes in this manner. Heused early experimental tubes which JanRajchman supplied to him in the late 30s.Before photomultipliers, phototubes withhigh -gain amplifiers were used. (Today,many astronomers are intensifying imageswith image -intensifier tubes which not onlyprovide high quantum -efficiency detectionbut image information as well.)

The 931 and a similar tube, the type 1P28with an ultraviolet -transmitting window,were also useful in spectroscopy. The sizeand shape of the photocathode were quitesuitable for the observation of line spectra,and the very wide range of available gainproved very useful." Note in Fig. 9 how thegain can be varied over a range of fiveorders by varying the voltage applied to thephotomultiplier tube. By using the voltagerequired for constant output current, anapproximately logarithmic calibration isachieved which nicely accommodates thewide range of intensities encountered inline spectra.

The photomultiplier became a radar jammerduring the war.

A totally unexpected application for thenew photomultiplier tube occurred duringWorld War II. The development of radarfor detecting and tracking aircraft led tothe simultaneous need for widebandelectronic -noise sources as radar jammers.Gas tubes were tried, but were deficient inhigh -frequency noise. Thermionic-emission shot noise was suitable, but re-quired very high amplification. The newlydeveloped photomultiplier tubes, however,provided high gain and wide bandwidth.During the early 40s, RCA participated ina contract sponsored by the Office ofScientific Research and Development.Participants in this contract were AlanGlover (later RCA vice president, nowretired), myself, and W.J. Pietenpol (wholeft RCA after the war to obtain a PhD inPhysics, and who is now teaching at theUniversity of Colorado, Boulder). Opera-tion of the 93I A as a noise source was doneby illuminating the photocathode with a dclight source and running the tube at a highgain.

The amplified shot noise from thephotocathode is given approximately by

(2eik B)I

where Ai is the gain of the tube, e is theelectron charge, ik is the photocathodecurrent, and B is the bandwidth of thesystem. Note that the noise current variesdirectly with the gain and with the squareroot of the photocathode current. Thebandwidth of the 931A is several hundredMHz (10 dB down at 500 MHz) and, thus,was adequate for jamming World War IIradars. The gain of the tube could be ashigh as 107, but too high a gain led to

Contract No. OEM sr 1060 (Final Report. 3/7/45).

107

ins

104

103

102025 50 15 100 125

VOLTAGE PER STAGE -VOLTS150 175

Fig. 9Logarithmic gain characteristic of the 931Aphotomultiplier as a function of voltage per stage.Note that the gain can be varied over five orders byvarying voltage. (The 9 -stage 931A is a modernmodification of the early type 931).

unwanted breakdown in the tube.Photocathode current could be increasedreadily by means of the light source, but asthe photocathode current increased so didthe output current. Too high an outputcurrent led to short life and loss of noiseoutput. One of the tasks of the contract wasto improve tube life; the modified 931 (the"A" version) was developed and run at anoutput current of 2.5 mA.

A particular advantage of the 93I A for thejamming application was the fact that itsnoise spectrum was "white." The RCAphotomultiplier design group was elatedand amused at the first report of thejammer's application in Italy: allied radaroperators who were not told about the testshut down their equipment because of"atmospheric conditions."

During part of the noise contract, amodified photomultiplier was tried inwhich a planar grid was mounted close tothe photocathode. We hoped to achieve ahigher noise output by feeding back part ofthe output noise to the grid. This schemewas not exploited because, although it didprovide increased noise, the output noisespectrum was no longer flat.

Prior to this jamming application, produc-tion of photomultipliers was measured inonly hundreds per year. The wartimeapplication required thousands ofphotomultipliers per month. J.J.Polkosky, RCA's phototube manufac-

21

turing manager at the time, did a

remarkable job of increasing our produc-tion in a very short time.

Scintillation counting was the next majorphotomultiplier development.

The scintillation counter is a direct exten-sion of Rutherford's very early "spinthar-iscope" device in which alpha particlesimpinging on a fluorescent screen causedlight flashes which were observed by meansof a microscope. In the modernscintillation -counter, gamma rays producelight flashes in a scintillator such asthalium-activated sodium iodide, Nal(T1).A photomultiplier optically coupled to thescintillator detects the light flashes, allow-ing them to be counted and theirmagnitude measured.* The count rateprovides a measure of the intensity of thegamma -ray source. The magnitude of theflash, or pulse height, is proportional to theenergy of the gamma ray. This latterproperty provided a means for the nuclearphysicist to identify energy levels in nucleardisintegrations. It also enabled the user, bymeans of pulse -height discrimination, toselect the scintillations of interest andeliminate other background.

Invention of the scintillation counter short-ly after World War II is generally creditedto Kallmann" working in Berlin and usingmoth balls (naphthalene) as a scintillator.At nearly the same time, Coltman andMarshall" of Westinghouse published asimilar application paper.

Following the invention of the scintillationcounter, scientists and engineers in thiscountry and in Europe began a feverisheffort to improve the device and to exploitits capabilities. The scintillation counterbecame the most important measurementinstrument in nuclear physics, nuclearmedicine, and radioactive tracerapplications of a wide variety.

During this early period, Hofstadter"investigated various scintillators and dis-covered that by doping Nal with thallium,an excellent scintillator resulted. Thisscintillator was used almost exclusively forthe next two decades in scintillationcounter applications. Hofstadter had pa -

Incidentally, the geiger counter is only useful in counting eventsand does not measure the energy of the gamma ray. Because ituses ionization of a gas, with relatively low absorption ofgamma -ray energy, to trigger the count, the efficiency ofcounting is much less than that of a scintillation counter.

NN

Fig. 10Scintillation counting stimulated the development of several photomultiplier types. Left toright (starting on top) are the 5819, probably the first tube designed specifically forscintillation counting; the 6199, a smaller tube used by early uranium prospectors and alsoplaced aboard a satellite by Dr. Van Allen and used to discover the radiation belts that nowbear his name; the 6342A, which has a flat faceplate to provide an easier interface with thecrystal; the 6810A, a 14 -stage linear dynode tube for higher gain and less feedback betweenphotocathode and collector (the circular arrangement is limited to 10 stages and thephotocathode and anode are close to each other); the 7850, with improved electron opticsfor higher speeds needed to time certain nuclear events; the 8054, which handles largercrystals (about 3 inches) for better absorption of high-energy gamma rays; the 8055, whichaccomodates 5 -inch crystals aid contains RCA's "venetian blind" optics; the 8575, animproved 7850 with a flat faceplate; and the 8854, developed for the AEC for use in highspeed measurements with plastic scintillators.

tent rights for the Nal(T I) and is generallybelieved to have done very well on theroyalties.

Because of the early work onphotomultipliers, RCA was in a favorableposition to profit by the expansion of thenew scintillation -counting field. RCA'sposition was advanced rapidly by GeorgeMorton and his associated laboratorygroup in Princeton, frequently financed bythe AEC. The Lancaster developmentgroup took advantage of Morton's in-itiative and commercialized numeroustypes of photomultipliers. My own timelyapplication paper" served as a primer formany new users and advertised RCA'sinvolvement.

Hofstadter's Nobel Prize in 1961 did not relate to the Nalscintillator for which he is so well known, but to bit studies of thesize and internal structure of the atomic nucleus.

RCA's competitors-DuMont, EMI, andLEP (Philips)-were also very busy, andduring the next decade there was a prolif-eration of photomultipliers developedparticularly for scintillation counting.Much of the work during this period wasreported by RCA and its competitors in thebiannual meetings of the ScintillationCounter Symposium. These symposiahave been reported fully in the IRE (andlater the IEEE) Transactions on NuclearScience beginning with the meeting inWashington, January 1948.

A number of the RCA tubes that weredeveloped for scintillation -countingapplications are shown in Fig. 10. One ofthe early design problems was to improvethe light -coupling system between thescintillator and the photomultiplier tube.Although the 931 A was used early as ascintillation -counter tube, the fact that thephotocathode was located some distance

22

within the envelope prevented really goodlight collection.

For a precise measure of energy in agamma -ray spectrum, it is important todetect as many of the photons per scintilla-tion as possible. The first commercial tubedesigned for better light collection from thescintillator was the 5819, which provided asemitransparent Cs3Sb photocathode onthe inner surface of the end -on window.The tube had a drawback; the window tothe photocathode was not quite flat, thusrequiring the user to shape the NaI(T I )crystal to match. The 6342A provided a flatcoupling surface. The 6199, a 11/2 -inch end -on tube, reasonably priced, was used inlarge numbers in applications such asportable scintillation counters, whichmany prospectors used to find uraniumores.

A series of tubes with different diameterswas made to accommodate differentscintillator dimensions. The very largesttubes were made on an experimental basisonly because of the great expense of thevery large Nal(T I) crystals. The re-quirements for the large crystals wasrelated to the absorption of high-energygamma rays. (Even a crystal 10 -cm thickabsorbs only about 90% of I MeV gammarays.)

Nuclear physicists were particularly in-terested in obtaining photomultiplier tubesthat had a very fast response. Specialscintillators were also required becauseNal(T1) has a relatively long time constantof about 0.25 ms. Organic plasticscintillators, such as p-terphenyl inpolyvinyltoluene, which has a decay timeof 3 ns, were developed. Typical ex-periments in nuclear physics involvedmeasuring time of flight and lifetime ofdecay products. The discovery of the anti-proton with the Bevatron machine at theUniversity of California, Berkeley, in 1957,was accomplished with the RCA type 6810,one of the early high-speed tubes (Fig. 10).Time discrimination with this tube waspossible to better than I ns. Design forshort -time discrimination in photo -multipliers was a matter of electron opticsand equalizing electron paths.2°

Scintillation counting has revolutionizedmany areas of science, engineering, andmedicine.

An agronomist may observe the locationand uptake of trace elements in growing

plants without destroying the plant.Archeological dating of very old samples ofwood is accomplished by measuring theresidual '4C, a radioactive isotope of '2C,which decays with a half life of about 5000years. The original equilibrium ratio of10c:12c

is established in the living treethrough interaction with atmospheric CO2.An engineering investigation of tool wearmay be made by incorporating a smallamount of radioactive isotope in the metal.Detergents can be evaluated by producinga "trace" oil from '4C, combining it withother oils and measuring the remainingradioactivity in a fabric or in the rinsewater. Scintillation counting is regularlyused to log the type of subsoil in oil -wellexploration.

In medicine, scintillation counting was firstapplied experimentally to the location ofbrain tumors in the 1950s in a few hospitalssuch as the Massachusetts General,associated with the Harvard MedicalSchool. The patient was given an in-travenous injection of radioactive -arsenic -doped chemical which was known to con-centrate more actively in carcinogenoustumors than in healthy parts of the body.The tumor was then located by scanningthe patient's head with a scintillation -counter probe. These techniques have beenactively pursued in many fields, andparticularly in medicine. The latest tech-niques, the associated devices and therelated business will be described later inthis chronicle. Let us now deviate to apractical application of photomultipliersunrelated to scintillation counting.

RCA developed a $5 photomultiplier for useas a headlight dimmer.

The General Motors Company (Guide -Lamp Division) had worked periodicallyon a photoelectric headlight dimmer sincethe early 1940s. But a successful device wasnot produced until the 1950s. RCAcollaborated closely with General Motorsduring this period. The headlightdimmer-first made available only onCadillacs and Oldsmobiles -was basicallya 931A, redesigned and tested to G.M.'sparticular requirements. The opticalengineering problem was to sense theoncoming headlights or tail -lights beingfollowed without responding to street andhouse lights. Vertical and horizontalangular sensitivity was designed to matchthe spread of the high beams of theautomobile. A red filter was installed in theoptical path to provide a better balancebetween sensitivity to oncoming headlights

and to tail -lamps being followed. Lifecharacteristics again were a criticalproblem because slight changes in eachdynode gain-taken to the 9th power-could easily result in an annoyingmaladjustment!

Despite a:I the problems, the deviceachieved a remarkable success, partlybecause of the novelty, although manythought the device would have been moresuccessful if it had caused the dimming ofheadlights on the oncoming car! For a fewyears, RCA sold General Motors manythousands of these photomultiplier tubesper month. There was continual pressureby G.M. on RCA to reduce price. Theprice, less than $5.00, was the lowest of anyphotomultiplier RCA ever marketed. Sub-sequently, G.M. developed their ownsupplier and then went to a simplerphotocell pick-up. Today, one rarely sees aheadlight dimmer. Perhaps, manual opera-tion is not too difficult, after all.

Developments in photocathode andsecondary -emission materials kept pacewith other photomultiplier developments.

Much of the development work onphotomultiplier tubes that has been

described in this history relates to physicalconfiguration and electron optics. But avery important part of the development ofphotomultiplier tubes was related to thephotocathode and secondary -emission sur-faces and their processing. RCA was veryfortunate through these years in having onits staff, probably the world's foremostphotocathode expert, Dr. A.H. Sommer.*He joined the Associated PrincetonLaboratories in Dr. Morton's group in1952, and during the next two decades,contributed not only improved photo-cathode formulations but provided con-siderable help in the processing of photo-cathodes to the Lancaster developmentand factory groups. His treatise onPhowernissive Materials2' continues toprovide a wealth of information to allphotocathode -process engineers.

Sommer explored the properties ofnumerous photocathode materials-particularly alkali-antimonides. Perhapshis most noteworthy contribution was themultialkali photocathode (S-20 spectralresponse). This photocathode, (Cs)Na2K-Sb, is important because of its high

Since his retirement from RCA in 1974. Dr. Sommer has beenconsulting at Thermo Electron Corporation in Wellesley.Massachusetts.

23

sensitivity in the red -and near -infrared; theearlier Cs3Sb photocathode spectralresponse barely extends through the visi-ble, although it is very sensitive in the bluewhere most scintillators emit.

Bialkali photocathodes were alsodeveloped by Sommer and have proven tobe better in some applications than theCs,Sb photocathode. Thus, the Na2K-Sbphotocathode has been found to be stableat higher temperatures than Cs3Sb and, inaddition, has a very low dark (thermal)emission. It has been particularly useful inoil -well -logging applications. Anotherbialkali photocathode, K2Cs-Sb, is moresensitive than Cs3Sb in the blue and is,therefore, used by RCA to provide a bettermatch to the Nal(T1) crystals used inscintillation counting.

Negative -Electron Affinity (NEA) is an ex-citing new concept applied to materialsused in photoemission and in secondaryemission.

Electron affinity, Fig. 1 la, is the requiredenergy, in electron volts, for an electron atthe conduction -band level to escape to thevacuum level. By suitably treating the

BOTTOM OFCONDUCTION BAND

FERMI LEVEL

TOP OF VALENCEBAND

surface of a p -type semiconductormaterial, typically with Cs, the band levelsat the surface can be bent downward sothat the effective electron affinity is actual-ly negative with reference to the bulkconduction -band level, Fig. I lb. Ther-malized electrons in the conduction bandare normally repelled by the electron -affinity barrier; the advantage of NEAmaterials is that these electrons can nowescape into the vacuum as they approachthe surface. In the case of secondaryemission, secondary electrons can becreated at greater depths in the materialand still escape, thus providing a muchgreater secondary emission yield (Fig. 2)than can be accomplished with ordinarymaterials. In the case of photoemission, ithas been possible to achieve extended -redand infrared sensitivities greater than thoseobtainable with any other knownmaterials. The impact of these new NEAmaterials has so far been primarily inscientific applications.

The NEA concept was the brainchild of Dr.Ralph E. Simon while working at the RCAPrinceton Associated Laboratory. The ear-ly work was sponsored by the U.S. ArmyResearch and Development Laboratories,

ELECTRONAFFINITY

VACUUMLEVEL

Fig. 11aEnergy -band model of a typical semiconductor photoemitter; elec-tron affinity is the energy required to allow an electron to escape theconduction band.

BOTTOM OFCONDUCTION BAND

FERMI LEVEL IN p- TYPESEMICONDUCTORTOP OF VALENCEBAND

BENT -BAND REGION

NEGATIVEELECTRONAFFINITY

CESIUMLAYER

Fig. 11bEnergy -band model of a p -type semiconductor with a surface layerof cesium. Note that the band levels at this surface are bent so thatthe effective electron affinity is negative.

Fig. 11Negative Electron Affinity (NEA) materials allow secondary elec-trons to be released from greater depths.

40

z 10Ua.a.

U0

U

2

0.1

0 01

Fort Monmouth, NJ, and was reported inthe first quarterly report.22 In this report,Simon defined the concept of NEA and theconditions necessary for NEA togetherwith data on photoemission from Cs -covered polycrystalline GaP. The firstpublication of the NEA concept in the openliterature was by Scheer and Laar23 ofPhilips, Eindhoven. The early work atRCA was carried on particularly by R.E.Simon, B.F. Williams, and W.E. Spicer.

The first practical application of the NEAconcept was to secondary emission. Anearly paper by Simon and Williams'described the theory and early experimen-tal results of secondary -emission yields ashigh as 130 at 2.5 kV for GaP(Cs), Fig. 2.

The particular advantage of the high -gainGaP dynode is in the improved statistics ofmultiplication when the dynode is used inthe first stage of a photomultiplier tube.Thus, the distribution in the number ofsecondary electrons at the first dynode isproportional to the square root of theaverage secondary emission. For the GaPdynode, the number of secondaries may be36 as compared with 4 for a conventionaldynode. Such GaP-dynode tubes can beused effectively to discriminate singlephotoelectron events,24 and are used togreat advantage in astronomy and spec-troscopy.

The high gain of the GaP dynode also led tothe development of very fastphotomultipliers because high gain couldbe achieved with fewer stages.25 For exam-ple, the RCA developmental type C31024is a five -stage tube with GaP dynodesproviding time measurement capability

2,,z

8

AllikK2Cs SD

11 '14'

6 4794D4

2 Q, 4: 94,4

8

6,4s

6 A4 ,C1.t.tt--02

86

4

2

200 400 600 BOO 1000 1200WAVELENGTH - NANOMETERS

92CS -30710

Fig. 12Spectral responses of several NEA photocathodes (graylines) compared with those of an Ag-O-Cs, a bialkali, and amodified multialkali photocathode.

24

down to 300 ps (as determined byvariations in pulse transit time).

Much of the early work on NEA materialsfor photocathodes was done at the RCALaboratories in Princeton.26 The initialtechnology transfer to the Lancaster opera-tion was primarily the responsibility ofFred Hughes!' The impetus for the workwas the requirement for near -infraredsensitivity, particularly in relation toapplications involving the Nd:YAG lasersoperating at 1.06 Am. The evolution of thedevelopment was from GaAs to variouscombinations of InGalAs. Longerwavelength threshold was achieved with anincreased percentage of In.

Some of the early work on NEAphotocathodes for the near -infrared wasdone in relation to their application toimage tubes.28 The technology was, ofcourse, readily applied to photomultiplierdevelopment. Fig. 12 shows the range ofresponsivities of the new NEA

The RCA work on NEA photocathodes for image tubes wassponsored by the U.S. Army Electronics Command, NightVision Laboratory, Fort Belvoir, Va.

LEAD CAP(1" THICK)

PHOTOMULTIPLIERTUBES

PLASTIC LIGHT PIPE(DIVIDES THE LIGHTAMONG PMT's)

GAMMA-RAYSHIELD(2" THICK)

photocathodes (GaAs and In laGas2As)compared with those for the Ag-O-Cs,multialkali [Na2KSb(Cs)] and bialkali(K2CsSb) photocathodes.

Gamma cameras have been built with asmany as 91 photomultipliers.

The gamma camera originally described byAnger29 is a more sophisticated version ofthe scintillation counter, but used forlocating tumors or other biological abnor-malities. The general principle of the gam-ma camera is illustrated in Fig. 13. Aradioactive isotope combined in a suitablecompound is injected into the biood streamor ingested orally by the patient. Certaincompounds or elements are taken uppreferentially by tumors or by specificorgans of the body-such as iodine in thethyroid gland. The radioactive materialdisintegrates and gamma rays are ejectedfrom the location of the concentration.

A lead collimator permits gamma rays topass through it only when they are parallelto the holes in the lead; gamma rays atother angles are absorbed in the lead. Inthis way, the location of the gamma -ray

Fig. 13Gamma -ray cameras are usedworldwide to provide early diagnosis ofcirculatory disorders, cancer, lungclots, coronary problems, and braintumors.

SODIUM IODIDE CRYSTAL_ CONVERTS GAMMA-RAYS

TO LIGHT FOR PMTDETECTION

source may be determined because gammarays originating on the left side of the organare caused to impact the left side of thescintillation crystal, etc. The crystal is quitelarge and covers an area about 10 inches ormore in diameter.

Behind the crystal are, perhaps, 19

photomultiplier tubes in an hexagonalarray. The light of the individual scintilla-tion is not collimated but spreads out to allof the 19 tubes. The location of the point ofscintillation origin is obtained by analgorithm depending upon the individualsignals from each of the photomultipliers.Resolution is obtained in this manner toabout IA inch. Each scintillation is thencorrespondingly located by a single spot ona CRT. Counting is continued until severalhundred thousand counts are obtained andthe organ in question is satisfactorilydelineated. Fig. 14 is a reproduction ofsuch a scintigram. The particular advan-tage of the gamma camera over other tech-niques such as the CAT scanner (describedbelow) is that the gamma camera providesfunctional information. For example: Tc-99m polyphosphate is used to reveal bonediseases; 1231 is used in thyroid studies;127Xe is inhaled to provide information onlung ventilation.

The photomultiplier business resultingfrom the proliferation of gamma camerasthroughout the major U.S. hospitals hasbeen quite significant because of thenumber of tubes per camera-in some

Fig. 14These scintigrams were obtained by Lancaster GeneralHospital using a gamma camera. The scintiphoto on theleft shows multiple emboli in the right lung; photo on rightshows lungs after clearing. The isotope Technitium-99mwas used to tag Albumin microspheres-a colloidal form ofthe albumin protein, with particles ranging in size from 2 to50 microns. These particles are injected into thebloodstream and are fi'tered and trapped in the lungcapillary bed; the scan can then determine those areaswhere the capillary bed is intact. Areas of diminished bloodflow show as "cold spots."

25

cases as many as 91. In 1977, RCAdelivered many thousand 2 -inch and 3 -inchend -on photomultipliers for this purpose.The business has been profitable and arewarding humanitarian effort.

The CAT scanner is another relatively newmedical instrument that is making a majorimpact on photomultiplier sales.

The Computerized Axial Tomographic(CAT) scanner was introduced to thiscountry in 1973.3° The device uses a pencilor fan -beam of X-rays which rotatesaround the patient providing X-raytransmission data from many directions. Ascintillator coupled to a photomultiplierdetects the transmitted beam -as anaverage photomultiplier current ratherthan by a pulse count -and a computerstores and computes the cross sectiondensity variation of the patient's torso orskull. The photomultipliers are 1/2 -inch and3/4 -inch end -on tubes which couple to thescintillator, commonly BGO (bismuth ger-manate). The CAT scanner is being sold tomajor hospital units at prices exceeding$500K. Each unit may be equipped with600 or more photomultipliers. The demandfor these small photomultipliersaccelerated rapidly and facilities in Lan-caster were expanded to manufacture tubesat a high rate. Recently, the demand hasdropped because of governmental concernover increased hospitalization costs.

Conclusion

RCA's investment in the photomultiplierbusiness began with Zworykin's interest inutilizing the tube as a pickup for soundmovies. That business did not materialize,but a small specialized business did developin scientific applications, such asastronomy and spectroscopy. RCA main-tained research and development programsthroughout the years since 1940 that havekept RCA's photomultiplier productsdiversified and among the best in theworld. At the outset, sales and designefforts were modest but self-supportingand profitable. Competition was minimal.

Unpredicted markets developed --radarjammers, headlight dimmers, scintillationcounters, and CAT scanners -which great-ly stimulated the development of newtypes, sales, and competition. Some ofRCA's present competitors have a broadrange of photomultiplier product. Thesecompetitors include Philips (and theirvarious subsidiaries), EMI, HamamatsuTV, DuMont, Centronic and ELORG (an

entry from the Soviet Union). There arealso those who supply specialphotomultipliers: Galileo, EMR, Varian,ITT, Photoelectric, and SRC.

Today, RCA competes on a broad front fora very diversified photomultiplier business.The photomultiplier sales were negligiblein the early 40s. By 1956, the annual saleswere about $0.6M. RCA's presentphotomultiplier business is very large bycomparison.

The photomultiplier is a good example ofRCA's technology base paying off eventhough no major business opportunitieswere anticipated at the outset. Withpatience and cultivation, perhaps othersizable and unforeseen markets woulddevelop from some of RCA's smaller butprofitable areas of technical superiority.

ReferencesI. Bruining, H.; Physics and applications of secondary electron

emission. (McGraw-Hill Book Co., Inc.; 1954).

2. Kollath, R.; "The secondary emission of solid bodies.Physik. Z.. Vol. 38 (1937) pp. 202-224.

3. Simon. R.E. and Williams, B.F.; "Secondary -electron emis-sion," IEEE Transactions on Nuclear Science. Vol. NS -15(1968) pp. 166-170.

4. Farnsworth, P.T.; "Television by electron image scanning."J. Franklin Inst.. Vol. 218. (1934) pp. 411444.

5. Weiss, Georg; "Ober sekundarelektronen-vervielfacher,"Zeitschrtft fur Technisch Physik. Vol. 17, (1936) pp. 623-629.

6. lams, H.E. and Salzberg. B.; "The secondary emissionphototube." Proc. I.R.E. Vol. 23 (1935) p. 55.

7. Zworykin, V.K.; Morton. G.A.: and Maher, L.; "Thesecondary emission multiplier -a new electronic device."Proc. I . R. E.. Vol. 24 (1936) pp. 351-375.

8. Pierce. J.R.; "Electron -multiplier design," Bell LaboratoriesRecord. Vol. 16 (1938) pp. 305-309.

9. Rajchman, J.A.; "Le courant residuel dans les multiplicateurs&electrons electrostatiques," These. L'Ecole PolytechniqueFederate (Zurich; 1938). Also, Zworykin, V.K. and Rajchman,J.A.; "The electrostatic electron multiplier," Proc. I. R.E..Vol. 27 (1939) pp. 558-566.

10. Rajchman, J.A. and Snyder. R.L.; "An electrically -focussedmultiplier phototube," Electronics. Vol. 13 (Dec 1940) pp. 20-23 and 58-60.

I I. Janes, R.B. and Glover, A.M.; "Recent developments inphototubes," RCA Review Vol. 6 (1941) pp. 43-54. Also,Glover. A.M.; "A review of the development of sensitivephototubes," Proc. I . R. E.. Vol. 29 (1941) pp. 413-423.

12. Gorlich, P.; "Ober zusammengesetzte, durchsichtigePhotokathoden," Z. Physik. Vol. 101 (1936) p. 335.

13. Wood. F.B. (Editor); "Astronomical photoelectricphotometry" (American Association for the Advancement ofScience; Washington. D.C.; 1955).

14. Kron, G.E.; "Application of the multiplier phototube toastronomical photoelectric photometry." Ap. J.. Vol. 103(1946) p. 326.

15. Sweet, M.H.; "Logarithmic photomultiplier tube photo-meter," JOSA. Vol. 37 (1947) p. 432.

16. Kallmann, H.; Natur u Technik (Jul 1947).

17. Coltman, J.W. and MarsLall, F.H.; "A photomultiplierradiation detector," Phys. Rev., Vol. 72 (1947) p. 582.

18. Hofstadter, R.; "Alkali halide scintillation counters," Pkvs.Rev.. Vol. 74 (1948) p. 100.

19. Engstrom. R.W.; "Multiplier phototube characteristics:Application to low light levels,"JOSA.Vol. 37 (1947) p. 420.

20. Morton, G.A.; Matheson, R.M.; and Greenblatt. M.H.;"Design of photomultipliers for the sub-millimicrosecondregion." IRE Trans. on Nuclear Science, Vol. NS -5 (Dec1958) p. 98.

21. Sommer, A.H.; Photoemissive materials (John Wiley &Sons; 1968).

22. Simon, R.E.; Research in electron emission from semicon-ductors. Quarterly Report, Contract DA 36 -039 -AMC -02221 (E) (1%3).

23. Scheer, J.J. and van Laar, J.; "GaAs -Cs: A new type ofphotoemitter," Solid State Communications Vol. 3 (1965)pp. 189-193.

24. Morton. G.A.; Smith, H. M.; and Krall, H.R.; "Pulse -heightresolution of High -Gain First -Dynode Photomultiplier,"Amt. Phys. Letters. Vol. 13(1968) p. 356. Also, Kral!, H.R.;Helvy, F.A.; and Persyk, D.E.; "Recent developments inGaP(Cs)-dynode photomultiplier," IEEE Trans. on NuclearScience. Vol. NS -17 (1970) pp. 71-74.

25. Persyk. D.E.; "New photomultiplier detectors for laserapplications," Laser J. (Nov -Dec 1969). Also. Krall, H.R.and Persyk, D.E.; "Recent work on fast photomultipliersutilizing GaP(Cs) dynodes." IEEE Trans. NuclearScience, Vol. NS -19 (1972) pp. 45-49.

26. Williams, B.F.; and Tietjen, J.J.; "Current status of negativeelectron affinity devices." Proc. IEEE. Vol. 59 (1971) pp.1489-1497.

27. Persyk, D.E. and Hughes, F.R.; "Photomultipliers withGalnAs photocathodes for use in the near infrared," Proc.electro-optical systems design conference (1971).

28. Hughes, F.R.; Savoye. E.D.; and Thoman, D.L.;"Applications of negative electron affinity materials toimaging devices," J. of Electronic Materials, Vol. 3. No. I(1974) pp. 9-23. Also, Savoye, E.D.; Williams, B.F.; andHughes. F.R.; "Near -infrared and low -light -level imagingusing negative -electron -affinity photocathodes," Proc. firstEuropean electron -optics markets and technology con-

ference. (Geneva; 13-15 Sep 1972); published by IPC Scienceand Technology Press, pp. 200-204.

29. Anger. H.O.; "Scintillation camera" RSI, Vol. 29 (1958) pp.27-33. Also, Wagner. H.N.. Jr.; Nuclear medicine(H PPublishing Co.; New York City; 1975).

30. See. for example: Engstrom. R.W.; "Computerizedtomographic x-ray scanners." RCA Engineer Vol. 22, No. 6(1977) pp. 18-20.

on

Reprint RE -24-1-2Final manuscript received March 22, 1978.

Ralph Engstrom, as staff consultant, is in-volved with most of the photosensitivedevices developed by RCA's Electro-Opticsand Devices activity. Since joining RCA in1941, Dr. Engstrom's work on such deviceshas included photomultipliers, image tubes,and camera tubes.

Contact him at:Electro-Optics and DevicesSolid State DivisionLancaster, Pa.Ext. 2503

26

Optical video disc for very largedigital memories

R.F. Kenville

The need to store and retrieve largequantities of tv programming at modestcost has generated the requirement for aneconomical, compact, high-speed massdata -storage medium. Significant im-provements in the state of the art in massstorage have resulted from the develop-ment of video disc systems throughout theworld. Developmental systems typicallyuse 12 -inch discs to store approximately 30minutes of tv video information at a datarate of about 5 MHz. Information packetsapproximately one micrometer in diameterare stored serially in tracks with a pitch ofapproximately 1.5 micrometers. Mostvideo -disc developments to date have beendirected at consumer -oriented systems,with emphasis on playback -only of un-alterable prerecorded discs. RCA,however, has developed two video discsystems, which employ completelydifferent techniques: capacitive pickupwith mechanical tracking; and opticalpickup with servo tracking. Table I com-pares the two systems.

Dick Kenville began working with advancesdata recording systems in 1959 when hejoined RCA. In his current position asManager of the Applied Physics Laboratoryin ATL, he directs advanced development ofoptical disc recorders, laser scanners,reconnaissance equipment, sensors, CCDsignal -processing subsystems, andphotovoltaic energy systems. His groupsupports both commercial and governmentproduct divisions with mechanical, thermal,and optical design and analysis.Contact him at:Applied Physics LaboratoryAdvanced Technology LaboratoriesGovernment Systems DivisionCamden, N.J.Ext. 3297

Reprint RE -24-1-10Final manuscript received April 27, 1978

The ultra -high packing density of the optical disc-up to5 x 1010 bits per disc side-makes it a prime candidate forlarge information -storage systems.

Table IComparison of two video disc systems. Optical system has better bandwidth and signal-to-noise qualities than capacitive system, but requires relatively complex tracking andfocusing systems.

ParameterHome

entertainmentOptical

video disc

Master recording methodMedia processingPlaybackBandwidthSignal-to-noise ratioR minPlaying timeStop actionTrackingFocus control

Electron beamMaster -stampMechanical icapacitive)3 MHz40 dB450Two hoursNoMechanical (grooves)None

Laser beamNoneLaser beam5 MHz to 30 MHz50 dB1800 (nominal)30 minutesYesServoed galvanometerServoed lens

Author Kenville and a prototype optical video disc merrory system.

27

CONSUMEROPTICAL VIDEOPLAYERS(NON RCA)

ELECTRONICNEWSPAPER

COMPUTERASSISTEDTRAINING

LOW COSTEQUIPMENT1011 BITS/DISC(DRUM REPLACE

LOW COSTDATADISTRIBUTION

PRESENT RCATECHNOLOGYTHRUST

RECORD/REPRODUCE1011 BITSSINGLE DISC

LIMITEDREPLICATIONCAPABILITY

RECORD/REPRODUCEJUKE BOX1013 BIT100 DISC

DIGITALDATARECORDERHIGH DATARATE

COMMERCIALVIDEORECORDER/REPRODUCER

REC100DISC

LAYINDISC

Fig. 1Low in cost relative to magnetic recording systems, optical video disc may find its best application in mass memory systems.Other applications are possible, however.

The capacitive pickup system wasdeveloped for home -entertainmentapplications. Its low-cost playback -onlyrecords are made by a master -stampprocess, with an electron beam used togenerate the disc master. This systemgenerates full -color programming (withsound), using a unique color encodingscheme that requires only 3 M Hz of video -signal bandwidth. A one -hour recording,with a 40 -dB signal-to-noise ratio, is storedon each side of the disc.

The RCA optical video disc (OVD) systemwas initially developed for use inbroadcast -studio applications. This systemrecords and plays back high -quality videowith a 5.3 -MHz bandwidth and a 50-d13signal-to-noise ratio. The extension ofOVD techniques to storing digital data, thesubject of this paper, is one of the moreinteresting applications of this technology.

Optical video discapplications

The optical disc has two primary advantagesover magnetic recording systems for digitaldata storage-the cost of the storage mediais low and the optical disc is more amenableto automatic information handling.

These advantages give the optical disc a bigedge in mass data storage systems and ourpresent technical effort is directed toward

013 No. 3DATA RETRIEVALSTAGING

00 No. 2DATARETRIEVAL

OD No. 1DATA STORAGE

CONTROL CONSOLEDATA INDEXINGRETRIEVAL LOCATOR

Fig. 2Archival data storage system is one potential application. Discs would be filed in cabinets(as magnetic discs now are stored in some systems) and retrieved by clerks as needed.

this type of application, as indicated in Fig.I. The principal objective of this effort is todemonstrate 10'1 bits of storage on a singledisc. With this basic capability, an archivalstorage system holding 1015 bits of data willbe possible.

The archival data -storage system shown inFig. 2 would function in a semiautomaticfashion. Clerks would pull discs from thelibrary for information retrieval, whilemonitoring a recording unit. Such a systemcould be housed in a 100-m2 room.

28

A more automatic system could use the"jukebox" configuration shown in Fig. 3.Here 10" bits could be stored on 100 discscontained within a single unit. Access toinformation would be less than 3 secondswith this arrangement. For additional on-line capability, several jukeboxes couldoperate in parallel under the control of ahost computer.

100 DISKSTORAGE MODULE

CARTRIDGEIN STORAGEPOSITION

ELECTRONICS

'Y' BALL SCREWDRIVE

RCA is not working on replication tech-niques for the optical disc at present, butreplication is a near -term possibility. Withthis added capability, the electronicnewspaper, computer -assisted training,and low-cost data -distribution applica-tions could be addressed. In theseapplications, the user would have a low-cost "play only" machine and prerecorded

'X' BALL SCREWAND DRIVE

POSITION SENSORSTRIP ASSEMBLY

LASERPLAYBACKDETECTOR

CARTRIDGE INCHANGER CARRIAGEPOSITION

,DISC CARTRIDGEIN READ POSITIONASSEMBLY

OPTICSPLATFORM

EXTREMECHANGEPOSITION(REF)

VIEWPORT

AIR INTAKE

Fig. 3"Jukebox" configuration operates more automatically than the system shown in Fig. 2. With100 discs stored i each self-contained unit, 10'3 bits of information could be stored with a 3 -second access time.

discs would be supplied. The disc would beplayed on an interactive basis under thecontrol of :he user. The inherent ability tostill -scan segments of the disc and toautomatically hop from segment to seg-ment is an advantage for this type ofapplication. High -rate digital datarecording and broadcast video recordingare other obvious uses for this technology.

OVD system descriptionA modulated laser beam records a string ofspots on a proprietary recording medium.

Fig. 4 shows how the system would work ineither analog or digital storage systems.The disc is placed on a turntable which issupported by a high -precision air bearing.Recording takes place when the disc isexposed to modulated laser radiation,which is passed through beam -formingoptics and directed towards the disc by thetrack mirror. The laser beam is focusedonto the disc to form a series of very smallspots, whose spatial relationships aredetermined by modulation.

The RCA OVD system uses a proprietarymaterial 12 for the recording/ playbackmedium. This medium can be deposited onseveral types of substrate; however, thevinyl substrate now used is similar to anungrooved phonograph record. Duringplayback the disc rotates at 1800 r/ min andstores the equivalent of 30 minutes of

PLAYBACKDETECTOR

OPTICS

RECORD DISC

TRANSLATIONSTAGE

Fig. 4System works by usingmodulated laser beam torecord a series of verysmall spots on the disc.For playback, laser poweris reduced, and reflectedmodulation is picked upby a photodetector. Servosystem maintains veryclosely fixed distancebetween focusing lensand disc surface.

29

programming when used as a videorecorder/ reproducer. A blank disc is in-serted onto the turntable and recorded live,using a modulated laser beam. Playbackcan take place immediately followingrecording or at any time thereafter byoperating the laser at lower power andusing a photodetector to pick up thereflected modulation.

Since the optical system has a very shortdepth of focus, constant repositioning ofthe focusing lens is required to keep thelaser spot focused onto the microscopicallyuneven disc surface. The focus servo main-tains a fixed distance between the focuslens and the disc surface by sensing theinstantaneous distance and then driving anobjective lens mounted on a speaker -typecoil.

The position of the laser spot on the disc isdetermined by the position of the trackingmirror. During recording, this position isdetermined by the motor -driven transla-tion stage. The control logic can move thetranslation stage so that either a spiral or acircular track can be traced.

Playback is done by reflecting low -powerlaser light off the spots and to a photodiode.

During a playback, the laser power isreduced by changing its operating mode,and the illumination level to the disc is heldconstant by the optical modulator. Lightreflected from the recorded disc passesback through the focus lens, trackingmirror, and optics to an avalanchephotodetector.

Here again, the translation stage is used todetermine what portion of the record isplayed back. The translation stage controlsthe readout laser's position on the disc tothe nominal track position. Since thetracks were not recorded as perfect circles,a small amount of disc runout occurs as aresult of removing and imperfectly replac-ing the disc. Finer tracking control is

obtained by a dither tracking servo, whichwobbles the tracking mirror to modulatethe position of the readout laser beam. Thiswobbling introduces modulation into thedetected return signal, and is used to closethe loop and to keep the optics aligned tothe center of the recorded track.

The recording and playback signal -

processing electronics determine thestorage format and the type of informationstored. Since the recorded informationconsists of a series of spots (determined by

laser modulation), information is coded byvariations in the local arrangement of thespots. In the case of analog video, informa-tion is impressed by an fm coding scheme.

Digital -data recordingUsing the optical video disc system forrecording digital information requires theoptimum selection of modulation, codingformat, and error detection and correction(EDAC) codes.

The information on the OVD is a two -levelsignal (binary coded)-recording is doneby using the "write" laser to removematerial from the disc. The recordedsignals are formed in a series of spots ofvarying length and spacing. Therefore,pulse coding must be used to store informa-tion on the disc efficiently, and a com-plementary form of decoding must be usedto recover the information duringplayback.

Fig. 5 is a block diagram of the generalsignal processing that would be used inrecording and playing back digital data onthe optical video disc. Serial input informa-tion must first be buffered to accommodatenonsynchronous data entry and then for-matted into appropriate data blocks. Thebuffering handles both asynchronous dataand data rates lower than the maximumrate of the OVD. Data blocking and theaddition of appropriate header informa-tion make it easier to organize the data intoconvenient segments for addressing anderror control.

The next step in the signal processing addserror detection and correction (EDAC)information. This step is most important indigital recording systems where bit errorrates (BER) of 10-7 and lower are desired.

The objective of EDAC is to recover theoriginal input data even though variousundesired changes have been made in thedata processes of recording, storing,and/ or reading the data. In the recordingmode, check bits are added to the data andthe resulting information is redistributedwith respect to itself. In some EDACsystems, a 1000 -to- I improvement in BERcan be achieved with check bits adding onlya 10% overhead.

The next step in the signal processingencodes the information for recording ontothe OVD. We are currently examiningseveral encoding formats for applicabilityto the OVD system. Some formats requirethe addition of timing information toensure proper decoding upon playback.The encoding portion of the system canalso be set up to feed the data into the discon parallel tracks. The number of tracksused here can be different than that used indemultiplexing and EDAC.

Retrieving information recorded on a discinvolves the inverse of the processing usedfor recording. Information read out of thetracks is first decoded to yield the basicinput information with EDAC bits. Timinginformation obtained during playback isfed back to the disc drive to adjust

OPTICAL

VIDEO DISC

DATA IN

DATA

ENCODER

RECORD

EDAC

0.

p

0'

INPUT/OUTPUT TIMINGBUFFER

AND

DATA

FORMATTER

DATA OUT

0 DATA

DECODER PLAYEDAC

4

4/4

TIMING

TRACKING CONTROL At

Fig. 5Signal processing includes buffer for nonsynchronous data. Error detection and correction(EDAC) step can reduce error rates by as much as a factor of 1000.

30

RELATIVE

OUTPUT

1.0

8

3 dB

6

6 dB

4

9 dB

THEORETICAL

0 MEASUREMENTS

10 15 20 25 30 35

FREQUENCY IMHz AT 8" MA I

Fig. 6Frequency response of the RCA optical video disc. Curve is for 1800 r/min.

rotational speed for optimum readout. TheEDAC circuitry checks the informationout of the decoder and corrects the errorswithin the capability of the EDAC code.Finally, the signals out of the EDACcircuitry are stripped of their EDAC bitsand fed to the multiplexer for recombina-tion into the original input data format.

Modulation and codingBefore selecting the optimum data formatfor high density, high data rate, and lowBER, certain characteristics of the basicrecord/ playback mechanism must bedetermined. Fig. 6 is a plot of relativeoutput vs. frequency for the RCA -proprietary optical video disc. The curve isfor a disc rotational speed of 30 r/ s. Higherspeeds will produce wider bandwidths, butnot necessarily with a one-to-one cor-respondence.

High signal-to-noise values are possiblewith the optical disc.

The signal-to-noise characteristics of theoptical video disc are also importantparameters to be factored into a systemdesign. Used as television signal recorders,present discs exhibit exceptionally highSI N of about 50 dB. These measurementswere made by using an fm recordingscheme, with the carrier at 9 MHz and avideo bandwidth of 6 MHz. Disc speed was30 r/ s, and the SI N was measured at thedemodulator output. Recent measure-

ments made of high -frequency tones havedemonstrated that the total informationcapacity of the disc is 40 dB SI N over a 30 -MHz bandwidth. The disc noise doesincrease at low frequencies, so it is notdesirable to use a modulation scheme thatrequires dc or low -frequency response.

Delay modulation presently appears to bethe best coding method for the optical disc.

In our work at RCA, we are investigatingcoding approaches to identify the optimumscheme. The encoding techniques thatappear most applicable are phase -shiftkeying, pulse -amplitude modulation -fm,multi -track nonreturn to zero, and delaymodulation. Although the scope of thisarticle does not permit thorough discussionof tradeoffs, it is appropriate to describesome of the important features of theleading candidate: delay modulation.

Delay modulation has been used in severalhigh -density magnetic recorders and ourpreliminary studies indicated that it will beoptimum for the optical disc as well. It usesa coding relationship whereby transitionsat the centers of bit cells are "Is" and theabsence of transitions at the centers are"Os." When a sequence of "Os" occur,transitions are made at the bit cell barrier.The principal features of delay modulationare that it requires only a half -cycle ofbandwidth per bit, it is self -clocking, andits dc response is low. Operating at 30 r/ s,

we expect to achieve a 50 -Mb/ s data ratewith an uncorrected BER of 110-5. Analysisand testing are currently being conductedto verify our selection.

We are concerned about developing theoptimum coding scheme because it willultimately lead to a system having thehighest bit -packing density and the lowestcost/ bit. We believe that eventually a costof 10-e cent/ bit is possible for the OVDstorage medium.

ConclusionThe ultra -high packing density, widebandwidth, and high signal-to-noisecapability of the RCA optical disc, coupledwith its potential for ease of handling andinstant recording and playback, have madethe system attractive for manyinformation -storage applications.

The user requirement that will lead theinitial development of digital OVD systemswill probably be for mass data storage. Wehave seen a number of future systemsrequiring 1015 bits of storage capacity. Ifwe use $10 as the disc cost, such arequirement would result in storage mediacost of $100,000. The equivalent cost for ahigh -density magnetic system would be atleast ten times higher. Thus, in a 10 '5 -bitstorage system, the savings in storagemedia alone justify the OVD approach.

For the mass data -storage application, weenvision a "jukebox" configuration thatwould contain 100 to 1000 discs on line. Inthis system, depending on the data trafficrequirements, there would be a number ofread/ write stations. The entire mass data -storage system could be configured in asquare room, each side 75 ft. long.

Our research is concentrating on how tomake best use of the disc's uniqueproperties of signal-to-noise ratio andpacking density. We see the high -data -rateOVD application paving the way for otherones, such as mass data archives, electronicnewspapers, expanded memory for mini -and microprocessors, as well as thetraditional role of the OVD as a televisiondevice.

ReferencesI. Bartolini, R.A.; Weakliem, H.A., Williams, B.F.; "Reviewand analysis of optical recording media," Optical Engineering.Vol. 15 SMar Apr 1976) pp. 99-108.

2. Bell, A.E.: Bartolini, R.A.; Bloom, A.; Spury, F.W.; OpticalSoc. of Am. Annual Meeting, Toronto, Ont. (Oct 1977).

31

Government -related programs inthe Solid State Division

E.J. Schmitt

The Government marketSolid-state devices for military and satellite -system use represent a very significant partof the total sales of the semiconductorindustry.

The U.S. semiconductor market in 1977was about $2.8 billion, with the Govern-ment market approximately $400 million(14%); the 1978 market promises to beeven higher. In the electro-optics area, theFederal market was about $76 million(about 35% of the U.S. market). Whilemuch of this market can be serviced bycommercial and industrial -grade products,a large portion still requires product that isscreened to custom specifications or thatmust be redesigned to meet specific missionrequirements. In some cases, entirely u-nique devices are needed, and the onlyrelationship between the Governmentdevice and a commercial/ industrialproduct is that a similar technology is usedin manufacture.

The Solid State Division (SSD) selectivelyundertakes the development of specialproducts for Government end use if 1) themarket represents a profitable enterprise or2) if the combination of the military marketneed and the spinoff of the technology intocommercial products together warrant areturn consistent with the investment inengineering and manufacturing costs.

In the past few years, military and spacehardware procurement budgets have beendeliberately reduced and, in addition, haveeroded as a result of inflation. It hasbecome more and more obvious to theFederal agencies that they must concen-trate their remaining financial resources byusing or adapting mainstream commercialtechnologies and products whenever possi-ble and by coordinating their requirementsso that tri-service or multi -agency fundingis used to develop specific technologies orclasses of devices. One consequence of the

Making special devices for the Government produces profitand technological spinoffs for commercial devices.

funding decline is that Governmentprograms have become more directed andapplications -oriented. Specifically, twonew approaches to funding have recentlybeen advocated by the Department ofDefense.

In R&D block funding, the Governmentcommits a major block of funds to onecompany for a few years instead ofsimultaneously pursuing two or threesmaller contracts with different companies.

This new concept (favored by the Navy)implies that a company desiring toparticipate in the Government R&Dmarket must make larger investments inthe initial development cycle of atechnology or device to show a clear, earlysuperiority over the competition. TypicalR&D block funding is in the range of $200kto $500k over a two -to three-year period.

The Government is also increasing itsemphasis on Manufacturing Methods andTechnology (MMT) Programs.

This is the next evolutionary step in the lifecycle of the new device or process afterfeasibility has been demonstrated in R&Dprograms. All services have been directedto significantly increase MMT fundingover the next five years, and the key to acompany's success in winning MMT con-tracts will be its ability to clearlydemonstrate a large return -on -investmentto the Government for their funding. Theelectronics part of the tri-service MMTbudget is presently about $15 million peryear. This level is expected to rise to $50million per year by 1981, and thus willpresent major opportunities for industryparticipation.

It is not always necessary for a company toparticipate in a Government -sponsoredR&D program from its very inception toqualify for the MMT programs. It is

possible that a company, because it desires

to keep a development proprietary forcommercial -market and patent -licensingreasons, will fund most of the R&D fromits own resources. Then, once it has es-tablished its commercial/ industrial marketbase, it may choose to expand its marketopening by soliciting Government MMTprograms to further develop the product ormanufacturing facility to qualify formilitary applications.

The Government has two basic reasons forinvesting in an M MT program. The first isan urgent, critical need within the Govern-ment for the device and/or process re-quired to make the device. Usually, theintent of the Government program is toaccelerate the introduction of the deviceinto production and field use. In situationsin which an industrial concern is alreadyinvesting in production, the Governmentwill frequently either cost -share or identifyadditional Government -related require-ments and create the funding necessary toaccelerate the preproduction phase. Muchof the funding of this nature is directedtoward producing higher -reliabilitydevices, radiation -hard devices, and partsthat meet unusual shock, hermeticity, andother environmental requirements.

The second reason for the Government'sinvestment in MMTs is the need for aproduct for which no comparable in-dustrial or commercial requirement exists.Typical electronic -component and sub-system programs that fall into this categoryare fuzes, seekers for missiles, and circuitsthat will operate reliably in severe nuclear -radiation environments.

Fortunately, MMT programs pay largereturns to both industry and government.Industry benefits because many of themanufacturing techniques developed aredirectly applicable to the industrial andcommercial product base. The Govern-ment benefits because their supplement to

32

industry's capital -facility investment is

highly leveraged and results in a very largereturn -on -investment to them. MMTprogram funding typically ranges from$500k to $1.5M over a two- to three-yearperiod.

Some of the more significant R&Dprograms, MMT programs, and productdevelopments presently underway in SSDand the Solid State Technology Center(SSTC) are described in succeedingsections. These programs can have signifi-cant commercial/ industrial product im-pact and technology spinoffs in the future.

COS/MOS semiconductorprograms

The principal Government agency in-volved in the selection and approval ofhigh -reliability (MIL -M 38510) parts is theDefense Electronics Supply Agency(DESC). However, concurrence of eachmilitary service and NASA is also solicited.

SSD is presently the major supplier ofhigh -reliability MIL -M 38510 Class ACOS/ MOS circuits, with twenty-sevendevice types presently on the Governmentqualified parts list (QPL). Additional cir-cuits are in the qualification cycle.

Among the more visible applications ofhigh -reliability COS/ MOS circuits aretheir use in NASA and Department ofDefense satellite systems such as Satcom,the Atmospheric Explorer Series, Nimbus,TIROS, the Defense MeteorologicalSatellite Program (DMSP), and theVoyager satellites, which will fly pastJupiter and Saturn. Other major Govern-ment application areas include fuzes,remote sensors, mines, and secure digitalcommunications systems.

Predicting the reliability of COS/MOS ICdevices is an area of great concern to themilitary user.

The Government maintains large staffs ofreliability R&D engineers at most agenciesto continually monitor device reliabilityand to provide R&D contract support toindustry in the development of newmethods for testing and assessing thereliability of electronic parts.

The Solid State Technology Center is

presently under contract to the MarshallSpace Flight Center to determine thevalidity of the presently used acceleratedtests (life tests run at elevated tempera-

tures, with and without electrical bias, inoperating and non -operating modes) forpredicting the reliability of COS/ MOSdevices. The problem being investigated isthe possible existence of new failure modesthat could arise as a result of acceleratedtesting but that are unrelated to failuremodes that would eventually occur when adevice is operated within its specifiedrating. In other words, how far canaccelerated testing be extended as a validmeans of determining normal end -of -operational -life failure characteristics.

The radiation hardening of COS/MOSdevices is extremely important in manyGovernment applications.

The radiation hardness of a device is

generally specified in two ways-total-dosesusceptability and transient -dose suscep-tability. Total -dose susceptability is

measured in rads (silicon) and is associatedprimarily with spacecraft applications inwhich cumulative radiation exposure levelsoccur over a period of months or years as aresult of natural radiation environmentsexisting around the earth, the planets, or asa result of solar activity. The transient -dosesusceptability is usually associated withapplications in which radiation fromnuclear bursts is involved.

As a result of considerable in-house R&Dand a joint program with the NASA JetPropulsion Laboratory, a radiation -hardened high -reliability COS/ MOSproduct capable of routinely achieving 2 x10' rads (Si) has been achieved in produc-tion. This hardness was accomplished byoptimizing the gate -oxide annealingmethod, a first -order effect for total -dosehardness in COS/ MOS devices.

In continuing R&D programs sponsoredby the Wright Paterson Air Force BaseMaterials Laboratory and the NavalResearch Laboratory, it now appears thatthe standard commercial RCA CD4000series COS/ MOS circuits can be hardenedto levels in excess of 106 rads (Si), and acomprehensive program is underway toachieve this capability, in productiondevices, in 1978. This COS/ MOS hardnesscapability is now seriously challenging the107 rads (Si) level of the bipolar devicespresently used in many systems.

Additional radiation -hardening studies arealso underway to characterize the hardnessof the new microprocessor and memorycircuits. The microprocessor circuit andsome of the peripheral circuits use the new

Closed COS/ MOS Cell (CCL) geometrywith a silicon gate structure.

These devices appear to be somewhat"softer" than the aluminum -gate devicesused in the CD4000 series, but recentprocessing modifications made under con-tract to the Naval Research Laboratoryand Sandia have resulted in significantimprovements in hardness. In addition,work is underway to increase the total dosehardness of the silicon -on -sapphire (SOS)COS/ MOS circuits used primarily for thehigh-speed random-access memories in themicroprocessor family. The SOS -type cir-cuits have a much greater capability tosurvive transient -dose radiation environ-ments (i.e.. SCR -type high -current latchupinduced by nuclear -burst radiation), andadditional R&D programs are underwayto optimize both processing techniques anddevice geometry to achieve the highestpossible transient -dose hardness level.

The Radiation Effects group at the NavalResearch Laboratory recently awardedRCA the first phase of a contract to studyand evaluate the impact of ionic impuritiesand structural defects on the radiationhardness of MOS devices, and to studytechniques for eliminating or modifyingthe impurities so that maximum radiationhardness can be achieved. This study will,in the first phase, evaluate bulk siliconstructures. In later phases, silicon -on -sapphire devices will be investigated. RCAalso has had government contracts toimprove the radiation hardness of linearcircuits.

While standard high -reliability devices fulfillthe majority of military COS/MOSapplications, an important class ofapplications requires custom circuits.

Military systems use custom circuitsbecause of serious cost or packaging con-straints. Essentially three options are nowavailable to military users for circuitdevelopment: the hand-crafted custom cir-cuit, a Universal or APAR Array, or amicroprocessor -based approach.

The hand-crafted design resuB in thesmallest functional circuits and is the mostcost effective if large -volume production isanticipated (i.e., typically 20,000 or morecircuits). Fig. 1 shows a typical hand-crafted circuit.

The Universal/ APAR Array, Fig. 2, hasthe lowest initial design cost, a short designand fabrication (and rework) cycle, and is

33

Fig. 1 (near right)A hand-crafted custom LSI general -processor circuit, the TCS-074. Thisgeneral-purpose unit is an 8 -bit slice of aself -aligned, silicon -gate, silicon -on -sapphire circuit capable of being linked withadditional GPU units to construct larger -word -size ALUs. It, in conjunction with theSOS TCS-075 ROM (1024 bits), the SOSTCS-093 (632 Gate Universal Array, shownin Fig. 2), and the CDP1821 (1024x1 SOSRAM), form the basic building blocks of afamily of high-speed COS/MOS/SOS cir-cuits that are being considered for use asprocessors in military applications. Circuitspeeds up to 50 MHz will be possible. Thethree TCS circuits were developed as part ofa Manufacturing Methods and Technologyprogram for Wright Paterson Air ForceBase.

Fig. 2 (top far right)Universal array using aself -aligned silicon -gateCOS/MOS device on asapphire substrate. This235x235 -mil chip, theTCC-093, has 632 gatesand can be fabricated ina package having up to64 pins.

Fig. 3 (bottom)A typical four -circuitCDP1802 microproces-sor system that, in ahigh -reliability andradiation -hardened ver-sion, will meet the re-quirements of manymilitary applications.

ROM

a 4 a a -al 4 4 411 I 1 111

11;7;1

..lit lU Eft trl trl L.f tmtu ti Lika

ADDR BUS

TPA

MRD

CEO

RAM

CLEAR WAIT

ADDR BUS

TPA

MRD

MWR

8BIT BIDIRECTIONAL DATA BUS

11- T

'qiiir;t11r,

N LINES, MRD

TPB

0

SCO, SC

INTERRUPT

DMA -IN, DMA -OUT

EFI- EF4

I/O

particularly applicable to small -quantityproduction or prototype equipment.

The microprocessor, Fig. 3, has the advan-tage of extreme versatility in permitting therapid introduction of design and/ or fieldchanges and of being adaptable to mul-tifunction system operations while usingthe same basic hardware.

In practice all three techniques are in wideuse, each being justified on the basis of itsparticular application and cost goals.

In 1974, the SSTC won a key MMTprogram for the development of customsilicon -on -sapphire circuits; the programpromises significant long-term benefits forboth user and manufacturer. The MMTprogram goals were the establishment of aproduction process for silicon -on -sapphire

LSI technology and the development ofthree LSI circuits to demonstratereproducibility and reliability. Theprocessing effort was successfully com-pleted in 1975, and three basic circuits of anew family of high-speed COS MOS SOSLSI devices were then developed. Two ofthe circuits are shown in Figs. I and 2.

The three circuits, plus the 1024 -bitCOS MOS SOS random access memory,can form the basic building blocks of ageneral-purpose processor with extremelyhigh performance capability. It could befast enough to effectively emulate otherprocessors for which software presentlyexists. This means that an SOS processorcan be designed as a pin -for -pin replace-ment, or as a functional replacement, forcomputers that are in existing systems,particularly in TTL-based systems, with a

DATA

CONTROL

considerable reduction in power consump-tion and equipment size. Some of thefuture major military systems that may beable to use this new SOS logic familyinclude the Advanced Minuteman Missile(MX), the Defense MeteorologicalSatellites, and the AF Global PositioningSystem.

Fuzes are a uniquely Government market.

While almost all military applications arecandidates for the custom -circuit ap-proach, the fuze area is one that highlightsmany of the unique requirements andconstraints faced by military users. Thehigh -volume fuzes (artillery, mortar,bomb, and mine applications, etc.) usuallyemploy hand-crafted custom circuitsbecause of the cost savings inherent in thehigher yields associated with the smallest

34

possible IC die size. However, since all fuzeprograms start as exploratory develop-ment programs, and since most areterminated early, the low-cost, fast turn-around of the universal array design ap-proach is attractive in the early develop-ment cycle.

Because physical space is limited in a fuze,hybrid circuits using IC chips are frequent-ly specified. Sometimes, because of theacceleration in the gun firing phase, plasticICs are specified since the encapsulationmaterial supports the wire -bond con-nections.

In the future, particularly in remote -setfuze applications, the flexibility of micro-processors will be desirable because onecircuit can sequentially provide thefunctions that now require threemechanical/electrical subsystems (i.e., safe

Wire -bonded hybrid

and arm, time of flight count, and detona-tion control). However, it is possible thatall three of the fuze -circuit design optionsdescribed above will be used at some timebetween the exploratory developmentphase and the final production phase.

Typical fuze programs using SSDCOS/ MOS devices include the Navy'sFMU I I7B bomb fuze and the 5 -inchcannon -launched guided projectile, the AFFMU 112 bomb fuze, and the Army'sStinger missile fuze.

Some special Government applications re-quire very dense packaging.

In most cases Government applications usestandard high -reliability COS / MOSdevices in the dual -in -line ceramic packageor the smaller ceramic flat package.However, many applications have only

Dual in-linceramic

zChip before bonding

Hi -ref trimetalplastic DIP

115=-11- _r:W1111

Tape hybrid

limited space available for electronics andso require very dense packaging. In theseapplications a ceramic substrate is usedwith either a beam lead or a wire -bondedchip.

In addition, two new approaches presentlybeing developed show good potential forfuture miniaturized -equipment applica-tion. One is the leadless flat pack, and theother is the tape -carrier trimetal inter-connect system. Fig. 4 shows the variouspackaging approaches; there is significantGovernment funding for the developmentof all of the approaches.

Ion -implantation research is an example ofGovernment funding for special re-quirements.

The Government is funding semiconductorprocessing programs so that higher -performance, lower -cost, and more -

Beam -leadtimetal device

Fig. 4Package types for government high -reliability circuits. Dual -in -line and flat -pack ceramicpackages are relatively standard, other types shown are the leadless flat -pack, high -reliability trimetal. and beam -leaded and wire -bonded hybrids.

LeadkisaInvertedmasa*

(top)

(bottom)

35

reliable, radiation -hard circuits will beavailable for its more sophisticated re-quirements. A key program presently beingconducted by SSTC for the MarshallSpace Flight Center involves the use of ion -implantation techniques for silicon -on -sapphire COS/ MOS circuits. Ion implan-tation is a process that introduces con-trolled numbers of impurity atoms into thesurface of a semiconductor substrate bybombarding it with ions in the keV to MeVenergy range. Fig. 5 is a photograph of anion -implanter system.

Until recently, most impurity doping ofbulk silicon and silicon -on -sapphire cir-cuits was done either by exposing thesilicon surface to the impurities containedin a dopant gas flowing past the substrate,or by applying a doped oxide to thesubstrate surface. A heating cycle was thenused to diffuse the impurities, typically

Fig. 6Beam -lead trimetal microbridge substrateusing two -level interconnect. Some of thetechnology used to produce this type ofcircuit resulted in the spinoff product shownin Fig. 7.

Fig. 5Ion -Implantation systemis being explored for usein SOS circuits. It shouldprovide better impuritydoping at high and lowlevels.

boron and phosphorus, into the bulk of thesubstrate.

This latter technique has two basiclimitations. For light impurity doping con-centrations (1 part in 106), the processvariables are such that precise control ofgas concentrations, flow rates, andtemperatures is difficult. For high dopantconcentrations (1 part in 104), the diffusionprofiles tend to spread laterally, resultingin limitations on device size and, conse-quently, the restriction of device operatingfrequency. Ion implantation avoids theseproblems.

Linear integrated circuitsPackaging improvements have produced animportant technology spinoff.

Because of the dense packaging re-quirements of the electronic subsystems on

PLATINUM

SILICON CHIP ( CONTACSILICIDETS

SILICON DIOXIDE CIRCUITELEMSILICON

ENT ONLIC

SILICON NITRIDE CHIP I(SEALING LAYER)

most strategic missiles, the radiation -harddielectrically isolated linear and digitalTTL circuits used have been developed ashermetic chips with either beam -lead orwire -bonding interconnections to ceramicsubstrates and PC boards. The hermeticcharacter of the chip and the corrosion -freecharacteristic of the gold -lead interconnectsystem on a microbridge substrate (i.e., thesubstrate printed wiring is also gold) resultsin a hybrid -circuit package that is free ofcorrosion under severe environmental con-ditions and that need not be sealed in alarger hermetic hybrid package.

The major military application for thistechnology is presently the Trident missilesystem, for which SSD is under contract toLockheed. Other potential applicationsinclude the advanced Air Force StrategicMissile System (MX) and the Site DefenseMissile and Radar Support Systems (i.e.,SPRINT/ Safeguard Class).

Fig. 6 shows the beam -lead circuitsmounted on a microbridge substrate.These circuits are unique in that there ispresently no commercial or industrial cir-cuit requirement of the same nature (i.e.,fabrication on a dielectrically isolated sub-strate). It has, however, been possible toadapt some of the processing, metalliza-tion, and passivation technology to thebulk silicon substrate devices that arewidely used in both military andcommercial/ industrial applications. Thistechnology spinoff has resulted in the RCAGold Chip plastic -packaged IC product,Fig. 7.

In 1976 the Navy Electronics SystemsCommand awarded SSD a $1.4 millioncontract to further improve the Gold Chip

TITANIUM

PLATINUM

GOLD INTERCONNECTION LEAD

CHEMICALLY VAPOR -DEPOSITEDPHOSPHORUS- SILICATE -GLASS(CV D -PSG)

Fig. 7"Gold Chip" plastic -packaged IC has a silicon-nitride-passivated trimetal structure. Devicesare useful in intermediate -reliability applications.

36

NEAT PIPE

ANODE(COLLECTOR)

LEAD

CERAMIC TO METALSEAL ENVELOPE

SILICON

WICK

GATE LEAO

CATHODE(EMITTER)

LEAD

Fig. 9Transcalent thyristor has very good heat -dissipationqualities because of its heat -pipe cooling. The gate lead mayalso be brought through the wall of the ceramic insulatorbetween the emitter and collector of the cevice by brazing afeedthrouqh to it.

plastic -packaged IC so that it could meetthe intermediate -level Class B military -reliability specification (0.005% failure rateat 60% confidence level). In addition, thecontract requires that the price of a devicebe no more than 20% higher than acomparable IC made using the standardaluminum metal, Si02 passivated, plastic -

Fig. 8Elements used In automated assembly ofthe high -reliability trimetal system. Fromtop: mounted, scribed, trimetal wafer(thinned and presliced) ready for automatedbonding to beam tape; ICs on copper -cladKapton film (beam) tape (inner -leadbonded): IC on metal frame; and final HiRelinjection -molded plastic IC.

/ ,t .diot '

Fig. 10Transcalent thyristors (left and rig -it) are rated at 400 A (rms). Devicein center is a competitive method of cooling high -current solid-statedevices in which a disc -shaped device (like the one in the foreground)is clamped between the large fins in the background. The two methodshave the same current capability when the case temperature and theheat -pipe temperature of the respective devices are both 100°C.

package approach. The steps taken tofulfill this goal involved the use of silicon -nitride junction passivation, titanium -platinum -gold metallization, and thedevelopment of automated assemblysystems. Fig. 8 shows the various elementsused in the automated, beam -tape, chip -carrier assembly.

In an attempt to develop as broad an areaof application as possible for the trimetalplastic -package IC system, the circuitsselected by the Navy for this contractincluded specific digital bipolar TTL cir-cuits, linear bipolar circuits, andCOS/ MOS devices. All of the circuits werealready on the Government Qualified PartsList (QPL) as aluminum metal, Si02passivated, ceramic -package devices. Themanufacturing and reliability studiesplanned as part of this contract will resultin quantitative comparison data betweenlarge numbers of trimetal high -reliabilityplastic -encapsulated parts and aluminum-metalized devices in ceramic packages, forwhich a large Government and industryreliability data base already exists.

Power devicesThe Government is a major user of powertransistors in a wide variety of applicationsincluding avionics equipment, missilesystems, communications sets, and powerconditioners.

SSD, under contractLaboratory, recentl!,

to the Draperdeveloped the

TA9107 radiation -hard power transistor.This device can operate in radiationenvironments with cumulative neutronflux levels to I x neutrons/ cm2 andgamma intensity to 2 x 108 rad (Si)/ second.

Heat -pipe cooling has improved the thermaldissipation abilities of power devices.

A new configuration called the transcalentsolid -slate power device promises to reducedevice size and weight and simplifysupplementary thermal dissipation equip-ment (e.g., fans, liquid coolants) an orderof magnitude over present, equivalentpower devices.

The transcalent device consists of a siliconwafer bonded to a heat pipe. It achieves itshigh thermal conductance by the evapora-tion of a heat -transfer liquid from thesilicon surface. The liquid is delivered tothe silicon surface by wicking action in aself-contained thermodynamic system. Across section of a transcalent thyristor isshown in Fig. 9; Fig. 10 compares presentdevices with their transcalent equivalents.

Development work for transcalentrectifiers, thyristors, and transistors hasbeen funded through R&D programs andM MT programs from the Army (MERDCand ERA DCOM) and the Navy (NADC).These devices have the potential for wideapplication in solid-state contactors,motor speed controls, dc/ac inverters, andac/dc converters.

37

Electro-optics

Government -sponsored R&D in siliconvidicon tube development for tv camerashas had a long history.

The RCA vidicon presently being used insystems such as Maverick, Walleye,Hobos, and Pave Strike, and in NASAsystems such as Apollo, Landsat and JPLplanetary satellite programs are results ofthis sponsorship.

RCA Electro-Optics and Devices (EO&D)is presently under contract to the ArmyMissile Command to develop a ruggedsilicon vidicon assembly for use in futureArmy missile systems. This tube is of metaland aluminum ceramic construction withthe silicon target bonded to the glassfaceplate. The gun is of stacked brazedceramic construction to minimize tube size,and the focus coil and deflection com-ponents are potted. This 1 -inch tube willoperate in the 0.71 to I.1 -micrometer

Ed Schmitt is Manager of GovernmentMarket Development, Southeast U.S., forthe Solid State Division. He has hadprevious experience as Manager of the GSDAdvanced Technology Laboratory-West,and as Project Manager for various com-mercial and defense computer systems.Contact him at:Government MarketingSolid State DivisionSomerville. New JerseyExt. 6143

spectral range and has an a ntibloomingcapability. Fig. 11 is a photograph of one ofthe first tubes developed under this con-tract.

Charge -coupled -device cameras have ad-vantages for reconnaissance and sur-veillance systems.

Recognizing the potential of small,monolithic structures for direct digitalreadout and possible remote monitoring ofthe image, the Army Night VisionLaboratory at Ft. Belvoir has becomeinterested in the charge -coupled device(CCD), a solid-state sensor with its ownintegral internal scanning system on thearray that, consequently, needs noelectron -beam scan assembly or associatedvacuum enclosure. The Night Vision Labhas given RCA EO&D a contract tomodify the commercial CCD imager so asto extend its detection capability into theinfrared region and to enhance its ability tooperate with covert illumination systems inreconnaissance and guided -missile deliverysystems.

Fig. 12 is a photograph of the present CCDimager, which has the capability of produc-ing the same picture quality as a con-ventional 2/3 -inch silicon vidicon. Othermilitary applications for the CCD imagerinclude aircraft gunfire recording systemsand star trackers.

In addition to optical imaging, the Electro-Optics and Devices activity is developingsilicon photodetectors, which find wide usein the military for the less sophisticated

Fig. 11Rugged silicon vidicon tubes are used inmany military tv surveillance and recon-naissance applications.

guided munitions with optical seekingcapability. The Army (ECOM) has fundeddevelopment work in silicon avalanchedetectors, the type of detector that iscompatible with the requirement for theArmy's cannon -launched guided pro-jectile, Copperhead.

Another area of Government need is thatof surveillance and security systems. Aprogram called BISS (Base IntrusionSecurity System) is currently a jointservices project; the purpose of the projectis to evaluate imaging and display productsand to define Government requirementsfor security systems. Commercial RCACCTV cameras and video display andswitching equipment are being evaluatedfor and adapted to the severe environmen-tal conditions encountered in military fieldoperations.

Government -supportedR&D programs in the future

The developmental areas described beloware candidates for future GovernmentR&D program support.

Sapphire materials-Reducing the cost ofsemiconductor substrates, particularly thesapphire substrates used in the manufac-ture of silicon -on -sapphire COS/ MOS cir-cuits, is a major goal in the semiconductorindustry. A new process, edge -pulled film -fed growth (EFG), is being developed inwhich the substrate material is pulled froman aluminum -oxide melt through a set ofdies as a flat ribbon rather than being

Fig. 12Silicon imaging device intended primarilyfor use in generating standard -interfaced525 -line television pictures. This self -scanned device uses an array of charge -coupled device (CCD) shift registers forphotosensing and readout. The device con-tains 512 vertical x 320 horizontal pixels(163,840 picture elements); chip size is only500 by 750 mils.

grown as a large cylindrical boule whichmust then be sliced into wafers andpolished. Fig. 13 shows a portion of theEFG process.

Automated processing systems-One ofthe major factors in the yield of LSI devicesis the initial defect density of the substrateand the additional defects that occur as thewafer proceeds through the processingcycle. The additional defects are caused bydust particles that settle on the surfaces,scratches from contact masks, residue fromphotoresists and oxides, metal im-perfections, and handling. Techniques andprocesses that minimize the sources ofdefects and damage are mandatory if LSIdevices are to be manufactured in thefuture with reasonable yields.

A generic name for a system presentlyunder study and development by SSTC forSSD is Production Monitoring Control.Key elements of this system include anenvironment with extremely low airborne -particle concentrations, automated wafermovement through photolithographicoperations, doping, and subsequentchemical processing steps; sensors tomeasure the electrical and physicalcharacteristics of the devices on the wafersas they proceed through processing; andcomparison of measured data to knownreference parameters contained in a com-puter data base.

The National Bureau of Standards isfunding RCA Laboratories and SSTC fordevelopment work to define a family of testcircuits that can be used to perform the in -process device evaluation. The techniqueswill permit pinpoint assessments of thespecific processing step (e.g., gas flow, filmthickness, material resistivity) that may becausing devices to deviate from theirprescribed values.

Advanced design-automation-Presently,most of the design -automation techniquesused in the industry consist of separateelemental computer programs forsimulating logic, reducing logic to artwork,and generating test patterns to exercise thelogic. A new area of development is theconsolidation of these programs withrespect to a common database and acommon data specification and commandlanguage. A system that can be used toextract data from the database in a formatcompatible with the requirements of eachservice program is required. A user of thisnew system should be able to specify, in amacro language, a description of the circuit

Fig. 13Sapphire ribbons being pulled from the meltduring edge -defined film -fed growthprocess. Sapphire ribbon will be used formaking substrates for silicon -on -sapphireintegrated circuits, which are a primecandidate for very -large-scale integration.

or logic net that he wishes to fabricate. Theoperating system should, in turn, developthe logic, test the logic, simulate andexercise the actual circuit parameters forthat particular configuration, evaluate thecircuit response, and transform the logicinto artwork. The operating system couldthen return from the artwork through areverse set of simulations to assure that thefinal artwork did indeed reflect the macrodefinitions developed in the initialspecification of the logic. Once this is

proven, the artwork would be transformedinto the final masks.

Another goal of this development is tocreate a sophisticated graphic system thatwill permit real-time interaction betweenthe software and the designer as he movesthe design through the development cycle.

Advanced lithographic techniques-Asdevices become smaller, and as line widthsapproach one micrometer, the mask aper-ture size approaches the diffraction limit ofthe light used to expose the mask. Conse-quently, the exposed patterns on the wafersurface lose their sharpness, and the

metallization and impurity diffusion stepsare less controlled.

A radically new technique designed toovercome this limitation is the use ofelectron -beam exposure techniques tomake the masks or to expose patterns onthe wafer. The new electron -beam resiststhat are being developed are much moresensitive and have considerably greaterresolution than resists sensitive to visiblelight. The SSTC recently began operatingan electron -beam mask -making systemand is already providing extremely preciseVLSI microprocessor and memory cir-cuits.

Additional areas for future Governmentsupport include: software systems to con-vert present artwork generators intogenerators capable of controlling theelectron -beam pattern, further develop-ment of electron -sensitive resists, and ul-traviolet and short -wavelength radiationsources for mask -alignment systems com-patible with the very small geometricpatterns made by the electron -beam ex-posure.

SummaryGovernment support of semiconductorR&D has been one of the main reasons forthe sophistication of the capability of theU.S. in military electronics technology.The support has also stimulated a vigorouscommercial semiconductor industry.Every evidence indicates that a strong andmutually beneficial program will continuebetween industry and the Government inthe future.

BibliographyI. Khajezadsh, H., and Rose, A.; "Advances in integrated

circuit reliability ," RCA Engineer, Vol. 22 No. 3 (Oct/ Nov1976)

2. Gallace, L., and Whelan, C.D.; "Accelerated testing ofCOS/ MOS integrated circuits," Report ST -6379, RCA SolidState Doision, Somerville, N.J. (Apr 1975)

3. "Manufacturing methods for silicon devices on insulatingsubstrates," Contract No. F336I-73-C-5043

4. Kessler. S.W.; Reed, R.E.; Shoemaker, H.; and Strater, K.;"Transcident solid state power devices," RCA Engineer. Vol.21 No. 6 (Apr/ May 1976)

5. Vincoff, M.N.; "Radiation resistance of the COS/ MOS CD4000A and CD 4000B series," 1CAN-6563, RCA Solid StateDivision, Somerville, N.J. (Oct 1976)

6. Jarl, R.B.; "Radiation hardness capability of RCA siliconpower transistors," AN -6320, RCA Solid State Division,Somerville, N.J. (Sep 1974)

7. "High -reliability low-cost integrated circuits," Dept. of theNavy, Contract N00039 -76-C-0240

8. Rodgers, R.L.; "Charge -coupled imager for 525 -line televi-sion," RCA Engineer, Vol. 20 No. I (Jun/Jul 1974)

9. Young, A.W.; "The CDP1802: a powerful 8 -bit CMOSmicroprocessor," RCA Engineer. Vol. 22 No. S (Feb/ Mar1977)

10. Berg)inan , R.H.; Aguikra, M.; and Skorup, G.E.; "MOSarray design: universal array, APAR, or custom," RCAEngsnerr. Vol. 21 No. I (Jun/Jul 1975)

Reprint F1E-24-1-13Final manuscript received February 15. 1978.

39

t- ""REY WEST

AUTEC technology at the tip of the tongue

Andros-largest of Bahama's picturesque Out Islands-is homefor more than 700 RCA Service Company personnel

who operate and maintain the U.S. Navy'sAtlantic Undersea Test and Evaluation Center

(AUTEC).*

GRANO BAHAMA ISLAND

AN DROSISLAND

AUTEC

GREAT ASACO

r'

ata.°NEW PROVIDENCE

GS

\

44, "G%

ELEUTHERA

TEST FACILITY

E. Erickson1R. KennedyiG. Virgin

AUTEC is a deep -water test facilityused to track surface ships, aircraft,submarines, and weapons precisely, tomeasure their underwater acousticprofiles, and to test and calibrate sonarsystems. These functions are per-formed in support of R&D projects aswell as Anti -Submarine Warfare (ASW)missions using three separateranges-the weapons range, theacoustic range, and the FORACSrange.**

RCA Service Company operates and maintains AUTECunder contract to the Naval Underwater Systems Center,Newport, R I

"FORACS-Fleet Operational Readiness AccuracyCheck Site

4

The AUTEC rangesThe AUTEC ranges are located about180 miles southeast of West PalmBeach on Andros Island, along the"tongue of the ocean," a unique 6,000 -foot deep trench. This area is ideallysuited for AUTEC's mission because itprovides low ambient noises, freedomfrom shipping lanes, and a year-roundmild climate. More specifica'ly, theranges are adjacent to the easternshore of Andros Island within the 20 -mile -wide, 100 -mile long tongue -of -the -ocean, surrounded by coral banks.The main base and five downrangesites are on Andros, along the westernborder of this basin.

....."" I \\

....

.......,, .--- \ ......

... VITA GI T ,--, - ...Ts I

(_..-

...-- i

)rreT A )C -0'- ''''',may -i)14-- ----

I N -- ,,Y1 . -

r \ ,- --

wEAPONS RANGE

The weapons range uses sonar, radar,and optical systems to track severaltargets in water and in air.

For in -water tracking, the weaponsrange uses a large array of bottom -moored hydrophones augmented by asmall array of hydrophones locatedfurther downrange. Acoustic pulsesemitted by target -mounted pingers arereceived by the various hydrophones,amplified, and transmitted via cable toshore -based terminals where they arerecorded, discriminated for targetidentification, and digitized fortransmission to the main base forprocessing and display. Targets aredistinguished from one another by us -

40

of the ocean

.-1

,- ACOUSTIC RANGE

CC BLDG

MAN BA

fORACS RANGE

d"'

ing unique frequencies and pulserepetition rates for each. Two com-puters at tne main base select optimumgroups of hydrophones for tracking,determine real-time position, andprocess post-test data.

In -air tracking is accomplished by tworadars operating in beacon or skinmode and equipped with closed-circuittelevision for initial acquisition Radarangle and range data are alsotransmitted, in digital form, to the ma nbase computers for real-time trackirgand post-test data reduction.

Five cine'heodolites produce filmswhich, when processed and read, alsoprovide post-test data.

The acoustic range can accuratelyprofile the noise signature of a ship.

The acoustic range consists of a

bottom -moored hydrophone array,backed -up ty a computerized dataacquisition, processing, and displaysystem Tie underwater array includesnoise -monitoring hydrophonesmounted along a vertical cable as wel-as tracking hydrophones mountedho.-izontally along tracking arms at thebase of the main array. Acoustic andtracking data are transmitted via cablefrom the array to tne main base.

A tracking system carried aboard thetest vessel uses a pinger and otherelectronics to keep the test vessel on an

5

4

3

6

1. The tongue of the ocean-a mile -deep trenchextending southward through the Bahamas about180 miles southeast of West Palm Beach, Florida-is an almost ideal spot for testing underseaweapons and technology. The Bahama islandchains Protect it from the turbulence of theAtlantic; it runs parallel to, and within a half mile of,tie easiern shore of the largest of the Bahama OutIslands-Andros: and it is relatively undisturbed byship traffic. 2. The three AUTEC test ranges-Weapons, Acoustic. and FORACS-border An-aros Island on the east; these ranges provide anideal a -ea for R&D and ASW (anti-submarinewarfare) tests conducted at AUTEC. 3. Aerialview of the main base at Andros Island in theBahamas, viewed from the south. Marine depart-ment and pier can be seen at right center.Operations Control Center is at lower right withrange support at lett-center. Housing, dining hallsand administration offices are located at far left.

4. CCC 3400 computer systems located at thecommand control building on Andros performreal-time and post-test processing tasks in supportof test of the AUTEC range. 5. The OperationsControl Center is the center for all testing andterminus of all data acquired by instrumentationsystems over the entire test range. 6. Thiscinetheodolite tracking camera is one of five usedon the \Aeapons range to film test events.

4l

10

7. A mobile target being loaded aboard AUTEC's 182 -foot vessel,the IX -306, in preparation for firing during an on -range test. In thebackground is a torpedo recovery vessel. 8.AUTEC's IX -306launch vessel has two torpedo tubes: one for surface and one forsubmerged launch capabilities. Additional electronics on boardalso enable this vessel to simulate a target. 9.RCA diverssurround a torpedo preparing to fasten a harness for helicoptertransport back to main base. 10. RCA divers performing routineinspection on deep -moored acoustic hydrophone. In addition toinspections and maintenance, divers recover torpedoes andprovide underwater photographic services. 11.AUTEC diversurfacing.

accurate course during a test run,which is necessary to determine ac-curately the level of the vessel'sradiated noise.

The acoustic range is used to evaluatenoise -quieting techniques and to es-tablish criteria for noise reduction.Data from this range have been used tostudy radiated noise directivity patternand cavitation, to analyze comparativedata collected by other facilities, and tosystematically collect and correlatenoises from ships of a particular class.

The FORACS range uses manydifferent sensors to measure perfor-mance of shipboard ASW, radar, ESM,and navigation systems.

The FORACS (Fleet OperationalReadiness Accuracy Check Site) rangeconsists of three optical trackingstations one-half mile offshore near thereef line, active and passive radartargets, ESM (electronic supportmeasures) targets, and one deep andtwo shallow sonar transducers capable

...111...

_41

of providing signals for testing activeand passive sonars. The sonartransducers are connected via un-derwater cable to the main base. Themission of the FORACS range is ac-complished in three phases: docksidecalibration and setup, on -range test,and post-test data reduction.

During the dockside phase, base per-sonnel mount a deck transit cross -leveled to the reference plane of theship's fire control system. They alsodetermine ship's centerline forward,mount a deck transit over thiscenterline, and transfer the transit tothe centerline aft for use during on -range tests. The gyrocompass settleerror is measured and parallax data onthe ship's systems are obtained. Theship's crew is thoroughly briefed ontest procedures. If the ship is a sub-marine, the periscope is calibrated.

On range, the ship travels a

predetermined course while 250 to 400data points (range and azimuth) arecollected. An optical marker on the

ship's bow is tracked by the threeshore -based tracking stations whilethe ship's crew use on -board equip-ment and the aft transit to track theirrespective targets. A series of marksare called and, on each mark, allstations record bearing and/or rangeinformation to their respective targets.This procedure is repeated until suf-ficient data, over a wide spectrum ofranges and bearings, are acquired.

After the on -range phase, the dataacquired is keypunched, prccessed,analyzed, and furnished via Navychannels to the vessel for action.

42

IIIIIIIITIFPF

Range -support facilitiesMain base houses range managementand is the nerve center for rangeoperations.

In addition to being the terminus for alltest and tracking information collectedby instrumentation at the downrangesites, the main base houses twoCDC3400 real-time computer systemswhich collect, process, and distributedata. To increase the real-timeprocessing availability of thesesystems, most MIS -type programs, in-cluding payroll, have been placed onthe RCA Spectra 70 in West PalmBeach. The associated peripheral andcommunication equipment needed tosupport test operations are alsolocated at main base.

Public works: 16,000 Megawatts ofelectricity and 40 million gallons offresh water.

RCA employees perform in all publicworks functicns in support of the mainbase and the five down -range sites,equivalent to a city in the United Stateswith a population of 1,000. Within thepublic works department are the civilengineering function, utilities, trans-portation, and work control groups.During the course of a year, 40,000,000gallons of fresh water are processedand 16,000,000 kW of electricity aregenerated.

Logistics: 52.000 meals each month,over 4 million pounds of supplies eachyear.

Logistical support for the AUTECranges is a continually varying require-ment filled by RCA personnel. Housingis provided for 820 U.S. Navy, Govern-ment, and RCA employees, and theirdependents, and hotel services for adaily average of 60 transient personnel.An average of 52,000 meals are servedeach month by the dining servicesdepartment. The supply departmenthandles an average of 4,290,000pounds of goods in the course of ayear. RCA personnel also operate thelounge and the retail store.

Fire, security, and medical facilities:around -the -clock operation.

A fire and security force provides forthe physical security of the Androsranges. This includes access control,visitor control, fire prevention, and afire fighting and guard capabilityaround the clock, 365 days per year.

A fully equipped medical dispensary,staffed with a registered residentphysician, nurse, and Navy medicalcorpsmen, provides for routine andemergency medical treatment forAUTEC personnel on Andros.

Marine operations: boats, divers, andrepair and calibration facilities.

RCA operates and maintains a fleet often vessels, up to 180 feet in length.These vessels launch and retrievetorpedo -type targets and weapons,provide logistics support to the down-range sites, and are used to transferpersonnel to and from test vessels atsea.

Within the framework of marineoperations is the range -support shopwhich provides diving services, elec-tronic equipment calibration and repairservices, marine electronics and radarmaintenance, and machine -shopcapabilities. In support of on -rangetesting, the range -support shop helpsthe range users prepare, check out,retrieve, maintain, and load torpedoesor targets.

Air operations: three helicopters and atwin turbo -prop aircraft

RCA owns the Fairchild FH 227 twinturbo -prop aircraft which daily fliesfrom West Palm Beach to AndrosIsland. Capable of carrying 44passengers in addition to a crew ofthree, th s aircraft converts readily tocarry cargo when needed to haultorpedoes back to West Palm Beach orto transport fresh foods and cargo toAndros Island.

RCA, through an aviation subcontract,operates and maintains three twinjethelicopters which are used to launchand recover torpedoes and performlogistics and personnel transferfunctions between main base anddown -range sites.

Torpedoes may be recovered byhelicopters using either of tworecovery methods: diver or diverless. Indiver recovery, AUTEC divers, who areRCA employees, are dropped from thehelicopter in the vicinity of the floatingtorpedo. They attach a harness to thetorpedo and are then lifted back aboardthe helicopter to return to main base.To prevent damage, the torpedo isdeposited on a trampoline.

43

13

12

44

14

In diverless recovery, the helicopterhovers at low altitude and lowers afunnel -shaped cage over the nose ofthe torpedo. (A spent torpedo floatsvertically with its nose above the sur-face of the water.) Compressed gas isreleased causing retaining rings tocapture and hold the torpedo within thecage. The cage is then pivoted to ahorizontal position and the entireassembly is raised out of the water andreturned to main base.

Mainland operations: overall projectdirection from West Palm Beach.

West Palm Beach, Florida, is the head-quarters site for overall projectmanagement, interfacing directly withthe U.S. Navy.

Test Planning Engineering, TechnicalServices, Purchasing, and MainlandTransportation functions are alsolocated in West Palm Beach. Personnelwithin these areas commute to AndrosIsland as required to plan, support, anddocument the on -range operations.

Financial administration and projectadministration functions are alsobased in West Palm Beach. TheFinance department operates andmaintains an RCA -owned Spectra70/46 computer system used to providefinancial, MIS, and support computerservices.

ConclusionUnderstandably, this brief descriptioncannot give all the facets of a projectsuch as AUTEC. The variety of theskills and the depth of RCA manage-ment required to maintain and operatea vital Navy test range of this size couldfill several volumes if described fully.However, the information presentedhere should provide a fresh perspectiveon one of RCA's exciting programs,being carried out in a beautiful setting,by dedicated people.

12. AUTEC boat working with sub-marine to transfer personnel at sea. 13.Loading of MK 48 torpedoes on RCA'sFH227 twin turbo -prop aircraft. 14. Aftera diverless recovery, an AUTEC helicoptertransports the torpedo within its retrievalcage directly to main base. RCA andaviation subcontract personnel were in-strumental in perfecting this safe, fastermethod of torpedo recovery.

A typical day on the rangeputting a submarine through its paces

Preparations

Before the submarine that will undergo tests arrives at Andros, the SchedulingOfficer at AUTEC has already interfaced and coordinated test requirements.Government representatives, program managers, and RCA test planningengineers then meet to verify the nature of the on -range test, range resourcesto be committed, personrel scheduling, and data desired from the test.

After the planning meeting, the test planning engineer writes an OperationsDirective (OD) which contains detailed information pertinent to the test. TheOD specifies arrival times on the range and establishes schedules for supportvessels, radars, hydrophones, communications networks, theodolites,helicopters, and targets needed for the test.

The submarine crew anc test personnel are fully briefed beforehand to verifythat they understand all phases of the test and what will be required of themfrom test commencement (COMEX) to test completion (FINEX).

Torpedoes to be used in the :ests have been loaded aboard the submarine atstateside Naval bases. While :his was being done, the Range Support Shop atAndros was busily preparing a mobile target-an instrumented device thatsimulates a submarine in maneuverability, speed, and acoustical character-istics and is shaped somewhat like a torpedo.

For this test, the mobile target "submarine" is transported to the pier andloaded aboard the IX -306, a 182 -foot vessel containing two torpedo tubes. (Insome tests, the target m'ght be carried to the heliport and attached, via aspecial launching harness, to one of the SH-3G helicopters for air launching.)

Countdown to COMEXT minus 24 hours ... check all instrumentation-radar, communi-

cation sonar, optical, computer, etc.

T minus 3 hours ... thoroughly exercise the entire instrumentationnet to be used for this test.

T minus 60 minutes ... all stations report manned and ready; begincountdown to first target launch.

T minus 30 minutes ... target launch vessel (the X-306) arrives onstation and prepares for first target launch.The submarine maintains its scheduled posi-tion, course, and speed (as outlined in theOperations Directive) and prepares to detecta running target and launch its torpedo.In the Operations Control Center, real-timetracking data is being displayed to the TestConductor, Range Safety Officer, ProgramManager, and Planning Engineers.

Plotboards, with input data from the real-timecomputers, are providing three-dimensionaltracks of all on -range participants. Radardata and In -water tracking data are displayedto the Range Safety Officer to enable him tomonitor course, speed, and depth of all participants to ensure range safety.

T minus 0 ... COMEX (commence testing) is announced by theTest Conductor.

The target is launched by the IX -306 to beginits preprogrammed run.

The testThe submarine searches for themobile target, which is nowsimulating another submarine. Thesubmarine acquires the target andlaunches a torpedo which searches,acquires, and tracks the targetthrough a number of predeterminedmaneuvers. At end -of -run, thesubmarine -launched torpedo and thetarget torpedo surface for recovery.The torpedo fired by the submarine isretrieved by a torpedo recoveryvessel. The mobile target is retrievedby helicopter and transported to mainbase for turnaround and preparationfor later tests.

After the testAs soon as the test is completed, theData Processing Departmentproduces an abbreviated datapackage and delivers it directly to thesubmar ne via a rendezvous vessel. Ata later date, the fully documented testdata and results are sent to the rangeuser.

Ray Kennedy, al left, joined RCA in 1972 as an Engineer on theFORACS Range. In 1974, he became the leader of FORACSand later was reassigned to Test Planning, where he wasresponsible for the coordination of ASW test programs. In hispresent position as Manager, FORACS Range, he is responsi-ble for ASW sensor performance measurements on Navalvessels of the United States and Allied Countries.

Ed Erickson, center, joined RCA in 1965 as a Technical Writerwith RCA Service Company in Cherry Hill, New Jersey andtransferred to the Computer System Division in Palm BeachGardens, Florida as Manager, Maintenance Documentation.He joined RCA AUTEC in 1975 as Manager, Technical Ser-vices and worked there until this year Recently, Mr. Ericksonleft RCA.

George Virgin, at right, joined RCA in 1968 as a Leader,Technical Writers with the RCA Management Services Projectin Huntsville, Alabama. In 1969, he transferred to the RCAAUTEC Project in West Palm Beach, Florida, where hepresently is Leader, Publications Services and responsible forproduction, control and publication of documentation.

Reprint RE 24-1-121Fina1 manuscript received April 17, 1978.

45

A computerized drawing controlsystem by and for the engineer

W.S. SepichIN.C. Lund

The Engineering Drawing InformationControl System (EDICS) is a computer -based system that was developed to helpprepare and revise Broadcast Systemsengineering drawings and to provide ac-curate, instantly available informationabout these drawings. It is expected toincrease the efficiency of managingproduct development and increaseproductivity by aiding in the documenta-tion of design information and by im-proving engineering and manufacturingaccess to information about drawings.

EDICS is a real-time terminal -based (time-sharing) program that stores informationabout product drawings such as:

complete drawing trees for all products;

status of drawings;

revision levels and dates of drawings;

codes to indicate if parts are standard,nonstandard, or custom-made;

production shop orders for products;

spare -part stock numbers;

labor, test, and material costs for com-ponents and assemblies, which may besummarized at any level;

codes for special handling of com-ponents where required; and

for components on printed wiringboards, the standard lead mountinglength and a key for automatically inser-table parts.

EDICS will print out parts lists either in abatch mode or directly at a terminal, andperhaps of most significance, EngineeringChange Packages, ECPs,' will be "written"by the engineer at a terminal.

The ECPs (Engineering Change Packages) used in BroadcastSystems are similar to ECNs (Engineering Change Notices) atmany other locations, except that a number of related drawingchanges may be entered on an E('P, as opposed to the commonpractice of limiting one drawing change per ECN.

EDICS, an on-line drawing information/control system, letsengineers find out about updatings and revisions instantly.The system saves engineering time and effort, and also has anumber of side benefits.

Historical backgroundSeveral years ago Broadcast Systems, withthe aid of a study performed by CorporateResearch and Engineering Staff,recognized the need for a modern com-puterized drawing control system for itsproducts. A committee was formed con-sisting of representatives from Engineer-ing, Operations Control, MIS, Parts &Accessories, Industrial Engineering,Materials, QC, Production Administra-tion, Test Engineering, and CentralEngineering. The committee was chairedby a member of Corporate Research andEngineering Staff and a final report wasissued in January of 1974. That reportrecommended a real-time terminal basedsystem as opposed to the batch systemsbeing used at many other RCA businessunits. The system was called EDICS, anacronym for Engineering Drawing Infor-mation Control System.

The EDICS system is a modification ofcommercially available software.

Because of the prohibitive cost of develop-ing the software for such a system, the idearemained in limbo until late 1974. At thattime, an outside software company, MitrolInc.,' was found to have a program con-taining many of the features desired forEDICS. Their program is called MIMS( Mitrol Industrial Management System).With the assistance of Corporate Researchand Engineering Staff, Mitrol was con-tracted to modify their software to performthe functions of EDICS, and place themodified program on the Cherry Hill370/ 168 V M / CMS computer. Actually,EDICS is a "module" that has been addedto the MIMS software, thus making bothEDICS and MIMS available to RCAusers. The detailed specificationdelineating the requirements and usage of

'Mitrol Inc., 1050 Waltham St., Lexington, MA. 02173

EDICS was written by the authors andfollowed closely with Mitrol. Since the firstpriority for EDICS was to benefit thedesign engineering organization, it wasimportant for people with design engineer-ing experience to write the specification.The benefits to manufacturing, materials,etc., although important, were neverthelessconsidered of secondary priority. Thesystem became operational in late 1976 andis presently being used in BroadcastSystems on a limited number of projects inorder to assess its cost effectiveness.

The system is operational now.

standard and commonlyused parts have been incorporated into theEDICS database, along with the productstructures, or drawing trees, for one largeproduct and several smaller products. Allearly releases of components have beenmade for one product using the computerto simplify this task. Output wastransmitted via magnetic tape for directinput into the Manufacturing computersystem.

The major distinctionsof EDICSThe system is on-line, rather than batch.

The EDICS system is, in many respects,similar to computer programs being usedin other RCA business units for storingdrawing information and generating partslists-CRP in GCS, MCS in MSR, andEMCE in AS. The difference betweenEDICS and these systems, however, issignificant; so much so that the writers feelEDICS can be classified as the next genera-tion of programs for controlling drawings.

An examination of EDICS vs the batchsystems used in other business units showsa similarity in basic software structure: i.e.,I) a drawing or document file, 2) a part file,

46

and 3) a product structure or breakdownfile. However, as presently configured, thereal-time EDICS system is designedprimarily to support an engineering/ designactivity, while in the writers opinion mostof the batch systems are of greatest benefitto a manufacturing/ production environ-ment. With these alternatives, the on-linesystem was chosen out of a desire :osupport the engineering organization. :nthe early stages of a design, and particular-ly when parts are being released tomanufacturing, it is imperative to keeptimely data about revision levels ofdrawings, and changes, additions, ordeletions to parts lists. This allowsengineering to have better control of the

drawing system for their programs and alsopermits rapid and accurate transmittal ofthis data to manufacturing. The produc-tion organization must be quickly advisedof changes so that the cut -in of thesechanges does not cause wasted effort andunnecessary expense.

The following advantages should explainwhy we chose an on-line system instead of abatch system:

Engineers will write ECPs on the terminal.

EDICS gives engineers immediate andaccurate information required for the

ECP, such as latest drawing revision levels.

Norman Lund, with Broadcast Systemssince 1969, has worked on the mechanicaldesign of television cameras and broadcastantennas. He is responsible for mechanicalstandards in the division, in addition to hisposition as EDICS Database Administrator.His interest in computer -aided designbegan with his earlier work on electronmicroscopes where he developed a com-puter program for spring design, and con-tinued later with his use of RCA's FiniteElement Analysis program to perform stressanalysis on antenna radomes.Contact him at:Broadcast Engineering StaffBroadcast SystemsCamden, N.J.Ext. 4941

Bill Sepich has been Manager of Engineer-ing Technical Support in BroadcastSystems since 1974. He has been responsi-ble for bringing a number of computer -aided systems into Broadcast engineering,such as the Applicon interactive graphicssystem, the EDICS software program, and aword-processing system. Prior to hisassignment in Broadcast he had been inMSR for 20 years, where his last positionwas Manager of Mechanical Engineering.Contact him at:Broadcast Engineering StaffBroadcast SystemsCamden, N.J.Ext. 2156

Authors Sepich (left) and Lund at an EDICS terminal. Wall charts explaining systemoperation are at rear.

pattern revision levels, production shoporders, and "where -used" data. Becausethis information is accurate, it eliminatesthe errors commonly made and theresulting ECP re -issues or rewrites re-quired. An additional benefit is that thedatabase will be automatically updated assoon as the ECP is approved ThereforeECPs can no longer be the cause ofdrawing information not being up-to-date,since it is now contained in the computer.

Engineers can create, revise, and print partslists rapidly at the terminals.

Although EDICS will input long parts listsby keypunching to save computer connecttime, verifying inputs, correcting the ever-present errors, adding components, andprinting is done at the terminal. Thisreduces the time cycle to manufacturingrelease and gets the finished parts lists toengineering in the shortest possible cycle.An additional advantage is that engineerscan build up parts lists and drawing trees atthe terminal in the initial design phase.

Information about drawings can be ob-tained instantly and accurately.

Anyone with access to a terminal canimmediately obtain current informationabout drawings from the drawing file, thepart file, or the product structure filewithout referring to a printout that may beseveral days old. This is particularly impor-tant to engineering and drafting manage-ment during the design and early manufac-turing cycle, when drawings are beingadded or revised daily. It is also an aid tomanufacturing, since they can determine ata terminal, without walking to the printroom, if one or more tracings have beenchanged to reflect the latest ECP.

Engineers can produce and revise costestimates rapidly and accurately on theterminal.

While other systems can perform cost -estimating functions in the batch mode, thetime cycle from input, through revisions tooutput. is not comparable. When changesare made to component or labor costs,where these are used throughout the

product, the effect on overall cost can beimmediately determined. This is especiallyuseful in early cost estimates, when manyiterations are required in an effort to meetproduct cost goals.

It's possible to get immediate, up-to-date,specialized reports (without programming).

The unusual flexibility of the EDICSsoftware does not limit the user to standard

47

Type -NumberFile

DrewlagFile

Pert Shop -Order

FileFile

ECP Historical Owg-number

so)PG -number Shop -order

Records Tills Outstrip Number

Owg-Type Unit-of-NueStatus Std.mtg-length

Rey -level's'

flatelel

Make/buyStd/Custom

Pattern Roos Auto -Insert

Cost

SW -number

Spcl-Symbol

ProductStructure

File

Assembly -Number

Rem -number

Qty !Numeric'Qty !Alpha'Non -Std Mtg Length

Non -Std Special Symbol

Shop

Item

File

Shop -Order

item -number

Quantity

Fig. 1All drawing and product -structure Information is stored in three EDICS files-the drawingfile, part/group file, and product -structure file. Fields in each file are shown in f.gure. Theother files are not used directly during design engineering.

requests or commands for printouts ofinformation contained in the database.Any type of data listing desired, selected topass certain criteria and sorted on anyfields, may be requested at the terminalusing simple, English -like language. If thedatabase structure must be modified, forexample to add new data fields, there is noneed to schedule a programmer to rewritethe program; the database administrator'does this in a few minutes directly on theterminal. Major changes, of course, wouldrequire program modification.

These reports have proven to be of im-measurable value. During product design,changes are so frequent that reportsproduced in the batch mode would besufficiently outdated to be of questionableusefulness. Also, if engineers and managerswere required to go through an MISactivity to obtain special reports and thenwait one to several days for the output, theysimply would not bother. This is one of thekey features that sets EDICS apart frombatch systems and makes it a useful tool toincrease productivity.

Commonly used reports have beenassigned one -word macro names, whichallow them to be called up with a one -wordcommand.

Examples of reports obtainable in thiscategory are:

'The database administrator is the person who has access to theentire database as opposed to limited access for general users-see section on database security.

lists of components with accumulatedquantities for building prototypes ofmodules (assembled printed wiringboards);

lists of standard or non-standard parts,used for design review;

lists of drawings still preliminary orhaving pre -assigned numbers (used bymanagement to monitor progress on aprogram and to assure completion ofdocumentation prior to manufacturingrelease);

lists of automatically -insertable parts(useful to manufacturing);

lists of drawings of a particular type(schematics, detail -assemblies, etc.) in agiven assembly or product; and

lists of parts having a certain specialsymbol that identifies those componentsor assemblies requiring special handlingin manufacturing.

Other reports can be prepared for specificneeds.

Here a series of commands must be enteredusing the system reference manual as aguide. Note, however, that this does notinvolve programming; furthermore, if afrequent need develops for such a report,the series of commands is simply given amacro name, thereafter becomingavailable for anyone's use.

Examples in this category that have beengenerated and used include:

a list of all drawings in Broadcast Stan-dards Books;

three separate lists (ICs, transistors, anddiodes) in the database sorted by typenumber;

a list of all electrical components on anew product for advance ordering byPurchasing;

lists of a particular type of component ona product such as ICs, crimped contacts,sockets, etc.; and

a list showing the count of electricalcomponents on selected parts lists to aidan engineering cost estimate.

Most of the foregoing were spur-of-the-moment needs; without instant results, thereports could not have served their in-tended purpose, that is to say, the userswould have proceeded to compile the databy hand at significant cost in time andschedule.

Structure of theEDICS database

All of the drawing information andproduct -structure data is stored in threeEDICS files called the drawing file, thepart/group file and the product -structurefile.

The drawing file contains informationabout drawings that are common to all ofthe parts or groups that are owned by adrawing. See Fig. I for a listing of the fieldsin this file.

The part/ group file contains data uniqueto the particular parts or groups ondrawings. Fig. I lists the fields in this file.

The product structure file may be referredto as the drawing family trees of allproducts manufactured by BroadcastSystems. For each product, it starts withthe top MI, and stores information relatingto a top -down breakdown of all drawingsused to manufacture the product. For eachpart/ group that owns components, i.e.,parts lists and detail assemblies, there is alink to each of its components. Additional-ly, each component has a link back to eachassembly that owns it. This is a significantfeature since a "Where -Used" request doesnot require a search through many partslists; the owning assembly is read directlyusing the product structure records.

In addition to the three major files, thereare also a type-numberfile for ECP month -to -date and year-to-date historical recordsfor a product, a shop -order file containingproduction shop order numbers, and a

48

All 10:

lCW

WailsECPs

1/W

Direct Accesslull MIME

Oils lase

Fig. 2What are the major EDICS functions? Thepossibilities are grouped in columns bymajor activity; many of the sub -activities areactually initiated from the commandenvironment, but are shown as occurring insequence for clarity.

Adiministrolor Only

Command Environment

I IWoolPt Oita

1-711

Prini

hrte Lists

All Nen

Edit Flies

1

Verily Files

kw 1Run

Ratified Files

I ilECP Eneineem is

CW

Amore, Chop

VerilyECP Flies

Print

ECP Fenn

CW

Ws ley EnvirennenI

Standard

Repots

Special Repels

Abandon (CPs

Oalilieu Access II Nana I Read Only CW-Codrelled Write R/W.Full Read/Writs

shop -item file listing the major assembliesand their quantities for a given shop order.See Fig. 1.

Database accessand securityOnly the Database Administrator hasgeneral access to the database via a uniqueID, which gives him complete READ andWRITE access.

All other users have complete READaccess to the database, but are limited intheir WRITE access. Examples of WRITEaccess capability for general users include:

creation of drawings, parts, and partslists, and changing them, provided theirstatus is not "final" (approved);

ECPs (but not ECP approvals)-this isthe only way that general users canchange final drawings, i.e., by writing anECP at the terminal and having it

approved ("executed") by the ECPAdministrator; andcost fields, stock numbers, and produc-tion shop -order numbers.

In order to allow for multiple user access,the program has been structured tominimize the time that a user is in theWRITE mode, since other users arenecessarily locked out of READ/ WRITEoperations during such periods.

An engineer wishing to create or changerecords first constructs a file with the

computer's editor (known as the CMSEditor) using standardized "Create" and"Change" requests. A "Verify" processthen checks for duplications, syntax errors,etc., and when the file is error -free it is

"RUN" into the database in a controlledWRITE mode. The time to perform theWRITE operation is thereby kept to aminimum since there is no need to wait forinteractive input from the user.

Fig. 2 shows the various activities that auser may perform after logging in toEDICS, together with the database accessrequired for each. For most READ re-quests the process is a simple matter oftyping a pre -defined one- or two -wordrequest (macro). This procedure has beenprinted on two large cards that are placedon the wakl at each EDICS terminal. Onecard describes the steps for logging on tothe system and the second card details themost frequently used requests. Nothingfurther is required. For ECPs, the use ofthe CMS Editor and Verify process isrequired. To simplify this, a third largecard is also placed at each terminal. Thiscard leads the engineer in step-by-stepinstructions through the ECP procedure.

In general, engineers need not concernthemselves with the creation of newdrawings, parts, and parts lists, althoughthey are able to do so if they wish to createpreliminary product trees for such pur-poses as cost estimating. Normally, thefunctions of creating new drawings, parts,

and parts lists are handled by the databaseadministrator.

Using EDICS inproduct designIn order to describe how EDICS maytypically be used, let's follow a productthrough its development cycle. The use ofEDICS on a new project is initiated by theentry of an outline of the drawing tree. Thisis done by entering pre -assigned partnumbers for major assemblies or sub-assemblies in their proper structuralrelationship. The tree indicates therelationships between parts and is useful inplanning a project well before any of theactual drawings are prepared. It will alsoaid in performing the first cost estimate,which will be refined throughout thedevelopment cycle.

As the product is developed, additionaldrawings and part/ groups will be added tothe tree as they are identified, and com-ponents will be added to subassembly partslists. For new drawings, the new drawingnumbers and part numbers will be entered.If an existing drawing is being used in a newdesign, its part/ group number need only beadded tc the new parts list; all fieldsassociated with that drawing or part/ groupwill already exist in the database.Preliminary parts lists are entered intoEDICS, one or more items at a time, asthey are developed. In preliminary form,these parts lists can be freely edited andeasily modified without using any controldocuments.

49

3417908

QUANTITY

504 503 502 501

UNIT

OfMEAS

ITEM

OR

SYMBOL

DRAWING

OR

SPECIFICATION

PART

OR

GROUP

DESCRIPTIONMTG

LENGTH

SIN

SYM

1

REVISION

2 PRELIM

50

R401R402R402R403R403R404R404R405T1

T1

T2T2T3T415T6TP1TP2TP3TP3TP4TP4TP5TP6TP7TP8TP9TP10TP11U1U1

U2U2U3U3

U4U4

99041334162499904139904139904139904139904019904133417166XFMR341716734171683417169341716934171653417165341696434169643416964341696434169643416964341696434169643416964341696434169643416964341696433337333333700333373633337333333733333372933337

225 RES FXD FILM 1000 OHMS 5% 1/4W105 RES FXD WM 4 OHMS 5% 34249 RES FXD FILM 10K OHMS 5% 1/4W202 RES FXD FILM 110 OHMS 5% 1/4W249 RES FXD FILM 10K OHMS 5% 1/4W208 RES FXD FILM 200 OHMS 5% 1/4W369 RES FXD FILM 5110 OHMS 1% 1/4W208 RES FXD FILM 200 OHMS 5% 1/4W

1 TRANSFORMERTRANSFORMER

1 TRANSFORMER1 TRANSFORMER1 TRANSFORMER1 TRANSFORMER1 TRANSFORMER1 TRANSFORMER2 JACK TEST SINGLE PC MOUNT RED2 JACK TEST SINGLE PC MOUNT RED4 JACK TEST SINGLE PC MOUNT BLU2 JACK TEST SINGLE PC MOUNT RED4 JACK TEST SINGLE PC MOUNT BLU2 JACK TEST SINGLE PC MOUNT RED2 JACK TEST SINGLE PC MOUNT RED2 JACK TEST SINGLE PC MOUNT RED2 JACK TEST SINGLE PC MOUNT RED2 JACK TEST SINGLE PC MOUNT RED4 JACK TEST SINGLE PC MOUNT4 JACK TEST SINGLE PC1 JACK TEST SINGL1 IC TIMER 8D

47 IC CMOS1 IC P

.4

.8

.4

.4

.4

.4

.4

.4

P

P

P

P

P

P

P

P

.3

.3

.3

.3

.3

.3

.3

.3

.3

Fig. 3Parts lists are typed out by computer on apreprinted form. Simple editing takes careof changes until list is signed off as final.Formal ECP procedure is then required.

TYPE NO:TK-47

ENGINEERING CHANGE PACKAGEDWG:

TITLE: CH DEMUR

DATE:REASON FOR CHANGE: DWG ERRENGINEERING S.O.: 871703TIME OF ECP: 08:07 06/28/78

PROD SO'S8030

AFFECTED DWGSPARTS LIST

APPROVALS:ENGINEER:ENG SUPV:

NEXT HIGHER MISSMI -570601 -AlMI -570601-B1

NEXT HIGHER PL'S3417071

REV

ECP NO'S: TK-47-81

ACTION REQUIREDFIELD BULL: NOTEST ENG: NOPKG DES: NOIB: NO

X/Y DWG NUMBER CURRENT REV NEW REV3417904 2 3

NAME PHONE DATE MATERIAL DISP CODE DISTRIBUTION1 NOT USABLE STD CODE:

PROD CONTR:OTHER:

2 REWORK3 USE AS IS OTHERS:4 SEE NOTE5 CHANGE QTY6 NEW PART7 DELIN ONLY

DRAWING(S) TO BE CHANGED AS FOLLOWS:

P/G: 3417904-501 CH DEMUR MI -570904

ADD ITEM 20, 3417092-18 QTY 1 INSULATOR SIL-PAD 4 PIN

ADD ITEM 21, 82244-101 QTY 4 NUT HEX 6-32 STL

ADD ITEM 22, 990106-109 QTY 4 SCREW PH CR 6-32X.375 STL

ADD ITEM 23, 999784-104 QTY 4 WASHER FLAT 06 STL

ADD ITEM 24, 8954869-76 QTY AR COMPOUND

CURRENT ITEM Q16i 3414712-1 QTTRANSISTOR 310-92 2N3CHANGE REF TOCHANGE DE

CU

MAT'L DISP

PROD D8SPDCUT IN

Fig. 4Engineering Change Package output istyped by computer at the terminal.

The data on part/ groups stored in EDICSwill eventually include the most recentmaterial cost from purchasing and anindication of the labor costs to assemblethe part and test the assembly. The es-timated cost of a part/group never usedbefore will be entered by engineering orcost estimating. Parts lists will include anindication of the costs attributable to eachitem totalized by the computer and anindication of the total cost of the assembly.EDICS will help the engineer maketradeoffs to obtain the required productperformance at the lowest cost. The ex-istence of readily retrievable preliminaryparts lists will help the cost -estimatingactivity prepare cost estimates for theproduct being developed. Separate fieldsare provided for these formal cost es-timates as opposed to the preliminary costestimates.

A parts list becomes final when theengineer signs the preliminary parts list,which had been typed by the computer on apreprinted form (Fig. 3). The engineer'sapproval for parts lists or any other draw-ing is entered into EDICS via a terminal,thereby changing its status from prelim tofinal. After the drawing status is final, thisdrawing may no longer be changed by asimple editing procedure. A formalengineering change package (ECP) must beprepared and approved. Since the prepara-tion and handling of ECPs is the key tocontrolling changes during the productionof a product, the procedure for preparingthese documents is described below.

ECP procedureThe ECP procedure is inherently interactivebetween the computer terminal and theengineer.

The contents of the computer's database isused to remind the engineer of thesituations that must be covered by theECP, and the engineer decides how toresolve each of these situations. Specifical-ly, when the engineer has decided on thechanges needed to achieve the desiredproduct performance, he identifies himselfthrough an on-line terminal and enters thenumber(s) of the drawing(s) to be changed.The computer responds with the products,MIs, next -higher PLs, and current produc-tion shop orders where this drawing isused. It also identifies the associateddrawings or parts lists for PWB drawings.For example, if the drawing is a PWB partslist, the associated drawings would be theschematic/ assembly and board detail,which includes photo masters, soldermasks, and marking information.

In response to computer prompting, theengineer enters the reason for change, andanswers questions as to whether fieldbulletins, test engineering, packing design,or instruction books are affected.

The engineer then indicates whichdrawings he will be changing. This, ineffect, reserves those drawings for his ECP.The computer responds with the newrevision levels of these drawings. If thechange is to a parts list, the engineer typesin the actual changes to be made.

With the above data, the computer cantype a formal version of the actual ECP(Fig. 4.) If the drawing to be changed isother than, or in addition to, a parts list, theengineer writes the instructions for theneeded change, by hand, in the spaceprovided on the form.

The engineer now obtains the necessaryapprovals for the ECP. When it is fullyapproved, it is checked by an ECPAdministrator, who enters a code signify-ing approval through a terminal, "ex-ecuting" the ECP. When the ECP affects aparts list, "execute" initiates the actualchange to that parts list stored in thedatabase. A new copy of this parts list canthen be obtained at the terminal on a pre-printed form. The drawing revision level inEDICS would be automatically increasedby the ECP approval. Changes to tracingsother than PLs would have to be made by adraftsman in accordance with the in-structions on the ECP. When the tracinghas been changed and approved, this infor-mation will also be entered into the EDICSdatabase.

If the preliminary ECP does not get ap-proved, the engineer deletes it from theEDICS database using the "abandon"command. The system will not accept asecond initiation of an ECP to a particulardrawing by a different engineer before thefirst is approved. EDICS provides thename of the engineer who reserved thedrawing to the second engineer.

Additional benefitsBesides the advantages of using EDICSduring product design, the system has anumber of "fallout" benefits:

EDICS can be an important aid for designreviews.

The data field that indicates whether acomponent is standard can be used inseveral ways. Not only does EDICSidentify whether a part is contained in theBroadcast Standards Books, it furtheridentifies parts defined as "RestrictedUsage" (high -cost components) and partsthat have been superseded by a differentdrawing number. Before conducting designreviews, the design review chairman canrequest, at a terminal, a listing of allnonstandard, restricted usage, orsuperseded parts in the assembly he will bereviewing. The designer must then justifyhis need to use these parts. The usagefrequency of particular classes of nonstan-

dard parts may indicate a need for furtherstandardization. This can be checked byrequesting, at the terminal, a count of howmany times parts are called for on partslists on aq products.

EDICS can also check nonstandardmounting lengths for printed wiring boards.

The listing of standard lead -mountinglengths for PWB-mounted axial lead com-ponents is used as a check against thedraftsman's layout. Parts lists arekeypunched together with the mountinglength used by the draftsman. The com-puter automatically flags the nonstandardones.

EDICS provides a better way to select partsfor spares.

During the course of new -product develop-ment, the engineer places an "X" in thisfield for all components he recommends tobe spared. When stocking takes place,D&SPD can use this as a guide, andreplace the "X" with their stock number.

Summary and conclusionIn the development and production stages,EDICS can provide a number of benefitssuch as:

improved management control ofprograms through the use of computer -generated status reports of drawings;

better control of product cost during theearly stages of design;

increased use of standard components byreadily identifying non-standard partusage;

improved availability of component costinformation to aid engineers in cost -performance tradeoffs; and

decreased errors in preparing parts listsand ECPs.

There is no question that EDICS will costmore to operate than existing batchsystems. However, the features availablefrom an on-line real-time system shouldhave significant value to Engineering insaving time, allowing a better -quality jobto be done, and in reducing some costs.Results obtained from EDICS to date aregratifying. However, many features such ascompiling component cost data have notyet been tried in a production environment.The computer costs, with the limited usageso far, have been approximately aspredicted. For 1978 it is planned to expandthe usage to cover another major product,in addition to the one presently beingprocessed. In this way additional ex-perience will be gained aimed at weighingthe benefits against a better projection ofcost.

Reprint RE -24-1-15Final manuscript received May 26. 1978.

51

Developing and managing largecomputer programs

R. W. Howery

The challenges and problems associatedwith managing the development of large,complex systems have become the focus ofincreasing attention in recent years. This isespecially true in the defense/ aerospaceindustry, where the need for the systems,the need for assurance that systems underlong-term development will actually per-form as required, and the potential forserious cost overruns are all critical.

A great deal of study and analysis of thismanagement process has yielded an assort-ment of management tools and techniquesto help maintain control and direction overdevelopment efforts. (Success in theapplication of these tools and techniquesgenerally shows a direct correlation withthe maturity and experience of themanagers responsible.)

The standard problems of managing amajor system development apply to aneven greater extent to large computer -program development tasks. As a relativelynew kid on the block and one that hasgrown "like Topsy," the computer -program development process looms as abehemoth, frightening in size and scope,and frustratingly difficult to understandand control.

Notwithstanding these well publicizeddifficulties, the fact remains that computerprogram development must be controlledand managed. And the truth is that largecomputer program development projectshave been managed with success, alongwith many small ones.

What's different aboutmanaging alarge software effort?The differences between developing smalland large computer programs are signifi-cant, but not as great as the similarities.

When software projects get very large, something must bedone to keep them from getting very inefficient. Thatsomething is management.

Quite simply, the development of largecomputer programs includes all of thetechnical and management tasks requiredin the development of a small computerprogram. The difference is in the size andcomplexity of the system, plus theadditional problem of managing largenumbers of people on a task that must becoordinated as if it were produced by asmall, well -run chief programmer team.

Unless good management is used, moreprogrammers means less efficiency.

Behavioral scientists have determined thatthe largest size team for coordinatedeffective work is nine members and thatteam efficiency falls with each added teammember.' This would imply that a largeteam of hundreds of programmers wouldaccomplish nothing. Since a number oflarge computer -program developmentshave been completed successfully,something must have been done to avoidthe team limitations described above. Thissomething is management, and it consistsof:

effective team organization;

formal division and organization oftasks;

use of a systems development approach;

standardization of design practices;

thorough technical and managementreview;

management visibility and control; and

timely decision for change.

Government and industry both realize thatsoftware management must be improved.

During the 1970s, the growing emphasis onimproving management of computer -program development projects has resulted

in a spate of articles and a number ofacademic and professional society con-ferences and seminars. Examples are theseries of seminars at the University ofSouthern California on "Modern Tech-niques for the Design and Construction ofReliable Software" and the three-phase"Software Management Conference"2sponsored by the American Institute ofAeronautics and Astronautics and theData Processing Management Associa-tion, with speakers from government andindustry. (The terms "software" and "com-puter program" are used interchangeablyin the computer -program development in-dustry.)

Because of its annual investment of about$3 billion in computer programs, theDepartment of Defense is also intimatelyinvolved. DoD is interested becausecomputer -program development has

become a greater factor than hardware inthe total weapon -systems costs of somesystems. Also, serious cost and scheduleoverruns have occurred on manycomputer -program development projects.Accordingly, DoD directives have beenaimed at increasing management visibilityinto the process for both government andindustry. In addition, DoD has sponsoreda number of software -management con-ferences to communicate the use and valueof new practices and techniques to allgovernment/ industry team members!'

Throughout this paper, the AEGIS Com-bat System computer -program develop-ment is used as an example since AEGIS isa major system (see box) comprising multi-ple systems with embedded computers, allintegrated by a central computer suite.Although the management techniquesdescribed are producing outstandingresults on AEGIS, they are applicable inprinciple and in practice to both large andsmall projects.

52

a large-scale software svo*c.rri

The AEGIS Ship Combat System

For the past several years Congress and DoD have beenworking to determine the size and makeup of the Navy'snew surface fleet. In support of this effort, the Navy hasestablished an AEGIS development plan to support the longsystem -development lead time and handle charging Com-bat System configurations with minimum impact on alreadydeveloped systems. The plan provides for unified andstandardized ship baseline systems for cruisers (CGN-42class) and destroyers (DDG-47 class), with potential forselected fleet modernization of current ship classes.

AEGIS Combat System development began during the1960s with a number of Navy studies on approaches tocountering the airborne missile threat of the 1980s andbeyond. Following an industry contract design competi-tion, RCA was awarded a contract in Decemoer 1969 todevelop the AEGIS Weapon System.

The AEGIS Ship Combat System development programimplementation has been set up in four major phases:

AIR SEARCH

AN /SPS-49 ---`111"RADAR SYSTEM--- - -WI TH ADT

C1:210SURFACE SEARCH

AN /SPS-55 = RADAR SYSTEMWITH ADT

14:4 ESAIDENTIFICATION

SYSTEMAN 'UPX 29

AN SLU 32

[ AOSG II 1

AN/SOR 19 ,TOWED ARRAY c UNDERWATER

SURVEILLANCE &_COMMUNICATIONS

SYSTEM

%111110C3AN/SOS 53ASONAR

NAVIGATIONSYSTEM

SUPPORTSYSTEMS

r 1

LINK j TSES14

1---tLINK

11

1) Concept development and system prototype for thebasic AEGIS anti -air warfare weapon system-completedin the early 1970s.

2) At -sea prototype tasting of this system-conductedduring 1974-1976.

3) Engineering development of the total AEGIS ShipCombat System-currently in progress.

4) Ship and system production of AEGIS destroyers andcruisers-underway April 1978.

The system

The AEGIS Ship Combat System consists of virtually everyfunctional combat element on a ship, excluding the hull,propulsion, and auxilary equipment, but including allsensors, communications equipments, and weapons. The24 principal elements illustrated in Fig. A indicate themultiplicity of functions and complexity of the overallCombat System.

1161

AN/UYK7 AN/UYK.20COMPUTER M MINICOMPUTER

UNDERWATERCOUNTERMEASURES

SYSTEM

WEAPONSCONTROL

SYSTEM MARK 1

LAMPSMARK

LAMPS MARK III HELO

LINK4A

AN/UYK-I9MINICOMPUTER

GUNWEAPONSYSTE M

HARPOONWEAPONSYSTEM

TOMAHAWKWEAPCNSYSTEM

UNDERNATERWEAPONSYSTiM

AN /SPO 9

MARK 45

5"/54 121

ifCANISTER

...hr HARPOONIIICANISTER

4-'"TOMAHAWK

121

PHALANX 121

AN/SPG-62 14)

rGMLS MARK 2612)

SMR20. ScIA 1

.1,,

#01111 121

L

r

TORPEDOMARK 06

121

OTS TORPEDOTUBES MARK 32

Fig. AComplexity of the AEGIS Combat System should be evident from this system diagram. There are 24 combat -system elements inall. (Don't try to follow all the abbreviations-the point of the diagram in this discussion is to show the complexity of the softwarework, and thus the good management system needed, for AEGIS and other complex software operations.)

53

Controlling this diversity of elements is a central CombatDirection System, which includes the Command and Deci-sion System for sensor control and tactical commandfunctions, and the Weapons Control System for schedulingand managing all weapon systems. The AN/SPY-1A radar isshown separately since it functions as the major ship sensorand provides missile tracking and control as an integral partof the anti -air warfare weapon system.

Immediately obvious in the figure is the proliferation ofcomputers, from the three centralized 4 -bay standard Navygeneral-purpose AN/UYK-7 computers to the ten individualAN/UYK-20 minicomputers and two AN/UYK-7 single -baycomputers applied in the various other system elements. Asa major ingredient of the system, the central computercomplex has been designed to support the very high systemavailability required, taking into account equipment andcomputer -program failure/repair/recovery characteristics.

The three 4 -bay AN/UYK-7 computers used in the centralcomputer suite are standardized. Each computer consistsof four bays containing four central processing units(CPUs), eight standard single -density memories, fourdouble -density memories, and four input/output con-trollers, providing a total core of 262,000 32 -bit words foreach 4 -bay computer. The standardization includesdouble -density memory locations, backplane wiring, andinput/output controller fast/slow quantities and locations.

The computer programsThe AEGIS Ship Combat System has a number of computerprograms embedded in the channelized sensor, weapons,and communications systems. The programs underdevelopment by RCA for Combat System EngineeringDevelopment are those which reside in the central com-puter suite. They include: 650,000 words of instruction anddata that are part of the tactical system; about 2 millionwords of disk -based Operational Readiness Test Systemtest routines, tables, and fault detection and isolationmessages; and about 1 million words of development andtest support computer programs. Fig. B illustrates therelative size of these computer programs.

The complexity of the five tactical computer programsranges from the AN/SPY-1A real-time radar controlprogram (where computer execution time is critical) to theCommand and Decision and Operational Readiness Test

2,000 K2,000 K

1,000 K

750 K

"D.

.7

iu

500 K

0 250 K

TacticalPrograms

Operational SupportReadiness P ograms

Test SystemData Base

Fig. BSheer size of the AEGIS software -development taskrequires special management techniques.

programs (where functional requirements result in criticalcore -allocation problems). The Weapons Control Systemand Fire Control System computer programs also havetime -critical real-time requirements in support of totalAEGIS Weapon System reaction time. Another dimensionof programming complexity is the interface complexity-program design complexity increases with the power of thenumber of interfaces, rather than linearly.

Development support programs include the OperationalReadiness Test System Data Base Generator and Analyzer;the AEGIS Tactical Utility System, which supports programdebug and unit test and builds tactical load tapes and disks;and the AEGIS Source Code Preprocessor, which providesindented source -code listings, narrative flow, indentednarrative, and a structured flowchart. Test -supportprograms include a data reduction and analysis system toreduce, analyze, and print test reports and graphs from thedata extraction tapes produced during tactical programtesting, and the Interface Simulator System, which providesscripted test scenarios for program testing in a computer -system -simulated environment.

54

The development processThe basis for understanding the computer -program development process for largesystems is a clear perception of how majorsystems are developed. Accordingly, thissection reviews, in summary form, thedevelopment sequence for a major system,and then relates this process to computer -program development. The system -

development sequence and computer -program development phases describedhere are based on AEGIS experience andformal Navy requirements.'

The system development sequence has sixsteps.

The development process starts with theinitial system concept development and isonly complete when the system is testedand operational. The six steps in thesystem -development sequence of a majorsystem are shown in Table I, with thecomputer -program development phaseshighlighted.

The project plan breaks the job down intomanageable tasks and assigns taskresponsibility to specific organizations.

System definition establishes performancerequirements for major system elements(for our AEGIS example, one of the majorsystem elements is the three-dimensionalelectronically scanned multiphased arrayradar) to support the total mission. Thesystem specification documents specificperformance and test requirements andallocates these requirements to the majorsystem element specifications.' This ele-ment specification further defines thesystem performance requirements andallocates them to the specific equipment,computer program, or human interfacespecifications.

During the element design process thetradeoff analysis is made (state of the art,cost, computer execution time, etc.)between allocating the specific functionalperformance requirements to equipment,computer program, or operator. Elementdesign includes development of the systemarchitecture and design of the algorithmsimplemented in the computer program.

Computer -program development is onlyone of the system -development phases, butit has eight phases of its own.

The first phase of computer -programdevelopment is program requirementsdefinition. Both performance and test re -

Table ISoftware development for large systems must be considered as a system -developmentsequence (left), of which computer -program development is only one phase of the program -development sequence.

System developmentsequence

Computer programdevelopment sequence

Project planning/ task allocation

System definition

Element design

Computer -program development

Element integration

System integration and test

Program requirements definition

Top-level design

Detailed design

Coding and debugging

Unit testing

Program integration

Multi -program integration

Acceptance test

ELSE

THEN

COMPUTEBIN NO.

COMPUTE BINCALC TERM

ELSE

THEN

COMPUTEBIN NO.

M7 ITTRF

COMPUTE BINCALC TERM

COMPUTEBIN NO.

MTPVXG = MTPCO

MTPVBI = MTPCI 'AZ

MTPVXG = MTPC3MTPVBI

MTPVB2 = MTPC2'EL

MTPVXG = I MTPVBI64'MTPVB2

Fig. 1Flowcharting is part of the computer -program design process. The AN/SPY-1A flowchartshave seven levels of indentation and provide a continuous flow of control through thecomplete subprogram. Coding and debugging are much faster when flowcharts are used.

quirements are documented in individualspecifications.'

Top-level design of the program includesdesign of the program architecture, thedata -base structure and the allocation offunctional requirements to both theprogram modules and the computerprocess units. This is beginning to look likea top -down design approach. Watch out! Itisn't this simple. System and computer -

program design are iterative creative tasks.The key to good design is not eliminationof iteration, but orderly review and im-provement with careful control of theprocess.

The "build -a -little, test -a -little" approachmakes for manageable -size programmingtasks.

For large programs it is not practical todevelop the total program all at once. This

55

would cause a large peak load on bothprogrammer and computer resources andresult in a massive integration problemwhen the results of the many programmingteams were brought together all at once.The build approach (build -a -little, test -a -little) has been established to solve thisproblem. A computer program build is afunctional and operational subset of thefinal computer program, with stubs for themissing procedures and data. The buildapproach allows the establishment ofmanageable -size tasks, supports early in-tegration with equipments and otherprograms, and provides early visibility intoprogram performance in critical high -riskareas.

The detailed design of the programmodules and data bases is done on a buildbasis. Detailed design includes flow-charting the program logic (Fig. 1) anddesigning the private and common databases, which are documented' andreviewed by system engineers and seniorprogrammers prior to release for coding.

Coding and debugging form the shortestand simplest phase of the developmentprocess because of the thoroughness of thedesign process.

Unit testing validates the module operationto the design requirements in the environ-ment of the executive and the commondata base. Test points, inserted at allprogram branch points, provide checks forassuring that all code is executed during thetest.

The next phase is program integration,which is performed on a build basis. Theintegration process starts with the programshell. (In AEGIS, with 4 -bay AN/ UYK-7computers, the shell is a 4 -bay version ofthe executive, with stubs for all tacticalapplication modules.) After this shellprogram checks out in the computer,groups or clusters of modules supporting aspecific functional capability are addedand checked out, again with stub (ordummy) modules filling in for missingmodules. Finally, the total program buildis tested with a simulator test program thatemulates the environment (external andequipment) and supports scripted test

scenarios similar to actual missions.

The computer -program test system coversboth element computer -program integra-tion and test and multi -program integra-tion, where message interfaces are verifiedand total weapons system functional per-

formance is evaluated. Builds of the in-dividual element programs are informallydemonstrated prior to delivery to multi-program integration and element integra-tion, which is conducted at the system testsite in parallel with multi -program integra-tion.

After the computer programs are complete,they must be put into the overall system.

This brings us back to the system -

development sequence. Element integra-tion interfaces the computer system withthe addition element equipment sub-systems (for example, radar signalprocessor, receiver, arrays, transmitter) ina methodical stepwise fashion similar tothe one used in the program integration.

Program acceptance test is performed onthe total program in a stand-alone environ-ment after the major portion of multi-program and element integration is com-plete to assure a stable baseline forprogram validation to the program perfor-mance requirements.

The final validation of the computerprograms is part of total system integrationand test. This covers functional perfor-mance testing, system load testing, andoperational evaluation against real -lifescenarios. In AEGIS, this takes place at aland -based test site.

Managing the processManaging the development of largeembedded computer systems and programsis both a system -development and acomputer -program -development task.

Success in this undertaking depends onestablishing a task structure and thenbuilding an organizational structure withappropriate business and technical skills toperform those tasks. In AEGIS, the tasksrange from highly technical tactical -program design in critical, real-time radarcontrol loops to maintenance and schedul-ing of programming production facilities.With a large, multi -disciplined team, majorproblems can arise in the areas of com-munication, visibility, and control. Tech-niques and practices to help the managersand developers meet these problems in-clude both project management and thesystem engineering and programming dis-ciplines. These aids apply in appropriatedegree to system and program develop-ments of all sizes and complexities.

As would be expected, the selection ofqualified team members is a major con-

tributor to success. A sound developmentapproach and good tools will not make itwith an unskilled or inexperienced team.

OrganizationA large project such as AEGIS is first andlast a system -development task, withequipment and program development sup-porting the rest of the system's perfor-mance requirements. System engineering isemphasized throughout the system andcomputer -program development process.In addition, a strong project -managementorganization is required to plan, task,monitor, and control the total team effort.

The matrix organization is an effectiveorganizational structure for large systemdevelopments.

In the matrix organization, responsibilityfor each major element of the system isassigned to an individual project manager.Each element project manager carriesresponsibility for: cost; schedule andtechnical performance; tradeoff decisionsamong equipment, computer programs,and operator functional allocation; andsystem element integration and accep-tance. He implements the tasks usingspecialists in skill areas, where a stan-dardized development approach can beapplied to all elements of the total system.

A central system engineering activity (undera system engineering manager) ties the totalsystem design together.

Again, a central programming projectmanager assures that a standardizedcomputer -program development approachis used throughout the project. The systemtest and evaluation activity acts as anindependent test and evaluation agency forthe project, to certify test completeness atthe lower levels and to perform the finalsystem test and evaluation.

The element programming team providessupport functions such as the centralcomputer -program library, the programgeneration center, documentation andpublications activities, configurationmanagement, product assurance, and pro-ject management. The technical program-mer teams, which develop the programs,are structured into skill and taskresponsibilities. The element programdesign team acts as the interface withsystem engineering, produces the programdesign specification, and monitors theprogram -development and integrationactivities for technical integrity. Specially

56

trained senior program -integration teamsperform the multiprogram integrating ele-ment integration and support final systemintegration and test.

The tools and techniquesA number of management techniques andpractices have proved successful over themore than seven years of AEGIS systemdevelopment, including an increase in dis-ciplined management of computer -program development. These tools andtechniques can conveniently be organizedinto three groups (Table II): systemengineering techniques; modernprogramming practices; and projectmanagement tools.

The AEGIS project uses standard systemengineering techniques and some new onesdeveloped specially for AEGIS.

The following system -engineering tech-niques are among the most important onesused in AEGIS. Functional'.low diagramsand descriptions (F2132) show thefunctional flow of the system at differentlevels or tiers of system functional alloca-tion. For example, the tier -2 level showsthe functional allocation and interfacesamong the equipment, computerprograms, and operators, and shows themessage and signal flow through thesystem. System activity timing diagramsshow the time sequence of operationalscenarios through the system. Errorbudgeting allocates system errors tospecific parts of the system so that errorperformance can be managed.

Two computer models of the AEGISsystem have been set up for the evaluationand prediction of system performance.These systems (a high -accuracy, single -target simulation and a medium -accuracymultitarget simulation) allow verificationof algorithm designs before they are im-plemented in tactical computer programs.Interface definition is achieved in a formalinterface design specification; interfacesare controlled by requiring engineersresponsible for each side of an interface tosign off all changes.

Finally, to support communicationbetween system engineering andprogrammers, a systems definitionrequest response system has been set up sothat if the programmer doesn't understandthe performance requirement, he can statehis problem and the system engineer canrespond with a clarification or correction.Design reviews, principally by system

Table IITools of the trade. These managementtechniques are discussed in text.

System engine?ring techniques

Functional flow diagramsand descriptions

System activiti timing diagrams

Error budgeting

System computer modeling

Interface defirition

System definition requests

Design reviews

Modern programming practices

Computer -program development plan

Top-down/structured design

Program mocularity

Programming standardsStructured programmingDocumentationTesting methodsModule and data structuresNaming conventionsCoding and commenting

Code verification

Formal prob.em/ error reporting

Core/ time resource management

Project mancgement tools

Work -breakdown structure

Task work packages

Earned value system

Baseline/ cor figuration controls

engineering, verify that the program designand code meet the performance re-quirements and support the overall systemdesign.

"Modern programming practices" haveshown improving results as the program-ming community becomes experiencedwith the new disciplines.

The programming techniques listed belowhave been used successfully on a number ofprojects in addition to AEGIS. We haveincorporated all of these practices inAEGIS, with especially effective results insome areas. The programming plan shouldinclude a description of what programs arebeing developed and the who, how, when,and where of the development process.Top -down structured design allows orderlydevelopment with sequential addition offunctional capability and localization of

impact caused by changes. Programmodularity is a direct result of effectivestructured design. Programming standardssupport effective communication and adisciplined large team development-without standards, the programs wouldnot interface properly.

Structured walkthroughs of the detaileddesign by system engineers and programtop-level designers are a must prior toformal release to coding, with all correctiveaction spelled out in the written release tocoding. After coding is complete, peer orimmediate supervisor code verificationassures that the program is as designed. Aformal problem error reporting and cor-rective action system is necessary, with thedegree of formality and level of approvalincreasing at higher levels of integrationand test.

Most programs get in trouble by overrun-ning their allocated computer resources inorder to meet performance requirements.A critical management tool that is also amodern programming practice is computerresource management; that is, thebudgeting and continued estimating ofcore and execution time use as the programmatures.

Project -management tools are used topredict potential trouble spots.

Project management of computer -program development includes task alloca-tion to formal statements of work per aproject work breakdown structure. Costmanagement systems are used to providemonthly cost reporting, supported byschedule charting and an earned valuesystem for measurement. Work packagesdefine tasks several levels below the work -breakdown structure for high -resolutionmeasurement of work -completion statusand mole accurate measurement of earnedvalue. For example, completion status isevaluated for flowcharting by measuringprocedures flowcharted versus the totalestimated number of procedures required;code status is measured as the number oflines of error -free code compiled versus theestimated required code; and unit test ismeasured on number of functional threadstested versus the number in the module.Work measurement should not be used togenerate historical records, but shouldpredict trouble spots for corrective action.

Baseline and configuration control is

critical on a large program development.Baselines under control should include theequipment configuration, computer

57

resource allocations, interface designspecifications, program documentation, allcode, problem/ error reports and correctiveactions, and all test materials (testprograms, test scripts, data sets,

procedures, and test reports). Systemreview and baseline control boards shouldbe established at the working level tosupport the project Change Control Board.

Applying the toolsA good set of tools is not enough. New andeven excellent tools, practices, or tech-niques have a cultural lag between theirintroduction and effective use.

Two approaches seem to help shorten thiscultural lag: I) involving key teammembers in establishing practices to beused and the timing and methodology oftheir use; and 2) investing in user manualsand training in their effective use. A mostimportant management task in this area isto listen carefully to complaints aboutpractices or tools, and then take timelycorrective action.

People management, a major part of anymanagement task, is even more importanton a large development project. And sinceboth systems engineering and computerprogramming are highly individualistictasks, this aspect of management is themost difficult. Most seminars and con-ferences have touched only slightly on thismajor factor in team success, butWeinberg" provides a thorough -goingcoverage of the subject.

Reviews are necessary at all levels of design.

The management task is plan, task,monitor, and control. Accurate knowledgeof technical and cost status and timelysolution to problems are vital to goodmanagement results. Regular design andproject status reviews with carefullyplanned agendas aid this process. Thesereviews should cover specific subjects to beeffective; the general "cover anything thatcomes up" meeting may be fun, but doesn'tcontribute much to results. AEGIS has aseries of technical reviews that run throughthe whole development sequence and in-clude Navy project and user personnel,Navy technical laboratory personnel, RCAsystem engineers and computer systemscientists, plus programmers and computerscientists from the program designactivities. Project reviews range from ex-ecutive sessions involving senior Navyofficials down to weekly reviews of theprogramming activities by the responsible

58

programming project engineer. Weeklyproject reviews are held by the RCAProject Manager, providing a review ofeach element of the Combat System ap-proximately once each month. Formalfollowup of action items from all reviewsincreases the effectiveness of the reviewsystem.

Management information centers arevaluable as a means of increasing visibilityand improving communication.

AEGIS uses a number of informationcenters or work rooms. The AEGIS Com-puter Programs Development ControlCenter has posted on its walls currentconfiguration, schedule and core and timemanagement information, and problemaction followup reporting. More detailedschedule and manpower planning is postedin rooms maintained by the programmingteams. Special workrooms have posted thesystem engineering F2D2 diagrams, theCommand and Decision data -base tables,and other subjects where high visibility andcareful following are required.

Although much of the managementprocess involves oral reviews, discussionsand conversations, it is important to putmanagement direction in writing. Thereshould be written statements of work andtask assignments as well as documentedstandards, methods, design analysis, andspecifications. Programmers, systemsengineers, and test engineers should havedesign documents for all programs, in-cluding flowcharts and well annotatedstructured listings. Problem/ error reportsand corrective actions should have a for-mal documentation process. Good dis-tribution of these documents aids com-munication and makes future accuratereference to the material possible.

One of the most critical areas for decisionsis the program -development facilities; thatis, how many computers and peripheralsare required to support the developmenttasks. Since this requires money up front,the wrong decision is often made, saving oncomputers at the front end of the job anddriving the development costs up manytimes over the savings. AEGIS has set upfour program -development centers usingapproximately twenty single- and multi -bay computers: the Program GenerationCenter; the Computer Program Test Site;the Data Reduction Center; and the Com-bat System Engineering Development Site.

Operating and maintaining these centersrequires an experienced crew, which in-

cludes field engineers from the majorequipment suppliers and computer -program systems specialists to support theoperating systems.

Finally, management must take time toevaluate all of this and make timelydecisions. Problems solved promptlyalways cost less. This is especially true on alarge team where continuing openproblems grow in confusion and misdirec-tion. Managers must invest the time in themanagement process.

References

I. Horowitz, E. (Editor); Practical Strategies for DevelopingLarge Software Systems. Addison-Wesley Publishing Co.,(1975).

2. American Institute of Aeronautics and Astronautics, LosAngeles Section; Proceedings of Software ManagementConference: Phase 1, 1976; Phase 2, 1977; Phase 3, Winter1977-78.

3. Office of the Assistant Secretary of Defense; "Defense SystemSoftware Management Plan."

4. Department of Defense; Instruction 5000.29, "Management ofComputer Resources in Major Defense Systems."

5. Department of Defense; Instruction 5000.31. "Interim List ofDOD Approved High Order Programming Languages(HOL)."

6. Office of the Secretary of Defense; "Embedded ComputerResources and the DSARC Process."

7. Dept. of the Navy; SECNAVINST 3560.1, "Tactical digitalsystems documentation standards," (Aug 1974).

8. Dept. of Defense; MIL ST D 490, "Military standard specifica-tion practices," (Oct 1968).

9. Weinberg. G.; The Psychology of Computer Programming,Van Nostrand Reinhold (1971).

Reprint RE -24-1-16Final manuscript received May 1, 1978.

Dick Howery has held a number of engineer-ing management positions, mostly indefense -related areas. He was ProjectManager of the AEGIS OperationalReadiness Test System from 1971 to 1973,when he moved up to his present position ofProject Manager for AEGIS computer -program development.Contact him at:AEGIS StaffMissile and Surface RadarMoorestown, N.J.Ext. 2592

VLSI tools, technology, and trends

What to make? How to use? These questions have assumedequal importance with the traditional "How to make?" andhave pushed the software technologies of design andapplications to supplement the hardware technology ofsilicon.

Very Large Scale Integration of tens ofthousands components per package is

showing a profound qualitative impact onthe application of electronics to the needsof our society.

SSI to MSI to LSI to VLSIDuring the past fifteen years, integrationlevels have progressed from a few gates perchip to thousands of gates and tens ofthousands of components per device(Table 1). This remarkable progress is theresult of several technological advances:improved photolithographic processes arepermitting finer geometries; refinements inthe manufacturing process have resulted infewer defects per wafer; and innovativecircuit design and system organizationhave resulted in high functional density.

Processing perfection is a crucial elementdetermining the practical level of deviceintegration.

If we define D as being the density of fataldefects per unit area associated with a givenprocess and A as the area of the devicebeing fabricated, then the process -limitedyield can be approximated by the expres-sion:

Y=I + A D

Table IIncreasing integrated circuit complexity hasled to significant changes in the applicationof electronics to our society. Starting withthe first small-scale integrated circuits in theearly 1960s, we have witnessed an order ofmagnitude jump in device complexity aboutevery five years.

Integration level Complexity (gates)

Small scale (SSI) 1 - 10Medium scale (MSI) 10 - 100

Large scale (LSI) 100 - 1000

Very large scale (VLSI) >1000

This function is plotted in Fig. 1. TypicalVLSI devices are between four and sixmillimeters on edge. Thus, to obtainpractical yields on such devices, defectdensities below twenty per squarecentimeter are required.

Design innovations have improvedfunctional density.

Problems of packaging as well as those ofdefect density and chips -per -wafereconomics have limited increased chip sizeas a prime approach to VLSI. The majorefforts have focused on achieving max-imum functional density within chip sizesof about sixteen to twenty squaremillimeters. Innovative circuit design andsystem organization have produced a highlevel of functional density without makingmajor demands on the technology.

The one -transistor -per -cell dynamicmemory is perhaps the best example ofcircuit innovation that improves functionaldensity. The use of integral read -only -memories and programmable logic arrayshave also been approaches for increasing

Fig. 1Functional complexityand hence the economicvalue of an integratedcircuit is related to itsarea. The defect densityis determined by a com-bination of processcomplexity and proces-sing environment. Thusthe practical attainmentof VLSI requires theapplication of refinedtechnologies in con-trolled environments.

100

90

60

70

60be

0 50_LLa

30

20

10

O

I.H. Kalish

functional complexity. After all thesetechniques have been applied, however, thebasic approach to improving functionalcomplexity is to put more-but smaller-transistors on a chip.

The push forsmaller transistorsCurrent MOS production technologies areutilizing transistor channel lengths of fourto seven microns. Intense effort is currentlybeing applied to the fabrication and un-derstanding of channel lengths in the onemicron region for the next generation ofVLSI. One of the new tools being appliedto the achievement of this next generationof technology is the electron beam.

The RCA Solid State Technology Centerrecently acquired the Manufacturing Elec-tron Beam Exposure System (MEBES) tosupport RCA's LSI and VLSI efforts.

This equipment makes high resolutionphotomasks for conventional processing,but it also has the capability of supportingengineering efforts on direct wafer ex -

45 16(3)2 (4)2(2)

AREA(simi/

25(5)2

36(6)2

59

This VLSI device illustrates the state of the art ofSOS manufacturing technology. This 20 -square -millimeter SOS memory contains over 25,000transisters and thus requires low defect density inboth photomasks and processing environment. Thehigh efficiency of this layout (over 60% of the chiparea is memory) has been accomplished by in-novative sensing techniques making the use of a fivetransistor memory cell feasible.

One -micron lines and spaces drawn from a MEBEStest pattern. The VLSI technology of the next decadewill be designed with micron and submicronphysical features and thus be dependent upon suchadvanced approaches as the electron beamtechnology for device fabrication.

posure. A one -quarter -micron resolutioncan be achieved by the MEBES electronbeam, and its laser interferometer position-ing controls ensure address repeatability towithin one -eighth of a micron.

Competition -NMOS vs.SOSThe need for low defect density in VLSIprocessing has narrowed the choice oftechnologies available. Table II comparestwo of the leading contenders. The lowernumber of masks required to implementN MOS designs has always made it popularfor LSI applications. As the industrymoves to even higher levels of complexity,however, the low -power potential of theSOS technology becomes more and more

RCA's Manufacturing Electron Beam Exposure System (MEBES) iscurrently being used to generate high resolution photomasks in supportof RCA's VLSI efforts. Complex data processing within the MEBEScomputer section permits the elimination of reticle and step -and -repeatoperations to minimize defect density and pattern run out.

a singlememory cell

Burled -contact five -transistor memory cellmade using MEBES-generated masks. A newgeneration of COS/MOStechnology, the buriedcontact approach, willmake a significant in-crease in circuit complex-ity practical. This 1000 -square -micron memory issmall enough to be thebuilding block in RCA's16k SOS static RAMprogram; it is less thanone third the area of cellsbeing fabricated withcurrent productiondesign rules.

important. Reliability and economic con-siderations make it advisable to keep thedissipation of complex devices below onewatt per package and this will therefore bea limit on NMOS VLSI. The numbers inTable II anticipate that, by 1980, theadvent of the buried contact SOStechnology will close the gap on packingdensity and from then on the higherperformance potential of the CMOS/ SOStechnology will dominate.

The evolution of the RCA CMOStechnology into the region of VLSIapplications is projected in Fig. 2. Theoriginal CMOS technology could noteconomically attack LSI applications andhence most of its original use was concen-

trated in MSI applications that requiredlow power and wide power supply range.LOVAG (low voltage aluminum gate) wasthe first LSI technology to be applied to thetimekeeping area while the high perfor-mance C2L (closed CMOS logic) providedan entry into the microprocessor market.The projection now is for the buried -contact SOS technology to support RCA'sentry into the VLSI marketplace.

The application of evolving technologytrends to SOS product generation is sum-marized in Table III. The generation of150,000 transistor chips using electron -beam technology and two -micron designrules is a straightforward extrapolation ofpresent efforts.

60

Table IIComparison of technologies. The introduction of the aggressiveburied -contact design rules will make CMOS/SOS directly com-petitive with advanced NMOS techniques in function complexity.From that point, the system advantages of CMOS-low power andwide voltage range-should give it an advantage in new VLSIapplications.

Table IIIFuture SOS product direction. Memories and microprocessors are the mostnatural directions for CMOS VLSI product development. Beyond these,however, the advent of customer -programmable circuits with the EAROM(Electrically Alterable Random Access Memory) will result in another genera-tion of user -determined system applications of VLSI.

1980 1982NMOS SOS

1978 1980 1978 1980

Process innovations Buried -contactdouble poly

E -beamphotolithography

Density (gates mm') 140 200 110 200Design rules 3V: micron 2 micron

Gate delay (ns) I 0.5 2 I

Die sizes (typical) 5 rnm2 6 mm2Power dissipation (mW gate) I 0.4 0.1 0.05

Wafer sizes 4" squares 5" squaresSpeed, power (pj) I 0.2 0.2 0.05

Defect density 1 cm" 1/2 cm'Mask levels (No.) 6 6 7 8 (per level)

Critical dimensions (microns) 4 2 5 3.5 Die complexity (typical) 80k transistors 150k transistors

Static memory cell (µ") 2400 1100 3300 1000 Potential products 16k RAM 64k RAM8k EAROM 28k EAROM

64k ROM I26k ROM

The futureI6 -bit micro-processor

32 -bit micro-processor

One aspect of VLSI can be predicted withconfidence-VLSI devices will be low costdevices. It will take a lot of engineeringeffort and time to generate a new VLSIdevice, but once it has been made anddebugged, all of semiconductor historyindicates that it will be a low-cost mass-produced device. Custom VLSI willprobably not be worth the economic risksince the enormous engineering expensesinvolved in the design, evaluation, and thedevelopment of testing procedures cannotbe amortized by volume production.

The implications of this are quite signifi-cant for the evolution of new equipmentdesign. In the MS1 world of integratedcircuits, the equipment designer hadthousands of circuits to choose from with avariety of manufacturers and a variety oftechnologies. The VLSI world is morelikely to be similar to the world of vacuumtubes where there were a relatively smallnumber of multisourced well -understoodand characterized designs that users

learned to be comfortable with. Theapplication of a given family of VLSI partswill require a major educational invest-ment by the user-an investment he will beunwilling to cast away too soon. Thus itseems likely that the VLSI world will bemade up of relatively standardized formsof memories and microprocessors withelectrically programmable logic arrays andROMs giving the user his potential for acreative contribution to his product.

Reprint RE -24-1-6Final manuscript received April 21. 1978.

1961

1960

1978

19764

0

11974

a

1972

1970

4006

- 4115

40207

4. 7

C2". -.1.iD.

°

183 31802

I6K STATICCr-7

41 STATIC

IK STATIC

Fig. 2Technology evolution. VLSI tools andtechnology are broadening the market forCMOS devices from the specialty low powerand mi'itary areas to the memory andmicroprocessor markets heretoforedominated by the NMOS technologies.

196610100 1000 10,000 100,000 1.000,000

GATES FLIP SMALL RAMS SMIFT TIMEKEE PING MICROCOMPUTER LARGE - COMPLEXITYFLOPS BUFFERS REG. COUNTERS MEMORY MEMORY COMPLEX (GATES)

MICROPROCESSORS TIMEKEEPING

lz Kalish has been involved in the design anddevelopment of semiconductor devicessince joining RCA in 1953. He is oresentlyManager of the IC Design and ProcessDevelopment group of the Solid StateTechnology Center where he is responsiblefor the development of the next generationof VLSI technology needed to implementhigh performance memories and micro-processors in both bipolar and field effectdevices. Iz has authored several papers onsemiconductor devices and a boox,Microminiature Electronics.Contact him at:IC Design and Process DevelopmentSolid State Technology CenterSomerville, NJExt. 6243

61

Modern IC applications in communications systems

M.V. Hoover

New generations of ICs are becomingavailable to aid communications systemsdesigners in their work. The pervasivenessof these IC applications can be best il-lustrated by descriptions of specific ex-amples taken from various IC productcategories. The examples here use ICs forfrequency division and fm -if systems.

A new IC forfrequency divisionat vhf and uhf

A new IC' has been developed for use infrequency division ("prescaling") at vhfand uhf, with particular applicability infrequency -synthesized tuning systems. TheIC divides by 256 in the uhf mode and by 64in the vhf mode, and operates over thefrequency band of 90 to 1000 MHz. Thelogic diagram and pin configuration (dual -in -line package) are shown in Fig. I.Independent input terminals are providedfor ac coupling of vhf and uhf input signals.

The mode of operation is band -switchselectable by means of a logic signalapplied to terminal 3. In the uhf mode,which is activated by applying a logic 1signal to terminal 3, all of the eight dividerstages are operated, resulting in division by256. In the vhf mode, with logic 0 at

ICs can provide relatively simple and economical solutions tocomplex communications design problems.

terminal 3, two divider stages are by-passed, resulting in division by 64.

Complementary output terminals (4 and 5)are provided, typically delivering a I -V

peak -to -peak signal with controlled slewrate for harmonic suppression. By main-taining a balanced load and by limitingoutput -signal rise and fall times (typically70 ns), harmonic output is limited aboveabout 40 MHz. Table I gives some typicalelectrical characteristics of the prescalingIC.

A new fm -if system ICIn 1970, the industry's first comprehensivefm -if system IC (CA3089) was introduced.'It contains an if amplifier, quadraturedetector, of preamplifier, and specific cir-cuits for agc, afc, muting (squelch), andtuning -meter control. This circuit has at-tracted a world-wide following of users.Recently, a generically similar new IC (theCA3189)' has been introduced, incor-porating a number of additional featuresand improvements that permit the designof communications systems with enhancedperformance. Following is a listing ofadditional features and advantages incor-porated into the new IC:

1) typical (S+ N)I N > 70 dB;

-I V001

N/C

SAND

VHF 14INPUT

VHF13

INPUT

Nuxl +2 ri

N/C +2 -2 -3 SWITCHBIAS 2 CONTROL

11 -UHFINPUT OUTPUT

UHF -4 -2 -2INPUT 10.-- OUTPUT

TTN/C 9- -6 N/C

BIAS I

Val VEE 2

Fig. 1Frequency -dividing IC divides by 256 in uhf mode and by 64 in the vhfmode. Logic 0 signal applied to terminal 3 bypasses two dividerstages for vhf division.

2) adjustable agc threshold voltage forthe tuner's rf stage;

3) signal -strength -meter drive voltagedepressed at low signal levels;

4) deviation -muting logic provides on -channel step control voltage, whichsimplifies design of signal -search typereceivers and permits use of circuitry tominimize "audio thumping" when tun-ing onto or through a station; and

5) externally programmable audio out-put voltage.

Fig. 2 contains a block diagram of the newIC (CA3189) together with a schematicdiagram of the comparatively few externalparts the IC requires. Its 3 -dB limitingsensitivity is typically 12 µV, and it typical-ly delivers audio in excess of 500 mV at ±75kHz deviation with the suggestedquadrature-detector components. Fig. 3depicts typical limiting and noisecharacteristics for three different receiver -if configurations.

The IC achieves typical (S + N)/N greaterthan 70 dB.

The input stage is a cascode circuit, whichprovides low input capacitance, an assetwhen the device is being driven by a

Table IElectrical characteristics of RCA's new frequency -division IC. Logic pulse changes operation from uhf tovhf.

UHF input signal

VHF input signal

= 450 to 950 MHz80 mV (rms) (max)

f, = 90 to 275 MHz40 mV (rms) (max)

Band -switch control voltage logic I = 2.4V (min)logic 0 = 0.8V (max)

Power -supply voltage*

Device current

(VDD VEE) = 5.0 V

60 mA

Terminals 7 and 8 are normally tied together as the negativesupply terminal.

62

ALL RESISTANCE VALUES ARE IN OHMSL TUNES WITH 100 pF IC) AT 10 7 MHz

0091 751701(0 No. KACS K586141,1 OREQUIVALENT.)

0.02

I

DELAYEDAGC FORRF AMP.

33 K

IFINPUT

AGCTHRESHOLD

"AAR2

TO INTERNALREGULATORS

CA31139E

IF AMPLIFIER1ST ipAMPL

LEVELDETEC TO

DELAYEDRF AGC

2ND IFAMPL

LEVELDETECTOR

FRAME SUBSTRATE

4 7 K

QUADRATURE c piV INPUT 22 H

REFERENCEems

df 110 0F

'14) UWE T'AER

3RD IFAMPL

LEVELDETEC TO

TUNINGMETERCIRCUIT

33K

FMDETECTOR

LEVELDETECTOR

DEV-IATIONMUTELOGIC

MUTE (SQUELCH!DR VE

C IRCVIT

AUDIOMUTE

ISOUELC041CONTROLAMPL

_Co TTOHRSETSEHROLEOD

TUNING METER OUTPUTLOGIC CIRCUI-S

OV -ON-CTIANNELSTEP CONTROL- VOLTAGE

R38.2 K

AFCOUTPUT

OE 470

DRIVECI

R SK

AF

MUTINGSENSITIVITY

p.F

RI10K

Fig. 2Improved fm -if subsystem IC has special circuits for signal -strength -meter drive, adjustable agc threshold voltage,deviation -muting logic, and programmable audio output level.

ceramic filter element. The bandwidth ofthe amplifier has been set at 15 M Hz. Sincea majority of the broadcast -band fm -ifcircuits operate at 10.7 MHz, thebandwidth of the amplifier is notrestrictive. This choice of bandwidthreduces wideband noise, as the amplifierdoes not employ interstage filter elements.Restricting the if bandwidth reduces thecriticality of pc -board layout from thestandpoint of the need to avoid parasiticinstabilities.

The agc threshold voltage for the tuner's rfstage is adjustable.

Many modern fm receivers employ a dual -gate MOS field-effect transistor( MOS FET) in the rf stage. The circuit usedin the new IC provides the user withflexibility in selecting the threshold voltageat which agc operation commences. Itprovides a dc output suitable for providingan agc range in the order of 40 dB whenapplied to G2 of a dual -gate MOSFET. Thepoint at which agc to the rf stage com-mences can be chosen by applying asuitable control voltage at terminal 16, thetypical agc threshold voltage being 1.25 V.Potential to drive the agc threshold -adjusting potentiometer R2 can be takenfrom terminal 13. The characteristics of theagc dc voltage for the tuner as a function ofinput signal voltage can be determined byreferring to Fig. 4.

With this IC, the signal -strength metershows a true reading at zero and low signalstrengths.

The on -chip tuning -meter circuit consistsof three amplitude detectors (fed from theoutput of the three limiter stages), a currentsummer, and an output -level shifter. Underideal zero -signal conditions, the currentoutput should be essentially zero. Inpractical applications, however, noise frompreceding stages (e.g., the rf and mixerstages) presents significant drive to themeter. To obviate this problem, a dc offsethas been built into the meter drive circuit sothat the meter will show a true zero readingunder zero -signal conditions. Fig. 4 showsthe voltage at meter -driving terminal 13 asa function of input signal.

The deviation -muting circuits eliminate an-noying "thumps" that can otherwise occurwhile tuning through stations.

Two types of muting circuits may be usedwith the new IC. The first is a classicalapproach, i.e., noise -muting, which can beused to mute noise between stations. Fig. 4shows the muting characteristics: muting isapplied in accordance with thischaracteristic curve as the input signaldecreases below about 40 µV. Potentio-meter RI, in Fig. 2, is used to adjust themuting threshold. This type of circuitry iseffective in combating interstation noise,

Of0

7 1

SIGNAL (DEVIATION i 75 kHz)si,/CA318SEI 20

I

ONLYPRECEDED BY

AND GAIN STAGE',\ '\

CA31139EFILTERC 119E PRECEDED BY -\ \\t 2 FIA31LTER AND GAIN STAGES

40 \Ii \

'?I 50, V- \ NOISE

IIJa60

\\ 1N

70

i itt t ill Z'.......".....t ill.Z.-BO

10 102 103

SIGNAL LEVEL -0V

(04

Fig. 3Typical noise characteristics are very good-greater than 70 dB at normal signal levels.Limiting characteristics are also good.

I0

IIAMBIENT

0s=viIs

0 _.1 1Qu

1 i --i .1g

§ 40.

5SO

-60

DC VOLTAGE SUPPLY V+ 12 VTEMPERATURE (TA)+26C

RECOVERED AU010 FROM FULLOUTPUT (LEFT COORDINATE)

ylTUNER AGC DC,VOLTAGE ATTERMINAL No. IS(RIGHT COORDINATETIMES 2)

tSS

3 'EtIPIN IS 13/

....00".072;41Pl, Cr.-

NICP ols>081t

I 11

VP *0 004'0OW." II 1 I I1 I i II I i 1

0 100 I I4 101g

INPUT SIGNAL-PV100k

Fig. 4Tuner agc has an adjustable threshold voltage.Figure also shows IC's low drive -signal output forsignal -strength meter at low signal strength.

63

I DC SUPPLY VOLTAGE (V') 12 VAMBIENT TEMPERATURE (TA)25

0 5 10 15 20 25LOAD RESISTANCE (BETWEEN TERM 7 AND TERM 10)-113

Fig. 5Deviation -muting circuit gives sharperresults than standard demuting circuits.Threshold value for demuting dependsupon value of resistor R3, placed betweenterminals 7 and 10 in Fig. 2. With the valueused in Fig. 2, graph shows that receivermust be tuned within 40 kHz of station'scenter frequency before demuting occurs.Other circuits can have partial demuting asfar away as 300 kHz from a station's centerfrequency.

Merle Hoover joined the RCA Electron TubeDepartment as an Advanced DevelopmentEngineer in 1948 and the RCA IntegratedCircuit activity in 1966. His specialty in ICshas been in applications engineering. He iscurrently Leader, Technical Staff, in theBipolar IC Engineering Department.Contact him at:Bipolar IC EngineeringSolid State DivisionSomerville, N.J.Ext. 7326

but its performance leaves something to bedesired when tuning toward or away from astrong signal. Under these conditions (e.g.,within the order of 300 kHz from thecorrect tuning point for a strong signal),there is "signal splatter" from the strongstation to produce partial or completedemuting; the result is a combination ofnoise and distorted audio as a consequenceof the off -channel tuning. Furthermore,when the receiver is tuned through astation, there is a sudden change in theoutput dc level at the audio output terminal(terminal 6), producing the familiar low -frequency "thump" noise. These classicalannoying imperfections are characteristicof noise -muted systems.

These annoying characteristics can be con-siderably diminished by using so-called"deviation -muting" circuitry, which hasbeen incorporated in the new IC.Deviation -muting logic is connectedbetween the afc and mute -drive outputterminals. This circuitry consists of twoaccurately determined reference levelssymmetrically placed about the correct -tuning afc reference; the result is a marked"sharpening" of the demuting action. Forexample, with this circuit in operation, thereceiver must be tuned within about 40 kHzof the center frequency before demutingsuddenly occurs. Furthermore, thedeviation -muting threshold depends uponthe value of the load resistor R3 connectedbetween terminals 7 and 10; thesecharacteristics are portrayed in Fig. 5. Theuse of this circuit permits the incorporationof an integrating circuit (C1, R4. C2 in Fig.2) to greatly reduce the "thumping" noisementioned above when tuning through astation. As shown in Fig. 2, the deviation -muting feature also provides a sharply -formed step control voltage (typically 5.6V) whenever muting or demuting occcurs.This "on -channel" signal (zero volts) isuseful in controlling receivers withchannel -searching features.

The audio output voltageprogrammable.

voltage reference generated at terminal 10by an internal zener diode, residual noisegenerated by the zener may be bypassed toground by means of a capacitor.

MicroprocessorsMicroprocessor lCs are being used in anincreasing abundance of applications in awide diversity of fields because they areflexible, economical building blocks. Forexample, many communications systemsrequire coordinated functioning betweenreal-time clocks (used for systems timing)and frequency -measurement circuits.RCA's COSM AC CMOS IC micro-processor has been designed into a systemto perform these functions; the design willbe the subject of a future RCA Engineerarticle.

ConclusionsNew -generation ICs are becomingavailable for a number of applicationcategories. A new IC for frequency -division at vhf and uhf has been described,as have the features of a next -generation ICfor fm -if systems. The fm -if IC provides anon -channel "stop -function" when used insearch -mode receivers, as well as improveddeviation muting. Perhaps these examplesbest illustrate the major reason for themodern IC's influence in communicationssystems-the IC offers the designer acomparatively simple and economicalmeans of solving design problems formerlyranked as complex.

AcknowledgmentThe writer acknowledges that his efforts inconnection with this paper have been moreeditorial than authorial. The credit for theorigination of the material contained herebelongs to a great many more people thanthe few mentioned in the reference sectionbelow. It is hoped that the many will favorthis rendition of their work.

is externally References

The audio output at terminal 6 is a currentsource, and requires the use of an externaloutput load resistor; it is preferablyreferenced back to the 5- to 6-V referencevoltage at terminal 10. In this manner, themagnitude of the recovered audio voltagemay be programmed by selecting an ap-propriate load resistor, limited only by themaximum distortion -free voltageamplitude within the device capability.Since the load resistor R3 is returned to a

I. The RCA Developmental Type TAI0535 VHF UHFPrescaler, RCA Solid State Division, Somerville. N.J. Thecircuit was designed by H.R. Beelitz and D.R. Preslar of theRCA Laboratories Solid State Technology Center.Somerville, N.J.

2. Data File No. 561 on the CA3089E and "Application of theCA3089E FM -IF Subsystem." 1CAN-6257, L.S. Baar, RCASolid State Division, Somerville. N.J.

3. Data File No. 1046 on the CA3I89E, RCA Solid StateDivision. Somerville, N.J.

Reprint RE -24-1-18Final manuscript received June 6. 1978.

64

drr- -0 T-

A

gn .ah@ lcWoff +11 n jobThe lure of electronic fishing

N. Brooks

Fishing can be frustrating as well as fun. The catch to fishing is first finding the fish.Once you've found them, then you've got a fighting chance.

The solid-state digital depth sounder described here gives you that break.

I grew up fishing the streams and rivers of southern Indianaand gradually expanded my range to cover many otherstates and quite a bit of Canada. My most "exotic" trips havebeen to Chantrey Inlet, N.W.T., approximately 100 milesnorth of the Arctic Circle where the Back River empties intothe Arctic Ocean, and to Clearwater Fjord of Baffin Islandwhere the Clearwater River empties into CumberlandSound right on the Arctic Circle. These places have laketrout, arctic char, arctic grayling, and whitefish.

All of my fishing has been in fresh water, and I have enjoyedcatching many species from bluegills to lake trout, salmon,and muskies. Being relatively close, I have done muchfishing in Lake Michigan and the states of Michigan andWisconsin. Both of these states have had very active fishconservation programs resulting in good fishing for severalspecies such as bluegill, perch, crappie, bass, trout, salmon,pike, walleye, and musky.

Fishing techniquesSeveral methods are used to catch fish, such as fly fishing,spinning, bait casting, trolling, and jigging; any of them maybe done with artificial or natural baits or lures. I have alwaysbelieved the object of fishing is to catch fish, and I have usedany or all legal methods in trying. Most of my fishingpartners will verify that I get my share of fish. Having theproper equipment and learning how to use it is veryimportant. Learning the habits and characteristics of thespecies of fish you are trying to catch also is a big help.Some are normally found near the surface, and some arenormally near the bottom, with others in between. Forexample, wind and weather, especially temperature, willaffect the depth of salmon in Lake Michigan.

Michigan has done a tremendous job in making LakeMichigan the best fishing hole in this part of North America.In recent years the other adjoining states have pitched inand fishing for coho and chinook salmon, rainbow trout,brown trout, and lake trout is very good in most all areas ofthe lake. Since salmon are not native to this area, we had tolearn new fishing methods and even developed new equip-ment such as downriggers and baits. A downrigger is madeup of a large diameter spool with strong (150 lb test) wireline used to lower a lead weight called a cannon ball (5 to 10lb common) to depths of up to 200 (or more) feet. The baitand line from the fishing rod is connected to the cannon ballby means of a release system so that when a fish strikes the

lure, the system releases the line from the cannon ball.Hence, there is no weight on the line for the fish to fightagainst. Many special baits, mostly odd shaped, and bentpieces of metal of all colors were tried, some with goodsuccess. Large flies Veiled from doggers (bent pieces ofmetal which wobble) also work very well.

Why a fish finder?

A fish finder is a necessity if you are serious about catchingfish in Lake Michigan or most other bodies of waterbecause, for reasons known only to the fish, they are not inall areas of the water at the same time. You must first find thefish before you can catch them. The fish finder will alsoindicate the depth at which 10 run your lures and provide anindication of the number and size of the fish. After the fishhave been located, it's up to the fisherman to determine thebest lure, trolling speed, etc., to catch them.

In recent times fishing has evolved into basically two types;commercial and sport fishing, each with its associatedtypes of equipment. Several years ago, electronics, in the

Neal Brooks joined RCA'sMarion plant Quality andReliability operations in1949, just prior to the startof tv picture tube produc-tion, with responsibilityfor finished -tube qualitytesting, life testing, andrecorded readings. Hereceived an award fordeveloping a portable pic-ture tube tester in 1955.From 1955 to 1974 he wasManager of Quality andReliability operations forthe Marion plant. Current-ly he is the Manager ofCustomer Liaison and therepresentative for theMarion plant in the inter-national markets.

Contact him at:-- Customer Liason

Picture Tube DivisionMarion, Ind.Ext. 5522

65

form of sonar fish locators and depth sounders, began toplay an important role in the taking of fish. These units werefirst used by commercial fishermen but later were adaptedto use by sportfishermen. Beside being expensive, the earlymodels were not without their problems. They were bulkyand heavy with their tubes and vibrator -type powersupplies. On most of them, the depth accuracy dependedon the speed of a motor usually controlled by a "circuitbreaker" type of governor, which caused reliabilityproblems. Then, transistors came into practical use. Thissolved some of the size, weight, battery drain (especially onportable units), and reliability problems; but it left a basicproblem area which is found today on most fish finders-the electric motors which determine the accuracy andreliability of the units.

All fish finders I know of work on the basic principle oftransmitting a pulse of ultrasonic sound, then receiving anddisplaying any echoes from fish or the bottom in terms oftime or depth. This is usually done by firing a neon bulbattached to a motor -driven disc with the motor speedcontrolled so that one revolution is equal to a given time orfeet of depth. I know of two units recently placed on themarket that have no moving parts. But one of these usesneon bulbs, and neither has the depth resolution required.

Searching for the ideal fish finder

A few years ago, four of us were on a salmon fishing trip toLake Michigan with three fish finders, one almost new,among us. Within three days none of the units was useable;one developed an open motor, another gave a 60 -foot depthreading in water we knew to be at least 90 feet deep (themotor running slow), and the third one had a shorted outputtransformer. (The transformer is used to develop the high -voltage signal to operate the neon bulb depth -indicatorlight.)

Fig. 1

Digital depth sounder has three ranges: 0-75 ft, 0-150 ft, and 0-300 ft, corresponding to the time -base generator output of4800, 2400. and 1200 pulses per second, respectively. Eachtime the "0" LED is scanned, the 200 -kHz transmitter is keyedand the pulse is radiated by the transducer. The echo return isreceived and rectified to a single pulse which lights the LEDcorresponding to the depth of the fish.

It was almost hopeless to go out without a finder, so wefinally borrowed one from another group of fishermen andfinished out our trip. This event started much discussion onwhat would constitute an ideal fish finder. As we also makefly -in trips to remote places in Canada, size, weight, andbattery life were also important considerations besidesreliability. If the indicated depth accuracy is not withinabout one percent, depending on the type of fishing, youwill do more boat riding than fish catching. Another point isthat in addition to having a good fish finder, you must learnits capabilities and how to use it. I have been with guidesand charter boat operators who were missing some of thefundamentals.

After establishing the basic requirements, I started to lookfor parts and materials that would satisfy my objectives.About this time LEDs were coming into use, and I decidedthese should make a reliable readout because they requiredno high -voltage transformer or motor and offeredsignificantly lower current drain on the batteries. Theanswer to accuracy was a crystal -controlled time basewhich triggers the sonar sounding pulse and controls thescanning time of the LED echo pulse displays. Low -powerICs were used for dividers in the time base and counters forthe LEDs where possible. A system of multiplexing wasworked out for the display LEDs to keep the number of ICsand the battery drain to a minimum.

Circuit design and operationThe depth sounder shown in Fig. 1 is of the type that excitesa 200 -kHz transmitter to generate an acoustical wavethrough a transducer. Reflected acoustical waves fromobjects such as fish and the bottom are received at thetransducer and converted by the receiver into a single pulseor series of pulses. The transmitter frequency of 200 kHz ischosen to be compatible with conventional transducercharacteristics.

480 IIHA CRYSTAL OSCAND DIVIDERS

20TRANS.

TRANSDUCER

RECEIVER

40400 PPS2400 PPS DECODE DECODE1200 PPS '7 DIGITS TENSQ0Q,1 Q

TRANS. KEYER

0LED

ABC A B C

DIG T DRIVER I 70 ISMULTIPLEXER

# V

ECHO PULSES

MPLX. KEYERI OF IS

SIGNAL KEYER

FISHAL ARM

TO KEYERS FORIS GROUPS OF 10LEDS IN PARALLELWITH THOSE SHOWN

,r- IV

LED (BATTTEST TEST

112V

66

As illustrated in Fig. 2, the time -base generator of thesounder is formed by a combination of a conventionalcrystal oscillator and frequency dividers. The generatorprovides three outputs of 4,800, 2,400, and 1,200 pulses persecond, corresponding to the three depth ranges of 0-75, 0-150, and 0-300 feet, respectively. Using one digit driver, one1 -to -15 multiplexer, and fifteen multiplex keyers, the LEDsare scanned sequentially at the rate determined by therange switch. The LEDs are arranged in fifteen groups withten LEDs in each group. In any one group, each LED hasone terminal connected to the corresponding output of thedigit driver. The other terminal is connected to the collectorof the multiplex keyer transistor for that group.

The 150 LEDs are arranged in a 15 -inch dial scale so that onthe 150 -foot range each LED equals one foot of depth.

Each time LED "0" is scanned, the output of the digit driverfor LEDs "0" - "9" and the 1 -to -15 multiplexer activates thetransmitter keyer circuitry (see Fig. 1). A 200 -kHz trans-mitter is then energized and transmits only for the time LED"0" is scanned. A receiver tuned to 200 kHz receives andrectifies the echo return so that the output is a single pulse.This echo return pulse closes the signal keyer transistorswitch and places a voltage on the emitter of all fifteenmultiplex keyer transistors, which in turn, lights the cor-responding LED.

Bottom depth and fish depth are clearly distinguishablefrom each other by the pattern observed in the LEDs, asshown in Fig. 3. Normally, for constant depths for fish or the

/.3N /.SK

/0A' .0/ /OA'

Q

/ON

4430NC .0/01-1(

G/400374C/54 OR

741_/54

W./ 4-24-2.3 H3

oc,Pc,7 TO 4 4- 4LEDs S<-5

Mi.L XAEYERS

7 (-07BHB9 (--- 9/0Hro/I +- /I

-.412

2423

22 P

"I

820

4.74(

r

SD x X -X f 1 I MITIOTATIAITTAFISH BC rrcm

TRANSDUCER -9-4,

,FISH ECHO

BOTTOM ECHO

TRANSMITTED -- - - FISH = 25 FT.SOUND

-.. -

BOTTOM - 30 FT

TYPICAL NEON DISPLAY

ZERO DEPTH FISH

BOTTOM

Fig. 3The solid-state digital depth sounder (top) uses 150 LEDs in itsdisplay; each one indicates a segment of depth. This systemprovides for greater accuracy aid reliability in determining depthsof fish at various levels than the typical motor -driven neon displaytype (bottom) with its inherent instability. Middle diagram showsthe cone -shaped sound -wave pattern produced by the transducer(used in both display systems).

74L'70

-2- 34

20/i/e3

/7

/6->/5 OUTPUT TO1S-->14 LEO'S

MPL x'46-* '3 RE YEA'S/3H042Imo Ps

74L 93

7

9

/0 5

//3

/3 2

4.74'

0-04N7

44

74L 90 / /I

-4 /3- 5 /2 4

( 5 0-- 7

4-5V

1800ce.1° 2400

74L90

/200

CU 'PVT 7

741/45

Li /40- 800-2 /3

0-03

ToLEO'S ".

9/0 7/I 5

//

5 /0,-(-304.4

4 3 OUTPUTTO ceo's13

-4 3129-7 IS t1

/.3.0/

5.64' 4-5V

Fig. 2Time -base generator circuit for the sounder combines a conventional crystal oscillator with frequency dividers. Sequentialscanning of the LEDs is controlled by multiplexers and digit drivers.

67

bottom, one or two LEDs will be lighted depending upon theconstancy of the depth. As rapid changes in depth occur,whether for fish or bottom soundings, the LEDs will light insequence corresponding to changes in depth.

Why LEDs for display?

To my knowledge, LEDs had never been used for thisapplication at the time I was considering the design. Somethought was given to numerical readouts, but because Iwanted a constant indication of the bottom depth and oftenthere are fish anywhere from the bottom to the surfacewhich are to be displayed at the same time, I decided to use150 one -tenth -inch diameter LEDs and let each one indicatea segment of depth. For used, thismeant that on the 75 -foot depth range each LED represents6 inches, on the 150 -foot depth range, 12 inches, and on the300 -foot range, 24 inches. Depending on crystal accuracythese segments can be made to indicate within an inch or soat a hundred -foot depth as far as the unit is concerned.There are other things, temperature, type of water, etc., thataffect readings, but these are not a factor in my fishing.

Interpreting the displayBy gating the LEDs with the counters and the echo pulses,any number of fish at any depth and also the bottom depthare displayed in sequence, which appears as a continuousdisplay to the eye. To properly interpret the display youmust take into account the pattern of the transmitted"sound" and the received echo of this "sound." The depthindicated on the display is a measure of the time it takes thesound to travel (in water, approximately 4800 feet persecond) from the transducer to the target and return to thetransducer.

The transducer normally sends out a short pulse of soundwaves in a conical pattern, which means that increasinglylarger areas are covered as the depth increases. Thispattern is shown in Fig. 3. Transducers vary widely inradiated patterns from a few degrees to several degrees. Atypical one with a 25 -degree cone will cover approximatelya 4 -foot diameter circle at a 10 -foot depth, a 23 -footdiameter at 50 feet, and 45 -foot diameter at 100 feet. Thus,not all fish, or other targets, that show on the display aregoing to be directly under the boat and may appear at a

depth slightly greater than the actual depth. As youapproach a fish lying still in the water, his depth appears tobecome shallower and, as you leave, it will appear to godeeper when you are on a line directly over the fish. If thefish is at the edge of the cone on either side of the boat, thedepth indication will not change noticeably but will beslightly greater than the actual depth.

Salmon, especially the chinook, run at depths of over 100feet, and it has been my experience that you must get yourlure within a foot or two of the actual depth of the fish butnever below them. This narrows the range of effective luredepth and requires an accurate depth indicator, especially ifyou are using it for location and navigation by water depth.In addition, you must select a good lure and move it at theright speed with the proper action. In short, a fish finder maylocate the fish but it's still up to the fisherman to make themopen their mouths.

Preventing false alarmsOn the days when you're not into large schools of fish, youlose interest in constantly watching for fish on the indicator.A fish alarm, an audible signal, was incorporated in the unitso an alert would sound in case of a potential strike. The fishalarm idea was not new, but after a few days of use in LakeMichigan, I decided it could be improved. The problem wasthat in some of the good fishing areas, as you might expect,there were many small fish near the surface covering largeareas. So the alarm was on almost constantly, requiringcareful watching of the display to see if there were good fishbelow, which there were many times. All fish alarm systemsthat I know of, even on units made today, have anadjustment to prevent alarm on a bottom echo but have nocontrol over echoes near the surface.

To overcome this problem I added a circuit and a control(see Fig. 4) with which I can vary the triggering sensitivity ofthe alarm circuit down to about 30 feet without affecting thesensitivity at the greater depths. The sensitivity is decreasedon a slope from the depth of the starting point, determinedby the control setting, to the surface. To do this I used twotransistors in series (a type of gate) so that if the combinedecho pulse and alarm gating pulse equals a given value itwill switcn on the following stage, turning on the alarm. Thecontrol marked "top" in Fig. 4 shapes the leading edge of the

68

S470K ro 711.

L. RULS.0 472OUT ItIT .22

2

2.2

2.002 470K

- .003/N

ro 4.5e zo .227. RC V R

DE TEC TOR FV3N TO SeALARM DEPTK 33K

'2V

44

5

}VKR

0 r

Fig. 4Fish alarm circuit incorporates dual triggering sensitivity to prevent alarm sounding fomsmall fish near the water's surface as well as from the bottom echo.

alarm gating pulse into a sawtooth and varies the amplitudeof it so that as it is reduced, it requires a larger echo (fish)pulse to turn on the alarm. Fig. 5 shows how the pulseshaping works. With proper adjustment the small fish are nobother but large fish near the surface are not missed.

Alarm circuit operationIn my alarm circuit, Q1 is driven by the keying pulse to the200 -kHz oscillator and generates a gate pulse of varyingwidth depending on the setting of the control marked"bottom." This pulse is used to turn on Q2. The echo pulsesturn on 03. Both Q2 and Q3 must be on to drive Q4, turningon the multivibrator driving the speaker. This is theconventional alarm circuit. When the bottom control isadjusted so that the gate pulse to Q2 shuts off Q2 before thebottom echo pulse is received, the alarm will not sound onthe bottom echo, but will sound on any echoes above thesetting of the bottom control.

By adding the control marked "top" and a 0.1-microfaradcapacitor, the leading edge of the gate pulse is shaped anddelayed so that small echo pulses on Q3 will not turn on thealarm. In this mode of operation, 02 receives an increasingamplitude (sawtooth) gate pulse and being only partiallyturned on, a correspondingly larger echo pulse must bereceived by 03 to turn on 04. Since the echo pulses aresomewhat related to fish size, with proper adjustment of the"top" control small fish will not trigger the alarm but largeones will, as shown in Fig. 5.

Echoes from fish are not normally as strong as the bottomechoes, depending of course on the type of fish and the typeof bottom and assuming the fish is near the bottom. Toobtain good readings on small fish (those approximately 5pounds in weight) I experimented with receiver sensitivity,

;PULSE LEVEL REQUIREDTURN ON Q4

ECHO PULSESro Q3

NORMAL GATEPULSE TO Q2

INPUT TO

Q4 - NORMAL

THIS POR770N ADJUSTEDBY "TOP" coNrRoL

DELAYED ANDSHAPED GATEPULSE TO Q2

INPUT TOQ4- DELAYED GATE

0 DEPTH INCREASING

Fig. 5The addition of a "TOP" control to the fish alarm circuit(Fig. 4) helps discriminate large fish from small fish near'he water's surface. The TOP control shapes and delays thegate pulse so that small echos (from small fish) will not turnon the alarm.

vs. transmitted power to the extent that I decided to run thetransducer at the maximum rating for voltage input. Thereason for maximum transducer drive vs. receiver sensitivi-ty is that more sensitivity gives more problems withunwanted signals such as ignition, motor, and propellernoise. Because transducers resonate at a specific frequen-cy, the output circuit driving them must be tuned to theexact frequency and, consequently, the receiver must betuned to the same frequency with a reasonably narrowbandpass.

Toward perfectionI worked on this project, along with others, in my spare timefor approximately a year. The unit was put into a box 3x4x6inches (almost pocket size). The power source is a 12 -voltlantern battery, and the unit has a current draw of 125 mAwith both "0" and bottom signal displayed.

I have had much enjoyment trom using this unit for the pastthree years. It is completely solid state, has no moving parts,and will maintain its accuracy for many years to come. But Iexpect to keep adding new developments to keep it ahead ofanything on the market.

Reprint RE -24-1-11Final manuscript received March 6, 1978.

69

Mandatory licensing of engineers: pros and cons

RCA Engineer Staff

A controversy regarding regulating thepractice of engineering is taking placeacross the nation. Up until now, mostengineers employed by industrial com-panies have been exempt from man-datory state licensing. During the pastfew years, however, mounting pressurehas been exerted on state legislaturesto remove this "industrial exemption"for many engineers. Expanded licens-ing requirements would strongly affectengineers, companies, and the public.In this article, arguments of the con-troversy will be presented along withenough background to inform readersabout the issues.

State laws governing the practice ofengineering exist in all fifty states, theDistrict of Columbia, the Canal Zone,Guam, and Puerto Rico. Generally, theintent of these laws is to safeguard life,health, and property, and to promotethe public welfare. The laws pertain toengineering that is performed in a

practitioner -client relation, generallyreferred to as consulting engineering.

The first engineering state licensinglaw was enacted by the state of Wyom-ing in 1907. Its impetus came from theState Engineer, who was determined tostop abuses in land surveys concern-ing water rights. Such surveys at thetime were being prepared by lawyers,real estate brokers, notaries, insuranceagents and even pawn brokers who hadbeen posing as engineers or as sur-veyors.19 Other states and territoriesfollowed suit in a slow but steadyprogression.

Throughout the development of licens-ing legislation, focus has remained ondirect practitioner -client consultingapplications. Legislation has notgoverned large numbers of engineersemployed in other capacities. Ex-emptions from mandatory licensinginclude U.S. Government employees,employees and subordinates of

Proposed state laws requiring mandatory licensing ofengineers in "responsible charge" could affect every workingengineer.

licensed engineers, public utilityemployees, industrial employees, andindividuals performing engineeringservices for other than public use.While engineering licensing hasproliferated among states, there is nofederal law on engineering licensing.

The authority to regulate engineeringpractice derives from the police powerof the state. Each state exercises fullautonomy in establishing andadministering its licensing re-quirements. Some progress has beenmade in promoting uniform standards,but wide differences currently existfrom state to state. Implementation oflicensing regulations is administeredby state boards that represent anofficial arm of state government andare created by legislatures. Stateboards are charged by legislatures toestablish criteria for licensing, to setminimum levels of performance, toadminister licensing statutes, and topropose changes needed to maintaincurrency. Individuals on various stateboards have formed a voluntaryassociation now called the NationalCouncil of Engineering Examiners(NCEE), which has participation by allU.S. states and territories.

NCEE has drafted proposed legislationwhich would revise existing statutesgoverning the licensing of engineers.This proposed legislation is known asthe NCEE "Model Law."' The ModelLaw contains no provision for an in-dustrial exemption per se. It defines the"practice of engineering" broadly, in amanner which would likely encompassall work in RCA that is typically referredto as engineering. It requires thatanyone who is in "responsible charge"of such work be licensed. Responsiblecharge is defined as "... direct controland personal supervision of engineer-ing work." The requirements for licens-ing include experience, examination,and graduation from an engineering

curriculum. The only industrialemployees practicing engineering whowould be exempt from licensing areemployees or subordinates of licensedengineers.

Lobby efforts have been organized toencourage the adoption of the NCEEModel Law. Spurred by the con-sumerism movement, recent lobby ef-forts have intensified. In 1976, NSPE,the National Society of ProfessionalEngineers (the professional associa-tion of licensed engineers) issued apolicy statement in support of theNCEE Model Law. NSPE Policy No. 36-B reads as follows:'

"NSPE believes that the stateengineering registration laws shouldapply to all engineers responsible forengineering design of products,machines, buildings, structures,processes, or systems that will beused by the public; and urgesmanagement to engage onlyregistered professional engineers forresponsible engineering positions.Management is also urged topromote registration among all of itsengineering staff. The Society is op-posed to proposals to exemptengineers in industry from the stateregistration laws and recommendsthe phasing out of existing stateexemptions in state registrationlaws."

Recently, the United States ActivitiesBoard, the political -action arm of IEEE,proposed to the IEEE Board of Direc-tors that IEEE should support removalof the industrial exemption. By itsaction of Feb. 19-20, 1977, the IEEEBoard adopted the USAB policy state-ment (IEEE Policy Statement 7.3).22However, this IEEE policy has not yetbeen implemented. Memberpreference is not clear. For example,response to the IEEE 1977 U.S.Member Opinion Survey26 appearedsomewhat contradictory:

70

16. Do you agree with the IEEE Boardof Directors' recommendation to es-tablish mandatory licensing by thestates for practitioners who areresponsible for their activities or forthe activities of their subordinates?

Agree 55.8%Disagree 34.0%Not Sure 10.1%

19. Would you favor eliminating allexemptions which allow engineers

working for industry towithout a lice -Ise?

Yes 29.4%No 57.5%Not Sure 13.1%

pract ce

18. Should use of the title "Engineer"be restricted to those persons whoare registered or licensed?

Yes 50.2%No. 44.0%Not Sure 5.8%

21. Should IEEE require either a

deg ee or a P.E. license for all grades ofmembership above student grade?grade?

Yes 45.5%NO 48.6%Not Sure 5.8%

Reprint RE -24-1-20Final manuscilpt received June 20.1978

Arguments favoring and opposing expanded licensing are numerous, often heated and emotional, sometimes based onfact and sometimes based on opinion. A sample of these arguments follows. In order to convey their intensity, the languagechosen is similar to that actually used. including quotes. Therefore, what is stated as fact may or may not be verifiable.

Pro-arguments forexpanded licensing"There is a public need."

The news of the day is replete with instances of actual, near,or potential injury or death resulting from actual or potentialmalfunction or inadequate protection of manufacturedproducts. The consumerism movement and the increasingnumber and size of product liability suits establish beyondany doubt that product safety is inadequate. Neither the"conscience" of industry nor product liability legislationhave proven sufficient. It is not sufficient to "pay" if

someone sues you, which has been industry's stand."... weneed the engineer to assume responsibility for doing it rightin the first place... and registration puts that responsibilityon him."6 Manufactured articles such as autos,refrigerators, tvs, gas furnaces, and microwave ovensimpinge directly on the daily lives of citizens. Faultyengineering in their design or fabrication may be just asdangerous as faults in bridges or buildings.

Expanded licensing might reduce the current cost ofproduct liability agencies. "Has the industrial exemption ofengineers and their engineering activity been in the public'sbest interest over the years? Two answers to the negativecan be seen in the recent enactment into Federal law of theOccupational Safety and Health Act (OSHA) and the U.S.Consumer Product Safety Act (CPSA). The public's electedlegislative representatives in Washington, D.C., deemed itto be in the public's safety interest to create these two verylarge agencies, provide millions of dollars in budgets, andplace thousands of government people to work in enforcingwhat industry through the years failed to do voluntarily-that is, to take steps necessary to provide safe workingconditions (OSHA) and to prevent unreasonably unsafeproducts from getting into the hands of consumers (CPSA).It appears now, using hindsight, that had industry'sengineers and their engineering activity not been exemptedfrom registration laws but rather had been legally boundand permitted to practice professional responsibility, theneed for OSHA and CPSA might well have been avoided.'

Con-argumerts againstexpanded licensing"Expanded 1.-censing would not increase public protection."

Within industry, engineering competency is demonstrateddaily on specific tasks. Business success depends oncompetent performance. It is inconceivable that licensingcriteria (edJcation, experience, exam) will be able to predictcompetence in generalized tasks adequately.

In the case of manufactured products, an abundance ofregulations, state and federal, already apply to theproducts, to the manufacturers, and to manufacturingprocesses. Corporations bear full legal responsibility for,and their survival depends on, the performance of theirproducts. Products are shaped by an extensive variety ofinfluences from many occupations. Licensing engineerswould not be beneficial; a survey conducted by NSPEindicated that its members believe the expanded licensingwould not improve the designs of consumer products."There is no significant and ciscernible threat to the publicposed by engineers in industry that is not far moreeffectively handled by product safety regulations.

"Expanded licensing would create dysfunction and increasedcosts."

Holding certain individual engineers legally responsiblewould result in their taking a protective -defensive posture,based on the potential of having to defend their judgmentand actions from a witness stand. Such a posture wouldreduce the efficiency of mult -discipline teamwork, which isessential to modern design.

With expanded registration requirements, assignment tocertain tasks could not be made solely on the basis ofcompetency, career development, reward, efficiency,availability, etc. Who is licensed and who is not alwayswould be an overriding concern.

Individuals and organizations would face increasedadministrative burdens, increased bureaucracy, and in -

71

Pro (continued)

"Expanded licensing of engineers in industry would be responsiveto the public need."

Licensing is an assurance for the public that engineersresponsible for designing equipment and providingengineering services that have potential public effect havemet at least minimum standards of competence. With theindustrial exemption, the public has no such assurance.Licensing is not a guarantee of the competence ofengineers any more than it is for medical practitioners. Itdoes, however, assure a minimum standard and provide fordisciplining offenses. Also, licensing creates a climateconducive to responsible actions. Licensed engineers havesworn oaths to be morally responsible in protecting thepublic's interest. They are very much aware that they arelegally and morally responsible to the public.

Licensing engineers in industry is entirely consistent withtheir expanding social consciousness. Today's engineer isvitally concerned and involved with consumerism,environmentalism, product safety litigation, and othersocial and legal forces. Engineers want to design productsand provide services that are inherently safe, but they aresometimes pressured by their employers to cut corners."The engineer attempting to practice ethically in industrytoday, whether a registered PE or an engineer professional,has virtually no legal or professional backing in the realworld when faced with taking responsible action to protectthe public, when this action is in disagreement with hisemployer. In effect, the so-called industrial exemption isdenying 82 percent of all engineers equal protection underthe same law that provides protection to a PE who practicesas a consultant."' Since the Model Law would require thatthe engineer not approve anything which, in his or herexperienced judgment, would be inconsistent with thepublic well-being, it would protect engineers in suchsituations. Given such backing, engineers would act ininstances where they cannot now act to provide additionalpublic protection.

Note that the "Model Law" requires licensing for only thoseengineers in "responsible charge." The objective is for thedesign process and final designs and decisions to beapproved by the judgment of licensed professionals.

"Expanded licensing will enhance the engineering profession."

"In the final analysis, whether engineering is to be fullyrecognized as a leading force in the nation's growth andprogress turns upon the status of engineering in theindustrial economy. Engineering as a whole cannot obtainits rightful place in the hierarchy of the professions until it isdefined and recognized on a legal basis, and that means theregistration of all qualified engineers in industry, wheremost of them are to be found."' Licensing carries with it asocial value and esteem. It enables an occupation to limitaccess to only those prescribed.

Con (continued)

creased costs. The state would have to administer licensingboards and disciplinary activities. Engineers would have toundergo additional educational and examination processesand pay license fees. Industry would have to deal withadditional record keeping and admiristrative reporting, plusreduced efficiency through lost flexibility. Ultimately, thesecosts would be borne by the public and by engineers. Costsfor license fees to U.S. engineers could amount to almost$200 million per year.22

"Many engineers and other technical professionals would be put ata disadvantage."

Graduates of non -engineering curricula would not beeligible for licensing. Many engineering graduates may notwish to obtain a license or may be unable to do so for manyreasons other than job competence. These individualswould never be in "responsible charge." Therefore,promotions, transfers, employment, and task assignmentswould be limited for them. Simply put, if you're content to bea subordinate engineer all your life, you won't need alicense. But if you aspire to responsibility and promotion,you would have to be licensed. Or, as put by Morton S. Fine,Executive Director, NCEE, "If the industrial exemption isremoved broadly, as is being proposed, then registrationwill be necessary to insure advancement, security, andmobility...."'9

Modern products are engineered by teams of professionalsfrom many disciplines for which licensing is not proposed(physics, chemistry, software, etc.). To arbitrarily subjugatesuch individuals to the direction and approval of a licensedengineer destroys the essence of effective teamwork, whichallocates influence to competence.

Older engineers would be forced to participate in a processdesigned primarily to screen new engineers wanting toenter the profession. "Grandfathering" may enable someengineers to obtain licenses who would not otherwise doso. But movement toward periodic recertification andrelicensing would eliminate this advantage. It is unlikelythat non -engineer professionals would be permitted to"grandfather.'"°

"Government regulation should be limited to necessity."

Legislative regulation of a profession should be used onlywhen no other satisfactory method exists for solving asignificant problem. Proponents have not establishedevidence for the following with regard to engineers inindustry: Have there been complaints of abuse? Howserious? How widespread? How many people involved?What harm is the public now suffering because engineers inindustry are not presently regulated? Would licensing haveprevented instances where abuse has been demonstrated?Would other remedies have worked as well?

Regulation of occupations has always been limited to directpractitioner -to -client or -public relationships. For example,regulation does not encompass physicians engaged in

72

Pro (continued)

A part of the licensing regulations orohibits others notlicensed frcm using the title of the occupation. Those nowusing the title engineer inappropriately would be preventedfrom doing so (e.g., stationary engineer, railroad engineer,service engineer).

-The industrial exemption is a mere political expediency."

There are many exemptions from required licensing -n thevarious state laws. Some are legal (engineers in the Federalgovernment), and some are policy (public utility engineers).Industrial exemptions, however, were originally grantedbecause state legislatures made political tradeoffs to in-dustry pressure in order to be able to enact engineeringregistration laws.' Therefore, industry exemptions neverwere based on nonpolitical reasoning.

It is inconsistent and illogical to require engineers to belicensed in order to design roads, bridges, tunnels, airportrunways and facilities, and radio and tv broadcastingbuildings and at the same time exempt from licenserequirements engineers who design the automobiles,trucks, trains, airplanes, and electrical and mechanicalequipmen: used by and in these facilities. When pol tics isset aside, the rational solution is the licensing of allengineers in responsible charge.

There is no expectation that universal registrat on ofengineers will be accomplished "overnight." The ModelLaw is proposed as a starting point on a state -to -state basis.Many clarifications and adjustments will have to be workedout through a trial -and -error process. However, it is arational process and it is important to start. Many com-panies encourage voluntary registration and thus admit toits having value.

Con (continued)

medical research, lawyers devoted to within -company legalactivity, or industrial accountants. Nor does it encompassplumbers and electricians em cloyed for piping or wiring onproducts. Regulation of direct -to -public relationships islegitimate because the public has no intermediate orintervening agency to protect its interests and governmen-tal regulation can be o significant, and perhaps :he only,protection.

Proponents are advocating the expansion of regulatorylaws into nontraditional coverage. Focus on the phrase"removing the industrial exemption" conveys the impres-sion that the industrial exemption is some sort of"loophole." It isn't. It is a needed and legitimate limitationthat brings engineering regulation into the same coveragefound with other regu ated c'ccupations-namely, direct -to -client relationships.

"Expanded licensing is sell -serving."

There are almost one million engineers in the U.S.; ap-proximately 30% are register= -d. Less than 7/s are membersof NSPE.9 Thus, a very small minority is pressinc stronglyfor states to impose legal restrictions on the majority.

It is typical for the biggest oromoters of licensing to bethose who are licensed:6 Their interests are very likely self-serving. NSPE's Position Paper openly admits to "... a dualpurpose and motivat on .. to protect the public health,safety, and welfare..." and "... there is no denying thatmuch of the impetus for universal registration of engineersderives from the true faith that engineering will not achieveits rightful place as a recognized profession unless and untilall of its practitioners and title holders have qualVied underthe legal standard of the law."'

Enhancing the status and prestige of an occupation is notan appropriate use of the police powers of a state.Arguments appealing to status so derived are improper.

ConclusionExpanded licensing would affect everyworking engineer either directly orindirectly. Some have classified tieproposed legislation as the most im-portant ever for engineers. At this po ntit is not predictable whether suchlegislation will be adopted.

Ed. Note' The terms "licensing" and"registration" are both used in referring tostate regulation. While there may be

technical differences, the two terms areusually used interchangeably in mostplaces.

In the artic e starting on the next page,RCA's official position on the issue isstated and explained.

BibliographyThis bibliography is arranged in threecategories: articles favoring expandedlicensing, articles opposing expandedlicensing, and articles that present in-formation or description without takinga strong position pro or con.

Favoring expanded licensing

1 National Society of Professional Engineers. Industrialexemption in registration laws for professional

engineer;: NSPE position paper. NSPE, Washington.D C , (Pv:ar 1976)

2. Nationa' Council of Engineering Examiners. A modellaw. Rev through August 1975, Morton S. Fine, Ex-ecutive Director. NCEE, P 0. Box 5000, Seneca, S C29678

3 "Industry exemption hit by changingpublic iew of technology uses, and engineeringactivism." Professional Engineer, (Sec 1975)

4. Eldon. Walter. "Eliminating the industrial exemptionfrom e"igineering registration laws. I." ProfessionalEngineer. (Jan 1976)

5 Kettler, G J , "The industry exemption from engineeringregistration laws-against the industry exemption,"Professional Engineer, (Mar 1976).

6 Lehnesis, Horace. "Professional engineering timefor update," Proc., ASEE 1976 College Industry Educa-tion Conference

1' Lunch, Milton F "The industry exemption and the drivefor unwersal registration." Professional Engineer. (Nov1974)

6. Nattress. Jr LeRoy Wm consideration of registra-tion and licensure,- prepared for the USAB of the IEEE.(Sep 1.3, 1977)

73

Opposing expanded licensing

9. National Electrical Manufacturers Association. NEMApolicy on professional engineer licensing. EngineeringDepartment Bulletin No. 80. National ElectricalManufacturers Association, (Aug 10, 1976).

10. Boulden, Larry F., "The politicians are stealing yourtitle," Automation, (Mar 1976)

11. Fax, David H.: "The case for the industrial exemption instate laws for the licensing of engineers." Proc., ASEE1976 College Industry Education Conf.

12. Kolhoff, Marvin J.; "Occupational freedom in engineer-ing within industry," IEEE Engineers Forum onEngineering Registration, Washington. D.C., (Sep 23.1977).

13. Kolhoff, Marvin J.; "Double jeopardy: proposed regula-tion of industry's internal engineering as well as productoutput," Commission on the future of the Society ofPlastics Engineers, Greenwich, Conn., (Sep 15. 1976).

14. Little, Jr.. Stanley M.. "The Boeing Company's positionregarding proposed revisions to the Professional

Engineers Registration Act, RCW Chapter 18.43," Sub-mitted to Subcommittee on Professional Engineers andLand Surveyors, Commerce Committee, WashingtonState Senate (Jun 16. 1976).

15. Maher, Robert G.. "Bell System position on engineeringlegislation." Internal report, (Oct 12, 1976).

16, Sims, Joe: "State regulation and the federal antitrustlaws.... The Justice Department's view of licensing,"National Council on Occupational Licensing (Aug 4.1975).

17. Young, J.F.: Engineering registration in manufacturingindustries. General Electric internal presentation. (Sep12, 1975).

Informational and descriptive

18. "Full steam ahead on registration? Notyet. Here's an update on the debate," The Institute. IEEE(Dec 1977)

19. Fine, Morton S., "Engineering registratior and the law,"IEEE Spectrum, (Oct 1977)

20. Hollander, Lawrence J., "The industry exemptions -forelimination." The Bent of Tau Beta Pi (Spring 1978).

21. Moench, Herman A.; "Eliminate the industry P.E. ex-emption?" Proc. ASEE, 1976 College Industry Educa-tion Conf.

22. Rubinstein, Ellis; "IEEE backs 'universal registration.'Where do you stand? Here are some pros and cons,"The Institute. IEEE, (Oct 1977).

23. Schindler. Louis. "Engineering registration laws: theexemption picture state by state," ProfessionalEngineer, (Oct 1976)

24. Shimber. Benjamin; "Mandatory licensing forengineers: the tough questions," ProfessionalEngineer, (Apr 1978).

25 Wright. Gayle N.: "Survey shows NSPE members favorremoval of the industrial exemption," ProfessionalEngineer. (Apr 1978).

26. "Response to the 1977 U.S. MemberOpinion Survey," The Institute, IEEE (May 1978).

What is it and how did we arrive at it?

RCA's position on licensing engineers

W.J. Underwood Faced with legislation that may remove the industryexemption from state engineering licensing laws, theCorporation has taken an official position.

The accompanying article described the issue and con-troversy concerning expanded mandatory licensing ofengineers. As indicated, large numbers of technicalprofessionals, both engineers and non -engineers, would besignificantly affected by enactment of NCEE's "Model Law."Also significantly affected would be the companies whoemploy those technical professionals and the public, whowould ultimately pay the increased cost. Because of itsimportance, RCA Corporate Engineering managementconsiders it desirable to adopt an official company positionon the issue and to work from this position in thecontroversy. That policy is given in the "box" on the nextpage.

We believe our Corporate position reflects the interests ofthe majority of RCA's technical staff. In the spring of 1977,as a part of the Corporate Engineering Information Survey,we asked RCA's professional technical staff, "How impor-tant do you think it will be in the future for RCA engineers toobtain a professional license?" The 2,290 responsesreceived to the question reflected a 75% cross-section ofRCA's engineering population. Results:

18% - Very Important or a Necessity82% - Slightly Important or Not Important

Those who already had state licenses (307) answered:

3O% - Very Important or a Necessity70% - Slightly Important or Not Important

Those who planned to become licensed (251) answered:

30% - Very Important or a Necessity70% - Slightly Important or Not Important

Any controversy usually has some merit on both sides. Werespect the interest in improving the professional status ofengineers and of improving the engineering profession. Wesuspect this goal accounts for most of the motivation toexpand licensing. We regard this position as the use of awrong method to attain a very desirable goal. We cannotsubscribe to "using" government regulation as a means ofstatus improvement.

RCA is justly proud of its technical staff, professional andsub -professional alike. RCA's professional technical non -engineers (chemists, physicists, mathematicians, computerscientists, etc.) have made major contributions to bothscience and engineering, many of world-wide repute. Theyhave held and continue to hold positions of significant andmajor responsibility. We would not subscribe to any actionwhich would diminish their status in any way.

RCA regards its technical staff as mature, responsiblepersons who do exercise integrity in their work. We regardour company in a similar manner. Therefore, we cannotsubscribe to a position that is based on an assumption ofirresponsibility. We believe that a careful study of productfailures would indicate an insignificant percentage causedby faulty engineering design that could be cured by

74

licensing. We regard arguments linking product failure tolack of licensing so overdrawn as to be unworthy of seriousconsideration.

RCA will continue supporting individuals who voluntarilyseek state licensing as a matter of their personal interest.We willingly conform to licensing requirements in anydirect client -practitioner relationship. We will, however,oppose attempts to force licensing on Close engineers nowpracticing under the very legitimate industrial exemption.

Bill Underwood is Director of Engineering ProfessionalPrograms for the Corporate Ergineering activity.Contact him at:Engineering Professional Prog3msCorporate EngineeringCherry Hill. N.J.Ext. 4383

Reprint RE -24.1-19Final manuscript received June 19. 1978.

Mandatory registration of engineers- RCA Corporat on position

RCA Corporation supports the licensing of engineers whooffer their services on a direct client -practitionerrelationship, and it supports the voluntary registration ofengineers employed by others where the engineers believethis would serve their personal interests. However, RCAopposes proposed revisions to state statutes that wouldrequire all "responsible" practitioners of engineering to belicensed. It believes such mandatory registration is not inthe best interests of the engineering profession, the public,or technological progress.

The proposed revision to the statutes governing registra-tion of engineers is based on a "model law" drafted by theNational Council of Engineering EKaminers. As broadlydefined in the "model law," the "practice of engineering,"for a technologically -based company, would likely includework performed by chemists, physicists, computerscientists, mathematicians, and others who are notgraduate engineers. However, the proposed requirementswould restrict licenses to graduates of an engineeringcurriculum and to those who pass an extensive engineeringexamination under state auspices. In addition, the "modellaw" would require licensing all those in "responsiblecharge" of engineering work.

For engineers in industry, the revised law would impose anunnecessary burden of time and money to obtain licensesand renewals to continue the work they have been doing foryears. For many it would require a -eeducation process. Itwould also limit leadership opportunities to the licensedengineers.

According to tne "model law " the purpose of mandatorylegislation is "...to safeguard life, health and property andto promote the public welfa -a." There is no convincingevidence that licensing engineers in industry would ac-complish ths. Industry is resconsible for its products andservices. Its professional technical people, regardless ofwhether they are licensed, a -e generally unknown to thepublic. It would be difficult, 1 not impossible, to pinpointindividual resoons bility in what is generally today a teameffort combining many specialized talents in many fields.

For industry, the proposed revisions would substantiallyrestrict flexibility in work assignments and responsibilities,and reduce internal efficiency without any compensatingbenefits. Additional administrative burdens would have tobe imposed to maintain control over who is licensed, inwhich disciplines and in which states, and personnel wouldhave to be asssigned accordingly. Given the complexity ofmodern technology, it would require artificial sign -off ondesigns even where it is impossible for any one person todetermine where his work began and left off.

Expanded licensing requirements would increase taxpayercosts for licensing and enforcement. They would introducecomplex bureaucratic procedures in industry; these costswould also ultimately have to be borne by the public.Because they are neither necessary nor in the publicinterest, RCA, therefore, opposes the proposed revisions.

75

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.

AUG 28-31, 1978 -Laser Applications andOptical Communication (Soc. of Photo -Optical Instrumentation Engineers) Townand Country, San Diego, CA Prog Info:SPIE, PO Box 10, Bellingham, WA 98225

SEP 5-8, 1978-COMPCON FALL (IEEE)Washington, DC Prog info: COMPCONFALL, P.O. Box 639, Silver Spring, MD 20901

SEP 5-7, 1978 -Intl. Optical ComputingConf. (IEEE) Imperial College, London,Eng. Prog Info: S. Horvitz, Box 274, Water-ford, CT 06385

SEP 6-8, 1978-Fiberoptic CommunicationConf. 8 Exhibit (Info. Gatekeepers Inc.)Hyatt Regency O'Hare, Chicago, IL ProgInfo: Information Gatekeepers, 167 CoreyRd., Suite 212, Brookline, MA 02146

SEP 6-8, 1978 -OCEANS '78 (IEEE, OEC,MTS) Sheraton Park, Washington, DC ProgInfo: Myra Binns, Marine Tech. Society,1730 "M" Street NW, Washington, DC 20036

SEP 12-14, 1978 -Western Electronic ShowConv.-WESCON (IEEE) Los Angeles

Convention Ctr., Los Angeles, CA Prog Info:W.C. Weber, Jr., 999 N. Sepulveda Blvd., ElSegundo, CA 90245

SEP 12-14, 1978 -Automatic SupportSystems for Advanced Maintainability(AUTOTESTCON) (IEEE) San Diego, CAProg Info: Bob Aquais, General DynamicsElectronic Div., Mail Stop 7-98, PO Box81127, San Diego, CA 92138

SEP 17-20, 1978 -Joint Fall Mtg. AmericanCeramic Soc. and IEEE Subcommittee onFerroelectricity (Am. Cer. Soc., IEEE) DallasHilton Inn, Dallas, TX Prog Info: Relva C.Buchanan, Dept. of Ceramic Engrg., 208Ceramics Bldg., U. of Illinois, Urbana, IL61801

SEP 21-23, 1978 -Interactive Techniques inComputer Aided Design (IEEE) Palazzo delCongressi, Fiera di Bologna, Italy Prog Info:Dr. Bertram Herzog, Computer Ctr., U. ofColorado, Boulder, CO 80303

SEP 24-27, 1978 -Electronic and AerospaceSystems Cony. (EASCON) (IEEE) SheratonIntl., Arlington, VA Prog Info: Bette English,At -Your Service, Inc., 821 15th Street NW,Washington, DC 20005

SEP 25-27, 1978 -Ultrasonics Symp. (IEEE)Cherry Hill Hyatt House, Cherry Hill, NJProg Info: F.S. Welsh, Bell Telephone Labs,555 Union Blvd., Allentown, PA 18103

OCT 1-4, 1978 -Design EngineeringTechnical Conf. (ASME) Radisson Hotel,Minneapolis, MN Prog Info: Technical Af-fairs Dept., ASME, 345 E. 47th St., New York,NY 10017

OCT 1-5, 1978 -Industry Application Socie-ty Annual Meeting (IEEE) Royal York Hotel,Toronto, Ont. Prog Info: W. Harry Prevey,4141 Yonge Street, Willowdale, Ont., MSP1N6

OCT 15-19, 1978-ISA Intl. Conf. and Ex-hibit (ISA) Philadelphia, PA Prog Info:Instrument Society of America, 400 StanwixSt., Pittsburgh, PA 15222

OCT 16-17, 1978 -Joint EngineeringManagement Conf. (IEEE) Regency,Denver, CO Prog Info: Henry Bachman,Hazeltine Corp., Greenlawn, NY 11740

OCT 16-18, 1978 -Electronics Conferenceand Natl. Communications Forum Chicago,IL Prog Info: National Electronics Conf.,Oak Brook Executive Plaza #2, 1211 W. 22St., Oak Brook, IL 60521

OCT 18-20, 1978 -Joint Automatic ControlConf., (IEEE) Civic Center, Philadelphia, PAProg Info: Dr. Harlan J. Perlis, New JerseyInstitute of Tech., 323 High Street, Newark,NJ 07102

OCT 18-20, 1978 -Canadian Communica-tions and Power Conf. (IEEE) QueenElizabeth Hotel, Montreal, P.Q. Prog Info:Jean Jacques Archambault, CP/P0 757,Montreal, Quebec H2L 4L6

OCT 21-25, 1978 -Engineering in Medicineand Biology (IEEE) Marriott Hotel, Atlanta,GA Prog info: Walter L. Bloom, M.D.,Georgia Institute of Tech., Atlanta, GA30302

OCT 23-25, 1978 -Digital Satellite Com-munications (IEEE) Hotel Reine Elizabeth,Montreal, Que. Prog Info: Marcel Perras,Teleglobe Canada, 680 Sherbrooke StreetW., Montreal, Que. H3A 2S4

OCT 23-25, 1978 -Fourth Intl. Conf. onDigital Satellite Communications, Montreal,Que. Prog Info: Mrs. Sara Bairn, Teleglobe,680 Sherbrooke St. West, Montreal, Que.H3A 2S4

OCT 24-26, 1978 -Biennial DisplayResearch Conf. (IEEE,SID) Cherry Hill Inn,Cherry Hill, NJ Prog Info: Lawrence Good-man, RCA Laboratories, Princeton, NJ08540

OCT 25-27, 1978-Inteiec (Intl. TelephoneEnergy Conf.) Sheraton Park, Washington,DC Prog Info: J.J. Suizzi, Bell Laboratories,Room 5D-178, Whippany, NJ 07981

OCT 29-NOV 3, 1978-SMPTE TechnicalConf. and Equipment Exhibit, AmericanaHotel, New York, NY Prog Info: SMPTE, 862Scarsdale Ave, Scarsdale, NY 10583

OCT 30-NOV 1, 1978 -SemiconductorLaser Conference (IEEE, QEA) Sheraton atFisherman's Wharf, San Francisco, CA ProgInfo: T.L. Paoli, Bell Laboratories, 600Mountain Ave., Murray Hill, NJ 07974

NOV 3-6, 1978 -Audio Engrg. Soc. Conf.(AES) New York, NY Prog Info: Almon H.Clegg, Panasonic Corp., One PanasonicWay, Secaucus, NJ 07094

NOV 6-9, 1978 -Advanced Techniques inFailure Analysis (IEEE) Marriott Hotel, LosAngeles, CA Prog Info: Robert J. Kolb, TRWDefense & Space Systems, 1 Space ParkR6/2184, Redondo Beach, CA 90278

DEC 4-6, 1978 -Intl. Electron DevicesMeeting (IEEE) Washington Hilton,Washington, DC Prog Info: Susan Henman,Courtesy Associates, 1629 "K" Street, NW,Washington, DC 20006

DEC 4-6, 1978 -National Telecom-munications Conf. (IEEE) Hyatt Hotel,Birmingham, AL Prog Info: H.T. Uthlaut, Jr.,South Central Bell., P.O. Box 771,Birmingham, AL 35201

DEC 12-14, 1978-MIDCON (IEEE) DallasConvention Ctr., Dallas, TX Prog Info: W.C.Weber, Jr., EEEI, 999 N. Sepulveda Blvd., ElSegundo, CA 90245

JAN 23-25, 1979 -Reliability and Main-tainability (IEEE) Shoreham Americana,Washington, DC Prog Info: D.F. Barber,POB 1401, Branch PO, Griffiss AFB, NY13441

Calls for papers

JAN 23-25, 1979 -ATE Seminar/Exhibit:Automated Testing for ElectronicsManufacturing Marriott, Los Angeles, CADeadline Info: 9/1/78 250 -word abs. toSheila Goggin, ATE Seminar/Exhibit, 1050Commonwealth Ave., Boston, MA 02215

SEP 23-26, 1979 -3rd World Telecom-munication Forum (ITU) Geneva,Switzerland Deadline Info: 9/30/78 100-200word abs. to A.E. Joel, Jr. Bell TelephoneLaboratories, Room 2C-632, Holmdel, NJ07733

76

AutomatedSystemsJ.E. FayATE operating systems, high orderlanguage and standardization-Industry/Joint Services Automatic Test Conf. andWorkshop, San Diego, CA (4/6/78)

J.C. HaggisATE test program development-ATESymp., U. of Alabama, Huntsville, AL(6/15/78)

R.E. HartwellATE for maintenance of non -electronicsystems-ATE Symp., U. of Alabama.Huntsville, AL (6/15/78)

R.J. MonisAutomated test system-ATE Symp., U. ofAlabama, Huntsville, AL (6/15/78)

J.A. MurnaneObservations and examples of ATEacquisition-Industry/Joint ServicesAutomatic Test Conf. and Workshop, SanDiego, CA (4/6/78)

S.P. PatrakisPRICE for design to unit product cost-PRICE Users' Meeting, San Francisco, CA(4/17/78)

E.W. RichterBroadband discriminators for ATEmeasurement of sideband noise-IEEETrans. on Instrumentation and Meas-urements (6/78)

Advanced TechnologyLaboratoriesM. CorringtonImplementation of fast cosine transformsusing real arithmetic-Natl. Aerospace &Electronics Conf., Dayton, OH (5/16-18/78)

A. FellerA fully automatic layout program for LSIdevices that optimizes speed per-formance-Lehigh Valley SemiconductorSymp., Bethlehem, PA (5/12-13/78)

Pen and Podium

Recent RCA technical papers and presentations

To obtain copies of papers, check your library or contact the author or his divisional Technical PublicationsAdministrator (listed on back cover) for a reprint. For additional assistance is locating RCA technical literature. contactRCA Technical Communications, Bldg. 204-2, Cherry Hill, N.J., extension 4256.

A. FellerA speed oriented fully automatic layoutprogram for random logic VLSI devices-Natl. Computer Conf., Anaheim, CA (6/5-8/78)

D. GandolfolJ. TowerL. ElliottlE. HerrmannProgress in programmablecorrelators-IEEE Workshop onSignal Processing, New York, NY16/78)

CCDCCD

(5/15 -

GovernmentCommunications SystemsC.D. FisherFrom ASPRs to DARs: a change or more ofthe same?-American DefensePreparedness Assoc., Technical Documen-tation Division, 20th Annual Meeting, NewOrleans, LA (6/7-9/78)

S. KlurmanA predictive analysis for optimum contour-after-wear for magnetic heads-IEEE Inter -magnetic Conf., Florence, Italy (5/10/78)

B. TyreeGround mobile forces tactical satellite SHFground terminals-Conf. Record, Intl. Conf.on Communications, Toronto, Canada (6/4-7/78)

GovernmentSystems Division StaffJ. HillibrandDesign tools for custom LSI -Industry/SAMS° Conf. and Workshop onMission Assurance, Los Angeles, CA (4/25-27/78)

LaboratoriesJ.J. HanaklV KorsunGranular metal barrier contacts inamorphous silicon solar cells-Proc., 13thIEEE Photovoltaic Specialists Conf.,Washington, DC (6/5-8/78)

P.H. RobinsonIR.V. D'AielloD. RichmanIB.W. FaughnanEpitaxial solar cells on low-cost siliconsubstrates-Proc., 13th IEEE PhotovoltaicSpecialists Conf., Washington, DC (6/5-8/78)

Missile andSurface RadarJ.A. BauerState of the art leadless chip carrierapplications for avionics packaging-IEEEComputer Packaging Symp., Skytop, PA(5/18/78)

M.W. BuckleyProject/program management-CourseDirector, Berkeley, CA (6/14-16/78)

M.W. BuckleyDecision making techniques for engineer-ing managers-Conf. Record, Electro (IEEEElectronic Show and Convention), Boston,MA (5/78)

C. GumacosWeighting coefficients for certain maximallyflat nonrecursive digital filters-Letter, IEEETrans. on Circuits and Systems (4/78)

D.L. PruittHybrid SCR switch-Proc., Thirteenth PulsePower Modulator Symp., Buffalo, NY (6/20-22/78)

D.L. PruittHigh voltage dc power conditioner-Proc.,Thirteenth Pulse Power Modulator Symp.,Buffalo, NY (6/20-22/78)

A. SchwarzmannThe high power performance of a 5 kW MICdiode phase shifter-Digest, GMTT Symp.,Ottawa, Ont. (6/29/78)

R.J. SmithDistributed microprocessing in radarsystems-Proc., 1978 IEEE Intl. Symp. onCircuits and Systems, New York, NY (5/78)

G.W. SunyReal-time simulation as a tool for systemdevelopment-Proc., Ninth Annual Model-ing and Simulation Conf., Pittsburgh, PA(4/78)

77

PatentsAdvanced TechnologyLaboratoriesM.L. LeveneCreating a closed image from segments -4088867

Astro ElectronicsC.A. BerardGuard circuit for high impedance signalcircuits -4091430

Automated SystemsW.J. HannanFormat for color diffractive subtractivefilters -4094584

Broadcast SystemsB.M. PradalIP.L. BuessFrequency linearization and sensitivityequalization of a frequency modulatedcrystal oscillator -4088968

Consumer ElectronicsS. MikoFerroresonant transformer structure -4088942

GovernmentCommunications SystemsL.V. HedlundChrominance signal transcodingapparatus -4093959

K.R. KellerForming patterned polycrystalline silicon -4090915

LaboratoriesF. AschwandenMaster oscillator synchronizing system -4092672

P.K. BaltzerParallel access memory system -4092728

S. Berkman IK.M. KimIH.E. TempleSi 3N4 coated crucible and die means forgrowing single crystalline silicon sheets -4090851

G. DenesRandom access -erasable read only memorycell -4095281

J. Dresnerl B. GoldsteinMagnesium oxide dynode and method ofpreparation -4088510

J. Dresnerl K.W. HangArticle with electrically -resistive glaze foruse in high -electric fields and method ofmaking same -4091144

M. EttenbergMulti -layer reflector for electroluminescentdevice -4092659 (assigned to J.S. Govern-ment)

A.M. GoodmanIC.E. WeitzelMethod of adjusting the leakage current ofsilicon -on -sapphire insulated -gate field-effect transistors -4091527

L.A. GoodmanLiquid crystal matrix display device withtransistors -4094582

W.H. GroenewegIA.V. Tumal L.A. HarwoodControlled oscillator with increased im-munity to parasitic capacitance -4095255

P.E. HaferlPincushion correction circuit -4088931

F.Z. Hawrylol H. KresselElectroluminescent semiconductor devicewith passivation layer -409501'

E.P. HerrmannCharge coupled device with diode reset forfloating gate output -4090095

A.C. IpriSilicon implanted and bombarded withphosphorus ions -4092209

H.F. Lockwood11-1. KresselStripe contact providing a uniform currentdensity -4092561

K.D. PetersDisc caddy and disc player systemtherefor -4093152

P.M. RussoPriority vectored Interrupt uing directmemory access -4090238

F.N. SechiConnection of a plurality of devicescircular waveguide-4091334

P.N. YocomRare earth activated lanthanum andlutetium oxy-chalcogenide-RE29662

Missile andSurface RadarM.L. BardashiC.P. ClasenlR.M. ScudderL.H. SimonIC.S. SorkinIR.O. YavneR.W. EkislA.I. MintzerGround -controlled guided missile system -4093153 (assigned to U.S. Government)

C.E. ProferaBeam forming network -4091387

Picture Tube DivisionF.C. Farmer, Jr.Electrical continuity test apparatus having aforward biased diode across the testterminals -4088947

R.L. BarbinConvergence apparatus for in -line beams -4091347

S.A. HarperPhotographic method for printing particlepattern with improved adherence utilizingvanadates-4089687

Solid State DivisionA.A. AhmedVoltage supply regulated in proportion tosum of positive- and negative -temperature -coefficient offset voltages -4095164

R.R. BrooksIJ.E. WojslawowiczDirection reversing direct current motorsand their control -4095155

L.F. Heckman, Jr.Coaxial cavity microwave oscillator withmanually acliustable capacitive feedback

to a element -4091337

W.W. SiekanowiczIC.H. AndersonT.L. CredelleFlat display device with beam guide -4088920

M. TodalS. OsakaSurface acoustic wave absorber -4090153

C.C. Wang1T.C. Lausman I R.F. BatesMethod of regenerating a lead monoxidetarget layer of a camera tube -4090758

C.F. Wheatley, Jr.Semiconductor device having symmetricalcurrent distribution -4091409

C.F. Wheatley, Jr.Voltage reference circuits -4088941

J.P. WittkeMethod of aligning optical fibers with lightemitting optical devices -4090777(assigned to U.S. Government)

M.V. HooverComplementary symmetry FET mixercircuits -4090139

L.A. KaplanCircuit for reducing ripple voltage -4092609

W.N. LewisCombination glass/low temperaturedeposited Si wfilx1-1yOz passivating overcoatwith improved crack and corrosionresistance for semiconductor devices -4091406 (assigned to U.S. Government)

J. 011endorflF.J. CestoneMethod of performing contactiessphotolithography -4088406

O.H. Schade, Jr.Amplifier circuits -4092612

H. SorkinProcess for filling dynamic scattering liquidcrystal cells -4091847

78

Engineering News and HighlightsRace is new EdRepat Consumer Electronics

Steve Race has been appointed an EditorialRepresentative for Consumer Electronics inIndianapolis. He is Manager, ManufacturingProjects, in Manufacturing Technology, andhas held test design and quality controlengineering and management positions inthe Bloomington and Juarez plants.

As EdRep, Steve will assist CE manufac-turing engineers with papers for the RCAEngineer and inform the editors of newdevelopments, professional activities,awards, pu5lications and promotions in hisarea.

Martin appointedto GSD post

Thomas A. Martin has been appointed aStaff Technical Advisor, Engineering, forthe Government Systems Division. He willadvise on the design of hardware andsoftware systems being produced forgovernment agencies. GSD's four businessorganizations are Missile and SurfaceRadar, Moorestown; Astro-Electronics,Princeton; Goverment CommunicationsSystems, Camden; and AutomatedSystems, Burlington. Since joining RCA in1970 he has held several positions in

systems management.

Promotions

AlascomRonald G. Weber from Senior Engineer toManager, Systems Engineering.

GovernmentCommunications Systems

Ray H. Brader from Senior Member,Engineering Staff, to Unit Manager, Com-munication Systems Engineering.

Kenneth J. Prost from Member, TechnicalStaff, to Unit Manager, Defense Com-munications Laboratory.

Edward Van Keuren from Unit Manager toManager, Command and Control Systems.

Astro-ElectronicsR. Manc.iso from Engineer to Manager(Specialty) Engineering.

Advanced TechnologyLaboratoriesAlba F. Cornish from Senior Member,Engineering Staff, to Unit Manager,Processor Design Group, Digital SystemsLaboratory.

Stanley E. Ozga from Senior Member,Engineering Staff, to Unit Manager,Processor Design Group, Digital SystemsLaboratory.

James E. Saultz from Manager, MarketingDevelopment, to Manager, Digital SystemsLaboratory.

Degrees granted

GovernmentCommunications Systems

David Sheby-PhD, Engineering Science;California Institute of Technology,Pasadena, CA.

Staff announcements

RCA LaboratoriesDavid D. Holmes, Director, TelevisionResearch _aboratory, appointed Robert H.Dawson, Manager, New TechnologyApolications Research.

Service CompanyEdward B. Campbell was appointedManager of Industrial Electronic ServicesMarketing.

Picture Tube DivisiorDonald R. Bronson, Division Vice President,Internet oval, appointed Gene W.Duckworth, Director, Operations, SovietEc uipment Contract.

Charles W. Thierfelder, Division Vice Presi-dEnt, Product Safety, Quality and Reliabili-ty appointed J. Paul Sasso, Administrator,Field Quality Program Coordination.

National BroadcastirgCompany, Inc.Julian Goodman, Chairman of the Board,announced the election of Fred Silvermanas President and Chief Executive Officerand a Member of the Board.

President andChief Executive OfficerEdgar H. Griffiths, President and ChiefExecutive Officer, announced the electioncf Herbert S. Schlosser, Executive VicePresident, to the Board of Directors.

Edgar H. Griffiths, President and ChiefExecutive Officer, announced the electionof Richard W. Sonnenfeldt to Vice President,' SelectaVision" VideoDisc Project.

Licensed engineers

When you receive a professional license,send your name, PE number (and state inwhich registered). RCA division, location,and telephone number to RCA Engineer,Bldg. 204-2, RCA, Cherry Hill, N.J. Newlistings (and corrections or changes toprevious listings) will be published in eachissue

Picture Tube Division

Lyons, K.C., Midland, Ont.; ONT-

79

Professional activities

Hempel on Critical TechnologyCommittee

Dan Hempel, Manager, Defense Com-munication Laboratory, GovernmentSystems Division, recently served as one often members of the Critical TechnologyExpert Group on Array Processors. Thisgroup of experts was formed at the sugges-tion of industrial leaders to aid the govern-ment in defining export controls on arrayprocessors.

University of Torontohonors Hillier

Dr. James Hillier, retired Executive VicePresident and Senior Scientist, will beawarded an honorary Doctor of Sciencedegree by the University of Toronto. He isbeing recognized for his pioneering work onthe electron microscope.

Kaye on IEEE committee

Leo Kaye, who is an engineer and scientiston the technical staff of Data SystemsEngineering, Automated Systems,Burlington, has been appointed to fill theStandards Affairs Seat of the SystemsTechnology Technical Interest Council(TIC), IEEE Computer Society. In this func-tion he will represent the TIC at StandardsCommittee meetings, ensure that eachTechnical Committee has appropriaterepresentation on the Standards Com-mittee, identify those areas within each TCthat requires standards activity, assist theTC Chairman as needed, and in generalpromote all TIC standards related activities.

Good things happen ...

Two artists who were responsible fordesigning RCA Engineer covers haverecently received professional recognition.

Andy Whiting received a third -place ribbonin the annual Society of CommercialPhotographers of Delaware Valley printcompetition for his photo of an ear of cornwith two hybrid circuits (our Aug/Sep 1977cover.) Andy is the Leader at PhotographicServices of Missile and Surface Radar,Moorestown.

Bob Canary, who designed the cover for thecommunications issues (Oct/Nov, 1977 andDec/Jan, 1977-78) was recently promotedfrom Illustrating Leader of the ATLEngineering Communications group toAdministrator, Proposal and PresentationGraphics, Government Systems Division,Marketing.

Burlington has Technical Recognition Dinner

Forty-one engineers and scientists and theirspouses from Automated Systems,Burlington, were recently honored at thatactivity's annual Authors' and PatentAwards Dinner. Of the forty-one people, tenhad published papers in the RCA Engineerin 1977.

At the Technical Recognition Dinner, DickHanson, the Manager of Design Engineer-ing, Non -Electronic Test Engineering gothis ten -patent -award; Steve Hadden, aSenior Member of the Technical Staff of thesame group, got his five -patent -award.Shown are (I. to r.) Harry J. Woll, Div. V.P.and General Manager; Dick Hanson; SteveHadden; and John Regan, Vice President,Patent Operations.

GCS honors authors/inventors

Government Communications Systems recently held its annual Authors and Inventorsreception in Camden. Of the 85 who attended, fifty-five were people who wrote or presentedpapers or filed patent disclosures in 1977.

CommunicationAward-Chet Sall, anEditorial Represen-tative for the RCAEngineer, who recent-ly retired after 35 yearswith RCA, wasawarded this plaqueby the IEEE. He is acharter member of thatorganization's Groupon Professional Com-munications.

. . -a. a la titsn tron. CopoofC' 0", "'.,....a'.

CHESTER SALL

. ,ocogotioon of on, ...NO and .natdottasts,wte consincuous by Os ...oh dogma of

ol 000nol,...1 and qualdy as furthest's,

onfoccouts botenson ononeers

tn consunsat olecttontcs by ha litany

,oars of wince as tddat.

4.4,n ..,44%-Net lira

80

Editorial RepresentativesContact your Editorial Representative, at the extensions listed here, to scheduletechnical papers and announce your professional activities.

Commercial CommunicationsSystems Division

Broadcast SystemsBILL SEPICH* Camden, N.J. Ext. 2156KRISHNA PRABA Gibbsboro, N.J. Ext. 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

Cablevision SystemsJOHN OVNICK* N. Hollywood, Cal. Ext. 241

Government Systems Division

Astro-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. 3081HARRY KETCHAM Camden, N.J. Ext. 3913

Government EngineeringMERLE PIETZ* Camden, N.J. Ext. 2161

Missile and Surface RadarDON HIGGS* Moorestown, N.J. Ext. 2836JACK FRIEDMAN Moorestown, N.J. Ext. 2112

Solid State DivisionJOHN SCHOEN* Somerville, N.J. Ext. 6467

Power DevicesHAROLD RONAN Mountaintop, Pa. Ext. 633-827SY 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

Consumer ElectronicsCLYDE HOYT*Indianapolis, Ind. Ext. 5208FRANCIS HOLT Indianapolis, Ind. Ext. 5217PAUL CROOKSHANKS Indianapolis, Ind. Ext. 5080STEVE RACE Indianapolis, Ind. Ext. 5636DON WILLIS Indianapolis, Ind. Ext. 5883

SelectaVision ProjectROBERT MOORE Indianapolis, Ind. Ext. 3313

RCA Service Company

JOE STEOGER* Cherry Kill, N.J. Ext. 5547RAY MacWILLIAMS Cherry Hill, N.J. Ext. 5986DICK DOMBROSKY Cherry Hill, N.J. Ext. 4414

Distributor andSpecial Products DivisionCHARLES REARICK' Deptford, N.J. Ext. 2513

Picture Tube DivisionED MADENFORD* Lancaster, Pa. Ext. 3657NICK MEENA Circleville, Ohio Ext. 228JACK NUBANI Scranton, Pa. Ext. 499J.R. REECE Marion, Ind. Ext. 566

Alascom

PETE WEST' Anchorage, Alaska Ext. 7657

Americom

MURRAY ROSENTHAL' Kingsbridge Campus, N.J. Ext. 4363

GlobcomWALT LEIS* New York, N.Y Ext. 3655

RCA Records

JOSEPH WELLS* Indianapolis, Ind. Ext. 6146

NBC

BILL HOWARD* New York, N.Y. Ext. 4385

Patent OperationsJOSEPH TRIPOLI Princeton, N.J. Ext. 2992

Research and Engineering

Corporate EngineeringHANS JENNY' Cherry Hill, N.J. Ext. 4251

LaboratoriesMAUCIE MILLER Princeton, N.J. Ext. 2321LESLIE ADAMS Somerville, N.J. Ext. 7357

'Technical Publications Administrator, responsible forreview and approval of papers and presentations.

RCA EngineerA technical journal published by Corporate Technical Communications"by and for the RCA engineer"

Printed in USAForm No RE -24-I